Alarm calling is known to be the most typical acoustic behaviour in the Marmot. It is evoked by the aerial passage of a potential predator (single whistle, SW,4) or when terrestrial animals or men approach (multiple whistle, MW,4). The present research has analysed the different answer degrees to aerial stimuli and temporal and spectral features of single whistle in order to define if there is a correlation between signals and greater or minor risk gravity. To investigate situational variations, Marmots alarm calling was collected using a Marantz CP 130 analogical tape recorder provided with Sennheiser microphone; The analysis was performed by means of IBM compatible PC supplied with programs designed to obtain timewave, spectrum and graph resulting from the cepstrum analysis. The study area was sited in the Parco Naturale Orsiera Rocciavr (Western Alps) at about 2000 m a.s.l.
Natural interactions between Marmots and their predators are very uncommon to observe, therefore wooden and pasteboardmade models were built and used as stimulus for the rodents. The models, representing silhouettes of predator birds (Golden Eagle, Aquila chrysaetos; Ghoshawk, Accipiter gentilis and Buzzard, Buteo buteo) and non predator (Kestrel, Falco tynnunculus and Chough, Pyrrocorax graculus) of natural dimensions, ran over a 200 m long nylon line strained from the top of a natural rocky wall over the Marmots burrows. For every models descent, more than one single whistle was uttered by different Marmots living about 400 m all around the line.
From the spectral point of view Marmots whistles are complex harmonic sounds of two or three components, the fundamental, the second and rarely the third harmonic. The whistle is produced without glottic periodic opening and closing, however the keeping glottis more or less open during the emission, changes the vocal tract shape and lenght, and influences signal harmonic structure. The amplitude of the second harmonic frequency is different from one whistle to another and changes from 0 to about that of the first. In order to analyse the Marmots whistle it is necessary to choose some phonetical parameters describing sound features; in this work they were seven: duration, starting (Fs), maximum (Fmx), final (Ff), mean frequency (Fme) and frequency band of the first harmonic. Moreover it was measured the amplitude difference (a.d.) between second and first harmonic. Mean duration of whistles is the same in every experimental condition and is no different from the duration of notes composing multiple whistle; so duration seems more related to anatomy and physiological
condition rather than the Marmots modulation of sound.
Regarding the other parameters, Fme is surely the most important frequency measuration (in fact it is the more reliable and however related to the others) while a.d. indirectly reflects glottic conformation during emission. During every experiment performed were recorded only single whistles; this fact underlines the strong correlation between this vocalization and aerial danger. Preliminary considerations show that predators silhouettes were able to elicit more single whistle than non predators (average 4.8 whistles for run versus 1.8; with P<0.01 using Fisher test). Moreover non predators models were frequently unable to elicit calls (42% of the runs for Ghough and 25% for Kestrel shapes). The Golden Eagle model was the more efficient stimulus of all (average 7.8 versus 3.8 of Goshawk and Buzzard, considered as occasional predators, with P<0.05).
Different models are able to induce different levels of excitation on the colony and this is related probably to dimension and silhouette represented.