THIS PLUME BUSINESS

Seismology has yet to resolve narrow (<200 km) deep plume conduits beneath active ends of volcanic chains. Instead, improved tomography (1, 2) confirms that broad regions (1000-2000 km wide) of reduced velocity in the shallow mantle parallel, but at depth are often offset from, volcanic ridges (3) and originate in the transition zone (4). These mantle updrafts rise slowly or are neutrally buoyant, and likely are upper-mantle responses to displacements caused by subduction and delamination of continental crust and underlying mantle. Conduits, if they exist, must be within these updrafts, which do not extend to >1000 km. Mass-balances suggest that updrafts rise at <several cm/yr, whereas the same volume of mantle in ~20 narrow plume conduits requires several m/yr. Updrafts are not smeared out, narrow, intense conduits.

The most important geochemical consequence of plate-driven convection is that the sources of both MORB and OIB are contained within updrafts or are acquired in the 200-km-thick boundary layer (BL) with which they interact. There, they are comingled by shear. Low-degree partial melts preferentially incorporate enriched components from the surrounding matrix. E-MORB erupt at spreading ridges and near-ridge seamounts. From there, enriched melts tend to concentrate beneath plates, so that seamount lavas become more enriched away from ridge axes. Rising enriched melts shear into the low-velocity sub-lithospheric mélange as plates cool, thicken, subside and age. Buoyant, delaminated residual harzburgite is concentrated along the crests of both sub-ridge and mid-plate updrafts and is responsible for wide linear island-chain swells and focusing of enriched lavas on islands, moats and arches. Updrafts at ridges do not interact with mature plates and are thus less enriched at their tops. Plate thickening drives depths of partial melting downward, so that mafic lavas with varying degrees of enrichment characterize all stages of volcanism at island chains. Mafic lava on thick plates is more variable than on thin plates. Within a BL, stress-driven melt segregation may be as important as buoyant flow.

1. Styles,E., et al., 2011. EPSL 311: 351–363.

2. French, S.W. et al., 2013. Science 342: 227-230.

3. Anderson, D. L., et al., 1992. Science 256: 1645–1650.

4. Ritsema, A., & Allen, R.M. (2003). EPSL 207: 1-12.