Continental
rifting and mantle upwellings are key components in our understanding of plate
tectonics and the asthenospheric mantle. While volcanism is common as rifts
mature from continental rifting to ocean-spreading, our current understanding
of how volcanism changes during rift evolution is limited. The Afar triangle,
in East Africa, is a rift-rift-rift system that meets in a triple junction and
has been proposed to be underlain by a mantle plume. Given that Afar has
already undergone extensive rifting, yet remains subaerial, the region presents
a unique opportunity to investigate the physical and chemical characteristics
of volcanism associated with a proto-oceanic rift. Here, the volcanic
evolution of three volcanoes, Alu, Dalafilla and Borale, are characterised
through a combination of detailed remote-sensing mapping, petrology, major
element, trace element and radiogenic isotope geochemistry. The approximate
eruptive frequency and style is constrained as well as highlighting the changes
in melt production and storage over the recorded eruptive history of the
volcanoes. The results show that each of the volcanoes has a compositional
cyclicity (basalt-trachy-andesite/rhyolite), which is likely controlled by the
time required for fractional crystallisation to occur in an interconnected
stacked-sill system (∼1-4 km depth). The variation in the melting extent and
depth among the basalts can be attributed to potential variations in the
rifting rate throughout the evolution of the volcanoes. Following this,
an integrated approach combining geochemistry (>100 new observations),
geophysics (shear wave velocities and Moho depth) and statistical modelling is
applied to investigate the melt production beneath the region, assessing the
spatial and chemical characteristics of mantle upwelling across the region. The
statistical modelling shows the favoured predictive model to explain melt
characteristics observed is a single heterogeneous plume, which is asymmetric
around the triple junction. This, combined with results from K-means cluster
analysis, indicate that different spreading rates across each rift arm may be
the cause of the plume’s asymmetry.