Jump to ContentJump to Main Navigation
Yellowstone's Destabilized EcosystemElk Effects, Science, and Policy Conflict$

Print publication date: 2006

Print ISBN-13: 9780195148213

Published to Oxford Scholarship Online: September 2007

DOI: 10.1093/acprof:oso/9780195148213.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 15 November 2019



(p.280) 15 Synthesis
Yellowstone's Destabilized Ecosystem

Frederic H. Wagner

Oxford University Press

Abstract and Keywords

Northern-range elk numbers have changed through four successive population phases over time: low numbers in prehistory up to ~1883; censuses of >6,000 up to 20,000-35,000 from 1884-1958; <6,000 from 1959-1970 during park reductions; and >6,000 from 1971 to the low 2,000s after reductions stopped. Varying sources of evidence indicate alternating responses to each of these phases by all measured components of the northern-range biota. The pre-1872 northern-range ecosystem can be characterized by a hierarchical model constrained at the top by human and large-carnivore predators that limited elk numbers, prevented heavy herbivory, and allowed development of a vegetation that provided ample food for other herbivores, habitat for other faunal components, and constraints on hydrology. Removing the predatory constraints allowed elk increase that altered the structure of the system and progressively reduced biodiversity. These changes have disproved the ecological-effect tenets of the natural-regulation hypothesis. Expansion of the northern range since 1988 and the reestablishment of wolves could return the system structure back to its 1872 state, although it is too early to ascertain whether this will occur. Meanwhile, climate warming could alter system structure to conditions not previously experienced.

Keywords:   hierarchical model, system responses, reduced biodiversity, natural regulation, northern range, elk numbers

Consilience: The explanations of different phenomena most likely to survive are those that can be connected and proved consistent with one another.

—Edward O. Wilson


The central purpose of this book is to synthesize the available evidence toward generalizing the effects of elk on structure and function of the northern range ecosystem and by implication to test the validity of the natural-regulation ecological hypothesis. The amount of evidence is immense. But contrary to statements in park publications, the research has not had an ecosystem orientation. There has been no conceptual model of the system that would guide systematic selection of projects addressing its structural and functional composition and would allow their integration into inferences about system behavior.

Rather, the research has largely proceeded by periodic emphases in response to the public relations or policy sensitivities of the moment: natural regulation, “is the northern range overgrazed?,” the 1988 fires, wolf introduction. Emphasis has largely been placed on a few components of the system, primarily charismatic animal and plant species: elk, bison, aspen, willows, and currently wolves. There has been little or no research on most of the fauna—small mammals, birds, invertebrates—or on soil microfauna and flora and on several aspects of the vegetation.

Except for ungulate demography, herbaceous and sagebrush production, erosion, and some aspects of nitrogen cycling, there has been little attention to system processes. Most projects have spanned only 1 to 2 years; the ungulate (p.281) censuses and periodic measurements inside and outside the large exclosures are exceptions.

The temporal dimension is important for two reasons. First, the several changes in elk numbers over the past 132 years (refer to figure 5.1) constitute a quasi-experimental manipulation, with two replications each of low and high numbers, of the independent variable. In an ideal world, continuous measurements of key system variables through this period would have given a clear, quantitative picture of elk effects; it would have given a basis for predicting the quantitative consequences of management alternatives.

A second reason for the importance of continuing measurements over time is in synthesizing the character of the northern range ecosystem. O'Neill et al. (1986) and Allen and Hoekstra (1992) conceptualize ecosystems as dual hierarchies of components and processes, each structured by systems of constraints. They argue that one cannot reconstruct system behavior knowing only component behavior.

Without temporal measurements of processes, it has been necessary to improvise with component information not originally obtained for that purpose but that can be used to reconstruct at least qualitative trends, and in some cases semi-quantitative ones. These include comparison of early anecdotal accounts with contemporary observations; early photographs matched with contemporary ones of the same locations (Kay 1990; Meagher and Houston 1998); dendrochronology of aspen and sagebrush (Romme et al. 1995; Ripple and Larsen 2000; Wambolt and Hoffman 2001); Keigley's plant architectural analyses; and inside-outside measurements of the large exclosures over a 33-year period. I infer system function in these cases over the intervals between two or more measurements or descriptions of state variables.

I have correlated or related these functions, based on what I have discerned in the preceding nine chapters about changes in northern range state variables, with the four elk population phases over the past ∼132 years. This covers the period from prehistory—some unspecified period prior to park establishment in 1872—to the mid- or latter 1990s. The newer annual winter movement from the park of, on average, a third of the herd may be creating a new set of conditions that have not yet been adequately researched. However, the effects of two-thirds of an equilibrium population of 16,000 censused animals remaining in the park may be no different from the effects of 10,000 censused animals during the 1930s–1950s (figure 5.1). Thus the question remains open.

A second, new variable is what the wolf effects will be—impossible to say at this point in time. Hence the reasons for not extending inferences beyond ca. 1999, except for the calculations of elk population equilibrium in 2, and speculations on other major environmental changes under way.

I have also grouped elk influences into first- and higher-order effects on major northern range system components and processes for which there is some evidence. This is a first step in conceptualizing the system in the O'Neill et al. (1986) and Allen and Hoeckstra (1992) hierarchical framework and arraying the constraints in this structure.

(p.282) For some readers, my evidence sources may not justify this attempt at synthesis. They do indeed vary in quantity and quality. But they converge on a paradigm in which inferences about the constituent parts and functions are internally consistent. And the inferences are consistent with what is known about the biology of those parts and contemporary ecological theory. Like all science, this paradigm is a pro tem judgment based on the prevailing evidence to date and remains the most probable unless and until overwhelming new evidence points to an alternative model. Thus my purpose in this chapter is to review the major conclusions of the preceding chapters and synthesize them into a conceptual model for the northern range ecosystem.


As discussed in chapters 25, the weight of the evidence points to a northern herd that occurred at four levels of abundance over the period from before park establishment to the beginning of the new millennium (figure 5.2). The northern herd, fluctuating in the short term as well as changing through these four periods of abundance, is that group of animals summering in the higher elevations of what is now the park. Prior to 1872 it moved north of present park boundaries to winter at lower elevations. From the latter 1800s to 1988, it wintered inside the park in an 83,000 ha area and an adjacent 17,000–ha area outside, collectively considered the northern range. Since the winter of 1988–89, on average a third of the herd has wintered in an additional 52,663 ha outside the park adjacent to the original 100,000.

For the first period of abundance (pre-1872 to ∼1884) I offer a speculative estimate of ∼5,000–6,000 animals on average (3). A reasonable hypothesis is that numbers were held at this level by some combination of aboriginal hunting and predation by large mammalian carnivores. Variations in winter severity probably induced some fluctuations in the population, allowing increases in mild winters, reducing numbers in severe ones.

Between approximately 1884 and 1962, the second population period, the herd exceeded censuses of 6,000, increasing rapidly from 1884 to a reported 20,000–35,000 (4) by the second decade of the 1900s. These may be conservative numbers, and the herd may well have exceeded them by some margin. Numbers declined after 1917, initially as a result of a series of severe winters but subsequently in response to increasing hunting removals outside the park, increasing herd reductions by the park, and I suspect increasing nutritional deprivation owing to forage depletion by the large herd.

This original increase and the continuing high numbers, even if declining, were facilitated by removal of the prepark constraints: ending Native American hunting of the animals, prohibition against Euro-American hunting of animals in the park, and reduction of predator numbers. They were facilitated by blockage of what had been the annual exodus out of the park area to lower-elevation winter range.

(p.283) In the third period, the herd was reduced to ∼ 5,000 censused animals in the winter of 1961–62 by park control efforts and outside hunting. It was reduced further through 1967–68, the last year of park control, to a censused 3,172. With cessation of park control, the herd immediately began increasing, rising to 9,981 censused animals by 1972–73. Thus the censuses counted fewer than 6,000 animals from 1959 to 1970, an 11-year period.

The herd grew continuously after 1970, with occasional severe winter setbacks, through the 1970s, 1980s, and early 1990s to censuses of 16,019 by 1981–82 and 19,045 in 1993–94. It declined somewhat in the latter 1990s and early 2000s, but continued at numbers above 10,000. As discussed in 2, the calculated equilibrium number for 1988–89 to 2002–2003 is 16,800 censused animals.

The increase since 1968 has occurred in the absence of any park control efforts. But it has proceeded in the face of increasing outside hunting kills and increasing density-dependent pressures imposed by growing nutritional deprivation. These two constraints together have set the number at which the herd, on average, has achieved equilibrium.

I have stressed that these are “censused” numbers to standardize the seasons of the counts (winter), taken after the fall hunting and park removals. Fall numbers have thus exceeded the censused numbers. And I have specified the census numbers because of the several studies that have shown the censuses to undercount actual population size by approximately 25% (2). The latter could be obtained by dividing the census numbers by 0.75.

Thus the herd has varied through alternating periods of low and high abundance, here placed at numbers below and above 6,000 censused animals. The dates of these four periods are slightly different from, but their population levels have resulted from, the four policy phases outlined in chapter1. Human actions have been major determinants of all the levels.


First-Order Effects on the Vegetation

I estimated in 6 that aspen woodland at the time of park establishment occupied ∼ 8,000–12,000 ha and 10–15% of what is now the park portion of the northern range. By the 1960s and 1970s, at the end of more than 75 years with high elk numbers, Houston and Barmore estimated the area of aspen at 1,418 ha, and 2% of the park portion of the northern range. Thus the type may have been reduced by 80–85%. There is fragmentary evidence of aspen response during the 1962–73 elk-population low. But the Romme et al. (1995) and Ripple and Larsen (2000) dendrochronological analyses showed no tree recruitment during the postreduction herd recovery.

Aspen stand structure is now comprised of low densities of superannuated trees at or near the limits of their longevity and with understories largely of grasses, (p.284) especially exotic species. This contrasts with stands in early photographs, in contemporary exclosures, and outside the park that had/have several strata composed of multiple aspen age classes and diverse understories of shrubs and forbs. With continuation of present trends, aspen as a species and as an ecological subsystem could largely disappear from the northern range in coming decades.

At 43,900 ha, the sagebrush-steppe is the most extensive of the major vegetation types and provides most of the forage for the several ungulate species. With no early measurements, and sagebrush too short-lived to provide dendrochronological data, the photographic record provides the only evidence of early sagebrush abundance. That record (see figure 7.3) shows it to have been a substantial component of the northern range sagebrush-steppe. By the 1920s, park biologists were expressing concern over heavy browsing impacts; the first measurements of its abundance in the vicinity of eight new exclosures in 1958 and 1962 showed it to have half or less the canopy cover that it would develop under subsequent protection (figure 7.4). The photographic record similarly showed a decline (figure 7.3) between 1920 and 1960.

Periodic measurements at the higher-elevation exclosures showed shrub canopy-cover increases, starting during the elk population low, and continuing to 1990 to levels ∼2 to 3 times those of 1967 (figure 7.4). But the 1990 levels inside the exclosures were ∼twice those on the outside. Photographic evidence suggested the same increase (figure 7.3). Cover measurements inside the Gardiner exclosures also showed increases to 1990, but continuing decline on the outside to virtual elimination of the species in the area (figure 7.4, 7.5). In 1994, Wambolt and Sherwood (1999) measured sagebrush density, production per plant, and production per unit area inside the higher-elevation exclosures at 1.61, 1.88, and 3.03 times those on the outside, respectively.

Thus sagebrush abundance has waxed and waned over time with changes in elk numbers. It is significantly more abundant inside exclosures than outside.

There are no sources of evidence that allow inferences on the condition of northern range herbaceous vegetation prior to park establishment when elk numbers were low. It cannot be judged from photographs.

The first accounts of herbaceous vegetation condition were expressions of alarm by park investigators in the latter 1920s, following more than 40 years of high elk numbers. Houston (1982) attributed these conditions to the drought of the 1930s, but the early accounts preceded the drought years. By 1958 and 1962 at the end of ∼75 years of high elk populations, two sets of herbaceous vegetation measurements inside and outside the new exclosures produced low values (figure 7.8).

Parker transects measurements both inside and outside the exclosures increased about twofold between 1967 and 1981, the increase starting during the elk population low (see figures 7.8). Cover measurements were not repeated until 1986, by which time they had also increased to more than twice the 1967 measurements (figure 7.8).

The Parker measurements declined between 1981 and 1990, the cover measurements between 1986 and 1990 (figure 7.8). Inside-outside measurements have (p.285) not differed significantly at all exclosures, although measurements at the Gardiner exclosures have consistently been lower than those at the higher elevations.

Park investigators have generally inferred that the evidence indicates no significant effects of elk use on the herbaceous vegetation, generally accepting Houston's discount of the early reports. Indeed comparisons of vegetation composition inside and outside the exclosures have shown little difference. Yet the early accounts, the low 1958–62 exclosure measurements, and the subsequent increases cannot be dismissed out of hand, especially because they coincide with major differences in elk numbers.

I hypothesize the following scenario for the effects of elk on the northern range herbaceous vegetation. Elk have a negative direct (first-order) effect on the herbaceous vegetation by grazing it, and a positive indirect (second order) effect by suppressing sagebrush competition. After ∼75 years of grazing and browsing by a massive elk herd, both sagebrush and herbaceous vegetation had been driven to low levels by 1958–62 (figures 7.4, 7.8).

With reduction of the herd and construction of the exclosures, sagebrush increased from 1967–90 (figure 7.4), herbaceous vegetation from 1967–81 (figure 7.8), both inside and outside the exclosures. But by 1981, both grazing pressure and sagebrush competition had increased outside the exclosures to the point of reducing the herbaceous between 1981 and 1990 even though the sagebrush was only about half as abundant as that on the inside. Herbaceous abundance on the inside, although freed of elk herbivory, was forced to coexist with twice the sagebrush competition on the outside and could increase no more than that on the outside. Despite comparable herbaceous abundance inside and outside the exclosures, the lower sagebrush density on the outside effects a more open plant community and allows invasion of nonnatives, something prevented by the dense vegetation on the inside.

Thus the evidence implies that the abundance, production, composition, and physical structure of the northern range sagebrush-steppe have varied during park history in response to alternating levels of elk use. These changes include the invasion of exotic plant species facilitated by that use. Whether the herbaceous decline of the 1980s (figure 7.8) would continue, and whether a sagebrush decline would set in with continuation of elk numbers of the 1990s, thus moving the vegetation back toward the conditions of 1958–62, is not known.

Although conifers occupy some 41% of the park portion of the northern range, they do not contribute a significant amount of forage for browsing ungulates as shown by food habits studies and implied by the virtually universal highlines that remove the foliage from browsing reach. But they provide resources for other components of the biota, and the highlining chronology gives another indication of elk impacts on the northern range ecosystem.

The photographic record does not show significant highlining before 1900. Reports of park personnel and photographs indicate its appearance in the early 1900s, some 2 to 4 decades into the first period of high elk numbers. Demographic research is needed to ascertain whether there are young replacement trees in the mature conifer stands.

(p.286) There are no data on the early abundance of the 20+ species of deciduous shrubs and small trees that are minor components of the northern range vegetation. At least eight of these produce berries edible for humans and wildlife. Early explorers commented on their prevalence, but early park investigators 30–50 years later commented on heavy browsing impacts on these largely palatable species. Kay (1990, 1995) measured extreme inside-outside differences in these at the exclosures and near absence of berry production on the outside plants.

All observers agree that the photographic record and reports of early observers indicate robust willow growth in moist areas throughout the northern range prior to and in the early years after park establishment, and that the area occupied by the species has declined sharply during park history.

Early photographs and observers' comments indicate full-statured willow growth in suitable habitats throughout the northern range. This growth endured in a period when, as the evidence shows, elk were present in low numbers and migrated out of the northern range in winter.

The first reports of browsing impacts appeared in 1915 and the 1920s. Kittams commented on and photographed willow hedging and disappearance in the 1940s. Following construction of the large exclosures in 1957 and 1962 at the beginning of the elk reductions, Barmore (1980) measured immediate growth response of willows inside, but continued suppression outside. But he, O'Gara (personal communication, 1996), and Patten (1968) all inferred some regrowth during the population reduction.

Willow suppression continued through the latter 1900s during the elk population recovery (figure 10.4). Kay (1990) shows a 1965 park photograph with willow growth that was no longer evident in his 1988 retake of the site. He also comments on growth of willow inside the Tower Junction exclosure constructed in 1957 but its disappearance by 1973 following dismantlement in 1971.

Thus riparian willows are another vegetation component of the northern range that has waxed and waned with the changing phases of elk abundance: full-statured and widely distributed during the elk population low at the time of park establishment; heavily impacted during the first population high at the end of the nineteenth and first half of the twentieth century; slight response during the short period of reduced numbers; then continued suppression during the postreduction population recovery.

Park publications have stated that willow has declined 50% during park history. Kay (1990) has claimed 95% reduction. Unfortunately these numbers fail to distinguish between areas occupied by full-statured shrubs or to areas where the species has been eliminated completely (figure 10.1). It is unfortunate that there has not been a systematic study of this question by comparing early photographs with field inspection as Kay and Wagner (1996) conducted on aspen clones.

Early park investigators commented on browsing impacts on cottonwoods by the 1920s. Kittams (1948) compared early photographs of cottonwood stands with his 1940s retakes and found that they had “virtually disappeared.” Dendrochronological analyses of three samples of narrowleaf cottonwoods by Keigley (p.287) (1997b, 1998) showed active tree establishment between 1840 and 1894, then only one established between 1894 and 1934, and none between 1952 and 1962. He aged 17 trees that grew between 1963 and 1974, the approximate period of low elk numbers, and then none established between 1974 and 1992 when he conducted his study.

More recently, Beschta (2003) examined all cottonwood trees in a 9.5–km2 area of the Lamar River valley in 2001 that were >5 cm in diameter. He estimated the ages of ∼497 narrowleaf cottonwoods. None of these had developed during the past ∼60 years.

Summation of First-Order Vegetation Effects

I have now reviewed the evidence available on northern range vegetation in four sequential, chronological periods: prior to and in the early years following park establishment; approximately 1884 to 1959, when the northern herd numbered more than 6,000, in some years 3 to 4 times this number or more; 1959–70, the period of extreme herd reduction and first few years of recovery; and 1971 to early 2000s when the herd had returned to high numbers. Except in 1 or 2 cases of vegetative components in 1 or 2 periods, there is evidence of change in each of the 6 vegetative components in all of the 4 time periods.

In some cases the evidence is fragmentary. In most cases it is circumstantial. But the sequence in essence constitutes two pseudo-replicates each of two treatments for each component. The inside-outside exclosure comparisons and inside-outside park comparisons constitute further pseudo-replication. In all cases the patterns are consistent with what would be expected of large variations in elk herbivorous pressure and with the extensive literature on herbivory, including the immense range-ecology literature, which has been largely ignored in park publications.

As discussed in 1, the northern range may be entering a fifth replicate in response to wolf effects. Singer et al. (2001a) observed increase in willow heights, possibly in response to changes in elk distribution driven by wolf hunting. Ripple and Beschta (2003) observed differences between 1995–96 and 2001–2003 photographs in the heights of cottonwoods in “low-risk” and “high-risk” sites along Soda Butte Creek and the Lamar River. Cottonwood heights increased on low-risk sites but not on high-risk ones. And Ripple and Beschta (2004) photographed willow regrowth on Blacktail Creek.

The evidence overwhelmingly indicates alteration of all components of the vegetation when the northern herd increases to levels above 6,000–10,000. The effects range from nearly complete elimination of some components, as with aspen woodland, to altered structure as with sagebrush-steppe and conifers.

With pronounced changes occurring in all components of the vegetation in response to different levels of elk numbers, it is to be expected that the other sectors of the ecosystem that interact with the vegetation would also vary. Though most of these sectors have had little or no research attention, there is evidence of response in those that have been studied, particularly the other ungulate (p.288) species. I now summarize that evidence that has been presented in previous chapters, grouped according to the services provided by the vegetation.

Second-Order Effects via the Vegetation

Reduced Food Resources for Herbivores

I discussed at considerable length in 9 an apparent population decline in three species of upland ungulates during park history as elk increased an estimated threefold by the 1990s and possibly four-to sevenfold by the early 1900s. Bison may be an exception, possibly increasing during this period when freed of the same constraints as those limiting the elk in prehistory.

Collectively, the declining upland species are two browsing and one grazing form. Pronghorn, for which sagebrush is a winter staple, formerly occurred in large numbers in the northwestern area of the northern range where sagebrush has been profoundly reduced (figures 7.5, 7.6). I suggested pronghorn decline on the order of 90%. Mule deer appear to have declined initially to the 1940s to 1960s (9.5) when they wintered inside the park, but then recovered numbers during the 1980s and '90s when they began wintering outside the park and evading elk competition.

Bighorns are grazers, probably experiencing competition with elk and bison for a heavily usurped herbaceous vegetation. I suggested ∼ 90% reduction during park history. They showed some evidence of population recovery during the elk population reduction (figures 9.1), but declined again in the latter twentieth century.

In 10 I discussed two browsing riparian species that probably declined in response to the sharp reduction in the riparian zone. Whitetailed deer disappeared completely as northern range residents by the 1920gs. Moose largely disappeared from the northern range following their late appearance in the park early in the twentieth century. Singer et al. (1998a) agreed that this species' decline was probably driven by elk competition.

The virtual disappearance of beaver from the northern range may be partially the result of food deprivation. The willows and aspens that are their primary food sources are also their major building materials. When removed from or near aquatic sites, the animals are unable to build the dams that form the ponds that provide colony habitats.

Beyond ungulates and beaver, there has been no work examining the effects of vegetation alteration on the food sources of small mammals and invertebrates. There almost certainly have been extensive influences.

Habitat Reduction

The changing vegetative structure undoubtedly has altered habitat for the many species that use the vegetation for that purpose. But there have only been a handful of studies in the northern range to document this effect. I cited several studies in other areas of the Greater Yellowstone Ecosystem in 6, 7, and 10 that doubtless reflect what is occurring on the northern range.

(p.289) These studies have shown several avian species that are unique to aspen and disappear with elimination of the woodland. Other species decline as the structural diversity of the vegetative type is reduced (see Dobkin et al. 2002). Similar effects occur in cottonwood stands.

Jackson (1992, 1993) documented declines in avian numbers and species as browsing intensity increased in northern range willows. Berger et al. (2001) observed the same effect in Teton National Park willows in response to moose browsing.

I am not aware of any studies that have measured effects on the avifauna of structural modifications in the northern range sagebrush-steppe. But in 7 I cited work elsewhere that indicated such effects. One can see an extreme example by observing the avian activity inside the Gardiner exclosures and the virtual absence of such activity outside (figure 7.5).

As discussed in 10, Debinski (1994, 1996) discussed the tenuous position of the Yellowstone checkerspot butterfly in the northern range riparian zone. She also observed (Debinski and Brussard 1994; Debinski et al. 1999) correlations between butterfly and avian diversity suggesting that browsing impacts on riparian vegetation has the same effects on butterfly diversity as on avian diversity. I cited G. E. Beetle's (1997) work on snails in aspen stands in 6, with habitat reduced to what she called “decadent” stands for the 8–11 snail species she observed.

Beyond these few studies, I am not aware of any research on the hundreds of northern range invertebrate species that would give clues to their population trends during park history and the possible elk effects on those trends. Nor am I aware of any studies on the small mammals, although research elsewhere has explored their relationships to sagebrush-steppe vegetative structure. I also do not know of any northern range studies on the effects of beaver pond disappearance on the aquatic fauna of this type.

First- and Second-Order Effects on Carnivores

As discussed in 9, the northern range has long had higher coyote densities than most of the northern portion of the western United States. Their conspicuous presence prompted Murie's (1940) study of the species in the 1930s. Murie observed that elk carrion was a major source of winter food for the northern range animals. Knowlton (1972; Knowlton et al. 1999) has concluded that availability of winter food is a major determinant of coyote abundance. Hence the high northern range elk population has in all probability been a major determinant of the high coyote densities.

The high coyote population, in turn, has influenced other system components. In 9 I cited the O'Gara (1968) and Barmore (1980) observations of high pronghorn fawn mortality exacted by coyotes. Thus the large elk population applies the dual pressures on pronghorns of forage competition, and what Holt (1977) calls “apparent competition” by supporting significant coyote predation.

The high coyote population may also suppress populations of other medium-sized carnivores (Wagner 1988). Gese et al. (1996) have observed aggressive (p.290) encounters between coyotes and red fox (Vulpes vulpes) on the northern range, a relationship that has been observed elsewhere (Gosselink et al. 2003).

There is also reason to infer dual effects on grizzly bear. Kay (1990) has pointed out that riparian zones in the eastern slopes of the Rocky Mountain system are important travel lanes and food sources for the species and postulated that the zones' impoverishment on the northern range has been detrimental to grizzlies. To this I would add the heavy impacts on berry-bearing shrubs, discussed in 8. But other authors (Mattson et al. 1991; Green et al. 1997) have pointed out the heavy use of elk carcasses and calf kills by grizzlies emerging from hibernation in spring and the likely benefits of these.

All of these interactions are subject to change, depending on the effects that wolves will have on both elk and coyotes. This will be touched on in the next section.


Ecosystem function is driven by myriad processes acting over a range of phenomenological integration: physiological, population, community, and system among the biota and chemical and physical processes among the abiotic components. Because the purpose here is to synthesize northern range ecosystem structure and function, I will comment here only on system-level processes.

There have been only a few studies of northern range system processes, and these have spanned only 1 to 2 years. Hence there is no basis for correlating variations in process measurements with the phases of elk abundance. Some tentative inferences can be drawn from periodic standing-crop measurements over a period of time.

Evaluating how primary production of the northern range system today compares with the level prior to 1872 requires a combination of production values for the several vegetation types and for these two time periods. There are no data for such a comparison. So only a crude perusal of numbers, mostly from standing-crop measurements, can be brought to bear on the question.

I cited available data on herbaceous primary production on the northern range in 13. All are from either 1- or 2-year studies. But one can perhaps infer production trends from the standing-crop measurements shown in figure 7.8. These showed low levels of herbaceous standing crop in 1958, 1962, and 1967, the first two measured at the end of a seven-decade period with a large elk population.

The measurements increased between 1967 and 1974, the latter few years with reduced herd size. They approximately leveled off from 1974 to 1981, then declined in 1986 and 1990 when the northern herd had returned to high levels. Thus, to the extent that standing-crop measurements at the end of the growing seasons are indices of herbaceous annual production, the latter have varied with changing herd size.

(p.291) But as I have discussed, there is some evidence that the herbaceous vegetation is influenced by sagebrush competition, and that the herbaceous trends may in part be a function of sympatric sagebrush trends as well as elk grazing. Thus, trends in herbaceous vegetation are probably determined by these two influences, and one cannot meaningfully speculate on how primary production in this type today compares with its level prior to park establishment. It certainly did not experience the grazing pressure that it had during the twentieth century. I reviewed the work of D. A. Frank in 13 and found the evidence for overcompensation mixed at best, and not persuasive that it is occurring in the northern range grasses. In total, it seems unlikely that herbaceous production today is higher than it was in 1872.

Despite sagebrush regrowth since 1962 (figure 7.4), the inside-outside exclosure comparisons indicated that browsing constrained the recovery through 1990. And Wambolt and Sherwood's (1999) extensive measurements in 1994 showed sagebrush outside the exclosures producing at one-third the inside levels per unit area.

An 80% reduction in aspen clearly implies an equivalent decline in production from 1872 levels by a species that formerly occupied on the order of 10% of the northern range. Woody riparian vegetation has evidently declined by a comparable order of magnitude, although it is a smaller component of the vegetation, perhaps originally occupying no more than 1%.

A major contributor to northern range primary production is the conifers, for which, to my knowledge, there has been no research on the northern range. There have been three documented significant changes in this type during park history. One was its spread of ∼5% reported by Houston (1982). The second was the 1988 fire. Although there are estimates of the total proportion of the park's area that was burned, I am not aware of any similar value for the northern range, at least for the coniferous vegetation. The third change has been the ubiquitous highlining during the twentieth century. Thus, whether there have been changes in total coniferous production, what is within reach of browsing ungulates and smaller animals has decreased.

In total, there is no way of calculating primary production for the entire northern range vegetation today or how that would compare with production levels at the time of park establishment. But with the clear decline in aspen and riparian production, evident reduction in sagebrush, either similar or reduced production in herbaceous vegetation, and significant decline in coniferous growth within reach of ungulates, the total primary production available to these and smaller animals is in all probability reduced from the levels present in 1872.

Whatever the trends in primary production during park history, it seems clear that the speculative estimate of a 2.3× increase in ungulate biomass since park establishment (9.2) implies an increase in ungulate consumption and secondary production of similar magnitude. Except for the near disappearance of northern range beaver and decline of some avian species, nothing can be said about the remainder of the fauna.

(p.292) Coughenour (1991) commented, on the basis of inside-outside exclosure measurements, that grazing diverts herbaceous plant material into ungulate production. The material would otherwise become litter—presumably much of it did prior to park establishment—and courses through the detritivory channels of the ecosystem.

If only a limited amount of research has been conducted on northern range bioenergetics, even less has been done on nutrient cycling and that largely the work of D. A. Frank and co-workers on nitrogen. These authors make a convincing case for ungulates' role in accelerating the nitrogen cycle on the northern range, and Coughenour (1991) measured higher nitrogen content in vegetation outside exclosures than inside. The result may be more rapid nitrogen cycling on the northern range during the twentieth century than before 1872, when few ungulates wintered there.

But the input-output processes have not been systematically studied to work out the system budget. There is evidence of accelerated surface erosion since park establishment and the destruction of microphytic crusts. The latter could reduce nitrogen fixation. In total no mass balance can be calculated at this time.

Hamilton, Keigley, and I discussed in chapters 11 and 12 the evidence indicating that surface erosion has increased during park history. Although some of this could be the result of hoof impacts, it is in all probability induced at least in part by vegetation alteration and released constraints on soil movement. Singer (1995) measured 2 to 8 times more bare ground outside exclosures than inside.

We also discussed evidence of stream-bank destabilization that resulted from the widespread elimination of riparian vegetation. The fluvial geomorphology of the Soda Butte and Lamar Rivers today are quite different from what is shown in early photographs.


Concepts Clarified

With essentially all of the measured components of the northern range ecosystem responding to variations in the elk population, it must follow that the synthetic characteristics of the entire system are also responding. I will examine those responses in this section. But before doing so, I need to clarify concepts to be discussed to ensure clear communication.

I use the term ecosystem in the standard textbook form: a group of interacting components, biotic and abiotic, that collectively perform one or more functions that the individual parts could not perform separately. The concept has the additional implication that if one part of the system is changed, the other parts, and hence the system, are changed.

Ecosystems have spatial limits, usually arbitrarily set for purposes of study or management. I have accepted the borders of the northern range (figure 1.1) (p.293) as the limits of the northern range ecosystem. These delimit the area in which the northern herd winters.

There are no closed ecosystems. All have inputs and outputs of energy and material that cross or penetrate the limits. Even the Earth is not a closed system. But contrary to Chase's (1986) rejection of the concept because no ecosystems are closed, the existence of inputs and outputs does not invalidate ecosystem analysis as long as they are measured and included in any mass-balance formulations.

Ecosystem analysis investigates processes that involve entire systems, as opposed to the functioning of the individual parts. Such processes include acquisition, containment, and exchange of energy and material. They have had very limited research attention on the northern range.

However, energy and materials are acquired, contained, and exchanged by the components that have received research attention. Hence system processes can be inferred to some degree from the component dynamics, which also provide the system's physical structure, and the constraints on both structure and function. In the sections that follow, I will attempt to infer structure and function of the northern range ecosystem in terms of the components discussed up to this point in the book.

The Northern Range Ecosystem Structure

A Hierarchical Structure

As stated at the beginning of this chapter, O'Neill et al. (1986) and Allen and Hoekstra (1992) have generalized ecosystems as hierarchies of functional components fundamentally structured by systems of constraints. Though O'Neill et al. considered that there may be several different forms of hierarchies, they emphasized a dual structure:

the ecosystem is a dual organization arising from the structural constraints that operate on organisms and functional constraints that operate on processes. … Once the constraints are lost, the hierarchical organization is lost. When a system goes unstable, it is the normal functioning of unconstrained components that tears the system apart. (O'Neill et al. 1986:210–11)

Allen and Hoekstra (1992:34, 90) further emphasize the important role of pathways:

frequency and constraints are the most important criteria for ordering levels. Upper levels constrain lower levels by behaving at a lower frequency. … We define the parts and explanatory principles of ecosystems as pathways of processes and fluxes between organisms and their environment … the critical parts are the pathways that may involve organisms, not the organisms themselves.

The northern range system can be conceptualized according to these criteria into a hierarchy. But for illustrative purposes, the concept must be simplified to (p.294) the few, major driving forces affecting the system's structure and function. Smith et al. (2003) have characterized the northern range biota with a network of organisms connected by trophic or competitive links that is highly instructive in representing the complexity of the system. But as mentioned in 10, Holling (1995) generalizes that ecology is moving toward a paradigm in which most of the functioning of complex ecosystems is driven by a small number of its components and processes. This appears to be the case with the northern range.

The northern range ungulate guild is a case in point. Technically, all five upland ungulate species apply herbivorous pressure to the vegetation, and each could be represented as a single block in the flow of influence in a hierarchy. But as discussed in 9, elk constituted 84% of upland ungulate numbers, made up 91% of ungulate biomass during the 1990s (9.2), and are the major herbivorous force affecting the vegetation. Moreover, the major interaction between the four nonelk species with the vegetation is not their first-order impact on it but its usurpation by elk and the consequent second-order competitive pressure on them by the elk. The position of bison, one of the four, may be intermediate between the extremes of elk impact on the vegetation and the other three species heavily influenced by elk competition.

Hence, I have simplified representation of the system into a hierarchy comprised of the few major driving forces affecting its structure and function subject to the following provision:

  1. 1. Elk are positioned by themselves in the hierarchy for the above reasons.

  2. 2. The O'Neill et al. (1986) and Allen and Hoekstra (1992) constraints and the latter authors' pathways are in some cases one and the same.

  3. 3. The magnitudes of the interactions, or in the above authors' lexicon the strength of the constraints, have varied between the different phases of elk abundance. Hence the structure of the system has changed, and the chronology of the hierarchy must be specified.

Consequently, the evidence points to the northern range ecosystem represented by a trophic hierarchy shown in figure 15.1. Prior to park establishment the top predators—humans, wolves, cougars, grizzly bears, coyotes—and winter migration out of what is now the northern range exerted the top-level constraints and prevented heavy ungulate pressures on the vegetation. Kay (1990, 1994a, 1998) has stressed the importance of aboriginal hunting and identified Native Americans as a keystone species.

Though the indices of wolf abundance discussed in 3 pointed to low wolf numbers prior to 1872, they were not necessarily inconsequential predators on elk and other ungulates. The latter were also present in low numbers. Hence the prey:predator ratios may have been low enough to allow the wolves to exert a significant predation rate.

The predatory constraint on the elk herd and other ungulates minimized their impacts on the vegetation and allowed it to develop the profuse growth shown in the early photographs and commented on by the early observers. At (p.295)


Figure 15.1 Hierarchical structure of Yellowstone's northern range ecosystem at park establishment and in the 1990s. Rectangles are major system components; ovals are constraints and/or pathways as described by O'Neill et al. (1986) and Allen and Hoeckstra (1992). Arrows represent the directions of influence.

that point, the major constraints operating on the vegetation probably were competition functioning within localized conditions of climate, topography, soils, and periodic fires.

The full expression of the vegetation then provided ample food resources for the ungulates, beaver, and other herbivores; habitat for birds, small mammals, and invertebrates as discussed in chapters 7, 8, and 10; and constraints on watershed and fluvial processes as discussed in chapters 11 and 12. The entire pattern is consistent with the quote from Allen and Hoekstra (1992): The structure and function of the entire hierarchy is maintained by the constraints at the top level.

With removal of the top predators—transfer of Native Americans away from the area, control of the large carnivores, and protection from Euro-American hunting by the park boundaries—the top-level constraints were removed. Along with cessation of winter migrations, the number of elk wintering on what became the northern range increased to several times prepark numbers. It then exerted the heavy pressures on the vegetation discussed in previous chapters, and the elk and other components of the system became limited by the reduced vegetation. Kay (1998) characterized these changes as conversion from top-down to bottom-up limitation, or as change from a predator-limited to resource-limited system. Again it coincides with the Allen and Hoeckstra (1992) quote on what happens to a system when the top-level constraints are removed.

(p.296) In an analogous model, Ripple and Beschta (2004) represent the changes in the northern range system as a trophic cascade occasioned by the removal of wolves. They suggest that wolves, and associated predation risk, can structure the system. They emphasize system responses (vegetation recovery, increase in beaver) in areas from which elk have moved to avoid predation risk by the presence of wolves: “Can predation risk structure ecosystems? Our answer … [on the basis of evidence and theory] is yes.”

This is indeed part of the story. But the evidence also points to a likely role of aboriginal hunting, perhaps a major one. Wolves were still present, though being killed, through the period between park establishment and the 1920s when the elk herd burgeoned. Constraints on aspen ramet development began in the 1890s, and severe impacts on other components of the vegetation were evident at least by 1914.

Moreover, as the evidence shows, it is the number of elk and the duration of use that are the main determinants of the degree of impact. Changing the distribution of the animals may provide system releases in areas they avoid, but this only shifts the distribution of the herbivorous pressure on the system as a whole. That pressure is not reduced unless elk numbers are reduced.

Declining Biodiversity

The changes in the northern range ecosystem since park establishment can also be characterized as progressive reduction in biodiversity. Boyce's (1998) comments on diversity seriously oversimplify this aspect of changes in the system. He is correct that Chadde and Kay (1988) found higher species richness outside northern range willow exclosures than inside. And this is probably an example of other range ecology findings that grazing mediates competition between plant species and permits coexistence of more species, much as Paine's (1966) classic starfish- predation study showed maintenance of diversity in intertidal invertebrates. But the pattern has not been consistent on the northern range: Singer (1995) and Reardon (1996) did not find inside-outside differences in herbaceous species composition at the large exclosures.

Beyond these cases, diversity is a far more complex matter. It can be measured at a range of biotic levels—genetic, species, communities as in the above cases, and landscapes—and spatial scales.

The northern range is a tapestry of subsystems: aspen woodland, sagebrush-steppe, coniferous stands, riparian zones. Some of the more mobile animal species use several or all of these, variously for feeding, cover, reproduction, and movement. The landscape thus provides what a number of authors are calling “habitat complementation” (Pulliam 1988; Pulliam and Danielson 1991; Dunning et al. 1992). Changes in their abundance and proximity within the animals' mobility ranges may foreclose the northern range for some species. Kay (1989, 1990) points out that riparian strips are important travel lanes for grizzly bears. Their near elimination from the northern range may have contributed to the species' decline in that portion of the park.

(p.297) More directly, the widespread elimination of aspen, willows, and cottonwoods constitutes a reduction in habitat and landscape diversity. The almost certain decline in avian and insect species obligate to these types, and likely decline of more generalist forms using them as discussed in chapters 6 and 10, must reduce the area's species diversity. Two documented mammalian cases are the disappearance of white-tailed deer and virtual disappearance of beaver. Boyce's disclaimer that these are of no consequence because they are abundant outside the park is not relevant to the current question of diversity trends on the northern range.

A more subtle scale is the situation of low-mobility animal species existing in meta-populations within fragmented landscapes. An entire symposium (Ehrenfeld 1995) discusses species persistence in fragmenting landscapes where habitat patch sizes decline, interpatch distances increase, and corridors between patches disappear. Local extinction probabilities increase in arthropods, plants, amphibians, birds, and small mammals (Fahrig and Merriam 1995), and repopulation probabilities decline, with increasing landscape fragmentation.

These questions have received no systematic research attention on the northern range, the few exceptions being Jackson's (1992, 1993) work on willow avifauna, Debinski's (1994, 1996) on butterflies, and Romme's (see Romme and Knight 1982) and Turner's (Turner et al. 1994) work on disturbance effects on landscape dynamics. The fact that the few studies that have been conducted have shown the expected effects on species diversity suggests that they may be common in the northern range biota.

The NRC review (Klein et al. 2002) made a number of statements about the northern range relevant to this discussion on changing biodiversity that need to be addressed. They are not helpful in and of themselves, but they prompt more meaningful conceptualization of the changes that have taken place. Thus the review stated that: “Yellowstone is not in ecological trouble … not on the verge of crossing some ecological threshold beyond which conditions might be irreversible … has not been associated with ecological disaster… dramatic ecological change does not appear to be imminent.” Without ecological definition, these terms are not meaningful and thus do not convey the reality of what has occurred in the northern range ecosystem. What constitutes ecological trouble, ecological disaster, dramatic ecological change, or thresholds in terms of the system's components and processes is not clear or probably conceptualized.

What has occurred is a progressive biotic impoverishment in the sense of a continuum. In that perspective, what is a disaster? Has a threshold been crossed when a species (e.g., whitetailed deer) is eliminated? Or two species (whitetails and beaver)? Or if bighorn and pronghorn also disappear? Or if a habitat is eliminated? Aspen and its obligate fauna, now reduced by ∼80%, could disappear with continuation of the trends of the 1990s. Has an ecological threshold, beyond which conditions might be irreversible, been crossed with loss of several centimeters of topsoil? Or with alteration of the fluvial geomorphology of the Lamar River? The reality is a progressive reduction of biodiversity and change (p.298) in system structure. Given a hypothetically long enough period of time with continuation of trends of the 1990s, the system could be markedly changed. But some form of ecosystem would continue to function. Would the change then have been “dramatic,” or a “disaster,” or would a threshold have been passed?

In a similar case of ambiguity, Boyce (1998) misquoted Wagner and Kay (1993) in ascribing to them the statement that “Yellowstone will somehow self destruct.” If trends into the 1990s were to continue, the northern range might be expected to continue to lose species—pronghorn and bighorn might be next—and landscape diversity. But some form of ecosystem would continue to persist. Wagner and Kay never called this self-destruction. And if we designate the prepark state as the “natural” (i.e., without Euro-American influence) state, the system today is well out of the range of natural variation.

Finally, the point needs to be made that all of the changes that have occurred cannot be attributed to the natural-regulation policy. Much (perhaps most) of the change had occurred by 1967 after ∼75 years of use by an enlarged elk herd. The effects of the policy have been to resume the changes underway before the herd reductions and probably to accentuate them further.

Alternate Steady States?

The question arises as to whether the northern range ecosystem has occupied one or more stable states. Most authors define stability as the tendency for a system to return to some “persistent configuration” following perturbation away from that configuration (see May 1973, Case 2000). Recent literature has questioned the reality of stability and its related concept equilibrium in natural systems, often citing its treatment by Botkin (1990).

Part of the debate is semantic, confused on the question of time scale. As O'Neill et al. (1986), Allen and Hoeckstra (1992), and Wagner et al. (1995a) point out, equilibrium is a matter of time scale. On geological, evolutionary, or climatic time scales or during continuing human disturbance there are no equilibria. But for periods of time scaled in years or decades, many systems remain in roughly constant ranges of values over time.

A related problem is confusion with the condition of stationarity, the total absence of change. Natural systems in variable environments fluctuate over time and may rarely, if ever, settle precisely on single equilibrium values. But negative feedback processes place probability limits on and restrict the range of values within which a system fluctuates. They exert continuing pressures to move systems back toward (if not precisely to) mathematical equilibrium points. The density-dependent pressures operating on the northern elk herd function in this manner, as discussed in 2. Equilibrium of natural systems in varying environments is thus absence of net or mean trend over specified periods of time, maintained within roughly constant limits of variation.

The state of the northern range system was clearly different prior to park establishment from its state during the 1900s. Whether that state for some decades or centuries prior to 1872 could be considered stable obviously can only be speculated on. There were climatic variations including the Little Ice Age and (p.299) the Medieval warming period. Yet temperature proxies for the past ∼400 years do not show any secular trends or extreme anomalies until the mid- to latter nineteenth century, initiating the contemporary warming period (Mann et al. 1999; Crowley and Lowery 2000).

Kay (1990) has pointed out the relative scarcity of ungulate remains down through the western North American archaeological columns for what must be centuries into the past suggesting low ungulate densities through that period. And the northern range system changes associated with the 1959–70 low elk numbers, slight and short-lived though they were, tended to move the northern range back toward the prepark state.

Whether the prepark system existed in some roughly stable state, it was clearly different from its state during the 1900s. The change was elicited, the earlier state destabilized, by establishment of the park and associated management actions that removed the earlier constraints. The change is symbolized by the shift from left to right in figure 15.1 Moreover the evidence does not suggest any stability during the 1900s. The system has undergone continuing change during the period.

What for the Future?

What the future, perhaps 50–75 years, holds in store for the northern range can, of course, only be speculated on. If one were to predict only on the basis of trends up to the mid-1990s, no significant changes in the northern range environmental circumstances, and continuation of current policies, one could reasonably project continuation of the trends of the preceding 20–30 years or indeed most of the twentieth century. The northern herd would probably continue for a time to fluctuate around an equilibrium value of 15,000–20,000 censused animals, but this might decline later into the period. Aspen and cottonwoods could well disappear, as would most willows. Sagebrush would likely decline further with grasses becoming dominant steppe vegetation. Nonnative plant species would probably increase and faunal diversity would surely decline.

But three recent changes in environmental circumstances may well change this scenario or may already be doing so. The first is the 41% expansion of the winter range that began in 1988, and the movement of, on average, one-third of the herd into this extended area. This could be lightening the elk pressure on the original 108,553 ha somewhat. But two-thirds of 16,500 censused animals were sufficient to alter the vegetation in the ways discussed in previous chapters on the original 108,553 ha. Hence, the relief might not be great.

The expanded range could allow some herd increase, as the Taper and Gogan (2002) calculations suggest. In that case the long-term trend could be eventual depletion of the new 52,663 ha, and resumption of the trends in the entire 152,663 ha that prevailed before 1988 on the 108,553.

A second change is the wolf reintroduction. Smith et al. (2003) provide an excellent overview of the growth and distribution of the wolf population since release in 1995, and early indications of possible northern range changes. How (p.300) many (if any) significant changes occur will depend on how much effect the wolves have on elk and other components of the system. The wolf population may have stabilized at approximately 14 packs and 132 animals in the park and 8 packs and 77 wolves in the northern range. Hence, whatever effects occur are likely to result initially from the current population size.

Recent modeling exercises (Eberhardt et al. 2003) project potential effects ranging from slight (∼ 10%) reduction in the elk herd to almost complete elimination of both elk and wolves, depending on the equations and parameter values used. If the wolves succeed in sharply reducing the elk herd, it is likely that the wolf population would decline as well. In 3, I cited evidence presented by Kay (1990) and Schullery and Whittlesey (1992) indicating that wolf numbers were low at the time of park establishment, an indication of low ungulate numbers.

There is still no unequivocal evidence, as this is written in 2004, of significant wolf effects on the northern herd. Calves have been 43% of wolf-killed elk (Smith et al. 2003), and winter-end calf:cow ratios in the late 1990s and early 2000s have been half or less the values in preceding years according to T. O. Lemke (personal communication, July 18, 2001). The herd has declined in the early 2000s, but as discussed in 1, these have been drought years, and Merrill and Boyce (1991) and Coughenour and Singer (1996) have shown strong correlations between northern herd recruitment and population growth rates and precipitation. Lemke has pointed out to me that calf:cow ratios and elk numbers have been down in these years elsewhere in western Montana, where there are no wolves.

Some recent northern range system changes appear to be associated with wolf activity, particularly due to alterations in elk distribution and movement. Several investigators have observed increase in willow growth in the vicinity of wolf dens (Singer et al. 2001a; Smith and Guerny 2002). Ripple et al. (2001) measured greater aspen sucker heights in high wolf-use areas than in low. Fecal pellet group indices showed lower elk use in the high wolf-use areas than in the low.

Ripple and Beschta (2003) measured higher browsing intensity and lower heights of cottonwood in open terrain with clear surrounding views than in sites where visibility was blocked by stream banks and vegetation. The authors hypothesized that elk occupy the more open sites which permit visibility of wolves. As stated before, Ripple and Beschta (2004) observed willow regrowth in response to wolf-driven shifts in elk distribution. However, until and unless there is significant reduction of the northern herd, these changes may only represent redistribution of browsing pressure and not actual reduction over the area as a whole.

Crabtree and Sheldon (1999) observed a 50% reduction in the northern range coyote population following wolf reintroduction. Wolves have been observed killing coyotes. Smith et al. (2003) reported some increase in pronghorn fawn survival in recent years, possibly associated with the coyote reduction.

If these wolf-induced changes continue, and particularly if they become significant, they will in essence move the system back toward the pre-1872 state. (p.301) This will be accomplished by restoration of some of the top-level constraints on the system hierarchy.

A third environmental change currently under way is the incipient global warming and climate change. This has enormous implications for the northern range and the entire park. Two major general circulation models (GCMs)—the British Hadley Centre Circulation Model 2 and the Canadian Coupled General Circulation Model 1—project mean annual temperature increase of 3.6°C (6.5°F) and 6.3–6.5°C (11.3–11.7°F) by 2080–2100 in the Intermountain West of the United States (Wagner 2003). They project annual precipitation increases of 54–184% (Wagner 2003).

Two recent analyses of twentieth-century weather records in the region show that these trends had already begun during the 1900s. Annual, average temperatures in the northern Rockies rose 0.6°C (1.1°F) (Baldwin 2003) during the 1900s and annual average monthly minimum temperatures rose 0.859°C (1.55°F) in the same area (T. G. F. Kittel, unpublished data). Annual precipitation in the northern Rockies rose 6% in the region (Baldwin 2003), and summer precipitation rose 29.5% (Kittel et al. 2002).

The changes are likely to affect every aspect of the physical and biotic resources of the region. Two Canadian climatologists predict zero snowpacks in the northern Rockies by 2070 (Fyfe and Flato 1999). In fact, the area occupied by glaciers today in Glacier National Park is only a third of the area of glaciers at the time of that park's establishment in 1910. The glaciers are predicted to disappear completely in about 30 years. That alpine snowpacks are beginning to melt earlier in the spring is reflected in a 10-day advance in the run-off peaks of several western streams (Baldwin et al. 2003).

If winter snowpacks no longer form (i.e., if snow changes to rain) at the YNP higher elevations, the ungulates may remain on what is now summer range throughout the year. The immediate effect would be to ease the herbivorous pressures now applied to the northern range. But the same changes would likely occur with all of the YNP elk herds that now summer at the park's higher elevations and winter at lower areas. At the same time the current constraints of winter weather and winter forage exhaustion would ease and allow population increase that could eventually impact the entire park.

The climate changes would also be likely to stimulate wide spatial shifts in vegetation types (Romme and Turner 1991). With temperature increases alone, forested zones would be expected to move upslope and reduce or eliminate alpine zones. The whitebark pine zone, a species important to grizzly bear, would be expected to shrink or be eliminated. With increase in both temperature and precipitation, the coniferous zone would be expected to expand into the lower elevation shrub steppe (Reiners 2003) and significantly change the character of the northern range.

In total, what appears to be imminent climate change would be likely to alter the ecology of the northern range profoundly. There is therefore an urgent need for extensive monitoring and well-designed research to provide an understanding of these changes and a knowledge base for future management policies.


I stated in 1 that a secondary purpose of this book (after marshaling the evidence on the truth of the northern range situation) is to test the natural-regulation ecological hypothesis. In fact, it has already been implicitly tested by many of the studies cited in the previous chapters and explicitly tested by several previous authors.

Kay's (1990) massive study falsified most of the projections on vegetation effects and competitive exclusion of other herbivores. In his closing chapter, he concludes, “Since, I can find no evidence to support any of these hypotheses and all of the available data support the opposite conclusions, I feel compelled to reject the entire ‘natural regulation’ paradigm.”

Boyce (1991) begins a discussion of natural regulation by stating “existing evidence suggests that each of these premises has been violated, and therefore, one might think that the natural regulation hypothesis should be rejected.” He next comments that each of the “premises”—for example, assuming that an ungulate population would not affect the vegetation—“is inappropriate or misleading.” By some logical turn that I don't follow, he then concludes that because the premises supporting the hypothesis are inappropriate or misleading, “no basis exists for rejecting the natural-regulation hypothesis.”

Boyce complicates the discourse further by adding such value considerations as “how many elk should be in the … Park” (emphasis added) and whether the “consequences of high elk numbers appear to be socially unacceptable.” This is an excellent example of the problem discussed in 1, the commingling of the natural-regulation ecological hypothesis, a scientific question, with the natural-regulation management policy, a value question.

What Boyce calls premises are in fact the several predictions of the hypothesis and the tenets to be tested. That he considers them “inappropriate” or “misleading” is his concession that they were improbable when they were posed, as judged by the accumulated ecological evidence and theory of the time and the existing evidence on the northern range. Subsequent research has falsified them and the hypothesis as a whole, and has confirmed the a priori probability that he conceded.

Singer et al. (1998a) state: “We conclude that the YNP natural-regulation management model was internally inconsistent because … [it] predicted almost no [vegetation] changes would occur.” The authors also critique the hypothesis for underestimating the significance of predators. I assume that the intent here is a comment on the ecological hypothesis, but it is another example of failure to make the above distinction.

The NRC study (Klein et al. 2002) also implicitly falsified the hypothesis. In the statement “vegetation changes observed in the past 130 years or so appear to have been influenced more by ungulate browsing than by climate change,” The panel in essence challenged the no-vegetation-effect tenets of the hypothesis.

(p.303) The wolf reestablishment may provide yet one more test. If the wolves succeed in significantly reducing the northern herd, and the vegetation responds by returning toward pre-1872 conditions, the hypothesis will have been falsified. As already discussed, some of those vegetation changes may already be occurring.

I listed the tenets of the hypothesis in 1, and there is no need to repeat them here. On the basis of the evidence presented in the preceding chapters, I agree with the authors that all have now been falsified except the one predicting that the elk herd would equilibrate. But as mentioned in 2, this was essentially a foregone conclusion. Moreover, equilibration occurred at higher levels than Houston's (1971, 1974) predictions. Furthermore, it has occurred significantly through usurpation of and effects on the forage resource, rather than through some self-regulatory mechanisms that would prevent vegetation effects, as implied by the hypothesis. Contrary to the hypothesis, predation has evidently played a significant role, as Singer et al. (1997b) have shown.

Beyond these population questions, the hypothesis had in essence been tested before it was posed in 1971. There was already an abundance of northern range evidence before that date attesting to its improbability. That the ensuing 33 years of research have falsified the hypothesis is thus not surprising although the park position has come around to this realization only gradually and haltingly, and even today reservedly. (p.304)