Cordilleran orogens, such as the central Andes, form above subduction zones, and their evolution depends on both continental shortening and oceanic plate subduction processes, including arc magmatism and granitoid batholith formation. Arc and batholith magma compositions are consistent with partial melting of continental lithosphere and magmatic differentiation, whereby felsic melts rise upward through the crust, leaving a high-density pyroxenite root in the deep lithosphere. We study gravitational removal of this root using two-dimensional thermal-mechanical numerical models of subduction below a continent. The volcanic arc position is determined dynamically based on thermal structure, and formation of a batholith-root complex is simulated by changing the density of the arc lithosphere over time. For the model lithosphere structure, magmatic roots with even a small density increase are readily removed for a wide range of root strengths and subduction rates. The dynamics of removal depend on the relative rates of downward gravitational growth and lateral shearing by subduction-induced mantle flow. Gravitational growth dominates for high root densification rates, high root viscosities, and low subduction rates, resulting in drip-like removal as a single downwelling over 1–2.5 m.y. At lower growth rates, the root is removed over >3 m.y. through shear entrainment as it is carried sideways by mantle flow and then subducted. In all models, >80% of the root is removed, making this an effective way to thin orogenic mantle lithosphere. This can help resolve the mass problem in the central Andes, where observations indicate a thin mantle lithosphere, despite significant crustal shortening and thickening.

This study explores a conceptual model for mantle convection in which buoyant and low-viscosity asthenosphere is present beneath the relatively thin lithosphere of ocean basins and regions of active continental deformation, but is less well developed beneath thicker-keeled continental cratons. We start by summarizing the concept of a buoyant plume-fed asthenosphere and the alternative implications this framework has for the roles of compositional and thermal lithosphere. We then describe the sinks of asthenosphere made by forming compositional lithosphere at ridges, by plate cooling wherever the thermal boundary layer extends beneath the compositional lithosphere, and by drag-down of buoyant asthenosphere along the sides of subducting slabs. We also review the implied origin of hotspot swell roots by melt-extraction from the hottest portions of upwelling plumes, analogous to the generation of compositional lithosphere by melt-extraction beneath a spreading center. The plume-fed asthenosphere hypothesis requires an alternative to “distinct source reservoirs” to explain the differing trace element and isotopic characteristics of ocean island basalt (OIB) and mid-ocean ridge basalt (MORB) sources; it does so by having the MORB source be the plum-depleted and buoyant asthenospheric leftovers from progressive melt-extraction within upwelling plumes, while the preferential melting and melt-extraction of more-enriched plum components is what makes OIB of a given hotspot typically fall within a tubelike geometric isotope topology characteristic of that hotspot. (The distinct plum components result from the subduction of chemically differing sediments, basalts, and residues to hotspot and mid-ocean ridge melt extraction.) Using this conceptual framework, we construct a thin-spherical-shell finite element model with a ∼100-km-scale mesh to explore the possible structure of global asthenosphere flow. Lubrication theory approximations are used to solve for the flow profile in the vertical direction. We assess the correlations between predicted flow and geophysical observations, and conclude by noting current limitations in the model and the reason why we currently neglect the influence of subcontinental plume upwelling for global asthenosphere flow.

Asthenosphere plume-to-ridge flow has often been proposed to explain both the existence of geochemical anomalies at the mid-ocean ridge segments nearest an off-axis hotspot and the existence of apparent geochemical provinces within the global mid-ocean spreading system. We have constructed a thin-spherical-shell finite element model to explore the possible structure of global asthenosphere flow and to determine whether plume-fed asthenosphere flow is compatible with present-day geochemical and geophysical observations. The assumptions behind the physical flow model are described in the companion paper to this study. Despite its oversimplifications (especially the steady-state assumption), Atlantic, Indian, and Pacific mid-ocean ridge isotope geochemistry can be fit well at medium and long wavelengths by the predicted global asthenosphere flow pattern from distinct plume sources. The model suggests that the rapidly northward-moving southern margin of Australia, not the Australia-Antarctic discordance, is the convergence zone for much plume material in the southern hemisphere. It also suggests a possible link between the strike of asthenosphere flow with respect to a ridge axis and along-axis isotopic peaks.