Sounds, acoustic behaviour and gillnet entanglement of Hector's dolphin.
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Vocalisations of free-ranging Hector's dolphins were recorded with wide-band recording equipment. Preliminary analyses of these sounds showed them to be high-frequency clicks centred around 120-125kHz. Digital signal processing methods were used to automatically measure many features of the sounds and, in combination with multivariate statistical methods, to provide a quantitative analysis of the acoustic repertoire of Hector's dolphins. Almost all of the sounds made were narrowband, high-frequency clicks of comparatively simple structure. Hector's dolphins make very few audible sounds, the most common of which is made up of high-frequency clicks emitted at such high repetition rates that the repetition rate is audible as a tonal "cry" or "squeal". Multivariate analyses of the automatically measured data revealed different types of high-frequency clicks according to their frequency and timing characteristics. The sounds are described in detail, as are the techniques used to automate the digital measurement process. To gain an insight into the possible role of these signals as echolocation signals, the ambiguity functions of different types of click are presented. With any simple sonar signal, the structural demands of range measurement and resolution of target velocity are in conflict. These analyses show that Hector's dolphin sonar signals are poorly suited to determining target velocity, but are well suited to resolving target range. Information about target velocity is accessible to the dolphin only from the trend of range measurements during a sequence of clicks. To explore whether click types have different communicative meaning, I analysed whether certain click types were used disproportionately in different behavioural contexts. Hector's dolphin clicks do not appear to be used solely in sonar. Click types with complex spectra were used more often in large groups than in small ones, and double pulses (in the time domain) were used more often in "surface active" groups than in "long-diving" ones, suggesting they have some social significance. High repetition rate sounds ("cries") were much more commonly associated with aerial behaviours than with feeding, and appear to indicate excitement. I conclude that there is a general association between sounds and behaviour, and hypothesise that dolphins may have developed a communication system based on the ability to interpret each other's sonar echoes. Several workers have suggested that gillnet entanglement is essentially an acoustic problem, as the dolphin's sonar apparently fails to detect the nets. The Pegasus Bay/Canterbury Bight gillnet fishery was studied to investigate gillnet entanglement. Over the four years of the study, 230 Hector's dolphins were reported killed in gillnets. Most dolphins (89%) were caught within four miles of the shore, and over the summer months of November to February (91%). The acoustic aspects of this problem were explored in an analysis of proposals to reduce entanglement by modifying gillnets. I show that neither making the nets more reflective to dolphin sonar nor warning of their presence by attaching sound emitters has proved successful, and argue that they are unlikely to be successful because of logical and practical difficulties with the concepts. I conclude that the best management strategy for the reduction of gillnet entanglement is the closure of specific areas to gillnetting.