Individual signatures outweigh social group identity in contact calls of a communally nesting parrot

Despite longstanding interest in the evolutionary origins and maintenance
of vocal learning, we know relatively little about how social dynamics
influence vocal learning processes in natural populations. The “social
group membership” hypothesis proposes that socially learned calls evolved
and are maintained as signals of group membership. However, in
fission-fusion societies, individuals can interact in social groups across
various social scales. For learned calls to signal group membership over
multiple social scales, they must contain information about group
membership over each of these scales, a concept termed “hierarchical
mapping”. Monk parakeets (Myiopsitta monachus), small parrots native to
South America, exhibit vocal mimicry in captivity and fission-fusion
social dynamics in the wild. We examined patterns of contact call acoustic
similarity in Uruguay to test the hierarchical mapping assumption of the
signaling group membership hypothesis. We also asked whether geographic
variation patterns matched regional dialects or geographic clines that
have been documented in other vocal learning species. We used visual
inspection, spectrographic cross-correlation and random forests, a machine
learning approach, to evaluate contact call similarity. We compared
acoustic similarity across social scales and geographic distance using
Mantel tests and spatial autocorrelation. We found high similarity within
individuals, and low, albeit significant, similarity within groups at the
pair, flock and site social scales. Patterns of acoustic similarity over
geographic distance did not match mosaic or graded patterns expected in
dialectal or clinal variation. Our findings suggest that monk parakeet
social interactions rely more heavily upon individual recognition than
group membership at higher social scales.