Stability and change in the vocal signatures of common marmoset mobbing calls

Abstract

Alarm calls that carry information about the identity of the caller may help the receiver decide how to react. We recorded the tsik calls of six captive adult marmosets (Callithrix jacchus) in two different social groups in response to snake models at two time points (summer 2014 and January 2015). We measured eight acoustic variables including duration, inter-call interval, minimum and maximum frequency, and starting, maximum and ending peak frequency. Discriminant function analyses (DFA) confirmed that calls were individually distinct at both time periods (78.88 and 79.89% correctly classified at time one and time two, respectively). Stability of the vocal signatures was assessed using the DFA model for summer 2014 to classify calls elicited in January 2015. Although the classification rates were lower in 2015, calls were still classified more than would be expected by chance (64.50%). This suggests that acoustic signatures of common marmoset tsik calls remain fairly stable over time and therefore remain recognizable by their groupmates. However, during that six-month period, at least three (out of seven) acoustic parameters changed such that they were significantly higher or lower in all six marmosets; in two marmosets six out of seven parameters changed. Changes to individual ‘voices’ of animals, despite overall stability reflected in above-chance matching of calls over time in a DFA analysis, may have implications for acoustic research.