The largest ever study of howling in the 'canid' family of species — which includes wolves, jackals and domestic dogs — has shown that the various species and subspecies have distinguishing repertoires of howling, or "vocal fingerprints": different types of howls are used with varying regularity depending on the canid species. Researchers used computer algorithms for the first time to analyse howling, distilling over 2,000 different howls into 21 howl types based on pitch and fluctuation, and then matching up patterns of howling. They found that the frequency with which types of howls are used — from flat to highly modulated — corresponded to the species of canid, whether dog or coyote, as well as to the subspecies of wolf.