Ontogeny of vocal learning in a hummingbird

Abstract

Hummingbirds have evolved to learn song from conspecifics independently of other bird clades that show this behaviour, such as parrots and oscine passerines. Little is known about how vocal learning occurs in hummingbirds. To determine the conditions necessary and sufficient for learning, we raised young Costa's hummingbirds, Calypte costae, in isolation in two experiments. The first experiment tested whether three pairs of males, raised either in silence or with only acoustic exposure to song, would produce wild-type song. They did not: each pair instead produced ‘isolate’ songs dissimilar to wild-type song. Each pair produced a unique isolate song with subtle differences from other pairs. Thus, exposure to song alone is insufficient to trigger learning. The second experiment tested whether individually housed males would learn songs when exposed to a live male model plus playback of song (1 h/day, 3 days/week, for 3 months). Seven of eight birds that heard Costa's-like songs learned to sing their tutor song. Each individual produced song that matched idiosyncratic features of the individual songs played to them, demonstrating attention to, and learning of the exact acoustic stimuli received. Two control birds exposed to acoustically dissimilar Anna's hummingbird, Calypte anna, song developed isolate song. Thus, exposure to species-specific song with a conspecific male is sufficient to trigger learning. The sensory phase began after fledging, around 23 days posthatch, and the sensorimotor phase began around 50 days posthatch, when singing also began. Song crystallization occurred approximately 125 days posthatch. Costa's also appear to be statistical learners; they learn to sing the songs they hear most often. Hummingbird song ontogeny appears to be similar to song ontogeny in passerines, despite its convergent origin.