Geochemical Evolution and Tectonic Significance of Boninites and Tholeiites from the Koh Ophiolite, New Caledonia

TECTONICS, VOL. 15, NO. 1, PAGES, 67-83, FEBRUARY 1996

/ / Geochemical evolution and tectonic significance of boninites and tholeiites from the Koh ophiolite, New Caledonia

S•bastien Meffre Departmentof Geologyand Geophysics, University of Sydney,New SouthWales, Australia

Jonathan C. Aitchison Departmentof Earth Sciences,University of Hong Kong,Hong Kong

Anthony J. Crawford Departmentof Geology,University of Tasmania,Hobart, Tasmania, Australia

Abstract. The Central Chain ophiolites in New Caledonia event was driven by adiabatic decompressionof hot depleted are fragmentsof a supra-subductionzone (SSZ) ophiolite,now mantle residual from the production of the lower tholeiites, preservedfrom the upper layeredgabbros through to volcanics during initiation of rifting of young oceaniccrust intimately and overlain by pelagic cherts and a thick Middle Triassic to associatedwith propagation of a back arc basin spreading Upper Jurassicvolcaniclastic sequence. Most of the fragments centre. The occurrence of a thick blanket of calc-alkaline were formed by a single tholeiitic magmaticepisode, but one volcaniclastic sediments above the ophiolite indicates of these, the Koh ophiolite, was formed by two tholeiitic proximity to a maturearc and suggeststhat the Koh boninites magmatic episodesseparated by boninites.The first event in were not associated with the initiation of subduction. A close the Koh ophiolite formed cumulate gabbros, dolerites, modern analogy for the Koh ophiolite exists on the Hunter plagiogranites, and the first pillow lava sequence from a Ridge protoislandarc between southernmostVanuatu (New tholeiitic magma with strongdepletion in the light rare earth Hebrides island arc) and the Fijian islands; there, high-Ca elements(LREE) and abnormallylow TiO2 (0.5% at Mg#=60). boninites lacking positive Zr spikes occur together with low- Shortly after their eruption, these tholeiitic lavas were Ti tholeiites and more typical BABB tholeiites where the overlain by a high-Ca boninitic unit with a basal section of southern spreading centre of the North Fiji Basin is boninite pillows, flows, and brecciasand an upper sectionof propagatinginto the protoarc crust of the Hunter Ridge. boninitic dacites and tuffs. The last magmatic phaseinvolved eruption of evolved tholeiitic basalts, as pillows above the boninites and as dykes and sills intruding the older plutonic Introduction and volcanic sectionsof the ophiolite. This secondphase of Many ophiolites are believed to have formed as oceanic- tholeiitic magmatism is compositionally distinct from the type crust within western Pacific-type arc-back arc basin first and is closest to back arc basin basalts (BABB) erupted systemsabove subductionzones. Those with arc geochemical during the early rifting historyof modernback arc basins.The signatures have been termed supra-subductionzone (SSZ) boninitic volcanics belong to a high-Ca series with slightly ophiolites [Pearce et al., 1984] and often contain a lower SiO2, A1203, and TiO2 comparedto thosefrom modern geochemical stratigraphy which includes boninitic series island arc systems,and they lack the positiveZr spikerelative volcanics (BSV) as well as island arc tholeiites (IAT) and lavas to adjacentrare earth elements(REE) in normalisedelement transitional between IAT and mid-ocean ridge basalts variation patterns.These boniniteswere formed shortly after (MORB). Although some ophiolites may well representback the production of back arc basin crust representedby the arc basin crust, the forearc crust and upper mantle of modern depleted tholeiites and shortly before a second spreading event which caused 40-60% extension of the initial basin intraoceanic island arc systems may provide the best analoguesfor many ophiolites[Casey and Dewey, 1984; Stern crust and eruption of the upper tholeiites.The dominanceof and Bloomer, 1992; Taylor et al., 1992]. BABB-like tholeiites throughoutthe Central Chain ophiolites Recent studies of western Pacific island arc systems [e.g., in New Caledonia, the restricted occurrence of boninites, and Pearce et al., 1992; Taylor, 1992; Hawkins, 1994; Pearce et the stratigraphyand chemistry of the Koh ophiolite suggest al., 1994] have provided much new information on the that the boninites were erupted in responseto an exceptional characteristicsisland arc systems and on the way in which tectonic situation. We suggest that this boninite generation they evolve. However, there are still many problems associated with interpreting the tectonic setting of Copyright1996 by the AmericanGeophysical Union. ophiolites. One such problem centres on the petrogenetic scenario and tectonic events responsible for the common Paper number95TC02316. ophiolitic associationof refractory boninitic series volcanics 0278-7407/96/95TC-02316510.00 with relatively fertile tholeiitic basalts of broadly MORB or

67 68 MEFFRE ET AL.: GEOCHEMICAL EVOLLrHON OF THE KOH OPHIOLITE back arc basin basalt (BABB) affinity [Cameron et al., 1979; Pre-Cretaceousrocks can be assignedto one of three arc- Crawford and Keays, 1987; Cameron, 1989; Coish, 1989]. related terranes. The Central Chain terrane, the focus of this Another example of this boninite-tholeiite association paper, covers 15% of the island and outcropsin the mountains occurs in the Koh ophiolite in New Caledonia. This paper and on the east coast (Figure 1) [Gudrangdet al., 1975; Paris, gives new details of SSZ ophiolites from New Caledonia, 1981]. It includes a late Paleozoic to Late Jurassicbasin with a compares them with volcanic suites from modern western basal ophiolite in which boninitic and tholeiitic volcanicsare Pacific arc systems to constrain their tectonic environment, overlain by a thick volcaniclasticsedimentary sequence. This and discussesaspects of the sequenceand nature of tectono- terrane was probably formed in an intraoceanic island arc- magmatic events that produced this late Paleozoic or Early related systemaway from the Gondwanamargin. Triassic ophiolite. The T6remba terrane outcropson the mid-westerncoast and contains Late Permian calc-alkaline arc volcanics overlain by Geological Setting a sequenceof shallow water pyroclasticsand volcaniclastics The geology of New Caledoniarecords three major plate of Triassic to Jurassic age [Campbell, 1984]. The tectonic stages:island-arc convergent margin tectonicsfrom relationshipsof this arc-related terrane to the Central Chain the Late Carboniferousor Early Permian to latest Jurassic, terrane are uncertain;as the sequencesare fault bounded,they culminatingin a major accretionand obductionevent in the were deposited in different environmentsand differ in their earliest Cretaceous; mid-Cretaceousto early Eocene passive stratigraphy, tectonic evolution, and faunal content. The margin extensionaltectonics; and mid-Eoceneto Oligocene Boghen terrane is an undated, regionally metamorphosed convergentmargin tectonics[Cluzel et al., 1994; Aitchison et sequenceof deformed volcanics and sedimentaryrocks in the al., 1995]. central part of the island. Overlap sedimentsindicate that all

20øS I

New•:•'- Caledonia Australia

Cantaloupai

Zealand 21oS New.0

50 16o 170 18 ITarøuimb%1I 1 (peddotiteultramaficterran( nappe) Koh ------Pouebo terrane • (eclogite)

...... '....:•(schist)Diahot terrane Koua

I isedimentaryCretaceous to rocksEocene Pocquereux (basaltPoyaterrane nappe) 22oS Nassi

.':.,'.•ß (volCentral ca nic Chainla stics terrane)

• (ophiolites)CentralChain terrane

•..."•T6rernba(arc volcanicsterrane & volcaniclastics) 0 km 50

I I •] (schist)Boghen terrane

164 øE 165 øE 166øE 167øE

Figure 1. Geologicalmap of New Caledonia.Terrane names are from Cluzelet al. [1994] andAitchison et al. [ 1995]. Geologyis modifiedfrom Paris [1981]. MEFFRE ET AL.: GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE 69

three terranes had amalgamated and accreted to the east by other magmatic episodes as they always occur Gondwana margin by the end of the Early Cretaceous stratigraphicallybelow the boninitic extrusive rocks and are [Aitchison et al., 1995]. intrudedby dykesof the secondtholeiitic magmatic episode. Shortly after the accretion and obduction of the pre- Tholeiitic lavas of this unit comprisethe lowest volcanic Cretaceousterranes, subsidence began in New Caledonia as a unit within the ophiolite(Figure 3). Approximately200 m of result of the breakup of the easternAustralian margin and the pillows and flows with minor pillow brecciasare intercalated opening of the Tasman Sea around 80 Ma [Aitchison et al., with subordinatered cherts.The pillows are mostly small, 1995]. A thick, deepening-upwardsequence developed on the spherical,and separatedby thin bands (1 cm) of green New Caledonian Platform, from Cretaceous sandstones hyaloclastite.Some pillows show radially arranged, elongated interbedded with basic and felsic volcanics to Paleocene and amygdalesoriented toward the pillow rims. The lowermost lower Eocene pelagic cherts and limestones. pillows are intruded by narrow dykes of dolerite and Passivesubsidence of the platform was disturbedin the mid- plagiogranite.The transition zone between the extrusivesand Eocene by collision with an island arc [Aitchison et al., the underlying plutonic rocks is less than 250 m thick and 1995], resulting in the emergenceof New Caledonia and the containsmany small dykes, as well as largerintrusions (up to obduction of a thin nappe of basalts and dolerites, overthrust 50 m in width) of gabbro,dolerite, and plagiogranitewith in turn by a mantle sequence,presently 1500 m thick, of complex relationships. Below these shallow intrusives, a depleted harzburgite above a serpentinised horizontal narrowtransition occurs into alteredlayered cumulate gabbros detachment[Prinzhofer et al., 1981; Nicolas, 1989]. The most and gabbronorites,chemically related to the lower tholeiites. recent event in the geological history of New Caledonia was The lower tholeiites and their intrusive equivalentsare the postorogeniccollapse of the island and the unroofing of extensively altered by greenschist facies subseafloor an eclogite-faciesmetamorphic core complex in the north of alteration,although original texturesare generallywell- the island [Cluzel•t al., 1994; Aitchison et al., 1995; Cluzel preserved.Most extrusiverocks contain rare phenocrystsof et al., 1995]. augitein a groundmassof alteredplagioclase laths and altered clinopyroxene microphenocrysts. The dolerites have interlocking altered crystals of clinopyroxene and Central Chain Ophiolites: Geology and plagioclase,and the cumulategabbros have very altered, Stratigraphy aligned plagioclase with large clinopyroxene, rare The Central Chain ophiolites are the basement of a orthopyroxene,and interstitial titanomagnetitecrystals. sedimentary sequence (>7 km thick) dominated by Occasionalfresh augite is preserved,but in mostsamples, volcaniclasticsandstones and siltstones.Four well-preserved pyroxenes are pseudomorphedby actinolite and/or chlorite. sections and many smaller fragments of the Central Chain Plagioclaseis albitised or replacedby fine-grainedclay ophiolites are scatteredover 180 km along the length of the mineralsand titanomagnetiteby titanite. New Caledonia [Paris, 1981; Maurizot et al., 1985; Cameron, 1989] (Figure 1). The most completeof these sectionsis in Boninites Pillows and Felsic Volcanics the vicinity of the tribal village of Koh, where the ophiolite Boninitic volcanics stratigraphically overlie the lower is 20 km long and 5 km thick. UndatedBoghen terrane schists tholeiites.In somecases, the boundarybetween the two units lie to the west acrossa steeply dipping fault, and to the east is markedby a zoneup to 2 m thickof pelagicradiolarian and the Koh ophiolite is conformably overlain by hydrothermal cherts, indicating a hiatus in the volcanic volcaniclastics.The peridotitesand cumulateultramafic rocks activity.Evidence for this is not alwayspresent and in some which are present at the base of many ophiolites are missing placesthe boundaryis markedby primitivelow-Ti high-Mg from Koh. They were probably removed either during lower tholeiites.The boninitesare approximately250 m obductionor subsequentstrike-slip faulting. The extrusiveand thick and generally occur as long pillow tubes, often fragmental ophiolitic volcanic rocks consist of pillow lavas, associated.- with minor green interpillow cherts, breccias, flows, and pyroclastic deposits that can be hyaloclastites,and breccias. The boninites are often vesicular subdivided into three units based on differences in outcrop (up to 30%) and are commonly porphyritic with altered style, petrography and geochemistry. These units can be orthopyroxenephenocrysts. These are succeededby 150 to traced in river sectionsand road cuttingsand over ridgesalong 250 m of daciteswith prominentquartz and clinopyroxene strike for more than 18 km between the villages of Sarram6a phenocrystsand fine vitric tuffs with minor reworkedtuffs. No and M6chin (Figure 2). Cameron [1989] provided useful boniniticdykes or their plutonicequivalents were observedin petrographicand geochemicaldata for the Koh boninites and the ophiolite. some tholeiites but did not distinguish between the various The boninites contain large chlorite+serpentine-altered tholeiitic suites we have identified. orthopyroxene phenocrysts in a groundmass of clinopyroxene laths, altered glass, and occasional small Koh Lower Tholeiite Pillows and Related Plutonic crystalsof chromite.Felsic volcanicsoverlying the boninites Rocks generallycontain large embayedquartz phenocrysts,fresh The oldest magmatic unit in the Koh ophiolite outcropsas augite, titanomagnetite, and small altered plagioclase cumulate gabbros with minor gabbronorite and pyroxenite, phenocrystsset either in a trachytic-texturedgroundmass isotropic gabbros, dolerites, plagiogranites, and tholeiitic containingsmall alteredplagioclase laths or in a groundmass basalts. These rocks can be distinguishedfrom those formed of devitrified glass. 70 •• ET •.: GE••C• EVOL•ON OF • KOH OP•OL•

1165;45'E • 165ø50'E

0 I 2 km

21 ø30'S ß: buildings

...... • sealed roads L__ dirt roads .. . .:.:...... ,....,,.::./ rivers N

21 ø35'S

iMt. Key I serpentiniteand peridotite L•I• Paleogenelimestones and chert ;..-:-•ß Cretaceous sandstone L,..,,,( -:• Cretaceousbasalt nappe, I---] schists

Koh ophiolite volcaniclastic sandstone black siltstone redpelagic siltstone .• orientionof bedding uppertholeiite boninitic felsic volcanic •, overturnedbedding •. faults boninite __ lowertholeiite '• possiblefaults plagiogranite • lowangle thrust faults isotropicgabbin & dolerite layeredgabbin 165o45'E 165ø50'E

Figure 2. Geologicalmap of the Koh ophiolite. MEFFREET AL.: GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE 71

Upper Tholeiite Basalts and Intrusions (Figure 3). Approximately 1100 m above the ophiolite the volcaniclastic rocks become interbedded with black siltstones A secondsequence of tholeiitesoverlies the boniniteseries felsic volcanicsand crops out as large, elongatedpillows up of mid-Triassic age (dated using ammonite fauna; (H.J. to 3.5 m long, interbeddedwith rare red pelagiccherts. The Campbell, Institute of Geological and Nuclear Sciences,New Zealand, written communication, 1990) which constrain tholeiites vary in thicknessfrom 100 m in the northern section of the ophiolite up to 300 m thick in the south. formation of the Koh ophiolite to before the Middle Triassic. Intrusiverocks related to the uppertholeiites occur throughout The black siltstonesare approximately 1000 m thick and are the other volcanic and plutonic units, either as small dykes overlain by volcaniclastic sandstones and conglomerates (-1 m wide) or intrusionsup to 70 m wide of alteredgabbros or which continueup sectionand form a Middle Triassic to Upper diorites.These intrusionscomprise approximately 25% of the Jurassic sedimentary sequence thousands of metres thick plutonicrocks within the ophiolitebut are less commonin [Gudrangdet al., 1975; Paris, 1981; Maurizot et al., 1985]. the volcanic section. The uppertholeiites are the leastaltered rocks of the Koh Other Central Chain Ophiolites ophiolite.They contain small glomerophyricaggregates of Several other ophiolitic sequencesare exposed along the small augire,plagioclase, and titanomagnetitephenocrysts in Central Chain; however, boninitic volcanism is unknown a groundmass of small plagioclase laths, equant outsidethe Koh area. Many of these ophiolitesare overlain by clinopyroxene, and devitrified glass altered to chlorite, red pelagic cherts and volcaniclasticsediments similar to that pumpellyite, and magnetite.They also contain prehnite, describedfrom Koh (Figure 3). The Sphinx ophiolite is the quartz, calcite and bright green, and brown pleochroic most similar to the Koh ophiolite and occurs 50 km to the pumpellyitewithin amygdalesand veins. south. It includes cumulate plutonic gabbros, plagiogranites, dolerite intrusions,and a pillowed sectionup to 850 m thick, similar to the lower tholeiites at Koh. The Tarouimba Sedimentary Rocks ophiolite is a related fragment offset by 10 km north-west The upper tholeiites are conformably overlain by green and along a sinistral strike-slip fault. The Cantaloupa¾ophiolite red pelagic siliceoussiltstones up to 130 m thick which drape in the north of the island contains both igneous and over the uppermost pillows and are succeededby tuffaceous sedimentaryrocks which have a strong metamorphicfoliation siltstones, volcaniclastic sandstones, and conglomerates and have crystallised lawsonire and fine-grained blue

sedimentarysequence continues for >3000 m volcaniclasticsandstone pelagic chert & siltstone CentralChain tholeiite uppertholeiite boniniticfelsic volcanics 1 m: boninite lowertholeiite isotropicgabbro &dolerite plagiogranite doleritedyke complex rn layeredgabbro uppertholeiite intrusion basalserpentinite rn i '--'-- -F 750 850 rn

100 rn 500 500 1150rn 1000 100•mv 1500 1

Cantaloupai Tarouimba Sphinx Koh Pocquereux Nassirah Koua

Figure 3. Stratigraphiccolumn of the Central Chain ophiolites.Diagonal lines representfaults; horizontal lines representconformable contacts; arrows show that conformablesedimentary sequence continues upward 72 MEFFREET AL.' GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE

'•r'-•docS•eqcScSmo• V --oom

•d•oeqo 'ddm• --•m MEFFREET AL.: GEOCHEMIC• EVOLUTION OF THE KOH OPHIOLITE 73 amphibole during a Palaeogene metamorphic deformation Silica has probably been removed from the boninites during event. The Pocquereux,Koua, and Nassirah ophiolitesin the chlorite-serpentine alteration of groundmass glass and south are volcanic and shallow intrusive sections broken up orthopyroxenephenocrysts. This may explain the relatively into small fragments(1 to 3 km in length) by low angle thrust low SiO2 content of the less evolved, orthopyroxene-phyric faults related to Eocene tectonics. These fragments include boninites. well-developed dyke complexes, a feature absent from the Plots of Mg# versusTi, Y, and Zr clearly separatethe Koh Koh, Sphinx and Tarouimbaophiolites (Figure 3) magmatic suites, and, unlike the tholeiitic suites, which show strong Fe-enrichment,the boninites define a trend of strongly increasing SiO2 and decreasing FeO* with advancing Geochemistry fractionation. Fifty three least altered samples were chosen for x ray fluorescence (XRF) analysis in order to investigate the Fractionation History chemical evolution and variation throughout the Central Tholeiites Chain ophiolites (Table 1). XRF analysis was undertakenat the University of New South Wales in Sydney and the Differentiation within the upper and lower tholeiites is University of Tasmania in Hobart using the methods of typical of MORB, with rapidly increasing Fe/Mg and high Norrish and Hutton [1969]. Rare earth elements (REE) from field strength element concentrations(HFSE; Ti, P, Zr, Y) selected samples (Table 2) were analysed by instrumental with increasing fractionation. Decreasing A120 3 with neutron activation analysis at Becquerel Laboratoriesin Lucas fractionation reflects plagioclase separation over the Heights, Sydney. crystallisationrange sampled.SiO2 tendsto be low (45-54%), Thirty-three sampleswere analysedfrom the Koh ophiolite typical of ocean floor basaltsand basalticandesites. and 20 from the rest of the Central Chain ophiolites.The Koh The three tholeiitic suites have distinctive major element ophiolite samplesbelong to three geochemicalseries, which compositions.The two suitesat Koh have parallel liquid lines correspondto the three extrusive suitesdescribed above. The of descent, with the upper tholeiites having generally lower samples from the remainder of the Central Chain ophiolites FeO* for a given MgO and being generally more evolved can be grouped into a single geochemicalsuite, referred to (Mg# 47-34) than the lower tholeiites (Mg# 65-46). The here as the Central Chain tholeiites. The major element tholeiites from the remainder of the Central Chain ophiolites characteristicsfrom each unit are illustrated in Figure 4, for are compositionally transitional between the two tholeiite whichMg# (100Mg2+/(Mg2++Fe2+tøtal))waschosen as the suites at Koh and probably commenced titanomagnetite abscissa as it provides the clearest indication of crystallisationat slightly lower TiO 2 (2.2%) and FeO* (13%) differentiation trends for the tholeiites. Compositionsof the than in the upper tholeiitic basalts (>2.4% and >14% Central Chain ophioliteshave been alteredduring mainly low- respectively) (Figure 4). grade (greenschist and subordinate prehnite-pumpellyite The two analysed layered gabbros contain only 10.8 and facies) sub seafloor and regional metamorphism.The effects 8.6% MgO (Table 1) and are a good match for the average of greenschist facies alteration on the geochemistry of crystallisation assemblage deduced from the lower tholeiite basaltic rocks have been extensively documented [e.g., compositionaltrends (Figure 4). This indicatesthat the modal Humphris and Thompson, 1978; Coish et al., 1982]. These mineralogy of the cumulate gabbro, with 50-60% plagioclase, studiesand many otherssuggest that Ti, Zr, Nb, Y, A1, and Ni 40-50% clinopyroxene, and minor orthopyroxene,is a close can be considered immobile in most circumstances. Plots of approximation to the fractionation assemblagewhich caused major elements againstTi (not shown) suggestthat alteration the liquid lines of descent. More detailed major element has affected Na, Si, Ca, K, Ba, Sr, Rb, and Mn to such an modelling was not attempted because of alteration-affected extent that magmatic trends are obscured.In contrast,Mg, Fe, CaO and alkali abundances. A1, Cr, Ni, V, and P show good trendswith significantscatter, Each of the tholeiite suitesspans a relatively small range of and Y and Zr seem to be the least mobile; Nb contents are composition(differences in Mg#< 20), equivalentto the range generally close to detectionlimit, but Nb can be expectedto observed in most MORB and BABB suites [Stakes et al., behave like Zr during low-grade hydrothermal alteration. 1984; Sinton and Fryer, 1987]. This suggeststhat these suites probably formed by magma mixing in a relatively open magma chamberundergoing periodical replenishment,similar Table 2. Rare Earth Element (REE) CompositionsFrom the to those postulated for MORB or back arc basin basalts Koh Ophiolite (BABB) [e.g., Sinton and Detrick, 1992]. boninite boninite L thol Lthol Lthol Lthol Lthol Uthol pillow pillow pillow dyke pillow flow pillow pillow 70944 70954 70958 70959 70960 70938 70962 70948 Boninites and Felsic Volcanics La 0.72 1.16 0.90 0.70 1 25 1 42 3.27 4.04 Ce 1.68 3.09 2.81 2.64 4 51 585 10 30 13.50 The Koh boninites differ considerablyfrom the tholeiites. Nd 1.14 2.23 3.30 2.87 5 81 6 20 11 40 12.60 The A1203 content is significantly lower, increasing until Sm 0.35 0.73 1.56 1.15 2 51 2 67 421 4.33 Eu 0.14 0.31 0.71 0.33 0 97 1 02 1 23 1.66 Mg#-60, while FeO* decreases, reflecting the late Tb 0.09 0.23 0.51 0.36 0 60 0 65 1 02 1.00 crystallisationof plagioclasein this suite (Figure 4). In the Ho 0.14 0.36 0.65 0.62 0 92 1.05 1 52 1.54 Yb 0.41 1.07 1.81 1.76 2 61 2.70 3.99 4.24 more evolved rocks (Mg#<60), A1203 decreasesrapidly and Lu 0.06 0.17 0.25 0.26 0 38 0.35 0.56 0.62 FeO* remains at 5-6%, reflecting continued mafic mineral Values are in ppm. Analysed at Becquerel Laboratories, crystallisationtogether with plagioclase.SiO2 also increases, Sydney. Abbreviations as in Table 1. along with a rapid decrease in Ni with advancing 74 MEFFRE ET AL.' GEOCI-IE•C• EVOLUTION OF THE KOH OPHIOL1TE

'4-

45 •,,,

1.5

0.5 A• -•+ + 0

18 ß 16

"'12

"• 10

•8

2 .,,, ,I .... I .... I,, ,[I •,,[ [,,• , . 16 _

14 -

& 10

0 8 ß • 6

2 -- I I i I I I I I I I I I ] I I I I [ I I I I ] I I I I

12 o••o o 8

30 40 50 60 70 80 90 30 40 50 60 70 80 90 Mg# Mg# Figure4. Mg# variationdiagrams for selectedmajor and trace elements showing the four magmatic units from the CentralChain ophiolites. Symbols are as follows:solid diamonds, lower tholeiite;crosses, boninitic series volcanics;open circles, upper tholeiite; open triangles, Central Chain tholeiite. Dashed lines are CentralChain tholeiitefield. Data for the Koh ophiolitefrom Cameron[1989] are included. MEFFREET AL.: GEOCHEMICALEVOLUTION OF THE KOH OPHIOLITE 75

fractionation, suggesting initial crystallisation of olivine clinopyroxene, followed by plagioclase and clinopyroxene at down to Mg#--65. The early rapid depletion of Cr confirms Mg#<65. The fractionation of quartz from evolved liquids that chromite is an important early fractionating mineral. cannot be inferred from the compositionaltrends, as SiO2 is Although CaO must be mobile to some extent in these rocks too mobile in the alteration of these largely glassy rocks. during the alteration of the glass, the scatter of composition The differentiation range (48-68% SiO2, Mg# 78-50) and is relatively small comparedto the tholeiite suites (Figure 4). the chemical zonation of the Koh boninites indicate that part This may be becausea large fraction of CaO in the boninites of the boninitic sequence is likely to have formed in a resides in the unaltered groundmass clinopyroxene relatively closed system. The phenocryst-richnature (average microphenocrysts,rather than in (albitised) plagioclase laths •-20%) and moderate Mg# of the stratigraphically lowest as in the tholeiites. CaO in the boninites remains essentially boninites indicate that the initial primitive magmas are not constantwith increasing differentiation until Mg# --60; in the exposedand may not have escapedthe magma chamber.This more evolved rocks, CaO decreasesrapidly with increasing chamber was probably replenished during extrusion of the fractionation, reflecting plagioclase and clinopyroxene boninites, as there are no systematic changes in extent of crystallisation. differentiation in the lower section of the sequence.However, The liquid line of descentis compatiblewith the sequenceof the overlying boninitic felsic volcanics record an upward mineral crystallisation deduced from the phenocryst increase in differentiation from plagioclase-phyricandesires assemblage. This sequence is characterised by early at the baseto quartz-phyricdacites and possiblyrhyolitic tuffs crystallisation of olivine, orthopyroxene, chromite, and at the top. This suggeststhat replenishmentof the chamber

r• 50

0.4

•0.3

I--0.1

16 , i , i , i i ......

.• 10 o

lO 9 8

Li. 3

1 , i , • , , , , , , . ß 2 4 6 8 10 12 14 16 18 MgO (wt. %) MgO (wt. %) Figure 5. Diagram showingthe chemicalthree groupsfrom the boninite seriesvolcanics. Symbols are as follows: open square,cumulus enriched boninites (Group 1); solid diamonds,aphyric and porphyriticboninites (Group 2); crosses,boninitic volcanics possibly contaminated with tholeiitic magma(Group 3). Data for the Koh ophiolite from Cameron [1989] are included. 76 MEFFREET AL.:GEOCHEMICAL EVOLLrI'ION OF THE KOH OPHIOLITE

lOC by boniniticmagma stopped or slowedafter the eruptionof Koh lower tholeiite 200 to 300 m of boniniticlavas, allowingessentially closed- systemfractionation to proceedthrough to increasinglyfelsic compositions. ..•:•...;•A t'"Hunter RidgeIow-Ti tholeiite The boninitic volcanics can be subdivided into three ..... /\ \ , -Lau.asinsite geochemicalgroups (Figure 5). Group 1 containsabundantly orthopyroxene-phyric(20-25 modal %) boninites in which the highCr, Ni, andSiO2 and relatively low A1203are likely due to accumulation of early formed phenocrysts of orthopyroxene (with chromite inclusions). Addition of lO( varying amountsof these minerals to liquids at different Koh boninite stagesof evolution explains the lack of correlation between Cr and Ni with MgO. Group 2 boninitesinclude aphyric samplesand coverthe main liquid line of descentbetween 16 and 6% MgO. This line of descentcontains little scatterin all but the mostmobile elements, indicating that a large part of '•_• /•\ A •' HunterRidge boninite the variation was causedby magma chamberfractionation processes. Group 3 includesall the evolved samplesand one more mafic sample(11% MgO). Relativeto Groups1 and 2, these rocks containtwice as much TiO2 as well as higherY, Zr, A1203 and FeO and lower SiO2 at a given Mg# and form a separateliquid line of descentfrom Groups1 and2. In general, uppertholeiite its geochemicalcharacteristics are offsettoward the liquidline •"•'•'"'••'•.'•;{/•111!,•_ ,•Lau Basinsite834 of descentof the tholeiites.This suggeststhese Group 3 lavas may have been contaminatedby liquids derivedfrom a more fertile source,possibly as a result of an overlap between magmageneration associated with the beginningof the upper tholeiiticevent and the end of the boniniticmagmatism. This LLD effect is most notablefor Ti, Zr, and Y, probablybecause theseelements are very low in the boninitesin comparisonto the uppertholeiites. Simple least squares mixing models show that the patternof enrichmentin Group3 boninitesis similar Figure 7. Mid-ocean ridge basalt (MORB) normalisedtrace to that producedby mixing -15% of typicalupper tholeiitic element patterns for the tholeiitic rocks from the Koh basaltwith an evolvedGroup 2 boninite. ophiolite and modernisland arc systems.Data for the Koh In summary,the boniniteswere derivedfrom liquidswhich ophiolite from Cameron [1989] are included. LLD is lower evolvedin an opensystem magma chamber into whichsupply limit of detectionfor Nb. Symbolsare solid diamonds,lower tholeiite; crosses,boninitic seriesvolcanics; open circles, upper tholeiite. Data are from the following sources:Lau lOO Basin, Ewart et al. [1994]; Hunter Ridge, A.J. Crawford (unpublisheddata, 1995). Normalisationdata are from Sun and McDonough [ 1989].

upper tholelites

•k%-• '"'"'""" '•'""•'••.'.': ';••.•.•::• .••..: ,.:•i:...'•..... '+..•'.'•.';•.•• •. •...•• .••.•----.-•-..• •.••.••.•••••::•:•:•:.••:•:::•:.• of boniniticparental magmas eventually ceased. A final batch of evolvingboninitic magma mixed with a volumetrically muchsmaller, newly arrived batch of uppertholeiite magma, ::::::::::::::::::::::::::::::::::::::::::::: :•.. :: ::::: ...... ½• ...... :::::::::::::::::::::.•. •.•.• .• ;•:•.::::::::::::::::::::::::::.:•:•%:.•::::.:::.:.:.:.:.::•...•.:..:.•.•:..•.:.•.:...... •.• and this hybrid magmafractionated further to producethe boninitic felsic volcanics.Implied is a relatively sudden significantchange in the fertility of the magma source betweenthe boninitesand the uppertholeiites. i•a - ' Boninites Sourcesof the Magma Suites I I I I I I I I La Ce Nd Sm Eu Tb Dy Ho Yb Lu The geochemicalcharacteristics of the threemagmatic units of the Koh ophiolite suggestthat they were derived from Figure 6. Rare earthelement (REE) patternsfrom the three separatesources. The lower tholeiiteswith 7-9% MgO show volcanic units in the Koh ophiolite. Data for the Koh strong depletion of the light rare earth elements (LREE) ophiolite from Cameron [1989] are included. Normalisation (Figure6) andhave significantly lower TiO2 (0.55-1.25%),Zr dataare from Sun and McDonough[ 1989]. (29-64ppm), and Y (18-31ppm)contents than N-type MORB MEFFRE ET AL.: GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE 77

(Figure 7). Their Nb contentsare generally below the XRF detection limit (0.5-1ppm), indicating La/Nb values significantlyabove thoseof N-MORB; this is reflectedin the 14 - unterRidge negative Nb anomaly in the MORB-normalised element I /• •/• boninites variation patterns in Figure 7. The lower tholeiites are I northTonga •/' '• •J significantlymore depletedthan typical BABB or MORB and 12I- bøn•ite.•I ///•• based on this, and the negative Nb anomaly would normally be classified as island arc tholeiites. However, some of the more evolved lower tholeiite (5-7% MgO) are significantly higher in TiO2 (1.13-2.16%) and FeO* (10.9-13.8%) and • 8 I- _•+' • J boninites/ lower in A1203 (12.8-14.6%) than island arc tholeiites [e.g., Ewart et el., 1977; Woodhead, 1988] and similar to those OI '••++ • /' /ODPsite786B,'' producedduring the openingof young back arc basinsand in 6- G•• •O.•• + ,/,/ arc rift settings[e.g., Hochstaedteret el., 1990; Hawkins et el., 1990]. During ODP Leg 135 in the Lau Basin, basalts more depleted than MORB, with pronouncedarc tholeiites ,/,/' affinities, were drilled [Ewart et el., 1994]; these BABB are very similar to the Koh lower tholeiites(Figure 7). o The boninites require a highly depleted mantle source to 4 6 8 10 12 14 16 18 20 account for the very high Cr/(Cr+A1) values of chromite AI203(wt %) (Figure 8), the very low Ti, Zr, and Y abundances,and the low Figure 9. Comparisonof CaO and A1203 betweenboninitic levels of heavy rare earth elements (HREE) (Table 2 and volcanics from Koh and modern arc systems.The samples Figures4 and 6). The relatively high CaO contentand the high with low CaO, high A1203 are from the boninitic felsic CaO/A120 3 values (Figure 9) suggestthat they are high-Ca volcanics and the Group 3 boninites. Data sources are for boninites[Crawford et el., 1989]. The slightly U-shapedREE northTonga, Falloon and Crawford [1991]; for OceanDrilling patterns are taken to indicate the addition of a LREE-enriched Program(ODP) site 786B, Arculuset el. [1992]; Hunter Ridge, component to a refractory peridotite source before or during Crawford (unpublisheddata, 1995). boninite genesis [Crawford et el., 1981; Cameron et el., 1983].

1.0 Whole rock compositionsof the upper tholeiites indicate a sourcemore fertile than, but similar in many respects,to that 0.9 which yielded the lower tholeiites. As for the lower tholeiites, MORB-normalised element distribution patterns of the upper 0.8 tholeiites (Figure 7) show negative Nb anomalies. Furthermore, the REE patternsof the upper tholeiites (Figure 0.7 6), with peaksat Nd-Sm, are typical of BABB producedduring the earlieststages of back arc basin opening[Crawford et el., 0.6 1986; Hochstaedter et el., 1990; Hawkins and Alan, 1994]. The upper tholeiites have higher end values (7.8-8.4 at 250 0.5 Me) compared with the single primitive lower tholeiite • ) Site analysed(6.7 [Cameron, 1989]) and high La/Sm at equivalent J834&836 TiO2 contents(Figure 10). Such differencesare unlikely to be u 0.4 la• •-J_auBasin related to fractionation, as they persist across the /Spreading/ fractionetlonrange within each suite; rather, they are taken to 0.3 Centre/ reflect source/primarymagma differences. The tholeiites from the remainder of the Central Chain 0.2 ophiolites are similar to those from Koh, but the lack of REE and isotope data makes it difficult to compare their sources. 0.1 However, it is likely that the similarities of the major and trace elementsshould extend to the REE and Nd isotopes. 1 .o 0.5 o.o All tholeiite suites show strong Fe-enrichment but little tendency to SiO2 enrichment with increasing fractionation, Mg# (Mg/(Mg+fe2+)) very high Ti/Zr values in their least evolved members (140- Figure 8. Comparisonof the mineralchemistry between 160), LREE depletion, and significantnegative Nb anomalies. chromites from Koh boninitic rocks (solid squares) and Although the K20 abundancesof theserocks are unlikely to be chromitesfrom modernisland arcs and the Troodosophiolite. primary, they are uniformly low (usually less than 0.2%), and Fe2+ is calculatedfrom stoichiometry.Data sourcesare for this, taken together with the compositionalcharacteristics of Chichijima,Umino[1986]; for Troodos,Cameron [1985]; for the eruptive units and dykes discussedabove, suggeststhat Lau Basin, Allan [1994]. they were low-K tholeiiteserupted in a SSZ setting. 78 MEFFRE ET AL.: GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE

interval between the eruptionof the units at Koh would be of the order of 1-2 m.y. 4.5 // 1 [*Kohtholeiites i I+Kohboninite During the time interval between the lower and upper [ouppertholeiite tholeiite events, the ophiolite may have migrated away from the lower tholeiite extension zone, or spreading may have 3.5 o become temporarily inactive. These two possibilities have $ different implications for the spreading history of the 2.5 / •,• 9..•, / ophiolite but are difficult to distinguish based on available data. In either case, the upper tholeiite extension apparently never developed into a full spreading event based on the outcrops examined at Koh, as the upper tholeiites comprise 1.5 only 20-30% of the plutonic sequence.However, the small Hunte'+• L•a•• •,•n• '•• RidQe k • • • ...... thicknessof the lower tholelite unit comparedto the extrusive sections from the other Central Chain ophiolites suggests 0.5 •onlnltes/• , that tectonic attenuation was important and that extension T•n•a-•e•ma•e•a•c ...... may have been as large as 70-80 %. The evolved 0.5 1 1.5 2 compositionsof the upper tholeiitesmay be the result of high TiO2 (wt%) cooling rates and low rates of magma supply, similar to the Figure 10. Comparisonof La/Sm ratios and TiO2 for the highly fractionatedbasalts from the tip of propagatingrifts in three volcanic units in the Koh ophiolite and modern island the Lau Basin [Pearce et al., 1994]. The upper tholeiitesmay arc systems. Note the low La/Sm of the lower tholeiites have formed on the edge of a propagatingrift zone or after a comparedto the upper tholeiites.Data for modem arc systems lengthy hiatus of magmatic activity at a sporadic spreading are from the following sources: Lau basin, Falloon et al. centre. [1992], Ewart et al. [1994]; N. Tonga,Falloon and Crawford The eruption of the boninites between these two [1991]; Izu-Bonin-Mariana (IBM) arc, Woodhead [1988]; extensional events does not constrain the extension history. Tonga-Kermadec (TK) arc, Ewart and Hawkesworth [1987]; The lack of interbeddingbetween the lower tholeiitesand the MarianaTrough, Hawkins et al. [1990],Sinton and Fryer boninites implies that they developed after generationof the [1987]; ODP Site 786, Arculus et al. [1992]; MORB, selected initial oceanic floor; however, the lack of boninitic intrusive analyses from Schilling et al. [1983]; Hunter Ridge, A.J. rocks in the plutonic sequence suggests that they formed Crawford (unpublisheddata, 1995). before the main phaseof extensionassociated with the upper tholeiite intrusions. The lack of boninites in the rest of the Central Chain ophiolites, which contain dyke complexesand indicate a single episode of extension, suggests that the Spreading History boninites may have been generatedonly in localised areas. The Koh and Central Chain ophiolites probably formed during an episode of extensional magmatism in an SSZ Tectonic Setting and Petrogenetic Scenario for environment. The screen of dykes which separatesthe basal the Koh Boninites tholeiitic lavas from the uppermost gabbroic rocks (better developedat Pocquereuxand Cantaloupa¾than at Koh) and the Boninites occur in the forearc regions of modem island arc occurrenceof thick tholeiitic sequencesof MORB- or BABB- systemsin the western and south western Pacific. They have type basalt-ferrobasaltpillows which were produced from been reported from Eocene sectionsof the Izu-Bonin-Mariana open magma chambers indicate that tholeiitic magmatism (IBM) arc system [Shiraki et al., 1978; Meijer et al., 1982; probably occurredin an extensionalenvironment rather than Sharaskin et al., 1983; Crawford et al., 1986; Umino, 1986; in stratocone-type arc volcanoes. This suggeststhat the Bloomer and Hawkins, 1987; Hickey-Vargas and Reagan, ophiolites were formed by a process similar to seafloor 1987; Hickey-Vargas, 1989; Arculus et al., 1992; Pearce et spreadingin a SSZ environment. al., 1992], from the northerntermination of the Tonga Trench The three geochemical units from the Koh ophiolites [Falloon et al., 1987; Falloon and Crawford, 1991; Sobolev formed during a period of complex spreadingand extension, and Danyushevsky,1994], and from the Hunter Ridge between with magmatism occurring during two separateevents, the Fiji and southernmostVanuatu [Sigurdssonet al., 1993; A.J. first forming the lower tholeiitesand the secondforming the Crawford et al., unpublisheddata, 1995]. A featurecommon to upper tholeiites. No exposed intrusive equivalents of the each of these locations is their proximity to the trench (<150 boniniteswere found; they may have formed either at the end km). of the lower tholeiite extensionalevent or at the beginningof The Koh boninites are similar to the high-Ca boninites the upper tholelite event. The time interval separatingthe two from north Tonga [Falloon and Crawford, 1991] and IBM extension episodes was probably short, as only small Ocean Drilling Program(ODP) site 786B [Pearce et al., 1992] amounts of pelagic or volcaniclastic sedimentsoccur within in their SiO2, CaO, FeO and MgO content,but they are lower the pillowed sequence,the largest being a lens <20 m thick in A1203, TiO 2, and Zr (Figures9 and 11). The low A1203 between the boninitic felsic volcanics and the upper comparedwith that of modernboninite suitesmay be due to a tholeiites. Assuming typical rates of pelagic sedimentation low water content, causing early plagioclase fractionation. (e.g., 14 m/m.y. [Rothwell et al., 1994]), the maximum time This possibility is supportedby the inflection of the A1203 MEFFRE ET AL.: GEOCHEMICAL EVOLUTION OF THE KOH OPI-[IOLITE 79

0.8 The closespatial and temporalassociation of the Koh high- Ca boninites with the upper and lower tholeiites, the 0.7 similarity of the Koh upper tholeiites to basalts erupted in immature or propagatingB AB, plus the similarities noted 0.6 abovebetween the magmasuites erupted in the Lau Basinand Hunter Ridge where back arc basin spreadingridges are 0.5 transecting preexisting protoarc or back arc basin crust all suggestthat the Koh boniniteswere eruptedin a SSZ setting where a back arc basin spreadingridge was propagatinginto older protoarc or young back arc basin crust. The absenceof boninitesin other Central Chain ophiolitesin New Caledonia probablyreflects their derivation from segmentsof back arc

0.2 basincrust that were unaffectedby later ridge propagation. The boninite generation model which correspondsmost closelyto the situationat Koh is that outlinedby Crawford et 0.1 al. [1981], Coish et al. [1982], Crawford et al. [1989], Hickey-Vargas [1989] and Falloon et al. [1992] for the generationof boninitesduring the rifting of arc lithosphere

Figure 11. Comparison of Zr and TiO 2 for boninitic lower tholeiite volcanicsfrom Koh and modern arc systems.Symbols and spreading data are as for Figure 13. Data from Cameron [1989] are included.Data for modernarc systemsare from the following sources:north Tonga, Falloon and Crawford [1991]; IBM forearc, Arculus et al. [1992], Wood et al. [1982]; Hunter Ridge, A.J. Crawford (unpublisheddata, 1995). and CaO liquid lines of descentat higherMgO andlower A1203 than in the Marianas ODP site 786B. The Koh boninites have very high Ti/Zr relative to other lower tholeiite shallowdepleted magmatism spreading boninite suites(Figure 11), more similar to typical BABB and possiblyassociated with plate MORB. They lack the characteristicZr enrichment of many drivenmantle upwelling Tertiary boninite suites. However, the slightly U-shaped REE ...: ..:.:.:.:.:.:.:.:.:...:... patterns(Figure 6), the high La/Sm ratios (Figure 10), and the Nd isotoperatios [Cameron, 1989] show that a LREE-enriched .'-•:•:•<•:•:•':•:::•:.::::::'•'::••%•.•;:::•::::::•:•::•:•:•::•:::•:•:::•?•::•!•::•!•::•!•;•::•i•::•:•::•:::::::::::::::::::::: ::::::::::::::::::::::::::::::•::•!•!•::•:5•:•::i5i.:•;•:•:-::--'-•,• ß lOWENd componentwas presentthe mantle source. •.., ,•,,.;• • ,•,., The recent drilling of refractory tholeiites with boninitic characteristicsin the Lau Basin [Ewart et al., 1994; Hawkins, 1994] and the dredging of boninites from off-axis seamounts in the central (Sunkel, [1990] as given by Falloon et al. [1992]) and northern [Hawkins, 1976; Hawkins and Melchior, partialmelting of fertile lowertholeiite mantleresumes, spreading 1985; Falloon et al., 1992] sectionsof the Lau Basin suggest beginingof upper • that boninites may form in young back arc basins relatively far from the trench, during off-axis magmatismand spreading ridge propagationinto arc and back arc crust [Falloon et al., tholeiitespreadin••,••r• • 1992]. Similarly, high-Ca boninites are recorded from the Hunter Ridge, where the protoarc is being transected by a propagating ridge tip associated with the southernmost spreadingcentre in the North Fiji Basin [Sigurdssonet al., 1993; Monzier et al., 1993; A.J. Crawford et al., unpublished data, 1995]. The boninites at the latter locality occur in close association with both low-Ti tholeiitic basalts and more Figure 12. Schematic mantle melting model for the fertile BABB associated with the propagating ridge tip, formationof depletedmelts during rifting of youngoceanic although the stratigraphic sequence of magmatic suites crust,based loosely on the modelsof Langmuir et al. [1992], remains unknown. Furthermore, the Hunter Ridge boninites Pearce et al. [1992], and Turcotteand PhippsMorgan [1992]. show no positive Zr anomalies, have slightly lower SiO2 at Depleted mantle (harzburgite)is representedby a dark shade; MgO>8% than typical high-Ca boninites (e.g., from north fertile mantle(lherzolite) is representedby a light shade.Zone Tonga forearc), and are remarkably close compositionallyto of partial melting is representedby solid lines in the mantle, the Koh boninites (Figures 7, 9, 10 and, 11). and mantle flow is representedby arrows. 80 MEFFREET AL.: GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE

and the initiation of back arc basins. An even closer analogy relation exists between the nature of boninites (high-Ca may be providedby the site of intersectionof the Hunter versuslow-Ca) and the natureof the associatedtholeiites. Ridgeprotoarc by the propagatingsouthern tip of the North Fiji Basinspreading centre (A.J. Crawfordet al., manuscript Conclusions in preparation).The Koh boninitesand upper tholeiites were The model proposed here to explain the geochemical probablyformed when mantle upwelling was initiatedin stratigraphyof the Koh ophiolite involvesthe productionof residual mantle close to the lower tholeiite spreadingzone or boninitesduring upwelling of depletedmantle at the initiation at the samesite duringreactivation by a propagatingback arc of rifting in young oceaniclithosphere in a back arc basinin basin spreadingcentre. The partial meltingwhich occurs responseto theinvasion by a propagatingback arc basin rift- duringadiabatic decompression of depletedmantle depends tip (Figure12). The modelassumes that a hotresiduum of largelyon how far belowits solidusthe mantleis, whichin depletedmantle underlies the recently formed lower tholeiite turn is a functionof age,pressure, and water content.Both lithosphere.Mantle upwellingduring rifting will cause depletedmantle residuum at the initiationof subductionin depletedmantle residuum from the lower tholeiitic spreading youngoceanic crust, such as thatoutlined by Pearceet al. to riseand partially melt. The production of thesemelts will [1992]for ODP leg 125boninites, and the residuum at theend be stronglycontrolled by how far theserocks are below of the lower tholeiite event at Koh would be close to their solidus,by the structureof the mantleupwelling, and by the respectivesolidus temperatures, as they form only a short drivingmechanism of mantleupwelling. In depletedmantle time after productionof oceaniccrust. The main difference which has had little time to cool, such as the residuum from betweenthese two settingsis that slab-derivedfluids play a the lower tholeiite-producingmelting event, partial melting greaterrole in the leg 125 boninitesas a resultof their mayresume soon after the initiation of upwelling.Subsequent proximityfrom the trench. melts are drawn from increasingly fertile MORB source asthenosphericmantle beneath the region.Unless extraction is rapid,the refractoryliquids will mix with morefertile material.Rapid extractionof the depletedmelts may occur Implications for Other Ophiolites whenthe rate shallowmantle flow is greaterthan that of the Boninites and boninitic rocks are well known in Mesozoic deepermantle, such as predictedfor plate-drivenupwelling and Tertiary ophiolitessuch as Troodos,Cyprus [Cameron, [Turcotteand PhippsMorgan, 1992]. 1985; Kostopoulosand Murton, 1992]; Pindos,Greece [Jones The specialconditions required for boninitegeneration in and Robertson, 1991]; Koh, New Caledonia [Cameron, 1989; sucha settingare restricted to thebeginning of reactivationof this paper];Cape Vogel, PapuaNew Guinea[Walker and mantleupwelling. At Koh, the boninite-productionevent Cameron, 1983]; and Zambales,Philippines [Hawkins and probablyoccurred on the edgeof the rift zone,where it Evans, 1983]. They are also presentin lower Palaeozoic propagatedinto the lower tholeiite crust without the ophiolites,such as BettsCove, Newfoundland [Coish, 1989]; productionof new seafloor,although the old floor was Thefiord Mines, Appalachians[Coish, 1989;Laurent and extendedand thickened.If the upper tholeiitic event later H•bert, 1989]; Karm0y, Norway [Pedersenand Hertogen, developedinto a full-scalespreading centre, this occurred at a 1990]; Ballantrae, Scotland[Smellie and Stone, 1989]; Khan- positionnot presentlyoutcropping within the Kohophiolite. Taishir, Mongolia [Zonenshain and Kuzmin 1978]; and The presenceof a singleextension event in the restof the Victoria and Tasmania, Australia [Crawford and Cameron, CentralChain ophioliteshows that the conditionswhich 1985; Crawfordand Keays,1987; Brown and Jenner,1989]. formedthe Koh uppertholeiites and boninites were restricted In all these ophiolites, boninites are accompaniedby in timeand space. Further spreading during the upper tholeiite tholeiitic volcanics issued from a less refractory source.The event would causethe formationof new oceanfloor, possibly length of time betweenthe generationof the tholeiitesand indistinguishablefrom either the lower or CentralChain boninites is short, as thick pelagic or clastic sedimentary tholeiites. sequencesrarely separatethe tholeiitesfrom the boninites. The coarse calc-alkaline arc-derived sedimentsoverlying This suggestsa geneticlink betweenthe tholeiitesand the the Koh and related ophiolitesindicate that this crust was generationof boninites [Meijer, 1980]. In some of these locatednear arc volcanoesduring the longhistory of the basin ophiolites (e.g., Troodos, Khan-Taishir, Tasmania) the which the ophiolitefloored. This implies a narrow and tholeiites have depletedincompatible element contents(and probablyyoung arc related basin, since spreading close to the are strikinglysimilar to the Koh lower tholeiites);however, arc mainlyoccurs in backarc basinsshortly after arc rifting in many others,the boninitesare associatedwith MORB- or (Hawkins, 1994). Thesevolcaniclastic sediments also imply BABB-type tholeiites (e.g., Victoria, Thetford Mines, thatthe Koh boniniteswere not producedduring the initiation Karm0y Betts, Cove, Zambales,Pindos), similar to the upper of the island arc, as calc-alkalinemagmatism is normally tholeiites from Koh. As this association has no well-studied restricted to mature intraoceanic arcs. equivalentin modernisland arcs, it presentsone of the main difficultiesin determiningtectonic setting. Similar modelsto Acknowledgements.The authorswould especially like to thankD. that proposedfor Koh may apply to other ophiolites,but Cluzelfor manyinteresting discussions onthe tectonic development of careful investigationof the stratigraphy,spreading history, the New Caledonia.This researchwas supportedby the Australian andchemistry is required.In particular,it is criticalto notethe PostgraduateAwards (S.M), grantsfrom the AustralianGovernment nature and location (before, or after boninites) of tholeiites Departmentof Trade, Industryand Commerce(J.C.A.) and the associatedwith boninitic magmatismand also whether any Australian Research Council (A.J.C.). MEFFRE ET AL.: GEOCHEMICAL EVOLUTION OF TIlE KOH OPHIOLrHE 81

References

Aitchison, J.C., G.L. Clarke, D. Cluzel, and S. Meffre, and C. Picard, D6nudation termination of the Tofua magmatic arc, Meffre, Eocene arc-continent collision in tectonique du complexe h noyau Tonga and adjacent Lau Basin, Aust. J. New Caledonia and implications for m6tamorphiquede haute pressionTertiaire Earth Sci., 34, 487-506, 1987. regional SW Pacific tectonic evolution, du Nord de la Nouvelle-Ca16donie Falloon, T.J., A. Malahoff, L.P. Zonenshain, Geology,23, 161-164, 1995. (Pacifique, France) donn6escin6matique, and Y. Bogdanov, Petrology and Allan, J.F., Cr-spinelin depletedbasalts from C. R. Acad. Sci., Ser. 2, in press,1995. geochemistry of back-arc basin basalts the Lau backarcbasin: Petrogenetic history Coish, R.A., Boninitic lavas in Appalachian from the Lau Basinspreading ridges at 15ø, from Mg-Fe crystal-liquidexchange, Proc. ophiolites, in Boninites, edited by A. J. 18ø and 19øS, Mineral. Petrol., 47, 1-35, Ocean Drill. Program Sci. Results, 135, Crawford, pp. 264-287, Unwin Hyman, 1992. 565-584, 1994. Boston, Mass., 1989. Gu6rang6, B., R. Lille, and J. Lozes, Etude Arculus, R.J., J.A. Pearce, B.J. Murton, and Coish, R.A., R. Hickey, and F.A. Frey, Rare g6ologique des t6rrain ant6-Oligoc•ne de S.R. Van der Laan, Igneous stratigraphy earth element geochemistryof the Betts la chaine centtale N6o-Ca16donienne: and major-elementgeochemistry of holes Cove ophiolite, Newfoundland: Stratigraphie, r6gime de s6dimentation, 786A and 786B, Proc. Ocean Drill. Complexities in ophiolite formation, 6volution structurale et m6tamorphisme, ProgramSci. Results,125, 143-169, 1992. Geochim. Cosmochim. Acta, 46, 2117- Bull. B.R.G.M., set. 2, 127-137, 1975. Bloomer, S.H., and J.W. Hawkins, Petrology 2134, 1982. Hawkins, J.W., Petrologyand geochemistryof and geochemistry of boninite series Crawford, A.J., and W.E. Cameron, Petrology basaltic rocks of the Lau Basin, Earth volcanic rocks from the Mariana trench, and geochemistry of Cambrian boninites Planet. Sci. Lett., 28, 283-297, 1976. Contrib. Mineral. Petrol., 97, 361-377, and low-Ti andesites from Heathcote, Hawkins, J.W., Petrologic synthesis: Lau 1987. Victoria, Contrib. Mineral. Petrol., 91, 93- Basin transect (Leg 135), Proc. Ocean Brown, A.V., and G.A. Jenner, Geological 104, 1985. Drill. Program Sci. Results,135, 879-905, setting, petrology and chemistry of Crawford, A.J., and R.R. Keays, Petrogenesis 1994. Cambrian boninite and low-Ti tholeiite of Victorian, Cambrian tholeiites and Hawkins, J.W., and J.F. Allan, Petrogenetic lavas in western Tasmania, in Boninites, implications for the origin of associated evolution of the Lau Basin sites 834 edited by A.J. Crawford, pp. 232-263, boninites, J. Petrol., 28, 1075-1109, 1987. through 839, Proc. Ocean Drill. Program Unwin Hyman, Boston,Mass., 1989. Crawford, A.J., L. Beccaluva, and G. Serri, Sci. Results, 135, 427-470, 1994. Cameron, W.E., E.G. Nisbet, and V.J. Tectono-magmaticevolution of the West Hawkins, J.W., and C.A. Evans, Geology of Dietrich, Boninites, komatiites and Philippine-Marianaregion and the origin of the Zambales Range, Philippine islands: ophiolitic basalts, Nature, 280, 550-553, boninites, Earth Planet. Sci. Lett., 54, 346- ophiolite derived from island arc-back-arc 1979. 356, 1981. basin pair, in The Tectonic and Geologic Cameron, W.E., M.T. McCulloch, and D.A. Crawford, A.J., L. Beccaluva, G. Serri, and J. Evolution of the SoutheastAsian Seas and Walker, Boninite petrogenesis:Chemical Dostal, Petrology, geochemistry and Islands: Part 2, edited by D. E. Hayes, pp. and Nd-Sr isotopic constraints, Earth tectonic implications of volcanicsdredged 95-123, Geophys. Monogr. Set., vol. 27, Planet. Sci. Lett., 65, 75-89, 1983. from the intersection of the Yap and AGU, Wahington,D.C., 1983. Cameron, W.E., Petrology and origin of Mariana trenches, Earth Planet. Sci. Lett., Hawkins, J.W., and J.T. Melchior, Petrology primitive lavas from the Troodosophiolite, 80, 265-280, 1986. of Mariana Trough and Lau Basin basalts, Cyprus, Contrib. Mineral. Petrol., 89, 239- Crawford, A.J., T.J. Falloon, and D.H. Green, J. Geophys.Res., 90, 11,431-11,468, 1985. 255, 1985. Classification, petrogenesisand tectonic Hawkins, J.W., P.F. Lonsdale, J.D. Cameron,W.E., Contrastingboninite-tholeiite settingof Boninites,in Boninites,edited by MacDougall, and A.M. Volpe, Petrologyof associations from New Caledonia, in A. J. Crawford, pp. 1-49, Unwin Hyman, the Mariana Trough backarc spreading Boninites, edited by A.J. Crawford, pp. Boston Mass., 1989. centre, Earth Planet. Sci. Lett., 100, 226- 314-336, Unwin Hyman, Boston, Mass., Ewart, A., and C.J. Hawkesworth, The 250, 1990. 1989. Pleistocene-RecentTonga-Kermadec arc Hickey-Vargas, R., Boninites and tholeiites Campbell, H.J., Petrography and lavas: interpretation of new isotopic and from DSDP Site 458, Mariana forearc, in metamorphism of the T6remba Group rare earth data in terms of a depleted Boninites, edited by A.J. Crawford, pp. (Permian-Lower Triassic) and the Baie de mantle source model, J. Petrol., 28, 495- 339-356, Unwin Hyman, Boston, Mass., St.-Vincent Group (Upper Triassic-Lower 530, 1987. 1989. Jurassic),New Caledonia, J. R. Soc. N. Z., Ewart, A., R.N. Brothers, and A. Mateen, An Hickey-Vargas, R., and M.K. Reagan, 14, 335-349, 1984. outline of the geology and geochemistry, Temporal variation of isotope and rare Casey, J.F., and J.F. Dewey, Initiation of and the possiblepetrogenetic evolution of earth element abundances in volcanic subduction zones along transform and the volcanicrocks of the Tonga-Kermadec rocks from Guam, Contrib. Mineral. accretingplate boundaries,trippie-junction arc, J. Volcanol. Geotherm. Res., 2, 205- Petrol., 97, 497-508, 1987. evolution, and forearc spreading centres 250, 1977. Hochstaedter, H., J.B. Gill, M. Kusakabe, S. -Implications for ophiolitic geology and Ewart, A., W.B. Bryan, B.W. Chappell, and Newman, M. Pringle, B. Taylor, and P. obduction,in Ophiolites and the Oceanic R.L. Rudnick, Regional geochemistryof Fryer, Volcanism in the Sumisu Rift, I, Lithosphere, edited by I.G. Gass, S.J. the Lau-Tonga arc and backarc systems, Major element, volatile, and stableisotope Lippard, and A.W. Shelton, Geol. Soc. Proc. Ocean Drill. Program Sci. Results, geochemistry,Earth Planet. Sci. Lett., 100, Spec.Publ. London,13, 269-290, 1984. 135, 385-425, 1994. 179-194, 1990. Cluzel, D., J.C. Aitchison, G. Clarke, S. Falloon, T.J., and A.J. Crawford, The Humphris, S.E., and G. Thompson, Meffre, and C. Picard, Point de vue sur petrogenesisof high-calciumboninite lavas Hydrothermal metamorphism of oceanic l'6volution tectoniqueet g6odynamiquede dredged from the northern Tonga ridge, basaltsby seawater,Geochim. Cosmochim. la Nouvelle-Ca16donie, C. R. Acad. Sci., Earth Planet. Sci. Lett, 102, 375-394, 1991. Acta, 42, 107-125, 1978. Ser. 2, 319, 683-690, 1994. Falloon, T.J., D.H. Green, and A.J. Crawford, Jones, G., and A.H.F. Robertson, Tectono- Cluzel, D., J.C. Aitchison, G. Clarke, S. Dredged igneous rocks from the northern stratigraphicand evolutionof the Mesozoic 82 MEFFRE ET AL.: GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE

Pindos ophiolite and related units, study of magma genesisduring the initial Murton, and P. Browning,Geol. Soc. Spec. northwestern Greece, J. Geol. Soc. stages of subduction,Proc. Ocean Drill. Publ. London, 60, 171-178, 1989. London, 148, 267-288, 1991. Program Sci. Results,125, 623-659, 1992. Sobolev, A.V., and L.D. Danyushevsky, Kostopoulos,D. K., and B. J. Murton, Origin Pearce, J.A., M. Ernewein, S.H. Bloomer, Petrologyand geochemistryof boninites and distributionof componentsin boninite L.M. Parson, B.J. Murton, and L.E. from the north termination of the Tonga genesis: Significance of the OIB Johnson, Geochemistry of Lau Basin Trench: Constraints on the generation component,in Ophiolitesand their Modern volcanic rocks: influence of ridge condition of primary high-Ca boninite Oceanic Analogues, edited by L. M. segmentation and arc proximity, in magmas,J. Petrol.,35, 1183-1211,1994. Parson, B. J. Murton, and P. Browning, Volcanism Associated with Extension at Stakes, D.S., J.W. Shervais,and C.A. Hopson, Geol. Soc. Spec. Publ. London, 60, 133- ConsumingPlate Margins, edited by J. L. The volcanic-tectonic cycle of the 154, 1992. Smellie,Geol. Soc.Spec. Publ. London,20, FAMOUS and AMAR valleys, mid- Langmuir, C.H., E.M. Klein, and T. Plank, 53-76, 1994. Atlantic ridge (36ø47'N): Evidence from Petrological systematics of mid-ocean Pedersen,R.B., and J. Hertogen, Magmatic basaltglass and phenocrystcompositional ridge basalts: Constraints on melt evolution of the Karm0y Ophiolite variations for a steady state magma generation beneath mid-ocean ridges, in Complex, SW Norway: Relationships chamber beneath the valley midsections, Mantle Flow and Melt Generation at Mid- between IAT-MORB-boninitic-calc- AMAR 3, J. Geophys.Res., 89, 6995-7028, Ocean Ridges,edited by J. PhippsMorgan, alkaline and alkaline magmatism,Contrib. 1984. D. K. Blackman,and J. M. Sinton,pp. 183- Mineral. Petrol., 104, 277-293, 1990. Stern, R.J., and S.H. Bloomer, Subduction 280, Geophys. Monogr. Set., vol. 71, Prinzhofer, A., A. Nicolas, D. Cassard, J. zone infancy: Example from the Eocene Washington,D.C., 1992. Moute, M. Leblanc, J.-P. Paris, and M. Izu-Bonin-Mariana and Jurassic California Laurent, R., and R. H6bert, The volcanic and Rabinovitch, Structures in the New arcs, Geol. Soc. Am. Bull., 104, 1621-1636, intrusive rocks of the Qu6bec Appalachian Caledonianperidotite-gabbro: Implications 1992. ophiolites (Canada) and their island-arc for oceanic mantle and crust, Sun, S.S., and W.F. McDonough, Chemical setting,Chem. Geol., 77, 287-302, 1989. Tectonophysics,69, 85-112, 1980. and isotopic systematicsof ocean floor Maurizot, P., J.-P. Paris, D. Feignier, and T. Rothwell, R.G., P.P.E. Weaver, R.A. basalts: Implications for the mantle C., Articulation de la Chaine Centrale et le Hodkinson, C.E. Pratt, M.J. Styzen, and compositionand processes,in Magmatism domaine m6tamorphique du Nord N.C. Higgs, Clayey nanofossil ooze in the Ocean Basins, edited by A.D. Ca16donien, Gdol. Fr., 1, 45-52, 1985. turbidites and hemipelagitesat sites 834 Saunders,and M.J. Norry, Geol. Soc. Spec. Meijer, A., Primitive arc volcanism and a and 835 (Lau Basin, southwest Pacific), Publ. London, 42,313-345, 1989. boninitic series:examples from the western Proc. Ocean Drill. Program Sci. Results, Taylor, B., Rifting and the volcanic-tectonic Pacific, in The Tectonic and Geologic 135, 101-130, 1994. evolution of the Izu-Bonin-Mariana arc, Proc. Ocean Drill. Program Sci. Results, Evolutionof the SoutheastAsian Seasand Schilling, J.-G., M. Zajac, R. Evans, T. 126, 627-651, 1992. Islands,edited by D.E. Hayes,pp. 269-282, Johnston, W. White, J.D. Devine, and R. Geophys. Monogr. Set., vol. 23, AGU, Taylor,R.N., B.J. Murton,and R.W. Nesbit, Kingsley, Petrologic and geochemical Washington,D.C., 1980. Chemical transects across intra-oceanic variations along the mid-Atlantic ridge island arcs: implicationsfor the tectonic Meijer, A., E. Anthony, and M. Reagan, from 29øN to 73øN, Am. J. Sci., 283, 510- Petrologyof volcanicrocks from the fore- settingsof ophiolites,in Ophiolitesand 586, 1983. arc sites,Initial Rep. Deep Sea Drill. Proj., Their Modern Analogues,edited by L.M. Sharaskin, A.Y., I.K. Pustchin, S.K. Zlobin, 60, 709-729, 1982. Parson, B.J. Murton, and P. Browning, and G. M. Kolszov, Two ophiolite suites Monzier, M., L.V. Danyushevsky, A.J. Geol. Soc. 3•pec.Publ. London,60, 117- from the basementof the northern Tonga Crawford, H. Bellon, and J. Cotten, High- 132, 1992. arc, Ofiliti, 8, 411-430, 1983. Mg andesitesfrom the southerntermination Turcotte, D.L., and J. Phipps Morgan, The of the New Hebrides island arc (SW Shiraki, K., S. Kuroda, S. Maruyama, and H. physics of mantle migration and mantle Urano, Evolution of the Tertiary volcanic Pacific), J. Volcanol. Geotherm. Res., 57, flow beneatha mid-oceanridge, in Mantle 193-217, 1993. rocks in Izu-Mariana arc, Bull. Volcanl., Flow and Melt Generation at Mid-Ocean 41,548-562, 1978. Nicolas, A., Structures of Ophiolites and Ridges, editedby J. PhippsMorgan, D. K. Dynamicsof OceanicLithosphere, 367 pp., Sigurdsson, I.A., V.S. Kamenetsky, A.J. Blackman, and J. M. Sinton, pp. 155-182, Kluwer Acad., Norwell, Mass., 1989. Crawford, S.M. Eggins, and S.K. Zlobin, Geophys. Monogr. Set., vol. 71, AGU, Primitive island arc and oceanic lavas from Norfish, K., and J.T. Hutton, An accurate X- Washington,D.C., 1992. ray spectrographicmethod for the analysis the Hunter Ridge-Hunter Fracture Zone. Umino, S., Magma mixing in boninite of a wide range of geological samples, Evidence from glass, olivine and spinel sequenceof Chichijima, Bonin Islands,J. Geochim. Cosmochim. Acta, 33, 431-453, compositions, Mineral. Petrol., 47, 149- Volcanol. Geotherm. Res., 29, 125-157, 1969. 169, 1993. 1986. Paris, J.-P., La Gdologie de La Nouvelle- Sinton, J.M., and R.S. Detrick, Mid-ocean Walker, D.A., and W.E. Cameron, Boninite Calddonie, Un Essai de Synth•se, Mem. ridge magma chambers,J. Geophys.Res., primary magmas: Evidence from Cape B.R.G.M, 133, 250 pp., 1981. 97, 197-216, 1992. Vogel Peninsula,PNG, Contrib. Mineral. Pearce, J.A., S.J. Lippard, and S. Roberts, Sinton, J.M., and P. Fryer, Mariana Trough Petrol, 83, 150-158, 1983. Characteristicsand tectonicsignificance of lavas from 18øN: Implications for the Wood, D.A., N.G. Marsh, J.-L. Tarney, P. supra-subduction zone ophiolites, in origin of back arc basin basalts, J. Fryer, and M. Treuil, Geochemistry of Marginal Basin Geology, edited by B. P. Geophys.Res., 92, 12, 782-12, 802, 1987. igneous rocks recovered from a transect Kokelaar, and M. F. Howells, Geol. Soc. Smellie, J.L., and P. Stone, Geochemical across the Mariana Trough, arc, forearc, Spec.Publ. London,16, 77-94, 1984. controlson the evolutionaryhistory of the and trench, sites453 though461, Deep Sea Pearce, J.A., S.R. van der Laan, R.J. Arculus, Ballantrae Complex, SW Scotland, from Drilling Project Leg 60, Initial Rep. Deep B.J. Murton, T. Ishii, D.W. Peate, and I.J. comparison with recent analogues, in Sea Drill. Proj. 60, 709-730, 1982. Parkinson,Boninite and harzburgitefrom Ophiolites and Their Modern Oceanic Woodhead, J.D., The origin of geochemical leg 125 (Bonin-Marianasforearc): A case Analogues, edited by L.M. Parson, B.J. variationsin the Mariana lavas: A general MEFFREET AL.:GEOCHEMICAL EVOLUTION OF THE KOH OPHIOLITE 83

Tony.Crawford @ geol.utas.edu.au model for petrogenesisin intra-oceanic J.C. Aitchison, Department of Earth S. Meffre, Departmentof Geologyand island arcs, J. Petrol., 29, 805-830, 1988. Sciences,University of HongKong, Pokfulam Geophysics,University of Sydney,New South Zonenshain, L.P., and M.I. Kuzmin, The Road, Hong Kong. (e-mail: Wales, 2006 Australia. (e-mail: Khan-Taishir ophiolitic complex, origin [email protected]) [email protected]) and comparisonwith other ophiolitic A.J. Crawford, Department of Geology, complexes,Contrib. Mineral. Petrol., 67, Universityof Tasmania,GPO 252C, Hobart, (Received October 18, 1994; 95-109, 1978. Tasmania 7001, Australia. (e-mail: revisedJuly 24, 1995; acceptedJuly 28, 1995.)