Negotiating a stable solution for vigilance behaviour.

Most, if not all, animals are highly responsive organisms that are capable of adjusting their behaviour during the course of their interactions with others [1]. This observation may seem obvious, yet evolutionary theory has traditionally modelled cooperation and conflict as a ‘sealed bid’, assuming that both parties decide on a fixed strategy without any knowledge of the other’s behaviour [2]. Recently, theoreticians have started to consider the so-called negotiation rules animals use to respond to, and potentially manipulate, each other in real time, but so far this work has focused exclusively on family conflicts over offspring care [3]. A new study by Sirot [4] applies this method for the first time to group foraging, exploring how the threat of predation may drive patterns of vigilance behaviour. Anti-predator vigilance is key to survival in many animal species and clearly falls within the scope of negotiation models because individuals living in groups are known to adjust their vigilance patterns depending on the behaviour of their companions [5,6]. A striking illustration of this is the sentinel behaviour seen in some social vertebrates, where group members suspend foraging, adopt a raised position and look out for danger [7–9]. Sentinel bouts within groups are tightly coordinated, with individuals typically taking turns and there rarely being more than one or two engaging in this activity at any one time [8,9]. In such situations, vigilance decisions are anything but a sealed bid. Recognizing the high degree of responsiveness in these interactions, Sirot modelled vigilance in a pair of foraging animals that continuously monitor each other’s behaviour. At any given time, each animal could choose either to feed or to scan for predators, but in contrast to previous models they could flexibly adjust this decision depending on the current behaviour of their companion. Sirot then varied the mode of predatory attack, to see how this influenced the evolutionarily stable vigilance pattern of the foragers. If predators randomly target either one of the pair, both benefit equally from the other’s vigilance behaviour, which favours separate bouts of scanning with no overlap (‘coordination’). By contrast, if predators direct their attacks towards a non-vigilant animal when its companion is vigilant, it is too risky to be the only one feeding. In this situation, when one animal starts scanning its companion immediately follows suit, resulting in long bouts of scanning together interspersed with long bouts of feeding together (‘synchronization’). The model therefore predicts that the extent to which vigilance behaviour is coordinated in a group depends on how predation risk is spread across those individuals that are vigilant and those that are not (see [10,11]). In general, one might expect that actively scanning individuals