Discrimination of song types and variants in song sparrows

Male song sparrows,Melospiza melodia, sing highly variable songs. Traditionally, researchers have partitioned this variability by assigning songs to discrete categories termed ‘song types’, but researchers also have recognized that songs classified as the same song type are themselves variable. Territorial playbacks were used in a habituation/recovery design to investigate whether songs sparrows perceive songs classified as separate song types to be more distinct from one another than songs classified as variants of the same song type. Cluster analysis was used to classify playback songs as song types or variants. Playbacks consisted of focal songs repeated for 1 h, followed by a switch to 6 min of either (1) a second variant of the focal song type from the same source male or (2) a second song type of the same source male. Twelve sets of playback tapes were used in a total of 96 trials. Subjects showed habituation in the response measure (distance to the speaker) during playback of the focal songs. Response recovered for between-song type switches, with mean distance to the speaker decreasing significantly between the last 3 min pre-switch and the first 3 min post-switch, and between the last 6 min pre-switch and the 6 min post-switch. For within-type switches, recovery was significant when comparing only the last and first 3 min. Recovery was significantly greater for between-type switches than for within-type switches on both measures. These results suggest that in the perception of male song sparrows, different song types are more distinct than are different variants of a single type. The problem of how to categorize behaviour is a general one. Ideally, we would like categories that not only have clear and objective definitions, but that also mirror the way in which the behaviour patterns in question are categorized by the animals that use them. A good example of this problem is encountered in classifying the songs produced by male birds. In many species of songbirds, individual males sing variable songs, and it is customary for human observers to partition this variation by dividing songs into discrete categories termed ‘song types’ (Kroodsma 1982). Much research assumes the biological relevance of song-type classifications, for example, studies correlating the number of song types sung by males with their mating success (Howard 1974; Searcy 1984) or reproductive success (McGregor et al. 1981; Hiebert et al. 1989). Song-type classifications typically are arrived at by sorting spectrograms of songs according to relatively subjective criteria of similarity and dissimilarity. The end product of such procedures is a classification that accounts for much of the variation in an individual’s songs, but often there is still variation remaining within the resulting song-type categories (Kroodsma 1982). Song-type classifications, accordingly, are vulnerable to two criticisms: first, that a more objective classification of songs might produce different categories; and, second, that even if an objective classification did produce the same categories, the birds singing and responding to the songs might not recognize those categories, that is, they might not classify songs in the same way. The first criticism was addressed recently for our study species, the song sparrow, Melospiza melodia, by Podos et al. (1992). In the present paper, we address the second criticism, by investigating whether songs classified by us as separate song types are perceived by song sparrows as being more different from one another than are songs classified as variants of the same song type. Early workers (Wheeler & Nichols 1924; Nice 1943) began the practice of classifying song sparrow songs into song types. Nice (1943), for example, described a sample of 40 males in her 0003–3472/95/051219+08 $08.00/0 ? 1995 The Association for the Study of Animal Behaviour