HOW THE EARTH WORKS

Ray Price has contributed greatly to understanding the evolution of mountain belts, which are complexly related to plate motions. So what drives plates? Popular notions are based on bad assumptions—lower mantle is unfractionated; convection, subduction, and “plumes” operate from core to surface; plates are propelled from below; net continental crust enlarges with time; oceanic heatflow is twice continental—that reflect disconnects between disciplines and are incapable of accounting for actual plate motions and interactions and for observed crust and mantle petrology and evolution. Plate convergence commonly is wrongly viewed as crumpling of an overriding plate against a fixed hinge over which oceanic lithosphere rolls and slides down a slot.

Despite its rapid spreading, the Pacific shrinks slowly (at about the rate the Atlantic, almost subduction-free, enlarges) because bounding hinges roll back into it as slabs sink more steeply than they dip. Oceanic lithosphere is formed by top-down cooling of asthenosphere, and subduction is the falling away of this lithosphere to right the resulting density inversion. Forearc basins atop thin leading edges of most overriding plates show that little or no plate-front crumpling occurs, except in collisions. Slabs younger than about 60 m.y. when they begin to sink mix into the middle part of upper mantle, whereas older slabs are plated down on the uncrossable(?) 650 km discontinuity. Broadside-sinking slabs push all sublithospheric mantle above 650 km back under subduction-bounded oceans, forcing rapid internal spreading, and suck overriding arc or continental lithosphere forward to follow retreating hinges. Downplated subducted lithosphere is overpassed like basal tank treads and recycled in nonsubducting oceans, which spread slowly because net transfer to them consists only of subducted lithosphere and entrained material. Ridges necessarily migrate to tap fresh asthenosphere.

Plate motions are primarily responses to subduction. Upper-mantle convection is driven by cooling from the top. Plate interactions make sense in these terms when motions are displayed in a framework of Antarctica fixed relative to a sluggish lower mantle, but are nonsensical in popular no-net-rotation and hotspot frames. (Most of these concepts are developed in GSA Today, November 2003.)