The global ridge system is dominated by oceanic rises reflecting large variations in axial depth associated with mantle hotspots. The little studied Marion Rise is as large as the Icelandic, considering length and depth, but has an axial rift rather than a high nearly its entire length. Uniquely, along the SW Indian Ridge systematic sampling allows direct examination of crustal architecture over its full length. Unlike Iceland, peridotites are extensively exposed high over the rise. This shows for the 1 st time that the crust is generally thin, and often missing over a rifted rise. Thus the rise must be largely an isostatic response to ancient melting events that created low-density depleted mantle beneath the ridge rather than thickened crust and/or a large thermal anomaly. The likely origin for the depleted mantle is that emplaced into the African asthenosphere during the Karoo and Madagascar flood basalt events.
Following Morgan1 , common wisdom equates oceanic rises to a hot fertile mantle plume producing a flow of plume-derived mantle to the ridge and down the sub-axial asthenospheric channel, resulting in elevated mantle temperature, ridge topography, and thickened igneous crust (the mantle wedge hypothesis, e.g.:2 ). Studies of the Reykjanes Ridge support this, with seismic crust thickening from ~6 km near the Gibbs FZ for 1600 km to ~18 km at the Reykjanes Peninsula e.g.: 3 . The Icelandic Rise, also has a long axial high consistent with such robust magmatism 4 (Table 1). The current consensus is the igneous crust thickens from east to west up the Marion Rise, and indeed up all rises (e.g.: 5 6 ). Basalt sodium contents, like other rises, decrease systematically towards Marion, and are interpreted to represent higher degrees of mantle melting and thicker crust 7 . Dick et al. 8 , however, point out that such correlations can be due to variable mantle temperature, or an increasingly depleted mantle source composition, and do not require thicker crust. The Marion Rise, in particular, has numerous large-offset transforms that would block sub-axial asthenospheric flow (e.g.: 9 ), and its deep rift valley indicates anemic rather than robust magmatism 4 . Niu and O'Hara 10 , though assuming thick crust over rises, suggest the global correlation of basalt chemistry and ridge depth is best explained by mantle composition variations, with many rises supported by depleted chemically buoyant mantle as first proposed by O'hara 11 for the Iceland Rise and Presnall and Helsley 12 for the Azores.2 14 . c) Maximum and minimum elevations are measured from shallowest point on axis, or midpoint of Iceland plateau to the rift valley axis point where average depth stabilizes or blocking transform. d) The 200-km average represents average elevation about the highest or lowest point. e) Iceland average plateau height; crest of Azores Rise, SWIR @ 36°15Ed) MAR @ Charlie Gibbs FZ; axial valley floor, MAR @ 24°N, axial valley floor @61°45'E. e) Depth anomaly calculated for the 200-km averages to eliminate local topographic e...