Large Igneous Provinces:

Origin and Environmental

Consequences

Thousand-metre basalt
cliffs, Kivioqs Fjord,
East Greenland.
PHOTO IAN PARSONS

Andrew D. Saunders1

pisodically, the Earth erupts large quantities of basaltic magma in

geologically short periods of time. This results in the formation of large
igneous provinces, which include continental flood basalt provinces,
volcanic rifted margins, and giant oceanic plateaus. These fluctuations in the
Earths system are still poorly understood. Do they owe their origin to mantle
plumes, meteorite impacts, or lithosphere-controlled processes? Whatever
their origin they correlate closely with major changes in oceanic and
atmospheric chemistry and may trigger global mass extinctions.

the main LIP, are an important

part of the story, but will not be
considered in detail here.

The locations and ages of the main

LIPs are shown in FIGURE 1. This
selection is biased: it does not
include the large silicic provinces
(e.g. Chon Aike in southernmost
South America or the Sierra Madre
in Mexico); the majority of the
LIPs in Figure 1 are predominantly
KEYWORDS: Continental flood basalts, oceanic plateaus, mass extinctions, basaltic. It is also ageist: it does not
mantle convection and temperature include any LIP older than 250 Ma,
for example, the Emeishan
Province in China (Permian) and
INTRODUCTION
the numerous Proterozoic and Archaean LIPs. It is imporIt has been nearly 15 years since the term large igneous
tant to stress that LIP formation has occurred throughout
province was introduced by Mike Coffin and Olaf Eldholm
Earth history and not just in the Mesozoic and Cenozoic,
(1991, 1994). An umbrella term to include continental
although they may occur in cycles (e.g. Ernst et al. 2005);
flood basalt provinces, oceanic plateaus, volcanic rifted
there is an increase in LIP formation in the Cretaceous (Larmargins and aseismic ridges, it rapidly entered common
son 1991), and Prokoph et al. (2004) have suggested cycles
parlance even though it was, and remains, loosely defined.
of LIP formation.
The key aspect of large igneous provinces (LIPs) is that they
represent anomalously high magmatic fluxes. The magma The information database is also strongly skewed in favour
is usually basaltic, but may be rhyolitic. They are large in of the continental flood basalt provinces. Accessibility to
area, covering many thousands if not millions of square the deeper parts of the dissected and faulted volcanic pile
kilometres, and they testify to unusual geological processes, means that a greater range of compositions and a wider
involving large amounts of thermal energy. Where this range of ages can be sampled in the continental sequences
energy comes from deep within the Earth as a mantle than in the oceanic plateaus. Some oceanic plateaus have,
plume, from a meteorite impact, or from sinking of dense however, subsequently collided with a volcanic arc or a
roots from the base of the continental crust or lithosphere continental margin, with the result that important infor(delamination) is a matter of considerable debate. As will mation can be obtained from the deeper crustal sections.
be seen from the papers in this issue of Elements, there is Thus, the collision of the Caribbean Plateau with South
also debate about their environmental effects. Could the America has provided a wealth of information about its petformation of such provinces cause the collapse of ecosys- rogenesis (Kerr et al. 1997). Similarly, the collision of the
tems, either by interrupting oceanic circulation systems or Ontong Java Plateau with the Solomon Islands allows us to
by releasing large masses of volcanic aerosols, and trigger walk through the top three kilometres of the plateaus
mass extinctions? Certainly, the timing of LIPs and mass basaltic crust. Otherwise we would be totally reliant on
extinctions suggests some causality, but we do not, as yet, cored material from the top few hundred metres of crust
metaphorically, pin-pricking the elephant (Tejada et al.
understand its nature.
2004). The basements of many other plateaus are, however,
sampled entirely by drilling and dredging (e.g. Kerguelen
LIPS AND LIPS
Plateau) or remain unsampled.
No two LIPs are the same. Just as Read (1948) recognised
that there are Granites and Granites, the term LIPs encompasses a wide range of geological structures and processes. FORMATION
For the purpose of this issue, we are focusing on the conti- There are almost as many theories and models for the fornental flood basalt provinces and their oceanic equivalents, mation of LIPs as there are individual provinces. I have
the oceanic plateaus. Volcanic trails (forming aseismic attempted to summarise these models, and some of the preridges across the ocean floor), which often lead away from dictions that arise, in TABLE 1. It is likely that no single
model can account for all LIPs, and the predictions from
each model are not fully understood or known. Most models agree that large amounts of thermal energy are required
1. Department of Geology
University of Leicester
in order to produce large volumes of magma over a geologLeicester LE1 7RH, UK
ically short period of time. Given that most LIPs are
E-mail: ads@leicester.ac.uk

ELEMENTS, VOL. 1,

PP.

259263

259

D ECEMBER 2005

Map showing the distribution of the main Mesozoic and

Cenozoic large igneous provinces (continental flood
basalts and basaltic oceanic plateaus), modified after Coffin and Eldholm (1991). Also included are present-day hotspots (red spots), which
may be related to individual LIPs (e.g. Runion to the Deccan Traps;
Galapagos to the Caribbean Plateau) via plate reconstructions and
volcanic chains. The Siberian Traps are shown buried (striped ornamentation) beneath the West Siberian Basin. CAMP: Central Atlantic Magmatic Province, located along the eastern edge of North and South
America, and western edge of North Africa and southern Europe. Cretaceous Plateaus include the Hess and Shatsky Rises, and Ontong Java
and Manihiki Plateaus.

FIGURE 1

basaltic, this requires some form of energy source in the

mantle. The required energy may be reduced slightly if the
mantle source is highly fertile (see glossary). This would be
the case if, for example, it contains large amounts of eclogite (see glossary and Anderson this issue) or is volatile rich.
What is the source of this energy? And how much is
required? Over time, the worlds oceanic ridge system supplies a remarkably constant amount of basaltic magma. The
thickness of the ocean crust ignoring the sediments is
very uniform, at about seven kilometres. There are differences. Very slow-spreading ridges, such as the Southwest
Indian Ridge, create anomalously thin crust, to the point
where it may even be absent. Similarly, ocean crust near
major transform faults may be thin. And, conversely, in
some areas (remarkably rare), such as Iceland, the crust is
anomalously thick (perhaps as much as 35 km). But the
bulk of the ocean crust is broadly uniform in thickness and
composition, which is remarkable given that two important
variables source temperature and source composition
can have a dramatic effect on the volume and type of basalt
produced. An increase of 100C in the potential temperature (see glossary) of the mantle source will more than double the amount of melt produced, and hence double the
thickness of the ocean crust. The source temperature of normal mid-ocean ridge basalts is unlikely, therefore, to vary
significantly.

ELEMENTS

So how do we generate the high crustal thicknesses found

in LIPs (typically 35 km for an oceanic plateau)? There are
several ways of doing this. First, we can increase the temperature of the source. A straightforward increase in source
potential temperature from 1300C to 1500C can produce
a 30+ km thick layer of melt. This is at the heart of the
plume model (White and McKenzie 1989; Campbell this
issue), where heat energy is transferred in a mass of mantle
ascending from a thermal boundary layer deep in the Earth
(e.g. the coremantle boundary). Second, we can increase
the rate at which source material is processed through the
zone of partial melting. Rather than passive upwelling (as is
thought to occur at mid-ocean ridges), the mantle rock
actively convects into and through the zone of partial melting. Combined with higher temperatures, this provides a
potent model for large-volume melt generation, and is
again implicit in the plume model. Rapid fluxing may be
particularly important during the start-up phase of the
mantle plume (Richards et al. 1989; Campbell this issue),
when the LIP is created, but it is also a key feature of the
edge model, discussed below. The impact model (Jones
this issue) also invokes high source temperatures, induced
by kinetic energy following meteorite impact. And third, we
can increase the fertility and volatile content of the source
to create more melt. None of these three factors temperature, mantle ascent rates, and source composition are
exclusive; indeed there is every reason to believe they may
occur together.
What is the evidence for high mantle temperatures during
LIP formation? The most direct evidence is the occurrence
of highly magnesian melts (preserved in high-Mg basalts,
picrites, and komatiites). These are found in several LIPs
but, importantly, not all, and this has been used as evidence
against an excessively hot mantle source (Anderson this
issue). To counter this, it should be remembered that magnesian melts are more dense than normal basalt and may be
trapped in magma chambers in the deep crust; in effect,
they are filtered out. Thus, absence of evidence for picrites,
the products of crystallization of magnesian melts, is not
necessarily evidence for the absence of them.

260

D ECEMBER 2005

TABLE 1
PREDICTIONS ARISING FROM VARIOUS MODELS OF LIP FORMATION

Prediction

Regional
uplift/doming

High-T magmas
(e.g. picrites
and komatiites)

Extraterrestrial
material and
impact breccias

Hotspot trail
leading from LIP

Currently
active hotspot

Model

Excess
mantle-derived
magmatism

Mantle plume

Yes

Likely

Likely

No

Likely

Possible

(unless the plume

impinges on the base
of thick lithosphere)

(before and/or
during magmatism)

(but dense melts may

not reach the surface)

Yes

Possible

Possible

Meteorite impact

Likely

Likely

Yes

(during magmatism)

(probably abundant)

Unlikely

No

Unlikely

Unlikely

Unlikely

No

Unlikely

Unlikely

Unlikely

No

Unlikely

Unlikely

Edge model, with

enhanced mantle
convection

Possible

Likely

(if mantle can ascend

sufficiently to
decompress and melt)

(during magmatism)

Delamination

Possible

Likely

(if mantle can ascend

sufficiently to
decompress and melt)

(could be substantial
during or after
magmatism

Possible

Unlikely

Melting of
fertile mantle
without excess
heat

Arguments against very high mantle temperatures are supported by limited uplift in the vicinity of some LIPs. A hot
mantle source beneath the lithosphere would be expected
to cause significant uplift, especially if it were dynamically
emplaced by a mantle plume (Campbell this issue), but in
some provinces the evidence for such uplift is ambiguous
(Anderson this issue). Readers may wish to read the recent
work by Burov and Guillou-Frottier (2005), however, which
suggests that the amount of uplift above a plume may be
small, or absent, in some circumstances.
If mantle plumes or impacts are not the main generators,
what other mechanisms may be responsible for LIP formation? King and Anderson (1995) noted that many continental flood basalt provinces lie close to the edges of
Archaean cratons. They proposed that thermal insulation
by the craton raises the temperature of the underlying
asthenosphere, which then flows sideways out from
beneath the craton and under thinner lithosphere. As it
ascends, the mantle decompresses and melts. In this model
the mantle need not be as hot as in the plume model. Furthermore, King and Anderson (1998) argued that edgedriven convection, where a secondary convection cell is
established at the craton margin, could increase magma
production rates and volumes. An alternative model, but
also involving displacement of upper mantle, is the delamination model, in which dense lower crust and the attached
lithospheric mantle sink into the mantle, and upper mantle flows into the ensuing space, decompressing and melting (Elkins-Tanton 2005; Anderson this issue). This model
predicts substantial uplift as the lithosphere rebounds, but
again does not necessarily produce high-Mg melts. It is also
debatable whether the required volumes of magma, and
magma of the right composition, could be produced in this
way from normal-temperature asthenospheric mantle.
My personal view is that most, if not all, LIPs can be
explained by mantle plumes, and that the best evidence
picritic rocks for the high source temperatures is often
hidden in the deeper crust or upper mantle. Variations in
source composition doubtless play a role in the composition and amounts of liquid that are generated, but at the
heart of the model is a mechanism for the release of

ELEMENTS

thermal energy originating from deep within the Earth. The

lithosphere plays a crucial role, capping (even preventing)
melting and redirecting the hot mantle towards thin spots.
Some LIPs may be entirely driven by lithospheric processes,
and some may be impact generated, but the majority of
them appear to be plume generated.

ENVIRONMENTAL EFFECTS
AND MASS EXTINCTIONS
There are several ways in which a LIP could affect the global
environment. One is through the immediate, eruptive
release of gas and aerosols. As shown by Self et al. (this
issue), a large basaltic lava eruption can release prodigious
quantities of SO2, CO2 and halogens, the effects of which
we are only beginning to appreciate. The initial release of
SO2 and its injection into the stratosphere could trigger a
global volcanic winter, akin to the models of nuclear winters, reducing photosynthesis through light occlusion and
cooling. Long-term accumulation of CO2 may lead to subsequent warming a volcanic summer especially if the
biologically driven carbon-capture mechanisms are compromised by the preceding volcanic winter. However, as
pointed out by several workers, including Self et al. (this
issue) and Wignall (this issue), the average flux of CO2
released by a LIP over its entire history is not large much
less than the current annual production of anthropogenic
CO2, for example.
The evidence that there is a link between LIPs and the environment is indicated by the close coincidence between LIPs
and mass extinctions (FIG. 2), as noted by Vincent Courtillot in 1994 and subsequently developed in his book Evolutionary Catastrophes (1999). But how does a LIP, or flood
basalt event, trigger a mass extinction? What other indicators are there of climate change? One is a rapid shift in the
carbon isotope record at the time of the extinctions. Such
isotopic variations indicate massive changes in seawater
and atmospheric composition, requiring the addition of billions of tonnes of carbon to the atmosphere and ocean
reservoirs. This carbon, it is argued, as CO2 in the atmosphere, leads ultimately to powerful global warming, loss of
habitat, and mass extinction (Kiehl and Shields 2005). But

261

D ECEMBER 2005

Extinction rate versus time (continuous line, blue field)

(multiple-interval marine genera, modified from Sepkoski
1996) compared with eruption ages of continental flood basalts (red
bands). Three of the largest mass extinctions, the Permo-Triassic, Triassic
Jurassic and the CretaceousTertiary, correspond to eruptions of the

Siberian Traps, the Central Atlantic Magmatic Province (CAMP), and the
Deccan Traps, respectively. Three oceanic plateaus, the Caribbean (CP),
Kerguelen (KP), and Ontong Java (OJP), are shown. Modified after White
and Saunders (2005).

where does this carbon come from? Some undoubtedly

comes from the basalts themselves but, given the low average rates of CO2 production, this is unlikely to be the entire
story. An alternative, mentioned by both Wignall and Kerr
in this issue, is that the greenhouse effects of CO2 from the
LIPs slowly raise the atmospheric and oceanic temperatures,
and this triggers release of methane previously trapped in
permafrost and methane hydrates on the seafloor. In effect,
a threshold is reached, potentially leading to a runaway
greenhouse. (Intriguingly, it has recently been reported that
Siberian permafrost is melting due to anthropogenically
driven global warming; perhaps the flood basalts offer a
model for current climate change.) An alternative explanation is that near-surface intrusions that accompany LIP formation are injected into carbon-rich sedimentary layers
(methane- or coal-bearing) and that these then release their
carbon into the ocean and atmosphere (Svensen et al. 2004;
McElwain et al. 2005).

modifying gases and aerosols. 40Ar/39Ar and zircon U/Pb

dating offer increasingly precise methods for improving
this knowledge, but even a precision of better than 0.1%
still leaves a lot to be desired. Mass extinction events may
occur in periods of 100,000 years or less, which is still
outside the precision offered by the best radiometric techniques for dating events that occurred during the mass
extinctions at the Permo-Triassic, TriassicJurassic, and
CretaceousTertiary boundaries.

FIGURE 2

The key to understanding these processes is knowing the

duration and flux rates of LIP magmatism, because from
these we can calculate the flux rates of the climate-

REFERENCES
Burov E, Guillou-Frottier L (2005) The
plume head-continental lithosphere
interaction using a tectonically realistic
formulation for the lithosphere.
Geophysical Journal International 161:
469-490
Coffin MF, Eldholm O (eds) (1991) Large
Igneous Provinces: JOI/USSAC Workshop
Report. The University of Texas at Austin
Institute for Geophysics Technical Report
114: 79 pp

ELEMENTS

FURTHER READING
The International Association of Volcanology and Chemistry of the Earths Interior (IAVCEI) has established the LIPs
Subcommision, which maintains a webpage with up-todate information about studies on large igneous provinces.
See http://www.largeigneousprovinces.org
For information on specific provinces, see Mahoney JJ,
Coffin MF (1997) Large Igneous Provinces: Continental,
Oceanic, and Planetary Flood Volcanism. American Geophysical Union Monograph 100, 438 pp. .

Coffin MF, Eldholm O (1994) Large

igneous provinces: crustal structure,
dimensions, and external consequences.
Review of Geophysics 32: 1-36
Courtillot V (1994) Mass extinctions in the
last 300 million years: one impact and
seven flood basalts? Israeli Journal of
Earth Sciences 43: 255-266
Courtillot V (1999) Evolutionary Catastrophes: The Science of Mass Extinctions.
Cambridge University Press, 188 pp

262

Elkins-Tanton LT (2005) Continental

magmatism caused by lithospheric
delamination. In: Foulger GR, Natland
JH, Presnall DC, Anderson DL (eds)
Plates, Plumes and Paradigms. Geological
Society of America Special Paper 388, pp
449-462
Ernst RE, Buchan KL, Campbell IH (2005)
Frontiers in large igneous province
research. Lithos 79: 271-297
Kerr AC, Tarney J, Marriner GF, Nivia A,
Saunders AD (1997) The CaribbeanColombian Cretaceous igneous province:

D ECEMBER 2005

Glossary
13C Carbon isotope values (and other
stable isotopes such as oxygen and
hydrogen) are expressed relative to a reference standard. The standard used for
carbon is a Peedee Formation belemnite
(PDB). The difference from the standard
is expressed as the delta function, which
may be positive (i.e. the carbon has a relatively higher abundance of heavy 13C
than the standard), the same, or negative
(a higher proportion of light 12C). Results
are expressed in parts per thousand. A
shift or spike in the seawater isotope
curve indicates a geologically rapid
change in the relative amounts of light
and heavy carbon. Organic carbon, especially biogenic methane, has low 13C
values; the 13C of marine carbonate is
about zero.
Basalt A basic igneous rock of volcanic
origin with between 45 and 52 wt% SiO2
and less than 5 wt% total alkalis. The
mineralogy typically comprises clinopyroxene (augite) and plagioclase feldspar.
Olivine, and an opaque mineral such as
magnetite may also be present. The plutonic equivalent of basalt is gabbro.
Delamination The collapse and peeling of
large layers of dense material from the
base of the lithosphere or crust. The
delamination process allows the
asthenosphere to ascend into the resulting space, triggering decompression
melting and magmatism.
Eclogite A high-pressure, high-density
metamorphic rock composed mainly of
garnet and clinopyroxene. It is the highpressure equivalent of basalt or gabbro.
Emplacement of basaltic magma into the
lower crust may lead to the formation of
dense eclogite, which may become
buoyantly unstable and collapse into the
underlying mantle (delamination). Subducted ocean crust is thought to convert
to eclogite and be entrained in the mantle,
eventually returning to the near-surface
by convection processes.

the internal anatomy of an oceanic

plateau. In: Mahoney JJ, Coffin M (eds)
Large Igneous Provinces: Continental,
Oceanic, and Planetary Flood Volcanism.
American Geophysical Union, Geophysical Monograph 100: 123-144
Kiehl JT, Shields CA (2005) Climate simulation of the latest Permian: implications
for mass extinction. Geology 33(9):
757-760
King SD, Anderson DL (1995) An
alternative mechanism of flood basalt
formation. Earth and Planetary Science
Letters 136: 269-279
King SD, Anderson DL (1998) Edge-driven
convection. Earth and Planetary Science
Letters 160: 289-296
Larson RL (1991) Latest pulse of Earth:
Evidence for a mid-Cretaceous superplume and geological consequences of
superplumes. Geology 19: 547-550
McElwain JC, Wade-Murphy J, Hesselbo SP
(2005) Changes in carbon dioxide during
an oceanic anoxic event linked to
intrusion into Gondwana coals. Nature
435: 479-482

ELEMENTS

Flow field The total lava products of one

effusive eruption, however long-lasting.
Mantle fertility The relationship between
mantle composition and its ability to produce melt. During partial melting, peridotite (the main mantle rock) can produce only so much melt before it
exhausts its supply of basalt producing
elements, such as Ca and Al. The more of
these elements present in the original
rock, the more melt can be produced it
may be said to have an increased fertility.
The presence of eclogite, which is chemically equivalent to basalt, substantially
increases the fertility of the source. Note,
however, that energy, in the form of
latent heat of melting, is still required to
generate melt. Increasing source fertility
will not substantially increase the volume
of melt unless that energy is also present.
Optical depth A measure of how opaque
a medium such as air is to the radiation passing through it. Solar radiation is
partially scattered and absorbed by fine
particles in the atmosphere, and so the
amount of incident light is always greater
than the amount of transmitted light. A
completely transparent medium has an
OD of zero. An OD of 1 results in ~40%
of light reaching the ground.
Peridotite An ultrabasic rock, with a mineralogy dominated by olivine and with
variable amounts of clinopyroxene (e.g.
diopside) and orthopyroxene (e.g. enstatite). Garnet or spinel may also be present. It is the predominant rock in the
Earths mantle. Peridotite comprising
mostly olivine and orthopyroxene is
termed harzburgite. Peridotite with
olivine, orthopyroxene and clinopyroxene is termed lherzolite.
Picrite A rock of volcanic origin with
between 12 and 18 wt% MgO, less than
3 wt% total alkalis (K2O + Na2O), and an
SiO2 content between 30 and 52 wt%.
Picrites are more magnesian and more

Prokoph A, Ernst RE, Buchan KL (2004)

Time-series analysis of large igneous
provinces: 3500 Ma to present. Journal
of Geology 112: 1-22
Read HH (1948) Granites and granites.
In: Gilluly J (ed) Origin of Granite,
Geological Society of America Memoir
28: 1-19
Richards MA, Duncan RA, Courtillot VE
(1989) Flood basalts and hot-spot tracks:
plume heads and tails. Science 246:
103-107
Sepkoski JJ (1996) Patterns of Phanerozoic
extinction: a perspective from global data
bases. In: Walliser OH (ed) Global Events
and Event Stratigraphy, Springer, Berlin,
pp 35-51

primitive than basalts and may be

indicative of a high-temperature parental
magma.
Plume buoyancy flux A measure of the
strength of a plume, given by the difference in density between the plume mantle
and the surrounding mantle, multiplied
by the buoyancy-driven volume flux of
the plume.
Potential temperature Rising, convecting
material (plastic mantle rock, or melt, or
air) cools slightly by adiabatic decompression. This defines an adiabatic cooling line (or, conversely, a heating line if
the material descends) part of the
Earths geotherm. For rock, the adiabatic
gradient is about 0.5C km-1. The theoretical intersection of this line with the
Earths surface is called the potential temperature. Thus, mantle with an actual
temperature of 1400C at 100 km depth
will have a potential temperature of
1400 (0.5 100) = 1350C, assuming
that the adiabatic gradient is linear.
Potential temperature, or Tp, is a convenient shorthand to describe how hot
the mantle is regardless of depth or, put
another way, how much energy it contains. Unfortunately we can only approximate the actual Tp of the upper mantle.
Some workers (e.g. Anderson this issue)
argue that the normal upper mantle has
a large temperature range, varying from
place to place by 100C, whereas others
argue that the normal mantle is much
more restricted in temperature, with a Tp
of about 1300C, and with localised
hotspots (plumes) where the Tp may
exceed 1500C.
Viscosity The resistance to flow within a
liquid (alternatively, a measure of the
internal friction of a liquid, or dynamic
viscosity). Kinematic viscosity is the
dynamic viscosity of a liquid divided by
its density.

the Ontong Java plateau from Pb-Sr-HfNd isotopic characteristics of ODP Leg
192 basalts. In: Fitton JG, Mahoney JJ,
Wallace PJ, Saunders AD (eds) Origin
and Evolution of the Ontong Java
Plateau. Geological Society of London
Special Publication 229: 133-150
White R, McKenzie D (1989) Magmatism
at rift zones: the generation of volcanic
continental margins and flood basalts.
Journal of Geophysical Research 94 (B6):
7685-7729
White RV, Saunders AD (2005) Volcanism,
impact and mass extinctions: incredible
or credible coincidences? Lithos 79: 299316 .

Svensen HS, Planke S, Malthe-Srenssen A,

Jamtveit B, Myklebust R, Eidem TR, Rey
SS (2004) Release of methane from a
volcanic basin as a mechanism for initial
Eocene global warming. Nature 429:
542-545
Tejada MLG, Mahoney JJ, Castillo PR, Ingle
SP, Sheth HC, Weis D (2004) Pin-pricking
the elephant: evidence on the origin of

263

D ECEMBER 2005

Large Igneous Provinces

and the Mantle Plume
Hypothesis

Columnar jointing in a
postglacial basalt flow at
Aldeyarfoss, NE Iceland.
PHOTO JOHN MACLENNAN

Ian H. Campbell1

antle plumes are columns of hot, solid material that originate deep
in the mantle, probably at the coremantle boundary. Laboratory
and numerical models replicating conditions appropriate to the
mantle show that mantle plumes have a regular and predictable shape that
allows a number of testable predictions to be made. New mantle plumes are
predicted to consist of a large head, 1000 km in diameter, followed by a
narrower tail. Initial eruption of basalt from a plume head should be preceded
by ~1000 m of domal uplift. High-temperature magmas are expected to
dominate the first eruptive products of a new plume and should be concentrated near the centre of the volcanic province. All of these predictions are
confirmed by observations.
KEYWORDS: mantle plume, large igneous

INTRODUCTION
The plate tectonic hypothesis provides an elegant explanation for Earths two principal types of basaltic volcanism,
mid-ocean ridge and island arc volcanism, both of which
occur at plate boundaries. Mid-ocean ridge basalts form
new ocean crust along the tensional zones that develop
where adjacent plates, with divergent motions, are pulled
apart, and island arc magmas form along regions of compression, where plates sink back into the mantle. However,
a third significant form of volcanism occurs away from
plate boundaries and therefore cannot be explained by
plate tectonics. The most volumetrically significant of these
are continental flood basalts, giant oceanic plateaus, and
aseismic ridges. Continental flood basalts and giant oceanic
plateaus, their oceanic equivalent, are massive outpourings
of basalt that erupt in 1 to 5 Myr. They cover an equidimensional area typically 20002500 km across (White and
McKenzie 1989). Collectively they are referred to as Large
Igneous Provinces (LIPs). Aseismic ridges are chains of volcanoes that stretch across the sea floor. FIGURE 1 shows the
Deccan Traps, a typical flood basalt, and the Chagos Lacadive
RidgeMascarene Plateau, a typical aseismic ridge. Notice
that the Deccan Traps are connected by the 200300 km
wide ChagosLacadive Ridge, across the CarlsbergCentral
Indian Ridge spreading center, through the Mascarene
Plateau to an active volcano at Runion. The plume
hypothesis attributes flood basalts and giant oceanic
plateaus to the melting of the large spherical head of a new
plume (Richards et al. 1989; i.e. Campbell and Griffiths
1 Earth Chemistry
Research School of Earth Sciences
The Australian National University
ACT 0200 Australia
and
Institute for Study of the Earths Interior,
Okayama University at Misasa, Tottori, 682-0193, Japan
E-mail: Ian.Campbell@anu.edu.au

ELEMENTS, VOL. 1,

PP.

265269

1990) and aseismic ridges, like the

ChagosLacadive Ridge, to the
melting of a plume tail (Wilson
1963; Morgan 1971).

THE MANTLE PLUME

HYPOTHESIS

Convection in fluids is driven by

buoyancy anomalies that originate
in thermal boundary layers.
Earths mantle has two boundary
layers. The upper boundary layer is
the lithosphere, which cools
provinces, uplift, picrite through its upper surface. It eventually becomes denser than the
underlying mantle and sinks back
into it, driving plate tectonics. The lower boundary layer is
the contact between the Earths molten ironnickel outer
core and the mantle. High-pressure experimental studies of
the melting point of ironnickel alloys show that the core
is several hundred degrees hotter than the overlying mantle.
A temperature difference of this magnitude is expected to
produce an unstable boundary layer above the core which,
in turn, should produce plumes of hot, solid material that
rise through the mantle, driven by their thermal buoyancy.
Therefore, from theoretical considerations, mantle plumes
are the inevitable consequence of a hot core.
The material in the lower boundary layer will be lighter
than the overlying mantle, but before it can rise at a significant rate, it must gather enough buoyancy to overcome the
viscosity of the mantle that opposes its rise. As a consequence, new plumes have a large head followed by a relatively narrow tail (FIG. 2). The tail or feeder conduit is comparatively narrow because hot, relatively low-viscosity
material following up the existing pathway of the tail
requires less buoyancy to rise than the head, which must
displace cooler, high-viscosity mantle. As the head rises
through the mantle, it grows for two reasons (Griffiths and
Campbell 1990). First, material in the high-temperature,
low-viscosity tail rises faster than the head and feeds a constant flux of hot mantle into the head of the plume. When
this material reaches the stagnation point at the top of the
plume, it flows radially with a spiraling motion to give the
head its characteristic doughnut shape. Second, the head
grows by entrainment. As the plume rises, heat is conducted into the adjacent mantle. In the boundary layers
adjacent to the head and tail, the temperature increases and
the density decreases, eventually becoming the same as the
plume; the boundary layers then begin to rise with the
plume. This material becomes part of the plume and is
swept into the base of the head by its recirculating motion.
The head is therefore a mixture of hot material from the

265

D ECEMBER 2005

Flattened plume heads should be

2000 to 2500 km in diameter
The diameter of a plume head (D) depends on the temperature difference between the plume and the adjacent mantle (, the plumes excess temperature), its buoyancy flux
(Q), the kinematic viscosity of the lower mantle (), and its
height of rise (), as described in equation (1)
D = Q1/5(/g )1/52/53/5

(1)

where g is gravitational acceleration, is the coefficient of

thermal expansion of the mantle, and is its thermal conductivity (Griffiths and Campbell 1990). Note that the
plume height of rise is raised to the power 3/5, whereas
most other terms are raised to the power 1/5. Therefore the
height of rise of the plume, which in the case of Earth is the
depth of the mantle, is the dominant factor influencing the
size of a plume head. If of a plume is assumed to be
300C and its buoyancy flux to vary between 3 103 and
4 104 N s-1, the calculated diameter of a plume head originating at the coremantle boundary is 1000 to 1200 km.
The plume head should flatten to produce a disk with a
diameter between 2000 and 2400 km when it reaches the
top of the mantle. The calculated time that a plume head
takes to rise from the coremantle boundary to the top of
the mantle is about 100 Myr.

Map of the western Indian Ocean showing the distribution

of volcanic rocks associated with the RunionDeccan
plume. The Seychelles were part of the Deccan Traps prior to separation
caused by spreading on the CarlsbergCentral Indian Ridge. Note that
the Deccan Traps are connected via the 200300 km wide Chagos
Lacadive Ridge, across the Carlsberg Ridge spreading center, to the
Mascarene Plateau and eventually to Runion Island, the current position of the plume. Between 60 and 40 Ma, the Runion plume was
located under the Carlsberg Ridge. It produced a volcanic ridge on both
sides of the spreading center, before leaving the ridge and appearing to
backtrack on itself towards Runion, which is its current position (adapted
from White and McKenzie 1989).

FIGURE 1

plume source region and cooler entrained mantle (FIG. 2A).

When the plume head reaches the top of its ascent, it flattens to form a disk with a diameter twice that of the head
(FIG. 2B). Note that growth of the plume head as it rises
through the mantle occurs because mantle in the plume tail
rises faster than the plume head, which is a direct consequence of the strong temperature dependence of the mantles
viscosity.

Plumes must originate from a hot boundary

layer, probably the coremantle boundary

PREDICTIONS AND OBSERVATIONS

The plume hypothesis makes the following testable
predictions.

New plumes consist of a large head

followed by a small tail
Flood basalts and oceanic plateaus, the oceanic equivalent
of flood basalts, are the first eruptive products of a new
mantle plume. The volumes of basalt produced during
these events are enormous, and Richards et al. (1989) have
shown that eruption rates of flood basalts are one to two
orders of magnitude higher than those of the associated
ocean island chain, which connects them to the current
position of the plume. This observation fits well with the
plume hypothesis, which attributes the high eruption rates
of flood basalts to melting of plume heads and the lower
eruption rates of ocean island chains to melting of narrower
plume tails.
ELEMENTS

When a plume head rises beneath continental crust, the

associated buoyancy anomaly lifts the lithosphere, placing
it under tension. This can lead to runaway extension and to
the formation of a new ocean basin (Hill 1991). During the
initial stages of rifting, the hot mantle in the underlying
plume head is drawn into the spreading centre to produce
thickened oceanic crust. If the line along which the continent splits lies close to the centre of the plume head, as was
the case when the North Atlantic opened above the Iceland
plume head, the length of thickened oceanic crust should be
equal to the diameter of the plume head 2000 to 2400 km.
FIGURE 3 shows the zone of thickened oceanic crust along
the east Greenland coast and the RockallVring plateaus,
which are associated with the break-up of the North
Atlantic. Notice that the zone of thickened oceanic crust on
both sides of the North Atlantic is ~2400 km long. When a
new oceanic ocean basin opens above a plume head, the
first oceanic crust to form is always anomalously thick, typically at least twice as thick as normal oceanic crust. Other
examples of plume-related thickened oceanic crust are the
margins of South America and Africa above the Paran
plume and India above the Deccan plume (White and
McKenzie 1989). Anomalous thickening of oceanic crust
does not occur when a new ocean basin forms away from
the path of a plume.

The obvious way to show that plumes originate from the

coremantle boundary is to use seismic methods to trace
plume tails from the top of the upper mantle to their
source. However, the small diameter of plume tails (100 to
200 km in the upper mantle, although wider in the lower
mantle) makes them difficult to resolve and, as a consequence, attempts to use seismic methods to image plume
tails have met with little success. A notable exception is the
pioneering use by Montelli et al. (2004) of a new finitefrequency technique to resolve plume tails. They have used
this method to trace the Ascension, Azores, Canary, Easter,
Samoa, and Tahiti plumes to the coremantle boundary. A
number of other plumes, including the Iceland plume, disappear in the lower mantle, and the Yellowstone plume has
no resolvable signature. Montelli et al. (2004) note that it is
more difficult to image plumes in the lower mantle than in

266

D ECEMBER 2005

the upper mantle, which may explain the discrepancies

between the Montelli et al. (2004) images and those
expected from plume theory. Nevertheless the Montelli et
al. method shows great promise and may eventually allow
unambiguous imaging of plume tails in both the upper and
lower mantles.

Both heads and tails should erupt

high-temperature picrites
The of mantle plumes can be estimated from the maximum MgO content of their erupted magmas because, for
dry melts, there is a linear relationship between the MgO
content and magma temperature. As a rough rule, a 4 wt%
increase in MgO in the melt equates to a 100C increase in
mantle potential temperature (see glossary). The maximum
MgO content of plume-derived picritic (high-Mg basaltic)
liquids varies between 18 and 22 wt%, suggesting that the
temperature excess for mantle plumes is between 150C and
250C. Examples of volcanic provinces that have been
attributed to plumes and that include high-MgO picritic
melts are RunionDeccan, Paran, North Atlantic Province,
Karoo, Emeishan, GalapagosCaribbean, and Hawaii.

The temperature excess of a plume head is

highest at the centre of the head and decreases
towards the margin
When new oceanic crust opens up above a plume head, the
thickness of the oceanic crust produced is dependent on the
temperature of the mantle that is drawn into the new midocean ridge spreading centre. The plume hypothesis predicts that the temperature of a plume head should be highest near the plume axis, where the tail continues to rise
through the centre of the head (FIG. 2). At the center of the
head, is expected to be 300 100C. The in the
remainder of the head, which is a mixture of hot material
from the boundary layer source and cooler entrained lower
mantle, varies with the plume buoyancy flux but must be
less than at the centre. For typical plume buoyancy fluxes,
the average is ~100C. Hopper et al. (2003) have used
seismic reflection and refraction data, obtained from four
traverses located between the plume head and its margin, to
determine the thickness of the first oceanic crust to form
when the North Atlantic opened above the Iceland plume.
They obtained thicknesses of 33 and 30 km for traverses T-I
and T-II, close to the plume axis, and 18 and 17 km for traverses T-III and T-IV, closer to the margin of the head (see
FIG. 3). The required to produce these crustal thicknesses, based on the work of McKenzie and Bickle (1988), is
350C and 100C at the centre and margin of the head,
respectively, which is consistent with the plume hypothesis.

Picrites should erupt early during flood

volcanism and be most abundant near the
centre of the plume head and less abundant
towards the margin
The hottest material in the head is the mantle from the
plume source (the dark colored fluid in FIG. 2), which is
300 100C hotter than the entrained mantle. Although
this temperature difference decreases with time, as adjacent
layers exchange heat, the high temperatures of the hot layers will persist for millions to tens of millions of years,
depending on their thickness, which in turn depends on
their distance from the plume axis (see FIG. 2). When a
plume head melts to form a flood basalt province, only the
top of the head ascends to a level in the mantle where the
pressure is low enough to allow melting. Note in FIG. 2 that
the hot layer at the top of the plume thickens towards the
centre, where it grades into the tail. Provided the first magmas for a new plume do not pond and fractionate in crustal
magma chambers, picrites should dominate the early melting products of plume heads and become less abundant as
the cooler, second layer ascends to a level where it can
begin to melt. Picrites should also be most abundant
towards the hot centre of the plume head and become less
abundant towards the margins. The predicted early picrites
have been documented for the ParanEtendeka, Deccan,
Emeishan, North Atlantic, and Karoo flood basalts. Examples of picrites that are abundant at the centre of a flood
basalt province and less abundant towards the margins are
the Letaba picrites, as seen along the Lobombo monocline
of the Karoo, and those in the deeply dissected valleys of
the Emeishan flood basalt province in China.

Photograph of a laboratory model of a starting thermal

plume (A) mid-way through its ascent and (B) after the
head flattens at the top of its ascent. The dark fluid represents hot material
from the plume source and the lighter fluid is cooler entrained material.
White arrows show motion within the plume and black arrows the direction of motion in the boundary layer adjacent to the plume; the boundary layer has been heated by conduction so that its density is approximately the same as that of the plume (after Griffiths and Campbell
1990).

FIGURE 2

ELEMENTS

Flood volcanism should be preceded by domal

uplift of 500 to 1000 m at the center of
the dome
The arrival of the hot plume head in the upper mantle will
produce domal uplift at the surface, the magnitude of
which depends on its average temperature. The area of
maximum uplift is predicted to have a radius of ca. 200 km
and to be surrounded by a zone, with a radius of ca. 400 km,
in which uplift is still significant (FIG. 4).

267

D ECEMBER 2005

sibly exceed 1000 m; an intermediate zone with a radius of

425 km, in which the average uplift is ca. 300 m; and an
outer zone with a radius 800 km, in which uplift is minimal.
The magnitude and shape of the uplift agree remarkably
well with that predicted by Griffiths and Campbell (1991)
(FIG. 4) and Farnetani and Richards (1994). Other examples
of uplift prior to volcanism include the Wrangellia province
of northwest Canada and southwest Alaska, the Natkusiak
province in northwest Canada, the Deccan flood basalt of
India, and the Ethiopian flood basalt. Uplift started 3 to 5 Myr
before the Emeishan and Wrangellia volcanism.

THERMOCHEMICAL PLUMES

Geological map of the North Atlantic region after Hopper

et al. (2003). Areas shaded black correspond to onshore
basalts, which erupted from the plume head prior to the opening of the
North Atlantic. The dark grey areas locate offshore seaward-dipping
reflectors, indicating areas of thickened oceanic crust. Light grey area
with vs between the Rockall Plateau and the Vring Plateau shows the
areal extent of basalts associated with the initial opening of the ocean.
T-I to T-IV are seismic transects along the East Greenland coast (from
Hopper et al. 2003). BTP: British Tertiary Province; FIR: FaeroesIceland
Ridge; GIR, GreenlandIceland Ridge; JMR: Jan Mayen Ridge; JMFZ: Jan
Mayen Fracture Zone.

FIGURE 3

An obvious weakness of the thermal plume hypothesis is

that it fails to explain a number of minor volcanic chains
that stretch across the ocean basin and cannot be linked to
LIPs. These appear to be the product of plume tails without
heads. It has been suggested that the plumes responsible for
these volcanic chains may originate in the mid-mantle.
Theoretically the mantle could divide into two convecting
layers separated by a boundary layer at the 670 km seismic
discontinuity; this boundary layer is a potential source of
plumes, which would rise a small distance and therefore
have small heads. However, seismic studies and numerical
models showing that slabs and plumes, respectively, can
pass through the 670 km discontinuity make this interpretation unlikely. A recent computer model of thermocompositional plumes (Farnetani and Samuel 2005) provides a
possible solution. They have shown that if the head of a
weak plume contains a high proportion of a dense mantle
component (e.g. subducted former basaltic crust), the
ascending head can stall at the 670 km discontinuity and
separate into low- and high-density components. Only the
light component penetrates the discontinuity and gives rise
to a weak plume with a smaller temperature excess than the
strong plumes associated with LIPs. The suggestion that
plumes may have an above-average mantle concentration
of dense, subducted former oceanic crust is consistent with
the usual interpretation of D a heterogeneous, seismically fast layer of variable thickness (300 300 km) at the
base of the mantle as a zone where subducted slabs have
concentrated. It is also consistent with the isotopic (both
stable and radiogenic) and trace element characteristics of
many plume-derived magmas, which also require a higher
than average mantle concentration of former oceanic crust.

CONCLUSION

Uplift above a plume head, as predicted by Griffiths and

Campbell (1991), compared with the uplift observed at
the centre of the Emeishan flood basalt (shown in pink) by He et al.
(2003). The timing and uplift shape predicted by Farnetani and Richards
(1994) is similar, but they predict more uplift because they model a
plume head with a higher excess temperature: 350C as opposed to
100C. Predicted profiles are given for maximum uplift (t = 0), when the
top of the plume is at a depth of ~250 km, and 2 Myr later (t = 2 Myr),
when flattening of the head is essentially complete. The uplift for the
Emeishan flood basalt province is the minimum average value for the
inner and intermediate zones as determined from the depth of erosion
of the underlying carbonate rocks.

FIGURE 4

The best-documented example of domal uplift occurs in

association with the Emeishan flood basalt province in
China (He et al. 2003). Here the carbonate beds of the
underlying Maokou Formation have been systematically
thinned by erosion towards the centre of the flood basalt
province. Isopachs of the Maokou Formation show that
uplift was broadly dome shaped as expected. He et al. recognize three zones: an inner zone with a radius of ca. 200 km,
where uplift is estimated to be at least 500 m and could posELEMENTS

The mantle plume hypothesis provides a simple explanation for all of the essential features of classic LIPs, such as
the Deccan Traps and the North Atlantic Igneous Province,
and all of its predictions have been confirmed by observation. Five days of intense scrutiny during The Great Plume
Debate (Fort William, 28 Aug1 Sept 2005) failed to land a
telling blow on the mantle plume chin, and no creditable
alternative emerged to explain the principal features of
LIPs, especially their high eruptive volumes and high-temperature magmas. However this does not mean that all
intraplate volcanoes and volcanic chains are due to plumes.
Each case must be considered on its own merits.

ACKNOWLEGMENTS
I thank Ross Griffiths and Guust Nolet for their comments
on the manuscript and Charlotte Allen for drafting the
diagrams. .

268

D ECEMBER 2005

REFERENCES
Campbell IH, Griffiths RW (1990)
Implications of mantle plume structure
for the evolution of flood basalts. Earth
and Planetary Science Letters 99: 79-93
Farnetani CG, Richards, MA (1994)
Numerical investigations of the mantle
plume initiation model for flood basalt
events. Journal of Geophysical Research
99 (B7): 13813-13833
Farnetani CG, Samuel H (2005) Beyond the
thermal plume paradigm. Geophysical
Research Letters 32: L03711, doi:
10.1129/2005GL022360
Griffiths RW, Campbell IH (1990) Stirring
and structure in mantle starting plumes.
Earth and Planetary Science Letters 99:
66-78
Griffiths RW, Campbell IH (1991)
Interaction of mantle plume heads with
the Earths surface and onset of smallscale convection. Journal of Geophysical
Research 96 (B11): 18295-18310

ELEMENTS

He B, Xu Y-G, Chung S-L, Xiao L, Wang Y

(2003) Sedimentary evidence for a rapid
kilometer-scale crustal doming prior to
the eruption of the Emeishan flood
basalts. Earth and Planetary Science
Letters 213: 391-405
Hill IR (1991) Starting plumes and
continental break-up. Earth and
Planetary Science Letters 104: 398-416
Hopper JR, Dahl-Jensen T, Holbrook WS,
Larson HC, Lizarralde D, Korenaga J, Kent
GM, Kelemen PB (2003) Structure of the
SE Greenland margin from seismic
reflection and refraction data: Implications for nascent spreading centre
subsidence and asymmetric crustal
accretion during North Atlantic opening.
Journal of Geophysical Research 108(B5):
2269, doi:10.1029/2002JB001996
McKenzie D, Bickle MJ (1988) The volume
and composition of melt generated by
extension of the lithosphere. Journal of
Petrology 29: 625-679

269

Montelli R, Nolet G, Dahlen FA, Masters G,

Engdahl ER, Hung S-H (2004) Finitefrequency tomography reveals a variety
of plumes in the mantle. Science 303:
338-343
Morgan WJ (1971) Convective plumes in
the lower mantle. Nature 230: 42-43
Richards MA, Duncan RA, Courtillot VE
(1989) Flood basalts and hot-spot tracks;
plume heads and tails. Science 246:
103-107
White R, McKenzie D (1989) Magmatism
at rift zones: The generation of volcanic
continental margins and flood basalts.
Journal of Geophysical Research 94 (B6):
7685-7729
Wilson JT (1963) A possible origin of the
Hawaiian islands. Canadian Journal of
Physics 41: 863-870 .

D ECEMBER 2005

Large Igneous Provinces,

Delamination, and
Fertile Mantle

Columbia River basalts,

Washington State. PHOTO
STEPHEN SELF

Don L. Anderson1

hen continental crust gets too thick, the dense eclogitic bottom
detaches, causing uplift, asthenospheric upwelling, and pressurerelease melting. Delamination introduces warm blocks of lower
crust with a low melting point into the mantle; these eventually heat up,
ascend, decompress, and melt. The mantle below 100 km depth is mainly
below the melting point of dry peridotite, but its temperature will be above
the melting point of recycled fertile (basaltic or eclogitic) components,
obviating the need for excess temperature to form hotspots or melting
anomalies. When plates pull apart or delaminate, the mantle upwells;
entrained crustal fragments of various ages are fertile and create melting
anomalies along developing mid-ocean ridges, fracture zones, and old suture
zones. Eclogites associated with delamination are warmer and less dense than
subducted oceanic crust and more susceptible to melting and entrainment.
KEYWORDS: LIPs, delamination, plumes, hotspots, lithosphere, eclogite

INTRODUCTION
Large igneous provinces (LIPs) are generally attributed to
hotter-than-normal mantle. It is important, therefore, to
know the normal range of mantle temperatures. Convection calculations for a fluid with mantle-like properties that
is heated internally and cooled from above predict temperature fluctuations of at least 100C (Anderson 2000). Geophysical evidence suggests that the mantle temperature
under most LIPs was in this normal range while the LIPs
were erupting (Clift 2005; Korenaga et al. 2002), and, where
measured, the present heatflow is also normal (i.e. appropriate for the age of the underlying crust). The sedimentary
records from a range of swells and plateaus of various ages
from all major ocean basins (North Atlantic, Mid-Pacific
Mountains, Shatsky Rise, Ninetyeast Ridge, Ontong Java
Plateau) are compatible with eruption of magma from mantle in the normal range of temperature, followed by conductive cooling of the type associated with regular oceanic
crust. The North Atlantic Ocean is particularly shallow, but
modeling by Clift (2005) indicates that the depth is consistent with a temperature anomaly of +100C or less. Furthermore, he shows many LIP sites where the depth to the
base of the sediments implies colder than average mantle
temperatures.
The largest LIPs (Siberian Traps, Ontong Java Plateau) were
arguably erupted at, below, or near sea level rather than at
an elevation of one or two kilometres above the surrounding terrain, as predicted by the plume hypothesis. LIP uplift,
if it occurs, appears to be syn- or post-volcanism, rather

1 Seismological Laboratory
Caltech
Pasadena, CA 91125, USA
E-mail: dla@gps.caltech.edu

ELEMENTS, VOL. 1,

PP.

271275

than millions of years prior to volcanism, as predicted by

thermal models. There is no indication from uplift, heatflow, or seismic tomography for mantle temperatures more
than 100C above the mean under these LIPs (Czamanske
et al. 1998; Korenaga 2005; Roberge et al. 2005; Gomer and
Okal 2003).
The crust under continental LIPs is thinner than average
continental crust (Mooney et al. 1998). Many LIPs occur
atop deep sedimentary basins, in back arcs, or on old convergent margins. The chemistry of LIPs is highly variable,
with compositions ranging between mid-ocean ridge basalt
(MORB) and ocean island basalt; many continental LIPs
show clear evidence of input from continental crust or lithospheric mantle. Picritic melts are rare. These observations are
all enigmatic in the context of the usual high-temperature
or plume explanations for LIPs. The alternative explanation
is that the mantle is hotter than generally assumed (Anderson
2000) or is compositionally heterogeneous on a large scale,
or both.

GONDWANA, ATLANTIC,
AND INDIAN OCEAN LIPS
The breakup of Gondwana was preceded by extensive volcanism along the future Atlantic and Indian ocean margins
(FIG. 1). Volcanism continued during breakup, along the
continental margins and on the separated fragments (in
and around the North Atlantic and on and near Madagascar, Kerguelen, and the Rio Grande Rise; see FIG. 1). Plate
reconstructions (Mller et al. 1993) show that the currently
continent-hugging plateaus (~1000 km offshore) were
mainly formed at ridges and triple junctions in the newly
opened Atlantic and Indian oceans, some tens of millions
of years after breakup of the supercontinent. This can be
illustrated by calculating the approximate half-width of the

271

D ECEMBER 2005

South America and in the wake of the drifting continents;

they are not connected to chains of volcanic islands. An
unstable stress regime, plate reorganizations, and complex
triple junction jumps may be responsible for the formation
of Pacific seamount fields and plateaus (Natland and Winterer 2005). The implications are that the mantle is close to
the melting point and is variable in composition and melting point (FIG. 2) and that magmatism is focused by lithospheric architecture and stress.

CONSTRAINTS FROM SEISMIC

TOMOGRAPHY

Distribution and ages of LIPs in the Gondwana hemisphere. Also shown are the ages of continental breakup
and uplift. This figure illustrates the geometry and relative timing of
supercontinent breakup and magmatism, and LIPs in the newly opened
ocean. Delamination of lower crust and consequent asthenospheric
upwelling may be responsible for the LIPs in the continents (ages are
given in small black numbers), which usually occur in mobile belts, arcs,
suture zones and accreted terranes. If the continents move away from
the delamination sites, it may be possible to see the reemergence of fertile delaminated material in the newly formed ocean basins (e.g. oceanic
plateaus; red patches), particularly where spreading ridges allow
asthenospheric upwellings. CFB: continental flood basalt. Continents are
a mosaic of fragments, some of which are shown with thin and thick
lines as borders. The names of the CFB and oceanic LIPs are left off for
clarity (see http://www.largeigneousprovinces.org/ and http://www.
mantleplumes.org/).

FIGURE 1

Some LIPs have small-diameter, seismic low-velocity zones

(LVZ) in the upper 200350 km, rarely deeper, of the underlying mantle (e.g. Christiansen et al. 2002; Allen and Tromp
2005). LVZs are regions in the mantle where seismic waves
are slowed. This happens because the mantle is hot and/or
partially molten or has a mineralogy different from that of
the normal ambient mantle. For example, eclogite in the
mantle has a low melting point (FIG. 2) and can have slow
seismic wave speeds (e.g. FIG. 3 in Anderson 1989a). It is
usually assumed that these LVZs are related to the overlying
volcanism and that they are thermal in nature. But some
ancient LIPs, such as Paran and the Ontong Java Plateau
(OJP), retain seismic low-velocity zones in the upper
200300 km of the mantle (Vandecar et al. 1995; Gomer
and Okal 2003) even though they have traveled thousands
of kilometres away from the putative source and the mantle has had more than 120 Myr to cool. The OJP low-velocity zone slows seismic waves down but does not attenuate
their amplitude significantly, as would be the case if the
mantle were hot or partially molten (Gomer and Okal
2003). Compositional, rather than thermal, effects are
implied (e.g. as in FIG. 3). Both upwelling asthenosphere
and sinking eclogite can have low seismic velocities. Sinking eclogite, however, melts at lower temperatures than
peridotite (FIG. 2). (Note that a decrease in seismic velocity
is not the same thing as attenuation. Cold material, such as
cold eclogite, can have a low seismic velocity but can be

ocean at the time of peak LIP volcanism, based on magnetic

anomaly maps and ages obtained from the plateaus. Some
of these plateaus (and half-widths of the ocean, in kilometres) are as follows: Azores (1300), Bermuda (1100), Cape
Verde (900), Crozet (1000), Discovery (2000), Iceland
(1000), Jan Mayen (900), Madagascar (1200), Mozambique
(800), Rio Grande Rise (1200), Broken Ridge (1100), Kerguelen (1100), and Walvis Ridge (1200). The delay between
continental breakup and the main stage of plateau formation is usually about 2050 Myr (FIG. 1). Although some
LIPs are currently intraplate, they formed, without known
exception, at plate or craton borders, at triple junctions, or
on a spreading ridge about 1000 km offshore from newly
separated continents.

THE PACIFIC PLATEAUS

The Pacific plate originated as a roughly triangular
microplate, antipodal to Pangea and surrounded by ridges
(Natland and Winterer 2005). It grew by the outward migration of ridges and triple junctions. The growing Pacific plate
may have been stationary because of the absence of bounding trenches. The great oceanic plateaus in the Pacific were
being constructed at the boundaries of the expanding
Pacific plate between the times of Pangea breakup and the
construction of the large igneous plateaus in Africa and
ELEMENTS

Melting relations in dry lherzolite and eclogite based on

laboratory experiments. The dashed line is a reference
1300C mantle adiabat. Eclogite will melt as it sinks into normal-temperature mantle, and upwellings from the shallow mantle will extensively melt any entrained gabbro and eclogite. Eclogite will be about
70% molten before dry lherzolite starts to melt (compiled by J. Natland,
personal communication). The two vertical lines separate the gabbro
field from the eclogite field (in between is the mixed phase region).

272

FIGURE 2

D ECEMBER 2005

transparent to seismic waves, i.e. have low attenuation. An

LVZ with low attenuation implies that it is compositional
rather than thermal in origin.)

CONTINENTAL CRUST IN THE MANTLE?

Fragments of continental crust have been found along the
Mid-Atlantic Ridge (e.g. Bonatti et al. 1996) and in hotspot
and LIP magmas. Schaltegger et al. (2002) found continental zircon xenocrysts in basalts from Iceland and Mauritius.
Continental crust is inferred to exist at the Seychelles, the
Faeroes, Rockall Bank, Jan Mayen, Kerguelen, the Ontong
Java Plateau, Cape Verde, and the Cameroon Line (e.g. Frey
et al. 2002; Ishikawa and Nakamura 2003). The widespread
isotopic characteristics of Indian Ocean basalts have been
attributed to the presence of lower continental crust
entrained during Gondwana rifting (Hanan et al. 2004) or
to delamination of lower continental crust (Escrig et al.
2004).

THE DELAMINATION MECHANISM

The challenge is to find mechanisms that can explain the
volume of basalt, the uplift history, and the ubiquitous evidence for involvement of both continental and mid-ocean
ridge material in LIP magmas. Some igneous provinces are
built on top of rafted pieces of microcontinents or abandoned island arcs, but is there any mechanism for putting
large chunks of continental material into the source regions
of LIPs? Lower crustal delamination is such a mechanism,
although it has been basically unexplored in this context.
The lower continental crust thickens by tectonic and
igneous processes (Kay and Kay 1993; Rudnick 1995),
including magmatic underplating. Presumably the same
thing can happen at intra-oceanic arcs. Below about 50 km,
mafic crust (basalt, dolerite, gabbro) transforms to dense
garnet pyroxenite (eclogite; see glossary. Arc eclogites in
FIG. 3). Histograms of the thickness of the continental crust
show a sharp drop-off at a thickness of 50 km (Mooney et
al. 1998). I suggest that this is controlled by delamination.
Once a sufficiently thick eclogite layer forms, it will detach
and founder because its density is 3 to 10% greater than
that of normal mantle peridotite (FIG. 3). Delamination of a
10 km thick eclogite layer can lead to 2 km of uplift and
massive melt production within 10 to 20 Myr (Vlaar et al.
1994; Zegers and van Keken 2001). Density contrasts of 1%
are enough to drive downwelling instabilities (Elkins-Tanton 2005). Thus, delamination is a very effective and non
thermal way of thinning the lithosphere, extending the
melting column, and creating massive melting and uplift.
In contrast to thermal models, uplift occurs during and
after volcanism, and crustal thinning is rapid.
Lee et al. (2005) estimate that it takes 1030 Myr for a
lower-crustal mafic layer to reach critical negative buoyancy
and for foundering to take place. The thickness of the mafic
layer at the time of foundering ranges between 10 and
35 km, resulting in significant size heterogeneities in the
mantle. When the lower crust is removed, the underlying
mantle upwells to fill the gap and melts because of the
effect of pressure on the melting point. This results in a
pulse of magmatism and an episode of rapid uplift. The
lower crust then rebuilds itself and cools, and the cycle
repeats. The delamination mechanism creates multiple
pulses of magmatism separated by tens of millions of years,
a characteristic of some LIPs. If the crustal thickening is due
to compressional tectonics, the time scales will be dictated
by convergence rates. In a typical convergent belt, thickening and delamination may take 2535 Myr.

ELEMENTS

The neutral density profile of the crust and mantle. The

materials of the crust and mantle are arranged mostly in
order of increasing density. A is the region of over-thickened crust that
can transform to eclogite and become denser than the underlying mantle.
C is the top part of the transition region. Eclogites having STP densities
between 3.6 and 3.7 g/cc may be trapped in C. Eclogites and peridotites have similar densities in this interval but different seismic velocities. In C, MORB-eclogite and low-MgO arc-eclogites reach density
equilibration. Trapped eclogite and dense eclogite sinkers can be lowvelocity zones. The order approximates the situation in an ideally chemically stratified mantle, except for the region just below the continental
Moho where potentially unstable lower crustal cumulate material is
formed. In such a stably stratified structure the temperature gradient will
be greater than the adiabatic gradient. TZ = transition zone; Lhz = lherzolite; Coe = coesite; St = stishovite.

FIGURE 3

There are several ways to generate massive melting: one is

to bring up hot material adiabatically from depth until it
melts; another is to insert fertile material with a low melting pointdelaminated lower arc crust, for exampleinto
the mantle from above and allow the mantle to heat it up
by conduction. Eclogite that was subsolidus at lower crustal
depths can melt extensively when placed into ambient
mantle (FIG. 2). Both mechanisms may be involved in LIP
formation. The time scale for heating and recycling lowercrust material is much less than for subducted oceanic crust
because the former starts out much hotter and does not
sink as deep (FIG. 3). The total recycle time, including
reheating, may take 30 to 75 Myr. If delamination occurs
near the edge of a continent, say along a suture belt (Foulger
et al. 2005), and the continent moves off at 3.3 cm per year
(the average opening velocity of the Atlantic Ocean), the
delamination site will have moved 1000 to 2500 km away
from a vertically sinking root in the time since
delamination.

273

D ECEMBER 2005

BROAD DOMAL UPLIFT

Broad domal uplift is a characteristic of delamination (Kay
and Kay 1993). The magnitude is related to the density and
thickness of the delaminating column. Crustal domes of
~1000 km in lateral extent and elevations of ~2 km above
background, with no heat flow anomaly, can be explained
by such shallow processes (Petit et al. 2002). The Mongolian
dome, for example, is underlain by a small anomaly (5%
seismic velocity reduction and a density reduction of only
0.01 g/cc) of limited vertical extent, 100200 km deep, yet
it has the same kind of domal uplift that has been assumed
to require a large and deep thermal perturbation. The
removal of a dense eclogitic root and its replacement by
upwelling peridotite can create regional uplift and a shallow
low-velocity zone; the eclogite sinker also has low seismic
velocities. This process is essentially a top-down athermal
process.
One of the best-documented examples of delamination,
uplift, and volcanism is the Eastern Anatolia region (Keskin
2003), which was below sea level between ~50 and ~13 Ma.
It was then rapidly elevated above sea level. Uplift was followed by widespread volcanic activity at 78 Ma, and the
region acquired a regional domal shape comparable to that
of the Ethiopian High Plateau. Geophysical, geological, and
geochemical studies support the view that domal uplift and
extensive magma generation were linked to the mechanical
removal of the lower crust accompanied by upwelling of
normal-temperature asthenospheric mantle to a depth of
50 km.
The above examples are important in showing that wellunderstood shallow processes can generate regional domal
structures and large volumes of magma. The LVZ under
some LIPs, including ancient ones, and domal structures
may be associated with cold eclogite rather than hot
upwellings (FIG. 3).

DO WE NEED TO RECYCLE OCEANIC CRUST?

Recycled oceanic crust is often considered to be a component of ocean-island and LIP magma, although this view is
disputed. Once in the mantle, subducted MORB-eclogite
reaches neutral buoyancy at depths of 500650 km (Anderson 1989b; Hirose et al. 1999) (FIG. 3). Very cold MORB may
sink deeper (Litasov et al. 2004). If current rates of oceanic
crust recycling operated for 1 Gyr (Stern 2005), the total
subducted oceanic crust would account for 2% of the mantle, and it could be stored in a layer only 70 km thick. The
surprising result is that most subducted oceanic crust need
not be recycled or sink into the lower mantle in order to satisfy any mass-balance constraints (see also Anderson
1989a). The MORB-like component in some LIPs may simply be due to passive asthenospheric upwelling, as in the
delamination model.
The recycling rate of lower-crustal mafic rocks (Lee et al.
2005) implies that about half of the continental crust is
recycled every 0.6 to 2.5 billion years. In contrast to oceanic
crust, one can make a case that eroded and delaminated arc
and continental material is not stored permanently, or over
the long term, or very deep in the mantle; it is reused and
must play an important role in continental mass balance,
global magmatism, and shallow-mantle heterogeneity.

range permits partial melting of peridotite, the formation

(in places) of high-MgO magmas, and extensive melting of
eclogite. Melting experiments (e.g. Yaxley 2000) suggest
that 6080% melting of eclogite is required to reproduce
compositions of some LIP basalts (Natland, personal communication). FIG. 2 suggests that this is plausible and that
lherzolite will start to melt under these conditions. The
interaction of melts from eclogite and lherzolite is implied.
The model discussed here does not imply low mantle temperatures. In fact, an internally heated (i.e. by radioactive
decay), chemically stratified mantle achieves higher temperatures than a uniform mantle heated from below. One
must explain, then, not the existence but the rarity of
picrites. The eruption of high-temperature MgO-rich magmas may require special circumstances because of their
high density. This special circumstance may be the rapid
upwelling of asthenosphere following a delamination event
and edge-driven convection (King and Anderson 1998).

DISCUSSION
Some LIPs may simply be due to passive upwelling of inhomogeneous asthenosphere as continental fragments diverge
(McHone 2000). Some are the result of reactivated suture
zones or other weaknesses of the lithosphere, combined
with the variable fertility and melting point of the underlying mantle (Foulger et al. 2005). In this paper, I have
focused on a new mechanism that augments these other
processes. Delaminated lower crust sinks into the mantle as
eclogite, where it has a relatively low seismic velocity and
melting point compared to normal mantle peridotite.
Although delaminated continental crust enters the mantle
at much lower rates than oceanic crust, the rates are comparable to LIP production rates. I speculate that the large
melting anomalies that form on or near ridges and triple
junctions may be due to the resurfacing of large fertile
blobs, including delaminated continental crust. Ponded
melts may contribute to magmatism at new ridges and
triple junctions. Delaminated eclogites may form a unique
component of hotspot and ridge magmas (Lee et al. 2005),
but I suggest that lower continental crust is not just a contaminating agent. Blocks of it are responsible for the melting anomalies themselves, including the Kerguelen Plateau
and other features in the Indian Ocean. The massive
plateaus, such as the Ontong Java Plateau, may be due to a
combination of delamination (Korenaga 2005), excess mantle fertility, slightly higher average mantle temperatures
than usually assumed (~100C), slightly lower melting
temperatures (~100C), and focusing of magma at a triple
junction.
The crustal delamination, variable mantle fertility model,
combined with passive asthenospheric upwelling, has the
potential to explain the tectonics and compositions of LIPs,
including heatflow and uplift histories. Apparently, no
other model explains the formation of LIPs and uplifted
domes so elegantly, with so few contradictions. But the
model needs to be tested further and quantified. .

MELTING OF ECLOGITE
Geophysical estimates of the potential temperature (see
glossary) of the mantle are about 13501400C (Anderson
2000), with statistical and geographic variations of at least
100C. These temperatures are about 100 degrees higher
than generally assumed by petrological modelling. This

ELEMENTS

274

D ECEMBER 2005

REFERENCES
(Current information on the subject
matter of this article can be found at
www.mantleplumes.org)
Allen RM, Tromp J (2005) Resolution of
regional seismic models: Squeezing the
Iceland anomaly. Geophysical Journal
International 161: 373-386
Anderson DL (2000) Thermal state of the
upper mantle; no role for mantle plumes.
Geophysical Research Letters 27(22):
3623-3626
Anderson DL (1989a) Theory of the Earth.
Blackwell Scientific Publications, Boston,
366 pp http://resolver.caltech.edu/CaltechBOOK:1989.001
Anderson DL (1989b) Where on Earth is
the crust? Physics Today 42: 38-46
Bonatti E, Ligi M, Borsetti AM, Gasperini L,
Negri A, Sartori R (1996) Lower
Cretaceous deposits trapped near the
equatorial Mid-Atlantic Ridge. Nature
380: 518520
Christiansen RL, Foulger GR, Evans JR
(2002) Upper-mantle origin of the
Yellowstone hotspot. Bulletin of the
Geological Society of America 114:
1245-1256
Clift PD (2005) Sedimentary evidence for
moderate mantle temperature anomalies
associated with hotspot volcanism. In:
Foulger GR, Natland JH, Presnall DC,
Anderson DL (eds) Plates, Plumes, and
Paradigms. Geological Society of America
Special Paper 388, pp 279-288
Czamanske GK, Gurevich AB, Fedorenko V,
Simonov O (1998) Demise of the Siberian
plume: paleogeographic and paleotectonic reconstruction from the prevolcanic
and volcanic record, North-Central
Siberia. International Geology Review 40:
95-115
Elkins-Tanton LT (2005) Continental
magmatism caused by lithospheric
delamination. In: Foulger GR, Natland
JH, Presnall DC, Anderson DL (eds)
Plates, Plumes, and Paradigms, Geological
Society of America Special Paper 388, pp
449-462
Escrig S, Capmas F, Dupr B, Allgre CJ
(2004) Osmium isotopic constraints on
the nature of the DUPAL anomaly from
Indian mid-ocean-ridge basalts. Nature
431: 59-63
Foulger GR, Natland JH, Anderson DL
(2005) A source for Icelandic magmas in
remelted Iapetus crust. Journal of
Volcanological and Geothermal Research
141: 23-44

Hirose K, Fei Y, Ma YH, Mao H-K (1999)

The fate of subducted basaltic crust in the
Earths lower mantle. Nature 397: 53-56
Ishikawa A, Nakamura E (2003) Dupal
signature recorded in quartz-garnet
clinopyroxenite xenolith from subOntong Java Plateau mantle. Geochimica
et Cosmochimica Acta Supplement
67: 175
Kay RW, Kay SM (1993) Delamination and
delamination magmatism. Tectonophysics 219: 177-189
Keskin M (2003) Magma generation by
slab steepening and breakoff beneath
a subduction-accretion complex: An
alternative model for collision-related
volcanism in Eastern Anatolia, Turkey.
Geophysical Research Letters 30 (24):
8046, doi:10.1029/ 2003GL018019
King SD, Anderson DL (1998) Edge-driven
convection. Earth and Planetary Science
Letters 160: 289-296
Korenaga J (2005) Why did not the Ontong
Java Plateau form subaerially? Earth and
Planetary Science Letters 234: 385-399
Korenaga J, Kelemen PB, Holbrook WS
(2002) Methods for resolving the origin
of large igneous provinces from crustal
seismology. Journal of Geophysical
Research 107 (B9): 2178,
doi:10.1029/2001JB001030
Lee C-T, Anderson DL, Cheng X (2005)
Continental crust formation at arcs, the
garnet pyroxenite delamination cycle,
and the origin of fertile melting
anomalies in the mantle. Earth and
Planetary Science Letters (submitted)
Litasov K, Ohtani E, Suzuki A, Kawazoe T,
Funakoshi K (2004) Absence of density
crossover between basalt and peridotite
in the cold slabs passing through 660 km
discontinuity. Geophysical Research
Letters 31 (L24607),
doi:10.1029/2004GL021306
McHone JG (2000) Non-plume magmatism
and rifting during the opening of the
central Atlantic Ocean. Tectonophysics
316: 287-296
Mooney WD, Laske G, Masters TG (1998)
CRUST 5.1: A new global crustal model at
5 5. Journal of Geophysical Research
103: 727-747
Mller RD, Royer J-Y, Lawver LA (1993)
Revised plate motions relative to the
hotspots from combined Atlantic and
Indian Ocean hotspot tracks. Geology
21: 275-278

Natland JH, Winterer EL (2005) Fissure

control on volcanic action in the Pacific.
In: Foulger GR, Natland JH, Presnall DC,
Anderson DL (eds) Plates, Plumes, and
Paradigms, Geological Society of America
Special Paper 388, pp 687-710
Petit C, Dverchre J, Calais E, Sankov V,
Fairhead D (2002) Deep structure and
mechanical behavior of the lithosphere
in the Hangai-Hvsgl region, Mongolia:
new constraints from gravity modeling.
Earth and Planetary Science Letters 197:
133-149
Roberge J, Wallace PJ, White RV, Coffin MF
(2005) Anomalous uplift and subsidence
of the Ontong Java Plateau inferred from
CO2 contents of submarine basaltic
glasses. Geology 33: 501-504
Rudnick RL (1995) Making continental
crust. Nature 378: 571-578
Schaltegger U, Amundsen H, Jamtveit B,
Frank M, Griffin WL, Grnvold K,
Trnnes R, Torsvik T (2002) Contamination of OIB by underlying ancient
continental lithosphere: UPb and Hf
isotopes in zircons question EM1 and
EM2 mantle components. Geochimica
et Cosmochimica Acta 66 (supplement 1):
A673
Stern RJ (2005) Evidence from ophiolites,
blueschists, and ultrahigh-pressure
metamorphic terranes that the modern
episode of subduction tectonics began
in Neoproterozoic time. Geology 33:
557-560
Vandecar JC, James DE, Assumpo M
(1995) Seismic evidence for a fossil
mantle plume beneath South America
and implications for plate driving forces.
Nature 378: 25-31
Vlaar NJ, van Keken PE, van den Berg AP
(1994) Cooling of the earth in the
Archaean: Consequences of pressurerelease melting in a hotter mantle. Earth
and Planetary Science Letters 121: 1-18
Yaxley GM (2000) Experimental study
of the phase and melting relations
of homogeneous basalt + peridotite
mixtures and implication for the
petrogenesis of flood basalt. Contributions to Mineralogy and Petrology 139:
326-338
Zegers TE, van Keken PE (2001) Middle
Archean continent formation by crustal
delamination. Geology 29: 1083-1086 .

Frey FA, Weis D, Borisova AYu, Xu G

(2002) Involvement of continental crust
in the formation of the Cretaceous
Kerguelen Plateau: new perspectives from
ODP Leg 120 sites. Journal of Petrology
43: 1207-1239
Gomer BM, Okal EA (2003) Multiple-ScS
probing of the Ontong-Java plateau.
Physics of the Earth and Planetary
Interiors 138: 317331
Hanan BB, Blichert-Toft J, Pyle DG,
Christie DM (2004) Contrasting origins
of the upper mantle as revealed by
hafnium and lead isotopes from the
Southeast Indian Ridge. Nature 432:
91-94

ELEMENTS

275

D ECEMBER 2005

Meteorite Impacts
as Triggers to Large
Igneous Provinces

Shocked quartz showing

decorated planar
deformation features
from newly discovered
impact deposits in SW
China, of unknown age.
Partially crossed polars,
field of view approximately 1 mm. PHOTO
ADRIAN JONES

Adrian P. Jones1

meteorite impacting on the surface of the Earth produces not only

a crater but also, if the impactor is sufficiently large, high melt
volumes. Computer simulations suggest that, in addition to shockinduced melting produced by impact, additional decompression melting
of the hot target mantle beneath the crater can produce melt volumes
comparable to those found in large igneous provinces (LIPs). The coincidence
between the expected frequency of such impact events combined with the
similarity in magma volumes of LIPs suggests that large meteorite impacts
may be capable of triggering LIPs and mantle hotspots from a point source
which is subsequently buried. Can the impact model explain any LIP? What
are the distinctive macroscopic criteria predicted from an impact model, and
how may they be recognised or rejected in the geological record of the Earth?

500 Myr. Thus, terrestrial impact

events capable of generating global
ejecta layers should occur relatively frequently in the geological
record, and some LIPs may overlap
in age with an unrelated impact
event (White and Saunders 2005).
LIP volumes are typically ~106 km3,
but some are smaller, e.g. the
Columbia River basalt plateau
(~105 km3), and some are larger,
e.g. the Siberian Traps and the
Ontong Java Plateau (106107 km3).

This paper summarises recent

modelling to test the impact forKEYWORDS: meteorite impact, decompression melting, mantle, hotspot
mation of the Ontong Java Plateau
(OJP), which due to its size is an
INTRODUCTION
extreme case. The case for cause and effect between a speThe idea that large meteorite impacts may trigger volcanic cific impact event and a LIP would be strengthened if the
activity (impact-induced volcanism) has been around for stratigraphy demonstrates that the impact event immediseveral decades. A meteorite impact was proposed to ately predated the LIP, and if the initial chemistry of the
explain the differentiated melts of the Sudbury Igneous LIP, where underlain by oceanic mantle, was ultramafic (e.g.
Complex, including the currently unfavoured idea that the high-magnesium basalt or picrite). Such an impact event
nickel-rich deposits are cosmogenic (Dietz 1972). Hot melt- preceding a LIP has been described from West Greenland
ing of impacted peridotite mantle was proposed by Green and is profiled below. The volcanic expression of impacts is
(1972) as an explanation for the very-high-temperature unknown, but key igneous features of a large differentiated
lavas called komatiites. Rogers (1982) proposed that large impact melt at Sudbury provide some clues.
oceanic basalt plateaus represent the relics of moderatesized meteorite impact craters [diameter (D) > 100 km] and IMPACTS AND MELTING
promoted Grieves (1980) idea that upwarping of the
Typically, the relationship between the body size of the
asthenosphere might be sufficient to initiate a long-lived
impactor and the diameter of the transient crater is 1:10.
thermal plume in the mantle. Melt productivity beneath
Rebound and gravity-assisted relaxation produce a final
impact craters may be significantly increased by decomcrater much shallower than the transient crater and with D
pression melting of the mantle and high ambient geotherapproximately twice that of the transient crater. A compilamal gradients.
tion of estimated melt volumes for terrestrial impact craters
Very large terrestrial impact craters (D > ~500 km) can cre- in crystalline rocks by Grieve and Cintala (1992) shows a
ate more melt than the volume of the impact crater, i.e. suf- simple logarithmic correlation with crater diameter (FIG. 1A).
ficient for LIPs (~106 km3, FIG. 1A), and simulations of the The conventional view is that melt volume scales with the
largest conceivable impact, the giant moon-forming event, kinetic energy (1/2mv2) of the impactor (Pierazzo et al.
predict reorganisation and substantial melting of the man- 1997) and that the volume of melt produced by a meteorite
tle and lithosphere on a global scale (Canup and Asphaug impact (for a crater with D ~100200 km) is insufficient to
2001). Glikson (2001) estimates that 390 36 craters (D > explain the amount of melt typical of a LIP (Ivanov and
100 km) and 45 4 craters (D > 250 km) formed on Earth Melosh 2003). Until recently, most impact cratering models
in the last 3.8 Gyr. Even when adjusted for a reduction in simulated the effect of impacts into cold targets. Since hot
the number of hits with time, this translates to one large targets melt at lower shock pressures, modelling the true
impact (D > 250 km) every ~100150 Myr during the last thermal structure of the target is vital for understanding the
effects of impacts on Earth. This is particularly true for large
impactors that penetrate through the Earths crust and into
1 Department of Geological Sciences
the mantle. Jones et al. (2002) used hydrodynamic comUniversity College London
puter modelling to demonstrate that the melting response
Gower Street
of the Earths peridotite mantle to decreasing pressure
London WC1E 6BT
United Kingdom
(decompression melting) beneath a large impact crater
Adrian.jones@ucl.ac.uk
might increase the total melt volume, depending on the
ELEMENTS, VOL. 1,

PP.

277281

277

D ECEMBER 2005

A
(A) Theoretical correlation of impact melt volume versus
crater diameter for terrestrial impact craters (sloping line)
with locus of terrestrial impact melt below this line (after Grieve and
Cintala 1992). Also shown is the melt volume typical of a LIP (horizontal dashed line at 106 km3). (B) Hypothetical increase in impact melt volume above the theoretical sloping line, due to additional decompression
melting of lithospheric mantle for large crater diameters, not precisely
determined (see Jones et al. 2002, 2003).

FIGURE 1

target thermal profile and lithology. For impacts into the

thermally active Earth, we contend that above some threshold crater diameter (not yet determined), where decompression melting becomes significant, the volume of melt
produced might be considerably greater than that predicted
by the conventional relationship between melt volume and
crater diameter. Melt volumes of approximately 106 km3
comparable to LIP volumes might be produced (FIG. 1B).

GEOLOGICAL EVIDENCE OF IMPACT

Amongst an array of geological impact signatures, two
important criteria are (1) the presence of shock metamorphic effects in mineral and rock inclusions in breccias and
melt rocks, and (2) evidence for a minor extraterrestrial geochemical component in these rocks (Koeberl 2002).
Shocked quartz is the foremost mineralogical criterion for
recognition of shock metamorphism in terrestrial materials;
it is petrographically distinctive and recognisable even as
rare fragments (FIG. 2A). However, shocked quartz can only
be produced in target rocks containing quartz, ruling out
parts of the Earths crust and lithosphere composed of
quartz-free rock like basalt (e.g. oceanic crust). Therefore
ejecta products of oceanic impacts will be free of shocked
quartz and must be identified by other mineralogical and
geochemical criteria. These include quenched melt droplet
spherules (FIG. 2B), Ni-spinel (FIG. 2C, D) and geochemical
anomalies for the siderophile elements, especially platinumgroup elements (e.g. iridium, osmium) and chromium. In
addition, there may be field evidence for unusual geological
activity, such as tsunami deposits.

IMPACT MODEL FOR THE ONTONG

JAVA PLATEAU
The mid-Cretaceous Ontong Java Plateau (OJP) is the largest
oceanic LIP. It is thought by many to have formed from a
deep mantle plume, although this is not universally
accepted. It may instead have been triggered by a meteorite
impact (Rogers 1982; Ingle and Coffin 2004) as examined
here. Jones et al. (2005a) modelled the first few hundred
seconds of a vertical impact between a hypervelocity meteorite projectile and a dry peridotite lithosphere target using
geotherms for young oceanic crust at the onset of the OJP.
At the scale of tens to hundreds of kilometres, the complexities of atmosphere and ocean are ignored. In the simulation, changing the target to hot oceanic lithosphere has
a dramatic effect and produces massive melting by both
ELEMENTS

D
(A) Shocked quartz from Tertiary breccia, Antrim,
Northern Ireland. Field of view 2.5 mm, plane polarized
light (PPL). (B)(D) From Tertiary spherule bed, Nuussuaq, West
Greenland (after Jones et al. 2005b). (B) Impact melt glass spherules,
field of view 2.5 mm (PPL). (C) Skeletal quench-textured Ni-spinel
occurring as radiating christmas trees (PPL). (D) Detail of (C). Backscattered electron image (width ~50 m) showing Ni-spinel crystals,
which have an irregular core of nearly pure Ni metal.

278

FIGURE 2

D ECEMBER 2005

0
50
100
300

200

100

0
kilometres

shock and decompression. The OJP was formed on young

oceanic crust close to a spreading ridge, and the precise age
of this crust determines the geothermal gradient, which
controls the amount of melt in our impact model. Simply
changing the age of the oceanic crust from 20 to 10 Ma produces about five times more melt, and for our 20 km
impactor, can produce ~106 km3 melt, equivalent to a LIP.
The largest event modelled is extreme, and involved a 30 km
diameter dunite projectile impacting at 20 km/s. Within
~10 minutes of impact, the melt was distributed predominantly as a giant sub-horizontal disc with a diameter in
excess of 600 km at ~150 km depth in the upper mantle,
although most of the melt was shallower than ~100 km
(FIG. 3). The total volume of melt produced is ~2.5 106 km3
and ranges from 100% melt (superheated liquid >500C
above the solidus) within 100 km of ground zero, to nonequilibrium partial melts varying in amount with depth
and distance (FIG. 3). Some of this melt will quench, but
most will crystallise slowly, taking perhaps tens of thousands of years to solidify (Jones et al. 2005a). Massive reorganisation of the affected upper mantle, driven by largescale physical disturbances such as displaced crust,
juxtaposed hot and cold materials and mobile melts, is virtually inevitable (Price 2001). This has not been modelled,
but thermodynamic relaxation of heterogeneously melted
mantle could be redistributed through a much larger
mantle volume as a conventional partial melt, to a maximum of approximately ~7.5 106 km3 of basalt, assuming
2030% partial melting.

ELEMENTS

The map shows the outline of the Ontong Java Plateau. Its
size can be compared with the impact melt derived from
a hydrodynamic simulation of a large impact. The round bulls eye at
the same scale as the map represents the area of melt generated ~20
minutes after vertical impact of a 30 km meteorite into young oceanic
crust. The simulation assumes a young oceanic mantle geotherm, minimal dry melting and maximum credible projectile size (Jones et al.
2005a). The enlarged cross section beneath, with horizontal equal to
vertical scale (in kilometres), shows melt distribution in the mantle on
the left and deformation paths as disturbed layers on the right. Colours
on the left-hand side show the extent of melting: red, up to 100% melt
(corresponding to superheated conditions at temperature >500 degrees
above solidus); yellow 50% melt; blue >1% melt. In peridotite mantle
the colours red-yellow-blue correspond approximately with regions
where the products of melting are peridotite or komatiite, picrite and
basalt, respectively. The maximum depth of melting for this model is
approximately 150 km. Most of the melt is confined to a diameter of
300 km, but a thin surface layer and a deeper disc at ~5070 km extend
to >600 km. The total melt volume generated both by shock and
decompression melting is ~2.5 106 km3. Massive reorganisation of the
affected upper mantle is expected, and could trigger further mantle
upwelling and additional melting.

FIGURE 3

This volume of impact melt exceeds that of most LIPs but is

slightly less than the total volume of the OJP (>30 106 km3).
Larger melt volumes will result from hotter mantle potential temperatures (>1500 degrees, e.g. Chazey and Neal
2004). Various other parameters could be changed projectile dimensions, velocity, hydrous mantle, or anomalous
mantle composition. Part of the impact-induced melt
would be buoyant and erupt rapidly. This episode might be
followed by an extended secondary period of additional
melting during mantle upwelling, as envisioned by Grieve

279

D ECEMBER 2005

(1980) and described in more detail in the lunar mare

model of Elkins-Tanton et al. (2004). A primary feature of
the oceanic impact model is that voluminous ultramafic or
mafic melts are expected early in the igneous activity.
The geochemistry of the OJP and the absence of thermal
uplift have proved difficult to reconcile with deep mantle
plume models, but these features may be consistent with an
impact origin (see Ingle and Coffin 2004). The major and
trace element chemistry of the OJP lavas is notably uniform, with many incompatible element concentrations
comparable to those for shallow mid-ocean ridge basalt
(MORB), but with isotopic characteristics more like those of
ocean island basalts. A shallow-melting origin (~100 km), is
compatible with the impact model.
The subject of correlations between specific impact events,
LIPs and mass extinctions has been reviewed by Alvarez
(2003). Given the likely size of the impact required, a global
record is expected, but will it be recognisable? Oceanic
impacts away from the continental shelf will not yield
shocked quartz, and shocked ejecta of glasses and silicates
are highly susceptible to replacement by clays. Global fallout layers of fine glass-rich ash and dust could be similar to
conventional volcanic deposits. Volcanic clay layers and
Fullers Earth horizons are distinctive marker bands in
BarremianAptian geological sequences of northern
Europe, coinciding with the ~120 Ma age of the OJP event;
might one of these contain the distinctive geochemical signature of extraterrestrial components linked to a distal OJP
impact layer? However, no mass extinction correlates with
the date of formation of the OJP, although there was a
global oceanic anoxic event and a global negative Sr isotope
anomaly (Jones and Jenkyns 2001).

IMPACT EJECTA BENEATH A LIP?

The early Tertiary (~62 Ma) lavas forming Disko Island in
central West Greenland and the Nuussuaq Peninsula immediately to the north have been correlated with those on
Baffin Island and form the earliest western extremity of the
North Atlantic Tertiary Igneous Province, in which volcanic
activity continues today in Iceland. These highly magnesian lavas require high temperature melting of shallow
mantle (6090 km) and may constitute a precursor to the
plume which became established under East Greenland
(Gill et al. 1995). A distinctive spherule bed horizon crops
out over a 3 m interval in shallow water sediments approximately 10 m beneath the local base of the flood lava pile
on the Nuussuaq Peninsula. The glassy spherule layers have
many of the hallmarks of impact ejecta: immiscible melt
textures, distinctive Ni-spinel, and high Ir, PGE and
siderophile element anomalies (Jones et al. 2005b). The
iron-rich silicate glass spherules (~3 wt% NiO, ~35 wt%
FeO) are circular in cross section and show evidence of surface dissolution, smectite replacement and calcite infilling
of vesicles, though many glasses are optically unaltered.
Their pronounced FeNi correlation is dissimilar to volcanic
suites, but can be explained by mixing of a basaltic melt
and an ironnickel source. Distinctive Ni-spinel grains
(~710 wt% NiO) possess very nickel-rich cores (FIG. 2D).
Rare glass spherules show compositional gradients towards
resorbed silicates (plagioclase, clinopyroxene); shocked plagioclase (maskelynite glass) has an anomalous texture comparable to that seen in impact-melted lunar breccias.
Although anomalously high copper and sulphur concentrations (up to ~1% in spherules) have led other researchers to
suggest terrestrial explanations, such as the possibility that
they are products of fire-fountaining of exotic or hot,
picritic Disko lavas (see Jones et al. 2005b for details), a
strong case can be made that they are impact deposits.
Delicate preservation features rule out substantial sedimenELEMENTS

tary reworking, and spherule sizes and bed thicknesses

imply a large source crater. The age of the spherule beds is
constrained by nannofossils and magnetostratigraphy to be
close to the age of initiation of the West Greenland flood
lavas (~62 Ma).

EXPRESSIONS OF IMPACT VOLCANISM?

Although there is no direct evidence that large volumes of
extrusive rock have been produced by impact, the Sudbury
structure shows that large volumes of subsurface magma
can be generated by impact. The Sudbury structure is a large
(D ~200 km, 1850 Ma), deformed and eroded impact crater,
whose central region was occupied by melt. An eight-year
multidisciplinary study by Stoffler et al. (1994) concluded
that the impact excavated deep into the crust, almost to the
mantle (~30 km), before collapse and rebound. The present
eccentric shape is due to subsequent tectonism. The melt
(> ~12,000 km3), possibly superheated, formed by impact
melting of crust within just a few minutes. The magma differentiated by gravity settling of crystals and immiscible
sulfides to produce hundreds of metres of noritic cumulates
(norite is a type of gabbro). Early formed pyroxene and sulfides were swept into basal depressions to form mineralised
norite, overlain by slowly cooled igneous-textured rocks
with differentiated compositions.
There is no record of volcanism at Sudbury but it may have
been spectacular. The high temperatures implied by coexisting immiscible melts and mafic magmas are comparable
to those of many large igneous intrusions, representing the
mid- to upper-crustal reservoirs feeding surface volcanism.
At Sudbury, the presence of pseudotachylites (veins of
shock-induced glassy rock), contact zone breccias and an
array of peripheral shock features is well established. Any
mantle melt component is thought to have been small, but
could have been delivered almost instantly via crust-spanning dykes with rapid post-crater closure (Price 2001).
Rapid closure of fractures may explain the absence of feeders in impact-induced melt bodies such as Sudbury. The
Sudbury nickel deposits are crudely concentrated around
the margins of the impact cavity and form the largest nickel
mining district ever mined. The source of the nickel could
be the impactor in terms of mass balance, although isotope
data suggest a crustal source for the accompanying sulphur.
In other impact craters, there is evidence for associated
downwards and outwards injection of magma, forming
dykes, breccias, and pseudotachylites, and for the establishment of vigorous hydrothermal systems. The Sudbury rootless (?) impact melt, the likelihood of superheat, and the
formation of immiscible sulphides are valuable lessons for
mainstream igneous petrology and global ore prospecting.

FURTHER WORK
Hypervelocity impact models predict that large oceanic
meteorite impacts can generate melts with volumes comparable to those of typical LIPs. Future modelling should
incorporate longer time scales in order to allow for mantle
flow. Impact-triggered mantle melts are expected to have
geochemical signatures like those of mantle plumes. Rapid
mixing of melts from sub-horizontal sub-crater reservoirs to
depths where pyrope garnet and/or diamond are stable is
possible (Jones et al 2003). Impact melting of the mantle
can generate peridotitic melts like komatiites and other
high-degree partial melts (Jones 2002). Reprocessing of
parts of the upper mantle via large bolide impacts is consistent with models of planetary accretion following the late
heavy bombardment and provides an alternative explanation for primitive geochemical signatures currently attributed to plumes entraining material from the core. For LIPs
with mafic to ultramafic initial volcanic products, and

280

D ECEMBER 2005

especially for oceanic targets, impact volcanism is a testable

hypothesis. Continental LIPs (e.g. the Siberian Traps; Jones
et al. 2002) may also be the result of impacts and might be
expected to show an admixed component from melted
crust; such LIPs need to be modelled. Evidence of an initiating catastrophe, such as an ejecta layer at the base of a
LIP, may in principal be found, but in practice may be missing because of burial or lack of exposure, as in the case of
the OJP. The West Greenland spherule horizon appears to

REFERENCES
Alvarez W (2003) Comparing the evidence
relevant to impact and flood basalt at
times of major mass extinctions.
Astrobiology 3: 153-161
Canup RM, Asphaug E (2001) Origin of
the Moon in a giant impact near the end
of the Earths formation. Nature 412:
708-712
Chazey WI, Neal CR (2004) Large igneous
province magma petrogenesis from
source to surface: platinum-group
element evidence from Ontong Java
Plateau basalts recovered during ODP legs
130 and 192. In: Fitton JG, Mahoney JJ,
Wallace PJ, Saunders AD (eds) Origin and
Evolution of the Ontong Java Plateau.
Geological Society of London Special
Publication 229, pp 219-238
Dietz RS (1972) Sudbury astrobleme, splash
emplaced sub-layer and possible cosmogenic ores. In: Guy-Bray JV (ed) New
Developments in Sudbury Geology.
Geological Association of Canada Special
Paper 10, pp 29-40
Elkins-Tanton LT, Hager BH, Grove TL
(2004) Magmatic effects of the lunar late
heavy bombardment. Earth and Planetary
Science Letters 222: 17-27

satisfy the impactLIP criteria of superposition and mafic

LIP startup geochemistry and is accessible to future scrutiny.

ACKNOWLEDGMENTS
I thank colleagues at UCL for many discussions, and particularly David Price and Paul DeCarli for mainstream partnership central to the theme of this review. Ian Parsons and
Stephanie Ingle are thanked for providing most thoughtful
reviews. .

Jones AP, Price GD, Price NJ, DeCarli PS,

Clegg RA (2002) Impact induced melting
and the development of large igneous
provinces. Earth and Planetary Science
Letters 202: 551-561

Jones CE, Jenkyns HC (2001) Seawater

strontium isotopes, oceanic anoxic
events, and seafloor hydrothermal
activity in the Jurassic and Cretaceous.
American Journal of Science 301: 112-149

Jones AP, Price GD, De Carli PS, Price NJ,

Clegg RA (2003) Impact decompression
melting: a possible trigger for impact
induced volcanism and mantle hotspots?
In: Koeberl C, Martinez-Ruiz F (eds)
Impact markers in the Stratigraphic
Record. Springer, Berlin, pp 91-120

Koeberl C (2002) Mineralogical and

geochemical aspects of impact craters.
Mineralogical Magazine 66: 745-768

Jones AP, Wunemann K, Price GD (2005a)

Modeling impact volcanism as a possible
origin for the Ontong Java Plateau. In:
Foulger GR, Natland JH, Presnall DC,
Anderson DL (eds) Plates, Plumes and
Paradigms. Geological Society of America
Special Paper 388: 711-720
Jones AP, Kearsley AT, Friend CRL, Robin E,
Beard A, Tamura A, Trickett S, Claeys P
(2005b) Are there signs of a large
Palaeocene impact, preserved around
Disko bay, W. Greenland?: Nuussuaq
spherule beds origin by impact instead of
volcanic eruption? In: Kenkman K, Hrz
F, Deutsch A (eds) Large Meteorite
Impacts III. Geological Society of America
Special Paper 384: pp 281-298

Gill RCO, Holm PM, Nielsen TFD (1995)

Was a short-lived Baffin Bay plume active
prior to initiation of the present Icelandic
plume? Clues from the high-Mg picrites
of West Greenland. Lithos 34: 27-39

Pierazzo E, Vickery AM, Melosh HJ (1997)

A reevaluation of impact melt production, Icarus 127: 408-423
Price NJ (2001) Major Impacts and Plate
Tectonics. Routledge, London, 416 pp
Rogers GC (1982) Oceanic plateaus as
meteorite impact signatures. Nature 299:
341-342
Stffler D, Deutsch A, Avermann M,
Bischoff L, Brockmeyer P, Buhl D,
Lakomy R, Mller-Mohr V (1994) The
formation of the Sudbury structure,
Canada: toward a unified impact model.
In: Dressler BO, Grieve RAF, Sharpton VL
(eds) Large Meteorite Impacts and
Planetary Evolution. Geological Society
of America Special Paper 293: 303-318
White RV, Saunders AD (2005) Volcanism,
impact and mass extinctions: incredible
or credible coincidences? Lithos 79:
299-316 .

Glikson AY (2001) The astronomical

connection of terrestrial evolution:
crustal effects of post-3.8 Ga mega-impact
clusters and evidence for major 3.2 0.1
Ga bombardment of the EarthMoon
system. Journal of Geodynamics 32:
205-229
Green DH (1972) Archaean greenstone
belts may include terrestrial equivalents
of lunar maria? Earth and Planetary
Science Letters 15: 263-270
Grieve RAF (1980) Impact bombardment
and its role in proto-continental growth
on the early earth. Precambrian Research
10: 217-247
Grieve RAF, Cintala MJ (1992) An analysis
of differential impact melt-crater scaling
and implications for the terrestrial impact
record. Meteoritics 27: 526-538
Ingle S, Coffin MF (2004) Impact origin for
the greater Ontong Java Plateau? Earth
and Planetary Science Letters 218:
123-134
Ivanov BA, Melosh HJ (2003) Impacts do
not initiate volcanic eruptions: Eruptions
close to the crater. Geology 31: 869-872
Jones AP (2002) Komatiites: new information on the type locality (Barberton), and
some new ideas. Geology Today 18: 23-25

ELEMENTS

281

D ECEMBER 2005

Gas Fluxes from Flood

Basalt Eruptions
Stephen Self1, Thorvaldur Thordarson2, Mike Widdowson1

ubaerial continental flood basalt volcanism is distinguished from all

other volcanic activity by the repeated effusion of huge batches of
basaltic magma (~102103 km3 per eruption) over short periods of
geologic time (<1 Myr). Flood basalt provinces are constructed of thick stacks
of extensive pahoehoe-dominated lava flow fields and are the products of
hundreds of eruptions. Each huge eruption comes from a dyke-fed fissure
tens to hundreds of kilometres long and lasts about a decade or more. Such
spatial and temporal patterns of lava production do not occur at any other
time in Earth history, and, during eruptions, gas fluxes of ~1 Gt per year of
SO2 and CO2 over periods of a decade or more are possible. Importantly, the
atmospheric cooling associated with aerosols generated from the SO2 emissions of just one flood basalt eruption is likely to have been severe and would
have persisted for a decade or longer. By contrast, warming due to volcanogenic CO2 released during an eruption is estimated to have been insignificant because the mass of CO2 would have been small compared to that
already present in the atmosphere.

Back-scattered electron
map of olivine (blue) in
glass (yellow). The
Stapafell eruption was
subglacial, and rapid
cooling has caught melt
inclusions being
entrapped. PHOTO JOHN
MACLENNAN

terized by the repeated effusion of

huge magma batches over a short
period of geologic time (<1 Myr).
These short-lived, main pulses of
eruptive activity (Rampino and
Stothers 1988) consist of many
large-volume and prolonged eruptions, each commonly yielding
lava flow fields (see glossary) of
102103 km3 (FIG. 1). Both the volume of magma emitted during
these individual eruptions and the
total volume of magma released
during the main eruptive pulses
(up to several million km3) are
exceptional in Earth history.

This contribution examines the

potential for volcanic gas release
KEYWORDS: flood basalt volcanism, eruption volumes, during flood basalt eruptions and
the manner in which this gas
gas release, atmospheric impact
release might then affect regional
and global environments. The
INTRODUCTION
obvious major way is through the effects of gases and resultMass extinction events have fundamentally shaped the his- ant aerosols on the Earths atmosphere and surface
tory of life on Earth, and many recent studies focus upon processes during and immediately after an eruption.
catastrophic factors as the cause. Two types of event have Previous studies have not investigated the species of gas
been invoked: episodes of continental flood basalt volcan- released, which are carbon dioxide (CO2), sulfur dioxide
ism and major bolide impacts (asteroids or comets). Several (SO2), nitrous oxides (NOx), and the dominant volcanic gas,
studies have compared dates of extinction events of various H2O, or the quantities. Important considerations are how
magnitudes with dates of flood basalt episodes and found often these gases were released during the main eruptive
significant correlations, thus supporting a possible cause- pulse (i.e. what was the eruption frequency?) and the manand-effect connection (e.g. Stothers 1993; Courtillot 1999). ner in which they were introduced into the atmosphere by
The apparent coincidence of eruptions in the Siberian flood flood lava activity. Clearly, such information will enable
basalt lava flow province with the severe extinctions at the better constrained simulations to be made regarding the cliend of the Permian, ~250 Ma ago, and the near-coincidence matic and environmental impact of flood basalt volcanism.
of activity in the Deccan flood basalt province (India) and
the extinctions at the end of the Cretaceous suggest that
flood basalt eruptions may have contributed significantly
to some mass extinction events, as summarized by Wignall
(2001 and this issue).
Continental flood basalt (CFB) provinces are a type of large
igneous province (LIP) and are erupted subaerially for the
most part. CFB province formation is unique with respect to
all other subaerial basaltic magmatism because it is charac1 Volcano Dynamics Group, Department of Earth Sciences,
The Open University, Milton Keynes, MK7 6AA, UK
2 Department of Geology and Geophysics
University of Hawaii at Manoa
1680 East-West Road, Honolulu HI 96822, USA

Here, we consider only sulfur (S), as its common gas SO2,

and CO2, although it is increasingly recognized that NOx
emissions may also have affected local and regional ecosystems (e.g. Mather et al. 2004). We assess the potential
amount of gases released during past flood basalt eruptions
in order to determine the degree of atmospheric pollution
and hence potential environmental deterioration attributable to this activity. The work builds on the one study made
so far that obtained a gas-release budget for an individual
flood basalt eruption, that of the Roza lava flow field in the
mid-Miocene Columbia River basalt province of eastern
Washington State, USA (Thordarson and Self 1996),
together with recent advances in knowledge concerning
similar basaltic eruptions in Iceland.

Corresponding author: Stephen.Self@open.ac.uk

ELEMENTS, VOL. 1,

PP.

283287

283

D ECEMBER 2005

ESTIMATING VOLATILE RELEASE FROM

FLOOD BASALT ERUPTIONS
As with the extensive ash- and spatter-fall deposits associated with Icelandic basaltic fissure eruptions, such as Eldgja
in AD 934 and Laki in AD 1783 (Thordarson et al. 2003),
the occurrence of deposits of spatter, spatter-fed lava, and
scoria mounds along eruptive fissures in the Columbia
River province suggests that violent fire-fountaining
occurred during prolonged flood basalt eruptions
(Thordarson and Self 1998). Calculations assuming magma
volatile contents of 11.5 wt% suggest that fountains during periods of peak eruptive output may have exceeded 1.5
km in height. Further, model estimates for convective
plumes rising above the fountains indicate eruption column heights in excess of 13 km. Thus, at tropical to midlatitudes, flood eruptive activity is readily capable of lofting

Wall of Dry Falls (about 200 m high), Grand Coulee gorge,

Washington, USA, in the mid-Miocene Columbia River
flood basalt province, showing a section through four pahoehoe lava
flow fields (14), each the product of a huge-volume eruption of >1000
km3 of lava. PHOTO BY S. SELF

FIGURE 1

gases and small amounts of fine ash to mid- to uppertropospheric altitudes, and to the base of the stratosphere
for high-latitude (or higher intensity) eruptions. In addition, flood lava flows themselves are also capable of releasing gas from their surfaces as they move away from the vent
system (FIG. 2), thus creating a lower level atmospheric gas
and aerosol cloud. The combined release of gas from fire
fountains and the surface of lava would be near-continuous
throughout the period of eruptive activity, which may last
from a few to perhaps >100 years. Clearly, the impact of such
huge, long-term degassing events is likely to be significant.
Flood basalt events have been considered to influence the
environment in two ways: (1) by atmospheric cooling
caused by sulfuric acid (H2SO4) aerosols generated from the
SO2 released, which scatter and absorb incoming solar radiation, thus increasing the opacity of the atmosphere (e.g.
Rampino and Self 2000); and (2) by atmospheric warming
through the addition of the greenhouse gas CO2 (e.g.
McLean 1985; Olsen 1999). These two simple cause-andeffect scenarios are a simplification of much more complex
phenomena. For instance, SO2 may act as a greenhouse gas
if present in sufficient concentrations at low altitudes,
whereas widespread climatic cooling would be primarily a
result of the formation and spread of stratospheric sulfuric
acid (H2SO4), or sulfate, aerosols formed by oxidation of the
SO2 with water and hydroxyl OH- radicals in the atmosphere (Robock 2000). In addition, while anthropogenic

ELEMENTS

CO2 emission is now known to cause measurable global

warming, the actual quantities likely to be released during
an individual CFB eruption, although prodigious, are very
small when compared with the natural atmospheric reservoir.

Sulfur Degassing and Sulfate Aerosols

Estimates of S emissions from past eruptions can be made
by measuring the pre-eruption S concentration in glass
(melt) inclusions trapped within crystals (FIGS. 2 AND 3) and
the post-eruption concentration (remaining in the melt
after degassing) in glassy tephra or lava (Devine et al. 1984).
This petrologic method of determining degassing budgets
has recently been validated for modern basaltic eruptions
(Sharma et al. 2004) and from degassing estimates during
recent flood lava flows such as Laki (Thordarson and Self
2003; Thordarson et al. 2003). These studies demonstrate
that ~75% of volatile S species present in the rising melt are
released at the vents, largely as SO2. The gas is then rapidly
transported into the atmosphere where it instantly begins
to convert to sulfuric acid aerosols, and continues to do so
over periods of days to about 1 month. The months-long
Laki eruption is estimated to have released ~120 Mt of SO2,
delivered by quasi-continuous emission but with maximum
fluxes reaching 6 Mt per day in the early stages [1 megatonne (Mt) = 1012 grams (g)]. At this rate, three days of Laki
emissions would have released the same amount of SO2 as
the entire 1991 Mount Pinatubo eruption!
Sulfur gas releases from ancient flood basalt eruptions, the
youngest of which is the ~16 Ma Columbia River basalt
province, are more difficult to determine because the lavas
have typically been affected by the action of weathering
and ground water alteration, which can lead to the removal
of mobile volatile elements. In addition, the petrologic
method requires phenocrysts containing glass inclusions,
but flood basalt successions are dominated by crystal-poor
flows. However, the well-preserved and porphyritic Roza
lava flow is one case where the S loss during eruption can
be measured (FIG. 2), and determinations indicate that concentrations of S in the pre-eruption magma were similar to
that of the Laki magma. The eruption of this 1300 km3
pahoehoe lava flow field is estimated to have released about
12 Gt of SO2 [1 gigatonne (Gt) = 1000 Mt], along with significant amounts of HF and HCl (Thordarson and Self
1996). If the Roza event lasted about 10 years, then the
annual average SO2 burden added to the atmosphere could
have been as much as 1.2 Gt, with a significant portion
injected into the lower stratosphere above a 12 to 14 km
high tropopause by fire fountains and associated convective
columns (up to 15 km above the vents). This equates to ~3
Mt of SO2 per day at a lava effusion rate of ~0.3 km3 per day,
a value similar to the longer term SO2 emission rate maintained from the Laki vents for several weeks during the
summer of 1783.
For crystal-poor flood basalt lavas, or in instances where
alteration renders the petrological method untenable (for
instance, the 65 Ma Deccan lavas of India), a chemical ratio
proxy can be used to estimate the original S content of the
lavas. This method is based upon an observed relationship
between the composition (i.e. FeO content) of recent and
historic basaltic lavas and their observed S content derived
either by direct degassing measurement or from glass inclusion data (Thordarson et al. 2003). The relationship can be
further refined by substituting the FeO/TiO2 ratio of the
glass or the bulk lava composition (FIG. 4). Quantitative values for the pre-degassed content of ancient lavas may be
estimated for the first time using this proxy technique.
Both the petrologic and proxy approaches indicate that vast
volumes of SO2 are likely to have been released continuously during eruption and emplacement of individual flood

284

D ECEMBER 2005

basalt flow fields. Such huge gas emissions, maintained for

many years, are totally unprecedented in any preconceived
models of volcanic gas release and sulfate aerosol generation (cf. Pinto et al. 1989). As a crude estimate taking into
account the range of magma compositions involved, for
each cubic kilometre of lava emitted during basaltic flood
lava eruptions, between 5 and 10 Mt of SO2 would have
been released into the atmosphere.
Volcanogenic SO2 is converted to H2SO4 aerosols in the troposphere and stratosphere; in many eruptions, magmatic
water and entrained tropospheric water entering the stratosphere with the eruption columns are sufficient for this
conversion. Very large burdens of SO2 would, however,
have the potential to greatly deplete stratospheric H2O and
OH- (Bekki 1995). The global stratosphere contains ~1 Gt of
H2O, but how an enormous amount of aerosols would manifest itself on atmospheric chemistry and dynamics cannot
yet be assessed. The balance between a quasi-continuous
volcanogenic source of S gas, the resulting H2SO4 aerosol
formation rates, and the deposition rates for aerosols is only
just beginning to be tested by atmospheric chemistry models (Stevenson et al. 2003). From modern eruptions it is
known that significant volcanic S injection into the stratosphere results in measurable global cooling (e.g. Blake
2003). Volcanic aerosol clouds at any altitude in the atmosphere should cool the Earths surface if the aerosol particle
size is in the normal range (Lacis et al. 1992), and a maximum cooling effect is achieved by low-level aerosols.
Therefore, volcanogenic aerosols in the troposphere, as well
as the stratospheric aerosols, will lead to cooling of the surface. However, after most volcanic eruptions, tropospheric
aerosols have a very short lifetime of only about one week
before they are washed out by cloud formation and rain,
while stratospheric residence times are much longer (typically up to 2 years).

Schematic illustration showing the key features of a twostage degassing model for a flood basalt eruption (modified from Thordarson et al. 2003). The amounts of SO2 degassed are

FIGURE 2

ELEMENTS

Estimates suggest that single flood basalt eruptions can generate such large amounts of sulfate aerosols that they could
create regional optical depths (a measure of atmospheric
opacity, see glossary) of >10. At such high levels of aerosol
loading, it is arguable that the immediate climate cooling
might suppress convection in the lower atmosphere to the
extent that rain-out might become less effective at removing the aerosols. Similarly, stratospheric dehydration would
occur due to the high-altitude portion of the injected gas
and delay the conversion to aerosols of further masses of
injected gas (Savarino et al. 2003) during the prolonged
eruptions. Clearly, even though the cooling effect resulting
from a massive continuous flux of S gas resulting from an
average flood lava eruption can perhaps be crudely calculated (Jolley and Widdowson 2005), global climate models
will be required in order to further explore the details
regarding the extent and severity of the effect.

Carbon Dioxide
To date, there are no direct measurements of CO2 concentrations in glass (melt) inclusions within crystals from flood
basalt lavas that may give an indication of pre-eruption
values. CO2 is relatively insoluble in basaltic melts, and
even CO2 abundances in glass inclusions may not reflect
the original (mantle) values (Wallace and Anderson 2000).
Some arc-setting basalts contain up to 2120 ppm CO2 (i.e.
0.21 wt%), and work on Mexican and Hawaiian basaltic
lavas suggests between 0.2 and 0.5 wt% (Cervantes and
Wallace 2003). Nevertheless, the universally low CO2 concentrations (usually <0.03 wt%) measured in the matrix
glass of recent subaerially erupted lavas indicate that
degassing of CO2 is a highly efficient process.

determined by a study of the Roza flow, Columbia River basalt province

(after Thordarson and Self 1996); the absolute masses of SO2 are based
on a flow volume of 1300 km3.

285

D ECEMBER 2005

Thin section (plane polarized light) of a sample of the

glassy margin of the Thrs lava flow, Iceland, (~8500
years old) showing a plagioclase phenocryst set in a light brown glassy
matrix with smaller plagioclase groundmass crystals; the phenocryst
hosts a glass inclusion (center, white arrow) with a shrinkage bubble.
The inset (at same scale) shows another phenocryst with several brown
glass inclusions up to 60 micrometres long and brown glass attached at
the top with vesicles (bubbles). The petrologic method (see text) relies
upon analyses of S in inclusions and matrix glass (see also Fig. 2).

FIGURE 3

Accepting 0.5 wt% as a reasonable but possibly high value

for pre-eruptive CO2 concentration, and assuming 100%
degassing, approximately 13 Mt of CO2 could be released
for every km3 of basaltic lava erupted. Thus the total CO2
release from a 1000 km3 eruption might be about 13 Gt.
While this is a very large mass, it represents less than 1/200
of the amount of CO2 present in the modern atmosphere
(~3000 Gt), and only about 3% of the current annual
landatmosphere CO2 flux. Moreover, because the gas
would actually be released throughout the eruptive event,
the annual fluxes to the atmosphere during such an eruption would be considerably smaller. A more realistic estimate might be to assume 80% efficient degassing of CO2
during a 1000 km3 eruption; this would yield a flux of ~1
and ~0.2 Gt per year for 10 and 50 year eruption durations,
respectively. Clearly, such annual fluxes are negligible
when compared with the natural atmospheric reservoir.

CONCLUSIONS
We have shown that during decade-long flood basalt eruptive events, and over the whole period of generation of a
flood basalt province, immense and sustained degassing
events should occur on a periodic basis. These events are
incomparably greater in scale than volcanic gas releases at
any other time in Earth history. While the impact of volcanic S gas release may be profound, the mass of CO2
directly released by individual flood lava eruptive events is
tiny in comparison to the normal mass in the troposphere
and stratosphere. The predicted increases in atmospheric
concentration are a fraction of the current anthropogenic
CO2 released from hydrocarbon burning (~25 Gt per year).
Moreover, while the amount of CO2 in the atmosphere is
currently ~3000 Gt, it was perhaps double this value during
the late Cretaceous (i.e. ~6000 Gt). It is therefore unlikely
that volcanic CO2 had a direct effect on mechanisms of
global warming, supporting earlier findings by Caldeira and
Rampino (1990). In addition, there would have been more
than sufficient time for the extra mass of CO2 added to
equilibrate, given that the lava-forming eruptive events
must have been spaced at least hundreds, and probably
thousands, of years apart. By contrast, SO2 emissions and
the atmospheric burden of sulfate aerosols generated during
ELEMENTS

An empirical model gives a proxy for estimating the sulfur released from eruptions of tholeiitic basalt magmas.
S concentration is plotted against the TiO2/FeO ratio for basalt lava samples of various ages from Iceland and other regions (MORB is mid-ocean
ridge basalt; Roza is discussed in text). A best fit line (A) is shown
through fields of data derived from glass inclusions in crystals from
Icelandic and other eruptions, indicating pre-eruption S concentrations
in magma. Best fit lines B and C are through fields of degassed Icelandic
vent tephra and lava flows. The example plotted is for a sample from a
flood lava eruption with a TiO2/FeO ratio of 0.2. The difference between
1650 ppm (line A) and 480 ppm (line B) represents the amount of S
degassed at vents; the difference between 480 ppm and 270 ppm (field
C) represents amount of S degassed from the lava flow. All data are from
electron microprobe analyses; after Thordarson et al. 2003 and references therein).

FIGURE 4

flood basalt events appear to be unprecedented at any other

time in Earth history. Acid rain may also have been widespread. What is less certain is whether affected biota would
have had time to recover from the deleterious effects of sulfate aerosol clouds and acid rain, although quiescent intervals lasting millennia appear to offer ample time for the
recovery of local biological and environmental systems
(Jolley 1997).
The picture emerging from work on the Columbia River
and Deccan flood basalt provinces is of a rapid, very voluminous eruptive pulse consisting of hundreds of eruptions,
with average repose times of thousands of years between
successive eruptions. However, in reality, these eruptive
hiatuses must have varied considerably in duration. The
SO2 released during each eruption would have formed considerable amounts of sulfate aerosols, with effects lasting at
least as long as the eruptions persisted (decades and
possibly longer). The potential impact of such huge, longduration degassing events under palaeoclimatic and palaeoatmospheric conditions warrants much further investigation.

ACKNOWLEDGMENTS
Thanks to an anonymous reviewer for helpful comments
on the initial draft of this article. Support to the first author
during this work came from the UK Natural Environment
Research Council. .

286

D ECEMBER 2005

REFERENCES
Bekki S (1995) Oxidation of volcanic SO2:
a sink for stratospheric OH and H2O.
Geophysical Research Letters 22: 913-916
Blake S (2003) Correlations between
eruption magnitude, SO2 yield, and
surface cooling. In: Oppenheimer C, Pyle
DM, Barclay J (eds) Volcanic Degassing.
Geological Society of London Special
Publication 213, pp 371-380
Caldeira K, Rampino, MR (1990) Carbon
dioxide emissions from Deccan
volcanism and a K/T boundary greenhouse effect. Geophysical Research
Letters 17: 1299-1302
Cervantes P, Wallace PJ (2003) Role of H2O
in subduction-zone magmatism: New
insights from melt inclusions in high-Mg
basalts from central Mexico. Geology 31:
235-238
Courtillot V (1999) Evolutionary
Catastrophes: The Science of Mass
Extinctions. Cambridge University Press,
Cambridge, 188 pp
Devine JD, Sigurdsson H, Davis AN, Self S
(1984) Estimates of sulfur and chlorine
yield to the atmosphere from volcanic
eruptions and potential climatic effects.
Journal of Geophysical Research 89 (B7):
6309-6325

Lacis A, Hansen J, Sato, M (1992) Climate

forcing by stratospheric aerosols. Geophysical Research Letters 19: 1607-1610
Mather TA, Pyle DM, Allan AG (2004)
Volcanic source for fixed nitrogen in the
early Earths atmosphere. Geology 32:
905-908
McLean DM (1985) Deccan Traps mantle
degassing in the terminal Cretaceous
marine extinctions. Cretaceous Research
6: 235-259
Olsen PE (1999) Giant lava flows, mass
extinctions, and mantle plumes. Science
284: 604-605
Pinto JR, Turco RP, Toon OB (1989) Selflimiting physical and chemical effects in
volcanic eruption clouds. Journal of
Geophysical Research 94 (D8): 1116511174
Rampino MR, Self S (2000) Volcanism and
Biotic Extinctions. In: Sigurdsson H et al.
(eds) The Encyclopedia of Volcanoes,
Academic Press, London, pp 263-269
Rampino MR, Stothers R (1988) Flood
basalt volcanism during the past 250
million years. Science 241: 663-668
Robock A (2000) Volcanic eruptions and
climate. Reviews of Geophysics 38:
191-219

Jolley DW (1997) Palaeosurface palynofloras of the Skye lava field and the age of
the British Tertiary volcanic province.
In: Widdowson M (ed) Palaeosurfaces,
Recognition, Reconstruction and
Palaeoenvironmental Interpretation.
Geological Society of London Special
Publication 120, pp 67-94

Savarino J, Bekki S, Cole-Dai J, Thiemens

MH (2003) Evidence from sulfate mass
independent oxygen isotopic compositions of dramatic changes in atmospheric
oxidation following massive volcanic
eruptions. Journal of Geophysical
Research 108 (D21): doi:10.1029/
2003JD003737

Jolley DW, Widdowson M (2005) Did

Paleogene North Atlantic rift-related
eruptions drive early Eocene climate
cooling? Lithos 79: 355-366

Sharma K, Blake S, Self S, Krueger AJ (2004)

SO2 emissions from basaltic eruptions,
and the excess sulfur issue. Geophysical
Research Letters 31: L13612, doi:10.1029/
2004GL019688

ELEMENTS

287

Stevenson D, Johnson C, Highwood E,

Gauchi V, Collins W, Derwent R (2003)
Atmospheric impact of the 1783-1784
Laki eruption: Part 1, Chemistry modelling. Atmospheric Chemistry and Physics
Discussions 3: 551-596
Stothers RB (1993) Flood basalts and
extinction events. Geophysical Research
Letters 20: 1399-1402
Thordarson T, Self S (1996) Sulphur,
chlorine and fluorine degassing and
atmospheric loading by the Roza
eruption, Columbia River Basalt group,
Washington, USA. Journal of Volcanological and Geothermal Research 74:
49-73
Thordarson T, Self S (1998) The Roza
Member, Columbia River Basalt Group:
A gigantic pahoehoe lava flow field
formed by endogenous processes? Journal
of Geophysical Research 103 (B11):
27411-27446
Thordarson T, Self S (2003) Atmospheric
and environmental effects of the 178384
Laki eruption: A review and reassessment.
Journal of Geophysical Research 108
(D1): 4011, doi:10.1029/2001JD002042
Thordarson T, Self S, Miller JD, Larsen G,
Vilmundardottir EG (2003) Sulphur
release from flood lava eruptions in the
Veidivtn, Grimsvtn, and Katla volcanic
systems. In: Oppenheimer C, Pyle DM,
Barclay J (eds) Volcanic Degassing.
Geological Society of London Special
Publication 213, pp 103-122
Wallace P, Anderson Jr AJ (2000) Volatiles
in magmas. In: Sigurdsson H et al. (eds)
The Encyclopedia of Volcanoes.
Academic Press, London, pp 149-170
Wignall PB (2001) Large igneous provinces
and mass extinctions. Earth-Science
Reviews 53: 1-33 .

D ECEMBER 2005

Oceanic LIPs:
The Kiss of Death
Oceanic plateaus have
been drilled from the
JOIDES Resolution drill
ship. PHOTO ODP

Andrew C. Kerr1

ceanic plateaus represent large areas (~1 106 km2) of thickened

oceanic crust formed from rapidly erupted lava (<3 Myr). These
plateaus have formed throughout most of geological time. They
generally correlate with periods of environmental catastrophe characterised
by oceanic anoxia, leading to black shale formation and mass extinction
events. Such correlations are particularly evident in the Cretaceous and can
be partly attributed to the release of CO2 during oceanic plateau formation,
which ultimately resulted in a runaway greenhouse effect. Additionally, sea
level rise and disruption of oceanic circulation patterns by displacement of
seawater during plateau formation contributed to increased environmental
stress and biotic extinction.

(93.5 Ma) and during the Aptian

(124112 Ma) (Sliter 1989;
Bralower et al. 1993; Jahren 2002).
This link appears to extend back to
the Precambrian: Condie et al.
(2001) have noted that significant
black shale events occurred at ~1.9
and 2.7 Ga and that these correlate
with the formation of mantle
plume-derived large igneous provinces (LIPs) and warmer palaeoclimates. Thus, there seems to be a
temporal association, throughout
a significant proportion of geologKEYWORDS: mass extinction, oceanic plateau, ical time, between periods of
black shale, anoxia, mantle plume global oceanic environmental
crises and oceanic plateau formaINTRODUCTION
tion. In the remainder of this contribution the possible
Although the potentially devastating environmental causal links between oceanic plateaus and oceanic environimpact of continental flood basalts has been extensively mental change will be reviewed, with particular reference to
discussed by many authors (Wignall 2001; articles by Self et Cretaceous events.
al. and Wignall this issue), the global environmental effects
of oceanic plateaus the marine equivalent of continental
OCEANIC PLATEAU VOLCANISM AND
flood basalts have received comparatively little attention.

Oceanic plateaus represent over-thickened areas of oceanic

crust (>10 km) in which the bulk of the >1 106 km3 lava
volume appears to have erupted in less than 23 Myr. These
plateaus generally cover an area in excess of 1 106 km2
and result from anomalously high rates of melt production
in the mantle. These high melt-production rates are most
likely due to the excess heat from a deep-rooted mantle
plume (Campbell this issue).
Oceanic plateaus are more buoyant than oceanic crust of
normal thickness generated at a mid-ocean ridge. This
buoyancy means that oceanic plateaus are more resistant to
subduction, a feature which results in partial accretion onto
continental margins. In this way oceanic plateaus can be
preserved in the geological record and can be recognised
back to the earliest Archaean.
Periods of oceanic environmental crisis can be identified in
the geological record by the occurrence of black shales,
which are indicative of low-oxygen or oxygen-absent conditions in the deep ocean. Vogt (1989), Sinton and Duncan
(1997), and Kerr (1998) have noted the coincidence of
global oceanic anoxia, black shale deposition, mass extinction and oceanic plateau formation in the Cretaceous, particularly around the CenomanianTuronian boundary
1 School of Earth, Ocean and Planetary Sciences
Cardiff University, Main Building
Park Place, Cardiff, Wales
CF10 3YE, UK
E-mail: kerra@cf.ac.uk

ELEMENTS, VOL. 1,

PP.

289292

GLOBAL OCEANIC ANOXIA AT THE

CENOMANIANTURONIAN BOUNDARY

Oceanic Volcanism
Arguably, the clearest link between oceanic plateau volcanism and environmental perturbation can be seen around
the CenomanianTuronian (CT) boundary. This time
period is marked by the formation of the Caribbean
Colombian oceanic plateau (eastern Pacific) and parts of
both the Kerguelen Plateau (Indian Ocean) and possibly the
Ontong Java Plateau (western Pacific) (FIG. 1). Also around
this time, India and Madagascar were beginning to rift
apart, and this event was associated with volcanism at the
Marion hotspot, which resulted in flood basalt eruptions on
Madagascar and basaltic lavas offshore. Due to the continued breakup of Gondwana, the length of the global ridge
system increased in the midlate Cretaceous, resulting in a
significantly greater volume of lava erupted globally at midocean ridges. Kerr (1998) has calculated that the peak production of oceanic crust (both intrusive and extrusive)
around the CT boundary was of the order of 45 106 km3,
with ~10 106 km3 of this erupted on the seafloor (FIG. 2).

Stratigraphic Characteristics
The stratigraphic succession around the CT boundary is
characterised by black organic-rich shales, signifying anoxic
oceanic conditions. This black shale event was associated
with a second-order mass extinction event marked by the
demise of 26% of genera (Sepkoski 1986). The CT boundary is also characterised by a sharp increase in 13C from 1.5

289

D ECEMBER 2005

CenomanianTuronian plate tectonic reconstruction

showing the location of ~9093 Ma and 123110 Ma
large igneous provinces. CenomanianTuronian boundary black shale
deposits are also shown. The black shale localities are taken from Arthur
et al. (1987), Herbin et al. (1987), Summerhayes (1987), and Yurtsever et
al. (2003).

FIGURE 1

to 4 in pelagic limestones, a decline in 87Sr/86Sr of seawater and evidence for a significant transgression (FIG. 2).
Oxygen isotope evidence reveals that globally averaged surface temperatures at the CT boundary were 6 to 14C
warmer than now (Kaiho 1994). This temperature rise was
caused by elevated CO2 levels in the atmosphere. Modelling
by Berner (1994) suggests atmospheric CO2 levels at this
time were up to six times greater than pre-industrial levels
and reached a peak around the CT boundary.

Links between Oceanic Plateau Volcanism

and Environmental Catastrophe
The coincidence of extensive oceanic plateau volcanism
and the physical and chemical phenomena outlined above
demand that we look for the causal links between oceanic
plateau volcanism, global oceanic anoxia and warming, and
mass extinction events.
The formation of LIPs on both the continents and the
oceans is often accompanied by lithospheric uplift and
doming (Larson 1991; Nadin et al. 1997). Under the oceans,
this elevation, in combination with the displacement of
water due to the eruption of ~10 106 km3 of lava onto the
ocean floor, results in a significant rise in sea level (FIG. 3).
This mechanism may provide an explanation for at least
part of the estimated ~100 m sea level rise which reached a
maximum at the CT boundary. The elevation of the sea
floor, from both plume uplift and voluminous lava extrusion, during oceanic plateau formation could also have disrupted important oceanic circulation systems around the
CT boundary. The CaribbeanColombian oceanic plateau
formed close to the proto-Caribbean seaway between North
and South America (FIG. 1). At this time, the only major
source of deep, cold, oxygenated water for the juvenile
Atlantic was the Pacific, and the water had to pass through
this proto-Caribbean seaway. The formation of such a
major volcanic edifice and the associated shallowing of seawater so close to this oceanic gateway would have restricted
the flow of deep oxygenated water from the Pacific to the
Atlantic and thus increased the extent of oceanic anoxia in
the Atlantic (de Boer 1986).
Although volcanism undoubtedly contributed to the elevated CO2 contents in the CT atmosphere, it is doubtful if
volcanism alone could have released enough CO2 to cause
the higher temperatures calculated to exist at this time (Self
et al. this issue). However, it is likely that a complex positive

ELEMENTS

Diagrams showing changes in key environmental indicators, sea level, and oceanic crust production between 110
and 80 Ma. The dotted horizontal line represents the Cenomanian
Turonian boundary (CTB). Diagram updated from Kerr (1998).

FIGURE 2

feedback mechanism triggered by the volcanically derived

CO2 led to increased CO2 and elevated temperatures (FIG. 3).
The initial emission of CO2 from oceanic plateau volcanism
was probably accompanied by the release of a considerable
amount of SO2 and halogens (Self et al. this issue), which
would have made the oceans locally more acidic (FIG. 3).
This increased acidity would have led to the dissolution of
shallow-water carbonates, thus releasing more CO2 to the
atmosphere. [Significantly, Arthur et al. (1987) have noted
that the CT boundary is characterised by a lack of carbonates.] Thus, the addition of carbonate-derived CO2 plus volcanic CO2 to the atmosphere at the CT boundary would
have caused global warming of both the atmosphere and
the oceans. Since the solubility of CO2 in seawater decreases
by 4% for every 1C rise in temperature, warming would
have resulted in the release to the atmosphere of yet more
CO2 which was previously dissolved in the oceans. Thus, a
positive CO2 feedback mechanism would have been established (FIG. 3). Kerr (1998) has proposed that such a scenario
would relatively rapidly result in the establishment of a
runaway greenhouse effect.

290

D ECEMBER 2005

Increased atmospheric CO2 levels, in combination with disrupted oceanic circulation patterns and associated
upwelling of nutrients from the deep ocean, would have
resulted in increased biogenic productivity in ocean surface
waters (FIG. 3). This increased biological activity led to
removal of CO2 from the atmosphere and provided a mechanism for reducing the amount of atmospheric CO2. The
13C peak in shallow ocean sediments at the CT boundary
reflects increased burial of marine organic carbon in the
oceans and is due to the preference of organic matter for
isotopically light carbon.
The decrease in 87Sr/86Sr in the stratigraphic record (from
0.70753 to 0.70735), which started in the late-Cenomanian
and continued until the mid-Turonian (FIG. 2), may be a
reflection of the addition to seawater of hydrothermal fluids with a low 87Sr/86Sr from oceanic plateau volcanism.
Conversely, the rise in 87Sr/86Sr from the mid-Turonian
onwards may signify increased continental weathering
resulting from global warming and its associated climatic
disturbance. Continental weathering is another mechanism
which can reduce the amount of atmospheric CO2.
Higher oceanic temperatures would also have contributed
to oceanic anoxia since the solubility of O2 in seawater
decreases by 2% for every 1C temperature rise (de Boer
1986). However, given that globally averaged ocean temperatures appear to have increased by at most 6C (FIG. 2),
the consequent ~10% reduction in the solubility of O2 in
seawater is not enough to explain widespread oceanic
anoxia. Several additional mechanisms by which oceanic
plateaus can contribute to the depletion of dissolved O2
have been discussed by Sinton and Duncan (1997). The first
of these is the reduction of dissolved O2 in seawater by the
reaction of trace metals and sulphides in hydrothermal
fluids with the O2. Although basaltic lava flows can be oxidised by hydrothermal fluids, both during and after eruption, this process is volumetrically insignificant when compared to the much greater effect of the oxidation of metals
in hydrothermal fluids in lowering the amount of dissolved
oxygen in seawater (FIG. 3). The eruption of a 1 104 km3
oceanic plateau basalt lava flow would release a similar volume of hydrothermal fluids at 350C into the ocean (Cathles, cited in Sinton and Duncan 1997). Sinton and Duncan
(1997) have calculated that the complete oxidation of the

ELEMENTS

Flow diagram of the likely physical and chemical environmental effects of oceanic plateau formation (see text for
a detailed description).

FIGURE 3

Fe2+, Mn2+, H2S, and CH4 in 1 x 104 km3 of hydrothermal

fluid would use up ~6% of the total dissolved oxygen in the
present-day ocean. However, as we have seen, the CT
ocean was significantly warmer and contained less dissolved oxygen than the present-day ocean. Therefore, the
proportion of dissolved oxygen removed by 1 104 km3 of
hydrothermal fluid at the CT boundary would have been
significantly greater than 6%.
Another mechanism for removing dissolved oxygen from
seawater, as discussed by Sinton and Duncan (1997), is the
stimulation of the growth of living organisms (organic productivity) in the oceans by the injection of hydrothermal
iron into surface waters. Vogt (1989) has suggested that
hydrothermal plumes, even those several orders of magnitude smaller than 1 104 km3, would have been capable of
rising into oceanic surface waters. Thus, as noted by Sinton
and Duncan (1997), a large (1 104 km3) hydrothermal
plume (with a low brine content) could easily rise through
the water column and spread laterally over a significant
proportion of the ocean surface. The trace metalrich
waters of such massive hydrothermal plumes may well have
stimulated increased levels of organic productivity in nutrient-poor surface waters (Sinton and Duncan 1997). Coale et
al. (1996) have shown that the addition of Fe into ocean
surface waters can result in a rapid increase in the amount
of phytoplankton. If a similar hydrothermal fluidinduced
phytoplankton bloom occurred around the CT boundary,
the net effect would have been a further reduction in the
amount of dissolved O2 as organic material decayed and
sank through the seawater column (FIG. 3).

FURTHER LINKS BETWEEN OCEANIC

PLATEAU VOLCANISM AND
ENVIRONMENTAL DISTURBANCE
The link between oceanic plateau volcanism and global
oceanic anoxia is given further credence by the occurrence
of Aptian (124112 Ma) black shales, which probably represent one of the most extensive concentrations of organicrich black shales in the geological record (Jenkyns 1980;

291

D ECEMBER 2005

Hallam 1987; Bralower et al. 1993). It is no coincidence that

one of the most extensive periods of plume-related oceanic
plateau formation occurred in the Pacific and Indian oceans
during this period, including the Ontong Java Plateau, Hess
Rise, Manihiki Plateau, the East Mariana and Nauru basins
and a significant proportion of the Kerguelen Plateau
(Eldholm and Coffin 2000). Although many of the chemical and physical characteristics of the Aptian oceanic
anoxic event are very similar to those of the CT boundary,
one of the black shale horizons in the early Aptian exhibits
a sharp decrease in 13C, unlike the CT boundary, which
shows a sharp increase in 13C (Jahren 2002).
Present-day methane hydrates buried in ocean floor sediments possess very low 13C (around 60). Thus the dissociation and catastrophic release of hydrates may have
caused the sharp decrease in 13C and contributed to global
warming and oceanic anoxia in the earlymid Aptian
(Jahren 2002). It has been proposed that methane release
was triggered by tectonic events related to mantle plume
uplift (Jahren 2002). However, while tectonic processes
undoubtedly played a role, I contend that LIP-induced
global warming was of greater importance and was probably of sufficient magnitude to cause dissociation of
methane hydrates and consequent release of methane. This
release would have accelerated global warming and anoxia
since methane is a much more potent greenhouse gas than
carbon dioxide. Furthermore, atmospheric oxidation of
methane would consume a significant amount of free oxygen. Methane release may also have occurred around the
CT boundary, but its distinctive? 13C signal may have
been diluted by higher 13C resulting from volcanism

REFERENCES
Arthur MA, Schlanger SO, Jenkyns HC
(1987). The Cenomanian-Turonian
oceanic anoxic event; II: Palaeoceanographic controls on organic-matter
production and preservation. In: Brooks
J, Fleet AJ (eds) Marine Petroleum Source
Rocks. Geological Society of London
Special Publication 26, pp. 401-420
Berner RA (1994) Geocarb II: a revised
model of atmospheric CO2 over
Phanerozoic time. American Journal
of Science 294: 56-59
Bralower TJ, Sliter WV, Arthur MA, Lekie
RM, Allard D, Schlanger SO (1993)
Dysoxic/anoxic episodes in the AptianAlbian (Early Cretaceous). In: Pringle MS,
Sager WW, Sliter WV, Stein S (eds) The
Mesozoic Pacific: Geology, Tectonics, and
Volcanism, American Geophysical Union
Monograph 77, pp. 5-37
Coale KH, Johnson KS, Fitzwater SE,
Gordon RM, Tanner S, Chavez FP, Ferioli
L, Sakamoto C, Rogers P, Millero F,
Steinberg P, Nightingale P, Cooper D,
Cochlan WP, Landry MR, Constantinou
J, Rollwagen G, Trasvina A, Kudela R
(1996) A massive phytoplankton bloom
induced by an ecosystem-scale iron
fertilisation experiment in the equatorial
Pacific Ocean. Nature 383: 495-501
Condie KC, DesMarais DJ, Abbott D (2001)
Precambrian superplumes and supercontinents: a record in black shales, carbon
isotopes, and paleoclimates? Precambrian
Research 106: 239-260
de Boer PL (1986) Changes in the organic
carbon burial during the Early
Cretaceous. In: Summerhayes CP,
Shackleton NJ (eds) North Atlantic
Palaeoceanography. Geological Society
of London Special Publication 21,
pp. 321-331

ELEMENTS

(around 10). Alternatively, the late Cenomanian ocean

may have been too warm for extensive methane deposits to
accumulate.
Increased oceanic volcanism as a cause of oceanic anoxia
and the deposition of organic-rich sediments has important
implications for the location of potential oil source rocks. It
is likely that many of the worlds most important occurrences of mid-Cretaceous oil source rocks owe their existence to the formation of oceanic plateaus in the Pacific
and Indian oceans and the resultant global anoxia.
This is obviously an interesting model, but how applicable
is it to older oceanic plateaus and black shale sequences?
Major black shale deposits occur throughout the Mesozoic
(Hallam 1987), and some of these correlate with oceanic
plateau volcanism. For example, important Kimmeridgian
to Tithonian (155146 Ma) oil source rocks correlate with
the formation of the Sorachi Plateau in the western Pacific
(Kimura et al. 1994). Furthermore, the formation of
Toarcian (187178 Ma) black shales corresponds with the
eruption of the Karoo, Ferrar and Weddell Sea large igneous
province during the breakup of Gondwana (Riley and
Knight 2001).
Thus, global oceanic anoxia (and the consequent formation
of oil source rocks) is frequently associated with the development of oceanic plateaus. It is contended that the formation of oceanic plateaus perturbs the oceanic and atmospheric environments and sets in motion a chain of events,
often leading to global warming, oceanic anoxia, black
shale deposition and ultimately mass extinction. .

Eldholm O, Coffin MF (2000) Large

igneous provinces and plate tectonics.
In: Richards MA, Gordon RG, van der
Hilst RD (eds) The History and Dynamics
of Global Plate Motions, American
Geophysical Union Monograph 121,
pp. 309-326
Hallam A (1987) End-Cretaceous mass
extinction event: argument for terrestrial
causation. Science 238: 1237-1242
Herbin JP, Montadert L, Muller C, Gomez
R, Thurow J, Wiedmann J (1987)
Organic-rich sedimentation at the
Cenomanian-Turonian boundary in
oceanic and coastal basins in the North
Atlantic and Tethys. In: Summerhayes
CP, Shackleton NJ (eds) North Atlantic
Palaeoceanography, Geological Society
of London Special Publication 21,
pp 389-422
Jahren AH (2002) The biogeochemical
consequences of the mid-Cretaceous
superplume. Journal of Geodynamics
34: 177-191
Jenkyns HC (1980) Cretaceous anoxic
events: from continents to oceans.
Journal of the Geological Society of
London 137: 171-188
Kaiho K (1994) Planktonic and benthic
foraminiferal extinction events during
the last 100 m.y. Palaeogeography,
Palaeoclimatology, Palaeoecology 111:
4571
Kerr AC (1998) Oceanic plateau formation:
A cause of mass extinction and black
shale deposition around the CenomanianTuronian boundary. Journal of the
Geological Society of London 155,
pp 619-626
Kimura G, Sakakibara M, Okamura M
(1994) Plumes in central Panthalassa?
Deductions from accreted oceanic fragments in Japan. Tectonics 13: 905-916

292

Larson RL (1991) Geological consequences

of superplumes. Geology 19: 963-966
Nadin PA, Kusznir NJ, Cheadle, MJ (1997)
Early Tertiary plume uplift of the North
Sea and Faeroe- Shetland Basins. Earth
and Planetary Science Letters 148:
109-127
Riley TR, Knight KB (2001) Age of prebreak-up Gondwana magmatism. Antarctic
Science 13: 99-110
Sepkoski JJ (1986) Phanerozoic overview of
mass extinction. In: Raup DM, Japlonski
D (eds) Pattern and Processes in the
History of Life, Springer-Verlag, Berlin,
pp. 277-295
Sinton CW, Duncan RA (1997) Potential
links between ocean plateau volcanism
and global ocean anoxia at the
Cenomanian-Turonian boundary.
Economic Geology 92: 836-842
Sliter WV (1989) Aptian anoxia in the
Pacific Basin. Geology 17: 909-912
Summerhayes CP (1987) Organic-rich
Cretaceous sediments from the North
Atlantic. In: Brooks J, Fleet AJ (eds)
Marine Petroleum Source Rocks,
Geological Society of London Special
Publication 26, pp. 301-316
Vogt PR (1989) Volcanogenic upwelling
of anoxic, nutrient-rich water: A possible
factor in carbonate-bank /reef demise and
benthic faunal extinctions? Geological
Society of America Bulletin 101: 1225-1245
Wignall PB (2001) Large igneous provinces
and mass extinctions. Earth-Science
Reviews 53: 1-33
Yurtsever TS, Tekin UK, Demirel IH (2003)
First evidence of the Cenomanian/
Turonian boundary event (CTBE) in the
Alakiray Nappe of the Antalya Nappes,
southwest Turkey. Cretaceous Research
24: 41-53 .

D ECEMBER 2005

The Link between Large

Igneous Province Eruptions
and Mass Extinctions
Paul Wignall1

n the past 300 million years, there has been a near-perfect association
between extinction events and the eruption of large igneous provinces,
but proving the nature of the causal links is far from resolved. The associated environmental changes often include global warming and the development of widespread oxygen-poor conditions in the oceans. This implicates a
role for volcanic CO2 emissions, but other perturbations of the global carbon
cycle, such as release of methane from gas hydrate reservoirs or shut-down of
photosynthesis in the oceans, are probably required to achieve severe greenhouse warming. The best links between extinction and eruption are seen in
the interval from 300 to 150 Ma. With the exception of the Deccan Trap
eruptions (65 Ma), the emplacement of younger volcanic provinces has been
generally associated with significant environmental changes but little or no
increase in extinction rates above background levels.

Basaltic scoria cone at

the southern end of the
1783 Lakigigar fissure
eruption, SE Iceland.
PHOTO JOHN MACLENNAN

Calcification crises
Mass extinction
A sharp decrease in the 13C values recorded in limestones; this is
usually interpreted as a record of
methane release from gas hydrate
reservoirs.

The KarooFerrar eruptions around

180 million years ago present all of
these features, and the Siberian
Traps eruptions 250 million years
ago show many of them. The other
LIP eruptions show several features
(TABLE 1). Volcanogenic cooling
has also been proposed for several
extinction events, but the eviKEYWORDS: Extinctions, ocean anoxia, volcanic eruptions
dence is insubstantial. The preponderance of evidence favours warming. This strongly suggests that
INTRODUCTION
CO2 emissions do all the damage, although in many extincCan large igneous province (LIP) eruptions cause environ- tion scenarios this effect is envisaged as a trigger (FIG. 1).
mental and climatic effects that are sufficiently severe to Indeed other factors are almost mandatory given the volcause extinctions? The answer to this question has changed ume of CO2 likely to be released during LIP eruptions (Self
significantly over the past two decades, from probably no et al. this issue). The amount of CO2 released during the
to probably yes. The change of view is due to advances in eruption of the largest LIPs is unlikely to have exceeded
radiometric dating of both extinction events and the ages 1013 tonnes of CO2, with the amount released during indiof LIPs (see Courtillot and Renne 2003). The close age cor- vidual flow events likely to have been at least two orders of
respondence often demonstrated by dating now provides magnitude lower (Wignall 2001). Thus, the gas released
the most compelling link between the two phenomena. during a major flow of 1000 km3 (which may have occurred
Thus, four mass extinctions have occurred in the 300 mil- as frequently as every few thousand years) is unlikely to
lion year interval between the Permian and the present day, have greatly exceeded the current anthropogenic CO2
together with a similar number of minor biotic crises. All of release rate of 25 109 tonnes per annum. This modern flux
these crises coincide with LIP eruptions (Table 1). No other comes not even close to recreating the conditions during
phenomenon shows such a 100% correlation; certainly not these ancient catastrophes. It is possible that LIP eruptions
meteorite impact. However, not all LIP eruptions in this are associated with excessively CO2-rich volcanism, reflectinterval coincide with extinctions. For example, the espe- ing a mantle source still rich in volatiles. However, this has
cially large ParanEtendeka Province was erupted 133 Ma yet to be demonstrated. Alternatively, the volcanism may
ago, in the early Cretaceous, at a time marked by extremely serve as a trigger for something else, such as the release of
low extinction rates. Thus, the claim that all extinction methane from clathrates buried at shallow depths beneath
events coincide with giant volcanic eruptions is well sub- the seafloor. Methane is a much more effective greenhouse
stantiated, but the converse that all such eruptions coincide gas although it is rapidly oxidised to CO2 in the atmoswith extinctions is not true.
phere. However, the tell-tale evidence for methane release
the rapid decrease of the 13C/12C ratio recorded in organic
carbon
and limestones, known as a negative 13C anomaly
TOWARDS A KILL MECHANISM
(see glossary) is seen less frequently than the other eviIn general, LIP eruptions are associated with some or all of
dence for warming.
the following climatic and environmental effects:
Rapid global warming
A Volcanic Greenhouse Scenario
Oceanic anoxia or increased oceanic fertilisation or both
Ideas concerning the role of volcanic gases have been devel1 School of Earth and Environment
University of Leeds, Leeds LS2 9JT, UK
wignall@earth.leeds.ac.uk

ELEMENTS, VOL. 1,

PP.

293297

oped primarily from events during the end-Permian and

Early Jurassic mass extinctions (e.g. Plfy and Smith 2000;
Wignall 2001). It is proposed that the volcanic extinction

293

D ECEMBER 2005

Comparison of environmental effects observed during

eruption of large igneous provinces. Dark blue indicates
a clear development of the effect; pale blue denotes effects for which
there is some evidence, although this may be controversial or less clearcut; red denotes effects which were clearly not developed at the time
of volcanism; and blank cells indicate a lack of data. Note that only the
KarooFerrar volcanism of the Early Jurassic is associated with all of these
effects, while oceanic anoxia is the only effect to coincide with the age
of every volcanic province.

TABLE 1

mechanism is triggered by the release of CO2 during the

eruption of the giant lava flows that form LIPs (FIG. 1). The
resultant increase of atmospheric CO2 levels would have
several deleterious consequences for the oceans. For example, an increase of CO2 concentrations in surface waters
causes a pH decline and thus problems for carbonate-secreting organisms (Gattuso and Buddemeier 2000). This is
known as a calcification crisis and is manifest as a decline
in carbonate content in many sections. Global warming can
also cause the development of oceanic anoxic events (and

ELEMENTS

therefore marine extinctions). These

are intervals of time when large
areas of the oceans and shelf seas
were either oxygen poor or oxygen
free (anoxic). Modern oceans typically have around 56 ml of oxygen
dissolved in a litre of water, but conditions become stressful for most
organisms if values decline below
1.0 ml/L, and no metazoan life can
survive below values of 0.3 ml/L.
Low
oxygen
conditions
are
restricted to only a few small areas
of modern oceans but they became
much more widespread during
oceanic anoxic events due to several
feedback factors associated with
global warming. First, warmer
waters hold less dissolved oxygen
than colder waters. Second, the
oceans circulation system is primarily driven by the temperature gradient between the equator and the
poles, with deep circulation driven
by the generation of cold and dense
waters in the polar regions. This system slows down as polar waters
warm up, thus decreasing the supply of oxygen to the oceans deeper
waters. A possible third factor may
relate to the supply of nutrients
from land, which will increase with
global warming due to increased
rainfall and runoff in a warmer,
more humid climate. Increased
nutrient flux to the seas will foster
increased biological productivity,
which in turn will decrease oxygen
levels in sea water as the plankton
biomass decays the same phenomenon is seen in many modern shelf
seas over-supplied with anthropogenic nutrients such as fertilisers and sewage. Evidence of
increased global runoff during LIP
eruptions is substantial and
includes several lines of geochemical evidence, such as an increase in
the trace metals rhenium and
osmium (Ravizza and PeuckerEhrenbrink 2003). However, many mass extinction events
coincide with a collapse, not an increase, of primary productivity (Hallam and Wignall 1997), and this third factor
may not be significant until after the mass extinction has
run its course. Indeed its main significance may be as a vital
negative feedback loop for drawing down atmospheric CO2
(FIG. 1). As already noted, there is debate as to whether volcanic CO2 emissions are sufficient on their own to cause
these environmental changes; other phenomena such as
gas hydrate release may also contribute to increasing greenhouse gas concentrations.
The volcanic greenhouse scenario is currently a working
hypothesis for several marine extinction events, but its relevance to contemporaneous terrestrial extinction events
has not been explored to any great extent. Severe global
warming will obviously shift climatic belts and presumably
restrict habitat area for the most cold-adapted communities. However, tropical habitats should benefit rather than
suffer from such changes. Consequently most terrestrial
extinction mechanisms focus on other aspects of LIP erup-

294

D ECEMBER 2005

tions. Volcanic halogen emissions can potentially damage the ozone layer, thus raising the
spectre of UV radiation as a contributory cause
of extinctions. Mutant and deformed plant
spores and pollen in the end-Permian extinction interval may be a sign of this radiation
damage (Visscher et al. 2004); it is certainly evidence for extreme environmental stress, but it
has yet to be established if such phenomena are
a regular feature of LIP eruptions. Acid rain
from volcanogenic sulphate aerosols is another
potentially harmful effect of flood basalt eruptions, but there is, as yet, little direct evidence
for this.

VOLCANI
VOLCANIC

GREENHOUSE SCENARIO

1. Eruption of large igneous province

2. Release of large volumes of volcanic CO2

3. Global warming

CASE EXAMPLES
Volcanismextinction scenarios have been
developed primarily to explain end-Permian
and Early Jurassic extinction events, and it is
interesting to compare these with environmental changes seen during some other LIP eruptions of the past 300 Myr.

Central Atlantic Magmatic Province

(CAMP) (200 Ma)
The realisation that the Central Atlantic
Magmatic Province was both very large and of
the right age to be implicated in the endTriassic mass extinction suggested yet another
important volcanismextinction link (Marzoli
et al. 1999). This extinction event has proved
rather difficult to study, primarily because of a
dearth of complete marine boundary sections.
This paucity reflects the extremely low sea level
at this time, which could of course have contributed to the marine extinctions (Hallam and
Wignall 1999). Other changes at this time
include a brief warming pulse, carbon isotope
evidence for a significant release of methane
and possibly a calcification crisis (Hautmann
2004). Marine anoxia was widespread, but this
seems to have been the case both before and
after the extinctions.

4. Gas hydrate release,

methane oxidation to CO2

9. Calcification crisis
in ocean surface waters

7. Dissolved oxygen
levels decline

5. Negative
shift of 13C

8. MARINE MASS
EXTINCTION
12. Increase
of Re and Os
fluxes to oceans

10. Increased weathering

and global runoff

11. Increased nutrient flux

and ocean productivity levels

13. Increased burial

of marine organic
C, drawdown of
atmospheric CO2

Flow chart showing suggested chain of environmental

events caused by the eruption of large igneous provinces.
The chart is colour coded to distinguish between observed facts (yellow), inferred consequences for which there is overwhelming geological evidence (purple) and other, somewhat more tentatively inferred
consequences for which there is not always supporting evidence
(red).

FIGURE 1

Clearly the CAMP volcanism may have been responsible for

these environmental changes, but there are some problems
associated with a CAMPextinction link. First, there is the
detailed timing. The sedimentary record of the Newark
Basin in northeastern USA contains evidence for both terrestrial extinctions and flood basalt volcanism; however,
the first basalt occurs somewhat above the extinction horizon (Wignall 2001). Furthermore, it is not at all clear that
the end of the Triassic was marked by a sudden, single-pulse
mass extinction (Hallam 2002). Extinction rates were high
throughout the last few million years of the Triassic, suggesting a prolonged crisis that began considerably before
the CAMP eruptions.

ParanEtendeka Province (133 Ma)

The ParanEtendeka Province was erupted in southern
Gondwana (SE South America and Namibia) early in the
Cretaceous (Renne et al. 1992), during the later half of the
Valanginian Stage. Until recently, it was thought little of
interest happened in the oceans at this time. However,
recent studies have revealed an extremely watered-down
version of the effects seen during the Early Jurassic and endPermian crises. Thus, a thin, widespread, late Valanginian
black shale has been found in ocean cores, indicating an
episode of oxygen-poor deposition that has been called the
Weissert Event (Erba et al. 2004). The calcareous nanno-

ELEMENTS

6. Ocean thermohaline
circulation decreases

plankton fossil record shows a calcification crisis at the

same time, which may reflect an increase in oceanic nutrient levels (calcareous nannoplankton are thought to prefer
low-nutrient conditions) or acidification of ocean surface
waters (Erba 2004) or both. This crisis did not cause extinctions and in fact proved something of a spur to evolution
because planktonic foraminifera begin a Cretaceous-long
radiation of new species immediately after the Weissert Event.
Unlike other intervals marked by LIP eruptions, it is unclear
if any substantial global temperature changes occurred at
this time. A case can be made for cooling in the last stages of
the anoxic event, which probably reflects CO2-drawdown.
This could be due to organic matter burial during the
Weissert Event (Erba 2004), rather than be of volcanic origin. There is no negative carbon isotope anomaly associated
with this event, and so methane release from gas hydrate is
not likely.

295

D ECEMBER 2005

Ontong Java Plateau (120 Ma)

A substantial part of the vast Ontong Java Plateau of the SW
Pacific was erupted around the BarremianAptian boundary
of the Early Cretaceous (Courtillot and Renne 2003).
Eruption of this great volume of oceanic volcanic rocks
slightly predated the oceanic anoxic event, known as the
Selli Event, at 119 Ma (Larson and Erba 1999). However,
just prior to this event there was a nannoconid crisis during which these very small, calcareous plankton suddenly
became rather rare. This could reflect a calcification crisis,
due to volcanogenic CO2 input, or a fertilisation crisis that
did not favour the low nutrientadapted nannoconids (Erba
2004). Erba further suggests that direct warming of the
ocean water by the lava pile may have contributed to the
break-down of ocean stratification and expansion of the
oxygen-minimum zone. Both the warming and the fertilisation may have contributed to the anoxic event, but this
effect was curiously delayed relative to the eruptions. The
re-establishment of normal oceanic conditions after the
Selli Event saw the reappearance of the missing nannoconids. Thus, the crisis was only temporary and not an
extinction event.

CaribbeanColombian Plateau (90 Ma)

The 133 and 119 Ma oceanic anoxic events appear to have
precipitated only minor extinction crises when they are
compared to the enormous losses of earlier mass extinctions. The next volcanismanoxia event in the Cretaceous
occurred at the CenomanianTuronian boundary, and this
time it did coincide with rather more extinctions, notably
of several planktonic foraminifera species (Wan et al. 2003).
This interval also marks the culmination of Cretaceous
greenhouse warming and sea level rise. It thus has many of
the hallmarks of other volcanogenic crises, although there
is only weak evidence for a calcification crisis and no evidence for methane release. This 90 Ma event coincided with
the eruption of a LIP in the CaribbeanColombian region
and probably part of the Kerguelen LIP in the Indian Ocean,
and also with some flood basalts in Madagascar (Kerr 1998).
There were clearly a lot of volcanic culprits to choose from
at this time.
Proposed kill mechanisms for the CenomanianTuronian
extinctions include poisoning by trace metals derived from
oceanic volcanism (Erba 2004), but this proposition is
rather difficult to test. The anoxic event itself provides the
most obvious cause of the marine extinctions, and the contribution of volcanism to global warming and fertilisation
of the oceans provides a justification for linking volcanism
and anoxia (Sinton and Duncan 1997). Furthermore, the
oceanic volcanism may also have caused additional warming effects in addition to the direct input of volcanogenic
CO2 to the atmosphere. Warming of the oceans by the lavas
and oceanic acidification (by volcanic SO2 release) would
both have released CO2 to the atmosphere, thus exacerbating a warming trend (Kerr this issue).

The Deccan Traps (65 Ma)

Evaluating the global environmental influence of the
Deccan Trap eruptions in India is problematic due to the
difficulty of disentangling the effects of the well-known,
coeval Chicxulub impact event. However, thanks to prolonged and intensive study, the detailed chronology of volcanism and climate change in the Maastrichtian Stage, during the lead up to the end-Cretaceous mass extinction, are
now established with some clarity. The mid-Maastrichtian
was rather a cool interval, but a rapid phase of warming
began around 400 kyr before the KT boundary
(Abramovich and Keller 2003). This was reversed by a rapid
cooling trend around 100 kyr before the boundary, when

ELEMENTS

the 45C temperature gain was lost. The cooling coincides

with a sharp sea level fall, and a lowstand was reached
shortly before the KT boundary. Thereafter, sea level began
rising again across the boundary (Hallam and Wignall
1999).
These substantial oscillations in climate and sea level did
not cause much in the way of extinctions. Keller (2003) has
shown that the latest Maastrichtian warming pulse was
associated with a destabilisation of planktonic foraminiferal
populations and short-lived blooms of stress-tolerant
species. According to Keller these may reflect the expansion
and intensification of the mid-water oxygen-minimum
zone. However, interesting though they are, these changes
cannot compare with the near-total and abrupt mass
extinction of planktonic foraminifera (and various other
groups) at the end of the Cretaceous.
The possibility that the Deccan Trap eruptions were implicated in some or all of these changes has of course been
known for some time. However, only recently has it been
appreciated that the main eruptive phase coincided with
the late Maastrichtian warm pulse (Ravizza and PeuckerEhrenbrink 2003). The release of volcanic CO2 is the most
likely driver of environmental change, with a calcification
crisis in the oceans and global warming of the order of 4C
the most direct consequences. Thus, like the other LIP eruptions of the Cretaceous, the Deccan Trap eruptions appear
to have caused significant climatic effects, but only modest
biotic effects, perhaps because the oceans did not become
anoxic. It has been argued that the biosphere was already
rather stressed at the moment of meteorite impact, but
without that impact one suspects the end-Maastrichtian
event would only have ranked alongside minor Cretaceous
crises such as the Selli Event (White and Saunders 2005).

North Atlantic Igneous (Brito-Arctic)

Province (55 Ma)
The climatic events at 55 Ma, around the Palaeocene
Eocene (PE) boundary, have received ample study and are
reasonably well understood. Thus, a sharp negative 13C
excursion is generally taken as the signature of gas hydrate
release, which in turn is held responsible for the brief (120
kyr) warming pulse at this boundary (Kennett and Stott
1991). Contemporary changes in the oceans include a calcification crisis and the development of oxygen-poor deep
waters, which caused the extinction of many of the species
living there. However, this was not a time of mass extinction by any stretch of the imagination. In fact extinction
rates at this time were some of the lowest ever recorded.
These climatic and oceanic changes are very similar to the
changes observed during Cretaceous LIP eruptions, and in
this case they may relate to the eruption of the North
Atlantic Igneous Province. However, this province seems to
have been formed in two discrete pulses, with the younger
pulse coinciding with the PalaeoceneEocene thermal maximum at 55 Ma, and the older eruptive phase coinciding
with a rather cool interval (Courtillot and Renne 2003). In
a recent study of three marine PE boundary sections,
Schmitz et al. (2004) noted that the thermal maximum
coincides with the onset of an unusual phase of intense,
explosive basaltic volcanism in the North Atlantic region.
Curiously the release of dust and aerosols should have produced cooling rather than the observed warming.

FINAL THOUGHTS
Perhaps the most intriguing question arising from the link
between LIPs and environmental changes concerns the
remarkably different magnitudes of the supposed volcanogenic effects. Thus, the Early Jurassic climatic and envi-

296

D ECEMBER 2005

ronmental changes are closely comparable to those proposed for the end-Permian crisis. Very similar changes also
occurred during the PalaeoceneEocene thermal maximum,
but a mass extinction event has not been recorded.
However, this last event was of much briefer duration and
may not have lasted long enough to wreak the devastation
of the earlier events.
In summary, large igneous province eruptions can cause
changes that range from interesting but benign
(PalaeoceneEocene boundary), to severely damaging (Early
Jurassic), to utterly catastrophic (end-Permian). A partial
solution to this problem of variable influence may be found
in modelling work. For example, Dessert et al. (2001) have
suggested that factors such as pre-eruption atmospheric
CO2 levels and the rate of eruption are key variables in any
climatic changes. The closest correspondence between
eruptions and extinctions coincides with the Pangean
world, when most of the continents were part of a single
supercontinent. It may be that such a configuration was less
able to cope with sudden influxes of CO2 into the atmosphere because chemical weathering (the main mechanism
of CO2 drawdown over geological timescales) would have
been more limited in the arid interior of such a vast
continent.

WANTED
The Hudson Institute of Mineralogy, a not-for-profit
organization chartered by the Board of Regents of the State
University of New York is seeking used analytical equipment, thin sections and mineral specimens for its descriptive mineralogical laboratory and educational programs.
We are dedicated to classical mineralogical research,
preservation of mineral specimens, and educational outreach to primary and secondary school teachers and students. If your institution is upgrading its analytical equipment, we want your used, working devices. Further, if you
are disposing of minerals, thin sections or similar geological artifacts, let us put them to good use; aesthetics are
unimportant, labels are! Please contact:

The Hudson Institute of Mineralogy

PO Box 2012 Peekskill, NY 10566-2012
www.hudsonmineralogy.org

ACKNOWLEDGMENTS
This paper has benefited from the comments of Gerta Keller
and Andrew Kerr. .

REFERENCES
Abramovich S, Keller G (2003) Planktonic
foraminiferal response to the latest
Maastrichtian abrupt warm event: a case
study from South Atlantic DSDP Site
525A. Marine Micropaleontology 48:
225-249
Courtillot V, Renne PR (2003) On the ages
of flood basalt events. Compte Rendus
Acadmie des sciences 335: 113-140
Dessert C, Dupr B, Francois LM, Schott J,
Gaillardet J, Chakrapani G, Bajpai S
(2001) Erosion of Deccan Traps determined by river geochemistry: impact on
the global climate and the 87Sr/86Sr ratio
of seawater. Earth and Planetary Science
Letters 188: 459-474
Erba E (2004) Calcareous nannofossils and
Mesozoic oceanic anoxic events. Marine
Micropaleontology 52: 85-106
Erba E, Bartolini A, Larson RL (2004)
Valanginian Weissert oceanic anoxic
event. Geology 32: 149-152
Gattuso J-P, Buddemeier RW (2000) Ocean
biogeochemistry: calcification and CO2.
Nature 407: 311-313
Hallam A (2002) How catastrophic was the
end-Triassic mass extinction? Lethaia 35:
147-157
Hallam A, Wignall PB (1997) Mass
Extinctions and their Aftermath. Oxford
University Press, Oxford, 320 pp
Hallam A, Wignall PB (1999) Mass
extinctions and sea-level change. EarthScience Reviews 48: 217-250
Hautmann M (2004) Effect of end-Triassic
CO2 maximum on carbonate sedimentation and marine mass extinction. Facies
50: 257-261

ELEMENTS

Keller G (2003) Biotic effects of impacts

and volcanism. Earth and Planetary
Science Letters 215: 249-264
Kennett JP, Stott LD (1991) Abrupt deepsea warming, palaeoceanographic
changes and benthic extinctions at the
end of the Palaeocene. Nature 353:
225-229
Kerr AC (1998) Oceanic plateau formation:
a cause of mass extinction and black
shale deposition around the CenomanianTuronian boundary? Journal of the
Geological Society of London 155:
619-626
Larson RL, Erba E (1999) Onset of the MidCretaceous greenhouse in the BarremianAptian: Igneous events and the biological, sedimentary, and geochemical
response. Paleoceanography 14: 663-678
Marzoli A, Renne PR, Piccirillo EM, Ernesto
M, Bellieni G, DeMin A (1999) Extensive
200-million-year-old continental flood
basalts of the central Atlantic magmatic
province. Science 284: 616-618
Plfy J, Smith PL (2000) Synchrony
between Early Jurassic extinction, oceanic
anoxic event, and the Karoo-Ferrar flood
basalt volcanism. Geology 28: 747-750
Ravizza G, Peucker-Ehrenbrink B (2003)
Chemostratigraphic evidence of Deccan
volcanism from the marine osmium
isotope record. Science 302: 1392-1395

volcanism, but no comet impact, at

the PaleoceneEocene boundary: highresolution chemical and isotopic records
from Egypt, Spain and Denmark. Earth
and Planetary Science Letters 225: 1-17
Sinton CW, Duncan RA (1997) Potential
links between ocean plateau volcanism
and global ocean anoxia at the
Cenomanian-Turonian boundary.
Economic Geology 92: 836-842
Visscher H, Looy CV, Collinson ME,
Brinkhuis H, van Konijnenburg-van
Cittert JHA, Krschner WM, Sephton MA
(2004) Environmental mutagenesis
during the end-Permian ecological crisis.
Proceedings of the National Academy of
Sciences 101: 12952-12956
Wan X-Q, Wignall PB, Zhao W-J (2003)
The Cenomanian-Turonian extinction
and oceanic anoxic event: evidence from
southern Tibet. Palaeogeography Palaeoclimatology Palaeoecology 199: 283-298
White RV, Saunders AD (2005) Volcanism,
impact and mass extinctions: incredible
or credible coincidences. Lithos 79:
299-316
Wignall PB (2001) Large igneous provinces
and mass extinctions. Earth-Science
Reviews 53: 1-33 .

Renne PR, Ernesto M, Pacca IG, Coe RS,

Glen JM, Prvot M, Perrin M (1992)
The age of Parana flood volcanism,
rifting of Gondwanaland, and the
Jurassic-Cretaceous boundary. Science
258: 975-979
Schmitz B, Peucker-Ehrenbrink B,
Heilmann-Clausen C, berg G, Asaro F,
Lee C-TA (2004) Basaltic explosive

297

D ECEMBER 2005