9

A study on time and frequency analysis of a double-click by the dolphin Tursiops truncatus is reported. By simple modification of mirror image double-clicks, which were first used in experiments with human subjects by Ronken (1970), I designed mirror-image double-clicks which differed in interclick intervals, but had the same long term amplitude spectrum and practically the same phase spectrum within dolphin's hearing range. The bottlenose dolphin was found to distinguish such stimuli at approximately a 10% difference in the interclick intervals. The shortest interclick interval of about 25 µms at which the dolphin could discriminate between the double- clicks can be considered as the estimate of the actual time resolution of its hearing. Frequency domain representation of a double-click is supposed to be based on some sort of interaction between the first and the second click, similar to an interference between the clicks in frequency filters. In order to measure the recovery time for dolphin hearing filters I used double-clicks which differed in amplitude spectrum but had the same interclick intervals. At equal interpulse intervals, the amplitude spectra of the pairs are rippled with the same period, but the maxima of one spectrum corresponds to the minima of the other. The largest interclick interval at which two dolphins were able to discriminate the pairs was found to be 100-110 µs. Thus, the representation of a double-click in the dolphin's auditory system can be described by its extremely high time resolution of less than 25-30 µs, and frequency filter's recovery time of 100-110 µs. The real time resolution of the dolphin hearing proved to be practically equal to the theoretical time resolution of sonar's clicks. The 'recovery time seems to represent the dolphin's limit for frequency analysis of double-clicks at both audio and echolocation frequencies. The sonar's high time resolution suggests that target discrimination cues are available for the dolphin in the time domain. Still, at least in experimental conditions it is possible to force the dolphin to process double-clicks in the frequency domain, even the ones which it normally processes in time domain, provided the conditions for frequency analysis are fulfilled. Experimental data were collected at the Karadag Department of Institute of Biology of Southern Seas, Crimea, USSR in 1978-1982.

Echolocation signals of dolphins (clicks) consist of short duration and high frequency ultrasonic pulses. Click intervals of dolphins in captivity are less than two-way transiting time of a target range. Click intervals of free-ranging dolphins are thought to provide underwater sensory range of echolocation. A/D converter driven by a data acquisition program on Windows 95 © with a small signal processing circuit (Clicker 45) enabled real-time data acquisition of click intervals. Clicker 45 converted clicks to rectangular signals which duration was 0.5 ms. The amplitude of the rectangular signals were in proportion to sound pressure levels of clicks. When the amplitude of the rectangular signals were more than the threshold level (127 dB re 1µPa), the serial time and the amplitude were measured by the data acquisition system whose accuracy were 5 mV and 100 ms, respectively. Consequently, the click intervals and sound pressure level in click trains could be retrieved automatically. The rectangular signals from Clicker 45 could be recorded by an ordinary band-limited data recorder, so that the present system could be applied not only for a laboratory works but also for an open sea observation of dolphins' sonar.

The time resolution constant for Tursiops truncatus clicks is between 12 and 15 µs (Au 1993). However this extremely high theoretical value has never been considered as the actual time resolution of the dolphin auditory system. The "critical interval'' of 260 + 25 µs as a measure of the time resolution was derived from the series of echolocation and hearing experiments on dolphin discrimination between correlated stimuli (Dubrovskiy and Velmin, 1975). Having learned by experience how wrong one could be in construing results of dolphin's discrimination of correlated stimuli, we introduced uncorrelated noise stimuli (Zaslavskiy, Ryabov and Titov 1979, Zaslavskiy and Ryabov, 1991). The temporal masking, time intervals and pulse envelopes discrimination for the noise stimuli were studied. The actual time resolution of dolphin's auditory system of about 20-30 µs consistently manifested itself from these experiments. The dolphins have shown remarkable capability in analysing different characteristics of very short pulses in the time domain. The time resolution of the dolphin auditory system proved to be practically equal to the theoretical time resolution of echolocation clicks and at least ten times shorter than auditory integration time (Au 1993). Results gave no indication of the "critical interval". Experimental data were collected at the Karadag Department of Institute of Biology of Southern Seas, Crimea, USSR in 1977- 1989.

The "Centro Interdisciplinare di Bioacustica e Ricerche Ambientali'' was founded in 1988 by the University of Pavia. Since 1989 it is endowed with a Laboratory of Marine Bioacoustics granted by the "lspettorato Centrale per la Difesa del Mare'' of the Italian Ministry of the Environment. The Cetacean Sound Library created at the Centro holds recordings made in research cruises organised to study the acoustic behaviour, distribution and biology of cetaceans in the Mediterranean Sea. More than 130 hours of recordings belong to Sperm whales (which till now has been the target species), Striped dolphins, Risso's dolphins, Bottlenose dolphins, Common dolphins, Pilot whales, and other sound sources including man-made noises. A catalogue based on a widely used database format was created in order to allow an easy and reliable access to the collected recordings and to related data such as digitised photographs of individual Sperm whales, video clips, behavioural observations, cruise tracks, bathymetric data, and others. In the last two years a useful partnership with the Italian Navy has been arranged, and the Centro is asked to evaluate biological sounds recorded in ASW (Anti Submarine Warfare) operations and surveys.

Studying the vocalisations of dolphins provides an insight into their underwater behaviour whereas visual methods are normally restricted to surfacing behaviour. Apart from whistles, which seem to be produced in a social context, the sonar emissions of dolphins consist of pulsed signals used for different echolocation purposes. By analysing the pulse periodicity, frequency and spectral components of 'clicks' and relating these to video recorded surfacing behaviour it should be possible to extract characteristic patterns which indicate different kinds of dolphin behaviour. Almost all the pulsed emissions recorded during a series of 24 hour long intensive studies suggest behaviour related to foraging. These recordings provide unambiguous conditions where the environmental constraints can be observed and understood. Goodson and Datta (1992) found that recognisable patterns occurred in the repetition rate of the echolocation signals during these sequences. They partitioned these pulse sequences into 4 distinct phases. These sub-classifications have been further examined using an extended data set from the same source. Graphical and statistical analysis of six complete sequences, recorded immediately prior to a visually observed fish capture, allows a better definition of the presence of these foraging partitions. An additional fifth foraging phase was identified. The sequences could be partitioned into separate identifiable segments and characterised as foraging phases. Possibilities exist for automating this analysis for on-line use in the field as well as for the application of this analysis approach to behaviour types other than foraging.

Ferry companies are increasingly utilising high-speed wave-piercing catamarans to provide fast alternatives to conventional services. The number of such ferries operating in the UK has doubled in the last 5 years, but the environmental impacts of this trend, including possible cetacean disturbance arising from noise pollution, have received little attention. In March 1997 a new high-speed ferry service began operating from Poole in Dorset, England, passing through the Durlston Marine Research Area, the site of a long-term bottlenose dolphin monitoring project. Recordings of the ferry were obtained, from portable and seabed mounted fixed hydrophones, in order to assess the potential for disturbance of the study animals. The most significant sound outputs are two sharp peaks around 500Hz. Apart from these, machinery noise also produces a continuous spectrum across the range 100Hz to above 5kHz. The other major noise source is from displaced water, contributing to noise levels in the higher part of the spectrum, particularly above 10kHz. For bottlenose dolphins, the ferry would appear unlikely to cause disturbance on acoustic grounds. In keeping with this, comparison of bottlenose dolphin sightings data before and since the commencement of the ferry service found no discernible change in the timing or frequency of dolphin activity in the study area. However, this was very much a preliminary short-term study and further data are required before firm conclusions can be made.

Most species of marine mammals seem highly reliant on and sensitive to underwater sounds. Sounds important to marine mammals may include calls from conspecifics, odontocete echolocation sounds, predator and prey sounds, and environmental sounds (e.g. surf or ice noise). Some man-made noises are known or suspected to have negative effects on marine mammals, including noise-induced masking, disturbance, hearing impairment, and possibly stress. However, marine mammals are adapted to a variable and often naturally noisy environment. Also, even when levels of man-made noise are well above natural ambient levels, negative effects on marine mammals are not always obvious. Data available up to early 1995 were summarised in the book "Marine Mammals and Noise'' (Richardson et al. 1995, Academic Press). Since then, advances have occurred in some but not all areas of particular concern:

(1) When can marine mammals hear man-made noise? Additional data are becoming available for some small- and moderate-sized odontocetes, pinnipeds, and manatees. There is still an urgent need for direct audiometric data from baleen and sperm whales.

(2) Does man-made noise mask important natural sounds? Data are available on masking in a few species of captive odontocetes and pinnipeds. However, we need data on masking processes and significance when free-ranging marine mammals are exposed to typical man-made sounds, including variable, non-tonal, and directional sounds.

(3) When does man-made noise disturb mammals, and when is disturbance strong enough to constitute harassment? Disturbance effects are graduated, not ''all or none''. Sometimes no disturbance is apparent even at short ranges with high received levels (RQ. At other times there is strong disturbance even at long ranges with low RLs. Strong and/or prolonged disturbance may have negative biological elects even if there is no physical damage. However, infrequent brief disturbances may have no biological significance, and if so should not be considered "harassment''. Additional controlled studies, both field and captive, are needed.

(4) What are the thresholds for noise-induced auditory impairment and non- auditory effects, and what types of man-made sounds could elicit them under field conditions? The first data on Temporary Threshold Shift (TTS) in marine mammals have been released. recently. TTS work with additional species and exposure conditions is needed. However, TTS results have limitations in establishing damage risk criteria (DRC), and relationships between TTS and harassment are uncertain.

(5) Noise-induced stress in marine mammals is almost entirely unstudied.

Mitigation measures sometimes used to reduce noise effects include seasonal and geographic restrictions, ramping up, and real-time monitoring plus localised mitigation. We need more data on the effectiveness of ramping up, visual and/or acoustic monitoring, and localised measures such as minimum approach distances, minimum altitudes, and shutdown radii. Progress is being made toward understanding noise effects on marine mammals, in focusing on the most serious issues, and in devising mitigation approaches. However, the issues are complex and the needed studies are often difficult. Some major emitters of underwater sound remain reluctant . to become involved in the process. It will take time, money and cooperation to conduct the needed studies, to determine which situations need mitigation, and to devise, test and implement effective yet practical mitigation measure.

This paper describes the first use of the Transmission Line Modelling (TLM) method for the time-domain numerical modelling of sound propagation in odontocete acoustic tissue. The validity of the technique is assessed by performing simulations on the highly specialised lipid materials distributed within the melon of the species Phocoena phocoena. The geometrical data for the simulations was obtained from Computer X-ray Tomography (CT) scans published by T W Cranford. A time-discrete waveform based on the output from the animal was synthesised for injection into the TLM simulation. The software described accepts a 24-bit bitmap for the geometrical data, and the injected samples in a spreadsheet file. Output is available as pressures in another spreadsheet and a series of bitmaps showing the propagating waves at different points in time. A number of bitmaps may be generated from the program and combined using a readily available commercial program into moving wave visualisations. The model clearly shows the acoustic energy contained by the waveguide effect of the graded-index composition of the melon. The investigation supports the assertion that the melon is a wave guiding structure rather than a conventional lens. The tissue structures prior to the melon therefore appear responsible for generating the oscillatory waveform.

The harbour porpoise, one of the smallest marine cetaceans of the Northern hemisphere, is a frequent victim of the continental shelf bottom-set gill-net fisheries with many thousands of animals being killed annually. The reasons for this high mortality are not clear as these animals possess a sophisticated active sonar system with which they detect, track and intercept small fish targets. The performance and limitations of this animal's sonar are therefore being studied in order to develop techniques that will minimise cetacean/ gillnet interactions. Sonar source levels and spectra were carefully determined in an enclosed environment from signals transmitted by two juvenile animals. This paper examines these signals and, in the context of the detailed structure of the vestibular air sacs, presents a new hypothesis suggesting that a passive biological mechanism exists in these animals which can explain the formation of their unusually narrow band echolocation signals. The knowledge gained to date in this and associated studies suggests that the harbour porpoise operates a relatively low power, short range sonar capable of detecting (ingestible size) individual fish out to a maximum range of about 30 m. The strongest acoustic emissions of this animal are expected to be around 120-140 kHz with Source Levels, in open water conditions, around 160-170 do re 1µpa at 1m. These sonar signals will be confined by the projecting aperture to a very directional beam pattern (approaching 8° in azimuth and 14° in elevation) projected directly ahead of the animal. The spectrum of the harbour porpoise's sonar pulse exhibits a single power peak and the -3 dB bandwidth is typically around 13 kHz. The harbour porpoise possesses a much shorter range, lower power, sonar than its larger delphinid cousins. In addition its more limited bandwidth may be expected to generate target echoes with less spectral coloration which in turn may impair its ability to correctly classify unfamiliar hazardous targets such as fishing nets from a safe distance.

In anti-submarine warfare (ASW), sound is the foremost mechanism for the detection of potential threats. To detect a potential enemy underwater the sonar operator must be familiar not only with the acoustic signatures of the threats, but also with the many and varied noises from non-threat sources in the ocean. Some of the most prolific of these sources are biological. SANDS is a research tool aimed primarily at providing a reference set of biological noises for training purposes. As well as examples of the sounds, it contains stills and movies for species identification purposes. It also includes diagrams and textual information for each species on particular subjects at different levels of detail. Facilities are provided to search for specific species and examples on various criteria. Over 6,500 digitised sound examples, covering 34 species, have been collated through research organisations in America and Europe. These are stored as digitised time-series files, along with associated textual information. While the system has been designed primarily to support aural training, it has application to many other areas of research. It is probably the most comprehensive database of its kind in the United Kingdom, possibly in Europe.