Ah-AAAAAH! Ah-AAAAAH! The peacock’s alarm call pierced the oppressive premonsoon heat at midday in Jim Corbett National Park, one of the remaining tiger reserves in northern India. My friend Nancy and I saw many types of birds and played with an orphaned elephant baby that was being hand-reared. We’d ridden on domesticated elephants that provided a sufficiently high vantage point to look down into the ten-foot-high vegetation—mostly wild Cannabis—that covered the unforested parts of the park. We’d found tiger kills and staked them out. We’d seen fresh tiger footprints, larger than my hand, and looked up and around with some trepidation. But after three days we still hadn’t seen any tigers. Then, on the fourth day we walked to a viewing tower near the campground and climbed up to the shaded platform to look for wildlife along a river that bisected the park.
Ah-AAAAAH! Peacock calls usually signal something they find disturbing, like the presence of an elephant, a boar, or possibly a tiger. The peacock flew noisily away. Minutes later, the riverbank was filled with monkeys—rhesus macaques—who looked back and forth before swimming quickly to the other side of the river. Once across the river, they climbed bushes and screamed while looking nervously back to the bank. A few moments later, a huge tiger emerged from the dense vegetation and slowly sauntered to the water’s edge. Pausing a moment, the tiger waded in and stretched out for a cooling soak. The tiger was enormous, at least twelve feet long. All who could fall prey to it gave it a wide berth. I was fascinated that day back in 1987 by how the warning calls of one species influenced others. How common is such interspecific communication? And if animals routinely warn other species, what happens when we begin losing species from a historically stable ecosystem?
In this chapter we will learn about how animals gain information about predators through vocalizations; both from members of their own species and from other species. We will learn what modulates the production of alarm vocalizations and what they may mean, which has implications for the evolution of human language. We’ll think about the reliability of alarm calls and the implications for others. Finally, we’ll apply this knowledge about alarm communication to better understand how we might effectively acquire information and respond to threats.
If you don’t have an opportunity to visit Corbett Park and hear the peacock’s alarm, perhaps take a stroll through the woods or another natural area with a dog. Depending upon where you are, you’ll likely hear the chattering of squirrels, the snorting of deer, and the trilling of birds. These vocalizations may serve three purposes: to communicate to predators, to communicate to other members of the same species (which we call “conspecifics”), and to communicate to other vulnerable heterospecifics (other prey species). While a vocalization may be intended for one purpose, targets are not mutually exclusive; a caller could be directing calls to all of them. It’s possible to warn your relatives that a predator is around while also signaling to a stealthy predator that it’s been detected. One of the many things that fascinates me about studying antipredator communication is that different recipients may select for varied features of alarm calls. To understand how, we first need to consider what we mean by communication.
Communication occurs when a caller produces a signal designed by natural selection to influence the behavior of a receiver. While animals respond to all sorts of cues for risk, as the monkeys did after hearing the peacock’s alarm call, communication requires the evolution of specific signals that have been selected for. These signals change the behavior of others. But therein lies a paradox; why emit signals that advertise both your presence and your location to a threatening predator?
Much of the scientific literature on alarm communication focuses on how these calls function to warn members of the caller’s own species. Specifically, a caller may benefit if relatives are warned. Recall the aim of evolution: to ensure that genes are passed on to the next generation. Because relatives share genes, the immediate answer to why prey would emit a potentially risky alarm call is that relatives are warned. By warning relatives, a caller protects its genes.
Classic research has shown that Belding’s ground squirrels increase their probability of emitting an alarm call in response to a predator based on the specific composition of their audience. When more of their close relatives are present, squirrels are more likely to emit calls. When distant relatives are present, squirrels are less likely to warn. And when only nonrelatives are present, the ground squirrels are least likely to sound the alarm. But it can be a bit more complex than this; all relatives may not be treated equally. For instance, we have found that yellow-bellied marmots are most sensitive to the presence of their vulnerable young. Mothers with pups present are most likely to emit warning calls, while other individuals do not. In sum, one reason animals emit alarm calls is to ensure that their genes make it to the next generation, either directly (through their direct descendants—their offspring and grand-offspring) or indirectly (through their siblings, cousins, nieces, and nephews). Regardless of the specifics, the family that calls together stays together.
Of course, another reason to emit an alarm call is to create pandemonium. Imagine you are being stalked by a predator. If you sound the alarm, other members of your own species will scatter, and the sudden, rapid, and chaotic movement around you may both shield you from the predator and make it substantially more difficult for the predator to focus on you. Admittedly this response is a bit more self-centered, but it still achieves the evolutionary purpose of survival—at least, survival long enough to pass on genes to the next generation.
Finally, calls can be directed at other species. Recall in Chapter 3 we discussed Gunther’s dik-diks, the gracile ungulates that are eaten by about thirty-six species of predators on the savannahs of Kenya. A student project studied whether dik-diks respond to the alarm calls from white-bellied go-away birds, a true sentinel of the savannah. They are large birds that perch on acacia trees and emit alarm calls when they detect predators. It’s currently unknown why they provide this public service, but many species seem to respond to their calls. We aimed to find out if dik-diks were among the species that responded to the white-bellied go-away birds. By broadcasting go-away bird alarm calls to dik-diks and contrasting the dik-dik’s response to nonthreatening bird song, we found that when dik-diks heard go-away bird alarm calls, they responded by running to cover, looking around more often, and foraging less. Thus a small, fearful ungulate can respond to alarm vocalizations from a bird.
Such eavesdropping is commonly seen in a variety of species and may be one of the benefits to species from living in what are referred to as “mixed-species groups.” In most cases it’s likely that the caller is directing its vocalization to either the predator or to its conspecifics, and that the eavesdroppers learn to respond to others’ alarm calls. For instance, the diminutive but colorful Australian superb fairy wren has been shown to learn to recognize previously uninformative sounds as alarm calls. My Australian colleague Rob Magrath and his students trained fairy wrens to specifically respond to these sounds. They did so by broadcasting them through a hidden speaker and immediately showing the birds a model raptor. After two days of training, the fairy wrens responded to the sounds as they did to their alarm calls; they had learned to associate a novel sound with the information that it conveyed about the presence of a predator.
But what actually happens when an animal hears an alarm call? Depending on how deep down you go into an individual’s physiological responses, hearing an alarm call can cause different genes to copy themselves, change the levels of stress hormones like catecholamines circulating in the bloodstream, or cause an animal to stop its current behavior and look around or flee, as we’ve just seen with dik-diks. We now know a lot about the conditions under which animals produce these calls, the physiological correlates with alarm calling, and the meaning of these calls. Based on my own work on marmots as well as key examples from research findings from other animals, including monkeys, meerkats, and rodents, we have learned a great deal about the immediate causes, meaning, and evolution of alarm calls. Let’s begin our study with marmots—specifically, at the end of the marmot’s digestive tract.
The marmots I study are herbivorous. Thus, they must eat a lot of vegetation, digest it a bit, eliminate it, and then eat some more. They spend their summer days eating, resting and digesting, and defecating. We, on the other hand, spend our days watching and live-trapping the marmots. Thus, it’s not unexpected to walk up to a marmot resting quietly in its cage trap and be greeted by a fecal sample waiting to be collected. If we’re lucky we collect even more when we reach into the conical handling bags we use to keep the marmots relaxed and still while we make measurements and collect data.
Feces are full of all sorts of good stuff to study. We can identify what marmots eat by looking at plant cell structure under a microscope, and we can find a variety of intestinal parasites by their eggs, which float to the top of a fecal slurry. Some folks extract DNA from sloughed-off intestinal cells in feces and use this to identify species and even count individuals of cryptic (difficult to see) carnivores. Feces also contain the digested remains of hormones—including stress hormones—such as the cortisol and corticosterone that we discussed in Chapter 1.
Quantifying these fecal glucocorticoid metabolites (FGMs) is a relatively noninvasive way to study physiological stress in animals. We’ve conducted a number of studies in our marmots that used FGMs. We know that we are looking at stress hormones in the feces because we conducted a simple experiment with a colleague’s captive marmots. First we injected adrenocorticotropic hormone (ACTH), a hormone that stimulates the anterior pituitary gland to produce stress hormones. Then we waited and collected all the feces from the captive animals, noting the time elapsed from the ACTH injection. About twenty-four hours later we found a peak in excreted glucocorticoid metabolites. Thus, we infer that fecal samples tell us something about an individual’s stress levels the day before we collected the fecal sample. While most individuals vary considerably both in the magnitude of and variation in stress hormone levels measured across multiple samples, some animals show multiple high-stress fecal samples. They are considered to have chronic stress.
Capitalizing on this knowledge, we began to note whether or not an adult female emitted an alarm call when we trapped her, and, if she was willing, we collected a fecal sample. Using this set of paired observations, we wanted to know whether a female was more likely to call when she had higher glucocorticoid levels. We discovered, rather unsurprisingly, that when females had higher background levels of circulating stress hormones, they were more likely to emit alarm calls. Therefore, stress hormones can be said to potentiate alarm calls, a finding that has been shown in a few other species, including rhesus macaques, a nonhuman primate. In the same way that you might be more likely to shout when you are under stress or scream when you are wandering through a haunted house, marmots and monkeys with higher levels of stress hormones are more likely to emit alarm calls.
Further, more recent work with our marmots at the Rocky Mountain Biological Laboratory has shown that socially isolated individuals are more likely to emit alarm calls. To determine this, we looked at individual marmots’ position in their social network. You could say that marmots, like many humans, have Facebook profiles: they interact with different numbers of other marmots and have different types of social relationships. Perhaps, like the employees at certain social media sites, we quantified marmot social interactions and calculated formal social network statistics to snoop on our clients. We described how socially connected each individual marmot was with other individuals in a given year. By doing so over many years, we found that those marmots who were less popular, measured by fewer interactions with other marmots, and those that had weaker relationships, interacting relatively less often with others, were more likely to emit calls when approached.
These findings match another series of results suggesting that socially isolated marmots are uniquely vulnerable. Like marmots, we might be more vulnerable to being mugged when among people we don’t know versus among a group of friends. We assume that our good friends will protect us if we are attacked, or that others won’t attack a group. Our marmot results also suggest that vulnerable individuals might direct their calls either to the predator to discourage pursuit, perhaps, or to other marmots to gain status by informing them of the threat. More work is needed to understand this intriguing possibility of status signaling.
But what do the calls mean? Unlike birdsong, which communicates identity and presence, and may provide information about the singer’s quality, alarm calls, and to some extent food calls, uniquely have the ability to be referential. Referential signals are signals that can refer to external objects or events. Thus, one type of call could tell others that there is a predatory fox around, while a different call would signal the presence of a hunting eagle. If so, alarm calls could function as basic words, and animals could use this information to develop a rich understanding of the world around them.
To understand why the study of referential signaling in animals may be a big thing, let’s go back to Darwin. Darwin noted that while humans had language, nonhumans were able to communicate only their emotions and not information about specific objects or events outside their body. Thus, nonhumans were not expected to have referential signals. Later, when we learned that honeybee waggle dances can encode the direction and distance from the hive to a patch of flowers and communicate this information to other foraging bees, people at first were not too fussed—it’s an exception to the rule in one species of tiny insects. Referential signaling in nonhumans challenges our belief that human language is unique—or at least it challenges the idea that only humans have the ability to communicate about external objects or events.
To test this potential, the quest for wordlike communication led to the development of a subdiscipline of researchers studying the meaning of the calls of primates, rodents, and birds. Evidence for wordlike communication, what has been called referential signaling, would be seen if two criteria were met. First, there had to be a high degree of production specificity; each call type was reliably produced by a signaler in response to the specific stimulus. For instance, upon detecting a fox, animals would always produce a specific-sounding fox-elicited alarm call. This call needed to sound different from the call elicited by an eagle. If, by contrast, a fox was detected very close to the caller and elicited a more eaglelike alarm call, the calls might be construed to be communicating the degree of risk and urgency of the response. In other words, the close presence of a fox is as risky as an eagle, which can fly swiftly and strike like a bolt of lightning. The second criteria was that there had to be some degree of contextual independence, or what’s been referred to as “response specificity.” For example, an animal hearing a fox-elicited call should respond as though a fox were nearby, while an animal hearing a raptor-elicited call should respond as though there was a raptor present. For animals with uniquely different escape strategies this is, in theory, relatively straightforward to quantify.
One of the earliest examples of such referential signaling came from a series of studies by Thomas Struhsaker, Peter Marler, Dorothy Cheney, and her husband and research partner Robert Seyfarth. Struhsaker and Marler’s initial observations of vervet monkeys in Uganda led to a series of detailed experiments later conducted in Kenya. These cat-sized, black-faced monkeys live in the savannahs and seek protection in trees. Vervet monkeys may be recognizable even if their name is not immediately familiar. Males are notable for their quite distinctive brilliant blue testes and bright red penis. Females are notable because they inherit their social rank from their mothers; a rather depressing example of transgenerational inheritance. Their initial observations in Uganda suggested that the vervet monkeys produced and responded in different ways to the uniquely distinctive alarm calls elicited by raptors, snakes, and terrestrial mammalian predators—in ways that suggested that these alarm calls functioned as simple words that communicated to others the type of predator detected.
Dorothy, Robert, and Peter aimed to see if a Kenyan population of vervet monkeys showed the same ability. Their experiments are summarized in Dorothy and Robert’s classic book How Monkeys See the World. Vervet monkeys spend a lot of time each day foraging on the ground, which is risky because the savannah is filled with predators, and trees provide a quick refuge. But vervets must travel between trees, and during this time they are quite exposed. They have evolved a variety of adaptations to reduce these risks: they can live in large, female-dominated social groups, which allows for increased security, and, of course, they emit alarm calls. The researchers found that vervets in the Kenyan savannah show an ability for referential signaling.
Upon detecting a python, vervets chutter. Upon detecting a leopard, vervets emit a series of short tonal calls. Upon detecting an eagle, vervets emit a grunt. Further, vervets hearing these predator-specific calls respond in unique ways. Their behavior changes based on the call. Upon hearing chutters, emitted once a vervet sees a snake, vervets get up on their hind legs and bipedally approach the caller, looking in the vegetation for a snake. Then, if they see the snake, they circle around it and harass it with chutters. They may even attack. Letting a snake know it has been detected and trying to get it to move away is a form of mobbing that many species engage in with predators that particularly rely on stealth for their success. Upon hearing leopard alarm calls, vervets run to trees and move to the most peripheral branches, where they might be safe from a heavier leopard, which can’t climb out to those distal locations. And, finally, upon hearing the raptor-elicited grunts, vervets on the ground run to trees, aiming for the center of the crown to be safe from raptors who can’t dive into the dense area without risking personal injury. Upon hearing the call, the vervets in trees drop down to the tree’s center for safety.
Seeing vervets or other monkeys respond to a predator is quite exciting. I first experienced this in 1986 when I was studying blue monkeys in the Kakamega forest in western Kenya. Whenever a martial eagle flew over, it literally rained monkeys! Each monkey either jumped or rapidly climbed down from high branches to seek the safety of the central part of the tree, where the eagle could not easily reach them. More recently, in Costa Rica, I heard but sadly did not see (due to the lack of visibility) a troop of spider monkeys respond to what must have been a large predatory cat. They all moved to a single location and collectively produced loud, rapidly repeated vocalizations. The rain forest was filled with these cries, and they seemed to work. Eventually it sounded as though the now-detected predator slinked away in search of less-alert prey.
Dorothy and Robert conducted further experiments in the Kenyan savannah specifically designed to determine whether the monkeys responded in unique ways to the unique alarm calls. After they hid the speakers, they played back specific calls and filmed the vervets’ responses. Through these broadcasts of predator-elicited vocalizations, they discovered a degree of response specificity that suggested the monkeys had simple words for predators. Thus, vervet monkeys can be said to produce referential alarm calls. The calls act as though they communicate predator type. Similar sorts of referential signaling have also been reported in other monkeys, in chickens and some species of birds, in Gunnison prairie dogs, and in meerkats.
But not all species have these referential abilities, and there may be some variation in this fascinating cognitive ability even among populations of the same species. A recent study with a South African population of vervets has failed to replicate Dorothy and Robert’s initial findings. In the South African population, vervets did not immediately flee in predator-appropriate ways. Instead, they looked toward the hidden speaker broadcasting the alarm calls. When they did flee, it was not always with the expected predator-specific response. The authors—Nicholas Ducheminsky, Peter Henzi, and Louise Barrett—suggested this may reflect the substantially larger group sizes of these vervets and the greater distance at which individuals heard natural alarm calls. Thus, they proposed that the referential responses that Cheney and Seyfarth discovered may not be a property of a species or a fundamental cognitive ability. Rather, referential responses may be more malleable and reflect social and ecological conditions.
In the case of the South African vervets, the calls from a greater distance implied vervets hearing them were at relatively less risk of predation. Much as you might immediately get down if someone next to you yelled “DUCK!,” you might look around for the source of the risk if someone far away yelled “duck.” The larger vervet group size implied vervets may have been generally more safe than those in Kenya. If so, we can hypothesize that an individual’s relative safety may be an important factor that influences cognitive abilities. Animals living under relatively higher risk may need to be more cognitively sophisticated than those living under lower risks. Finally, in larger and more spread-out groups, there may be more false alarms. Vervets hearing false alarms should be more discerning and look first to identify the true threat before engaging in any predator-specific escape behavior.
Others have questioned the utility of studying the meaning of calls, and whether information is a valuable metric to quantify when studying communication. On the one hand, referential calls may not necessarily resemble words that label predators; calls could be instructions for others, such as “run to a tree” or “get up on your toes and look around.” On the other hand, those that question the value of information contained in animal signals often ask what’s in it for the signaler to provide meaningful information if communication is about manipulating the behavior of others. Besides, as these people note, information isn’t an entity; it’s an abstract concept. Evolution shouldn’t act to optimize information transmission; it should act to optimize fitness.
My view is that signals contain potential information, and it’s vital for animals to acquire information about potential risks. Even if it’s difficult to imagine what information is precisely, you know it when you see it. If your behavior changes as a function of a detecting a signal or having an experience, you have obtained information about something that is potentially valuable. This is true for nonhumans too—if their behavior changes after hearing a vocalization, sniffing a scent, or seeing a particular image, those stimuli contained information.
I also find the quest for wordlike communication worthy of study, even if we find that it’s relatively restricted. This is because it’s an attribute of communication that is not found in all species, and this variability demands explanation. What explains the evolution of cognitive abilities? Why are some species “smarter” (admittedly a loaded word) than others? Ultimately, I think we should be studying the conditions under which communication referentially evolves or is favored by natural selection.
I’ve spent over a decade studying the evolution of referential communication in eight of the fifteen species of marmots. Based on previous reports suggesting that some species of marmots had a high degree of production specificity (seen when a single predator type elicits a single type of alarm call and different predator types elicit different types of alarm calls) while other species were not reported to have this ability, I set out with microphones, tape recorders, and speakers in tow to study both production and response specificity in marmots. And, ultimately, I aimed to study the evolution of referential communication in them as well.
My studies began in Pakistan, with golden marmots, then to the Alpine marmots in the Berchtesgaden Alps in Germany, to Ohio and Kansas for research on groundhogs, and Utah and Colorado to learn about yellow-bellied marmots. From there I traveled to the Olympic Peninsula and Mount Rainier National Park in Washington for my work studying Olympic and hoary marmots, respectively, to the steppes of Russia for work with steppe marmots, and on to the central mountains on Vancouver Island for research on the critically endangered Vancouver Island marmot. My wife, Janice, and I spent thousands of hours watching marmots in lovely Alpine settings and conducted a variety of experiments. We walked toward the marmots to observe their reaction. We flew an eagle-sized radio-controlled glider above them. We named our glider Eagle Knievel because it was difficult to land a glider on rocky hillsides. I spent hours fixing it, much like Evel Knievel’s long recoveries after his adventures on rocket-powered motorcycles in the American West. We drove a remote-controlled stuffed badger, RoboBadger, at the marmots and broadcast different alarm calls back at them. We found, somewhat unexpectedly, that none of the species produced predator-specific calls. Indeed, a human walking toward marmots could elicit many of the different call types that these species produced. So instead of producing predator-specific alarm calls, the marmots seemed to communicate degree of risk they experienced. And they did so in a remarkable variety of ways.
The golden marmots in Pakistan varied the number of calls they packaged together. These animals, living in a vibrant and intact predator community, would begin calling at me when I was over 100 meters away. As I approached, the number of calls packaged together into multinote utterances decreased. Thus, golden marmots tracked risk by varying the number of calls produced. As risk increased, they produced less conspicuous calls. By contrast, yellow-bellied marmots increased the rate of calling as risk increased. They did so with people but also called at higher initial rates when they detected an eagle or our model eagle.
As previously described, my more recent studies suggest that one of the functions of marmot calls is likely to communicate directly with the predator, perhaps to signal that it’s been detected. Indeed, Erin Shelley, then an undergraduate honors student in my lab, and I conducted an evolutionary analysis of alarm signaling and concluded that the initial function of alarm signaling was likely directed to predators to discourage pursuit. We found that in 209 species of rodents, the evolutionary origin of alarm signaling was not associated with the evolutionary origin of sociality but rather with the evolutionary origin of being active during the day. Rodents active during the day seemingly benefited by emitting alarm calls directed to their predators. Further, while none of the species had referential abilities they had evolved a variety of ways to communicate risk. By packaging calls, varying the rate, and yes, by producing different call types, marmots communicated degree of risk to other marmots and potentially communicated to their predators that they had been detected.
Another interesting finding was the variation in call repertoire size. After recording hundreds of alarm calls in each species and making spectrograms (voice prints) that plot time and frequency (the pitch) and amplitude, I could sort calls by type. The critically endangered Vancouver Island marmot produced five different types of alarm calls in response to predators. By contrast, most other species of marmots produced one or two call types. Although Vancouver Island marmot alarm calls were not produced with sufficient production specificity to be considered referential, remarkably the order that we broadcast calls back to marmots influenced the magnitude of their response. It was as if Vancouver Island marmots had simple syntax! To note, only a few studies suggest that nonhumans have syntactical abilities—and syntax is one of the key traits that characterizes human language.
To expand our discussion of alarm calls, let’s consider meerkats—small, adorable, and highly social carnivores. They live in the deserts of southern Africa where they forage for insects, cooperatively defend vulnerable young, and face existential threats from a variety of aerial and terrestrial predators. My friend and colleague Marta Manser, a professor at the University of Zurich, has studied meerkats for years in the Kalahari Desert of northern South Africa. She now directs the Kalahari Meerkat Project, the site of the popular television show Meerkat Manor. In this program cameras followed around Flower and her fellow soap opera stars through the trials and tribulations of life as a meerkat in the Kalahari. In the background, Marta was conducting elegant experiments to understand their communication system and the meaning of their vocalizations. Marta discovered that meerkats have referential signals—they produce different calls in response to terrestrial and aerial predators. And importantly, they can simultaneously communicate risk. Marta aimed to discover the types of information they communicate in their calls. As she has shown, meerkat calls get noisier (remember, fear-inducing sounds are noisy) as risk increases. Thus, meerkats produce either a low-risk or high-risk call in response to an aerial predator. Meerkats that hear these different calls respond accordingly.
So what have we learned from this collection of studies? First, nature is grand in its diversity, and detailed study is needed to tease out the specifics. There are a number of putative adaptations to avoid predation, many that involve emitting alarm calls. Presumably, all work sufficiently as most species are not going extinct due to a faulty antipredator alarm system. Communication systems illustrate the diversity of adaptations that have evolved to solve similar problems.
While much of my comparative marmot work was designed to understand the rudimentary steps of the evolution of language-like communication, in retrospect this focus may blur the remarkable diversity of life that surrounds us. This diversity is ripe for study and as a source for creative inspiration. Biomimicry, a term coined by Janine Benyus, is a field that uses ideas from nature to solve human problems. For instance, engineers created body suits for Olympic-class swimmers after observing the sleek movement of sharks and analyzing their smooth skin. Geckos are renowned for their ability to climb vertical glass surfaces, and the structure of their footpads has inspired new adhesives. Much of biomimicry seeks novel solutions from physics and chemistry.
What about biomimetic lessons from the diversity of antipredator behavior? Can our knowledge of behavior be applied to increase our human security and defense? One such lesson is almost trivial—don’t overreact. All successful animals have evolved from ancestors that were able to manage their risks well. And we know that across the scale of biological organization—from immunological responses to disease threats to behavioral responses upon detecting predators—it’s extremely costly to overreact to a threat.
Physiological reactions include autoimmune diseases, which may be a fatal overreaction of the immune system to noninfectious stimuli. The immune system has evolved to respond to infectious challenges. Normally, it tolerates commonly encountered stimuli. But when your immune system is primed, these harmless stimuli are suddenly viewed as threats. Energy is misallocated to unnecessary defense, and the body itself is attacked. Those who study Evolutionary Medicine are exploring novel therapies to counter autoimmune diseases. And some of these therapies may capitalize on providing immune systems with benign threats, like nonlethal parasites, and thus, in a sense, occupy or otherwise distract the immune system from attacking the body itself.
On a much larger scale, consider the US response to 9-11. Over 3,000 people were killed, more than 6,000 injured, and countless more lives were ruined in the aftermath of these terrorist attacks. But according to the US National Highway Traffic Safety Administration, 10,874 people were killed from drunk-driving related crashes in 2017 alone. We will discuss in Chapter 12 some reasons why we react in horror to terrorist attacks in New York but not the constant and widespread toll exacted by drunk driving. If we had the mental clarity of Star Trek’s Mr. Spock we might be puzzled by the American reaction to 9-11. Spock may have suggested we further reduce legal blood alcohol levels for drivers, impose stiffer penalties for drunk or alcohol-impaired drivers, and better enforce seatbelt laws. Rather, in our quest for security, we have created what has become an endless war in Afghanistan, resulting in substantially more pain and suffering. By early 2020, there had been over 2,300 US military deaths and over 20,300 US military wounded. This doesn’t include deaths and injuries of US government contractors, American allies, or the long-suffering Afghan people, nor the horrific effects of PTSD suffered by returning service members or by Afghans. Wars, once started, are often difficult to end. Overreaction to a threat can be costly, leading to death and impoverishment.
We should not forget that a marmot which overreacts to an alarm call by never leaving its burrow will eventually starve, and a marmot that forages foolishly may be killed. These trade-offs became very clear in March and April 2020, as the world reacted to the COVID-19 pandemic by shutting down most movement. If we listen to our inner marmot, we will know that individuals and nations that neither cut themselves off from others nor overinvest in defense will ultimately outcompete those that do. Successful marmots keep in touch and work together, something that was tragically lost in the initial nationalistic responses to the coronavirus pandemic.
Another lesson from risk communication is that alarm calls are often individually distinctive. Why would such an individual signature matter, and what purpose would it serve? Of course it could be an unselected by-product of how sounds are produced. In mammals and birds, air is forced through a vibrating sound production organ (the larynx or syrinx), and the sounds resulting from these vibrations are filtered by the vocal track to create the sounds that we ultimately hear. Morphological variation in the vocal tract could be responsible for idiosyncratic differences in vocalizations. Work with Kim Pollard, then a PhD student, revealed that in marmots, ground squirrels, and prairie dogs, species that live in typically larger social groups have more individually distinctive vocalizations. And I found that the acoustic features that seemingly communicate individual differences among marmots do not degrade as much when they are broadcast through the environment as those features that communicate risk. Both lines of evidence suggest that individuality in alarm calls has been subjected to natural selection and is not simply the unselected by-product of morphological variation in the vocal tract. In other words, a distinctive vocalization serves a purpose. We can infer past selection when we see variation in some trait—in this case, individuality—map nicely onto environmental or social variation—in this case, group size. Indeed, it makes sense that contact calls—vocalizations given by some primates, meerkats, and social birds that are used to keep track of others of the same species (conspecifics)—will be individualistic because you can’t keep track of different individuals if they all sound the same. If there is a need to sound different, we expect there may be selection on the signaler to produce individually distinctive contact calls and selection on the receiver to distinguish them. One great example of the necessity of distinctive contact calls can be found in crèching penguins and marine mammals, who must find their young among a noisy and smelly colony with hundreds to thousands of other screaming young so that they can feed them when they return from foraging sorties. Both adults and chicks produce and respond to individually distinctive vocalizations.
But how does all of this relate to our understanding of alarm calls? We know that marmots and other ground squirrels have individually distinctive calls, but what’s the benefit to either the signaler or the receiver in producing or distinguishing them?
Reliability assessment may be the key. In Chapter 7 we learned that all individuals may not be reliable. Recall Aesop’s fable about the shepherd boy who cried wolf. It is a clear representation that individuals vary in reliability. Moreover, we have a mechanism to explain how this may come about. We know that stress hormone levels can influence the probability of emitting calls, thus individuals could have different calling thresholds, which are modulated by stress hormone levels. If so, then some individuals may be generally more likely to call in response to both real predators and nonpredators, while others may call only when there are predators around. Let’s call these individuals Nervous Nelly and Cool-Hand Lucy. Cool-Hand Lucy is a reliable caller, whereas Nervous Nelly is unreliable. If Aesop’s fable is accurate, then we should expect Nervous Nelly to be ignored, much as the boy who cried wolf was eventually ignored.
We asked whether this hypothesis could explain why and how marmots respond to alarm calls. Because we were unable to properly estimate reliability, as it’s really hard to know what stimulus triggers alarm calls in many situations, we conducted a playback experiment. We created a reliable marmot by pairing her calls with the presence of a stuffed badger and an unreliable marmot by pairing her calls with the presence of no badger. We employed a nifty experimental approach that Dorothy Cheney and Robert Seyfarth originally borrowed from developmental psychologists who study preverbal children. The technique is called a habituation-recovery protocol. This is the same technique used in Chapter 2 to study spider recognition in human infants. Again, when properly used, the technique permits us to make inferences about how animals classify different stimuli. In this case we first asked how an individual marmot responded to a random set of calls from a novel marmot. We did this for a number of different individual marmots, and we called this the pretest stage of the experiment. We then created a set of reliable (R) and unreliable (U) callers by broadcasting their calls with or without the badger present to the same set of individuals we had just pretested.
We then tested to see if marmots could distinguish callers based solely on their reliability. We broadcast either a new call from R or a new call from U. If R and U represented reliable and unreliable callers, we would expect two things if marmots had learned about caller reliability. First, they would increase their responsiveness to the reliable caller. Second, they would decrease their responsiveness to the unreliable caller. In the case of our playback experiments, we baited animals to a central location, broadcast the sound, watched them respond by looking, and quantified the time it took them to resume foraging. If the boy-who-cried-wolf hypothesis about reliability was to be supported, we expected to see marmots who heard the unreliable caller resume foraging sooner following the playback.
What we found was the exact opposite of what we expected: marmots hearing the unreliable caller stopped foraging for a longer period of time, mostly using this time to look around, while those that heard the reliable caller initially looked around and then resumed foraging. After dismissing the results as a silly mistake, we thought deeply about the reasons why marmots might respond more to unreliable callers. After all, these results are inconsistent with the boy-who-cried-wolf hypothesis, which has been suggested to explain individual specificity in the alarm calls of so many animals, including steppe marmots, Richardson’s ground squirrels, bonnet macaques, and rhesus macaques.
What, after all, does reliability mean? It means that you can reliably infer something. By contrast, unreliability means that it will be difficult to reliably infer something. If what you’re inferring is the presence of a predator, then perhaps it does make sense that after hearing an unreliable individual, marmots looked around more to independently assess what the true risk of predation was. It’s as if they knew that even reliable animals make the occasional mistake, but you really can’t trust the unreliable ones. Thus, unreliable callers or situations induce independent investigation.
Other lines of evidence support this idea. Marmots pay more attention or are more responsive when they encounter uncertain situations. They forage more after hearing alarm calls from older and presumably more reliable marmots compared to potentially unreliable calls from pups. Marmots also forage more after hearing undegraded calls compared to acoustically degraded calls. All sounds degrade as they are transmitted through space, and the farther they are transmitted, the more they degrade. The idea, therefore, is that undegraded calls simulate a situation where the caller is nearby; the sound was transmitted a short distance and was minimally degraded. Thus, after an immediate look up and around for the source of the alarm call, marmots resume foraging. When they hear degraded calls, they look around more, presumably because they are less certain about the true risk of predation. A degraded call could mean a caller is looking at you or away from you—and one looking at you might mean that the predator is nearby! So the take-home message is that unreliable individuals or situations are unreliable specifically because it’s difficult to assess the true risk of predation. Given this difficulty, risky situations elicit independent investigation. And as with the production of alarm calls, we should generally expect evolutionary flexibility in mechanisms of communication.
Reliability assessment is likely to be a general explanation for the evolution of the ability to discriminate among callers. This illustrates another possible biomimetic response—when a situation is truly uncertain, some animals allocate more effort to assessing it correctly. This may (or should) sound a little familiar. Let’s think about how we humans respond to uncertain sources of information. We are confronted with a twenty-four-hour news and spam cycle that churns out vast amounts of potentially contradictory information, including some that is entirely erroneous. Assessing the risk and reliability of this information is now more important than ever. We all have an inner marmot; we have evolved mechanisms to believe trusted sources. But our evolved systems can’t keep pace; we face an evolutionary mismatch in which our evolved mechanisms have broken down, and we are apt to believe untruths or exaggerations. There’s simply too much potential information to process. Mindful of this, I suggest that we scrutinize our news sources. If it sounds too good to be true, it could be a scam. News sources that follow strict journalistic practices and fact check their sources are, without question, going to be more reliable on average than those that simply aggregate information. The rise of fake news means that we must relearn how to trust but verify. And we must support reliable journalism that properly fact-checks sources because there simply isn’t enough time for each of us to fact check everything we hear. We need good information to make informed decisions.
Animals use a variety of ways to communicate alarm signals. Anything that is produced in response to a threatening situation has potential signal value for an information-seeking recipient. Some rodents, such as kangaroo rats, rapidly beat their hind feet against the ground in elaborate bouts of foot drumming. Some of this foot drumming has been shown to signal territory ownership, but species also foot drum in response to detecting predators or cues from predators. Indeed, snakes have rather poor hearing but feel vibrations quite well. They are often the target of foot drumming by rodents keen to move them out of their territories. Conspecifics hearing a bout of foot drumming may wisely increase their vigilance and look around for snakes.
Besides foot drumming there are a variety of other sounds associated with increased risk. Some pigeons and doves, when alarmed, fly up and away and produce whistles with their wings. Rob Magrath’s group and my students and I have both studied the signal value of these whistles. Rob and his students worked with the crested pigeon, an Australian pigeon with a distinctive spiny crest of feathers coming off its head. They discovered that when pigeons feeding on the ground were suddenly startled, they took off at a steeper angle than they would have if they were not as alarmed. These pigeons’ specially modified feathers produced a whistle when they took off at a steep angle—an honest indicator of perceived threat. When the researchers broadcast alarmed wing whistles versus other flight sounds, they found that wing whistles were much more likely to cause a flock to take flight than nonalarmed wing sounds. Thus, wing whistles have the ability to communicate risk nonvocally.
In the course of research in the Virgin Islands, my students and I studied zenaida doves, the national bird of the nearby Caribbean territory of Anguilla. Like crested pigeons, zenaida doves produce mechanical wing whistles when they take off quickly. We wished to know how doves responded to them compared to other possible sources of risk information. We conducted a playback experiment and broadcast wing flaps to doves that contained whistles and wing flaps that did not contain whistles. We found that doves increased vigilance significantly more in response to wing whistles than to wing flaps without whistles (or in response to control playbacks). These results indicate that conspecifics interpret wing whistles as alarm signals. We then conducted another playback experiment which demonstrated that playbacks from a potential predator, the red-tailed hawk, elicited higher levels of dove vigilance than the wing whistles of fellow zenaida doves. Taken together, these results suggest that, contrary to other species, zenaida doves seemingly consider predator vocalizations more informative than conspecific alarm signals. This means that the quest for reliable information should be broad and not restricted to a single species or source of information.
As we have seen, wise prey use any cues they can to accurately assess predation risk, including the alarm calls that other species produce or the cues produced by predators themselves. What’s particularly interesting about communication between species is that different species face different risks. Calls from another species may provide less reliable information about the true risk of predation. A number of playback studies have shown that mammals are likely to respond to alarm calls of other mammals as well as alarm calls from birds. What general factors might influence whether or not one responds to another species?
As a rough approximation, the size of an individual or species explains much about what it should fear. Small fishes are incredibly vulnerable to predation, and as they grow, they have fewer predators. Small mammals also may be vulnerable to a larger suite of predators. If so, a larger body size may be protective. As an aside, this likely explains why our corgi, Theo, was fear conditioned by the direct hit of nearby lightning—the extremely loud clap of low-frequency thunder could be interpreted as a nearby HUGE animal. Thus, vulnerability is related, at some level, to body size, and we see a variety of adaptations at both the individual and species level to counter these risks.
If the enemy of my enemy is my friend, how does body size influence the value of information? Imagine a small golden-mantled ground squirrel hearing an alarm call from a larger-sized marmot. What information could the squirrel infer from this? Or what information about predation risk could a marmot obtain from a hearing a squirrel? It turns out that wise squirrels should listen to marmots, but wise marmots may sometimes ignore squirrels. This is because everything that eats a marmot will eat a squirrel, but the opposite is not necessarily true. Marmots are about seventeen times larger than these squirrels. If this size discrepancy alone was responsible for explaining response, then mule deer, which are about sixteen times larger than yellow-bellied marmots, should ignore marmot alarm calls. It turns out that mule deer hearing marmot alarm calls immediately looked toward the speaker and sometimes fled. But why should deer listen to marmots when marmots don’t listen to squirrels?
The answer may lie in sharing important common predators. At our Colorado study site both deer and marmots are preyed upon by coyotes and, before wolves were hunted to extinction, wolves. Thus, sharing important predators may trump size discrepancies. Information is potentially valuable even if it comes from a much smaller animal. This lesson is highlighted by the fact that vervet monkeys respond to alarm calls by the diminutive superb starling, a brilliantly colored bird found in East Africa. Even though vervet monkeys are sixty-three times as large as starlings, they share some subset of both aerial and terrestrial predators. Vervets are able to respond to starling alarm calls, an ability that is certainly the result of learning since the species do not coexist throughout their ranges. However, they are quite discriminating and can be trained to ignore all calls from unreliable starlings.
Much as a peacock’s call can be used to find a tiger, all organisms naturally seek information about risk from a variety of sources. But all sources are not equally reliable. It’s essential to evaluate a source’s likely accuracy before acting on it. Acquiring information about risk is essential, even if it comes from other species. Alliances, whether evolved or learned, should be expected. Thus, maintaining intact communities of animals may be vital for ecosystem stability and species persistence. We’ll learn much more about the importance of fear in ecosystems in the next chapter. Regardless, a lesson for humans is that those tasked with assessing the risk of fearful events will benefit from seeking information from all who share similar risks.