Data from: Space between houses influences movement and habitat selection of ungulates: Width as a novel metric of development
Data files
Jun 04, 2026 version files 6.42 GB
-
GPS_ModelData_Migration.csv
1.12 GB
-
GPS_ModelData_Winter.csv
5.29 GB
-
README.md
7.50 KB
-
WidthRaster_Cody.tif
7.62 MB
-
WidthRaster_Jackson.tif
5.99 MB
Abstract
Wildlife often lose access to habitat due to housing development. The magnitude of indirect habitat loss can be conditional on the configuration of individual houses, but commonly used metrics (i.e., density or distance) can overlook the configuration of development.
We introduce a novel framework to index the configuration of development based on the width of space between houses and associated structures. We use resource selection functions to assess the degree that GPS collared elk (Cervus canadensis), mule deer (Odocoileus hemionus), pronghorn (Antilocapra americana), and moose (Alces alces) on winter range and on migration routes use space between houses within northwest Wyoming, USA near the towns of Cody and Jackson. Further, to help inform regulations aiming to promote wildlife movement across a gradient of land uses, we differentiated between individuals residing in primarily rural and exurban areas.
Rural populations of elk, mule deer, pronghorn, and moose avoided spaces narrower than 2 km and never used spaces narrower than 50 m between houses, whereas exurban populations of elk, mule deer, and moose selected for spaces narrower than 2 km but avoided spaces narrower than 50 m. We identified cutoffs in rural and exurban areas where space may become too narrow for most animals to use.
Through this metric, managers and policy makers can inform the necessary width to maintain wildlife movement through corridors. Our width metric can be applied to other systems and our workflow is publicly available (https://wildlifemovetools.org/width-calculator) so users can estimate the width of space between structures in their focal areas.
Synthesis and applications. In rural areas, maintaining spaces between houses >500 m will likely facilitate ungulate movement. To fully conserve functional habitat, such as unimpeded habitat use along migration corridors, maintaining spaces approximately >2.5 km between houses will likely be necessary.
Dataset DOI: 10.5061/dryad.4xgxd25rk
Description of the data and file structure
The data release here comprise a movement analysis of big game wildlife (deer, elk, pronghorn, and moose) in northwestern Wyoming and how big game movement is affected by the width of space between houses and other buildings. Included to the data release are four different datasets. The study included 2 study areas in Wyoming: Cody and Jackson, Wyoming. The csvs are derived GPS collar data, cleaned and stripped of location due to data sensitivities.
- GPS_ModelData_Migration.csv - the data used for both univariate and multivariate RSFs of big game during migration.
- GPS_ModelData_Winter.csv - the data used for both univariate and multivariate RSFs of big game while on winter range.
- WidthRaster_Cody.tif - the raster used to extract the narrowest width between houses and other buildings for big game within the Cody study area.
- WidthRaster_Jackson.tif - the raster used to extract the narrowest width between houses and other buildings for big game within the Jackson study area.
Files and variables
File: GPS_ModelData_Migration.csv
Description: Model data used to derive univariate and multivariate RSFs for animals while on migration.
Variables
- AID_bioyear: Character. Animal ID and the biological year that the animal was tracked, where biological year was defined as starting on July 1st.
- studyarea_sp: Character. The study area and species.
- season: Character. Season the animal was tracked.
- case: Binary variable, whether a used or available sampled point. Used as the response variable in RSFs.
- Width_buildings: Numeric, 0 - 10,000 m. The narrowest width between houses and other buildings, in meters.
- type_used: Character, indicates whether the given row is a used or available sampled point for the RSF.
- jday: Numeric. Julien day. Will be NA for sampled available points.
- AID: Character. Animal ID.
- bio_year: Integer. Biological year.
- n_build: Numeric. The number of houses and other buildings observed in the given animal-year available domain. This number was used to differentiate between animals in primarily "rural" and "exurban" spaces.
- class: Character. Either rural (n_build/km2 <= 7) or exurban (n_build/km2 > 7)
- studyarea_sp2: Character. Grouping variable used to differentiate between each population/rural and exurban populations. This is a merge of the columns "studyarea_sp" and "class:
- sp_study: Character. Grouping variable, identical to studyara_sp2 just reformatted. Necessary for visualization.
- Width_buildf: Factor. Width between buildings rounded to one of 21 categories.
- slope: Numeric. Slope of the given point. Units are degrees.
- tpi: Numeric. Topographic position index.
- dist_ag: Numeric. Nearest distance to agriculture. In meters.
- dist_wetland: Numeric. Nearest distance to wetlands. In meters.
- shrub_cover: Numeric. Fractional shrub cover. Percent.
- perennial_biomass: Numeric. Perennial biomass (lb/acre).
- build_dens: Numeric. Number of buildings per km2. Result of n_build/area of available domain. Used to categorize animal-years as "rural" or "exurban".
- near_distbuild: Numeric. Nearest distance to houses and other buildings. For supplemental analysis. In meters.
- build_density: Numeric. Building density of the given used or available point based on a 1km2 radius focal window. For supplemental analysis. In buildings/km2.
File: GPS_ModelData_Winter.csv
Description: Model data used to derive univariate and multivariate RSFs for animals while on winter range.
Variables
- AID_bioyear: Character. Animal ID and the biological year that the animal was tracked, where biological year was defined as starting on July 1st.
- studyarea_sp: Character. The study area and species.
- case: Binary variable, whether a used or available sampled point. Used as the response variable in RSFs.
- Width_buildings: Numeric, 0 - 10,000 m. The narrowest width between houses and other buildings, in meters.
- type_used: Character, indicates whether the given row is a used or available sampled point for the RSF.
- jday: Numeric. Julien day. Will be NA for sampled available points.
- AID: Character. Animal ID.
- bio_year: Integer. Biological year.
- n_build: Numeric. The number of houses and other buildings observed in the given animal-year available domain. This number was used to differentiate between animals in primarily "rural" and "exurban" spaces.
- class: Character. Either rural (n_build/km2 <= 7) or exurban (n_build/km2 > 7)
- studyarea_sp2: Character. Grouping variable used to differentiate between each population/rural and exurban populations. This is a merge of the columns "studyarea_sp" and "class:
- sp_study: Character. Grouping variable, identical to studyara_sp2 just reformatted. Necessary for visualization.
- Width_buildf: Factor. Width between buildings rounded to one of 21 categories.
- slope: Numeric. Slope of the given point. Units are degrees.
- tpi: Numeric. Topographic position index.
- dist_ag: Numeric. Nearest distance to agriculture. In meters.
- dist_wetland: Numeric. Nearest distance to wetlands. In meters.
- shrub_cover: Numeric. Fractional shrub cover. Percent.
- perennial_biomass: Numeric. Perennial biomass (lb/acre).
- build_dens: Numeric. Number of buildings per km2. Result of n_build/area of available domain. Used to categorize animal-years as "rural" or "exurban".
- near_distbuild: Numeric. Nearest distance to houses and other buildings. For supplemental analysis. In meters.
- build_density: Numeric. Building density of the given used or available point based on a 1km2 radius focal window. For supplemental analysis. In buildings/km2.
File: WidthRaster_Cody.tif
Description: Raster dataset generated within Cody, Wyoming, to estimate the width of space between houses and other buildings. Units are in meters. Maximum possible value is 10 km.
File: WidthRaster_Jackson.tif
Description: Raster dataset generated within Cody, Wyoming, to estimate the width of space between houses and other buildings. Units are in meters. Maximum possible value is 10 km.
Code/software
Deriving all datasets based on program R version 4.4.0. Needed packages included "glmmTMB" (Brooks et al 2017), 'car' (Fox and Weisberg 2019), and 'sf' (Pebesma and Bivand 2023).
Fox J, Weisberg S (2019). _An R Companion to Applied Regression_, Third edition. Sage, Thousand
Oaks CA. https://www.john-fox.ca/Companion/.
Pebesma, E., & Bivand, R. (2023). Spatial Data Science: With Applications in R. Chapman and
Hall/CRC. https://doi.org/10.1201/9780429459016
Mollie E. Brooks, Kasper Kristensen, Koen J. van Benthem, Arni Magnusson, Casper W. Berg, Anders
Nielsen, Hans J. Skaug, Martin Maechler and Benjamin M. Bolker (2017). glmmTMB Balances Speed
and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. The R
Journal, 9(2), 378-400. doi: 10.32614/RJ-2017-066.
Access information
Data was derived from the following sources:
