Search USGS⌕ Search

Geology topics

Sarah Nelson Sells

Publications and source records attributed to Sarah Nelson Sells.

12 recordsLinked to original sources
USGS2026

Predicted habitat use for reintroduced grizzly bears in the transboundary North Cascades ecosystem

Grizzly bears ( Ursus arctos ) were once numerous in the North Cascades transboundary region of Washington State (United States) and British Columbia (Canada); however, few remain today. To support ongoing reintroduction evaluations, we used simulations based on movement models developed in the Northern Rocky Mountains to predict habitat use by a small founding group of grizzly bears in the North Cascades during the early stages of reintroduction. These simulations represent movements across a spatially explicit landscape based on individual-specific selection and movement parameters. We first evaluated predictive performance in three nearby populations of grizzly bears in the Squamish-Lillooet, McGillvary Mountains, and North Stein-Nahatlatch regions of the Coast Mountains, British Columbia. After applying an elevation-based calibration, predicted habitat use showed strong agreement with GPS location data from 73 collared bears. Across population–sex subgroups, Spearman rank correlations were ≥0.95, with 69.2%–83.3% of locations occurring within the top five habitat classes (covering 50% of mapped habitat) and 17.1%–36.2% in the top class (representing 10%). Overall, 24.5% and 77.9% of locations fell within the top class and top five classes, respectively. Seasonal validation showed strongest predictive performance from May to late summer or fall. Model predictions for the North Cascades indicate that habitat use during early reintroduction is likely to be concentrated in the central and northern mountainous portions of the ecosystem along the US–Canada border. These maps can guide recovery planning when no local bear data are yet available in the North Cascades transboundary region.

British Columbia, Washington↗
USGS2026

Using structured decision-making to develop a communications strategy for the U.S. Geological Survey Cooperative Research Units Program

Communication regarding the mission of the U.S. Geological Survey (USGS) Cooperative Research Units Program (CRU) can take many forms, yet clear and concise messaging for various audiences is critical to highlight program accomplishments and increase visibility. Before the work described in this report, CRU did not have a communication strategy; therefore, CRU leadership supported a structured decision-making (SDM) workshop to develop a comprehensive strategy for multiple audiences. The workshop was conducted in November 2024, in Nebraska City, Nebraska. The working group for this SDM process included CRU Program leadership, the CRU Communications Team lead, Unit scientists, a Unit administrative assistant, a representative of the Wildlife Management Institute (WMI), a member of the USGS Ecosystems Mission Area (EMA), Office of Communications and Publishing (OCAP) team, and the team lead for the CRU Program strategic planning process, as well as three facilitators who were also unit scientists as well as experts in SDM. Over the course of a week, the SDM team followed the PrOACT framework which identified the problem, objectives, alternatives, consequences, and tradeoffs to guide us toward a strategy for implementation of a set of actions for CRU communications. Results of the SDM workshop included the development of a problem statement, an objectives hierarchy, a suite of alternatives that were evaluated using a consequences table and a clear process for assessing tradeoffs among alternative communication actions and strategies. Through the evaluation of consequences of each action or campaign, the team developed both the assessment tool (for the future) and an immediate plan for communication product development and distribution. The consequences table for this problem was meant to be flexible to accommodate changes in CRU thematic priorities and can be easily updated with new objectives, measures, and alternatives. In addition, the weight placed on objectives may change as the Team moves forward; the ranking and scoring system used in the workshop can be easily updated. Overall, the working group identified three different actions or campaigns—Fact Sheets, Who Are We Campaign, and Alumni Campaign—that scored high in the prototype decision framework. However, the tradeoffs analysis indicated that each action(s) performed better on some objectives than others. The working group identified a need to therefore develop an implementation plan that is composed of individual actions that each target different objectives to potentially create a holistic and feasible communications strategy that performs well for all objectives. In addition, the SDM prototype developed a scalable, objective-based framework for effectively communication of the value and accomplishments of the CRU program.

Preprints.org↗
USGS2026

Where will the cat cross the road? Comparing camera and GPS-based models for identifying wildlife corridors

Designing effective wildlife corridors is a critical conservation challenge in fragmented landscapes. GPS-based step selection functions strongly predict dispersal corridors and connectivity, but GPS collaring can be expensive and invasive. Camera-based occupancy models are widely used for connectivity analyses but may involve trade-offs in data resolution. Despite widespread use of both approaches, few studies have directly compared them using concurrent datasets. We developed a stacked single-species, single-season occupancy model and a Circuitscape connectivity surface for mountain lions (Puma concolor) on Washington’s Olympic Peninsula, USA, and compared them with a connectivity surface from an existing integrated step selection function. Both models predicted mountain lion GPS locations well, with binned Spearman rank correlations of 1 for Circuitscape and 0.96 for the step selection function, though step selection better identified habitat use by dispersers. Connectivity predictions were moderately correlated across the landscape ( r = 0.26), but agreement was strongest in human-dominated areas most critical for corridor planning. We conclude that GPS-based approaches are advantageous when data collection is feasible and the focus is on dispersal or fine-scale movement. However, camera-based approaches may be preferable for multi-species monitoring, large spatial and temporal scales, noninvasive sampling, when resources are limited, or when fine-scale or dispersal-specific inference is not required.

Washington↗
USGS2026

Appendix 6: Overview of iPOM 2.0: Updates to wolf abundance estimation in Montana

Montana Fish, Wildlife and Parks (MFWP) uses the integrated Patch Occupancy Model (iPOM) to estimate statewide and regional wolf abundance. During the early years of recovery, wolf abundance was tracked using minimum counts. As the population increased, limitations of this method became apparent, motivating development of a model-based approach for estimating abundance. iPOM development began in 2006, was operationalized in 2013, and has been used in its current form (iPOM 1.0) since 2021. Since its inception, iPOM has been developed and presented as an adaptive modeling framework intended to evolve as new data and analytical methods become available1. Here, we briefly present ongoing work to produce the next iteration of iPOM, hereafter iPOM 2.0, and summarize major model updates and resulting preliminary wolf abundance estimates. iPOM uses models for occupancy, territory size, and pack size, along with monitoring data collected by MFWP wolf specialists and potential wolf sightings reported by deer and elk hunters during the 5-week general rifle season each Nov-Dec. These monitoring data, wolf sightings, and covariates for habitat and survey effort are used to model occupancy of packs on a statewide grid of 600-km 2 cells. The area used by wolves is then estimated as each cell’s occupancy estimate multiplied by grid cell area. iPOM estimates pack abundance by dividing this area by predicted territory sizes, and in-pack wolf abundance by multiplying pack abundance by estimated group size. This is combined with a lone wolf rate to estimate total wolf abundance (see Sells et al. 1 for a full description of iPOM 1.0). This technical summary provides a high-level overview of the proposed iPOM updates and their implications for wolf abundance estimates. The estimates presented here represent results as of June 2026 and are provided to inform upcoming discussions surrounding harvest management and regulations (e.g., season dates, quotas, methods). Results are preliminary and remain subject to final model refinement and peer review.

Montana↗
USGS2025

Investigating the influence of forest disturbance on grizzly bear habitat ecology and fitness in the Northern Continental Divide Ecosystem

The Northern Continental Divide Ecosystem (NCDE) grizzly bear (Ursus arctos horribilis) population of northwest Montana was listed as threatened in 1975 under the Endangered Species Act. Recovery objectives for this population have centered predominantly on limiting humancaused mortality and protecting habitat. Consequently, empirically informed habitat management has been an integral component of grizzly bear recovery in the NCDE and will remain a major component of future conservation and management efforts. The NCDE has a long history of forest management and wildfire activity. Studies have demonstrated both positive and negative effects of timber harvest on grizzly bears. Grizzly bears are associated with forest edges and early successional forests (Munro et al. 2006, Stewart et al. 2013), which often have higher abundances of preferable forage (Mace and Jonkel 1986, Souliere et al. 2020). Forest management activities that reinstate early successional vegetation can thus have a positive effect on forage quality for grizzly bears (Nielsen et al. 2004). However, roadbuilding and human access associated with timber harvest and silvicultural treatments can have a negative effect on grizzly bear habitat selection and survival (Kasworm and Manley 1990, Mace et al. 1996, 2012, Schwartz et al. 2010, Boulanger and Stenhouse 2014), potentially negating the positive effects of timber harvest and certain silvicultural treatments. Wildfire disturbance is likely to have variable effects on grizzly bears. Post-fire, early successional vegetation provides high quality forage, and grizzly bears likely select these postfire stands to take advantage of increases in graminoids, forbs, roots, ungulate prey, and fruiting shrubs that emerge post disturbance (Mace and Jonkel 1986, Hamer and Herrero 1987, Best et al. 2024). On larger temporal scales, historical variation in spatial patterning of infrequent, predominantly stand-replacing fires in the NCDE, has contributed to heterogeneous landscape conditions (Hood et al. 2021). Such conditions likely benefit grizzly bears by providing various landscape features required for denning, access to foods, and security. However, more frequent, and larger high severity fires are projected to increase throughout the Rocky Mountains (Abatzoglou and Williams 2016, Parks and Abatzoglou 2020, Higuera et al. 2021), which could reduce cover and impact regrowth of herbaceous vegetation (Bard and Cain 2020). Future fire activity could change the distribution of food resources across the landscape and potentially reduce expected fire benefits to grizzly bears. Our objectives are to investigate the effects of forest disturbance from timber harvest and fire on grizzly bears in the NCDE to understand conditions that promote or degrade grizzly bear habitat quality. Balancing grizzly bear conservation while managing forest disturbances is a complex and dynamic challenge for land managers. This proposed research could help address this challenge by providing information to assist managers in developing strategies that balance species conservation and forest management objectives.

Montana↗
USGS2024

Leveraging local wildlife surveys for robust occupancy trend estimation

Natural resource agencies are frequently tasked with monitoring populations of at-risk species to ensure management activities do not negatively affect the viability of wildlife populations. Typically, these monitoring efforts evaluate trends in a population’s abundance, occupancy, or geographic distribution. Often, surveys provide local information, but results are generally not incorporated into broad-scale monitoring efforts that focus on range-wide population changes due to their variable nature in both spatial extent and effort. We investigated whether aggregating these local (hereafter “variable”) surveys can generate enough statistical power to estimate broad-scale population trends using simulations of declining populations of fishers ( Pekania pennati ) over a 10-year time horizon. Our simulations included three population sizes which we refer to as abundant, common, and rare ( N 0 = 700, 350, and 100 individuals, respectively) with each declining at a rapid and moderate pace ( λ = 0.933, and 0.977, respectively). For each population, we simulated variable surveys using an occupancy framework to subsample the population with parameters that mimic combining multiple independent monitoring efforts which vary annually in location, and effort. Regardless of spatial consistency of annual sampling, there was minimal variation in statistical power under both high and low detection probability simulations. However, when sampling effort varied each year, statistical power was lower for most populations and sampling scenarios when compared to consistent sampling effort unless some baseline level of sampling effort was reliably achieved in all years. In many cases, adding low-level consistent baseline sampling to variable surveys resulted in statistical power close to that of consistent sampling efforts. Our results suggest statistical power is driven by annual consistency in the proportion of landscape sampled rather than spatial consistency in sampling locations. This result indicates that current variable surveys could be leveraged and combined to detect population declines for at-risk species at broad-scales if a baseline proportion of landscape is robustly sampled. The level of baseline sampling is highly dependent on population size and magnitudes of population change. In simulations with a common or abundant population experiencing a rapid decline, a baseline survey effort of at least 5% of the landscape in combination with variable surveys resulted in statistical power consistently above the standard threshold of 0.80 for occupancy monitoring. Leveraging existing local efforts to achieve high detection probability and baseline sampling would reduce financial and logistical burdens of broad-scale wildlife monitoring efforts.

Idaho, Montana↗
USGS2024

Predicting future grizzly bear habitat use in the Bitterroot Ecosystem under recolonization and reintroduction scenarios

Many conservation actions must be implemented with limited data. This is especially true when planning recovery efforts for extirpated populations, such as grizzly bears ( Ursus arctos ) within the Bitterroot Ecosystem (BE), where strategies for reestablishing a resident population are being evaluated. Here, we applied individual-based movement models developed for a nearby grizzly bear population to predict habitat use in and near the BE, under scenarios of natural recolonization, reintroduction, and a combination. All simulations predicted that habitat use by grizzly bears would be higher in the northern half of the study area. Under the natural recolonization scenario, use was concentrated in Montana, but became more uniform across the northern BE in Idaho over time. Use was more concentrated in east-central Idaho under the reintroduction scenario. Assuming that natural recolonization continues even if bears are reintroduced, use remained widespread across the northern half of the BE and surrounding areas. Predicted habitat maps for the natural recolonization scenario aligned well with outlier and GPS collar data available for grizzly bears in the study area, with Spearman rank correlations of ≥0.93 and mean class values of ≥9.1 (where class 10 was the highest relative predicted use; each class 1–10 represented 10% of the landscape). In total, 52.4% of outlier locations and 79% of GPS collar locations were in class 10 in our predicted habitat maps for natural recolonization. Simulated grizzly bears selected habitats over a much larger landscape than the BE itself under all scenarios, including multiple-use and private lands, similar to existing populations that have expanded beyond recovery zones. This highlights the importance of recognizing and planning for the role of private lands in recovery efforts, including understanding resources needed to prevent and respond to human-grizzly bear conflict and maintain public acceptance of grizzly bears over a large landscape.

Idaho, Montana↗
USGS2024

Wolf harvest management strategy evaluation: Annual Report, 2024

Wolf harvest season setting is complicated and controversial. State law requires Montana Fish, Wildlife and Parks (MFWP) to both reduce the wolf population and avoid federal relisting under the Endangered Species Act (Montana Fish, Wildlife and Parks, 2002). Disparate stakeholder groups each have different objectives for wolf management. For instance, big game advocates want to see improved big game populations and hunting opportunities in northwest Montana, while wolf advocates want to see regulations that minimize wolf mortality. Decision making about season setting tries to balance these objectives. Wolf hunting and trapping season decisions are made by the Montana Fish and Wildlife Commission and are informed by annual wolf abundance estimates from an integrated patch occupancy model (iPOM, Sells et al., 2022c) as well as the predictions of wolf abundance into the future under potential constant harvest levels. Parametric uncertainty (uncertainty surrounding the value of a parameter) from the iPOM estimates is propagated through to future projections, providing the Commission with plausible and worst-case outcomes of different levels of public harvest over the short term, i.e., five years into the future, on the wolf population in Montana (Parks et al., 2024). An alternative approach to inform wolf management and harvest decisions is through adaptive management. Adaptative management is appropriate for decisions that are made iteratively and when monitoring data are collected to learn about the outcomes from decisions, where monitoring data help to reduce critical uncertainties regarding ecosystem function or management outcomes (Walters, 1986; Williams, 2011). Management strategy evaluation (MSE) is one way to develop an adaptive management framework. MSE was developed by fisheries managers and scientists to more accurately and fully incorporate various forms of uncertainty, consider long-term time horizons, and add more transparency in a fisheries context (Punt et al., 2016). It has been used routinely and has become a standard approach for complicated and contentious marine fisheries management situations, yet it has been underutilized in wildlife management (but see Bunnefeld et al., 2013, 2011). MSE is a forward simulation approach for testing prospective management options or strategies over a wide range of possible states (Punt et al., 2016). A MSE framework captures the ‘truth’ or what is happening in the system (termed the ‘the operating model’) and the information available to the decision makers (termed ‘the estimation model’ or ‘management strategy’). More precisely, there are four main processes modeled. First, models are constructed based on current understanding and data to represent ‘truth’. Second, the collection of monitoring data is simulated from the ‘truth’ model. Third, the simulated monitoring data are fit to an estimation model and the next time step’s population metrics are predicted from the estimated parameters. Fourth, based on the estimation model results and the predictions, the decision-making process is simulated following a management strategy, whereby a decision is made and the implementation of this decision feeds back into the ‘truth’ model (Figure 1). This process continues through time. Additionally, each simulation through time is repeated to capture the full range of stochasticity and uncertainty.

Montana↗
USGS2023

Grizzly bear movement models predict habitat use for nearby populations

Conservation planning and decision-making can be enhanced by ecological models that reliably transfer to times and places beyond those where models were developed. Transferrable models can be especially helpful for species of conservation concern, such as grizzly bears ( Ursus arctos ). Currently, only four grizzly bear populations remain in the contiguous United States. We evaluated transferability of previously derived individual-based, integrated step selection functions (iSSFs) developed from GPS-collared grizzly bears in the Northern Continental Divide Ecosystem by applying them within the nearby Selkirk (SE), Cabinet-Yaak (CYE), and Greater Yellowstone Ecosystems (GYE). We simulated 100 replicates of 5000 steps for each iSSF in each ecosystem, summarized relative use into 10 equal-area classes for each sex, and overlaid GPS locations from bears in the SE, CYE, and GYE on resulting maps. Spearman rank correlations between numbers of locations and class rank were ≥ 0.96 within each study area, indicating models were highly predictive of grizzly bear space use in these nearby populations. Assessment of models using smaller subsets of data in space and time demonstrated generally high predictive accuracy for females. Although generally high across space and time, predictive accuracy for males was low within some watersheds and in summer within the SE and CYE, potentially due to seasonal effects, vegetation, and food assemblage differences. Altogether, these results demonstrated high transferability of our models to landscapes in the Northern Rocky Mountains, suggesting they may be used to evaluate habitat suitability and connectivity throughout the region to benefit conservation planning.

Idaho, Montana, Washington, Wyoming↗
USGS2023

Predicted connectivity pathways between grizzly bear ecosystems in western Montana

Habitat and corridor mapping are key components of many conservation programs. Grizzly bear populations in the continental US are fragmented and connectivity among federal recovery areas is a conservation goal. Building on recent work, we modeled movements to predict areas of connectivity, using integrated step selection functions (iSSFs) developed from GPS-collared grizzly bears (F = 46, M = 19) in the Northern Continental Divide Ecosystem (NCDE). We applied iSSFs in a >300,000 km 2 area including the NCDE, Cabinet–Yaak (CYE), Bitterroot (BE), and Greater Yellowstone (GYE) Ecosystems. First, we simulated directed movements (randomized shortest paths with 3 levels of exploration) between start and end nodes across populations. Second, we simulated undirected movements from start nodes in the NCDE, CYE, or GYE (no predetermined end nodes). We summarized and binned results as classes 1 (lowest relative predicted use) – 10 (highest relative predicted use) and evaluated predictions using 127 outlier grizzly bear locations. Connectivity pathways were primarily associated with mountainous areas and secondarily with river and stream courses in open valleys. Values at outlier locations indicated good model fit and mean classes at outlier locations (≥7.4) and Spearman rank correlations (≥0.87) were highest for undirected simulations and directed simulations with the highest level of exploration. Our resulting predictive maps can facilitate on-the-ground application of this research for prioritizing habitat conservation, human-bear conflict mitigation, and transportation planning. Additionally, our overall modeling approach has utility for myriad species and conservation applications.

Montana↗