large igneous provinces: origin and environmental...

December 2005Volume 1, Number 5

ISSN 1811-5209

The Mantle Plume Hypothesis

Large Igneous Provinces, Delamination,

and Fertile Mantle

Meteorite Impacts as Triggers to

Large Igneous Provinces

Gas Fluxes from Flood Basalt Eruptions

Oceanic LIPs: The Kiss of Death

Large Igneous Provinces and

Mass Extinctions

Large Igneous Provinces:Origin and Environmental

ConsequencesThe Mantle Plume Hypothesis

Large Igneous Provinces, Delamination,

and Fertile Mantle

Meteorite Impacts as Triggers to

Large Igneous Provinces

Gas Fluxes from Flood Basalt Eruptions


Large Igneous Provinces and

Mass Extinctions

http://www.elementsmagazine.org/

Large Igneous Provinces:Origin and EnvironmentalConsequences Andrew D. Saunders, Guest Editor

Large Igneous Provinces and the Mantle Plume Hypothesis Ian H. Campbell

Large Igneous Provinces, Delamination,and Fertile Mantle Don L. Anderson

Meteorite Impacts as Triggers to Large Igneous Provinces Adrian P. Jones

Gas Fluxes from Flood Basalt Eruptions Stephen Self, Thorvaldur Thordarson, and Mike Widdowson

Oceanic LIPs: The Kiss of DeathAndrew C. Kerr

The Link between Large Igneous Provincesand Mass Extinctions Paul Wignall

DepartmentsEditorial . . . . . . . . . . . . . . . . . . . . . . . . 251From the Editors . . . . . . . . . . . . . . . . . . . 2522006 Preview . . . . . . . . . . . . . . . . . . . . . .253Letters to the Editors . . . . . . . . . . . . . . . . 255Triple Point . . . . . . . . . . . . . . . . . . . . . . . 257Meet the Authors . . . . . . . . . . . . . . . . . . 264People in the News . . . . . . . . . . . . . . . . . . 276Book Review – Atlas of Meteorites . . . . . . . . . . 282The Editors’ Pick . . . . . . . . . . . . . . . . . . . 288Society News . . . . . . . . . . . . . . . . . . . . . . 298Publication Forum – GeoScienceWorld . . . . . . . 313Conference Report – 13th ICC . . . . . . . . . . . . 314Travelogue – Antarctica . . . . . . . . . . . . . . . . 316Calendar . . . . . . . . . . . . . . . . . . . . . . . . 317Parting Shot . . . . . . . . . . . . . . . . . . . . . . .319Advertisers in this Issue . . . . . . . . . . . . . . . 320

249

Volume 1, Number 5 • December 2005ABOUT THE COVER:

Stacks of pahoehoelava sheet lobes (dark

layers) from a largenumber of flow fields make up the lava pile of

the Deccan floodbasalt province, seenhere in the Western

Ghats of centralwestern India where a

thickness of more thanone kilometre isexposed. PHOTO

M. WIDDOWSON

259

271

265

277

289

293

Elements is published jointly by TheClay Minerals Society, the EuropeanAssociation for Geochemistry, theMineralogical Society of America, theMineralogical Society of Great Britainand Ireland, the International Associationof GeoChemistry, the MineralogicalAssociation of Canada, and theGeochemical Society. It is provided asa benefit to members of these societies.

From 2006, Elements will be published sixtimes a year. Individuals are encouragedto join any one of the participatingsocieties to receive Elements. Institutionalsubscribers to any of the followingjournals – American Mineralogist, TheCanadian Mineralogist, Clays and ClayMinerals, Mineralogical Magazine andClay Minerals – will also receive Elementsas part of their 2006 subscription.Institutional subscriptions are availablefor US$125 a year in 2006. Contact themanaging editor ([email protected]) for information.

Copyright ©2005 by the MineralogicalSociety of America

All rights reserved. Reproduction in anyform, including translation to otherlanguages, or by any means – graphic,electronic or mechanical, includingphotocopying or information storageand retrieval systems – without writtenpermission from the copyright holderis strictly prohibited.

Publications mail agreement no. 40037944

Return undeliverable Canadian addresses to:PO Box 503 RPO West Beaver Creek Richmond Hill, ON L4B 4R6

Printed in Canada ISSN 1811-5209www.elementsmagazine.org

283

INTRODUCTIONIt has been nearly 15 years since the term ‘large igneousprovince’ was introduced by Mike Coffin and Olaf Eldholm(1991, 1994). An umbrella term to include continentalflood basalt provinces, oceanic plateaus, volcanic riftedmargins and aseismic ridges, it rapidly entered commonparlance even though it was, and remains, loosely defined.The key aspect of large igneous provinces (LIPs) is that theyrepresent anomalously high magmatic fluxes. The magmais usually basaltic, but may be rhyolitic. They are large inarea, covering many thousands if not millions of squarekilometres, and they testify to unusual geological processes,involving large amounts of thermal energy. Where thisenergy comes from – deep within the Earth as a mantleplume, from a meteorite impact, or from sinking of denseroots from the base of the continental crust or lithosphere(‘delamination’) – is a matter of considerable debate. As willbe seen from the papers in this issue of Elements, there isalso debate about their environmental effects. Could theformation of such provinces cause the collapse of ecosys-tems, either by interrupting oceanic circulation systems orby releasing large masses of volcanic aerosols, and triggermass extinctions? Certainly, the timing of LIPs and massextinctions suggests some causality, but we do not, as yet,understand its nature.

LIPS AND LIPSNo two LIPs are the same. Just as Read (1948) recognisedthat there are ‘Granites and Granites,’ the term LIPs encom-passes a wide range of geological structures and processes.For the purpose of this issue, we are focusing on the conti-nental flood basalt provinces and their oceanic equivalents,the oceanic plateaus. Volcanic trails (forming aseismicridges across the ocean floor), which often lead away from

the main LIP, are an importantpart of the story, but will not beconsidered in detail here.

The locations and ages of the mainLIPs are shown in FIGURE 1. Thisselection is biased: it does notinclude the large silicic provinces(e.g. Chon Aike in southernmostSouth America or the Sierra Madrein Mexico); the majority of theLIPs in Figure 1 are predominantlybasaltic. It is also ageist: it does notinclude any LIP older than 250 Ma,for example, the EmeishanProvince in China (Permian) and

the numerous Proterozoic and Archaean LIPs. It is impor-tant to stress that LIP formation has occurred throughoutEarth history and not just in the Mesozoic and Cenozoic,although they may occur in cycles (e.g. Ernst et al. 2005);there is an increase in LIP formation in the Cretaceous (Lar-son 1991), and Prokoph et al. (2004) have suggested cyclesof LIP formation.

The information database is also strongly skewed in favourof the continental flood basalt provinces. Accessibility tothe deeper parts of the dissected and faulted volcanic pilemeans that a greater range of compositions and a widerrange of ages can be sampled in the continental sequencesthan in the oceanic plateaus. Some oceanic plateaus have,however, subsequently collided with a volcanic arc or acontinental margin, with the result that important infor-mation can be obtained from the deeper crustal sections.Thus, the collision of the Caribbean Plateau with SouthAmerica has provided a wealth of information about its pet-rogenesis (Kerr et al. 1997). Similarly, the collision of theOntong Java Plateau with the Solomon Islands allows us towalk through the top three kilometres of the plateau’sbasaltic crust. Otherwise we would be totally reliant oncored material from the top few hundred metres of crust –metaphorically, pin-pricking the elephant (Tejada et al.2004). The basements of many other plateaus are, however,sampled entirely by drilling and dredging (e.g. KerguelenPlateau) or remain unsampled.

FORMATIONThere are almost as many theories and models for the for-mation of LIPs as there are individual provinces. I haveattempted to summarise these models, and some of the pre-dictions that arise, in TABLE 1. It is likely that no singlemodel can account for all LIPs, and the predictions fromeach model are not fully understood or known. Most mod-els agree that large amounts of thermal energy are requiredin order to produce large volumes of magma over a geolog-ically short period of time. Given that most LIPs are

259E L E M E N T S , V O L . 1 , P P . 2 5 9 – 2 6 3 DECEMBER 2005259

Andrew D. Saunders1

1. Department of Geology University of LeicesterLeicester LE1 7RH, UKE-mail: [email protected]

Episodically, the Earth erupts large quantities of basaltic magma ingeologically short periods of time. This results in the formation of largeigneous provinces, which include continental flood basalt provinces,

volcanic rifted margins, and giant oceanic plateaus. These fluctuations in theEarth’s system are still poorly understood. Do they owe their origin to mantleplumes, meteorite impacts, or lithosphere-controlled processes? Whatevertheir origin they correlate closely with major changes in oceanic andatmospheric chemistry and may trigger global mass extinctions.

KEYWORDS: Continental flood basalts, oceanic plateaus, mass extinctions,

mantle convection and temperature

Thousand-metre basaltcliffs, Kivioqs Fjord,

East Greenland. PHOTO IAN PARSONS

Large Igneous Provinces:Origin and EnvironmentalConsequences

Large Igneous Provinces:Origin and EnvironmentalConsequences

260E L E M E N T S DECEMBER 2005

basaltic, this requires some form of energy source in themantle. The required energy may be reduced slightly if themantle source is highly fertile (see glossary). This would bethe case if, for example, it contains large amounts of eclog-ite (see glossary and Anderson this issue) or is volatile rich.

What is the source of this energy? And how much isrequired? Over time, the world’s oceanic ridge system sup-plies a remarkably constant amount of basaltic magma. Thethickness of the ocean crust – ignoring the sediments – isvery uniform, at about seven kilometres. There are differ-ences. Very slow-spreading ridges, such as the SouthwestIndian Ridge, create anomalously thin crust, to the pointwhere it may even be absent. Similarly, ocean crust nearmajor transform faults may be thin. And, conversely, insome areas (remarkably rare), such as Iceland, the crust isanomalously thick (perhaps as much as 35 km). But thebulk of the ocean crust is broadly uniform in thickness andcomposition, which is remarkable given that two importantvariables – source temperature and source composition –can have a dramatic effect on the volume and type of basaltproduced. An increase of 100°C in the potential tempera-ture (see glossary) of the mantle source will more than dou-ble the amount of melt produced, and hence double thethickness of the ocean crust. The source temperature of nor-mal mid-ocean ridge basalts is unlikely, therefore, to varysignificantly.

So how do we generate the high crustal thicknesses foundin LIPs (typically 35 km for an oceanic plateau)? There areseveral ways of doing this. First, we can increase the tem-perature of the source. A straightforward increase in sourcepotential temperature from 1300°C to 1500°C can producea 30+ km thick layer of melt. This is at the heart of theplume model (White and McKenzie 1989; Campbell thisissue), where heat energy is transferred in a mass of mantleascending from a thermal boundary layer deep in the Earth(e.g. the core–mantle boundary). Second, we can increasethe rate at which source material is processed through thezone of partial melting. Rather than passive upwelling (as isthought to occur at mid-ocean ridges), the mantle rockactively convects into and through the zone of partial melt-ing. Combined with higher temperatures, this provides apotent model for large-volume melt generation, and isagain implicit in the plume model. Rapid fluxing may beparticularly important during the start-up phase of themantle 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 (Jonesthis issue) also invokes high source temperatures, inducedby kinetic energy following meteorite impact. And third, wecan increase the fertility and volatile content of the sourceto create more melt. None of these three factors – tempera-ture, mantle ascent rates, and source composition – areexclusive; indeed there is every reason to believe they mayoccur together.

What is the evidence for high mantle temperatures duringLIP formation? The most direct evidence is the occurrenceof highly magnesian melts (preserved in high-Mg basalts,picrites, and komatiites). These are found in several LIPsbut, importantly, not all, and this has been used as evidenceagainst an excessively hot mantle source (Anderson thisissue). To counter this, it should be remembered that mag-nesian melts are more dense than normal basalt and may betrapped 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 notnecessarily evidence for the absence of them.

260

Map showing the distribution of the main Mesozoic andCenozoic large igneous provinces (continental flood

basalts and basaltic oceanic plateaus), modified after Coffin and Eld-holm (1991). Also included are present-day hotspots (red spots), whichmay be related to individual LIPs (e.g. Réunion to the Deccan Traps;Galapagos to the Caribbean Plateau) via plate reconstructions andvolcanic chains. The Siberian Traps are shown buried (striped ornamen-tation) beneath the West Siberian Basin. CAMP: Central Atlantic Mag-matic Province, located along the eastern edge of North and SouthAmerica, and western edge of North Africa and southern Europe. Cre-taceous Plateaus include the Hess and Shatsky Rises, and Ontong Javaand Manihiki Plateaus.

FIGURE 1


Arguments against very high mantle temperatures are sup-ported by limited uplift in the vicinity of some LIPs. A hotmantle source beneath the lithosphere would be expectedto cause significant uplift, especially if it were dynamicallyemplaced by a mantle plume (Campbell this issue), but insome provinces the evidence for such uplift is ambiguous(Anderson this issue). Readers may wish to read the recentwork by Burov and Guillou-Frottier (2005), however, whichsuggests that the amount of uplift above a plume may besmall, or absent, in some circumstances.

If mantle plumes or impacts are not the main generators,what other mechanisms may be responsible for LIP forma-tion? King and Anderson (1995) noted that many conti-nental flood basalt provinces lie close to the edges ofArchaean cratons. They proposed that thermal insulationby the craton raises the temperature of the underlyingasthenosphere, which then flows sideways out frombeneath the craton and under thinner lithosphere. As itascends, the mantle decompresses and melts. In this modelthe mantle need not be as hot as in the plume model. Fur-thermore, King and Anderson (1998) argued that ‘edge-driven convection’, where a secondary convection cell isestablished at the craton margin, could increase magmaproduction rates and volumes. An alternative model, butalso involving displacement of upper mantle, is the delam-ination model, in which dense lower crust and the attachedlithospheric mantle sink into the mantle, and upper man-tle flows into the ensuing space, decompressing and melt-ing (Elkins-Tanton 2005; Anderson this issue). This modelpredicts substantial uplift as the lithosphere rebounds, butagain does not necessarily produce high-Mg melts. It is alsodebatable whether the required volumes of magma, andmagma of the right composition, could be produced in thisway from normal-temperature asthenospheric mantle.

My personal view is that most, if not all, LIPs can beexplained by mantle plumes, and that the best evidence –picritic rocks – for the high source temperatures is oftenhidden in the deeper crust or upper mantle. Variations insource composition doubtless play a role in the composi-tion and amounts of liquid that are generated, but at theheart of the model is a mechanism for the release of

thermal energy originating from deep within the Earth. Thelithosphere 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 ofthem appear to be plume generated.

ENVIRONMENTAL EFFECTS AND MASS EXTINCTIONSThere are several ways in which a LIP could affect the globalenvironment. One is through the immediate, eruptiverelease of gas and aerosols. As shown by Self et al. (thisissue), a large basaltic lava eruption can release prodigiousquantities of SO2, CO2 and halogens, the effects of whichwe are only beginning to appreciate. The initial release ofSO2 and its injection into the stratosphere could trigger aglobal volcanic winter, akin to the models of nuclear win-ters, reducing photosynthesis through light occlusion andcooling. Long-term accumulation of CO2 may lead to sub-sequent warming – a volcanic summer – especially if thebiologically driven carbon-capture mechanisms are com-promised by the preceding volcanic winter. However, aspointed out by several workers, including Self et al. (thisissue) and Wignall (this issue), the average flux of CO2

released by a LIP over its entire history is not large – muchless than the current annual production of anthropogenicCO2, for example.

The evidence that there is a link between LIPs and the envi-ronment is indicated by the close coincidence between LIPsand mass extinctions (FIG. 2), as noted by Vincent Cour-tillot in 1994 and subsequently developed in his book Evo-lutionary Catastrophes (1999). But how does a LIP, or floodbasalt event, trigger a mass extinction? What other indica-tors are there of climate change? One is a rapid shift in thecarbon isotope record at the time of the extinctions. Suchisotopic variations indicate massive changes in seawaterand atmospheric composition, requiring the addition of bil-lions of tonnes of carbon to the atmosphere and oceanreservoirs. This carbon, it is argued, as CO2 in the atmos-phere, leads ultimately to powerful global warming, loss ofhabitat, and mass extinction (Kiehl and Shields 2005). But

Prediction Excess Regional High-T magmas Extraterrestrial Hotspot trail Currentlymantle-derived uplift/doming (e.g. picrites material and leading from LIP active hotspot

Model magmatism and komatiites) impact breccias

Mantle plume Yes Likely Likely No Likely Possible (unless the plume (before and/or (but dense melts may impinges on the base during magmatism) not reach the surface)of thick lithosphere)

Meteorite impact Likely Likely Yes Yes Possible Possible (during magmatism) (probably abundant)

Edge model, with Possible Likely Unlikely No Unlikely Unlikelyenhanced mantle (if mantle can ascend (during magmatism) convection sufficiently to

decompress and melt)

Delamination Possible Likely Unlikely No Unlikely Unlikely(if mantle can ascend (could be substantialsufficiently to during or afterdecompress and melt) magmatism

Melting of Possible Unlikely Unlikely No Unlikely Unlikelyfertile mantle without excess heat

PREDICTIONS ARISING FROM VARIOUS MODELS OF LIP FORMATION

TABLE 1

where does this carbon come from? Some undoubtedlycomes from the basalts themselves but, given the low aver-age rates of CO2 production, this is unlikely to be the entirestory. An alternative, mentioned by both Wignall and Kerrin this issue, is that the greenhouse effects of CO2 from theLIPs slowly raise the atmospheric and oceanic temperatures,and this triggers release of methane previously trapped inpermafrost and methane hydrates on the seafloor. In effect,a threshold is reached, potentially leading to a runawaygreenhouse. (Intriguingly, it has recently been reported thatSiberian permafrost is melting due to anthropogenicallydriven global warming; perhaps the flood basalts offer amodel for current climate change.) An alternative explana-tion is that near-surface intrusions that accompany LIP for-mation are injected into carbon-rich sedimentary layers(methane- or coal-bearing) and that these then release theircarbon into the ocean and atmosphere (Svensen et al. 2004;McElwain et al. 2005).

The key to understanding these processes is knowing theduration and flux rates of LIP magmatism, because fromthese we can calculate the flux rates of the climate-

modifying gases and aerosols. 40Ar/39Ar and zircon U/Pbdating offer increasingly precise methods for improvingthis knowledge, but even a precision of better than 0.1%still leaves a lot to be desired. Mass extinction events mayoccur in periods of 100,000 years or less, which is stilloutside the precision offered by the best radiometric tech-niques for dating events that occurred during the massextinctions at the Permo-Triassic, Triassic–Jurassic, andCretaceous–Tertiary boundaries.

FURTHER READINGThe International Association of Volcanology and Chem-istry of the Earth’s Interior (IAVCEI) has established the LIPsSubcommision, which maintains a webpage with up-to-date 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 Geo-physical Union Monograph 100, 438 pp. .


Extinction rate versus time (continuous line, blue field)(multiple-interval marine genera, modified from Sepkoski

1996) compared with eruption ages of continental flood basalts (redbands). Three of the largest mass extinctions, the Permo-Triassic, Triassic–Jurassic and the Cretaceous–Tertiary, correspond to eruptions of the

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

FIGURE 2

REFERENCESBurov E, Guillou-Frottier L (2005) The

plume head-continental lithosphereinteraction using a tectonically realisticformulation for the lithosphere.Geophysical Journal International 161:469-490

Coffin MF, Eldholm O (eds) (1991) LargeIgneous Provinces: JOI/USSAC WorkshopReport. The University of Texas at AustinInstitute for Geophysics Technical Report114: 79 pp

Coffin MF, Eldholm O (1994) Largeigneous provinces: crustal structure,dimensions, and external consequences.Review of Geophysics 32: 1-36

Courtillot V (1994) Mass extinctions in thelast 300 million years: one impact andseven flood basalts? Israeli Journal ofEarth Sciences 43: 255-266

Courtillot V (1999) Evolutionary Catastro-phes: The Science of Mass Extinctions.Cambridge University Press, 188 pp

Elkins-Tanton LT (2005) Continentalmagmatism caused by lithosphericdelamination. In: Foulger GR, NatlandJH, Presnall DC, Anderson DL (eds)Plates, Plumes and Paradigms. GeologicalSociety of America Special Paper 388, pp449-462

Ernst RE, Buchan KL, Campbell IH (2005)Frontiers in large igneous provinceresearch. Lithos 79: 271-297

Kerr AC, Tarney J, Marriner GF, Nivia A,Saunders AD (1997) The Caribbean-Colombian Cretaceous igneous province:


the internal anatomy of an oceanicplateau. In: Mahoney JJ, Coffin M (eds)Large Igneous Provinces: Continental,Oceanic, and Planetary Flood Volcanism.American Geophysical Union, Geophysi-cal Monograph 100: 123-144

Kiehl JT, Shields CA (2005) Climate simu-lation of the latest Permian: implicationsfor mass extinction. Geology 33(9): 757-760

King SD, Anderson DL (1995) Analternative mechanism of flood basaltformation. Earth and Planetary ScienceLetters 136: 269-279

King SD, Anderson DL (1998) Edge-drivenconvection. Earth and Planetary ScienceLetters 160: 289-296

Larson RL (1991) Latest pulse of Earth:Evidence for a mid-Cretaceous super-plume and geological consequences ofsuperplumes. Geology 19: 547-550

McElwain JC, Wade-Murphy J, Hesselbo SP(2005) Changes in carbon dioxide duringan oceanic anoxic event linked tointrusion into Gondwana coals. Nature435: 479-482

Prokoph A, Ernst RE, Buchan KL (2004)Time-series analysis of large igneousprovinces: 3500 Ma to present. Journalof Geology 112: 1-22

Read HH (1948) Granites and granites.In: Gilluly J (ed) Origin of Granite,Geological Society of America Memoir28: 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 Phanerozoicextinction: a perspective from global databases. In: Walliser OH (ed) Global Eventsand Event Stratigraphy, Springer, Berlin,pp 35-51

Svensen HS, Planke S, Malthe-Sørenssen A,Jamtveit B, Myklebust R, Eidem TR, ReySS (2004) Release of methane from avolcanic basin as a mechanism for initialEocene global warming. Nature 429: 542-545

Tejada MLG, Mahoney JJ, Castillo PR, IngleSP, Sheth HC, Weis D (2004) Pin-prickingthe elephant: evidence on the origin of

the Ontong Java plateau from Pb-Sr-Hf-Nd isotopic characteristics of ODP Leg192 basalts. In: Fitton JG, Mahoney JJ,Wallace PJ, Saunders AD (eds) Originand Evolution of the Ontong JavaPlateau. Geological Society of LondonSpecial Publication 229: 133-150

White R, McKenzie D (1989) Magmatismat rift zones: the generation of volcaniccontinental margins and flood basalts.Journal of Geophysical Research 94 (B6):7685-7729

White RV, Saunders AD (2005) Volcanism,impact and mass extinctions: incredibleor credible coincidences? Lithos 79: 299-316 .

δδ13C – Carbon isotope values (and otherstable isotopes such as oxygen andhydrogen) are expressed relative to a ref-erence standard. The standard used forcarbon is a Peedee Formation belemnite(PDB). The difference from the standardis expressed as the delta function, whichmay be positive (i.e. the carbon has a rel-atively higher abundance of heavy 13Cthan the standard), the same, or negative(a higher proportion of light 12C). Resultsare expressed in parts per thousand. Ashift or spike in the seawater isotopecurve indicates a geologically rapidchange in the relative amounts of lightand heavy carbon. Organic carbon, espe-cially biogenic methane, has low δδ13Cvalues; the δδ13C of marine carbonate isabout zero.

Basalt – A basic igneous rock of volcanicorigin with between 45 and 52 wt% SiO2and less than 5 wt% total alkalis. Themineralogy typically comprises clinopy-roxene (augite) and plagioclase feldspar.Olivine, and an opaque mineral such asmagnetite may also be present. The plu-tonic equivalent of basalt is gabbro.

Delamination – The collapse and peeling oflarge layers of dense material from thebase of the lithosphere or crust. Thedelamination process allows theasthenosphere to ascend into the result-ing space, triggering decompressionmelting and magmatism.

Eclogite – A high-pressure, high-densitymetamorphic rock composed mainly ofgarnet and clinopyroxene. It is the high-pressure equivalent of basalt or gabbro.Emplacement of basaltic magma into thelower crust may lead to the formation ofdense eclogite, which may becomebuoyantly unstable and collapse into theunderlying mantle (delamination). Sub-ducted ocean crust is thought to convertto eclogite and be entrained in the mantle,eventually returning to the near-surfaceby convection processes.

Flow field – The total lava products of oneeffusive eruption, however long-lasting.

Mantle fertility – The relationship betweenmantle composition and its ability to pro-duce melt. During partial melting, peri-dotite (the main mantle rock) can pro-duce only so much melt before itexhausts its supply of ‘basalt producing’elements, such as Ca and Al. The more ofthese elements present in the originalrock, the more melt can be produced – itmay be said to have an increased fertility.The presence of eclogite, which is chem-ically equivalent to basalt, substantiallyincreases the fertility of the source. Note,however, that energy, in the form oflatent heat of melting, is still required togenerate melt. Increasing source fertilitywill not substantially increase the volumeof melt unless that energy is also present.

Optical depth – A measure of how opaquea medium – such as air – is to the radia-tion passing through it. Solar radiation ispartially scattered and absorbed by fineparticles in the atmosphere, and so theamount of incident light is always greaterthan the amount of transmitted light. Acompletely transparent medium has anOD of zero. An OD of 1 results in ~40%of light reaching the ground.

Peridotite – An ultrabasic rock, with a min-eralogy dominated by olivine and withvariable amounts of clinopyroxene (e.g.diopside) and orthopyroxene (e.g. ensta-tite). Garnet or spinel may also be pres-ent. It is the predominant rock in theEarth’s mantle. Peridotite comprisingmostly olivine and orthopyroxene istermed harzburgite. Peridotite witholivine, orthopyroxene and clinopyrox-ene is termed lherzolite.

Picrite – A rock of volcanic origin withbetween 12 and 18 wt% MgO, less than3 wt% total alkalis (K2O + Na2O), and anSiO2 content between 30 and 52 wt%.Picrites are more magnesian – and more

primitive – than basalts and may beindicative of a high-temperature parentalmagma.

Plume buoyancy flux – A measure of thestrength of a plume, given by the differ-ence in density between the plume mantleand the surrounding mantle, multipliedby the buoyancy-driven volume flux ofthe plume.

Potential temperature – Rising, convectingmaterial (plastic mantle rock, or melt, orair) cools slightly by adiabatic decom-pression. This defines an adiabatic cool-ing line (or, conversely, a heating line ifthe material descends) – part of theEarth’s geotherm. For rock, the adiabaticgradient is about 0.5°C km-1. The theo-retical intersection of this line with theEarth’s surface is called the potential tem-perature. Thus, mantle with an actualtemperature of 1400°C at 100 km depthwill have a potential temperature of1400 − (0.5 × 100) = 1350°C, assumingthat the adiabatic gradient is linear.Potential temperature, or Tp, is a con-venient shorthand to describe how hotthe mantle is regardless of depth or, putanother way, how much energy it con-tains. Unfortunately we can only approx-imate the actual Tp of the upper mantle.Some workers (e.g. Anderson this issue)argue that the normal upper mantle hasa large temperature range, varying fromplace to place by ±100°C, whereas othersargue that the normal mantle is muchmore restricted in temperature, with a Tpof about 1300°C, and with localisedhotspots (plumes) where the Tp mayexceed 1500°C.

Viscosity – The resistance to flow within aliquid (alternatively, a measure of theinternal friction of a liquid, or dynamicviscosity). Kinematic viscosity is thedynamic viscosity of a liquid divided byits density.

Glossary

Don L. Anderson isProfessor Emeritus in theDivision of Geological andPlanetary Sciences atCaltech. He received a BSin geology and geophysicsand a DSc (Hon) from RPI,and a PhD in geophysics

and mathematics from Caltech. He receivedthe Crafoord Prize at the Royal SwedishAcademy of Sciences and the National Medalof Science at the White House. He is pastpresident of the AGU. He is interested in theorigin, evolution, structure, and compositionof Earth and other planets. His work inte-grates seismological, solid state physics,geochemical, and petrological data. He wasdirector of the Seismological Laboratory ofCaltech from 1967 to 1989. See http://www.gps.caltech.edu/~dla

Ian H. Campbell isa Senior Fellow at TheAustralian NationalUniversity and is currentlyvisiting the Institute forStudy of the Earth’sInterior, University ofOkayama at Misasa, Japan,

where he is supported by the Centre ofExcellence for the 21st Century. He is co-chairof the Commission on Large Igneous Pro-vinces and was a co-convener of the GreatPlume Debate held at Fort William betweenAugust 28 and September 1, 2005. Hisprincipal research interests are the relation-ship between igneous processes and oredeposits, and the evolution of the Earth’scrust and mantle.

Andrew C. Kerr, anative of Northern Ireland,grew up a few miles fromthe legendary GiantsCauseway columnarbasalts. He left NorthernIreland to study geochem-istry at the University of

St Andrews , followed by a PhD on the geo-chemistry of lavas from the Isle of Mull, atDurham University. This was followed by twopostdoctoral positions at the University ofLeicester: one studying the accreted portionsof the Caribbean–Colombian oceanic plateauand the other assessing the environmentalimpact of large igneous provinces. He is cur-rently a senior lecturer at Cardiff University,treasurer of the Commission on Solid EarthChemistry and Evolution, and a councilmember of the Mineralogical Society ofGreat Britain.

Adrian P.Jones is readerin petrology inthe Departmentof EarthSciences atUniversityCollege

London. With connections to Durham,Chicago, Caltech, and Kingston, he hassupervised approximately 20 graduatestudents. His research group explores themineralogy, geochemistry, and petrology ofcarbon in the Earth’s mantle, including theorigin of diamond, and of carbonatite andkimberlite melts. From 2000 to 2003 he waschairman of the ESF Eurocarb Network, whichexamined solid carbon reservoirs and theircontribution to CO2 in the atmospherethrough volcanism. He explained carbonatemelts in the Chicxulub crater and since 2000,has worked on large-scale impact melting.

Andrew D. Saundersis professor of geochemistryat the University ofLeicester. He has studiedlarge igneous provinces inMadagascar, the NorthAtlantic, and in the Pacificand Indian Oceans, using

a variety of techniques, but mostly igneousgeochemistry. He is currently investigatingthe role of the Siberian Traps in the end-Permian mass extinction and asking the basicquestion, how does such volcanism trigger aworldwide extinction event? Andy is a strongadvocate for mantle plumes, but is perfectlywilling to accept that this amazing planet issufficiently large and complex to accommo-date a variety of LIP-forming mechanisms.

Steve Self is professorof volcanology and headof the Volcano DynamicsGroup at the OpenUniversity in the UK.He has studied volcanicactivity and volcanic rocksin many parts of the world,

concentrating on explosive eruptions, large(flood) lava effusions, and the impact ofvolcanism on the atmosphere. He has writtenover 170 articles and reviews in the scientificpress, for books, and in other sciencemagazines. He is a fellow of the GeologicalSociety of America and the Geological Societyof London, and a member of the AmericanGeophysical Union, the InternationalAssociation of Volcanology, the MineralogicalSociety, and the International Association ofSedimentology. He is currently head of theUK’s Volcanic and Magmatic Studies Group.

Thorvaldur (Thor)Thordarson is a seniorresearcher in volcanologyat the University of Icelandand University of Hawaii.He has conducted vol-canological research onmodern to Archean

successions across the globe, with emphasison flood basalt and komatiite volcanism,volcano–climate interactions, and lava flowemplacement mechanisms. He has writtenmore than 50 articles in scientific journalsand co-authored the book Iceland in the seriesClassical Geology in Europe. He is a memberof the Geological Society of America, theAmerican Geophysical Union, the Interna-tional Association of Volcanology, the Geo-logical Society of Iceland, and the Glaciologi-cal Society of Iceland. He has been appointed,as of March 1, 2006, to the post of seniorlecturer in volcanology at the Universityof Edinburgh, Scotland.

Mike Widdowson islecturer in volcanology atthe Open University, UK.He received his degree ingeology and mineralogy atOxford (1985) where, afterstudying the uplift historyof the Deccan Traps of

India, he also obtained his D Phil (1990).He has over 20 years of experience studyingmany aspects of the Deccan continental floodbasalt province. His broader fields of researchinclude the volcanology, geochemistry andgeochronology of large igneous provinces,and the wider role of volcanism upon climatechange in the geological record. He has alsopublished papers on the evolution of passivecontinental margins and the geochemistryof lateritization and tropical weatheringprocesses. He is a fellow of the GeologicalSociety of London and the Geological Societyof India.

Paul Wignall is aprofessor of palaeoenviron-ments at the School ofEarth and Environment,University of Leeds. Hehas been interested in theorigin of the end-Permianmass extinction, ever since

he first visited the sections of the western US,in 1989. In recent years he has also becomeinterested in other mass extinctions, notablythe end-Triassic and Early Jurassic events. Hisresearches have taken him all over the worldto places such as Greenland, China, NorthAmerica and even his own backyard inYorkshire. He has published around 70 papersand two books, mostly on mass extinctions,but also on black shales, his other mainresearch interest.


Meet the Authors

INTRODUCTIONThe plate tectonic hypothesis provides an elegant explana-tion for Earth’s two principal types of basaltic volcanism,mid-ocean ridge and island arc volcanism, both of whichoccur at plate boundaries. Mid-ocean ridge basalts formnew ocean crust along the tensional zones that developwhere adjacent plates, with divergent motions, are pulledapart, and island arc magmas form along regions of com-pression, where plates sink back into the mantle. However,a third significant form of volcanism occurs away fromplate boundaries and therefore cannot be explained byplate tectonics. The most volumetrically significant of theseare continental flood basalts, giant oceanic plateaus, andaseismic ridges. Continental flood basalts and giant oceanicplateaus, their oceanic equivalent, are massive outpouringsof basalt that erupt in 1 to 5 Myr. They cover an equidi-mensional area typically 2000–2500 km across (White andMcKenzie 1989). Collectively they are referred to as LargeIgneous Provinces (LIPs). Aseismic ridges are chains of vol-canoes that stretch across the sea floor. FIGURE 1 shows theDeccan Traps, a typical flood basalt, and the Chagos– LacadiveRidge–Mascarene Plateau, a typical aseismic ridge. Noticethat the Deccan Traps are connected by the 200–300 kmwide Chagos–Lacadive Ridge, across the Carlsberg–CentralIndian Ridge spreading center, through the MascarenePlateau to an active volcano at Réunion. The plumehypothesis attributes flood basalts and giant oceanicplateaus to the melting of the large spherical head of a newplume (Richards et al. 1989; i.e. Campbell and Griffiths

1990) and aseismic ridges, like theChagos–Lacadive Ridge, to themelting of a plume tail (Wilson1963; Morgan 1971).

THE MANTLE PLUMEHYPOTHESISConvection in fluids is driven bybuoyancy anomalies that originatein thermal boundary layers.Earth’s mantle has two boundarylayers. The upper boundary layer isthe lithosphere, which coolsthrough its upper surface. It even-tually becomes denser than theunderlying mantle and sinks back

into it, driving plate tectonics. The lower boundary layer isthe contact between the Earth’s molten iron–nickel outercore and the mantle. High-pressure experimental studies ofthe melting point of iron–nickel alloys show that the coreis several hundred degrees hotter than the overlying mantle.A temperature difference of this magnitude is expected toproduce an unstable boundary layer above the core which,in turn, should produce plumes of hot, solid material thatrise through the mantle, driven by their thermal buoyancy.Therefore, from theoretical considerations, mantle plumesare the inevitable consequence of a hot core.

The material in the lower boundary layer will be lighterthan the overlying mantle, but before it can rise at a signif-icant rate, it must gather enough buoyancy to overcome theviscosity of the mantle that opposes its rise. As a conse-quence, new plumes have a large head followed by a rela-tively narrow tail (FIG. 2). The tail or feeder conduit is com-paratively narrow because hot, relatively low-viscositymaterial following up the existing pathway of the tailrequires less buoyancy to rise than the head, which mustdisplace cooler, high-viscosity mantle. As the head risesthrough the mantle, it grows for two reasons (Griffiths andCampbell 1990). First, material in the high-temperature,low-viscosity tail rises faster than the head and feeds a con-stant flux of hot mantle into the head of the plume. Whenthis material reaches the stagnation point at the top of theplume, it flows radially with a spiraling motion to give thehead its characteristic doughnut shape. Second, the headgrows by entrainment. As the plume rises, heat is con-ducted into the adjacent mantle. In the boundary layersadjacent to the head and tail, the temperature increases andthe density decreases, eventually becoming the same as theplume; the boundary layers then begin to rise with theplume. This material becomes part of the plume and isswept into the base of the head by its recirculating motion.The head is therefore a mixture of hot material from the


Ian H. Campbell1

1 Earth ChemistryResearch School of Earth SciencesThe Australian National UniversityACT 0200 AustraliaandInstitute for Study of the Earth’s Interior,Okayama University at Misasa, Tottori, 682-0193, JapanE-mail: [email protected]

Mantle plumes are columns of hot, solid material that originate deepin the mantle, probably at the core–mantle boundary. Laboratoryand numerical models replicating conditions appropriate to the

mantle show that mantle plumes have a regular and predictable shape thatallows a number of testable predictions to be made. New mantle plumes arepredicted to consist of a large head, 1000 km in diameter, followed by anarrower tail. Initial eruption of basalt from a plume head should be precededby ~1000 m of domal uplift. High-temperature magmas are expected todominate the first eruptive products of a new plume and should be concen-trated near the centre of the volcanic province. All of these predictions areconfirmed by observations.

KEYWORDS: mantle plume, large igneous provinces, uplift, picrite

Columnar jointing in apostglacial basalt flow atAldeyarfoss, NE Iceland.PHOTO JOHN MACLENNAN

Large Igneous Provincesand the Mantle PlumeHypothesis


plume source region and cooler entrained mantle (FIG. 2A).When the plume head reaches the top of its ascent, it flat-tens to form a disk with a diameter twice that of the head(FIG. 2B). Note that growth of the plume head as it risesthrough the mantle occurs because mantle in the plume tailrises faster than the plume head, which is a direct conse-quence of the strong temperature dependence of the mantle’sviscosity.

PREDICTIONS AND OBSERVATIONSThe plume hypothesis makes the following testablepredictions.

New plumes consist of a large headfollowed by a small tail Flood basalts and oceanic plateaus, the oceanic equivalentof flood basalts, are the first eruptive products of a newmantle plume. The volumes of basalt produced duringthese events are enormous, and Richards et al. (1989) haveshown that eruption rates of flood basalts are one to twoorders of magnitude higher than those of the associatedocean island chain, which connects them to the currentposition of the plume. This observation fits well with theplume hypothesis, which attributes the high eruption ratesof flood basalts to melting of plume heads and the lowereruption rates of ocean island chains to melting of narrowerplume tails.

Flattened plume heads should be 2000 to 2500 km in diameter The diameter of a plume head (D) depends on the temper-ature difference between the plume and the adjacent man-tle (∆Τ, the plume’s excess temperature), its buoyancy flux(Q), the kinematic viscosity of the lower mantle (υ), and itsheight of rise (Ζ), as described in equation (1)

D = Q1/5(υ/g α ∆Τ)1/5Κ2/5Ζ3/5 (1)

where g is gravitational acceleration, α is the coefficient ofthermal expansion of the mantle, and Κ is its thermal con-ductivity (Griffiths and Campbell 1990). Note that theplume height of rise is raised to the power 3/5, whereasmost other terms are raised to the power 1/5. Therefore theheight of rise of the plume, which in the case of Earth is thedepth of the mantle, is the dominant factor influencing thesize of a plume head. If ∆Τ of a plume is assumed to be300°C and its buoyancy flux to vary between 3 × 103 and4 × 104 N s-1, the calculated diameter of a plume head orig-inating at the core–mantle boundary is 1000 to 1200 km.The plume head should flatten to produce a disk with adiameter between 2000 and 2400 km when it reaches thetop of the mantle. The calculated time that a plume headtakes to rise from the core–mantle boundary to the top ofthe mantle is about 100 Myr.

When a plume head rises beneath continental crust, theassociated buoyancy anomaly lifts the lithosphere, placingit under tension. This can lead to runaway extension and tothe formation of a new ocean basin (Hill 1991). During theinitial stages of rifting, the hot mantle in the underlyingplume head is drawn into the spreading centre to producethickened oceanic crust. If the line along which the conti-nent splits lies close to the centre of the plume head, as wasthe case when the North Atlantic opened above the Icelandplume head, the length of thickened oceanic crust should beequal to the diameter of the plume head – 2000 to 2400 km.FIGURE 3 shows the zone of thickened oceanic crust alongthe east Greenland coast and the Rockall–Vøring plateaus,which are associated with the break-up of the NorthAtlantic. Notice that the zone of thickened oceanic crust onboth sides of the North Atlantic is ~2400 km long. When anew oceanic ocean basin opens above a plume head, thefirst oceanic crust to form is always anomalously thick, typ-ically at least twice as thick as normal oceanic crust. Otherexamples of plume-related thickened oceanic crust are themargins of South America and Africa above the Paranáplume and India above the Deccan plume (White andMcKenzie 1989). Anomalous thickening of oceanic crustdoes not occur when a new ocean basin forms away fromthe path of a plume.

Plumes must originate from a hot boundarylayer, probably the core–mantle boundary The obvious way to show that plumes originate from thecore–mantle boundary is to use seismic methods to traceplume tails from the top of the upper mantle to theirsource. However, the small diameter of plume tails (100 to200 km in the upper mantle, although wider in the lowermantle) makes them difficult to resolve and, as a conse-quence, attempts to use seismic methods to image plumetails have met with little success. A notable exception is thepioneering use by Montelli et al. (2004) of a new finite-frequency technique to resolve plume tails. They have usedthis method to trace the Ascension, Azores, Canary, Easter,Samoa, and Tahiti plumes to the core–mantle boundary. Anumber of other plumes, including the Iceland plume, dis-appear in the lower mantle, and the Yellowstone plume hasno resolvable signature. Montelli et al. (2004) note that it ismore difficult to image plumes in the lower mantle than in

Map of the western Indian Ocean showing the distributionof volcanic rocks associated with the Réunion–Deccan

plume. The Seychelles were part of the Deccan Traps prior to separationcaused by spreading on the Carlsberg–Central Indian Ridge. Note thatthe Deccan Traps are connected via the 200–300 km wide Chagos–Lacadive Ridge, across the Carlsberg Ridge spreading center, to theMascarene Plateau and eventually to Réunion Island, the current posi-tion of the plume. Between 60 and 40 Ma, the Réunion plume waslocated under the Carlsberg Ridge. It produced a volcanic ridge on bothsides of the spreading center, before leaving the ridge and appearing tobacktrack on itself towards Réunion, which is its current position (adaptedfrom White and McKenzie 1989).

FIGURE 1


the upper mantle, which may explain the discrepanciesbetween the Montelli et al. (2004) images and thoseexpected from plume theory. Nevertheless the Montelli etal. method shows great promise and may eventually allowunambiguous imaging of plume tails in both the upper andlower mantles.

Both heads and tails should erupt high-temperature picrites The ∆Τ of mantle plumes can be estimated from the maxi-mum MgO content of their erupted magmas because, fordry melts, there is a linear relationship between the MgOcontent and magma temperature. As a rough rule, a 4 wt%increase in MgO in the melt equates to a 100°C increase inmantle potential temperature (see glossary). The maximumMgO content of plume-derived picritic (high-Mg basaltic)liquids varies between 18 and 22 wt%, suggesting that thetemperature excess for mantle plumes is between 150°C and250°C. Examples of volcanic provinces that have beenattributed to plumes and that include high-MgO picriticmelts are Réunion–Deccan, Paraná, North Atlantic Province,Karoo, Emeishan, Galapagos–Caribbean, and Hawai‘i.

The temperature excess of a plume head ishighest at the centre of the head and decreasestowards the margin When new oceanic crust opens up above a plume head, thethickness of the oceanic crust produced is dependent on thetemperature of the mantle that is drawn into the new mid-ocean ridge spreading centre. The plume hypothesis pre-dicts that the temperature of a plume head should be high-est near the plume axis, where the tail continues to risethrough the centre of the head (FIG. 2). At the center of thehead, ∆Τ is expected to be 300 ± 100°C. The ∆Τ in theremainder of the head, which is a mixture of hot materialfrom the boundary layer source and cooler entrained lowermantle, varies with the plume buoyancy flux but must beless than at the centre. For typical plume buoyancy fluxes,the average ∆Τ is ~100°C. Hopper et al. (2003) have usedseismic reflection and refraction data, obtained from fourtraverses located between the plume head and its margin, todetermine the thickness of the first oceanic crust to formwhen the North Atlantic opened above the Iceland plume.They obtained thicknesses of 33 and 30 km for traverses T-Iand T-II, close to the plume axis, and 18 and 17 km for trav-erses T-III and T-IV, closer to the margin of the head (seeFIG. 3). The ∆Τ required to produce these crustal thick-nesses, based on the work of McKenzie and Bickle (1988), is350°C and 100°C at the centre and margin of the head,respectively, which is consistent with the plume hypothesis.

Picrites should erupt early during floodvolcanism and be most abundant near thecentre of the plume head and less abundanttowards the margin The hottest material in the head is the mantle from theplume source (the dark colored fluid in FIG. 2), which is300 ± 100°C hotter than the entrained mantle. Althoughthis temperature difference decreases with time, as adjacentlayers exchange heat, the high temperatures of the hot lay-ers will persist for millions to tens of millions of years,depending on their thickness, which in turn depends ontheir distance from the plume axis (see FIG. 2). When aplume head melts to form a flood basalt province, only thetop of the head ascends to a level in the mantle where thepressure is low enough to allow melting. Note in FIG. 2 thatthe hot layer at the top of the plume thickens towards thecentre, where it grades into the tail. Provided the first mag-mas for a new plume do not pond and fractionate in crustalmagma chambers, picrites should dominate the early melt-ing products of plume heads and become less abundant asthe cooler, second layer ascends to a level where it canbegin to melt. Picrites should also be most abundanttowards the hot centre of the plume head and become lessabundant towards the margins. The predicted early picriteshave been documented for the Paraná–Etendeka, Deccan,Emeishan, North Atlantic, and Karoo flood basalts. Exam-ples of picrites that are abundant at the centre of a floodbasalt province and less abundant towards the margins arethe Letaba picrites, as seen along the Lobombo monoclineof the Karoo, and those in the deeply dissected valleys ofthe Emeishan flood basalt province in China.

Flood volcanism should be preceded by domaluplift of 500 to 1000 m at the center ofthe dome The arrival of the hot plume head in the upper mantle willproduce domal uplift at the surface, the magnitude ofwhich depends on its average temperature. The area ofmaximum uplift is predicted to have a radius of ca. 200 kmand to be surrounded by a zone, with a radius of ca. 400 km,in which uplift is still significant (FIG. 4).

Photograph of a laboratory model of a starting thermalplume (A) mid-way through its ascent and (B) after the

head flattens at the top of its ascent. The dark fluid represents hot materialfrom the plume source and the lighter fluid is cooler entrained material.White arrows show motion within the plume and black arrows the direc-tion of motion in the boundary layer adjacent to the plume; the bound-ary layer has been heated by conduction so that its density is approxi-mately the same as that of the plume (after Griffiths and Campbell1990).

FIGURE 2

B

A

The best-documented example of domal uplift occurs inassociation with the Emeishan flood basalt province inChina (He et al. 2003). Here the carbonate beds of theunderlying Maokou Formation have been systematicallythinned by erosion towards the centre of the flood basaltprovince. Isopachs of the Maokou Formation show thatuplift was broadly dome shaped as expected. He et al. recog-nize three zones: an inner zone with a radius of ca. 200 km,where uplift is estimated to be at least 500 m and could pos-

sibly exceed 1000 m; an intermediate zone with a radius of425 km, in which the average uplift is ca. 300 m; and anouter zone with a radius 800 km, in which uplift is minimal.The magnitude and shape of the uplift agree remarkablywell with that predicted by Griffiths and Campbell (1991)(FIG. 4) and Farnetani and Richards (1994). Other examplesof uplift prior to volcanism include the Wrangellia provinceof northwest Canada and southwest Alaska, the Natkusiakprovince in northwest Canada, the Deccan flood basalt ofIndia, and the Ethiopian flood basalt. Uplift started 3 to 5 Myrbefore the Emeishan and Wrangellia volcanism.

THERMOCHEMICAL PLUMESAn obvious weakness of the thermal plume hypothesis isthat it fails to explain a number of minor volcanic chainsthat stretch across the ocean basin and cannot be linked toLIPs. These appear to be the product of plume tails withoutheads. It has been suggested that the plumes responsible forthese volcanic chains may originate in the mid-mantle.Theoretically the mantle could divide into two convectinglayers separated by a boundary layer at the 670 km seismicdiscontinuity; this boundary layer is a potential source ofplumes, which would rise a small distance and thereforehave small heads. However, seismic studies and numericalmodels showing that slabs and plumes, respectively, canpass through the 670 km discontinuity make this interpre-tation unlikely. A recent computer model of thermocom-positional plumes (Farnetani and Samuel 2005) provides apossible solution. They have shown that if the head of aweak plume contains a high proportion of a dense mantlecomponent (e.g. subducted former basaltic crust), theascending head can stall at the 670 km discontinuity andseparate into low- and high-density components. Only thelight component penetrates the discontinuity and gives riseto a weak plume with a smaller temperature excess than thestrong plumes associated with LIPs. The suggestion thatplumes may have an above-average mantle concentrationof dense, subducted former oceanic crust is consistent withthe usual interpretation of D” – a heterogeneous, seismi-cally fast layer of variable thickness (300 ± 300 km) at thebase of the mantle – as a zone where subducted slabs haveconcentrated. It is also consistent with the isotopic (bothstable and radiogenic) and trace element characteristics ofmany plume-derived magmas, which also require a higherthan average mantle concentration of former oceanic crust.

CONCLUSION The mantle plume hypothesis provides a simple explana-tion for all of the essential features of classic LIPs, such asthe Deccan Traps and the North Atlantic Igneous Province,and all of its predictions have been confirmed by observa-tion. Five days of intense scrutiny during The Great PlumeDebate (Fort William, 28 Aug–1 Sept 2005) failed to land atelling blow on the mantle plume chin, and no creditablealternative emerged to explain the principal features ofLIPs, especially their high eruptive volumes and high-tem-perature magmas. However this does not mean that allintraplate volcanoes and volcanic chains are due to plumes.Each case must be considered on its own merits.

ACKNOWLEGMENTSI thank Ross Griffiths and Guust Nolet for their commentson the manuscript and Charlotte Allen for drafting thediagrams. .


Geological map of the North Atlantic region after Hopperet al. (2003). Areas shaded black correspond to onshore

basalts, which erupted from the plume head prior to the opening of theNorth Atlantic. The dark grey areas locate offshore seaward-dippingreflectors, indicating areas of thickened oceanic crust. Light grey areawith v’s between the Rockall Plateau and the Vøring Plateau shows theareal extent of basalts associated with the initial opening of the ocean.T-I to T-IV are seismic transects along the East Greenland coast (fromHopper et al. 2003). BTP: British Tertiary Province; FIR: Faeroes–IcelandRidge; GIR, Greenland–Iceland Ridge; JMR: Jan Mayen Ridge; JMFZ: JanMayen Fracture Zone.

FIGURE 3

Uplift above a plume head, as predicted by Griffiths andCampbell (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 aplume head with a higher excess temperature: 350°C as opposed to100°C. Predicted profiles are given for maximum uplift (t = 0), when thetop 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 theEmeishan flood basalt province is the minimum average value for theinner and intermediate zones as determined from the depth of erosionof the underlying carbonate rocks.

FIGURE 4

REFERENCESCampbell IH, Griffiths RW (1990)

Implications of mantle plume structurefor the evolution of flood basalts. Earthand Planetary Science Letters 99: 79-93

Farnetani CG, Richards, MA (1994)Numerical investigations of the mantleplume initiation model for flood basaltevents. Journal of Geophysical Research99 (B7): 13813-13833

Farnetani CG, Samuel H (2005) Beyond thethermal plume paradigm. GeophysicalResearch Letters 32: L03711, doi:10.1129/2005GL022360

Griffiths RW, Campbell IH (1990) Stirringand structure in mantle starting plumes.Earth and Planetary Science Letters 99:66-78

Griffiths RW, Campbell IH (1991)Interaction of mantle plume heads withthe Earth’s surface and onset of small-scale convection. Journal of GeophysicalResearch 96 (B11): 18295-18310

He B, Xu Y-G, Chung S-L, Xiao L, Wang Y(2003) Sedimentary evidence for a rapidkilometer-scale crustal doming prior tothe eruption of the Emeishan floodbasalts. Earth and Planetary ScienceLetters 213: 391-405

Hill IR (1991) Starting plumes andcontinental break-up. Earth andPlanetary Science Letters 104: 398-416

Hopper JR, Dahl-Jensen T, Holbrook WS,Larson HC, Lizarralde D, Korenaga J, KentGM, Kelemen PB (2003) Structure of theSE Greenland margin from seismicreflection and refraction data: Implica-tions for nascent spreading centresubsidence and asymmetric crustalaccretion during North Atlantic opening.Journal of Geophysical Research 108(B5):2269, doi:10.1029/2002JB001996

McKenzie D, Bickle MJ (1988) The volumeand composition of melt generated byextension of the lithosphere. Journal ofPetrology 29: 625-679

Montelli R, Nolet G, Dahlen FA, Masters G,Engdahl ER, Hung S-H (2004) Finite-frequency tomography reveals a varietyof plumes in the mantle. Science 303:338-343

Morgan WJ (1971) Convective plumes inthe 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) Magmatismat rift zones: The generation of volcaniccontinental margins and flood basalts.Journal of Geophysical Research 94 (B6):7685-7729

Wilson JT (1963) A possible origin of theHawaiian islands. Canadian Journal ofPhysics 41: 863-870 .


INTRODUCTIONLarge igneous provinces (LIPs) are generally attributed tohotter-than-normal mantle. It is important, therefore, toknow the normal range of mantle temperatures. Convec-tion calculations for a fluid with mantle-like properties thatis heated internally and cooled from above predict temper-ature fluctuations of at least ±100°C (Anderson 2000). Geo-physical evidence suggests that the mantle temperatureunder most LIPs was in this normal range while the LIPswere erupting (Clift 2005; Korenaga et al. 2002), and, wheremeasured, the present heatflow is also normal (i.e. appro-priate for the age of the underlying crust). The sedimentaryrecords from a range of swells and plateaus of various agesfrom all major ocean basins (North Atlantic, Mid-PacificMountains, Shatsky Rise, Ninetyeast Ridge, Ontong JavaPlateau) are compatible with eruption of magma from man-tle in the normal range of temperature, followed by con-ductive cooling of the type associated with regular oceaniccrust. The North Atlantic Ocean is particularly shallow, butmodeling by Clift (2005) indicates that the depth is consis-tent with a temperature anomaly of +100°C or less. Fur-thermore, he shows many LIP sites where the depth to thebase of the sediments implies colder than average mantletemperatures.

The largest LIPs (Siberian Traps, Ontong Java Plateau) werearguably erupted at, below, or near sea level rather than atan elevation of one or two kilometres above the surround-ing terrain, as predicted by the plume hypothesis. LIP uplift,if it occurs, appears to be syn- or post-volcanism, rather

than millions of years prior to volcanism, as predicted bythermal models. There is no indication from uplift, heat-flow, or seismic tomography for mantle temperatures morethan 100°C above the mean under these LIPs (Czamanskeet al. 1998; Korenaga 2005; Roberge et al. 2005; Gomer andOkal 2003).

The crust under continental LIPs is thinner than averagecontinental crust (Mooney et al. 1998). Many LIPs occuratop deep sedimentary basins, in back arcs, or on old con-vergent margins. The chemistry of LIPs is highly variable,with compositions ranging between mid-ocean ridge basalt(MORB) and ocean island basalt; many continental LIPsshow clear evidence of input from continental crust or lithos-pheric mantle. Picritic melts are rare. These observations areall enigmatic in the context of the usual high-temperatureor plume explanations for LIPs. The alternative explanationis that the mantle is hotter than generally assumed (Anderson2000) or is compositionally heterogeneous on a large scale,or both.

GONDWANA, ATLANTIC, AND INDIAN OCEAN LIPSThe breakup of Gondwana was preceded by extensive vol-canism along the future Atlantic and Indian ocean margins(FIG. 1). Volcanism continued during breakup, along thecontinental margins and on the separated fragments (inand around the North Atlantic and on and near Madagas-car, Kerguelen, and the Rio Grande Rise; see FIG. 1). Platereconstructions (Müller et al. 1993) show that the currentlycontinent-hugging plateaus (~1000 km offshore) weremainly formed at ridges and triple junctions in the newlyopened Atlantic and Indian oceans, some tens of millionsof years after breakup of the supercontinent. This can beillustrated by calculating the approximate half-width of the


Don L. Anderson1

1 Seismological LaboratoryCaltechPasadena, CA 91125, USAE-mail: [email protected]

When continental crust gets too thick, the dense eclogitic bottomdetaches, causing uplift, asthenospheric upwelling, and pressure-release 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 mainlybelow the melting point of dry peridotite, but its temperature will be abovethe melting point of recycled fertile (basaltic or eclogitic) components,obviating the need for excess temperature to form “hotspots” or “meltinganomalies”. When plates pull apart or delaminate, the mantle upwells;entrained crustal fragments of various ages are fertile and create meltinganomalies along developing mid-ocean ridges, fracture zones, and old suturezones. Eclogites associated with delamination are warmer and less dense thansubducted oceanic crust and more susceptible to melting and entrainment.

KEYWORDS: LIPs, delamination, plumes, hotspots, lithosphere, eclogite

Columbia River basalts,Washington State. PHOTO

STEPHEN SELF

Large Igneous Provinces,Delamination, and Fertile Mantle


ocean at the time of peak LIP volcanism, based on magneticanomaly maps and ages obtained from the plateaus. Someof these plateaus (and half-widths of the ocean, in kilome-tres) are as follows: Azores (1300), Bermuda (1100), CapeVerde (900), Crozet (1000), Discovery (2000), Iceland(1000), Jan Mayen (900), Madagascar (1200), Mozambique(800), Rio Grande Rise (1200), Broken Ridge (1100), Ker-guelen (1100), and Walvis Ridge (1200). The delay betweencontinental breakup and the main stage of plateau forma-tion is usually about 20–50 Myr (FIG. 1). Although someLIPs are currently intraplate, they formed, without knownexception, at plate or craton borders, at triple junctions, oron a spreading ridge about 1000 km offshore from newlyseparated continents.

THE PACIFIC PLATEAUSThe Pacific plate originated as a roughly triangularmicroplate, antipodal to Pangea and surrounded by ridges(Natland and Winterer 2005). It grew by the outward migra-tion of ridges and triple junctions. The growing Pacific platemay have been stationary because of the absence of bound-ing trenches. The great oceanic plateaus in the Pacific werebeing constructed at the boundaries of the expandingPacific plate between the times of Pangea breakup and theconstruction of the large igneous plateaus in Africa and

South America and in the wake of the drifting continents;they are not connected to chains of volcanic islands. Anunstable stress regime, plate reorganizations, and complextriple junction jumps may be responsible for the formationof Pacific seamount fields and plateaus (Natland and Win-terer 2005). The implications are that the mantle is close tothe melting point and is variable in composition and melt-ing point (FIG. 2) and that magmatism is focused by lithos-pheric architecture and stress.

CONSTRAINTS FROM SEISMICTOMOGRAPHYSome LIPs have small-diameter, seismic low-velocity zones(LVZ) in the upper 200–350 km, rarely deeper, of the under-lying mantle (e.g. Christiansen et al. 2002; Allen and Tromp2005). LVZs are regions in the mantle where seismic wavesare slowed. This happens because the mantle is hot and/orpartially molten or has a mineralogy different from that ofthe normal ambient mantle. For example, eclogite in themantle has a low melting point (FIG. 2) and can have slowseismic wave speeds (e.g. FIG. 3 in Anderson 1989a). It isusually assumed that these LVZs are related to the overlyingvolcanism and that they are thermal in nature. But someancient LIPs, such as Paraná and the Ontong Java Plateau(OJP), retain seismic low-velocity zones in the upper200–300 km of the mantle (Vandecar et al. 1995; Gomerand Okal 2003) even though they have traveled thousandsof kilometres away from the putative source and the man-tle has had more than 120 Myr to cool. The OJP low-veloc-ity zone slows seismic waves down but does not attenuatetheir amplitude significantly, as would be the case if themantle were hot or partially molten (Gomer and Okal2003). Compositional, rather than thermal, effects areimplied (e.g. as in FIG. 3). Both upwelling asthenosphereand sinking eclogite can have low seismic velocities. Sink-ing eclogite, however, melts at lower temperatures thanperidotite (FIG. 2). (Note that a decrease in seismic velocityis not the same thing as attenuation. Cold material, such ascold eclogite, can have a low seismic velocity but can be

Distribution and ages of LIPs in the Gondwana hemi-sphere. Also shown are the ages of continental breakup

and uplift. This figure illustrates the geometry and relative timing ofsupercontinent breakup and magmatism, and LIPs in the newly openedocean. Delamination of lower crust and consequent asthenosphericupwelling may be responsible for the LIPs in the continents (ages aregiven in small black numbers), which usually occur in mobile belts, arcs,suture zones and accreted terranes. If the continents move away fromthe delamination sites, it may be possible to see the reemergence of fer-tile delaminated material in the newly formed ocean basins (e.g. oceanicplateaus; red patches), particularly where spreading ridges allowasthenospheric upwellings. CFB: continental flood basalt. Continents area mosaic of fragments, some of which are shown with thin and thicklines as borders. The names of the CFB and oceanic LIPs are left off forclarity (see http://www.largeigneousprovinces.org/ and http://www.mantleplumes.org/).

FIGURE 1

Melting relations in dry lherzolite and eclogite based onlaboratory experiments. The dashed line is a reference

1300°C mantle adiabat. Eclogite will melt as it sinks into normal-tem-perature mantle, and upwellings from the shallow mantle will exten-sively melt any entrained gabbro and eclogite. Eclogite will be about70% molten before dry lherzolite starts to melt (compiled by J. Natland,personal communication). The two vertical lines separate the gabbrofield from the eclogite field (in between is the mixed phase region).

FIGURE 2


transparent to seismic waves, i.e. have low attenuation. AnLVZ with low attenuation implies that it is compositionalrather than thermal in origin.)

CONTINENTAL CRUST IN THE MANTLE?Fragments of continental crust have been found along theMid-Atlantic Ridge (e.g. Bonatti et al. 1996) and in hotspotand LIP magmas. Schaltegger et al. (2002) found continen-tal zircon xenocrysts in basalts from Iceland and Mauritius.Continental crust is inferred to exist at the Seychelles, theFaeroes, Rockall Bank, Jan Mayen, Kerguelen, the OntongJava Plateau, Cape Verde, and the Cameroon Line (e.g. Freyet al. 2002; Ishikawa and Nakamura 2003). The widespreadisotopic characteristics of Indian Ocean basalts have beenattributed to the presence of “lower continental crustentrained during Gondwana rifting” (Hanan et al. 2004) orto “delamination of lower continental crust” (Escrig et al.2004).

THE DELAMINATION MECHANISMThe challenge is to find mechanisms that can explain thevolume of basalt, the uplift history, and the ubiquitous evi-dence for involvement of both continental and mid-oceanridge material in LIP magmas. Some igneous provinces arebuilt on top of rafted pieces of microcontinents or aban-doned island arcs, but is there any mechanism for puttinglarge chunks of continental material into the source regionsof LIPs? Lower crustal delamination is such a mechanism,although it has been basically unexplored in this context.

The lower continental crust thickens by tectonic andigneous processes (Kay and Kay 1993; Rudnick 1995),including magmatic underplating. Presumably the samething can happen at intra-oceanic arcs. Below about 50 km,mafic crust (basalt, dolerite, gabbro) transforms to densegarnet pyroxenite (eclogite; see glossary. Arc eclogites inFIG. 3). Histograms of the thickness of the continental crustshow a sharp drop-off at a thickness of 50 km (Mooney etal. 1998). I suggest that this is controlled by delamination.Once a sufficiently thick eclogite layer forms, it will detachand founder because its density is 3 to 10% greater thanthat of normal mantle peridotite (FIG. 3). Delamination of a10 km thick eclogite layer can lead to 2 km of uplift andmassive 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-Tan-ton 2005). Thus, delamination is a very effective and nonthermal way of thinning the lithosphere, extending themelting column, and creating massive melting and uplift.In contrast to thermal models, uplift occurs during andafter volcanism, and crustal thinning is rapid.

Lee et al. (2005) estimate that it takes 10–30 Myr for alower-crustal mafic layer to reach critical negative buoyancyand for foundering to take place. The thickness of the maficlayer at the time of foundering ranges between 10 and35 km, resulting in significant size heterogeneities in themantle. When the lower crust is removed, the underlyingmantle upwells to fill the gap and melts because of theeffect of pressure on the melting point. This results in apulse of magmatism and an episode of rapid uplift. Thelower crust then rebuilds itself and cools, and the cyclerepeats. The delamination mechanism creates multiplepulses of magmatism separated by tens of millions of years,a characteristic of some LIPs. If the crustal thickening is dueto compressional tectonics, the time scales will be dictatedby convergence rates. In a typical convergent belt, thicken-ing and delamination may take 25–35 Myr.

There are several ways to generate massive melting: one isto bring up hot material adiabatically from depth until itmelts; another is to insert fertile material with a low melt-ing point—delaminated lower arc crust, for example—intothe mantle from above and allow the mantle to heat it upby conduction. Eclogite that was subsolidus at lower crustaldepths can melt extensively when placed into ambientmantle (FIG. 2). Both mechanisms may be involved in LIPformation. The time scale for heating and recycling lower-crust material is much less than for subducted oceanic crustbecause the former starts out much hotter and does notsink as deep (FIG. 3). The total recycle time, includingreheating, may take 30 to 75 Myr. If delamination occursnear the edge of a continent, say along a suture belt (Foulgeret al. 2005), and the continent moves off at 3.3 cm per year(the average opening velocity of the Atlantic Ocean), thedelamination site will have moved 1000 to 2500 km awayfrom a vertically sinking root in the time sincedelamination.

The neutral density profile of the crust and mantle. Thematerials of the crust and mantle are arranged mostly in

order of increasing density. A” is the region of over-thickened crust thatcan transform to eclogite and become denser than the underlying mantle.C’ is the top part of the transition region. Eclogites having STP densitiesbetween 3.6 and 3.7 g/cc may be trapped in C’. Eclogites and peri-dotites have similar densities in this interval but different seismic veloci-ties. In C”, MORB-eclogite and low-MgO arc-eclogites reach densityequilibration. Trapped eclogite and dense eclogite sinkers can be low-velocity zones. The order approximates the situation in an ideally chem-ically stratified mantle, except for the region just below the continentalMoho where potentially unstable lower crustal cumulate material isformed. In such a stably stratified structure the temperature gradient willbe greater than the adiabatic gradient. TZ = transition zone; Lhz = lher-zolite; Coe = coesite; St = stishovite.

FIGURE 3

BROAD DOMAL UPLIFTBroad domal uplift is a characteristic of delamination (Kayand Kay 1993). The magnitude is related to the density andthickness of the delaminating column. Crustal domes of~1000 km in lateral extent and elevations of ~2 km abovebackground, with no heat flow anomaly, can be explainedby such shallow processes (Petit et al. 2002). The Mongoliandome, for example, is underlain by a small anomaly (5%seismic velocity reduction and a density reduction of only0.01 g/cc) of limited vertical extent, 100–200 km deep, yetit has the same kind of domal uplift that has been assumedto require a large and deep thermal perturbation. Theremoval of a dense eclogitic root and its replacement byupwelling peridotite can create regional uplift and a shallowlow-velocity zone; the eclogite sinker also has low seismicvelocities. This process is essentially a top-down athermalprocess.

One of the best-documented examples of delamination,uplift, and volcanism is the Eastern Anatolia region (Keskin2003), which was below sea level between ~50 and ~13 Ma.It was then rapidly elevated above sea level. Uplift was fol-lowed by widespread volcanic activity at 7–8 Ma, and theregion acquired a regional domal shape comparable to thatof the Ethiopian High Plateau. Geophysical, geological, andgeochemical studies support the view that domal uplift andextensive magma generation were linked to the mechanicalremoval of the lower crust accompanied by upwelling ofnormal-temperature asthenospheric mantle to a depth of50 km.

The above examples are important in showing that well-understood shallow processes can generate regional domalstructures and large volumes of magma. The LVZ undersome LIPs, including ancient ones, and domal structuresmay be associated with cold eclogite rather than hotupwellings (FIG. 3).

DO WE NEED TO RECYCLE OCEANIC CRUST?Recycled oceanic crust is often considered to be a compo-nent of ocean-island and LIP magma, although this view isdisputed. Once in the mantle, subducted MORB-eclogitereaches neutral buoyancy at depths of 500–650 km (Ander-son 1989b; Hirose et al. 1999) (FIG. 3). Very cold MORB maysink deeper (Litasov et al. 2004). If current rates of oceaniccrust recycling operated for 1 Gyr (Stern 2005), the totalsubducted oceanic crust would account for 2% of the man-tle, and it could be stored in a layer only 70 km thick. Thesurprising result is that most subducted oceanic crust neednot be recycled or sink into the lower mantle in order to sat-isfy any mass-balance constraints (see also Anderson1989a). The MORB-like component in some LIPs may sim-ply be due to passive asthenospheric upwelling, as in thedelamination model.

The recycling rate of lower-crustal mafic rocks (Lee et al.2005) implies that about half of the continental crust isrecycled every 0.6 to 2.5 billion years. In contrast to oceaniccrust, one can make a case that eroded and delaminated arcand continental material is not stored permanently, or overthe long term, or very deep in the mantle; it is reused andmust play an important role in continental mass balance,global magmatism, and shallow-mantle heterogeneity.

MELTING OF ECLOGITEGeophysical estimates of the potential temperature (seeglossary) of the mantle are about 1350–1400°C (Anderson2000), with statistical and geographic variations of at least100°C. These temperatures are about 100 degrees higherthan generally assumed by petrological modelling. This

range permits partial melting of peridotite, the formation(in places) of high-MgO magmas, and extensive melting ofeclogite. Melting experiments (e.g. Yaxley 2000) suggestthat 60–80% melting of eclogite is required to reproducecompositions of some LIP basalts (Natland, personal com-munication). FIG. 2 suggests that this is plausible and thatlherzolite will start to melt under these conditions. Theinteraction of melts from eclogite and lherzolite is implied.The model discussed here does not imply low mantle tem-peratures. In fact, an internally heated (i.e. by radioactivedecay), chemically stratified mantle achieves higher tem-peratures than a uniform mantle heated from below. Onemust explain, then, not the existence but the rarity ofpicrites. The eruption of high-temperature MgO-rich mag-mas may require special circumstances because of theirhigh density. This special circumstance may be the rapidupwelling of asthenosphere following a delamination eventand edge-driven convection (King and Anderson 1998).

DISCUSSIONSome LIPs may simply be due to passive upwelling of inho-mogeneous asthenosphere as continental fragments diverge(McHone 2000). Some are the result of reactivated suturezones or other weaknesses of the lithosphere, combinedwith the variable fertility and melting point of the underly-ing mantle (Foulger et al. 2005). In this paper, I havefocused on a new mechanism that augments these otherprocesses. Delaminated lower crust sinks into the mantle aseclogite, where it has a relatively low seismic velocity andmelting point compared to normal mantle peridotite.Although delaminated continental crust enters the mantleat much lower rates than oceanic crust, the rates are com-parable to LIP production rates. I speculate that the largemelting anomalies that form on or near ridges and triplejunctions may be due to the resurfacing of large fertileblobs, including delaminated continental crust. Pondedmelts may contribute to magmatism at new ridges andtriple junctions. Delaminated eclogites may form a uniquecomponent of hotspot and ridge magmas (Lee et al. 2005),but I suggest that lower continental crust is not just a con-taminating agent. Blocks of it are responsible for the melt-ing anomalies themselves, including the Kerguelen Plateauand other features in the Indian Ocean. The massiveplateaus, such as the Ontong Java Plateau, may be due to acombination of delamination (Korenaga 2005), excess man-tle fertility, slightly higher average mantle temperaturesthan usually assumed (~100°C), slightly lower meltingtemperatures (~100°C), and focusing of magma at a triplejunction.

The crustal delamination, variable mantle fertility model,combined with passive asthenospheric upwelling, has thepotential to explain the tectonics and compositions of LIPs,including heatflow and uplift histories. Apparently, noother model explains the formation of LIPs and uplifteddomes so elegantly, with so few contradictions. But themodel needs to be tested further and quantified. .


REFERENCES(Current information on the subjectmatter of this article can be found atwww.mantleplumes.org)

Allen RM, Tromp J (2005) Resolution ofregional seismic models: Squeezing theIceland anomaly. Geophysical JournalInternational 161: 373-386

Anderson DL (2000) Thermal state of theupper 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/Cal-techBOOK:1989.001

Anderson DL (1989b) Where on Earth isthe crust? Physics Today 42: 38-46

Bonatti E, Ligi M, Borsetti AM, Gasperini L,Negri A, Sartori R (1996) LowerCretaceous deposits trapped near theequatorial Mid-Atlantic Ridge. Nature380: 518–520

Christiansen RL, Foulger GR, Evans JR(2002) Upper-mantle origin of theYellowstone hotspot. Bulletin of theGeological Society of America 114: 1245-1256

Clift PD (2005) Sedimentary evidence formoderate mantle temperature anomaliesassociated with hotspot volcanism. In:Foulger GR, Natland JH, Presnall DC,Anderson DL (eds) Plates, Plumes, andParadigms. Geological Society of AmericaSpecial Paper 388, pp 279-288

Czamanske GK, Gurevich AB, Fedorenko V,Simonov O (1998) Demise of the Siberianplume: paleogeographic and paleotec-tonic reconstruction from the prevolcanicand volcanic record, North-CentralSiberia. International Geology Review 40:95-115

Elkins-Tanton LT (2005) Continentalmagmatism caused by lithosphericdelamination. In: Foulger GR, NatlandJH, Presnall DC, Anderson DL (eds)Plates, Plumes, and Paradigms, GeologicalSociety of America Special Paper 388, pp449-462

Escrig S, Capmas F, Dupré B, Allègre CJ(2004) Osmium isotopic constraints onthe nature of the DUPAL anomaly fromIndian mid-ocean-ridge basalts. Nature431: 59-63

Foulger GR, Natland JH, Anderson DL(2005) A source for Icelandic magmas inremelted Iapetus crust. Journal ofVolcanological and Geothermal Research141: 23-44

Frey FA, Weis D, Borisova AYu, Xu G(2002) Involvement of continental crustin the formation of the CretaceousKerguelen Plateau: new perspectives fromODP Leg 120 sites. Journal of Petrology43: 1207-1239

Gomer BM, Okal EA (2003) Multiple-ScSprobing of the Ontong-Java plateau.Physics of the Earth and PlanetaryInteriors 138: 317–331

Hanan BB, Blichert-Toft J, Pyle DG,Christie DM (2004) Contrasting originsof the upper mantle as revealed byhafnium and lead isotopes from theSoutheast Indian Ridge. Nature 432: 91-94

Hirose K, Fei Y, Ma YH, Mao H-K (1999)The fate of subducted basaltic crust in theEarth’s lower mantle. Nature 397: 53-56

Ishikawa A, Nakamura E (2003) Dupalsignature recorded in quartz-garnetclinopyroxenite xenolith from sub-Ontong Java Plateau mantle. Geochimicaet Cosmochimica Acta Supplement67: 175

Kay RW, Kay SM (1993) Delamination anddelamination magmatism. Tectono-physics 219: 177-189

Keskin M (2003) Magma generation byslab steepening and breakoff beneatha subduction-accretion complex: Analternative model for collision-relatedvolcanism in Eastern Anatolia, Turkey.Geophysical Research Letters 30 (24):8046, doi:10.1029/ 2003GL018019

King SD, Anderson DL (1998) Edge-drivenconvection. Earth and Planetary ScienceLetters 160: 289-296

Korenaga J (2005) Why did not the OntongJava Plateau form subaerially? Earth andPlanetary Science Letters 234: 385-399

Korenaga J, Kelemen PB, Holbrook WS(2002) Methods for resolving the originof large igneous provinces from crustalseismology. Journal of GeophysicalResearch 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 meltinganomalies in the mantle. Earth andPlanetary Science Letters (submitted)

Litasov K, Ohtani E, Suzuki A, Kawazoe T,Funakoshi K (2004) Absence of densitycrossover between basalt and peridotitein the cold slabs passing through 660 kmdiscontinuity. Geophysical ResearchLetters 31 (L24607),doi:10.1029/2004GL021306

McHone JG (2000) Non-plume magmatismand rifting during the opening of thecentral Atlantic Ocean. Tectonophysics316: 287-296

Mooney WD, Laske G, Masters TG (1998)CRUST 5.1: A new global crustal model at5° × 5°. Journal of Geophysical Research103: 727-747

Müller RD, Royer J-Y, Lawver LA (1993)Revised plate motions relative to thehotspots from combined Atlantic andIndian Ocean hotspot tracks. Geology21: 275-278

Natland JH, Winterer EL (2005) Fissurecontrol on volcanic action in the Pacific.In: Foulger GR, Natland JH, Presnall DC,Anderson DL (eds) Plates, Plumes, andParadigms, Geological Society of AmericaSpecial Paper 388, pp 687-710

Petit C, Déverchère J, Calais E, San’kov V,Fairhead D (2002) Deep structure andmechanical behavior of the lithospherein the Hangai-Hövsgöl 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 subsidenceof the Ontong Java Plateau inferred fromCO2 contents of submarine basalticglasses. Geology 33: 501-504

Rudnick RL (1995) Making continentalcrust. Nature 378: 571-578

Schaltegger U, Amundsen H, Jamtveit B,Frank M, Griffin WL, Grönvold K,Trönnes R, Torsvik T (2002) Contamina-tion of OIB by underlying ancientcontinental lithosphere: U–Pb and Hfisotopes in zircons question EM1 andEM2 mantle components. Geochimicaet Cosmochimica Acta 66 (supplement 1):A673

Stern RJ (2005) Evidence from ophiolites,blueschists, and ultrahigh-pressuremetamorphic terranes that the modernepisode of subduction tectonics beganin Neoproterozoic time. Geology 33: 557-560

Vandecar JC, James DE, Assumpção M(1995) Seismic evidence for a fossilmantle plume beneath South Americaand implications for plate driving forces.Nature 378: 25-31

Vlaar NJ, van Keken PE, van den Berg AP(1994) Cooling of the earth in theArchaean: Consequences of pressure-release melting in a hotter mantle. Earthand Planetary Science Letters 121: 1-18

Yaxley GM (2000) Experimental studyof the phase and melting relationsof homogeneous basalt + peridotitemixtures and implication for thepetrogenesis of flood basalt. Contribu-tions to Mineralogy and Petrology 139:326-338

Zegers TE, van Keken PE (2001) MiddleArchean continent formation by crustaldelamination. Geology 29: 1083-1086 .



BALZAN PRIZE TO HEMLEY AND MAO

People in the News

The International BalzanFoundation awarded one of itsprestigious prizes to MSAmembers Russell J. Hemley andHo-Kwang Mao for “the impres-sive impact of their joint workleading to fundamental break-throughs, theoretical andexperimental, in the field ofminerals submitted to extremephysical conditions.” The goal ofthe International E. Balzan Prizesis fostering culture, the sciencesand the most meritorioushumanitarian initiatives of peaceand brotherhood among peoples,regardless of nationality, race orcreed. Each prize has a value of

one million Swiss francs (about650,000 euros); half of the prizemust be designated by the prize-winner for research work,preferably involving youngscholars. The International BalzanFoundation was established in1956 by Angela Lina Balzan inmemory of her father, EugenioBalzan, who was co-publisher formany years of the Corriere dellaSera, an influential Italian news-paper. It awards four prizes everyyear in the fields of naturalsciences, humanities, socialsciences, and art. We reproducepart of the citation.

CMS PEOPLE IN THE NEWS

“What happens at extremely high pressures? The two physicistsRussell J. Hemley and Ho-kwang (David) Mao have focused on thisquestion in their research. Hemley and Mao both work at the Geo-physical Laboratory of the Carnegie Institution in Washington, USA,and devote themselves to investigating the properties of substancesunder extreme conditions. This particularly means at high tempera-tures, but also at pressures of up to 2.5 megabars.

In 1976, Ho-kwang (David) Mao and his colleagues were the first tocreate a static pressure of 1 megabar – one million times the ambientpressure at sea level and double what had previously been achievedin a laboratory. Since 1985, in collaboration with Russell J. Hemley,he has further improved the technique of creating such pressures aswell as the methods of analyzing what happens to substancesexposed to them. Hemley and Mao have observed and describednumerous extreme-pressure phenomena such as the occurrence ofnew types of molecular bonds, the creation of new, extremely hardmaterials, superconductors and magnetic structures, as well as pres-sure-induced crystallization and amorphisation.

The two scientists are also particularly interested in planetary mate-rials, leading to conclusions about processes taking place withinEarth and other planets. Both researchers have already won manyawards: Hemley was most recently honored with the HillebrandMedal of the American Chemical Society in 2003, while in 2005 Maoreceived both the Gregori Aminoff Prize of the Royal Swedish Acad-emy of Sciences and the Roebling Medal of the Mineralogical Societyof America.”

Dewey Moore, Past President, signsoff to Cliff Johnston 2005, Presidentof The Clay Minerals Society (CMS)

Pete Ryan and Michelle Hluchy, organizersof the 2005, 42nd AnnualCMS Meeting

Andreas Bauer organized the 2005CMS Workshop

Richard Brown, photographerfor the 42nd Annual CMSMeeting

Robert J. Bodnar (VirginiaPolytechnic Institute and StateUniversity, Blacksburg, Virginia,USA) will be recognized by theAmerican Geophysical Union

(AGU) for his landmark contribu-tions to studies of fluid inclusionsin minerals and to aqueous andhydrothermal geochemistry. Hewill be presented with the 2005Bowen Award at the 2006 JointAssembly in Baltimore, Maryland.The Bowen Award is given annu-ally for outstanding contributionsto volcanology, geochemistry,or petrology.

Russell J. Hemley and Ho-Kwang Mao

BODNAR TO RECEIVE N.L. BOWEN AWARD

INTRODUCTIONThe idea that large meteorite impacts may trigger volcanicactivity (impact-induced volcanism) has been around forseveral decades. A meteorite impact was proposed toexplain the differentiated melts of the Sudbury IgneousComplex, including the currently unfavoured idea that thenickel-rich deposits are cosmogenic (Dietz 1972). Hot melt-ing of impacted peridotite mantle was proposed by Green(1972) as an explanation for the very-high-temperaturelavas called komatiites. Rogers (1982) proposed that largeoceanic basalt plateaus represent the relics of moderate-sized meteorite impact craters [diameter (D) > 100 km] andpromoted Grieve’s (1980) idea that upwarping of theasthenosphere might be sufficient to initiate a long-livedthermal plume in the mantle. Melt productivity beneathimpact craters may be significantly increased by decom-pression melting of the mantle and high ambient geother-mal gradients.

Very large terrestrial impact craters (D > ~500 km) can cre-ate more melt than the volume of the impact crater, i.e. suf-ficient for LIPs (~106 km3, FIG. 1A), and simulations of thelargest conceivable impact, the giant moon-forming event,predict reorganisation and substantial melting of the man-tle and lithosphere on a global scale (Canup and Asphaug2001). Glikson (2001) estimates that 390 ± 36 craters (D >100 km) and 45 ± 4 craters (D > 250 km) formed on Earthin the last 3.8 Gyr. Even when adjusted for a reduction inthe number of hits with time, this translates to one largeimpact (D > 250 km) every ~100–150 Myr during the last

500 Myr. Thus, terrestrial impactevents capable of generating globalejecta layers should occur rela-tively frequently in the geologicalrecord, and some LIPs may overlapin age with an unrelated impactevent (White and Saunders 2005).LIP volumes are typically ~106 km3,but some are smaller, e.g. theColumbia River basalt plateau(~105 km3), and some are larger,e.g. the Siberian Traps and theOntong Java Plateau (106–107 km3).

This paper summarises recentmodelling to test the impact for-mation of the Ontong Java Plateau(OJP), which due to its size is an

extreme case. The case for cause and effect between a spe-cific impact event and a LIP would be strengthened if thestratigraphy demonstrates that the impact event immedi-ately predated the LIP, and if the initial chemistry of theLIP, where underlain by oceanic mantle, was ultramafic (e.g.high-magnesium basalt or picrite). Such an impact eventpreceding a LIP has been described from West Greenlandand is profiled below. The volcanic expression of impacts isunknown, but key igneous features of a large differentiatedimpact melt at Sudbury provide some clues.

IMPACTS AND MELTINGTypically, the relationship between the body size of theimpactor and the diameter of the transient crater is 1:10.Rebound and gravity-assisted relaxation produce a finalcrater much shallower than the transient crater and with Dapproximately twice that of the transient crater. A compila-tion of estimated melt volumes for terrestrial impact cratersin crystalline rocks by Grieve and Cintala (1992) shows asimple logarithmic correlation with crater diameter (FIG. 1A).The conventional view is that melt volume scales with thekinetic energy (1/2mv2) of the impactor (Pierazzo et al.1997) and that the volume of melt produced by a meteoriteimpact (for a crater with D ~100–200 km) is insufficient toexplain the amount of melt typical of a LIP (Ivanov andMelosh 2003). Until recently, most impact cratering modelssimulated the effect of impacts into cold targets. Since hottargets melt at lower shock pressures, modelling the truethermal structure of the target is vital for understanding theeffects of impacts on Earth. This is particularly true for largeimpactors that penetrate through the Earth’s crust and intothe mantle. Jones et al. (2002) used hydrodynamic com-puter modelling to demonstrate that the melting responseof the Earth’s peridotite mantle to decreasing pressure(decompression melting) beneath a large impact cratermight increase the total melt volume, depending on the


Adrian P. Jones1

1 Department of Geological SciencesUniversity College LondonGower Street London WC1E 6BTUnited [email protected]

Ameteorite impacting on the surface of the Earth produces not onlya crater but also, if the impactor is sufficiently large, high meltvolumes. Computer simulations suggest that, in addition to shock-

induced melting produced by impact, additional decompression meltingof the hot target mantle beneath the crater can produce melt volumescomparable to those found in large igneous provinces (LIPs). The coincidencebetween the expected frequency of such impact events combined with thesimilarity in magma volumes of LIPs suggests that large meteorite impactsmay be capable of triggering LIPs and mantle hotspots from a point sourcewhich is subsequently buried. Can the impact model explain any LIP? Whatare the distinctive macroscopic criteria predicted from an impact model, andhow may they be recognised or rejected in the geological record of the Earth?

KEYWORDS: meteorite impact, decompression melting, mantle, hotspot

Shocked quartz showingdecorated planar

deformation featuresfrom newly discoveredimpact deposits in SW

China, of unknown age.Partially crossed polars,

field of view approxi-mately 1 mm. PHOTO

ADRIAN JONES

Meteorite Impactsas Triggers to LargeIgneous Provinces


target thermal profile and lithology. For impacts into thethermally active Earth, we contend that above some thresh-old crater diameter (not yet determined), where decom-pression melting becomes significant, the volume of meltproduced might be considerably greater than that predictedby the conventional relationship between melt volume andcrater diameter. Melt volumes of approximately 106 km3 –comparable to LIP volumes – might be produced (FIG. 1B).

GEOLOGICAL EVIDENCE OF IMPACTAmongst an array of geological impact signatures, twoimportant criteria are (1) the presence of shock metamor-phic effects in mineral and rock inclusions in breccias andmelt rocks, and (2) evidence for a minor extraterrestrial geo-chemical component in these rocks (Koeberl 2002).Shocked quartz is the foremost mineralogical criterion forrecognition of shock metamorphism in terrestrial materials;it is petrographically distinctive and recognisable even asrare fragments (FIG. 2A). However, shocked quartz can onlybe produced in target rocks containing quartz, ruling outparts of the Earth’s crust and lithosphere composed ofquartz-free rock like basalt (e.g. oceanic crust). Thereforeejecta products of oceanic impacts will be free of shockedquartz and must be identified by other mineralogical andgeochemical criteria. These include quenched melt dropletspherules (FIG. 2B), Ni-spinel (FIG. 2C, D) and geochemicalanomalies for the siderophile elements, especially platinum-group elements (e.g. iridium, osmium) and chromium. Inaddition, there may be field evidence for unusual geologicalactivity, such as tsunami deposits.

IMPACT MODEL FOR THE ONTONGJAVA PLATEAUThe mid-Cretaceous Ontong Java Plateau (OJP) is the largestoceanic LIP. It is thought by many to have formed from adeep mantle plume, although this is not universallyaccepted. It may instead have been triggered by a meteoriteimpact (Rogers 1982; Ingle and Coffin 2004) as examinedhere. Jones et al. (2005a) modelled the first few hundredseconds of a vertical impact between a hypervelocity mete-orite projectile and a dry peridotite lithosphere target usinggeotherms for young oceanic crust at the onset of the OJP.At the scale of tens to hundreds of kilometres, the com-plexities of atmosphere and ocean are ignored. In the sim-ulation, changing the target to hot oceanic lithosphere hasa dramatic effect and produces massive melting by both

(A) Theoretical correlation of impact melt volume versuscrater diameter for terrestrial impact craters (sloping line)

with locus of terrestrial impact melt below this line (after Grieve andCintala 1992). Also shown is the melt volume typical of a LIP (horizon-tal dashed line at 106 km3). (B) Hypothetical increase in impact melt vol-ume above the theoretical sloping line, due to additional decompressionmelting of lithospheric mantle for large crater diameters, not preciselydetermined (see Jones et al. 2002, 2003).

FIGURE 1

(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, WestGreenland (after Jones et al. 2005b). (B) Impact melt glass spherules,field of view 2.5 mm (PPL). (C) Skeletal quench-textured Ni-spineloccurring as radiating “christmas trees” (PPL). (D) Detail of (C). Back-scattered electron image (width ~50 µm) showing Ni-spinel crystals,which have an irregular core of nearly pure Ni metal.

FIGURE 2

A

A

B

C

D

B


shock and decompression. The OJP was formed on youngoceanic crust close to a spreading ridge, and the precise ageof this crust determines the geothermal gradient, whichcontrols the amount of melt in our impact model. Simplychanging the age of the oceanic crust from 20 to 10 Ma pro-duces about five times more melt, and for our 20 kmimpactor, can produce ~106 km3 melt, equivalent to a LIP.The largest event modelled is extreme, and involved a 30 kmdiameter dunite projectile impacting at 20 km/s. Within~10 minutes of impact, the melt was distributed predomi-nantly as a giant sub-horizontal disc with a diameter inexcess 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 >500°Cabove the solidus) within 100 km of ground zero, to non-equilibrium partial melts varying in amount with depthand distance (FIG. 3). Some of this melt will quench, butmost will crystallise slowly, taking perhaps tens of thou-sands of years to solidify (Jones et al. 2005a). Massive reor-ganisation of the affected upper mantle, driven by large-scale physical disturbances such as displaced crust,juxtaposed hot and cold materials and mobile melts, is vir-tually inevitable (Price 2001). This has not been modelled,but thermodynamic relaxation of heterogeneously meltedmantle could be redistributed through a much largermantle volume as a conventional partial melt, to a maxi-mum of approximately ~7.5 × 106 km3 of basalt, assuming20–30% partial melting.

This volume of impact melt exceeds that of most LIPs but isslightly less than the total volume of the OJP (>30 × 106 km3).Larger melt volumes will result from hotter mantle poten-tial temperatures (>1500 degrees, e.g. Chazey and Neal2004). Various other parameters could be changed – projec-tile dimensions, velocity, hydrous mantle, or anomalousmantle composition. Part of the impact-induced meltwould be buoyant and erupt rapidly. This episode might befollowed by an extended secondary period of additionalmelting during mantle upwelling, as envisioned by Grieve

0100200300

0

50

100

kilometres

The map shows the outline of the Ontong Java Plateau. Itssize can be compared with the impact melt derived from

a hydrodynamic simulation of a large impact. The round “bulls eye” atthe same scale as the map represents the area of melt generated ~20minutes after vertical impact of a 30 km meteorite into young oceaniccrust. The simulation assumes a young oceanic mantle geotherm, mini-mal dry melting and maximum credible projectile size (Jones et al.2005a). The enlarged cross section beneath, with horizontal equal tovertical scale (in kilometres), shows melt distribution in the mantle onthe left and deformation paths as disturbed layers on the right. Colourson the left-hand side show the extent of melting: red, up to 100% melt(corresponding to superheated conditions at temperature >500 degreesabove solidus); yellow 50% melt; blue >1% melt. In peridotite mantlethe colours red-yellow-blue correspond approximately with regionswhere the products of melting are peridotite or komatiite, picrite andbasalt, respectively. The maximum depth of melting for this model isapproximately 150 km. Most of the melt is confined to a diameter of300 km, but a thin surface layer and a deeper disc at ~50–70 km extendto >600 km. The total melt volume generated both by shock anddecompression melting is ~2.5 × 106 km3. Massive reorganisation of theaffected upper mantle is expected, and could trigger further mantleupwelling and additional melting.

FIGURE 3

(1980) and described in more detail in the lunar maremodel of Elkins-Tanton et al. (2004). A primary feature ofthe oceanic impact model is that voluminous ultramafic ormafic melts are expected early in the igneous activity.

The geochemistry of the OJP and the absence of thermaluplift have proved difficult to reconcile with deep mantleplume models, but these features may be consistent with animpact origin (see Ingle and Coffin 2004). The major andtrace element chemistry of the OJP lavas is notably uni-form, with many incompatible element concentrationscomparable to those for shallow mid-ocean ridge basalt(MORB), but with isotopic characteristics more like those ofocean island basalts. A shallow-melting origin (~100 km), iscompatible 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 globalrecord is expected, but will it be recognisable? Oceanicimpacts away from the continental shelf will not yieldshocked quartz, and shocked ejecta of glasses and silicatesare highly susceptible to replacement by clays. Global fall-out layers of fine glass-rich ash and dust could be similar toconventional volcanic deposits. Volcanic clay layers and«Fullers Earth» horizons are distinctive marker bands inBarremian–Aptian geological sequences of northernEurope, coinciding with the ~120 Ma age of the OJP event;might one of these contain the distinctive geochemical sig-nature of extraterrestrial components linked to a distal OJPimpact layer? However, no mass extinction correlates withthe date of formation of the OJP, although there was aglobal oceanic anoxic event and a global negative Sr isotopeanomaly (Jones and Jenkyns 2001).

IMPACT EJECTA BENEATH A LIP?The early Tertiary (~62 Ma) lavas forming Disko Island incentral West Greenland and the Nuussuaq Peninsula imme-diately to the north have been correlated with those onBaffin Island and form the earliest western extremity of theNorth Atlantic Tertiary Igneous Province, in which volcanicactivity continues today in Iceland. These highly magne-sian lavas require high temperature melting of shallowmantle (60–90 km) and may constitute a precursor to theplume which became established under East Greenland(Gill et al. 1995). A distinctive spherule bed horizon cropsout over a 3 m interval in shallow water sediments approx-imately 10 m beneath the local base of the flood lava pileon the Nuussuaq Peninsula. The glassy spherule layers havemany of the hallmarks of impact ejecta: immiscible melttextures, distinctive Ni-spinel, and high Ir, PGE andsiderophile element anomalies (Jones et al. 2005b). Theiron-rich silicate glass spherules (~3 wt% NiO, ~35 wt%FeO) are circular in cross section and show evidence of sur-face dissolution, smectite replacement and calcite infillingof vesicles, though many glasses are optically unaltered.Their pronounced Fe–Ni correlation is dissimilar to volcanicsuites, but can be explained by mixing of a basaltic meltand an iron–nickel source. Distinctive Ni-spinel grains(~7–10 wt% NiO) possess very nickel-rich cores (FIG. 2D).Rare glass spherules show compositional gradients towardsresorbed silicates (plagioclase, clinopyroxene); shocked pla-gioclase (maskelynite glass) has an anomalous texture com-parable to that seen in impact-melted lunar breccias.Although anomalously high copper and sulphur concentra-tions (up to ~1% in spherules) have led other researchers tosuggest terrestrial explanations, such as the possibility thatthey are products of fire-fountaining of exotic or hot,picritic Disko lavas (see Jones et al. 2005b for details), astrong case can be made that they are impact deposits.Delicate preservation features rule out substantial sedimen-

tary reworking, and spherule sizes and bed thicknessesimply a large source crater. The age of the spherule beds isconstrained by nannofossils and magnetostratigraphy to beclose to the age of initiation of the West Greenland floodlavas (~62 Ma).

EXPRESSIONS OF IMPACT VOLCANISM?Although there is no direct evidence that large volumes ofextrusive rock have been produced by impact, the Sudburystructure shows that large volumes of subsurface magmacan 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-yearmultidisciplinary study by Stoffler et al. (1994) concludedthat the impact excavated deep into the crust, almost to themantle (~30 km), before collapse and rebound. The presenteccentric shape is due to subsequent tectonism. The melt(> ~12,000 km3), possibly superheated, formed by impactmelting of crust within just a few minutes. The magma dif-ferentiated by gravity settling of crystals and immisciblesulfides to produce hundreds of metres of noritic cumulates(norite is a type of gabbro). Early formed pyroxene and sul-fides were swept into basal depressions to form mineralisednorite, overlain by slowly cooled igneous-textured rockswith differentiated compositions.

There is no record of volcanism at Sudbury but it may havebeen spectacular. The high temperatures implied by coex-isting immiscible melts and mafic magmas are comparableto those of many large igneous intrusions, representing themid- to upper-crustal reservoirs feeding surface volcanism.At Sudbury, the presence of pseudotachylites (veins ofshock-induced glassy rock), contact zone breccias and anarray of peripheral shock features is well established. Anymantle melt component is thought to have been small, butcould have been delivered almost instantly via crust-span-ning dykes with rapid post-crater closure (Price 2001).Rapid closure of fractures may explain the absence of feed-ers in impact-induced melt bodies such as Sudbury. TheSudbury nickel deposits are crudely concentrated aroundthe margins of the impact cavity and form the largest nickelmining district ever mined. The source of the nickel couldbe the impactor in terms of mass balance, although isotopedata suggest a crustal source for the accompanying sulphur.In other impact craters, there is evidence for associateddownwards and outwards injection of magma, formingdykes, breccias, and pseudotachylites, and for the establish-ment of vigorous hydrothermal systems. The Sudbury root-less (?) impact melt, the likelihood of superheat, and theformation of immiscible sulphides are valuable lessons formainstream igneous petrology and global ore prospecting.

FURTHER WORKHypervelocity impact models predict that large oceanicmeteorite impacts can generate melts with volumes compa-rable to those of typical LIPs. Future modelling shouldincorporate longer time scales in order to allow for mantleflow. Impact-triggered mantle melts are expected to havegeochemical signatures like those of mantle plumes. Rapidmixing of melts from sub-horizontal sub-crater reservoirs todepths where pyrope garnet and/or diamond are stable ispossible (Jones et al 2003). Impact melting of the mantlecan generate peridotitic melts like komatiites and otherhigh-degree partial melts (Jones 2002). Reprocessing ofparts of the upper mantle via large bolide impacts is consis-tent with models of planetary accretion following the lateheavy bombardment and provides an alternative explana-tion for primitive geochemical signatures currently attrib-uted to plumes entraining material from the core. For LIPswith mafic to ultramafic initial volcanic products, and


REFERENCESAlvarez W (2003) Comparing the evidence

relevant to impact and flood basalt attimes of major mass extinctions.Astrobiology 3: 153-161

Canup RM, Asphaug E (2001) Origin ofthe Moon in a giant impact near the endof the Earth’s formation. Nature 412: 708-712

Chazey WI, Neal CR (2004) Large igneousprovince magma petrogenesis fromsource to surface: platinum-groupelement evidence from Ontong JavaPlateau basalts recovered during ODP legs130 and 192. In: Fitton JG, Mahoney JJ,Wallace PJ, Saunders AD (eds) Origin andEvolution of the Ontong Java Plateau.Geological Society of London SpecialPublication 229, pp 219-238

Dietz RS (1972) Sudbury astrobleme, splashemplaced sub-layer and possible cosmo-genic ores. In: Guy-Bray JV (ed) NewDevelopments in Sudbury Geology.Geological Association of Canada SpecialPaper 10, pp 29-40

Elkins-Tanton LT, Hager BH, Grove TL(2004) Magmatic effects of the lunar lateheavy bombardment. Earth and PlanetaryScience Letters 222: 17-27

Gill RCO, Holm PM, Nielsen TFD (1995)Was a short-lived Baffin Bay plume activeprior to initiation of the present Icelandicplume? Clues from the high-Mg picritesof West Greenland. Lithos 34: 27-39

Glikson AY (2001) The astronomicalconnection of terrestrial evolution:crustal effects of post-3.8 Ga mega-impactclusters and evidence for major 3.2 ± 0.1Ga bombardment of the Earth–Moonsystem. Journal of Geodynamics 32: 205-229

Green DH (1972) Archaean greenstonebelts may include terrestrial equivalentsof lunar maria? Earth and PlanetaryScience Letters 15: 263-270

Grieve RAF (1980) Impact bombardmentand its role in proto-continental growthon the early earth. Precambrian Research10: 217-247

Grieve RAF, Cintala MJ (1992) An analysisof differential impact melt-crater scalingand implications for the terrestrial impactrecord. Meteoritics 27: 526-538

Ingle S, Coffin MF (2004) Impact origin forthe greater Ontong Java Plateau? Earthand Planetary Science Letters 218: 123-134

Ivanov BA, Melosh HJ (2003) Impacts donot initiate volcanic eruptions: Eruptionsclose to the crater. Geology 31: 869-872

Jones AP (2002) Komatiites: new informa-tion on the type locality (Barberton), andsome new ideas. Geology Today 18: 23-25

Jones AP, Price GD, Price NJ, DeCarli PS,Clegg RA (2002) Impact induced meltingand the development of large igneousprovinces. Earth and Planetary ScienceLetters 202: 551-561

Jones AP, Price GD, De Carli PS, Price NJ,Clegg RA (2003) Impact decompressionmelting: a possible trigger for impactinduced volcanism and mantle hotspots?In: Koeberl C, Martinez-Ruiz F (eds)Impact markers in the StratigraphicRecord. Springer, Berlin, pp 91-120

Jones AP, Wunemann K, Price GD (2005a)Modeling impact volcanism as a possibleorigin for the Ontong Java Plateau. In:Foulger GR, Natland JH, Presnall DC,Anderson DL (eds) Plates, Plumes andParadigms. Geological Society of AmericaSpecial 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 largePalaeocene impact, preserved aroundDisko bay, W. Greenland?: Nuussuaqspherule beds origin by impact instead ofvolcanic eruption? In: Kenkman K, HörzF, Deutsch A (eds) Large MeteoriteImpacts III. Geological Society of AmericaSpecial Paper 384: pp 281-298

Jones CE, Jenkyns HC (2001) Seawaterstrontium isotopes, oceanic anoxicevents, and seafloor hydrothermalactivity in the Jurassic and Cretaceous.American Journal of Science 301: 112-149

Koeberl C (2002) Mineralogical andgeochemical aspects of impact craters.Mineralogical Magazine 66: 745-768

Pierazzo E, Vickery AM, Melosh HJ (1997)A reevaluation of impact melt produc-tion, Icarus 127: 408-423

Price NJ (2001) Major Impacts and PlateTectonics. Routledge, London, 416 pp

Rogers GC (1982) Oceanic plateaus asmeteorite impact signatures. Nature 299:341-342

Stöffler D, Deutsch A, Avermann M,Bischoff L, Brockmeyer P, Buhl D,Lakomy R, Müller-Mohr V (1994) Theformation of the Sudbury structure,Canada: toward a unified impact model.In: Dressler BO, Grieve RAF, Sharpton VL(eds) Large Meteorite Impacts andPlanetary Evolution. Geological Societyof America Special Paper 293: 303-318

White RV, Saunders AD (2005) Volcanism,impact and mass extinctions: incredibleor credible coincidences? Lithos 79: 299-316 .


especially for oceanic targets, impact volcanism is a testablehypothesis. Continental LIPs (e.g. the Siberian Traps; Joneset al. 2002) may also be the result of impacts and might beexpected to show an admixed component from meltedcrust; such LIPs need to be modelled. Evidence of an initi-ating catastrophe, such as an ejecta layer at the base of aLIP, may in principal be found, but in practice may be miss-ing because of burial or lack of exposure, as in the case ofthe OJP. The West Greenland spherule horizon appears to

satisfy the impact–LIP criteria of superposition and maficLIP startup geochemistry and is accessible to future scrutiny.

ACKNOWLEDGMENTSI thank colleagues at UCL for many discussions, and partic-ularly David Price and Paul DeCarli for mainstream part-nership central to the theme of this review. Ian Parsons andStephanie Ingle are thanked for providing most thoughtfulreviews. .


When neophytes first gaze at the coloredrectangles in one of Piet Mondrian’s paintings,they may muse that the image is pretty, butpuzzle over whether it means anything. TheDutch artist is widely known for formulatingthe neoplasticism style of painting duringWorld War I, having successively given up onnaturalism, impressionism, pointillism, andcubism.

Colored geometric shapes also appear in thepetrographic microscope when one is viewingnon-opaque minerals in transmitted lightusing crossed polarizers. To the novice, theimages may seem pretty, but puzzling. Besidesa course in optical microscopy, what a begin-ner needs is a guide to the appearance of min-erals and rocks in thin section. In the early1980s, two such guides were published: Atlasof Rock-forming Minerals in Thin Section byMacKenzie and Guilford and Atlas of IgneousRocks and their Textures by MacKenzie,Donaldson, and Guilford. These are terrificbooks, filled with colorful images of rocks andmineral grains; adjacent photomicrographsare made in plane-polarized light and withcrossed polars. Short descriptions are providedand magnifications are included. The aspiringpetrographer can compare an uncharacterizedsample on the microscope stage with thephotos in the book and come up with atentative identification. At least two genera-tions of petrology students have used thesebooks to aid their studies of terrestrial rocks.

But what of the fledging meteoriticist? TheMacKenzie books are of limited help—mostof the illustrated minerals and rocks are notrepresented in meteorites. In anticipation ofthis need, Gustav Tschermak published TheMicroscopic Properties of Meteorites in 1885,replete with 100 black and white photomicro-graphs of stony meteorites in thin section. Ifwe use the modern classification system, wefind that Tschermak included images of CO,CV, and CR carbonaceous chondrites, equili-brated and unequilibrated ordinary chondrites,HED samples (howardites, eucrites, anddiogenites), aubrites, mesosiderites, silicatesfrom the IVA iron Steinbach, and two Martianmeteorites (Shergotty and Chassigny). Shortdescriptions are given for each figure.

Tschermak’s book was a fine one in its day. Itwas translated by John and Mathilde Woodand republished in the Smithsonian Contri-butions to Astrophysics in 1964. I keep a copyon my shelf, but I don’t often refer to it.There are informative images in the book tobe sure, but a newcomer to meteoritics wouldneed more. He or she would need an atlas in

color, one thatwould show all themajor chondritegroups, a book thatincluded everythingin Tschermak plusadditional non-chondrite groups,like ureilites,acapulcoites, lodra-nites, winonaites,brachinites,angrites, and lunarmeteorites. Ideally,the book shouldinclude photomicro-graphs made atdifferent magni-fications, and illustrate the same field of viewin plane-polarized light, with crossed polars,in reflected light, and with back-scatteredelectrons (BSE). Short descriptions would beessential and a brief introduction to thevarious meteorite groups would be an addedbonus. I would have greatly benefited fromsuch a book when I was a graduate student.I’m sure I wasn’t the only tyro wishing therewas a cosmochemistry edition of MacKenzie.

Dante Lauretta and Marvin Killgore haveheard our lament (we weepeth sore in thenight, and our tears were on our cheeks) andproduced A Color Atlas of Meteorites in ThinSection. With nearly 300 pages of photomicro-graphs and BSE images, it fills an essentialniche. It is not just for graduate students,dilettantes, and interested amateurs—I canalso recommend it to professional meteoritescientists. Few such folk are familiar with thetextures and mineralogy of all the meteoritegroups. For example, a quick flip of the pageswould demonstrate to the curious that lodra-nites are much coarser grained thanacapulcoites.

The photomicrographs and BSE images in theatlas are all in sharp focus (a feat not alwaysachieved in research papers), and the colors inthe crossed-polar images are vibrant. Theauthors found a printing house in South Koreathat could reproduce the images faithfully.

I like the book and recommend it, but thereare a few minor shortcomings. A peculiarityof the volume is the reliance on the secondauthor’s meteorite collection for so many(56%) of the illustrated specimens. In somecases, thin sections of less-weathered samplescould have been readily obtained from insti-tutional collections. An unfortunate omissionis the thin section number; this would havebeen of lasting value to petrologists who mightspot an interesting feature in the photomicro-graphs and want to request sections. In theIntroduction, the authors maintain that theyfollowed the classifications listed in the

authoritative Catalogueof Meteorites, edited byMonica Grady, but I can’tfigure out where theyobtained the unrealisticallylow subtypes of the EH3and EL3 chondrites—they’re not listed in Grady,the Antarctic MeteoriteNewsletter, or TheMeteoritical Bulletin. Themean diameters of H, L,and LL chondrules givenon page 11 (~0.3, ~0.7, and~0.9 mm, respectively) areout of date; the mostrecent estimates are ~0.3,~0.5, and ~0.6 mm (Rubin

2000). And as long as I’m being picayune, Hand L chondrites are named for their hightotal iron and low total iron contents, respec-tively, not for their high metal and low metalabundances as indicated on page 10. Thewell-known mesosiderite (stony-iron mete-orite) Vaca Muerta (i.e. Spanish for “DeadCow”) is misspelled in the Table of Contentsand the body of the atlas. But these flaws andthe rare typo (e.g. on p. 26) do not detractfrom the overall usefulness of the book.

At $98 this hardcover is not cheap, but neitheris a graduate school education. I recommendthat the atlas be available near the micro-scopes of all cosmochemistry laboratories sothat graduate students and their mentors willhave a handy reference when they find them-selves gazing at unfamiliar features in a thinsection. The legions of meteorite collectorswould also enjoy this book. If they are willingto shell out more than $40 per gram for adiogenite (meteoritic orthopyroxenite) oneBay, they should invest a hundred bucksand buy this new atlas.

REFERENCESGrady MM (2000) Catalogue of Meteorites, 5th

edition. Cambridge University Press, Cambridge,UK, 689 pp

MacKenzie WS, Guilford C (1980) Atlas of Rock-forming Minerals in Thin Section. LongmanGroup, New York, 98 pp

MacKenzie WS, Donaldson CH, Guilford C (1982)Atlas of Igneous Rocks and their Textures. Wiley,New York, 148 pp

Rubin A.E (2000) Petrologic, geochemical andexperimental constraints on models of chondruleformation. Earth-Science Reviews 50: 3-27

Tschermak G (1885) The Microscopic Properties ofMeteorites. In Smithson. Contrib. Astrophys. 4,(1964), 137-239

Alan E. RubinInstitute of Geophysics and Planetary Physics

University of CaliforniaLos Angeles, CA 90095-1567, USA

Book Review

A Color Atlas of Meteoritesin Thin Section1

1 Lauretta DS, Killgore M (2005) A Color Atlasof Meteorites in Thin Section. Golden RetrieverPublications, Tucson, AZ and Southwest MeteoritePress, Payson, AZ, 301 pp, ISBN 0-9720472-1-2, $98

INTRODUCTIONMass extinction events have fundamentally shaped the his-tory of life on Earth, and many recent studies focus uponcatastrophic factors as the cause. Two types of event havebeen invoked: episodes of continental flood basalt volcan-ism and major bolide impacts (asteroids or comets). Severalstudies have compared dates of extinction events of variousmagnitudes with dates of flood basalt episodes and foundsignificant correlations, thus supporting a possible cause-and-effect connection (e.g. Stothers 1993; Courtillot 1999).The apparent coincidence of eruptions in the Siberian floodbasalt lava flow province with the severe extinctions at theend of the Permian, ~250 Ma ago, and the near-coincidenceof activity in the Deccan flood basalt province (India) andthe extinctions at the end of the Cretaceous suggest thatflood basalt eruptions may have contributed significantlyto some mass extinction events, as summarized by Wignall(2001 and this issue).

Continental flood basalt (CFB) provinces are a type of largeigneous province (LIP) and are erupted subaerially for themost part. CFB province formation is unique with respect toall other subaerial basaltic magmatism because it is charac-

terized by the repeated effusion ofhuge magma batches over a shortperiod of geologic time (<1 Myr).These short-lived, main pulses oferuptive activity (Rampino andStothers 1988) consist of manylarge-volume and prolonged erup-tions, each commonly yieldinglava flow fields (see glossary) of102–103 km3 (FIG. 1). Both the vol-ume of magma emitted duringthese individual eruptions and thetotal volume of magma releasedduring the main eruptive pulses(up to several million km3) areexceptional in Earth history.

This contribution examines thepotential for volcanic gas releaseduring flood basalt eruptions andthe manner in which this gasrelease might then affect regionaland global environments. The

obvious major way is through the effects of gases and result-ant aerosols on the Earth’s atmosphere and surfaceprocesses during and immediately after an eruption.Previous studies have not investigated the species of gasreleased, which are carbon dioxide (CO2), sulfur dioxide(SO2), nitrous oxides (NOx), and the dominant volcanic gas,H2O, or the quantities. Important considerations are howoften these gases were released during the main eruptivepulse (i.e. what was the eruption frequency?) and the man-ner in which they were introduced into the atmosphere byflood lava activity. Clearly, such information will enablebetter constrained simulations to be made regarding the cli-matic and environmental impact of flood basalt volcanism.

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 ecosys-tems (e.g. Mather et al. 2004). We assess the potentialamount of gases released during past flood basalt eruptionsin order to determine the degree of atmospheric pollutionand hence potential environmental deterioration attributa-ble to this activity. The work builds on the one study madeso far that obtained a gas-release budget for an individualflood basalt eruption, that of the Roza lava flow field in themid-Miocene Columbia River basalt province of easternWashington State, USA (Thordarson and Self 1996),together with recent advances in knowledge concerningsimilar basaltic eruptions in Iceland.


Stephen Self1, Thorvaldur Thordarson2, Mike Widdowson1

1 Volcano Dynamics Group, Department of Earth Sciences,The Open University, Milton Keynes, MK7 6AA, UK

2 Department of Geology and Geophysics University of Hawai‘i at Manoa1680 East-West Road, Honolulu HI 96822, USA

Corresponding author: [email protected]

Subaerial continental flood basalt volcanism is distinguished from allother volcanic activity by the repeated effusion of huge batches ofbasaltic magma (~102–103 km3 per eruption) over short periods of

geologic time (<1 Myr). Flood basalt provinces are constructed of thick stacksof extensive pahoehoe-dominated lava flow fields and are the products ofhundreds of eruptions. Each huge eruption comes from a dyke-fed fissuretens to hundreds of kilometres long and lasts about a decade or more. Suchspatial and temporal patterns of lava production do not occur at any othertime in Earth history, and, during eruptions, gas fluxes of ~1 Gt per year ofSO2 and CO2 over periods of a decade or more are possible. Importantly, theatmospheric cooling associated with aerosols generated from the SO2 emis-sions of just one flood basalt eruption is likely to have been severe and wouldhave persisted for a decade or longer. By contrast, warming due to volca-nogenic CO2 released during an eruption is estimated to have been insignifi-cant because the mass of CO2 would have been small compared to thatalready present in the atmosphere.

KEYWORDS: flood basalt volcanism, eruption volumes,

gas release, atmospheric impact

Back-scattered electronmap of olivine (blue) in

glass (yellow). TheStapafell eruption was

subglacial, and rapidcooling has caught melt

inclusions beingentrapped. PHOTO JOHN

MACLENNAN

Gas Fluxes from FloodBasalt Eruptions


ESTIMATING VOLATILE RELEASE FROMFLOOD BASALT ERUPTIONSAs with the extensive ash- and spatter-fall deposits associ-ated with Icelandic basaltic fissure eruptions, such as Eldgjain AD 934 and Laki in AD 1783 (Thordarson et al. 2003),the occurrence of deposits of spatter, spatter-fed lava, andscoria mounds along eruptive fissures in the ColumbiaRiver province suggests that violent fire-fountainingoccurred during prolonged flood basalt eruptions(Thordarson and Self 1998). Calculations assuming magmavolatile contents of 1–1.5 wt% suggest that fountains dur-ing periods of peak eruptive output may have exceeded 1.5km in height. Further, model estimates for convectiveplumes rising above the fountains indicate eruption col-umn heights in excess of 13 km. Thus, at tropical to mid-latitudes, flood eruptive activity is readily capable of lofting

gases and small amounts of fine ash to mid- to upper-tropospheric altitudes, and to the base of the stratospherefor high-latitude (or higher intensity) eruptions. In addi-tion, flood lava flows themselves are also capable of releas-ing gas from their surfaces as they move away from the ventsystem (FIG. 2), thus creating a lower level atmospheric gasand aerosol cloud. The combined release of gas from firefountains and the surface of lava would be near-continuousthroughout the period of eruptive activity, which may lastfrom a few to perhaps >100 years. Clearly, the impact of suchhuge, long-term degassing events is likely to be significant.

Flood basalt events have been considered to influence theenvironment in two ways: (1) by atmospheric coolingcaused by sulfuric acid (H2SO4) aerosols generated from theSO2 released, which scatter and absorb incoming solar radi-ation, thus increasing the opacity of the atmosphere (e.g.Rampino and Self 2000); and (2) by atmospheric warmingthrough the addition of the greenhouse gas CO2 (e.g.McLean 1985; Olsen 1999). These two simple cause-and-effect scenarios are a simplification of much more complexphenomena. For instance, SO2 may act as a greenhouse gasif present in sufficient concentrations at low altitudes,whereas widespread climatic cooling would be primarily aresult of the formation and spread of stratospheric sulfuricacid (H2SO4), or sulfate, aerosols formed by oxidation of theSO2 with water and hydroxyl OH- radicals in the atmos-phere (Robock 2000). In addition, while anthropogenic

CO2 emission is now known to cause measurable globalwarming, the actual quantities likely to be released duringan individual CFB eruption, although prodigious, are verysmall when compared with the natural atmospheric reservoir.

Sulfur Degassing and Sulfate AerosolsEstimates of S emissions from past eruptions can be madeby measuring the pre-eruption S concentration in glass(melt) inclusions trapped within crystals (FIGS. 2 AND 3) andthe post-eruption concentration (remaining in the meltafter degassing) in glassy tephra or lava (Devine et al. 1984).This ‘petrologic method’ of determining degassing budgetshas recently been validated for modern basaltic eruptions(Sharma et al. 2004) and from degassing estimates duringrecent flood lava flows such as Laki (Thordarson and Self2003; Thordarson et al. 2003). These studies demonstratethat ~75% of volatile S species present in the rising melt arereleased at the vents, largely as SO2. The gas is then rapidlytransported into the atmosphere where it instantly beginsto convert to sulfuric acid aerosols, and continues to do soover periods of days to about 1 month. The months-longLaki eruption is estimated to have released ~120 Mt of SO2,delivered by quasi-continuous emission but with maximumfluxes reaching 6 Mt per day in the early stages [1 mega-tonne (Mt) = 1012 grams (g)]. At this rate, three days of Lakiemissions would have released the same amount of SO2 asthe entire 1991 Mount Pinatubo eruption!

Sulfur gas releases from ancient flood basalt eruptions, theyoungest of which is the ~16 Ma Columbia River basaltprovince, are more difficult to determine because the lavashave typically been affected by the action of weatheringand ground water alteration, which can lead to the removalof mobile volatile elements. In addition, the petrologicmethod requires phenocrysts containing glass inclusions,but flood basalt successions are dominated by crystal-poorflows. However, the well-preserved and porphyritic Rozalava flow is one case where the S loss during eruption canbe measured (FIG. 2), and determinations indicate that con-centrations of S in the pre-eruption magma were similar tothat of the Laki magma. The eruption of this 1300 km3

pahoehoe lava flow field is estimated to have released about12 Gt of SO2 [1 gigatonne (Gt) = 1000 Mt], along with sig-nificant amounts of HF and HCl (Thordarson and Self1996). If the Roza event lasted about 10 years, then theannual average SO2 burden added to the atmosphere couldhave been as much as 1.2 Gt, with a significant portioninjected into the lower stratosphere above a 12 to 14 kmhigh tropopause by fire fountains and associated convectivecolumns (up to 15 km above the vents). This equates to ~3Mt 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 main-tained from the Laki vents for several weeks during thesummer of 1783.

For crystal-poor flood basalt lavas, or in instances wherealteration renders the petrological method untenable (forinstance, the 65 Ma Deccan lavas of India), a chemical ratioproxy can be used to estimate the original S content of thelavas. This method is based upon an observed relationshipbetween the composition (i.e. FeO content) of recent andhistoric basaltic lavas and their observed S content derivedeither by direct degassing measurement or from glass inclu-sion data (Thordarson et al. 2003). The relationship can befurther refined by substituting the FeO/TiO2 ratio of theglass or the bulk lava composition (FIG. 4). Quantitative val-ues for the pre-degassed content of ancient lavas may beestimated for the first time using this proxy technique.

Both the petrologic and proxy approaches indicate that vastvolumes of SO2 are likely to have been released continu-ously during eruption and emplacement of individual flood

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 lavaflow fields (1–4), each the product of a huge-volume eruption of >1000km3 of lava. PHOTO BY S. SELF

FIGURE 1


basalt flow fields. Such huge gas emissions, maintained formany years, are totally unprecedented in any preconceivedmodels of volcanic gas release and sulfate aerosol genera-tion (cf. Pinto et al. 1989). As a crude estimate taking intoaccount the range of magma compositions involved, foreach cubic kilometre of lava emitted during basaltic floodlava eruptions, between 5 and 10 Mt of SO2 would havebeen released into the atmosphere.

Volcanogenic SO2 is converted to H2SO4 aerosols in the tro-posphere and stratosphere; in many eruptions, magmaticwater and entrained tropospheric water entering the strato-sphere with the eruption columns are sufficient for thisconversion. Very large burdens of SO2 would, however,have the potential to greatly deplete stratospheric H2O andOH- (Bekki 1995). The global stratosphere contains ~1 Gt ofH2O, but how an enormous amount of aerosols would man-ifest itself on atmospheric chemistry and dynamics cannotyet be assessed. The balance between a quasi-continuousvolcanogenic source of S gas, the resulting H2SO4 aerosolformation rates, and the deposition rates for aerosols is onlyjust beginning to be tested by atmospheric chemistry mod-els (Stevenson et al. 2003). From modern eruptions it isknown that significant volcanic S injection into the strato-sphere results in measurable global cooling (e.g. Blake2003). Volcanic aerosol clouds at any altitude in the atmos-phere should cool the Earth’s surface if the aerosol particlesize is in the normal range (Lacis et al. 1992), and a maxi-mum cooling effect is achieved by low-level aerosols.Therefore, volcanogenic aerosols in the troposphere, as wellas the stratospheric aerosols, will lead to cooling of the sur-face. However, after most volcanic eruptions, troposphericaerosols have a very short lifetime of only about one weekbefore they are ‘washed out’ by cloud formation and rain,while stratospheric residence times are much longer (typi-cally up to 2 years).

Estimates suggest that single flood basalt eruptions can gen-erate such large amounts of sulfate aerosols that they couldcreate regional optical depths (a measure of atmosphericopacity, see glossary) of >10. At such high levels of aerosolloading, it is arguable that the immediate climate coolingmight suppress convection in the lower atmosphere to theextent that ‘rain-out’ might become less effective at remov-ing the aerosols. Similarly, stratospheric dehydration wouldoccur due to the high-altitude portion of the injected gasand delay the conversion to aerosols of further masses ofinjected gas (Savarino et al. 2003) during the prolongederuptions. Clearly, even though the cooling effect resultingfrom a massive continuous flux of S gas resulting from anaverage flood lava eruption can perhaps be crudely calcu-lated (Jolley and Widdowson 2005), global climate modelswill be required in order to further explore the detailsregarding the extent and severity of the effect.

Carbon Dioxide To date, there are no direct measurements of CO2 concen-trations in glass (melt) inclusions within crystals from floodbasalt lavas that may give an indication of pre-eruptionvalues. CO2 is relatively insoluble in basaltic melts, andeven CO2 abundances in glass inclusions may not reflectthe 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 basalticlavas suggests between 0.2 and 0.5 wt% (Cervantes andWallace 2003). Nevertheless, the universally low CO2 con-centrations (usually <0.03 wt%) measured in the matrixglass of recent subaerially erupted lavas indicate thatdegassing of CO2 is a highly efficient process.

Schematic illustration showing the key features of a two-stage degassing model for a flood basalt eruption (modi-

fied from Thordarson et al. 2003). The amounts of SO2 degassed are

determined by a study of the Roza flow, Columbia River basalt province(after Thordarson and Self 1996); the absolute masses of SO2 are basedon a flow volume of 1300 km3.

FIGURE 2

Accepting 0.5 wt% as a reasonable but possibly high valuefor pre-eruptive CO2 concentration, and assuming 100%degassing, approximately 13 Mt of CO2 could be releasedfor 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 1/200of the amount of CO2 present in the modern atmosphere(~3000 Gt), and only about 3% of the current annualland–atmosphere CO2 flux. Moreover, because the gaswould actually be released throughout the eruptive event,the annual fluxes to the atmosphere during such an erup-tion would be considerably smaller. A more realistic esti-mate might be to assume 80% efficient degassing of CO2

during a 1000 km3 eruption; this would yield a flux of ~1and ~0.2 Gt per year for 10 and 50 year eruption durations,respectively. Clearly, such annual fluxes are negligiblewhen compared with the natural atmospheric reservoir.

CONCLUSIONSWe have shown that during decade-long flood basalt erup-tive events, and over the whole period of generation of aflood basalt province, immense and sustained degassingevents should occur on a periodic basis. These events areincomparably greater in scale than volcanic gas releases atany other time in Earth history. While the impact of vol-canic S gas release may be profound, the mass of CO2

directly released by individual flood lava eruptive events istiny in comparison to the normal mass in the troposphereand stratosphere. The predicted increases in atmosphericconcentration are a fraction of the current anthropogenicCO2 released from hydrocarbon burning (~25 Gt per year).Moreover, while the amount of CO2 in the atmosphere iscurrently ~3000 Gt, it was perhaps double this value duringthe late Cretaceous (i.e. ~6000 Gt). It is therefore unlikelythat volcanic CO2 had a direct effect on mechanisms ofglobal warming, supporting earlier findings by Caldeira andRampino (1990). In addition, there would have been morethan sufficient time for the extra mass of CO2 added toequilibrate, given that the lava-forming eruptive eventsmust have been spaced at least hundreds, and probablythousands, of years apart. By contrast, SO2 emissions andthe atmospheric burden of sulfate aerosols generated during

flood basalt events appear to be unprecedented at any othertime in Earth history. Acid rain may also have been wide-spread. What is less certain is whether affected biota wouldhave had time to recover from the deleterious effects of sul-fate aerosol clouds and acid rain, although quiescent inter-vals lasting millennia appear to offer ample time for therecovery of local biological and environmental systems(Jolley 1997).

The picture emerging from work on the Columbia Riverand Deccan flood basalt provinces is of a rapid, very volu-minous eruptive pulse consisting of hundreds of eruptions,with average repose times of thousands of years betweensuccessive eruptions. However, in reality, these eruptivehiatuses must have varied considerably in duration. TheSO2 released during each eruption would have formed con-siderable amounts of sulfate aerosols, with effects lasting atleast as long as the eruptions persisted (decades andpossibly longer). The potential impact of such huge, long-duration degassing events under palaeoclimatic and palaeo-atmospheric conditions warrants much further investigation.

ACKNOWLEDGMENTSThanks to an anonymous reviewer for helpful commentson the initial draft of this article. Support to the first authorduring this work came from the UK Natural EnvironmentResearch Council. .


Thin section (plane polarized light) of a sample of theglassy margin of the Thórsá lava flow, Iceland, (~8500

years old) showing a plagioclase phenocryst set in a light brown glassymatrix with smaller plagioclase groundmass crystals; the phenocrysthosts a glass inclusion (center, white arrow) with a shrinkage bubble.The inset (at same scale) shows another phenocryst with several brownglass inclusions up to 60 micrometres long and brown glass attached atthe top with vesicles (bubbles). The petrologic method (see text) reliesupon analyses of S in inclusions and matrix glass (see also Fig. 2).

FIGURE 3

An empirical model gives a proxy for estimating the sul-fur released from eruptions of tholeiitic basalt magmas.

S concentration is plotted against the TiO2/FeO ratio for basalt lava sam-ples of various ages from Iceland and other regions (MORB is mid-oceanridge basalt; Roza is discussed in text). A best fit line (A) is shownthrough fields of data derived from glass inclusions in crystals fromIcelandic and other eruptions, indicating pre-eruption S concentrationsin magma. Best fit lines B and C are through fields of degassed Icelandicvent tephra and lava flows. The example plotted is for a sample from aflood lava eruption with a TiO2/FeO ratio of 0.2. The difference between1650 ppm (line A) and 480 ppm (line B) represents the amount of Sdegassed at vents; the difference between 480 ppm and 270 ppm (fieldC) represents amount of S degassed from the lava flow. All data are fromelectron microprobe analyses; after Thordarson et al. 2003 and refer-ences therein).

FIGURE 4


REFERENCESBekki S (1995) Oxidation of volcanic SO2:

a sink for stratospheric OH and H2O.Geophysical Research Letters 22: 913-916

Blake S (2003) Correlations betweeneruption magnitude, SO2 yield, andsurface cooling. In: Oppenheimer C, PyleDM, Barclay J (eds) Volcanic Degassing.Geological Society of London SpecialPublication 213, pp 371-380

Caldeira K, Rampino, MR (1990) Carbondioxide emissions from Deccanvolcanism and a K/T boundary green-house effect. Geophysical ResearchLetters 17: 1299-1302

Cervantes P, Wallace PJ (2003) Role of H2Oin subduction-zone magmatism: Newinsights from melt inclusions in high-Mgbasalts from central Mexico. Geology 31:235-238

Courtillot V (1999) EvolutionaryCatastrophes: The Science of MassExtinctions. Cambridge University Press,Cambridge, 188 pp

Devine JD, Sigurdsson H, Davis AN, Self S(1984) Estimates of sulfur and chlorineyield to the atmosphere from volcaniceruptions and potential climatic effects.Journal of Geophysical Research 89 (B7):6309-6325

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

Jolley DW, Widdowson M (2005) DidPaleogene North Atlantic rift-relatederuptions drive early Eocene climatecooling? Lithos 79: 355-366

Lacis A, Hansen J, Sato, M (1992) Climateforcing by stratospheric aerosols. Geo-physical Research Letters 19: 1607-1610

Mather TA, Pyle DM, Allan AG (2004)Volcanic source for fixed nitrogen in theearly Earth’s atmosphere. Geology 32:905-908

McLean DM (1985) Deccan Traps mantledegassing in the terminal Cretaceousmarine extinctions. Cretaceous Research6: 235-259

Olsen PE (1999) Giant lava flows, massextinctions, and mantle plumes. Science284: 604-605

Pinto JR, Turco RP, Toon OB (1989) Self-limiting physical and chemical effects involcanic eruption clouds. Journal ofGeophysical Research 94 (D8): 11165-11174

Rampino MR, Self S (2000) Volcanism andBiotic Extinctions. In: Sigurdsson H et al.(eds) The Encyclopedia of Volcanoes,Academic Press, London, pp 263-269

Rampino MR, Stothers R (1988) Floodbasalt volcanism during the past 250million years. Science 241: 663-668

Robock A (2000) Volcanic eruptions andclimate. Reviews of Geophysics 38: 191-219

Savarino J, Bekki S, Cole-Dai J, ThiemensMH (2003) Evidence from sulfate massindependent oxygen isotopic composi-tions of dramatic changes in atmosphericoxidation following massive volcaniceruptions. Journal of GeophysicalResearch 108 (D21): doi:10.1029/2003JD003737

Sharma K, Blake S, Self S, Krueger AJ (2004)SO2 emissions from basaltic eruptions,and the excess sulfur issue. GeophysicalResearch Letters 31: L13612, doi:10.1029/2004GL019688

Stevenson D, Johnson C, Highwood E,Gauchi V, Collins W, Derwent R (2003)Atmospheric impact of the 1783-1784Laki eruption: Part 1, Chemistry model-ling. Atmospheric Chemistry and PhysicsDiscussions 3: 551-596

Stothers RB (1993) Flood basalts andextinction events. Geophysical ResearchLetters 20: 1399-1402

Thordarson T, Self S (1996) Sulphur,chlorine and fluorine degassing andatmospheric loading by the Rozaeruption, Columbia River Basalt group,Washington, USA. Journal of Volcano-logical and Geothermal Research 74: 49-73

Thordarson T, Self S (1998) The RozaMember, Columbia River Basalt Group:A gigantic pahoehoe lava flow fieldformed by endogenous processes? Journalof Geophysical Research 103 (B11):27411-27446

Thordarson T, Self S (2003) Atmosphericand environmental effects of the 1783–84Laki 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) Sulphurrelease from flood lava eruptions in theVeidivötn, Grimsvötn, and Katla volcanicsystems. In: Oppenheimer C, Pyle DM,Barclay J (eds) Volcanic Degassing.Geological Society of London SpecialPublication 213, pp 103-122

Wallace P, Anderson Jr AJ (2000) Volatilesin magmas. In: Sigurdsson H et al. (eds)The Encyclopedia of Volcanoes.Academic Press, London, pp 149-170

Wignall PB (2001) Large igneous provincesand mass extinctions. Earth-ScienceReviews 53: 1-33 .


Aqueous fluids play a crucial rolein mineral weathering and trans-port in sediments, as well as influid–rock interaction deeper inthe Earth’s crust. For decadesresearchers have sought generaldescriptions of mineral dissolu-tion kinetics as a means toquantify and predict behaviorover time spans greatly exceedingthose accessible in the laboratory.However, a relationship incorpo-rating thermodynamic variablesand time that “works” over arange in saturation level andtemperature has been elusive—even for a single mineral underlaboratory conditions.

In a study published recently inthe Proceedings of the NationalAcademy of Sciences, Patricia Doveand Nizhou Han at Virginia Techand their colleague Jim De Yoreoat Lawrence Livermore NationalLaboratory have shown thatmineral dissolution can be under-stood through the same mecha-

nistic theory of nucleationdeveloped previously to describemineral growth. By viewing disso-lution as the formation of“negative crystals,” the resear-chers were able to generalize thenucleation-rate equations toobtain a model that predictsprogressive changes in thedissolution mechanism withdegree of undersaturation—fromstep retreat to defect-controlled

to homogeneous nucleation ofetch pits. The underlying reasonfor the transitions is that the rateat which a crystal dissolves iscontrolled by the density of stepson the surface. Whicheverprocess creates the greatest stepdensity dominates the dissolutionprocess.

Prof. Dove and her colleaguestested the model using a newset of experiments performed onpure quartz sand, in which theymeasured the dependence ofdissolution rate on undersatura-tion. Complementary AFMobservations of the steady statesurfaces that developed onnatural faces showed that ratesconform to model predictionsover a wide range of thermody-namic conditions. They alsoshowed that the “salt effect,”recognized almost 100 years ago,results from a transition in thedominant mechanism fromdislocation- to nucleation-drivendissolution.

It turns out that the model (seefigure) also explains dissolutionbehavior previously reported forfeldspar. It also reconciles appar-ently conflicting data sets forkaolinite dissolution at twodifferent temperatures byshowing the basis for a tempera-ture-activated transition in thedissolution process. Thesedifferences suggest that thepractice of extrapolating high-temperature data to ambienttemperatures may over-predictrates of dissolution when theextrapolation comes from ratesdriven by etch-pit nucleation.

The researchers believe that theirnucleation model may alsopredict dissolution rates of spar-ingly soluble salts, perhaps open-ing up the possibility to under-stand “demineralization” ofbiological materials under certainconditions.

Dove PM, Han N, De Yoreo JJ(2005) Mechanisms of classicalcrystal growth theory explainquartz and silicate dissolutionbehavior. Proceeding of theNational Academy of Sciences102: 15357-15362

This article was picked by Bruce Watson.

The Editors’ Pick

MECHANISMS OF CLASSICAL CRYSTAL GROWTHTHEORY EXPLAIN QUARTZ AND SILICATEDISSOLUTION BEHAVIOR

RARE MINERALS FOR RESEARCH

We maintain a stock of over 200,000 mineral specimens from

worldwide localities, encompassing more than 3100 differentspecies. If you require reliably identified samples for your researchwork, let us supply your needs. We and our predecessor

companies have been the suppliers of choice for researchers since1950! Inquiries via e-mail are encouraged.

DIAMOND CRYSTALSWe have acquired the extensive diamond crystal inventories ofDavid New and others, amassing one of the largest collections oflocality-specific diamond specimens available in today’s market.

Over twenty-five different localities are represented in this inventoryof several thousand rough diamond specimens. We can supply avariety of crystal habits, colors and localities to meet your research

needs. Individual specimens as well as comprehensive referencecollections are on hand. See our website, and specific inquiries viae-mail are welcomed.

Excalibur Mineral

Corporation1000 North Division Street – Peekskill, NY 10566-1830

Tel: (914)739-1134 Fax: (914)739-1257www.excaliburmineral.com - www.diamondcrystals.net

E-mail: [email protected]

A mechanistic model that takes intoaccount the competition between stepcreation at dislocations and by nucleationhighlights the analogy between growthand dissolution and explains apparentdiscrepancies in measured growth anddissolution rates. A) Rates measured at80°C by Nagy et al. 1991, AJS, areexplained by the dislocation model;whereas B) the dependence of rate onchemical driving force at the higher tem-perature of 150°C by Devidal et al. 1997,GCA, is predicted by the nucleationmodel.

INTRODUCTIONAlthough the potentially devastating environmentalimpact of continental flood basalts has been extensivelydiscussed by many authors (Wignall 2001; articles by Self etal. and Wignall this issue), the global environmental effectsof oceanic plateaus – the marine equivalent of continentalflood basalts – have received comparatively little attention.

Oceanic plateaus represent over-thickened areas of oceaniccrust (>10 km) in which the bulk of the >1 × 106 km3 lavavolume appears to have erupted in less than 2–3 Myr. Theseplateaus generally cover an area in excess of 1 × 106 km2

and result from anomalously high rates of melt productionin the mantle. These high melt-production rates are mostlikely due to the excess heat from a deep-rooted mantleplume (Campbell this issue).

Oceanic plateaus are more buoyant than oceanic crust ofnormal thickness generated at a mid-ocean ridge. Thisbuoyancy means that oceanic plateaus are more resistant tosubduction, a feature which results in partial accretion ontocontinental margins. In this way oceanic plateaus can bepreserved in the geological record and can be recognisedback to the earliest Archaean.

Periods of oceanic environmental crisis can be identified inthe geological record by the occurrence of black shales,which are indicative of low-oxygen or oxygen-absent con-ditions in the deep ocean. Vogt (1989), Sinton and Duncan(1997), and Kerr (1998) have noted the coincidence ofglobal oceanic anoxia, black shale deposition, mass extinc-tion and oceanic plateau formation in the Cretaceous, par-ticularly around the Cenomanian–Turonian boundary

(93.5 Ma) and during the Aptian(124–112 Ma) (Sliter 1989;Bralower et al. 1993; Jahren 2002).This link appears to extend back tothe Precambrian: Condie et al.(2001) have noted that significantblack shale events occurred at ~1.9and 2.7 Ga and that these correlatewith the formation of mantleplume-derived large igneous pro-vinces (LIPs) and warmer palaeocli-mates. Thus, there seems to be atemporal association, throughouta significant proportion of geolog-ical time, between periods ofglobal oceanic environmentalcrises and oceanic plateau forma-

tion. In the remainder of this contribution the possiblecausal links between oceanic plateaus and oceanic environ-mental change will be reviewed, with particular reference toCretaceous events.

OCEANIC PLATEAU VOLCANISM ANDGLOBAL OCEANIC ANOXIA AT THECENOMANIAN–TURONIAN BOUNDARY

Oceanic Volcanism Arguably, the clearest link between oceanic plateau volcan-ism and environmental perturbation can be seen aroundthe Cenomanian–Turonian (C–T) boundary. This timeperiod is marked by the formation of the Caribbean–Colombian oceanic plateau (eastern Pacific) and parts ofboth the Kerguelen Plateau (Indian Ocean) and possibly theOntong Java Plateau (western Pacific) (FIG. 1). Also aroundthis time, India and Madagascar were beginning to riftapart, and this event was associated with volcanism at theMarion hotspot, which resulted in flood basalt eruptions onMadagascar and basaltic lavas offshore. Due to the contin-ued breakup of Gondwana, the length of the global ridgesystem increased in the mid–late Cretaceous, resulting in asignificantly greater volume of lava erupted globally at mid-ocean ridges. Kerr (1998) has calculated that the peak pro-duction of oceanic crust (both intrusive and extrusive)around the C–T 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 C–T boundary ischaracterised by black organic-rich shales, signifying anoxicoceanic conditions. This black shale event was associatedwith a second-order mass extinction event marked by thedemise of 26% of genera (Sepkoski 1986). The C–T bound-ary is also characterised by a sharp increase in δ13C from 1.5


Andrew C. Kerr1

1 School of Earth, Ocean and Planetary Sciences Cardiff University, Main BuildingPark Place, Cardiff, Wales CF10 3YE, UKE-mail: [email protected]

Oceanic plateaus represent large areas (~1 ×× 106 km2) of thickenedoceanic crust formed from rapidly erupted lava (<3 Myr). Theseplateaus have formed throughout most of geological time. They

generally correlate with periods of environmental catastrophe characterisedby oceanic anoxia, leading to black shale formation and mass extinctionevents. Such correlations are particularly evident in the Cretaceous and canbe partly attributed to the release of CO2 during oceanic plateau formation,which ultimately resulted in a runaway greenhouse effect. Additionally, sealevel rise and disruption of oceanic circulation patterns by displacement ofseawater during plateau formation contributed to increased environmentalstress and biotic extinction.

KEYWORDS: mass extinction, oceanic plateau,

black shale, anoxia, mantle plume

Oceanic plateaus havebeen drilled from theJOIDES Resolution drill

ship. PHOTO ODP



to 4‰ in pelagic limestones, a decline in 87Sr/86Sr of sea-water and evidence for a significant transgression (FIG. 2).Oxygen isotope evidence reveals that globally averaged sur-face temperatures at the C–T boundary were 6 to 14°Cwarmer than now (Kaiho 1994). This temperature rise wascaused by elevated CO2 levels in the atmosphere. Modellingby Berner (1994) suggests atmospheric CO2 levels at thistime were up to six times greater than pre-industrial levelsand reached a peak around the C–T boundary.

Links between Oceanic Plateau Volcanismand Environmental Catastrophe The coincidence of extensive oceanic plateau volcanismand the physical and chemical phenomena outlined abovedemand that we look for the causal links between oceanicplateau volcanism, global oceanic anoxia and warming, andmass extinction events.

The formation of LIPs on both the continents and theoceans is often accompanied by lithospheric uplift anddoming (Larson 1991; Nadin et al. 1997). Under the oceans,this elevation, in combination with the displacement ofwater due to the eruption of ~10 × 106 km3 of lava onto theocean floor, results in a significant rise in sea level (FIG. 3).This mechanism may provide an explanation for at leastpart of the estimated ~100 m sea level rise which reached amaximum at the C–T boundary. The elevation of the seafloor, from both plume uplift and voluminous lava extru-sion, during oceanic plateau formation could also have dis-rupted important oceanic circulation systems around theC–T boundary. The Caribbean–Colombian oceanic plateauformed close to the proto-Caribbean seaway between Northand South America (FIG. 1). At this time, the only majorsource of deep, cold, oxygenated water for the juvenileAtlantic was the Pacific, and the water had to pass throughthis proto-Caribbean seaway. The formation of such amajor volcanic edifice and the associated shallowing of sea-water so close to this oceanic gateway would have restrictedthe flow of deep oxygenated water from the Pacific to theAtlantic and thus increased the extent of oceanic anoxia inthe Atlantic (de Boer 1986).

Although volcanism undoubtedly contributed to the ele-vated CO2 contents in the C–T atmosphere, it is doubtful ifvolcanism alone could have released enough CO2 to causethe higher temperatures calculated to exist at this time (Selfet al. this issue). However, it is likely that a complex positive

feedback mechanism triggered by the volcanically derivedCO2 led to increased CO2 and elevated temperatures (FIG. 3).The initial emission of CO2 from oceanic plateau volcanismwas probably accompanied by the release of a considerableamount of SO2 and halogens (Self et al. this issue), whichwould have made the oceans locally more acidic (FIG. 3).This increased acidity would have led to the dissolution ofshallow-water carbonates, thus releasing more CO2 to theatmosphere. [Significantly, Arthur et al. (1987) have notedthat the C–T boundary is characterised by a lack of carbon-ates.] Thus, the addition of carbonate-derived CO2 plus vol-canic CO2 to the atmosphere at the C–T boundary wouldhave caused global warming of both the atmosphere andthe oceans. Since the solubility of CO2 in seawater decreasesby 4% for every 1°C rise in temperature, warming wouldhave resulted in the release to the atmosphere of yet moreCO2 which was previously dissolved in the oceans. Thus, apositive CO2 feedback mechanism would have been estab-lished (FIG. 3). Kerr (1998) has proposed that such a scenariowould relatively rapidly result in the establishment of arunaway greenhouse effect.

Cenomanian–Turonian plate tectonic reconstructionshowing the location of ~90–93 Ma and 123–110 Ma

large igneous provinces. Cenomanian–Turonian boundary black shaledeposits are also shown. The black shale localities are taken from Arthuret al. (1987), Herbin et al. (1987), Summerhayes (1987), and Yurtsever etal. (2003).

FIGURE 1

Diagrams showing changes in key environmental indica-tors, 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


Increased atmospheric CO2 levels, in combination with dis-rupted oceanic circulation patterns and associatedupwelling of nutrients from the deep ocean, would haveresulted in increased biogenic productivity in ocean surfacewaters (FIG. 3). This increased biological activity led toremoval of CO2 from the atmosphere and provided a mech-anism for reducing the amount of atmospheric CO2. Theδ13C peak in shallow ocean sediments at the C–T boundaryreflects increased burial of marine organic carbon in theoceans and is due to the preference of organic matter forisotopically light carbon.

The decrease in 87Sr/86Sr in the stratigraphic record (from0.70753 to 0.70735), which started in the late-Cenomanianand continued until the mid-Turonian (FIG. 2), may be areflection of the addition to seawater of hydrothermal flu-ids with a low 87Sr/86Sr from oceanic plateau volcanism.Conversely, the rise in 87Sr/86Sr from the mid-Turonianonwards may signify increased continental weatheringresulting from global warming and its associated climaticdisturbance. Continental weathering is another mechanismwhich can reduce the amount of atmospheric CO2.

Higher oceanic temperatures would also have contributedto oceanic anoxia since the solubility of O2 in seawaterdecreases by 2% for every 1°C temperature rise (de Boer1986). However, given that globally averaged ocean tem-peratures appear to have increased by at most 6°C (FIG. 2),the consequent ~10% reduction in the solubility of O2 inseawater is not enough to explain widespread oceanicanoxia. Several additional mechanisms by which oceanicplateaus can contribute to the depletion of dissolved O2

have been discussed by Sinton and Duncan (1997). The firstof these is the reduction of dissolved O2 in seawater by thereaction of trace metals and sulphides in hydrothermalfluids with the O2. Although basaltic lava flows can be oxi-dised by hydrothermal fluids, both during and after erup-tion, this process is volumetrically insignificant when com-pared to the much greater effect of the oxidation of metalsin hydrothermal fluids in lowering the amount of dissolvedoxygen in seawater (FIG. 3). The eruption of a 1 × 104 km3

oceanic plateau basalt lava flow would release a similar vol-ume of hydrothermal fluids at 350°C into the ocean (Cath-les, cited in Sinton and Duncan 1997). Sinton and Duncan(1997) have calculated that the complete oxidation of the

Fe2+, Mn2+, H2S, and CH4 in 1 x 104 km3 of hydrothermalfluid would use up ~6% of the total dissolved oxygen in thepresent-day ocean. However, as we have seen, the C–Tocean was significantly warmer and contained less dis-solved oxygen than the present-day ocean. Therefore, theproportion of dissolved oxygen removed by 1 × 104 km3 ofhydrothermal fluid at the C–T boundary would have beensignificantly greater than 6%.

Another mechanism for removing dissolved oxygen fromseawater, as discussed by Sinton and Duncan (1997), is thestimulation of the growth of living organisms (organic pro-ductivity) in the oceans by the injection of hydrothermaliron into surface waters. Vogt (1989) has suggested thathydrothermal plumes, even those several orders of magni-tude smaller than 1 × 104 km3, would have been capable ofrising into oceanic surface waters. Thus, as noted by Sintonand Duncan (1997), a large (1 × 104 km3) hydrothermalplume (with a low brine content) could easily rise throughthe water column and spread laterally over a significantproportion of the ocean surface. The trace metal–richwaters of such massive hydrothermal plumes may well havestimulated increased levels of organic productivity in nutri-ent-poor surface waters (Sinton and Duncan 1997). Coale etal. (1996) have shown that the addition of Fe into oceansurface waters can result in a rapid increase in the amountof phytoplankton. If a similar hydrothermal fluid–inducedphytoplankton bloom occurred around the C–T boundary,the net effect would have been a further reduction in theamount of dissolved O2 as organic material decayed andsank through the seawater column (FIG. 3).

FURTHER LINKS BETWEEN OCEANICPLATEAU VOLCANISM ANDENVIRONMENTAL DISTURBANCEThe link between oceanic plateau volcanism and globaloceanic anoxia is given further credence by the occurrenceof Aptian (124–112 Ma) black shales, which probably repre-sent one of the most extensive concentrations of organic-rich black shales in the geological record (Jenkyns 1980;

Flow diagram of the likely physical and chemical environ-mental effects of oceanic plateau formation (see text for

a detailed description).

FIGURE 3

Hallam 1987; Bralower et al. 1993). It is no coincidence thatone of the most extensive periods of plume-related oceanicplateau formation occurred in the Pacific and Indian oceansduring this period, including the Ontong Java Plateau, HessRise, Manihiki Plateau, the East Mariana and Nauru basinsand a significant proportion of the Kerguelen Plateau(Eldholm and Coffin 2000). Although many of the chemi-cal and physical characteristics of the Aptian oceanicanoxic event are very similar to those of the C–T boundary,one of the black shale horizons in the early Aptian exhibitsa sharp decrease in δ13C, unlike the C–T boundary, whichshows a sharp increase in δ13C (Jahren 2002).

Present-day methane hydrates buried in ocean floor sedi-ments possess very low δ13C (around –60‰). Thus the dis-sociation and catastrophic release of hydrates may havecaused the sharp decrease in δ13C and contributed to globalwarming and oceanic anoxia in the early–mid Aptian(Jahren 2002). It has been proposed that methane releasewas triggered by tectonic events related to mantle plumeuplift (Jahren 2002). However, while tectonic processesundoubtedly played a role, I contend that LIP-inducedglobal warming was of greater importance and was proba-bly of sufficient magnitude to cause dissociation ofmethane hydrates and consequent release of methane. Thisrelease would have accelerated global warming and anoxiasince methane is a much more potent greenhouse gas thancarbon dioxide. Furthermore, atmospheric oxidation ofmethane would consume a significant amount of free oxy-gen. Methane release may also have occurred around theC–T boundary, but its distinctive? δ13C signal may havebeen diluted by higher δ13C resulting from volcanism

(around –10‰). Alternatively, the late Cenomanian oceanmay have been too warm for extensive methane deposits toaccumulate.

Increased oceanic volcanism as a cause of oceanic anoxiaand the deposition of organic-rich sediments has importantimplications for the location of potential oil source rocks. Itis likely that many of the world’s most important occur-rences of mid-Cretaceous oil source rocks owe their exis-tence to the formation of oceanic plateaus in the Pacificand Indian oceans and the resultant global anoxia.

This is obviously an interesting model, but how applicableis 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 oceanicplateau volcanism. For example, important Kimmeridgianto Tithonian (155–146 Ma) oil source rocks correlate withthe formation of the Sorachi Plateau in the western Pacific(Kimura et al. 1994). Furthermore, the formation ofToarcian (187–178 Ma) black shales corresponds with theeruption of the Karoo, Ferrar and Weddell Sea large igneousprovince during the breakup of Gondwana (Riley andKnight 2001).

Thus, global oceanic anoxia (and the consequent formationof oil source rocks) is frequently associated with the devel-opment of oceanic plateaus. It is contended that the forma-tion of oceanic plateaus perturbs the oceanic and atmos-pheric environments and sets in motion a chain of events,often leading to global warming, oceanic anoxia, blackshale deposition and ultimately mass extinction. .


REFERENCESArthur MA, Schlanger SO, Jenkyns HC

(1987). The Cenomanian-Turonianoceanic anoxic event; II: Palaeoceano-graphic controls on organic-matterproduction and preservation. In: BrooksJ, Fleet AJ (eds) Marine Petroleum SourceRocks. Geological Society of LondonSpecial Publication 26, pp. 401-420

Berner RA (1994) Geocarb II: a revisedmodel of atmospheric CO2 overPhanerozoic time. American Journalof Science 294: 56-59

Bralower TJ, Sliter WV, Arthur MA, LekieRM, Allard D, Schlanger SO (1993)Dysoxic/anoxic episodes in the Aptian-Albian (Early Cretaceous). In: Pringle MS,Sager WW, Sliter WV, Stein S (eds) TheMesozoic Pacific: Geology, Tectonics, andVolcanism, American Geophysical UnionMonograph 77, pp. 5-37

Coale KH, Johnson KS, Fitzwater SE,Gordon RM, Tanner S, Chavez FP, FerioliL, Sakamoto C, Rogers P, Millero F,Steinberg P, Nightingale P, Cooper D,Cochlan WP, Landry MR, ConstantinouJ, Rollwagen G, Trasvina A, Kudela R(1996) A massive phytoplankton bloominduced by an ecosystem-scale ironfertilisation experiment in the equatorialPacific Ocean. Nature 383: 495-501

Condie KC, DesMarais DJ, Abbott D (2001)Precambrian superplumes and supercon-tinents: a record in black shales, carbonisotopes, and paleoclimates? PrecambrianResearch 106: 239-260

de Boer PL (1986) Changes in the organiccarbon burial during the EarlyCretaceous. In: Summerhayes CP,Shackleton NJ (eds) North AtlanticPalaeoceanography. Geological Societyof London Special Publication 21,pp. 321-331

Eldholm O, Coffin MF (2000) Largeigneous provinces and plate tectonics.In: Richards MA, Gordon RG, van derHilst RD (eds) The History and Dynamicsof Global Plate Motions, AmericanGeophysical Union Monograph 121,pp. 309-326

Hallam A (1987) End-Cretaceous massextinction event: argument for terrestrialcausation. Science 238: 1237-1242

Herbin JP, Montadert L, Muller C, GomezR, Thurow J, Wiedmann J (1987)Organic-rich sedimentation at theCenomanian-Turonian boundary inoceanic and coastal basins in the NorthAtlantic and Tethys. In: SummerhayesCP, Shackleton NJ (eds) North AtlanticPalaeoceanography, Geological Societyof London Special Publication 21,pp 389-422

Jahren AH (2002) The biogeochemicalconsequences of the mid-Cretaceoussuperplume. Journal of Geodynamics34: 177-191

Jenkyns HC (1980) Cretaceous anoxicevents: from continents to oceans.Journal of the Geological Society ofLondon 137: 171-188

Kaiho K (1994) Planktonic and benthicforaminiferal extinction events duringthe last 100 m.y. Palaeogeography,Palaeoclimatology, Palaeoecology 111:45–71

Kerr AC (1998) Oceanic plateau formation:A cause of mass extinction and blackshale deposition around the Cenomanian-Turonian boundary. Journal of theGeological Society of London 155,pp 619-626

Kimura G, Sakakibara M, Okamura M(1994) Plumes in central Panthalassa?Deductions from accreted oceanic frag-ments in Japan. Tectonics 13: 905-916

Larson RL (1991) Geological consequencesof superplumes. Geology 19: 963-966

Nadin PA, Kusznir NJ, Cheadle, MJ (1997)Early Tertiary plume uplift of the NorthSea and Faeroe- Shetland Basins. Earthand Planetary Science Letters 148: 109-127

Riley TR, Knight KB (2001) Age of pre-break-up Gondwana magmatism. AntarcticScience 13: 99-110

Sepkoski JJ (1986) Phanerozoic overview ofmass extinction. In: Raup DM, JaplonskiD (eds) Pattern and Processes in theHistory of Life, Springer-Verlag, Berlin,pp. 277-295

Sinton CW, Duncan RA (1997) Potentiallinks between ocean plateau volcanismand global ocean anoxia at theCenomanian-Turonian boundary.Economic Geology 92: 836-842

Sliter WV (1989) Aptian anoxia in thePacific Basin. Geology 17: 909-912

Summerhayes CP (1987) Organic-richCretaceous sediments from the NorthAtlantic. In: Brooks J, Fleet AJ (eds)Marine Petroleum Source Rocks,Geological Society of London SpecialPublication 26, pp. 301-316

Vogt PR (1989) Volcanogenic upwellingof anoxic, nutrient-rich water: A possiblefactor in carbonate-bank /reef demise andbenthic faunal extinctions? GeologicalSociety of America Bulletin 101: 1225-1245

Wignall PB (2001) Large igneous provincesand mass extinctions. Earth-ScienceReviews 53: 1-33

Yurtsever TS, Tekin UK, Demirel IH (2003)First evidence of the Cenomanian/Turonian boundary event (CTBE) in theAlakirçay Nappe of the Antalya Nappes,southwest Turkey. Cretaceous Research24: 41-53 .

INTRODUCTIONCan large igneous province (LIP) eruptions cause environ-mental and climatic effects that are sufficiently severe tocause extinctions? The answer to this question has changedsignificantly over the past two decades, from “probably no”to “probably yes.” The change of view is due to advances inradiometric dating of both extinction events and the agesof LIPs (see Courtillot and Renne 2003). The close age cor-respondence often demonstrated by dating now providesthe most compelling link between the two phenomena.Thus, four mass extinctions have occurred in the 300 mil-lion year interval between the Permian and the present day,together with a similar number of minor biotic crises. All ofthese crises coincide with LIP eruptions (Table 1). No otherphenomenon shows such a 100% correlation; certainly notmeteorite impact. However, not all LIP eruptions in thisinterval coincide with extinctions. For example, the espe-cially large Paraná–Etendeka Province was erupted 133 Maago, in the early Cretaceous, at a time marked by extremelylow extinction rates. Thus, the claim that all extinctionevents coincide with giant volcanic eruptions is well sub-stantiated, but the converse that all such eruptions coincidewith extinctions is not true.

TOWARDS A KILL MECHANISMIn general, LIP eruptions are associated with some or all ofthe following climatic and environmental effects:• Rapid global warming• Oceanic anoxia or increased oceanic fertilisation or both

• Calcification crises• Mass extinction• A sharp decrease in the δ13C val-ues recorded in limestones; this isusually interpreted as a record ofmethane release from gas hydratereservoirs.

The Karoo–Ferrar eruptions around180 million years ago present all ofthese features, and the SiberianTraps eruptions 250 million yearsago show many of them. The otherLIP eruptions show several features(TABLE 1). Volcanogenic coolinghas also been proposed for severalextinction events, but the evi-dence is insubstantial. The prepon-derance of evidence favours warm-ing. This strongly suggests that

CO2 emissions do all the damage, although in many extinc-tion scenarios this effect is envisaged as a trigger (FIG. 1).Indeed other factors are almost mandatory given the vol-ume of CO2 likely to be released during LIP eruptions (Selfet al. this issue). The amount of CO2 released during theeruption of the largest LIPs is unlikely to have exceeded1013 tonnes of CO2, with the amount released during indi-vidual flow events likely to have been at least two orders ofmagnitude lower (Wignall 2001). Thus, the gas releasedduring a major flow of 1000 km3 (which may have occurredas frequently as every few thousand years) is unlikely tohave greatly exceeded the current anthropogenic CO2

release rate of 25 × 109 tonnes per annum. This modern fluxcomes not even close to recreating the conditions duringthese ancient catastrophes. It is possible that LIP eruptionsare associated with excessively CO2-rich volcanism, reflect-ing a mantle source still rich in volatiles. However, this hasyet to be demonstrated. Alternatively, the volcanism mayserve as a trigger for something else, such as the release ofmethane from clathrates buried at shallow depths beneaththe seafloor. Methane is a much more effective greenhousegas although it is rapidly oxidised to CO2 in the atmos-phere. However, the tell-tale evidence for methane release –the rapid decrease of the 13C/12C ratio recorded in organiccarbon and limestones, known as a negative δ13C anomaly(see glossary) – is seen less frequently than the other evi-dence for warming.

A Volcanic Greenhouse ScenarioIdeas concerning the role of volcanic gases have been devel-oped primarily from events during the end-Permian andEarly Jurassic mass extinctions (e.g. Pálfy and Smith 2000;Wignall 2001). It is proposed that the volcanic extinction


Paul Wignall1

1 School of Earth and EnvironmentUniversity of Leeds, Leeds LS2 9JT, [email protected]

In the past 300 million years, there has been a near-perfect associationbetween extinction events and the eruption of large igneous provinces,but proving the nature of the causal links is far from resolved. The asso-

ciated environmental changes often include global warming and the develop-ment of widespread oxygen-poor conditions in the oceans. This implicates arole for volcanic CO2 emissions, but other perturbations of the global carboncycle, such as release of methane from gas hydrate reservoirs or shut-down ofphotosynthesis in the oceans, are probably required to achieve severe green-house warming. The best links between extinction and eruption are seen inthe interval from 300 to 150 Ma. With the exception of the Deccan Traperuptions (65 Ma), the emplacement of younger volcanic provinces has beengenerally associated with significant environmental changes but little or noincrease in extinction rates above background levels.

KEYWORDS: Extinctions, ocean anoxia, volcanic eruptions

Basaltic scoria cone atthe southern end of the

1783 Lakigigar fissureeruption, SE Iceland.

PHOTO JOHN MACLENNAN

The Link between LargeIgneous Province Eruptionsand Mass Extinctions


mechanism is triggered by the release of CO2 during theeruption of the giant lava flows that form LIPs (FIG. 1). Theresultant increase of atmospheric CO2 levels would haveseveral deleterious consequences for the oceans. For exam-ple, an increase of CO2 concentrations in surface waterscauses a pH decline and thus problems for carbonate-secret-ing organisms (Gattuso and Buddemeier 2000). This isknown as a calcification crisis and is manifest as a declinein carbonate content in many sections. Global warming canalso cause the development of oceanic anoxic events (and

therefore marine extinctions). Theseare intervals of time when largeareas of the oceans and shelf seaswere either oxygen poor or oxygenfree (anoxic). Modern oceans typi-cally have around 5–6 ml of oxygendissolved in a litre of water, but con-ditions become stressful for mostorganisms if values decline below1.0 ml/L, and no metazoan life cansurvive below values of 0.3 ml/L.Low oxygen conditions arerestricted to only a few small areasof modern oceans but they becamemuch more widespread duringoceanic anoxic events due to severalfeedback factors associated withglobal warming. First, warmerwaters hold less dissolved oxygenthan colder waters. Second, theocean’s circulation system is prima-rily driven by the temperature gra-dient between the equator and thepoles, with deep circulation drivenby the generation of cold and densewaters in the polar regions. This sys-tem slows down as polar waterswarm up, thus decreasing the sup-ply of oxygen to the ocean’s deeperwaters. A possible third factor mayrelate to the supply of nutrientsfrom land, which will increase withglobal warming due to increasedrainfall and runoff in a warmer,more humid climate. Increasednutrient flux to the seas will fosterincreased biological productivity,which in turn will decrease oxygenlevels in sea water as the planktonbiomass decays – the same phenom-enon is seen in many modern shelfseas over-supplied with anthro-pogenic “nutrients” such as fertilis-ers and sewage. Evidence ofincreased global runoff during LIPeruptions is substantial andincludes several lines of geochemi-cal evidence, such as an increase inthe trace metals rhenium andosmium (Ravizza and Peucker-

Ehrenbrink 2003). However, many mass extinction eventscoincide with a collapse, not an increase, of primary pro-ductivity (Hallam and Wignall 1997), and this third factormay not be significant until after the mass extinction hasrun its course. Indeed its main significance may be as a vitalnegative feedback loop for drawing down atmospheric CO2

(FIG. 1). As already noted, there is debate as to whether vol-canic CO2 emissions are sufficient on their own to causethese environmental changes; other phenomena such asgas hydrate release may also contribute to increasing green-house gas concentrations.

The volcanic greenhouse scenario is currently a “workinghypothesis” for several marine extinction events, but its rel-evance to contemporaneous terrestrial extinction eventshas not been explored to any great extent. Severe globalwarming will obviously shift climatic belts and presumablyrestrict habitat area for the most cold-adapted communi-ties. However, tropical habitats should benefit rather thansuffer from such changes. Consequently most terrestrialextinction mechanisms focus on other aspects of LIP erup-

Comparison of environmental effects observed duringeruption of large igneous provinces. Dark blue indicates

a clear development of the effect; pale blue denotes effects for whichthere is some evidence, although this may be controversial or less clear-cut; red denotes effects which were clearly not developed at the timeof volcanism; and blank cells indicate a lack of data. Note that only theKaroo–Ferrar volcanism of the Early Jurassic is associated with all of theseeffects, while oceanic anoxia is the only effect to coincide with the ageof every volcanic province.

TABLE 1


tions. Volcanic halogen emissions can poten-tially damage the ozone layer, thus raising thespectre of UV radiation as a contributory causeof extinctions. Mutant and deformed plantspores and pollen in the end-Permian extinc-tion interval may be a sign of this radiationdamage (Visscher et al. 2004); it is certainly evi-dence for extreme environmental stress, but ithas yet to be established if such phenomena area regular feature of LIP eruptions. Acid rainfrom volcanogenic sulphate aerosols is anotherpotentially harmful effect of flood basalt erup-tions, but there is, as yet, little direct evidencefor this.

CASE EXAMPLESVolcanism–extinction scenarios have beendeveloped primarily to explain end-Permianand Early Jurassic extinction events, and it isinteresting to compare these with environmen-tal changes seen during some other LIP erup-tions of the past 300 Myr.

Central Atlantic Magmatic Province(CAMP) (200 Ma)The realisation that the Central AtlanticMagmatic Province was both very large and ofthe right age to be implicated in the end-Triassic mass extinction suggested yet anotherimportant volcanism–extinction link (Marzoliet al. 1999). This extinction event has provedrather difficult to study, primarily because of adearth of complete marine boundary sections.This paucity reflects the extremely low sea levelat this time, which could of course have con-tributed to the marine extinctions (Hallam andWignall 1999). Other changes at this timeinclude a brief warming pulse, carbon isotopeevidence for a significant release of methaneand possibly a calcification crisis (Hautmann2004). Marine anoxia was widespread, but thisseems to have been the case both before andafter the extinctions.

Clearly the CAMP volcanism may have been responsible forthese environmental changes, but there are some problemsassociated with a CAMP–extinction link. First, there is thedetailed timing. The sedimentary record of the NewarkBasin in northeastern USA contains evidence for both ter-restrial extinctions and flood basalt volcanism; however,the first basalt occurs somewhat above the extinction hori-zon (Wignall 2001). Furthermore, it is not at all clear thatthe end of the Triassic was marked by a sudden, single-pulsemass extinction (Hallam 2002). Extinction rates were highthroughout the last few million years of the Triassic, sug-gesting a prolonged crisis that began considerably beforethe CAMP eruptions.

Paraná–Etendeka Province (133 Ma)The Paraná–Etendeka Province was erupted in southernGondwana (SE South America and Namibia) early in theCretaceous (Renne et al. 1992), during the later half of theValanginian Stage. Until recently, it was thought little ofinterest happened in the oceans at this time. However,recent studies have revealed an extremely watered-downversion of the effects seen during the Early Jurassic and end-Permian crises. Thus, a thin, widespread, late Valanginianblack shale has been found in ocean cores, indicating anepisode of oxygen-poor deposition that has been called theWeissert Event (Erba et al. 2004). The calcareous nanno-

plankton fossil record shows a calcification crisis at thesame time, which may reflect an increase in oceanic nutri-ent levels (calcareous nannoplankton are thought to preferlow-nutrient conditions) or acidification of ocean surfacewaters (Erba 2004) or both. This crisis did not cause extinc-tions and in fact proved something of a spur to evolutionbecause planktonic foraminifera begin a Cretaceous-longradiation of new species immediately after the Weissert Event.

Unlike other intervals marked by LIP eruptions, it is unclearif any substantial global temperature changes occurred atthis time. A case can be made for cooling in the last stages ofthe anoxic event, which probably reflects CO2-drawdown.This could be due to organic matter burial during theWeissert Event (Erba 2004), rather than be of volcanic ori-gin. There is no negative carbon isotope anomaly associatedwith this event, and so methane release from gas hydrate isnot likely.

VOLCANIVOLCANIC GREENHOUSE SCENARIO

1. Eruption of large igneous province

2. Release of large volumes of volcanic CO2

3. Global warming 6. Ocean thermohaline circulation decreases

4. Gas hydrate release, methane oxidation to CO2

7. Dissolved oxygen levels decline

9. Calcification crisis in ocean surface waters

5. Negative shift of δ 13C

8. MARINE MASSEXTINCTION

10. Increased weathering and global runoff

11. Increased nutrient flux and ocean productivity levels

12. Increase of Re and Osfluxes to oceans

13. Increased burial of marine organicC, drawdown ofatmospheric CO2

Flow chart showing suggested chain of environmentalevents caused by the eruption of large igneous provinces.

The chart is colour coded to distinguish between observed facts (yel-low), inferred consequences for which there is overwhelming geolog-ical evidence (purple) and other, somewhat more tentatively inferredconsequences for which there is not always supporting evidence(red).

FIGURE 1

Ontong Java Plateau (120 Ma)A substantial part of the vast Ontong Java Plateau of the SWPacific was erupted around the Barremian–Aptian boundaryof the Early Cretaceous (Courtillot and Renne 2003).Eruption of this great volume of oceanic volcanic rocksslightly predated the oceanic anoxic event, known as theSelli Event, at 119 Ma (Larson and Erba 1999). However,just prior to this event there was a “nannoconid crisis” dur-ing which these very small, calcareous plankton suddenlybecame rather rare. This could reflect a calcification crisis,due to volcanogenic CO2 input, or a fertilisation crisis thatdid not favour the low nutrient–adapted nannoconids (Erba2004). Erba further suggests that direct warming of theocean water by the lava pile may have contributed to thebreak-down of ocean stratification and expansion of theoxygen-minimum zone. Both the warming and the fertili-sation may have contributed to the anoxic event, but thiseffect was curiously delayed relative to the eruptions. There-establishment of normal oceanic conditions after theSelli Event saw the reappearance of the missing nanno-conids. Thus, the crisis was only temporary and not anextinction event.

Caribbean–Colombian Plateau (90 Ma)The 133 and 119 Ma oceanic anoxic events appear to haveprecipitated only minor extinction crises when they arecompared to the enormous losses of earlier mass extinc-tions. The next volcanism–anoxia event in the Cretaceousoccurred at the Cenomanian–Turonian boundary, and thistime it did coincide with rather more extinctions, notablyof several planktonic foraminifera species (Wan et al. 2003).This interval also marks the culmination of Cretaceousgreenhouse warming and sea level rise. It thus has many ofthe hallmarks of other volcanogenic crises, although thereis only weak evidence for a calcification crisis and no evi-dence for methane release. This 90 Ma event coincided withthe eruption of a LIP in the Caribbean–Colombian regionand 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 fromat this time.

Proposed kill mechanisms for the Cenomanian–Turonianextinctions include poisoning by trace metals derived fromoceanic volcanism (Erba 2004), but this proposition israther difficult to test. The anoxic event itself provides themost obvious cause of the marine extinctions, and the con-tribution of volcanism to global warming and fertilisationof the oceans provides a justification for linking volcanismand anoxia (Sinton and Duncan 1997). Furthermore, theoceanic volcanism may also have caused additional warm-ing effects in addition to the direct input of volcanogenicCO2 to the atmosphere. Warming of the oceans by the lavasand oceanic acidification (by volcanic SO2 release) wouldboth have released CO2 to the atmosphere, thus exacerbat-ing a warming trend (Kerr this issue).

The Deccan Traps (65 Ma)Evaluating the global environmental influence of theDeccan Trap eruptions in India is problematic due to thedifficulty of disentangling the effects of the well-known,coeval Chicxulub impact event. However, thanks to pro-longed and intensive study, the detailed chronology of vol-canism and climate change in the Maastrichtian Stage, dur-ing the lead up to the end-Cretaceous mass extinction, arenow established with some clarity. The mid-Maastrichtianwas rather a cool interval, but a rapid phase of warmingbegan around 400 kyr before the K–T boundary(Abramovich and Keller 2003). This was reversed by a rapidcooling trend around 100 kyr before the boundary, when

the 4–5°C temperature gain was lost. The cooling coincideswith a sharp sea level fall, and a lowstand was reachedshortly before the K–T boundary. Thereafter, sea level beganrising again across the boundary (Hallam and Wignall1999).

These substantial oscillations in climate and sea level didnot cause much in the way of extinctions. Keller (2003) hasshown that the latest Maastrichtian warming pulse wasassociated with a destabilisation of planktonic foraminiferalpopulations and short-lived blooms of stress-tolerantspecies. According to Keller these may reflect the expansionand intensification of the mid-water oxygen-minimumzone. However, interesting though they are, these changescannot compare with the near-total and abrupt massextinction of planktonic foraminifera (and various othergroups) at the end of the Cretaceous.

The possibility that the Deccan Trap eruptions were impli-cated in some or all of these changes has of course beenknown for some time. However, only recently has it beenappreciated that the main eruptive phase coincided withthe late Maastrichtian warm pulse (Ravizza and Peucker-Ehrenbrink 2003). The release of volcanic CO2 is the mostlikely driver of environmental change, with a calcificationcrisis in the oceans and global warming of the order of 4°Cthe most direct consequences. Thus, like the other LIP erup-tions of the Cretaceous, the Deccan Trap eruptions appearto have caused significant climatic effects, but only modestbiotic effects, perhaps because the oceans did not becomeanoxic. It has been argued that the biosphere was alreadyrather stressed at the moment of meteorite impact, butwithout that impact one suspects the end-Maastrichtianevent would only have ranked alongside minor Cretaceouscrises 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 (P–E) boundary, have received ample study and arereasonably well understood. Thus, a sharp negative δ13Cexcursion is generally taken as the signature of gas hydraterelease, which in turn is held responsible for the brief (120kyr) warming pulse at this boundary (Kennett and Stott1991). Contemporary changes in the oceans include a cal-cification crisis and the development of oxygen-poor deepwaters, which caused the extinction of many of the speciesliving there. However, this was not a time of mass extinc-tion by any stretch of the imagination. In fact extinctionrates at this time were some of the lowest ever recorded.

These climatic and oceanic changes are very similar to thechanges observed during Cretaceous LIP eruptions, and inthis case they may relate to the eruption of the NorthAtlantic Igneous Province. However, this province seems tohave been formed in two discrete pulses, with the youngerpulse coinciding with the Palaeocene–Eocene thermal max-imum at 55 Ma, and the older eruptive phase coincidingwith a rather cool interval (Courtillot and Renne 2003). Ina recent study of three marine P–E boundary sections,Schmitz et al. (2004) noted that the thermal maximumcoincides 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 pro-duced cooling rather than the observed warming.

FINAL THOUGHTSPerhaps the most intriguing question arising from the linkbetween LIPs and environmental changes concerns theremarkably different magnitudes of the supposed vol-canogenic effects. Thus, the Early Jurassic climatic and envi-



ronmental changes are closely comparable to those pro-posed for the end-Permian crisis. Very similar changes alsooccurred during the Palaeocene–Eocene thermal maximum,but a mass extinction event has not been recorded.However, this last event was of much briefer duration andmay not have lasted long enough to wreak the devastationof the earlier events.

In summary, large igneous province eruptions can causechanges that range from interesting but benign(Palaeocene–Eocene boundary), to severely damaging (EarlyJurassic), to utterly catastrophic (end-Permian). A partialsolution to this problem of variable influence may be foundin modelling work. For example, Dessert et al. (2001) havesuggested that factors such as pre-eruption atmosphericCO2 levels and the rate of eruption are key variables in anyclimatic changes. The closest correspondence betweeneruptions and extinctions coincides with the Pangeanworld, when most of the continents were part of a singlesupercontinent. It may be that such a configuration was lessable to cope with sudden influxes of CO2 into the atmos-phere because chemical weathering (the main mechanismof CO2 drawdown over geological timescales) would havebeen more limited in the arid interior of such a vastcontinent.

ACKNOWLEDGMENTSThis paper has benefited from the comments of Gerta Kellerand Andrew Kerr. .

REFERENCESAbramovich S, Keller G (2003) Planktonic

foraminiferal response to the latestMaastrichtian abrupt warm event: a casestudy from South Atlantic DSDP Site525A. Marine Micropaleontology 48: 225-249

Courtillot V, Renne PR (2003) On the agesof flood basalt events. Compte Rendus –Académie des sciences 335: 113-140

Dessert C, Dupré B, Francois LM, Schott J,Gaillardet J, Chakrapani G, Bajpai S(2001) Erosion of Deccan Traps deter-mined by river geochemistry: impact onthe global climate and the 87Sr/86Sr ratioof seawater. Earth and Planetary ScienceLetters 188: 459-474

Erba E (2004) Calcareous nannofossils andMesozoic oceanic anoxic events. MarineMicropaleontology 52: 85-106

Erba E, Bartolini A, Larson RL (2004)Valanginian Weissert oceanic anoxicevent. Geology 32: 149-152

Gattuso J-P, Buddemeier RW (2000) Oceanbiogeochemistry: calcification and CO2.Nature 407: 311-313

Hallam A (2002) How catastrophic was theend-Triassic mass extinction? Lethaia 35:147-157

Hallam A, Wignall PB (1997) MassExtinctions and their Aftermath. OxfordUniversity Press, Oxford, 320 pp

Hallam A, Wignall PB (1999) Massextinctions and sea-level change. Earth-Science Reviews 48: 217-250

Hautmann M (2004) Effect of end-TriassicCO2 maximum on carbonate sedimenta-tion and marine mass extinction. Facies50: 257-261

Keller G (2003) Biotic effects of impactsand volcanism. Earth and PlanetaryScience Letters 215: 249-264

Kennett JP, Stott LD (1991) Abrupt deep-sea warming, palaeoceanographicchanges and benthic extinctions at theend of the Palaeocene. Nature 353: 225-229

Kerr AC (1998) Oceanic plateau formation:a cause of mass extinction and blackshale deposition around the Cenomanian-Turonian boundary? Journal of theGeological Society of London 155: 619-626

Larson RL, Erba E (1999) Onset of the Mid-Cretaceous greenhouse in the Barremian-Aptian: Igneous events and the biolog-ical, sedimentary, and geochemicalresponse. Paleoceanography 14: 663-678

Marzoli A, Renne PR, Piccirillo EM, ErnestoM, Bellieni G, DeMin A (1999) Extensive200-million-year-old continental floodbasalts of the central Atlantic magmaticprovince. Science 284: 616-618

Pálfy J, Smith PL (2000) Synchronybetween Early Jurassic extinction, oceanicanoxic event, and the Karoo-Ferrar floodbasalt volcanism. Geology 28: 747-750

Ravizza G, Peucker-Ehrenbrink B (2003)Chemostratigraphic evidence of Deccanvolcanism from the marine osmiumisotope record. Science 302: 1392-1395

Renne PR, Ernesto M, Pacca IG, Coe RS,Glen JM, Prévot M, Perrin M (1992)The age of Parana flood volcanism,rifting of Gondwanaland, and theJurassic-Cretaceous boundary. Science258: 975-979

Schmitz B, Peucker-Ehrenbrink B,Heilmann-Clausen C, Åberg G, Asaro F,Lee C-TA (2004) Basaltic explosive

volcanism, but no comet impact, atthe Paleocene–Eocene boundary: high-resolution chemical and isotopic recordsfrom Egypt, Spain and Denmark. Earthand Planetary Science Letters 225: 1-17

Sinton CW, Duncan RA (1997) Potentiallinks between ocean plateau volcanismand global ocean anoxia at theCenomanian-Turonian boundary.Economic Geology 92: 836-842

Visscher H, Looy CV, Collinson ME,Brinkhuis H, van Konijnenburg-vanCittert JHA, Kürschner WM, Sephton MA(2004) Environmental mutagenesisduring the end-Permian ecological crisis.Proceedings of the National Academy ofSciences 101: 12952-12956

Wan X-Q, Wignall PB, Zhao W-J (2003)The Cenomanian-Turonian extinctionand oceanic anoxic event: evidence fromsouthern Tibet. Palaeogeography Palaeo-climatology Palaeoecology 199: 283-298

White RV, Saunders AD (2005) Volcanism,impact and mass extinctions: incredibleor credible coincidences. Lithos 79: 299-316

Wignall PB (2001) Large igneous provincesand mass extinctions. Earth-ScienceReviews 53: 1-33 .

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 equip-

ment, thin sections and mineral specimens for its descrip-

tive mineralogical laboratory and educational programs.

We are dedicated to classical mineralogical research,

preservation of mineral specimens, and educational out-

reach to primary and secondary school teachers and stu-

dents. If your institution is upgrading its analytical equip-

ment, we want your used, working devices. Further, if you

are disposing of minerals, thin sections or similar geologi-

cal artifacts, let us put them to good use; aesthetics are

unimportant, labels are! Please contact:

The Hudson Institute of MineralogyPO Box 2012 • Peekskill, NY 10566-2012

www.hudsonmineralogy.org

large igneous provinces: origin and environmental...

Documents