干旱区生态环境知识资源中心

We present data on variation in frequency and duration characters of the advertising song of Zonotrichia capensis, the Rufous-collared sparrow, and information on the qualitative structure of the introductory "theme." These data are analyzed with respect to their relationships with altitude, habitat type, body size, syrinx size and the dialect variation shown by the terminal trill.

Principal components analysis shows that the major axis of variation (PC1) in the frequency and duration variables is one primarily of increasing frequency and bandwidth of the trill (mainly due to increasing maximum trill frequency), increasing frequency of the theme, and of increasing song length, mainly contributed by theme length; increasing values on PC2 correspond to an increasing theme bandwidth and maximum theme frequency, with decreasing theme length and increasing trill length.

PC1 scores from this analysis are negatively correlated with altitude over the whole sample. Higher-altitude habitats are usually structurally open, lower-altitude habitats usually mixed or closed. Songs from the nine categories of original, natural vegetation differ significantly in their PC1 scores, while contemporary vegetation structure has no significant effect: songs from the open-country habitats, desert, puna and grassland are shorter, have lower frequency and narrower bandwidth than all woodland, thornscrub and forest songs.

With the exception of the very slow-trilled songs from the Monte desert dialect, there is a positive relationship between trill interval and PC1 score: slower-trilled songs (longer trill interval; lower trill rate) are longer, of higher frequency and broader bandwidth. The slow-trill Monte dialect songs are anomalous in having PC1 characteristics like the fastest-trilled dialect songs (puna and grassland).

There is a significant negative relationship between body size and PC1, though it is nonlinear: birds from all habitats but puna are very similar in having smaller body size. Syrinx size is not correlated either with measures of body size or with habitat. It is concluded that most variation in song modal frequency and bandwidth is due to learning processes, rather than to size constraints of the body or organs.