Starling Flock Networks Manage Uncertainty in Consensus at Low Cost

Flocks of starlings exhibit a remarkable ability to maintain cohesion as a group in highly uncertain environments and with limited, noisy information. Recent work demonstrated that individual starlings within large flocks respond to a fixed number of nearest neighbors, but until now it was not understood why this number is seven. We analyze robustness to uncertainty of consensus in empirical data from multiple starling flocks and show that the flock interaction networks with six or seven neighbors optimize the trade-off between group cohesion and individual effort. We can distinguish these numbers of neighbors from fewer or greater numbers using our systems-theoretic approach to measuring robustness of interaction networks as a function of the network structure, i.e., who is sensing whom. The metric quantifies the disagreement within the network due to disturbances and noise during consensus behavior and can be evaluated over a parameterized family of hypothesized sensing strategies (here the parameter is number of neighbors). We use this approach to further show that for the range of flocks studied the optimal number of neighbors does not depend on the number of birds within a flock; rather, it depends on the shape, notably the thickness, of the flock. The results suggest that robustness to uncertainty may have been a factor in the evolution of flocking for starlings. More generally, our results elucidate the role of the interaction network on uncertainty management in collective behavior, and motivate the application of our approach to other biological networks.     

Some causes of the variable shape of flocks of birds

Flocks of birds are highly variable in shape in all contexts (while travelling, avoiding predation, wheeling above the roost). Particularly amazing in this respect are the aerial displays of huge flocks of starlings (Sturnus vulgaris) above the sleeping site at dawn. The causes of this variability are hardly known, however. Here we hypothesise that variability of shape increases when there are larger local differences in movement behaviour in the flock. We investigate this hypothesis with the help of a model of the self-organisation of travelling groups, called StarDisplay, since such a model has also increased our understanding of what causes the oblong shape of schools of fish. The flocking patterns in the model prove to resemble those of real birds, in particular of starlings and rock doves. As to shape, we measure the relative proportions of the flock in several ways, which either depend on the direction of movement or do not. We confirm that flock shape is usually more variable when local differences in movement in the flock are larger. This happens when a) flock size is larger, b) interacting partners are fewer, c) the flock turnings are stronger, and d) individuals roll into the turn. In contrast to our expectations, when variability of speed in the flock is higher, flock shape and the positions of members in the flock are more static. We explain this and indicate the adaptive value of low variability of speed and spatial restriction of interaction and develop testable hypotheses.     

Diffusion of individual birds in starling flocks

Flocking is a paradigmatic example of collective animal behaviour, where global order emerges out of self-organization. Each individual has a tendency to align its flight direction with those of neighbours, and such a simple form of interaction produces a state of collective motion of the group. When compared with other cases of collective ordering, a crucial feature of animal groups is that the interaction network is not fixed in time, as each individual moves and continuously changes its neighbours. The possibility to exchange neighbours strongly enhances the stability of global ordering and the way information is propagated through the group. Here, we assess the relevance of this mechanism in large flocks of starlings (Sturnus vulgaris). We find that birds move faster than Brownian walkers both with respect to the centre of mass of the flock, and with respect to each other. Moreover, this behaviour is strongly anisotropic with respect to the direction of motion of the flock. We also measure the amount of neighbours reshuffling and find that neighbours change in time exclusively as a consequence of the random fluctuations in the individual motion, so that no specific mechanism to keep one''s neighbours seems to be enforced. On the contrary, our findings suggest that a more complex dynamical process occurs at the border of the flock.     

Intraflock variation in the speed of escape-flight response on attack by an avian predator

The benefits of flocking to prey species, whether through collective vigilance,dilution of risk, or predator confusion, depend on flock members respondingin a coordinated way to attack. We videotaped sparrowhawks attackingredshank flocks to determine if there were differences in thetiming of escape flights between flock members and the factorsthat might affect any differences. Sparrowhawks are surpriseshort-chase predators, so variation in the time taken to takeflight on attack is likely to be a good index of predation risk.Most birds in a flock flew within 0.25 s of the first bird flying,and all birds were flying within 0.7 s. Redshanks that werevigilant, that were closest to the approaching raptor, and thatwere close to their neighbors took flight earliest within aflock. Birds in larger flocks took longer, on average, to takeflight, measured from the time that the first bird in the flockflew. Most birds took flight immediately after near neighbors tookoff, but later flying birds were more likely to fly immediatelyafter more distant neighbors took flight. This result, alongwith the result that increased nearest neighbor distance increasedflight delay, suggests that most redshanks flew in responseto conspecifics flying. The results strongly suggest that thereis significant individual variation in predation risk withinflocks so that individuals within a flock will vary in benefitsthat they gain from flocking.     

The Effects of Predation Risk, Food Abundance, and Population Size on Group Size of Brown-Headed Cowbirds (Molothrus ater)

Social and ecological conditions can influence flock formation (e.g. number of flocks, flock size, etc.) depending on the degree of social attraction of a species. We studied group formation in brown‐headed cowbirds (Molothrus ater) over short time periods (30 min) in two semi‐natural experiments conducted under controlled conditions. First, we determined the shape of the relationship between intake rate and flock size by manipulating group size in a single enclosure. Second, we assessed the role of population size, food abundance, and predation risk, and their interactions, in flock size formation in a system of four enclosures (two with and two without food) connected to a central refuge patch. In the first experiment, we found that pecking rates peaked at intermediate flock sizes (three to six individuals), which was influenced by greater availability of foraging time and more aggressive interactions in large groups. In the second experiment, flock sizes in the patches with food increased with population size likely due to the benefits of patch exploitation in groups. Flock size decreased after predator attack probably because refuge availability reduced perceived predation risk more than flocking in larger groups. Food abundance had minor effects, varying flock sizes between the two patches with food, under high food availability conditions when population size was high, probably due to social cohesion effects. Our results suggest that: (1) this species has an inverted‐U food intake–group size relationship with a range of intake‐maximizing flock sizes rather than a single peak, (2) the presence of a near refuge modifies the expected benefits of group patch exploitation under high predation risk, and (3) an increase in population size would more likely be translated into rapid increases in the size of the flocks rather than in more new flocks.     

Self-organization of collective escape in pigeon flocks

Bird flocks under predation demonstrate complex patterns of collective escape. These patterns may emerge by self-organization from local interactions among group-members. Computational models have been shown to be valuable for identifying what behavioral rules may govern such interactions among individuals during collective motion. However, our knowledge of such rules for collective escape is limited by the lack of quantitative data on bird flocks under predation in the field. In the present study, we analyze the first GPS trajectories of pigeons in airborne flocks attacked by a robotic falcon in order to build a species-specific model of collective escape. We use our model to examine a recently identified distance-dependent pattern of collective behavior: the closer the prey is to the predator, the higher the frequency with which flock members turn away from it. We first extract from the empirical data of pigeon flocks the characteristics of their shape and internal structure (bearing angle and distance to nearest neighbors). Combining these with information on their coordination from the literature, we build an agent-based model adjusted to pigeons’ collective escape. We show that the pattern of turning away from the predator with increased frequency when the predator is closer arises without prey prioritizing escape when the predator is near. Instead, it emerges through self-organization from a behavioral rule to avoid the predator independently of their distance to it. During this self-organization process, we show how flock members increase their consensus over which direction to escape and turn collectively as the predator gets closer. Our results suggest that coordination among flock members, combined with simple escape rules, reduces the cognitive costs of tracking the predator while flocking. Such escape rules that are independent of the distance to the predator can now be investigated in other species. Our study showcases the important role of computational models in the interpretation of empirical findings of collective behavior.     

The impact of individual perceptual and cognitive factors on collective states in a data-driven fish school model

In moving animal groups, social interactions play a key role in the ability of individuals to achieve coordinated motion. However, a large number of environmental and cognitive factors are able to modulate the expression of these interactions and the characteristics of the collective movements that result from these interactions. Here, we use a data-driven fish school model to quantitatively investigate the impact of perceptual and cognitive factors on coordination and collective swimming patterns. The model describes the interactions involved in the coordination of burst-and-coast swimming in groups of Hemigrammus rhodostomus. We perform a comprehensive investigation of the respective impacts of two interactions strategies between fish based on the selection of the most or the two most influential neighbors, of the range and intensity of social interactions, of the intensity of individual random behavioral fluctuations, and of the group size, on the ability of groups of fish to coordinate their movements. We find that fish are able to coordinate their movements when they interact with their most or two most influential neighbors, provided that a minimal level of attraction between fish exist to maintain group cohesion. A minimal level of alignment is also required to allow the formation of schooling and milling. However, increasing the strength of social interactions does not necessarily enhance group cohesion and coordination. When attraction and alignment strengths are too high, or when the heading random fluctuations are too large, schooling and milling can no longer be maintained and the school switches to a swarming phase. Increasing the interaction range between fish has a similar impact on collective dynamics as increasing the strengths of attraction and alignment. Finally, we find that coordination and schooling occurs for a wider range of attraction and alignment strength in small group sizes.     

Japanese Macaques Depend not Only on Neighbours but also on More Distant Members for Group Cohesion

A well‐known behavioural model for group aggregation is that an individual depends on a few neighbouring individuals to adjust its movement, such as departure (repulsion) from and approach (attraction) to neighbours. However, an individual may rely not only on a few closest neighbours, but also on more distant individuals, in a group of stable membership. We measured temporal changes in the local density of individuals around a focal individual and changes in distance to other focal individuals in a group of wild Japanese macaques to determine whether the macaques depended only on a few neighbours or also on more distant individuals for adjustments in cohesiveness. We used simultaneous focal animal sampling, with two observers recording the individuals' locations using a global positioning system (GPS), over three seasons. Numbers of individuals within 20 m from an animal tended to increase after 10 min when there were a small number of individuals around the animal. However, the number tended to decrease when there was a larger number of individuals. It remained similar when there were an intermediate number of individuals. The two focal animals tended to separate after 10 min when the interindividual distance was short. However, they tended to move closer when far apart. They remained a similar distance apart when they were at an intermediate distance. Contact calls, which are suggested to function as locating group members and keeping cohesiveness, were emitted more frequently when the distance between the two focal animals was very large in two seasons. However, the rate of contact calls was not influenced by the number of individuals within 20 m from an animal. These results suggest that individual Japanese macaques do not only rely on a few closest neighbours, but also rely on more distant group members. Japanese macaques may know the general whereabouts of the whole group, and when they stay at the periphery of the group, they may emit contact calls frequently and move towards the central zone so as not to become separated from the group.     

How the Spatial Position of Individuals Affects Their Influence on Swarms: A Numerical Comparison of Two Popular Swarm Dynamics Models

Schools of fish and flocks of birds are examples of self-organized animal groups that arise through social interactions among individuals. We numerically study two individual-based models, which recent empirical studies have suggested to explain self-organized group animal behavior: (i) a zone-based model where the group communication topology is determined by finite interacting zones of repulsion, attraction, and orientation among individuals; and (ii) a model where the communication topology is described by Delaunay triangulation, which is defined by each individual''s Voronoi neighbors. The models include a tunable parameter that controls an individual''s relative weighting of attraction and alignment. We perform computational experiments to investigate how effectively simulated groups transfer information in the form of velocity when an individual is perturbed. A cross-correlation function is used to measure the sensitivity of groups to sudden perturbations in the heading of individual members. The results show how relative weighting of attraction and alignment, location of the perturbed individual, population size, and the communication topology affect group structure and response to perturbation. We find that in the Delaunay-based model an individual who is perturbed is capable of triggering a cascade of responses, ultimately leading to the group changing direction. This phenomenon has been seen in self-organized animal groups in both experiments and nature.     

THE COMPOSITION AND BEHAVIOUR OF SOME MIXED-SPECIES BIRD FLOCKS IN SARAWAK

Mixed-species flocks of birds feeding on insects were observed mainly in forest in Sarawak at a time of year when insect availability is known to be near its annual minimum. The approximate individual and specific composition of most flocks were noted. Twenty-six species occurring in 25% of flocks were regarded as occasional members, 15 species occurring in 40% of flocks being classed as regular members. Recording species' feeding behaviour was the main priority and few observations of inter- and intra-specific interactions within the flock, horizontal distribution or vertical stratification were made, although the last proved to be of potential significance. From the analysis of feeding behaviour two groups of species were distinguished. The larger group contained those usually having an exclusive common feeding pattern and showing very little overlap with the other such species. Fewer species used a wide range of feeding methods, all of which were shared with other species, but the least overlap was with other members of the same category. The observed composition of these flocks, in terms of regular members, might be interpreted as ensuring a low level of inter-specific competition and it is suggested that the less specialized foragers may occur in the flocks by utilizing the ‘gaps’ between, and absences of, specialists. The possible advantages of membership of mixed-species flocks are briefly considered and the likelihood of the selective advantage in any situation being the result of a balance of factors emphasized. In the apparent absence of regular potential predators the existence of these flocks is interpreted primarily as an adaptation for augmenting available insect food, particularly perhaps at critical times of year, by flushing insects as a result of the foraging activities of flock members. It is suggested that the varied responses of insects on being disturbed coupled with the different and fairly specialized feeding techniques of the birds could ensure benefit for all members of the flock.     

The feeding success of cattle egrets in flocks

The feeding rates of grouped (<1.5 m from conspecifics) and solo (>5 m from conspecifics) cattle egrets (Bubulcus ibis) in loose flocks away from cows were compared, to test the hypothesis that grouped cattle egrets benefit from feeding on prey flushed inadvertently by nearby conspecifics. The flock feeding rates were also compared to those of grouped and solo egrets near cows, to determine the effects of flock membership on feeding rates. Birds in flocks captured prey faster than those with cows, and tended to capture larger prey, but field observations and captive experiments failed to show that the feeding success of flock members was enhanced by the hypothesized ‘beater’ effect. Increases in prey density, however, always resulted in higher feeding rates, so some cattle egret groups may form in response to local concentrations of prey. Prey size may also play a role in group formation, because birds in the field tended to feed at greater distances from their neighbours when larger prey were captured, regardless of prey density. When small groups did form among cattle egrets feeding on relatively large prey, group members occasionally captured prey items that had been discovered by nearby conspecifics. This behaviour was not observed among birds in dense aggregations, which fed on small, highly abundant prey. These data indicate that there is a potential cost associated with feeding too near others unless the prey are relatively small and abundant.     

Collective Decision-Making in Homing Pigeons: Larger Flocks Take Longer to Decide but Do Not Make Better Decisions

Social animals routinely are challenged to make consensus decisions about movement directions and routes. However, the underlying mechanisms facilitating such decision-making processes are still poorly known. A prominent question is how group members participate in group decisions. We addressed this question by examining how flocks of homing pigeons (Columba livia) decide their homing direction. We released newly formed flocks varying in size and determined the time taken to choose a homing direction (decision-making period) and the accuracy of that choice. We found that the decision-making period increases exponentially with flock size, which is consistent with a participatory decision-making process. We additionally found that there is no effect of flock size on the accuracy of the decisions made, which does not match with current theory for democratic choices of flight directions. Our combined results are better explained by a participatory choice of leaders that subsequently undertake the flock directional decisions. However, this decision-making model would only entirely fit with our results if leaders were chosen based on traits other than their navigational experience. Our study provides rare empirical evidence elucidating decision-making processes in freely moving groups of animals.     

Resilience and Controllability of Dynamic Collective Behaviors

The network paradigm is used to gain insight into the structural root causes of the resilience of consensus in dynamic collective behaviors, and to analyze the controllability of the swarm dynamics. Here we devise the dynamic signaling network which is the information transfer channel underpinning the swarm dynamics of the directed interagent connectivity based on a topological neighborhood of interactions. The study of the connectedness of the swarm signaling network reveals the profound relationship between group size and number of interacting neighbors, which is found to be in good agreement with field observations on flock of starlings [Ballerini et al. (2008) Proc. Natl. Acad. Sci. USA, 105: 1232]. Using a dynamical model, we generate dynamic collective behaviors enabling us to uncover that the swarm signaling network is a homogeneous clustered small-world network, thus facilitating emergent outcomes if connectedness is maintained. Resilience of the emergent consensus is tested by introducing exogenous environmental noise, which ultimately stresses how deeply intertwined are the swarm dynamics in the physical and network spaces. The availability of the signaling network allows us to analytically establish for the first time the number of driver agents necessary to fully control the swarm dynamics.     

Experimental analysis of the social value of flocking by starlings (Sturnus vulgaris) in relation to predation and foraging

In groups of ten, indidual starlings, Sturnus vulgaris, spent significantly less time in surveillance than did individuals in smaller groups and responded more quickly than single birds to a flying model hawk. Captive starlings in flocks reduce their individual surveillance efforts, but their combined efforts still enable them to be more effective than single birds in the detection of predators. Foraging behaviour of flocks was observed by placing single starlings with groups of tricoloured blackbirds, Agelaius tricolor; the starlings reduced the time they devoted to surveillance at the same rate as if they were with other starlings.     

The content and availability of information affects the evolution of social-information gathering strategies

Social animals can gather information by observing the other members of their groups. Strategies for gathering this type of social information have many components. In particular, an animal can vary the number of other animals it observes. European starlings (Sturnus vulgaris) in flight pay attention to a number of neighbors that allows the flock to reach consensus quickly and robustly. The birds may do this because being in such a flock confers benefits on its members, or the birds may use the strategy that is individually beneficial without regard for the flock’s structure. To understand when individual-level optimization results in a group-level optimum, we develop a model of animals gathering social information about environmental cues, where the cue can be about either predators or resources, and we analyze two processes through which the number of neighbors changes over time. We then identify the number of neighbors the birds use when the two dynamics reach equilibrium. First, we find that the equilibrium number of neighbors is much lower when the birds are learning about the presence of resources rather than predators. Second, when the information is about the presence of predators, we find that the equilibrium number of neighbors increases as the information becomes more widespread. Third, we find that an optimization process converges on strategies that allow the flock to reach consensus when the information is about the presence of abundant resources, but not when it is about the presence of scarce resources or predators.     

Collective motion from local attraction

Many animal groups, for example schools of fish or flocks of birds, exhibit complex dynamic patterns while moving cohesively in the same direction. These flocking patterns have been studied using self-propelled particle models, most of which assume that collective motion arises from individuals aligning with their neighbours. Here, we propose a self-propelled particle model in which the only social force between individuals is attraction. We show that this model generates three different phases: swarms, undirected mills and moving aligned groups. By studying our model in the zero noise limit, we show how these phases depend on the relative strength of attraction and individual inertia. Moreover, by restricting the field of vision of the individuals and increasing the degree of noise in the system, we find that the groups generate both directed mills and three dynamically moving, ‘rotating chain’ structures. A rich diversity of patterns is generated by social attraction alone, which may provide insight into the dynamics of natural flocks.     

Multi-species territoriality and dynamic of neotropical forest understorey bird flocks

1. Eleven contiguous mixed-species bird flocks, with colour-banded individuals, were monitored continuously during 3 years in a 132-ha study area of primary rainforest in French Guiana.
2. Flock members were divided into six categories according to their flocking propensity and occurrence: 10 core or permanent species and 56 regular, occasional or incidental species. Each core species was represented by a single breeding pair with their fledglings and extra 'floaters' (unmated subadults and adults).
3. Flock home ranges overlapped slightly, but were communally defended by all core species in areas of overlap. Their size varied from 3·2 to 14·3 ha and was inversely correlated with vegetation density, but not flock size or species composition.
4. Flock number, size and composition, as well as boundaries were highly stable between seasons and years. Each flock had a single permanent gathering site and bathing site in late afternoon, the latter sometimes shared by 2–3 flocks.
5. Core species produced 0·18–0·73 fledglings per pair per year, which stayed in their natal flock for 200 to over 421 days. Then, these individuals usually moved between two and six different flocks, sometimes for up to 3 years, before finding a mate and a flock where they could settle and breed. Once breeding, they probably remained for life in the same flock. The mean annual survival rate was at least 0·75.
6. This highly evolved and stable organization, associated with a low breeding success and high survival rate was a critical factor maintaining low species density, delayed reproduction and a proportion of floating individuals buffering population fluctuations.
7. These social groups with their multi-species territoriality and co-evolved roles of flock members were similar to those described elsewhere in South America. They seem to be a general phenomenon in neotropical lowland rainforests.     

Evidence for a rule governing the avoidance of superfluous escape flights

When an imminent attack by a predator on a group of birds is signalled to non-detectors only by the departure of the detector, non-detectors may make time-wasting false-alarm flights in response to mistaken or non-predator-driven departures. The frequency of false-alarm flights might be reduced if group members assess the reason for single departures before responding. Immediate flights should only occur after multiple simultaneous departures, because these are only likely to be generated by an attack. The response delay between the detectors' departure and the next birds that respond should then be dependent on the number of detectors. On sparrowhawk attack, response delays in redshanks decreased significantly as detector number increased, controlling for raptor conspicuousness and proximity, and flock size and spacing. If response delay is modified because of risk dilution, it should increase with flock size and, consequently, the rate of alarm flights due to mistakes should decrease. However, response delay did not increase and flight frequency due to misidentification of non-raptors or non-predator-driven departures did not decrease with flock size. Significantly more feeding time was lost by birds in small flocks, suggesting that the dilution effect decreased the cost of each false-alarm flight rather than their frequency.     

On the feeding and social behaviour of the laying hen

This paper describes a study of the influences of early husbandry conditions, social attraction and social rank on various aspects of the feeding behaviour of laying hens.Birds were raised in flocks of 10, 60 or 500. Groups of 3 birds, selected from flocks of the same size, were then housed in pens. Some groups consisted of hens raised in the same flock, and some of birds raised in different flocks. The feeding and agonistic behaviour of each group at each of 5 types of feeder was observed and compared with the behaviour shown when the birds had free access to a 1-m long food trough. Each of the 5 feeders offered the same area of feeding space, but differed in its partitioning and spatial distribution. One feeder had a single unpartitioned feeding space. The other 4 feeders had 3 partioned feeding spaces which were adjacent, or separated by distances of 10, 20 or 40 cm, respectively.For the 5 feeders, total feeding times and lengths of feeding bouts were greatest, and the number of feeding bouts least, when the feeding space was unpartitioned. Synchrony of feeding behaviour was low when the feeding space was unpartitioned or the partitioned spaces adjacent, but was comparable to that at the 1-m long food trough when the distance between partitioned feeding spaces was 10 cm or greater. When feeding space was partitioned, the likelihood that 2 birds would eat together at the same site increased with the distance between feeding space. Dominant birds always exhibited the longest feeding bouts and greatest total feeding times, but were less likely to feed in the same space as another bird, and exhibited less synchrony of feeding behaviour than subordinates. The size of the flock in which the birds were raised, and whether or not the birds in a group had been raised together or apart, had no clear effect on behaviour.These results indicate that, within the limits of this experiment, early husbandry conditions do not influence behaviour shown during feeding in later life, and that social attraction has a greater influence on the feeding behaviour of hens than is generally assumed. In view of this latter finding, it is postulated that in attempting to determine the requirements of laying hens for feeding space, attention must be paid to social attraction as well as to competition at the feeder.     

The prevalence of Campylobacter amongst a free-range broiler breeder flock was primarily affected by flock age

Campylobacter successfully colonizes broiler chickens, but little is known about the longer term natural history of colonization, since most flocks are slaughtered at an immature age. In this study, the prevalence and genetic diversity of Campylobacter colonizing a single free-range broiler breeder flock was investigated over the course of a year. The age of the flock was the most important factor in determining both the prevalence and diversity of Campylobacter over time. There was no correlation with season, temperature, the amount of rain and sunshine, or the dynamics of colonization amongst geographically and temporally matched broiler flocks. The higher prevalence rates coincided with the age at which broiler chickens are typically slaughtered, but then in the absence of bio-security or other intervention methods, and despite changes in flock management, the prevalence fell to significantly lower levels for the remainder of the study. The genetic diversity of Campylobacter increased as the flock aged, implying that genotypes were accumulated within the flock and may persist for a long time. A better understanding of the ecology of Campylobacter within commercial chicken flocks will allow the design of more effective farm-based interventions.