13

Preliminary results of a detailed acoustic analysis of vocalisations in a captive group of Varecia variegata variegata are discussed. Digital audio and video recording of naturally occurring vocalisations of this species were collected at the St. Louis Zoo. The sound recording equipment included a Sony TCD-D8 DAT recorder with a Sony ECM-MS907 microphone. Sound recordings were processed with Sonic Foundry SoundForge 4.5 and analysed with Voxys 3.1 for PC. Comparative analyses of affiliative vocal communication in Varecia variegata variegata revealed significant differences in frequency modulation between Mews and Whines. Statistical categorisation based on the first harmonic and the modulation of the fundamental and the first harmonic correctly assigned vocalisations to their predicted categories (100% accuracy). Univariate analysis underlined that differences between Mews and Whines (Gamba et al. 2000: Folia Primatologica: 71, 297) consist mainly of frequency modulation.. Investigating both modulated and stable parts of the signals, we show there is no difference in this group between spectral measures of the stable portion of Mews and Whines, but significant differences between the modulated parts. The fundamental frequency, first and second harmonics and most of the parameters related to frequency modulation were significant. Vocalisations of juveniles (3 and 11 months-old) differ significantly from adults' in the lndex of Harmonicity measured in the stable portion of the signals.

Animal studies indicate that many primate species recognise the individual characteristics of sounds of both adult and young. In human newborns, the successful individual perceptive attribution of the cries by the adult care-giver indicate that it may be considered an "acoustic signature'' (Green & Gustafson 1983: Developmental Psychobiology, 16, 485-493). In this study we attempted to find whether some sound parameters are better than others in individual discrimination. We recorded 64 cries induced by cynetic stimulation from 32 infants in their first and second day of life successively, and preliminarily analysed 11 parameters from 1628 wails. In phonated cries (i.e. wails with clear fundamental frequency) we found that 17 (53%) infants did not significantly change more than two parameters in the two successive recordings. A discriminant function analysis performed with SPSS software showed that 62% of the cries could be correctly discriminated on the base of sex (F5,759 = 7.91, P < 0.000). Twenty-two (69%) infants, subdivided in six smaller groups, could be individually discriminated. However, only 53% of the cries could be correctly assigned to an individual infant. Consequently, infants varied greatly in discriminability: from 97% of the cries to 0%. In individual discrimination, none of the sound parameters was privileged in entering the discriminant function. As regards individual discriminability, the number of cries from an infant was the only significant parameter for successful discrimination (F1,30 = 28.07, P < 0.000). Results are discussed within a comparative frame of hominid maternal care evolution and sound comparison with other primate species.

Agaltepec is a Mexican island, located in Catemaco Lake (Veracruz), where a wild group of mantled howler monkeys was introduced in 1988. At the moment more than 60 individuals live on the island allowing a good opportunity to undertake studies on this interesting colony. The first authors studied the vocal repertoire of the howler monkeys living in Agaltepec. Several groups of A. palliata mexicana were observed for a total of 220 hours and their naturally occurring vocalisations were digitally recorded on a Sony TCD-D7 DAT recorder with a Sony ECM-707 microphone. Sound recordings were processed and then analysed by Sonic Foundry SoundForge 4.5 (PC), Canary 1.4 (Mac) and Voxys 5 (PC). We proceeded to the qualitative and quantitative description of the vocalisations emitted by males by measuring a large set of acoustic parameters, including pulse number, pulse and inter-pulse duration and spectral properties of the emission at the beginning, in the middle and at the end of the signal. As in previous studies, we found several different vocalisations in the study group repertoire (Baldwin and Baldwin 1976: Folia Primatologica, 26, 81-108) but we focussed our study on the descriptive analysis of Roars, Woofs and lncipient Roars. This preliminary analysis of the Alouatta palliata mexicana vocalisations showed that the pulsed structure of the Woofs and of the lncipient Roars shows remarkable difference in pulse number (6.88 vs 16.50), pulse duration (0.6 s vs 0.24 s), interpulse (0.3 s 0.23 s) and total duration (1.22 s vs 3.02 s). Interesting differences in spectral features of the Roar inhaled and expired units are suggested but should be studied in detail in future analyses.

We describe the spectro-temporal features of begging calls uttered by Red-billed Chough Pyrrhocorax pyrrhocorax females in five localities (Scotland, Western Alps, Central Apennines, Sicily and Morocco), attempting to detect the causative factors responsible for the present pattern of geographic diversification. Females from different populations were significantly discriminated by their begging calls. Differences in the frequency of the first band of the call might be explained by body size. The largest-bodied populations uttered the calls with the lowest pitch, thus supporting the reverse correlation between body size and call frequency, already demonstrated in the trill call of the species. Comparison with a previous study carried out on the trill call suggests that the pattern of differentiation of the two calls does not always match: (a) most differences between populations were accounted by the duration in the begging call and by the frequencies in the trill call; 09 the frequency carrying most energy was affected by body size in the trill but not in the begging call; (c) in the begging call we found no relationship between acoustic similarity and geographic distance, as in the trill call. We suggest that the forces driving the evolution of the spectrotemporal features of the Red-billed Chough calls might differently act (or have acted) on the two call types.

We measured differences in behavioural responses of territorial Tawny Owls to four stimuli: (i) hootings of a neighbouring conspecific male, (ii) hootings of a stranger conspecific male, (iii) calls of a Long-eared Owl male and (iv) barking of a Roe Deer Capreolus capreolus (L.) male (as a control). In the breeding season, responses to the playbacks of Tawny Owls' (both stranger and neighbour) and Long-eared Owls' vocalisations did not differ from one another, but they were all different (i.e. stronger) in the responses to the control stimulus (iv). Outside the breeding season, responses to the hootings of Tawny Owls (b0th stranger and neighbour) differed from responses to the other two stimuli (with no differences within the pairs). This suggests that, while conspecifics (both neighbours and strangers) are 'enemies' of Tawny Owls throughout the year, Long-eared Owls belong to that category only in the breeding season, when they establish territories and become competitors. On the other hand, Long-eared Owls in our study area are extremely reluctant to respond even to playbacks of conspecifics. Both facts seem to support Murray's hypothesis about non-adaptive interspecific territorialism.

Some animals that use sound to communicate have evolved adaptations to counteract the masking effects of environmental background noise. In the short-term, one such adaptation is the Lombard effect, in which a signaller increases the amplitude of its vocalisations in response to an increase of the background noise amplitude. Such a capacity has been demonstrated for zebra finches (Cynx et al. 1998: Anim. Behav. 56, 107-113), a non-territorial songbird. Territorial songbirds, on the other hand, may benefit from singing as loud as possible to defend territories and attract females. Therefore, they may maximise the amplitude of their songs rather than regulating it according to the background noise level. I addressed this issue by experimentally manipulating the background noise and measuring the sound level of the vocalisations of a territorial songbird, the Common Nightingale Luscinia megarhynchos. This species is a good model to investigate the mechanisms of vocal amplitude regulation in detail, because males typically produce their territorial songs with specific amplitude differences between the diverse song sections (Hultsch, personal communication). Subjects (n = 4) increased the mean sound level of their songs in response to increased sound levels of white noise. Within songs, such regulation was found in both the loudest and the softest notes. One bird increased its call amplitude in response to increased levels of noise, showing that nightingales may also exhibit the Lombard effect in non-territorial signals. These findings show that nightingales do not maximise song amplitude but regulate vocal intensity dependent on the level of environmental noise. Possibly, such adjustment of vocal amplitude serves to maintain a specific signal-to-noise ratio that is favourable for signal production. Concurrently, amplitude regulation may ensure maintaining a given active space for communication. Thus, vocal amplitude in a territorial songbird is be a flexible trait, which is regulated according to ecological demands from signal transmission, as recently discussed by Brumm and Hultsch (2001, Anim. Behav., 61, 747-754).

When returning to their breeding colony from a foraging trip at sea, mothers of Antarctic fur seals Arctocephalus gazella have to find and identify their own young among hundreds of pups. Before meeting and suckling can occur, both mothers and pups call to each other at a distance and then, coming into contact, exchange nose-sniffs. To investigate the way in which pups recognise their mothers acoustically, we carried out playback experiments in the field with different kinds of modified maternal calls. Frequencies (pitch, relative levels of harmonics), amplitude and frequency modulations (AM and FM of calls were modified using synthesis techniques. According to our tests, it appears that pups pay attention to the FM shape and to the spectral profile (pitch + relative levels of harmonics, i.e. timbre) of the call to identify their mothers. It appears also that finding mothers is a two stage process for pups. At long range (more than 20m), a maternal call broadcast repeatedly often attracts several pups. Our propagation tests suggest that, at this distance, pups can only analyse the FM shape of the signal and this would explain the misidentifications observed. At a shorter distance (about 8m), only the right pup (exceptionally two pups) replies to the playback.  In this case, it seems that pups can analyse simultaneously a combination of two acoustic parameters of the maternal call: the FM shape and the spectral profile. This combination improves the accuracy of the identification process. The final verification by scent made by the mother appears as an additional means to secure fully the identification of the right pup and to prevent extra nourishment of non-offspring.

In colonial mammals, individual voice recognition between mother and offspring is a key factor for mother-pup reunions among numerous other individuals. The pup's ability to recognise its mother's voice is crucial for its survival: in most species, except phocids, females only feed their offspring and reject any attempt of milk stealing by strange pups. In the subantarctic fur seal Arctocephalus tropicalis, females have to leave their newborn a few days after birth to feed at sea. Pups recognise their mother by voice when she comes back from the sea. We investigate at what age a fur seal young is able to respond specifically to its mother's voice. Using playback experiments, we show that the subantarctic fur seal pup needs to be two to five days old in order to develop recognition of the mother's voice. Through investigating the relationship between the mother departure date and the duration of the young's learning process, we show that the mother delays her departure for her first feeding sea-trip until her pup responds specifically to her calls. In spite of the strong energetic constraints which may encourage her to go to sea as soon as possible after parturition, a mother leaves her offspring only when it is able to recognise her voice.

Social animals such as Steller sea lions Eumetopias jubatus often produce vocalisations that are very consistent and easy to discriminate, thus reducing potentially hazardous social errors in reunions between mothers and pups, between neighbouring territorial males, or other social relationships where individual recognition is beneficial. We have developed a system for accurately categorising calls to the appropriate individual, using parallel processing software algorithms (neural networks). An ability to quickly and easily identify individuals by voice will greatly aid researchers monitoring social behaviour on crowded rookeries where visual identification can often be difficult and uncertain.

CIBRA is participating in the design of an acoustic experiment to be carried out with an underwater platform deployed by INFN (lstituto Nazionale di Fisica Nucleare) on the sea floor 25 km off Catania (Sicily). The platform, placed at 2500m depth, is connected by a fibre optic cable to the laboratories in Catania. The platform will host an acoustic module (to be deployed in spring 2002) with 4 broadband phones and related electronics for transmitting acoustic data (96 kHz sampling, 24 bits, 4 channels) to the coastal lab in real-time. Dedicated software will provide continuous monitoring with either continuous or scheduled recording of acoustic data. Then, through the fast INFN network it will be possible to have high speed access to the system located in Catania to get sounds for post-processing. The special design of the instrumentation will allow high sensitivity, low noise and a great dynamic range to support a series of experiments to measure the ocean noise and biological sound sources in order to evaluate the feasibility of a further INFN project, named NEMO, concerned with the detection of acoustic pulses generated by high energy neutrinos passing through water. The experiment will allow the long term study of underwater noise and of biological sound sources, fin whales and deep divers like sperm whales in particular. CIBRA is participating in the design of the acoustic instrumentation and of data processing methodologies for the classification of biological sounds, identification of man-made sounds and long term monitoring of underwater noise. One of the goals is to map the seasonal variation of sounds emitted by fin whales and sperm whales and possibly to assess their seasonal movement patterns.