Involvement of Continental Crust in the Formation of the Cretaceous Kerguelen Plateau: New Perspectives from ODP Leg 120 Sites

Transcription

1 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 PAGES Involvement of Continental Crust in the Formation of the Cretaceous Kerguelen Plateau: New Perspectives from ODP Leg 120 Sites FREDERICK A. FREY 1, DOMINIQUE WEIS 2, ANASTASSIA YU. BORISOVA 2,3 AND G. XU , DEPARTMENT OF EARTH, ATMOSPHERIC AND PLANETARY SCIENCES, MASSACHUSETTS INSTITUTE OF TECHNOLOGY, CAMBRIDGE, MA 02139, USA 2 DEPARTMENT OF EARTH AND ENVIRONMENTAL SCIENCES, CP 160/02, UNIVERSITÉ LIBRE DE BRUXELLES, AV. F. D. ROOSEVELT, 50, B-1050 BRUSSELS, BELGIUM 3 VERNADSKY INSTITUTE OF GEOCHEMISTRY AND ANALYTICAL CHEMISTRY, KOSYGIN ST. 19, , MOSCOW, RUSSIA RECEIVED MAY 7, 2001; REVISED TYPESCRIPT ACCEPTED DECEMBER 17, 2001 The basaltic basement of the large igneous province formed by the edge of the Southern Kerguelen Plateau and Leg 183 Site 1137 Kerguelen Plateau and Broken Ridge in the southeastern Indian on Elan Bank. The continental components in the Kerguelen Plateau Ocean has been sampled by three Ocean Drilling Program cruises basalts may have resided in a heterogeneous mantle plume that was (Legs 119, 120 and 183). Although the Cretaceous parts of this formed, in part, by deep recycling of crust. It is more likely, however, plateau formed in the embryonic Indian Ocean basin, presumably that slivers of Gondwana lithosphere reside within the lithosphere by melting associated with the Kerguelen plume, trace element and asthenosphere of the Indian Ocean mantle where they contaminate abundances and isotopic ratios of Sr, Nd and Pb of Cretaceous both plume-derived and mid-ocean ridge basaltic magmas. basalt from several drill sites indicate that continental lithosphere was involved in their petrogenesis. On the basis of relative depletions in Nb, Ta and Th, and isotopic characteristics similar to those of EMI ocean island basalt, lavas from Leg 120 Site 747 in the KEY WORDS: Ocean Drilling Program Leg 120; large igneous provinces; Central Kerguelen Plateau contain a component derived from lower Kerguelen Plateau; basalt geochemistry continental crust. On the basis of relative abundances of Sr and Eu and EMI-like Pb isotopic ratios, the source of basalt from Leg 120 Site 750 in the northeastern part of the Southern Kerguelen Plateau also contained a component derived from lower continental INTRODUCTION crust; in this case, the crustal component formed as a plagioclase- The submarine Kerguelen Plateau is a dominantly Crerich, clinopyroxene-bearing cumulate. Basalts from Leg 120 Site taceous large igneous province in the southern Indian 749 define two distinct isotopic (Sr, Nd and Pb) groups which Ocean that stands 2 4 km above the surrounding deep differ from the isotopic fields for Site 747 and 750 basalts. Among ocean basins. The plateau has been divided into distinct Site 749 lavas, there is subtle evidence for a continental component, domains: Southern Kerguelen Plateau (SKP), Central broadly similar (i.e. moderate 206 Pb/ 204 Pb >18 0) to that expressed Kerguelen Plateau (CKP), Northern Kerguelen Plateau more obviously in basalt from Leg 119, Site 738 on the southern (NKP), Elan Bank and Labuan Basin (Fig. 1). Together Corresponding author. Fax: fafrey@mit.edu Oxford University Press 2002

2 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 with the formerly contiguous Broken Ridge, this large contrast, the 206 Pb/ 204 Pb ratios of these Cretaceous dredge igneous province covers an area of km 2 (Coffin basalts are lower than those of Kerguelen Archipelago & Eldholm, 1994). It has been interpreted as an igneous lavas. Davies et al. (1989) found that these dredged basalts structure formed from the initial volcanism associated have higher Th/Ta and La/Nb than typical for ocean with the Kerguelen mantle plume (e.g. Morgan, 1971; island basalts (OIB), and Storey et al. (1989) proposed Duncan & Storey, 1992; Coffin et al., 2002). Abundant that these high ratios may indicate a component derived geophysical data are consistent with a dominantly basaltic from continental lithosphere. A compelling conclusion to gabbroic, >20 km-thick crust [see references given by could not be made from these few data for dredged Shipboard Scientific Party (2000)]. Volcanic features basalts, but subsequent drilling by ODP Leg 119 at Site attributed to the Kerguelen plume include the 738 in the SKP ( Fig. 1) recovered tholeiitic basalt that > Ma SKP, the > Ma CKP and Broken clearly contains a continental crust component; these Ridge (Coffin et al., 2002; Duncan, 2002), and the basalts have ε Nd (T) = 13 to 7, ( 87 Sr/ 86 Sr) T = Ma Ninetyeast Ridge, which is interpreted as the , very high 207 Pb/ 204 Pb relative to 206 Pb/ Kerguelen plume hotspot track (e.g. Weis et al., 1992). 204 Pb and relative depletion in Nb and Ta (Alibert, 1991; Cenozoic to recent volcanism attributed to the Kerguelen Mahoney et al., 1995). Most recently, Weis et al. (2001) plume includes part of the NKP, the Kerguelen Ar- and Ingle et al. (2002a) showed that tholeiitic basalts from chipelago and Heard and McDonald Islands (Fig. 1 and ODP Leg 183 Site 1137 on Elan Bank (Fig. 1) also Shipboard Scientific Party, 2000). It is likely that the contain a continental crust component. Moreover, at this Kerguelen plume is now beneath the Cretaceous Ker- site clasts of continental crust occur in a conglomerate guelen Plateau (Weis et al., 2002); hence, the spatially intercalated with the basalts (Frey et al., 2000a; Nicolaysen diffuse recent island volcanism ( Fig. 1) may reflect the et al., 2001; Ingle et al., 2002b). Although they are sepdifficulty of establishing magma ascent paths through the arated from the surrounding continental margins by deep thick plateau lithosphere (Frey et al., 2000b; Weis et al., ocean basins, basalts at Sites 738 and 1137 erupted on 2002). the periphery of the Kerguelen Plateau closest to the Despite formation within the Indian Ocean basin, there rifted continental margins ( Fig. 1). has been a long-term discussion (e.g. Dietz & Holden, The objective of this paper is to evaluate the role of 1970) regarding the role of continental lithosphere in continental crust during basalt petrogenesis at other drill forming the Kerguelen Plateau. In some areas, such as sites within the CKP and SKP; specifically ODP Leg the SKP and Elan Bank, a western salient of the plateau, 120, Site 747 in the SE CKP, Site 749 in the interior of geophysical data provide evidence for continental fragthe northern part of the SKP, and Site 750 at the NE ments within the crust of the plateau (Operto & Charvis, edge of the SKP (Fig. 1). Tholeiitic basalt was recovered 1996; Charvis et al., 1997). Moreover, at two locations, at these sites (Storey et al., 1992). Alkalic basalt was ODP Site 738 on the SKP and ODP Site 1137 on Elan Bank, tholeiitic basalts contain a component derived from recovered at the base of Hole 748C, but this basalt is not continental crust (Mahoney et al., 1995; Weis et al., 2001). interpreted as part of the igneous basement (Shipboard Determining the role of continental lithosphere in the Scientific Party, 1989a). Isotopic (Sr, Nd, Pb and Hf ) petrogenesis of Kerguelen Plateau basalt as a function data for the tholeiitic basalt from Sites 747, 749 and 750 of eruption age and location is essential to understanding showed that each drill site is isotopically distinct, but the breakup of eastern Gondwana and subsequent evolubeing consistent with an oceanic origin (Salters et al., these isotopic characteristics were initially interpreted as tion of the Indian Ocean basin. Dredges from the Labuan Basin, 77 E Graben, and 1992). Storey et al. (1992), however, noted that basalts Skiff Bank ( Fig. 1) have recovered metamorphic and from Sites 747 and 750 are offset to low 206 Pb/ 204 Pb from granitic rocks, but these have been interpreted as icehave a relative Nb depletion. They suggested that the the field of Kerguelen Archipelago lavas and that some rafted debris (Ramsay et al., 1986; Bassias et al., 1987; Montigny et al., 1993; Weis et al., 2002). The first intaminated by components derived from the Gondwana source of these Kerguelen Plateau basalts was condisputable samples of the Cretaceous igneous crust formcontinental lithosphere. These data are not sufficient for ing the Kerguelen Plateau were dredged from the 77 E Graben in the SKP (Bassias et al., 1987; Leclaire et al., a rigorous assessment of the role for a continental crustal 1987; Davies et al., 1989) (Fig. 1). These dredge samples component in these lavas because there are too few are dominantly altered, tholeiitic basalt. Leclaire et al. isotopic and trace element abundance data for the same (1987) reported a K/Ar age of 114 Ma for plagioclase samples. We obtained a comprehensive dataset for major separated from a basalt. Isotopic ratios of Sr and Nd in and trace element abundances and isotopic ratios for the basalts from the 77 E Graben largely overlap with basalt from Sites 747, 749 and 750 to use these data to the broad field defined by lavas from the Cenozoic evaluate the role of a continental crustal component in Kerguelen Archipelago (Weis et al., 1989, 1993). In the interior parts of the Kerguelen Plateau. 1208

3 FREY et al. FORMATION OF CRETACEOUS KERGUELEN PLATEAU Fig. 1. Bathymetry (contour interval 500 m) of the Kerguelen Plateau showing locations of major features. ODP Leg 119, 120 and 183 drill sites that recovered igneous basement are indicated by filled stars. ODP Sites indicated by open stars bottomed in sediment. This paper focuses on basalt from Site 747 on the Central Kerguelen Plateau (CKP) and Sites 749 and 750 on the northern part of Southern Kerguelen Plateau (SKP). There is unambiguous evidence for a component derived from continental crust in basalt from Site 738 at S (Mahoney et al., 1995) and Site 1137 at S (Weis et al., 2001; Ingle et al., 2002a). In contrast, at Sites 1136, 1138, 1139 and 1140 (Neal et al., 2002; Kieffer et al., 2002; Weis & Frey, 2002), evidence for a continental crust component in the basement basalt is absent or very subtle. SAMPLING AND PETROGRAPHY Site 747 This site is in the Central Kerguelen Plateau (Fig. 1) where 53 9 m of basement basalt was penetrated (20 5 m recovered) below m of sediment. Although the basalt sediment contact was not recovered, the oldest sediment is early Santonian in age (>87 Ma) (Shipboard Scientific Party, 1989b). The basalts are highly altered and difficult to date using K Ar or 40 Ar/ 39 Ar techniques (Whitechurch et al., 1992). An 40 Ar/ 39 Ar age of 85 Ma was reported by Pringle et al. (1994), but this age is now considered unreliable (Duncan, 2002). The basement core is interpreted to consist of up to 12 subaerially 1209

4 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 erupted basaltic lava flows separated by zones of basaltic different lava flows; Units 3 and 4 may belong to a single breccia (Shipboard Scientific Party, 1989b). flow. Site 747 lavas range from aphyric to moderately phyric Site 750 lavas are moderately phyric basalt ( Table 1). (<10% phenocrysts) to lavas with 10 15% phenocrysts The Unit 3 lava is highly altered with clays replacing forming two assemblages: olivine plagioclase and plagioclase and glass and filling cavities. As discussed olivine plagioclase clinopyroxene ( Table 1). The vechanged below, its chemical composition has been significantly sicularity of the basalts, up to 10%, indicates subaerial by post-magmatic processes (e.g. K 2 O/Na 2 O = or shallow-water eruption. To variable extents all samples 6 4, Table 2). are altered and contain veins. Vesicles are filled with Units 5 and 6 are massive, moderately plagioclase- clay minerals and veins are filled with zeolites. Olivine and clinopyroxene-phyric basalts. In Unit 5, 11 5 m thick, phenocrysts are euhedral (0 7 2 mm) but replaced by the plagioclase phenocrysts (An and mm in clay minerals, serpentine and iddingsite. Plagioclase length) are euhedral tabular crystals forming a variolitic phenocrysts (An ) occur as unaltered euhedral tabular texture. Unit 5 lavas have two different textures: mediumcrystals (0 5 2 mm) and as pseudomorphs replaced by to coarse-grained, centimeter-scale subophitic bands or zeolite and smectite. Unaltered augite phenocrysts clots separated by fine-grained basalt (see Shipboard (0 5 2 mm) are subhedral to anhedral. The texture is Scientific Party, 1989d, fig. 34). The bands are phenocryst microlitic to subophitic with the matrix formed of plagio- rich and show grading in size distribution. In the subclase laths, augite and interstitial Fe Ti oxides. ophitic bands euhedral clinopyroxene (up to 2 mm) poikilitically encloses plagioclase. These bands have been interpreted as dislodged cumulates, or mechanical segregations produced during flow or as products of magma Site 749 mixing (Shipboard Scientific Party, 1989d). In Unit 6 This site is in the northwestern part of the Southern lavas, tabular plagioclase crystals (0 7 1 mm in length) Kerguelen Plateau (Fig. 1) where 47 5 m of basement are associated with subhedral clinopyroxene phenocrysts. basalt was penetrated (23 3 m recovered) below 202 m Unit 6 lavas are more altered than Unit 5 lavas; secondary of sediments. The age of the oldest sediment is Ma, minerals include clay minerals, zeolites and calcite filling but the sediment basalt contact was not recovered (Shipvesicles and veins. board Scientific Party, 1989c). K/Ar and 40 Ar/ 39 Ar ages of >110 Ma have been determined for the basalt (Whitechurch et al., 1992; Pringle et al., 1994; Storey et al., 1996). The recovered basaltic basement consists of five subaerial lava flows intruded by a dike. SAMPLE PREPARATION AND Site 749 lavas (Table 1) range from aphyric (Units 1, ANALYTICAL PROCEDURES 3 and 4) to moderately plagioclase-phyric (5 10% of To analyze the least altered samples, small rock chips 1 2 mm phenocrysts in Unit 5) to highly plagioclasefrom the interior of the core were selected to avoid large phyric (20 30% of An 62 to An 84 laths ranging from 0 5 filled vesicles, veins and regions of intense alteration. to 20 mm in Unit 6). These lavas are vesicular and These chips were powdered in an agate shatterbox. As variably altered; some plagioclase phenocrysts are rein our previous studies of lavas related to the Kerguelen placed by clay minerals. The microlitic matrix consists plume (e.g. Frey et al., 1991, 2000b) abundances of major of plagioclase (An ), augite and titanomagnetite. oxides and some trace elements were determined by X- ray fluorescence (XRF) spectrometry ( Table 2) at the University of Massachusetts (Rhodes, 1996). Other trace Site 750 elements were determined by instrumental neutron activation This site is in the northeastern part of the Southern analysis (INAA) ( Table 2) at the Massachusetts Kerguelen Plateau (Fig. 1) where 34 m of basement basalt Institute of Technology ( MIT) (Ila & Frey, 1984, 2000). was penetrated (22 9 m recovered) below m of In addition, we report trace element analyses determined sediment. The oldest sediments are Albian terrigenous at MIT by inductively coupled plasma mass spectrometry deposits containing abundant fossil wood and other landwith (ICP-MS) (Table 3) using a Fisons VG Plasmaquad 2+S derived organic matter (Shipboard Scientific Party, both internal and external drift monitors. In general, 1989d; Francis & Coffin, 1992; Mohr & Gee, 1992). compared with data determined by XRF and INAA, the Whitechurch et al. (1992) reported ages of 101 Ma (K Ar) ICP-MS data agree within ±10% (La, Ce, Nd, Sm, Eu, and 118 Ma ( 40 Ar/ 39 Ar) for a Site 750 basalt; Coffin et Rb, Sr, Ba, Nb, Ta, Zr and Hf ); however, at Th <0 5 al. (2002) reported a ± 0 4 Ma age ( 40 Ar/ 39 Ar) for ppm and Rb <1 ppm, the discrepancies are higher plagioclase from a Site 750 basalt. The basement basalts because the INAA (Th) and XRF (Rb) data are not precise are divided into five units consisting of at least four at these low abundances. Abundances of Y (>15%) and 1210

5 FREY et al. FORMATION OF CRETACEOUS KERGUELEN PLATEAU Table 1: Petrographic description of Leg 120 basalts Sample number Rock Phenocryst abundances, Matrix texture Vesicles and veins Specific features of (unit number) texture shape and size secondary alteration 747C11R-1, 0 5 cm (1) Sparsely <5%; Ol, Cpx, PL subhedral to Glassy Vesicles 1 2 mm in size (<5%) Ol and matrix replaced by phyric euhedral (1 2 mm) filled with calcite iddingsite, clay minerals (celadonite and smectites) 747C12R-1, cm (8) Sparsely <5%; Ol and Pl euhedral (2 mm); Glassy Vesicles (5%) filled with calcite Ol, Pl and matrix are phyric Cpx subhedral (1 mm) completely altered, Cpx is partially altered 747C12R-2, cm (10) Moderately 5 10%; Pl euhedral laths Microlitic Vesicles mm in size Fresh Pl Cpx assemblages; phyric (1 mm); Cpx subhedral (<5%) filled with brown clay Pl is replaced by reddish (0 7 mm), Cpx Pl assemblage minerals brown clay minerals 747C12R-4, cm (10) Moderately 5 10%; Ol Pl euhedral assemblages Microlitic Vesicles (<5%) filled with clay Ol Pl assemblage is altered phyric (1 2 mm), Cpx Pl subhedral minerals microphenocrysts 747C13R-1, cm (10) Moderately 10%; Ol Pl euhedral assemblages Microlitic, Vesicles (<5%) filled with clay Ol and glass is replaced by phyric (<2 mm), Cpx Pl subhedral (1 mm) Pl Cpx microlites, minerals reddish brown clay minerals and intersertal Fe Ti oxides 747C13R-3, cm (17) Aphyric Microlitic Highly altered 747C14R-1, cm (19) Sparsely <5%; Pl, Ol euhedral crystals Microlitic, Vesicles (5%) Highly altered phyric (1 2 mm) Pl Cpx microlites, and intersertal Fe Ti oxides 1211

6 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 Table 1: continued Sample number Rock Phenocryst abundances, Matrix texture Vesicles and veins Specific features of (unit number) texture shape and size secondary alteration 747C14R-2, cm (21) Sparsely <5%; Pl Cpx euhedral Microlitic, Pl Cpx Vesicles (<5%) filled with clay Matrix is replaced with brown phyric microphenocrysts (<1 mm) microlites, and minerals clay mineral intersertal Fe Ti oxides 747C15R-1, cm (23) Highly 10 15%; Pl euhedral laths; Pl Cpx assemblage Vesicles (10%) filled with clay Ol replaced by iddingsite and phyric Ol euhedral (1 1 5 mm) and intersertal minerals, veins filled with serpentine, Pl is altered Fe Ti oxides zeolites 747C15R-2, cm Moderately 10%; Pl euhedral laths; Subophitic Ol is replaced by green (25) phyric Ol euhedral (1 5 2 mm) mineral, and Pl is altered 747C15R-3, cm (27) Sparsely <5%; Ol euhedral (1 1 5 mm); Subophitic Vesicles filled with clay Ol is replaced by serpentine, phyric Pl Cpx subhedral (1 mm) minerals, zeolites Cpx is altered 747C16R-2, cm (29) Moderately 10%; Ol Pl euhedral (1 5 2 mm); Subophitic Vesicles (10%) filled with clay Ol is replaced by clay minerals; phyric Cpx Pl subhedral (<0 5 mm) minerals, zeolites Pl is partially replaced 747C16R-5, cm (30) Sparsely <5%; Ol euhedral (1 1 5 mm); Subophitic Pl Cpx assemblage is fresh; phyric Pl Cpx subhedral (1 mm) reddish brown clay mineral 1212

7 FREY et al. FORMATION OF CRETACEOUS KERGUELEN PLATEAU Sample number Rock Phenocryst abundances, Matrix texture Vesicles and veins Specific features of (unit number) texture shape and size secondary alteration 749C12R3, cm (1) Aphyric Microlitic, Vesicles and veins filled with 749C12R5, cm (3) Aphyric Pl Cpx microlites, up to 10% calcite, 749C15R4, cm (4) Aphyric Fe Ti oxides and green clay minerals 749C16R1, cm (5) Moderately 5 10%; Pl euhedral laths (1 2 mm) Microlitic, Pl Cpx Vesicles filled with calcite Pl is partially altered by phyric microlites, Fe Ti brown clay minerals oxides 749C16R5, cm (6) Highly 20 30%; Pl euhedral laths Microlitic, Pl Cpx Vesicles are filled with clay Pl is relatively unaltered; 749C16R6, cm (6) phyric ( mm) microlites, Fe Ti minerals brown and green clay oxides minerals 750B14R1, cm (3) Moderately 3 5%; Pl euhedral laths (1 mm) Microlitic, Pl Cpx Vesicles filled with clay Highly altered, brown and phyric microlites, Fe Ti minerals, calcite (10%) green clay minerals oxides 750B14R2, cm (4) Moderately 7 12%, 10% Pl; Pl euhedral laths Microlitic, Pl Cpx Vesicles, veins filled with clay Green and brown clay phyric (0 5 2 mm); Pl Cpx subhedral microlites, Fe Ti minerals minerals phenocrysts (1 5 2 mm) oxides 750B15R1, cm (5) Moderately 5 10%, 7 5% Pl; Pl euhedral laths Heterogeneous Calcite, chlorite (?), brown clay Moderately altered, 750B16R2, cm (5) phyric (0 7 1 mm); Pl Cpx assemblage from microlitic to mineral glass is replaced by green (1 mm) subophitic clay minerals Pl Cpx microlites, Fe Ti oxides 750B15R3, cm (5) Moderately 5 10%, 7 5% Pl; Pl euhedral laths From microlitic to Vesicles are filled with clay Brown clay minerals, 750B15R7, cm (5) phyric (<1 mm); Pl Cpx assemblage subophitic Pl Cpx minerals plagioclase is replaced by 750B16R7, cm (5) (1 5 2 mm) microlites, Fe Ti zeolites oxides 750B17R2, cm (6) Moderately 5 7%, 6% Pl, Pl Cpx assemblage Microlitic, Pl Cpx Vesicles, veins filled with clay Highly altered, brown, green 750B17R3, cm (6) phyric (<2 mm) microlites, Fe Ti minerals, zeolites clay minerals, zeolites oxides Number within parentheses indicates unit number from Shipboard Scientific Party (1989b, 1989c, 1989d) and asterisk indicates samples chosen for Sr Nd Pb isotopic analyses. Pl, plagioclase; Cpx, clinopyroxene; Ol, olivine; Fe Ti oxides, magnetite or ilmenite. 1213

8 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 Table 2: Major element (wt % oxide) and trace element (ppm) abundances in lavas from Sites 747, 749 and 750 Site 747 Unit: Sample: 11R1 12R1 12R2 12R4 13R1 13R3 14R1 14R2 15R1 15R2 15R3 16R2 16R SiO TiO Al 2 O Fe 2 O MnO MgO CaO Na 2 O K 2 O P 2 O Total LOI Sc Cr Co La Ce Nd Sm Eu Tb Yb Lu Hf Ta Th Rb Sc Ba V Ni Zn Ga Y Zr Nb

9 FREY et al. FORMATION OF CRETACEOUS KERGUELEN PLATEAU Site 749 Unit: Sample: 12R2 12R3 12R5 15R4 16R1 16R5 16R SiO TiO Al 2 O Fe 2 O MnO MgO CaO Na 2 O K 2 O P 2 O Total LOI Sc Cr Co La Ce Nd Sm Eu Tb Yb Lu Hf Ta Th Rb Sr Ba V Ni Zn Ga Y Zr Nb

10 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 Table 2: continued Site 750 Unit: Sample: 14R1 14R2 15R1 15R3 15R7 16R2 16R4 16R7 17R2 17R SiO TiO Al 2 O Fe 2 O MnO MgO CaO Na 2 O K 2 O P 2 O Total LOI Sc Cr Co La Ce Nd Sm Eu Tb Yb Lu Hf Ta Th Rb Sr Ba V Ni Zn Ga Y Zr Nb Abundances of major elements (all iron as reported as Fe 2 O 3 ) and the trace elements Rb, Sr, Ba, V, Ni, Zn, Ga, Y, Zr and Nb determined by X-ray fluorescence at the University of Massachusetts following the procedures of Rhodes (1996). Only a subset of samples were analyzed for TE because of difficulty in making coherent disks for some of the clay-rich samples. Other trace element data determined by instrumental neutron activation analysis at MIT following the procedures of Ila & Frey (1984, 2000). LOI (loss on ignition) indicates weight loss after heating to 1020 C for 30 min. Although 30 units were identified in the Site 747 basement core, these units include breccia zones associated with up to 12 lava flows (Shipboard Scientific Party, 1989b). Most of the Site 750 samples are from Unit 5, the thickest (11 5m) flow unit at this site. 1216

11 FREY et al. FORMATION OF CRETACEOUS KERGUELEN PLATEAU Yb and Lu (8%) determined by ICP-MS are higher than to a lesser extent Si are mobile during post-magmatic those determined by XRF and INAA, respectively. alteration, the tholeiitic nature of basalts from Sites 747, Samples for radiogenic isotope analyses (Table 4) were 749 and 750 in a Na 2 O + K 2 O vs SiO 2 classification selected on the basis of their relative freshness and to plot ( Fig. 2) is consistent with the relative abundances of encompass the largest range of chemical composition. immobile incompatible elements. For example, at a given The chemical procedure used is similar to that described MgO content, lavas from these sites vary systematically by Weis et al. (1987) with improvements discussed by in TiO 2 content, with TiO 2 increasing in the order 750 Weis & Frey (1991). To remove secondary or alteration to 747 and 749, to 748 (Fig. 3). This sequence is the phases all samples were acid-leached repeatedly with same as that for increasing alkalis at a given SiO 2 content warm 6N HCl until a clear solution remained. Total (Fig. 2). A dramatic example of alkali mobility is shown blank values were <1 ng for all three isotopic systems by the two analyses from Unit 3 at Site 750. These considered. Such values are negligible in view of the samples have a weight loss on ignition of 7 and 10%, elemental concentrations in the samples (Table 3). Sr high K 2 O( %) and atypically low Na 2 O( %) and Nd isotopic ratios were measured on single Ta contents; in addition, they have relatively low SiO 2, filaments and triple Re Ta filaments, respectively, in the high total iron, and they have lost most of their CaO dynamic mode, on a VG Sector 54 multicollector mass ( %) (Table 2, Fig. 2). In general, there is a trend spectrometer with an internal precision better than 1 for SiO 2 /Al 2 O 3 to decrease with increasing LOI. In 10 5 unless specified in Table 4. Sr isotopic ratios were contrast, TiO 2 was apparently relatively immobile, as the normalized to 86 Sr/ 88 Sr = and Nd isotopic ratios TiO 2 contents of Unit 3 lavas are similar to those of (147, 146, 145, 144, 143 and 142), measured as metal other Site 750 basalt (Table 2). Also distinctive is the to 146 Nd/ 144 Nd = The average 87 Sr/ 86 Sr value plagioclase-rich basalt from Unit 6 at Site 749 (Table of the NBS 987 Sr standard was ± 14 (2σ on 1). These lavas have high Al 2 O 3 (>20%) and CaO 62 samples) and analyses of Rennes Nd standard (Chauvel ( %) coupled with low total iron and TiO 2 & Blichert-Toft, 2001) yielded 143 Nd/ 144 Nd = abundances (Table 2, Fig. 3). These compositional fea- ± 12 (2σ on 61 samples). The precision for Sr and Nd tures, which reflect the abundance of plagioclase (20 isotopic ratios can be evaluated by comparing duplicate 30%, Table 1), indicate that this basalt contains cumulate analyses, which agree within the 2σ uncertainties for six plagioclase. of seven samples ( Table 4). Pb isotopic ratios were measured on single Re filaments using the H 3 PO 4 silica gel technique. All Pb isotopic ratios were corrected for mass fractionation (0 12 ± 0 024% per a.m.u.) on the Trace element abundances basis of 72 analyses of the NBS 981 Pb standard for a Abundances of the highly incompatible elements Th, La, temperature range between 1090 C and 1150 C. Ce, Nb and Ta in Leg 120 lavas from Sites 747, 749 Between-run precisions were better than >0 1% for and 750 are strongly correlated; e.g. in trace element vs 206 Pb/ 204 Pb and 207 Pb/ 204 Pb and better than >0 15% for Th plots the correlation coefficients exceed 0 96 (Fig. 4). 208 Pb/ 204 Pb. Abundances of U, Ba, Pb, Zr and Hf are also positively correlated with Th content, although there are a few aberrant points, which probably reflect the effects of postmagmatic GEOCHEMICAL RESULTS processes [e.g. the anomalously high Ba ( Fig. 4) and K (Table 2) in the clay-rich sample R1 Major element compositions from Unit 3]. Abundances of the mobile alkali metals K It is apparent from their petrography, weight loss on and Rb are not correlated with Th content; K/Rb ranges ignition (LOI; typically 1 5% but some up to 10%, Table from 425 to This wide range reflects the effects of 2), and weight loss after the acid leaching before isotopic post-magmatic processes; hence we do not use K and Rb analyses, that the samples have been strongly affected for petrogenetic interpretations. Abundance variations of by post-magmatic alteration. In addition, zeolite-facies Yb and Sr with Th are also poorly correlated. As discussed minerals are abundant in amygdules (Sevigny et al., 1992). below, the scatter in Sr reflects the well-known com- Despite the mineral and compositional changes caused patibility of Sr in feldspar; whereas lavas from each site by alteration and metamorphism, these lavas are un- define different Yb Th trends. doubtedly tholeiitic basalt. Firstly, their petrographic Primitive mantle-normalized plots show several imcharacteristics and inferred phenocryst crystallization portant features of the relative abundances of in- sequence, olivine and plagioclase followed by clino- compatible elements in these Kerguelen Plateau basalts pyroxene, are distinct from the alkalic analcite-bearing (Fig. 5). Most of the lavas have (Ba/Rb) PM much greater basalt overlying basement at Site 748 (Shipboard Scientific than unity (PM denotes normalized to primitive mantle Party, 1989a). Secondly, although Na, K and ratio of 11) and (Ba/Rb) PM varies widely ( ). As 1217

12 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 Table 3: Trace element abundances (ppm) in lavas from Sites 747, 749 and 750 (data obtained by ICP-MS at MIT) Site 747 Site 749 Unit: Sample: 11R1 12R1 12R2 12R4 13R1 13R3 14R1 14R2 15R1 15R2 15R3 16R2 16R5 12R2 12R3 12R5 15R4 16R1 16R5 16R Rb Sr Ba Y Zr Nb Hf Ta La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Pb Th U

13 FREY et al. FORMATION OF CRETACEOUS KERGUELEN PLATEAU Table 3: continued Site 750 BHVO-2 Unit: Sample: 14R1 14R2 15R1 15R3 15R7 16R2 16R4 16R7 17R2 17R Rb Sr Ba Y Zr Nb Hf Ta La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Pb Th U Units are in ppm. The concentrations for BHVO-2 are the average values of 11 analyses, with a 1σ standard deviation of Ζ3%. with K/Rb, we infer that such anomalous ratios reflect (Fig. 5b). Among Leg 120 lavas, those from Site 750 the effects of subaerial and submarine alteration processes have the lowest abundances of incompatible elements, that have affected the Rb contents of these basalts. For and they also have significant relative enrichments in Sr Site 747 lavas, the slope in Fig. 5a is negative from La and Eu and a depletion in Ti, Zr and Hf ( Fig. 5c). to Yb, i.e. relative enrichment in the more incompatible Surprisingly, these lavas do not contain a large abundance elements. However, the highly incompatible elements of plagioclase phenocrysts (e.g. lavas from Units 5 and 6 Th, Nb and Ta are relatively depleted compared with at Site 750 typically contain 6 7 5% plagioclase, Table the adjacent elements Ba and La; sample 14R2 has a 1, Fig. 6b). Relative to the other flow units at Site 750, particularly high La/Nb ( Fig. 5a). In addition, Sr is Unit 6 basalts have higher abundances of rare earth relatively depleted. Compared with Site 747 basalts, Site elements (REE) and several other incompatible elements 749 basalts define a shallower negative slope in a primitive ( Fig. 5c). mantle-normalized plot ( Fig. 5b). However, the two In summary, the relative abundances of Sr and Eu plagioclase-rich samples from Unit 6 have very marked provide constraints on the petrogenesis of several basaltic enrichments in Sr and Eu; also the aphyric samples from units from each site. Except for the highly altered Site Unit 1 (Table 1) have relative enrichments in Sr and Eu 750 sample 14R1, the relative enrichments of Sr and Eu 1219

14 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 Table 4: Sr, Nd and Pb isotopic ratios and parent/daughter abundance ratios of lavas from ODP Leg 120, Sites 747, 749 and Rb/ ( 87 Sr/ 2σ ( 87 Sr/ 147 Sm/ ( 143 Nd/ 2σ ( 143 Nd/ 238 U/ 235 U/ 232 Th/ ( 206 Pb/ ( 206 Pb/ ( 207 Pb/ ( 207 Pb/ ( 208 Pb/ ( 208 Pb/ 86 Sr 86 Sr)m 86 Sr)i 144 Nd)m 144 Nd)m 144 Nd)i 204 Pb 204 Pb 204 Pb 204 Pb)m 204 Pb)l 204 Pb)m 204 Pb)i 204 Pb)m 204 Pb)i Site R1, R4, R3, R1, R2, R1, R2, R2, R5, Site R2, (1) R3, (1) R5, (3) R4, (4) R1, (5) R5, (6) R6, (6) Site R2, (4) R1, (5) R2, (5) R4, (5) R7, (5) R3, (6) For Sites 749 and 750 the number within parentheses in the sample number indicates the Unit number. The 2σ uncertainties for ( 87 Sr/ 86 Sr) m and ( 143 Nd/ 144 Nd) m (m indicates measured) indicate variation in last significant digits. Initial ratios (subscript i) are calculated for 85 Ma (Site 747), 113 Ma (Site 749) and 111 Ma (Site 750). Parent/daughter abundance ratios are calculated from data in Table 3. All isotopic data were obtained on acid-leached samples but samples were not acid-leached before elemental analysis. Full-procedural duplicate, including leaching. 1220

15 FREY et al. FORMATION OF CRETACEOUS KERGUELEN PLATEAU Fig. 2. Total alkalis (Na 2 O + K 2 O) vs SiO 2 (all in wt %) basalt classification plot with Fe 2+ as 85% of total iron. Alkalic tholeiitic dividing line from Macdonald & Katsura (1964). Data are for basalts from Site 747 (circles), 748 (plus), 749 (squares) and 750 (triangles); filled symbols are data from this paper and open symbols are data from Storey et al. (1992). The effects of post-magmatic alteration are typically to increase total alkalis and decrease SiO 2 content; the extreme examples are two analyses of Unit 3 from Site 750, which have weight loss on ignition of 7 10 wt %; that is, they are highly altered tholeiitic basalts that plot in the alkalic field. All other samples from Site 750 are clearly tholeiitic basalt. The alkalic nature of the analcite-bearing basalts from Site 748 is a primary feature. Basalt from Sites 747 and 749 is generally within the tholeiitic field but some points straddle the dividing line. are highly correlated ( Fig. 6a). In lavas from Site 747, for nine Site 747 lavas form two clusters, with four (Sr/Nd) PM decreases from >1 to 0 5 with increasing samples from cores 15 and 16 at the bottom of the hole abundances of incompatible elements, such as La ( Fig. having slightly higher Pb isotopic ratios than the overlying 6c). This correlation is expected from plagioclase frac- basalt ( Fig. 8). tionation; that is, La and Nd are incompatible and Sr is In general, the isotopic data of Salters et al. (1992) compatible in plagioclase. In contrast, for the Site 750 agree with our new data. An important exception are lavas (Sr/Nd) PM increases from 1 to 2 5 with only a slight Pb isotopic data reported by Salters et al. (1992) for decrease in La content ( Fig. 6c). This correlation is three samples powdered on the ship (samples Rconsistent with accumulation of plagioclase. 2, cm; R-2, cm; and R3, cm). These samples yielded aberrant highly radiogenic Pb ratios, which we did not find in our Pb Isotopic ratios of Sr, Nd and Pb isotopic analyses of 21 acid-leached samples powdered Site 747 in agate (see insets in Fig. 8). We infer that these three aberrant samples were contaminated with Pb during Isotopic data are available for 14 samples [nine in Table 4 shipboard sample preparation; therefore we exclude the and five from Salters et al. (1992; personal communication, Salters et al. (1992) Pb isotopic data for these three 1992)]. The initial Sr and Nd isotopic ratios form a samples from our subsequent discussion. tightly grouped cluster that overlaps in 87 Sr/ 86 Sr but is offset to lower 143 Nd/ 144 Nd than the <10 Ma lavas in the Kerguelen Archipelago ( Fig. 7a). The Pb isotopic ratios of Site 749 Site 747 lavas differ substantially from those of Kerguelen Isotopic data are available for 12 samples [seven in Table Archipelago lavas (Fig. 8). In detail, the Pb isotopic data 4 and five from Salters et al. (1992)]. There is general 1221

16 JOURNAL OF PETROLOGY VOLUME 43 NUMBER 7 JULY 2002 Fig. 3. TiO 2 vs MgO (wt %) for basalt from Sites 747, 749 and 750. Symbols are defined in Fig. 2 caption. At a given MgO content, TiO 2 abundances increase in the order 750 to 749 and 747, to 748, and in general this is an order of increasing abundances of incompatible elements (Tables 2 and 3). The plagioclase-rich lavas from Unit 6 at Site 749 (Table 1) are an exception in that their low TiO 2 contents are within the field for Site 750 lavas. agreement between the two datasets for Sr and Nd. The agreement between the two datasets, with Unit 6 having measured Sr and Nd data for Site 749 basalts are at the distinctly lower 143 Nd/ 144 Nd and higher 87 Sr/ 86 Sr than depleted end (i.e. relatively high 147 Nd/ 144 Nd and low the other units (Fig. 7a). Unit 6 lavas are also relatively 87 Sr/ 86 Sr) of the array defined by lavas attributed to the enriched in incompatible elements (Fig. 5). A surprising Kerguelen plume; the age-corrected ratios, however, are result is that basalts from Sites 749 and 750 have similar displaced to lower 143 Nd/ 144 Nd at a given 87 Sr/ 86 Sr (Fig. 143 Nd/ 144 Nd, but all Site 750 basalts have distinctly higher 7a). In detail, Sr and Nd isotopic data for Site 749 lavas 87 Sr/ 86 Sr (Fig. 7). Although 87 Sr/ 86 Sr age corrections are define two groups, with basalts from Units 3 and 4 having uncertain because of Rb mobility, the sample from Unit higher 87 Sr/ 86 Sr and lower 143 Nd/ 144 Nd than basalts from 6 at Site 750 with normal Ba/Rb (11 2, Fig. 5c) is also Units 1, 5 and 6 (Table 4 and Fig. 7a). These two groups offset to high 87 Sr/ 86 Sr (Fig. 7a). The Unit 6 sample that do not have different Pb isotopic ratios (Table 4 and is distinctive in Sr and Nd has slightly higher 208 Pb/ 204 Pb Fig. 8). Compared with Kerguelen Archipelago lavas, and 207 Pb/ 204 Pb than other Site 750 basalts (Table 4). Site 749 basalts have 206 Pb/ 204 Pb and 207 Pb/ 204 Pb most Despite their differences in Sr and Nd isotopic ratios similar to the field for <10 Ma lavas from the Kerguelen (Fig. 7a), the Pb isotopic fields for Site 750 and 747 Archipelago (Fig. 8), but they are not similar to these basalts overlap, with both sites having low 206 Pb/ 204 Pb Kerguelen Archipelago lavas in 208 Pb/ 204 Pb or in Sr and (>17 5) compared with younger lavas associated with Nd isotopic ratios (Figs 7a and 8). In Pb Pb isotopic the Kerguelen plume (Fig. 8). plots Site 749 lavas form vertical arrays that lie along a downwards extension of the vertical trends defined by basalts from Leg 183 Site 1137 on Elan Bank (Fig. 8). This vertical trend defined by Site 1137 lavas has been DISCUSSION interpreted as reflecting contamination by continental Mantle source for basalt forming the crust (Weis et al., 2001; Ingle et al., 2002a). Kerguelen Plateau and Broken Ridge Site 750 A working hypothesis has been that much of the Kerguelen Plateau and Broken Ridge formed when the Isotopic data are available for 10 samples [six in Table Kerguelen plume head impinged upon recently formed 4 and four from Salters et al. (1992; personal communication, Indian Ocean lithosphere at >115 Ma (e.g. Shipboard 1992)]. For Sr and Nd there is general Scientific Party, 2000). The uppermost igneous surface 1222