A case for human mobility data applications in wildlife management
Data files
May 08, 2025 version files 890.31 KB
-
README.md
4.19 KB
-
SierraInteractions_Centered.csv
886.11 KB
Abstract
Human activities have significantly altered terrestrial ecosystems, leading to biodiversity loss and habitat fragmentation. Traditional methods for measuring human impacts often lack the precision required for localized assessments, fail to capture temporal dynamics, or are scale-limited. Human mobility data (HMD) from GPS-enabled smartphone applications offers a valuable approach to understanding human movement patterns, overcoming many of these limitations.
We present case studies that demonstrate the use of HMD in assessing human activity within ecologically sensitive habitats (e.g., winter ranges and breeding grounds) and in evaluating where human-wildlife interactions are likely to occur to inform proactive conflict management. We also discuss how HMD can improve inference from habitat connectivity analyses and provide detailed, timely insights on how HMD can broadly support conservation and wildlife management goals.
HMD revealed patterns of recreational use in bighorn sheep's (Ovis canadensis) winter range and helped inform the timing and scope of protective measures for sheep. HMD also allowed us to quantify the frequency and timing of potential wildlife-human interactions, such as cougar (Puma concolor) proximity to human activity, identifying high human-wildlife conflict areas and opportunities to mitigate mortality risks (e.g., road crossings).
We discussed how future uses of HMD can improve conservation and connectivity, allowing managers to assess barriers to movement, identify critical thresholds of human disturbance, and refine strategies for mitigating impacts on species sensitive to human disturbance. We provided a simple example by highlighting differences in human use of a migration corridor for pronghorn (Antilocapra americana).
Synthesis and applications: HMD provides a transformative tool for wildlife management by offering scalable, dynamic insights into human activity that traditional methods cannot fully capture. These data enable managers to prioritize intervention areas, improve compliance with management zones, mitigate conflict risks, and enhance connectivity for sensitive species, supporting effective conservation strategies in a human-dominated world.
Dataset DOI: 10.5061/dryad.sqv9s4nfb
Description of the data and file structure
This dataset includes the anonymized cougar data used in the cougar–human interaction analysis. To assess spatial and temporal patterns of human–cougar interactions, GPS collar data from eight mountain lions were collected at 3-hour intervals in California’s eastern Sierra Nevada mountains as part of an ongoing California Department of Fish and Wildlife study. These data were paired with smartphone-based Human Mobility Data (HMD) from anonymized users to identify spatiotemporal proximity events between cougars and people. Potential interactions were classified into high, medium, and low probability categories based on distance and time thresholds (≤75m/30min, ≤150m/60min, ≤300m/120min, respectively). This design allowed researchers to evaluate where, when, and how frequently cougars came into proximity with humans, to identify patterns of human recreation that may increase interaction risk and inform proactive conflict mitigation strategies. All animal handling followed approved welfare protocols.
NOTE: Due to the sensitive nature of wildlife GPS data, particularly for large carnivores like cougars in California, location information has been modified to protect individual animals and conservation efforts. Specifically, we subtracted the mean X-coordinate of the cougar locations involved with an interaction from all cougar and HMD X-coordinates and then repeated the process with the Y-coordinates. HMD used in the analysis are proprietary and cannot be shared due to legally binding data-sharing agreements with a commercial provider (Supporting Information 1). As such, we cannot release the raw HMD underlying the analysis. However, we provide derived variables from the HMD, including spatial and temporal differences between cougar and human locations, and the distance from the nearest anthropogenic features (e.g., buildings, roads) as used in the interaction modelling in case study two. Other raw datasets, including winter closure zone data provided by Grand Teton National Park, are similarly restricted from public release due to their sensitive nature.
Files and variables
File: SierraInteractions_Centered.csv
Description:
Variables
- cougarID: Unique cougar ID number
- distance_m: Distance (m) between the cougar and HMD locations
- month: Month of the potential interaction
- hour: Hour of the potential interaction
- DiffTime_s: Difference (in seconds) between the cougar and HMD locations
- dist_road: Distance (m) of HMD location to the nearest roadway
- dist_build: Distance (m) of HMD location to the nearest building
- X_cougar: Centered X-location of the cougar
- Y_cougar: Centered Y-location of the cougar
Code/software
All scripts were run in the RStudio environment (v2022.12.0+353), which facilitates project organization and reproducibility.
The workflow included:
Importing and preprocessing: Cougar GPS collar data and Human Mobility Data (HMD) were imported as .rds and spatial .shp files. Timestamps were standardized using lubridate, and spatial layers were projected using sf.
Interaction classification: Distances and time differences between cougar and smartphone user locations were calculated. Based on predefined spatiotemporal thresholds, proximity events were classified into high, medium, or low probability categories using dplyr and custom R functions.
Summarization and visualization: Summary statistics and monthly or hourly interaction trends were computed with data.table, and plots were generated using ggplot2.
sf (v1.0-14) – for spatial data manipulation and projection
dplyr (v1.1.2) – for data wrangling
lubridate (v1.9.2) – for date-time processing
ggplot2 (v3.4.2) – for visualization
data.table (v1.14.8) – for efficient data handling of large datasets
Access information
Other publicly accessible locations of the data:
- None
Data was derived from the following sources:
- Humans: Unacast
- Cougars: California Department of Fish and Wildlife
