Stationary boundaries of sedentary species may belie dynamic processes that form them. Our aim was to test an implication of an evolving body of theory, that such boundaries are manifestations of complex regulatory dynamics. On rocky shores of British Columbia, large-scale field experiments altered the densities of predatory sea stars (Pisaster ochraceus), causing shifts in the location of the lower vertical boundaries of their prey, sea mussels (Mytilus californianus). While control mussel beds remained unchanged, experimental reductions of sea star densities caused the downward extension of the lower boundaries, and experimental increases in sea stars densities caused the upward recession of the lower boundary well into the zone presumed to be a spatial refuge from predation. Cleared plots prepared within the initial boundaries were recolonized to varying degrees, depending on predator densities. After 30 months, plots on sea star removal sites showed high densities of adult mussels, control plots showed intermediate densities, and sea star addition plots showed only a sparse cover of alternative prey. Observations by divers at high tide showed that as small prey were depleted progressively from removal, to control, to addition sites, correspondingly larger mussels were attacked, including very large individuals comprising the lower boundary of addition sites. The findings contradict classic theory of zonation based on static prey refuges and support an alternative theory in which boundaries are maintained by complex, spatially structured equilibria.