Abstract: This paper proposes an extension of generalized linear models to the analysis of longitudinal data. We introduce a class of estimating equations that give consistent estimates of the regression parameters and of their variance under mild assumptions about the time dependence. The estimating equations are derived without specifying the joint distribution of a subject's observations yet they reduce to the score equations for niultivariate Gaussian outcomes. Asymptotic theory is presented for the general class of estimators. Specific cases in which we assume independence, m-dependence and exchangeable correlation structures from each subject are discussed. Efficiency of the pioposecl estimators in two simple situations is considered. The approach is closely related to quasi-likelihood.

Abstract: We spent two winters studying the social behaviour of wild woodland caribou (Rangifer tarandus caribou) at a time when their main food (ground lichens; Cladina sp.) is available only at snow craters dug by the animals. The competition for access to such craters was severe, the animals constantly trying to take over the craters of others. During a two-month period when a group maintained a constant size (20) and composition (all age-sex classes represented), we could rank the animals in a rather linear dominance hierarchy (Landau's index = 0.87). Rank was correlated with access to resources, percent of time spent active, and percent of time feeding in craters. It was also correlated with age and antler size. However, rank is not an attribute of individuals, but of a relationship between individuals. As such it is only an intervening variable between physical attributes and access to resources, a variable whose value has meaning only within a given group. Among the three attributes studied (age, sex, antler size), the latter was by far the best predictor of the occurrence and outcome of interactions. Between two individuals within any of the three age-sex classes studied (adult and yearling males and adult females), the one with larger antlers initiated significantly more often, escalated its aggression (to the point of hitting the target) less often, and enjoyed a higher success rate in obtaining resources. When their antlers were larger than those of an adult male target (i.e. males that had shed their antlers), adult females won almost all their interactions with adult males even though they escalated only one fourth of them. This clarifies the long-standing speculation that female caribou have antlers and shed them later than males, in order to overcome their sexual handicap in competition for food in the winter. We conclude that the link between rank and dominance of an individual on one hand, and some of its attributes on the other (e.g. sex, age, weight, antler size) is fundamentally realized by the animal itself through its active preference for targets it is likely to beat, i.e. targets with smaller antlers.

Abstract: We investigated the foraging producer-scrounger system of a captive flock of feral pigeons (Columba livia) by monitoring the number of food patches each individual produced. In one experiment, three different patch types were tested on the whole flock while, in a second, flock composition was varied for one patch type. In all cases we found non-uniform distributions of the number of patches produced per individual, which suggests the existence of producer and scrounger roles. This result could not be explained by either dominance or variability in individual learning ability. Individuals switched roles in response to changes both in food patch type and flock composition. These results are discussed in light of the skill pool hypothesis, which suggests that, in a group, different foraging specialists will profit by parasitizing each other's food discoveries.