Individual vocal signatures show reduced complexity following invasion

Abstract

The manner in which vocal learning is used for social recognition may be sensitive to the social environment. Biological invaders capable of vocal learning are useful for testing this possibility, as invasion alters population size. If vocal learning is used for individual recognition, then individual identity should be encoded in frequency modulation patterns of acoustic signals. Furthermore, frequency modulation patterns should be more complex in larger social groups, reflecting greater selection for individual distinctiveness. We compared social group sizes and used supervised machine learning and frequency contours to compare contact call structure between native range monk parakeets (Myiopsitta monachus) in Uruguay and invasive range populations in the U.S. Invasive range sites exhibited fewer nests and simpler frequency modulation patterns. Beechers statistic revealed reduced individual identity content and fewer possible unique individual signatures in invasive range calls. Lower estimated social densities and simpler individual signatures are consistent with relaxed selection on the complexity of calls learned for individual recognition in smaller social groups. These findings run counter to the traditional view that vocal learning is used for imitation, and suggest that vocal learning can be employed to produce individual vocal signatures in a manner sensitive to local population size.