Effects of Grizzly Bear Predation on

Ungulate Populations

Grizzly bears in the northern Rocky Mountains eat lots of meat obtained from large herbivores, notably elk, bison, deer, and moose, principally during spring and fall in drier colder regions. Most of this meat is consumed by scavenging carrion, but at times and in places much is also obtained by outright predation, principally on elk and moose, including both calves and adults. When reckoned in sheer numbers of killed animals, though, predation is directed principally at calves during their first months of life. 

Given the prevalence of grizzly bear predation on calves, it is reasonable to expect that elk and moose populations would be affected. But any direct assessment of population-level impacts is beset with logistical, analytical, and political difficulties. There are few issues more controversial and beggared by complexity than the impacts of predation by large carnivores on the large herbivores that hunters love to shoot. As much as the linear-thinking point-and-shoot types believe that the calculus is zero-sum, it isn't. One calf killed by a bear does not directly translate into one less antler-heavy bull for them to euphamistically harvest.


There is the tricky bit about whether predation is additive or compensatory--whether the animal that was killed would have otherwise survived during the subsequent year(s), or whether it would have died regardless from some other cause such as disease or starvation. There is also the issue of "reproductive value," which is another way of saying that not all animals have equal per capita effects on population growth rates primarily because of differences in sex, age, reproductive potential, and survival rates. There are also the hard-to-account-for effects of hunter harvest, summer forage conditions, winter severity, wildfires, disease, and intraspecific competition that confound any simple reckoning of predation effects.

But this complexity does not obviate the need to effectively address it if a person is to isolate the effects of grizzly bear predation on population dynamics of elk and moose. Fortunately, ample data are available of relevance to understanding population dynamics of elk in the northern US Rocky Mountains, and numerous researchers have deployed all sorts of statistical gymcrackery to crunch these data in efforts to illuminate the comparative importance of different drivers. Plus, these elk-specific efforts are augmented by a comparable corpus of work with moose, but primarily in farther-afield boreal and subarctic Alaska and Yukon.

What follows is my attempt to synthesize what we know about elk population dynamics in the northern US Rocky Mountains as necessary context for, then, illuminating the effects of grizzly bear predation...

Short Sweet Summary

The bar graphs at left summarize the current state of knowledge regarding the magnitude of various effects on various rates that either drive or directly manifest growth of elk populations in our region. Researchers typically express magnitude in terms of coefficients (called beta terms) produced by various models, with a larger beta denoting a greater responsiveness of the modeled rate to changes in the putative driver (or explanatory variable), with plus or minus signs denoting either a positive or negative effect. In the graph at left, any bar that extends below the central horizontal line corresponding to "0" represents a negative effect, and any bar above, a positive effect. The larger or longer the bar, the greater the effect.

John Vecutich and co-authors directly modeled rate of change in population size for elk on Yellowstone's northern range, thus integrating but also subsuming all constituent demographic building blocks. His results are represented by bars with the darkest hue. Kelly Proffitt and her co-authors modeled several phenomena relevant to the culminating rate at which elk were recruited to adulthood, including pregnancy rates, survival of calves to 7 months of age (using an index), and survival of calves through their first year--also for northern range elk. Finally, Jedediah Brodie and a laundry list of co-authors modeled effects of various factors on survival of cow elk using data from multiple studies across the northwestern US, including several from the northern Rockies.

In short, regulated (some might say, dysregulated) hunting by humans had by far the greatest effect on population growth rate (negative) followed by summer growing conditions and resulting forage quality and abundance (positive). Research by Arthur Middleton on the east side of the Yellowstone ecosystem affirmed this substantial effect of summer forage, mediated principally through the effects of female body fat on pregnancy rates (see the figure at right), although Kelly estimated that this effect was of a lesser magnitude. The perhaps intuitively obvious negative effect of winter severity on multiple demographic rates was also substantiated, primarily mediated through survival of older calves and adult females. A major effect of winter severity was also documented in an analysis by Scott and Michael Creel using annual estimates of population size for elk herds throughout western Montana. Insofar as predators are concerned, wolf predation apparently had a major effect on survival of calves to 1 year of age and a minor effect on survival of cows, whereas bear predation had a substantial effect on survival of calves to 7 months of age. That being said, John did not detect any signal of significance in his analysis of net change in population size that could be attributed to predation by large carnivores. So...predation may not be that big a deal, especially compared to human harvest and weather.


But there is more to the story...

Wolves versus People

The dominant effect of human harvest on elk population dynamics deserves to be unpacked a bit, especially in contrast to the lesser effect of predation by large carnivores such as wolves. The graph at left is from an article published by Dan MacNulty in Yellowstone Science, again with a focus on the northern range elk of Yellowstone. This graph is relevant because it shows how kills of cow elk by humans and wolves break out relative to age of the victim; also how pregnancy rates vary with age. All of this is relevant to assessing the relative "reproductive value" of elk characteristically killed by humans versus predators operating without high-powered firearms. Clearly, people kill proportionately far more young elk compared to wolves, which is what you would expect. Older elk are probably a lot easier to catch and kill if you depend on four paws for getting around and your bare teeth to kill things. The upshot is, when you put together information on the prospective future contribution of females to population growth, cow elk killed by people are roughly 1.6 times more valuable than the more senescent cow elk killed wolves, meaning that, per capita, human harvest (of females) will have a greater effect on growth of elk populations. 

More about People

The graph at right shows the characteristic magnitude and seasonality of cow elk kills by people (in gray, labeled sport-hunting) versus other causes ("background natural," in burgundy), again thanks to Jedediah Brodie. The top graph, A, shows a scenario where the only large predators are people, in which the bulk of elk mortality is from sport hunting concentrated during the fall. By contrast, the bottom graph, B, shows a scenario with 4 largish predators (cougars, wolves, and black and grizzly bears) plus people. Predators such as wolves and cougars kill elk essentially year-round, which results in a larger fraction of the total occurring during times of year other than fall. In this scenario, sport hunting is also considerably diminished to accommodate the effects of natural predation--which some hunters might consider to be justification for slaughtering large carnivores they see as competitors. Perhaps, if the only purpose being served is producing a harvestable surplus of elk for white guys with an ego deficit to kill. 

Importance of Calves versus Adult Females

It has become close to a truism that, per capita, the death of an adult female elk will have more effect on population growth than will the death of a female calf. Intuitively, once a calf has run the gauntlet of youth, where chances of survival can be low as 15-20%, as an adult with an 80-90%+ annual chance of survival it will make a much greater contribution to population growth through the production of calves. This notion is statistically bolstered through analyses of elasticity in growth rates--basically, the comparative extent to which population growth changes with a unit change in any given vital rate. These sorts of analyses consistently give prime age cow elk the highest value.

But an analysis published by Jarod Raithel added a twist--the nub of which is illustrated by the figures at left, where each dot is the result of a simulation which varied vital rates fed into a matrix model that produced a resulting rate of total population change (denoted by lambda). Relations between calf survival rates and population growth are shown in A; between prime-age cow survival and population growth in B. The slopes (beta-coefficients) of the regression lines affirm the results of elasticity analyses in that there is nearly twice the rate of change in population growth for any unit change in cow survival compared to calf survival. Yet calf survival explained roughly 75% of the variation in population growth compared to only 16% for prime-aged cow survival . In addition, the absolute magnitude of population growth was equal for the natural range of variation in calf and prime-age cow survival rates, primarily because calf rates were potentially 3-4 times more variable.

The upshot? Even though the per capita loss of a prime-age cow is more consequential than that of a calf (i.e., higher elasticities of population growth), population growth rate more closely tracks change in survival of calves, with potentially equal overall effect, primarily because elk calf survival can vary so much, from less than 10% to over 70%. And, in fact, we have seen substantial changes in calf survival rates in the Yellowstone ecosystem, primarily in response to changes in the rates of predation by bears (see below). Which is another way of saying that, despite all of what I presented immediately above, there is a good chance that bear and wolf predation on calves could be driving a significant part of the decline of elk populations that we've seen in especially the Yellowstone ecosystem, consistent with the extent to which Kelly Proffitt found predation to be a driver of calf survival to both 7 and 12 months of age.

For example, Kathy Griffin and her co-authors found that the effects of predation on calf survival varied not only with the number of predators in the environment, but also with summer forage conditions (i.e., May precipitation), with these two effects interacting. I provide more on the effects of adding more predator species below, but the graph at left shows the interaction between forage conditions and the predator suite. Basically, as you increase numbers of predator species, calf survival declines (as you would expect), but this decline is less in more productive environments, as indicated by more May rainfall. The explanation? Basically, with greater productivity, elk can be more dispersed and less predictable in their location, and thus less vulnerable to predators such as bears.

Parenthetically, Proffitt offered detailed information on how the various rates she modeled varied with the magnitude of key explanatory variables. This information is show at right in the form of x-y diagrams, with rates on the y axis and explanatory variables on the x.

Both the probability that a cow would be pregnant and that its calf would survive to 7 months of age (basically, mid-May to December) were governed by the cow's age and, in both instances, probability reached a peak for females that were roughly 7-15 years of age (A, at right). This mid life could be considered the age of both maximum competence and fecundity for cow elk, during which they were not only most likely to produce a calf, but also best able to tend it.

The middle graph (B) shows the monotonic relations associated with weather, including the strong negative effect of winter severity on calf survival and the weaker positive effect of summer precipitation (i.e., summer forage) on likelihood of pregnancy, presumably mediated by body fat (see above). Finally, the bottom graph (C) shows the monotonic effects of predator abundance on calf survival. Grizzly bears have a somewhat weaker effect on survival to 7 months compared to the effect of wolf predation on calf survival to 1 year. But, then, there are complexities...

More Details on Driver of Elk Demography

Predation on Elk Calves

Some more of Kathy Griffin's results are shown at left, along with results of research on elk calf survival from Yellowstone's northern range by Shannon Barber-Meyer. Kathy's analysis of elk calf survival used data from 12 populations in the northwestern US, which was how she could tease out the effects of different predator communities. Shannon's research spanned 2003-2005. All of these graphs express a measure of elk calf mortality on the y axis and time since birth on the x--essentially calf age in days.

The top graph shows the total number of calf mortalities recorded by Shannon as a function of age. The mean age of death, at least during the first 100 days of life, was 21 days, the median age of death closer to 8. The not surprising point is that calves die disproportionately more often when they are the most vulnerable, encompassing the time period when calves rely primarily on hiding and immobility for protection rather than flight. And, as it turns out, predation accounts for roughly 45-94% of all elk calf deaths during the first 100 days of life...depending on the study area and time period.

The bottom three graphs (B-D), show the rate of calf deaths caused by different predators (bears, cougars, and wolves) as a function of age, but with each of these rates differentiated by whether they occur in areas with 3, 4, or 5 large predators. A 3-predator community consists of coyotes, cougars, and black bears; a 4-predator community of the previous three species plus wolves; and a 5-predator community, of the previous four, plus grizzly bears. In essence, results for a 5-predator community are tantamount to the direct and indirect effects of having grizzly bears in the mix.

One key point is that bears kill calves at a consistently much higher rate during their first 100 days than do any other predators. Notice the difference in scaling of the y-axis for the various predators, from a daily maximum rate near 0.020 for bears versus a daily maximum rate nearer 0.003 for cougars and wolves. Maximum rates of bear predation are 6-7-fold greater, consistent with the fact that bears have accounted for between 52 and 85% of all calf deaths in the Yellowstone ecosystem attributable to predation. And notice also that bear-caused deaths are concentrated in the first 20-30 days of life, at a mean age of 10 days. By contrast, cougar- and wolf-caused deaths are dispersed with respect to calf age, averaging 35 days of age for those that are wolf-caused and nearer 100 days for those that are cougar-caused. This difference suggests that bears may be more effective at ferreting out hiding calves compared to cougars and wolves.

The other wrinkle has to do with changes in the magnitude and timing of death rates with the addition of wolves to the system and then, on top of that, grizzlies (i.e., changes from a 3- to a 4- to a 5-predator system). Bear-caused deaths jump dramatically with the addition of grizzlies--implying that this species tends to be a more aggressive and effective predator compared to black bears. By contrast, rates of cougar predation drop with the addition of wolves and then grizzlies. This pattern suggests that cougars are essentially crowded out of the field as a calf predator when wolves and grizzly bears are around. Cougars are easily displaced from kills by both wolves and grizzlies, and vulnerable to ending up as prey themselves if they stand their ground. In the case of wolves, peak predation on calves is shifted to younger calf ages in the presence of grizzlies--for reasons I and others can only guess at.

One additional aspect of all this is similarities between bear predation on elk and moose. There has been comparatively little insightful research on moose demography during recent decades in the northern US Rockies. By contrast, there has been a long and sustained research focus on moose in areas with grizzly bears in Alaska and the Yukon. Aggregating the results of 7 different studies, the 608 moose calves monitored in these areas were killed at a 63% rate during their first 100 days of life. The mean rate attributable solely to bear predation was 45%, which was 72% of the total attributable to all species of predators. The figures for elk calves in the Yellowstone ecosystem, at least during the last 20 years, are a remarkably similar 64%: 40%: 66%. The point of this is that bear predation probably affects moose populations in the northern US Rockies in similar ways and to a similar extent as it affects elk populations.

Predation on Moose Calves

Changes in Bear Predation on Calves

Given that bears cause much, if not most, elk calf mortality during the calves' first 4-7 months of life, it is noteworthy that rates of bear predation have increased dramatically in the Yellowstone ecosystem during the last 25 years or so, with much of that increase concentrated between the early 1990s and early 2000s.

The bar graph at left is illustrative. The total bar height corresponds to the total elk calf mortality rate documented in four different studies concentrated during two different periods of time--farthest left, in the Jackson, Wyoming, area are during the early 1990s; next farthest left, on the northern range during the late 1980s; next to the right, in the Gallatin drainage during 2005; and, farthest right, in the northern range again, but during 2003-2005. The burgundy plus brown portion of each bar denotes the portion of total mortality caused by predation and of that, in brown, the portion caused by bears.

Elk calf mortality rates have increased, but the most striking change has been the dramatic upsurge in the portion caused by predation, largely driven by increases attributable to bears. In fact, raw rates of bear-caused elk calf mortality have jumped by around 3.5-fold between c. 1990 and c. 2005. As Arthur Middleton and Jennifer Fortin first speculated and as I elaborate elsewhere, these non-trivial increases in bear predation almost certainly reflect bears eating more meat from large herbivores to compensate for losses of whitebark pine and cutthroat trout

What to make of all this? My attempt at a graphic distillation is at left. There is little doubt that human-caused mortality, more than any other factor, governs the fates and trajectories of elk populations, especially when managers unleash hunters on prime-aged cow elk. Weather also has a substantial effect: growing season moisture primarily through effects on forage quality and quantity and, through that, pregnancy rates; winter severity primarily through effects on calf survival and, during exceptional years, even survival of adult females. Predation also has its direct and indirect effects--bears primarily through predation on calves during their first 1-month of life; wolves primarily through effects on survival of calves during their first 12 months and, secondarily, through predation on cow elk.

Non-human predators, among them grizzly bears, no doubt affect trajectories of elk populations. The magnitude of this effect seems to vary with grizzly bear densities and dietary alternatives. The less of other foods, the more bears are likely to prey on elk calves--with greater consequent effects on size of elk populations.

Many hunters and state wildlife managers would assume from this that populations of predators--including grizzly bears--need to be "controlled" sufficient to generate the surpluses of adult elk that hunters have come to assume is their right and privilege. This attitude has been amply and unambiguously demonstrated in state-level management of wolves, cougars, and black bears. The metric for management often ends up being inflated elk population goals that are artifacts of predator-depauperate environments predating restoration of wolf and grizzly bear populations. Rarely is any allocation of the putative harvestable surplus made for predators, and this despite the fact that most people in the United States would support such a move. I won't elaborate my critique of state-level wildlife management here other than to refer you to these treatises on the despotic nature of this institution and the problematic ethos it embodies. Suffice to say, much more could be said about how institutional pathologies translate into problematic policies for managing predators.