Résumé
We present a dynamic model of the evolution of host resistance to avian brood parasites, when the latter can retaliate against hosts that reject parasitic eggs. In a verbal model, Zahavi (1979, American Naturalist,113 , 157–159) suggested that retaliatory cuckoos might prevent the evolution of host resistance by reducing the reproductive success of rejecter hosts (i.e. by destroying their eggs or nestlings). Here we develop a model based on the association between the great spotted cuckoo, Clamator glandarius, and its main host, the European magpie, Pica pica, because this is the only system that has provided supportive evidence, to date, for the existence of retaliatory behaviour. Our aims were (1) to derive the conditions for invasion of the retaliation strategy in a nonretaliatory parasite population and (2) to investigate the consequences of retaliation for the evolution of host defence. If we assume a cost of discrimination for rejecter hosts in the absence of parasitism, and a cost paid by a retaliator for monitoring nests, our model shows cyclical dynamics. There is no evolutionarily stable strategy, and populations of both hosts and parasites will cycle indefinitely, the period of the cycles depending on mutation and/or migration rate. A stable polymorphism of acceptors and rejecters occurs only when parasites are nonretaliators. The spread of retaliator parasites drives rejecter hosts to extinction.