Abstract

The petrography, mineralogy, and geochemistry of a suite of lavas from the northwestern part of Epi Island in the Vanuatu Arc, southwest Pacific Ocean, are described. The more primitive members of this suite are rich in clinopyroxene phenocrysts and are strikingly similar to primitive lavas from Merelava Is. in the same arc. These primitive, clinopyroxene-rich lavas are designated arc ankaramites to differentiate them from primitive, olivine-rich arc picrites which also occur in this arc system. The primitive Epi lavas are shown to have evolved from low-K primary melts which were saturated in both olivine and clinopyroxene. The most Mg-rich olivine (mg-number 92·2) and clinopyroxene (mg-number 94·4) in the ankaramites represent cotectic crystallization with Cr-rich spinels. Initial plagioclase (An94) crystallized in equilibrium with olivine (mg-number 78–80) and the plagioclase-olivine cotectic path extends to mg-number 50 and An58. The ankaramitic parent magma composition is calculated from the most primitive olivine phenocryst composition and the liquid line of descent, and has 14·5% MgO, 11% A12O3, 14·8%CaO, 0·29% K2O, and flat REE patterns. The origin of this parent magma has been modelled with Ghiorso & Carmichael's (1985) program SILMIN. An assimilation model involving a clinopyroxenite or wehrlite assimilate and a low-K picrite host requires ca. 90% assimilate to match the phase chemistry and bulk-rock chemistry of the parental ankaramite. The required degree of superheating necessary to achieve this, and the apparent restriction of low-K picrites to Anatom Island in the far south of the arc, renders this model unsatisfactory. Partial melting models involving typical upper mantle lherzolite also fail to give satisfactory results, but partial melting of a wehrlite source (mg-number 87-88) with < 10% normative (mol.) orthopyroxene, at 5·10kb and 1325°C, closely matches the parental ankaramite composition. These results can be reconciled with melting of lower crustal cumulates by an ascending peridotite diapir, a hypothesis which accounts for the very low Ni contents of the parental melts and primitive phenocrysts. The more evolved lavas define two distinct assemblages: a relatively tight grouping of high-K andesites straddling the high-K-‘shoshonite’ boundary, characterized by low Zr/Rb (2·2) and high K2O/Na2O ratios (1·3–0·9), and a relatively coherent fractionation pathway to dacites straddling the ‘calc-alkaline’-high-K boundary, with Zr/Rb = 2·9 and K2O/Na2O=0·6. Numerical modelling demonstrates that the dacite trend is compatible with fractionation from an ankaramite parent, whereas the high-K andesites are incompatible with open- or closed-system fractionation from ankaramitic or picritic sources and may represent fractionated, hybrid magmas, largely derived from melting of lower crustal gabbros.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call