Learning specificity in antipredator training

Reintroductions are becoming an increasingly important conservation tool. However, captive bred individuals that are released back into the wild have no experience with predators and may have lost their abilities to recognise and respond efficiently to them. Several previous studies have shown that antipredator behavour can be enhanced by pairing predator models with aversive stimuli. However, if such acquired fear responses are not specifically evoked by the predator for which training is undertaken, there is a risk that they will extinguish very quickly once the animals are released back into the wild. We enhanced responses of tammar wallabies (Macropus eugenii) to a realistic model fox by pairing its presentation with an aversive stimulus (a human who simulated a capture procedure). To determine whether the animals' acquired responses were specifically controlled by the fox, we also quantified their behaviour to an array of visual models, both before and after training. Following training, animals presented with the fox reduced foraging, grooming and sitting behaviour. This response was significantly different from that of a control group matched for both total exposure to predator stimuli and experience of humans simulating capture. Although never paired with the aversive stimulus, a model cat evoked a pattern of response similar to that obtained with the fox, while the effects of a wallaby presentation were relatively ephemeral. In contrast, wallabies did not alter their behaviour towards a model goat. This pattern of results suggests that the animals learnt that a predator-like model predicted the onset of a capture procedure. They could potentially have responded to features, such as frontally placed eyes, that are shared by many predators. Thus, our antipredator training techniques appear to successfully inculcate a relatively specific fear response to a model predator.