Does restoring apex predators to food webs restore ecosystems? Comment

Abstract

Trophic cascades, which represent the indirect effects of predators that extend downward through multiple trophic levels and food webs, have gained significant recognition in the scientific community (Ripple et al., 2016; Terborgh & Estes, 2010). In northern Yellowstone's ungulate winter range, or northern range, intensive herbivory by Rocky Mountain elk (Cervus canadensis) was a dominant force across the landscape during the seven decades when gray wolves (Canis lupus) were absent, suppressing the growth of deciduous woody plants and affecting the availability of resources and physical habitat for other species (Barmore, 2003; Jonas, 1955; Kay, 1990). However, following the 1995–1996 reintroduction of wolves, thus completing the park's large carnivore guild, studies of willows (Salix spp.) in the northern range reported that some of them were again growing taller due to a decrease in herbivory—a trophic cascade resulting from large carnivore restoration (e.g., Baril et al., 2011; Beyer et al., 2007).

Hobbs et al. (2024), hereafter referred to as Hobbs et al., conducted research to evaluate the effects of a restored large carnivore guild on willow dynamics by studying their heights over time in northern Yellowstone riparian zones and concluded that apex predator restoration was unsuccessful in restoring riparian willow communities. This conclusion relied on a narrow and arbitrary definition of a trophic cascade and restoration success. Specifically, they indicated that since tall willows were largely lost from the northern range within two decades following apex predator loss, a similarly abrupt recovery of tall willows following apex predator restoration was necessary for evidence of a trophic cascade and willow restoration. We demonstrate, based on published research, that a trophic cascade in northern Yellowstone is reversing the effects of large predator loss to multiple taxa and food webs. Although specific historical conditions in riparian areas have not been fully replicated or restored, important ecological processes have been changed and a trajectory of ecosystem restoration is underway.

Hobbs et al. conducted their 20-year study (2001–2020) of willows growing on riparian floodplains and terraces along streams in the park's northern range. Based on those results and their interpretation of what constitutes restoration, they developed far-reaching conclusions: “Our two decades of research revealed that the restoration of apex predators to a food web after a long absence failed to reverse the effects of their loss from the food web” (p. 18). At face value, their comments would lead readers to believe the restored large carnivore guild has had little, if any, influence on the ecosystem structure and function of northern Yellowstone's riparian areas.

Hobbs et al. (p. 27) indicated a number of studies have claimed ecosystem restoration has occurred in northern Yellowstone following the return of gray wolves, further suggesting such claims might be false or exaggerated. However, a recent evaluation of seven of these Yellowstone studies identified by Hobbs et al. found none of them had claimed that the northern Yellowstone ecosystem has been restored to a former state (Beschta et al., 2023); thus, Hobbs et al. critiqued them for something they did not assert.

While the future is unlikely to duplicate the past, northern range studies nevertheless have shown and continue to show evidence of important vegetation changes underway resulting from a trophic cascade, one that is contributing to the ecological recovery of riparian areas and other aspects of the ecosystem (see synthesis in Beschta & Ripple, 2016; Ripple & Beschta, 2012). Even so, ecosystem recovery will take time given the extent to which riparian plant communities, and their stream channels were altered by decades of intensive elk herbivory, as well as the new constraints associated with a changing climate (Beschta & Ripple, 2016) and a surging population of bison (Bison bison) (Beschta et al., 2020; Kauffman et al., 2022; Painter et al., 2023).

The primary conclusion from Hobbs et al., as well as in previous publications of their study, was that willows growing on relatively dry riparian floodplains and terraces associated with incised channels require beavers to fully restore the hydrology that allowed these plants to exist there previously. We would note that the restoration of a beaver-willow state is normally a four-stage process: (1) the historical cause of willow suppression, intensive long-term ungulate use, must be reduced; (2) willows, along with quaking aspen (Populus tremuloides) and other deciduous woody species, can then grow into tall thickets capable of supporting beavers; (3) beavers return and build dams that expand willow habitat; and (4) this expanded habitat is capable of supporting future generations of beavers. Indeed, various studies of riparian vegetation, including that of Hobbs et al., have shown that all of these stages have emerged in some parts of northern Yellowstone following the restoration of large carnivores and a reduction in elk herbivory. We contend that these changes are ecologically significant and suggest Hobbs et al. are incorrect in concluding this complex situation is a recovery failure and that the trophic cascade has been ineffective. In fact, it is the opposite.

A recent analysis evaluating changes in willow crown volume provides another example challenging the assertion of Hobbs et al. that the trophic cascades experiment in Yellowstone's northern range is a failure. In that analysis, Ripple et al. (2025) utilized the 2001–2020 willow height data of Cooper and Hobbs (2023) to estimate temporal changes in the crown volume of individual willows. Since willow crown biomass and crown volume are highly correlated (Yao et al., 2021), any increased crown volume is a reasonable surrogate for changes in biomass, which is often used to evaluate the strength of a trophic cascade. Results of the Ripple et al. (2025) analysis indicated a 1500% increase in willow crown volume (Figure 1), an effect size greater than the vast majority of trophic cascades studies summarized in a meta-analysis by Borer et al. (2005).

Hobbs et al. hypothesized that the incision of streams after the loss of beavers would prevent willow recovery even with reduced herbivory, and identified this as an alternative stable state that is resilient to apex predator restoration. The core of the Hobbs et al. study was the selection of four riparian sites with willows, historical beaver presence, and an incised channel. At each site, a control plot was established where willows were accessible to ungulates. Three treatment plots were also established (i.e., (1) fenced to prevent ungulate herbivory, (2) in-channel dams to raise water tables, and (3) fenced and dammed), treatments expected to increase willow growth rates over those of the controls. Over the course of their experiment, many willows within the treatment plots grew to heights expected for restored communities (i.e., ≥200 cm average height, see Marshall et al., 2013). By comparison, willows experiencing ambient conditions (i.e., control plots) increased less in height. Because the control sites were chosen for dry, incised conditions, it is inappropriate to use them to generally represent “ambient conditions” for willow sites in northern Yellowstone. Additional “observational sites” also were chosen based on limited criteria.

Even with these constraints on site selection, in 2020 nearly 30% of the Hobbs et al. control and observational sites had average willow heights ≥200 cm, with many of the remaining sites on an upward trajectory. Given the consistent upward trend in average willow heights for their “ambient” sites, it would appear that this 200-cm threshold would likely have been exceeded on additional sites if the study had continued a few more years. Consequently, even for these relatively dry willow sites, the conclusion of restoration failure based solely on this height threshold appears arbitrary and premature.

Hobbs et al. indicate that restoration of a degraded ecosystem can be complicated and that the successional pathway of going from elk-grassland to beaver-willow states is likely to display hysteresis, a conclusion with which we agree. Depending on the degree of impact to both vegetation and channels from decades of intensive ungulate use and the loss of beavers, ecosystem recovery may take various pathways and may not soon return to the exact conditions that were present historically. Thus, we suggest the Hobbs et al. vision that only a fully restored beaver-willow state can currently be used as the primary measure of restoration success is an overly simplistic perspective for evaluating trophic cascades effectiveness and recovery processes across the park's riparian areas, and discounts many of the cascading trophic effects of large carnivores currently underway.

The conclusion of Hobbs et al. regarding a lack of restoration success is inconsistent with observed changes occurring along Blacktail Deer Creek (Figure 1), as well as with their own data. It is also inconsistent with the extensive evidence from various vegetation field studies that followed the restoration of large carnivores. A synthesis of 24 studies of deciduous plants in northern range riparian areas that were published within two decades of wolf reintroduction (Beschta & Ripple, 2016) included studies of willow, aspen, cottonwood (Populus spp.), thinleaf alder (Alnus incana ssp. tenuifolia), and various berry-producing shrubs. All but two of these studies found increases in plant height, stem diameter, stem establishment, canopy cover, fruit presence, or recruitment. This coherent pattern of results led Beschta and Ripple (2016) to conclude that: “This early stage of vegetation recovery in northern Yellowstone, although not occurring everywhere, represents a major departure from the widespread suppression of woody plants by elk browsing that occurred in the decades when wolves were absent” (p. 1). More than half of the studies measured ungulate browsing and, in all instances, increased growth and cover of woody plants occurred concurrently with a decrease in browsing. Similarly, Hobbs et al. found a strong downturn in herbivory from elk, the dominant large herbivore in the northern range during the early years of their study, as previously suppressed willows on ambient sites began to grow taller. In general, the results of various northern range vegetation studies, including the data of Hobbs et al., consistently point to an effective large carnivore trophic cascade, one that has contributed significantly to restoring riparian plant community dynamics and processes.

The direct effects of a restored large carnivore guild upon elk and coyotes (Canis latrans) have also been documented in northern Yellowstone and the Greater Yellowstone ecosystem, as have indirect effects on pronghorn, small mammals, upland plants, stream morphology, beaver, birds, berry production, scavengers, and bears (Ripple et al., 2014). Elsewhere, in the Canadian national parks of Banff (Hebblewhite et al., 2005) and Jasper (Beschta & Ripple, 2007), deciduous plant communities previously suppressed by intensive elk herbivory have begun to recover following the return of wolves. These examples of trophic cascades underscore how, and to what extent, the restoration of a large predator guild can affect the behavior and density of prey species, thereby indirectly benefiting the establishment, growth, and recruitment of willows and other palatable species.

However, the trajectory of riparian vegetation recovery in northern Yellowstone has been complicated in recent years by the nearly fivefold expansion of its bison population, from fewer than 700 animals in 2001 to over 3400 by 2020 (National Park Service, 2024). This large, year-round bison presence represents a novel ecological factor that is increasingly contributing to the suppression of recovering willows, thus dampening the effects of the predator-elk-vegetation trophic cascade (Beschta et al., 2020; Kauffman et al., 2022; Painter & Ripple, 2012; Painter & Tercek, 2020).

One of the challenges in understanding ecosystem responses during a large predator trophic cascade is that changes in plant community dynamics can result from various factors and produce varying results. It is clear that, prior to the restoration of wolves and cougars (Puma concolor) in northern Yellowstone, high levels of browsing by elk had been limiting the ability of young willows and other deciduous plants to grow taller (Barmore, 2003; Jonas, 1955; Kay, 1990). However, this began to change when browsing levels started to decline following the return of wolves, leading to increased heights of young woody plants (Beschta & Ripple, 2016; Ripple & Beschta, 2012).

Even so, the increase in young plant heights has not been uniform across the landscape, and this variation is not apparent in a simple assessment of mean height. Regarding the Hobbs et al. study sites, in 2001 the difference in average willow heights between the 1st and 3rd quartiles was only 4 cm (Figure 2). While average heights approximately doubled over the course of their study, by 2020 the difference between the 1st and 3rd quartiles had increased to 87 cm, representing more than a 20-fold increase in the distribution range within a period of two decades. The increased variance in willow heights between sites is likely due to factors such as differing levels of elk and bison herbivory, differences in channel incision and groundwater sources, and other variables. At some sites, high herbivory may still limit willow growth, while at others, reduced herbivory has allowed taller willows and beavers to return. Thus, the trophic cascade and willow recovery in the northern range is a complex process; nevertheless, it is currently a process exhibiting a general trend of increased willow height, as well as increased variance in height, across both time and space.

Hobbs et al. concluded that the restoration of the large carnivore guild “failed” to reverse the effects of large predator loss, despite evidence to the contrary, including: (1) consistent reports of recovering riparian plant communities in other northern range studies (see synthesis in Beschta & Ripple, 2016; Ripple & Beschta, 2012); (2) an approximate doubling of willow heights on the Hobbs et al. ambient sites (i.e., unfenced and undammed) to an average of 178 cm, a significant change both statistically and ecologically; and (3) additional ecological effects identified following the return of wolves (Ripple et al., 2014).

Hobbs et al. selected their four experimental sites for measuring riparian willow height using several criteria (paraphrased here): (1) evidence of historical beaver activity, (2) presence of willows, (3) summer water tables lacking recharge from adjacent hillslopes, and (4) incised channels. They expanded their results from these four experimental sites by adding 21 randomly selected observational sites in 2009, and it appears that similar criteria were used in selecting those sites. Regarding the first criterion, Persico and Meyer (2009) found that less than 30% of mainstream reaches in the northern range had historically experienced beaver-related aggradation, thus effectively eliminating a large proportion of stream reaches from site selection. This single criterion renders the restoration endpoint of a beaver-willow state unsuitable for the majority of stream reaches in the northern range. The third and fourth criteria eliminate many additional reaches—those where groundwater inflows from adjacent hillslopes significantly influence water tables, or where channels are not deeply incised. Thus, the study design of Hobbs et al. excluded the most resilient, diverse, and productive streamside sites where willows are supported by groundwater or high-water tables, missing key elements of processes important for restoring the beaver-willow state. Once herbivory is reduced, areas that were excluded from their study can support productive willows even without beaver dams raising local water tables (Painter & Tercek, 2020).

The above concerns demonstrate significant limitations regarding the scope of inference of the Hobbs et al. study as well as the applicability of their results across northern range stream systems. Given their sweeping generalizations about the ineffectiveness of trophic cascades based solely on willow heights on the sites they selected for specific conditions, as well as their dismissal of other northern range vegetation studies, a thorough evaluation of the potential weaknesses and limitations of their study, including their scope of inference, would have been in order.

In a synthesis of carnivore restoration effects on vegetation in northern Yellowstone, Peterson et al. (2020) identified three general patterns of changing plant dynamics: (1) young deciduous woody plants have grown taller in some areas, (2) increased plant heights exhibit temporal and spatial variation due to several factors (e.g., decreased browsing, groundwater availability), and (3) reductions in elk browsing have been counteracted in some areas by increased bison use. The results of Hobbs et al. confirm a consistent upward trend in average willow heights over the duration of their study, with some sites exceeding their height threshold, thereby indicating successful willow restoration. Furthermore, the results from other studies of vegetation dynamics in both riparian areas and upland settings provide multiple lines of evidence that many northern range plant communities, in addition to willows, are on a fundamentally different trajectory than they were when wolves were absent, a trophic cascade driven by the park's restored large carnivore guild.

The authors declare no conflicts of interest.