海洋底ダイナミクス 2018 Ocean Floor Geodynamics 2018

13. プレート内火成活動：海山と海台 Intraplate volcanism: seamounts and plateaus

プレート内火成活動 Intraplate volcanism Type1 古典的なホットスポット：不動か？“Classic” type hot spot Type2 スーパースウェル Super swell

Type3 リソスフェアのクラック＋プチスポット Lithosphere cracks + petit spot

巨大火成岩岩石区 Large Igneous Provinces (LIPs) 巨大海台 Oceanic plateaus 環境・地球史への影響 Environmental impact

1

地球の火成活動（現世） volcanism on earth

(Press, Understanding Earth, 2003

2 FEATURE | FOCUS

Mantle plumes persevere

Anthony A. P. Koppers

The ocean floor is littered with hundreds of thousands of mostly extinct volcanoes. The origin of at least some of these seamounts seems to rest with mantle plumes.

recent census suggests that Morgan’s mantle plumes conduits are difcult to resolve using seamounts1 — typically extinct Some hotspots form seamount trails seismic data4,5, their existence has been A underwater volcanoes — are along the surface of tectonic plates, far difcult to confrm. Many question whether numerous. It has been estimated that away from their volcanically active plate all hotspot volcanism is formed by mantle about 125,000 seamounts with a height boundaries. Forty years ago, W. Jason plumes6,7 and some doubt whether mantle of more than one kilometre exist on our Morgan introduced the concept of mantle plumes exist at all8,9. ocean foors. Most of these are postulated plumes to explain this kind of hotspot Perhaps the most captivating aspect to form at volcanic hotspots that are the volcanism2,3. According to his theory, of Morgan’s plume model is that he could surface expressions of mantle plumes plumes of hot material upwell from the explain the formation of the Hawaiian– — hot material upwelling from Earth’s deep mantle. During ascent and on impact Emperor and three other seamount trails interior. Yet, many seamounts do not show with the overlying tectonic plates, these along the Pacifc Ocean foor. Tese three the typical characteristics expected for plumes drive melting and the production trails track each other in such a way that volcanoes that have formed above a mantle of magma, which erupts to form volcanic they can be explained by the rotation of plume. So, debate about the feasibility of seamounts at the plate surface. Morgan a rigid Pacifc plate that is drifing over the mantle plume hypothesis is ongoing. proposed that the migration of tectonic four plumes fxed in the mantle (Fig. 1a). Te most straightforward explanation is plates over stationary and long-lived With this observation, Morgan supplied that not all hotspot volcanoes are alike, mantle plumes would generate chains of compelling support for the existence of and that some groups of seamounts are volcanoes on the ocean foor. However, plumes and also provided independent better explained by mechanisms other than because mantle plumes themselves cannot proof for the motion of tectonic plates mantle plumes.intraplate volcanism be directly sampled and the thin plume relative to the underlying mantle. Te mantle plume model also opened up プレート内火成活動についての最近の認識 potential new avenues of research into the Earth’s deepest regions10. Specifcally, Hotspot if the volcanic seamounts are the surface type 2 expression of a mantle plume, their Mid-ocean erupted lavas could potentially preserve a Hotspot spreading Hotspot type 3 centre type 1 record of long-lived variations in mantle composition and could provide insights into mantle convection.

Holes in the theory

67 Te plume model calls on an extensive 0 km global network of long-lived stationary mantle plumes that are continually delivering hot material from deep in the Earth. However, such a global network never fully materialized. It now seems that there aren’t many active hotspot systems CMB around the world, maybe a few dozen — too few to have produced all of the (Koppers, 2011 world’s seamounts. Of those seamounts that do seem to have 3 formed above a mantle plume, some show Figure 1 | Models of ocean-island and seamount-trail formation: Courtillot’s framework6 of three evidence that the underlying plume was hotspot types. The first, a classical Morgan-style long-lived mantle plume, originates from as deep neither long-lived nor stationary. Improved in the mantle as the core–mantle boundary (CMB). The second hotspot type includes short-lived, mapping of seamount trails using satellite smaller plumes originating from shallower parts of the mantle, probably as ofshoots from large altimetry reveals that most seamount trails superplumes. These secondary hotspots are more common. The third type of hotspot is not related to have typical life spans of just 30 million any kind of mantle ウィルソンの定義したホットスポットplume and may form where the oceanic lithosphere cracks or extends. This kind is years11. And samples of lava collected from the least investigateddefinition and may overlap of hotspot considerably by with Wilson the other hotspot(1963) types. the Emperor seamounts during the Deep

816 NATURE GEOSCIENCE | VOL 4 | DECEMBER 2011 | www.nature.com/naturegeoscience

© 2011 Macmillan Publishers Limited. All rights reserved

(Wilson, 1963

relatively small, long-lasting, and hot regions -- called hotspots -- must exist below the plates that provide localized sources of high heat energy (thermal plumes) to sustain volcanism.

4 固定ホットスポット仮説と海山列 fixed hotspot hypothesis and hotspot track

• Hotspots are fixed in deep mantle = no relative motion among hotspots

• Seamounts chain extending from hotspot is a kind of “track” of plate motion.

• support ‘plate tectonics’

• provide past plate motions

(Understanding Earth, 2003)

5

ハワイー天皇海山列 Hawaii hotspot and Emperor seamount chain

(Press, Understanding Earth, 2003

6 ホットスポットの分布 Global distribution of hotspot

Based on UTIG hotspot list 7

ホットスポット火成活動の特徴 Hotspot volcanism : Features

• Courtillot et al.(2003)’s five criteria

• long-lived tracks

• traps at their initiation

• magma flux > 103 kg/s

• High 3He/4He or 21Ne/22Ne

• anomalously low shear velocities (Vs) in the mantle below

8 noticeable for the Hawaiian-Emperor seamount trail (e.g., Watts, 1976) and is believed to be directly related to the buoyancy of a plume and its interac- FLEXURAL tion with the overlying Pacifc Plate MOAT (Figure 1). Mapping of these swells using satellite-derived gravity and geoid data, for instance, allowed scientists to equate the sizes of these swells to vertical plume fuxes. Although the MID-PLATE volume of active intraplate volcanism is SWELL small compared to island arc volcanism and the formation of the oceanic crust 0 Myr 1.4 Myr 4.2 Myr 7.0 Myr 15.215 2 MyMyrr 23.623 6 MyMyrr ホットスポット軌跡 at the mid-ocean ridges, plume fuxes 0 1 ranging from 1.0 Mg s-1 (Canary) to track and mantle plume Thermal Anomaly 8.7 Mg s-1 (Hawai`i) become signifcant when integrated over geological time and including all known hotspot systems (Davies, 1988; Sleep, 1990). Further observations that support the presence of mantle plumes include evidence that these lithospheric swells diminish away from active hotspots, the formation of linear age-progressive seamount trails, and the volcanic extinction of seamounts Figure 1. Te Hawaiian-Emperor hotspot trail is our textbook example of the classical mantle (Van Keken, 1997 when plate motions move them away plume model explaining the formation of intraplate seamounts. In this map of the Northwest Pacific (D. Sandwell and W.M. Smith: Gravity Anomaly Map based on Satellite Altimetry, from their hotspot locations. Version 15.2), this archetypical seamount trail is exemplified by a deep flexural moat along its But mantle-plume behavior is not entire length and a significant mid-plate swell only prevalent toward the young southeastern end. Te linearity of this seamount trail, in combination with a large mid-plate swell and a quite so simple, as the latest numerical systematic age progression (with radiometric ages increasing toward the older northwestern mantle convection models suggest end; see Figure 2), provides strong evidence for the existence of a mantle plume, maybe origi- that a simple density-driven upwelling nating deep in the mantle from a thermal anomaly (see simulation at the bottom by Van Keken [1997]). In this model, the seamount trail only forms after the plume head has dissipated and (Figure 1) is very unusual (but not the narrow plume stem starts interacting with the lithosphere. Because the older Emperor implausible) and that the resulting seamounts have all been subducted into the Aleutian trench to the north, the fate of the plume plumes mostly are not vertically straight, head and any link to large igneous province volcanism are unidentified. narrow, and continuous, but ofen

(more or less linear) seamount trails that 1500-m high, and correlating with long- Anthony A.P. Koppers (akoppers@coas. form as the plates constantly move over wavelength gravity and geoid anomalies) oregonstate.edu) is Associate Professor, the “fxed” loci of the upwelling mantle have been found at the leading edges of College of Oceanic and Atmospheric plume stems (Richards et al., 1989). many active seamount trails. Te correla- Sciences, Oregon State University, Corvallis, Many observations are consistent tion implies that the swells are supported OR, USA. Anthony B. Watts is Professor with the existence of mantle(Press, plumes. Understanding at depth Earth, by 2003low-density subcrustal of Marine Geology and Geophysics, Large mid-plate topographic swells mantle material. Tis large-scale warping Department of Earth Sciences, University (typically 1500–3000-km wide, up to of otherwise rigid lithosphere is most of Oxford, Oxford, UK. 9

44 Oceanography Vol.23, No.1

洪水玄武岩 Flood basalt

玉木、岩波地球惑星科学

10 地球化学的な特徴 Geochemical (isotopic) features

typical values for R/Ra MORB 8+-2 OIB 5~42 Continent << 1

primordial isotope alpha decay of U and Th accumulate over time Anderson et al. http://www.mantleplumes.org/HeliumFundamentals.html

11

マントルプルームの深部構造 Deep structure of mantle plume

•Dense network of Ocean Bottom Seismometers

•low-velocity zone extends to lower mantle = mantle plume from D”? (Wolf et al., 2009)

12 V. Courtillot et al. / Earth and Planetary Science Letters 205 (2003) 295^308 297 Type 1

Table 1 Courtillot’s primaryseven Scores for 49 hotspots with respect to ¢ve criteria used to diagnose a potentially deep origin (see text) Hotspot Lat Lon Track Flood/plateau Age Buoy. Reliab. 3He/4He Tomo Count (‡E) (Ma) (500) Afar 10N 43 no Ethiopia 30 1 good high slow 4 Hawaii Ascension 8S 346 no no / na na na 0 0+? Louisville Australia E 38S 143 yes no / 0.9 fair na 0 1+? Azores 39N 332 no? no / 1.1 fair high? 0 1+? Baja/Guadalupe 27N 247 yes? no / 0.3 poor low 0 0+? Balleny 67S 163 no no / na na na 0 0+? Bermuda 33N 293 no no? / 1.1 good na 0 0+? Bouvet 54S 2 no no / 0.4 fair high 0 1+?

Easter Bowie 53N 225 yes no / 0.3 poor na slow 2+? Cameroon 4N 9 yes? no / na na na 0 0+? Canary 28N 340 no no / 1 fair low slow 2 Cape Verde 14N 340 no no / 1.6 poor high 0 2 Caroline 5N 164 yes no / 2 poor high 0 3 Comores 12S 43 no no / na na na 0 0+? Crozet/Pr. Edward 45S 50 yes? Karoo? 183 0.5 good na 0 0+? Darfur 13N 24 yes? no / na poor na 0 0+? Discovery 42S 0 no? no / 0.5 poor high 0 1+? Easter 27S 250 yes mid-Pac mnt? 100? 3 fair high slow 4+? Eifel 50N 7 yes? no / na na na 0 0+? Fernando 4S 328 yes? CAMP? 201? 0.5 poor na 0 0+? Galapagos 0 268 yes? Carribean? 90 1 fair high 0 2+? Great Meteor/New England 28N 328 yes? no? / 0.5 poor na 0 0+? Hawaii 20N 204 yes subducted? s 80? 8.7 good high slow 4+? Tristan

Iceland Hoggar 23N 6 no No / 0.9 poor na slow 1 14 13 Iceland 65N 340 yes? Greenland 61 1.4 good high slow 4+? Jan Mayen 71N 352 no? yes? / na poor na slow 1+? Juan de Fuca/Cobb 46N 230 yes no / 0.3 fair na slow 2+? Juan Fernandez 34S 277 yes? no / 1.6 poor high 0 2+? Kerguelen(Heard) 49S 69 yes Rajmahal? 118 0.5 poor high 0 2+? Louisville 51S 219 yes Ontong-Java 122 0.9 poor na slow 3+? Lord Howe (Tasman East) 33S 159 yes? no / 0.9 poor na slow 1+? Afar Macdonald (Cook-Austral) 30S 220 yes? yes? / 3.3 fair high? slow 2+? Marion 47S 38 yes Madagascar? 88 na na na 0 1+?

Reunion Marqueses 10S 222 yes Shatski? ??? 3.3 na low 0 2+? Martin/Trindade 20S 331 yes? no / 0.5 poor na fast 0+? Meteor 52S 1 yes? no / 0.5 poor na 0 0+? Pitcairn 26S 230 yes no / 3.3 fair high? 0 2+? Raton 37N 256 yes? no / na na na slow 1+? Reunion 21S 56 yes Deccan 65 1.9 poor high 0 4 St Helena 17S 340 yes no / 0.5 poor low 0 1 Samoa 14S 190 yes no? 14? 1.6 poor high slow 4 San Felix 26S 280 yes? no / 1.6 poor na 0 1+? Socorro 19N 249 no no / na poor na slow 1+? Tahiti/Society 18S 210 yes no / 3.3 fair high? 0 2+? Tasmanid (Tasman central) 39S 156 yes no / 0.9 poor na slow 2 Tibesti 21N 17 yes? no / na poor na 0 0+? Tristan 37S 348 yes Parana 133 1.7 poor low 0 3 Vema 33S 4 yes? yes? (Orange R.) / na poor na 0 0+? Yellowstone 44N 249 yes? Columbia? 16 1.5 fair high 0 2+?

EPSL 6470 3-1-03 Cyaan Magenta Geel Zwart articles

Obviously, it is also possible to construct models with substantially the Pacific rim. The relative motions of the two groups can be larger hotspot motion (for example, with lower viscosity, and/or determined from observations of the sea floor in the South Pacific larger density anomalies in the mantle15); however, such models are and southern Indian Ocean, with the caveat that deformation is not consistent with observations,RESEARCH such as hotspot tracks, whereas knownARTICLES to have occurred in continental regions of Antarctica and the models just summarized are broadly consistent. The directions New Zealand, but this is hard to quantify. of hotspot motion are less sensitive to model differences than are the For times younger than chronEmperor 20 (43 chain Myr should ago), match substantial the present- speeds of motion. Australia–Pacific plate motion (throughday latitude New of Zealand) Hawaii (ϳ cannot19°N) if be the hot- The Emperor Seamounts: Southward spot has remained fixed with respect to determined with sufficient precisionEarth’s from spin local axis. dataThe most for reliable it to be indica- Relative plate motions and hotspotMotion tracks of the Hawaiianuseful in Hotspot this analysis. However, seafloortors of paleolatitude spreading are in basaltic the South rocks, but ホットスポットは移動するのか？We now account for predicted hotspot motion and compute hot- Pacific is accurately quantified18, andtheir motion reliability between depends East on and each West section spot tracks from specified relative plate motions.Plume We in cannot Earth’s give Antarctica Mantle has been determined fromspanning a consideration enough time to of sample seafloor geomag- formal uncertainties, although on the basis of the spread of model geometry19. The motion of Africa,netic relative secular to East variation. Antarctica Recovery since of such Does hotspot move?— 1 2 3 results as just discussed we can judgeJohn whether A. Tarduno, a discrepancy* Robert seems A. Duncan,the LateDavid Cretaceous, W. Scholl, is determinedsamples from sea requires floor ocean-drilling in the Southwest technology, — 1 4 and only a few seamounts have been sam- to be significant. The global tectonics of theRory Earth D. is Cottrell, characterizedBernhard by Indian Steinberger, Ocean20. Hence the motion of Africa relative to the Pacific Thorvaldur Thordarson,5 Bryan C. Kerr,3 Clive R. Neal,6 pled to date. a group of plates that diverge from the Indian and Atlantic oceans, plate is determined by a plate motion chain running from Africa Hawaiian hotspot track is predictedFred A. Frey,from7 Masayukia global Torii, plate8 Claire motion Carvallo chain9 based Paleomagnetic analyses of 81-million- and a Pacific group of oceanic plates that converge around most of through East Antarctica and WestAntarcticayear-old basalt to the recovered Pacific (Figsfrom Detroit 3, 4). Sea- on relative plate motion data, and it is assumed that the Hawaiian hotspot is mount (Site 884) yielded a paleolatitude of fixe relative to African hotspots,The Hawaiian-Emperor it does hotspotnot fit trackthe hasobserved a prominent track bend, which has ϳ36°N (10), which is discordant with served as the basis for the theory that the Hawaiian hotspot, fixed in the Hawaii. Data from ϳ61-million-year-old deep mantle, traced a change in plate motion. However, paleomagnetic and basalt (9)fromSuikoSeamountdefinea radiometric age data from samples recovered by ocean drilling define an paleolatitude of 27°N (11). These data sug- age-progressive paleolatitude history, indicating that the Emperor Sea- gest that the Emperor Seamounts record mount trend was principally formed by the rapid motion (over 40 milli- southward motion of the hotspot plume in meters per year)R ESEARCH of the HawaiianA RTICLES hotspot plume during Late Cretaceous to the mantle (10). early-Tertiary times (81 to 47 million years ago). Evidence for motion of the A paleomagnetic test. The Ocean Dril- have been less frequent at Site 884 relative to calculation (25) requires a mantle density and Hawaiiana large-scale plume upwelling. affectsmodels Thus, in of these mantle mod- convection and plate tectonics, ling Program (ODP) Leg 197 (12)soughttotest the other sites because of its flank position. viscosity model and a surface-velocity changingels, fast hotspot our understanding motion corresponds of terrestrial to slower dynamics. the hypothesis of southward motion of the Ha- The Site 1204 (Hole B) lavas record a low boundary condition (13). A tomography mantle flow rates (ϳ10 to 20 mm yearϪ1). waiian hotspot by drilling additional basement angular dispersion, but these rocks might car- model was used to infer mantle density vari-The conceptMost computationsof an age-progressive (25) yield set a hotspot of define mo- the directional change at 43 million sites in the Emperor chain (Fig. 1). We collect- R ESEARCH A RTICLES ry a CRM, explaining the agreement of their ations (26), and a viscosity structure basedvolcanic on tion islands, of 5° atolls,to 10° andtoward seamounts the south pro- to southeastyears ago (Ma) (7)thatwouldbeexpected ed detailed stepwise alternating field (AF) de- mean inclination with that of the Site 884 an optimized fit to the geoid (with additionalduced byduring a hotspot the past plume 100 fixed million in the years deep (My).ifduction It such is a of large volcanic change edifices in plate older motionthan the hadspectmagnetization to the spin axis data since aboard the Early the Creta- drillingand ship more gradual than previously thought. basalt section and the Site 1203 sediments. constraints from heat flow) (27) was applied.mantle waspossible first to developed achieve a to good explain fit to the the paleo-occurred.oldest extant There seamount, was also Meiji a Guyot) general gener- lack ofceousJOIDES Epoch. Similarly, Resolution some.Althoughtheseshipboard changes in the Given the central role the Hawaiian-Emperor Because of potential inclination shallowing, Both moving- and fixed-plume sourcesHawaiian (28, magnetic Islands data, (1). The because bend the separating age of the initia-circum-Pacifically yield the best tectonic fits. For the events fixed-source (8)docu-morphologydata are of of the high geomagnetic resolution, field they with alonebend are in- has played as an example of plate mo- the mean inclination from the Site 1203 sed- 29) that originate at the top of the low-the westward-trendingtion of the Hawaii Hawaiian hotspot is island unknownmented (andmodel, the for computed this time. hotspot Recent motion age consists data sug-time (sufficient34)thathavereliedonfixedhotspots to define paleolatitudes. Magnetiction change, these observations now raise the of two distinct phases. During the first phase, to anchor data from global sites are proba- question of whether major plates can undergo iments should be a minimum. This suggests viscosity layer at the base of the mantle werechain fromcan the hence northward-trending be used as a free Emper- parameter).gestwhich For a lastsslightly 100 to older 150 My, age southward for the bend motion [ϳ47bly artificial.minerals One with recent intermediate analysis to that high has coercivities,large changes in direction rapidly, and wheth- further that the mean derived from the basalts considered (13). or Seamountsthe moving-source has most often model, been southward inter- motionMacan (be9)], rapid. but The this second revised phase timing begins still when doesnot reliedcarrying on magnetizations the fixed-hotspot resistant reference to AF demag-er plate boundary forces alone can play a at the same site is shallower because the Fast motion occurs when a conduitpreted is astends an example to be faster of if a an change earlier in plume plate originnotthe is firstcorrespond conduit elements to an episode that arise of from profound the framenetization, has called are for commonly a significant formed axial duringdominant sub- role in controlling plate motion. 3 available lavas underrepresent higher inclina- sheared and tilted in the large-scale flowmotion and recordedassumed. in However, a fixed-hotspot plume frame initiation of plate agesfixed source motion reach change the surface. recorded The computed within theoctopoleaerial contribution or seafloor ( weathering.g 0)tothetime- The magnetiza-The similarity of the Hawaiian and Lou- tion values. the tilted conduit rises to the surface aidedreference by from (2). 180 to 120 Ma (which imply thePacifichotspot sub- basin motion or is on slow its during margins. the second averagedtions field of these (35). mineral This conclusion phases are is con- easily resolv-isville hotspot tracks implies that the motion phase and, in the example shown (Fig. 4), troversial, but if correct, it would imply we are tracking by the new paleomagnetic We consider two scenarios: one in which However, global plate circuits suggest One approachSteinberger to examine hotspot et al.(2004) fixity able in thermal demagnetization data, which we the paleomagnetic results from the Site 884 large relative motions between Hawaii and issomewhat to determine toward the the age north and (13 paleolatitude). ofthat ouralso paleolatitude discuss here calculations (13). underes- data is of large scale. This Late Cretaceous to Overall, the results of the large-scale flow timate the true hotspot motion. early-Tertiary episode of hotspot motion was basalts, 1204 (Hole B) basalts, and Site 1203 hotspots in the Atlantic and Indian Oceans volcanoes that form a given hotspot track. The geomagnetic field at a radius r, co- 15 modeling approach described above are con- Backtracking the position of early-Tertiary not isolated; motion of the Atlantic hotspots sediments best represent the field (paleolati- (3– 6). Improved mapping of marine mag- Forsistent the with Hawaiian the Leg 197 hotspot, paleomagnetic the paleolati- data. and olderlatitude Pacific␪, basin and sites, longitude an essential␾ can aspect be describedrelative to those in the Pacific occurred at ϩ12.4° netic anomalies in the Pacific has failed to tudes of extinct volcanic edifices of the by the gradient of the scalar potential (⌽): tude model A; IT ϭ 56.5° Ϫ12.6°, N ϭ 3) and Potentially important differences lie in the of some paleoclimate and tectonic studies, re- similar rates during mid-Cretaceous times another in which we combine all the individ- total motion predicted since ϳ80 Ma (13) and quires rethought, given that previous efforts (39). These data sets indicate a much more ual basalt inclination units from Detroit Sea- in the need to incorporate in the modeling have also relied on fixed hotspots. The north- active role of mantle convection in control- results a change in plate motion at or near the erly position of the Late Cretaceous Hawaiian ling the distribution of volcanic islands. At mount into a mean (Model B; I ϭ 52.9° 1 Fig. 1. Hawaiian-Emperor chain shown T Department of Earth and Environmental Sciences, time of the bend. The paleomagnetic data do hotspot (23)castsdoubtonthesouthernoption times, it is this large-scale mantle convection ϩ3.7°, N ϭ 32). With either model, the paleo- with ODP Leg 197 sites (12) and marine Ϫ6.9° University of Rochester, Rochester, NY 14627, USA. not require a change in plate motion, al- for the Kula-Farallon ridge [a plate configura- that is the principal signal recorded by hot- 2 magnetic-anomaly identifications (40). latitude and age data yield average rates College of Oceanic and Atmosphere Science, Oregon though a small change is not excluded. tion that is typically called on to create high spot tracks. (Model A: 57.7 Ϯ 19.2 mm yearϪ1; Model B: State University, Corvallis, OR 97331–5503, USA. 3 Implications of hotspot motion. The rates of northward transport for tectonostrati- Ϫ1 ハワイホットスポットの南進Geophysics Department, Stanford University, Stan- 43.1 Ϯ 22.6 mm year ) that are consistent ford, CA 94305, USA. 4Institute for Frontier Research hotspot motion defined by the new paleomag- graphic terranes in Alaska and British Colum- References and Notes with the hypothesis that the Hawaiian hotspot netic and radiometric age data has implica- bia (36, 37)]. 1. J. T. Wilson, Can. J. Phys. 41, 863 (1963). southward motionon Earth Evolution,of Hawaiian Japan Marine Science plume and Tech- 2. W. J. Morgan, Nature 230, 42 (1971). moved rapidly southward from 81 to 47 Ma nology Center, Yokosuka 237–0061, Japan. 5Depart- tions for a wide variety of issues, including The fixed-hotspot interpretation of the 3. P. Molnar, T. Atwater, Nature 246, 288 (1973). (10). The values are consistent with updated ment of Geology and Geophysics–School of Ocean true polar wander (TPW) (30), the morphol- Hawaiian-Emperor bend implies that huge 4. P. Molnar, J. Stock, Nature 327, 587 (1987). and Earth Science and Technology, University of Ha- ogy of the past geomagnetic field, and the plates can undergo large changes in direction 5. C. A. Raymond, J. M. Stock, S. C. Cande, in The History estimates of hotspot motion based on inde- 6 and Dynamics of Global Plate Motions, vol. 121, Paleomagnetic and waii,radiometric Honolulu, HI 96822, age USA. dataDepartment from of Civil history of plate motions. Some investigators rapidly. But such changes cannot be associ- Geophysical Monograph Series, M. A. Richards, R. G. pendent relative plate motions (5). Both pa- Engineering and Geological Sciences, University of (31) have proposed that as much as 30° of ated with internal buoyancy forces (such as Gordon, R. D. van der Hilst, Eds. (American Geophysi- samples recovered by ocean drilling define an7 leolatitude models suggest that most of the Notre Dame, Notre Dame, IN 46556, USA. Depart- TPW (rotation of the entire solid Earth) has subducting slabs) because these require many cal Union, Washington, DC, 2000), pp. 359–375. 6. V. DiVenere, D. V. Kent, Earth Planet. Sci. Lett. 170, motion occurred before the time of the ment of Earth, Atmospheric and Planetary Sciences, accumulated during the past 200 My. How- millions of years to develop. This has led to age-progressive paleolatitudeMassachusetts Institute history. of Technology, Cambridge, 105 (1999). Hawaiian-Emperor bend (Model A, Ͼ44 Ma; MA 02139, USA. 8Department of Biosphere- ever, a fixed-hotspot reference frame is used the suggestion that plate-boundary forces 7. T. Atwater, in The Eastern Pacific Ocean and Hawaii, to define TPW in these studies. The data might be responsible (38). The new paleolati- vol. N of The Geology of North America, E. L. Win- Model B, Ͼ43 Ma). This is further supported Geosphere System Science, Okayama University of terer, D. M. Hussong, R. W. Decker, Eds. (Geological 9 presented here, together with other tests (32, tude and radiometric age data (9) suggest that by the mean paleolatitude value from Koko Science, Okayama 700–0005, Japan. Department of Society of America, Boulder, CO, 1989), pp. 21–72. Tarduno et al.(2003) 33), indicate that TPW has been overestimat- changes of plate motion at the time of the Seamount (based on both thermoremanent Figure 2 Computed hotspot motionPhysics, and Geophysics tracks for Division, the moving-source University of model. Toronto, Hotspot four tracks for the two plate motion chain models as indicated; parameters are African 8. I. O. Norton, Tectonics 14, 1080 (1995). Mississauga, ON L5L1C6 Canada. ed; Earth has been relatively stable with re- Hawaiian-Emperor bend were much smaller 9. W. D. Sharp, D. A. Clague, Eos 83, F1282 (2002). magnetizations and CRMs), which is only motion is shown as rainbow-coloured lines (right colour bar), tracks as single-coloured rotations 0–47, 47–62 and 62–83 Myr ago. Dotted lines, hotspots assumed fixed; short-10. J. A. Tarduno, R. D. Cottrell, Earth Planet. Sci. Lett. *To whom correspondence should be addressed. E- 153, 171 (1997). 2.5° north of the fixed-hotspot prediction. lines (shown for the past 83 Myr);mail: tickmark [email protected] interval is 10 Myr for both. Hotspot tracks are dashed line, only hotspot motion on African hemisphere considered (shown for Hawaii);11. M. Kono, Init. Rep. Deep Sea Drill. Proj. 55, 737 Modeling of hotspot motion. Crust computed for the motion of the plate where they are located, except for Hawaii, where long-dashed line, only hotspot motion on Pacific hemisphere considered (shown for (1980). ages available from marine magnetic anom- 12. J. A. Tarduno et al., Proceedings of the Ocean Drilling tracks computed for motion of the Pacific plate are plotted beyond its boundary.22 AUGUST Tracks 2003Hawaii); VOL continuous 301 SCIENCE lines, all www.sciencemag.org hotspot motions considered. A least-squares method7 is Program, Initial Report (Ocean Drilling Program, Col- alies, radiometric age data from drill sites, 1064 lege Station, TX, 2002), vol. 197. are plotted regardless of whether a hotspot was actually beneath a given plate at a given used to optimize the fit to locations and radiometric ages (see Fig. 5) of seamounts. Free13. Material and methods are available as supporting and geochemical data (22) indicate that the material on Science Online. 44 Hawaiian hotspot was close to a spreading time. Red lines (model 1, shown for Hawaii and Louisville), and the purple line (model 2, air gravity shows actual hotspot tracks. This, rather than topography, was chosen 14. P. L. McFadden, A. B. Reid, Geophys. J. R. Astron. Soc. 69, 307 (1982). ridge during the formation of Detroit Sea- shown for Hawaii), are computed for absolute plate motions such that the fit to Tristan and because it should better distinguish between actual hotspot tracks (that is, seamounts 15. P. L. McFadden, R. T. Merrill, M. W. McElhinny, S. Lee, mount (23). Hence, asthenospheric channel- Re´union hotspot tracks is optimized for the two plate motion chain models as indicated; formed above a plume, in an intraplate location) and material erupted at a ridge through J. Geophys. Res. 96, 3923 (1991). 16. A. V. Cox, Geophys. J. R. Astron. Soc. 20, 253 (1970). ing of the plume (24) from a position to the optimization parameters are African plate rotations 0–47 and 47–83 Myr ago. Green lines plume–ridge interaction: the former are not expected to be in isostatic equilibrium and17. J. A. Tarduno, Geophys. Res. Lett. 17, 101 (1990). south toward a more northerly ridge could 18. W. A. Berggren, D. V. Kent, C. C. Swisher III, M.-P. (shown for Re´union and Tristan) are for optimizing the fit to Hawaii and Louisville for plate should therefore be readily visible on a gravity map.Fig. The4. (A latter) Computed is expected Hawaiian to be hot- Aubry, Soc. Econ. Paleontol. Mineral. Spec. Publ. 54 have played some role in the difference spot motion for fixed-source model (1995), pp. 129–212. between the paleomagnetic data and the motion chain model 1; parameters are Pacific plate rotations 0–25, 25–47, 47–62 and approximately in isostatic equilibrium, hence without(colored strong line), gravity and trackssignature. for fixed- 19. R. A. Fisher, Proc. R. Soc. London Ser. A 217, 295 62–83 Myr ago. Black (model 1) and blue (model 2) lines are for optimizing jointly to all source models (continuous line; (1953). prediction of the fixed-hotspot model. The plume initiation at 160 Ma) and 20. G. B. Dalrymple, M. A. Lanphere, D. A. Clague, Init. monotonic age progression of lavas recov- moving-source models (dashed line; Rep. Deep Sea Drill. Proj. 55, 659 (1980). NATURE|VOL 430|8 JULY 2004|www.nature.com/nature plume initiation at 170 Ma). Tickmark16921. J. McKenzie, D. Bernoulli, S. O. Schlanger, Init. Rep. ered from Detroit to Koko Seamounts, © 2004 Nature Publishing Group Deep Sea Drill. Proj. 55, 415 (1980). however, leads us to believe that this po- interval is 10 Ma for both. (B) Com- 22. R. A. Keller, M. R. Fisk, W. M. White, Nature 405, 673 puted changes of hotspot latitude for (2000). tential channeling of plume material was fixed-source plume model (13) (con- 23. R. D. Cottrell, J. A. Tarduno, Tectonophysics 362, 321 limited to the region at or north of Detroit tinuous red lines) for plume-initiation (2003). ages of 150, 160, and 170 Ma (upper 24. C. J. Ebinger, N. H. Sleep, Nature 395, 788 (1998). Seamount. Furthermore, the similarity of to lower). Moving-source model (13) 25. B. Steinberger, Geochem. Geophys. Geosyst. 3, the Hawaiian-Emperor chain with the Lou- Fig. 3. (A) Paleolatitude data from ODP Leg 197 sites (1206, Koko Seamount;16 1205, Nintoku results (dashed purple lines) are 10.1029/2002GC000334 (2002). shown for plume initiation at 180, 26. T. W. Becker, L. Boschi, Geochem. Geophys. Geosyst. isville chain of the South Pacific suggests Seamount; 1204B and 1203, Detroit Seamount), ODP Site 884 (Detroit Seamount) (10), and DSDP 3, 10.129/2001GC000168 (2002). that asthenospheric channeling was not the Site 433 (Suiko Seamount) (11). Orange, results of thermal demagnetization; blue, results of 170, and 160 Ma (upper to lower). 27. B. M. Steinberger, A. R. Calderwood, paper presented alternating field demagnetization. Result from 433 is based on AF and thermal data. Magenta, Paleolatitude means for Koko, Nin- at the European Union of Geosciences XI meeting, sole cause of the paleolatitude progression. toku, Suiko, and Detroit Seamounts Strasbourg, France, 8 to 12 April 2001. We examined whether the observed pa- magnetization carried by hematite from weathered basalt from Site 1206. (B) Average paleolati- (Fig. 3) are also shown. 28. A. M. Jellinek, M. Manga, Nature 418, 760 (2002). tude value for Detroit Seamount (square), based on inclination groups derived from basalts of Sites leolatitude motion can be explained by a 884, 1203, and 1204B (Model B, see text), plotted with select values from other seamounts (A). geodynamic model of the interaction of a Also shown is a least-squares fit to the data (orange) and several paleolatitude trajectories1068 22 AUGUST 2003 VOL 301 SCIENCE www.sciencemag.org plume with large-scale mantle flow. The flow representing combinations of plate and hotspot motion.

www.sciencemag.org SCIENCE VOL 301 22 AUGUST 2003 1067 天皇-ハワイ海山列の屈曲年代 Age of Emperor Hawaiian bend

80 Detroit Detroit Tholeiitic Lava Miiji Seamount 50 Meiji Alkalic Lava EMPEROR SEAMOUNTS Suiko Paleontologic Data Rejuvinated Lava 60 Nintoku 8.6 ± 0.2 cm/yr Ojin Koko Suiko Seamount ge (Ma) HAWAIIAN-EMPEROR BEND (HEB) Nintoku Colahan 40 Seamount Ojin Seamount Midway otassium Argon A Koko HAWAIIAN-EMPEROR P Seamount BEND (HEB) 20

30 Midway Colahan French Frigate Nihoa 170 Seamount French Frigate HAWAIIAN Shoals HAWAIIAN Shoals Kauai RIDGE ISLANDS Nihoa Kauai 0 180 Oahu 2000 1000 0 Distance from Topographic High 170 Necker 20 Kilauea in Kilometers Active Volcanic Vent Niihau 160 Loihi Loci of Shield Volcanoes Kilauea Seamount

Figure 3. Linear Hawaiian age progression derived from ages of Hawaiian-Emperor seamounts plotted against their distance from the active Kilauea Volcano based on data available from Clague and Dalrymple (1987), Duncan and Keller (2004), and Sharp and Clague (2006). Although the highly linear morphology of this seamount trail is more intricate when viewed close up (Jackson et al., 1972), these K/Ar and 40Ar/39Ar age data show a systematic (and more or less linear) aging of the shields of these volcanic islands and seamounts to the northwest and across the sharp 120° Hawaiian-Emperor Bend (HEB). articles (Koppers, 2010 17 particular model; however, comparison with our previous qualitatively correct. The computation consists of two steps. First, a 7,9,15–17 knowledge about the motion of the allowed us to accurately determine dura- Keller,work 2004;shows Duncanthat etthese al.,ar 2006).e typical Tresults.e By referring to large-scale mantle flow field is computed, which is based on a these previous results, we show that the conclusions here do not mantle density structure inferred from seismic tomography, global plumes themselves as well as the best tions of seamount formation (sometimes directiondepend on andthe choicemagnitudeof one parof ticularthis drisetfof model parameters but plate motions and a radial mantle viscosity structure; this model fits are robust conclusions that are valid as long as our model is at least several observational constraints (see Supplementary Information possible seamount geochronology. up to ~ 10 million years) and rates of is similar to recent modeling of plume 1). The computed large-scale flow field is time dependent, although Recent improvements in 40Ar/39Ar age progression along seamount trails. advections within the context of whole it does not change much over the period considered here. In the lower part of the mantle, the computed flow field is dominated by geochronology have allowed us to start Even though early geochronological mantle convection (Koppers et al., 2004; structure of spherical harmonic degree two, with large upwellings under the Pacific Ocean and Africa, downwellings around the addressing many of the above-described studies squarely underwrote the hotspot Steinberger et al., 2004) and suggests that Pacific Ocean, and flow towards the large upwellings in the lower- challenges in hotspot geodynamics model, later studies (for the same or initiation ~ 50 million years ago of the most mantle. In the upper part of the mantle, flow is also related to マントルに風が吹く？ plate motions. In the uppermost part of the lower mantle, com- and intraplate volcanism. Sensitivity other seamount trails, based on more distinctive 120° Hawaiian-Emperor Bend puted flow is a combination of outward flow away from the large upwellings and plate return flow towards ridges. The latter might be Mantleimprovements in masswind spectrometry model and extensive data sets, and using today’s (HEB) partially or even entirely refects ARTICLES NATURE GEOSCIENCE DOI: 10.1038/NGEO1638 an important contribution close to spreading ridges. Figure 1 shows the construction of low-blank extraction analytical techniques) have revealed a a change in the timing and magnitude a cross-section through density and flow field for different times beneath the Pacific Ocean. lines, in combination with aggressive signifcantly more ofcomplicated ‘not applicable’ picture. because itof is hotspot uncertain motion whether (Steinberger clasts in these and In the second step, a plume conduit is inserted into the flow field. In the deep mantle beneath the northern Pacific, the It is taken to be initially vertical. Plume initiation times are assumed application of incremental heating proto- For example, age datingvolcanic and sediments paleo- have retainedO’Connell, their orientation 1998; Tarduno since eruption.et al., 2003, In this study we use results only from the most reliable lava flow to be the same as in ref. 7, except for Hawaii, where the age is mantle flows southward. Mantle plume from hotspot unknown and 170 or 150 Myr ago is used here. The base of the cols,150 have allowed a renaissance68°¬69° in the magnetic data fromunits Ocean and Drilling dikes (ISCI 3 or 2).2009; Steinberger et al., 2004). Redating plume (assumed at depth 2,620 km, at the top of a low-viscosity Site U1372 Results from Rigil= Guyot provide the best documented in- bendage datingalong of seamounts the flow, (e.g., Kopperscausing southwardnProgram = 21 archive (ODP) Leg rapid 197 demonstrated motion of of the Louisville seamount trail (Koppers layer) is assumed either to move with the horizontal component of 100 clination record, with 522 m penetration at Site U1374 and an flow or to be fixed in location (see Supplementary Information 1). et al., 2003, 2007, 2008;Present-day Sharp and nthat = 9 discrete the Hawaiian hotspot drifed south et al., 2004) has shown that a formerly Hawaiian hotspotLouisville and no rapid motion ofadditional Louisville 66 m drilled at Site U1373, located 10 km to the east The velocity of points along the plume conduit is computed as the Count on the summit plain. Site U1374 has dominantly⇠ steep negative vector sum of ambient mantle flow and a buoyant rising velocity Clague,50 2006; O’Connorhotspot et al., 2007). As from ~ 35°N to 20°N between 80 and “linear” age progression (Watts et al., hotspot. inclinations (normal polarity) with several reversed polarity flows (see Supplementary Information 1). It is assumed that plume a result, a more precise understanding 49 million years agoin (Figures the uppermost 3 and 454; m and1988) remarkably is, in consistentfact, nonlinear, inclinations with varia in - conduits do not influence larger-scale mantle flow. Hotspot surface 0 many volcaniclastic breccias, similar to that of intercalated in situ motion is computed from the positions where, over time, the plume of the timing of intraplate volcanism has LouisvilleTarduno et IODPal., 2003; Duncan and tions in both hotspot and plate motions, conduit reaches the base of the lithosphere (assumed at depth flows or dikes (Fig. 3). A total of 19 in situ flow unit means for Site 100 km). Our choice of modelling parameters and assumptions U1374 were determined from discrete samples (seven additional has been explained in previous work7,9,15,16. Sites U1373 units were not sampled at sea). After correction for deviation of Motion of plume conduits in a high-viscosity lower mantle is and U1374 the borehole from vertical (2.2 correction;Oceanography see Supplementary March 2010 47 dominated by advection, and in a low-viscosity upper mantle it is 100 n = 28 archive 2-cm Information) these yield a mean inclination of 68.7 8.4 us- dominated by buoyant rising. Consequently, hotspot motion tends archive-half n = 28 discrete ing inclination-only averaging20. An additional 9 flow units± from to be similar to the horizontal flow component at the depth at which the transition from low to high viscosity occurs (that is, the upper Site U1373 give a shallower mean inclination of 55.2 10.6, part of the lower mantle)16. For the viscosity structure used, the 50 similar to moderate inclinations recorded at both the top± and Count horizontal flow at this depth has a root-mean-square value of Discrete samples base of U1374. As lava flows of Site U1373 erupted 1.0 Myr 21 ⇠ ,1 cm yr . Beneath the Tristan and Re´union hotspots it is domi- later than lava flows in the deepest basement units of Site U1374, nated by outward flow from the large upwelling under Africa; in the 0 the shallower inclinations at Site U1373 are more likely to reflect case of Louisville it is a combination of flow away from the large palaeosecular variation rather than Pacific Plate motion or drift of upwelling under the Pacific Ocean and plate return flow towards the the Louisville hotspot. Pacific–Antarctic ridge17. The computed motion of these hotspots is We have therefore combined flow units from Sites U1373 indeed slow and is as expected from the flow pattern (Fig. 2). For the Hawaiian plume, another effect leads to faster (a few cm yr21) and U1374 to calculate an overall mean inclination for Rigil hotspot motion. Figure 1 shows a projection of the Hawaiian Site U1376 150 Guyot at 70 Myr ago. For this and other guyots, we have conduit; it is strongly tilted in a north–south direction. Flow in n = 8 archive ⇠ treated each flow unit as independent, because we were unable to the upper part of the mantle is to the north and in the lower part to n = 8 discrete 100 sample all in situ flow units at sea and the remaining unsampled the south, which tilts and distorts the plume. Subsequently, the units would probably affect identification of inclination groups18. rising of a tilted plume conduit, aided by a large-scale upwelling, Count 50 More importantly, the presence of intercalated sediments and causes comparatively rapid hotspot motion. Among the hotspots volcaniclastics suggests that many flow units in fact represent considered here, we find this effect only for the Hawaiian hotspot. 0 We have previously shown results for a larger number of models, independent samples of the geomagnetic field. Owing to the small and discussed the dependence of results on various model par- number of lava flows, we use a bootstrap resampling to provide the ameters7,9,15–17. In summary, these results yield the following. For most robust estimate of the mean inclination and its uncertainty Hawaii, the motion is in a southerly to southeasterly direction. The Kopper et al.(2012) Steinberger et al.(2004) 21 for each site (Fig. 4 and Table 1; see Supplementary Information for average speed is moderately slow (,1 cm yr ), but episodes of Site U1377 details). In contrast to earlier studies, we also explicitly incorporate faster motion (several cm yr21), lasting for several tens of Myr occur n = 12 archive 100 within-flow inclination dispersion (with a median kappa of 280 for some models. Computed hotspot motion tends to be faster, if an n = 3 discrete 18 Figure 1 North–south mantle cross-section at⇠1558 W. For different times, computed older age for the plume is assumed. For Louisville there is slow for discrete sample flow means). The resulting distributions for 21 2-cm archive-half measurementsmantle anddensity discreteanomalies samplesare shown in fromrainbow Rigilcolours, and north–south and vertical motion (,1 cm yr or less) in various directions—in most cases in 50 components of computed mantle flow as arrows. a, 120 Myr ago;b, 90 Myr ago;c, 60 Myr a southeasterly direction. For Re´union there is slow motion Guyot are similar and statistically indistinguishable (at the 1 21

Count ago;d, 30 Myr ago;e, present. Also shown is the projection of the predicted Hawaiian (,1 cm yr or less) in an easterly direction. For Tristan there is confidence level) from the geocentricplume conduit for axialsource moving dipolewith the inclinationflow (plume initiation age 170 Myr ago;thick red either very slow motion (less than 1 cm yr21) in a southeasterly to 0 ( 68) for the present-day hotspotline) and locationfixed source at (age51150 S.Myr ago;thick violet line) at those times. southwesterly direction, or an almost stationary hotspot. ¬90° ¬80° ¬70° ¬60° ¬50° ¬40° ± ⇠ The resulting palaeolatitude for Rigil Guyot is 47.0 S 8.0 Inclination 168 NATURE | VOL 430 | 8 JULY 2004 | www.nature.com/nature (n 28) and its distribution of mean flow inclinations is statistically± © 2004 Nature Publishing Group similar= to that expected from global geomagnetic field models Figure 4 Bootstrap inclination distributions for individual Louisville at 51 S (Fig. 5). This result alone suggests that the Louisville seamounts| drilled by IODP Expedition 330. Distributions for both 2-cm hotspot⇠ experienced limited latitudinal motion since 70 Myr ago. archive-half (red) and discrete sample (blue) flow means are shown, each Therefore, assuming the simplest possible palaeolatitude history, representing the bootstrap results of 1,000 resamplings with replacement it follows that the younger Louisville volcanoes are likely to have and incorporating within-flow dispersions, using only units with ISCI 3 or sampled the geomagnetic field at a similar latitude. Even though = 2. Inclination averages and 1 (filled rectangles) and 2 uncertainties we sampled only a limited number of flows from the 64- and (lines) exclude pseudosamples where inclination-only20 averaging failed to 50-Myr-old Burton and Hadar guyots, they each are statistically converge (yielding inclinations of 90 that are excluded in the indistinguishable from 51 S at 49.8 S 4.8 (n 9) and 52.3 S ± = ± uncertainties of our average palaeolatitude estimates). The vertical grey 20.2 (n 3). It is unlikely that the results from either guyot line indicates the expected geocentric axial dipole inclination for the adequately= average palaeosecular variation, but nevertheless these present-day hotspot location (50.9–52.4 S; refs 4,17,21,25). results are broadly compatible with the palaeolatitude estimate of Rigil Guyot and, therefore, a limited Louisville hotspot motion, at least since 70 Myr ago. Louisville seamounts is less straightforward than for sites in Together, the inclinations from Sites U1373–U1377 represent the Hawaiian–Emperor seamount trail5,18,19. We developed an a relatively large number of flows, from one dominantly normal objective, although qualitative, index (the in situ confidence polarity and two reversed polarity guyots, which may provide an index, ISCI; see Supplementary Information) to determine whether adequate sampling of geomagnetic palaeosecular variation. To test individual units were probably in situ, ranging from ISCI 3 for this, we compare the directional scatter (circular standard deviation, = units that are definitely in situ to ISCI 0 for those that are ✓63) of the combined data set to two statistical models of the probably not. Volcaniclastic units were assigned= an ISCI value geomagnetic field (CJ98, TK03; refs 22,23) that were designed

914 NATURE GEOSCIENCE VOL 5 DECEMBER 2012 www.nature.com/naturegeoscience | | | Type 2

スーパースウェル：フレンチポリネシアGeochemistry Geophysics 3 adam et al.: seafloor swells 10.1029/2004GC000814 SuperGeosystems Swell:G French Polynesia

Figure 10. Swell amplitudes (in meters). The general map represents the case in which two swells are found for the Cook Islands. The case when one swell only is found is represented in the top left corner inset. The black line is(Adam the et al., 2005 3000 m isobath. 水深が年齢に比して浅い water depth is shallower than expectedly age-depth curve chain at the scale of the alignment and this chain not correlated with any volcanic structure. This will not be discussed any further. demonstrates19 that the volcanoes contribution is efficiently removed by the MiFil method. For this 4.2. Society volcanic chain, the swell description corresponds to the one previously reported for hot spot swells, 4.2.1. Volcanic Chain Description created by the simple interaction of a plume with

B11407 FORSYTH ET AL.: GLIMPSE SEAMOUNTS AND RIDGES B11407 the lithosphere: the swell’s maximum is located [44] The Society islands (Figure 11a) are situated 215 kilometers downstream from the active volca- between latitudes 16 Sand19Sandlongitudes ° ° nism, the swell stretches along the volcanic chain 147 Wand153WonaseafloordisplayingagesType 3 ° ° and subsides along the direction of the plate between 65 and 95 Ma. They stretch along a motion. 200-km-wide and 500-km-long band orientated in the direction of the present Pacificリソスフェアの裂け目から漏れる？ plate motion: 4.2.3. Buoyancy Flux N115 ± 15°. The age progression is uniform from the youngest submarine volcano,lithospheric Mehetia [46]FortheSocietyvolcanicchainwefinda cracking? B11407 FORSYTH ET AL.:1 GLIMPSE SEAMOUNTS AND RIDGES B11407 (0.264 Ma [Duncan and McDougall,1976]),situ- buoyancy flux of 1.58 ± 0.15 Mg sÀ , using VL = 1 ated at the southeast extremity to the oldest dated 110 mm yrÀ .Previousestimations[Davies,1988; island, Maupiti (4.8 Ma [White and Duncan, 1996]). Sleep, 1990] are greater, mostly because the

B11407 Figure 1. Predicted bathymetryFORSYTH of the East Pacific Rise ET in the AL.: vicinity ofGLIMPSE the GLIMPSE study SEAMOUNTS area AND RIDGES authors overestimatedB11407 the swell volume in not based on satellite altimetry [Smith and Sandwell, 1997]. The broad shallow area immediately north of the Sojourn Ridge at about 116°W4.2.2. does not exist (see Swell Figure 2); it is an artifact stemming from a geoid or removing completely the volcanoes. gravity anomaly that does not have corresponding bathymetric expression. The inset map of the globe shows the location of the GLIMPSE study area in the box, with heavy black lines indicating plate boundaries. [45]Thetopographicanomalyassociatedwiththis [47]AspointedoutbyCourtillot et al. [2003], 2003]. Gans et al. [2003] suggested instead that cracking of lineations is to examine in detail the morphology of the the lithosphere is caused by bendingalignment of the plate under seafloor is and shown the distribution of in eruptive Figure centers and lava 11b. It stretches several criteria (plume duration, traps at their thermal stresses associated with cooling of the lithosphere. flows. One could argue that the evidence for cracks preced- Sandwell and Fialko [2004] showed thatalong thermal stresses theing volcanic volcanism has been chain.obscured by subsequent Itsmaximal burial amplitude, initiation, rare gas isotopic ratio, Vs at 500 km would lead to a preferred wavelength of bending and beneath the built-up edifices, but if we look at the process cracking, which could reproduce the initial 150 to 200 km in the early stages of formation, we should be able to wavelength of cross-grain gravity lineations980 [Haxby m, and isrecognize reached which type30 of model km is more northwest likely by searching of Tahiti, and is depth and the buoyancy flux) allow to establish the Weissel, 1986]. Both of these variations on the tectonic for faults or linear features and by establishing whether the cracking hypothesis are passive models in which the vol- initial volcanic activity is localized or widespread. The Puka canic ridges form in response to cracking; either the crack Puka chain is now a fossil feature, except perhaps for some 13 of 25 lets preexisting asthenospheric melt escape to the surface or scattered volcanism in the Rano Rahi seamount field the small amount of local extension induces some upwelling [Scheirer et al.,1996a,1996b],butwehaverecently that causes pressure release melting. mapped two other neighboring, linear volcanic features that [4] Alternative models for the origin of the volcanic are still actively forming to the west of the East Pacific Rise. ridges involve anomalous melting in the asthenosphere [6] This paper describes the distribution of recent intra- associated with small-scale convection, minihot spots, or plate volcanism and the morphology of the volcanic features return flow of the asthenosphere to the East Pacific Rise that in the Gravity Lineations Intraplate Melting Petrology and incorporates compositional and/or thermal anomalies within Seismic Expedition (GLIMPSE) study area, with an em- the returning flow field. In these models, the volcanic ridges phasis on distinguishing between lithospheric cracking and are caused by active processes beneath the lithosphere that asthenospheric origins for the volcanic activity. In this area create temperature variations and pressure-release melting. just to the north of the Rano Rahi seamount field (Figure 2), The ridge-like nature of the individual edifices could be young lava flows and the Hotu and Matua seamounts were attributed to dikes propagating under the influence of a first discovered in the early 1990s during mapping in remotely applied stress field or under the stresses associated preparation for the Mantle Electromagnetic and Tomogra- with loading of the lithosphere by a linear chain of sea- phy (MELT) Experiment. The existence of the Sojourn mounts [Hieronymus and Bercovici, 2000]. Ridge was first noted on high-resolution images of satellite [5] One way to distinguish between lithospheric cracking free-air gravity anomalies [Smith and Sandwell, 1997], and and asthenospheric models for the origin of the volcanic it was partially mapped from shipboard for the first time in

2 of 19

(Forsyth et al, 2006 Figure 3. (a) High-resolution bathymetry and (b) side-scan sonar images of Hotu Matua region(box A in Figure 2). Matua is dominated by a single conical feature, while Hotu is a coalescence of at least four 20 flat-topped volcanic centers. The most reflective patches representing the roughest and presumably youngest seafloor are scattered around and to the east of Matua. An example of botroidal topography is found to the east of the main edifice of Matua. Blue lines indicate older, less reflective, bathymetric features that separate the Hotu and Matua recent flows and which probably formed nearer the East Pacific Rise (EPR); they are labeled as preexisting features. Dredge locations with Ar/Ar ages are shown as red dots with age determinations indicated. In this and subsequent detailed images, pixel size is 200 m. Artificial illumination of the bathymetry is from the east.

6 of 19

Figure 2

3 of 19 GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO. 14, PAGES 2719-2722, JULY 15,2001

HIRANO ET AL.: A NEW TYPE OF INTRA-PLATE VOLCANISM 2721 A new type of intra-plate volcanism; young alkali-basalts Table 1. Bulk compositionsdiscoveredoffrom 10K#56 the R-001 subducting andR-002 Pacific by XRF analysis.Plate, northern Japan sample Trenchmajor element (wt%) N. Hirano et al. Total N.Hirano SiOl, K.2 Kawamura TiO 2 AI203?,M. Fe203 Hattori 3,FeOK. Saito MnO4and MgOY. Ogawa CaO-• Na20 K20 P205 H20+ H20- TypeR-001 3bulk 48.22 2.56 10.30 6.38 4.23 0.14 11.68 7.47 2.95 3.17 0.78 1.76 0.36 100.00 100.00 groundmassAbstract.a 48.84 Alkali 2.72 pillow basalts11.60were 10.44 collected from - the0.13 toe of 8.23of the7.68 northern 3.04Japan 3.88 Trench 0.83 (39023 ' 1.23N, 144016 1.39' E) during REPORTS R-002 bulk the oceanward48.45 slope2.84 of 11.26the northern 6.12 Japan4.36 Trench. 0.13 These 7.83 JAMSTEC8.14 (Japan3.28 Marine3.63 Science 0.84and Technology2.49 0.64 Center) 100.00 R/V groundmassalkali-basaltsa 49.27formed 2.84as a result12.18of a low10.00 degree of partial- 0.12melting 6.61 Kairei/ROV 7.61 KAIKO3.13 cruise4.11 KR97-11. 0.87 The1.47 slope is1.81 characterized 100.00 プチスポット by trench-parallel(N-S) normal faults with someNNW or NNE velocity region (a zone potentially representingof Pacific Ocean mantleby and rapid means rise of ofthe surfacemagma (no wave tomography, show- volumes of magma output over a large area sample fractionationtraceelement in shallow(ppm) magma chambers). Reconstructing faults,due to warpingof the downgoingSeamounts,Pacific Plate (the knollsage of and petit-spot monogenetic volcanoes on the subducting Pacific Plate Pacific Plate motionBa based Ce on 40Ar-•9ArCo age Cr datesGaof 5.95_Nb 0.31 Nithe PacificPb PlateRb here is EarlySr CretaceousTh [Kobayashi V Y et al., Zr amantleplume)ata410-kmdepthinthisarea ing that the thermal state1998]) of (Fig. the 1). These asthenosphere normal faults bound horstcould and graben easily be explained. R-001 petitbulk Ma forspot these1202.3basalts indicates94.4 56.7that they527.8 erupted outboard 17•7 of36.6 outer 418.1 7.2 47.0 1076.5 2.0 140.1 14.2 243.0 structuresthat are approximately5 km in horizontalextent with swell or forebulgeof the JapanTrench in the NW Pacific. We was reported by Obayashi et algroundmass.(17);d however, 1176.8 -154.0is notably 424.0 homogeneous- 37.4 163.1100 to 5006.9 throughout m vertical56.2 separations1212.4 the [Ogawa 4.7 Pa- and Kobayashi,- 18.8 An1993; 263.8 old and cold lithospheric plate behaves suggest that these alkali-basalts represent a new form of N. HiranoR-002et al . bulk 1209.5 108.7 46.7 400.5 20.2 40.3(a) 227.8 7.3 53.1 1092.4 2.0 143.4 15.3 259.9 ing to Koppers etal. (2000) this is due to the di¡erential de- there is no evidence for a conduit orintra-plate connectionvolcanism, whereby cificmagmatic Oceanactivity occurs (18).off the The temperature at a depth elastically and may be flexed because of 19.7 273.8 groundmassforebulged 1191.2of the downgoing - Pacific129.0 slab,353.0perhaps using - conduits 39.4 117.2 140øE 7.3 59.0 1140.4145'E 4.9 - 1501E gassing of alteration minerals at the low temperature steps of any type between this low-velocityrelated to region fracturing of ofthe slab 150 during km bending is estimated prior to to be between 1450- loading by an ocean island or seamount or by a Data for the samplesseparated the olivinephenocryst from the groundmass.In thisdata a, Fe203show the total Fe-oxide. and the shallower mantle. Our geochemicalsubduction. and 1480-C, which implies that the tempera- subduction-related plate flexureand (21). the In the degassing area of primary plagioclase and clinopyrox- 1. Introduction ene phases at the high temperature steps, all in combina- evidence strongly supportsTable 2. a depletedREE compositions mantleof the bulkture samples just belowby the ICP-MS the Pacific analysis. PlateAnalyzed is 50 by- Dr.to M. 150 Komuro-C andbetween Ms. K. Fujii, sites A and B, the bathymetric high (or Instituteof Geoscience,Alkali-basaltsUniversity occur of on Tsukubavarious parts(personal on the surfacecommunication). of the tion with various redistributions between Ar isotopes due (nonplume-like) source. Furthermore,earth, the mostparticularly vol- in lowercontinentaland than hotspot theareas. solidusThese of dry mantle mate- outer rise)0%• is aligned parallel to the Japan Trench. ppm ench 39 37 ume of magma produced at thealkali-basalts petit spotare productsofrials deep-originT (r19magma ). However,from the upper at this depth and temper- The Pacificench Plate flexes convexlyto Ar here, and asAr it recoil. These e¡ects may be more severe Y mantleLaor lowerCedepths. Occurrences Pr Ndof such alkali-basalt Sm areEu also Gd Tb [ Dy• •J.•_ Ho Er TmRise ,]Yb r Lu R-001 19.17 documented54.07 106.95from tectonically12.03 urunique i49.35l locations,9.41 such 2.98 as along 7.74 1.00 4.63 0.77 1.83 0.23 1.41 0.19 when the glass content of the groundmass is high, but volcanoes must have been severaldeep ordersfractures in oceanic of crust. KurilKature Trench in the asthenosphere, a small percent subductsT beneath Japan and yields a positive R-002 20.78 58.69Alkali olivine114.28 basalt 12.97and trachyandesite53.61 10.02representative 3.22of the 8.38 1.08 4.72 0.88 1.93 0.23 1.46 0.20 magnitude less than those typical ofoceanhot island spotbasalt series melthave beenshould documented existaround the in the presence of small gravity anomaly (22). This flexedwell-developed region is age plateaus are generally formed at the –4000 Japan Trench–8000 on the Joban, Erimo and Takuyo Seamounts uril –2000 volcanoes. We found 4.2 to 8.5 million-year-[Kobayashiet al., 1987; Cadetamounts et al., 1987]. (TheG 1%)40Ar•9Ar ofages H2O or CO2 (20), which elevated 9800 m above the normalintermediate ocean floor temperature steps in case of holocrystalline –5500 KurilK Trench old volcanoes at site A, suggestingof –6000these episodic volcanicrocks rangelowers from 120 theMa (Daiichi-Kashima melting temperature. The highly (È6000 m below sea level atgroundmass site B). Large samples. We see similar systematics in our 4. Tectonicinterpretation ErimoErSeamountimo in the and Joban geophysical Seamount Chain) to 104 Ma (Erimo eruption of magma over a distanceSeamountSeSeamount) ofam 400oun[Takigamit km et al.,vesicular 1989], indicating petitthat these spotare thebathymetric lavas probablychart of the northwest representPacific, curvatures this area corresponds imposed on the preflexedgroundmass lithosphere age spectra. However, two samples did not implication productsof Cretaceousoff-ridge seamountvolcanism (Fig. 1A).to a site just oceanwardof the currentouter swell or forebulge In contrast,the bathymetryNosappuNosappuof the studyarea doesnot show any of plate motion. Accordingly, the petit spotFractureFractureincipient partial melts(Hokkaido that rise), formed with an in inferred the as- paleo-depthmight of-6000 instigate m brittle fracturingproduce (that clear is, age plateaus, and two other samples only The alkali-basaltsdocumented evidencefor a largehere ZoneZon evolcanic are veryedifice young, or seamount, muchbut only(b) a–4000 volcanic province is characterizedsmall by mound several(Fig. lB). thenosphere in the(asterisk presencein Fig. of 1A). vola tiles, most bending-induced faults) (23). As for volcanic youngerthan the oceanfloor This inpaper thisdescribes area (identifiedan occurrenceisochron of youngalkali-basalt M9 on Enriched incompatible element concentrationsand REE formed narrow three-step age plateaus.This indicates that or M10; around 130Site Ma A the [Gradstein downgoingoceanicet al.,slab 1994;of a subduction Kobayashi–2000zone. et We al., present the –8000 million years of small-volume magma produc-–5500 plausibly–5500 CO2.Ifthemagmasweresuppliedpattern indicate that the magma source forcones these alkali-basalts at site A, the volcanic features, aligned in 1998]). Seamountsin thisgeologic region setting, are major also and Cretaceous, trace elementwith compositions, ages and alteration could not su⁄ciently be removed or that the tion over a large area. 40Ar-39Arage of these basalticfrom rocks. the We normalthen discuss asthenosphere,the formed as a result with of emplace- low degree awest-northwesttoeast-southeastdirection,areof partial melting. rangingfrom 120-104 Ma.tectonic In andcontrast, geophysical the implications rocks analyzedfor this firstin documentation this DisequilibriumFig.1 . Bathymetrical–5500 between olivine phenocrystsand groundmass groundmass samples contained too much devitri¢ed ba- ench of alkali-basalticmagmatism outboard of outer swell of a Temperatures ofstudy the have asthenosphere agesr of 5.95+0.31 beneath Ma (latestment Miocene), channels and–6000 are controlledolivinesmap of o¡shore suggest by NE Japan tectonic that the fractur-phenocrystsessentiallymay be xenocrysts perpendicular to hinge lines on the T subductingoceanic slab. using the data of Smith & saltic glass.We thus need to be cautious in our interpreta- the Pacific Plate havefound been in small-volumes mappedalong precisely seafloorSite B escarpments ing of therather overlying than transportedSandwell lithospheric (1997) from deep plate, in the mantle low withbending rapid rise plate of the (Fig. 1, A to C). Accordingly, it on seamountsor large volcanic constructions.By performinga alkali-basalticcontoured at 500 mmagma. If a fractureoccurs or is rejuvenatedin the tion of these age spectra and only use our results to JobanJoban2. Occurrenceand descriptionErimoEof r samplesimo intervals (upper panel) plate tectonicapan reconstruction we have determinedthat there is no appears that magmas are brought to the surface JapanJ Trench SeamountSeamouContinuousnts outcrops of pillow basalt were documentedand and a model for petit- Fig. 2 ...... distinguish between samples of Cretaceous age and sam- plausible hotspot that sampledcould athave depths producedof 7325 to–6000 7360 these m SeamountSon basalts. thee oceanwardam Weslope otoeu spot n relatedt to the plate along fractures parallel tothedirectionofthe havereconstructed the eruptionlocation of thesebasalts using the £exure (lower) modi¢ed ples that are younger than 10 Ma, the low age expected for DaiichiDaiichi– from HiranoFigureet al1.. (2006). Index maps of the dive site. A: The general IDoctralProgram –6000in Geoscience,University of Tsukuba,Tsukuba, Japan. maximum horizontal compression. However, 40Ar-39ArKashimaKasageh im aof 5.95 + 0.31 Ma and the present "absolute" White andbathymetric brown boxesNosappuNmap o ofs thea northwestppuPacific oceanfloor basedA.on SeamountSeamount Now at OceanResearch Institute, University of Tokyo,Tokyo, Japan. Kobayshiet al. (1998). Black area is trench floor deeperthan typical petit-spot volcanoes (Hirano et al., 2006). motionof the PacificJapaneseJap Plateane2Fukadas (10.29e Geological cm/yrInstitute, to 295.26Tokyo, Japan degrees [ Gripp show the area of Fig. 2 the surface compression is actually caused by SeamountsSeamounts and the7000 region m, of cross-and grey-shadedarea is the outer swell (<5400 m in 3JapanMarine Science and Technology Center, Yokosuka, Japan FractureFracture and Gordon, 1990]). Using this method we have derived a 03201.sectionaldepth). modelRadiometric shown by ages of Erimo (a), Ryofu (b), Daini-Kashima The petit-spot basalt of 6K#880-R3B, on the other 4Departmentof Earthand Environmental Sciences, Faculty of Science, Plateau age=5.95_0.31extensional Ma stresses on the base of the down- the lower(c) ¢gure, and Daiichi-Kashima(d) Seamountsare obtainedas follows position of approximatelyYamagata612+32 University, kmYamagata, ESEJapan off the northern respectively.Whiterespectively; heavyZoneZ(a) o 104n Mae and(d) 120 Ma 40Ar-39Arage [Takigami hand, does show a well-developed age plateau that is 83% 5Instituteof Geoscience,University of Tsukuba,Tsukuba, Japan warping Pacific Plate (Fig. 3C). Japan Trench, now approximatelyat 37øN, 149øE. As the lines showet al., the 1989], (b) 70-7239Ar=94 Ma and (c)% 81 Ma(n=4) K-Ar age [Ozima et al., volcanicfront in the NECopyright Japan2001 Arc by has the American scarcelyGeophysical shifted Union. since the Cretaceous1977]. fracture The zoneapproximateeruption site is plotted as asterisk.B: wide and includes eight heating steps. However, the q:10Seabeam bathymetric map of the dive site 10K#56 Post–erosional-stageby R/V lavas on some of the latest Miocene [Ohki et al., 1993], the kinematicsPacificof the Plate Pacific (Nakanishi & Winterer, 40 Papernumber 2000GL012426. 1998). AreaKAIREI of outer-rise at the northernJapan Trench. Contour interval Hawaiianis 250 m. Islands, as well as submarine2.54 0.76 lavas Ma plateau age is too high due to excess Ar slab at 5-6Site Ma A are the0094-8276/01/2000GL012426505.00 same as at present. According to the (c)is shownTrench by hatchedaxis area,is shownby white arrow. Æ 40 36 Site B Site A which is approximately 2719 on the flexural Hawaiian Archas indicated (because by of the Ar/ Ar ratio of 340 21that is high- correspond–5500 to shallower –5500 outer-rise 0 o 100 (Hirano et al., 2001; 2006; 2008) Æ area than 6000 meters39Ar cum. % 50 loading on the plate), have chemicaler than composi- the atmospheric ratio (295.5) in the inverse iso- Japan below sea level (mbsl) for lOO 0.2 0.4 39Ar/4øAr0.8 1.0 Trench 130^140 Ma ocean £oor at ß . tions, and tectonic emplacementchron. mechanisms The 1.76 0.58 Ma age from the inverse isochron the western side of 21 Æ fracture zone and than (24)thataresimilartothoseofthepetitspot3 diagram thus appears the best age estimate for sample 5900 mbsl"',k for120^125 Ma (4MSWD40 Ar/ 36 Ar),=o= = 1 .O6 304.9_9. -I- 4 B Fig.1. Bathymetrical ench •3ocean £oor at the eastern lavas. It has also been proposed6K#880-R3B that alkalic and is consistent with its morphological r side of the fracture zone, map of o¡shore NE Japan respectively. lavas in the western Samoan Islandsdesignation (located as on a young petit-spot volcano. T the opposite side of the main Samoan hotusing spot the data of Smith & Site B shield volcanoes)10• may be derived fromSandwell the (1997) r 2008 The Authors 544 Journal Compilation r Blackwell Publishing Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists asthenosphere during tectonic faulting (contoured25). If at 500 m true, a similar style of volcanismDISCUSSION may be present apan JobanJoban Age (Ma) lO 8 6 intervals (upper panel) o.1 JapanJ Trench SeamountSeamounts and a model for petit- sr Kアウターライズ屈曲で生まれれるか？ RI:BaThTaNI:CePZrHfSmTi Y YbScCr LaCePrNd SmEuGdT'bDyHoErTmYbLu Figure 6. Age determinationresults by 40Ar-39Ar method of Twostages of intra-plate volcanism characterize the NW sample 10K#56Fig. R-002. 2 Correctionfactors and J-value are as follows; (36Ar/37Ar)ca= (3.744-1-0.082)–6000X 10-4, (39Ar/37Ar)ca= Paci¢c Platespot spanning related to a thegeological plate time period from the Figure $. Spidergramsof trace elementconcentrations in (d) samplesOuter10K#56 R-001 rise and R-002. flexure A: Normalizedby average (9.30+0.44)X 10-4 and J = (3.412+_.0.063)x 10-3. Calculated Early Cretaceous£exure (lower) to the recentmodi¢ed past. In the ¢rst stage, the MORB [Pearce,DaiichiD 1982,aii 1983].chi B:– Chondrite-normalizedREE usingdecay constants and potassiumisotope ratios from Steiger seamounts of the West Paci¢c Seamount Province (WPSP) pattern[Evensen et al., 1978].U, Ta andHf werenot analyzed. and Jager (1977).–6000 from Hirano et al. (2006). KKashimaashima and JobanWhite and and Japanese brown Seamount boxes Trail (JJST) were SeamountSeamountJapaneseJapanese formed duringshow the the Latearea Jurassicof Fig. 2 and Cretaceous (Koppers etal., 2003).The new circular knolls described in this study SeamountsSeamounts appear inand the the Late region Cretaceous of cross- and thus seem to be part of this earlysectional stage ofmodel intra-plate shown volcanism. by However, as we Fig. 1. (A) Map of the northwestern Pacific will discussthe lower below, ¢gure, their morphology may also indicate Ocean (22)withsurveyedareasnotedinthe that theyrespectively.White were formed as o¡-ridge heavy volcanoes. In the second black boxes. A, site A; B, site B; C, site C. The stage, the irregularly shaped and low-volume petit-spots outer rise (G5500 m below sea level) is shaded. lines show the Bathymetric and side-scan sonar maps for site A appear, butCretaceous only during fracture the last zone 10 Myr and only at loca- (B and C)andsiteB(D and E).De 50-mpth (m and) 20-m tions on(Nakanishi the outer-rise & of Winterer, the subducting Paci¢c Plate (e.g. contours are indicatedFig. in3. (B) Bathymetrical and (D), respectively. maps (left ¢gures)PacificFig. and2. Plot sidescan ofPlate Ne sonar versus Ar isotopes.Hirano (20Ne/et1998).21 alNe)*., 2006).This Area of outer-rise latter kind of intra-plate volcan- 20 22 20 22 Site A In (B) and (C), whiteimages rectangles (right ¢gures) indicate of dive circular and knollsis shown equal each to site [( Ne/ in Fig.Ne) 2. –ism ( Ne/ seemsNe unrelatedAir)]/ to the massive hotspot volcanism that 21 22 21 22 is shown by hatched area, dredge sites; dashedBathymetrical yellow lines maps indicate are contouredvolcano at100[( Ne/ m intervals.Ne) – ( HorizontalNe/ NeAir)], showingcharacterizes the slope the West Paci¢c during the Cretaceous, distributions; and theresolutions redSite line ofB indicates bathymetrical the cross and sidescanof a mixing map are line 300 on and the 50 conventional m, instead neon itwhichthree- appears isapproximately to be directly related to the subduction 20 22 21 22 section described inrespectively. Fig. 3A. Black arrows in (D) isotope plot ( Ne/ Ne versus processNe/ Ne) andcorrespond ( the8). bulging to shallower of the downgoing slab (e.g.Watts outer-rise Data with 20Ne/22Ne ratios of less than 10 were and (E) depict the three dives with the two & Zhong,area 2000). than Here, 6000 we meters will further expand on the mor- dredges. not plotted (8). Literature and data for MORB and Japan age spectra for acid-leaching samplesOIB may of altered be found submarine in the followingphological onlinebelow data- characteristics sea level (mbsl) and seamount for ages that allow us Trench basalts show high apparent agesbases: and low Petrological Ca/K ratios Database for ofto the distinguish Ocean130^140 Floor between Ma ocean the large £oor seamounts at formed in the the low temperature steps, and low(www.petdb.org/) apparent ages and and Geochemistry high Cretaceous of Rocksthe of theand western the smallside of petit-spot volcanoes formed Ca/K ratios for the high temperatureOceans increments. and Continents Accord- (http://georoc.mpch-mainz.more recentlyfracture inrelation zone and to than the subducting Paci¢c Plate. gwdg.de/georoc/). 5900 mbsl for120^125 Ma r 2008 The Authors Journal Compilation r Blackwell Publishing Ltd, European Association of Geoscientists & Engineersocean and £oor International at the eastern Association of Sedimentologists 547 www.sciencemag.org SCIENCE VOL 313 8 SEPTEMBER 2006 side of1427 the fracture zone, respectively. (Hirano et al., 2001; 2006; 2008)

22

r 2008 The Authors 544 Journal Compilation r Blackwell Publishing Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 巨大火成岩岩石区 Large Igneous Provinces LIPs

Figure 1. Phanerozoic global LIP distribution, with transient (“plume head”) and persistent (“plume tail”) LIPs indi- cated in red and blue, respectively. LIPs are better preserved in the oceans where they are not subject to terrestrial erosional processes, off ering a prime target for scientifi c ocean drilling. Modifi ed from Coffi n and Eldholm (1994).

23 as ~ 40 km, determined from seismic Seismic wave velocities suggest that the beneath these three LIPs apparently re- and gravity studies of Iceland (Fig- middle crust is most likely gabbroic, and fl ect residual effects, primarily chemi- ure 1) (Darbyshire et al., 2000). Nearly that the lower crust is mafi c to ultra- cal and perhaps secondarily thermal, all of our knowledge of LIPs, however, mafi c, and perhaps metamorphic (e.g., of mantle upwelling (Gomer and Okal, 陸上の洪水玄武岩：デカン高原is derived from the most accessible la- Eldholm and Coffi n, 2000). 2003). If proven to be common, roots floodvas forming basaltthe uppermost on portions land of DeccanLow-velocity zonesTrap have been ob- extending well into the mantle beneath crust. This extrusive upper crust may served by seismic imaging of the mantle oceanic LIPs would suggest that such exceed 10 km in thickness (e.g.,height Cof- ~2km,beneath area ~500,000km2,the oceanic Ontong Java volume Plateau 512,000LIPs contribute km3 to continental initiation fi n and Eldholm, 1994). On the basis of 68-60(e.g., Richardson Ma , multiple et al., 2000), flood as well basalt as well as continental growth. Instead geophysical, predominantly seismic, data as under the continental Deccan Traps of being subducted like normal oce- from LIPs, and from comparisons with (Kennett and Widiyantoro, 1999) and anic lithosphere, LIPS may be accreted normal oceanic crust, it is believed that Paraná fl ood basalts (Vandecar et al., to the edges of continents, for example, the extrusive upper crust is underlain by 1995) (Figure 1). Interpreted as litho- obduction of Ontong Java Plateau ba- an intrusive middle crust, which in turn spheric roots or keels, the zones can salts onto the Solomon Island arc (e.g., is underlain by lower crust (“lower crust- extend to at least 500–600 km into the Hughes and Turner, 1977), and terrane al body”) characterized by compressional mantle. In contrast to high-velocity roots accretion of Wrangellia (e.g., Richards seismic wave velocities of 7.0–7.6 km s-1 beneath most continental areas, and the et al., 1991) and Caribbean (e.g., Kerr (Figure 2). Dikes and sills are probably absence of lithospheric keels in most et al., 1997) LIPs to North and South common in the upper and middle crust. oceanic areas, the low-velocity zones America, respectively.

152 Oceanography Vol. 19, No. 4, Dec. 2006

24 巨大海台：オントンジャワ海台

GeochemistryOceanic Plateau: Ontong Java Plateau Geophysics 3 inoue et al.: ontong java plateau construction 10.1029/2007GC001780 Geosystems G LIPs Origin - mantle plume - MOR+triplejunction - volcanic margin S. Ingle, - impact M.F. Co⁄n / Earth and Planetary Science Letters 218 (2004) 123^134 129

et al., 2004) or in subaerial permafrost volcanism has been the latitude at which istry (e.g., Devine et al., 1984; Stothers settings (Figure 5). A key factor affect- the LIP formed. In most basaltic erup- et al., 1986). Subaerial eruptions of large

Figure 1. Predictedlargesting the bathymetry magnitude flood [after of Smithvolatilebasalt and release Sandwell 4,270,000km2 has,1997]oftheOntongJavaPlateau,outlinedinblack tions, released volatiles remain in the volumes of basaltic magma at high-lati- [Mahoney et al.,been 2001], whether and surrounding eruptions region. were Three subaerial research cruisestroposphere. have acquired However, MCS data at on high the plateau:latitudes R/V tude LIPs over relatively brief geological Kairei KR05-01119-125Ma in 2005 (thick black lines),formation R/V Hakuho Maru KH98-1 Leg 2 (thin white lines) in 1998, and R/V Maurice Ewing90MaEW95-11or submarine; (thinmagmatic purple hydrostatic lines) in pressure 1995.pulse Deep in- Sea Drilling(e.g., Project Kerguelen (DSDP) Plateau), and Ocean the Drilling tropopause Program intervals, including phreatomagmatic (ODP) sites on thehibits Ontong vesiculation Java Plateau and and degassing in the Nauru of rela- Basin areis indicated relatively by low, squares; allowing those penetrating larger mass igneous fl ux eruptions (an explosive eruption result- basement are shown in red. Malaita (Solomon Islands) is indicatedInoue in red; other et land al.(2008) area appears in gray. Black circles indicate the start and end of composite MCS profiles (Figure 4). Red lines indicate data shownFig. 4. in Conceptual Figures 5–7.model showingIngle possible sequence and of eventsCoffin if an V20 (2003) km diameter bolide instigated the creation of the OJP. tively soluble volatile components (H2O, (via basaltic fi ssure eruption(a) t1 Momentplumes of for impact, watering column from is vaporized,the interaction 20 Myr old of oceanic magma lithosphere with (pink layer) at impact site is obliterated, Inset shows zoom-in of intersection of MCS lines KR05-01 Split and KH98-01 Leg 2 501 (see Figure 2 for velocity S, Cl, F) during deep-water submarine transport) of SO and otheruppermost volatiles asthenosphere into is penetrated,water) and(e.g., surrounding Ross et lithosphere al., 2005), fractures would[62]. (b) t2 Moment of maximum penetration, the analysis at the intersection and Figure 7 for MCS data at the intersection). 2 crater25 is completely formed and melting region becomes focused. (c) t3 In¢ll of void from bottom and sides, melt also migrates out along fractures in lithosphere, refractory surrounding mantle ¢lls space vacated by out£owing magmas. (d) V120 Ma, the eruptions, although low-solubility com- the stratosphere. SulfuricOJP acid at endaerosol of emplacement (pinkhave layer increased represents Vpotential35 km thick to crust). contribute (e) V90 Ma, to tectonism causing new pressure release ponents (CO , noble gases) are mostly particles that form in themelting. stratosphere Scale is maintained throughoutglobal theenvironmental diagram. effects. [6]OntheOJP,intrabasementreflectionshave2 2001], and correlation between seismic data and been observeddegassed along itseven northwestern at abyssal depths margin (e.g., field observationsafter such eruptions of obducted have OJP a longer sections resi- in the Highly explosive felsic eruptions, such [Hagen et al.,1993],alongitssouthernmargin Solomon Islands is difficult becausecally imaged of limited mantle root. Melt would over¢ll the generated solely by thermal expansion of ambient [Phinney et al.Moore,1999,2004],andonitscrestand and Schilling, 1973; Dixon and exposuredence of the time sections. and greater globalcrater duedispersal to thermal expansion,as those documented and it would frommantle the North (as opposed to excess temperatures and re- also propagate radially from the crater along frac- sultant dynamic buoyancy characterizing plumes), eastern flank (thisStolper, paper). 1995). What causes the OJP’s than if the SO2 remains in the tropo- and South Atlantic volcanic passive mar- intrabasement reflections remains enigmatic, how- [7] Early Cretaceous basalts ontures Malaita created Island, in the brittle, surrounding litho- relatively minor uplift and subsidence would be Another important factor determin-termedsphere; the Malaita therefore, Volcanic they Group havesphere greater (MVG), to erupt ef- com- in thegins, proximal the Red ocean Sea, basins the Kerguelenassociated Plateau, with emplacement of the greater OJP, ever, because unlike lava flows drilled on volcanic (Fig. 4c). Solid mantle, not involved in the melt- and it would be in isostatic equilibrium with sur- passive marginsing andthe environmental the Kerguelen impact Plateau, of all LIP prise afects monotonous on climate sequence and atmospheric of pillow and chem- mas- and continental fl ood basalts (e.g., Bryan LIPs sive lavas and sills with rare interbedded ing event, sediment would then both rise buoyantly from rounding lithosphere [5,18,69] (Fig. 1b). lavas drilled on the OJPの活動と地球環境変動 and sampled in the beneath and £ow from the sides to replace the A large meteorite impact could result in the Solomon Islands were erupted in a submarine and sedimentary rock (chert and a quite thin mudstone both calcareous andvolume noncalcareous), vacated by the erupting, melted mantle formation of a temporary magma lake. Such a environment [e.g., Michael,1999;Ingle and (Fig. 4d). Because the melt and underlying solid process may have been common on Earth during Coffin, 2004;LIPsRoberge etand al., 2005]. earth’s Furthermore, and environment minor microgabbro, gabbro,mantle and dolerite would ascend dikes adiabatically, with buoyancy heavy meteorite bombardment in Early Archean no drilling has penetrated these reflections, with [Petterson et al.,1997,1999;Petterson,2004]. the deepest in situ penetration of OJP basement Rates of lava effusion for obducted OJP basalts on being 217 m at ODP Site 1185 [Mahoney et al., Malaita Island were high to very high [Petterson et al.,1997,1999;Petterson,2004].The39Ar/40Ar

3of19 Increased Planetary Albedo Heterogeneous Chemistry N2O5 EPSL 6915 9-1-04 Cyaan Magenta Geel Zwart h & OH ClONO2 HNO3 HCl ClO SO2, HCl, Ash, [CH4]

SO2 H2SO4 Warming Stratosphere

Nucleation and Particle Growth Rainout Removal H2O, HCl, Ash Processes Troposphere Cirrus Modifcation Infrared Land Bridges Extrusives Sea Level Gateways Hydrothermal Activity Extrusives Upwelling Continental Crust Oceanic Crust Lower Crustal Body

Lower Crustal Body

Continental Flood Basalt Oceanic Plateau (off-axis)

Figure 5. Complex chemical and physical environmental eff ects associated with LIP formation. LIP eruptions can perturb the Earth-ocean-atmosphere system signifi cantly. Note that many oceanic plateaus form at least in part subaerially. Energy from solar radiation is hv, where h = Planck’s constant, and v = frequency of electromagnetic wave of solar radiation. Modifi ed from Coffi n and Eldholm (1994).

26 156 Oceanography Vol. 19, No. 4, Dec. 2006 and Gondwana (Atlantic and Indian oceans) the histograms for "world total" and "oceanic reported by Rea and Vallier (1983). Their pro- ridges, and oceanic plateaus are shown sepa- plateau" production are very similar. The mid- posed hiatus was centered on the Turanian and rately. The "world total" is the sum of the lower Cretaceous pulse peaked soon after its onset Coniacian stages, now thought to total only 4 three curves, excluding Tethys. The Deccan (between 120 and 100 Ma), after which it m.y., from 90.5 to 86.5 Ma (Harland et al., Traps (Courtillot et al, 1987; White and continued with reduced intensity from 100 to 80 1990). Thus, a very short hiatus may exist dur- Latest pulse of Earth:McKenzie , 1989) at 65 Ma are shown for com- Ma. This decay is partially the result of end-on ing these stages that is averaged out by the histo- parison of a well-known continental flood-basalt ridge subduction in the Pacific as opposed to gram intervals in Figure 1. Evidence for a mid-Cretaceous superplumsequence, but are noet include d in the world spreading-rate variations that cannot be meas- By 30 Ma, the world total and oceanic- total. Clearly, a pulse of ocean crustal produc- ured during the long Cretaceous normal polarity plateau production rates had nearly leveled off tion appears between 120 and 80 Ma in the interval. However, the oceanic plateau volumes at about the same rates seen prior to the mid- R. L. Larson world total that is predominantly the result of also decay in the same fashion. After 80 Ma, the Cretaceous pulse, although it must be remem- Graduate School of Oceanography, University of Rhode Island, Narragansett, Rhode Island 02882 contributions from the Pacific ridges and Pacific world total and oceanic plateau production con- bered that the exclusion of a potential Tethys oceanic plateaus. Maximum spreading rates cal- tinued to decline until about 30 Ma; a secondary ridge system may underestimate Early Creta- culated for individual ridges during the Creta- peak occurred near the Cretaceous/Tertiary ceous and Late Jurassic rates. During the past 30 ABSTRACT INTRODUCTION AND METHODS ceous pulse are about 17 cm/yr. This is, perhaps boundary at 65 to 60 Ma. This subsidiary pulse m.y., Pacific ridge output approximated Gond- ' A calculation of Earth's ocean crustal budget for the past 150 m.y. reveals a 50% to 75% Earth is a huge heat engine, fueled mainly by coincidentally, about the same as the present is mainly a result of increased production rates wana ridge output, while world total and oce- increase in ocean crust formation rate between 120 and 80 Ma. This "pulse" in ocean crust the decay of the radioactive isotopes of potas- world-maximum spreading rates observed at the in the Gondwana ridge system and oceanic pla- production is seen both in spreading-rate increases from ocean ridges and in the age distribu- sium, uranium, and thorium and by release of anic plateau formation rates remained constant. Pacific-Nazca ridge. Thus, individual ridges did teaus (Fig. 1). In particular, it results from the tion of oceanic plateaus. It is primarily a Pacific Ocean phenomenon with an abrupt onset, and the heat of crystallization at the inner/outer core fast spreading rates on the Indian Ocean ridge peak production rates occurred between 120 and 100 Ma. The pulse decreased in intensity notboundary spread . abnormallThe heat yfro mrapidl thisy engin in eth ies dissimid-- INTERPRETATION from 100 to 80 Ma, and at 80 Ma rates dropped significantly. There was a continued decrease Cretaceouspated mainl, buy t durinaveragg the e rateformatios on n thofe oceaniPacificc system associated with the breakup of Madagas- The mid-Cretaceous pulse in ocean-crust from 80 to 30 Ma with a secondary peak near the Cretaceous/Tertiary boundary at 65 Ma. For ridgescrus wert in eth clearle world'y highers ocea.n basins. It is generally car and India that also resulted in the Chagos- formation is evidenced mainly in Pacific ridge the past 30 m.y., ocean crust has formed at a nearly steady rate. Because the pulse is seen Thassumee onsed that otf thithse heamid-Cretaceout energy is produces episodd aet ais Laccadives, the Madagascar Ridge, and the and Pacific oceanic plateau production, began primarily in Pacific oceanic plateau and ridge production, and coincides with the long Creta-suddenearln yan constand is seet nrate in. alHoweverl three ,component the constancs y(Pa of- Deccan Traps flood basalts. relatively suddenly at 120-125 Ma, and de- heat-energy dissipation has been a source of ceous interval of normal magnetic polarity, I interpret it as a "superplume" that originated atcifi c ridges, Gondwana ridges, and oceanic pla- There is no evidence in Figure 1 for the hiatus creased over a long period to about 70 Ma. speculation for decades (Holmes, 1965), and an about 125 Ma near the core/mantle boundary, rose by convection throug h the entire mantle, teaus) of the world total. The general shapes of in mid-Cretaceous volcanism from 95 to 80 Ma Spreading rates during the Cretaceous lie within and erupte形成年代と地球の様相d beneath the mid-Cretaceous Pacific basin. The present-day South Pacific "super- episodically "pulsating" Earth could account for mountain-building episodes, climatic extremes, the present-day range, although average Pacific swell" under Tahiti is probably the nearly exhausted remnant of the original upwelling. How rates were higher. Present-day variation in this superplumAgee stoppe ofd magneti globalc field reversal LIPss for 41 m.y. is a matter of speculation, but it eustatic sea-level fluctuations, and abnormal ac- TABLE 1. OCEANIC PLATEAU AGES AND VOLUMES spreading rates is mainly a function of the avail- probably involved significant alteration of the temperature structure at the core/mantle cumulations of petroleum. The most recent of ability or absence of long subducting slabs to boundary and the convective behavior of the outer core. these pulses can now be confirmed and quanti- fied as a 50% to 75% increase in oceanic-crust provide driving forces for rapid spreading. Thus, productioOceanic n during mid-CretaceouAge s time. The in- Age Volume References it is possible that subduction-zone initiation or plateau (Biostratigraphy) (Ma) (K^KmS) itial suggestion (Larson and Pitman, 1972) of a rearrangement in the Mesozoic Pacific is respon- 35 "I mid-Cretaceous spreading pulse was speculative, sible for the pulse. However, what we know of Brokebutn neRidge-w evidenc1 e o>Turonian=Kerguelen magnetic lineation n map- 90-110 5.19* ODP 754 Mesozoic subduction from the geologic record Caribbean-ping, magneti2 c reversaTuronian-Campanial stratigraphic ncalibration , 75-90 20.41 DSDP 146, 149, 150, 152, 153 of the rim of the Pacific basin is that subduction and ocean-crustal dating allow a more quantita- Caroline Seamounts Miocene-Pleistocene 1-10 5.60 Keating et al. (1984) has been generally continuous since the Jurassic, Chagotives calculatioLaccadivesn of Earth'early Paleocene-earls ocean-crustay Eocenl budgee t 50-60 14.01* ODP 707, 712, 713,715 at least in the American Cordillera and in Japan. Crozefort thPlateae pasu t 150 m.y—r the• iassociation -i'-i- r i 'of i impactsi i Walvi ands Ridg masse LIPSlate Campanian-MaastA KEY IODP. INITIATIVE65-75 6.85 DSDP 525, 527, 528 gj 7 t: D New magnetic anomaly studies were sum- O o < H C-OJ s n w els in the mantle. The source material must also s o M £ £ U c/5 V < 03 > CO (e.g., Scaillet and MacDonald, 2006). i - i- —f extinctions.—r—'—r The most dramatici marizeexampled on worldUnderstanding maps of basement agthee b yformation Lar- of LIPs rise quickly and approximately adiabatically to 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 soNoten e:t Ageal. s(1985 arc mainl) anyd fro ofm magneti the oldesct sedimenlineationt ages bsy a t the base of Deep Sea Drilling Project (DSDP) or Ocean LIP formation may have been respon- is the eruption of the Siberian fl ood ba- constitutes a fi rst-order problem in the lithosphere in order to retain most of its DrillinCandg Prograe et alm . (ODP(1989)) dril. lWherea sites, frosm mandredgey ddetaile rocks, do r from tectonic associations with other features of known age MILLIONS OF YEARS (e.g., Broken Ridge=Kerguelen Plateau). Individual DSDP and ODP drill sites are not referenced in full except for the original heat for massive upper-mantle melting. sible for some of the most dramatic and salts ~ 250 million years ago (Figuresmosareat recents remai 1sites n, foODr PfuturEarth 803Larson, ean analysis dscience, 807 (Le, thg e 131991and world-wid0 Shipboar as such,de Scientifi is a cmajor, Party, 1990) , but their ages can be found in the Figure 1. World oceanic crust production tor past 1S0 m.y. partitioned into Pacific ridges, Gond- Sitedistributio Chapter fonr eacofh magnetisite in thec Initialineationl reports s oisf, thien Deethep Sea Drilling Project and in the Proceedings of the Ocean The pulse in both total production and wana rapidridges changes(Atlantic an ind theIndia globaln oceans) environment., and oceanic plateau ands tha 6),t su whichm to worl coincidedd total. Decca withn Drillinthemain largestg Program, well known. Plateahigh-priority. uWorkin volumesg derive version initiatived fros mo fSchuber thes fore t an thed Sandwel Integratedl (1989). Numbered plateaus are shown in 27 Figure 2. oceanic-plateau building correlates closely with Traps volume shown for comparison with continental flood basalts. Each point on these noncu- charts and various tectonic models were used by mulativBetweene histogram ~ s145 represent and ~s 50volum millione of ocea yearsn crus t produceextinctiond in 1 m.y of. "N plantso magneti andc reveranimals- •Volume in thes reduced Oceanfrom thei rDrilling total platea uProgram volumes by (IODP) removing a (Coffi6.5 km thicknesn, s of normal oceanic crust the long period of normal magnetic polarity in becausKomine thesz e (1984plateau, sTabl formee d 1a)t spreadinto compilg ridges.e table s of sals" representago, the sglobal long Cretaceou oceans swere interva characterizedl of normal polarit y enclosegeologicald by magneti recordc (e.g.,anomalie Rennes 34 and Basu, McKenzie, et al., 2001). Strong evidence the mid-Cretaceous, and the coincident onset of to M0. Geologic time scale from Harland et al. (1990). ridge+Volume cress tdouble lengthd tso accounvs. spreadint for potentiag ratel stwi fon rplateau the s on the Farallon plate that have been subducted. by chemical and isotopic variations (es- 1991). Ninety percent of all species be- exists that many LIPs manifest a form both these phenomena is especially striking. As GEOLOGY, v. 19, p. 547-550, June 1991 547 pecially in C and Sr isotope ratios, trace came extinct at that time. On548 Iceland, GEOLOGY , June of199 mantle1 dynamics not clearly related metal concentrations, and biocalcifi ca- the 1783–1784 eruption of Laki provides to plate tectonics; the processes involved tion),大絶滅イベントと関連？ relatively high temperatures, high the only historical record of the type of in their formation are critically impor- relative sea level, episodic deposition of volcanism that constructs transient LIPs tant for understanding both mantle and Extincition of marine genera & eruption of LIPs

Figure 6. Extinction of marine genera versus time (continuous line, blue fi eld), modifi ed from Sepkoski (1996), Siberia Deccan compared with eruption ages of LIPs (red columns). Strong temporal cor-

Ontong Java relation between LIPs and mass extinc- North Atlantic Columbia River Central Atlantic tions suggests causality. T ree of the Karoo and Ferrar Karoo Emeishan, Panjal largest mass extinctions—the Permo-

Ethiopia and Yemen Triassic, Triassic-Jurassic, and Creta- Paraná and Etendeka ceous-Tertiary—coincide with eruption Caribbean, Madagascar ages of the Siberian Traps, the Central Atlantic Magmatic Province, and the Kerguelen/Broken, Rajmahal Kerguelen/Broken, Deccan Traps, respectively. Modifi ed from White and Saunders (2005). See Figure 1 for locations of LIPs.

28 Oceanography Vol. 19, No. 4, Dec. 2006 157