Editor's Note: The following is an excerpt from Cristina Eisenberg's  book The Wolf’s Tooth.

A doe burst out of the forest and tore across the meadow, two wolves in close pursuit. This drama unfolded not twenty feet from where my young daughters and I knelt in our garden peacefully pulling weeds, our pant legs wet with morning dew. One black, the other gray, the black wolf in the lead, they closed in on the doe's haunches. In less than two heartbeats they pierced the deep wood on the far side of the meadow, leaving a wake of quaking vegetation.

We live at the base of a mountain in northwestern Montana. As wild as the wildest places in the lower forty-eight United States, it isn't quite paradise, although the handful of us who live here think it comes close. Midway up the mountains, overgrown clear-cuts show up as yellow-green rectangles against the darker green of old-growth forest. From our cabin you can walk due east beyond the state forest lands and not encounter much more than federally protected wilderness for a hundred miles.

Landscapes shape us and speak to us on a primal level. Most of us have a landscape we intuitively comprehend. This is mine. I open the front door of my cabin and find wolf tracks pressed into the snow. In spring, even before I see the grizzly lumber out of the forest to dig roots, I smell its ripe essence. These discoveries give me pleasure and an unspoken awareness of the natural order of things.

Humans also have a primal relationship with large predators. This relationship has been eloquently elucidated across the ages in Paleolithic petroglyphs of dire wolves and other creatures sharp of tooth and claw and in medieval paintings of wolves menacing sheep. Wolves began to recolonize our area in the early 1990s. Since then we had been hearing them howl from the shoulder of our mountain and occasionally finding their tracks. But we had never seen them— not until that misty August morning when they ran across our meadow. For some long moments after they passed we knelt motionless in the garden, at a loss for words. Then curiosity kicked in and we stepped outside our small fenced yard to follow the wolves' trail.

I marked one track, and from it we located others laid out in a gallop pattern. We even found the spot where one of the wolves had turned to look at us, a motion that had caused its left front foot to break forward. Fascinated, we continued to follow the subtleties of their trail—which sometimes consisted of little more than a few bent blades of grass that even as we watched were springing back upright. And I wondered how many other times wolves had run through our land and I'd missed the evidence.

In the fifteen years since wolves returned, the deer had been behaving differently—more wary, not standing in one place, eating all the shrubs down to nothing. After the first three years I seldom saw deer browsing in the meadow, and then only for brief periods. And after a decade the meadow was nearly gone, with shrubs and young aspens filling in what used to be open grass. Until we saw the wolves hunting, I had never actually observed a trophic cascade in action.

The Green World Hypothesis
In 1969 Loren Eiseley wrote an evocative story, "The Star Thrower," about a man who walked the strip of wet sand that marks the tide's ebb and flow, tossing sea stars that had washed ashore back into the ocean. Motivated by a need to save them from death, each day he returned the stars to the ocean. At one point in the narrative, Eiseley commented on the apparent futility of this task.

In the real world the star thrower is a scientist, and death is running even more fleet than he across every ecosystem on this earth. Like Eiseley's star thrower, Robert Paine is motivated by promoting life, although the results of his actions are far less futile and best described as utterly Eltonian. Considered the father of trophic cascades science, Paine has spent most of his life studying aquatic communities. I visited his University of Washington lab, which overflows with the products of a long and illustrious career: books and papers he's authored, photographs of the intertidal world he's studied for so many years. He sums up his work in two sentences: "You can change the nature of the world pretty simply. All you do is get rid of one species." As a young scientist he proceeded to do just that, experimentally removing sea stars in control plots, thereby creating profound changes in the intertidal community he studied.

In 1960 Nelson Hairston, Frederick Smith, and Lawrence Slobodkin (HSS) proposed that vegetation patterns are determined primarily by patterns of food consumption by herbivores. They further suggested that the act of predation shaped herbivory. Therefore, herbivores unchecked by predators would have a great influence on vegetation. Known as the green world hypothesis, their ideas provided an ingeniously simple explanation for why the earth is green. This hypothesis resulted from a seminar at the University of Michigan, in which scientists met to discuss central issues in terrestrial ecology and arrived at this remarkable conclusion. Their ideas were inspired by Charles Elton's food pyramid.

As a new professor at the University of Washington, Paine tested the green world hypothesis. The idea for a simple experiment that essentially involved playing God in the rocky intertidal zone came to him during a visit to the Scripps Institution Wharf in California, as he stood watching the carnivorous sea star Pisaster ochraceus devour the mussel Mytilus californianus. What if he removed sea stars to see what would happen? The next year, National Science Foundation funding in hand, he settled into the University of Washington and began his research. Month after month he traveled to Mukkaw Bay, at the northern tip of the Olympic Peninsula. There, on a rocky crescent of shore, he hurled sea stars into the ocean. In his control plot he did nothing. As he continued removing stars, the assemblage of species on the rocks gradually began to change. Within one year the sinister implications of his experiment became all too graphically obvious. Where Pisaster flourished, so did the vegetation. Where Pisaster had been removed, the mussels took over, crowding out other species and eating all the vegetation, until little more than a dark carpet of mussels and barnacles remained. The paper Paine produced for the American Naturalist about this research turned out to be one of the most influential journal articles in the history of ecology. It provided one of the first examples of an ecosystem that had been transformed by trophic cascades.

Mentored by Frederick Smith of HSS fame, Paine was also influenced by Charles Elton and Charles Darwin. He thinks the green world hypothesis may have had its genesis during Smith's 1957 course at the University of Michigan on the natural history of freshwater invertebrates, on the kind of spring day when professors don't really feel like teaching and students don't really feel like sitting in class. That day Smith looked out the window of the zoology building, which faced a courtyard, and said to his students, "We're not going to have a field trip today. I want to teach you how to think instead. Tell me, what do you see out there?"

"A tree," said some bright person.
"What color is the tree?" asked Smith.
"Well, it's got green leaves," said another student.
"And why aren't the grazers out there—the insects—eating all those leaves? What makes the leaves green?" asked Smith.
"Well, it's photosynthetic pigment," said someone.

That wasn't quite what Smith was after; he wanted them to explore ecological relationships between organisms. And so the conversation continued, back and forth. For Smith had been thinking about these ideas for quite some time: he was a deep thinker. So he looked out the window, and this argument of immense consequences followed linearly fromthat. What came to be known as the green world hypothesis was first articulated on that bright spring day in Smith's zoology class, and it fueled the fires of enthusiasm in the graduate students at the University of Michigan and rocked science.

The Architecture of a Trophic Cascade: Predator-Herbivore-Vegetation
When HSS formulated their hypothesis, the scientific community commonly accepted that bottom-up processes, mostly related to competition between species, were the primary forces shaping populations. The green world hypothesis provided an alternative view of population regulation driven by top predators, via carnivory. It enabled predation and grazing to play roles equally important to resources or habitat, with predation having the key role for some populations and resources having the dominant role in others.

HSS organized the food chain into three trophic levels. At the lowest trophic level they put producers (plants), with consumers (herbivores) at the next higher trophic level and predators at the top—as in Elton's food pyramid. Resources limit each trophic level. HSS noted interspecific competition among members of each level and concluded that because herbivores are seldom food limited, they appear to be most often predator limited and thus unlikely to compete for common resources. They offered the example of the vast carbon deposits that had accumulated globally as evidence that herbivores historically have been limited by predation, and they provided cases of the direct effect of predator removal on herbivore populations, such as the mule deer (Odocoileus hemionus) herd irruption on the Kaibab Plateau. This deer population, which lived on the northern rimof the Grand Canyon in Arizona, increased sharply almost immediately after extirpation of wolves (Canis lupus) and substantial removal of cougars (Puma concolor) in the 1920s, with estimates of deer numbers ranging from 4,000 in 1908 to 30,000 from 1923 through 1930. The powerful ramifications of HSS' hypothesis —that predators influence community dynamics at all trophic levels—inspired vital new hypotheses about community ecology.

Trophic cascades are based on HSS' three trophic levels (an odd number), in which a top predator consumes herbivores at the next lower level, and that in turn affects vegetation at the next level below. These cascades are essentially the indirect effects of predation. Direct effects occur via a predator killing prey, while indirect effects are mediated by a third species. An example would be the indirect effects of sea stars on vegetation in the rocky intertidal zone caused by changes in mussel density via predation. In some systems indirect effects of predation can also arise as a result of behavioral changes by prey in response to the threat of predation, which we will further explore in this chapter. In all cases, indirect and direct effects of predation interact to structure ecosystems.

Aldo Leopold was one of many who contributed to this more enlightened perspective, as was Elton. Although Elton and Leopold identified trophic cascades in primarily terrestrial systems, such as the Kaibab Plateau and northern Mexico, today the majority have been observed in aquatic systems. The sea otter (Enhydra lutris), sea urchin (Strongylocentrotus polyacanthus), and kelp (Laminaria spp. and Agarum cribrosum) cascade reported by marine ecologist James Estes in Alaska provides a classic example. When Vitus Bering explored the North Pacific in 1741 he found shorelines teeming with sea otters. By 1911 they were nearly extinct. Enough remained in sheltered rocky pockets along the coast that by the 1960s the otter population had recovered in places. Otters have a varied carnivorous diet and prey heavily on sea urchins. In areas without otters Estes found herbivorous sea urchins thickly carpeting the ocean floor—and no kelp. On reefs with otters he found a lush, green kelp forest and low numbers of sea urchins.

Top-Down versus Bottom-Up
Not everyone bought the green world hypothesis. William Murdoch's counterargument, termed the plant self-defense hypothesis by conservation biologist John Terborgh, suggests that food (bottom-up control) has the strongest influence, that the world may be green because not all plants are palatable to herbivores, and that predators are unnecessary for ecosystem regulation. Or, as evolutionary ecologist Stevan Arnold puts it, the world is green, but that doesn't mean it's edible.

Murdoch asserted that food shortage and plant defense strategies may be regulating herbivore numbers. He hypothesized that while plants are essential for the survival of the trophic levels above, the reverse is not true. He and other critics of the green world hypothesis, such as ecologists Donald Strong and Gary Polis, suggested that HSS' failure to address the full complexity of systems, which includes omnivory, weakened their hypothesis. Some have identified HSS' tri-level trophic model as a hypothetical construct because in the real world food webs are not so tidy and can have fewer or more than three levels. Still others, such as conservation biologist Michael Soulé, believe that top-down versus bottom-up, like all dualisms, is false, because the natural world is complex and bottom-up forces (nutrient flow) interact with top-down forces (the effects of predation). Other scientists, such as Rolf Peterson, concur; in the wolf–moose–balsam fir system he studies in Isle Royale National Park, it's never one or the other but a synergy of the two.

HSS' hypothesis was based on a terrestrial environment mostly involving insects and their predators, with the Kaibab mule deer irruption the sole example presented of a mammalian system. This provoked a strong rebuttal by wildlife ecologist Graeme Caughley, who suggested that because the factors that may have resulted in this irruption were "hopelessly confounded," a case study of the Kaibab provided an ineffective example of top-down control. These factors included herbivory by livestock and unreliable deer counts. HSS based evidence for top-down effects on things that control insects and on the connections between insects and plants. They observed that plants flourished in the absence of insects when insects were kept down by predation. Researchers working in various marine and mammalian systems subsequently tested the green world hypothesis. In all systems, whether pertaining to insects or mammals, they found a strong positive correlation between predator removal, plant community simplification, and reduced energy flow, which goes back to Paine's dictum about being able to change the world by removing one species.

Ecologists Lauri Oksanen and Stephen Fretwell proposed that if we treat trophic levels as units, systems having four or more trophic levels may have more than one level representing predation. In this scenario a predator at the fourth level will dominate a predator at the third level, and this will release the herbivore population at the second level frompredation, causing its numbers to increase. This in turn will cause overgrazing of plant communities at the first level. Trophic cascades in which a top predator controls its herbivore prey via top-down forces will therefore always have an odd number of trophic levels. Removal of the top trophic level in such systems will have a radical effect on lower levels, causing herbivore irruption and overconsumption of vegetation. In any food chain, energy flow alternates; in odd-linked systems, plants will be limited by resources available to them (top-down control); in even-linked systems, plants will be limited by grazers (bottom-up control). Thus systems with an odd number of levels will be green, while systems with an even number of levels will be brown or barren.

In the 1990s Estes saw a three-level trophic system flip into a four-level system in coastal Alaska. The apex carnivore Orcinus orca, the killer whale, began preying on sea otters, reducing their numbers significantly. Killer whales added a fourth level, reversing the flow of energy from top-down to bottom-up. The resulting cascade rippled through the North Pacific, causing an upsurge in sea urchin numbers and kelp consumption and denuding the ocean floor.

Researchers have studied how the green world hypothesis holds up under different levels of ecosystem productivity, called net primary production. In 1981 Oksanen and his colleagues developed the exploitation ecosystems hypothesis (EEH). In their model herbivory is greatest in relatively unproductive environments, with predation more important and the impact of herbivory reduced as ecosystem productivity increases. Deserts and arctic regions provide examples of systems where plant production is so scant that it can fail to support herbivores. At slightly higher productivity levels, such as on prairies and savannas, an ecosystem becomes able to support a consumer trophic level, with herbivory increasing as productivity increases. As productivity continues to increase, as in boreal forests, plant biomass (the total mass of living matter of a particular type) increases, herbivore biomass increases, and ecosystems become capable of sustaining a third trophic level—the predators—with this level controlling herbivores. Because it uses environmental productivity as a key driver of ecosystemdynamics, EEH incorporates both bottom-up and top-down forces.

The Keystone Species Concept
The keystone species concept lies at the heart of the HSS debate. When Robert Paine introduced it in 1969, he envisioned its mechanisms as a dominant predator consuming and controlling the abundance of a particular prey species and a prey species competing with other species in its trophic class and excluding them from the community. As long as one keeps these two processes in mind, the whole idea of keystone species fits into place—like the keystone of an arch. And when the keystone is removed, arches and ecosystems fall apart. This mechanism explains the ecological collapse Paine observed on Mukkaw Bay. One year after he began removing Pisaster, seven of the fifteen species he had inventoried at the start of his star-throwing experiment were gone, with the others declining rapidly. However, some scientists suggest that ecosystems may not exactly "collapse" when a keystone predator is removed. An ecosystem thus altered continues to function, albeit in a different manner, by moving into what is referred to in ecology as an alternative stable state.

The keystone concept has been applied to a variety of species, from large carnivores to herbivores and even plants. And it has created disagreement among ecologists, who have questioned this term's usefulness and generality. In response to mounting criticism of the overly broad use of this popular term, marine ecologist Bruce Menge defined a keystone species as "one of several predators in a community that alone determines most patterns of prey community structure, including distribution, abundance, composition, size, and diversity." Keystones selectively prey on dominant prey, have large body size relative to prey size, and are highly mobile, with a large foraging range. More recently Michael Soulé and his colleagues proposed the term strongly interacting species for those that have a strong ecological effect on communities. Next we will see how fear drives some of these effects.

The Ecology of Fear
Mid-May in Glacier National Park, Montana, is not a time or place for the fainthearted. The gunmetal sky was beginning to spit snow at me and my field technician, Dave Moskowitz, as we hurried to finish a track transect before a late season blizzard broke out. All day we had been hearing wolves howling. It had been difficult to pinpoint their location over the rising wind, but it seemed as if they were moving along a benchland one-tenth of a mile east of us. Hearing them like this provided a powerful reminder of the wildness of the system I was studying.

My work involved putting in fifty-seven miles of track transects in Glacier National Park's North Fork, arguably one of the most intact systems in the lower forty-eight states. This place harbored a full suite of large predators in one of the highest densities found south of Alaska, as well as abundant elk (Cervus elaphus) and deer (Odocoileus spp.). An expert tracker, Dave was writing a field guide to tracks of the Pacific Northwest. I was fortunate to have his help. Using track transects in which we were measuring all occurrences of elk, deer, moose (Alces alces), wolves, cougars, coyotes (Canis latrans), and bears (Ursus spp.), I was mapping their interactions and determining game densities in a system with so many predators. I was also measuring predation risk, to see whether wolves and other predators influence elk movements by causing them to avoid areas with escape impediments, such as downed wood and thick brush. In a 1999 article wildlife ecologist Joel Brown noted that the nonlethal effects of predators can be ecologically more important than the direct mortality they inflict. Evidence I'd found thus far created a compelling picture of the same sort of ecologically complete system Aldo Leopold had observed in 1936 in Mexico. I was also seeing patterns sharply etched on the landscape. These elk were spending more time on flat, open ground and less in riskier terrain, where they might have had more difficulty escaping predators.

As we raced to finish before the weather deteriorated further, we encountered a bachelor herd of a dozen skittish elk at a distance of fifty yards. They had shed their antlers and sported velvety antler buds. I gauged their ages from their body size and behavior. The younger bulls nervously ran around in circles. The mature bulls eyed us warily but stood their ground. Eventually they left, the young bulls taking cues from their elders, moving in the elegant head-high trot characteristic of their kind. As they vanished into the inky conifers below the benchland, I was left with an ominous feeling in the pit of my stomach. Something about their behavior made me uneasy, but I couldn't put words to it.

We finished just as the blizzard hit. When I looked back toward where we began the transect, I encountered a sight so unexpected, so shocking, that initially I thought I was imagining it. For there, conspicuous even through a scrim of falling snow, lay a fresh bull elk carcass. The animal looked impossibly huge. The wind carried clouds of steam from its gaping belly. I was amazed that we hadn't heard the takedown, even over the moaning wind. One minute the dead elk wasn't there and the next minute it was, having met death in an area I had moments earlier characterized as having very high predation risk.

I investigated the carcass—an old bull surrounded by crimson wolf tracks on snow, its flesh still warm beneath my hands. Already the wolves had removed most of its hindquarter meat and some organs. While Dave photographed it I looked up, my senses sharpened by the coppery scent of blood and this primal encounter with the ecology of fear. A fresh carcass soon draws cougars and bears; I lingered only sufficiently to record location coordinates.

The next morning a young grizzly fed on the carcass. I watched it patiently for an hour as it removed much of the remaining meat, made a vain attempt to haul the still heavy carcass up the bench, and then lumbered away to sleep off its meal. All the while a Steller's jay perched on a downed log just to the side of the carcass. It repeatedly tried to scavenge meat, each attempt met by a growl and swat from the grizzly. Later that day I observed a coyote trotting through the area, balancing a purloined elk leg in its mouth. Bears emerge from their dens in April, ravenously hungry. One of their survival strategies involves searching out wolf kills. In this park, as in Yellowstone National Park, many wolf kills end up usurped by grizzlies. Carcasses such as this one show how apex carnivore predation can support a wealth of species, from grizzlies to jays to coyotes.

Six months later, in mid-November, I worked in Waterton Lakes National Park. The wind blasted across the Alberta prairie, nearly knocking over my tripod and spotting scope. I steadied them with gloved hands and tucked my chin deeper into the collar of my down parka. It felt more like mid-January in this extreme landscape, where most years the only month I didn't experience snow at my field sites was July. I was out there using yet another method to determine whether elk fear wolves. Conservation biologist Joel Berger, who has done global research on the fear of predation, believes this phenomenon underscores more fundamental questions—the meaning of fear itself and how it can affect ecosystems.

My teenage daughter Bianca had joined me in the field. We were watching a herd of approximately four hundred elk cows, which stood on a high benchland on the southeastern border of the park. Most had their heads up, scanning the landscape rather than eating. They skittishly grazed on tawny dried remnants of prairie grasses that poked up through the thin snow, taking quick bites, looking up for some long moments before stealing another mouthful of food.

"What's up with the elk?" asked Bianca.
"You'd think there were wolves nearby," I said.
"You think?" she asked.

Soon the wolves would oblige us with an answer.

The ecology of fear has deep roots. Staying alive during the early Pleistocene epoch involved escaping large creatures with sharp teeth and claws. This meant that prey species evolved behavior driven by survival. Vigilance—time spent head up, looking for threats—is essential for survival in systems with top predators, but it comes at the expense of time spent eating. For the past two years I had been doing focal animal observations on elk. This involved watching one animal at a time, recording how long it spent with its head down feeding versus head up, scanning for predators. I had categorized my study sites as areas of high, medium, and low wolf presence. I wanted to know whether fear varied on the basis of wolf density and, from that, to learn how many wolves would be enough to trigger changes in herbivory patterns—a trophic cascade. Would one pack passing through an area occasionally, but not denning there, have the same effect as two very large packs that had produced multiple litters of pups in one year? How about one pack that kept hanging on despite losing half of its members annually as a result of human-caused mortality in an area where it was legal to shoot wolves outside the park? The elk I observed on that blustery November afternoon in my medium wolf density area were far more skittish than usual.

All at once five black shapes crested the bench, fluidly trotting through the elk. Wolves. We'd seen their tracks earlier that day, pressed into a skiff of snow on the Chief Mountain Highway. Few humans visited the park between late fall and spring; predators and other wildlife adapted to this by increasing their use of park roads that are closed to vehicle traffic in the off-season. Even through the blowing snow I could see that these wolves were muscular and well fed. They moved comfortably through the herd, shifting into a slow lope, tails high. The elk parted as the wolves passed, and then they regrouped a short distance away, heads up, bunched more tightly for safety. The wolves didn't stop but continued on their way, disappearing over an eskerine ridge. It was a long while before the elk settled and resumed feeding, and their vigilance level remained high for hours. Their earlier restlessness made perfect sense.

Aldo Leopold was among the first to observe the behavioral effects of lack of predation on his own land. In 1935 he bought an abandoned farm in southwesternWisconsin, to use as a hunting reserve. He subsequently dubbed the farm "the shack." This land, now known as the Leopold Memorial Reserve, lies 45 miles north of Madison, on the southern edge of Wisconsin's sand counties. The Leopold family spent every weekend there, restoring the land. Between 1939 and 1940, in his shack journals Leopold noted the effects of deer herbivory on herbaceous plants (plants whose leaves and stems die down to soil level at the end of the growing season) and trees he had planted on his land, which included oaks (Quercus spp.) and aspens (Populus tremuloides). He commented that some species were being nipped down to eighteen inches in height. In a game survey he also documented how humans had by then eliminated wolves from much of Wisconsin. Deer had exploded in northern Wisconsin, from several hundred in 1920 to at least 100,000, causing game managers to formally acknowledge the problem. Although things were not this bad at the shack, Leopold noted ongoing plant damage caused by deer, which in the absence of wolves calmly stood their ground and browsed young saplings down to nothing.

There is a saying that the more things change, the more they stay the same. And an older Romanian proverb from the Karpaten states that where wolves go around, the forest grows. Both pieces of folk wisdom came to mind when I visited the shack to see whether the current aspen growth pattern and deer behavior would fit Leopold's historical observations. It was mid-April as I drove along the rural road that runs through the Leopold Memorial Reserve, noting small herds of deer standing around with their heads down, eating shrubs. I spent the next few days examining the amount of browsing on aspen sprouts.

One day Aldo Leopold's daughter Nina Leopold Bradley joined me in the field. Her chocolate Labrador retriever, Maggie, ran glad circles around us as we examined the aspens around the shack for evidence of deer herbivory. Almost all the aspens below browse height (the height a deer can reach to eat) featured chisel-pointed ends where deer had bitten off the apical stem, the dominant growth bud. Many had zigzagging trunks, where they had been browsed and had healed, and then had grown in a different direction. I held my measuring rod to an aspen less than three feet tall and counted its browse wounds—eight in all, each marked by a crook in its trunk. I showed Nina how telltale signs on the aspen allowed us to distinguish browsing from disease, because the latter made the trees' growth tips atrophy. All aspens can sustain moderate browsing, but these bonsai aspens looked stunted and shrublike. With chronic herbivory they would eventually die. Indeed, we found many that had succumbed in this manner.

Nina and I reflected on how little things had changed at the shack since her dad's era. As we continued to walk she recalled his observations about the effect of wolf removal on deer behavior, and how deeply this awareness affected him. "My father always said it all had to do with relationships. But he couldn't convince managers of that. He was even unable to convince some of his best friends. He had found stacks of dead deer as big as a house in northern Wisconsin, and his colleagues would not vote for a doe-hunting season. I think we have the same problem today. I do not think that people, even at the highest level, quite understand the interdependence of all of these issues. We still have too many deer, we still have hunters thinking we don't have enough deer, and we still have no wolves here."

So how do these relationships work? Let's say you are a white-tailed deer foraging on the Leopold reserve. There are no large predators in the forest that could threaten you. So you feed steadily on shrubs and grasses, looking up only to interact with others of your kind or search for food. Now let's say you are a white-tailed deer foraging in the North Fork. You take a bite and look up, sacrificing food for safety, highly vigilant. You are living in a landscape of fear, where your ability to survive depends on your ability to detect and escape predators as well as obtain food. The resulting stealth and fear dynamics—and the relationship between top predators and their prey—have profound ecological implications.

Risky Business: Predation and Resource Selection
Predator-prey interactions have two components: predators killing prey and predators scaring prey. While the lethal effects of predation are well documented, nonlethal effects may have equally strong consequences. Joel Berger tested this by tossing snowballs imbued with predator scents, such as wolf urine and grizzly bear feces, at ungulates. In addition to pungent snowballs, he experimented with tape recordings of predator sounds (lion roars and wolf howls) and neutral sounds (water and monkeys). Where wolves had been absent for decades, such as in Rocky Mountain National Park, elk responded to the snowballs or predator sounds with some curiosity, but none became alarmed or ran. In Denali National Park and Preserve, where wolves had been present for many years, ungulates responded by becoming hypervigilant. Berger and colleagues continued this work on a circumpolar scale, working in Greenland and Siberia, where predators had long been present, and finding similar results with moose and caribou (Rangifer tarandus). Beyond individual responses, Berger wanted to know how prey animals acquire knowledge, how fear is transmitted, and how current behavior can help unravel the ambiguity of past extinctions and contribute to future conservation. Ultimately his work will help shed light on how predators shape prey behavior and landscapes.

Research about predation risk has the potential to inform human choices about which landscapes can be allowed to harbor dangerous animals. Berger and colleagues found that in Wyoming moose increased vigilance behavior in the presence of grizzly bears, keeping their heads up longer and staying on the move to avoid predation. This reduced browsing on willows, enabling the willows to flourish, thus improving habitat for songbirds and increasing biodiversity. Awareness of these landscape-scale effects can be used to make management decisions about grizzly bears, perhaps allowing them to expand their ranges.

Predation is the main driver of fear in prey because it can lead to death. Fear of predation involves a response to predation risk, whereby prey react to predator presence—or even to the mere threat of it. Fear causes the adrenal glands to secrete adrenaline, a short-acting substance that prepares the muscles and brain for flight. It also produces cortisol as part of an animal's long-term response to chronic stress. An elk uses all its senses to evaluate the threat of predation. Its ability to assess and control its risk of being preyed upon strongly influences habitat selection and feeding decisions. Prey animals establish an optimal baseline level of vigilance in the absence of direct evidence of predator presence. Individuals who successfully balance the benefits of risk avoidance against energy costs (missed opportunities to eat) have a greater chance of survival.

This response is not limited to large mammals. Working with animals at the opposite end of the size spectrum, Oswald Schmitz found a behavioral trophic cascade consisting of spiders and grasshoppers. The top predator he studied, the nursery web spider Pisaurina mira, preferentially preys on the grasshopper Melanoplus femurrubrum. In the absence of spiders, grasshoppers selected a diet composed almost entirely of grass rather than forbs (flowering plants that are not grasses). In this famed experiment Schmitz glued spiders' mouths shut to render them unable to prey on grasshoppers. In the presence of spiders with glued mouths, grasshoppers nevertheless reduced their feeding time and preferentially ate forbs, which provide greater cover and safety from predation. This shift resulted in a trophic cascade.

Behavioral adaptations are complex and variable and show an evolutionary relationship to landscapes. In large mammals, behavior that evolved over thousands of years underlies trophic cascades mechanisms. Elk originated in Asia, on high grassland steppes. They colonized North America about 10,000 years ago, crossing the Bering land bridge. Lacking competition from other elk species in North America, they spread widely across many habitat types, from Pacific Northwest rain forests to sagebrush deserts. Long-legged cursors, elk run with their heads up and a straight-legged gait (as opposed to bounding). They escape predators via rapid and sustained flight, an adaptation found in ungulates from open plains with low flight impediments. On landscapes with both open and closed habitat structure, they may use a combined strategy of hiding in forest cover to lower predator encounter rates and seeking open terrain, such as grasslands, where predation risk may be reduced.

Recent studies have examined factors that can render prey more vulnerable, such as differences in ungulate grouping behavior and terrain features. In Banff National Park, landscape ecologist Mark Hebblewhite found predation risk lowest in small groups of elk, with groups larger than twenty-five having the highest probability of being preyed upon by wolves, possibly because they are more likely to contain weak or sick individuals and are easier for predators to detect.

In Yellowstone, ecologist William Ripple developed his predation risk hypothesis while sitting on a high terrace in the Lamar Valley, where he noticed patchy willow growth. Out in the open, willows were browsed down, but where there were visual or terrain obstacles, willows flourished. He and a student, Joshua Halofsky, proceeded to measure elk behavior and found heightened vigilance in areas with escape impediments. Elk spent more time with their heads up, scanning for predators, in these areas, behaving more skittishly than when they were assured of a clear escape route.

According to Douglas Smith, leader of the Yellowstone Gray Wolf Restoration Project, the concept of predation risk eludes easy definition. For example, an area where wolves take down prey after a long chase may not necessarily be the site of greatest predation risk. That may actually be the site where prey first encounter predators. Additionally, Smith notes that most wolf kills occur between dusk and dawn. Because elk and other ungulates have poor vision, obstacles to their viewshed may not play a significant role in the dynamics of predation risk. 30 Presence of other predator species complicates matters. Wildlife biologist Kyran Kunkel found that avoidance of one species, such as the cougar, which hunts by stealth, makes prey more vulnerable to another, such as the wolf, which runs down its prey.

Elk have a sophisticated response to predation risk that includes gathering in larger groups in open areas. Landscape ecologist Matthew Kauffman and colleagues found that open areas enable wolves to detect prey more easily and thus present greater predation risk. Wildlife ecologists Stewart Liley and Scott Creel found that elk adjust their vigilance in response to the size of their group and the type of immediate threat they face from wolf presence, with environmental variables such as obstacles having a secondary influence on vigilance.
Some researchers recommend that trophic cascades studies incorporate radio-collar data to measure behavioral predation risk (i.e., wolf presence). According to Smith, the complexity of these interactions merits deeper investigation.

I ended up putting in 150 miles of track transects in the Crown of the Continent Ecosystem and doing 700 focal animal observations. In doing this work I found compelling evidence of a trophic cascade. Where wolf density was high, elk avoided areas with debris and other escape impediments. Most carcasses and the greatest amount of wolf sign, such as tracks and scat, occurred in thick forests, debris, ravines, and riverbanks, which I had characterized as high predation risk sites. My focal animal observations suggested that the more wolves there are in a landscape, the more wary elk become. This response may be triggering cascading effects in this ecosystem, enabling aspens to grow above browse height. Indeed, the ecology of fear may be behind the changes at my home, where shrubs and trees have reclaimed the meadow after wolves returned and deer, to stay alive, have had to act more like deer and less like livestock. These pervasive effects influence even small, nonkeystone predators, as we shall see.

Mesopredator Release
It was late May in Waterton Lakes National Park and the snow had just melted, the matted grasses still a winter-killed brown. My field crew and I were putting transects into an expansive rolling grassland dotted with aspen stands. My field technician Blake Lowrey was on point that day, doing dead reckoning with a compass and pulling the transect tape due east through a copse of stunted aspens, making good progress. All at once he stopped and said, "What's that smell?" I had cautioned everyone to be careful around carcasses because of bears. At the head of the transect line I found a dead coyote on a well-used game trail. This relatively fresh coyote carcass had been there for maybe one or two days. It lay on its back, limbs outspread and neck outstretched. Its throat had been ripped out and it had been eviscerated. No other flesh had been removed. All around it lay evidence of the perpetrator of this carnage: wolf feces and tracks. The coyote appeared to have been a young adult male in relatively good health that had perished because it had had the misfortune to come upon a wolf. In most systems wolves make it their business to kill coyotes. This particular carcass had been left on a primary game trail as a grisly marker and warning to other coyotes that wolves rule this system—they are the apex predator.

Two weeks later, in Glacier National Park, we found another coyote in the same position, also on a game trail, its throat and guts ripped out, no other flesh missing, wolf scats and tracks all around. By now my crew had become sufficiently accustomed to carcasses to find this fascinating. We took a break and discussed the possible pattern here—the powerful signature wolves were leaving on this landscape.

Wolf-coyote enmity is not new. Wolves recolonized Isle Royale National Park in the early 1950s. Within two years they had wiped out the resident coyote population. When wolves were functionally extirpated from Yellowstone, wildlife biologist Adolph Murie noted a corresponding steep rise in coyote numbers, which began to formlarger packs and hunt deer. Smith considers what happened next to Yellowstone's coyotes one of the best stories to emerge from the mid-1990s wolf reintroduction. After the wolf 's return, coyote numbers dropped by as much as 50 percent overall and by 90 percent in core wolf pack territories. To survive, they formed smaller groups and spent more time in the interstices between wolf territories and nearby roads. There Smith found coyotes killed by wolves in a similar manner as I'd observed. Most of the pre-wolf coyote population had occurred in packs. Since wolves, half the coyote population has consisted of what Smith calls "floaters," unaffiliated coyotes with higher survival rates. Breeding coyotes have the highest mortality because they are easiest for wolves to find and kill, since their behavior is more predictable and they live in territories.

One of the most powerful indirect effects of predation involves mesopredator release. Defined as medium-sized predators, mesopredators are controlled by top predators—often by direct mortality, as we have seen, but also via competition for shared resources. Humans commonly remove keystone species to protect economically valuable big game from predation. For example, upon removal of the wolf fromthe Endangered Species List in the northern Rockies, the state of Idaho proposed to eliminate 40 percent or more of its wolves to help create more elk for humans to hunt. This type of action causes mesopredators, such as coyotes, to increase and puts abnormal pressure on smaller species, such as game birds, which decline and can become extinct.

In the mid-1980s David Wilcove investigated the effects of human land use on songbirds. He studied small woodlots in rural and suburban sites in Maryland and larger forest patches in Tennessee in Great Smoky Mountains National Park. The biggest tract of virgin forest in the eastern United States, this park retained forest-dwelling mammals and birds long extinct in central Maryland. Wilcove wanted to test the effect of mesopredator release on songbird nest predation—what can be thought of as the empty nest hypothesis. To do this he filled experimental wicker nests with Japanese quail (Coturnix japonica) eggs and placed them in forest locations ranging from the midcanopy to the understory, to reflect native birds' nesting habits. One week later he measured the percentage of experimental nests raided by mesopredators. He found higher rates of nest predation in small woodlots near human communities because these areas had higher populations of raccoons (Procyon lotor) and squirrels (Sciurus spp.) and few, if any, large predators, such as cougars and bobcats (Lynx rufus). This led Wilcove to link nest predation to mesopredator release.

As Wilcove was finishing his work, on the opposite side of the country Michael Soulé began to study how coyote decline affected the ecology of Southern California's chaparral canyons. He lived north of San Diego, a few miles fromthe coast, in an area about to be developed, which would fragment wildlife habitat. Soulé spoke with the developers and suggested preserving some open lands and maintaining connectivity corridors between them. The developers wouldn't listen, so he got mad and began a research project in the chaparral, also called coastal sage scrub habitat.

Soulé and his students investigated what happens to birds in isolated patches of habitat. This landscape-scale project involved surveying thirty sites for the presence of seven species of chaparral-dependent birds, beginning in a two-acre habitat patch created forty years earlier by development. He recalls sitting in that patch for an hour, listening and looking for birds, and finding it silent except for mockingbirds (Mimus polyglottos) and jays (Aphelocoma insularis)—generalist species that do not need chaparral. As he noted the absence of quail (Coturnix sp.), roadrunners (Geococcyx californianus), sage thrashers (Oreoscoptes montanus), and other chaparral-dependent species, he realized that there had been significant extinction in some of these old patches.

Among the variables Soulé used to analyze his data were coyote presence or absence. Coyotes act as keystone predators in some systems and mesopredators in others. In Yellowstone, where wolves are keystones, coyotes take the mesopredator role. In Southern California, where there are no wolves and probably never were any, coyotes take the keystone role. Soulé hypothesized that falling coyote numbers in an area being developed by humans would result in the release of native and exotic mesopredators such as raccoons and housecats.

His analysis showed that patches with coyotes contained more chaparral-requiring birds and, on the other hand, patches without coyotes had fewer such birds. But his epiphany came when he realized that this was actually a cat effect. He owned cats and knew they had trouble surviving around coyotes because of predation. So what he was observing with the birds was the indirect effect of coyote absence (more cats) and thus a mesopredator release.

Ten years later Soulé's student Kevin Crooks deepened this study by radiocollaring cats. His data very graphically demonstrated the relationship between lack of coyotes, increased cat movements, and reduced populations of chaparral-dependent birds. Soulé's mesopredator study had been based on a statistical correlation showing the trophic cascade coyotes were producing, but Crooks confirmed this empirically using radio-collars. Accordingly, Crooks and Soulé concluded that coyotes were ecologically beneficial because they controlled mesopredators that preyed on birds while rarely preying on birds themselves.

Since this study, mesopredator release has been identified in the Dakotas, where coyote absence caused the red fox (Vulpes vulpes) population to surge, making survival far more challenging for prairie ducks. Similarly in Texas, coyote removal led to an increase in five species of mesopredators and a decrease in game birds. In these cases mesopredator release reduced biodiversity and demonstrated the ecological importance of the alternative food web pathways created by keystones. These relationships raised scientific awareness of what may be at stake ecologically when we lose a keystone.

Remembrance of Things Past: Megafaunal Extinctions
We live in a world of losses. Cave paintings dating back to the upper Paleolithic period, between 30,000 and 10,000 years ago, depict creatures long gone from the northern latitudes—mammoths, horses, cave bears, rhinoceroses (Rhinocerotidae family), and saber-toothed tigers. Today we are drawn to these haunting images of a lost world. What exists today in terms of large fauna is but half of what once was.

A guild is a group of species that have the same role and coexist in an ecosystem. For example, wolves, cougars, and grizzly bears form the predator guild. Predator guild extinctions have far-reaching associations that go back beyond near time into the distant past. These effects invite us to take a closer look at a lost world shrouded in mystery.

Most large mammals, termed megafauna (animals weighing more than 100 pounds), became extinct in the past 50,000 years, during the late Pleistocene epoch, when Homo sapiens colonized the earth. We have lost more than 150 genera of megafauna in this time span. Paleontologist Paul Martin and others believe that spear-wielding humans brought this time of mammalian giants to an end. But some scientists, such as R. Dale Guthrie and Aldo Leopold's youngest daughter, Estella Leopold, a renowned palynologist, believe that climate change and disturbance also may have had a major influence on these extinctions. To illustrate the role of climate on extinction, Estella recounted her experiences while doing research in the Mojave Desert:

"We were in Death Valley, in the third year of our study, measuring the impacts of drought. It was a barren landscape, with little growing and almost no wildlife. And then it started to rain during the fourth year, and everything recovered. Creatures we thought were gone returned, like quail, tortoises, snakes, and lizards. And then there was another drought, and the plants and animals disappeared again. I had been studying the paleosediments of these valleys, using drill cores, doing pollen analysis, and finding out that these kind of jerks went on down through time two million years. The bottom of the core, which is bedrock, is a long section of nothing, and then you begin to see the plants of the Sierra Nevada showing up, and then you begin to see plants disappearing and reappearing. The picture you get from these cores is the same as you get if you sit on the edge of the lake today, after one of these droughts, the intermittent wobbles. You see birds coming and going, flowers coming and going. This has been happening for nearly two million years. It very dramatically tells you about ecosystems and their evolution."

While understanding causes and effects of megafaunal extinctions will require a deeper knowledge of climate change and prehistoric human land use chronologies, everyone agrees that they had major impacts on plant communities. Imagine a world with five times as many herbivores as we have today and twice as many large predators. In North America extinct megafauna include a giant armadillo, three species of giant ground sloth, the Columbian mammoth, a mastodon, a giant tortoise, a camel, a saber-toothed cat, an American lion, and a woodland musk ox. With the exception of Africa, megafaunal extinctions occurred globally, leaving ecologists to wonder what it means when prairies grow silent in the absence of roaring lions and thundering bison hooves.

Although we will never know the impact this had on Pleistocene vegetation, we know from the few places where echoes of such assemblages still exist that these effects can be tremendous. In Africa, elephants (Loxodonta africana) mainly browse woody species, changing the landscape by pushing over, breaking, or uprooting trees, creating openings in the forest and helping maintain grasslands. Similarly, Hippopotamus amphibius, which primarily feeds on grasses, transforms tall grasslands into a mosaic of short and tall patches that support a rich variety of species. Pollen and fossil records suggest that grazing megaherbivores, in the form of mammoths and ground sloths, together with browsers such as mastodons, may have helped maintain the open parkland vegetation that covered much of North America. Conservation biologists have been profoundly struck by the implications of Paul Martin's work. Loss of large bodied terrestrial and aquatic fauna may have had a huge impact on systems and suggests many trophic cascades that formerly arose fromtop predators have vanished.

The ultimate keystone predator, humans alter their environments by eliminating species and modifying ecosystem structure and function, thereby contributing to extinction, altering evolution itself. Megafaunal extinctions have had enduring effects, simplifying ecosystems and eliminating large predators such as the dire wolf and saber-toothed tiger, large herbivores such as elephants and giant sloths, and the suite of large scavengers supported by these predator prey interactions.

Ecosystems have been truncated or decapitated by the loss of larger animals. Beyond evolutionary entanglements, when one views these extinctions through trophic cascades glasses, the profound ecological wreckage humans have inadvertently wrought on this planet begins to become apparent.

What does megafaunal extinction mean in terms of the distribution and characteristics of existing plant communities? It's about pattern and process. All the green growing things on this earth evolved over millennia, adapting in response to changing environmental conditions. Systems became simpler toward the end of the Pleistocene because of the loss of megaherbivores. The remaining top predators gained a more powerful keystone role, and plant distribution again adapted to new herbivory patterns.

Species loss has continued into the present day, driven by a variety of factors, including climate change and overharvest by humans. Andrea Laliberte and William Ripple found that some North American carnivores have experienced range contractions to less than 20 percent of their early eighteenth-century range. This has resulted in irruptions of ungulates unchecked by predation, followed in many systems by reduced diversity of plants and wildlife such as songbirds.

Landscapes evolve in time and space. These mechanisms leave patterns in a landscape. The elk I observed on that eskerine benchland in Waterton, heads high, taking cautious, quick bites of grasses and aspens and then looking up to scan for wolves, are shaping the way things grow, creating patterns with their behavior. Landscapes tell stories, if we listen. The trophic cascades mechanisms that underlie this behavior can teach us much of what we need to know in order to conserve biodiversity and create more resilient terrestrial and aquatic ecosystems.

From The Wolf's Tooth: Keystone Predators, Trophic Cascades, and Biodiversity by Cristina Eisenberg. © 2010 Cristina Eisenberg. Reproduced by permission of Island Press, Washington, D.C.