Ecology and environmental management is largely predicated on the view that ecosystems respond to environmental changes in a smooth and straightforward way. However, in Ecology Letters, May, Schmitz reports on a long-term field experiment that may prompt a hard, critical look at this reigning view. In the experimental system, top predators instrumentally control species diversity and productivity of the ecosystem. This control comes about by causing herbivores to feed on and suppress the abundance of a highly productive plant that otherwise would out-compete many other plant species. Alteration of this control by a predator removal experiment caused rapid increase in the abundance of the competitive dominant plant. It also caused an abrupt, catastrophic loss of plant species diversity. Attempts to restore plant diversity by reintroducing the predators back into the system failed. This failure happened because the ecosystem became locked into an alternate stable state in which herbivores were incapable of suppressing the now highly productive, competitively dominant plant. The study provides a cautionary tale that ecosystem management which alters predator abundances even slightly may cause dramatic and irreversible shifts in the productivity and diversity of ecosystems.