Acoustic identification of individual Rufous Bristlebirds, a threatened species with complex song repertoires

Abstract

Rufous Bristlebirds (Dasyornis broadbenti), possess song repertoires of between 12 and 30 different song-types, only a proportion of which are shared between different individuals. As a result, the comparison of songs among individuals is complicated by the use of multiple song-types. Using both discriminant function analysis (DFA) and spectrographic cross-correlation analysis (SPCC), those song-types that were shared among individuals demonstrated high levels of vocal individuality. DFA correctly classified an average of 98.3 ± 1.2% of male songs, whereas SPCC correctly classified an average of 87.4 ± 2.8% of songs. However, common song features among song-types showed limited value for individual identification. The information provided by the SPCC analyses was then used to discriminate among individuals over a distance of 7.8 km, the results of which supported those of previous studies with regard to the sedentary nature of Bristlebirds. Although the results of the present study show promise, the issue of song comparisons among multiple song-types remains a significant challenge for the development of acoustic identification systems for this species.