In this paper, modern time-series analysis methods are applied to the detection of possible nonlinear and chaotic features of gadfly petrel (Pterodroma sp.) sounds. Some of the recordings contain a considerable amount of noise due to wing flapping and wind. Poor signal-to-noise ratio usually impedes the chaos analysis, and thus noise-reduction methods play a significant role in the analysis. The results of two different tests are presented. First, the nonlinearity of the petrel vocalizations is tested using an information theoretic method –linear and nonlinear redundancies. Secondly, the local Lyapunov exponents are utilized as an indication of possible chaotic dynamics in the signal production system.
petrel vocalizations, nonlinear dynamics, chaos