Can plants evolve stable alliances with the enemies' enemies?

Rooted and unable to evade herbivores, plants gain by supporting the enemies of herbivores. They may provide food and refuge space to predators, but also release herbivory-induced chemical signals betraying herbivores to their predators. Can we expect such plant-predator alliances to resist invasion by mutants acting as saboteurs or cheaters? Examples of such mutants are 'stealthy' herbivores that do not alert the plant and 'cry wolf' plants that invest in acquiring predators even when they harbor no or only few herbivores? Mathematical models suggest alliances to build up and breakdown due to frequency-dependent selection. Hence, we predict that tritrophic systems are prone to exhibit waves of alliance build up and breakdown, and we point out the implications for pest control through crops with genetically engineered investments in alliances with predators.