Looking after your partner: sentinel behaviour in a socially monogamous bird

Introduction

Avoiding predation is an almost ubiquitous challenge for most species and natural selection favours those individuals with effective anti-predator defences (Ruxton, Sherratt & Speed, 2004; Caro, 2005). Longevity is an important determinant of lifetime reproductive success (Pettorelli & Durant, 2007) and accordingly, a wide range of sophisticated anti-predator defences have evolved amongst prey species (Caro, 2005). The presence of sentinels, whereby one individual assumes a prominent position and scans for predators whilst other individuals engage in some other activity, such as foraging, has evolved as an effective form of defence. Sentinel behaviours have been well described in species with stable social groups of birds and mammals within cooperative and polygynous breeding systems (Beletsky, Higgins & Orians, 1986; Knight & Temple, 1988; Hollén, Bell & Radford, 2008; Ridley, Raihani & Bell, 2010; Clutton-Brock et al., 1999). Indeed, amongst cooperatively breeding birds and mammals, the deployment of sentinels as a common form of anti-predator defence has long been perceived as a classic example of kin selection or reciprocal altruism (Caro, 2005; Ridley, Nelson-Flower & Thompson, 2013). Even in circumstances in which sentinels are believed to be acting in their own individual interests (Clutton-Brock et al., 1999), other group members benefit from their presence. When sentinels are present, group members enjoy increased foraging success (Hollén, Bell & Radford, 2008) and receive early warnings of approaching predators (Caro, 2005).

Surprisingly, given the cooperation between a male and female that is inherent at the heart of a socially monogamous relationship, unequivocal evidence of sentinels in socially monogamous breeding systems remains elusive, despite some earlier reports of their presence (Morton & Shalter, 1977; Tilson, 1980; D’Agostino, Giovinazzo & Eaton, 1981). Over the past couple of decades, most research on socially monogamous species has focused on aspects of conflict rather than cooperation. For example, there has been much effort invested in questions relating to genetic infidelity (Petrie & Kempenaers, 1998; Griffith, Owens & Thuman, 2002), the conflict over parental investment (Parker, Royle & Hartley, 2002; Royle, Hartley & Parker, 2002), and maternal effects, particularly from the conflict-centric differential allocation hypothesis (Burley, 1988; Gil et al., 1999; Pryke & Griffith, 2009). Despite this apparent neglect, cooperative behaviour between partners such as sentinelling should be expected to have evolved within socially monogamous breeding systems (Caro, 2005) where the evolutionary fitness of one’s partner is linked to one’s own fitness. When one member of the pair has a low probability of detecting a predator but a high probability of receiving advanced warning of a predator from the other pair member, then active sentinelling behaviour should be advantageous. Effective sentinelling within partnerships may improve partner survival and also the efficiency with which they engage in activities such as foraging or incubation.

The purpose of this study was to experimentally test for the presence and effectiveness of sentinels in a socially monogamous bird, the wild zebra finch (Taeniopygia guttata). Although domesticated zebra finches are a well-used model system in studies of sexual conflict (Burley, 1988; Gil et al., 1999; Royle, Hartley & Parker, 2002), in the wild, zebra finches form life-long genetically monogamous pair bonds (Zann, 1996; Griffith et al., 2010) with high levels of behavioural synchrony (Mariette & Griffith, 2012; Mariette & Griffith, 2013). Individuals increase their fitness by breeding with a familiar partner (Adkins-Regan & Tomaszycki, 2007) and, as the long-term evolutionary interests of both parents overlap significantly, cooperative anti-predator behaviours should be expected (Ruxton, Sherratt & Speed, 2004; Caro, 2005). Whilst foraging, adult zebra finches suffer high mortality rates from aerial predators, such as Australian hobbies (Falco longipennis) and collared sparrowhawks (Accipiter fasciatus) (Zann, 1996), and are vulnerable in their enclosed bottle-shaped nests to predation by cats and a number of snake, lizard and bird species (Zann, 1996; Griffith, Pryke & Mariette, 2008). Consequently, in this study, we tested the hypothesis that, when approached by a human observer during incubation, birds would flush from their nests at a higher rate and significantly greater distances when their reproductive partners were present and acting as a sentinel than when they were absent.

Methods

Results

Sentinel birds were observed in 163 (84%) of the 195 trials in which an individual was inside the nest incubating the eggs, and not observed during 32 (16%) of the trials. In the 163 trials in which sentinels were observed, males were found to be incubating in 67 (41%) trials (i.e., the sentinel was a female in these cases) and in the 32 trials without sentinels, males were found to be incubating in 17 (53%) trials. There was a strong positive correlation between the observer’s distance from the nest and the time taken by the observer to walk to the nest for those trials when the incubating bird left the nest (F_1,194 = 2698.255, P < 0.001), which meant that the observer was indeed moving towards the nest at a steady pace. In the 163 trials in which sentinel birds were present, incubating birds flushed from the nest when the observer was 11.8 ± 0.5 m from the nest, whereas in the 32 trials in which sentinel birds were absent, the observer was essentially able to catch the bird in the nest as incubating birds flushed from nest when the observer was only 0.4 ± 0.2 m from the nest. This means that incubating birds flushed from their nests when the observer was a significantly greater distance from the nest when their reproductive partner was present and observed acting as a sentinel (P < 0.001) (Table 1) (Figs. 1A and 1B). Further analyses indicated that there was a strong positive correlation between the total time from the beginning of each trial (which consisted of the sum of the random departure time between 0–90 s and the time from when the observer started walking to when the incubating bird flushed between 0–100 s) and the random departure time (F_1,194 = 2297.197, P < 0.001). This indicates that departure times by incubating birds were driven by the observer’s proximity to the nest rather than being random natural departures. Meanwhile, the distance at which incubating birds flushed from nests was not influenced by the approach direction of the observer, the gender of the focal bird, the presence or absence of conspecifics, the type of habitat, the size of the breeding colony or any of their two-way interaction terms (Table 1).

Discussion

We found that acting as a sentinel for an incubating partner was a common and effective form of anti-predator defence in the socially monogamous zebra finch. When a human observer approached nests during incubation, birds flushed from their nests at significantly greater distances when their partner was acting as a sentinel than when their partner was absent. We are confident that incubating birds did not flush as a result of their directly seeing the approaching experimenter because there was no difference in the flushing distance between those trials in which the observer approached from the front or the side. In the latter situation, a bird on the nest could not see through the side wall of the nestbox, whereas an observer approaching from the front could be seen through the nest box entrance hole.Therefore, by focusing on birds that were visually occluded by the nest box itself we have been able to demonstrate that individuals benefit from the presence of a sentinelling partner. Nevertheless, we believe that sentinel behaviour will be just as effective in birds in natural nests and certainly we see no reason to believe that it will have increased in frequency in this nest box nesting population. Although the nestboxes were attached to steel picket posts they were always positioned within a few metres of an acacia or wattle bush (Griffith, Pryke & Mariette, 2008) and sentinel birds were seen perching on the nearest bush, and within a few metres of the nest, providing them with good opportunity to scan for predators.

The sentinelling behaviour that we have observed will have enabled the adult in the nest to leave the nest before potentially being surprised and trapped by a predator. We have previously observed one direct predation of an adult bird being consumed after being caught on a natural nest by an eastern brown snake (Pseudonajat extilis), and found half a dozen more active natural nests that appear to have been pulled apart by predators and in which we have seen large numbers of adult feathers, providing suggestive evidence that an adult was predated from the nest (Griffith, Pryke & Mariette, 2008). An adult taken by surprise in a nest is very vulnerable. The nest is a light construction of thin grass stems readily broken apart by most predators, and it is hard for adults to escape as the only exit is the funneled entrance hole, scarcely wider than the bird itself. Exiting the nest before an approaching predator reaches the nest is therefore the best survival strategy. Our results demonstrate the effectiveness of this strategy and indeed on the 32 occasions when there was no sentinel, the observer reached the nestbox and was able to trap and capture the bird inside the nest.

Zebra finches do not mob, attack, and attempt to distract predators and instead, either flee or freeze to keep out of harm’s way (Zann, 1996). Even if the eggs or nestlings are vulnerable to a predator, the adults will not make any attempt to dispel the predator. Therefore, we can be reasonably confident that the sentinel behavior we observed is focused on preserving the life of the partner rather than being an incidental response to an anti-predator call that might have another functional response such as eliciting help from conspecifics to mount a mobbing response. For that reason, although we only studied the behavior on a single day of the incubation period, we expect such sentinelling to be maintained across the whole reproductive period, and indeed to occur throughout non-breeding periods as well. We believe that the primary function of the behavior is to maintain the partnership rather than the reproductive attempt itself. In this, our work contrasts nicely with one of the observations of sentinel behaviour in a non-cooperatively breeding bird – the red-winged blackbird. In this polygynous species, male sentinels gain direct fitness benefits by warning offspring about an approaching predator and/or instigating a mobbing defence against a predator (Beletsky, Higgins & Orians, 1986; Knight & Temple, 1988; Yasukawa, Whittenberger & Nielsen, 1992; Burton & Yasukawa, 2001). This sentinelling behaviour in the red-winged blackbird reduces predation on eggs or nestlings and is actively selected by females as one of their mate choice criteria because good males will provide better defence for their offspring (Knight & Temple, 1988). Zebra finches use sentinelling to protect a partner, whereas red-winged blackbirds appear to primarily use sentinel behaviour to protect offspring. However, when comparing our findings with those from the blackbirds, it is important to remember that the data of the two studies were collected during the incubation and nestling phases respectively. Sentinel behaviour (that is focused on the survival of the brood) might be expected to change throughout the reproductive cycle as investment in the brood increases and this will be worthy of future study.

Sophisticated forms of anti-predator defences such as sentinelling are expected to evolve in social species that suffer high levels of predation and live in habitats that facilitate some individuals with a good vantage point to sense approaching danger (Bednekoff, 2001; Caro, 2005). These criteria apply in the zebra finch, which typically live in small social groups and forage on the ground in semi-open habitat where they are vulnerable to aerial ambush predators such as Australian hobbies or collared sparrowhawks. Whilst sentinelling for each other will probably make sense throughout the whole year it makes particular sense during reproduction when one adult is required to spend time alone in the nest. Previous work has demonstrated that pairs have very synchronised behaviour and move around together almost habitually and visit the nest together during nestling provisioning (Mariette & Griffith, 2012). We recently found that the pairs that had the highest levels of behaviour synchrony (spending more time in each other’s company) hatched a higher proportion of their eggs resulting in higher fitness (Mariette & Griffith, 2012; Mariette & Griffith, 2013). Perhaps this may, in part, be due to a greater efficacy of incubation by individuals that are reassured by the presence of their partner on guard outside the nest. Another important determinant of fitness that has been demonstrated in a study of domesticated zebra finches was that pairs that have experience of breeding together are more efficient at initiating a reproductive attempt once conditions allow (Adkins-Regan & Tomaszycki, 2007). These two lines of evidence provide good motivation for a partner to look out for its incubating mate.

The costs of acting as a sentinel are not necessarily great and sentinelling is no longer seen as an altruistic behaviour in the way that it once was (Clutton-Brock et al., 1999; Wright et al., 2001; Wright, Maklakov & Khazin, 2001; Bednekoff & Woolfenden, 2003; Ridley, Raihani & Bell, 2010; but see Ridley, Nelson-Flower & Thompson, 2013). The incubation phase is one of the few times when an individual will find itself potentially physically separated from its mate, and alone in the environment (Zann, 1996; Elie et al., 2010). Isolating an individual zebra finch from other conspecifics (visually and acoustically) has been shown to be physiologically stressful (Perez et al., 2012), and therefore perhaps sentinels are partly motivated by a desire to remain in contact with their partner. Furthermore, taking up a sentinelling position should reduce an individual’s predation risk because a sentinel can choose a position that is near cover and provides good sight lines to approaching danger (Clutton-Brock et al., 1999).

The mechanism through which the incubating bird learned of a potential predator is currently unclear, although there are three possibilities. First, the sentinelling bird may have given an alarm call alerting the incubating bird. Second, the cessation of an ongoing duet between the pair (Elie et al., 2010) may have alerted the incubating bird to the possibility of danger, causing it to have approached the nestbox entrance to check for itself. Third, the incubating bird may have heard the flapping wings of the fleeing partner as this has previously been shown to be an effective anti-predator strategy in other species (Hingee & Magrath, 2009). Moreover, given that sentinel and other anti-predator behaviours have been shown to be plastic over spatial and temporal timescales, the cues observed in the zebra finches may differ between different contexts (Caro, 2005; Ridley, Nelson-Flower & Thompson, 2013).

Over the past few decades, much research effort in socially monogamous animals has focused on aspects of sexual conflict, and the zebra finch is one of the most widely used models in this area (Burley, 1988; Gil et al., 1999; Royle, Hartley & Parker, 2002; Forstmeier et al., 2011). We believe that our findings help to redress the balance a little and highlight the mutual benefits that a male and female provide for one another in the social bond of monogamy. Recent work on the wild zebra finch has demonstrated that pairs are reproductively faithful to one another (Griffith et al., 2010), are behaviourally coordinated to a high degree (Mariette & Griffith, 2012; Mariette & Griffith, 2013), and have subtle and highly ritualised acoustic communication (Elie et al., 2010). This emerging picture of the social bond of the wild zebra finch does contrast with much of the work focused on the domesticated zebra finch revealing various aspects of sexual conflict. However, we believe that the unpredictable ecological conditions of the Australian arid zone (Morton et al., 2012), and the selective and behavioural pressure exerted by predators helps to reinforce the strength and depth of the social bond in wild zebra finches.

Social monogamy is the predominant mating bond formed in birds (>90% of species, (Lack, 1968)), and the behavioural and physiological dynamics at the heart of this pair bond play an important role in life-history evolution. Recently breeding pairs were described as a ‘flock of two’ (Lima, 2009), and our findings support the idea that the individuals in a socially monogamous partnership have the opportunity to benefit mutually in a number of ways that effect their own individual fitness. Although sentinels are most classically observed in cooperatively breeding mammals and birds (Clutton-Brock et al., 1999; Ridley, Raihani & Bell, 2010) as one of the clearest manifestations of highly social and cooperative behaviour, our work demonstrates that it may be found in a much broader context.

We believe that our findings should lead to a renewed interest in the cooperative aspects of social monogamy that have been neglected for so long in favour of the more fashionable ideas focused on the conflict between the partners. These conflicts over investment, and individual fitness, are important but have been overstated in recent decades. The strength and duration of the partnership between a male and female has the potential to directly affect individual fitness in a number of ways that have yet to be thoroughly explored and we believe that our findings will help to establish the importance of those determinants of individual fitness.

Supplemental Information