To identify individual birds based on vocalizations, it is necessary to first confirm that individuals produce distinct vocalizations that are stable over time. Using vocal features, I was able to estimate the Ryukyu Scops Owls population on Minami-daito Island in 2003–2012 and confirmed that it is possible to identify individuals over time. I analysed 20 hoots per individual male in 2005–2007. Using 11 hoot parameters, the discriminant function analyses (DFAs) achieved a 97.4% probability of correct classification. Using 31 marked males, I examined the vocal features of 118 series of hoots over 1–6 years. Each male’s vocal features remained essentially stable over time and the calls of 14.9% (13/87) males were wrongly identified as those of other males. Individual misidentifications were the result of subtle differences in spectral features, or imperceptible differences in frequency and duration. The typological features (to visually classify variation in spectrograms according to the shape of syllables in the owl species) of the sound spectrograms of the misclassified males were found to be consistent over several years. When considering the efficacy of the DFA as a tool for identifying individuals, we rely on their vocal features, but should examine the visible typological features.
Discriminant function analysis, Otus elegans interpositus, owl population, small oceanic island, spectrogram, vocal identification