IGNEOUS PETROLOGY

We recovered 10 basement units totaling 144.2 m at Site 1137, with an average recovery of 73%. Seven units are subaerial basaltic lavas with a total thickness of 94.1 m; three are volcaniclastic sedimentary interbeds within the basalt sequence. For lavas not separated by sedimentary layers, we have identified units representing single lava flows by using marked differences in flow structure, vesicularity, and phenocryst content and type occurring over a short interval of recovered section. We have subdivided some of the lava-flow units into subunits corresponding to brecciated or highly vesicular and more massive parts (see "Physical Volcanology").

In this section we describe the lithology, petrography, primary mineralogy, and geochemistry of the seven igneous basement units at Site 1137. Volcanic structures are described in "Physical Volcanology" and secondary mineralogy is described in "Alteration and Weathering". We also describe the igneous constituents of the three volcaniclastic sedimentary units in the basement sequence. The sedimentological features of these units are described in "Lithostratigraphy" and "Physical Volcanology".

Curated depths of the units (Table T7; Fig. F35) assume that the top of a core comes from the top of the cored interval (see "Introduction" in the "Explanatory Notes" chapter), so the curated depths of flow boundaries (in mbsf) normally differ from their actual depths in the hole. Relatively good core recovery and the high quality of the logging results enabled us to determine true thicknesses of the basement units through core-log integration (see "Physical Volcanology" and "Downhole Measurements"). In this section we refer to sample locations and unit boundaries according to curated depths in mbsf as described above, but unit thicknesses were derived from core-log integration (Tables T9, T10; Fig. F35).

The upper four basement units form a 59.4-m-thick sequence of lava flows. Unit 1 is slightly altered, massive, nonvesicular to slightly vesicular, sparsely plagioclase-phyric basalt. The top of this unit is marked by an erosional unconformity; thus, the recovered 6.1 m is a minimum estimate for the original flow thickness. Unit 2 (10.1 m thick) is divided into a 2.9-m upper brecciated subunit (2A) and a 7.2-m lower massive subunit (2B). Subunit 2A is moderately to highly altered, very sparsely to highly vesicular, sparsely plagioclase-phyric basaltic breccia that is interpreted as a flow top. Subunit 2B is slightly to moderately altered, massive, aphyric to sparsely plagioclase-phyric basalt.

Unit 3 is 16.6 m thick, and its top is marked by an abrupt increase in vesicularity, vesicle size, and alteration. Subunit 3A is 9.6 m thick and consists mostly of highly vesicular aphyric basalt. It is highly oxidized and altered at the top, but the level of alteration decreases downward to slight at the base of the subunit. Subunit 3B is a 7.0-m-thick, massive, slightly to moderately altered, aphyric to sparsely plagioclase-phyric basalt. Beneath its base is an ~8-cm-thick laminated, brecciated volcanic siltstone (interval 183-1137A-29R-2, 64-73 cm) that blankets the top of the thick (26.6 m) underlying Unit 4 basalt flow (Fig. F19). Below a classic pahoehoe top, the moderately plagioclase + clinopyroxene + olivine-phyric basalt of Unit 4 varies from massive to highly vesicular. Bulk-rock alteration varies with vesicularity from slight to high. Vesicle fillings constitute most of the alteration, whereas the phenocrysts and groundmass tend to be only slightly altered. The basal chill zone of this flow and the baked contact with the underlying sediment of Unit 5 are well preserved near Section 183-1137A-33R-1, 122 cm (Fig. F20).

Units 5 and 6 are sediments and represent a significant hiatus of unknown duration between the eruption of Units 1-4 and 7-10. Unit 5 is a 4.4-m-thick sequence of interbedded crystal-lithic volcanic sandstone and siltstone (see "Lithostratigraphy"). Unit 6 is a 31-m-thick lithic volcanic conglomerate containing well-rounded, granule to small boulder-sized clasts embedded in a poorly to moderately sorted matrix of coarse to fine sand-sized crystal-lithic volcanic fragments, including altered glass (see "Lithostratigraphy" and "Physical Volcanology"). The clasts are predominantly volcanic lithologies and are moderately to highly altered; many have distinct, concentric oxidation rims (Fig. F9). In the upper part of the unit, the predominant clast lithologies are massive trachyte (sanidine-plagioclase-clinopyroxene-phyric), flow-banded rhyolite (sanidine-plagioclase-phyric), and highly plagioclase-phyric basalt. Minor lithologies include granitoid and garnet-biotite gneiss. In the lower part of the unit, aphyric and plagioclase-phyric basaltic pebbles are more abundant. Primary textures and mineralogy of the clasts within the conglomerate are described below.

Units 7, 8, and 10 form a basaltic sequence that is 34.7 m thick, interrupted only by the Unit 9 tuff. Unit 7 is a 13.0-m-thick, sparsely to moderately plagioclase-phyric basalt that has a 5.8-m-thick, highly to completely altered flow-top breccia (Subunit 7A) and a 7.2-m-thick, slightly to moderately altered massive flow interior (Subunit 7B). Subunit 8A is the upper 3.4-m-thick, brecciated and vesicular part of Unit 8 and consists of moderately to highly altered, sparsely to moderately plagioclase-phyric basaltic breccia that grades into massive basalt. Subunit 8B (7.6 m thick) is slightly to moderately altered, moderately plagioclase-clinopyroxene-phyric, massive basalt. Unit 10 is moderately to highly plagioclase-clinopyroxene-phyric basalt at least 10.7 m thick (drilling ceased before reaching the bottom of the flow), with a 3.5-m-thick moderately to highly altered, brecciated top (Subunit 10A) and a >7.2-m-thick slightly altered, massive underlying interior (Subunit 10B). Both subunits contain abundant plagioclase-dominated glomerocrysts, as large as 20 mm in size.

The lower basalt sequence is interrupted by Unit 9, a 14.9-m-thick, highly altered crystal-vitric tuff that contains ~40% feldspar crystals and <5% oxidized mafic phenocrysts (see "Physical Volcanology" for a detailed description of the tuff). Scattered (<5%), round to elongate, lithic fragments as large as several centimeters (Fig. F10) consist of basalt, trachyte, granitoid, and garnet-biotite gneiss, generally similar to those found in the conglomerate (Unit 6). Primary minerals in the tuff as well as several of the clasts within the tuff are described below.

Basalts (Units 1-4, 7-8, 10)

At least one massive flow interior and one finer-grained sample closer to a chilled zone were sampled from each flow for shipboard X-ray fluorescence (XRF) and thin-section analysis. Other interesting macroscopic features were also sampled for thin-section and, in some cases, XRF analysis. Table T10 summarizes the principal petrographic characteristics of each unit; phenocryst abundance for each of the basaltic units is shown in Figure F36. For the complete thin-section descriptions, see the "Core Descriptions" contents list.

Alteration is only slight to moderate in the interiors and bases of the units (Subunits "B"), but moderate to high in the vesicular and/or brecciated flow tops (Subunits "A") (Fig. F35). The major minerals are relatively unaltered within flow interiors, but glass and olivine are always replaced by clays. Skeletal habits of titanomagnetite and, to a lesser extent, plagioclase reflect rapid cooling during extrusion of the basalts, especially in the brecciated clasts and flow-boundary chill zones. Vesicles are generally, but not exclusively, concentrated at flow margins and are either partially or totally filled with clay and/or zeolite, with subordinate amorphous silica (see "Alteration and Weathering" and "Physical Volcanology").

The basalts are sparsely to highly porphyritic. As in basalt flows at Site 1136, phenocryst phases are present in different forms as (1) discrete, single crystals of plagioclase (0.5%-20%), clinopyroxene (0%-5%), and olivine (0%-8%) or (2) as compact glomeroporphyritic intergrowths (0%-23%), which are usually monomineralic (plagioclase) or occasionally bimineralic (plagioclase with subordinate clinopyroxene). Although there is a hiatus in volcanism between Units 4 and 7, we observed no significant differences in mineralogy or texture between the upper (Units 1-4) and lower (Units 7-8) basalts. However, the lowermost unit (Unit 10) is petrographically distinct from the other basalt units in that it contains much larger (and generally more abundant) plagioclase phenocrysts and glomerocrysts.

Normal and oscillatory zoning and resorption features are common in the glomeroporphyritic plagioclase throughout the basement sequence and also in the larger discrete plagioclase phenocrysts of Unit 10. Some of these larger plagioclase crystals have overgrowths and display sieve textures (Fig. F37). A spectacular example of oscillatory zoning is in Unit 10 (Sample 183-1137A-45R-4, 118-120 cm), where a 20-mm glomerocryst contains dramatically zoned plagioclase (Fig. F38). Subunit 2B contains a coarser-grained microgabbro xenolith (~1 cm diameter) with somewhat embayed, subhedral to anhedral clinopyroxene and plagioclase (~An₇₀) reacting with the basalt host (Fig. F39).

We estimated plagioclase phenocryst compositions to be between An₅₅ and An₆₅, except in Unit 8 and perhaps Unit 7, where the plagioclase is more calcic (An_70-75). In several units, plagioclase glomerocryst compositions, An₆₅ to An₇₅, appear to be slightly more calcic than smaller, discrete phenocryst plagioclase. In flow tops and basal contacts, plagioclase phenocrysts are commonly altered to a mixture of zeolite ± sericite ± clay and are locally replaced by carbonate. This alteration preserves Carlsbad twinning in places but destroys finer-scale albite twins, giving plagioclase the appearance of an alkali feldspar or feldspathoid (Fig. F40).

Clinopyroxene phenocrysts are found predominantly in the interior of flow units. Altered microphenocrysts of olivine and possibly clinopyroxene (replaced by clay or zeolite pseudomorphs) (Fig. F41) are present in fine-grained flow tops and bases. In some of the units, the abundances of clinopyroxene and olivine phenocrysts exhibit a broad inverse relationship through the sequence, with olivine being more abundant at flow margins and clinopyroxene in the flow interiors (Fig. F36).

The groundmass of all the basaltic rocks is fine-grained to aphanitic and consists of plagioclase (45%), pyroxene (35%), titanomagnetite (10%), and altered mesostasis. The most significant difference in primary mineralogy and texture of the groundmass is between massive flow interiors and the finer-grained flow-top breccias and vesicular flow bottoms. We observed intergranular to intersertal and rare subtrachytic textures in the interiors of the flows. Toward the flow margins, the grain size decreases dramatically and the textures become more hyalopilitic, almost vitrophyric, as the proportion of glass increases. This is particularly evident in the flow breccias where the groundmass is dominated by altered glass. In places, this glass contains a myriad of minute (<0.01 mm) titanomagnetite crystals (e.g., Subunit 10A).

The dominant opaque phase, titanomagnetite, is present in two forms: (1) skeletal, acicular crystals with serrated edges, and (2) tabular subhedral octahedra. Unlike Site 1136, we observed no maghemite exsolution, although in Subunit 2B (Sample 183-1137A-25R-5, 113-115 cm) ilmenite has exsolved from titanomagnetite in a few crystals (Fig. F42). Sulfides are rare in these basalts, but trace amounts of pentlandite were found in Units 2 and 4 and pyrrhotite in Unit 10. The most abundant sulfide was found in Subunit 2B (Sample 183-1137A-26R-2, 38-40 cm), where pentlandite is a late-stage vein segregation and pervades into the groundmass (Fig. F43).

Unit 5: Crystal-Lithic Volcanic Siltstone and Sandstone

We examined a sample from one of the coarsest layers (medium sand size) in the sandstone of Unit 5 in thin section and by XRF analysis (Sample 183-1137A-33R-3, 50-53 cm) (see Fig. F8). Volcaniclastic detritus dominates the section, with ~75% angular to subangular trachytic and basaltic lithic fragments (0.5 mm) and only minor alkali feldspar (~10%) and quartz (~5%). The alkali feldspar component contains both clear sanidine and inclusion-rich, perthitic grains. Subhedral garnet grains up to 0.4 mm in diameter (~1%) (Fig. F44) and rare biotite are also present. Thus, the mineralogy of the sandstone is similar to that in clasts in the immediately underlying conglomerate (Unit 6). In addition, unaltered, angular to subangular, brown hornblende up to 0.2 mm is present. This mineral is not present in the lavas and is distinct from the brown amphibole in the crystal-vitric tuff.

Unit 6: Conglomerate—Igneous Clasts

The conglomerate contains diverse igneous clasts. Some clasts appear similar to lavas in the basalt flow units and were not studied in further detail. Clasts of sanidine-phyric trachyte and rhyolite are relatively abundant, and we also found cobbles of granitoid. We studied these distinctive clasts in thin section, and two were analyzed by XRF.

Sanidine-Plagioclase-Clinopyroxene-Phyric Trachyte

This clast lithology is common in the upper two-thirds of recovered core from Unit 6 and decreases in abundance toward the base; the examined clast (Sample 183-1137A-33R-5, 10-12 cm) is from the top of the unit. The trachyte contains ~20% subhedral (rounded) to euhedral sanidine phenocrysts and ~4% subhedral plagioclase phenocrysts, some of which have a continuous overgrowth of either sanidine or relatively Na-rich plagioclase (Fig. F45). Clinopyroxene phenocrysts (2%) are moderately altered, although fresh cores persist. Minor euhedral titanomagnetite phenocrysts (0.5%) are present. The groundmass is dominated by feldspar with minor titanomagnetite and glass and trace amounts of zircon.

Sanidine-Phyric Flow-Banded Rhyolite

As with the trachytic clasts, this lithology is common throughout the upper two-thirds of the recovered unit and decreases in abundance in the bottom third. The examined clast (Sample 183-1137A-34R-2, 113-117 cm) contains subhedral to rounded sanidine phenocrysts (8%-15%) and subordinate plagioclase (1%-3%). No quartz phenocrysts were identified, but the high SiO₂ content of the rock (74 wt%) indicates a rhyolitic composition (see "Geochemistry: Major and Trace Element Compositions"). The entire groundmass is marked by conspicuous flow banding, a series of millimeter-scale alternating light and dark layers that bend around the phenocrysts (Fig. F46). The lighter, translucent bands are recrystallized quartz and feldspar, whereas the darker, opaque bands are devitrified glass containing microspherulites. A minor amount (1%-2%) of groundmass titanomagnetite is present, as well as trace amounts of zircon and apatite.

Granitoid

A granitoid clast from the bottom of Unit 6 consists of alkali feldspar (50%) and plagioclase (38%), with subordinate quartz (12%) and no significant mafic or hydrous phases. Symplectite intergrowths between the two feldspars are ubiquitous (Fig. F47), and the feldspars are generally untwinned. Titanomagnetite and zircon are present in trace amounts.

Units 6 and 9: Metamorphic Clasts

Metamorphic clasts are found in both the conglomerate of Unit 6 and the crystal vitric tuff of Unit 9. We studied four in thin section, two from each unit, and selected one for XRF analysis.

Garnet-Biotite Gneiss

Two clasts from Unit 9 and one from Unit 6 are garnet-biotite gneiss. Garnets are distinctively poikiloblastic in the clasts from the conglomerate (Sample 183-1137A-35R-2, 46-47 cm) (Fig. F48) and from the bottom of Unit 9 (Sample 183-1137A-44R-4, 44-46 cm) but are porphyroblastic in the other clast from Unit 9 (Sample 183-1137A-43R-4, 57-60 cm). This latter sample, unlike the other two, contains no obvious quartz and also exhibits symplectic intergrowths between plagioclase and K-feldspar. All three clasts have a xenoblastic groundmass with a weak gneissic fabric delineated by biotite (4%-20%) and plagioclase (5%-20%). The groundmass is dominated by alkali feldspar (45%-60%), with quartz (10%-20%), and subordinate amounts of plagioclase (5%-20%). Accessory zircon is distinctive in the two quartz-bearing gneisses and absent from the other clast.

Actinolite Gneiss

Another clast sampled from Unit 6 is actinolite-bearing gneiss (Fig. F49). It is dominated by xenoblastic alkali feldspar (70%), with subordinate quartz (20%) and minor actinolite (~3%), plagioclase, magnetite, and microcline. A weak gneissic fabric is evident.

Unit 9: Crystal-Vitric Tuff

We examined seven thin sections from different levels within the 14.9-m-thick crystal-vitric tuff sequence. In the upper half of the sequence, the original glassy matrix has been almost completely altered to relatively isotropic, pale green to brown clay, making it difficult to discern any original structure in the matrix (Samples 183-1137A-41R-1, 127-128 cm; 41R-2, 98-99 cm; 41R-3, 42-46 cm; and 43R-3, 42-44). However, in the lower part of the tuff (Samples 183-1137A-43R-4, 57-60; 44R-4, 6-9 cm; and 44R-4, 44-46 cm), outlines of 0.1- to 0.2-mm-wide cuspate and tricuspate glass shards are still apparent (Figs. F28, F50, F51, F52), including some patches of relatively unaltered, pale yellow isotropic glass (Fig. F52).

The modal abundance of the major phenocryst phases is relatively uniform throughout the tuff, with 35%-40% sanidine, minor but significant amounts of plagioclase and quartz (1%-3% each), and trace amounts of opaques, zircon, and kaersutite. Sanidine and quartz vary from euhedral to deeply embayed (Fig. F50); some are broken or disrupted (Fig. F51). Most of the symplectite growths on sanidine phenocrysts appear to be an alteration reaction with the groundmass. However, rare isolated garnet crystals and some of the symplectic and inclusion-rich alkali feldspar are probably xenocrystic. Mafic pseudomorphs after biotite(?) are now nearly completely altered to pleochroic brown-green clay and usually have inclusions of opaques, zircon, and kaersutite (Fig. F53). Relatively large (up to 0.15 mm) melt inclusions are present in some sanidine phenocrysts (Fig. F52).

Basalt Compositions

We analyzed 20 basalt samples by XRF (Table T11). Seven samples are from relatively unaltered flow interiors, three were specifically chosen to investigate the effects of alteration, and two were chosen to investigate macroscopic features (a microgabbro xenolith and a glass-rich segregation vein from Unit 7). To characterize intraflow variation, chilled margins, and vesicle filling, we also analyzed multiple samples from Units 3, 4, and 7 (Table T11).

Most Site 1137 basalts are slightly quartz-normative tholeiitic basalts; SiO₂ contents vary from 50.4 to 54.0 wt%, MgO from 4.4 to 7.4 wt%, and TiO₂ from 1.84 to 2.68 wt% (Table T11, Figs. F54, F55). The low MgO, Cr, and Ni contents of most samples indicate that these are not primary melts from peridotite (Fig. F55; Table T11). The three samples chosen specifically to investigate alteration have compositions in the alkalic field (see the square symbols in Fig. F54), presumably because postmagmatic alteration increased their alkali content, especially K₂O. The two other basalts with compositions in the alkalic field, as well as the three samples in the tholeiitic field closest to the Macdonald-Katsura line (see the diamond symbols in Fig. F54), are chilled flow-margin samples with up to 50% clay and zeolite after glass and very fine-grained groundmass phases. These samples are also likely to have elevated total alkali contents as a result of alteration.

The relatively unaltered nature of the flow interiors is reflected in their low loss on ignition (LOI). Of the seven flow interior samples, only one has LOI >1.6 wt%. In contrast, the more altered samples from Units 3, 4, 7, and 10 have generally higher LOI values (as much as 3.3 wt%), together with higher K₂O and Rb contents. The large concentration range of K and Rb shows that their abundances were affected by postmagmatic alteration (Fig. F56A, F56B). In the least-altered basalts (Fig. F56C, F56D), primitive mantle-normalized Ba abundance is, with one exception, greater than normalized values of the more compatible elements, and abundance ratios of Ba to these elements are relatively constant. On this basis, we tentatively conclude that high Ba content is an original magmatic feature of Site 1137 basalts. Abundances of Nb, Ce, Zr, and Y are very similar in the altered and least-altered basalts, indicating that secondary processes did not affect these elements (Fig. F56A, F56B).

Comparison of the basaltic flow units at Site 1137 shows few systematic downhole trends (Fig. F57). Mg numbers (Mg# = molar MgO/[MgO + FeO], where FeO is estimated as being 80% of the total iron [reported as Fe₂O₃] present) vary from 0.33 to 0.48 and show no systematic trend from Units 1 to 8. The lower sample from Unit 7 and the two samples from Unit 10 share a lower Mg# of 0.33 (Fig. F57A), surprisingly accompanied in Unit 10 samples by relatively higher Ni contents (59-61 ppm). Trace element ratios such as Zr/Y and Nb/Ce decrease down section (Fig. F57B). Relative to its host lava, a microgabbro xenolith from Unit 2 (Sample 183-1137A-26R-1 [Piece 12, 143-146 cm]) has significantly higher Mg#, CaO, and Cr and lower Ti, Zr, P₂O₅, and V contents (Table T11; Fig. F57A). The chilled margin of Unit 10 has much higher Y and Ce abundances than the other samples (Table T11; Fig. F57A), perhaps the result of interaction or contamination with sediment (see "Physical Volcanology").

The relatively evolved quartz-normative tholeiite of Unit 4, a 26.6-m-thick compound lava flow (see "Physical Volcanology"), has the largest internal variation in SiO₂ (50.6 to 54.0 wt%), Mg# (0.48 to 0.40), and TiO₂ (2.05 to 2.52 wt%). Units 8 and 10 show higher Mg#, Cr, and olivine abundances near flow margins but no significant variation in Ni (Figs. F35, F36; Table T11); the absence of olivine in the interiors of these flows (Fig. F36) may reflect olivine reacting out as the melt evolved during cooling.

Comparison with Other Sites

Site 1137 basalts differ from Cretaceous basalts recovered from other sites on the Kerguelen Plateau (Figs. F54, F55, F58, F59) in several significant ways. In general, Site 1137 basalts have unusually high concentrations of highly to moderately incompatible trace elements and slight depletion of Nb relative to Ba and Ce. Compared to Site 1136, concentrations of Ba, Nb, Ce, and Zr in most 1137 basalts exceed those of 1136 by a factor of two (Fig. F58, F59). Site 1137 basalts are in the low MgO, high SiO₂ and TiO₂ portion of the overall concentration range for Kerguelen Plateau basalts, but do not extend to the very low MgO contents of Site 738 basalts (Fig. F55). In comparison to all other Kerguelen Plateau basement sites, Site 1137 basalts have distinctly higher Zr, Nb, and Ce contents but comparable Y contents; only basalts from Sites 738, 747, and 1137 have distinctly high Ba (Fig. F58). The degree of trace element enrichment in Site 1137 basalts does not, however, reach the levels observed in the younger, more alkaline rocks on Heard Island, Kerguelen Archipelago, and Site 748 (see "Previous Sampling of Igneous Basement: Ages and Geochemical Characteristics" in "Study Area" in the "Leg 183 Summary" chapter). However, compared to other Cretaceous basalts from the Kerguelen Plateau, basalts from Sites 738 and 1137 have the highest Zr/Ti and Zr/Y (Fig. F60).

In primitive mantle-normalized incompatible element diagrams (Figs. F56, F59), the Site 1137 basalts show moderate negative Nb anomalies. A second perspective on relative Nb depletion (Fig. F60) shows that Site 1137 basalts have Nb/Ce similar to Site 747 and the high Zr/Ti ratios that characterize basalts from Sites 738 and 747. As described in "Previous Sampling of Igneous Basement: Ages and Geochemical Characteristics" in "Study Area" in the "Leg 183 Summary" chapter, similar ratios have been used together with isotope data as evidence for a continental crustal component in these basalts.

Fitton et al. (1997, 1998) used Nb/Y vs. Zr/Y to distinguish Icelandic plume-related basalts from depleted mid-ocean ridge basalt (MORB). In such a plot (Fig. F61), Kerguelen Plateau basement basalts range from the plume field (Sites 749 and 750) to near the plume-MORB boundary (Sites 1136 and 747) to well within the MORB field (Sites 738 and 1137). The Site 1137 lavas trend to the highest Nb/Y and Zr/Y; despite plotting within the MORB field, these lavas do not have the incompatible element depletion that characterizes MORB (cf. Fig. F56). Two possible interpretations are (1) contamination of plume-derived lavas with continental crust as proposed for Sites 738 and 747 and discussed above or (2) the mantle source of Site 1137 lavas had lower Nb/Y at a given Zr/Y compared to the source of basalts from Iceland.

Felsic Volcanic Rocks

Three of the analyzed volcanic rocks have nonbasaltic compositions (Table T12). Two are sanidine-phyric clasts from the Unit 6 volcanic conglomerate, and the other is a sample from the crystal-vitric tuff of Unit 9. We also analyzed a coarse layer in the lithic volcanic sandstone (Unit 5), which is composed almost entirely of lithic volcanic fragments and phenocrysts (Table T10, T12).

Moderate to high SiO₂ contents (64.9 to 73.5 wt%), total alkalis (~9.5 wt%), and high normative quartz and alkali feldspar abundances all reflect the evolved nature of these clasts. In petrologic classification diagrams such as total alkalis vs. SiO₂ (Fig. F54B) and normative quartz-orthoclase-albite (not shown), the sanidine-plagioclase-clinopyroxene-phyric clast (Sample 183-1137A-33R-5 [Piece 1, 7-9 cm]) is a trachyte and the sanidine-phyric flow-banded clast (Sample 183-1137A-34R-2 [Piece 1D, 111-113 cm]) is a rhyolite. The crystal-lithic volcanic sandstone has a very similar major element composition to the trachyte, suggesting that the lithic volcanic fragments are dominantly intermediate in composition. The crystal-vitric tuff exhibits an intermediate composition plotting on the line dividing andesite and trachyandesite. However, the tuff sample is highly altered (LOI = ~7%), and the high crystal content suggests that crystal enrichment may have resulted from pyroclastic and/or sedimentary processes (see "Physical Volcanology"). The phenocryst assemblage (sanidine >> plagioclase, minor quartz) is consistent with a highly evolved bulk composition, although the likely sedimentary redeposition of the tuff allows for incorporation of xenocrysts and other complications.

These samples have very high concentrations of incompatible trace elements (e.g., Zr from 582 to 1274 ppm, Ce from 164 to 373 ppm) (Table T12). For these evolved rocks, the mantle-normalized trace element patterns are distinctive, with levels of Rb, K, Nb, Ce, and Zr that are three to five times higher than in the basalts, and highly variable Ba abundance (Fig. F56E). The rhyolite and trachyte are relatively depleted in Sr and Ti, probably because of fractional crystallization of plagioclase and Fe-Ti oxides, respectively. The trace element compositions are very similar to those of Miocene to Holocene trachytes from Heard Island and the Kerguelen Archipelago (Fig. F62).

The evolved volcanic rocks at Site 1137 share some of the distinctive chemical characteristics of the basalts, such as high Zr/Y ratios (Fig. F61), but do not exhibit a Nb anomaly on a primitive mantle-normalized plot (Fig. F56E). The tholeiitic basalts at Site 1137 are not suitable parental magmas for the trachytes because low-pressure fractional crystallization does not produce alkalic magmas such as trachytes from silica-oversaturated parents, and magmas with relatively high Nb/Ce are not normally formed from parental magmas with low Nb/Ce.

Garnet-Biotite Gneiss

Sample 183-1137A-35R-2 (Piece 5, 44-46 cm) (Table T12) is a garnet-biotite gneiss clast from the Unit 6 conglomerate. With 73.3 wt% SiO₂ and 5.6 wt% Na₂O + K₂O, the sample is a granite (i.e., in a total alkalis-silica classification diagram, it is in the rhyolite field) (Fig. F54). The rock is metaluminous but contains corundum in the norm and has a relatively low Na₂O content. The primitive mantle-normalized trace element pattern for the garnet-biotite gneiss shares the low Sr and Ti contents of the trachytic and rhyolitic samples but is distinguished by the large negative Nb anomaly characteristic of continental crust (Fig. F56E). As seen on a Nb/Y vs. Zr/Y diagram (Fig. F61), the composition of the gneiss is distinct from average continental crust estimates, such as those of Rudnick and Fountain (1995).

Source Implications

Three distinctive aspects of the chemical compositions of basalts from Site 1137 hint at their origin: (1) the relative depletion of Nb (Fig. F56A, F56B, F56C, F56D), (2) the high concentrations of incompatible trace elements (Fig. F58), and (3) the fractionation of highly incompatible elements from moderately incompatible elements (e.g., Zr from Y) (Fig. F60).

The relatively low Nb contents indicate that Site 1137 basalts contain a component from the continental lithosphere. In the absence of other constraints, evidence for such a component in the source of a basalt erupted in an oceanic environment could indicate that the component was an integral part of the mantle source of the basalt. However, given the firm evidence of continental crust in the vicinity (clasts of garnet-biotite gneiss in the conglomerate and tuff, and detrital garnet in the sandstone), it is more likely that this component was introduced as a contaminant as the basalts passed through continental crust.

The presence of a continental component in the Site 1137 basalts may have influenced the relative abundances of elements other than Nb. However, the addition of continental crust to magmas with relatively high concentrations of incompatible elements, such as the basalts from Site 1137, would have only a minor effect on abundance ratios of elements such as Zr and Y because the Zr/Y ratios of continental crust and enriched magmas, such as those of Site 1137, are similar (Fig. F61). We conclude that the Zr/Y ratio in Site 1137 basalts is little affected by the crustal component and that the primary magma for Site 1137 basalt probably had a high Zr/Y ratio (i.e., ~3) (Fig. F61). This feature was either inherited from a mantle source that was relatively enriched in incompatible elements or was derived by degrees of melting that were lower than those that produced the other tholeiitic basalts of the Kerguelen Plateau. Investigating these hypotheses—the extent of continental contamination, variation in source compositions, and the degree of partial melting—will be the focus of shore-based geochemical studies on the origin of basalt at Site 1137.