Abstract:Individual recognition is an important element of social interactions among animals. While the presence of individually distinct vocalisations (providing a basis for individual recognition) has been widely tested across species, information about which components of a call encode this information is lacking. We investigated whether female alpaca ( Vicugna pacos ) vocalisations, particularly their hums, encode information about individual identity and explored which parameters contribute to this encoding. We recorded vocalisations from 9 adult female alpaca and extracted both spectro-temporal features (frequencies and duration) and mel-frequency cepstral coefficients (MFCC). Random forest analyses revealed clear individual differences in both datasets, with the spectro-temporal features allowing for slightly more accurate classification than MFCC (71% and 66.5% accuracy for spectro-temporal features and MFCC, respectively). These robust acoustic identities have the potential to provide a basis for individual recognition in alpaca, which could have important flow-on effects for alpaca communication, as it allows receivers to modulate their response to the callers' identity. Alpaca, as herd-living and vocal animals, provide an excellent model system for better understanding the mechanisms, causes and consequences of recognition and inter-individual communication.

Robust encoding of acoustic identity in alpaca hums – a basis for individual recognition