Sarah N. Sells Michelle McLellan Jason I. Ransom Andrea Lyons Mackenzie Clarke Clayton D. Apps 20260323 raster digital data https://doi.org/10.5066/P13YTTEJ This data release provides raster datasets of relative predicted habitat use for grizzly bears (Ursus arctos) reintroduced to the transboundary North Cascades Ecosystem (NCE) of Washington State (United States) and British Columbia (Canada). The datasets represent predicted post-release habitat use by a small founding group of translocated grizzly bears during the early stages of recovery. Predicted habitat maps were generated using individual-based movement simulations parameterized with integrated step selection functions developed for grizzly bears in the Northern Continental Divide Ecosystem, a likely source population for future translocations. Model transferability was evaluated using GPS collar data from nearby Canadian populations, and an elevation-based calibration was applied to reduce extrapolation into low-elevation environments not represented in the training data. This release includes uncalibrated (“no calibration”) and elevation-calibrated predicted habitat maps for the NCE, provided for both sexes combined and separately by sex. All datasets are summarized as raster layers of relative predicted habitat use, binned into ten equal-area quantile classes (1 = lowest, 10 = highest). These spatial data products are intended to support early-stage recovery planning, including evaluation of potential release areas and anticipation of zones where human–bear interactions may be more likely. The purpose of this data release is to provide spatially explicit predictions of relative habitat use for grizzly bears (Ursus arctos) reintroduced to the transboundary North Cascades Ecosystem (NCE). Because no resident grizzly bear population currently occupies the NCE, recovery planning must proceed in the absence of local movement data. These datasets are intended to support early-stage recovery planning by identifying areas where reintroduced bears are predicted to concentrate activity relative to surrounding landscapes during the initial period following release. Predicted habitat maps were generated using individual-based movement simulations parameterized with habitat selection models developed for a likely source population in the Northern Continental Divide Ecosystem. Model transferability was evaluated using GPS collar data from nearby Canadian populations, and an elevation-based calibration was applied prior to producing the final NCE predictions. The resulting raster datasets are intended for conservation and management applications such as evaluation of potential release areas, anticipation of zones where human–bear interactions may be more likely, and prioritization of mitigation and coexistence planning. These data represent relative predicted habitat use under the modeling assumptions described in the associated publication and are not deterministic forecasts of bear presence, occupancy, or population size. 20260215 ground condition Complete None planned Washington; British Columbia -122.7189 -119.0893 50.4948 47.0010 ISO 19115 Topic Category biota None connectivity grizzly bears Ursus arctos movement ecology integrated step selection functions North Cascades Ecosystem Northern Continental Divide Ecosystem USGS Thesaurus bears natural resource management habitat use reintroduction USGS Metadata Identifier USGS:6968fe5ed4be0249f340d7f9 None Washington State British Columbia North Cascades Ecosystem Northern Continental Divide Ecosystem Canada United States Ursus arctos Ursus arctos Species Ursus arctos Grizzly bear None. Please see 'Distribution Info' for details. None. Users are advised to read the dataset's metadata thoroughly to understand appropriate use and data limitations. Sarah N Sells U.S. Geological Survey, ECOSYSTEMS Assistant Unit Leader, Research Ecologist mailing address

Natural Science Bldg, Univ of MT

Missoula MT 59812 US 406-243-4393 406-243-6064 ssells@usgs.gov US Geological Service, Montana Cooperative Wildlife Research Unit, University of Montana. Validation site data was loaned by coauthors M. McLellan (Victoria University of Wellington) and C. Apps (Aspen Wildlife Research), along with S. Rochetta (Retired WLRS) and B. McLellan (Retired MFLNRO). Additionally, N. Bickerton (WLRS) assisted in collating data, and A. Hamilton, L. Casper, D. Brown and S. Senger (members of former Lillooet Grizzly Bear Working Group) contributed to initial program development for NSN data and part of the MM data. Contribution of the SL and part of MM data was granted by C. Apps and S. Rochetta. Data from the SL, MM, and NSN were collected on the traditional territories of the Lil’wat, St’at’imc, Sechelt, and Squamish Peoples. Windows 10; Program R version 4.5.2; R Studio version 2025.09.2. NCE_predicted_habitat_no_calibration_both_sexes.tif (107 kb) NCE_predicted_habitat_no_calibration_female.tif (119 kb) NCE_predicted_habitat_no_calibration_male.tif (129 kb) NCE_predicted_habitat_elevation_calibrated_both_sexes.tif (104 kb) NCE_predicted_habitat_elevation_calibrated_female.tif (113 kb) NCE_predicted_habitat_elevation_calibrated_male.tif (125 kb) Associated layer file version (.lyrx) are each 12 kb in size. Sarah N. Sells Cecily M. Costello Paul M. Lukacs Lori L. Roberts Milan A. Vinks 202212 publication Biological Conservation vol. 276 n/a Elsevier BV ppg. 109813 https://doi.org/10.1016/j.biocon.2022.109813 Sarah N. Sells Cecily M. Costello Paul M. Lukacs Frank T. van Manen Mark Haroldson Wayne Kasworm Justin Teisberg Milan A. Vinks Dan Bjornlie 202303 publication Biological Conservation vol. 279 n/a Elsevier BV ppg. 109940 https://doi.org/10.1016/j.biocon.2023.109940 Sarah N. Sells Cecily M. Costello Paul M. Lukacs Lori L. Roberts Milan A. Vinks 20230901 publication https://doi.org/10.1016/j.biocon.2023.110199 Sarah N. Sells Cecily M. Costello 20240725 publication Sarah N. Sells Michelle McLellan Jason I. Ransom Andrea Lyons Mackenzie Clarke Clayton D. Apps 20260215 publication No formal attribute accuracy tests were conducted. No formal logical accuracy tests were conducted. Data set is considered complete for the information presented, as described in the abstract. Users are advised to read the rest of the metadata record carefully for additional details. A formal accuracy assessment of the horizontal positional information in the data set has not been conducted. A formal accuracy assessment of the vertical positional information in the data set has either not been conducted, or is not applicable. Sells et al. (2022) developed 65 predictive integrated step selection functions (iSSFs) for 46 female and 19 male grizzly bears in the NCDE. Refer to Sells et al. (2022) for methods used to develop the iSSFs. These iSSFs provided the foundation for our present data release. As in Sells et al. (2022) and Sells et al. (2023a, 2023b), each bear’s iSSF was applied to the study landscape at a 300-m cell resolution to create a conductance surface for that individual bear. Conductance values were calculated as exp(βxᵢ), where β is the coefficient vector of the estimated iSSF and xᵢ is the vector of habitat covariates for cell i. Extreme values were trimmed using the 0.025 and 0.975 quantiles and remaining values were normalized to a 0–1 scale. 20240301 We simulated undirected movement following the approach described in Sells et al. (2022, 2023a, 2023b), using each bear’s iSSF in turn in Program R (R Core Team, 2020). Each simulated bear acted independently within a spatially explicit landscape defined by its bear-specific conductance surface. Simulated movement was implemented as a stepwise process, with each step representing approximately three hours of movement (+/− 45 minutes), consistent with the temporal resolution of the GPS data used to fit the models. For simulations applied to the North Cascades Ecosystem (NCE), each simulation iteration began by placing a simulated bear at a randomly selected start node drawn from within potential release sites identified in the 2024 Final Environmental Impact Statement for grizzly bear restoration in the NCE. Simulated bears were allowed to move freely within a broader simulation area defined as the NCE buffered by 100 km, with highly urbanized areas excluded to avoid extrapolation beyond the model training domain. At each movement step, we generated a sample of 11 candidate steps based on the individual bear’s observed step length and turn angle distributions. For each candidate step, we extracted the conductance value at the step endpoint, calculated as the iSSF value exp(βx). The next step was selected probabilistically, with selection probabilities proportional to these conductance values. This stochastic selection process was repeated sequentially for 5,000 steps per simulation iteration. For each individual bear model, we ran 100 simulation iterations for females and 242 iterations for males to yield approximately equal numbers of total iterations by sex (4,600 iterations for females and 4,598 iterations for males). To generate spatial predictions of habitat use, simulated step endpoints were summed across all individuals and iterations within each 300 × 300 m raster cell (separately by sex, and for both sexes combined). Raster cells with building densities exceeding 100 buildings per km² were masked to exclude highly urbanized areas. Resulting step-count surfaces were then binned into 10 equal-area quantile classes of relative predicted habitat use, with class 1 representing the lowest relative use and class 10 the highest. This data release includes only the NCE predicted habitat maps (uncalibrated and elevation-calibrated, described below), provided for both sexes combined and separately by sex. 20240301 To evaluate model transferability prior to application in the NCE, we applied the same simulation framework to a nearby validation site in British Columbia encompassing the Squamish–Lillooet (SL), McGillvary Mountains (MM), and North Stein–Nahatlatch (NSN) grizzly bear populations. Predicted habitat maps for the validation site were generated solely for the purpose of evaluating model transferability and selecting the elevation-rescaling parameter prior to producing the final NCE prediction maps; validation-site map products are not included in this data release. We evaluated predictive performance by overlaying GPS collar locations from 73 grizzly bears monitored in the validation populations and extracting predicted habitat classes at each location. Predictive performance was assessed using Spearman rank correlations between habitat class values and area-adjusted location densities, as well as by summarizing the proportion of locations falling within the highest predicted-use classes. Performance was evaluated by population, sex, and season. To address overprediction in low-elevation environments outside the training domain, we applied an elevation-based calibration to the simulation outputs. This post-simulation rescaling reweighted raster cell step counts as a function of elevation prior to quantile binning, reducing the contribution of low-elevation areas and increasing the contribution of higher-elevation areas in the final predicted habitat maps. The calibration affected only the aggregation of simulated steps and did not modify the underlying movement simulations or habitat selection models. The elevation-calibrated maps showed strong agreement with observed space use in the validation populations and were subsequently applied to the NCE simulations. Both uncalibrated (“no calibration”) and elevation-calibrated versions of the predicted habitat maps are included in this data release. Final datasets are provided as raster layers of relative predicted habitat use summarized into 10 equal-area quantile classes. 20240302 Raster Grid Cell 1159 621 1 Albers Conical Equal Area 29.5 45.5 -96.0 23.0 0.0 0.0 row and column 300.0 300.0 meters WGS_1984 WGS 84 6378137.0 298.257223563 NCE_predicted_habitat_elevation_calibrated_both_sexes.tif Raster geospatial data file. Producer Defined Value The Class attribute represents, for each pixel, the relative predicted habitat use of the cell (1 = lowest, 10 = highest), based on reintroduction simulations for male and female grizzly bears. Producer Defined NA NA indicates there were no simulation results here Producer defined 1.0 10.0 NCE_predicted_habitat_elevation_calibrated_female.tif Raster geospatial data file. Producer Defined Value The Class attribute represents, for each pixel, the relative predicted habitat use of the cell (1 = lowest, 10 = highest), based on reintroduction simulations for female grizzly bears. Producer Defined NA NA indicates there were no simulation results here Producer defined 1.0 10.0 NCE_predicted_habitat_elevation_calibrated_male.tif Raster geospatial data file. Producer Defined Value The Class attribute represents, for each pixel, the relative predicted habitat use of the cell (1 = lowest, 10 = highest), based on reintroduction simulations for male grizzly bears. Producer Defined NA NA indicates there were no simulation results here Producer defined 1.0 10.0 NCE_predicted_habitat_no_calibration_both_sexes.tif Raster geospatial data file. Producer Defined Value The Class attribute represents, for each pixel, the relative predicted habitat use of the cell (1 = lowest, 10 = highest), based on reintroduction simulations for male and female grizzly bears. Producer Defined NA NA indicates there were no simulation results here Producer defined 1.0 10.0 NCE_predicted_habitat_no_calibration_female.tif Raster geospatial data file. Producer Defined Value The Class attribute represents, for each pixel, the relative predicted habitat use of the cell (1 = lowest, 10 = highest), based on reintroduction simulations for female grizzly bears. Producer Defined NA NA indicates there were no simulation results here Producer defined 1.0 10.0 NCE_predicted_habitat_no_calibration_male.tif Raster geospatial data file. Producer Defined Value The Class attribute represents, for each pixel, the relative predicted habitat use of the cell (1 = lowest, 10 = highest), based on reintroduction simulations for male grizzly bears. Producer Defined NA NA indicates there were no simulation results here Producer defined 1.0 10.0 The Class attribute represents, for each pixel, the relative predicted habitat use of the cell (1 = lowest, 10 = highest), for simulations for female and male grizzly bears. Sells, S. N., M. McLellan, J. I. Ransom, A. Lyons, M. Clarke, C. D. Apps. 2026. Predicted Habitat Use for Reintroduced Grizzly Bears in the Transboundary North Cascades Ecosystem: spatial data. U.S. Geological Survey data release. GS ScienceBase U.S. Geological Survey mailing address

Denver Federal Center, Building 810, Mail Stop 302

Denver CO 80225 United States 1-888-275-8747 sciencebase@usgs.gov Unless otherwise stated, all data, metadata and related materials are considered to satisfy the quality standards relative to the purpose for which the data were collected. Although these data and associated metadata have been reviewed for accuracy and completeness and approved for release by the U.S. Geological Survey (USGS), no warranty expressed or implied is made regarding the display or utility of the data for other purposes, nor on all computer systems, nor shall the act of distribution constitute any such warranty. Digital Data https://doi.org/10.5066/P13YTTEJ None 20260323 Sarah N Sells U.S. Geological Survey, ECOSYSTEMS Assistant Unit Leader, Research Ecologist mailing address

Natural Science Bldg, Univ of MT

Missoula MT 59812 US 406-243-4393 406-243-6064 ssells@usgs.gov FGDC Biological Data Profile of the Content Standard for Digital Geospatial Metadata FGDC-STD-001.1-1999