The Importance of Invariant and Distinctive Features in Species Recognition of Bird Song

Two hypotheses concerning the relative importance of different acoustic features in species-song recognition were examined. The invariant-features hypothesis predicted that a feature's accuracy in song recognition is inversely proportional to its intraspecific variability, as expressed by the coefficient of variation. The sound-environment hypothesis assumes that both variability and a feature's central tendency relative to other species in the local acoustic environment determine a feature's importance in species-song recognition. These hypotheses were compared in an acoustic analysis of 14 features of songs of two focal species, the Field Sparrow (Spizella pusilla) and Chipping Sparrow (S. passerina), and 11 other species found in open habitats in the eastern United States. In both focal species, maximal note frequency was one of the two least-variable features within conspecific song, and the most accurate single feature in classifying songs in a canonical discriminant analysis. In neither species was a feature's accuracy in classifying songs correlated with its relative variability within conspecific song. These results, in conjunction with recent experimental studies, contradict the common assumption in the literature that invariant features provide the most important song recognition cues.