techniques

There is a growing concern in the literature about the effects of low frequency sounds (LFS) on marine mammals. A primary way to assess these effects on marine mammals involves the study of disturbance reactions. Detailed research of disturbance reactions of submerged marine mammals requires 3- dimensional localization and tracking of the animals. Animals such as sperm whales Physeter macrocephalus are localized passively with the use of travel time differences (TTD) of their vocalizations received by multiple hydrophones at known positions. Classically, straight-line paths of sound propagation between source and receiver are used to calculate source position. A more accurate calculation of source position involves naturally occurring non- constant sound speeds. This gives rise to arced paths of sound propagation between source and receiver. An algorithm is used to recursively pinpoint source position in a medium with a non-constant sound speed. 5 hydrophone array configurations are tested, each with 30 randomly generated source positions. Average errors of the 150 source position calculations (x, y, z) are (±1.58m, ±1.70m, ±10.44m) for the straight line, and ±0.76m, ±0.87m, ±1.10m) for the algorithm. On average, the algorithm improves the source depth calculation by an order of magnitude.

Determination of the delphinid sonar signal generation site has eluded cetologists for decades. Recent interest in developing a bionic sonar system has reinvigorated the effort to characterize the apparatus and its operation. We studied activities within the pharyngeal and nasal cavities of two bottlenose dolphins Tursiops truncatus during echolocation. A high-speed dual-camera video system provided synchronized windows for recording two concomitant events: movements visible through an endoscope and oscilloscope traces of acoustic pressure at a hydrophone placed near the animal's head. Dolphins have two tissue complexes (Cranford et al. 1996), one located on either side, and just above, the membranous nasal septum. They apparently generate acoustic pulses by pushing air across sets of internal flips.' The acoustic pulse occurs coincident with one oscillatory cycle of the lips. Changes in acoustic pulse repetition rate and the lip's vibration cycles are simultaneous, indicating that their rates and periods are synchronous. We did not find other structures in the airways vibrating synchronously with each acoustic pulse generation event. The palatopharyngeal muscle complex compresses air for the system. These observations settle a long-standing controversy over the site of biosonar signal generation in odontocetes and open a vista of potential avenues for future investigations. (Work supported by the Office of Naval Research)

This presentation provides an overview of instrumentation and software useful in animal bioacoustics studies. Researchers need to know what is important in the context of their own projects so they may choose sensors, recorders and processors. In addition to the microphones and hydrophones common to sound pressure sensing in air and water, particle motion sensors have their uses in directional sensing. Arrays of sensors are useful for source localization and signal enhancement. Audio cassette and digital audio tape (DAT) recorders are common, and computer memories (disks and flash memory cards) are being used increasingly. Analyzers have to be flexible to accommodate the computation of useful parameters to characterize animal sounds or sounds to which animals are exposed. There are general purpose signal analysers available, and there are systems of signal processing software routines like Signal and Canary. More researchers are using MATLAB to develop custom routines for themselves. Examples will be presented for time series and spectrum analyses of airgun pulses. Specific applications will be presented of moored buoy arrays and autonomous bottom recorders to acoustic studies of bowhead whales.

There are two methods of ultrasound acquisition and analysis with PC's: The first and most straightforward solution is a special data acquisition board with a sampling frequency high enough to capture the entire bandwidth of the signals to be analysed. This requires taking the computer into the field if you have to do field research. The best choice would be a PCMCIA data acquisition card for portable notebook PC's. Unfortunately the devices available on the market still have limited sampling frequencies of approximately 100 kHz. This allows a usable bandwidth of less than 50kHz, which is not sufficient for all kinds of ultrasound. The data streams generated by these data acquisition boards can be stored on hard-disk and can then be read by the Avisoft-SONAGRAPH software using one of its user-defined import formats. The alternative method is the usage of a digital time-expansion bat- detector to transform the ultrasound into low frequency sound which can be processed by conventional audio equipment. These transformed sounds can be stored on standard tape recorders and can be transferred into the computer using a common sound-card. In this case the transformed ultrasound is treated like any other audio signal. In order to get the original scaling of waveforms, spectra and sonagram displays, the time-expansion factor used on the bat-detector can be specified in the Avisoft -SONAGRAPH software. Conventional bat-detectors using heterodyne or frequency division technology are not suited for spectral analysis on a computer. These devices could only be used to get an idea of the time structure of the signals.

In June 1995, a 12 days research cruise was organized in the Ligurian and North Tyrrhenian Sea to record cetacean sounds with the towed array of the University of Pavia. The cruise was supported by the Italian Navy within the ENCY 95 (European Nature Conservation Year) activities. The hydrophone was towed for 111 hours (out of 12 cruising days) at speeds up to 14 km/h; listening stations were held on a 24h schedule for at least 10 min every half an hour. One sperm whale was detected and located. It was heard at night and acoustically tracked for the following 8 hours. Within this period the whale was sighted at the surface 5 times, while 8 complete dives were continuously recorded on DAT tapes (about 6 hours of recording). The recordings are now archived at the Cetacean Sound Library held at the Centro. New methods of sound analysis were developed to make the analysis of such long recordings easier and to give compact pictures of whole dives. Our real-time analysis software was modified with new procedures able to 1) automatically detect and count clicks, 2) measure and save inter-click intervals, 3) save packed representations of the click sequences and display autocorrelograms to show the evolution of inter-click intervals over long periods of time. The analysis of the recordings shows that all the recorded dives were characterized by a typical and constant clicking pattern at their beginning. The duration of the acoustical emission, measured from the first click to the last click of each dive, was on average 27 minutes 30 seconds, while the silence related to the surfacing was on average 13 minutes and 11 seconds.

For one year, starting on September 5th 1994, the behaviour of a captive new-born bottlenose dolphin Tursiops truncatus and its mother was monitored through both video and acoustic recordings. The main objective of such research was to study the behavioural evolution of the two during the first year of the calf's life. Attention was focused on acoustic behaviour and its use in relation to the contexts. In order to be able to discover any possible correlation, the acoustic signals were analysed with a Pc-based DSP Workstation developed at the University of Pavia, and the resulting real-time spectrographic analyses were superimposed on the live video recordings.

The problems of reducing small cetacean bycatch in fishing nets are many and complex, and acoustic solutions need to be tailored to suit individual species and net types. Active deterrent devices or alarms have recently been shown to be beneficial in reducing the bycatch of harbour porpoises Phocoena phocoena in some carefully monitored trials in North American sink-gillnet fisheries (Kraus et al. 1995). The methodology employed to date is still at an early stage of development; typical devices generate simple 10 kHz tonal pulses from small battery powered packages distributed at intervals along the fishing net. The long term effectiveness of this approach has already been questioned, as small cetaceans are known to habituate quite quickly to novel stimuli. A more sophisticated approach, now at the trials stage, uses higher frequency, wideband signals which have been determined to be more aversive to this species (Goodson et a1. above). To minimise habituation effects over time, such 'beacon alarm' signals need only be transmitted occasionally if the acoustic activity of an approaching echolocating animal can also be detected and used to trigger the device into a transponder type of operation. Such a device has now been designed at Loughborough University and implemented using digital micro-controller technology. As this circuitry is programmable almost any complex waveform can be generated to suit specific applications. The use of a digital micro-controller permits a number of additional features to be implemented in software without a significant increase in the overall component count and, despite the sophistication, the cost per device remains acceptably low. It may be argued that a silent animal could remain at risk. However, in the context of bottom set nets (sink-gillnets) this should not apply, as the harbour porpoise swimming close to the bottom in deep water is there to forage for prey and hence actively employing its echolocation sense. This interactive approach reduces spurious acoustic emissions which waste battery energy and, as most responses are triggered by the approach of an animal at risk of colliding with the net, the deterrent effect is expected to be longer lasting. The alarm response ceases quickly once the animal turns away. Provided that these devices are spaced apart along the net within detection range of each other, additional benefit is obtained if they also respond to a neighbour's alarm signal as this will result in a 'ripple-fire' of activity along the net. This linking of emissions provides better orientation information than can be given by simple randomly timed pingers. lnteractive devices are also intended for application to large pelagic trawls, where the delineation of the net's boundary by such a ripple-fire transmission is expected to give a clearer indication of the extent of the hazard to an animal which may have followed fish into the net. Since triggering can also be stimulated by a ship's echosounder, these devices should make the location and recovery of lost nets a relatively simple exercise.

Instrumentation recorders designed to record ultrasounds are very expensive and not well suited for field use; thus, cheaper devices to detect and possibly record ultrasound were developed for field research, mainly to study echolocation in bats. These are called bat detectors. Basically they employ three methods for making ultrasounds more manageable: heterodyne conversion, frequency count-down, and time-expansion. Only the last method enables the complete recording of all the features of the ultrasonic signals. It is based on digital recording with a high enough sampling rate (typically ranging between 300 and 400 ksamples/sec) and on the subsequent playback at a reduced sample rate to lower ultrasonic frequencies to within the range of conventional audio recorders. This procedure does not allow long recordings. Better suited to long recordings is digital acquisition on a PC, and several acquisition boards on the market allow for this. We developed a Pc-based Digital Signal Processing Workstation (DSPW), based on a Microstar DAP 2400E/6 acquisition board, to allow recording and playback of acoustic signals up to 150 kHz with a resolution of 12 bits (72 (IB dynamic range). The actually available Pentium CPUS allow, together with highly optimized custom-made software, analysis and display of spectrograms in real-time up to 150 kHz while performing hard-disk recording and other data analysis and logging tasks. A sharp anti-aliasing filter is required to prevent aliasing. When sampling at 312.5 ks/sec a 2 GB hard disk allows the continuous recording of up to 56 minutes. This instrumentation is very useful in laboratory experiments to monitor the ultrasonic activities of the research subjects and to optimize the instrumental setup (minimization of noise sources, microphone placement). Also it allows immediate evaluation of the results of an experiment instead of waiting for later analyses on the recordings. Special analysis procedures enable data to be logged in real-time (onset, offset, freq. tracking) and reports about the monitored signals to be produced.

Spectrographic display of animal sounds has been widely used since the first analogue analysis instruments were developed for military acoustic research. However, until recently, the high cost of the specialized hardware necessary for this type of analysis made it inaccessible to most researchers. The recent development of digital signal processing techniques and high-speed hardware at relatively low-cost has made the real-time visualization of acoustic signals an every-day invaluable tool for bioacoustic research and education. A new version of the DSPW software already described in previous reports has now been developed to use Creative^TM Sound Blaster and compatible boards. The software strictly requires 16-bit sound boards (Sound Blaster 16, Sound Blaster 32 and true compatibles, including those based on the Vibra 16 chipset); it is DOS based and uses a high DMA channel (4-7) to perform continuous gap-free transfer of samples Nom the board to the computer memory. Since several notebooks now incorporate Sound Blaster compatible sound devices, this software opens up new perspectives in field applications. The software is a powerful display and analysis tool. Depending on the CPU speed (a Pentium is recommended), overlap and zero-padding can be performed in real-time to get sm00th time-frequency plots up to 44.1 ks/sec. Both spectrograms and cepstrograms can be computed in real-time. Three basic real-time display modes are available: horizontal display with envelope (wrap around or scrolling display), four strips scrolling display and vertical scrolling display. Even if cheap sound boards seem to be adequate for music and games, some problems are still to be solved when analyzing sounds: signal to noise ratio, bit resolution, frequency response, anti-aliasing filters, available sampling frequencies (most boards allow sampling rates ranging from 5 ks/s to 44.1 ks/s) and sampling frequency accuracy change depending on the sound board model. Thus, great care must be taken when using digitizing devices which are not designed for great accuracy. Before starting to use a particular sound device, a series of tests must be made. For this purpose, a collection of freeware utilities have been developed by Philip van Baren to check the real performances of Sound Blaster boards. A freeware version of the real-time software will be soon available on the net.