Elodie F. Briefer
Institute of Agricultural Sciences, ETH Zürich, Universitätstrasse 2, 8092 Zürich, Switzerland
Acoustic communication is used by Arthropodes and Vertebrates, and particularly in species that move in the three-dimensional space (e.g. underwater or in forests), in order to communicate at both short and long distances. This mode of communication is highly developed in social species, and plays a crucial role in reproduction, parent-offspring communication, predator avoidance, territoriality, foraging and group communication. The modes of production of sounds and the structures used in sound production are very diverse and range from vibration of the wings in fruit flies (Drosophila) to stridulation in crickets, snapping of the swimbladder in fish, tongue clicks in some bats and vibration of the tympaniform membranes in birds or of the vocal cords in mammals. Several seminal papers on the modes of acoustic production and its link with the acoustic structure of vocalisations, as well as on the information content of vocalisations, have been published in Journal of Zoology. This free Virtual Issue gathers a selection of reviews and research papers, from various years, on the topic of sound production mechanisms in animals.
In the first selected paper, Parmentier et al. (2008) compared the structures involved in acoustic communication in three species of pearlfish (Carapidae; Carapus boraborensis, Encheliophis gracilis and Carapus homei). These small eel-like fish produce species-specific sounds that differ in temporal parameters, spectral frequencies, and sound intensity. In these species, sounds are produced through the swimbladder. The contraction of primary sonic muscles pulls the anterior bladder. When releasing the tension, the swimbladder snaps back to its resting position, producing sound. The authors analysed the sound production system as well as the acoustic features of the sounds produced. They were able to relate each part of the sounds to the action of swimbladder muscles. They also showed anatomical as well as acoustic differences between the three studied species.
With the second and third paper selected for this Virtual Issue, we move on to bird sound production. These two papers, by Thorpe (1958) and Warner (1972), are among the earliest detailed papers on the topic of sound production mechanism in birds. Thorpe (1958) is a review paper describing the vocal apparatus of birds and comparing it with human vocal apparatus. Both bird vocalisation and human voice are produced by the vibration of membranes during the exhalant phase of respiration (tympaniform membranes and vocal cords, respectively). This sound is then shaped in the resonance cavities (vocal tract above the syrinx and larynx, respectively). Warner (1972), after a historical review of previous work conducted on sound production mechanism in birds, compared the anatomy of the syrinx in several passerines birds (songbirds). His main finding is that the only demonstrable vibratile areas, hence the only sound sources, in the syrinx are the internal tympaniform membranes. These membranes, located one in each bronchus, can vibrate independently of each other, thus producing two harmonically unrelated tones at the same time (biphonation). These findings would be confirmed later on by many other studies.
Most bats produce echolocation signals, ranging from 20 to 200 kilohertz in frequency, using their larynx. However, some variation exists. For instance, a few species click their tongue, whereas horseshoe and leaf-nosed bats emit their echolocation calls through their nostrils, which are surrounded by a fleshy, horseshoe/leaf-like structure. In the fourth paper of this Virtual Issue, Robinson (1996) investigated the function of the noseleaf of horseshoe and leaf-nosed bats in echolocation. The author measured the echolocation frequency, noseleaf width, and forearm length of 14 species from the genera Rhinolophus and Hipposideros. This study revealed that noseleaf width is related to the wavelength of the echolocation signal, but not to forearm length. This suggests that noseleaf width is determined by wavelength rather than body size, thus highlighting the function of noseleaf in the production of echolocation signals.
The next paper selected for this Virtual Issue, by Sissom et al. (1991), investigates purring in cats. The mechanisms of cat purring have turned out to be challenging to understand. Indeed, the very low fundamental frequency of purring, notably in domestic cats (around 25 Hertz), suggests alternative mechanisms of sound production than flow-induced vocal cord vibration, as only very long cords could produce such low frequency. Proposed mechanisms include aerodynamic and hemodynamic vibration of the true and false vocal chords, the soft palate, and the arterial system, or muscular vibrations of the diaphragm and a repetitive closing of the glottis. The authors recorded domestic cats in a shelter and found that purring occurs during the entire respiratory cycle, with a fundamental frequency ranging between 23 and 31 Hertz. This frequency is higher during expiration than inspiration, but is quite stable throughout the life of an individual, and is not related to its weight or sex. Their results support the laryngeal mechanism but argue against the mechanical involvement of the diaphragm and intercostal muscles. Thus, purring could arise from the gating of respiratory flow by the larynx. This had been suggested also by Remmers and Gautier (1972), who showed that purring is in fact produced by active contractions of laryngeal muscles modulating the respiratory air flow passing through the vocal cords, as opposed to flow-induced self-sustaining oscillations found, for example, in humans.
According to the source-filter theory of voice production (Fant, 1960; Titze, 1994), the air flow coming from the lungs induces the oscillation of the vocal cords, thus producing the “source” sound. This sound is then filtered in the vocal tract (“filter”). Some frequencies, which correspond to the resonances of the vocal tract, will be amplified and other frequencies will be dampened. The source determines the lowest frequency of the voice (fundamental frequency) and its harmonics, while the filter determines the spectral peaks, called “formants”. In human voice, the pattern of the first two formants allows to distinguish between different vowels and is thus the basis of human speech. The source-filter theory framework has been recently adapted to other animals. The sixth paper of this Virtual Issue, by Taylor and Reby (2010), describes how mammal communication research benefited from this framework. By linking the structure of vocalizations to their mode of production, researchers have been able to highlight information, in vocalisations, about the sender’s body size, age, sex, hormonal levels, dominance status and even motivational or emotional state. This review paper describes the source-filter theory in details and the indices that are generated at the source and at the filter. The seventh paper of the Virtual Issue, by Briefer (2012), describes in more details how motivational or emotional states can affect vocalisations in humans and other mammals. In this paper, I reviewed the existing literature on vocal correlates of emotions in several mammals, in order to highlight common patterns of changes in vocal parameters and find the best source- and filter-related parameters that can reliably indicate the two main dimensions of emotions (arousal and valence).
The eighth paper selected for this Virtual Issue, by Fitch (1999), shows how the source-filter theory framework can be applied to birds, in order to explain the evolution of trachea elongation. More than 60 bird species possess an elongated trachea. Formant frequencies depend on the vocal tract length and shape, with lower frequencies indicating longer vocal tracts. Normally, the vocal tract is constrained by surrounding bones, so that its length strongly depends on body size. Formants are thus good indicators of body size in many species. However, some species possess either a mobile larynx that they can retract to lengthen the vocal tract during vocalisations (e.g. red deer and fallow deer), or possess an elongated vocal tract. Fitch suggests that these characteristics evolved through sexual selection to exaggerate perceived body size, as animals can produce vocalisations with lower formants than expected from their body size.
The ninth paper selected for this Virtual Issue investigated the source of vocal production in muskoxen (Ovibos moschatus), a large ungulate of the family Bovidae and subfamily Caprinae. Although highly sexually dimorphic (males are 1/3 heavier than females), both sexes of this species produce very low roars. In this paper, Frey et al. (2005) investigated the laryngeal anatomy and the roaring vocalizations of the muskox. Roars in both sexes are characterised by a pulsed structure, with a pulse rate of 20 Hz on average. The larynx size of adult male and female are remarkably similar (i.e. almost identical larynx size and vocal cord length). Both sexes possess a potentially inflatable ventrorostral laryngeal ventricle, which could serve to increase the amplitude of roaring or to act as an additional resonance space. The only difference between the sexes is a voluminous fat pad in the medial portions of the vocal cords of adult males, but not of females. However, this pad does not seem to induce important differences between the pitch of males and females. Muskoxen thus differ from other species with similar mating systems (e.g. fallow deer, red deer, elephant seals), in which strong sexual dimorphism is accompanied by distinct acoustic differences.
The two last papers selected for this Virtual Issue on sound production mechanism investigated the link between vocal tract length and formant frequencies in canids (Plotsky et al. 2013) and fallow deer (McElligott et al. 2006), respectively. Plotsky et al. (2013) present one of the first clear evidence that formants provide good, honest vocal cues of signaller size. The authors tested the link between vocal tract length, measured from X-ray images, and measures of body size in Portuguese water dogs (Canis lupus familiaris) and Russian silver foxes (Vulpes vulpes). They show that the oral component of the vocal tract, which determines formant frequencies, is strongly related to body size. McElligott et al. (2006) investigated the link between vocal tract elongation and formant frequencies in fallow deer (Dama dama). This species possesses a mobile larynx that can be retracted during vocalizations. The authors show, using audio and video recordings of mature, groaning fallow bucks, that individuals can increase their vocal tract length on average by 52% during vocalization. The highest formants (3-6), which strongly depend on the vocal tract length, are lowered, while the lowest formants (1-2), which depend more on the shape of the vocal tract, show minimal change during laryngeal retraction. This phenomenon could be used by males to exaggerate their perceived body size.
We hope that you enjoy reading this free collection of papers on various modes of sound production and adaptions published in Journal of Zoology.
Fant, G. (1960). Acoustic theory of speech production. The Hague: Mouton.
Remmers, J.E. and H. Gautier, Neural and Mechanical Mechanisms of Feline Purring. Respiration Physiology, 1972. 16: p. 351-361.
Titze, I.R. (1994). Principles of vocal production. Englewood Cliffs: Prentice-Hall.