13

Male and female Nezera viridula communicate in their premating period by substrate-borne calling, courtship and rival songs. The dominant and basic frequency peak of all their songs lies between 96 and 150 Hz, higher harmonics do not exceed 500 Hz, Q₁₀dB ranges from 1.5 to 3.2 and intensity measured at the back of a singing animal lies between 10-3 and 10-4 m/s. Songs are characterized by two types of subunits: short FM pulses are exchanged with longer narrow band ones. Within the song, both FM and narrow band pulses appear either as a sequence of single units or are grouped into pulse trains which contain one or both types of them. Songs of the green stink bug are transmitted in different host plants with different efficiency. Plants like Cyperus alternifolius enable vibrational communication over distances of several meters. By transmission through the Cyperus stem the amplitude of narrow band pulses oscillates between 6 dB amplification and 5 do attenuation; their spectral and temporal structure is less altered. On the other hand, the short FM pulses are transmitted in Cyperus stem with less attenuation and lower level of amplitude oscillation but their temporal and spectral characteristics vary significantly at different points. Due to different transmission in plants, short FM pulses probably carry information about presence and location and the narrow band ones about the sex and species of calling or courting partner.

The development of Digital Signal Processing with low cost high-speed computer hardware and large hard disks has made the computer analysis of bioacoustical signals an everyday invaluable tool for ethological research. The latest versions of our real-time Digital Signal Processing Workstation (DSPWI meets most of the needs for sound recording and analysing. The system can deal with signals ranging from infrasounds to ultrasounds both in the lab and field. It is based on standard PC hardware and can be fully configured depending on the research needs. The software package includes analysis, recording and display tools, with real-time spectrogram and cepstrogram, spectral averaging, frequency tracking, event counting, scheduled recording, etc. Sound files and spectrograms can be saved to allow further processing with other standard software. A portable version can be easily moved across laboratories or used in field applications, for example those requiring real- time visualisation and recording of acoustic events. Depending on the installed sound acquisition devices, analogue I/O is possible in the audio frequency range and/or in the ultrasonic range up to 180kHz. Digital I/O is supported for direct transfer from DAT recorders to the PC. Real-time capabilities, typically available in much more expensive instruments are very useful in laboratory experiments to monitor the acoustic activities of the subjects (immediate correlation between observed behaviours and emitted/received signals) and to optimise the instrumental set-up (minimisation of noise, microphone placement).These facilities enable the immediate evaluation of experimental results instead of waiting for later analyses on the recordings, and they make it easier to analyse long periods. To date, the DSPW has been extensively used by many researchers to study sounds produced by fishes, birds, marine and terrestrial mammals and ultrasounds emitted by ants, moths and rodents. Examples will be shown on the poster and in the technical demonstration.

Between-year repeatabilities for several song characteristics were measured in a population of willow warblers. The results show that repertoire size has a significant repeatability (0.497), and that the song characteristic with the highest repeatability was song versatility (0.770). However, an analysis of the evolvability of these traits suggests that, given a minimum heritability, repertoire size would have higher residual and additive genetic variance than song versatility. Father-son regressions for all song characteristics were not significant due to the small sample size and the low heritability that is expected from the repeatability results.

We consider the peculiarities of echolocation discrimination of targets by dolphins in the presence of interfering reflection from steel cylinders, at angles less than the width of the transmission beam pattern. New data are presented that when the animal performs an echolocation task his spatial hearing is provided by the monaural mode of reception, which represents the best (practically ideal) noise-immunity from interfering refection, compared with the possibilities of binaural hearing. Using monaural reception, during performance of an echolocation task, the dolphin finds the optimum position of the head (transmitter and receiver) in relation to targets and interfering cylinders? providing the best condition of mutual suppression interfering echo (destructive interference). Modelling the destructive interference has shown that 1) the resultant energy of interfering reflection decreases four fold; 2) at the frequencies of maximum hearing sensitivity in dolphins there is a complete suppression component. The most probable mechanism of monaural perception of echo seems to be provided by passing through the lower jaw "acoustic window'' (Norris 1964, 1968, 1969, Brill 1988, 1991). In this case, the base of reception apparently becomes much less than the distance between the external hearing meatus, which essentially reduces the interaural difference of intensity and phase in the inner ear. It has- also been shown that dolphins use the slopes of the beam patterns in both radiation and reception to increase the difference between intensities of echo from the targets and interfering cylinders.

Song in birds is thought to function mainly as an intraspecific signal. But it can also be used in interspecific communication, among others in cases of interspecific territoriality. However, the function and information content of interspecific singing is not well studied. We recorded song responses of territorial black-capped warbler S. atricapilla males to playbacks of conspecific song and to the garden warbler's S. borin song. A preliminary analysis suggested that some blackcap warbler males, while responding to the garden warbler's song, can alter phrasing of their song to resemble the song of the congener. We present quantitative analyses of sonograms and discuss potential significance of such behaviour.

Coughing is one of the most frequent overt symptoms of many diseases affecting the airways and lungs of man and other animals. The registration of the cough sound as a diagnostic tool is still absent from current medical and veterinary practice in comparison with the registration of ECG, EEG etc. Registration of coughs from four different pigs in a metallic chamber was performed in order to analyse the acoustical signals. In order to build a system that is capable of distinguishing coughs from other sounds a large number of samples were collected. The best performance was obtained with a hybrid classifier that classifies coughs and metal clanging separately from the rest, giving better results compared to a PNN (Probabilistic Neural Network) and an MLP (Multilayer Perceptron). The hybrid classifier gave high discrimination performance (91.3 %) in the case of runts and noise and a performance of 94.8% for correct classification in the case of coughs. In the case of metal clanging, 82.6 % for correct classification was obtained. The intelligent system is proposed for real-time cough registration in a pighouse, as an early alarm for possible viral infection, or for studying the pig behaviour.

Red-legged Alectoris rufa and rock A. graeca partridges hybridise at a contact zone. Hybrids are viable and fertile. Rally calls (implicated in species isolation mechanisms) have been studied in males of the two species and in F1 hybrids. 3 temporal (call duration, silence duration and sound/silence ratio) and 3 frequency parameters (fundamental frequency, frequency of the maximum amplitude and frequency band containing 80% of energy) were used in a Principal Component Analysis. This indicated a greater variability in rock partridges than in red-legged partridges. Two frequency parameters (frequency of the maximum amplitude and frequency band containing 80% of energy) appear to be the factors differing most between the two species. In order to understand the decoding system of the two species, broadcast experiments were made on males. Each species was tested with natural conspecific and heterospecilic rally calls. The responses in terms of latency, duration and number of calls to conspecific and heterospecific signals of males of the two species did not differ significantly. Although signal coding exhibits several differences, the decoding system seems to be broader, allowing interspecific recognition.

Vocal development of nestlings of Canary lsland Chiffchaffs leads from short simple elements to complex calls that show a large number of frequency bands. Calls do not develop at a uniform rate but change their duration, frequency and shape more rapidly during the last 4-6 days before fledging than before. Duration increases during development. Until 8 days before fledging the duration lengthens from 4 ms to 22 ms. Three days before fledging calls last 22-27 ms, increasing to 133-160 ms at fledging day. The frequency used by nestlings is higher than that of adult singing males. Maximum frequency is around 8 kHz but minimum frequency increases from 5 days before to fledging. Constant maximum frequency and rising minimum frequency leads to a smaller frequency range. The structure of calls also changes during development. Starting with simple hooked notes with rapid frequency changes until 8 days before fledging, they lead to element types that are modulated in frequency during the last part of voice development until fledging. Rising minimum frequency will be discussed as a protection against Sparrowhawks Accipiter nisus, the only avian predator on the islands. The ability to locate high frequencies is decreases with rising frequency [thanks to Hessische Landestrraduiertenforderung and Deutscher Akademischer Austauschdienst for support].

Recent advances in computer technology have opened new opportunities for recording and monitoring of animal sounds. High quality PCI-bus soundcards with up to 8 channels, initially designed for music production, are now available at a reasonable price. Such soundcards installed in a Windows-PC with a fast hard drive are the basis for a versatile sound capturing system. This software is able to manage up to four soundcards installed in one PC. Up to 8 channels, each sampling at up to 48 kHz, can be monitored simultaneously. For low frequency or ultrasound applications not supported by soundcards, special data acquisition boards could be used. Each channel can be triggered separately by different trigger events. A trigger event is usually the exceeding of a certain energy threshold within a defined frequency band. This prevents triggering caused by undesired disturbance signals outside the frequency range of the animal vocalisations. Alternatively, trigger events could be keystrokes or joystick button hits. Once a trigger event has occurred, the assigned channel(s) will be written into numbered WAV-files onto the hard drive as long as the trigger event is active. In order to capture sounds preceding the trigger event, a pre-trigger duration can be specified. Additionally a hold-time is implemented in order to prevent cutting up of related sound elements. The adapted Avisoft-SASLab sound analysis software and the LOG-file output support quick navigation through the captured sound files.

Baleen whales live at extended timescales. Study of how these animals use their vocalisations for communication requires massive data sampling over long periods. The volume of data precludes traditional hands-on analysis techniques, at least during the first stages of data reduction. This paper describes a system for automating the sampling and analysis of baleen whale calls. Blue Balaenoptera musculus and fin B. physalus whale calls are very stereotypical. Blue whale 'A' and 'B' calls have fundamental frequencies of approximately 17 Hz, narrow bandwidth and well-defined harmonic structure, and typical duration of 15-25 sec. Fin whale 'pulses' have fundamental frequencies of approximately 17 Hz, but are broadband in nature and short (~ 1 sec) duration. The homogeneous call structure lends itself to automated detection. Stable acoustical differences in call structure lend themselves to automated species identification. We have benchmarked a series of bioacoustical call identification algorithms against a set of blue and fin whale calls while systematically manipulating the signal to noise ratio. The results demonstrated a typical tradeoff of speed versus accuracy. The best algorithm was inserted into the underwater sound recording system and its signal-detection theoretic performance was quantified. Results will be discussed with respect to technological and ecological aspects of baleen whale bioacoustics (Prolect CS-1082 of the Strategic Environmental Research and Development Program).