Chloe E. Moore Meryl C. Mims Jason B. Dunham Joseph R. Benjamin Kristin L. Jaeger Patricia K. Haggerty 20250414 vector digital data www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/P13QTXJG Jason B. Dunham Joseph R. Benjamin Chloe E. Moore Meryl C. Mims Daniel J. Isaak Roy Sando Kristin L. Jaeger 2025 publication This data release provides spatial data for species distribution models developed for 24 stream fishes native to the Pacific Northwest, USA and associated model evaluation metrics used to determine the best supported model. These models support a larger analysis for understanding native stream fish vulnerability in the region, specifically population viability analyses built to determine species extinction risk. The models were developed using the Maximum Entropy (MaxEnt) machine learning approach. This approach is best suited for the presence only occurrence data available for our analyses. The point occurrences representing presences in the models were previously collated for the Pacific Northwest Rarity and Climate Sensitivity Index (Moore et al., 2025). The environmental predictors used to develop the models were chosen based on a combination of biological relevance to the included species and being uncorrelated with other selected predictors. Models were predicted across two possible extents for each species, the extent of the species' IUCN range polygon if available or the extent of a convex hull of occurrence records used to build the model. These species distribution models were developed to determine 24 native stream fishes' distributions in the Pacific Northwest, USA. Each species' shapefile contains the continuous probability of habitat suitability within individual streams across the region from 0 (low suitability) to 1 (high suitability), as well as the calculated environmental predictor values associated with each individual stream used in building the models. The occurrence records used to develop the models are mostly opportunistic collections, and although they were previously cleaned, filtered, and underwent expert review, and steps were taken to reduce spatial bias associated with opportunistic collections during model development, the nature of the input dataset must be considered for future use of model outputs. 1977 2020 ground condition Complete None planned HUC17 Pacific Northwest -124.7330 -110.8200 48.9946 41.1278 ISO 19115 Topic Category biota environment climatologyMeteorologyAtmosphere inlandWaters None multispecies study species distribution model USGS Thesaurus vertebrates fish river systems modeling biogeography geospatial analysis geospatial datasets USGS Metadata Identifier USGS:679d2f36d34eb38981c9c6d0 Common geographic areas North America United States Washington Oregon Idaho Integrated Taxonomic Information System Acrocheilus alutaceus Catostomus ardens Catostomus columbianus Catostomus macrocheilus Catostomus tsiltcoosensis Catostomus warnerensis Cottus beldingii Cottus gulosus Cottus leiopomus Cottus marginatus Cottus perplexus Cottus rhotheus Gila atraria Mylocheilus caurinus Oregonichthys crameri Percopsis transmontana Prosopium williamsoni Ptychocheilus oregonensis Ptychocheilus umpquae Rhinichthys falcatus Rhinichthys osculus Richardsonius balteatus Siphateles bicolor Cottus schitsuumsh Integrated Taxonomic Information System (ITIS) 2025 website https://www.itis.gov/ Integrated Taxonomic Information System (ITIS) 1)The cedar sculpin (Cottus schitsuumsh), described in Lemoine, et al., 2014, is not yet included in the Integrated Taxonomic Information System (ITIS) database. For reference, we include Cottus schitsuumsh with its congeners in the ITIS taxonomic hierarchy below. 2)We use Acrocheilus alutaceus (chiselmouth, TSN: 163531) in our datasets. Gila alutacea (chiselmouth, TSN: 1199065) is the name recognized by ITIS and is provided in the taxonomic hierarchy. Both names refer to the same fish. Dependent on source data. Dependent on source data. Kingdom Animalia Subkingdom Bilateria Infrakingdom Deuterostomia Phylum Chordata Subphylum Vertebrata Infraphylum Gnathostomata Superclass Actinopterygii Class Teleostei Superorder Ostariophysi Order Cypriniformes Superfamily Cyprinoidea Family Leuciscidae Subfamily Laviniinae Genus Gila Species Gila alutacea Species Gila atraria TSN: 163543 Genus Ptychocheilus Species Ptychocheilus oregonensis TSN: 163523 Species Ptychocheilus umpquae TSN: 163526 Genus Siphateles Species Siphateles bicolor TSN: 913989 Subfamily Pogonichthyinae Genus Mylocheilus Species Mylocheilus caurinus TSN: 163521 Genus Oregonichthys Species Oregonichthys crameri TSN: 163879 Genus Rhinichthys Species Rhinichthys falcatus TSN: 163386 Species Rhinichthys osculus TSN: 163387 Genus Richardsonius Species Richardsonius balteatus TSN: 163528 Superfamily Cobitoidea Family Catostomidae Subfamily Catostominae Tribe Catostomini Genus Catostomus Species Catostomus ardens TSN: 163899 Species Catostomus columbianus TSN: 163897 Species Catostomus macrocheilus TSN: 163896 Species Catostomus tsiltcoosensis TSN: 1171682 Species Catostomus warnerensis TSN: 163915 Superorder Acanthopterygii Order Scorpaeniformes Suborder Cottoidei Superfamily Cottoidea Family Cottidae Genus Cottus Species Cottus beldingii TSN: 167238 Species Cottus gulosus TSN: 167234 Species Cottus leiopomus TSN: 167246 Species Cottus marginatus TSN: 167247 Species Cottus perplexus TSN: 167248 Species Cottus rhotheus TSN: 167252 Species Cottus schitsuumsh cedar sculpin Superorder Paracanthopterygii Order Percopsiformes Suborder Percopsoidei Family Percopsidae Genus Percopsis Species Percopsis transmontana TSN: 164410 Superorder Protacanthopterygii Order Salmoniformes Family Salmonidae Subfamily Coregoninae Genus Prosopium Species Prosopium williamsoni TSN: 162009 No access constraints. Please see 'Distribution Info' for details. These data are marked with a Creative Common CC0 1.0 Universal License. These data are in the public domain and do not have any use constraints. Users are advised to read the dataset's metadata thoroughly to understand appropriate use and data limitations. Kristin L Jaeger U.S. Geological Survey, NW-PACIFIC ISLAND REG Research Hydrologist mailing address

934 Broadway

Tacoma WA 98402 US 253-552-1617 kjaeger@usgs.gov This project was funded by the USGS Northwest Climate Adaptation Science Center. Chloe E. Moore Meryl C. Mims Jason B. Dunham Joseph R. Benjamin 2025 tabular digital data www.sciencebase.gov U.S. Geological Survey Michael Lemoine Michael K. Young Kevin S. McKelvey Lisa Eby Kristine L. Pilgrim Michael K. Schwartz 2014 publication Zootaxa Magnolia Press Volume 3755, Issue 3, pages 241 - 258 http://doi.org/10.11646/zootaxa.3755.3.3 Model parameters (regularization multiplier and feature class) were optimized for each individual species by comparing model performance for all possible combinations of 6 feature classes and a regularization multiplier of 1 to 5. Model performance was statistically evaluated by using either Continuous Boyce Index (CBI), Area under the Receiver Operating Characteristic (AUC), or a combination of both, and selecting the highest performance evaluation value. Model parameter sets were first selected for maximum CBI, with maximum AUC used as a tie breaker if multiple models had equivalent CBI values. Final models released required a minimum CBI of 0.70 and AUC of 0.70, excluding Percopsis transmontana which we were only able to develop a best performing model with a CBI of 0.69. All data points were confirmed correct in ArcGIS Pro. The potential species included in this dataset is confined to only stream fishes that are native to the Pacific Northwest of the United States. Stream fishes native to the study extent were removed as candidates for analysis if only 25 % or less of the species total range (as defined by IUCN range maps or a 50 km buffer of all available occurrence points in GBIF) fell within the study extent. We also did not include any species that had fewer than 20 occurrence points. Chloe E. Moore Meryl C. Mims Jason B. Dunham Joseph R. Benjamin 2025 tabular digital data www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/P9WE05SV Digital and/or Hardcopy 1995 2015 ground condition PNW_fish_occ Point occurrence records for the 24 included stream fishes native to the Pacific Northwest, USA, used as species presence observations within species distribution models. US Geological Survey 2017 vector digital data https://www.epa.gov/waterdata/get-nhdplus-national-hydrography-dataset-plus-data Digital 2017 observed NHD_flowline NHD flowlines were used to pair occurrence records with environmental predictors within species distribution models. The models were also predicted to NHD flowlines using pair environmental predictors. United States Geological Survey 20200701 vector digital data https://www.usgs.gov/national-hydrography/access-national-hydrography-products Digital and/or Hardcopy 20200701 observed HUC12_watershed Hydrologic Unit Code (HUC12) watersheds were used to obtain spatially controlled background flowlines/points for species distribution models. USDA Forest Service 20220510 vector digital data data.fs.usda.gov USDA Forest Service https://data.fs.usda.gov/geodata/edw/datasets.php?xmlKeyword=hydro+flow+metric Digital 19771001 20060930 Contiguous United States, historical time period (1977-2006) US_streamflow_metric_dataset Provided environmental variable of 'winter flashiness' or the "average number of daily flows between December 1 and March 31 which exceed the 95th percentile of daily flows across the entire year" per source metadata. PRISM Climate Group 202111 raster digital data https://prism.oregonstate.edu/normals/ Digital 1991 2020 observed PRISM Climate Normals Used to calculate the mean annual precipitation and mean minimum air temperature within occupied and background HUC12 watersheds. Roy Sando August R. Schultz 20220215 tabular digital data https://www.usgs.gov/data/reach-scale-predicted-annual-streamflow-permanence-probabilities-predicted-monthly-mean-stream https://doi.org/10.5066/P94SMJKI Digital and/or Hardcopy 2004 2015 observed PROSPER Permanence probability variables were used to predict fish species distributions. USDA Forest Service 20191001 raster digital data USDA Forest Service Salt Lake City, UT https://data.fs.usda.gov/geodata/rastergateway/treecanopycover/ Digital 20070101 20111110 publication date USFS Tree Canopy Used to estimate mean percent canopy cover within occupied and background HUC12 watersheds. Jon Dewitz 20210604 remote-sensing image US Geological Survey Sioux Falls, SD https://doi.org/10.5066/P9KZCM54 Digital 2001 2019 publication date NLCD Used to calculate the percent developed land within occupied and background HUC12 watersheds. International Union for Conservation of Nature 2017 vector digital data www.iucnredlist.org Digital 20211129 observed IUCN Species range map shapefiles from IUCN were used to calculate percent of native range overlapping study extent and to filter occurrence records. Point occurrence records were spatially snapped to the nearest NHD flowline (previously filtered to only retain records within 500 meters to the nearest flowline. A species was determined to be 'present' within a stream if a flowline had a corresponding snapped occurrence point. All subsequent steps and model development used these flowlines as the presence location (rather than a single point coordinate). PNW_fish_occ NHD_flowline 20220825 The MaxEnt species distribution algorithm uses background points, sometimes referred to as pseudo-absences, rather than species' absences that are used in other species distribution algorithms. We selected a set of spatially controlled background points to minimize the effect of spatial bias associated with opportunistically collected data. To do this, we selected all HUC12 watersheds from the USGS Watershed Boundary Dataset that contained at least one occurrence record for each individual species. We then selected all NHD flowlines falling within the species presence delineated HUC12 watersheds. Finally, we randomly selected 10,000 flowlines from this filtered flowline set, or the maximum number of flowlines possible excluding presence flowlines, and used these flowlines and associated environmental predictors as our background 'points'. PNW_fish_occ NHD_flowline HUC12_watershed 20220825 Predictors were obtained, collated, and developed based first on biological relevance to the study's focal species. This included climatic, hydrologic, land cover, and stream morphology predictors from several different sources. Some data sources were originally compiled at the stream reach scale and assigned the associated COMID from the corresponding NHD feature (variables from NHD_flowline, PROSPER, US_streamflow_metric_dataset). These variables were not modified from the original source for use in the models. Other environmental variable sources' original format was raster data (PRISM Climate Normals, USFS Tree Canopy, NLCD), which were summarized at the watershed level. First, for each flowline established as containing a species presence or randomly selected as a background flowline, we identified the HUC12 watershed (HUC12_watershed) the flowline primarily fell within. Then, we calculated the weighted average of all raster cells for the target environmental variable that overlapped the HUC12 watershed. Weight was defined by the portion of the cell area overlapped by the watershed. For the land cover variables, prior to quantifying the weighted average, the land cover classes were recategorized and summarized into 9 larger categories (unclassified, water, snow, developed, barren land, forest, grasslands, agriculture, and wetland). The percent of each class within the watershed was calculated. Lastly, the correlation of all initial variables across all flowlines in the Pacific Northwest Region was estimated using Pearson's correlation. Variables with less than |0.7| Pearson's r were prioritized for final inclusion in the models. NHD_flowline US_streamflow_metric_dataset PROSPER PRISM Climate Normals USFS Tree Canopy NLCD 20220825 To develop the species distribution models, we used the Maximum Entropy (MaxEnt) modeling algorithm. We optimized each model to the individual species by testing and comparing a range of model parameters for each species. We tested 6 feature classes (linear, linear-quadratic, quadratic, product, threshold, and hinge) and 5 regularization multipliers (1, 2, 3, 4, 5), for a total of 30 candidate models per species. We used block cross-validation, where the data is split into four relatively equal - sized quadrants using a latitude and longitude split, to partition the occurrence and background records into training and evaluation bins and assess model performance. We selected the best performing model as determined by Continuous Boyce Index (CBI) and/or Area under the Receiver Operating Characteristic (AUC). Model parameter sets were first selected for maximum CBI, with maximum AUC used as a tie breaker if multiple models had equivalent CBI values. We aimed for a performance minimum CBI of 0.70 and AUC of 0.70 for final models used to predict distributions. 20220825 The selected candidate distribution model was then predicted to range extent constrained NHDplus v2 flowlines based on the flowlines' catchment level summarized environmental or climatic variables (as done in step 3). We constrained the potential flowlines for prediction by only predicting to flowlines that fell within a species' IUCN range shapefile with a classification of 'Extant (resident)' or the convex hull of the species' filtered occurrence points. For the IUCN range shapefile, we also selected flowlines within a 10 km buffer of the range and for the convex hull, we selected flowlines within a 50 km buffer of the hull. We used a larger buffer for the hull extent to allow for prediction to regions potentially missing from the opportunistically collected occurrence records. An IUCN range was not available for some of the species included in the study, and we only used the convex hull prediction extent for these species. NHD_flowline IUCN 20220916 Vector G-polygon 0.0000001 0.0000001 Decimal degrees North American Datum of 1983 (NAD 83) Geodetic Reference System 1980 6378137.000000 298.257222 PNWfishes_SDM_model_evaluation_results.csv Comma Separated Value (CSV) file containing model performance results for MaxEnt Species Distribution Models (SDMs) of 24 PNW fishes. Producer Defined species Species Identity Producer Defined Species scientific name described in this column. Species names were matched to the GBIF Backbone Taxonomy (GBIF Secretariat, 2019, https://doi.org/10.15468/39omei) and Integrated Taxonomic Information System (Integrated Taxonomic Information System (ITIS), www.itis.gov, CC0, 2020, https://doi.org/10.5066/F7KH0KBK). Both accessed on 20210716. feature_class Feature class used for species' MaxEnt model Producer Defined T Threshold feature class Producer defined L Linear feature class Producer defined H Hinge feature class Producer defined LQ Linear-quadratic feature class Producer defined P Product feature class Producer defined reg_multi Regularization Multiplier used for MaxEnt model Producer Defined 1 5 CBI_val Mean validation Continuous Boyce Index for feature class by regularization multiplier combination in corresponding row across all partitioned validation datasets. Producer Defined 0.6920000000000001 0.972 CBI_val_stdev Standard deviation of Continuous Boyce Index for feature class by regularization multiplier combination in corresponding row across all partitioned validation datasets. Producer Defined 0.012 0.25 AUC_val Mean validation Area under the Receiver Operating Characteristic curve for feature class by regularization multiplier combination in corresponding row across all partitioned validation datasets. Producer Defined 0.7070000000000001 0.899 AUC_val_stdev Standard deviation of Area under the Receiver Operating Characteristic curve for feature class by regularization multiplier combination in corresponding row across all partitioned validation datasets. Producer Defined 0.011000000000000001 0.21 SDMs_IUCNextent.zip Shapefiles of the species distribution model predictions for 18 PNW fishes. The predictions in this folder were constrained using NHD flowlines falling within the species' IUCN range shapefile plus a 10 km buffer. All shapefiles are named using [species scientific name]_IUCN.[dbf/prj/shp/shx] The attribute table columns and descriptions are the same for all species, but the actual values reported below are an example and do not encompass values for all species. Producer Defined COMID NHD flowline unique COMID to connect record with the nearest flowline if pairing with NHD data Producer Defined Each of the values represent the unique ID given to each flowline in the NHDplus dataset. W95_HIS Number of winter floods (Dec - Mar) exceeding 95th percentile of daily flows value from US stream flow metric (USFS) dataset for the associated flowline US_streamflow_metric_dataset NA No value available Producer defined 0 17.2483 count StrmOrd Strahler's Stream Order value from NHD dataset for the associated flowline NHD_flowline 1 8 numeric order AreSqKM Network catchment area that the flowline falls in. Value from the NHD dataset HUC12_watershed 0.0009 718.4007 kilometer squared Pthlngt Distance downstream to the network end for the associated flowline NHD_flowline 0 2106.4510 kilometers SLOPE Slope of the stream for the associated flowline (meters/meters) based on smoothed elevations. Value is from the NHD dataset. NHD_flowline NA No value available Producer defined 0 0.8086 VA_MA Mean annual velocity for flow from runoff of stream for the associated flowline. Value from the NHD dataset. NHD_flowline -9999 or NA No value available Producer Defined 0.1767 7.8919 feet per second MEANspp Mean annual predicted probability of stream permanence. Mean across all years available from the PROSPER dataset. PROSPER 0.0047 0.9715 probability AugTemp Mean predicted stream temperature of associated flowline in August. Average across predicted August temperature from 2004 to 2015 from the PROSPER dataset. PROSPER 1.6717 27.7842 degrees Celsius annl_fl Predicted mean annual flow as estimated by the sum of mean monthly streamflow discharge averaged across all years from 2004 to 2015 from the PROSPER dataset. PROSPER 0 829340.3000 cubic feet per second lowQ Mean predicted summer stream discharge for the associated flowline. Calculated as the sum of mean June, July, August, and September discharge divided by the sum of mean annual discharge across 2004 to 2015 from PROSPER dataset. PROSPER 0 1 cubic feet per second mn_ppt_ws Mean annual precipitation across 1991 to 2020. Value from the weighted average of all cells that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from PRISM dataset. PRISM Climate Normals 175.3021 4806.5620 millimeter mn_tmin_ws Mean minimum temperature across 1991 to 2020. Value from the weighted average of all cells that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from PRISM dataset. PRISM Climate Normals -14.8809 -5.2789 degrees celsius mn_tcc Mean percent tree canopy cover. Value from the weighted average of all cells that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from USFS Tree Canopy Geospatial dataset. USFS Tree Canopy NA No value available Producer defined 0 90.0616 percent p_dev Mean percent developed land cover. Value from the weighted average of all cells represent low, medium, and high developed land that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from NLCD. NLCD 0 93.2009 percent pred_iucn Species distribution model output. Predicted habitat suitability of associated flowline for selected species, with values closer to representing higher probability of fish being present in flowline. Value predicted flowlines within the species range as defined by convex hull of species occurrence points with 50 km buffer. Producer Defined NA No prediction made for flowline Producer defined 0 1 probability SDMs_HULLextent.zip Shapefiles of the species distribution model predictions for 24 PNW fishes. The predictions in this folder were constrained using NHD flowlines falling within a convex hull of the species' filtered occurrence points plus a 50 km buffer. All shapefiles are named using [species scientific name]_hull.[dbf/prj/shp/shx]. The attribute table columns and descriptions are the same for all species, but actual values reported below are only an example and do not encompass all values for all species. Producer Defined COMID NHD flowline unique COMID to connect record with the nearest flowline if pairing with NHD data NHD_flowline Each of the values represent the unique ID given to each flowline in the NHDplus dataset. W95_HIS Number of winter floods (Dec - Mar) exceeding 95th percentile of daily flows value from US stream flow metric (USFS) dataset for the associated flowline US_streamflow_metrics_dataset 0 17.2483 count StrmOrd Strahler's Stream Order value from NHD dataset for the associated flowline NHD_flowline 1 8 numeric order AreSqKM Network catchment area that the flowline falls in. Value from the NHD dataset HUC12_watershed 0.0009 718.4007 kilometer squared Pthlngt Distance downstream to the network end for the associated flowline NHD_flowline 0 2106.451 kilometers SLOPE Slope of the stream for the associated flowline (meters/meters) based on smoothed elevations. Value is from the NHD dataset. NHD_flowline 0 0.8086 VA_MA Mean annual velocity for flow from runoff of stream for the associated flowline. Value from the NHD dataset. NHD_flowline -9999 or NA No value available Producer Defined 0 17.2483 feet per second MEANspp Mean annual predicted probability of stream permanence. Mean across all years available from the PROSPER dataset. PROSPER 0.0047 0.9715 probability AugTemp Mean predicted stream temperature of associated flowline in August. Average across predicted August temperature from 2004 to 2015 from the PROSPER dataset. PROSPER 1.6717 27.7842 degrees Celsius annl_fl Predicted mean annual flow as estimated by the sum of mean monthly streamflow discharge averaged across all years from 2004 to 2015 from the PROSPER dataset. PROSPER 0 829340.3000 cubic feet per second lowQ Mean predicted summer stream discharge for the associated flowline. Calculated as the sum of mean June, July, August, and September discharge divided by the sum of mean annual discharge across 2004 to 2015 from PROSPER dataset. PROSPER 0 1 cubic feet per second mn_ppt_ws Mean annual precipitation across 1991 to 2020. Value from the weighted average of all cells that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from PRISM dataset. PRISM Climate Normals 175.3021 4806.5620 millimeter mn_tmin_ws Mean minimum temperature across 1991 to 2020. Value from the weighted average of all cells that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from PRISM dataset. Producer Defined -14.8809 -5.3020 degrees Celsius mn_tcc Mean percent tree canopy cover. Value from the weighted average of all cells that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from USFS Tree Canopy Geospatial dataset. USFS Tree Canopy 0 90.0616 percent p_dev Mean percent developed land cover. Value from the weighted average of all cells represent low, medium, and high developed land that fall within the HUC12 watershed encompassing associated flowline. Weighted by area of cell overlapped by watershed. Original data from NLCD. NLCD 0 93.2009 percent pred_hull Species distribution model output. Predicted habitat suitability of associated flowline for selected species, with values closer to representing higher probability of fish being present in flowline. Value predicted flowlines within the species range as defined by IUCN range shapefile with 10 km buffer. Producer Defined 0 1 probability U.S. Geological Survey GS ScienceBase 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. CSV, SHP https://doi.org/10.5066/P13QTXJG None 20250428 Data Manager Forest and Rangeland Ecosystem Science Center mailing address

777 NW 9th Street, Suite 400

Corvallis OR 97330 US 541-750-1030 fresc_outreach@usgs.gov FGDC Biological Data Profile of the Content Standard for Digital Geospatial Metadata FGDC-STD-001.1-1999