Abstract: Theory predicts that individuals living in fission-fusion societies, in which group members frequently change subgroups, should modify grouping patterns in response to varying social and environmental conditions. Spotted hyaenas, Crocuta crocuta, are long-lived carnivores that reside in permanent social groups called clans. Clans are complex, fission-fusion societies in which individual members travel, rest and forage in subgroups that frequently change composition. We studied two clans in Kenya to provide the first detailed description of fission-fusion dynamics in this species. Because social and ecological circumstances can influence the cohesiveness of animal societies, we evaluated the extent to which specific circumstances promote the formation of subgroups of various sizes. We found that cooperative defence of shared resources during interclan competition and protection from lions were cohesive forces that promoted formation of large subgroups. We also tested hypotheses suggesting factors limiting subgroup size. Mothers with small cubs avoided conspecifics, thereby reducing infanticide risk. Victims of aggression either reconciled fights or separated from former opponents to reduce the immediate costs of escalated aggression in the absence of food. As predicted by the ecological constraints hypothesis, hyaenas adjusted their grouping patterns over both short and long time scales in response to feeding competition. Crocuta were most gregarious during periods of abundant prey, joined clanmates at ephemeral kills in numbers that correlated with the energetic value of the prey and gained the most energy when foraging alone because cooperative hunting attracted numerous competitors. Overall, our findings indicate that resource limitation constrains grouping in this species.

Abstract: I develop guidelines for assessing the precision and power of statistical techniques that are frequently used to study nonhuman social systems using observed dyadic associations. Association indexes estimate the proportion of time that two individuals are associated. Binomial approximation and nonparametric bootstrap methods produce similar estimates of the precision of association indexes. For a mid-range (0.4-0.9) association index to have a standard error of less than 0.1 requires about 15 observations of the pair associated, and for it to be less than 0.05, this rises to 50 observations. The coefficient of variation among dyads of the proportion of time that pairs of individuals are actually associated describes social differentiation (S), and this may be estimated from association data using maximum likelihood. With a poorly differentiated population (S~0.2), a data set needs about five observed associations per dyad to achieve a correlation between true and estimated association indexes of r=~0.4. It requires about 10 times as much data to achieve a representation with r=~0.8. Permutation tests usually reject the null hypothesis that individuals have no preferred associates when S2H>5, where H is the mean number of observed associations per individual. Thus most situations require substantial numbers of observations of associations to give useful portrayals of social systems, and sparse association data inform only when social differentiation is high.

Abstract: It has been suggested that social dynamics affect social learning but empirical support for this idea is scarce. Here we show that affiliate relationships among kin indeed enhance the performance of common ravens, Corvus corax, in a social learning task. Via daily behavioural protocols we first monitored social dynamics in our group of captive young ravens. Siblings spent significantly more time in close proximity to each other than did nonsiblings. We subsequently tested birds on a stimulus enhancement task in model-observer dyads composed of both siblings and nonsiblings. During demonstration the observer could watch the model manipulating one particular object (target object) in an adjacent room. After removing the model, the observer was confronted with five different objects including the former target object. Observers from sibling dyads handled the target object for significantly longer periods of time as compared with the other four available objects, whereas observers from nonsibling dyads did not show a preference for the target object. Also, siblings matched the model's decision to cache or not to cache objects significantly more often than did nonsiblings. Hence, siblings were likely to attend to both, the behaviour of the model (caching or noncaching) and object-specific details. Our results support the hypothesis that affiliate relations between individuals affect the transmission of information and may lead to directed social learning even when spatial proximity has been experimentally controlled for.

Abstract: Our understanding of how, why, and the circumstances under which animals use social information has been facilitated by three principal areas of research, social learning, public information use and social eavesdropping. With few exceptions, these related concepts have remained remarkably distinct within the literature, with little discussion or integration among them. Are these distinctions warranted? We tackle the issue by exploring similarities and differences between the concepts with respect to how animals gather and use social information, the type of information gathered, how information is packaged, and the relative payoffs to individuals involved. We contend that none of the currently dominant paradigms, social learning, public information use, or social eavesdropping, provide a unifying theme for studying social information use. Instead, we favour the central characteristic of the three concepts, social information use, as the overarching umbrella, and advocate a broader conceptual framework for understanding more comprehensively how animals behave with their social environments. Our intention is not to revolutionize the fields of social learning, public information use or social eavesdropping, but rather to stimulate discussion among researchers investigating the abilities of animals to extract information from the social environment.

Abstract: The question whether a certain species is or is not able to imitate has received much recent attention. However, the ability to copy a demonstrated action might depend not only on the imitative ability of the observer but also on its attention, a factor widely neglected in discussions and experiments. Since attention differs between species as well as between individuals, it is likely to influence the amount and type of information different species and/or observers may extract from a given demonstration. We studied attention in common marmosets, Callithrix jacchus. In a series of experimental sessions, individual marmosets watched different conspecific models that were searching, manipulating an object and feeding. The observers could see the demonstration through two observation holes, which allowed us to measure exactly how often and for how long they watched the model. Marmosets were more attentive towards conspecifics engaged in a problem-solving task than an exploring model. Individual variation in attention was large, ranging from less than 10% to over 49%. Attention also depended on the identity of the model with an increase in attention if it was of the opposite sex. Overall, attention of marmosets was short with a median duration of 6 s while the model manipulated an object. This study measured the real interest of the observer towards a model, which could be an important variable in social-learning experiments.