A model developed by University of Michigan researchers and published in the Oct. 4 issue of Science provides a relatively simple explanation.
"It's been known that slabs (portions of plates that extend down into the Earth) drive convection in Earth's mantle, and ultimately the motion of the surface plates, but it hasn't been well established exactly how that happens---the ideas have been fairly vague," says Clinton Conrad, a postdoctoral fellow in the department of geological sciences. "In this paper, we've been able to describe more precisely how slabs interact with the plates."
When two plates collide, one is forced down beneath the other into the mantle (the plastic-like layer between Earth's crust and core that flows under pressure), creating what geologists call a subduction zone. Because subducting slabs are colder and denser than surrounding mantle material, they tend to sink like a lead ball in a vat of molasses.
There are two main ways these sinking slabs might influence plate motion. If a slab is attached to a plate, the slab can directly pull the plate toward the subduction zone. A slab that is not well attached to a plate, on the other hand, can't pull directly on the plate. Instead, as it sinks, it sets up circulation patterns in the mantle that exert a sort of suction force, drawing nearby plates toward the subduction zone much as floating toys are drawn toward the outlet of a draining bathtub.
To understand the relative importance of slab pull and slab suction forces, Conrad and assistant professor of geological sciences Carolina Lithgow-Bertelloni, with whom he worked on the project, developed models in which: 1) only slab suction was operating; 2) only slab pull was operating; and 3) both slab suction and slab pull were at work. Then they compared the plate motions that would result from each of these scenarios with actual plate motions. The best fit was the model that combined slab pull and slab suction forces.
The model also explained an observation that has baffled geodynamicists for some time. "The way the observation was originally framed was that plates that have continents on them are slow, compared to plates that are only oceanic," says Lithgow-Bertelloni. But the real issue is whether or not the plates have slabs attached, she explains. Overriding plates, which have no slabs, are slower than subducting plates, which have slabs. The explanation? Subducting plates move faster because the pull effect acts directly on them, making them move rapidly toward the subduction zone. Overriding plates are also drawn toward the subduction zone---by the suction effect---but at the same time, the pull effect creates forces in the mantle that counteract that motion. The net effect is that overriding plates move more slowly toward the subduction zone than subducting plates do.
"We've been able to explain that the difference in speed occurs because slab pull generates mantle flow that counteracts the motion of the overriding plate," says Lithgow-Bertelloni. "We also found that this effect is only important for slabs in the upper 600 to 700 kilometers of the mantle. Any slabs deeper than 700 kilometers do not contribute to this effect. They're important for driving flow in the mantle, but they're not important for the pull."