Two methods for using period length to study rhythmic phenomena

Summary

We have developed two distinct methods of biological rhythm analysis. The procedures are based on existing techniques for analysis of time series, Enright's periodogram and autocorrelation, and both of the new methods use the parameter, period length (τ), for defining oscillatory phenomena. We empirically evaluated the two types of analyses using real biological data from circadian rhythm studies in salamanders and sparrows.

The first method permits us to make a statistical comparison of period lengths between groups of animals in given treatments. This method is useful for data where the signal-to-noise ratio of the suspected rhythm is very low; and the method is not adequate for making a definitive judgment from single animals. It can best be applied to the question of whether a signal is entraining a rhythm or not and to questions of group differences in period length.

With the second method, we determined period length versus time. Using this procedure, we took into consideration the observation that the period length of many biological oscillations changes with time. The method is applicable to records from individual animals, and it can be used to compare treatment effects in individual animals. The technique can also be used to answer the common question of whether periodicityper se exists within a defined range in a time series.