Sequential predictability of the scanning behaviour of greater rheas, Rhea americana

Some models that predict the scanning pattern to maximise the probability of predator detection by prey assume sequential randomness (i.e. lengths of inter-scan intervals are independent of each other and of the length of scan events). Sequential randomness would prevent stalking predators synchronising their attacks with long inter-scan intervals. We analysed the presence of sequential randomness in the scanning behaviour of the greater rhea (Rhea americana), a large flightless bird that forages solitarily or in groups. We analysed sequences of behaviour of 40 wild rheas, foraging solitarily or in groups of 2 to 17 individuals, in two populations in eastern Argentina. We used parametric (auto and cross-correlations) and non-parametric (runs and Spearman rank correlation) tests to detect dependence among intervals. We detected the temporal dependence of inter-scan intervals in 30 of 40 behavioural series, but only 11 of 40 sequences showed consistent dependence when tested using parametric and non-parametric tests. The temporal dependence detected in some individual sequences would still be used by predators to synchronise their attacks with long inattentive periods of the prey. However, simple simulations showed that attack success would not be different from that of predators launching their attacks at the beginning of a randomly selected inattentive period of the prey. We propose that temporal dependence among the length of inattentive periods in greater rheas would be the result of simple mechanistic rules determined by the compromise between foraging and scanning behaviour, and it would be of little value to potential stalking predators.