In this section we describe the primary texture, mineralogy, and geochemistry of the igneous basement units at Site 1138 (Fig. F53; Table T12). Volcanic structures and secondary mineralogy are described briefly here and in greater detail in "Physical Volcanology" and "Alteration and Weathering".
We recognize 22 units within the recovered portion (48%) of the 145.2 m of igneous basement drilled at Site 1138. Unit 1 occupies the uppermost 19.5 m and is composed of rounded cobbles of aphyric to sparsely feldspar-phyric dacite (Fig. F54). Unit 2 is a 24.3-m succession of interbedded pumice-lithic breccias and clays that contain clasts of highly plagioclase-clinopyroxene-phyric basalt. A rather uniform sequence of thin basaltic lava flows, ranging in curated thickness from 0.7 to 9.6 m, comprises Units 3 to 22 and forms the bulk of the drilled basement section. We interpret these units to represent a series of subaerial eruptions. The presence or absence of oxidized flow tops, rounded clasts, and sediments filling pore space in breccias indicate variable time intervals between flows (see "Physical Volcanology"). Most flows exhibit both brecciated and massive portions, with wide variation in vesicularity. Alteration ranges from high to complete in brecciated zones and slight to locally high in massive intervals. Figure F53 illustrates the recovered section, the positions of unit boundaries, flow type, and mineralogy together with a schematic representation of flow structure.
We distinguish the individual basement units on the basis of volcanological features, principally on significant downward changes in flow structure from brecciated or highly vesicular flow tops to massive interiors in the case of lava flows and degree of welding and types of clasts within the (possibly reworked) pyroclastic flows. As at Sites 1136 and 1137, the tops of several flows are marked by a notable zone of oxidation, suggesting subaerial weathering between some eruptions. Marked changes in vesicularity, vesicle shape, and distribution are also evident within units. We have noted features characteristic of pahoehoe, aa, and transitional flows (Fig. F53; see "Physical Volcanology"). The sequence of relatively thin flows at Site 1138 shows similarities to Hawaiian lava sequences (see "Physical Volcanology") and contrasts with the generally thicker flows drilled at Sites 1136 and 1137.
Overall, Site 1138 basalts are moderately to highly vesicular and aphyric to sparsely plagioclase-phyric. Groundmass textures are generally intersertal to intergranular, although occasionally subtrachytic (Fig. F55). The massive interiors of several flows exhibit subhorizontal glassy mesostasis segregations up to 1 mm wide, at approximately centimeter-scale spacing (Fig. F56). The relatively fast cooling of these units is highlighted by the skeletal dendritic forms of the titanomagnetites (Figs. F57, F58). Green and brown clay has replaced large portions of the glassy mesostasis and groundmass clinopyroxene. Vesicles and veins are generally completely filled, with clay at the margins and with zeolite in the interiors.
Discrete clinopyroxene phenocrysts are present only in Units 9 and 19. Glomerocrysts in Units 3 and 4 consist of monomineralic aggregates of plagioclase. In Unit 9, glomerocrysts contain both clinopyroxene and plagioclase (Fig. F59). The percentage of phenocrysts and glomerocrysts correlates weakly with flow thickness. Plagioclase phenocryst compositions are An60-65, whereas microphenocrysts are ~An55. In many of the more altered flow margins, rounded plagioclase phenocrysts have been replaced by low-relief, weakly birefringent zeolites that have grown parallel to the feldspar cleavage and preserved its Carlsbad twinning.
A notable aspect of Units 5-16 and 19 is the presence of 1%-5% olivine microphenocrysts, now completely replaced by clay although original morphologies are still preserved (Fig. F60). The olivines are generally euhedral to subhedral and are slightly larger than the groundmass plagioclase, clinopyroxene, and titanomagnetite. Extensive maghemite has formed, apparently via exsolution from the titanomagnetite (or by alteration along crystallographic planes) in Units 4, 5, and 20, but it is rare in other units. Sulfides are generally absent, with the exception of trace amounts of chalcopyrite, pyrite, and/or pentlandite, which are associated with groundmass alteration and as inclusions in primary phases in Units 6, 15, 17, and 21.
We list XRF analyses of major and trace elements for 20 basalts and three felsic rocks in Table T13. Basalt compositions are tholeiitic to transitional (Fig. F61) and are slightly olivine to quartz normative. With the exception of one sample from Unit 18, which has anomalously high Na2O, the low alkali contents (Na2O + K2O = 2.5 to 3.5 wt%) and LOI (0.5 to 2.0 wt%) indicate that the massive portions of these units are relatively unaltered. Major element abundances are relatively uniform, with narrow ranges in MgO (4.5 to 7.0 wt%) and SiO2 (47.0 to 49.9 wt%) (Fig. F62). In detail, however, there are systematic changesdowncore in major and minor element abundances (Fig. F62) and in variation of abundance with Mg# (Mg# = Mg/[Mg + Fe2+] with Fe2+ estimated as 80% of total iron) (Fig. F63; Table T13). For Site 1138 basalt compositions, we found that smoother trends developed in plotting elemental abundance variations against Mg# rather than MgO. Overall, these downcore changes are rather continuous, but there appears to be a notable break halfway through the sequence, seen particularly in TiO2 and P2O5 and in a change from irregular to smooth variation in Mg# (Fig. F62). The boundary between the upper series (Units 3-13) and the lower series (Units 14-22) is marked by an oxidized breccia with rounded clasts, which is one of several possible time breaks within this eruptive sequence (see "Physical Volcanology").
MgO decreases downcore, whereas Fe2O3* (total Fe expressed as Fe2O3) increases, resulting in a trend toward lower Mg# (i.e., more evolved compositions with depth) (Fig. F62). Al2O3 is distinctly lower in the lower series (12.9 to 13.8 wt%) than in the upper series (13.9 to 16.8 wt%). As noted above, TiO2 and P2O5 contents are greater in the lower series (2.6 to 3.2 wt% and 0.30 to 0.38 wt%, respectively) than in the upper series (1.9 to 2.3 wt% and 0.19 to 0.26 wt%, respectively). Compositional variability within the entire sequence of lava flows is consistent with shallow-level fractionation of the phases observed in thin section (i.e., olivine, clinopyroxene, and plagioclase). Interestingly, however, the compositions reflect a magmatic system that moved toward more primitive melts, rather than evolved liquids, with decreasing age. The lower Al2O3 abundances in Units 14-22 (Fig. F63) suggest that plagioclase fractionation was important.
The major element trends are mirrored by variations in trace element abundances. Incompatible elements, such as Zr and Y, decrease upsection, whereas compatible elements, such as Ni and Cr, increase, reflecting diminishing extents of crystal fractionation with time (Fig. F64). Incompatible trace element ratios Zr/Ti and Zr/Y decrease very slightly upcore, in keeping with the slightly more incompatible behavior of Zr compared with Ti and Y (Fig. F64). Patterns of incompatible trace elements, normalized to primitive mantle values in Figure F65, are intermediate, relative to those from other CKP sites. The important role of plagioclase fractionation inferred from Al2O3 content and Al2O3/TiO2 ratio (Fig. F63) is corroborated by a depletion in Sr relative to Zr, especially in the lower series basalts (Figs. F64, F65).
Two cobbles from Unit 1 have very similar dacitic compositions, whereas a pumice lithic breccia sample from Unit 2 plots in the trachyte field (Fig. F61). These highly silicic compositions are consistent with formation during explosive subaerial volcanism at the waning stage of magmatic activity, inferred from the pyroclastic deposits in Unit 2 (see "Physical Volcanology").
The enrichment in TiO2 (up to 3.2 wt%), and especially Fe2O3* (up to 19.2 wt%) distinguishes the lower group of Site 1138 basalts from those of other sites on the plateau (Fig. F66). Such trends are similar to those for Fe-Ti-rich basalts from other large igneous provinces that are inferred to have evolved at low pressures from tholeiitic magmas (e.g., Imnaha basalts and Columbia River flood basalt province, Hooper, 1997; Mananjary basalts and Cretaceous Madagascar province, Storey et al., 1997; East Greenland ferrobasalts, Brooks et al., 1991). Another notable feature of Site 1138 basalts relative to basalts from Sites 1136 and 1137 is their low Al2O3/CaO (Fig. F66) and strong Sr depletions (Fig. F65), reflecting large amounts of plagioclase fractionation.
Trace element discriminants such as (Zr/Ti)N vs. Zr reveal that the Site 1138 basalts fall within the range of other central and southern Kerguelen Plateau basalts (Fig. F67). The (Zr/Ti)N values, in fact, are most similar to those for Site 747 basalts and fall between those of Site 1136 and 1137 basalts. This kinship is also shown in Figure F65, where the incompatible trace element patterns for Site 1138 basalts fall between the flatter Site 1136 patterns and the more elevated patterns from Site 1137 and most closely match the Site 747 patterns. With regard to Nb/Y vs. Zr/Y, we see that the Site 1138 basalts are again most similar to the Site 747 basalts (Fig. F68). The Site 1138 basalts plot mainly within the field defined by basalts derived from the Iceland mantle plume thought to be uncontaminated by continental crustal material or depleted asthenosphere (Fitton et al., 1997, 1998). This field also contains the Sun and McDonough (1989) estimate for the composition of primitive mantle, as well as compositions of basalts from the Kerguelen Plateau that do not show other evidence of continental crustal assimilation (Fig. F68). The most evolved basalt compositions from Site 1138 (lower series) trend from the plume field toward the Site 1137 basalts. Given that higher Zr/Ti and lower Nb/Y values may derive from contamination of melts with continental crust (see "Igneous Petrology" in the "Site 1137" chapter), one interpretation of this trend is that Site 1138 magmas have assimilated small amounts of continental material. However, removal of clinopyroxene, which has a significant partition coefficient for Y (Pearce and Norry, 1979), from more evolved tholeiitic magmas will drive the residual liquid away from the array at a shallow angle (constant Nb/Zr), in precisely the direction observed for the most evolved Site 1138 basalts. We conclude, then, that there is no strong evidence from shipboard analyses that Site 1138 magmas assimilated any continental crustal material, and we await further shore-based studies to resolve this issue.