This paper presents a simple stochastic model of the dynamics of waterfowl aggregation, investigates model fit, and considers
variance in the model’s parameter estimators. The model, a marked-point process with four parameters, describes a behavioral
process defined by the movements of animal groups. My approach provides new methods to explore animal social behavior. I
illustrate the fit of the model to field observations of 39 aggregations observed when they were not at equilibrium and outline a
procedure for determining which parameters, and hence which behaviors, are constant across the 39 aggregations. A comparison
of the predicted variablity in the model’s parameter estimates, assuming constant parameter values for all 39 aggregations, to
the observed variablity in the estimates suggests that waterfowl aggregation dynamics are affected by changing movement rates
and not changes in group size: simulations reveal that the estimates of the arrival and departure rates are more variable than
expected. In contrast, the estimates of the parameters describing movement group size do not show excessive variability. The fit
of a constant parameter modified-geometric distribution to the group size data indicates that the formation of movement groups
is not a complicated process. Analysis of the extra variability in the movement rates reveals two variables, migratory season and
overall waterfowl density, that influence the behavior of aggregating birds.
© 2003 Elsevier B.V. All rights reserved.
Keywords: Waterfowl; Aggregation; Social behavior; Marked-point process; Markov model