The juvenile life stage is a formative and perilous time for young animals. In songbirds, juveniles must also learn to sing. Although song learning is intensively studied in the laboratory, the behavior of wild juvenile birds is virtually unknown. Here I describe four research projects examining social, hormonal, and ecological factors affecting song learning patterns in song sparrows (Melospiza melodia).

I used radio telemetry to track the spatial movements of juvenile male sparrows and observe their behavior during song learning. Subjects rarely associated with adults but frequently associated with other juvenile males. Juveniles formed complex social networks and stable social relationships, which varied by season and by individual. Individuals that associated together often settled near each other and shared song types, indicating that early social associations among juveniles can have long-term effects on future behavior.

In the laboratory, I studied how testosterone affects song crystallization and learning. Short-term testosterone (T) implants in the autumn rapidly increased song rates and song stereotypy in juveniles and reduced further song learning. These results suggest that testosterone functions to promote song crystallization and terminate the sensitive phase for song memorization.

Laboratory research suggests that birds may learn song by eavesdropping on adult interactions. I tested this “social eavesdropping hypothesis” in the field by playing recordings simulating song interactions or solo singing to radio-tagged juveniles. Subjects approached song interactions but treated solo singing equivalently to heterospecific song indicating that juveniles actively eavesdrop on adult song interactions. By eavesdropping, birds may glean information important for song learning including song structure, rules for singing, or adult dominance levels.

Why juveniles copy songs from certain adults is a puzzle. I tested whether juveniles assess adult dominance or aggression levels when choosing song tutors. Using adult and juvenile mounts with age-appropriate song playback, I conducted simulated territorial intrusions with adults known to be good or bad tutors. Subjects responded more strongly to adult than juvenile intruders in the summer, but increased aggression toward juveniles in autumn and early spring. Good tutors responded more aggressively to territorial intrusion than bad tutors indicating that juveniles selectively learn from aggressive adults.