The issue of what maintains signal honesty is a central problem in the study of animal communication. Honesty can be maintained by the costs of signalling, but honesty can also be maintained if a signal’s production is linked to its meaning. An example of such a signal is the wing whistle of the crested pigeon, Ocyphaps lophotes, which is produced by the wings during flapping flight. When birds flee from danger they flap more rapidly and produce a whistle with a faster tempo. This acoustically distinct whistle acts as an index signal of alarm. However, it is unknown what feathers produce the high and low notes that compose the whistle, and which acoustic features signal danger.

We carried out feather-removal and playback experiments to examine both the mechanism of production and function of different wing whistle sounds. First we played back natural whistles with adjusted timing to test the importance of tempo in signalling danger. Second, we removed the highly modified 8^th primary wing feather and adjacent feathers to test their role in sound production. Finally, we tested the importance of each of these feathers in signalling alarm by playing back recordings of birds that were missing a feather.

Conspecifics did not respond to alarmed whistles that had reduced tempo, showing that high tempo is necessary to signal alarm. Feather removal experiments revealed that the 8^th primary feather produced the high notes during downstrokes and the 9^th produced the low notes during upstrokes. The playback experiment then showed that only the high notes produced by the highly modified 8^th primary are essential in signalling alarm.

Overall, we conclude that high whistle tempo – which is a consequence of escape flight – signals danger, and that the 8^th primary feather has evolved to produce the key sound by which listeners judge tempo.