THE TIMING OF MELT INITIATION AND PEAK MELTING IN EVOLVING CONTINENTAL RIFTS: INSIGHTS FROM 1-D GEODYNAMIC MODELS OF THE EAST AFRICAN RIFT SYSTEM

Geochemical and isotopic signatures of magmatism spanning both the eastern and western branches of the East African Rift System (EARS) suggest that sources of syn-rift magmas vary with time. The earliest magmas erupted during nascent rifting are likely produced by melting of fertile lithospheric metasomes while later magmatism occurring in more mature rifts are the result of asthenospheric derived melts. Transitional magmas produced in the intervening period may reflect a mixture of melts from the early- and late-stage sources. Potential fertile sources in the mantle (pyroxenite, eclogite, and amphibole bearing metasomes as well as wet lherzolite) responsible for the EARS geochemical signatures have been proposed to result from metasomatism. We test the viability of a multi-lithology model consisting of fertile metasomatic sources (pyroxenite, eclogite, or wet lherzolite) embedded in the lithospheric and/or asthenospheric mantle (dry lherzolite) using 1-D geodynamic models which calculate the pressure-temperature conditions within an extending lithosphere and vertically ascending underlying asthenosphere through time. The model estimates the amount, rate, and timing of melt generated from different source compositions in the lithospheric and asthenospheric mantle under extension scenarios similar to those in the EARS. Results predict that in the EARS, dry lherzolite mantle would begin melting at ~50 % extension and reach peak melt production at >110 % extension unless the mantle was hot (Tp ≥ 1500 °C) or the pre-rift lithosphere was thin (<100 km). In such cases, the onset of melting may begin by <10 % extension with peak melt production occurring by 60 % extension. With the exception of the models with very hot asthenosphere or thin pre-rift lithosphere, the time elapsed between the onset of rifting and the onset of melting for fertile sources (whether in the lithosphere or asthenosphere) predates melting of dry lherzolite by at least 50% extension. The fertile phases also reach peak melt production before dry lherzolite by at least 35 % extension. These results support the proposed metasomatic petrogenic models of the EARS, and suggest that melting beneath juvenile rifts may be dominated by fertile metasomes in the lithospheric mantle, and late stage extension dominated by less fertile sources.