D. Joanne Saher Michael S. O’Donnell Cameron L. Aldridge Julie A. Heinrichs 20211209 raster digital data Fort Collins Science Center U.S. Geological Survey https://doi.org/10.5066/P93WFW13 D. Joanne Saher Michael S. O’Donnell Cameron L. Aldridge Julie A. Heinrichs 202111 publication n/a Elsevier BV ppg. e01935 https://doi.org/https://doi.org/10.1016/j.gecco.2021.e01935 The Gunnison sage-grouse (Centrocercus minimus) habitat suitability surface for Dove Creek satellite population represented here reflects breeding season at a patch scale context (30 m x 30 m pixel and radius window extents [radius] of 45 m, 120 m, 180 m, 270 m, 390 m, and 570 m). Habitat suitability estimated for areas constrained within the thresholded landscape model (containing 95% of use locations) developed for Colorado Parks and Wildlife critical habitat extent (southwestern Colorado). We developed habitat selection models for Gunnison sage-grouse (Centrocercus minimus), a threatened species under the U.S. Endangered Species Act. We followed a management-centric modeling approach that sought to balance the need to evaluate the consistency of key habitat conditions and improvement actions across multiple, distinct populations, while allowing context-specific environmental variables and spatial scales to nuance selection responses. Models were developed for six isolated satellite populations (San Miguel, Crawford, Piñon Mesa, Dove Creek, Cerro Summit-Cimarron-Sims, and Poncha Pass) from use locations collected between 1991 and 2016 (see larger citation for map of population boundaries). For each population, models were developed at two life stages (breeding and summer) and at two hierarchical scales (landscape and patch). We used multi-scale and seasonal resource selection analyses to quantify relationships between environmental conditions and sites used by animals. These resource selection function models relied on spatial data describing habitat conditions at different spatial scales, where environmental conditions differ, and habitat selection occur at different spatial scales for different available resources. Identifying, protecting, and restoring habitats for declining wildlife populations is foundational to conservation and recovery planning for any species at risk of decline. Resource selection analysis is a key tool to assess habitat and prescribe management actions. Conservation, species recovery, and habitat management efforts are needed in six isolated satellite populations (San Miguel, Crawford, Piñon Mesa, Dove Creek, Cerro Summit-Cimarron-Sims, and Poncha Pass), where environmental conditions differ, and the already small number of birds is declining. Resource selection analyses contribute to our understanding of species management by identifying key factors influencing selection (e.g., strength of covariates) and predicted habitat suitability. Our Gunnison sage-grouse case study points to specific opportunities for habitat restoration or improvement that may otherwise be missed with conventional habitat suitability modeling approaches. The inclusion of hierarchical seasonal habitat selection, varying ecological contexts, and a range of scales of influence yielded general insights regarding the reliability of crucial management variables and associated actions, as well as context-specific considerations that could be vital to achieving meaningful habitat improvements. These habitat suitability maps (raster surfaces) are intended to help wildlife and land managers gauge how Gunnison sage-grouse currently use habitat within each population and how changes to existing habitat conditions through restoration might improve habitat most effectively. We provide a list and brief description of all habitat suitability maps developed for our Gunnison sage-grouse satellite population modeling effort (resource selection functions): 1. gusg_ccsmesa_breed_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Cerro Summit-Cimarron-Sims satellite population (breeding, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 2. gusg_ccsmesa_breed_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Cerro Summit-Cimarron-Sims satellite population (breeding, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 3. gusg_ccsmesa_breed_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Cerro Summit-Cimarron-Sims satellite population (breeding, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 4. gusg_ccsmesa_summer_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Cerro Summit-Cimarron-Sims satellite population (summer, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 5. gusg_ccsmesa_summer_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Cerro Summit-Cimarron-Sims satellite population (summer, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 6. gusg_ccsmesa_summer_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Cerro Summit-Cimarron-Sims satellite population (summer, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 7. gusg_crawford_breed_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Crawford satellite population (breeding, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 8. gusg_crawford_breed_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Crawford satellite population (breeding, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 9. gusg_crawford_breed_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Crawford satellite population (breeding, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 10. gusg_crawford_summer_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Crawford satellite population (summer, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 11. gusg_crawford_summer_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Crawford satellite population (summer, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 12. gusg_crawford_summer_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Crawford satellite population (summer, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 13. gusg_dovecreek_breed_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Dove Creek satellite population (breeding, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 14. gusg_dovecreek_breed_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Dove Creek satellite population (breeding, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 15. gusg_dovecreek_breed_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Dove Creek satellite population (breeding, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 16. gusg_pinyonmesa_breed_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Piñon Mesa satellite population (breeding, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 17. gusg_pinyonmesa_breed_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Piñon Mesa satellite population (breeding, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 18. gusg_pinyonmesa_breed_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Piñon Mesa satellite population (breeding, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 19. gusg_pinyonmesa_summer_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Piñon Mesa satellite population (summer, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 20. gusg_pinyonmesa_summer_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Piñon Mesa satellite population (summer, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 21. gusg_pinyonmesa_summer_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Piñon Mesa satellite population (summer, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 22. gusg_ponchapass_breed_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Poncha Pass satellite population (breeding, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 23. gusg_ponchapass_breed_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Poncha Pass satellite population (breeding, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 24. gusg_ponchapass_breed_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Poncha Pass satellite population (breeding, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 25. gusg_ponchapass_summer_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Poncha Pass satellite population (summer, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 26. gusg_ponchapass_summer_landscape_usfws_hsi.tif: Gunnison sage-grouse habitat suitability surface for Poncha Pass satellite population (summer, landscape): U.S. Fish and Wildlife Service critical habitat extent (southwestern Colorado) 27. gusg_ponchapass_summer_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for Poncha Pass satellite population (summer, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 28. gusg_sanmiguel_breed_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for San Miguel satellite population (breeding, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 29. gusg_sanmiguel_breed_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for San Miguel satellite population (breeding, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 30. gusg_sanmiguel_summer_landscape_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for San Miguel satellite population (summer, landscape): Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 31. gusg_sanmiguel_summer_patch_cpw_hsi.tif: Gunnison sage-grouse habitat suitability surface for San Miguel satellite population (summer, patch): Utilized a subset of the Colorado Parks and Wildlife critical habitat extent (southwestern Colorado) 1991 2016 observed Complete None planned Dove Creek satellite population -109.0680 -108.8048 37.8987 37.7166 ISO 19115 Topic Category biota environment None Conservation Gunnison sage-grouse Habitat selection Logistic regression Resource selection function Breeding season Patch scale USGS Metadata Identifier USGS:61ae3fead34eb622f69a7147 None United States Colorado CO Southwestern Colorado Dove Creek satellite population None Gunnison Sage-Grouse Kingdom Animalia Subkingdom Bilateria Infrakingdom Deuterostomia Phylum Chordata Subphylum Vertebrata Infraphylum Gnathostomata Superclass Tetrapoda Class Aves Order Galliformes Family Phasianidae Subfamily Tetraoninae Genus Centrocercus Species Centrocercus minimus TSN: 677540 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. Cameron Aldridge U.S. Geological Survey, ROCKY MOUNTAIN REGION Research Ecologist mailing address

2150 Centre Avenue Bldg C

Fort Collins CO 80526 US 970-226-9100 970-226-9230 aldridgec@usgs.gov Colorado Parks and Wildlife (CPW) provided Gunnison sage-grouse location data, which they collected over the past couple decades (1991 – 2016) using a variety of capture methods and transmitter types. We thank the Colorado Bureau of Land Management, U.S. Geological Survey and U.S. Fish and Wildlife Service Science Support Partnership Program for funding and supporting various aspects of environmental layer and model development. All statistical analyses derived from StataCorp® software (STATA™ version 12.1) and all mapping derived from Esri® software (ArcGIS™ versions 10.3-7). All analyses completed on Windows 10 operating system with approximately 16 GB memory and 16 logical processors. J. P. Donnelly D. E. Naugle C. A. Hagen J. D. Maestas 20160202 publication Ecosphere vol. 7, issue 1 n/a Wiley https://doi.org/10.1002/ecs2.1208 These data are generated from a resource selection function model describing seasonal Gunnison sage-grouse habitat. We verified the values occurred between 0 and 1, representing relative suitability of habitat. We used modeled results and use locations to identify threshold values where any value less than a threshold reflected a relative suitability of habitat. We selected the threshold value as the resource selection function value that captured 95% of known telemetry locations for a given population model, season, and scale. These data are generated from a resource selection function model describing seasonal Gunnison sage-grouse habitat. The values represent a relative rank of seasonal habitat suitability and occur between 0 and 1. Values are not comparable across or within populations (i.e., USFWS and CPW extents differ in their rankings). The habitat suitability surfaces reflect a single Gunnison sage-grouse satellite population for a given season (life stage) and scale (covariates summarized at landscape or patch scales). The predicted surfaces are defined for three different mapped extents: Colorado Parks and Wildlife critical habitat for Gunnison sage-grouse, U.S. Fish and Wildlife Service critical habitat for Gunnison sage-grouse, and a patch scale defined by thresholding the CPW extent predicted surface at a level that captured 95% of use locations. All covariate data sources primarily reflect 30 meter (m) by 30 m spatial resolution. Vector data inputs, such as National Hydrology Dataset High Resolution (1:24,000), mapped urban roads (1:5,000) and residential locations (1:5,000), all have at least 1:24,000 spatial scale accuracy. Climate interpolations reflect 800 m by 800 m spatial resolution, which informed a weighted compound topographic index. Our digital elevation model reflected 10 m by 10 m and resampled to 30 m by 30 m after derived products (heat load index, compound topographic index, and terrain vector measure). The overall absolute vertical accuracy calculated for the national elevation data is 2.44 meters root mean square error (RMSE). The assessment included examination of the accuracy as a function of specific terrain conditions. M. C. Rigge C. Homer L. Cleeves D. K. Meyer B. Bunde H. Shi G. Xian S. Schell M. Bobo 2020 raster digital data https://www.mrlc.gov/data/type/rcmap-basemap-(2016) https://doi.org/10.3390/rs12030412 24000 Digital and/or Hardcopy 2016 ground condition Rangeland vegetation These data represented multiple shrubland vegetation components, including shrub, herbaceous, bare ground, litter, sagebrush, big sagebrush and annual herbaceous, along with estimating shrub height and sagebrush height, as candidate model inputs. We used moving windows to assess mean and standard deviation of percent cover or height. Michael J. Falkowski Jeffrey S. Evans David E. Naugle Christian A. Hagen Scott A. Carleton Jeremy D. Maestas Azad Henareh Khalyani Aaron J. Poznanovic Andrew J. Lawrence 2017 raster digital data Rangeland Ecology & Management vol. 70, issue 1 n/a Elsevier BV ppg. 15-24 https://doi.org/10.1016/j.rama.2016.08.002 https://map.sagegrouseinitiative.com/ecosystem/tree-cover 100000 Digital and/or Hardcopy 2013 2011 ground condition Conifer PJ Discrete classes of Conifer-PJ 1-4% and 4-10% cover/acre included as a covariate. We used moving windows to assess mean and standard deviation of percent cover or height, as well as distance (linear or decay) to the presence of these classifications. U.S. Geological Survey 2018 raster digital data https://www.usgs.gov/core-science-systems/ngp/3dep/about-3dep-products-services https://viewer.nationalmap.gov/basic/?basemap=b1&category=ned,nedsrc&title=3DEP%20View 24000 Digital and/or Hardcopy 2018 publication date DEM The digital elevation model was used as a covariate and to derive additional terrain indices (e.g., compound topographic index, heat load index, and vector ruggedness measure). U.S. Geological Survey 2017 vector digital data U.S. Geological Survey, National Geospatial Program, 20170914, USGS National Hydrography Dataset (NHD) Best Resolution 20170914 for Colorado State or Territory FileGDB 10.1 Model Version 2.2.1: U.S. Geological Survey. https://nhd.usgs.gov/data.html https://apps.nationalmap.gov/downloader/#/ 24000 Digital and/or Hardcopy 2017 publication date Hydrology These data were used to represent permanent/perennial rivers and water bodies as well as intermittent rivers/streams and water bodies. We used moving windows to assess distance to features (linear and decay) and fractional cover (when water represented as an area). We used HUC4 subregions 1403, 1402, 1301, and 1408. U.S. Fish and Wildlife Service 2017 vector digital data https://www.fws.gov/wetlands/data/State-Downloads.html 24000 Digital and/or Hardcopy 2017 publication date Wetlands These data were used to represent wetland areas and determine how sage-grouse select for or avoid these habitat conditions. We used moving windows to assess distance to features (linear and decay) and fractional cover (when water represented as an area). Colorado Department of Natural Resources 2015 vector digital data https://cdss.colorado.gov/gis-data 24000 Digital and/or Hardcopy 2015 ground condition Agriculture 1 (irrigated/cultivated) These data represent irrigated and cultivated lands within Colorado based on 2015 conditions and we used these to assess selection for and avoidance of agriculture features in habitat models. We used moving windows to assess distance to features (linear and decay) and fractional cover.We downloaded Irrigated Lands (2015) data for divisions 3, 4, 5, and 7. Utah Department of Natural Resources 2017 vector digital data https://gis.utah.gov/data/planning/water-related-land/ 24000 Digital and/or Hardcopy 2017 publication date Agriculture 2 (irrigated/cultivated) These data represent irrigated and cultivated lands within Utah based on 2015 conditions and we used these to assess selection for and avoidance of agriculture features in habitat models. We used moving windows to assess distance to features (linear and decay) and fractional cover. C. G. Homer J. A. Dewitz L. Yang S. Jin P. Danielson G. Xian J. Coulston N. D. Herold J. D. Wickham K. Megown 2015 raster digital data https://www.mrlc.gov/data/ 100000 Digital and/or Hardcopy 2011 ground condition Agriculture (non-irrigated/non-cultivated) These data represent non-irrigated and non-cultivated agricultural lands based on 2011 conditions as described in the NLCD Land Cover (CONUS) All Years data set and we used these to assess selection for and avoidance of agriculture features in habitat models. We used moving windows to assess distance to features (linear and decay) and fractional cover. Prism Climate Group, Oregon State University 2015 raster digital data http://prism.oregonstate.edu/normals/ 1000000 Digital and/or Hardcopy 1981 2010 ground condition Climate The precipitation data were used to develop our compound topographic index (CTI). For CTI index, we defined our flow accumulation by weighting the index using 1980 – 2010 total precipitation values, allowing for geographic discrimination because CTI is otherwise informed from convex and concave shapes of terrain and therefore does not necessarily reflect a moisture surrogate. Michael J. Falkowski Jeffrey S. Evans David E. Naugle Christian A. Hagen Scott A. Carleton Jeremy D. Maestas Azad Henareh Khalyani Aaron J. Poznanovic Andrew J. Lawrence 2017 raster digital data Rangeland Ecology & Management vol. 70, issue 1 n/a Elsevier BV ppg. 15-24 https://doi.org/10.1016/j.rama.2016.08.002 https://map.sagegrouseinitiative.com/ 100000 Digital and/or Hardcopy 2013 2011 ground condition Forest 1 Discrete classes of Conifer-PJ >10% cover/acre (including piñon pine (Pinus edulis) and juniper (Juniperus osteosperma and Juniperus scopulorum); hereafter, conifer-PJ) was combined with NLCD Forest Service percent tree canopy (>30%) to define forest presence. We used moving windows to assess distance to features (linear and decay) and fractional cover. U.S. Geological Survey 2016 raster digital data https://www.mrlc.gov/data/ 100000 Digital and/or Hardcopy 2011 ground condition Forest 2 NLCD Forest Service percent tree canopy (>30%) was combined with Discrete classes of Conifer-PJ >10% cover/acre as a covariate to define forest presence. We used moving windows to assess distance to features (linear and decay) and fractional cover. U.S. Geological Survey 20170226 raster digital data https://earthexplorer.usgs.gov/ Digital and/or Hardcopy 20150601 20150626 ground condition Landsat (Mesic) Precision Terrain Correction (Level-1TP) provides systematic, radiometric, and geometric accuracy by incorporating ground control points while employing a Digital Elevation Model (DEM) for topographic accuracy. Tiles included: LC08_L1TP_034033_20150603_20170226_01_T1, LC08_L1TP_034033_20150619_20170226_01_T1, LC08_L1TP_034034_20150603_20170226_01_T1, LC08_L1TP_034034_20150619_20170226_01_T1, LC08_L1TP_035033_20150626_20170226_01_T1, LC08_L1TP_035034_20150626_20170226_01_T1, LC08_L1TP_036033_20150601_20170226_01_T1, LC08_L1TP_036033_20150617_20170226_01_T1, LC08_L1TP_036034_20150601_20170226_01_T1, LC08_L1TP_036034_20150617_20170226_01_T1. Landsat values were thresholded to define mesic areas and then masked based on the process steps described above (Step 2). U.S. Department of Agriculture Farm Service Agency 20170301 remote-sensing image https://nrcs.app.box.com/v/gateway/folder/20369878234 Digital and/or Hardcopy 2011 ground condition NAIP (Urban) These data are hosted on the USDA Geospatial Data Gateway. These images were used for heads-up digitizing of roads and residences. We provide a general workflow describing the process of developing Gunnison sage-grouse (Centrocercus minimus) habitat suitability surfaces for multiple seasons (breeding and summer) and multiple hierarchical scales (landscape and patch). Refer to the larger citation noted in the metadata for a complete description of our methods, results, and discussion. All statistical analyses derived from StataCorp® software (STATA™ version 12.1) and all mapping derived from Esri® software (ArcGIS™ versions 10.3-7). All analyses completed on Windows 10 operating system with approximately 16 GB memory and 16 logical processors. 2021 We developed a suite of candidate covariates reflecting Gunnison sage-grouse (Centrocercus minimus; hereafter GUSG) habitat conditions at multiple spatial scales that sage-grouse frequently select for or avoid. Landscape scales included window extents (radius) of 1 km, 3 km, and 6.4 km, and patch scales included 30 m x 30 m pixel, and radius windows of 45 m, 120 m, 180 m, 270 m, 390 m, and 570 m. The scales are used by applying a moving window via a specified distance and generating a new surface. In some instances, the scale is used to define decay distances (nonlinear distance from a feature such as a road) where the distance from the scale defines the furthest distance from the feature to calculate the decay. The decay function uses the form e^(-d/alpha), where d is the linear distance in meters from a given pixel to a feature of interest, and alpha represents the radius of the window extent. These raster datasets were used to extract pixel values at use locations and available locations in order to develop statistical models using logistic regression (resource selection functions). Derived environmental variables ------------------------------- Rangeland/Shrubland vegetation components: These included fractional cover of sagebrush (all species), shrub (all species), and big sagebrush (Artemisia tridentata), as well as height (cm) for sagebrush and shrubs. Other vegetation included mean and standard deviations of percent cover for annual herbaceous (typically cheatgrass), all herbaceous (a combination of forbs, perennial, and annual grasses), bare ground, and litter, which we derived from shrubland fractional components. We also derived fractional subcomponents of non-big sagebrush (all sagebrush minus big sagebrush) and non-sagebrush shrub species (shrub minus all sagebrush). Mesic: Mesic areas were defined by compiling 6 Landsat 8 scenes (06/01/2015 – 06/30/2015) of the normalized difference vegetation index (NDVI) products available with <10% cloud and shadow masking (tier 1 with radiometric corrections and precision terrain correction; level 1TP). WRS Path: 34-36 and WRS Row: 33-34. Access date 12/31/2019. https://earthexplorer.usgs.gov/. With the absence of publicly available data delineating mesic resources (Donnelly et al. 2016), we used a similar approach implemented by Donnelly et al. (2016) to define persistent mesic sites. To best capture persistent mesic sites (Donnelly et al. 2016), we identified an NDVI value >0.4, where Donnelly used 0.3 for individual annual timestamps between 1984 and 2012. We masked our mesic binary classification using irrigated land, NLCD tree canopy cover (>30%), and any hydrologic features captured in other predictors. Compound topographic index (CTI): The CTI represents a steady-state wetness index, a function of slope and upstream contributing area per unit width that is orthogonal to the flow direction (i.e., ratio of catchment area and slope). Flatter areas will have large values (typically not informative), while smaller catchments with steep slopes will reflect low values of the index. We used the D8 algorithm to define our catchment area (flow accumulation) and the following equation: ln(((flow accumulation+1)* (pixel area_meters ))/tan(slope_radians)). CTI is often used as a surrogate for soil moisture and vegetative productivity; however, this index does not work well for larger geographic extents where precipitation differs. Therefore, we defined our flow accumulation by weighting the index using 1980 – 2010 total precipitation normal. Heat load index (HLI): The HLI identifies the potential annual direct incident radiation based on equation three referenced in the literature and ideal for latitudes 30 – 60 degrees North (R2 = 0.983). This regression equation includes latitude, slope, and aspect identifying gradients of the coolest slopes on northeastern aspects and the warmest slopes on southwestern aspects. Higher values will indicate areas that will dry out more quickly relative to surrounding areas. The values can range from 0 (little to no variation in the terrain and therefore coolest) to 1 (significant variation and therefore hottest). Most natural terrain generally has a much smaller range of values. The index does not account for cloud cover, regional differences in the atmospheric coefficient, or shading due to topography. Therefore, this index captures a potential measure of annual radiation. Vector ruggedness measure (VRM): The VRM is like other various forms of calculating terrain ruggedness, but unlike others, this incorporates slope and aspect. Numerous wildlife habitat studies, including those for sage-grouse, have demonstrated this index as a useful predictor of habitat selection. We used the window extents in the VRM equation to derive the index at different scales. For all indices (HLI, VRM, and CTI), we calculated the mean and standard deviation across all window sizes and considered them as well as their quadratic forms for all populations and seasons at both patch and landscape scales. Urban roads and residential housing: The urban category includes both transportation (line vector) and residential (point vector) data, which we developed from heads-up digitizing at a scale of 1:5,000 within GIS software from 2011 NAIP imagery (National Agriculture Imagery Program). We classified each transportation feature as one of three road classes. Class 1 included all major primary paved interstates, US and State highways, and secondary paved county highways. Class 2 consisted of all other maintained roads and roads leading to residential areas, including major community roads that support regularly maintained, 2-wheel drive clearance, and spur roads leading to residences or businesses. Class 3 roads included non-maintained single lane features, requiring 4-wheel drive clearance such as forestry roads, ranch roads, and 2-track trails. These classes identify roads of similar footprint disturbance and use of vehicular types, for which sage-grouse have previously been shown to respond. We buffered residential points by 75 m, defining a zone of residential influence because most of the study area is rural, and houses often include multiple, adjacent uninhabited buildings. We did not include buildings that were questionable in function and did not have identifiable road access. Forest: We used the National Land Cover Database of tree canopy cover to identify all pixels with >30% canopy cover to represent forests. The conifer-PJ product did not capture all forest types because it was focused on providing a high-resolution mapped product of conifer-PJ that would help inform management. Therefore, we merged the NLCD forest (>30%) with the conifer-PJ product (category of 10 – 50% cover/acre), giving the conifer-PJ product priority as it was more accurate than the NLCD data to produce our forest layer. 2021 Colorado Parks and Wildlife (CPW) provided Gunnison sage-grouse location data, which they collected over the past couple decades (1991 – 2016) using a variety of capture methods and transmitter types. The CPW also distributed available locations within CPW population boundaries of each satellite and for each season. They used the use and available location data with the covariate data (candidate habitat conditions) that we developed to extract pixel values. The resulting dataframe, without location context, was then provided for our modeling efforts. 2021 To model habitat selection we used statistical models of resource selection functions: w(x) = exp (β1x1 + β2x2 + … βkxk). For each satellite population, season (breeding and summer), and hierarchical scale (landscape and patch), we developed models by considering all candidate covariates. In all models, we required three types (families) of management-relevant habitat variables to evaluate the influence of factors likely manipulated by management actions: Sagebrush and shrub (e.g., sagebrush/shrub fractional cover), Forest (including piñon pine, juniper, and other tree cover), and Mesic resources (access/availability of wet meadows and their associated herbaceous content). We screened all selected variables for correlation (Pearson’s r >|0.7|) to prevent multicollinearity. The model with the best Akaike Information Criterion was selected for each population, season, and hierarchical scale. To evaluate the broad landscape conditions selected by GUSG and identify areas for potential habitat improvement, we first assessed habitat conditions across a larger landscape (i.e., 1 to 6.4-km radius surrounding use locations). Then, within suitable landscape habitat, we evaluated patch-level habitat selection (30 – 570 m from use locations) to assess local resource conditions selected by GUSG. This conceptually aligns with how GUSG might perceive their resources, hierarchically making seasonal selection decisions across space. Standard deviation variables (SD) capture habitat heterogeneity with patchy and uniform habitat being indicated by positive and negative values, respectively. Compound Topographic Index (CTI), Heat Load Index (HLI), and Vector Ruggedness Measure (VRM) are variables from the terrain grouping. The scale at which each variable was included is indicated in brackets (e.g., []). Variables contained within braces (e.g., {}) are only considered in combination with each other. Quadratic terms could only be evaluated with their linear partner, and SD was only considered with its corresponding mean. Model: {(0.152 * % other sage [570m]) – (0.829 * SD % other sage [570m])} – (1.106 * distance to forest [180m decay]) + (0.009 * % mesic [570m]) – {(1.104 * % annual herb [390m]) + (0.675 * SD % annual herb [390m])} – {(0.395 * CTI [570m]) + (0.217 * SD CTI [570m])} + {(0.030 * % non-irrigated ag [45m]) – (<0.001 * % non-irrigated ag^2 [45m])} – (1.669 * distance to mesic [570m decay]) 2021 To generate the mapped surfaces based on the top model derived from STATA software, we have provided the equation without exponentiating or including the constant factor. The exponentiated form that includes the constant scale the data and result in very small numbers, which sometimes cannot be exponentiated. Therefore, our surfaces do not exponentiate or include the constant. 2021 The thresholds applied to the surface for defining the mapping extent of patch model were applied to the landscape scale models at the CPW extent. In some cases, a threshold was also applied to the patch model to demonstrate where higher quality habitat exists within the patch framework. The landscape threshold was used for defining patch extent models by thresholding the landscape model using values provided here. A complete list of thresholds of landscape and patch are provided in the larger citation (Appendix). Crawford breeding landscape: -17.418 Crawford summer landscape: 2.770 Cerro Summit-Cimarron-Sims: thresholds not established because use locations did not exist for this satellite population. Dove Creek breeding landscape: 2491.271 Piñon Mesa breeding landscape: -14.537 Piñon Mesa summer landscape: -0.859 Poncha Pass breeding landscape: 152.557 Poncha Pass summer landscape: 17.804 San Miguel breeding landscape: 529.180 San Miguel summer landscape: 595.318 2021 After applying thresholds, we rescaled the resource selection function values to fall between 0 and 1 to represent a habitat suitability index. 2021 We report model fit statistics for each top model (season, population, and scale) to assess how well the data used could be predicted (within-sample assessment) and report the deviation explained (pseudo R2) for each base and top-selected model. We also report the proportion of the landscape implicated when a given top model was thresholded to include 95% of known locations for both the landscape and patch models. A more discriminatory model will capture those locations with a smaller amount of the landscape being implicated. Thresholds: The landscape threshold was used for defining patch extent models by thresholding the landscape model using values provided here. A complete list of thresholds of landscape and patch (not used for developing data) are provided in the larger citation (Appendix). 2021 Raster Grid Cell 592 691 1 Albers Conical Equal Area 29.5 45.5 -96.0 23.0 0.0 0.0 row and column 30.0 30.0 meters WGS_1984 WGS 84 6378137.0 298.257223563 gusg_dovecreek_breed_patch_cpw_hsi.tif Raster geospatial data file. Producer Defined Value The relative suitability of habitat for Gunnison sage-grouse. Producer Defined -3.40282346639e+38 Predictions not made to these pixels (intentionally masked or pixel represent areas beyond critical habitat). Producer defined 0.0 1.0 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 on any other system or for general or scientific purposes, nor shall the act of distribution constitute any such warranty.Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Digital Data https://doi.org/10.5066/P93WFW13 None 20211209 FORT Metadata Specialist U.S. Geological Survey, Fort Collins Science Center FORT Metadata Specialist mailing address

2150 Centre Avenue Bldg C

Fort Collins CO 80526 US 970-226-9100 970-226-9230 fortdatamanagement@usgs.gov FGDC Biological Data Profile of the Content Standard for Digital Geospatial Metadata FGDC-STD-001.1-1999