Evidence for melt channelization in Galapagos plume-ridge interaction

Abstract:

Many present-day hot spots are located within ~ 1000 km of a mid-ocean ridge, either currently or in the geologic past, leading to frequent interaction between these two magmatic regimes. The consequent plume-ridge interactions provide a unique opportunity to test models for asthenosphere-lithosphere dynamics, with the plume acting as a tracer fluid in the problem, and excess magmatism reflecting otherwise unsampled sub-surface phenomena.

Galapagos is an off-ridge hotspot with the mantle plume located ~150-250 km south of the plate boundary. Plume-ridge interaction in Galapagos is expressed by the formation of volcanic lineaments of islands and seamounts – e.g., the Wolf-Darwin lineament (WDL) – providing a direct probe of the plume-ridge interaction process, especially in regards to geochemical data. Although several models have been proposed to explain plume-ridge interaction in Galapagos, none adequately explain the observed characteristics, especially the WDL. In particular, predicted lithospheric fault orientations and melt density considerations appear at odds with observations, suggesting that lithospheric extension is not the primary process for formation of these islands. Other off-ridge hotspots interacting with nearby spreading ridges, such as Reunion and Louisville, also exhibit volcanic lineaments linking the plume and the ridge. Thus these lineament-type features are a common outcome of plume-ridge interaction that are indicative of the underlying physics. We propose that the lineaments are surface expressions of narrow sub-lithospheric melt channels focused towards the spreading ridge. These channels should form naturally due to the reactive infiltration instability in a two-phase flow of magma and solid mantle as demonstrated in two-phase flow simulations (e.g., Katz & Weatherley 2012). For Galapagos, we show that melt channels can persist thermodynamically over sufficient length-scales to link the plume and nearby ridge segments. We also show that the relevant timescales for magma movement through these channels are sufficiently brief to account for observed U-Th systematics along the ridge (Kokfelt et al. 2005). We suggest more generally that plume-ridge interaction may be dominated by transport of magma in melt channels rather than by solid-state mantle flow.