ABSTRACT

Biologists frequently discuss behavioral attributes of species or more inclusive taxa, and organize these attributes into classifications for comparative studies. Such behavioral classifications may influence the kinds of questions biologists ask. Recently proposed revisions to a well–established definition of ‘eusocial’ behavior are intended to improve classifications of animal societies, and hence facilitate comparative studies. Crespi and Yanega (1995) propose to narrow the scope of eusociality, whereas Gadagkar (1994) and Sherman et al. (1995) aim to widen its scope. These new proposals, along with the traditional definition, are used here to discuss the important point that behavioral classifications are operational, and therefore different questions require different classifications. The revisions and the existing definition use arbitrary criteria, in the sense that a priori they do not specify ‘meaningful’ parameters. Moreover, social behavior can vary within taxa, so taxon–specific descriptions of social behavior may force a typological characterization that potentially masks interesting variability, especially for taxa that are not thoroughly studied. A natural classification would be advantageous in that it would take into account actual patterns of evolutionary history (phylogeny), and thus ensure the accurate assessment of convergent behavioral patterns. Without a natural classification, authors should precisely define terms in ways suited to their particular needs in a given study.