Evaluating the effectiveness of model predators in simulating real predators in tropical bird communities

Predatory attacks are difficult to observe in nature, and thus, researchers use model predators to simulate real predators. However, few studies have systematically compared the response of prey to models versus living predators. We flew a trained live raptor across mixed species flocks of Amazonian birds and compared the acoustic and behavioural response of individual species and the entire flock to that following a flight of a painted model glider. We found no statistically significant differences between treatments in alarm call bandwidth 50%, duration 90%, latency for most species, or the proportion of species producing alarm calls. A significant overall treatment effect was detected for alarm call frequency 75%, although this was not driven by any specific species and was small in magnitude. The number of urgent notes showed a marginal trend toward fewer notes in model trials. Latency to call differed between treatments overall, driven by White-flanked Antwren, which called more rapidly to the live hawk than to the model. About one-third of flock species responded with alarm calls, a proportion that did not differ between treatments. Models can indeed be an effective method to study anti-predator behaviour, but more research on how and why species may or may not discriminate between models and live predators is warranted.