Why Wolves Howl
by John & Mary Theberge
It was a deep-freeze January morning, with mist peeling back in strands off the open riffles of the Lamar River like a series of gossamer curtains hiding a stage, eventually revealing the willow flats of the far shore. Out there, initially invisible, was the big Druid Peak pack. Their howls filtered to us through the mist. When they became visible, most of the pack were bedded while others were drifting around. But the black beta male, by that time famous, or infamous, known simply as "302" seemed anxious to move on. He trotted briskly downriver, disappearing briefly behind a mist curtain, and then reappearing on a knoll.
After looking back at his pack, he threw up his head and howled. Short, deep howls, breaking both up and down in pitch. The pack ignored him. He trotted on and then howled again. This time a few wolves got up and started his way. Then others. Soon the whole pack was down in the willows and into a draw that led them out of sight.
Why did he howl? The off-the-cuff explanation, overheard from a woman standing nearby, was that he was telling his fellow pack mates to get off their butts and follow.
She may have been right. This situation was one where a clear expression of intent seems to have been involved, at least subjectively. We have seen it infrequently both before and since. Sometimes the pack moves, but not always. Most often a move is initiated in silence. Over the span of a few minutes, one wolf after another gets up and heads out the same way. Or, one wolf howls, the whole pack joins in, and then they move off more or less together. Regardless, as a prelude to a move, howling is used inconsistently.
That inconsistency is shared in most of the 22 social or environmental situations we identified in Yellowstone. And that inconsistency is based on considerable underlying motivational complexity, not only in wolf howls, but in all animal vocal communication.
Competing triggers that may cause wolves to howl include some basic, but surprisingly slippery concepts. Intent itself, if based on memory and learning, can trigger vocalization, as anyone knows whose dog barks to be let in. Was 302's intent based on remembering a similar, successful experience either learned directly or observed? Or, did he think it out, on-the-spot reasoning, which is much more problematical? Intent can be classified by levels of complexity (Dennett 1983) and involve recursion or sequential reasoning such as we humans use all the time (Chomsky 1988, Corballis 2011). These higher levels of intent require a well-developed prefrontal cortex that wolves simply do not have. A preponderance of biologists would assert that non-human animals cannot engage in complex reasoning, especially reasoning that involves conceptualizing a number of steps (Hauser et al. 2002, Chomsky 1953).
Instead, the primary basis of vertebrate vocal communication is believed to be emotion (Suddendorf 2013), a conclusion reached by Charles Darwin in The Expression of Emotion in Man and Animals (which is still widely quoted). Even in chimps, "the production of sound in the absence of the appropriate emotional state seems to be almost an impossible task" (Goodall 1986). We share many midbrain structures for emotions with other vertebrate species. Included in the term "emotion" are the concepts of "internal drives" (Grandin 2005) or "internal motivational states" (Lord et al. 2009). Were 302's howls that day really motivated wholly or in part by hunger, exploratory behaviour, or care soliciting? Possibly those innate feelings caused adaptive neurohormonal adjustments in him; excitation of the sympathetic nervous system resulted in more cortisol or other biochemicals in the blood, and the result was that he howled. The biochemistry of emotions is a complex and active research area. If the outcome of such howls often enough was that the pack joined in on what turned out to be successful hunts, then natural selection would favour howling in that particular context without having any reasoned-out intent. After all, even bacteria communicate and coordinate with no brain at all: "Some bacteria move in voracious swarms called wolf packs, as do lions, wild dogs, and killer whales…" (Moffett 2011).
Did 302's howls actually express anything? That is, did his howls contain any specific information? From research we conducted years ago largely with captive wolves, we learned wolves have an amazing ability to distinguish differences in harmonic overtones in each other's howls, which makes individual recognition possible from a distance (Theberge and Falls 1967). Similarly, Palacios et. al (2007) identified structural differences in the fundamental frequencies of the howls of different wolves. We also learned that a higher level of excitement is reflected in higher pitched vocalizations (Theberge and Falls 1967), a trait in common with most mammals (Morton 1977). So 302's howls conveyed that it was he who was howling and he was to some degree excited or disturbed. Perhaps that was enough. His pack mates could recognize the situation and read into it that he wanted them to follow. But was other information coded in his howls?
The vast majority of biologists define language as requiring sophisticated syntax, that is, grammar and sentence structure to convey meaning; and they attribute that ability solely to humans (Corballis 2011, Pinter 1994, Chomsky 1953, Kruglinski 2009). Neurobiology backs this up. Only primates have a Broca's area, situated in the left prefrontal cortex, and a Wernicke's area, in the left parietal cortex; both are fundamental to expressing and understanding language. What about signalling or simple sounds that convey specific information referring to specific objects (called "referential communication" by Seyfarth and Cheney  or "protolanguage" by Bickerton )? Such signalling is well-known for species of primates (Cheney and Seyfarth 1990) and rodents (Slobodchikoff 2012), where different calls refer to different predators and elicit different avoidance responses. These vocalizations represent situations under intense selection pressure. We do not know yet if wolves exhibit referential howling. To find out, we have made hundreds of digital recordings for computer-based sonographic analysis. Whether language or protolanguage, or emotion or reason-driven, it is clear that wolf howling touches on deep concepts. For centuries these concepts have occupied the thoughts and writings of psychologists, physiologists, neurobiologists, ethologists, and ecologists—included are intelligence, reason, cooperation, language, cognition, and consciousness. We ultimately hope our research may help shed light on these difficult topics. But the place to start is with descriptions of when wolves howl. The why, the really tough part, comes later.
Yellowstone is an ideal place to study wolf howling with known and radio-collared animals, and open habitats where animal behavior may be observed. So in 2001 we turned to Yellowstone. Our starting place, besides our own studies, were statements reported about howling playing a role in territoriality during the breeding season, but not nailed down with quantification (Peters and Mech 1975, Harrington and Mech 1979, Harrington and Asa 2003).
Disconcertingly, we soon learned it takes a considerable effort to amass large enough sample sizes to draw statistically valid conclusions. The main reason is the inconsistency mentioned earlier. Some field trips yielded few howls where context could be identified. Howling is probably secondary to scent (Harrington and Asa 2003), despite the richness of both motivational triggers for wolf howling and the richness of social and environmental situations that wolves get into.
The inconsistency springs from another source and has yielded some interesting conclusions. The other source is a dramatic seasonal variability. Howling is four times more common in February than in May, the extremes of a smooth curve of change, except for a sudden drop at the end of February. We abstracted this pattern from more than 11,000 howls noted over a 10-year period by Yellowstone Wolf Project's Rick McIntyre. This pattern mirrors the annual pattern of serum levels of testosterone and estradiol in wolf blood. This finding indicates the reproductive state underlies wolf howling, not only during the breeding season but all year. The same pattern is found in other social carnivores that are monestrous (have only one confined and regular breeding season per year), such as coyotes and dingoes. No such seasonal pattern of vocalizations exists in polyestrous species (e.g., African lion and spotted hyena.)
Even more noteworthy are the 1,509 howling responses by pack mates and foreign wolves. They, too, were seasonal. Foreigners answered increasingly more often than pack mates from October (the time when wolf packs travel more extensively in their territories) to the end of February (the end of the breeding season). We interpret this period is one dominated by between-pack territoriality and mate-finding concerns. Then, abruptly, the situation changes—almost all answers throughout the denning and summer seasons are by pack mates. Defended territoriality, so prominent in fall and winter, almost ceases to exist. Replacing it is a near-complete shift to within-pack concerns, likely accompanied by a different set of dominant emotions. Wolves are into pup rearing and except for hunting forays, show little concern over neighboring packs.
Paralleling the seasonal frequency of between-pack howling are aggressive encounters between packs quantified by Yellowstone Wolf Project's Kira Cassidy (2015), who drew almost exactly the same curve as ours. The drop in aggressive encounters is sudden at the beginning of the denning season, even though actual inter-pack killing is highest in April, likely because the presence of pups restricts the defensive behavior of the pack being attacked (Smith et al. 2015). Summer is a time of relative between-pack harmony on the range, and fall and winter is a time of territorial and mate-finding tensions.
Other conclusions about the howls of Yellowstone wolves are on the horizon. One features two qualitatively different types of group or pack howls, which may lead to quite solid evidence of emotion versus reason in their utterance. We are anxious to flesh that one out.
The End of 302
It was late September 2009, and 302 was several years older. By then he had left the Druid Peak pack and emerged as the alpha male of the large Blacktail Plateau pack that lived to the west. Snow had fallen on and off for several days. 302 and his pack were on the Swan Lake flats, and they howled frequently when split up between opposite sides of the busy park road. It was classical disturbance howling, one of the key and most consistent situations triggering howling, whether caused by humans, bears, vehicles, or by foreign wolves. To our amazement, disturbance howling appeared to trump discretion. He and his pack were well out of their normal defended territory, trespassing on lands claimed by the Quadrant pack. They had interfaced with that pack before. Wolf packs know their boundaries well. After several days the resident pack came, attacked, and killed 302.
We have seen howling lead to several other wolf deaths. It is quite clearly a two-edged sword. It can be both adaptive and maladaptive, further complicating interpretation.