Although songtypes are generally considered to be important functional units in birdsong, they have not been well‐characterized in terms of within‐ and between‐songtype variation. We analyzed the song repertoires of 12 adult male song sparrows (Melospiza melodia) from a population in New York. We identified minimal units of production (MUPs), and calculated the degree to which song variants within a bird's repertoire were similar to each other based on an analysis of MUP sharing. Using statistical techniques from numerical taxonomy, we assessed how song variants clustered into songtypes, and we derived quantitative measures of within‐songtype and between‐songtype similarity. We found that birds produced a limited number of songtypes, but constantly produced new song variants within the framework of these types. Most song variants were produced very rarely; over 43% of song variants were produced only once. Repertoires differed in the degree of between‐songtype similarity and in the number of songtypes defined, but there was comparatively less variation in within‐songtype similarity. Between‐songtype similarity and repertoire size were positively correlated. We argue that song sparrow songtypes are probabilistic units of song production, and discuss the functional and evolutionary implications of having vocal motor patterns organized in this way.

No abstract

No abstract