FIGURE CAPTIONS

Figure 2. Plate reconstructions of the southern Indian Ocean region (after Royer and Coffin, 1992; Royer and Sandwell, 1989) using a hot spot reference frame (Müller et al., 1993) and keeping Antarctica fixed. Reconstructed position of the Kerguelen hot spot (after Müller et al., 1993) is indicated by stars. Volcanic rock associated with the Kerguelen hot spot is indicated in gray shading, and lamprophyres are indicated as diamonds, as they have appeared through geologic time. Dashed line = a possible northern boundary for Greater India. IND = India; ANT= Antarctica; AUS = Australia. 130.9 (M10) and 118.7 (M0) Ma: Seafloor spreading initiates during Chron M11 (~133 Ma) between Western Australia and Greater India. The Bunbury Basalt (BB) of southwest Australia is erupted close to these breakup events in both time and space. Of the two Bunbury Basalt types, the Casuarina may be related to the breakup of Australia and India, influenced by the Kerguelen hot spot, and the Gosselin may represent magmatism associated with the breakup of Australia and Antarctica. Between 130.9 and 118.7 Ma, Antarctica migrates to the southeast relative to the Kerguelen hot spot. 110 and 100 Ma: Seafloor spreading continues among India, Antarctica, and Australia. Rajmahal (RAJ) rocks postdate breakup of India and Australia by ~15 m.y. (Markl, 1974, 1978) and the breakup of India and Antarctica by ~15-40 m.y. Indian and Antarctic lamprophyres (diamonds) also postdate major breakup events. The first massive pulse of Kerguelen magmatism creates the southern Kerguelen Plateau (SKP) (Figs. 3, 4) at ~110 Ma, as Indian Ocean lithosphere migrates southeast relative to the Kerguelen hot spot. 83 (C34) and 63.6 (C28) Ma: India continues its northward drift relative to Antarctica, and the Kerguelen hot spot is predicted to have remained close to the northeast edge of the central Kerguelen Plateau (CKP) (Figs. 3, 4) and Broken Ridge (BR) (Figs. 5, 6), which form at ~85 Ma. Subsequently, the hot spot generates the Ninetyeast Ridge (NER). 40.1 (C18) and 23.4 (C6c) Ma: At ~40 Ma, seafloor spreading commences between the CKP and BR. The hot spot generates the northern Kerguelen Plateau (NKP) (Figs. 3, 4), and since 40 Ma, as BR and the Kerguelen Plateau continue to separate, produces the Kerguelen Archipelago (KA), Heard and McDonald islands (Figs. 3, 4), the chain of volcanoes between Kerguelen and Heard (Figs. 3, 4).

Figure 3. Bathymetry of the Kerguelen Plateau. Leg 119, 120, and 183 drill sites that recovered igneous basement, are indicated by filled circles and stars, respectively; sites that bottomed in sediment are shown as open circles and stars, respectively. Contour interval = 500 m.

Figure 4. Satellite-derived, free-air gravity map of the Kerguelen Plateau (after Sandwell and Smith, 1997). The plateau consists of five sectors: northern, central, southern, Elan Bank, and Labuan Basin (outlined in white). Current (Leg 183) and previous (Legs 119 and 120) sites are indicated by stars and circles, respectively (solid symbols = basement sites; open = sediment sites). Squares indicate dredge and piston core sites where igneous rock (solid squares) and sediment (open squares) were recovered.

Figure 5. Bathymetry of Broken Ridge. Previous DSDP and ODP drill sites that recovered igneous basement are indicated by solid stars; sites that bottomed in sediment are shown as open stars. Dredge locations (DR-X) that recovered igneous basement are indicated by solid squares. Described symbols for Sites 1141 and 1142 for Leg 183 are indicated by open circles; seismic line used to select these locations is depicted by black line with cruise identifier (C = Robert Conrad). Contour interval = 500 m.

Figure 6. Satellite-derived gravity field for Broken Ridge (Sandwell and Smith, 1997) showing location of previous ODP drill sites, locations of the C dredge sites, which recovered basaltic basement, and ODP Leg 183 drill sites (black star). Bathymetric contours (500-m interval) are from Fisher (1997).

Figure 7. Kerguelen hot spot magma output since ~130 Ma.

Figure 8. Total alkalis (Na₂O + K₂O) vs. SiO₂ plot (wt% with FeO adjusted to 85% of total iron) for classifying tholeiitic and alkalic basalts. Boundary line separating alkalic and tholeiitic fields is from Macdonald and Katsura (1964). A. The open fields show the temporal evolution of Kerguelen Archipelago lavas from the ~30 Ma flood basalts, with compositions transitional between tholeiitic and alkalic lavas (Mts. Bureau and Rabouillere), to the slightly alkalic Southeast Lower Miocene Series (LMS) to the highly alkalic Southeast Upper Miocene Series [UMS]. Data sources are Weis et al., 1993, 1998; Yang et al., 1998). In contrast, lavas from Kerguelen Plateau ODP Sites 747, 749, and 750 are tholeiitic basalts; however, the above basement lavas from ODP Site 748 are alkalic basalts. B. Expanded panel showing more detail for basalts drilled and dredged from the Kerguelen Plateau and Broken Ridge. Lavas dredged from the central Kerguelen Plateau and Broken Ridge (open circles) and ODP Site 738 straddle the boundary line, largely because the total alkali contents of these lavas were increased during postmagmatic alteration. As an extreme example, the solid triangle indicates a highly altered sample from ODP Leg 120, Site 750 (Sample 120-750B-19R-1, 47-50 cm). Data from Davies et al. (1989), Storey et al. (1992), and Mahoney et al. (1995).

Figure 9. ¹⁴³Nd/¹⁴⁴Nd vs. ⁸⁷Sr/⁸⁶Sr showing data points for basalts recovered from the Kerguelen Plateau and Broken Ridge. The Broken Ridge samples are measured data corrected to an eruption age of 88 Ma; the dredged Kerguelen Plateau samples are measured data corrected to an eruption age of 115 Ma. The effects of age correction are shown by the two fields (measured and age-corrected) for Site 738 on the southern Kerguelen Plateau. Data for other sites are not age-corrected because parent/daughter abundance ratios are not available. Data for Kerguelen Plateau and Broken Ridge samples are from Weis et al. (1989), Salters et al. (1992), and Mahoney et al. (1995). Shown for comparison are fields for SEIR MORB (Hamelin et al., 1985/1986; Michard et al., 1986; Dosso et al., 1988; J.J. Mahoney, unpubl. data), St. Paul and Heard Islands (Heard data indicated by trajectory of solid line from Barling et al., 1994), the Ninetyeast Ridge (shaded fields labeled NER and NER DSDP Site 216 from Weis and Frey, 1991, and Frey and Weis, 1995), and the entire Kerguelen Archipelago. Most samples are from the Kerguelen Archipelago plot in the subfield labeled "Kerguelen Plume?" that Weis et al. (1993, 1998) interpreted as representative of the Kerguelen Plume.

Figure 10. ²⁰⁸Pb/²⁰⁴Pb vs. ²⁰⁶Pb/²⁰⁴Pb showing measured data points for basalts recovered from the Kerguelen Plateau and Broken Ridge. Shown for comparison are measured fields for SEIR MORB (two subfields: the field extending to high ratios includes samples near the Amsterdam-St Paul platform), lavas from St. Paul and Amsterdam Islands (diagonal line), and initial ratios for lavas from the Kerguelen Archipelago (shaded area) and the Ninetyeast Ridge. Data sources are as in Figure 7.

Figure 11. Initial ⁸⁷Sr/⁸⁶Sr vs. (La/Nb)_pm (subscript pm indicates normalized to primitive mantle ratios of Sun and McDonough, 1989), showing the positive correlation that arises from an increasing proportion of a continental crust component in the Bunbury Basalt of southwest Australia (Frey et al., 1996). By analogy, we infer that a significant amount of a continental crust component is in the Kerguelen Plateau basalts from ODP Site 738 and dredged basalts from the eastern Broken Ridge. In contrast, most oceanic island basalts and all basalts from the Kerguelen Archipelago and Heard Island have (La/Nb)_pm < 1.2; these lavas are interpreted to be representative of the Kerguelen Plume (e.g., Weis et al., 1993, 1998). Data sources are as for Fig. 7.

Figure 12. A. Abundance ratios of (Th/Nb)_pm vs. (La/Nb)_pm (subscript pm indicates normalized to primitive mantle ratios of Sun and McDonough, 1989). Most oceanic island basalts, including ~100 basalts from the Kerguelen Archipelago, have (Th/Nb)_pm < 1 and (La/Nb)_pm = ~1. In contrast, most continental crust, especially upper crust, is relatively depleted in Nb (and Ta) (e.g., Thompson et. al., 1984) with (Th/Nb)_pm = 5.46 and (La/Nb)_pm = 2.17 in average bulk crust (Rudnick and Fountain, 1995, except with Nb and Ta values of upper crust from Plank and Langmuir, 1998). As indicated, there are considerable differences between lower (LC) and upper (UC) crust composition with the lower crust value of Weaver and Tarney (1984) (W&T), which has much higher (La/Nb)_pm. Dredge 8 basalts from eastern Broken Ridge and Kerguelen Plateau basalts from Site 738 and dredge basalts from the 77° graben lie outside the oceanic basalt field, thereby showing that they contain a continental crustal component, an inference that is consistent with isotopic data (Figs. 7, 8). Also, Kerguelen Plateau samples from Site 747 are offset to high (La/Nb)_pm, but they have normal (Th/Nb)_pm; these basalts may have a smaller proportion of a different continental crustal component. Although all Kerguelen Archipelago flood basalts lie within the field for normal oceanic basalts, lavas of the Heard Island-Big Ben basaltic series trend to high (Th/Nb)_pm. This trend is accompanied by increasing ⁸⁷Sr/⁸⁶Sr (Fig. 9), and it is also a trend reflecting an increasing role for a continental crust component (Barling et al., 1994). Other data sources are Davies et al., 1989; Storey et al., 1992; Mahoney et al., 1995; Yang et al., 1998; Frey et al., 1999, in press. B. Same plot as in A, but with an enlarged scale to include data for continental basalts that have been attributed to result from the Kerguelen plume, that is, the Bunbury Basalt (southwest Australia) and Rajmahal Basalt (northeast India). These basalts show a trend of variable contamination by a continental crust component (Frey et al. 1996; Kent et al., 1997). Estimates of average, upper and lower continental crust are from Rudnick and Fountain (1998) and Weaver and Tarney (1984) and are labeled (R&F) and (W&T) in the figure. Also shown are North Atlantic MORBs recovered during Leg 152 in a transect away from Greenland. The lowermost lavas in Hole 917A, the drill site closest to Greenland, define a trend that is consistent with variable contamination by lower crustal granulites. In fact, Fitton et al., (1998a, 1998b) concluded that two different crustal components are present in lavas from Hole 917A; the oldest lavas contain a component derived from granulite-facies Archean crust, whereas some of the younger lavas contain a component derived from amphibolite-facies Archean crust. Relative to Kerguelen Plateau basalts, the much stronger continental signature in these North Atlantic MORBs probably reflects the lower abundances of incompatible elements in MORBs relative to plume magmas.

Figure 13. Ti/Zr vs. MgO (in weight percent) for basalts from Broken Ridge and the Kerguelen Plateau. The four Kerguelen Plateau basement drill sites, 738, 747, 749, and 750, have distinctive values ranging from unusually high to low ratios; the primitive mantle and normal MORB ratios are 116 and 103, respectively (Sun and McDonough, 1989). Kerguelen Plateau basalts from Site 738, which have an obvious continental crustal component, also have anomalously low Ti/Zr. Site 747 lavas also have relatively low Ti/Zr and high ⁸⁷Sr/⁸⁶Sr (Fig. 7). In many continental flood basalts, relatively low Ti/Zr results from contamination with continental crust. Abundances of Ti and Zr are precisely determined by the shipboard XRF analyses; therefore, Ti/Zr is a useful ratio for shipboard assessment of the role of continental crust in the Kerguelen Plateau and Broken Ridge basalts.

Figure 14. LIP drilling strategies. Age-composition transect sites penetrate ~150 m into volcanic basement, with intermediate and deep sites to be chosen following exploratory drilling. Offset sections provide windows into middle and deep crustal levels. On land sections permit detailed sampling, albeit of tectonized rocks. Reference sites on older oceanic crust record the volcanic and deformational history of LIP emplacement.

Figure 15. Rig Seismic RS180/201 multichannel seismic profile across Site 1135. CDP = common depth point. V.E. = vertical exagerration.

Figure 16. Marion Dufresne MD47/10 multichannel seismic profile across Site 1136. V.E. = vertical exagerration.

Figure 17. Composite stratigraphic section for Site 1135 showing core recovery, a simplified summary of lithology, lithologic unit boundaries, ages of units, names of lithologies, and total carbonate contents expressed as CaCO₃ weight percent.

Figure 18. Composite stratigraphic section for Site 1136 showing core recovery, a simplified summary of lithology, lithologic unit boundaries, ages of units, names of lithologies, and total carbonate contents expressed as CaCO₃ weight percent.

Figure 19. Compositions of volcanic rocks from Sites 1136, 1137, 1138, and 1140 on the Na₂O + K₂O vs. SiO₂ classification diagram. Alkalic and tholeiitic basalt fields are distinguished by the Macdonald-Katsura (1964) line. Compositions of basalts from southern and central Kerguelen Plateau locations are shown as fields for Sites 738, 747, 749, and 750 (Alibert, 1991; Mehl et al., 1991; Salters et al., 1992; Storey et al., 1992) and dredge samples (Davies et al., 1989; Weis et al., 1989). Open circles and diamonds indicate highly altered samples from Sites 1137 and 1138, respectively.

Figure 20. Nb/Ce, Nb/Zr, and Zr/Ti ratios vs. Zr/Y ratio or Zr content. Basalts from Site 1136 are shown as solid symbols; a highly altered sample is shown by the open circle. These ratios are sensitive to the presence of a component from the continental crust. Note that basalts from Site 738, which are thought to contain a continental component, have the lowest Nb/Ce and Nb/Zr and highest Zr/Ti values. Basalts from Site 1136 most closely resemble the upper compositional group from Site 749. Ratios are normalized using estimated primitive mantle abundances of Sun and McDonough (1989). Numbers indicate ODP site numbers.

Figure 21. Rig Seismic RS179/601 multichannel seismic profile across Site 1137. CDP = common depth point. V.E. = vertical exaggeration.

Figure 22. Composite stratigraphic section for Site 1137 showing core recovery, a simplified summary of lithology, lithologic unit boundaries, basement unit boundaries, ages of units, and names of lithologies.
Figure 23. Incompatible trace element compositions of basalts from Site 1137 compared with those from other parts of the Kerguelen Plateau. Data sources are given in caption to Figure 19.

Figure 24. Mantle-normalized Nb/Ce, Zr/Ti, and Nb/Zr vs. Zr/Y ratio or Zr content. These ratios are sensitive to the presence of a component from the continental crust. Note that basalts from Site 738, which are thought to contain a continental component, have the lowest Nb/Ce and Nb/Zr and highest Zr/Ti values. Site 1137 basalts have a comparable Zr/Ti range but higher Nb and Zr contents and, consequently, higher Zr/Y, Nb/Ce, and Nb/Zr. Ratios are normalized using estimated primitive mantle abundances of Sun and McDonough (1989). Symbols and data sources as in Figure 19. Numbers indicate ODP site numbers.

Figure 25. Nb/Y vs. Zr/Y diagram for Kerguelen Plateau basement basalt from Sites 738, 747, 749, 750, 1136, and 1137. The diagonal lines delimit the field of plume-derived Icelandic lavas; MORB plots to the right of this field. Fitton et al. (1998) argue that melts derived by a varying extent of melting from a common peridotite source define trajectories nearly parallel to these boundary lines. Average MORB and oceanic island basalt (OIB) (Sun and Mc Donough, 1989), average lower and upper crust (LCC and UCC, respectively, Rudnick and Fountain, 1995), alkalic basalts from Site 748 (recovered above basement), and the evolved rocks from Site 1137 are shown for reference.

Figure 26. Trace-element contents of basalts from Site 1137, normalized to primitive mantle values of Sun and McDonough (1989). All Site 1137 basalts are enriched in incompatible elements. The wide range of Rb and K shows the mobility of these elements during postmagmatic alteration.

Figure 27. Rig Seismic RS179/101 multichannel seismic profile across Site 1138. CDP = common depth point. V.E. = vertical exaggeration.

Figure 28. Composite stratigraphic section for Site 1138 showing core recovery, a simplified summary of lithology, lithologic unit boundaries, ages of units, and names of lithologies.

Figure 29. A. Site 1138 igneous rock compositions in the Na₂O + K₂O vs. SiO₂ clasification diagram (LeBas et al., 1986). Unit 1 cobbles are dacites, whereas the pumice-rich portion of the Unit 2 breccia has a trachytic composition. Units 3-22 are tholeiitic to transitional basalts. Alkalic and tholeiitic basalt fields are distinguished by the Macdonald and Katsura (1964) line.

Figure 30. Downhole variations in trace element abundances and primitive-mantle-normalized ratios for Site 1138 basalts. Symbols as in Fig. 29.

Figure 31. Nb/Y vs. Zr/Y has been proposed as a discriminant for contamination of mantle plume derived melts, which fall within the two sloping, parallel lines defined by uncontaminated basalts from the Iceland plume (Fitton et al., 1996). The Sun and McDonough (1989) primitive mantle composition is located by the cross. Continental crustal material typically plots to the right of the right line. Site 1138 basalt compositions are most similar to Site 747 compositions and trend from the mantle array toward Site 1137 basalts (modest continental crustal assimilation). However, this trend most probably resulted from crystal fractionation of clinopyroxene.

Figure 32. Marion Dufresne MD109-05 multichannel seismic profile across Site 1139. CDP = common depth point. V.E. = vertical exaggeration.

Figure 33. Composite stratigraphic section for Site 1139 showing core recovery, a simplified summary of lithology, lithologic unit boundaries, ages of units, and names of lithologies.

Figure 34. Interpretative summary diagram for Site 1139 showing 19 igneous units, which range from basalt to rhyolite, together with rudstone and volcaniclastic breccia. All units are highly altered, which obscures the primary vesicularity of the lava flows and marginal breccias. Sanidine xenocrysts are distributed throughout the more mafic flows. Although changes in potassium concentration from downhole natural gamma measurements correlate generally with the defined igneous units, the logging data show more complexity, indicating unrecovered changes in lithology.

Figure 35. A. Site 1139 igneous rock compositions on the Na₂O + K₂O vs. SiO₂ classification diagram (LeBas et al., 1986); alkalic and tholeiitic basalt fields are distinguished by the Macdonald and Katsura (1964) line. Note that there is one chemical analysis per igneous unit except for Units 5, 18, and 19 from which two, two, and four samples were analyzed, respectively. The less altered Units 18 and 19 samples are marked in bold and other samples from these units show considerable spread because of alteration. B. Site 1139 mafic compositions compared to samples from Kerguelen Plateau Sites 1136, 1137, 1138. The fields represent data from previous dredging and drilling on the southern and central Kerguelen Plateau (Sites 738, 747, 749, and 750 and dredge locations reported by Weis et al.,1989). Data sources are Davies et al. (1989), Weis et al. (1989), Alibert (1991), Mehl et al. (1991), Salters et al. (1992), Storey et al. (1992), Mahoney et al. (1995), and this study. Numbers indicate ODP sites.

Figure 36. Marion Dufresne MD109-06 multichannel seismic profile across Site 1140. CDP = common depth point. V.E. = vertical exaggeration.

Figure 37. Composite stratigraphic section for Site 1140 showing core recovery, a simplified summary of lithology, lithologic unit boundaries, ages of units, and names of lithologies.

Figure 38. Close-up photograph of Section 183-1140A-28R-3, 86-101 cm (Piece 2). Glassy pillow rind with calcite filling vesicles and veins. Open space-filled dolomite and baked white sediment is along the margin with the glass.

Figure 39. Photomicrograph (plane polarized light) of the glass from a chilled pillow margin from Unit 1 containing fresh, euhedral olivine phenocrysts (characteristic six-sided shape; center) with inclusions of two small chromites and glass (left of the crystal) in thin section 183-1140A-26R-1, 5 18 cm (Piece 1A). The chromite inclusions in the olivine indicate early oxide precipitation.

Figure 40. Primitive-mantle normalized Zr vs. Nb and Zr/Ti vs. Zr diagrams for Site 1140 basalts compared with data for central and southern Kerguelen Plateau basalts. Numbers indicate ODP sites.

Figure 41. JOIDES Resolution JR183-101 single-channel seismic profile across Sites 1141 and 1142. CDP = common depth point. V.E. = vertical exaggeration.

Figure 42. Composite stratigraphic section for Site 1141 showing core recovery, a simplified summary of lithology, lithologic unit boundaries, ages of units, and names of lithologies.

Figure 43. Composite stratigraphic section for Site 1142 showing core recovery, a simplified summary of basement lithology, basement unit boundaries, and names of basement lithologies. The interval from the seafloor to top basement rocks was washed, and the sedimentary section was not cored.

Figure 44. Color close-up photograph of Site 1137 conglomerate.

Figure 45. Photomicrographs of garnet gniess. A. Poikiloblastic garnet (gt) in Unit 6 (clast in conglomerate, Sample 183-1137A-35R-2, 46-47 cm). Field of view = 1.4 mm (plane-polarized light). B. Porphyroblastic garnet (gt) and biotite (bi) from Unit 9 (clast in tuff, Sample 183-1137A 44R-4, 44-46 cm). Field of view = 2.8 mm (plane polarized light).

Leg 183 Operations Synopsis
Leg 183 Table of Contents