A Localized Negative Genetic Correlation Constrains Microevolution of Coat Color in Wild Sheep

may be important locally, but they do not represent a dominant global pattern. Additionally, we have not evaluated the possible effects of 3D slab morphology on trench-parallel flow in the wedge (e.g., 26) or beneath the slab (14, 27). Such flow is likely important locally andmay serve to strengthen or otherwise modulate the trench-parallel flow induced by trench migration. Our model essentially requires a thin decoupling zone between the downgoing slab and the subslab mantle, because we observe no detectable anisotropic signal from slab-entrained flow. [Cascadia represents an exception (8) to this rule; this may be a consequence of the very young age of the subducting lithosphere or the region’s complicated subduction history and slab morphology.] This decoupling zone appears to be universal, in that slab-entrained flow is not generally detected even for subduction zoneswith a nearly stationary trench, and may be a consequence of the entrainment of a thin layer of buoyant asthenosphere (28). The inferred subslab flow field also requires a partial barrier tomantle flow beneath the slab, as proposed by (14), as well as an effective barrier to horizontal flow in the trench-normal direction (29), so that mantle material is forced to escape laterally in response to the motion of the trench. The barrier to entrained flow beneath the slab provides a constraint onmass transfer between the upper and lower mantle, which in turn has major implications for mantle evolution and dynamics. Finally, the existence of trench-parallel flow in the mantle wedge as a common (although not ubiquitous) phenomenon implies that along-strike transport of mantle material is important for many arcs, with consequences for magma genesis, volatile transport, and thermal structure.