Tertiary flood basalts in the Ethiopian Province are closely associated with the presence of the Afar mantle plume, extension in the Ethiopian Rift and continental breakup. Because this province is currently active it presents an ideal area to assess the roles of extension and mantle plumes in melt generation, the influence of the mantle lithosphere on basalt composition and the rates of melt generation in continental basalt provinces. This work focuses on new results from southern Ethiopia where magmatism spans the entire 45Ma magmatic history of the Ethiopian Province.
Basaltic magmatism in the Southern Rift of Ethiopia both pre-dates and is synchronous with extension. New Ar-Ar ages confirm previous K-Ar ages of 35-45Ma for pre-rift transitional tholeiites which have a stratigraphic thickness of 0.5-1km. After a 15-20My hiatus, these were followed by the syn-rift alkaline basalts which were erupted between 18 and 11Ma and are of a similar thickness. Localised alkali basaltic volcanism continues into Recent times at accommodation zones between border faults. Basalts from these stratigraphic divisions have distinct trace element characteristics. For example, the pre-rift transitional tholeiites have Zr/Nb ratios of 7.0-14.4 whereas the syn-and post-rift lavas both have lower values of 2.0-4.6. The pre-rift lavas also have high La/Nb and low Th/Tb, Nb/Y and La/Yb ratios relative to the syn- and post-rift basalts and the greater incompatible element enrichment of the syn-rift basalts suggests that they are the products of smaller degrees of melting. New Sr and Nd isotope data confirm this sub-division. The high Zr/Nb, pre-rift lavas have higher 87Sr/86Sr (0.7036-0.7046) and lower 143Nd/144Nd ratios (0.5125-0.5127) than the later basalts (87Sr/86Sr; 0.7032-0.7039 and 143Nd/144Nd; 0.5126-0.5128).
Observed variations between the two magma types cannot be easily reconciled by crustal contamination. Although a degree of crustal contamination is implied by elevated Th/Ta ratios of 2-2.5 in the oldest pre-rift basalts, the bulk of the sequence have higher La/Nb ratios (>1) and lower Th/Nb ratios (<1.15) than can be generated through mixing of a plume-derived melt with upper continental crust. Moreover trace element ratios similar to those found in the lower crust preclude this as a contaminant as this would require 80-100% assimilation, a value that is clearly at odds with the major element composition of the basalts.
The preferred interpretation is that the two magma types represent melting of two discrete mantle sources. The low Zr/Nb magma type shows similar trace element characteristics to ocean-island basalts (for example, La/Nb). This, accompanied with higher 143Nd/144Nd ratios than the pre-rift basalts, strongly suggests a mantle plume source for the syn-rift basalts. By contrast, the pre-rift basalts bear trace element abundances unlike any oceanic counterparts, and have lower 143Nd/144Nd ratios. However, La/Nb ratios approach those seen in the Parana CFB, and this observation lends support to the view that they were derived from the continental mantle lithosphere.
Melt production is clearly not continuous over the whole 45My period of magmatic activity and the observed variations in the pre-rift sequence appear to be dominated by lithospheric melts with only small melt fractions from the mantle plume. These observations essentially rule out a simple model of decompression melting of the underlying mantle plume to explain all of the basaltic magmatism in the Ethiopian Province. An alternative mechanism for generating the pre-rift basalts is one of conductive heating of the lithosphere by the mantle plume. This would explain their predominantly lithospheric compositional features and provides a mechanism for thinning the lithosphere prior to extension and eruption of the syn-rift basalts. Syn-rift magmatism, by contrast, is dominated by melts derived largely from the plume itself, hence the magmatism between 18 and 11 My can be adequately explained by decompression melting of the mantle plume associated with extension.
Melt production rates calculated for the Ethiopian province are in the region of 0.025-0.03 km3/y-1 for both the pre-rift and syn-rift sequences. These rates are similar to those determined for many oceanic mantle plumes, but are a factor of ten less than those predicted for other CFB such as the Deccan. This difference may reflect the contrast between the thermal budgets of impacting plume heads and those of already established plumes and suggests that magmatism in Ethiopia is not the consequence of plume impact but of the movement of the Afro-Arabian plate over the site of the already active Afar plume ~50My ago.