The effects of relaxed and reversed selection by predators on the antipredator behavior of the threespine stickleback, Gasterosteus aculeatus

Isolation from predators can lead to the reduction or loss of ancestral behavioral defenses in prey, but does not always do so. Predators introduced to populations that have experienced relaxed selection from some ancestral predators can favor the evolution of antipredator behavior that has been lost. We examined these possibilities by eliciting antipredator behavior in three populations of threespine stickleback fish, Gasterosteus aculeatus: an oceanic population thought to resemble the ancestral form, and two populations historically (up to 20 000 yr) devoid of piscine predators (relaxed selection), one of which has been stocked with salmonids for nearly 25 yr (reversed selection). We used three kinds of predator models: a sculpin (ambush predator), a rainbow trout (chasing predator), and an overhead silhouette of an arctic tern. Stickleback reacted differently to the three models, indicating that they distinguished among them. Individuals from all populations responded similarly to the tern model. The ancestral population showed the weakest response to the sculpin model despite being the only population that encounters these predators naturally. Stickleback from the trout-free population displayed slightly reduced responses to the trout model, and recovery times like those in the ancestral population providing only weak evidence for loss of the ancestral antipredator repertoire. Fish from the reverse-selected population exhibited fascinating, elevated responses to both the trout and sculpin models relative to the other two populations. These findings offer initial evidence of (1) a partial alteration of the ancestral behavioral repertoire during a long period of relaxed selection from piscine predators, and (2) rapid acquisition of extreme responses to piscine predators under reverse selection.