Geochemical variations at ridge-centered hotspots caused by variable melting of a veined mantle plume

We model the dynamics and melting of a ridge-centered mantle plume, and predict the geochemical composition of magma at the surface. The mantle source is a fine-scale mixture of a small fraction of hydrous peridotite that is relatively enriched in incompatible elements (“EC”) and is embedded in a drier peridotite (“DC”) matrix. We assume all magma erupts at the ridge and calculate the contribution of EC and DC to the pooled composition along the ridge. If viscosity increases as melting dehydrates the mantle, EC contributes more to the pooled magma at the hotspot center than anywhere else along the ridge. The magnitude of this EC anomaly increases with Rayleigh number, and the along-axis distance to normal ridge composition increases with Rayleigh number, plume radius, and thermal buoyancy flux. A subset of model calculations designed to simulate the Iceland hotspot and Mid-Atlantic Ridge predict variations in crustal thickness, 87Sr/86Sr, and La/Sm with magnitudes and widths along the ridge that are comparable to, but less than, those observed. Improved fits to the observations require the innermost plume mantle to be compositionally distinct from the ambient asthenosphere; for example, by having a slightly higher mass fraction of EC (13-16%), or with DC having slightly higher 87Sr/86Sr and La/Sm. The inferred bulk plume 87Sr/86Sr composition, however, is within the predicted range of the source of normal mid-ocean ridge basalts worldwide. The broader implication is that the source of the Iceland plume is more similar in composition to the ambient upper mantle than previously thought, as a large part of the variation in ridge basalt composition can be attributed to the dynamics of mantle flow and melting.