A COMPARISON OF SOME TECHNIQUES USED TO ASSESS VOCAL INDIVIDUALITY

Abstract

ABSTRACT In this paper we assess the applicability to the extraction of conservation information of three commonly used methods of examining the individuality of acoustic signals: qualitative assessment, spectrographic cross-correlation and discriminant function analysis. We tested the ability of human observers, with different levels of training, to sort and match spectrograms of bittern and fantail warbler vocalisations. This simulated a census and monitoring situation. We found that training had little effect on accuracy and high inter-individual variation made generalisations difficult. Cross- correlation provides an objective measure of similarity but was shown to be sensitive to background noise and the signal structure being compared. Discriminant function analysis is a powerful descriptor of individuality and functions well when developing predictive tools. However it should not be used to count a population and its use is constrained to re-identification in populations of known size.