plate tectonics at 3.8–3.7 ga: field evidence from the isua accretionary complex, southern west...

[The Journal of Geology, 1999, volume 107, p. 515–554] q 1999 by The University of Chicago. All rights reserved. 0022-1376/1999/10705-0001$01.00

515

ARTICLES

Plate Tectonics at 3.8–3.7 Ga: Field Evidence from the IsuaAccretionary Complex, Southern West Greenland

Tsuyoshi Komiya, Shigenori Maruyama, Toshiaki Masuda,1 Susumu Nohda,2

Mamoru Hayashi,3 and Kazuaki Okamoto

Department of Earth and Planetary Sciences, Tokyo Institute of Technology, Tokyo 152-8551, Japan(e-mail: [email protected])

A B S T R A C T

A 1 : 5000 scale mapping was performed in the Isukasia area of the ca. 3.8-Ga Isua supracrustal belt, southern WestGreenland. The mapped area is divided into three units bounded by low-angle thrusts: the Northern, Middle, andSouthern Units. The Southern Unit, the best exposed, is composed of 14 subunits (horses) with similar lithostratig-raphy, bound by layer-parallel thrusts. Duplex structures are widespread in the Isua belt and vary in scale from a fewmeters to kilometers. Duplexing proceeded from south to north and is well documented in the relationship betweenlink- and roof-thrusts. The reconstructed lithostratigraphy of each horse reveals a simple pattern, in ascending order,of greenstone with low-K tholeiitic composition with or without pillow lava structures, chert/banded iron-formation,and turbidites. The cherts and underlying low-K tholeiites do not contain continent- or arc-derived material. Thelithostratigraphy is quite similar to Phanerozoic “oceanic plate stratigraphy,” except for the abundance of maficmaterial in the turbidites. The evidence of duplex structures and oceanic plate stratigraphy indicates that the Isuasupracrustal belt is the oldest accretionary complex in the world. The dominantly mafic turbidite composition suggeststhat the accretionary complex was formed in an intraoceanic environment comparable to the present-day westernPacific Ocean. The duplex polarity suggests that an older accretionary complex should occur to the south of the Isuacomplex. Moreover, the presence of seawater (documented by a thick, pillow, lava unit at the bottom of oceanic platestratigraphy) indicates that the surface temperature was less than ca. 1007C in the Early Archean. The oceanic geothermfor the Early Archean lithosphere as a function of age was calculated based on a model of transient half-space coolingat given parameters of surface and mantle temperatures of 1007 and 14507C, respectively, suggesting that the Archeanoceanic lithosphere was rigid. These conclusions—rigidity and lateral plate movement—support the idea that themodern style of plate tectonics was in operation only 0.7–0.8 G.yr. after the formation of the Earth.

Introduction

The ca. 3.7-Ga Amıtsoq gneisses, along with en-claves of the Isua supracrustal belt at Isukasia, havelong been recognized to be among the oldest rockson Earth (Moorbath et al. 1977; Nutman et al. 1993,1996, 1997). The 3.8-Ga Isua supracrustal rocks aresignificant because they retain primary magmatic

Manuscript received December 30, 1998; accepted June 16,1999.

1 Institute of Geosciences, Faculty of Science, Shizuoka Uni-versity, Shizuoka 422-8529, Japan.

2 Department of Environmental Science, Faculty of Science,Kumamoto University, Kurokami 2-40-1, Kumamoto 860-8555,Japan.

3 Department of Earth Sciences and Astronomy, Universityof Tokyo at Komaba, Meguro-ku, Tokyo, 153-8902, Japan.

features such as pillow structures (Allaart 1976;Maruyama et al. 1991b, 1994) and original sedi-mentary structures such as rhythmic banding inbanded iron-formation (BIF) or graded bedding inturbidites (Nutman et al. 1984). The Isua supra-crustal belt, one of the oldest terrains in the world,presents a key to the understanding of tectonic pro-cesses on the early Earth.

Immediately after the establishment of the newglobal tectonic paradigm, the question arose as towhen the modern style of plate tectonics began(Dewey and Spall 1975). One school of thought heldthat it commenced at the earliest stage of the evolv-ing Earth, as suggested by the geochemistry ofgreenstones (Hart et al. 1970; Condie 1972, 1982),

Figure 1. Geologic map of the northeastern part of Isua supracrustal belt. The inset map is a geotectonic map ofsouthern West Greenland, simplified after McGregor et al. (1991). Three terranes, Akia (gray), Tasiusarsuaq (graywith horizontal bars), and Akulleq are shown. Akulleq terrane is composed of Ikkattoq gneiss (dark gray with whitebars), Amıtsoq gneiss (stippled), and supracrustal rocks (black). Postcollision Qorqut granite (crosses) intrude mainlythe Akulleq terrane. The Isua supracrustal belt is located in the northeastern corner.

Journal of Geology I S U A A C C R E T I O N A R Y C O M P L E X 517

Figure 2. Mafic sediment (MS4) on the left and the underlying carbonate layers on the right. The boundary betweenthe two units is a knife-edge layer-parallel fault (T2), recognizable because carbonate layers form isoclinal folding (F1b)and the folding axis is clearly cut by the fault, whereas the upper mafic sediments form rhythmic bedding withoutfolding. The scale is 60 cm by carpenter’s rule. The locality is horse IV2.

geophysical considerations (Bickle 1978; Arndt1983; Nisbet and Fowler 1983), and geological sim-ilarities with Phanerozoic orogenic belts (Burke etal. 1976). The other school, called the “millipedemodel” (Wynne-Edwards 1976), or ensialic verticaltectonics (Kroner 1977, 1981; Piper 1982), sug-gested that the tectonic regime until the late Prot-erozoic was typified by vertical movement-domi-nated orogenies. The latter school listed theabsence of the following features as precludingplate tectonics in the Archean: ophiolites, eitherFranciscan-type melange or accretionary com-plexes, and blueschists, all of which characterizedivergent or consuming plate boundary processeson the modern Earth. Geological evidence for platetectonics and oceanic crust accretion has recentlybeen reported in some Archean terrains (Kusky1989; Hoffman 1991; Calvert et al. 1995), but someare still controversial (Bickle et al. 1994; Kusky andWinsky 1995). Here we report evidence for ophio-lites and an accretionary complex within the Isuasupracrustal belt, propose criteria for accretion of

oceanic materials to continental lithosphere, anddiscuss the features of the accretionary complex inthe Isukasia region compared with those in thewestern Pacific region, especially the Japanese is-lands. We will use the technical terms of horse,duplex, duplexing, and roof-, floor-, and link-thrustas defined by McClay (1992).

Geological Outline. The Isua supracrustal belt,approximately 150 km northeast of Nuuk, Green-land (fig. 1), is part of the Akulleq terrane. Thisterrane is composed of 3880–3660-Ma Amıtsoqgneiss and 2820-Ma Ikkattoq gneiss. The formercontains enclaves of Early Archean supracrustalrocks (Akilia association), whereas the latter con-tains other enclaves of late Archean supracrustalrocks (Malene supracrustals) and an anorthositecomplex (Friend et al. 1987; Nutman et al. 1989;McGregor 1993). The Akulleq terrane was intrudedby the K-rich Qorqut postorogenic granite at about2550 Ma and stabilized at that time (Baadsgaard1976; Brown et al. 1981; Moorbath et al. 1981;McGregor 1993). Friend et al. (1987) and McGregor

518 T . K O M I Y A E T A L .

Figure 3. Conglomerate. This layer is located on the top of the lithostratigraphy of horse IV2. The gravel is polymictic,varying in size (pebble to cobble) and in lithology. The clasts are significantly flattened because of postdepositionaldeformation. This conglomerate-bearing unit is interlayered with carbonate layers in some places. The pen scale is15 cm long.

et al. (1991) have proposed a tectonic model for theAkulleq terrane, based on a plate-tectonic conceptof collision-accretion (Coney et al. 1980). Thenorthern 3230–2980-Ma Akia and the southern2920–2830-Ma Tasiusarsuaq, with the Akulleq ter-rane between, collided at 2820–2712 Ma. Most ar-eas in southern West Greenland are underlain bypolydeformed amphibolite to granulite facies rocks(McGregor 1993). The northeastern part of the Ak-ulleq terrane is an exception. Here, a relatively lessmetamorphosed segment, the Isua supracrustalbelt, is exposed (fig. 1, inset) as a fault-boundedtectonic slice and is in tectonic contact with thetonalitic Amıtsoq orthogneiss on both sides(McGregor 1973; Bridgwater et al. 1974).

The Isua supracrustal belt forms a 35-km-longarcuate tract, the largest fragment of Early Archeansupracrustal rocks in the Akulleq terrane (fig. 1,inset). Farther south, Early Archean supracrustalrocks equivalent to the Isua supracrustal rocks areexposed as enclaves in the Amıtsoq gneiss. Thissuggests that the Isua belt was originally intruded

by tonalitic rocks that later metamorphosed toform the Amıtsoq gneiss. The Isua supracrustal beltwas affected by Barrovian-type amphibolite-faciesmetamorphism during the Archean (Boak and Dy-mek 1982). In spite of metamorphism, metaso-matism (Rosing et al. 1996), and polyphase ductiledeformation, original protolith structures are wellpreserved in the northeastern part of the belt wherethe metamorphic grade is the lowest. Because theIsua belt displays well-preserved primary structuresthat yield information helpful in deciphering pre-3.8-Ga events on Earth, considerable interdisciplin-ary work had been conducted here (e.g., see recentsummaries by Nutman 1986; Schidlowski 1988;Nutman et al. 1993, 1996, 1997; Mojzsis et al.1996).

We have spent two field seasons in the Isua su-pracrustal belt, in 1990 and 1993, and undertook ageologic map covering the northeastern part of thebelt at a scale of 1 : 5000. During our field inves-tigations, the rock units were mapped according totheir premetamorphic state (Maruyama et al.


Figure 4. Pillow lavas. Some basalts preserve original pillow lava structure in this figure. Each pillow consists ofcore (green), mantle (darker green), rim (pale green to white), and matrix (darker green). Each pillow is flattened byregional deformation.

1991b, 1994; Komiya and Maruyama 1995). Wetherefore have used the names doleritic dike, pil-lowed basalt, carbonate, and conglomerate, insteadof homogeneous amphibolite, Garbenschiefer, calc-silicate rocks, etc. We have distinguished five majorlithologic units: (1) a mafic-felsic turbidite se-quence, including interlayered minor conglomerate(figs. 2, 3); (2) chert and BIF; (3) pillow lava (fig. 4),massive lava, pillow breccia, and hyaloclastite; (4)dolerite-gabbro intrusives related to unit 3 (fig. 5);and (5) ultramafic rocks. In addition, a large number

of sills and dikes were encountered. Some are mid-Archean Ameralik amphibolite dikes with char-acteristic large patches of plagioclase crystals, butthe majority are of an age as yet unknown. Thenoritic or high-Mg andesite dikes trending N-S arethe youngest rocks in the mapped area and havebeen dated at 2215 Ma by SHRIMP (Nutman et al.1995).

Mylonite zones a few meters wide, dipping about507 SE, run along both sides of the Isua belt andjuxtapose it against Amıtsoq gneiss (fig. 6; Nutman

520 T . K O M I Y A E T A L .

Figure 5. Basaltic dike, cutting pillow basalt, and overlying hyaloclastite layers. The hammer is on the boundarybetween the dike and pillowed basalt. Many drilled holes for paleomagnetism studies are visible along the sharpboundary of the intruded dike. See the location on figure 14.

1984). The mylonite zone on the western boundaryhas alternating quartz-rich and feldspar-micaceouslayers. Mineral lineations on the foliation dipsteeply (1507) to the ENE. Microtextures in thinsection indicate a down-dip shear sense. Under themicroscope, elongated ribbon quartz is warped intoa stair-stepping morphology around the roundedfeldspar porphyroclasts. Matrix quartz is dynami-cally recrystallized and aligned !407 oblique to thefoliation (Type II S-C mylonite of Lister and Snoke1984). The chlorite-rich parts, originating fromfoliation-parallel veining, show mica fish–like tex-tures. These asymmetric textures all indicate top-to-east (down-slip) sense of shear (vertical simpleshear, D1, of James 1976). Metamorphic conditionsduring mylonitization are assumed to be green-schist facies because of the presence of veinlets ofchlorite and the brittle deformation of feldspar.Drag folding (D2 phase of James 1976), caused bydextral lateral movement, has bent the myloniticfoliation.

The eastern boundary is also defined by a my-lonite zone with a 507 SE strike (fig. 7) and has dip-

slip mineral lineations. Under the microscope,ribbon-quartz showing stair-step shape and S-C my-lonite textures indicate top-to-ENE sense of shear.Biotite and white mica show fish structures and areoften recognized as tails of epidote porphyroblasts.Metamorphic conditions are assumed to havereached greenschist facies, as given above. Thesetextural features imply domal uplift of the innergneiss, causing simple shear around the boundarybetween the Isua supracrustal belt and the Amıtsoqgneiss.

Pillow structures in amphibolite were recognizedby geologists of Kyrolitselskabet Øresund A/S inthe early 1970s but were subsequently overlookedbecause rocks were classified according to theirmetamorphic structure and texture, instead of theirprotoliths (Allaart 1976). Therefore, the pillow la-vas in the Isua supracrustal belt have never beenwell described. The pillow lava structures are spo-radically preserved in areas of relatively low strain.Even in high-strain areas, pillow structures can berecognized if the tectonic stretching is parallel tothe elongated pillow tubes. The pillow lavas are


Figure 6. The east-dipping boundary between Amıtsoq gneiss (left, white) and the supracrustal belt (right, dark) onthe west side of the supracrustal belt. The contact is highly deformed and recrystallized to form a mylonite zone afew meters wide.

relatively small, 30–150 cm across, and are madeof oval with cusp through pear-shaped to crescent-shaped forms, depending on secondary deforma-tion. Relatively undeformed pillow lavas indicatethe way-up (Borradaile 1982) and are composed ofcore (green), rim (pale green to white colored), andmatrix (darker green); some have a mantle (darkergreen) between the core and rim (fig. 4). The pillowlavas are close-packed type with very little matrixmaterial. The pillows vary in size and shape de-pending on the flow unit. Vesicles and varioles arenot visible on rims, presumably because of the sec-ondary deformation and metamorphic recrystal-lization.

Barrovian-Type Progressive Regional Metamor-phism. Both the Isua supracrustal belt and theAmıtsoq gneiss were thought to have originally suf-fered amphibolite-facies metamorphism and thenrecrystallized under greenschist facies conditionsbecause of retrogressive metamorphism (Boak andDymek 1982; Nutman 1986). Our detailed petro-graphic analyses from 1750 rock samples from theNorthern, Middle, and Southern Units revealedthat, although garnet and hornblende partiallychange into chlorite along their rims or within fis-sures, no relict minerals of high-grade amphibolitic

metamorphism exist even in the least metamor-phosed zone of the greenschist facies mineral as-semblage. Moreover, Ca-amphiboles in the high-grade amphibolite zone possess knife-edge-boundedzoning of actinolite in the core and hornblende inthe rim, suggesting the presence and preservationof the prograde metamorphic stage.

The study area is divided into four metamorphiczones based on mineral assemblages and the com-positions of the metamorphic minerals (i.e., anor-thite content of plagioclase) in metabasites (fig. 8).Zone A is characterized by the greenschist mineralassemblage of chlorite (Chl), epidote (Ep), albite(Ab), and quartz (Qtz). Zone B is defined by theappearance of Ca-amphiboles, either actinolite(Act) and/or hornblende (Hbl), together with albiteand/or oligoclase. The mineral assemblage is Chl,Ep, two amphiboles, peristeritic plagioclase (Pl),and Qtz, showing the transition between green-schist and amphibolite facies. The distinctive fea-ture of Zone C is the disappearance of the peris-teritic gap in plagioclase and the presence of twoCa-amphiboles in intermediate magnesian samplesand single Ca-amphibole, actinolitic hornblende inmore magnesian samples, or hornblende in less

522 T . K O M I Y A E T A L .

Figure 7. Fault boundary between Amıtsoq gneiss (upper, white) and supracrustal belt (lower, strongly folded) onthe east side of the supracrustal belt. The hammer is on the side of the Amıtsoq gneiss. Supracrustal rocks are severelydeformed, whereas the Amıtsoq gneiss is relatively undeformed.

magnesian samples. The metabasites are composedof diagnostic minerals Chl, Ep, Act-Hbl, Pl, andQtz. Zone D is defined by the high anorthite con-tent of plagioclase and single Ca-amphibole. Themineral assemblage is Chl, Ep, Hbl, high-Ca pla-gioclase, and Qtz. Some rocks contain garnet andcummingtonite because of their anomalous com-position (enriched in MgO, FeO, and Al2O3 andMgO and FeO, respectively). The Mg value, definedby , of chlorite coexisting100 # MgO/(MgO 1 FeO)with Hbl, Act, Ep, Pl, and Qtz increases progres-sively from 52 in Zone B to 63 in Zone C. The Fe-Mg partition coefficient between garnet and biotitealso shows the progressive increase of metamorphictemperature: 4007C in the northern area of Zone Cthrough 4507C in the southern area of Zone C to5507C in Zone D. This suggests that metamorphiczonation represents the Barrovian-type progressivemetamorphism. Isograds trend NW-SE and cross-cut the NS-elongation of the Isua supracrustalbelt, which implies postmetamorphic tectonicjuxtaposition of the Isua belt into the Amıtsoqgneiss (fig. 8).

Geology of the Northeastern Part of the Isua Supra-crustal Belt. The mapped area is subdivided intothree units (the Northern, Middle, and SouthernUnits; fig. 1) by structural breaks defined by NS-trending/E-dipping or EW-trending/N-dipping low-angle thrust faults. Each unit is composed of im-bricated piles of oceanic material, such as pillowedor massive basaltic flows, hyaloclastites, BIF/chert,and mafic/felsic sediments.

The Northern Unit is juxtaposed with the MiddleUnit across a NS-trending fault with 207–307 dipsto the east. The lower half of the Northern Unit ishighly disrupted melange (Type II; Cowan 1985),with blocks of greenstone and chert, comparable toexamples in Phanerozoic Tethyan and circum-Pacific orogenic belts (fig. 9a). More than 37 angularto subrounded blocks of greenstone were randomlyembedded in a highly sheared muddy matrix, to-gether with several blocks of chert with or withoutBIF. Moreover, magmatic structures of pillow lava,dike, gabbro, and hyaloclastite have been identifiedin greenstone blocks, whereas bedding planes ofcherts are well preserved in chert blocks. The upper

Figure 8. Progressive metamorphism in the mapped area, which is subdivided into four zones based on mineralassemblages and anorthite content (An) of plagioclase in the metabasite. The metamorphic grade ranges from green-schist facies in the northernmost area through epidote-amphibolite transition to amphibolite facies in the SouthernUnit.

Figure 9a

Figure 9. a, Geological map of the Northern Unit. The western part of the unit is underlain by melange equivalentto those in circum-Pacific orogenic belts in the Phanerozoic. The randomly oriented blocks of cherts and greenstonesare distributed within mafic sediments. In contrast, the eastern part is underlain by five subunits with stratigraphicsequences of pillow lava flow and the overlying bedded cherts and banded iron-formation (BIF). Location for the sketchmap shown in figure 9a is labeled “Fig. 9a.” b, Sketch map of the greenstone-chert sequence of the Northern Unit.See the sketch location in figure 9a. It is underlain by the five subunits with pillow lava flow and hyaloclastite layercapped by chert/BIF sedimentary sequence.

Figure 9b

526 T . K O M I Y A E T A L .

Figure 10. a, Simplified cross section of piling subunits along (A)–(B) on b. Small-scale folds and dikes are removedbecause of simplification. b, Distribution pattern of duplexes bounded by layer-parallel faults in the Southern Unit.Eight duplexes are defined by both lithostratigraphy within a horse and fault distribution. Duplexes consist of severalhorses. The layer-parallel boundary thrusts among horses, both interduplexes and intraduplex thrusts, are left-lateral,and relative direction of fault displacement is shown by small arrows. They converge to the south or west. Thesestructures are characterized by convergence and divergence of layer-parallel faults and are duplex structures. Faultsat the bottom and top are called floor- and roof-thrusts, respectively. Proterozoic dikes and many Archean basalticdikes are removed from map for simplification.

half, which resembles the Southern Unit in termsof structure and lithostratigraphy, is composed ofmore than four subunits bounded by layer-parallelthrusts. Each subunit is composed of pillow lavaor sheet-flow greenstone at the bottom, overlain by

layered chert/BIF that is multiply deformed and cutby faults. Pillowed basalts covered with a hyalo-clastite layer are conformably capped by thickchert/BIF sequences. BIF layers occur at the bottomof the sedimentary sequence, whereas white chert


Figure 11. Representative example of lithostratigraphywithin horse III2. The lithostratigraphy is composed ofmetabasite, chert, muddy chert, and mafic sediment, inascending order. The metabasite unit consists of a pileof pillowed basalt with massive basaltic flow and hyalo-clastite. The reworked hyaloclastite with rough surfacesand numerous metamorphic garnet megaporphyroblastsoccurs at the top of the unit. The chert unit is dominantlycomposed of a white, massive type, whereas a white/bluetype occurs near the bottom. The chert unit was intrudedby several basaltic sills. Some carbonate pods sporadi-cally occur between the metabasite unit and the chertunit. Mafic sediment with graded bedding covers thechert unit at the top. Muddy chert occurs as transitionalsediments between the chert unit and mafic sediments.Some carbonate layers are interleaved into muddy chertand mafic sediments. The reconstructed lithologies arecompared with oceanic plate stratigraphy (see also fig-ure 17).

becomes more abundant through the transition ofwhite-blue chert at the upper part. A thin graphite-bearing chert layer also occurs at the muddy-cherthorizon, resting on bedded chert. Basaltic dikes/sills are intruded into BIF/chert sequences in sev-eral places. The rocks in this unit suffered fromrelatively low-grade progressive metamorphismfrom greenschist facies in the eastern region toepidote-amphibolite facies in the western region(fig. 8).

The Middle Unit is cut on the south by an EW-trending, gently N-dipping, fault. Although Qua-ternary glacial drift obscured the unit boundary,bedding planes change abruptly beyond the bound-ary from a NNE-strike in the Middle Unit to a NE-strike in the Southern Unit. Across the boundary,the lithostratigraphic discontinuity is obvious;chert and overlying mafic sediment predominate inthe Middle Unit, whereas massive lava, pillow lava,and hyaloclastite are abundant in the SouthernUnit. Although the Middle Unit is mainly com-posed of dominantly NS-trending chert layers, nearthe southern margin of the unit is a mafic mate-rial–enriched turbidite sequence associated withfelsic sandstone, mudstone, calcareous sandy shale,and even minor conglomerate. Chert is mainlywhite, and some layers are intercalated with im-pure (dolomitic) carbonate. More than 110 dikesand sills are intruded into the Middle Unit, varyingin thickness from tens of centimeters to tens ofmeters. Only two contain plagioclase patches, sug-gesting that they are Ameralik dikes. One of theAmeralik dikes intruded the unit-bounding fault,indicating postjuxtaposition dike emplacement.The dike itself has been metamorphosed to am-phibolite facies. Reconstruction of the original stra-tigraphy is difficult because there are no key bedsin the unit, but it probably consists of several sub-units, as do the other units.

Geology of the Southern Unit. The Southern Unitoccupies the major part of the mapped area. Majorlithologies are basaltic lava flows, with or withoutpillow structures. Because various lithologies (e.g.,conglomerate layers and felsic sediments) crop outthroughout the unit, we employ them as key bedsin geological and structural investigations. The dis-tribution pattern of the five major lithologies iscomplicated. Cherts and associated sediments formseveral belts, apparently converging in some casesand diverging in the others, as indicated by layer-parallel faults (fig. 10). Faults formed by layerparallel shortening are common in the Isua supra-crustal belt, and duplex structures are on scales ofmeters to a few kilometers. Duplex structures arecommon in all three units, but the truncated du-

Figure 12a

Figure 12. a, Sketch map showing the complex deformation patterns in the turbidite sequence. This sequence belongs to horseIV2 (see the location on fig. 1). The area is subdivided into four blocks, named I, II, III, and IV. The N-S cross sections and reconstructedlithostratigraphy are shown in figure 13. Inset figure is lower hemisphere equal-area projection of poles to tectonic layering fromall the blocks. Large shaded squares are eigenvectors based on a cylindrical best fit to data. Note the location of figure 2. b,Magnified sketch map of the eastern part. c, Magnified sketch map of the western part. d, Outcrop in this sketch is located 20m to the west of figure 12a. Although the outcrop is about 20 m far from that of figure 12a, the continuous stratigraphy can bereconstructed by tracing the key bed (the muddy chert layer).


Figure 12b

plex structure is most conspicuous in the SouthernUnit (fig. 10).

At least eight major subunits (duplexes) are de-fined by NE-SW trending layer-subparallel faults(fig. 10). Although the faults are subparallel to thelayering of the turbidite sequence, the layering is

clearly cut by those faults in many places, at out-crop scale. The duplexes are, in turn, composed ofseveral horses; at least 12 horses can be recognized.The lithostratigraphy and chert thickness are sim-ilar among horses in each particular duplex; thissimilarity is used to define a single duplex (a duplex

Figure 12c


Figure 12d

is indicated by a roman numeral, and the horsebelonging to that duplex is indicated by a subscriptArabic numeral, e.g., IV2).

We reconstructed the original lithostratigraphywithin each horse by removing folds and/or thrusts,based on detailed sketches of the outcrops. The re-constructed lithostratigraphy within a horse is sim-ple and consists of, in ascending order, pillow ba-salt, chert, and turbidite (fig. 11). Major tectonicbreaks separating horses on both sides are com-monly layer-parallel faults. The stratigraphic topwithin a horse always faces southeastward, exceptin a few strongly folded areas.

Reconstruction of Lithostratigraphy within a Horsein the Southern Unit. The lithostratigraphy withineach horse was reconstructed and compared. Cen-timeter-scale measurements of chert and turbiditesequences were completed on all horses. The chertand turbidite are, in general, severely deformedcompared with metabasite, and hence particularlycareful observation is necessary for the reconstruc-tion of the original lithostratigraphy. Chert hasbeen subdivided into the following six types: (1)BIF-bearing, (2) white, (3) white-blue (magnetite-bearing), (4) white-green (mafic tuff-bearing), (5)muddy chert, and (6) white-gray (or brown becauseof weathering) chert containing thin, laminated,carbonate layers. The type 3 and type 4 cherts are

characterized by millimeter- to centimeter-scalerhythmic alternations.

The mineral assemblages and abundances in therocks reflect not only their metamorphic grade butalso their chemical composition. Mafic sediments(MS) can also be classified into the following fivetypes based on the appearance and amount of gar-net. If mafic sediment is more enriched in FeO and/or MnO relative to MgO, garnet is more abundantat a given metamorphic grade. Thus, the modalamount of garnet is a marker of the relative abun-dance of FeO. MS1 is black and massive, with fewgarnets. MS2 is also black but is differentiated fromMS1 by its rough surface and is garnet free. MS3 hasquite large amounts of garnet megacrysts and isgreen and massive. MS4 is also black but differentfrom MS1 in terms of larger modal amounts of gar-net. MS5 is black to gray with a considerableamount of garnet.

In spite of complex folding, the lithostratigraphywithin a horse can be reconstructed by using keybeds of chert and mafic sediment lithologies. A rep-resentative example is shown in figure 12band c. A wide variety of lithologies occurs in an

-m2 area. These include rhythmically al-80 # 30ternating white-gray chert, muddy chert, alternat-ing felsic sandstone and mudstone, mafic sedi-ments, carbonate, and conglomerate. The area is

532 T . K O M I Y A E T A L .

Table 1. Summary of the Structural Relationships within the Horse IV2

D0

D1

D2Folding Thrust

S0 F1a F2Sedimentary layering Upright to moderately inclined Moderately plunging

southwardModerately plunging northward SE-NW trendingTight to isoclinalWNW-ESE trending

F1bUpright to moderately inclinedModerately plunging southwardClosed to isoclinalNE-SW trending

NE-SW trendingVerging northward

NE-SW trendingVerging northward

F1c S2Upright to moderately inclined NE-SW strikeModerately plunging southward Steeply to moderately

verging northwardClosed to isoclinalNE-SW trending

ENE-WSW trendingVerging northward

F1dUpright to moderately inclinedModerately plunging southwardOpen to isoclinalNE-SW trending

subdivided into four blocks (blocks I–IV) by threeNE-SW trending, folded thrusts (T1 to T3 in fig. 12a).

T1, with NE-trends and SE-dips, runs along theboundary between MS4 and conglomerate (fig. 12b).T2, with NE-trends and SE-dips, constitutes theboundary between the MS4/carbonate and MS4/MS1

in the western area (fig. 12c), whereas it constitutesthe boundary between MS4 and conglomerate in theeastern area (fig. 12b). T1 is cut by T2 in the middlearea. T3, with ENE-trends and SSE-dips, runs alongthe boundary between the chert and MS4, and trun-cates T2 in the western area (fig. 12c).

Although older folds are obscured by later defor-mation, and even the faults themselves are tightlyfolded, at least four separate generations (F1a to F1d)can be observed, based on the direction and plungeof fold axes and crosscutting relationships withthrusts. Sedimentary bedding planes (S0) such asgraded bedding, intercalation of conglomerate withcarbonate, and MS4 with carbonate are well pre-served in this area.

F1a folds have been traced throughout the easternpart of this area by using MS3 and sandstone as

structural markers (fig. 12b). They are noncylin-drical tight to isoclinal folds, with wavelenghts ofca. 2 m, upright to moderately dipping, WNW-strik-ing axial surfaces and hinge lines plunging WNW.F1b folds are noncylindrical, closed to tight, withupright to moderately dipping, NNE-striking axialsurfaces and hinge lines plunging SSW. F1b foldshave been traced throughout the northern part byusing intercalations of conglomerate with carbon-ate or MS4 (fig. 12b, c). F1c folds have very distinctforms. F1c folds are noncylindrical, closed to tight,with upright to moderately dipping, NE-striking ax-ial surfaces and hinge lines plunging SW (fig. 12a).F1d folds are noncylindrical tight to isoclinal, withupright to moderately dipping, NE-striking axialsurfaces and SW-plunging hinge lines (fig. 12a).

The axial surface and the eastern limb of the F1b

fold are sharply cut by T2 in the middle area of blockIV (fig. 2). Moreover, the axial surface and the east-ern limb of the F1b fold are cut by T1, and its westernrim is cut by T2 in block III (fig. 12b). The axialsurfaces of F1b folds sweep to the west along thrusts,and interlimb angles decrease near the thrusts. F1c

Figure 13. Schematic NS profile of chert-turbidite unit in figure 12 and the reconstructed lithostratigraphy. Note the stratigraphicchange from basalt and chert, through muddy chert to turbidite, including conglomerate, and the upward coarsening of thesedimentary sequence.

Figure 14. A sketch of horse III2 showing the stratigraphic relationship between the pillow lava flow unit, hyalo-clastite, and reworked hyaloclastite, dolomitic carbonate pods, and massive white chert beds on the top. Location offigure 5 is shown.


Figure 15. Well-preserved pillowed basaltic flows, hyaloclastite, reworked hyaloclastite, and the overlying beddedchert from the right to the left side. The geologist stands on the chert beds. The locality is shown on the sketch(horse III2) of figure 14.

folds bend the axial surfaces of F1a and even faultsurfaces of T1 and T2. In contrast, F1c folds are trun-cated by T3 (fig. 12c). F1d fold bends T3 in the west-ern area (fig. 12a). The foliations (S2) in this areaare mainly NE-SW striking, dip moderately tosteeply to the southeast, and appear to be folded,with steeply SE-plunging axes (F2; fig. 12a inset).All the rocks in this area uniformly suffered fromlower-amphibolite metamorphism, zone D (fig. 8).This means that juxtaposition of those blocks pre-dates regional amphibolite metamorphism. Super-position of the folds and thrusts allows us to re-construct the original stratigraphy. The thrust andfolding events are summarized in table 1.

Block I contains a 3–5-m-thick bedded chert layercapped by alternating felsic sandstone and mud-stone, through a muddy chert layer a few metersthick. The bedded chert is intercalated with thinMS1 layers in the lower part (fig. 12c). Block II isdominated by MS4 with interbedded MS1, MS3, MS5,sandstone, and carbonate. Felsic sandstone is a keybed to correlate the stratigraphy of block I withblock II where it is interlayered with MS4 in thelower part. Toward the stratigraphic top, mafic sed-iments dominate, exhibiting a variety of differentlithologies. MS3 layers occur in the stratigraphic

upper part, while MS1 and MS5 occur in the middlepart. Thin carbonate layers are interbedded in theupper part. Among them are 10–20-cm-thick Fe-rich mafic layers containing remarkably large (ca.3-cm across) garnet crystals (MS3, MS4), a markerhorizon. This aids not only analysis of the foldedstructures, but also comparison of the lithostratig-raphy between blocks (fig. 12b, c). The overturnedblock IV starts from MS4 interlayered with sand-stone layers through alternations of MS4 and car-bonate layers, finally up to a 4-m-thick conglom-erate layer (fig. 12b, c). Block III is composed ofalternating conglomerate and thin MS4 and MS2

layers with minor intercalations of carbonate (fig.12b). Although all blocks are metamorphosed toamphibolite grade, some sedimentary structures,such as cross lamination and graded bedding, arewell preserved (Nutman et al. 1984). Using theseto determine the directions of the stratigraphic top(Nutman 1986), the lithostratigraphic sequence canbe reconstructed based on the above thrust/foldinggenerations and comparison of the lithostratigra-phy within each block (fig. 13). The total thicknessof the turbidite sequence is estimated to be about70–80 m. However, because of considerable tec-tonic thinning, the original thickness may have

536 T . K O M I Y A E T A L .

Figure 16. Lithostratigraphies of duplexes I–VIII in comparison with that of Mesozoic Japan (Isozaki et al. 1990)

been much more. Deformation is measured usingshapes of pebbles in conglomerate layers, indicatinga strain ratio of 1 : 1.23 : 0.33. If such a strain ratiois applied to all lithologies, the original thicknesscan be estimated to be about 200–300 m.

Compared with reconstruction of the chert-turbidite association, it is easy to reconstruct thestratigraphy of the basaltic flow piles capped bychert because the basalts are less deformed thanthe other rocks, as shown on the geologic map. Aclose-up view of one example is shown in figures14 and 15. A N-dipping steep wall, m2, is80 # 300exposed, exhibiting a complete succession from pil-low at the bottom through the hyaloclastite layersto massive white chert on top. Along the boundarybetween chert and hyaloclastite are thin dolomiticcarbonate pods. The pillow lavas are relativelysmall, up to 80 cm across. Morphologies range fromoval with cusp to pear shaped, relatively undefor-med pillow lavas, indicating younging toward thesouthwest. The pillow lava-hyaloclastite flow unit

rests on another hyaloclastite layer in the footwall(fig. 14). The hyaloclastite layers on the top are di-vided into upper and lower hyaloclastite sections,respectively. The lower hyaloclastite layer is up to9 m thick and characterized by a smooth surfaceand lack of garnet megacrysts. It consists of poorlysorted pillow breccias set in matrix and conform-ably covers the pillow lavas. The upper hyaloclas-tite layer is up to 3 m thick, varying laterally inthickness, and composed of well-sorted, fine-grained metasediments, probably reworked. It ischaracterized by rough surfaces and garnet mega-crysts. High abundance of garnet suggests that therocks are enriched in Fe and Mn. It conformablycovers the lower hyaloclastite layer and is conform-ably overlain by white bedded chert layers. Fivesubparallel dikes cut all the pillow lava flows andthe overlying sedimentary units. Massive whitechert beds contain thin films of either carbonate ormafic sediments near the base.

Through detailed mapping, we reconstructed

Figure 17. a, Schematic diagrams of oceanic plate stratigraphy and the travel history of an oceanic plate, after Isozaki et al.(1990). When the plate is formed at a mid-ocean ridge, it is composed of basaltic lava flows with or without the overlying carbonatepods. The bedded chert is deposited on the oceanic crust during its migration from the ridge to the trench. Once ocean islandbasalt (OIB) volcanism occurs on the way between the ridge and trench, its product is deposited on or intruded into the basalticflow piles and the overlying bedded chert, and bedded chert is deposited on the complex again. Hemipelagic sediment includingchert and some terrigenous pelitic sediments is deposited near the outer trench, and finally turbidite covers all units at the trench.Note the great similarity of lithostratigraphy in the Isua belt to that of Japan. b, Mode of accretion of subducted oceanic materials,modified after McClay (1992). As a result, a duplex belt is formed from top to bottom, by successive accretion with time. Eachduplex consists of several horses with similar oceanic plate stratigraphy. The duplex is cut on the top and bottom by roof- andfloor-thrusts, respectively. Deformation concentrates in unconsolidated to semiconsolidated sediments, heavily deformed in theturbidite and deep-sea sediments, and less deformed in the competent oceanic crust. c, A schematic diagram of accretion of oceanicmaterials at trench.

538 T . K O M I Y A E T A L .

Table 2. Summary of the Age of the Supracrustal Rocks and Amıtsoq Gneiss

Age (Ma) Lithology Analytical methods

In the map area:3710 5 4 Pyroclastic rocks SHRIMPa

3708 5 3 Felsic volcanics SHRIMPb

3760 5 70 BIF (Northern Unit) Pb-Pb isochronc

3730 5 150 Metabasites (pillowed basalt unit) Sm-Nd isochrond

3750 5 40 Volcaniclastic metasediments Sm-Nd isochrond

Outside of our map:Isua supracrustal belt:

3707 5 6 BIF SHRIMPa

3806 5 2 Felsic volcanics SHRIMPa

3711 5 6 Detrital zircons from mafic sediments SHRIMPa

3900–3820 Detrital zircons from quartzite SHRIMPe

3807 5 1 Detrital zircons from conglomerates SHRIMPf

3798 5 4 Tonalite sheet cutting supracrustal rocks SHRIMPa

3791 5 4 Tonalite sheet cutting supracrustal rocks SHRIMPe

Amıtsoq gneiss:3695 5 5 Outer Amıtsoq gneiss SHRIMPg

3702–3627 Inner Amıtsoq gneiss SHRIMPg

3782–3811 Amıtsoq gneiss to the southern SHRIMPg

Note. iron-formation.BIF = bandeda Nutman et al. 1997.b Nutman et al. 1996.c Moorbath et al. 1973.d Hamilton et al. 1983.e Nutman et al. 1997.f Compston et al. 1986.g Nutman et al. 1993.

lithostratigraphic columnar sections for each horse(fig. 16). A group of horses with similar stratigraphyis regarded as belonging to a particular duplex.Eight duplexes were recognized. All of the horsesin the given duplex (e.g., III) share the same stra-tigraphy and chert thickness except for duplex IV,in which two share both the same stratigraphy andchert thickness, but the third shares only the stra-tigraphy with the others. Duplex II has a distinctstratigraphic feature, which implies an oceanic pla-teau or seamount origin (for details of lithography,see below).

Duplex I, at the structural base in the mappedarea, is cut by a 507 E-dipping fault across which itis juxtaposed against the inner Amıtsoq gneiss (fig.10). About 50-m-thick white chert beds and over-lying mafic turbidite sequences are exposed, but theunderlying pillow lava units are mostly cut off bythe fault (fig. 1).

Duplex III has three horses, each characterizedby thick-bedded chert (30 m). The basaltic layer iscomposed of a few pillow lava units, and a turbiditelayer marking the top of the sequence is domi-nantly mafic in composition. Along a lithologicboundary between chert and mafic turbidite, a fewlayers of carbonate are present, each a few tens ofcentimeters thick (fig. 16).

Duplex IV is formed by three horses, all withthick turbidite-containing felsic sandstone,

mudstone, and conglomerate in addition to mafic-carbonate sediments. Conglomerate layers associ-ated with this duplex are traceable 3–4 km alongthe strike. The conglomerate layers may be re-lated to dacite (calc-alkaline [CA] volcanics)boulder–bearing, thick felsic turbidite layers, about12 km to the southwest. Zircons from dacite peb-bles yielded a well-constrained igneous age of

Ma (Compston et al. 1986). The chert3807 5 1thickness in two of the horses is about 5 m, butthat of the third is 40 m (fig. 16). The chert litho-stratigraphy is also distinctive; white chert domi-nates in IV1 and IV2, while the chert sequence inIV3 is dominantly composed of blue/white chertwith some layers of green/white and white chertand even tuff and soapstone layers. Nevertheless,the prominent muddy chert layers at the top of thechert sequence and the lithostratigraphy of the tur-bidite sequence overlying the chert are the same inall horses, suggesting the simultaneous accretionof all three at a trench (Kimura 1994; Kusky et al.1997b).

Duplex V is characterized by white chert withan overlying thick blue-white chert unit (total: 40m). Near the base of the chert bed, fuchsite occursin some localities. Along the boundary betweenchert and underlying thick hyaloclastite layers, athin layer of talc-bearing ultramafic rock appears.

Figure 18. Travel history of Early Archean oceanic lithosphere in the Isua accretionary belt, West Greenland. This diagram showspaleogeographic reconstruction of accreted oceanic lithosphere, being followed by successive underthrusting of oceanic crust toform duplexes from duplex VIII to I with time.

540 T . K O M I Y A E T A L .

Figure 19. Comparative paleogeography of the Phaner-ozoic arc-trench system, ridge subduction (a) versus oldplate subduction (b) after Maruyama (1997). On the ridgesubduction, regional uplift and forearc calc-alkaline (CA)volcanism occur because of hot and buoyant plate sub-duction. Large amounts of sediments, including con-glomerate, are supplied to the trench. The wide lateralextent of slab melting leads to more active island arcvolcanism.

The turbidite layer is dominantly mafic in com-position with minor felsic sediments (fig. 16).

Duplexes VI, VII, and VIII are not well exposedand have been affected by thermal metamorphismof the 2.2-Ga (Nutman et al. 1995) boninitic dikeintrusions. This makes it difficult to unravel thedetailed lithostratigraphy and the distribution oftheir horses. However, a few critical exposuresclearly indicate the presence of different duplexeswith different lithostratigraphies. All are charac-terized by different chert thicknesses and differentlithologies. Duplex VI is characterized by thin-bed-ded blue-white chert and varied lithologies in theturbidite layers; centimeter- to meter-scale alter-nating mafic, carbonate, felsic, and ultramafic sed-iments (fig. 16). Duplex VII is characterized by thinwhite chert layers and both thick felsic and maficsediments with distinctive carbonate layers (fig.16). Thick chert layers and the very thick overlyingmafic sediments with ultramafic materials are pres-ent in duplex VIII (fig. 16).

Duplex II has a thick (up to 600 m) lithostratig-raphy, remarkably different from the others becausechert layers appear in two stratigraphic horizons(fig. 16). Between the lower 50-m-thick and the up-per 2-m-thick bedded chert layers is a 190-m-thickrelatively undeformed mafic-felsic alternationcharacterized by bimodal volcaniclastic felsic andmafic compositions. Mafic sediment contains nu-merous garnet megacrysts. Recent SHRIMP datingof separated zircons reveals that the felsic volcan-iclastic rocks were formed at Ma (Nut-3710 5 4man et al. 1997). In the lower section, the alter-nation is rhythmically and finely spaced. Near thebase are tourmaline boulders (Appel 1984, 1995).In the middle section, each mafic and felsic sedi-ment layer becomes up to 5 m thick. Four maficsills are intruded into felsic layers or the boundarybetween felsic and mafic sediments. In the uppersection, mafic sediment becomes dominant, inter-bedded with six felsic sediment layers, about 1 mthick. Thin mafic sediment layers without garnetmegacrysts are interbedded with thick mafic sed-iment layers with garnet megacrysts. A package ofthin magnetite and green phyllite layers is inter-calated into one of the thick mafic sediment layers.Near the top, a unit of very thick green phyllitewith felsic sediment layers on both sides covers thealternation of mafic sediment layers. At the top, amafic sill is intruded. The upper 70-m-thick sedi-ment layer overlying the thin chert layer is domi-nantly mafic, and one thin felsic sediment layer isinterlayered near the bottom. Mafic sediments con-tain no garnet megacrysts, suggesting a more Mg-rich bulk composition. Picritic lava flows are also

present. The uppermost part is a thin alternationof felsic sandstone and mudstone (fig. 16).

Comparison of Lithostratigraphy of the Isua Belt withOceanic Plate Stratigraphy. Petrographic exami-nation of the chert in the Isua duplexes indicatesthat no continental or arc materials occur withinchert beds and underlying pillow lava flow units.Detrital materials start to appear in thin laminatedchert beds precipitated after the end of major chertdeposition, and become common in the turbiditelayers. The sequential change of sedimentary as-pects, from chert to turbidite, is common to allhorses in the Isua belt. The reconstructed litho-stratigraphy is quite similar to that of circum-Pa-cific accretionary complexes, which were formedby oceanic plate subduction during the last 450m.yr. along the peripheral parts of the proto-Pacificto the present-day Pacific Ocean (Isozaki et al.1990; Matsuda and Isozaki 1991; fig. 16).

The reconstructed lithostratigraphy in an accre-tionary prism in those Phanerozoic orogens iscalled “oceanic plate stratigraphy.” The oldest ex-isting oceanic segment, off the Mariana arc, is 200m.yr. old and forms part of the Pacific plate, whichis being subducted under the Izu-Mariana intra-


oceanic arc. Along the western margin of North andSouth America, spreading centers themselves arebeing subducted under the continents (e.g., von Hu-ene et al. 1997). Therefore, crudely speaking, theaverage lifetime of oceanic lithosphere from itsbirth at a mid-oceanic ridge to its demise in thetrench is 100 m.yr. The deposition rate of beddedchert, estimated by radiolarian microfossils in Tri-assic to Jurassic bedded chert in Japan (Matsuda andIsozaki 1991), is 1 m m.yr.21 Thus, a 100-m-thickchert bed corresponds to chert deposited over 100m.yr. Figure 17a shows a simplified oceanic platestratigraphy and a travel history of the 100-m.yr.-old oceanic plate. When a plate is formed at a mid-oceanic ridge, the oceanic crust is composed oftholeiitic basaltic flow piles. The topmost crust ispartly covered by a thin bed of limestone where theridge crest is above the carbonate compensationdepth. Hydrothermal circulation of seawater at thespreading center alters the oceanic crust (oceanridge metamorphism) and forms massive sulfide de-posits through black and white smokers locally de-veloped on the ridge. Because of horizontal move-ment by ocean-floor spreading, the plate cools withtime, the water depth increases, and deep-sea sed-iments precipitate on the ocean floor until the platearrives at the trench. During the 100 m.yr. of travelfrom mid-oceanic ridge to subduction zone, onlydeep-sea sediments are precipitated on the oceanfloor. These sediments are bedded cherts formed byrelatively rapid and periodical deposition of “ma-rine snow” (radiolarians). Near the trench, conti-nent- or arc-derived turbidite starts to cover theocean floor. The turbidite is made of very fine-grained mudstones mixed with chert (hemipelagicsediments), outboard of the trench outer wall andcovered by thick coarse-grained turbidites at thetrench axis. The 100-m.yr.-old oceanic crust cappedby 100-m-thick chert and turbidite sequence is thensubducted and accreted to the arc either by off-scraping at shallow levels or underplating at deeperlevels.

The accretion of subducted oceanic materials ineither off-scraping or underplating is through layer-parallel shortening (fig. 17b). A competent basalticoceanic crust overlain by semiconsolidated to un-consolidated sediments is underthrust along the de-collement by oceanic plate subduction. The releaseof pore fluid, strain hardening, and compactionforce the frictional resistance along the decolle-ment to increase and cause down-stepping of thedecollement. Development of decollement down-stepping at this stage results in the accretion ofoceanic crust and the formation of duplex struc-tures made of thrust-bounded horses. The newly

accreted oceanic crust pushes up the already ac-creted horses and forces them to rotate landward.The repetition of the above process forms a duplex,which consists of several horses (fig. 17b).

As a result of duplex growth, the newly devel-oped decollement episodically advances oceanwardto push up the already accreted duplexes and causesthe formation of a younger duplex under the olderduplex (fig. 17c). Accretionary orogens developdownward and oceanward with time. Because ofthe contrasting physical properties of componentrocks, deformation concentrates in turbidite se-quences and deep-sea sediments, rather than inrigid basaltic crust at shallow levels. If seamounts,rises, or plateaus were formed on a plate, thosewould be sandwiched on the top and bottom bydeep-sea sediments before accretion at the trench.In addition, an abundance of sills and dikes wouldbe intruded into deep-sea sediments if oceanic is-land magmatism developed. All of those materialsformed by ocean island basalt (OIB) volcanismwould be finally exposed on land, but are quite dif-ferent from the normal oceanic crust and overlyingsediments. Systematic rules on the mode of occur-rence of accreted oceanic materials with differentorigins are summarized by Isozaki et al. (1990).

Travel History of Early Archean Oceanic Lithospherein the Isua Belt. In the Southern Unit, and possiblyin the whole Isua supracrustal belt, an accretionarycomplex has formed through accretion of oceaniccrust to the hangingwall by duplexing. A sequenceof duplexing scenarios can be reconstructed asfollows.

The structural bottom is duplex I, and the top isduplex VIII, as judged by the southward polarity ofduplexing. As shown in figure 17b, the youngerhorses develop downward; therefore, the oldest du-plex is VIII. Accretion proceeded from duplex VIII,through VII, VI, V, IV, III, and II, to I. Within a givenduplex, the southeastern horse is the oldest. Thesimilar lithostratigraphy within each horse sug-gests that those oceanic crust segments subductedin the same area and during the same period. Themode of formation of the duplex structure throwslight on the direction of the growth of the accre-tionary prism and the subduction of oceanic lith-osphere. The duplex polarity is defined by the di-rection of convergence of faults, which dividehorses, and coincides with plate convergence (fig.17b). The southward convergent morphology of theduplex structure is widespread over the mappedarea, indicating that the Isua accretionary prismgrew from south to north.

Radiometric ages are not fully available for es-tablishing the detailed chronology within each


Figure 20. a, Geothermal gradients of oceanic lithosphere for Earth and Venus at present and in the Archean. Thetransitions of ductile/brittle of peridotite and diabase (thick lines) are calculated as function of temperature andpressure for a given steady state strain rate of a representative tectonic strain rate, 10214 s21 (after Hoffman and Ranalli1988). Oceanic geotherms for different lithosphere ages given in units of millions of years on the plots are calculatedbased on model of transient half-space cooling (Hoffman and Ranalli 1988). Note that surface temperature on Venusis 4707C higher than on Earth. b, Strength profiles in compression for modern oceanic lithosphere as a function ofage given in units of millions of years on the plots. The 7-km-thick oceanic crust is modeled as diabase and the uppermantle as peridotite. c, Strength profiles in compression for Archean oceanic lithosphere, 10 m.yr. old (left) and 20–50m.yr. old (right). The 15-km-thick crust is modeled (Ohta et al. 1996). d, Strength profiles in compression for Venusianoceanic lithosphere, 30 m.yr. old (left) and 50 m.yr. old (right). The 20-km-thick crust is modeled (Buck 1992).

horse. Currently available constraints on the age ofthe supracrustal rocks and Amıtsoq gneiss near theIsukasia are shown on table 2. SHRIMP data revealthat felsic sediments (bimodal volcanism on sea-mount) in duplex II were formed at 3710 Ma, theonly utilizable constraint for the age of depositionof oceanic materials before accretion. In contrast,the igneous ages of the Amıtsoq tonalitic gneissesare 3627–3702 Ma in the inner area and 3695 Main the outer area. SHRIMP data are consistent withthe accretionary model because oceanic material isolder than tonalite. Judging from these data, the ageof subduction of oceanic crust must be before 3702Ma and is probably in the range of 3702–3710 Ma.Tonalitic granitoids, precursors of Amıtsoq gneiss,were subsequently intruded into the Isua accre-tionary complex around the Isua supracrustal beltat 3627–3702 Ma.

A paleogeographic reconstruction (fig. 18) illus-trates sequential events of accretionary prisms inthe Southern Unit at about 3700–3800 Ma. Thearrangement of oceanic lithosphere is estimated bythe accretionary scenario described above. Twomid-oceanic ridges are assumed because the chertthickness of duplexes VI and IV (except IV3) is verythin (4–6 m), whereas the others are all 30–50 m.Therefore, a very young oceanic plate must havesubducted at the trench to form duplexes VI andIV, suggesting the presence of a mid-oceanic ridgenearby. In spite of the difference in chert thicknessbetween IV1 and IV2 (4–6 m) and IV3 (up to 40 m),lithostratigraphy of the turbidite sequence in IV3 isvery similar to those of IV1 or IV2. This may suggestthe presence of a large transform fault with litho-sphere of differing ages on either side. Duplex II ispresumably a large oceanic plateau or seamountformed by intraoceanic off-ridge volcanism becausechert layers occur along two stratigraphic horizonsand bimodal volcaniclastic sediments are interca-lated with chert beds. The thick lower chert layerof duplex II suggests that its lifetime was long andthat an old plate subducted and accreted to the han-

gingwall. In contrast, the thin upper chert layersoverlying the mafic-felsic sediments suggest thatwithin-plate volcanism occurred near the trench.

The thin (5 m) chert beds in horses IV2 and IV3

imply subduction of a very young oceanic plate (fig.19a). If a sedimentation rate typical of Phanerozoicchert is applicable, it may have been ∼5 m.yr. In amodern tectonic setting, for such young plate sub-duction, the volcanic front of the CA rock seriesmust be very close to the trench and near-trench-volcanism must be present, as shown in presentand past examples (e.g., Tertiary southwestern Ja-pan; Takahashi 1986). When a young, hot, platesubducts, the hangingwall is uplifted on a regionalscale because of buoyancy. This has been well doc-umented in young orogenic belts around the cir-cum-Pacific region, e.g., Late Tertiary southwest-ern Japan (Takahashi 1986) and the Early TertiaryAlaska (DeLong et al. 1978; Kusky et al. 1997a). Inthese cases, thick piles of turbidite cap the sub-ducting oceanic crust before subduction. It is worthmentioning the lithostratigraphy of horses IV2 andIV3, which are characterized by a thick quartzo-feldspathic turbidite sequence that contains mafic,felsic, and silicic pebbles in several interlayeredconglomerates (fig. 3). Large dacite cobbles up to50 cm across have been described from this hori-zon, and their petrochemical characters show typ-ical Archean CA series rocks mentioned above(Nutman 1986). Figure 19 shows the arc-trench pa-leogeography during ridge subduction in compari-son to normal plate subduction. Both regional upliftand forearc volcanism enabled the input of felsicmaterials into the trench.

Rigidity of Oceanic Lithosphere in the Early Ar-chean. Evidence of accretion of oceanic materialsat the subduction zone indicates successive under-thrusting of oceanic crust and suggests that tec-tonics similar to Phanerozoic plate tectonics op-erated in the Early Archean. It does not, however,necessarily prove that modern-style plate tectonics,in a strict sense, were actually at work. Rigidity of

Figure 21. N-S cross section of the southwestern Japanese arc (a) in comparison with NW-SE cross section of the Akulleq terranenear Isukasia, West Greenland (b). Note the 100-Ma granite-greenstone terrane in Japan. Successive oceanward growth of arc causedby addition of new crust by TTG makes the older accretionary complexes “enclaves” within TTG plutons. Also note the 100-Ma gneiss terrane that has been derived from granite-greenstone terranes formed in the lower crustal levels by synchronous TTGplutonism.


oceanic plates is also necessary to discern whatkind of tectonics operated—modern-style plate tec-tonics or a hybrid of plume and plate tectonics(Head and Crumpler 1990).

Pillowed basalt at the bottom of the oceanic platestratigraphy in the Isua supracrustal belt indicatesthat oceanic crust was not komatiitic but basalticin composition even in the Early Archean, seawaterwas present, and open sea extended far away fromoceanic island arc. Numerical simulation of tem-perature and composition of ocean and atmospheresuggests that formation of a protoocean decreasesthe atmospheric pressure to ca. 10 atm and the sur-face temperature to ca. 207C (Tajika and Matsui1990). These numbers support the idea that the sur-face temperature was cooled down at !1007C in theEarly Archean. Moreover, recent petrological in-vestigation of the Middle Archean MORBs indi-cates that potential temperature of the MORB-source mantle was about 1207C higher than that ofthe modern upper mantle and that the thickness ofthe oceanic crust was about 15 km (Ohta et al.1996). Knowledge of composition and thickness ofoceanic crust and temperatures of the mantle andthe Earth’s surface allows us to reconstruct thestructure of the oceanic plate (fig. 20). The oceanicgeotherm for the Early Archean lithosphere as afunction of age was calculated based on a model oftransient half-space cooling (Turcotte and Schubert1982) at given mantle and surface temperature pa-rameters of 14507C (Ohta et al. 1996) and 1007C,respectively. The corresponding parameters for pre-sent-day Earth are 13307C and 07C (Iwamori et al.1995), and for Venus they are 14507C and 4707C(Buck 1992), for comparison. In addition, transi-tions of ductile-brittle of peridotite and diabase arecalculated as a function of temperature and pres-sure for a given steady state strain rate of a repre-sentative tectonic strain rate, 10214 s21 (after Hoff-man and Ranalli 1988). The rheological structuresof oceanic lithospheres as functions of age are es-timated from the oceanic geotherms and the tran-sitions of ductile/brittle (fig. 20b–d).

The estimated structure indicates that an oce-anic plate older than 20 m.yr. behaves rigidly evenin the Early Archean (fig. 20c, right), while a muchyounger plate has a ductile basaltic zone betweenbrittle basaltic crust and peridotitic mantle (fig.20c, left). It is well known, however, that youngocean crust is efficiently cooled by active hydro-thermal circulation around a spreading center (seethe summary by Alt 1995), which suggests that oce-anic plates !20 m.yr. were hardened because of ef-ficient freezing by extensive hydrothermal circu-lation at oceanic ridges analogous to modern

ridge-hydrothermal systems and on-land ophiolite(Ishizuka 1985). It is consistent with occurrencesof linear fractures in which basaltic dikes intrudeda pile of pillow lava flows (fig. 14). This evidencesuggests that oceanic plates in the Early Archeanwere rigid, in contrast to the substantial ductilesurface of Venus (fig. 20c).

Previous workers suggested that oceanic plate inthe Archean was thinner than present because ofhigh mantle temperature, assuming that komatiiterepresents an upper mantle magma (e.g., Arndt1983; Hoffman and Ranalli 1988). The present es-timate also suggests that oceanic lithosphere at agiven age in the Early Archean was about 80% thin-ner than at present.

Discussion

Except for a few pioneering works (Folinsbee et al.1968; Condie 1972; Hoffman 1973; Burke et al.1976), most Archean geologists in the 1970s did notbelieve in the presence of layer-parallel shorteningstructures in Archean orogens. Such structureshave now been described in many Archean oro-genic belts (Friend et al. 1988; Nutman et al. 1989;Maruyama et al. 1991b, 1994 for southern WestGreenland; Kimura et al. 1993, for the Superiorprovince Kusky 1989, 1990, 1991 for the Slave Prov-ince; de Wit et al. 1987a, 1987b, 1992 for the Bar-berton Group, South Africa; Isozaki et al. 1991; Ki-mura et al. 1991; Maruyama et al. 1991a for PilbaraCraton, Western Australia). To explain such nappe-stacked orogenic structures, plate tectonics hasbeen preferentially adopted as a possible explana-tion (e.g., Kusky and Kidd 1992). However, large-scale subhorizontal shortening does not necessarilymean subduction of oceanic or continental platesbecause such structures have been also describedwithin continental plates, for example, in the fore-land basins of the Cenozoic Alps–Himalayas andCaledonides in Europe and in the eastern UnitedStates. To define an accretionary complex, in ad-dition to documentation of layer-parallel shorten-ing structures in orogenic belts, the oceanic platestratigraphy within the accretionary prism is crit-ical because it means gradation of deposition of vol-canics and sediments from open sea to continentalmargin. Hoffman (1991) showed the lithologicalsimilarity of some Archean orogenic belts to thePhanerozoic southwestern Japan arc and speculatedthat Archean granite-greenstone belts were formedas accretionary complexes. However, evidence oflithological similarity is insufficient to define anaccretionary complex because basalt, chert, and ter-rigenous sediments are also deposited, even in con-

546 T . K O M I Y A E T A L .

Table 3. Simplified Summary of the Magmatic, Structural, and Metamorphic Events in the Isua Area

Age (Ma) Periods Event Notes

1600 (5) Events afteramalgamation

Ataneq Faulta

2214 Intrusion of Proterozoic noritic dikeb

2550 Emplacement of the Qorqut granitecomplexc

Postcollision granite

ca. 2700 Regional metamorphismd Greenschist to amphibolitefacies metamorphism

2712 (4–1) Collision-amalgama-tion of arcs

Intrusion of Qarusuk granitic dikese

Collision-accretion to Akia terranef

Collision-accretion to Tasiusarsuaqterranef

Syn-collision granite

2820 (4–2) Rifting of arcs Intrusion of the precursor of Ikkattoqgneissf

Rifting?

3100–3400 Intrusion Tarssartoq dikesg Rifting?(3) Tonalite intrusion Intrusion of polyphase tonalites (Amıtsoq

gneisses)hTTG intrusion

(2) Subduction-accretion Accretion through off-scraping/underplatingi

Formation of duplexstructure

∼3812–3600 Subduction of oceanic crusti

(1) Formation of oceanicplate

Deposition of trench-fill turbiditei

Deposition of BIF/chert in open seaenvironmenti

Ocean-floor metamorphism?i

Formation of oceanic crusti

Formation of oceanic platestratigraphy

a Nutman et al. 1989.b Nutman et al. 1995.c McGregor 1973.d Nutman and Collerson 1991.e McGregor et al. 1983.f Frend et al. 1988.g Nutman et al. 1983.h Nutman and Bridgwater 1986.i This article.

tinental rift zones, and because ubiquitous occur-rences of layer-parallel faults have not beendemonstrated.

A continental rifting model has been used to ex-plain the geology of the Isua supracrustal belt. Itsuggests that sedimentary sequences, includingconglomerates, were deposited conformably inshallow marine environments, with mafic volcan-ics intruded into or extruded on them (e.g., Nutman1986; Dymek and Klein 1988; Jacobsen and Dymek1988). More recently, the tectonic history of south-ern West Greenland was accounted for by the ter-rane concept of collision-accretion (Coney et al.1980), based on the presence of three fault-boundedblocks with a geological history distinct from thoseof adjacent terranes (Friend et al. 1987, 1996; Nut-man et al. 1989). Such a model takes into accountthe growth of continental crust through collision-amalgamation after the formation of the continen-tal crust. However, a continental rifting model isstill not ruled out because the terrane model doesnot account for the tectonic setting where basalticlavas, chert, and turbidite with conglomerate were

deposited or for the emplacement mechanism ofthose rocks onto the continental crust.

Nutman and Collerson (1991) first used the term“accretion complex” for the Isua supracrustal beltbased on the similarity of Isua supracrustal lithol-ogy to that of Phanerozoic accretionary complexes.But lithology alone is insufficient to define an ac-cretion complex. The present report is the first de-tailed documentation of duplex structure and oce-anic plate stratigraphy in an Archean orogenic belt,features necessary to define an Archean accretion-ary complex comparable to those in the Phan-erozoic.

Geologic records within turbidite sediments alsogive strong constraints on types of active plate mar-gins. Turbidites formed along a large continentalmargin would be dominantly quartzofeldspathic. Ifthey formed in intraoceanic environments, such asthe western Pacific domains, they would be dom-inantly mafic in composition because no big riverssuch as the Yangtze or the Yellow River would bepresent to transport huge amounts of continentalmaterials to trenches. The sedimentary petrology


of the turbidite sequences clearly indicates that theIsua accretionary complex was formed in an intra-oceanic environment because mafic sediments aredominant in the turbidite sequences. It has longbeen evident that isotope signatures of Isua meta-basites and Amıtsoq gneiss do not support a pre-existing continental crust (Moorbath and Taylor1981). During Archean times, especially in theEarly Archean, most orogenic belts presumablyformed adjacent to intraoceanic island arcs, thengrew gradually through the collision-amalgamationprocess of island arcs (Condie 1982). Our presentdata support the idea of Condie (1982), Nutman etal. (1996), and Kusky and Vearncombe (1997).

The 3.7–3.8-Ga accretionary complex of the Isuasupracrustal belt was formed near a CA volcano-plutonic region as indicated by the felsic turbiditesitting on the top of the turbidite sequence, as wellas dacite pebbles in a stratigraphic horizon of du-plex IV. Those petrochemical aspects are quite sim-ilar to those of Phanerozoic TTG (tonalite-tron-dhjemite-granodiorite) along the circum-Pacificorogenic belts, although details are different, e.g.,La/Yb ratio and Eu-anomaly (Martin 1986; Nutmanand Bridgwater 1986; Shimizu et al. 1988). TheAmıtsoq gneiss and dacite pebbles may be one ofthe oldest continental materials formed by oceanicplate subduction. Orogenic belts grow with timethrough plate subduction during which CA vol-cano-plutonism creates and develops new acidiccontinental crust. As the orogen grows, the trenchshifts oceanward with time. The Japanese islandsare such an example. They have grown about 400-km oceanward during the last 400 m.yr. by the for-mation of accretionary complexes and episodic in-trusion of TTG every 100 m.yr. Therefore, as aninevitable result, the older accretionary complexrocks are now present as enclaves within grani-toids. The Cretaceous southwestern Japan is thePhanerozoic analogue of granite-greenstone terrainin the Precambrian. The middle Jurassic and latePermian accretionary prisms occur as enclaveswithin huge batholiths of TTG (fig. 21). The Cre-taceous gneiss terrain is a regionally metamor-phosed granite-greenstone terrain, which is alsoequivalent to the Precambrian gneiss terrains (Hoff-man 1991).

Table 3 summarizes the Isukasia area’s geologicalevolution based on lithostratigraphy, deformation,and metamorphism. The Archean history of theIsukasia area spanned 11 G.yr. and can be subdi-vided into five periods: (1) formation of the oceanicplate at the oceanic ridge and its subsequent mi-gration to the trench; (2) subduction and accretionof oceanic crust to form duplex structures; (3) con-

tinental growth of tonalite-trondhjemite-grano-diorite (TTG) plutonism (Amıtsoq tonalite); (4) rift-ing and collision-amalgamation of arcs, includingrifting-related intrusion of basic magma (Tarssartoqdikes) and TTG plutonism (Ikkattoq granodiorite),collision-accretion of the Akulleq terrane with theAkia terrane and the Tasiusarsuaq terrane (Friendet al. 1987), and collision-related regional meta-morphism; and (5) events after terrane amalgama-tion, including postcollision granitic magmatism(McGregor 1973), Proterozoic noritic dikes (Nut-man et al. 1995), and Ataneq strike-slip fault (Nut-man et al. 1989).

The number of regional metamorphism events isstill controversial. Previous workers had suggestedthat the Isua supracrustal belt suffered regionalmetamorphism more than two times in the Earlyand Late Archean since intrusions of hot granitoidslikely left metamorphic imprints (Nutman 1986).Boak and Dymek (1982) and Boak et al. (1983) con-sidered the main stage of amphibolite facies meta-morphism to be Early Archean, but they did notpresent compelling evidence, as pointed out byNutman (1986). Recently, SHRIMP dating of detri-tal zircons revealed only two ages, Early Archeanigneous core and Late Archean metamorphic rimwithin overgrown zircons (Nutman and Collerson1991). In the present work, microprobe analysesshow no evidence of relict minerals of the preced-ing higher-grade metamorphism even in the lowest-grade area of the greenschist facies, although asmall amount of chlorite retrograded from garnetand hornblende occurs at higher grades (amphibol-ite facies). Moreover, Ca-amphiboles in the high-grade amphibolite facies zone possess knife-edge-bounded zoning of actinolite in the core andhornblende in the rim. This evidence suggests pres-ervation of prograde metamorphism instead of theoverprinting of retrogressive metamorphism. Thezonation pattern of progressive metamorphism andthe trend of the foliations is not consistent withthe shape of duplex structures (figs. 8, 10). The my-lonite zones between the Amıtsoq gneiss and theIsua supracrustal belt did not undergo greater meta-morphism than greenschist grade. This indicatesthat the regional greenschist- to amphibolite-grademetamorphism postdated the formation of the du-plex structures and predated the final emplacementof the Amıtsoq gneiss with contact of the Isua su-pracrustal belt, probably at 2.7 Ga.

The direction of zoned growth of accretionarycomplexes in the Isukasia, as inferred from unrav-eling the duplex structure, is northward with time.The Isua belt would therefore be the youngest inthe NE territory of Akulleq terrane in West Green-

548 T . K O M I Y A E T A L .

Figure 22. A speculative cartoon showing 3.8-Ga water-planet Earth and its satellite Moon. Plate tectonics hadalready started; however, there were no solid inner core, double-layered mantle convection, or large continents. Instead,a number of intraoceanic island arcs of ca. 1000 oceanic microplates, abundant off-ridge volcanism related to mantleupwellings, and plume- or bombardment-triggered spreading axis all characterized early Earth at 3.8 Ga. One platesatellite, Moon, was active only by heavy extraterrestrial bombardments and related Mare volcanism.

land (fig. 1). This enables us to predict the site ofthe oldest accretionary prism and TTG in this area.Figure 21 shows the N-S cross section of the Isua-Amıtsoq composite terrane in comparison with thezoned growth history of southwestern Japan arc. Inthe southwestern Japan arc, the younger accretion-ary complexes, the major parts of the Sanbagawaand Shimanto belts, have not been intruded byTTG yet, whereas the Paleozoic to Jurassic accre-tionary prisms have been extensively intruded bythem through the zoned growth of the orogen andoceanward shifting of the trench with time. Simi-larly, the older accretionary complexes should bepenetrated by TTG, remaining as enclaves or roofpendants within TTG plutons, and should be to thesouth of the Isukasia. The Akilia association de-

fined by McGregor (1973) is the expected materialfor the oldest accreted oceanic materials to thenorth of the Ivisartoq region mapped by Chadwickand Coe (1988) and Nutman (1986). Recently Nut-man et al. (1993) reported 3811 to 3812 Ma SHRIMPages on zircons separated from the Amıtsoq gneissca. 30 km south of the mapped area. The subduc-tion-accretion time of Akilia association must beearlier than 3812 Ma, and its formation at a mid-oceanic ridge or thereafter by off-ridge volcanismmust be much older.

The Isua supracrustal belt records the evidenceof 3.8–3.7-Ga tectonics similar to modern plate tec-tonics. It also records the materials formed by theplate boundary process at that time. These havesimilarities and differences with materials formed


by modern plate boundary processes. Differencesare (1) more frequent intraoceanic or off-ridge vol-canism, as demonstrated by the frequent intrusionswithin chert beds. This may reflect the higher-temperature Archean; hence more turbulent man-tle upwellings occurred in the Archean, as pre-dicted by many workers. (2) More frequent accre-tion of oceanic basement than in the Phanerozoic.This may reflect thicker oceanic crust than the pre-sent. If the crust were thicker than 10 km, the me-chanically weak plane easily developed withincrust during subduction of oceanic plate, accordingto recent rheological experiments (e.g., Shimamotoet al. 1993); therefore, accretion of oceanic mate-rials is not common during Phanerozoic subductionexcept during deep accretion. Although sections ofoceanic crust 110 km thick have not yet been re-ported in accretionary complexes, such crustalthickness must have been common in the Archean(Hoffmann and Ranalli 1988; Kusky 1993). (3) Theabsence of blueschist in the Archean accretionarycomplex. A high geothermal gradient at the sub-duction zone led to Barrovian-type metamorphisminstead of the high P/T-type metamorphism suchas that of the Sanbagawa belt (Grambling 1981; Ma-ruyama 1997a). This is consistent with progressivemetamorphism reflecting the subduction processin the Isua supracrustal belt.

In spite of those differences, many similaritiesexist between the Archean and the present. Al-though the chert was possibly derived from chem-ical precipitation at the mid-ocean ridge, the oc-currence of chert beds is quite similar toPhanerozoic bedded chert with regard to the ob-vious bedding planes and the lithological gradationfrom bottom to top. Mid-oceanic ridge volcanicsare not komatiite in composition but are low-Ktholeiite (Gill and Bridgwater 1979; Komiya andMaruyama 1995). Moreover, systematic composi-tional differences can be observed between the pil-low lavas underlying the thick chert and basalticdikes-sills intruding the chert layers (Komiya andMaruyama 1995). Pillow lavas are enriched inAl2O3, FeO, and Y relative to basic dikes-sills at agiven MgO content. The pattern of chondrite-normalized REE (rare earth elements) is LREE (lightrare earth elements) depleted for pillow lavas, flator slightly LREE enriched for the dikes-sills. Thisresult is quite different from the commonly heldspeculation that Archean mid-oceanic ridge vol-canics were komatiite (e.g., Takahashi 1990).

Evidence of accretion of oceanic materials at sub-duction zones suggests that a kind of tectonics sim-ilar to the Phanerozoic plate tectonics operated inthe Early Archean but does not necessarily prove

that modern-style plate tectonics worked, in astrict sense. Evidence for modern-style plate tec-tonics includes (1) rigid plates, (2) significant lateralmovement of the plate, (3) multiple plates sur-rounded by plate boundaries, and (4) linearity ofplate boundaries. However, 1 and 2 are necessaryand sufficient to identify the modern-style platetectonics in practice because horizontal passivemovement of rigid plate results in multiple plates(3) and in linearity of plate boundaries (4) (Olden-burg and Brune 1972; Turcotte and Schubert 1982).

A comparison with tectonics on Venus providessome perspective because Archean tectonics onEarth has been widely regarded as similar to mod-ern Venusian tectonics. Proposals have been madethat Earth-like crustal spreading and plate recyclingdo not occur or are not significant on Venus, on thebasis of topographic data by the Magellan mission,landers, and Earth-based radar telescope (e.g., Kaulaand Phillips 1981; Phillips et al. 1981, 1991; Solo-mon et al. 1991; Ford and Pettengill 1992), in spiteof much dispute (Schaber 1982; Crumpler et al.1987; Head and Crumpler 1987; McKenzie et al.1992; Sandwell and Schubert 1992). Volcanic fea-tures on Venus are broadly distributed globally andnot concentrated along linear boundaries (Head etal. 1992). The fact also supports the idea that Venusis governed by plume tectonics rather than platetectonics. Venus is comparable to the Earth in sizeand density but much higher in surface tempera-tures and pressures (4707C and about 98 bars). Thehigh temperature of Venusian surface prevents theplate from being hardened. Assuming that Venu-sian crust is basaltic in composition and 20 kmthick (Surkov et al. 1984; Buck 1992), and that thetemperature of the mantle is 1207C higher thanmodern Earth’s, strength profiles in compressionfor oceanic lithosphere were calculated as a func-tion of age (fig. 20d). The result shows the two dif-ferent modes of rheological structure depending onage. For the lithosphere !30 m.yr., only 5-km top-most of the crust is rigid, while the rest remainsductile (fig. 20d, left). For the lithosphere 130 m.yr.,the topmost rigid basaltic layer grows only a fewkilometers thick with age, while the major part ofthe crust remains ductile. But the uppermost man-tle, ca. 8 km thick at 50 m.yr., becomes rigid be-neath the crust. Formation of a thick ductile zonebetween the upper crust and mantle may lead todecoupling of mantle and crust (Buck 1992; fig. 20d,right). This does not support rigid plates on Venus.

Even though subduction at Aphrodite and IshtarTerra based on topographic image was proposed bysome authors (e.g., Crumpler et al. 1986; McKenzieet al. 1992), insignificant lateral movement is sug-

550 T . K O M I Y A E T A L .

gested from the absence of volcanic chains such asthe Hawaii-Emperor seamount chain (Head et al.1992). In contrast, in even the Early Archean, theoceanic lithosphere on Earth was hardened. Suc-cessive underthrusting of oceanic crust at subduc-tion zone means predominantly lateral movementof the Earth’s surface rather than vertical tectonics.Oceanic plate stratigraphy preserved in the Isua ac-cretionary complex displays a gradual change ofdepositional environment from open sea to trench,also proving significant horizontal movement of anoceanic plate. Since the Middle Archean, wander-ing of the palaeomagnetic pole of terranes has in-dicated a global scale of plate motion at a velocityat least as fast as in Phanerozoic (Kroner and Layer1992). In summary, there is evidence for both rigidplate and significant lateral movement of the platein the Early Archean accretionary complex, dem-onstrating that modern-style plate tectonics has op-erated since the Early Archean, quite different fromthe Earth’s satellite Moon, which was resurfacedby heavy extraterrestrial bombardment and the re-lated Mare volcanism (fig. 22).

Conclusions

Our detailed mapping of the northeastern part ofthe Isua supracrustal belt shows that the belt con-sists of duplex structures. The sketches of eachhorse allow us to reconstruct the original litho-stratigraphy, which is quite similar to Phanerozoic“oceanic plate stratigraphy,” and consists, in as-cending order, of pillow-massive lava units, beddedcherts, and turbidites.

The discovery of the duplex structures and oce-anic plate stratigraphy reveals that the Isua supra-crustal belt is the oldest known Archean accre-tionary complex. This indicates successiveunderthrusting of oceanic crust and significant hor-izontal motion of an oceanic plate.

Geological evidence shows that three distinctivemafic igneous suites are present in the belt. Thefirst is pillowed and massive lava flows with sub-ordinate amounts of hyalocalstite layers (mainlyGarbenschiefer units), in turn overlain by thickchert layers with no terrigenous materials, sug-gesting that the mafic rocks had been derived fromoceanic ridge volcanism. The second includes theabundant doleritic dikes and sills (some homoge-neous amphibolites), which nowhere intrude into

the turbidite sequences but into pillowed flowunits and the overlying chert layers, suggesting off-ridge volcanism origin. The third type is the abun-dant doleritic dikes (some homogeneous amphib-olites), which also intrude all of those unitsincluding the turbidite sequences. This may be re-lated to postaccretionary island arc volcanism orcontinental rifting volcanism thereafter.

The duplex polarity indicates that the Isua ac-cretionary prism grew from the south to the northand was followed by successive tonalitic intru-sions. This idea is consistent with recent SHRIMPdating (Nutman et al. 1993), suggesting that rem-nants of older accretionary complexes might be pre-served as enclaves and roof-pendants within Amıt-soq gneiss to the south.

The oldest pillow lavas in the Isua supracrustalbelt evidently show that seawater was present inthe Early Archean. Especially, a pillow lava unitcovered with thick-bedded chert at the bottom ofoceanic plate stratigraphy suggests that the opensea extended far way from an oceanic island or con-tinents. It proves that the surface temperature wasless than ca. 1007C in the Early Archean, in contrastto over 4707C on present-day Venus. The Early Ar-chean oceanic geotherm was calculated based on amodel of transient half-space cooling at given pa-rameters of temperatures of Earth’s mantle and sur-face. The rheological structure of oceanic litho-sphere as a function of age is estimated from theoceanic geotherm and the transient temperature ofductile/brittle of peridotite and diabase, indicatingthat oceanic lithosphere was rigid even in the EarlyArchean. The evidence for lateral movement ofrigid plates suggests that modern-style plate tec-tonics has already operated in the Early Archean.

A C K N O W L E D G M E N T S

We appreciate the fruitful discussions of Isukasiageology with V. R. McGregor and C. R. L. Friend.C. Parkinson improved the English language in thisarticle. Y. Isozaki and C. Parkinson are also to bethanked for discussions of accretionary geology inthe Phanerozoic. Sincere thanks are also due to K.Burke, A. P. Nutman, and T. M. Kusky, all of whomconstructively reviewed the manuscript. This workwas partly supported by grants from Cubas Com-pany and from the Ministry of Education, Science,and Culture, Japan (04302021).


R E F E R E N C E S C I T E D

Allaart, J. H. 1976. The pre-3760 m.y. old supracrustalrocks of the Isua area, central West Greenland, andthe associated occurrence of quartz-banded ironstone.In Windley, B. F., ed. The early history of the Earth.London, Wiley, p. 177–189.

Alt, J. C. 1995. Subseafloor processes in mid-ocean ridgehydrothermal systems. In Humphris, S. E.; Zieren-berg, R. A.; Mullineaux, L. S.; and Thompson, R. E.,eds. Physical, chemical, biological, and geological in-teractions within seafloor hydrothermal systems.Washington, D.C., Am. Geophys. Union, p. 85–114.

Appel, P. W. U. 1984. Tourmaline in the Early ArchaeanIsua supracrustal belt, West Greenland. J. Geol. 92:599–605.

———. 1995. Tourmalinites in the 3800 Ma old Isua su-pracrustal belt, West Greenland. Precambrian Res. 72:227–234.

Arndt, N. T. 1983. Role of a thin, komatiite-rich oceaniccrust in the Archean plate-tectonics process. Geology11:372–375.

Baadsgaard, H. 1976. Further U-Pb dates on zircons fromthe early Precambrian rocks of the Godthaabsfjordarea, West Greenland. Earth Planet. Sci. Lett. 33:261–267.

Bickle, M. J. 1978. Heat loss from the Earth: a constrainton Archaean tectonics from the relation between geo-thermal gradients and the rate of plate production.Earth Planet. Sci. Lett. 40:301–315.

Bickle, M. J.; Nisbet, E. G.; and Martin, A. 1994. Archeangreenstone belts are not oceanic crust. J. Geol. 102:121–138.

Boak, J. L., and Dymek, R. F. 1982. Metamorphism of theca. 3800 Ma supracrustal rocks at Isua, West Green-land: implications from early Archean crustal evolu-tion. Earth Planet. Sci. Lett. 59:155–176.

Boak, J. L.; Dymek, R. F.; and Gromet, L. P. 1983. Pe-trology and rare earth element geochemistry of clasticmetasedimentary rocks from the Isua supracrustalbelt, West Greenland. Rapp. Gronlands Geol. Unders.112:23–33.

Borradaile, G. J. 1982. Tectonically deformed pillow lavaas an indicator of bedding and way-up. J. Struct. Geol.4:469–479.

Bridgwater, D.; McGregor, V. R.; and Myers, J. S. 1974.A horizontal tectonic regime in the Archean of Green-land and its implications for early crustal thickening.Precambrian Res. 1:179–197.

Brown, M.; Friend, C. R. L.; McGregor, V. R.; and Perkins,W. T. 1981. The late Archean Qorqut granite complexof southern West Greenland. J. Geophys. Res. 86:10,617–10,632.

Buck, W. R. 1992. Global decoupling of crust and mantle:implications for topography, geoid and mantle viscos-ity of Venus. Geophys. Res. Lett. 19:2111–2114.

Burke, K.; Dewey, J. F.; and Kidd, W. S. F. 1976. Domi-nance of horizontal movements, arc and microconti-nental collisions during the later permobile regime.In Windley, B. F., ed. The early history of the Earth.London, Wiley, p. 113–129.

Calvert, A. J.; Sawyer, E. W.; Davis, W. J.; and Ludden, J.N. 1995. Archaean subduction inferred from seismicimages of a mantle suture in the Superior Province.Nature 375:670–674.

Chadwick, B., and Coe, K. 1988. 1 : 100000 Ivısartoq mapsheet (64 V.2 N). Copenhagen, Geological Survey ofGreenland.

Compston, W.; Kinny, P. D.; Williams, I. S.; and Foster,J. J. 1986. The age and Pb loss behaviour of zirconsfrom the Isua supracrustal belt as determined by ionmicroprobe. Earth Planet. Sci. Lett. 80:71–81.

Condie, K. C. 1972. A plate tectonics evolutionary modelof the South Pass Archaean greenstone belt, south-western Wyoming. Proc. 24th Int. Geol. Congr. 1:104–112.

———. 1982. Plate tectonics and crustal evolution. NewYork, Pergamon, 310 p.

Coney, P. J.; Jones, D. L.; and Monger, J. W. H. 1980.Cordilleran suspect terranes. Nature 288:329–333.

Cowan, D. S. 1985. Structural styles in Mesozoic andCenozoic melanges in the western Cordillera of NorthAmerica. Geol. Soc. Am. Bull. 96:451–462.

Crumpler, L. S.; Head, J. W.; and Campbell, D. B. 1986.Orogenic belts on Venus. Geology 14:1031–1034.

Crumpler, L. S.; Head, J. W.; and Harmon, J. K. 1987.Regional linear cross-strike discontinuities in westernAphrodite Terra, Venus. Geophys. Res. Lett. 14:607–610.

DeLong, S. E.; Fox, P. J.; and McDowell, F. W. 1978. Sub-duction of the Kula ridge at the Aleutian trench. Geol.Soc. Am. Bull. 89:83–95.

Dewey, J., and Spall, H. 1975. Pre-Mesozoic plate tec-tonics: how far back in Earth history can the WilsonCycle be extended? Geology 3:422–424.

de Wit, M. J.; Armstrong, R.; Hart, R. J.; and Wilson, A.H. 1987a. Felsic igneous rocks within the 3.3- to 3.5-Ga Barberton Greenstone Belt: high crustal levelequivalents of the surrounding tonalite-trondhjemiteterrain, emplaced during thrusting. Tectonics 6:529–549.

de Wit, M. J.; Hart, R. A.; and Hart, R. J. 1987b. TheJamestown ophiolite complex, Barberton mountainbelt: a section through 3.5 Ga oceanic crust. J. Afr.Earth Sci. 6:681–730.

de Wit, M. J.; Rosing, C.; Hart, R. J.; Armstrong, R. A.;de Ronde, C. E. J.; Green, R. W. E.; Tredoux, M.; Pe-berdy, E.; and Hart, R. A. 1992. Formation of an Ar-chaean continent. Nature 357:553–562.

Dymek, R. F., and Klein, C. 1988. Chemistry, petrology

552 T . K O M I Y A E T A L .

and origin of banded iron-formation lithologies fromthe 3800 Ma Isua supracrustal belt, West Greenland.Precambrian Res. 39:247–302.

Folinsbee, R. E.; Baadsgaard, H.; Cumming, G. L.; andGreen, D. C. 1968. A very ancient island arc. Geophys.Monogr. 12:441–448.

Ford, P. G., and Pettengill, G. H. 1992. Venus topographyand kilometer-scale slopes. J. Geophys. Res. 97:13103–13114.

Friend, C. R. L.; Nutman, A. P.; Baadsgaard, H.; Kinny,P. D.; and McGregor, V. R. 1996. Timing of late Ar-chaean terrane assembly, crustal thickening and gran-ite emplacement in the Nuuk region, southern WestGreenland. Earth Planet. Sci. Lett. 142:353–365.

Friend, C. R. L.; Nutman, A. P.; and McGregor, V. R. 1987.Late Archaean tectonics in the Færingehavn–TreBrødre area, south of Buksefjorden, southern WestGreenland. J. Geol. Soc. Lond. 144:369–376.

———. 1988. Late Archaean terrane accretion in the God-thab region, southern West Greenland. Nature 335:535–538.

Gill, R. C. O., and Bridgwater, D. 1979. Early Archaeanbasic magmatism in West Greenland: the geochem-istry of the Ameralik dykes. J. Petrol. 20:695–726.

Grambling, J. A. 1981. Pressures and temperatures in Pre-cambrian metamorphic rocks. Earth Planet. Sci. Lett.53:63–68.

Hamilton, P. J.; O’Nions, R. K.; Bridgwater, D.; and Nut-man, A. 1983. Sm-Nd studies of Archaean metasedi-ments and metavolcanics from West Greenland andtheir implications for the Earth’s early history. EarthPlanet. Sci. Lett. 62:263–272.

Hart, S. R.; Brooks, C.; Krogh, T. E.; Davis, G. L.; andNava, D. 1970. Ancient and modern volcanic rocks:a trace element model. Earth Planet. Sci. Lett. 10:17–28.

Head, J. W., and Crumpler, L. S. 1987. Evidence for di-vergent plate-boundary characteristics and crustalspreading on Venus. Science 238:1380–1385.

———. 1990. Venus geology and tectonics: hotspot andcrustal spreading models and questions for the Ma-gellan mission. Nature 346:525–533.

Head, J. W.; Crumpler, L. S.; Aubele, J. A.; Guest, J. E.;and Saunders, R. S. 1992. Venus volcanism: classifi-cation of volcanic features and structures, associa-tions, and global distribution from Magellan data. J.Geophys. Res. 97:13,153–13,197.

Hoffman, P. F. 1973. Evolution of an early Proterozoiccontinental margin: the Coronation geosyncline andassociated aulacogens of the northwestern Canadianshield. Philos. Trans. R Soc. Lond. A273:547–581.

———. 1991. On accretion of granite-greenstone ter-ranes. In Robert, F.; Sheahan, P. A.; and Green, S. B.,eds. Greenstone gold and crustal evolution. St. John’s,Geological Association of Canada, p. 32–45.

Hoffman, P. F., and Ranalli, G. 1988. Archean oceanicflake tectonics. Geophys. Res. Lett. 15:1077–1080.

Ishizuka, H. 1985. Prograde metamorphism of the Ho-rokanai ophiolite in the Kamuikotan Zone, Hokkaido,Japan. J. Petrol. 26:391–417.

Isozaki, Y.; Maruyama, S.; and Furuoka, F. 1990. Accretedoceanic materials in Japan. Tectonophysics 181:179–205.

Isozaki, Y.; Maruyama, S.; and Kimura, G. 1991. MiddleArchean (3.3 Ga) Cleaverville accretionary complexin northwestern Pilbara block, western Australia. EOSTrans. 72:542.

Iwamori, H.; McKenzie, D.; and Takahashi, E. 1995. Meltgeneration by isentropic mantle upwelling. EarthPlanet Sci. Lett. 134:253–266.

Jacobsen, S. B., and Dymek, R. F. 1988. Nd and Sr isotopesystematics of clastic metasediments from Isua, WestGreenland: identification of pre-3.8 Ga differentiationcrustal components. J. Geophys. Res. 93:338–354.

James, P. R. 1976. Deformation of the Isua block, WestGreenland: a remnant of the earliest stable continen-tal crust. Can. J. Earth Sci. 13:816–823.

Kaula, W. M., and Phillips, R. J. 1981. Quantitative testsfor plate tectonics on Venus. Geophys. Res. Lett. 8:1187–1190.

Kimura, G. 1994. The latest Cretaceous-early paleogenerapid growth of accretionary complex and exhumationof high pressure series metamorphic rocks in north-western Pacific margin. J. Geophys. Res. 99:22,147–22,164.

Kimura, G.; Ludden, J. N.; Desrochers, J.-P.; and Hori, R.1993. A model of ocean-crust accretion for the Su-perior province, Canada. Lithos 30:337–335.

Kimura, G.; Maruyama, S.; and Isozaki, Y. 1991. EarlyArchean accretionary complex in the eastern Pilbaracraton, Western Australia—detailed field occurrencein the Marble Bar area. EOS Trans. 72:542.

Komiya, T., and Maruyama, S. 1995. Geochemistry ofthe oldest MORB and OIB of the World, Isua (3.8 Ga),Greenland. EOS Trans. 76:700.

Kroner, A. 1977. The Precambrian geotectonic evolutionof Africa: plate accretion versus plate destruction. Pre-cambrian Res. 4:163–213.

———. 1981. Precambrian plate tectonics. In Kroner, A.,ed. Precambrian plate tectonics. Amsterdam, Elsevier,p. 57–90.

Kroner, A., and Layer, P. W. 1992. Crust formation andplate motion in the Early Archean. Science 256:1405–1411.

Kusky, T. M. 1989. Accretion of the Archean Slave prov-ince. Geology 17:63–67.

———. 1990. Evidence for Archean ocean opening andclosing in the southern Slave Province. Tectonics 9:1533–1563.

———. 1991. Structural development of an Archean or-ogen, western Point Lake, Northwest Territories. Tec-tonics 4:820–841.

———. 1993. Collapse of Archean orogens and the gen-eration of late- to postkinematic granitoids. Geology21:925–928.

Kusky, T. M.; Bradley, D. C.; and Haeussler, P. 1997a.Progressive deformation of the Chugach accretionarycomplex, Alaska, during a paleogene ridge-trench en-counter. J. Struct. Geol. 19:139–157.

Kusky, T. M.; Bradley, D. C.; Haeussler, P. ; and Karl, S.


1997b. Controls on accretion of flysch and melangebelts at convergent margins: evidence from the Chu-gach Bay thrust and Iceworm melange, Chugach ac-cretionary wedge, Alaska. Tectonics 16:855–878.

Kusky, T. M., and Kidd, W. S. F. 1992. Remnants of anArchean oceanic plateau, Belingwe greenstone belt,Zimbabwe. Geology 20:43–46.

Kusky, T. M., and Vearncombe, J. 1997. Structural as-pects. In de Wit, M. J., and Ashwal, L. D., eds. Green-stone belts. Oxford, Clarendon, p. 91–124.

Kusky, T. M., and Winsky, P. A. 1995. Structural rela-tionships along a greenstone/shallow water shelf con-tact, Belingwe greenstone belt, Zimbabwe. Tectonics14:448–471.

Lister, G. S., and Snoke, A. W. 1984. S-C Mylonites. J.Struct. Geol. 6:617–638.

Martin, H. 1986. Effect of steeper Archean geothermalgradient on geochemistry of subduction-zone mag-mas. Geology 14:753–756.

Maruyama, S. 1997a. Initiation of return flow seawaterinto mantle at 750 Ma and the Proterozoic/Phanero-zoic boundary. EOS Trans. 78:826.

———. 1997b. Pacific-type orogeny revisited: Miyashiro-type orogeny proposed. Island Arc 6:91–120.

Maruyama, S.; Isozaki, Y.; and Kimura, G. 1991a. Is theMid-Archean barite formation from the Pilbara cra-ton, Australia, under the deep-sea environment? EOSTrans. 72:532.

Maruyama, S.; Komiya, T.; and Appel, P. W. U. 1994. Theoldest (3.8 Ga) accretionary complex of the world,Isua, Greenland. EOS Trans. 75:691.

Maruyama, S.; Masuda, S.; and Appel, P. W. U. 1991b.The oldest accretionary complex on the Earth, Isua,Greenland. Geol. Soc. Am. Abs. Prog. 23:A429–A430.

Matsuda, T., and Isozaki, Y. 1991. Well-documentedtravel history of Mesozoic pelagic chert in Japan. Tec-tonics 10:475–499.

McClay, K. R. 1992. Glossary of thrust tectonics terms.In McClay, K. R., ed. Thrust tectonics. London, Chap-man & Hall, p. 419–433.

McGregor, V. R. 1973. The early Precambrian gneisses ofthe Godthab district, West Greenland. Philos. Trans.R Soc. Lond. A273:343–358.

———. 1993. Descriptive text to 1 : 100000 geologicalmap of Greenland Qorqut 64 V.1 Syd. Copenhagen,Geological Survey of Greenland, 40 p.

McGregor, V. R.; Bridgwater, D.; and Nutman, A. P. 1983.The Qarusuk dykes: post-Nuk, pre-Qorqut granitoidmagmatism in the Godthab region, southern WestGreenland. Rapp. Grønlands Geol. Unders. 112:101–112.

McGregor, V. R.; Friend, C. R. L.; and Nutman, A. P. 1991.The late Archean mobile belt through Godthabsfjord,southern West Greenland: a continent-continent col-lision zone? Bull. Geol. Soc. Denmark 39:179–197.

McKenzie, D.; Ford, P. G.; Johnson, C.; Parsons, B.; Sand-well, D.; Saunders, S.; and Solomon, S. 1992. Featureson Venus generated by plate boundary processes. J.Geophys. Res. 97:13,353–13,544.

Mojzsis, S.; Arrhenius, G.; McKeegan, K. D.; Harrison,

T. M.; Nutman, A. P.; and Friend, C. R. L. 1996. Ev-idence for life on Earth before 3,800 million years ago.Nature 385:55–59.

Moorbath, S.; Allaart, J. H.; Bridgwater, D.; and Mc-Gregor, V. R. 1977. Rb-Sr ages of early Archaean su-pracrustal rocks and Amıtsoq gneisses at Isua. Nature270:43–45.

Moorbath, S.; O’Nions, R. K.; and Pankhurst, R. J. 1973.Early Archaean age for the Isua Iron Formation, WestGreenland. Nature 245:138–139.

Moorbath, S., and Taylor, P. N. 1981. Isotope evidencefor continental growth in the Precambrian. In Kroner,A., ed. Precambrian plate tectonics. Amsterdam, El-sevier, p. 491–525.

Moorbath, S.; Taylor, P. N.; and Goodwin, R. 1981. Originof granitic magma by crustal remobilisation: Rb-Srand Pb/Pb geochronology and isotope geochemistry ofthe late Archaean Qorqut Granite Complex of south-ern West Greenland. Geochim. Cosmochim. Acta 45:1051–1060.

Nisbet, E. G., and Fowler, C. M. R. 1983. Model for Ar-chean plate tectonics. Geology 11:376–379.

Nutman, A. P. 1984. Early Archaean crustal evolution ofthe Isukasia area, southern West Greenland. InKroner, A., and Greiling, R., eds. Precambrian tecton-ics illustrated. Stuttgart, Schweizerbart’sche, p.79–93.

———. 1986. The early Archaean to Proterozoic historyof the Isukasia area, southern West Greenland.Grønlands Geol. Unders. Bull. Vol. 154, 80 p.

Nutman, A. P.; Allaart, J. H.; Bridgwater, D.; Dimroth,E.; and Rosing, M. 1984. Stratigraphic and geochem-ical evidence for the depositional environment of theearly Archaean Isua supracrustal belt, southern WestGreenland. Precambrian Res. 25:365–396.

Nutman, A. P.; Bennett, V. C.; Friend, C. R. L.; and Ros-ing, M. T. 1997. ∼3710 and ≥3790 Ma volcanic se-quences in the Isua (Greenland) supracrustal belt;structural and Nd isotope implications. Chem. Geol.141:271–287.

Nutman, A. P., and Bridgwater, D. 1986. Early ArchaeanAmıtsoq tonalites and granites of the Isukasia area,southern West Greenland: development of the oldest-known sial. Contrib. Mineral. Petrol. 94:137–148.

Nutman, A. P.; Bridgwater, D.; Dimroth, E.; Gill, R. C.O.; and Rosing, M. 1983. Early (3700 Ma) Archaeanrocks of the Isua supracrustal belt and adjacentgneisses. Rapp. Grønlands Geol. Unders. 112:5–22.

Nutman, A. P., and Collerson, K. D. 1991. Very earlyArchean crustal-accretion complexes preserved in theNorth Atlantic craton. Geology 19:791–794.

Nutman, A. P.; Friend, C. R. L.; Baadsgaard, H.; andMcGregor, V. R. 1989. Evolution and assembly of Ar-chean gneiss terranes in the Godthabsfjord region,southern West Greenland: structural, metamorphic,and isotope evidence. Tectonics 8:573–589.

Nutman, A. P.; Friend, C. R. L.; Kinny, P. D.; andMcGregor, V. R. 1993. Anatomy of an Early Archeangneiss complex: 3900 to 3600 Ma crustal evolution insouthern West Greenland. Geology 21:415–418.

554 T . K O M I Y A E T A L .

Nutman, A. P.; Hagiya, H.; and Maruyama, S. 1995.SHRIMP U-Pb single zircon geochronology of a Prot-erozoic mafic dyke, Isukasia, southern West Green-land. Bull. Geol. Soc. Denmark 42:17–22.

Nutman, A. P.; McGregor, V. R.; Friend, C. R. L.; Bennett,V. C.; Kinny, P. D. 1996. The Itsaq Gneiss Complexof southern West Greenland: the world’s most exten-sive record of early crustal evolution (3,900–3,600Ma). Precambrian Res. Spec. Issue 78:1–39.

Ohta, H.; Maruyama, S.; Takahashi, E.; Watanabe, Y.; andKato, Y. 1996. Field occurrence, geochemistry and pet-rogenesis of the Archean mid-oceanic ridge basalts(AMORBs) of the Cleaverville area, Pilbara craton,Western Australia. Lithos 37:199–221.

Oldenburg, D. W., and Brune, J. N. 1972. Ridge transformfault spreading pattern in freezing wax. Science 178:301–304.

Phillips, R. J.; Grimm, R. E.; and Malin, M. C. 1991. Hot-spot evolution and the global tectonics of Venus. Sci-ence 252:651–658.

Phillips, R. J.; Kaula, W. M.; McGill, G. E.; and Malin,M. C. 1981. Tectonics and evolution of Venus. Science212:879–887.

Piper, J. D. A. 1982. The Precambrian palaeomagneticrecord: the case for the Proterozoic supercontinent.Earth Planet. Sci. Lett. 59:61–89.

Rosing, M. T.; Rose, N. M.; Bridgwater, D.; and Thomp-son, H. S. 1996. Earliest part of Earth’s stratigraphicrecord: a reappraisal of the 13.7 Ga Isua (Greenland)supracrustal sequence. Geology 24:43–46.

Sandwell, D. T., and Schubert, G. 1992. Flexural ridges,trenches, and outer rises around Coronae on Venus. J.Geophys. Res. 97:16,069–16,084.

Schaber, G. G. 1982. Venus: limited extension and vol-canism along zones of lithospheric weakness. Geo-phys. Res. Lett. 9:499–502.

Schidlowski, M. 1988. A 3,800-million-year isotopic rec-ord of life from carbon in sedimentary rocks. Nature333:313–318.

Shimamoto, T.; Seno, T.; and Uyeda, S. 1993. A simplerheological model for comparative subductology. Geo-phys. Monogr. (Jeffreyes vol.) 76:39–52.

Shimizu, H.; Nakai, S.; Tasaki, S.; Masuda, A.; Bridg-water, D.; Nutman, A. P.; and Baadsgaard, H. 1988.Geochemistry of Ce and Nd isotope and REE abun-dances in the Amıtsoq gneisses, West Greenland.Earth Planet. Sci. Lett. 91:159–169.

Solomon, S. C.; Head, J. W.; Kaula, W. M.; McKenzie, D.;Parsons, B.; Phillips, R. J.; Schubert, G.; and Talwani,M. 1991. Venus tectonics, initial analysis from Ma-gellan. Science 252:297–312.

Surkov, Y. A.; Barsukov, V. L.; Moskalyneva, L. P.; Khar-yukova, V. P.; and Kemurdzhian, A. L. 1984. New dataon the composition, structure, and properties of Venusrock obtained by Venera 13 and Venera 14. J. Geophys.Res. 89:B393–B402.

Tajika, E., and Matsui, T. 1990. The evolution of theterrestrial environment. In Newsom, H. E., and Jones,J. H., eds. Origin of the Earth. New York, Oxford Uni-versity Press, p. 347–370.

Takahashi, E. 1990. Speculations on the Archean mantle:missing link between komatiite and depleted garnetperidotite. J. Geophys. Res. 95:15,941–15,954.

Takahashi, M. 1986. “Island arc” magmatism before andafter the spreading of Japan sea. Kagaku 56:103–111(in Japanese).

Turcotte, D. L., and Schubert, G. 1982. Geodynamics:applications of continuum physics to geological prob-lems. New York, Wiley, 450 p.

von Huene, R.; Corvalan, J.; Flueh, E. R.; Hinz, K.; Korst-gard, J.; Ranero, C. R.; Weinrebe, W.; and the CON-DOR Scientists. 1997. Tectonic control of the sub-ducting Juan Fernandez Ridge on the Andean marginnear Valparaiso, Chile. Tectonics 16:474–488.

Wynne-Edwards, H. R. 1976. Proterozoic ensialic oro-genesis: the millipede model of ductile plate tectonics.Am. J. Sci. 276:927–953.

plate tectonics at 3.8–3.7 ga: field evidence from the isua accretionary complex, southern west...

Documents