Original data can be downloaded from here. Another online version of the data can be found HERE.This version presented and hosted by CPAWS-NL allows for data extraction and analysis within ArcGIS Online Map Viewer."Kernel density estimation (KDE) utilizes spatially explicit data to model the distribution of a variable of interest. It is a simple non-parametric neighbor-based smoothing function that relies on few assumptions about the structure of the observed data. It has been used in ecology to identify hotspots, that is, areas of relatively high biomass/abundance, and in 2010 was used by Fisheries and Oceans Canada to delineate significant concentrations of corals and sponges. The same approach has been used successfully in the Northwest Atlantic Fisheries Organization (NAFO) Regulatory Area. Here, we update the previous analyses with the catch records from up to 5 additional years of trawl survey data from Eastern Canada, including the Gulf of St. Lawrence. We applied kernel density estimation to create a modelled biomass surface for each of sponges, small and large gorgonian corals, and sea pens, and applied an aerial expansion method to identify significant concentrations of theses taxa. We compared our results to those obtained previously and provided maps of significant concentrations as well as point data co-ordinates for catches above the threshold values used to construct the significant area polygons. The borders of the polygons can be refined using knowledge of null catches and species distribution models of species presence/absence and/or biomass." (DOI: 10.17632/dtk86rjm86.2)

The Kernel Density tool calculates the density of features in a neighborhood around those features.Kernel Density calculates the density of point features around each output raster cell. Conceptually, a smoothly curved surface is fitted over each point. The surface value is highest at the location of the point and diminishes with increasing distance from the point, reaching zero at the Search radius distance from the point. Only a circular neighborhood is possible. The volume under the surface equals the Population field value for the point, or 1 if NONE is specified. The density at each output raster cell is calculated by adding the values of all the kernel surfaces where they overlay the raster cell center. This layer is included in a storymap about the Panama City crayfish, a species listed as Threatened under the Endangered Species Act in 2022. Storymap link: https://fws.maps.arcgis.com/home/item.html?id=a791906fe3f8433eabadda5898184372

The attraction of tourism resources is very important to promote the sustainable development of tourism industry. This study takes China’s world cultural and natural heritage as the research object, and constructs an attractiveness evaluation system for China’s world cultural and natural heritage tourism resources by collecting user feedback data from three major travel OTA platforms. At the same time, ArcGIS 10.7 software was used for spatial autocorrelation analysis and kernel density analysis to explore the spatial distribution pattern of tourism resource attraction. The results show that China’s world cultural and natural heritage can be subdivided into 5 main categories and 10 sub-categories. From the perspective of spatial aggregation, only the Moran’s I index of tourist resource points showing a significant spatial aggregation feature. This study is helpful to reveal the weaknesses of tourism resource points and provide reference for sustainable development of attraction and optimization of tourism planning and management.

Please note that this data was selected from a larger dataset for use in the San Diego Ocean Planning Partnership, a collaborative pilot project between the California State Lands Commission and the Port of San Diego. For more information about the Partnership, please visit: https://www.sdoceanplanning.org/ The data was retrieved in May 2018 from OceanSpaces.org and is now available at https://data.cnra.ca.gov/dataset/spatial-and-economic-human-uses-california-south-coast-mpa-baseline-study-1992-to-2012 (Chen et al. An Economic and Spatial Baseline of Coastal Recreation in the South Coast of California. OceanSpaces.org. Retrieved May 2018).These data are a Kernel Density layers produced using ArcGIS. The Kernel analysis is a nonparametric statistical method for estimating probability densities from a set of points. Conceptually, a smooth raster surface is fitted over each point. The surface value is highest at the location of the point and diminishes with increasing distance (i.e.search radius), eventually reaching zero. A default value is calculated by the analysis tool for the search radius based on the input data; increasing the radius has little affect on the density value. Although more points will fall inside a larger search radius the number will be divided by a larger area when calculating density resulting in a more generalized output raster. The volume under the surface equals the weighted value for the point. The weights were created by Knowledge Networks and applied to the points based on demographics. The density of the output raster is calculated by adding the values of all the individual surfaces where they overlap. The point's weighted value determines the number of times to count the point. For example, a weighted value of 1.5 would cause the point to be counted one and half times. The resulting dataset is a smooth raster surface depicting the intensity use or density of an activity. Based on previous experience and after conducting some tests, all of the activity datasets were given a search radius of one mile. In discussing a similar project with the Oregon Department of Parks and Recreation we discovered that most visitors to the coast stay within a mile of their activity location. This distance also proved to be a good match to the mapped activities.

This study uses the methods of kernel density analysis and service area analysis in GIS to quantify the accessibility of public service facilities for residents within 15-minute living circles. This study takes the Ciqikou block as an example to analyze the distribution of public facilities and the polarization of services in historical blocks, quantifies and evaluates the accessibility of public facilities and the influencing factors of accessibility for residents' walking, and makes an assessment contribution to the sustainable renewal of social transportation and the maintenance of social equity in historical blocks in the later stage, with the aim of providing a useful reference for sustainable urban renewal.

The attraction of tourism resources is very important to promote the sustainable development of tourism industry. This study takes China’s world cultural and natural heritage as the research object, and constructs an attractiveness evaluation system for China’s world cultural and natural heritage tourism resources by collecting user feedback data from three major travel OTA platforms. At the same time, ArcGIS 10.7 software was used for spatial autocorrelation analysis and kernel density analysis to explore the spatial distribution pattern of tourism resource attraction. The results show that China’s world cultural and natural heritage can be subdivided into 5 main categories and 10 sub-categories. From the perspective of spatial aggregation, only the Moran’s I index of tourist resource points showing a significant spatial aggregation feature. This study is helpful to reveal the weaknesses of tourism resource points and provide reference for sustainable development of attraction and optimization of tourism planning and management.

This data gathers socioeconomic drivers at 1km2 grid cell spatial resolution to predict forest fires in a region in Spain. The response variable was fire density by grid cell from kernel density methods. It was produced under the Firemap project (https://geogra.uah.es/firemap/) by using GIS, spatial and statistical data sources at regional level in Spain. The resulting work was published at Vilar del Hoyo L, Martín Isabel MP, Martínez Vega FJ. 2011. Logistic regression models for human-caused wildfire risk estimation: analysing the effect of the spatial accuracy in fire occurrence data. European Journal of Forest Research. 130:983-96. doi: 10.1007/s10342-011-0488-2

description: These data are a Kernel Density layers produced using ArcGIS. The Kernel analysis is a nonparametric statistical method for estimating probability densities from a set of points. Conceptually, a smooth raster surface is fitted over each point. The surface value is highest at the location of the point and diminishes with increasing distance (i.e.search radius), eventually reaching zero. A default value is calculated by the analysis tool for the search radius based on the input data; increasing the radius has little affect on the density value. Although more points will fall inside a larger search radius the number will be divided by a larger area when calculating density resulting in a more generalized output raster. The volume under the surface equals the weighted value for the point. The weights were created by Knowledge Networks and applied to the points based on demographics. The density of the output raster is calculated by adding the values of all the individual surfaces where they overlap. The point's weighted value determines the number of times to count the point. For example, a weighted value of 1.5 would cause the point to be counted one and half times. The resulting dataset is a smooth raster surface depicting the intensity use or density of an activity. Based on previous experience and after conducting some tests, all of the activity datasets were given a search radius of one mile. In discussing a similar project with the Oregon Department of Parks and Recreation we discovered that most visitors to the coast stay within a mile of their activity location. This distance also proved to be a good match to the mapped activities.Link to the Dataset - ftp://ftp.dfg.ca.gov/R7_MR/NONCONSUMPTIVE/NCCSR/BeachActivities.zip; abstract: These data are a Kernel Density layers produced using ArcGIS. The Kernel analysis is a nonparametric statistical method for estimating probability densities from a set of points. Conceptually, a smooth raster surface is fitted over each point. The surface value is highest at the location of the point and diminishes with increasing distance (i.e.search radius), eventually reaching zero. A default value is calculated by the analysis tool for the search radius based on the input data; increasing the radius has little affect on the density value. Although more points will fall inside a larger search radius the number will be divided by a larger area when calculating density resulting in a more generalized output raster. The volume under the surface equals the weighted value for the point. The weights were created by Knowledge Networks and applied to the points based on demographics. The density of the output raster is calculated by adding the values of all the individual surfaces where they overlap. The point's weighted value determines the number of times to count the point. For example, a weighted value of 1.5 would cause the point to be counted one and half times. The resulting dataset is a smooth raster surface depicting the intensity use or density of an activity. Based on previous experience and after conducting some tests, all of the activity datasets were given a search radius of one mile. In discussing a similar project with the Oregon Department of Parks and Recreation we discovered that most visitors to the coast stay within a mile of their activity location. This distance also proved to be a good match to the mapped activities.Link to the Dataset - ftp://ftp.dfg.ca.gov/R7_MR/NONCONSUMPTIVE/NCCSR/BeachActivities.zip

As part of the cultural landscape, administrative toponyms do not only reflect natural and sociocultural phenomena, but also help with related management and naming work. Historically, county-level administrative districts have been stable and basic administrative regions in China, playing a role in the country’s management. We explore the spatio-temporal evolutionary characteristics of the county-level administrative toponyms cultural landscape in China’s eastern plains areas. A Geographical Information System (GIS) analysis, Geo-Informatic Tupu, Kernel Density Estimation, and correlation coefficients were conducted. We constructed a GIS database of county-level administrative toponyms from the Sui dynasty onward using the Northeast China, North China, and Yangtze Plains as examples. We then summarized the spatio-temporal evolutionary characteristics of the county-level administrative toponyms cultural landscape in China’s eastern plains areas. The results indicate that (1) the number of toponyms has roughly increased over time; (2) toponym densities on the three plains are higher than the national average in the corresponding timeframe since the Sui; and (3) county-level administrative toponyms related to mountains and hydrological features accounted for more than 30% of the total in 2010. However, the percentage of county-level administrative toponyms related to natural factors on the three plains has decreased since the Sui. To explore the factors influencing this spatio-temporal evolution, we analyzed the correlations between the toponyms and natural factors and human/social factors. The correlation degree between toponym density and population density is the highest, and that between toponym density and Digital Elevation Model (DEM) the lowest. Temperature changes were important in toponym changes, and population changes have influenced toponym changes over the last 400 years in China.

This dataset includes pre-processed geospatial data grids (.shp) and in high spatial resolution (50 m X 50 m) of 17 microscale built environment attributes. (in km per km2) The geospatial grids have been calculated using the Kernel density tool in ArcGIS Desktop, v.10.3 (ESRI, REDLANDS) applied to the raw GIS vectors (i.e., polylines), created during the street observation phase of our research. Please, also see the read_me.txt and desclaimer.txt files.

More information about the project can be seen: 1. on our website: http://geochoros.survey.ntua.gr/walkandthecitycenter/home 2. and our data article in 'Data in Brief' Journal: Bartzokas-Tsiompras, A., Photis, Y., Tsagkis, P., & Panagiotopoulos, G. (2021-under review). Microscale walkability indicators for fifty-nine European central urban areas: An open-access tabular dataset and geospatial web-based platform. Data in Brief.

Number of incidents counted within census tracts based on different spatial join approaches.

Analysis condicted by ABR Inc.–Environmental Research & Services.Data from ADFG/BLM/NSB and ConocoPhillips Alaska Inc.For Brownian Bridge Movement Models - Conducted dynamic Brownian Bridge Movement Models (dBBMM) to delineate movmeents on seasonal herd ranges. dBBMM models were run using the move package for r using the following methods.1. Locations within 30 days of first collaring were removed from the analysis. 2. Selected females from PTT and GPS collars during the date range July 1 2012–June 30 2017 and individuals having more than 30 locations per season.3. ran a dBBMMM model for each individual during each season using 1 km pixels. 4. Calculate the 95% isopleth for each individual.5. Overlap all 95% isopleths and calculate the proportion of animals using (as defined by 95% isopleth) each pixel. Value shown is proportion times 1000. Seasons used: Winter (Dec 1-Apr 15); Spring (Apr 16-May 31); Calving (June 1-15); postcalving (June 16-30); Mosquito (July 1-15); Oestrid Fly (July 16-Aug 7); late summer (August 8-Sept 15); Fall Migration (Sept 16-Nov 30). For Kernel Density Estimates - Conducted Kernel Density Estimation (KDE) to delineate seasonal herd ranges. Kernels were run using the ks package for r and the plugin bandwidth estimator. 1. Locations within 30 days of first collaring were removed from the analysis. 2. The mean latitiude and longitude for each animal was calculated for each day.3. A KDE utilization distribution was calculated for Julian day of the season (all years combined). 4. The daily KDE uds were averaged across the season. This method accounts for individual's movements during the seasons without the overfitting that results from using autocorrelated lcoations from individuals.Seasons used: Winter (Dec 1-Apr 15); Spring (Apr 16-May 31); Calving (June 1-15); postcalving (June 16-30); Mosquito (July 1-15); Oestrid Fly (July 16-Aug 7); late summer (August 8-Sept 15); Fall Migration (Sept 16-Nov 30).

In Dec 2017, Idaho made decision to use a 95% kernel density estimate on all their telemetry data as well as observation points provided by ID Department of Fish and Game. In March 2018, a national decision was made to consistently buffer occupied leks by 6.4 km. Idaho then merged the buffered leks with the KDE and maintains an occupancy layer for the State.

Snake River Plain Play Fairway Analysis - Phase 1 CRS Raster Files. This dataset contains raster files created in ArcGIS. These raster images depict Common Risk Segment (CRS) maps for HEAT, PERMEABILITY, AND SEAL, as well as selected maps of Evidence Layers. These evidence layers consist of either Bayesian krige functions or kernel density functions, and include: (1) HEAT: Heat flow (Bayesian krige map), Heat flow standard error on the krige function (data confidence), volcanic vent distribution as function of age and size, groundwater temperature (equivalue interval and natural breaks bins), and groundwater T standard error. (2) PERMEABILTY: Fault and lineament maps, both as mapped and as kernel density functions, processed for both dilational tendency (TD) and slip tendency (ST), along with data confidence maps for each data type. Data types include mapped surface faults from USGS and Idaho Geological Survey data bases, as well as unpublished mapping; lineations derived from maximum gradients in magnetic, deep gravity, and intermediate depth gravity anomalies. (3) SEAL: Seal maps based on presence and thickness of lacustrine sediments and base of SRP aquifer.
Raster size is 2 km. All files generated in ArcGIS. Kernel density function for volcanic vents weighted by Age and Size.

Predator-prey interactions can be profoundly influenced by vegetation conditions, particularly when predator and prey prefer different habitats. Although such interactions have proven challenging to study for small and cryptic predators, recent methodological advances substantially improve opportunities for understanding how vegetation influences prey acquisition and strengthen conservation planning for this group. The California Spotted Owl (Strix occidentalis occidentalis) is well-known as an old-forest species of conservation concern, but whose primary prey in many regions – woodrats (Neotoma spp.) – occurs in a broad range of vegetation conditions. Here, we used high-resolution GPS tracking coupled with nest video monitoring to test the hypothesis that prey capture rates vary as a function of vegetation structure and heterogeneity, with emergent, reproductive consequences for Spotted Owls in Southern California. Foraging owls were more successful capturing prey, including woodrats, in taller multilayered forests, in areas with higher heterogeneity in vegetation types, and near forest-chaparral edges. Consistent with these findings, Spotted Owls delivered prey items more frequently to nests in territories with greater heterogeneity in vegetation types and delivered prey biomass at a higher rate in territories with more forest-chaparral edge. Spotted Owls had higher reproductive success in territories with higher mean canopy cover, taller trees, and more shrubby vegetation. Collectively, our results provide additional and compelling evidence that a mosaic of large tree forests with complex canopy and shrubby vegetation increases access to prey with potential reproductive benefits to Spotted Owls in landscapes where woodrats are a primary prey item. We suggest that forest management activities that enhance forest structure and vegetation heterogeneity could help curb declining Spotted Owl populations while promoting resilient ecosystems in some regions. Methods See README DOCUMENT Naming conventions *RSF or prey refers to prey capture analysis *delivery in a file name refers to delivery rate analysis *repro in a filename means that file is for the delivery rate analysis

Setup *files with vegetation data should work with minimal alteration(will need to specify working directory) with associated R code for each analysis *Shapefiles were made in ArcGIS pro but they can be opened with any GIS software such as QGIS.

Locational data files

NOTE LOCATIONAL DATA IS SHIFTED AND ROTATED FROM THE ORIGINAL -due to the sensitive nature of this species. The locational_data includes: * All_2021_owls_shifted * Point file showing all GPS tag locations for prey capture analysis * Attributes include: * TERRITORY ID: Numerical identifier for each bird * Year: year GPS tag was recorded * Month: month GPS tag was recorded * Day: Day GPS tag was recorded * Hour: Hour GPS tag was recorded * Minute: minute GPS tag was recorded * All_linked_polygons_shifted * Polygon file showing capture polygons for prey capture analysis * Attributes include * Territory ID: numerical identifier for each bird * Polygon id: numerical identifier for each capture polygon for each bird * Shape area: area of each polygon * SBNF_camera_nests_shifted * Point file showing spotted owl nests for prey capture analysis * Attributes include * Territory id: numerical identifier for each bird * C95_KDE_2021_socal_shifted * Polygon file of owls 95% kernel density estimate for prey delivery rate analysis * Attributes include * Id: numerical identifier for each territory(bird) * Area: area of each polygon * San_bernardino_territory_centers * Point file showing Territory centers for historical SBNF territories – shifted for repro success analysis * Attributes include * Repro Territory id: unique identifier for each territory in broader set of territories

Besides the sifted locational data we have included - For the Resource selection function vegetation data, for the delivery analysis we have included an overview of prey deliveries by territory and vegetation data used, and for the reproductive analysis we have again included vegetation data as well as an overview of reproductive success. these are labled as follows:

Files for the prey capture analysis

Socal_RSF_data.txt

*description: Text file with vegetation data paired with capture locations both buffered polygons used in prey capture analysis and the unbuffered ones which were not used.(Pair with Socal_rsf_code R script) *format: .txt *Dimensions: 2641 X 35

*Variables: *ORIG_fid: completely unique identifier for each row *unique_id: unique identifier for each capture polygon(shared between a buffered capture location and its unbuffered pair) *territory_id: unique numerical idenifier of territory *Polygon_id: within territory unique prey capture polygon id *buff: bianary buffered or unbuffered (1=buffered, 0=unbuffered) *used: bianary used=1 available=0 *prey_type: prey species associated with polygon unkn:unknown, flsq:flying squirel, wora:woodrat, umou:mouse, pogo:pocketgopher, grsq: grey squirel, ubrd: unknown bird, umol:unknown mole, uvol, unknown vole. *area_sqm: area of polygon in square meters *CanCov_2020_buff: average canopy cover in polygon *CanHeight_2020_buff: average canopy height in polygon *Canlayer_2020_buff: average number of canopy layers in polygon *Understory_density_2020_buff: average brushy vegetation density in polygon *pix_COUNT: count of pixels in polygon (not needed for analysis) *p_chaparral: percent of polygon comprised of chaparral habitat
*p_conifer: percent of polygon comprised of conifer habitat *p_hardwood: percent of polygon comprised of hardwood habitat *p_other: percent of polygon comprised of other habitat types *Calveg_cap_CHt_gt10_CC_30to70_intersect_buff: percent of polygon comprised of trees taller than 10m with 30-70percent canopy cover (used to check data) *Calveg_cap_CHt_gt10_CCgt70_intersect_buff: percent of polygon comprised of trees taller than 10m with greater than 70percent canopy cover (used to check data) *Calveg_cap_CHt_lt10_intersect_buff:percent of polygon comprised of trees less than 10m (used to check data)
*p_sm_conifer: percent of polygon comprised of conifer trees less than 10m (used to calculate diversity)
*p_lrg_conifer_sc: percent of polygon comprised of conifer forests >10m tall with sparse canopy(used to calculate diversity) *p_large_conifer_dc: percent of polygon comprised of conifer forests greater than 10m tall with dense canopy (used to calculate diversity) *p_sm_hard: percent of polygon comprised of hardwood trees less than 10m (used to calculate diversity) *p_lrg_hard_sc: percent of polygon comprised of hardwood forests greater than 10m with sparse canopy(used to calculate diversity)
*p_lrg_hard_dc: percent of polygon comprised of hardwood forests greater than 10m dense canopy (used to calculate diversity) *p_forests_gt10_verysparse_CC: percent of polygon comprised of trees less than 10m with very sparse canopies (used to calculate diversity) *primary_edge: total distance in meters of primary edge in a polygon
*normalized_by_area_primary_edge: total distance in m of primary edge in a polygon divided by the area of the polygon
*secondary_edge: total distance in meters of secondary edge in a polygon *normalized_by_area_secondary_edge:total distance in m of secondary edge in a polygon divided by the area of the polygon *coarse_diversity: shannon diversity in each polygon (see methods below) *fine_diversity: shannon diversity in each polygon (see methods below) *nest_distance: distance from polygon center to nest for each polygon in meters

For the Delivery analysis

note: For information on determining average prey biomass see methods as well as zulla et al 2022 for flying squirels and woodrat masses Zulla CJ, Jones GM, Kramer HA, Keane JJ, Roberts KN, Dotters BP, Sawyer SC, Whitmore SA, Berigan WJ, Kelly KG, Gutiérrez RJ, Peery MZ. Forest heterogeneity outweighs movement costs by enhancing hunting success and fitness in spotted owls. doi:10.21203/rs.3.rs-1370884/v1. PPR:PPR470028.

prey_deliveries_byterritory.csv *Description: overview file of prey delivered to each nest *format: .csv *dimensions:332 x 8

*Variables: *SITE: Unique numerical identifier for each territory *DATE: date prey was delivered (in UTC) *CAMERA TIME: time in UTC prey was delivered *VIDEO TIME: time on video prey was delivered - unrelated to real time just original file
*PREY ITEM: prey species delivered to nest unkn:unknown, uncr: unknown if delivery(removed from eventual analysis due to

Indicators and calculation formulas for analyzing factors affecting the spatial distribution of traditional villages.

This hosted feature layer has been published in RI State Plane Feet NAD 83. This dataset was developed as an aid in the identification and protection of plant and animal species listed in the RI Natural Heritage Data.The Natural Heritage Areas were developed from a kernel density analysis of Heritage data element occurrences (EO). The calculation, based on a 30 meter pixel size, determines the mean number of EOs per square kilometer for each pixel within a one kilometer search radius. Non-statistically significant areas were eliminated and the remaining areas converted to a polygon dataset. Element Occurrences are discreet observations of a community or nesting site of State or Federally listed rare or threatened species OR species deemed noteworthy by the State.

The temple gardens are an important human landscape and have an important position in the Chinese garden system. Using GIS analysis tools, primarily the Nearest Neighbor Index, Kernel Density Estimation, and Spatial Autocorrelation, and employing a Geographic Detector model, we analyzed the spatial distribution characteristics and influencing factors of 4,317 temples and gardens in Jiangxi Province. Research shows that: 1) The spatial distribution type of temple gardens in Jiangxi Province is agglomeration type, with large spatial differences in distribution, forming a spatial distribution pattern of “generally dispersed and concentrated in some areas”; 2) the distribution of temple gardens in Jiangxi Province is uneven. They are mostly distributed in five prefecture-level cities: Ganzhou, Jiujiang, Shangrao, Fuzhou, and Nanchang; 3) The overall spatial distribution of temple gardens in Jiangxi Province has positive autocorrelation characteristics, and prefecture-level cities have significant proximity characteristics, forming a “high-high” “agglomeration” and “low-low agglomeration” distribution patterns; 4) Temple gardens in various regions are affected by geomorphological factors, and are mostly concentrated in the lower altitude range of 0–500 m and the gentle slope of 0°–30°. Most of the distribution density of temple gardens in various prefecture-level cities is within the buffer zone distance of the road network within the range of 0–1.5 km. 5) Economic, cultural, demographic, and historical factors have affected the development of temple gardens. Areas with more active economies have a denser number of temple gardens. The unique regional culture affects the distribution of temples and gardens in different regions. In places where the modern population is densely distributed, there are fewer temples and gardens, while in places where the population is less densely distributed, there are more temples and gardens. 6) The use of geographical detectors to detect influencing factors shows that the greatest impact on the spatial distribution of temple gardens in Jiangxi Province is the road network, followed by elevation, slope, GDP, and water systems. The research is conducive to scientific understanding of the distribution of temple gardens among prefecture-level cities in Jiangxi Province, and provides reference for strengthening the protection of temple gardens and exploring the tourism characteristics of temple gardens.

This shapefile contains four levels of occurrence (frequent, common, occasional, and rare range) for the Florida black bear (Ursus americanus floridanus) throughout the state of Florida. Range extent and levels of occurrence were created using research, management, and public-generated location data of black bears from 2009-2018. The four levels of occurrence were estimated by 90% kernel density estimator (KDE) isopleth (range of frequent occurrence), 97.5% KDE isopleth (common), concave hull model (occasional), and the remainder of Florida (rare).

Many species show range expansions or contractions due to climate-change-induced changes in habitat suitability. In cold climates, many species that are limited by snow are showing range expansions due to reduced winter severity. The European polecat (Mustela putorius) occurs over large parts of Europe with its northern range limit in southern Fennoscandia. However, it is to date unknown what factors limit polecat distribution. We thus investigated whether climate or land-use variables are more important in determining the habitat suitability for polecats in Sweden. We hypothesized that 1) climatic factors, especially the yearly number of snow days, drive habitat suitability for polecats, and that, 2) as the number of snow days is predicted to decline in the near future, habitat suitability in northern Sweden will increase. We used a combination of sightings data and a selection of national maps of environmental factors to test these hypotheses using MaxEnt models. We also used maps of future climate predictions (2021–2050 and 2063–2098) to predict future habitat suitability. The number of snow days was the most important factor, negatively determining habitat suitability for polecats, as expected. Consequently, the predictions showed an increase in suitable habitat both in the current distribution range and in northern Sweden, especially along the coast of the Baltic Sea. Our results suggest that the polecat distribution is limited by snow and that reduced snow cover will likely result in a northward range expansion. However, the exact mechanisms for how snow limits polecats are still poorly understood. Consequently, we expect the Scandinavian polecat population to increase in numbers, in contrast to many populations elsewhere in Europe, where numbers are declining. Due to polecat predation, the expansion of the species might have cascading effects on other wildlife populations. Methods Polecat sightings data To determine the distribution and habitat suitability of polecats in Sweden, we used sightings data of polecats gathered by volunteers and documented to the Swedish Species Information Centre between 1960 and 2020 (Figure 1; Swedish Species Information Centre 2020). We validated data at the edge of the distribution by contacting the person that reported the sighting, and removed data points if there was uncertainty about the sighting, we also removed data that was categorized as roadkill. As a result, we discarded 44 of the 425 sightings before analysis. Due to an increase of popularity of the sightings platform, the majority of sightings (78%) used in the analyses was from the period 2010–2020. Description of covariates We used nine covariates distributed over four different categories to test our hypotheses (Table 1). We included these covariates based on habitat and diet preferences of the polecat found in previous studies All parameters were rasterized and aggregated to 1-km2 grid cells over the whole of Sweden in ArcGis Pro version 2.6.0

Land cover and soil moisture

We selected several land-cover types and an index of soil moisture from the ungeneralised version of the National Land Cover Database (NMD) Sweden. This project provides a land cover map for the whole of Sweden divided into 24 different land-cover types, as well as a measure of soil moisture as a spectrum from dry to wet soil. We reclassified 17 of the 24 land-cover types into three different groups: coniferous forest, deciduous forest and open landscapes (Table S1.1). We selected these three groups for ease of analysis and due to previous studies showing that polecats selected or avoided these land-cover types at small spatial scales. Polecats were found to avoid coniferous forest (Baghli and Verhagen 2005, Zabala et al. 2005), select for deciduous forest (Jedrzejewski et al. 1993, Baghli et al. 2005), and use landscapes that were characterised by a variety of open habitat and forest (Blandford 1987, Lodé 1993, 2000a, Baghli et al. 2005). Furthermore, we used soil moisture as a variable that could further distinguish areas that would be wet during part of the year, which could result in increased amphibian populations, which are an important part of the polecat’s diet (Lodé 1993, 1997, 2000b, Hammershøj et al. 2004, Malecha and Antczak 2013). After reclassification, we determined the proportion of surface covered by each land-cover group, as well as the average soil moisture index, for each 1-km2 grid cell. Table S1.1: The landcover type clusters and which variables are merged from the original landcover data.

Cluster of land-cover types

Land-cover type number as presented in the NMD

Coniferous forest

111 (Pine forest not on wetland), 112 (Spruce forest not on wetland), 113 (Mixed coniferous not on wetland), 121 (Pine forest on wetland), 122 (Spruce forest on wetland), 123 (Mixed coniferous on wetland)

Deciduous forest

115 (Deciduous forest not on wetland), 116 (Deciduous hardwood forest not on wetland), 117 (Deciduous forest with deciduous hardwood forest not on wetland), 125 (Deciduous forest on wetland), 126 (Deciduous hardwood forest on wetland), 127 (Deciduous with deciduous hardwood forest on wetland)

Open landscapes

2 (Open wetland), 3 (Arable land), 41 (Non-vegetated other open land), 42 (Vegetated other open land), 118 (Temporarily non-forest not on wetland), 128 (Temporarily non-forest on wetland)

Snow cover and Minimum winter temperature The Swedish Meteorological and Hydrological Institute provides snow cover data for Sweden as the average number of days with snow with a depth above 20mm. The data is provided in 4 time periods, including two future projections (P1=1961–1990, P2 = 1991–2013, P3 = 2021–2050, P4= 2069–2098; Swedish Meteorological and Hydrological Institute 2021). The future projections we included are based on the 4.5 RCP scenario(Thomson et al. 2011). We included these data as a previous study showed that polecats had more difficulty catching prey when there is snow on the ground (Weber 1989). The data consists of interpolated data from 200 weather stations with an average given per municipality. We have rasterized the data giving average values for cells crossing municipality boundaries. We used data averages per cell of P2 for model building, while we used averages of P3 and P4 per cell for model projections of future scenarios.

Human pressure

We used the human footprint index as published by NASA in 2018 as a measure of human pressure. We did this as previous studies have shown that polecats tend to select for areas with extensive human use (Sidorovich et al. 1996, Rondinini et al. 2006) while avoiding urban centres (Zabala et al. 2005). The dataset is based on the global human footprint between 1995 and 2004. The human footprint is an index based on population density, land-use, infrastructure (buildings, lights, land use/cover) and human access (roads, railways; Venter et al. 2018). This raster dataset is publicly available and has a resolution of 1 km2. We only clipped the dataset to the borders of Sweden. Water availability We used the Water & Wetness geo data from Copernicus (CLMS 2018) as a measure of water availability. We did this as previous studies showed that polecats select for riparian habitat (Baghli et al. 2005) as amphibians are an important part of their diet (Lodé 1993, 1997, 2000b, Hammershøj et al. 2004, Malecha and Antczak 2013). This dataset includes all waterways and waterbodies with a resolution of 10 m. We have outlined all waterbodies and then made a buffer of 30 meters around all waterlines to represent near-water (riparian) habitat. We then calculated the proportion of near-water habitat in each 1-km2 grid cell. Elevation We used elevation data from the Copernicus Land Monitoring Service - EU-DEM project. We did this as previous studies showed that polecats avoid high-elevation areas. The dataset is provided as a raster with a spatial resolution of 25 meters. We calculated the average elevation for each 1-km2 grid cell. Bias correction for sampling intensity Due to the nature of citizen science data, it is prone to come with a bias. This bias manifests itself mostly in a discrepancy in spatial sampling effort. To account for this bias, we created a density kernel (as recommended by Kramer-Schadt et al. 2013 and Morelle and Lejeune 2015 and in line with Rutten et al. 2019) based on all mustelid sightings (n = 25686) reported to the Swedish Species Information Centre between 1972 and 2021 (Swedish Species Information Centre 2020), except for the Eurasian badger (Meles meles), the wolverine (Gulo gulo) and the polecat. We excluded the badger and wolverine as we expect this species to be much easier to identify and see compared to the polecat and other mustelids. Furthermore, badger and wolverine have a limited distribution in Sweden, while all other species – Eurasian otter (Lutra lutra), pine marten (Martes martes), American mink (Neovison vison), stoat (Mustela erminea), and weasel (Mustela nivalis) – have a distribution that covers the whole of Sweden (Swedish Species Information Centre 2020). We created the kernel with the ‘Kernel Density’ function in ArcGIS Pro (Esri 2021) with the mustelid sighting coordinates and the 1 km2 raster grid used for the covariates. The use of this density kernel is based on the assumption that people reporting other mustelids would also report a polecat if they saw one, and thus that the distribution of mustelid sightings is representative of the distribution of potential polecat reporters.