Predator metamorphosis and its consequence for prey risk assessment.

Living with a diverse array of predators provides a significant challenge for prey to learn and retain information about each predator they encounter. Consequently, some prey respond to novel predators because they have previous experience with a perceptually similar predator species, a phenomenon known as generalization of predator recognition. However, it remains unknown whether prey can generalize learned responses across ontogenetic stages of predators. Using wood frog tadpole (Lithobates sylvaticus) prey, we conducted two experiments to explore the extent of predator generalization of different life stages of two different predators: (1) predacious diving beetles (Dytiscus sp.) and (2) tiger salamanders (Ambystoma mavortium). In both experiments, we used chemical alarm cues (i.e., injured conspecific cues) to condition tadpoles to recognize the odor of either the larval or adult stage of the predator as risky. One day later, we tested tadpoles with either the larval or adult predator odor to determine whether they generalized their learned responses to the other life stages of the predator. Tadpoles generalized between larval and adult beetle odors but failed to generalize between larval and adult salamander odors. These results suggest that the odor of some predator species changes during metamorphosis to an extent that reduces their recognisability by prey. This "predator identity reset" increases the number of threats to which prey need to attend.

Uncertainty about old information results in differential predator memory in tadpoles.

As information ages, it may become less accurate, resulting in increased uncertainty for decision makers. For example, chemical alarm cues (AC) are a source of public information about a nearby predator attack, and these cues can become spatially inaccurate through time. These cues can also degrade quickly under natural conditions, and cue receivers are sensitive to such degradation. Although numerous studies have documented predator-recognition learning from fresh AC, no studies have explored learning from aged AC and whether the uncertainty associated with this older information contributes to shortening the retention of learned responses (i.e. the 'memory window'). Here, we found that wood frog tadpoles, Lithobates sylvaticus, learned to recognize a novel odour as a predator when paired with AC aged under natural conditions for up to 1 h. However, only tadpoles conditioned with fresh AC were found to retain this learned response when tested 9 days after conditioning. These results support the hypothesis that the memory window is shortened by the uncertainty associated with older information, preventing the long-term costs of a learned association that was based on potentially outdated information.

Evidence for the Predator Attraction Hypothesis in an amphibian predator-prey system.

Many species possess damage-released chemical alarm cues that function in alerting nearby individuals to a predator attack. One hypothesis for the evolution and/or maintenance of such cues is the Predator Attraction Hypothesis, where predators, rather than prey, are the "intended" recipients of these cues. If a predator attack attracts additional predators, these secondary predators might interfere with the predation event, providing the prey with a better chance to escape. In this study, we conducted two experiments to explore this hypothesis in an amphibian predator/prey system. In Experiment 1, we found that tiger salamanders (Ambystoma mavortium) showed a foraging attraction to chemical cues from wood frog (Lithobates sylvaticus) tadpoles. Salamanders that were experienced with tadpole prey, in particular, were strongly attracted to tadpole alarm cues. In Experiment 2, we observed experimental encounters between a tadpole and either one or two salamanders. The presence of the second predator caused salamanders to increase attack speed at the cost of decreased attack accuracy (i.e., increasing the probability that the tadpole would escape attacks). We also found that the mere presence of visual and chemical cues from a second predator did not affect this speed/accuracy trade-off but did cause enough of a distraction to increase tadpole survival. Thus, our findings are consistent with the Predator Attraction Hypothesis for the evolution and/or maintenance of alarm cues.