Is sentinel behaviour safe? An experimental investigation Amanda R. Ridley a, b, * , Martha J. Nelson-Flower b , Alex M. Thompson b a Department of Biological Sciences, Macquarie University, Sydney, Australia b DST/NRF Centre of Excellence at the Percy FitzPatrick Institute, University of Cape Town, Cape Town, South Africa article info Article history: Received 31 July 2012 Initial acceptance 12 September 2012 Final acceptance 9 October 2012 Available online xxx MS. number: 12-00592 Keywords: cooperative breeding kin selection pied babbler safe location selfish sentinel sentinel behaviour Turdoides bicolor Sentinel behaviour, where individuals take turns to watch for danger and give alarm calls to approaching predators, has been observed in a number of animal societies. However, the evolutionary causes of this behaviour remain unclear. There are two main, competing hypotheses regarding the evolution of sentinel behaviour. The first hypothesis is that it is a cooperative behaviour, where group members benefit from the detection of danger but share the workload of acting as a sentinel. The second is that it is a safe, selfish behaviour. Under the second hypothesis, once an individual is satiated, being a sentinel is safer because sentinels can detect threats more readily and can therefore escape from predators faster. We examined whether sentinels are safer than foragers in a wild, free-living cooperative bird (the pied babbler, Turdoides bicolor) with a well-described sentinel system. We found that sentinel behaviour was costly because (1) sentinels were targeted by predators more often, (2) they were further from cover than foragers, and (3) they took longer to reach the safety of cover following a predator alarm. These results suggest that individuals do not become sentinels because it is safer. This is the first study to demonstrate that sentinels are at greater risk of predator attack than foraging group members and suggests sentinel activity may have evolved as a form of cooperative behaviour. Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. The occurrence of a sentinel or guard is well known and wide- spread in group-living species, and involves individuals taking turns to watch for predators while other group members rest or forage. Sentinel behaviour has been observed in a number of cooperative vertebrates (Rasa 1986, 1987 , 1989; McGowan & Woolfenden 1989; Hailman et al. 1994; Clutton-Brock et al. 1999, Manser 1999; Wright et al. 2001a, b; Bednekoff & Woolfenden 2003, 2006; Ridley & Raihani 2007; Hollén et al. 2008; Bell et al. 2009; Ridley et al. 2010; Sharpe et al. 2010) and differs from personal antipredator vigilance in that (1) it is coordinated (usually only one individual is actively scanning for predators at any point in time while the rest of the group forages, McGowan & Woolfenden 1989; Bednekoff 1997) and (2) sentinels usually stand in an elevated position while on look-out, rather than the typical ‘head- up’ behaviour displayed during personal vigilance (Elgar 1989; Bednekoff & Lima 1998). Studies of sentinel activity have found that this behaviour is beneficial in terms of predator detection: sentinels are more effective at detecting predators than are foraging group members (Rasa 1987; McGowan & Woolfenden 1989; Manser 1999; Ridley et al. 2010), and foragers gain increased biomass intake when a sentinel is present because they spend less time investing in personal vigilance (Hollén et al. 2008). In addition, research has shown that group members are extremely responsive to the vocal information that sentinels provide, and adjust their behaviour accordingly (Rasa 1986; Manser 1999; Hollén et al. 2008; Bell et al. 2009; Radford et al. 2009). Despite this evidence of a benefit of sentinel activity for foraging group members, whether sentinel behaviour represents a cooper- ative or selfish act has been contested. Although this type of behaviour was originally presumed to have evolved as a form of kin selection or reciprocal altruism (Hamilton 1964; Trivers 1971), more recently Bednekoff (1997) suggested that becoming a sentinel was the safest behaviour for a satiated individual, and thus that sentinel activity represented a safe, selfish behaviour. Bednekoff (1997 , 2001) also suggested that by being a sentinel, individuals could detect predators sooner than if they were resting elsewhere, and could thus escape to the safety of cover more quickly. Empirical research provided support for Bednekoff’s ideas, finding that indi- viduals were more likely to become sentinels after supplemental feeding (Clutton-Brock et al. 1999; Wright et al. 2001b; Bednekoff & Woolfenden 2003), that sentinels tended to be closer to the safety of cover than other group members (Rasa 1989; Clutton-Brock et al. 1999) and that sentinels did not suffer higher predation rates than foragers (Clutton-Brock et al. 1999). Despite considerable empirical evidence for Bednekoff’s theory of safe, selfish sentinels, several issues remain unresolved and suggest that sentinel activity may, at * Correspondence and present address: A. R. Ridley, School of Animal Biology (M092), University of Western Australia, 35 Stirling Highway, Crawley 6009, WA, Australia. E-mail address: Amanda.ridley@uwa.edu.au (A. R. Ridley). Contents lists available at SciVerse ScienceDirect Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav 0003-3472/$38.00 Ó 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2012.10.017 Animal Behaviour xxx (2012) 1e6 Please cite this article in press as: Ridley, A. R., et al., Is sentinel behaviour safe? An experimental investigation, Animal Behaviour (2012), http:// dx.doi.org/10.1016/j.anbehav.2012.10.017