Abstract
Leopard seals are solitary animals, which communicate via acoustic sequences during their breeding season, producing a sequence of stereotyped vocalisations underwater. These vocalisation bouts consist of ordered sequences of discrete elements from a shared vocal repertoire, and the patterns that underlie the choice of sounds in these acoustic displays could be used for encoding individual identity. To investigate the properties of these sequences, we used information theory to estimate the entropy of the calling bouts of a population of 26 leopard seals from Davis Sea, Eastern Antarctica. We used three estimators for comparison – an independent identically distributed (i.i.d.) model, a first order Markov model, and a sliding window match length (SWML) estimator. The majority of the SWML estimates are lower than the i.i.d. and first order Markov estimates, suggesting that the leopard seal vocal sequences have more complex temporal structures than can be modelled by first order Markov chains. This finding suggests that the temporal structure of male leopard seal song conveys information to receivers of male mating displays. Our results highlight the potential of the temporal structure of these stereotyped sequences to convey cues of individual identity in leopard seal calling bouts.