M. Grégoire

GEMOC, Macquarie

The ultramafic and mafic xenoliths from the Kerguelen Islands occur in dykes, lava-flows and breccia pipes of the youngest and more alkaline magmas. They range in size from a few to 60 cm and most are very fresh. They include many types of ultramafic and mafic xenoliths entrained by within-plate basalts in continental and oceanic settings (except eclogites and garnet peridotites). Two main types can be distinguished. The first type consists of spinel-bearing harzburgites and dunites and associated composite xenoliths representing mantle wall-rock. The second type comprises a wide variety of metamorphosed igneous rocks ranging from peridotites to basic granulites. Three main mineralogical sub-types have been distinguished for the type II xenoliths. The first (IIa) is the clinopyroxene+orthopyroxene+spinel series which represents 45 % of the whole collection of xenoliths. This series of xenoliths always contains cpx, opx and spinel and shows a regular modal evolution from ultrabasic rocks such as wehrlites, clinopyroxene-rich lherzolites, websterites with or without olivine and plagioclase, to mafic rocks such as metagabbros with or without garnet and sapphirine. The second subgroup (IIb) is the clinopyroxene+ilmenite+spinel series represented by clinopyroxenites with or without garnet and by garnet+spinel+ilmenite metagabbros. The third subgroup (IIc) consists of the clinopyroxene+ilmenite series essentially represented by garnet-bearing metagabbros. The type II xenoliths are deep crustal and upper mantle segregates from the mantle derived tholeiitic-transitional basaltic magmas (type IIa and Type IIc xenoliths) and the mantle derived alkaline basaltic magmas (type IIb xenoliths) of the archipelago.

The harzburgites (type I) were equilibrated in the spinel peridotite stability field (T = 850-1150 °C). They have been divided into protogranular Cr-diopside-bearing harzburgites and poikilitic harzburgites which contain an interstitial magnesian-augite, sometimes associated with phlogopite and amphibole. Some scarce samples of the two types of harzburgites show an unusual mineral association consisting of feldspar + olivine (2) + Ti-chromite + rutile + Mg-ilmenite + armalcolite + Ca-Cr armalcolite (FORIAC paragenesis). These minerals occur in reaction zones replacing opx and spinel or as thin veins or dykelets cross cutting olivines. Trace element characteristics of type I Kerguelen xenoliths point out the fact that the whole samples have been affected at various degree by metasomatic processes. Most of the samples show LREE-enriched patterns normalised to primitive mantle value and the few samples which display LREE-depleted patterns have a La/Sm ratio too high to be explain only by depletion processes. The feldspar of the FORIAC mineral paragenesis is variable in composition, but is alkali-rich (K2O: 1-10.5 wt%). Rutile and Ca-Cr-Zr armalcolite may contain significant amounts of Nb2O3 (0.30-1.85 wt%), ZrO2 (0.15-3.00 wt%) and Ce2O3 (0.07-1.45 wt%). Kerguelen harzburgites result of two main processes: (1) high degree of partial melting (20-30 %), and (2) multiple metasomatic processes. The partial melting event is well established by the high refractory indices of both whole rocks and minerals (Fo up to 92, Cr-rich spinel, high MgO and low CaO, Al2O3, Na2O and total Fe2O3 contents in bulk-rock analyses). The multi-metasomatic events are associated with the activity of the Kerguelen plume and explain the modification of this previously depleted mantle to varying degrees. The main metasomatic event resulting in trace element enrichment of the two types of harzburgite and the crystallization of Mg-augite ± phlogopite and amphibole is probably related to the percolation into the upper mantle of melts ranging in composition from lamprophyric to carbonatitic. The minor metasomatic event (FORIAC metasiomatism) is probably related to the percolation of a Ti- and alkali-rich, H2O-poor fluid. The two-stage process which explains the petrological and geochemical characteristics of Kerguelen harzburgites may be related to the origin and evolution of the Kerguelen archipelago: (i) partial melting is related to formation of the Kerguelen oceanic lithosphere in the vicinity of the South East Indian Ridge, (ii) reaction between the harzburgitic residue and melts related to the activity of the Kerguelen mantle plume in the within-plate setting of the islands.

All the type II plagioclase-bearing rocks have reequilibrated in the granulite facies, i.e. in the P-T conditions of the Kerguelen oceanic lower crust and upper mantle. These mafic granulites are the first occurrence ever reported in an oceanic environment (Grégoire et al., 1994). The existence of oceanic granulites beneath the Kerguelen islands is consistent with the presence of a thickened crust evidenced by the seismic studies (14-20 km, Recq et al., 1990; Charvis et al., 1995; Operto & Charvis, 1996). Furthermore the preliminary calculated Vp values for basic granulites are consistent with those observed in the low-velocity region beneath the oceanic crust (Vp = 7.2-7.5 km s-1). Thus, the first study of the Kerguelen ultramafic and mafic xenoliths supports crustal thickening consistent with previously recorded seismic velocity characteristics. The Kerguelen thickened oceanic crust is thus inferred to have been generated by intrusions of basalts at different levels of the lithosphere and this may explain the occurrence of basic granulites in this oceanic setting. The synergy of the young East-Indian Ridge and of the Kerguelen hot spot (Å 45 Ma ago) is considered to have caused voluminous tholeiitic-transitional magmas with resultant intrusion and formation of associated cumulates in the vicinity of the Moho. This process of crustal thickening by deep intrusions was later amplified by the extrusive volcanic sequences related to the longevity of the hot spot activity (∼ 45 Ma). The upper mantle/lower crust cumulates equilibrated to granulite facies conditions.