Having processed the sounds of wolves and other members of the canine family with the help of computers, scientists found out that the howling of these animals has characteristic "dialects", depending on the species and subspecies. The new study is featured in the journal Behavioral Processes.
Arik Kershenbaum of the University of Cambridge and his colleagues have worked with howling canines (wolves, jackals and dogs) recorded both in captivity and in vivo. The six thousand howls were reduced to two and left to the machine learning algorithm to break them down into different types - based on the pitch and its vibration.
It turned out that the acoustic "prints" of the howl are not accidental - they differ depending on the species and subspecies of the canine. These features are of great importance from an environmental point of view. For example, overlaps in howl patterns (such as between a red wolf and a coyote, characterized by a modulated howl that turns into a whine) promote crossbreeding - which threatens the survival of red wolves as a species.
Although scientists have managed to classify the howling of canines, they know almost nothing about the meaning of various dialects - like the sounds of dolphins, howling of wolves is very difficult to study in natural conditions. Zoologists plan to install a variety of microphones in Yellowstone National Park and use them to determine whether certain signals are associated with communication at a distance or warning of danger.
