MEPS 425:261-268 (2011)  -  DOI: https://doi.org/10.3354/meps09022

ABSTRACT: Many species of animals live in groups, but the mechanisms responsible for the formation and cohesion of groups are not well known and have not often been quantified. Different models of aggregations, all making unique assumptions about rates of arrivals to and departures from groups, produce different group size distributions. Fitting such models to empirical distributions of group sizes can provide insights into the mechanisms that underlie aggregations in animals. Four models predicting the distribution of group sizes (the power law, Poisson, exponential and negative binomial models) were tested in 7 species of Northwest Atlantic seabirds using maximum-likelihood estimation and a random factor to control for differences in mean group size among seabird surveys. Of the 4 models tested, the power law ranked first in most species in terms of raw Akaike’s information criterion values. Nevertheless, likelihood-ratio tests revealed that the power law and the negative binomial model, which ranked second in most species, often provided a statistically similar fit to the data. Goodness-of-fit tests revealed a rather good fit of these models to the data. The finding that aggregation models can apply to the distribution of group sizes offers the opportunity to examine quantitatively the mechanisms or underlying processes that give rise to aggregations in seabirds and possibly other animals.

KEY WORDS: Seabirds · Power law · Maximum likelihood · Group size