Dynamic riskscapes for prey: Disentangling the impact of human and cougar presence on deer behavior using GPS smartphone locations
Data files
Apr 30, 2025 version files 94.52 MB
-
MD_dataset.RDS
94.52 MB
-
README.md
6.25 KB
Abstract
Dataset Summary
This dataset includes high-resolution animal movement and environmental data used to evaluate prey behavioral responses to spatial and temporal risk from predators and humans. Data were collected between 2015 and 2022 across two ecologically distinct study areas in the American Southwest (Book Cliffs, Colorado/Utah and Pine Valley, Utah). The dataset supports risk-sensitive habitat selection and movement analyses by integrating information on animal space use, human activity, and landscape features.
Structure and Contents
Wildlife GPS Data: Location data were collected from GPS-collared female mule deer (Odocoileus hemionus). Deer GPS fixes were recorded at 2-hour intervals. Each GPS location is associated with a phenological season (e.g., late gestation, rearing, hunting) and a diel period (day, night, crepuscular). To protect sensitive wildlife location information, all raw spatial coordinates have been removed or generalized. Derived environmental and behavioral metrics are retained for reuse. Full covariates are described in the README.
Environmental Covariates: Spatially explicit raster data include vegetation indices (NDVI), terrain ruggedness, and reclassified land cover categories (e.g., open, forested, shrub). These covariates were extracted for each animal location and included with the 'Wildlife GPS Data.'
Risk Surfaces: The dataset includes modeled probability surfaces representing the relative risk of human and cougar encounters, generated using resource selection functions. These covariates were extracted for each animal location and included with the 'Wildlife GPS Data.'
Derived Movement Data: Step lengths and turn angles were calculated between successive deer GPS fixes. These data were used in integrated step selection analyses to evaluate behavioral responses to risk. These covariates were calculated for each animal location and included with the 'Wildlife GPS Data.'
Reuse Potential
Raw human mobility data are licensed under a data use agreement prohibiting redistribution or public sharing. With permission only. Reuse potential statements in the associated publication refer to the full dataset, including withheld raw human mobility data, and do not apply to this data package.
Legal and Ethical Considerations
Appropriate institutional animal care and use committees approved all animal capture and handling procedures and followed professional guidelines. The dataset also includes anonymized human mobility data derived from GPS-enabled smartphones and obtained via a third-party provider (Gravy Analytics). These data were collected from consenting users in compliance with the California Consumer Privacy Act (CCPA) and the European General Data Protection Regulation (GDPR). However, raw human mobility data are licensed under a data use agreement prohibiting redistribution or public sharing. Users wishing to replicate or extend human mobility data analyses must secure access independently.
Dataset DOI: 10.5061/dryad.51c59zwkg
Description of the data and file structure
This dataset includes high-resolution animal movement and environmental data used to evaluate prey behavioral responses to spatial and temporal risk from predators and humans. Data were collected between 2015 and 2022 across two ecologically distinct study areas in the American Southwest (Book Cliffs, Colorado/Utah and Pine Valley, Utah). The dataset supports risk-sensitive habitat selection and movement analyses by integrating information on animal space use, landscape features, and modeled risk surfaces.
Important note: Only the wildlife GPS data and associated environmental covariates are included in this Dryad data package. The raw human mobility data (e.g., smartphone detections) and aggregated products used in the original study are not included in this repository due to privacy and data-sharing restrictions. The data were obtained through a commercial license and are therefore proprietary, subject to data-sharing restrictions, and handled in compliance with applicable privacy laws.
Reuse potential statements in the associated publication refer to the full dataset, including withheld raw human mobility data, and do not apply to this Dryad data package.
Files and variables
File: MD_dataset.RDS
Description:
Wildlife GPS Data: Location data were collected from GPS-collared female mule deer (Odocoileus hemionus). Deer GPS fixes were recorded at 2-hour intervals. Each GPS location is associated with a phenological season (e.g., late gestation, rearing, hunting) and a diel period (day, night, crepuscular). To protect sensitive wildlife location information, all raw spatial coordinates have been removed. Derived environmental and behavioral metrics are retained for reuse. Full covariates are described in Table 1.
Environmental Covariates: Spatially explicit raster data include vegetation indices (NDVI), terrain ruggedness, and reclassified land cover categories (e.g., open, forested, shrub). These covariates were extracted for each animal location and included with the 'Wildlife GPS Data.'
Risk Surfaces: The dataset includes modeled probability surfaces representing the relative risk of human and cougar encounters, generated using resource selection functions. These covariates were extracted for each animal location and included with the 'Wildlife GPS Data.'
Derived Movement Data: Step lengths and turn angles were calculated between successive deer GPS fixes. These data were used in integrated step selection analyses to evaluate behavioral responses to risk. These covariates were calculated for each animal location and included with the 'Wildlife GPS Data.'
Table 1. Variable key for wildlife GPS data.
| Variable | Description |
|---|---|
| burst_ | Unique movement burst identifier (each burst includes one used and multiple available steps) |
| sl_ | Step length (meters) between consecutive GPS locations |
| ta_ | Turn angle (radians) between consecutive steps |
| t1_ | Start time of movement step (POSIXct format) |
| t2_ | End time of movement step (POSIXct format) |
| dt_ | Duration between GPS fixes (e.g., "2 hours") |
| tod_end_ | Time-of-day category at the end of the step (day, night, or crepuscular) |
| case_ | Logical indicator for whether the step was used (TRUE) or available (FALSE) |
| step_id_ | Unique step identifier shared across used and available steps within a burst |
| cos_ta | Cosine of turn angle (used in movement modeling) |
| log_sl | Natural log of step length (used in modeling movement rate) |
| NDVI | Normalized Difference Vegetation Index at the end of the step |
| season | Biological season category (e.g., late_gestation, birthing_rearing, hunting) |
| elev_s2 | Elevation (meters) at the end of the step |
| TRI_s2 | Terrain Ruggedness Index at the end of the step |
| nlcd_s2 | Land cover classification at the end of the step (e.g., Forested, Shrub) |
| dt_rds | Distance (meters) to nearest improved road at the end location |
| lionRSF_s2 | Cougar encounter risk (relative probability) at the end of the step |
| season.year | Combined biological season and year identifier (e.g., birthing_rearing_2019) |
| hmd_avg_counts_season | Average number of smartphone detections (human mobility) during the season |
| hmd_prob_use_season | Modeled human encounter risk (relative probability) during the season |
| site | Study area identifier (BookCliffs or PineValley) |
| animalID | Anonymized individual animal identifier |
Code/software
Program R (version 4.3.1)
Human subjects data
The data provider has anonymized the data. Human data has been scaled such that personal information is no longer identifiable.
