Anthropogenically protected but naturally disturbed: a specialist carnivore at its southern range periphery

Forests provide the fabric of goods and services required by natural ecosystems and human societies. Yet, forests that will provide ecological and human welfare now and in the future are declining globally due to complex and varied interactions involving climate change and associated drought, harvest and deforestation, urbanization, disease and insect pathogens, and other natural and anthropogenic disturbances (Millar and Stephenson 2015; Seidl et al. 2017; Sommerfeld et al. 2018; Biedermann et al. 2019; Steel et al. 2022). For example, persistent warming and drying trends from climate change contributed to a doubling of area burned in the western continental United States from 1984 to 2015 (Abatzoglou and Williams 2016) and to an eightfold increase in areas burned at high severity (Parks and Abatzoglou 2020). Forest management and fire suppression actions may have compounded this impact through increased forest homogenization (Hagmann et al. 2021; Hessburg et al. 2021), whereas these same actions may enhance landscape restoration and resilience depending on field prescriptions (Stephens et al. 2018; Fettig et al. 2019). In addition, disturbance from outbreaks of tree-killing bark beetles (Coleoptera: Scolytinae) strongly impact forest structure and composition at broad scales, especially given how climate change impacts basic ecological drivers that regulate insect populations (Bentz et al. 2010; Seidl et al. 2017; Rodman et al. 2021). Finally, forest management is a significant source of forest disturbance as well (Gauthier et al. 2015; Kuuluvainen and Gauthier 2018) with harvest effects on forest structure that range from those somewhat similar to insect outbreaks (e.g., uneven-aged harvest) to outcomes more closely resembling fire (e.g., complete overstory removal via clearcut; Graham & Jain 1998; Savilaakso et al. 2021). In general, individual harvest disturbances impact forests at smaller scales than fire or insect outbreak (Smith 2000).

Forest disturbance is compounded by an expanding human footprint that includes high human density, land transformation (e.g., agriculture, urbanization, transportation systems, hydrologic modification), power infrastructure (e.g., sustainable and non-sustainable energy development) and other interrelated factors (Sanderson et al. 2002). Human impacts are also heightened by the demand for nature-based recreation that has increased sharply across the same natural landscapes required for conservation of species dependent on public lands in the United States (USA) (Larson et al. 2016).

Forests in the Southern Rocky Mountains of the western USA (interchangeably, “Southern Rockies”) support ecosystems with diverse natural disturbance regimes from fire and insects as well as anthropogenic caused disturbance. In general, increased temperatures from climate change in Rocky Mountain forests result in increased rain as winter precipitation, earlier snowmelts, and increased severity and duration of the fire season (Rocca et al. 2014). The historical fire regime in subalpine forests is characterized by infrequent but extensive stand-replacing fires that result in high tree mortality (Agee 2000; Schoennagel et al. 2004; Sibold et al. 2006). However, there has been a decades-long trend of increased fire in the Rocky Mountains since 1984, and fire is currently occurring at a higher rate than at any time over the past two millennia (Higuera et al. 2021). Heightened levels of fire disturbance were especially acute in 2020 across the western United States with 2.5 million ha burned in some of the largest recorded fires (Higuera and Abatzoglou 2020). Additionally, 39.5% (10,256 km²) of montane and subalpine forests across the Southern Rockies were impacted by tree-killing bark beetles with 19.3% of stands experiencing > 70% tree mortality from 1997 to 2019 (Rodman et al. 2021). Colorado, which includes most of the Southern Rockies, also has a rapidly expanding human footprint and supports the 9th fastest growth rate of human populations in the US from 2022 to 2023 (U. S. Census Bureau 2023). In addition, Colorado supports 28 developed ski resorts that are located primarily on public lands administered by the U.S. Forest Service, and the state experienced the highest economic contribution in the USA from winter sports that generated $2.56 billion in annual economic value and 43,000 jobs in 2016 (Hagenstad et al. 2018). However, climate impacts from mid to late century may reduce snow-related visitations from 40 to 60% (Parthum and Christensen 2022).

The challenge of managing these complex patterns of natural and human-caused disturbance is especially daunting in the western USA where forest managers must also promote the conservation of snow-dependent species like Canada lynx (Lynx canadensis; interchangeably “lynx”) in high elevation, subalpine forests. For these species, extant patches of high quality habitat that resist climate impacts through elevation and topography provide “stepping stones” of refugia that enhance the persistence of vulnerable species during periods of severe and extensive disturbance (Michalak et al. 2020; Morelli et al. 2020). As pressures mount on forested ecosystems, understanding the distribution, extent, and status of habitat refugia for species is fundamentally important for their continued existence with climate change (Keppel et al. 2012; Krawchuk et al. 2020; Morelli et al. 2020). An important initial step in identifying disturbance and climate refugia for vulnerable species is to delineate the current extent and configuration of high-quality habitat within a disturbance context.

From 1999 to 2006, Colorado Parks and Wildlife (CPW) reintroduced 218 wild-caught Canada lynx into Colorado with the goal of establishing a viable population at the species’ southern-most range periphery in the Southern Rockies (Devineau et al. 2010). All lynx currently present in the Southern Rockies are the result of this reintroduction. As a federally listed species in the contiguous USA under the U.S. Endangered Species Act (ESA; Federal Register. 2000. Vol. 65, No 58, 16052–16086), land managers must consider the conservation of Canada lynx populations in the design and implementation of management actions. Lynx in both northern boreal forests and populations at the range periphery preferentially use boreal or subalpine forests with high horizontal forest cover (e.g., conifer boughs at the snow or ground surface) (Mowat et al. 2000; Koehler et al. 2008; Squires et al. 2010; Simons-Legaard et al. 2013; Ivan and Shenk 2016); forests with high horizontal cover support populations of snowshoe hare (Lepus americanus), the primary prey of lynx (Hodges 2000; Mowat et al. 2000; Squires and Ruggiero 2007; Ivan and Shenk 2016). In the western USA, lynx preferentially use mosaics of spruce-fir forests dominated by mature and advanced regenerating (≈ 20–40 yr old trees) structure classes (Holbrook et al. 2017). Canada lynx may also use landscapes disturbed by large-scale fire (Vanbianchi et al. 2017) and forest insects (Squires et al. 2020), but with time-lags that depend on the nature of disturbance and management actions (Holbrook et al. 2018, 2019; Olson et al. 2023). Lynx also exhibit functional (Squires et al. 2019) and behavioral responses to motorized and non-motorized winter recreation, including an avoidance of high recreation-related disturbance at developed ski resorts (Olson et al. 2018).

Canada lynx in the Southern Rockies provide a valuable case study of managing a sensitive species in landscapes that are increasingly structured by complex disturbance patterns accelerated by climate change. Here, we model the distribution of lynx habitat in the Southern Rocky Mountains of the western United States based on GPS telemetry and environmental covariates that we validated with independent lynx detections (i.e., den locations, camera photos, telemetry data collected a decade prior to this study). We then spatially delineated a lynx habitat threshold that captures 95% of testing data to represent the extent and spatial configuration of high-quality lynx habitat in the Southern Rocky Mountains. Next, we determined how patterns of disturbance from fire, forest insects, past active forest management (e.g., tree harvest, thinning, controlled burning), and urbanization and ski resort development may overlap current lynx habitat, as well as determined how administratively protected landscapes, such as designated Wilderness areas, roadless areas, and national parks and monuments, relate spatially to lynx habitat. Thus, we evaluated how natural and human-caused disturbance overlaps high-quality, in situ habitat required by lynx to identify areas of highest conservation concern for management, as a foundation for the recognition of potential lynx refugia that resist disturbance and climate impacts.

For this research, we predicted the following: (1) lynx habitat in the Southern Rockies would consist of relatively small and spatially isolated patches given the complex mountain topography; (2) natural disturbance from fire and insect outbreaks are primary ecological forces that are impacting the current spatial and temporal configuration of lynx habitat; and (3) the expanding human footprint from urbanization, ski-resort development, and forest management is reducing the spatial extent of lynx habitat, and, as such, protected landscapes offer an important buffer to human-caused direct impacts.

The distribution of Canada lynx habitat at the species’ range periphery in the Southern Rocky Mountains of the western U.S. is spatially restricted to relatively small landscape patches of southwestern and central Colorado, within a larger matrix of unlikely habitat. Consistent with our original prediction, the Likely habitat (5,727 km²) was restricted to narrow bands of subalpine forests arranged in convoluted shapes, due to the complex mountain topography and the narrow elevation gradient that supports the subalpine forests preferred by lynx and their primary prey, snowshoe hares. The quality of lynx habitat followed a general pattern of good to poor from southern to northern extents of the Southern Rocky Mountains in western CO. The San Juan Mountains at the near southern-most extent of the Southern Rockies supported the highest quality habitat (e.g., the largest blocks of Likely habitat and Core habitat) characterized by cool moist subalpine forests dominated by subalpine fir and Engelmann Spruce. Habitat quality then declined to the north as dry forest became more dominant, comprised mostly of climax lodgepole pine or pine-dominated mixtures of lodgepole-spruce-fir forests.

The subalpine forests that sustain Canada lynx have been structured by natural disturbance from fire and insect outbreaks for millennia (Viereck 1973; Price and Apps 1995; Agee 2000). In recent years, these disturbance agents have increased in impact, particularly in extent and severity, associated with climate change (Bentz et al. 2010; Abatzoglou and Williams 2016; Parks and Abatzoglou 2020; Johnson and Haynes 2023). Thus, the pressing challenge for lynx in the Southern Rockies is the extent, frequency, and severity of impacts on habitat from these disturbances in a relatively short time frame of decades. For any given year, populations of lynx require habitat that promotes their prey species, snowshoe hare and red squirrels (Tamiasciurus hudsonicus). Current undisturbed patches of habitat and those previously disturbed but with vegetative recovery (e.g., > 25 years post-disturbance; Kosterman et al. 2018; Holbrook et al. 2019; Olson et al. 2023) will be the stronghold of lynx habitat  in the coming decades. With current trends in temperature, precipitation, and drought linked to climate change, the main concern for lynx in the Southern Rockies is the persistence of these relatively few, and somewhat isolated, patches of habitat within a time window that is sufficient to have demographic, and thus, conservation consequences (Meddens et al. 2018; Morelli et al. 2020). A natural question, then, is can we implement forest- and fire-management frameworks in and around lynx habitat  to instill habitat mosaics that promote resistance to disturbance? If so, what might that framework look like? Our work presents a static depiction of lynx habitat relative to disturbance agents, which provides the important fundamental information to guard against shifting baseline syndrome (Soga and Gaston 2018). Spatial depictions of Likely, in situ habitat are also needed to focus management and conservation actions in areas most apt to be effective. Thus, the next step for lynx conservation in the Southern Rockies is to understand the efficacy and ecological constraints associated with forest and fire/fuels management strategies that promote the dynamic spatial assortments of relatively small and isolated patches of Likely habitat mosaics that sustain lynx populations (Holbrook et al. 2019) across space and time in ways that are necessary for persistence within a backdrop of human-caused climate change.

Currently, natural and human-caused disturbance overlaps 37% of likely lynx habitat in the southern Rockies, and 24% of Core (Table 2). This result supported our prediction that disturbance factors are important considerations in conservation planning for the species, but the impact of specific disturbance elements to lynx habitat varied. Natural disturbance was the primary factor that overlapped current lynx habitat, whereas the spatial overlap of human-caused disturbance to lynx habitat was low. However, we recognize that natural and anthropogenic disturbances impact lynx habitat at different spatial and temporal scales, and to different degrees. For example, disturbance to forest structure and composition from fire and insects recover through successional processes, albeit over potentially very long-time frames, while human disturbance from developed recreation at ski resorts and urbanization from building structures represent permanent habitat loss.

Spruce beetle outbreaks were the primary disturbance factor (25%) overlapping Likely lynx habitat (Table 2; Appendix 5) and were most prominent in the south-most extent of the Southern Rockies. However, the legacy forests following insect outbreaks tend to retain an understory that provides high horizontal cover (conifer boughs touching the snow or ground surface) sufficient to support adequate snowshoe hare abundance (Ivan et al. 2018, 2023). Subalpine forests in the coming decades may shift increasingly to subalpine fir following beetle disturbance (Rodman et al. 2022) that may benefit snowshoe hare and lynx populations, given the positive association of both species to subalpine fir in the forest understory (Squires et al. 2020; Ivan et al. 2023). Canada lynx in the Southern Rockies preferentially select locations with higher proportions of beetle-killed trees in insect-impacted landscapes across spatial scales (Squires et al. 2020). The distribution of Canada lynx in the Southern Rockies has changed little before and after large-scale spruce beetle outbreaks primarily due to keystone habitat features, such as high horizontal forest cover that promote habitat suitability (Squires et al. 2022). Thus, despite high spruce-beetle overlap with Likely lynx habitat, these landscapes still provide the forest structure necessary for continued occupancy.

An important caveat to how beetle disturbance may impact lynx long-term is that red squirrels (Tamiasciurus hudsonicus), which provide > 70% of the lynx diet during periods of low hare abundance (Ivan and Shenk 2016), are negatively impacted by changes in forest structure from spruce-beetle outbreaks (Ivan et al. 2018, 2023). Thus, potentially for decades, Canada lynx are at heightened risk from reduced prey density, starvation, and potential increased dispersal movements until spruce-fir forests attain an age following insect outbreaks that cone-bearing trees are sufficient to support squirrel populations as alternative prey (Ivan et al. 2018, 2023; Squires et al. 2020, 2022). Area of forests impacted annually by spruce beetles in Colorado has steadily dropped from a peak of around 200,000 ha during 2014 to 12,000 ha in 2022 (Colorado State Forest Service 2022), and there are relatively few unimpacted areas into which the irruption can spread. Thus, effect of disturbance due to bark beetles on lynx habitat is likely waning, and most spruce-fir stands in the Southern Rockies will be in a recovery phase from forest insects over the next several decades.

We predicted a priori that fire would represent a primary disturbance factor impacting the extent and configuration of lynx habitat across the Southern Rockies. Our opinion was shaped by the fact that three of the largest fires recorded in Colorado (e.g., Cameron Peak—84,545 ha, East Troublesome—78,433 ha, and Pine Gulch—56,254 ha) burned during the 2020 fire season (Higuera and Abatzoglou 2020). However, burns from 1990 to 2022 overlapped just 5% of Likely habitat, with large-scale fires mostly at elevations below lynx occupancy and in Unlikely habitat that occurred primarily in northern portions of the Southern Rockies. However, the West Fork Fire Complex (Papoose, West Fork, and Windy Pass fires) burned at high severity across 442 km² of the San Juan Mountains in 2013 (MTBS Project 2022), impacting one of the most important patches of lynx habitat in the Southern Rockies. This wildfire illustrates the potential vulnerability of lynx habitat to fire impacts, given the limited spatial extent of lynx across the Southern Rockies (Fig. 3). In general, lynx avoid fire impacted landscapes for at least ~ 25 yrs (Olson et al. 2023), likely because stand-replacing fires of high severity that are common in subalpine systems reset much of the impacted area to a stand initiation stage (Agee 2000; Schapira et al. 2021). In the North Cascades Ecosystem of Washington, USA, fires in 2013 and 2020 burned 32% of lynx habitat, and fire disturbance from 2000 to 2020 reduced estimated carrying capacity of lynx by 66–73% despite active fire suppression (Lyons et al. 2023). That said, Canada lynx do use burns across spatial and temporal scales (Vanbianchi et al. 2017; Olson et al. 2023), including immediately after fire disturbance depending on site conditions. The degree that fires impact the persistence of populations is dependent on extent, frequency, and severity of disturbance events, given that lynx demography and resource-use are impacted by mosaics of forest structure and pattern at the range periphery (Holbrook et al. 2017, 2019; Kosterman et al. 2018). Therefore, despite low current overlap, a central conservation issue for lynx and forest management in the Southern Rockies is how to “defend” Likely, in situ habitat from frequent fire disturbance with climate change.

Like large carnivores (Carter and Linnell 2016), Canada lynx in the Southern Rockies occupy a mosaic of protected and multiple-use lands, mostly administered by the U.S. Forest Service (83.5%, 4811 km²). Protected areas in fragmented landscapes can promote larger patches of forests in ways that support species occurrence and conservation (Timmers et al. 2022). As originally predicted, both multiple-use lands and protected landscapes are central to the conservation of Canada lynx at the range periphery. Yet, the high degree (62% overlap) that protected areas insulated Likely and Core habitat from direct human disturbance (i.e., ski resort development, urbanization) was unexpected, given that many protected areas in the Southern Rockies include high-elevation landscapes dominated by alpine vegetation that lynx avoid. The interstitial areas between patches of Likely habitat are subject to permanent alteration from housing/building urbanization and ski-resort development (Fig. 6). Of protected landscapes that overlapped Likely lynx habitat, 21% of these lands (1,191 km²) are administered by the U. S. Forest Service as “roadless” areas (Appendix 5). Thus, the management of roadless areas in ways that enhance the conservation of lynx is important at the southern-most range periphery. As the name implies, “roadless” areas represent landscapes where the emphasis is to minimize new road construction. However, managers still have some flexibility in the management of roadless areas, such as tree harvest and potentially controlled burning, if done in the context of maintaining or restoring habitat for threatened or endangered species under the ESA. Therefore, active management in protected landscapes might reduce the risk of large-scale natural disturbance, especially large-scale fire, if consistent with the landscape mosaics that sustain lynx occupancy (Holbrook et al. 2017, 2019; Kosterman et al. 2018).

At the population level, movement paths for lynx in the Southern Rockies extended along the western Continental Divide through the Sawatch, Mosquito, and Front Ranges of western Colorado (Buderman et al. 2018). This population-level corridor crosses several important patches of Likely lynx habitat. Dispersing lynx often traverse large portions of non-typical habitat (i.e., xeric shrublands, lodgepole pine) during long-distance movement (Buderman et al. 2018). In addition, despite a general avoidance of developed ski resorts, lynx in the Southern Rockies can traverse these high-disturbance recreation sites at night and during snow-free periods (Olson et al. 2018). Finally, lynx in the Southern Rockies also readily cross linear features on the landscape, like 2-lane highways and on occasion, 4-lane interstate highways with high traffic volumes (Baigas et al. 2017). Therefore, we believe that connectivity among the relatively small and isolated patches of habitat within protected and multiple-use landscapes is intact at the population-level with current levels of human disturbance (Fig. 6). An important caveat is that higher human density in the future could curtail or block connectivity movements for Canada lynx. We demonstrated that lynx exhibit some avoidance behavior to building structures (Appendix 2) and anecdotal observations from the Northern Rocky Mountains suggest that lynx circumvent rather than cross suburban-type neighborhood developments (J. Squires, personal communication, 2023). Thus, future expansion of human-caused disturbance from urbanization and resort development in the relatively low-elevation valleys that intersect patches of lynx habitat in the Southern Rockies may threaten population connectivity and persistence, but that disturbance threshold is unknown.

The conservation insights gained from our overlap analyses are heavily reliant on the predictive ability of the underlying SDM model. While SDM models are a common method to estimate habitat suitability (Elith and Leathwick 2009) with modeling methods that are well-studied (Elith et al. 2006; Watling et al. 2015) and easy to implement (Thuiller et al. 2009), rigorous validation of their results is frequently lacking (Sofaer et al. 2019). This is often because small sample sizes preclude withholding testing data or, more importantly, there is a lack of independently collected test data. Spatially independent data collected from other data sources or in other areas provide a rigorous test of model performance when applied to common model assessment metrics such as AUC or other measures of model discrimination (Fourcade et al. 2018). A rich independent dataset thus provides this work, and a similar model we created for the Northwest Rocky Mountains (Olson et al 2021), with a rare opportunity for rigorous model testing and validation. Using independent data and multiple model assessment metrics, we were able to discriminate among several models based on predictive performance as well as to demonstrate that the predictions of our three different model hypotheses were highly spatially consistent, which increased our confidence in model outputs. Further, the different types of independent data, including different seasons and behaviors such as denning, long-distance movements, and movements within territories, allowed us to better understand the predictive strengths and weaknesses of our model to future field applications (Smeraldo et al. 2018). We also stress the importance of collaboration across different agencies and conservation organizations in assembling independent datasets, as was done for this study, since independent data is often expensive and time-consuming to collect. Furthermore, this study and Olson et al. (2021) benefited from data that were collected consistently across a long time period, primarily for other projects, which was then able to be fortuitously used for model testing. The conservation importance of collecting this type of multiple-use long-term data is high, especially given current rapidly changing environmental conditions due to climate change and human encroachment.

In conclusion, the combined SDM from this study for the Southern Rockies, in conjunction with Olson et al. (2021) for the Pacific Northwest, provides updated spatial understandings of the distribution of habitat required by Canada lynx across the species’ range periphery in the western United States. These remapping efforts are important because they focus conservation planning and management actions in areas that spatially and temporally have the greatest impact on the persistence of sensitive species, like Canada lynx at the range periphery. This is particularly important for the conservation and management of lynx in the continental USA because habitat was broadly mapped with limited data in 2000 when the species was federally listed under ESA (McKelvey et al. 2000; Interagency Lynx Biology Team 2013). In addition, Canada lynx at the time of federal listing were just being reintroduced into the Southern Rockies (Devineau et al. 2010). Over the intervening two decades, considerable research on the ecology of Canada lynx and key prey in the Southern Rockies justifies new models that spatially delineate the extent and pattern of lynx habitat across the region. We conclude that Canada lynx in the Southern Rockies depend on habitat that is small and relatively spatially isolated, with complex configurations determined by mountainous topography with little contiguous core habitat. We also conclude that natural disturbance, primarily from spruce beetles, has the greatest spatial overlap with lynx habitat. However, despite current low overlap, fire is likely the greatest threat to lynx habitat over a decades-long timeframe, given the current increase in fire size, severity, and frequency from climate change. Therefore, management approaches for Canada lynx in the Southern Rockies will require an integration of disturbance ecology with the species’ ecological requirements. Canada lynx, and their primary prey snowshoe hares, require subalpine forests with high horizontal cover, which are susceptible to large-scale fire. Therefore, an emerging challenge for lynx conservation in the Southern Rockies is to address the ecological risk of large scale disturbance in ways that sustain habitat mosaics (Holbrook et al. 2017, 2019; Kosterman et al. 2018) across spatial and temporal scales within the complex assortment of multiple-use and protected landscapes. Given that large-scale fire in subalpine forests are stand-replacing events driven by complex interactions of climate, topography, and fuel (Andrus et al. 2016; Hagmann et al. 2021; Prichard et al. 2021), fire management that reduces risk to these cool, moist forests is less about within-stand tree characteristics and more about instilling landscape-scale patterns across north and south aspects, forest openings, and moisture gradients from vegetation that reduce fire spread (Hessburg et al. 2015, 2019; Prichard et al. 2021). Finally, we conclude that although current human-caused disturbance (tree harvest, resort development, urbanization) overlaps little with Likely lynx-habitat in the southern Rockies, future permanent habitat loss from heightened development could limit population connectivity between habitat patches, especially in the southern and central portions of the Southern Rockies.