INTRODUCTIONInterest in rhythms of animal behavior derives from the recognition that the biological value of a behavior depends as much on when it occurs as on the particular form it takes (see Enright, 1970, and Chapter 15). It follows that a meaningful ethogram of any ~pecies should describe both species-typical motor patterns and species-typical timing of behavior. This chapter surveys the variety of behavioral rhythms that have been studied and the range of species in which they have been described.There are several problems with most of the available descriptions of behavioral rhythmicity. One is analogous to a classical problem in descriptive ethology: that of determining appropriate units of analysis. Ethologists have recognized that significant features of behavior, such as the degree of apparent stereotypy, are to a large extent determined by the temporal and spatial resolution of the recording procedure (Barlow, 1968). Reports of behavioral rhythms have generally ignored this issue and have included data that are grouped in a great variety of ways across hours, days, and individuals. The temporal resolution of such behavioral recording can determine whether particular rhythms are detected and what their apparent characteristics are. A simple example is that recording only the degree of "nocturnality" of a behavior may lead to the erroneous conclusion that a rhythm has been lost when only its phase relation to the light cycle has changed.In other situations the appropriateness of particular units of analysis is more ambiguous. In a field study of monkey troops (Nilgiri langurs, Presby tis johnii), Horwich (1976) recorded hourly means of feeding and calling averaged over several days; the data obtained showed clear dawn and dusk peaks, especially of feeding. However, a finer-grained analysis of individual troops during a single day revealed previously undetected ultradian behavioral cycles. Horwich concluded that the presence of significantly different individual (or troop) patterns and day-to-day variations can be obscured by grouping data across individuals and over days in search of the population mean. But the most detailed analysis is not always the best, since excessive attention to detail and individual differences can obscure biologically significant population trends and prevent the formulation of meaningful generalizations.In order to choose among these possible descriptive strategies, it is useful first to identify some of the sources of behavioral variability among individuals and across time. If individual variability is generated as the sum of independent, stochastic determinants, then the average pattern in the population will also be the modal (i.e., most common) pattern. But if individual variability reflects the presence of naturally separate subpopulations, then the "species-typical pattern" may have little biological meaning. To take an extreme example, a bimodal activity pattern in a population might result from one subpopulation's preying on dawn-active species and another's utilizin...