The geochemistry and petrogenesis of lavas from the Comores Archipelago, Western Indian Ocean

Master Thesis

1994

Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
License
Series
Abstract
The islands of the Comores Archipelago define a short, NNW-trending, volcanic lineament across the northern entrance to the Mozambique Channel in the western Indian Ocean and are considered to represent the surface expression of the Comores mantle plume. New mineral and bulk-rock analyses, as well as selected rare earth element (REE), 87Sr/86Sr and 143Nd/144Nd compositions for lavas from three of the four main islands of the archipelago are presented. The alkaline lavas of the Comores range in texture from sparsely to strongly porphyritic, containing phenocrysts of olivine, clinopyroxene, amphibole, feldspar, Fe-Ti oxide and in one sample nepheline and garnet, which are typically set in a fine-grained groundmass of clinopyroxene, feldspar or nepheline, Fe-Ti oxide, lesser olivine and accessory apatite and titanite (sphene). With the exception of a few hypersthene- and even quartz-normative samples, all of the rocks analysed are moderately to strongly silica-undersaturated, nepheline-normative lavas. On Grande Comore, the youngest island in the archipelago, a clear distinction exists between the generally alkali basaltic lavas erupted by Karthala volcano and the basanites of La Grille volcano. The samples from the island of Moheli are classified as alkali basalts, basanites and nephelinites. The lavas of Mayotte, the oldest Comorean island, were erupted during three phases of volcanic activity and display greater chemical variation than both Grande Comore and Moheli lavas. Mayotte lavas range in composition from alkali basalt through trachybasalt, basaltic trachyandesite and trachyandesite to trachyte, and from basanite and nephelinite through phonotephrite to phonolite. Comorean lavas have high incompatible element abundances and display strongly light rare earth element-enriched chondrite-normalised REE patterns. On the basis of their primitive mantle-normalised incompatible element patterns (spidergrams), all of the lavas analysed may be assigned to one of two very distinct Comorean lava types: La Grille-type (LGT) lavas display strong relative depletions in K (and sometimes Rb), whereas Karthala-type (KT) lavas do not show such depletions. Both LGT and KT lavas were erupted on Grande Comore, Moheli and Mayotte. With the exception of the lavas erupted by La Grille volcano, which exhibit the petrographic and geochemical characteristics expected of primary mantle magmas, all Comorean lavas analysed have experienced compositional modifications after they segregated from their source regions. Quantitative major and trace element modelling suggests that much of the compositional variation observed amongst Comorean lavas is explicable in terms of fractional crystallisation processes dominated by the early fractionation of olivine and clinopyroxene. Plagioclase appears to have been an additional fractionating phase during the evolution of Moheli basanites. The more advanced stages of the differentiation history of Mayotte lavas were dominated by the fractionation of increasing quantities of feldspar and amphibole, as well as smaller amounts of clinopyroxene, Fe-Ti oxide and accessory apatite and titanite. The presence of high-pressure clinopyroxene, amphibole and possibly garnet crystals in Comorean lavas suggests that the crystal fractionation processes involved in their differentiation may have been initiated at relatively elevated pressures. Partial melting in the presence of residual amphibole is proposed to be the most likely mode of origin for the K-depletions which characterise LGT Comorean lavas. It is suggested that primary LGT magmas were generated by small degrees of partial melting of an amphibole-bearing garnet-lherzolite mantle source at depths corresponding to pressures greater than 25 kb. Primary KT magmas, on the other hand, are considered to be the product of somewhat larger degrees of partial melting of an amphibole-free spinel-lherzolite source at shallower depths. The Nd and Sr isotopic compositions of Comorean lavas bear evidence for a timeaveraged depletion in incompatible elements, whereas the high incompatible element abundances of these lavas are proposed to reflect the effects of a recent mantle enrichment event. It is suggested that the garnet-lherzolite source of LGT Comorean lavas experienced recent modal metasomatism, resulting in the precipitation of amphibole, whereas the shallower spinel-lherzolite source of KT lavas underwent cryptic metasomatism only, without the introduction of new minerals. The ambient subComorean upper mantle is proposed to consist of a mixture between the HIMU mantle component and a depleted component (DMM, PREMA or a component on the mantle array). The Nd and Sr isotope signature of the majority of Comorean lavas (both LGT and KT) may be explained as a mixture dominated by the components comprising the sub-Comorean mantle with limited contributions from the EM I-type Comorean mantle plume itself. The lavas erupted by Karthala volcano (the youngest Comorean lavas), however, have significantly different isotopic compositions than all other Comorean lavas (lower 143N d/144N d and higher 87S r / 86S r ), suggesting significantly increased contributions from the Comores mantle plume. An alternative model, equally consistent with the present data, involves an ambient sub-Comorean mantle consisting of a depleted component that is mixed with a Comores mantle plume which contains both HIMU and EM I components. In this scenario, the character of the plume would have to change from one dominated by the HIMU component during most of the volcanic history of the Comores, to one dominated by the EM I component during the generation of the lavas erupted by Karthala. An internally consistent, broadly chronological petrogenetic model for the evolution of Comorean volcanism is presented.
Description

Reference:

Collections