A plume origin for hydrous melt at the lithosphere-asthenosphere boundary

Plate tectonics requires a low-viscosity layer beneath the lithosphere-asthenosphere boundary (LAB), yet the origin of this ductile transition remains debated(1,2). Explanations include the weakening effects of increasing temperature(3,4), mineral hydration(5 )or partial melt(6). Electrical resistivity is sensitive to all three effects(7), including melt volatile content(8), but previous LAB constraints from magnetotelluric soundings did not simultaneously consider the thermodynamic stability ofthe inferred amount of melt and the effect of uncertainty in the estimated resistivity(8-14). Here we couple an experimentally constrained parameterization of mantle melting in the presence ofvolatiles(15,16) with Bayesian resistivity inversion(17) and apply this to magnetotelluric data sensitive to a LAB channel beneath the Cocos Plate(9). Paradoxically, we find that the conductive channel requires either anomalously large melt fractions with moderate volatile contents or moderate melt fractions with anomalously large volatile contents, depending on the assumed mantle temperature. Large melt fractions are unlikely to be mechanically stable and conflict with melt-migration models(18). As large volatile contents require a highly enriched mantle source inconsistent with mid-ocean-ridge estimates(19), our results indicate that a mantle plume emplaced volatile-rich melts in the LAB channel. This requiresthe presence of a previously undetected nearby plume or the influence ofthe distant Galapagos hotspot. Plumesthat feed thin, hydrous melt channels(9,14,20) may be an unrecognized source of LAB anomaliesglobally.