When and how did plate begin? Theoretical and empirical considerations


Geosciences Department, University of Texas at Dallas, Box 830688, Richardson TX 75083-0688, USA (email: [email protected]) is the horizontal motion of ’s thermal boundary layer () over the con- vecting () and is mostly driven by lithosphere sinking in zones. Plate tectonics is an outstanding example of a self organizing, far from equilibrium complex (SOFFECS), driven by the negative buoyancy of the thermal boundary layer and controlled by dissipa- tion in the bending lithosphere and viscous mantle. Plate tectonics is an unusual way for a silicate planet to lose heat, as it exists on only one of the large five silicate bodies in the inner . It is not known when this mode of tectonic activity and heat loss began on Earth. All silicate planets probably experienced a short-lived . After this solidified, stagnant lid behavior is the common mode of planetary heat loss, with interior heat being lost by and “hot spot” and shallow intrusions. Decompression melting in the hotter early Earth generated a dif- ferent lithosphere than today, with thicker and thinner mantle lithosphere; such litho- sphere would take much longer than at present to become negatively buoyant, suggesting that plate tectonics on the early Earth occurred sporadically if at all. Plate tectonics became sustainable (the modern style) when Earth cooled sufficiently that decompression melting beneath spreading ridges made thin oceanic crust, allowing oceanic lithosphere to become negatively buoyant after a few tens of millions of . Ultimately the question of when plate tectonics began must be answered by informa- tion retrieved from the geologic record. Criteria for the operation of plate tectonics includes ophiolites, blueschist and ultra-high metamorphic belts, eclogites, passive margins, transform faults, paleomagnetic demonstration of different motions of different cratons, and the presence of diagnostic geochemical and isotopic indicators in igneous rocks. This record must be interpreted individually; I interpret the record to indicate a progression of tectonic styles from active Archean tectonics and magmatism to something similar to plate tectonics at ~1.9 Ga to sustained, modern style plate tecton- ics with deep subduction―― and powerful ―― beginning in Neoproterozoic time.

Plate tectonics, subduction, Precambrian,

In June 2006, a Geological Society of America Penrose motions and and is the principal way conference in Lander, Wyoming, USA, convened to Earth cools. The hotter early Earth may have had a discuss when plate tectonics began. Here is how the weaker and less dense lithosphere and so had a different meeting was described: “Earth is the only planet with style of . This theoretical background plate tectonics, and it is controversial why and when this is essential to provide context for empirical (field and began. Some argue that plate tectonics already operated lab analyses) results. Demonstrating that plate tectonics in Archean time, whereas others argue for a much later operated at any given time requires evidence for sub- beginning. Numerical experiments show that Earth’s duction and independent plate motions and rotations. thermal boundary layer, the lithosphere, controls mantle convection. We now understand that the sinking of dense Received September 25, 2006; accepted November 28, 2006 doi: 10.1007/s11434-007-0073-8 lithosphere in subduction zones is responsible for plate Supported by the US National Science Foundation (Grant No. 0405651)

578 R. J. STERN Chinese Science Bulletin | March 2007 | vol. 52 | no. 5 | 578-591 quences andarc rock assemblages. margin passive as such tectonics se- of plate diagnostic assemblages) and secular distribution of associations and diamond- andcoesite-bearing ultrahigh-pressure marks ofsubduction (ophiolite styles ofdeformation, temporal the distribution ofhall- topic and trace elem zones and ocean ridges, including , iso- as proxies for plate motion or operation of subduction measurements laboratory-based tions include the results of a wide range of field- and may have affected this . Empirical considera- powers the plates, and how Earth plate tectonicprocessesarepresentedanddiscussed. this explored. Finally, someof the geologic evidence of to those ofits neighbors is noted, and the implications of Then the unusual of Earth’s tectonic style relative modern plate tectonic/subduction system isdescribed. when plate tectonics began are explored below. First the understanding for considerations most pertinent The tectonic style must have evolved as the planet cooled. began late in Earth history, ~1Ga oryounger My conclusion that themodern style of plate tectonics dallas.edu/~rjstern/PlateTectonicsStart/presentations.htm. 2], and meeting presentations can be found at http://ut- meetin about the further details opinionsof rangeregarding when plate tectonics began; and interact across Earth’s nearlyspherical surface Plate tectonics describes how lithospheric sectorsmove the platesmove? Whatisplate tectonics?Whatmakes 1 considerations include Earth quires a creative and interdisciplinary effort. derstanding Earth; progress toward resolving this re- one of the most important unresolved problems in un- Understanding when and why plate tectonics began is what forces drive the plates. Early on, the drivingforce description; the theory was formulated without knowing ity. Plate tectonics isapowerfulbutpurely kinematic reference frameswitharotation pole and angular veloc- Plate motions aredescribed using absolute orrelative lier ference concluded that plate tectonics began much ear- nority opinion. Most geoscientists at the Penrose con- Sixty-one geoscientists from geoscientists Sixty-one [1, 4 ] , butthere wasgeneral agreement that Earth’s ent compositions of igneous rocks, R. J. STERN Chinese Science Chinese Bulletin R. J.STERN ’ g can be foundatrefs. [1, s thermal history,s thermal what 14 nations shared awide s, blueschists, eclogites, s, blueschists,eclogites, and analyses that serve ’ ” s progressive cooling

Theoretical [3]

is ami- [5] . | March 2007 | vol. 52 | no. 5| | no. |March2007vol. 52

fecting the thick lithospheric keels of west-moving plates.There may besome basal drag af- mid-ocean ridge, where it consists only of basaltic crust. Lithosphere is buoyant when itformsspreading by ata plates mid-ocean ridgeseparating the Australian andAntarctic domains meet and sink into the mantle beneath the (AAD), whereIndian and Pacific asthenospheric mantle spheric flow isthe -Antarctic Discordance ple of how independent are plate motions andastheno- of -bearing plates. but this flow impedes as well as encourages the motions growing (Atlantic-type) or shrinking (Pacific-type) away from thebasin, depending whether theocean is around the plate margins the trench and sinking in subduction zones (Figure1(a)) lithosphere subsiding from ridges () towards a ultimately result from thenegative buoyancy ofthe asthenospheremoves towards the ridge overlying plate. Beneath the interior of an oceanic plate, asthenosphere generally does notflow parallel to the sphere Some workersinfer an“eastward drift” of astheno- the lithosphere and has little effect on its motions motions. First, the asthenosphere isweakly coupled to tions ofrelationships between asthenospheric and plate namic considerations (outlined below) orfrom observa- geody- by modern supported isnot motions ers plate organize overallmantle convection and dominate theoceanbasins. allowgravitationally unstable oceanic lithospheretoform not have happened until the Earthcooled sufficientlyto subduction zones mainly powered by sinking of the oceanic lithosphere in our presentif understanding trast, Incon- faster plates. or more have would thenosphere, hotter, with a lower andrapidlymixing as- the reasonable conclusion that the ancient Earth, being the base of the plate”. This consideration, if true,leads to “Convective motion in the asthenosphere applies drag to textbook says that an important force forplate motions is taught to Introductory classes. One popular This is a deeply embedded idea that is 1(b)). still (Figure asthenosphere, the weak mantle beneath the lithosphere mistakenlywas thought toresult from convection of the The hypothesisthatasth It is now clear that the plat the It isnowclear that [13] [9] but east-moving plates do not move faster than . ―is correct, then plate tectonics could 578-591 enospheric convection pow- convection enospheric The most spectacular exam- asthenosphere flows into or es drive themselves asthey

― [1 4 thatare theplates ,15] . Plate motions [7] , whereas [6] [10 . The ― 579 12] [6] [8] . . ,


Oceanic crust alone is less dense than asthenosphere, but with less confidence down to about 1700 km but it is not the mantle part of lithosphere is colder and therefore clear yet whether such slabs sink all the way to the denser than asthenosphere. Oceanic lithosphere thickens core-mantle boundary[18]. Sinking of the lithosphere in by conductive cooling of the , which pro- subduction zones controls plate motions in two ways, by gressively increases bulk lithosphere density (crust plus directly pulling on the plate (slab pull) and by entraining mantle) as it ages (Figure 2). This results in a buoyancy the surrounding mantle to descend with it (). crossover, when oceanic lithosphere becomes denser Even detached subducted slabs help drive the plates, by than the underlying asthenosphere. For the modern Earth, entraining mantle that sinks with the slab (Figure 1(a)); this “buoyancy crossover time” happens relatively this is known as “slab suction”. Observed plate motions quickly and today lithosphere is denser than astheno- are best predicted if slab pull and slab suction forces sphere after 20―40 million years[16,17]. each account for about half of the subduction-related [19,20] driving force . The plates do not move parallel to the dip of the Wadati-Benioff zone, but also have a vertical component that results in trench roll-back, as required by shrinking of the Pacific Ocean[21]. The sinking of lithosphere in subduction zones is re- sisted by dissipation forces in the lithosphere and by viscous resistance of the asthenosphere (Figure 1(a)). Lithospheric resistance includes bending as plates de- scend into subduction zones (~40% of total dissipa- tion[22]), faulting, and sliding resistance at the outer trench [23]. Asthenospheric viscosity and resistance increases with mantle depth. Physical equilib- rium requires a balance of driving and resisting forces, resulting in a “plate tectonic speed limit”, which for the Cenozoic is approximated by the northward movement of India at 170 mm/a during Paleocene time, ~60 Ma. In summary, there is a strong consensus among geo- dynamicists that plate tectonics and subduction drive mantle convection, encapsulated as “Top-down Tecton- ics”[24]. This consensus warrants re-examining when in the history of the cooling Earth these forces were suffi- ciently great to overcome the resisting forces and estab- Figure 1 Summary diagram of plate-driving forces. (a) Modern view of plate-driving forces; (b) obsolete view of plate-driving forces. In spite of lish plate tectonics. First, we briefly examine the other the broad acceptance of (a) by the geodynamic community, (b) is consid- silicate planets to the point that plate tectonics is ered by many to be important. See text for further discussion. unusual in the space dimension, and so may also be un- usual in the time dimension. The negative buoyancy of aging oceanic lithosphere provides the potential energy that powers plate tectonics. 2 Plate tectonics is an unusual way for a This potential energy is partitioned between ridge push silicate planet to cool and slab pull and suction. Ridge push provides about 10% of the driving force, and descent of lithosphere in The geoscientific subdiscipline of tectonics is concerned subduction zones provides the rest; because of this, with understanding deformation of the outer conductive Earth’s present tectonic style is more aptly described as shell―the lithosphere―of Earth. This deformation oc- “Subduction Tectonics”[3]. Modern plate tectonics is curs at all scales, forming mountain chains and characterized by very deep subduction. Tomographic mid-ocean ridges as well as deforming mineral grains. images suggest that some subducted slabs can be traced Ultimately, tectonic style reflects how the planet loses its with some confidence down to 1100―1300 km, and internal heat, and plate tectonics is only one mode of

580 R. J. STERN Chinese Science Bulletin | March 2007 | vol. 52 | no. 5 | 578-591 examination. ing style of the largest silicate planet warrantscloser smaller silicate bodies. Nevertheless, the unusual cool- planet bodies, itmay not be surprising that the largest silicate mode andlarger bodies retain heat longer thansmall Because tectonic styles fundamentally express cooling episodes beginning brief to largelymolten (magma stage), butthese ocean were deep intheplanet. tive lid and high concentrati interior to heat up, particularly ifit has a thick conduc- gressively cool, but it isalso possiblefor aplanetary Earth lose more heat than they produce and thus pro- tectonics plate and zones subduction has Earth know (, , Earth, , and Mars), only planetary cooling. Ofthe 5 largest silicate bodies that we to lose 4.2×10 hydrothermal activity). Every yearthe Earth is estimated and advection (shallow intrusions, flows, and interiors produces heat. Interior heat is lost by conduction cially Th, and K, presence of now-extinct radionuclides (espe- planetaryaccretion, U, of radioactive greater abundance the inevitable result of tremendous early heat sources: mid-ocean ridges about 10 TWmay belost by hydrothermal activity at the surface thermal boundarylayer(lithosphere) and and perhaps the Sun’s T-tauri phase Early in their histories, l al the planets histories, werepartially their in Early Planets lose heat even as as even heat lose Planets ― 26 Al), continued bolide impacts,formation,core Earth ―has a different tectonic style from other 13 W of heat: 32 TW is conducted through [27] . Tectonically active planets like iue2 Figure 2 [28] R. J. STERN Chinese Science Chinese Bulletin R. J.STERN . The magma. oceanstage was ons of radioactive elements radioactive decay intheir Development ofnegativebuoyancy withincr [27 ― 29] . A magma . A [25,26] . | March 2007 | vol. 52 | no. 5| | no. |March2007vol. 52

face in order to compensate for magmamoving to the sur- sink― nant lid mode, where lithosphere nevertheless must cially do not understand the magmatically active stag- silicate planets, butthere aremany variants.We espe- planet a single lithospheric plate encompassing the entire lid forms, the bodymust cool differently. Stagnant lid lose heat, but oncethis ocean solidifies a conductive and ocean is the most effective wayforsuch ahot planet to ewe lt etnceioe. between platetectonicepisodes. magmatic activity, either as aprelude to orinterlude most certainly accompanied by abundant tectonic and 4.5 Ga history. Stagnant lid perienced onemore or stagnant lid episodes during its cate planets today, it seems likely that Earth also ex- that stagnant lid tectonics is thedominantmode ofsili- times because the planet is toosmall andcold. Given the underlying asthenosphere istoo viscous, and some- because the lithosphere is too strong,sometimes because sometimes lithosphere, buoyant reflects this sometimes change multiplechange times lifeofacoolingplanet overthe small bodies have thick, strong stagnant lids that are surface, which isaveryinefficient waytocool. These These bodies lose heat only This is the present situation ofMercury andmoon. our and magmatically inactive, ei alone; when this happens, the planet becomes tectonically Ultimately, it is a planet’s fate to cool by conduction The mode of heat loss and thus tectonic style may thustectonicstyle heatlossand modeThe of easing ageofmodern oceaniclithosphere. [30,31] ―seems tobethe dominant mode heat lossforthe perhaps by delamination . The reasons for stagnant lid behavior vary: 578-591 tectonics on Earth wasal- onEarth tectonics by conduction throughthe ther dead or hibernating.

― into the deep interior,

581 [25] ― .


very stable. It is possible for the interior of a stagnant lid planet to heat up, if the conductive lid is thick enough and the abundance of radioactive elements is great enough. The mantle potential (Tp)―the temperature of adiabatically decompressed mantle―will increase in this case, possibly to the melting point. In this scenario, an inactive stagnant lid planet may become magmatically active again; this may be the cause of periodic resurfac- ing events, as inferred for Venus, or for the formation of lunar mare. Stagnant lid behavior is thus a very stable mode of heat loss, characterizing dead (Moon, Mercury) as well as active (Mars, Venus) planets. In general, heat loss through a stagnant lid increases with increasing Tp, because the conductive lid thins with hotter interior . As the lithosphere thins, conductive cooling is supplemented by convective heat loss due to periodic advective breakthroughs such as hot spot/heat pipe volcanism or lithospheric delamination. Both delamination and hot spot volcanism are favored by higher Tp for a stagnant lid planet, and larger, warmer planets like Venus have unstable stagnant lids. The Mar- tian stagnant lid is moderately unstable, disrupted by the large Tharsis volcanoes and the Valles Marineris . Figure 3 Diagram showing evolution of cooling silicate bodies through [32] Linear magnetic anomalies and the presence of rocks three modes of planetary heat loss―magma ocean, plate tectonics, and similar to [33] in the ancient southern highlands stagnant lid―as a function of Urey Ratio (heat production/heat loss) and Tp (mantle potential temperature) modified from ref. [25]. (a) Stagnant lid of Mars suggest that plate tectonics may have operated encompasses a wide range of magmatic and tectonic styles for planets in the ancient past, but no more. Earth’s twin, Venus, has with only one lithospheric plate, from that of essentially ‘dead’ bodies (Mercury, Earth’s moon) to magmatically and tectonically active planets a stagnant lid that is very unstable. It is tectonically and such as Venus; this is simplified as stable (cold) and unstable (hot) stag- magmatically much more active than Mars, with periodic nant lids in (b). Dashed line shows two possible thermal histories for [34] silicate planets, one with four tectonic mode transitions, from magma and spectacular resurfacing events . Venusian resur- ocean to stagnant lid to plate tectonics to stagnant lid, the other remaining facing events may mark times of large-scale lithospheric as stagnant lid throughout planetary history. (b) Same as (a) but showing relative thermal states and tectonic styles of the terrestrial planets and failure, when volcanism is stimulated by sinking of por- moon. Dashed line illustrates thermal evolution of a planet with a pro- tions of the lower lithosphere. Similar lithospheric failure longed transition from unstable stagnant lid to plate tectonics, perhaps like the Proterozoic Earth. on Mars probably caused that planet’s hemispheric dis- continuity, with a densely cratered, ancient lithosphere thermal conditions exist. We can be confident that preserved in the south and a much younger lithosphere in magma ocean and stagnant lid tectonic scenarios exist, the north[35]. although we are only starting to investigate the range of These tectonic possibilities are explored in Figure 3, stagnant lid behaviors. These seem to range from very modified after ref. [25] to show some of the modes of strong, stable lids to unstable lids that are affected by planetary heat loss discussed above as a function of intense magmatism and deformation. The three funda- Urey ratio (relative importance of heat production to mental tectonic styles―magma ocean, stagnant lid, and heat loss) and Tp. Earth’s Urey ratio is estimated to range plate tectonics ― are further complicated by interior from 0.16[36] to 0.65―0.85[37], with most researchers processes that are independent of tectonic style, includ- accepting a value of ~0.4[38]. These curves show only ing mantle plumes, large igneous provinces, and litho- some simple possibilities, intended to demonstrate that spheric delamination[39]. It is not clear where some pro- plate tectonics can only occur when appropriate mantle posed pre-plate tectonic deformation styles fit in this

582 R. J. STERN Chinese Science Bulletin | March 2007 | vol. 52 | no. 5 | 578-591 lid ratherthanproto-platetectonics. these may bemanifestations ofa very unstable stagnant nieyta hs hudpritfrtelf ftepae. unlikely thattheseshouldpersistforthelifeofplanet. existed onEarth,butthesec might platetectonics have how longconditionsfavoring We tosubduct(Figure3). was toobuoyant donotknow generated thickoceaniccrustsothatlithosphere decompression occurred because Trench lock. oftrench tonics couldnotoccurbecause lock act asplateboundaries. inthepast,platetec- Sometime mid-oceanwill nolonger andthe ridges decompression melt coolto byadiabatic mantle becomes too when the plate tectonic regime willbeshutdownbyridgelock, Earth’s future, the effective In mode ofplanetarycooling. evolution, whereithappensatall,becauseissuchan only shortintervalsinplanetary befavoredforrelatively interior. coolthe rials to Nevertheless, plate tectonics may mate- subductionzonessame time bringcoldsurface that ridges efficiently deliversinteriorheattothesurface atthe mid-ocean at discussed above.Thisisbecausespreading reflected byestimatesEarth’s of heat loss,as ratio Urey Prigogine et al. TheSOFFECSconcept (SOFFECS). ganizing far from equilibrium system complex Plate tectonics isaspectacular example ofaself-or- from equilibriumcomplexsystem Platetectonicsasaself-organizingfar 3 diagram, suchaseclogitic “drip” tectonics difference between the mantle potential temperature and date and respond to heat loss. Because of the tremendous peculiarities of how magnesiasilicate solids accommo- 1). Plate tectonics is a very non-linear response to the tion zones and isviscously resisted by the mantle (Figure subduc- in bends the plate when occurs dissipation and system that draws on the great store of interior energy, tonics is now Earth’s way to export entropy, it is an open these requirements andthus isanSOFFECS.Platetec- crustal delamination Anderson it isactually organizedfrom above,by surfacetension. heatedstovean exampleself-organization, on aas of but dissipation, and a mechanism forexportingentropy. non-linear interconnectedness ofsystem components, system, alarge steady outside source of matter or energy, self organization requires an open physical or chemical Plate tectonicsisaveryeffectivemode of planetary [ 4 3] noted thatplate tectonicssatisfiesalso [ 4 2] used convection in a pan of ― mantle refertilization recycling mantle refertilization R. J. STERN Chinese Science Chinese Bulletin R. J.STERN onsiderations deem it very it very deemonsiderations of Earth’smantle hotter [ 4 2] considers that [ 4 0] orcatalytic [ 4 1] ;

| March 2007 | vol. 52 | no. 5| | no. |March2007vol. 52

crust ter Archean Earthwould generate thicker oceanic spreading and decompressionmantle melting of the hot- sition (and thus crustal density) iscontrolled by its (Figure 2).Oceanic from spreadingridgeinto thesubductionzone. founder, and bend, yet strong enough to remain coherent requires that lithosphere be weak enough to rupture, essary but not sufficient, because plate tectonics also gravitationallyunstable. Lithospheric instability isnec- until a significant fraction of the lithosphere became tectonicsbegan. Plate tectonics could not havestarted to understandover thepastweGawhen are if 4.5 plate how lithospheric density has changed as Earth cooled sphere drives plate motions today,must we understand Given that the negative buoyancy of old oceanic litho- plate tectonics Physical requirements forbeginning 4 htw e oa. what weseetoday. tectonics experienced multiple phases before evolving to blethat the giganticEarth SOFFECS known asplate reorganizing rapidly andunpredictably. seems It inevita- andstop, start they averydisequilibriumsystem, reflects point:BecauseSOFFECS this emphasizes SOFFECS temporally. The realization that plate tectonics is an the notion that plate tectonics is also likely to be limited the planets, and the present tectonics isan important way thatEarth is unique among special feature of an unusual Instead weshould appreciate that plate tectonics is a Earth” demonstratethe limitations uniformitarianism. of impact and extinction and the Neoproterozoic “Snowball Unique eventsand episodes tion of the history and mystery ofatruly dynamic Earth. plication ofthis principle impedes objective reconstruc- anism overly strict application of the principal of uniformitari- given amount mantleof cause thisdetermines how muchmelting accompanies a higher thantoday tial temperature forthe Archean Earth was significantly etnc sb a h ags OFC nti lnt tectonics isby farthelargest SOFFECSonthisplanet. stable, farfrom equilibriumsy that ofEarth’s exterior, our planet is an inherently un- Oceanic lithosphere density islargely determined by The SOFFECSconcept encourages us to resist an [ 4 ― 6] andthis would result in more buoyant litho- “The present isthe key tothe past”. Blind ap- [ 4 5]

, perhaps 300 by , perhaps 578-591 crustal thickness andcompo- uniqueness in space supports like the Cretaceous/Tertiary planet. We know that plate

stem. Inthis sense, plate [ 4 4 ] . The mantle poten- ― 500 ℃. Seafloor T p be- 583


sphere. It should have taken much longer for this litho- viscosity of the mantle[51,52]. Water also is needed to sphere to become negatively buoyant than it does at make serpentine which may be key for weakening present, making it very difficult to start subduction. lithospheric mantle. Certainly the abundance of water on Another question concerns when plate tectonics can Earth favors plate tectonics but only if lithosphere is be sustained. Because conductive cooling and thickening gravitationally unstable. is a function of age, oceanic lithosphere must ultimately become gravitationally unstable relative to astheno- 5 Geologic criteria for recognizing plate sphere, although for a hotter early Earth this may have tectonics taken many tens or even hundreds of millions of years, compared to 20―40 million years today. If the buoy- The theoretical considerations outlined above motivate critical examination of the geologic record for evidence ancy crossover time was much greater than it is today, of the earliest preserved record of plate tectonics. To do negatively buoyant lithosphere should have been re- this, we must first determine what are the most reliable moved by subduction much more rapidly than it could indications of plate tectonic activity, especially those have been produced, causing subduction to stop or be- that are likely to be preserved. We must recognize that come episodic. Today the mean age of oceanic litho- some evidence has been removed by surficial sphere is ~100 Ma[47], much older than the buoyancy and by tectonic erosion at subduction zones[53], or oblit- crossover time of 20―40 Ma, so that there is an inex- erated as a result of deformation or metamorphism. The haustible supply of negatively buoyant lithosphere and geologic record is surely incomplete, and we can never subduction can be sustained indefinitely. Earlier in Earth know how incomplete it is. Nevertheless, constraints history, the crossover time was surely greater so that about when plate tectonics began are at least partially subduction would have quickly exhausted the supply of preserved in the geologic record. Given these uncertain- negatively buoyant oceanic lithosphere. Subduction in ties, any indications for plate tectonics beginnings must this situation would not have been sustainable; it may be regarded as a minimum constraint; all we can say is have occurred episodically, after a protracted period of that plate tectonics began no later than the age of the oceanic lithospheric cooling, or it may not have hap- oldest reliable evidence. pened at all. We should also be conservative in establishing criteria Considerations of lithospheric density alone cannot for the operation of plate tectonics, avoiding especially resolve when plate tectonics began. The lithosphere circular reasoning. Care should be taken not to confuse must also be ruptured so that asthenosphere can rise observations with interpretations (e.g., the observation is above it, allowing the lithosphere to sink. The great pillowed with Nb depletions, the interpretation is strength of the lithosphere, which increases with age and formation of a back-arc basin). The mere existence of thus gravitational instability, is a fundamental problem , igneous and metamorphic rocks, and [48] for all models of subduction initiation . Many re- deformation of a certain age should not be taken to in- searchers conclude on this basis that subduction zones dicate that these must have been produced by plate tec- form where the lithosphere is already ruptured, such as tonic processes. We know that Venus and Mars do not [49,50] at transform faults or fracture zones . The transition have plate tectonics, yet they experience a lot of igneous from stagnant lid to plate tectonics also requires litho- activity and deformation, as should have the pre-plate spheric rupture to nucleate subduction, but it is not clear tectonic Earth. Presumably the pre-plate tectonic Earth what this would have been. Perhaps a meteorite impact was dominated by unstable stagnant lid tectonics and or major episode of delamination caused the initial rup- magmatism dominated by magmatism and de- ture that ultimately evolved into the first subduction lamination of the lower crust and mantle litho- zone. sphere[39,41,54]. Water is another important consideration. The fact In the following discussion, some of the most robust that most active plate boundaries on Earth are underwa- evidences for the operation of plate/subduction tectonics ter may be why we have plate tectonics and other plan- are presented and discussed: ophiolites, blueschist and ets do not. Water weakens rocks, lowers the melting ultra-high pressure metamorphic belts, eclogites, passive point and lowers the strength of the lithosphere and the margins, transform faults, paleomagnetism, igneous

584 R. J. STERN Chinese Science Bulletin | March 2007 | vol. 52 | no. 5 | 578-591 mlcda h pems nto ap tc. emplaced astheuppermost unitofanappestack. erosion, becauseby be the disrupted ophiolite. Ophiolites may be readily removed for inferring that a disrupted mafic-ultramafic suite is a containing Cr-rich spinels are aminimum requirement patterns and associated with harzburgitic ultramafics Pillowed tholeiitic basalts all of the components should be present, if disrupted. plate tectonics should include fragmentary ophiolites but zolite) arerare. Thesearch for the ophiolitic evidence of tectonized ultramafic residues(harzburgite and lher- , pillowed basalts, sheeted dikes, gabbros, and timing. lines indicatelesser degrees ofconfid lines representmore ambiguous indications thatare welldated. Dashed History. Solidlinesindicatestrong indicationsthatare welldated, thin Figure 4 start ofplatetectonics. of thelines of evidence suggest different times forthe discussed below. It is perhap dence aresummarized inFigur also when it first becomes common. These lines of evi- only when the assemblage or property first appeared but not canbe considered, evidence lines of geologic these , Th isotopes. and and emplacement settings forophiolites require importantfor thepresent discussion becausealltectonic ophiolites form iscontroversial; fortunately this is not convergence to emplace this floor spreading to form thecrustal section and plate lithospheric motion expected from plate tectonics: - plate tectonic activity. Ophiolites manifestmodes two of placed onthecontinents and (i) Ophiolites are sections of oceanic lithosphere em- Criteriaback through for Earth theoperationof platetectonics [50] . Complete ophiolites, with pelagic R. J. STERN Chinese Science Chinese Bulletin R. J.STERN with concave-downward REE

st preserved examples are ence abouttheindicator and/or its s not surprisingthatsome onthe continent.Where subductionforformation ere are different waysthat are reliable indicators of e 4 and their significance

| March 2007 | vol. 52 | no. 5| | no. |March2007vol. 52

synonymous with “B-type” UHP terranes tive subduction zones mélanges andisconfirmed studiesofac- subduction by ancient with association their on based is zones perception that blueschist only forms in subduction ported in western China re- wasalso age Neoproterozoic of apparent Blueschist to exist in South China Somewhat older (ca.940 Ma) blueschist was suggested waits confirmation (may haveformed inthe Paleozoic). challenged to have been subducted 15 subducted been have to those ofthe Franciscan and Sanbagawa terranes, appear returning Jormua ophiolite of Finland thought to be the oldest ophiolite oldest bethe to thought sode ofsubduction tectonics began, but this is not the edged as important evidence for whenthe modern epi- pre-Neoproterozoic record would bewidely acknowl- absence the that think and exhumed in pre-Neoproterozoic time. One might preserved somewhere on Eart rocks inS. preservation of ~3.2 Ga high zones, they are not likely to be removed byerosion. The Because blueschists are preserved in subduction absence does notseem tobea preservation problem. that pre-Neoproterozoic blueschists areunknown.This fied this important occurrence. It is widely recognized widely distributed,in West Indiaand Africa Neoproterozoic time, ca. 800 acknowledged that the oldest blueschists date from sult of high- asare- forms which amphibole, sodic abundant taining are not found in very ancient rocks ~2.0 Ga Purtuniq ophiolite ofCanada characteristic ofPaci Sea-like oceanicrift. tics and are interpreted to have formed inasmall Red metabasalts areenriched andhave noSSZcharacteris- show “suprasubduction-zone” (SSZ)affinities. Jormua proterozoic and younger ophiolites, which generally show interesting differences frommore common, Neo- that these are about the same age. Even these ophiolites The 2.5 Ga Dongwanzi ophiolite of NE China was (ii) Blueschist is a metamorphosedmafic rock con- It hasbeen known forahalf-century that blueschists tothesurface [1,56] P , low- . The oldestconvincing. ophiolites are [66] T indicates that blueschists would be [61 metamorphism fic-type orogenic belts [65] [60] ― 578-591 63] , but no further study has veri- . [59] . The best-studied blueschists, . Thebest-studied [58] but its metamorphic age of blueschistsfrom the ; it isprobably significant

h if they were produced ― ― /low P/low 70 km deep before before kmdeep 70 700 Ma.Theseare [6 [55] 4 [59] ] . Itis now widely , but ithasbeen, [57] . Blueschists are T metamorphic metamorphic and 1.96 Ga 1.96 and [59] andare [60] . The [59] 585 .


case. Instead, many geoscientists think that the absence of pre-Neoproterozoic blueschist indicates that the pre-Neoproterozoic Earth was hotter so that blueschist was not generated in subduction zones. Surely the early Earth had a somewhat higher mantle potential tempera- ture but this may not have affected the thermal structure of subduction zones and thus whether or not blueschist was produced. The thermal structure of subduction zones mostly reflects the age of the subducted litho- sphere and the convergence rate[67], not mantle potential temperature. This controversy might be resolvable by developing robust geodynamic models for subduction zones in a hotter Earth. (iii) Ultra-high pressure (UHP) metamorphic terranes are important indicators of ancient subduction zones.

These form when continental crust is subducted to depths >100 km and return to the surface. Metamorphic Figure 5 Plot of apparent thermal gradients versus age of peak P-T, for the three main types of metamorphic belts. Open circles: Ultrahigh tem- assemblages in UHP terranes include coesite and/or perature granulite metamorphism (G-UHTM) belts; gray circles: Medium diamond and indicate peak metamorphic conditions of temperature eclogite-high pressure granulite metamorphism (E-HPGM) ― ℃ ― [68] belts, and filled circles: lawsonite blueschist-eclogite metamorphism and ~700 900 and 3 4 GPa or more . In contrast to ultrahigh pressure metamorphism (HPM-UHPM) belts. From ref. [72]. blueschists, the significance of which has long been ap- preciated, UHP terranes have been recognized for only form when continents rift apart and a the last 15 years or so, and new localities will surely be long-lived ocean basin forms outboard of the continent. found in the future. The oldest, reliably dated UHP lo- They take tens to hundreds of millions of years to form, cality is in Mali, where coesite-bearing gneiss was allowing hundreds of meters of thermal subsidence that metamorphosed at ~620 Ma[69]. The oldest diamond- creates even greater accommodation space for sediments. bearing UHP terrane is found in Kazakhstan, where The proximity of passive margins to continental sedi- diamond- and coesite-bearing paragneiss of the Kok- ment sources such as rivers and glaciers ensures that chetav massif was subducted >120 km deep at ~530 they will contain enormous thickness of sediments. Ma[70]. The first evidence for deep subduction of conti- These sedimentary sources are built on very different nental crust is thus found in late Neoproterozoic and crusts: continental crust on the landward side, oceanic early Cambrian rocks. crust on the seaward side, and transitional crust beneath. (iv) Eclogites are often thought to reflect plate tec- Because passive margins are associated with thick con- tonic processes but there are many ways to generate tinental crust, they are difficult to subduct entirely and garnet-clinopyroxene metamorphic and igneous rocks. so can be expected to be partly preserved in ancient su- Lawsonite-bearing eclogites clearly manifest metamor- ture zones. Examination of the geologic record indicates phism in a subduction zone but these are only found in that the oldest passive margins date to ~2.7 Ga, but be- Phanerozoic terranes[71]. Medium T-high P eclogites and come common first about 2.0 Ga, again about 0.9 Ga granulites are known from as early as Neoarchean time, (Dwight Bradley, pers. comm. 2006). but these represent significantly higher geothermal gra- (vi) Transform faults ― plate bounding strike-slip dients than are found in modern subduction zones[72]. faults like the San Andreas ―are one of the three Figure 5 summarizes metamorphic styles through Earth fundamental types of plate boundaries, and the identifi- history, showing that the low T, high P metamorphic style cation of ancient transform faults clearly indicates op- characteristic of modern subduction zones has not been eration of plate tectonics. Sleep[73] argued that the de- observed for rocks older than Neoproterozoic. velopment of major strike-slip faults at a given time in- (v) The record is a powerful constraint. dicated rigid plates and weak plate boundaries and thus Passive continental margins such as those around the should be taken as evidence for plate tectonics.

586 R. J. STERN Chinese Science Bulletin | March 2007 | vol. 52 | no. 5 | 578-591 must beovercome. tectonics begin?”, but the aforementioned complications important foranswering the question of“When did plate et al. Cawood metamorphism. and alteration with associated printingof primary magnetization remagnetization by poles, the possibility oftruepolar wander, andthe over- not always been a dipole approximating the geographic result from thepossibility that Earth’s magnetic field has shearing of Venus’ stagnantlid. produce horizontal compressive stresses and horizontal welling mantle, which iscoupled to the lithosphere to tivity but may reflect differential stresses due to down- like those produced today in island arcs istaken as menting compositions ofancient igneous rocksthat are be used to infer tectonic setting. In particular, docu- [78 blocks in pre-Neoproterozoic times aregiven in refs. Other examples of differentialmovement ofcontinental independent plate motions and thus plate tectonics. independently between 1770 and 1500 Ma,requiring nevertheless indicate that Australia and Baltica drifted gitudes and polarities, but that the paleomagnetic results comparisons of coeval paleopoles yields ambiguous lon- continents inpre-Neoproteroz netic database demonstrates differential movements of Yilgarn and Pilbara cratons of Australia Well-documented Archean strike-slipfaultsin occurthe are often uncertain orambiguous and reconstructions of Prec farther back in time paleomagnetic measurements reach, crustal blocks began. However, uncertainty increases the constrain when significant differential motion between radar imagery ple, strike-slipfaultingha which may not reflect plate tectonic activity. Forexam- with smalleroffsets form inawide range ofsettings dreds of kilometer ofoffset. Shorter strike-slip faults dreds to thousands of kilometer long, with tens to hun- be short but those on continental crust are typically hun- cant length and offset. Transforms inoceanic crust can signifi- with those only are transforms, faults strike-slip analysismust be done with caution. We know that not all as ath interpreted it others Ma but (~3200) fault strike-slip documented oldest gued that the Inyoka fault in southern Africa was the (viii)The chemical composition ofigneous rockscan (vii) Paleomagnetic measurements could potentially ― [ 4 80]. Clearly, careful paleomagnetic studies are ] recently concluded that the existing paleomag- [75,76] . This is not due to plate tectonic ac- R. J. STERN Chinese Science Chinese Bulletin R. J.STERN s been inferred for Venuss beeninferredfor from ambrian continentalmotions oic times. Theynoted that rust. Clearly thissortof [77] . These uncertainties [7 4 ] . Sleep [7 4 ] ar- | March 2007 | vol. 52 | no. 5| | no. |March2007vol. 52

Mg-andesites, andadakites. boninites, on time, focusing Archean in associations gin would be very usefulto carry outasystematic study of there aremany goodICP-MS these plots, but this is now relatively common now that high-quality trace element analyses in order to generate indicate crustal contamination. It is necessary to have and because vertical trends arereadily interpreted to are not redistributed unless diagrams areespecially useful because these elements this (Cawood etal. formed by subduction, and there are many papers that do infering the existence of ancient magmatic arcs that preting ancient basalts using th et al. defined by oceanic basalts MORB and OIB parallel to but distinctly higher than the mantle trend the zone isto plot Th/Yb and Ba/Yb vs.Nb/Yb orTa/Yb on neous rocks formed inassociation with asubduction vincing demonstration that a given suite ofancient ig- trace elements, especially HFS cations. The most con- alteration; this concern results in reliance onimmobile on spider diagrams forancient rocks are not the result of taken to assure that spikes of fluid-mobile elements seen isotopic and geochronologic studies. Caremust alsobe contamination that does occur is likely to be revealed by contaminatedpre-existing by primitive basalts, because these are likely to be less proach applied toancient rocksis particularly useful for as isseen for many Phanerozoic floodbasalts. Thisap- subcontinentallithosphere, the from or areinherited tion, istics are not the result of alteration, crustal contamina- specifically toensurethat important chemical character- the operation of plate tectonics, care should be taken, the tectonic setting of ancient igneous rocks element “discriminant diagrams”, widely used to infer grams”) and are alsothe basis ofa wide range of trace from normalized plots of trace elements (“spider dia- Ta high-field strength (HFS) cations, especially Nb and ciated with strong relative depletions in incompatible relative toneighboring elem or ‘spikes’ in fluid-mobile elements (e.g.,K, Sr, Pb) intra-oceanicsystems arc show distinctive enrichments and thus plate tectonic processes. Basalts frommodern strong evidence for the operation ofsubduction zones [81] As isthecasefor other li X-axis for igneousrocks.magmasArc form arrays [8 . These distinctive. characteristicscan beinferred 4 ] inferredArchean examples ofconvergent mar- [ 4 ] offerd several examples). Kerrich 578-591 ents. These spikes areasso- the rocks are veryaltered nes of inferenceregarding continentalanyand crust

is approachfor is useful analytical facilities. It [82, 83] [85] . These . Inter- 587


igneous rocks through time using this kind of plot to see Neoprotoerozoic time, as indicated by the record for when the arc-diagnostic diagonal trends are first seen in blueschists, UHP terranes, and lawsonite eclogites. Earth history. Another approach to the problem is to try and recog- (ix) Isotopic compositions are uniquely powerful for nize the magmatic and tectonic fingerprints of a inferring times of crust-mantle differentiation and for pre-plate tectonic Earth in the geologic record. The identifying deeply recycled surficial materials. There is challenge here is the difficulty of imagining a magmatic abundant evidence for early recycling of crust and sedi- or tectonic feature that could not have formed by plate ments into the mantle from as far back as the early Ar- tectonic processes. We should also think broadly about chean, including the Pb-Pb array for MORB and OIB. what secondary effects the start of plate tectonics would Nd isotopic compositions are consistent with the forma- have had on the Earth System, and look for this as well tion of an early depleted reservoir in the mantle, indi- as the direct evidence. A tectonic revolution of this mag- cating very early differentiation of the Earth’s crust and nitude should have an effect on the (perhaps mantle[86]. Isotopic studies of inclusions in diamonds triggering glacial episodes such as Snowball Earth[90]), from eclogites derived from the mantle beneath Archean The start of plate tectonics may also have affected cratons have isotopic compositions of O, Sr and Pb. Earth’s moment of inertia and rotational behavior (per- These characteristics are best interpreted to reflect sea- haps triggering True Polar Wander[91]). Other Earth sys- floor alteration seafloor and recycling into the man- tems may also have been affected by the great tectonic tle[87,88]. Mass-independent 33S isotope anomalies in revolution. diamonds are also strong evidence of that these materi- als were originally at the surface early in Earth history 6 Concluding remarks and were recycled into the mantle[89]. The isotopic re- The grand question of when and how plate tectonics cord clearly indicates that the mantle differentiated very began offers a richly interdisciplinary avenue of solid early and that crust and sediments were also recycled Earth research, one that strengthens ties between many very early. The question of whether or not this requires strands of the geosciences at the same time that it yields subduction or was accomplished by another convective new insights into our planet’s history. Because this is an mode is still open. investigation into deep time, the principle of uniformi- The above exercise in identifying geologic criteria for tarianism should be regarded as only applying to plate tectonics is only a beginning. The list of criteria chemical and physical processes and not to indicate that should be evaluated by interested individuals and modi- SOFFECS like plate tectonics must have always existed. fied accordingly. Similarly the timeline for each criteria The exploration requires that we more fully engage our should be updated frequently and individually. In any imaginations and consult a wide range of disciplines. We case, presenting the criteria and timelines in this fashion should look at the other silicate planets for insights as is an excellent way to move the discussion forward. we consider more fully the range of tectonic styles that Given the criteria and timelines shown in Figure 4, how Earth might have experienced as it cooled and differen- are these best interpreted? Clearly the isotope record tiated. It is possible that plate tectonics began very early shows that there has been material moving from the in Earth history, perhaps as early as 100 Ma after Earth Earth’s surface and mixing into the mantle from very accreted[92], and has continued since then without inter- early in Earth history; this is permissive of but does not ruption. Another possibility is that plate tectonics began require subduction and plate tectonics. The evidence for relatively late in Earth history―similar to what refs. [3] arc-like geochemical signatures also dates from early in Earth history, strongly suggesting the presence of sub- and [93] conclude―and an active stagnant lid tectonic duction zones and thus the operation of plate tectonics. style existed prior to that time. A final possibility―and Paleomagnetic data indicates differential plate motions the one I prefer―is that the Earth has a more complex at least as far back as Paleoproterozoic time, and the tectonic history than we have heretofore appreciated. ophiolite record starts about this time. A very important Geodynamic considerations and the geologic record perspective is revealed by metamorphic rocks because support an interpretation whereby a tectonic style similar this reveals deep subduction. It is thus very important to in some ways to plate tectonics―especially in being note that there is no record of deep subduction prior to able to deeply recycle surface materials to depth in the

588 R. J. STERN Chinese Science Bulletin | March 2007 | vol. 52 | no. 5 | 578-591 formation of arc crust above subduction zones of continental growth is expected to mostly occur by terozoic time. the first great ophiolite graveyards date from Neopro- of years.Clearly the evidence fordeep subduction and became negatively buoyant af buoyancy cross-overtime, so Earth cooled sufficiently to allow arelatively short tectonics late in Earth history reflected the fact that the Establishment ofself-sustaining subduction and plate style) ofplate tectonics began in Neoproterozoic time. of years elapsed before themodern episode (modern plate tectonics pulse ended, several hundreds ofmillions oceanic lithosphere. After the Paleoproterozoic proto quickly exhausted the supply ofnegatively buoyant of along buoyancy cross-over time, whereby subduction have had deep subduction and was short-lived because subduction. ThePaleoproterozoic episodemight not lying asthenosphere, butagain with noevidence fordeep nally becomenegatively buoyant with respect to under- from collapse of ancient oceanic basins which hadfi- Ga during Paleoproterozoic time episode of ‘proto” plate tectonics occurred ~1.8 to 2.0 pendent plate motions indicates that ashort but intense ence ofophiolites and paleomagnetic evidence forinde- deep subduction by lithospheric negative buoyancy or associated with mantle and to generate arc-like ,but not driven the late Archean and Paleoproterozoic explains themajorpulsesof a step function. Furthermore, such an explanation better plate tectonicsmay have been aprotracted evolution, not such crust shouldgrow at an CawoodP A, KrönerA, Pisarevsky 4 CondieK C, Kröner A, Stern R J. Penrose Conference Report: When 2 Stern R from ophiolite J. Evidence 3 Witze The startNature,A.we knowit. oftheworld 2006, as 442: 1 6 Forsyth D, Uyeda S. Ontherelativ ForsythD,Uyeda 6 Cox A. Plate Tectonics. Oxford: Blackwell, 1986 5 MELT-Seismic-Team.Imaging thedeepseismic a of structure 7

These sorts of consideration indicate that the start of Criteria andevidence. GSA Today, 2006, 16(7): 4 did plate tectonics begin? GSAdid platetectonicsbegin? Today, 2006, 16(10): 40 557―560 tion tectonicsbeganinNeoprot metamorphicmodernpressurethe episode ofsubduc- terranes that 128―131 of platemotions. Geophys JR Astr Soc, 1975, 43:163 mid-ocean ridge.Science,1998,280:1215 [DOI] [DOI]

― occurred inArchean time. Thepres- R. J. STERN Chinese Science Chinese Bulletin R. J.STERN approximately constant rate crustalgrowth inferredfor erozoic time. Geology, 2005, 33: ter afewtens ofmillions e importance of the driving forces thedriving forces of e importance that oceanic lithosphere S. platetectonics: Precambrian and ultra-highs, blueschists,and ― ―128 perhaps resulting [9 4 [DOI] ―11 ,95] . Formation [DOI] ―200

―41 [96]

[DOI] , and

| March 2007 | vol. 52 | no. 5| | no. |March2007vol. 52

igneous activity. mantle and result in brief episodes of unusually vigorous umes ofwater andsediment are injected into a hotter brief bursts of plate-tectonic behavior, when large vol- of plate tectonics, but would be expected to accompany are difficult to reconcile w ning of modern-style plate tectonics wander and suspect that these arerelated to the begin- mate change and evidence for Neoproterozoic true polar pressed by the occurrence of major Neoproterozoic cli- that such a tectonic revolution might trigger. I am im- effects clude climaticwander) androtational(truepolar shouldin- events andlook forthisrecordaswell.These aboutthelikely secondary tonics began,weneedtothink This isUTDGeosciencescontribution#1108. reviews,for helpful Yaoling and prepare to invitation forthe Niu paper.this ference are alsogreatly appreciated. I thankD. Anderson and M. Brown Spring 2006. Insights provided byparticipants ofthe 2006 Penrose con- P. Asimow, R. Workman) whileI wasaTectonics Observatory fellowin (e2005 and withcolleaguesatCaltech Scholl, N. Sleep, T. Tsujimori) wh colleagues atStanford (especially G. Ernst, J. G. Liou,S. Klemperer, D. Much ofmyunderstanding oftheprobl plate tectonics begin? plate tectonicsbegin? clear answers tothe questions: When solid Earth system. Such an effort may someday leadto ises to lead to an improved understanding of the great and development of refined geodynamic models, prom- critical examinationtherecord Precambrianof geologic if itisto be successful.The overall effort, focused on plate tectonics began should be broadly interdisciplinary when stressed, however, of be exploration isthat the process for aslong as plate tectonics isthemajor crust-forming 0 Bird P. Testing hypotheseson plat 10 12 Liu Z 1 Bokelmann G H R. Which driveforces NorthGeology, America? 11 13 Gurnis M, Mueller R Mueller GurnisM, Theorig D. 13 DoglioniC, Carminati E, Cuffaro M. Simple kinematics of subduc- 9 AlvarezW. Geologicevidencefor 8 As weassess evidence for howand when plate tec- cordance. Tectonics, 1990, 9:1213 continental hypothesis a 2002GL016002 tion zones. Int GeolRev, 2006, 48:479 faults. JGeophys Res, 1998, 103(B5): 10115 lithospheremodels includingtopogr mantle flow. Geophys ResLett, 2002 2002, 30:1027 cordance from an ancient slab a an ancientslab cordance from [97] , Bird P.NorthAmerica westward plateisdriven bylower . Majorincreases orpulses of crustal growth ―1030 [DOI]

[DOI] 578-591

ile I was Blaustein fellow during Fall ile IwasBlausteinfellowduringFall ith the continuous operation in of theAustralianAntarcticDis- in nd mantle In: nd wedge. Hillis R R, specially D. Anderson, M. Gurnis, ―1220 e-drivingmechanisms withglobal em resulted from discussionswith the plate-driving mechanism: The

nd theAustralian-AntarcticDis- aphy, and structure, thermal ―493 , [98,99] ―10129 29(24): doi:10.1029/ ― and how and . The point. to [DOI]

―did 589


Mueller R D, eds. The Evolution and Dynamics of the Australian Condie K, eds. Geophysical Monograph 164. Washington D C: AGU, Plate. Boulder: Geological Society of America, 2003 2006. 7―32 14 Davies G F, Richards M A. Mantle convection. J Geol, 1992, 100: 37 Schubert G D, Stevenson D, Cassen P. Whole planet cooling and the 151―206 radiogenic heat source contents of the Earth and Moon. J Geophys 15 Lithgow-Bertelloni C, Richards M A .The dynamics of Cenozoic and Res, 1980, 85: 2531―2538 Mesozoic plate motions. Rev Geophys, 1998, 36: 27―78[DOI] 38 Butler S L, Peltier W R. Thermal evolution of Earth: Models with 16 Hynes A. Buoyancy of the oceanic lithosphere and subduction initia- time-dependent layering of mantle convection which satisfy the Urey tion. Int Geol Rev, 2005, 47: 938―951 ratio constraint. J Geophys Res, 2002, 107(B6): doi: 10.1029/ 17 Oxburgh E R, Parmentier E M. Compositional and density stratifica- 2000JB000018[DOI] tion in oceanic lithosphere-Causes and consequences. J Geol Soc 39 Anderson D L. Large igneous provinces, delamination, and fertile London, 1977, 133: 343―355 mantle. Elements, 2005, 1: 271―275 18 van der Hilst R, Engdahl R, Spakman W, et al. Tomographic imaging 40 Davies G F. Conjectures on the thermal and tectonic evolution of the of subducted lithosphere below Northwest Pacific island arcs. Na- Earth. Lithos, 1993, 30: 281―289[DOI] ture, 1991, 353: 37―43[DOI] 41 Bédard J H. A catalytic delamination-driven model for coupled gene- 19 Conrad C P, Lithgow-Bertelloni C. How mantle slabs drive plate sis of Archean crust and sub-continental lithospheric mantle. Geo- tectonics. Science, 2002, 298(5591): 207―209[DOI] chim Cosmochim Acta, 2006, 70: 1188―1214[DOI] 20 Conrad C P, Lithgow-Bertelloni C. The temporal evolution of plate 42 Prigogine I, Stengers I. Order Out of Chaos. New York: Bantam, driving forces: Importance of “slab suction” versus “slab pull” dur- 1984. 349 ing the Cenozoic. J Geophys Res, 2004, 109(B10407): doi: 10.1029/ 43 Anderson D L. Plate tectonics as a far-from-equilibrium self-organ- 2004JB0022991[DOI] ized system. In: Stein S, Freymueller J T, eds. Plate Boundary Zones. 21 Garfunkel Z, Anderson C A, Schubert G. Mantle circulation and the Washington DC: American Geophysical Union, 2002. 411―425 lateral migration of subducted slabs. J Geophy Res, 1986, 91(B7): 44 McKenzie D, Bickle M J. The volume and composition of melt gen- 7205―7223 erated by extension of the lithosphere. J Petrol, 1988, 29: 625―679 22 Buffett B A, Rowley D B. Plate bending at subduction zones: Con- 45 Nisbet E G, Cheadle M J, Arndt N T, et al. Constraining the potential sequences for the direction of plate motions. Earth Planet Sci Lett, temperature of the Archean mantle: A review of the evidence from 2006, 245: 359―364[DOI] komatiites. Lithos, 1993, 30: 291―307[DOI] 23 Ranero C R, Phipps M J, Mcintosh K, et al. Bending-related faulting 46 Davies G F. On the emergence of plate tectonics. Geology, 1992, 20: and mantle serpentinization at the Middle America trench. Nature, 963―966[DOI] 2003, 425: 367―373[DOI] 47 Parsons B. Causes and consequences of the relation between area 24 Anderson D L. Top-Down Tectonics? Science, 2001, 293: and age of the sea floor. J Geophys Res, 1982, 87: 289―302 2016―2018[DOI] 48 Gurnis M, Hall C, Lavier L. Evolving force balance during incipient 25 Sleep N L. Evolution of the mode of convection within terrestrial subduction. Geochem Geophys Geosyst, 2004, 5(Q07001): doi: planets. J Geophys Res, 2000, 105(E7): 17563―17578[DOI] 10.1029/2003GC000681[DOI] 26 Stevenson D J. Styles of mantle convection and their influence on 49 Toth J, Gurnis M. Dynamics of subduction initiation at preexisting planetary evolution. C R Geosci, 2003, 335: 99―111[DOI] fault zones. J Geophys Res, 1998, 103(B8): 18053―18067[DOI] 27 Anderson DL. Theory of the Earth. Oxford: Blackwell, 1989 50 Stern R J. Subduction initiation: spontaneous and induced. Earth 28 Abe Y. Thermal and chemical evolution of the terrestrial magma Planet Sci Lett, 2004, 226: 275―292[DOI] ocean. Phys Earth Planet Inter, 1997, 100: 27―39[DOI] 51 Mei S, Kohlstedt D L. Influence of water on plastic deformation of 29 Tonks W B, Melosh H J. Magma ocean formation due to giant im- olivine aggregates, 1. Diffusion creep regime. J Geophys Res, 2000, pacts. J Geophys Res, 1993, 98(E3): 5319―5333 105(B9): 21457―21470[DOI] 30 Solomatov V S, Moresi L N. Stagnant lid convection on Venus. J 52 Regenauer-Lieb K, Yuen D A, Branlund J. The initiation of subduc- Geophys Res, 1996, 101(E2): 4737―4754[DOI] tion: criticality by addition of water? Science, 2001, 294(5542): 31 Phillips R J, Hansen V L. Geological evolution of Venus: rises, 578―580[DOI] plains, plumes, and plateaus. Science, 1998, 279: 1492―1497[DOI] 53 Clift P, Vannucchi P. Controls on tectonic accretion versus erosion in 32 Connerney J E P, Acuña M H, Wasilewski P J, et al. Magnetic linea- subduction zones: implications for the origins and recycling of the tions in the ancient crust of Mars. Science, 1999, 284: continental crust. Rev Geophys, 2004, 42: doi: 10.1029/ 794―798[DOI] 2003RG000127[DOI] 33 Rieder R, Economou T, Wanke H, et al. The Chemical composition of 54 Jull M, Kelemen P B. On the conditions for lower crustal convective Martian soil and rocks returned by the mobile alpha proton X-ray instability. J Geophys Res, 2001, 106(B4): 6423―6446[DOI] spectrometer: preliminary results from the X-ray mode. Science, 55 Kusky T M, Li J H, Tucker R T. The Dongwanzi ophiolite: complete 1997, 278(5344): 1771―1774[DOI] Archean ophiolite with extensive sheeted dike complex, North China 34 Strom R G, Schaber G G, Dawson D D. The global resurfacing of craton. Science, 2001, 292: 1142―1145[DOI] Venus. J Geophys Res, 1995, 99: 10899―10926[DOI] 56 Zhai M, Zhao G, Zhang Q, et al. Is the Dongwanzi Complex an Ar- 35 Nimmo F. Tectonic consequences of Martian dichotomy modification chean ophiolite? Science, 2002, 295: 923―923[DOI] by lower-crustal flow and erosion. Geology, 2005, 33(7): 57 Scott D J, Helmstaedt H, Bickle M J. Purtuniq ophiolite, Cape Smith 533―536[DOI] Belt, northern Quebec, : A reconstructed section of early Pro- 36 Korenaga J. Archean geodynamics and the thermal evolution of terozoic oceanic crust. Geology, 1992, 20: 173―176[DOI] Earth, in Archean Geodynamic Processes. In: Benn K, Mareschal J C, 58 Peltonen P, Kontinen A, Huhma H. and geochemistry of

590 R. J. STERN Chinese Science Bulletin | March 2007 | vol. 52 | no. 5 | 578-591

71 Tsujimori T, Sisson V B, Liou J G, et al. Very-low-temperaturere- T,al. Tsujimori V Sisson B,Liou J G, et 71 5 KoenigE, AydinEvidence forA. large-scale strike-slip faulting on 75 2 Brown M.A duality ofthermal regi 72 66 Moyen J J Moyen F, G, subduction Stevens mid-Archean of Record A. Kisters 66 Maruyama J S,Liou G,Terabayashi M. Blueschistsandeclogitesof 59 7 ScoteseC R. A continentaldrift flipbook. JGeol, 2004, 112: 77 SleepN plate-tectoni Archean H. 73 62 Maekawa H, Fryer P,H, Fryer Maekawa metamor- Blueschist-facies A.Incipient Ozaki 62 6 Tuckwell G W, Ghail R C. A 400-km 76 JG.S, Liou Maruyama Ultrah 70 JahnB M, Caby Monie R, P. UHP The oldest eclogitesof the : 69 metamorphism UHPJ G, Liou Global Tsujimori T,al. Zhang RY, et 68 PeacockS sub- metamorphic evolutionof and structure Thermal M. 67 5 ShuL, CharvetKinematics J. a 65 RoeverW P distribution thepreferential the causeof d. On of certain 64 0 ErnstWG. High-pressuremetamorphic andultrahigh-pressure- belts 60 4 SleepN H. Evolution ofthec 74 63 Zhang H, Thurber C H, Shelly D, et al. High-resolution subduct- High-resolution al. et ThurberH, CH, ShellyD, Zhang 63 1 AbersG A,Van Keken P E, Kneller E The thermalstructure al. A, et 61

eclogites. Lithos, 2006, 609 cord of the subduction process: A review of worldwide lawsonite J Petrol, 1996, 37(6): 1359 1.95GaJorm metabasalts from the Venus. Geology, 1998, 26(6): 551 tonics sincetheNeoarchean. Geology, 2006, 34:961 2006, 442(3): 559 Africa. Nature, South terrain, metamorphism Barberton from inthe the world andtheir exhumation. Int GeolRev, 1996, 38, 490 nental geology? CanJEarthnental geology? Sci, 1992, 29:2066 1995. 281―289 WesternWashington Pacific. D C: TaylorJ, eds.Natland B, Active Margins andMarginal Basins ofthe phism intheactivesubduction zonebeneath theMariana . In: 45―55 boundaryThetisVenus. Regio, of Arc, 1998, 7:6 significance on the Proterozoic-Phanerozoic boundary. The Island cations. Chem Geol, 2001, 178:143 Age ofUHPof protolithsandtectonicimpli- metamorphism, nature Rev, 2004, 46:1 and continentalsubduction/collision: model. IntGeol TheHimalayan physical Monograph138. Washington DC: AGU, 2002.7 ducting slabs. In: Eiler J, eds. 1996, 267:291 melangeophiolitic (Jiangnan Region, zone Dongxiang-Shexian ductileshear schau, 1964, 54:933 mineralsmetamorphic inorogenic beltsof different age. GeolRund- of America. SpecialPaper 373, 2003. 365 ThougEvolution ofGeological and ophiolite studies. In:DilekY, Newcombeeds. S, Concept Ophiolite Subduction, recrystallization,e Planet Sci, 2005, 33:360 ble-difference tomography. Geology, 2004, 32(4): 361 ing-slab structure beneathnor 241: 387―397 for slabdehydration andwedgefl of subduction zones constrained by seismic imaging:Implications [DOI]

―35 [DOI] ―302 ―27 ―562 [DOI] ―941

[DOI] [DOI] ―393


R. J. STERN Chinese Science Chinese Bulletin R. J.STERN ― ontinental lithosphere. AnnuRev Earth

624 [DOI] Inside theSubductionFactory,Geo- nd ofthe Proterozoic thern Honshu, revealed by dou- igh-pressure metamorphism andits xhumation, andsignificancefor ―554 cs-what can be learned from conti- from learned be cs-what can [DOI] ua ophiolite, northeastern Finland. ophiolite,northeasternFinland. ua EarthLett, 2003, PlanetSci 211: ow.Earth PlanetSciLett, 2006, ―158 American GeophysicalUnion, mes is the hallmark of platetec- of thehallmark mes is ht. Boulder: GeologicalSociety

SouthChina)., -scale strike-slip zonenear the : With: HP metamorphism and [DOI]

―384 [DOI]


―964 ―364 ―22 [DOI] ―596 [DOI]

| March 2007 | vol. 52 | no. 5| | no. |March2007vol. 52

0 PowellC M, Jones D L, Pisarevs 80 84 Kerrich R, Polat A. Archean greenstone-tonalite duality: Thermo- greenstone-tonalite A. Archean Polat KerrichR, 84 Stern R SubductionJ.Geophys, zones. Rev 2002, 40:1012, doi: 81 0 Hoffman P F, Schrag D P.Earth hypothesis: TestingThe snowball the 90 4 CondieK models:Afterthoughts C. Episodic continentalgrowth and 94 Hamilton W. alternative Earth. GSA An Today,2003, 93 13(11): 92 Harrison T M, Blichert-ToftT J, MuM,ller Harrison W, Ha- et al. Heterogeneous 92 Goldreich P, Toomre remarksA. onpolar Some wandering. JGeo- 91 99 Maloof A C, Halverson G P, Kirschvink J L, et al. Combined paleo- Combined G P,al. Halverson Kirschvink JL, et A C, Maloof 99 AllenP A. SnowballEarth onTrial.EOSTrans Am Geophys Union, 98 2 Pearce J A, Cann J R. Tectonic settingofbasicvolcanic rocks deter- 82 78 Pesonen L J, Elming S A, Mertanen S, et al. Palaeomagnetic con- Palaeomagnetic al. et L J, Elming SA, MertanenS, Pesonen 78 9 Farquhar J, Wing B A, Mckeegan K D,Mass-independent etal. sulfur 89 6 Tatsumi Y.The subduction factory: theevolving Howitoperates in 96 BalashovY VGlaznev A, problemEndogenicN. and the cycles of 95 iso- oxygen crustal B, Harte Valley Extreme D E, J al. W, Schulze et 88 oce- Ancient eclogites: Victor Roberts I D, Manton W I. MacGregor 87 3 Shervais J W.T-V plots andthepetrogenesismodern of andophioli- 83 79 Pisarevsky S Pisarevsky A, M McElhinny W. 79 7 Scholl D W, Huene R V. recycling Crustal modern at subduction zones 97 6 Boyet M,Carlson R W. 86 5 PearceJ A, Peate D W. Tectonic Implications theComposition of Vol- of 85 canic Arc Magmas. Annu RevEarth PlanetSci, 1995, 23:251 729―741 ferentiation ofthesilicateEarth. Science, 2005, 309:576 brian Res, 2001, 110: 33 straints onthepositionofKalaharicratoninRodinia.Precam- Earth globaldynamics? Tectonophysics, 2006, 415:141 chemical mantle convection models 10.1029/2001RG000108 limits ofglobalchange. Terra Nova, 2002, 14:129 Extensions. Tectonophysics, 2000, 322:153 4 ence, 2005, 310:1947 dean hafnium:Evidence for continental crust at4.4 to 4.5 Ga. Sci- phys Res, 1969, 74(10): 2555 Geol Soc Am Bull, 2006, 118: 1099 from theNeoproterozoic Akademikerbreen Group, Svalbard, Norway. magnetic, isotopic, andstratigraphi 2006, 87(45): 495 290―300 Lett, 1973, Sci Earth Planet elementmined analyses. usingtrace 19: 192 2003, Union, 84: Geophys Trans Am EOS, form. visual its into veloped 2003, Tectonophysics, theProterozoic. during continents of figuration Science, 2002, 298:2369 of inclusions indiamond andsulfur recycling ontheearly Earth. Earth. GSA Today, 2005, 15(7): 4 crustal growth. Geochem Int, 2006, 44(2): 131―140 68―70 tope signatures preserved incoes anic crust. JGeophys Res, 1986, B91:14063 Boulder: Geological Society of America, SpecialPaper, 2007 eds. Crust. al, et Continental D, JR of Hatcher Framework 4D The mantle geochemistry,basement, In: reconstruction. and of continental preservation and growth of - issues past the to applied tic . Earth PlanetSciLett, 1982, 59:101 ―12 [DOI] 375: 289―324 [DOI] [DOI]

[DOI] ―1950 1 4 ―46 [DOI] 2 ―2372 Nd evidence early(> for 4.53 Ga) global dif- 578-591

[DOI] ―2567 [DOI]

[DOI] ―10 ite indiamond. Nature, 2003, 423: Global paleomagnetic data base de- Global paleomagneticbase data

ky ky S, et al. Paleomagnetic con- ―1124 c evidence for true polar wander truepolarwander evidence for c or plate tectonics in the early intheearly tectonics or plate [DOI]

[DOI] ―162

―14079 ―118

[DOI] ―155 [DOI] ― ―164 581

[DOI] ―

[DOI] 285 [DOI]

[DOI] 591