Integrating human trail use in montane landscapes reveals larger zones of human influence for wary carnivores
Data files
Nov 25, 2024 version files 126.92 MB
-
README.md
16.37 KB
-
spatial-zoi.zip
126.90 MB
Abstract
Coexistence between humans and wildlife is necessary for many conservation goals but is difficult to achieve in landscapes with increasing human populations and species that are often wary of people and may also threaten human safety. In these contexts, coexistence may be enhanced by identifying geographic areas where animal movement is particularly important and changes to human use via trail design could support both wildlife conservation and human safety.
We used camera trap data to monitor the spatial distribution of grizzly bears (Ursus arctos), gray wolves (Canis lupus), and humans within the central Canadian Rocky Mountains, where anthropogenic development and human activity have gradually encroached on limited wildlife habitat. We quantified spatial variation in human use, and then incorporated this output into models for the detection rates of bears and wolves. We interpolated metrics of human use throughout the study area using inverse distance weighted averages of human detection rates from cameras. This approach supported a novel estimate of the cumulative effects of human use at all nearby trails on animal space use. We used our models to estimate the zone of influence of human use on bears and wolves, determining the distance at which human-use on nearby trails no longer exhibited a measurable change in detection rates for each of grizzly bears and wolves.
The negative effects of human use on wildlife declined steeply with distance such that 50% of the decrease in detection rates immediately adjacent to trails would be expected to occur at 267 m for grizzly bears and 576 m for wolves. Weak effects, 5% as strong as the effect adjacent to trails, extended up to 1.8 km and 6.1 km for grizzly bears and wolves, revealing the importance of cumulative measures of human use.
Our work shows how the distribution of human activity over entire landscapes can alter wildlife detection rates. Our results identify target buffer distances for protected areas near trails, and the modelling framework could be used by land managers to predict how altering trail networks and modifying human activity could affect wary wildlife species and advance coexistence.07-Nov-2024 --
https://doi.org/10.5061/dryad.s7h44j1h8
Description of the data and file structure
This Dryad repository contains all data necessary to replicate the analyses conducted by Thompson et al. (2024)* Journal of Applied Ecology*. These data were acquired through a decade-long study that involved deploying over 1,000 remote cameras that photographed wildlife and humans passing in front of them. Camera images were processed by countless volunteers and Parks staff.
Files and variables
File: spatial-zoi.zip
Description: This ZIP file contains the following files:
human_detections_SAMPLE.csv
: CSV file containing all detections of human recreationalists from a subset of cameras in our study area. Column headings:- t: time of observation (in days)
- rec_type: type of recreation associated with each observation; not relevant to this analysis but can be useful
- dogsonleash: number of dogs on leash recorded with each observation; not relevant to this analysis but can be useful
- dogsoffleash: number of dogs off leash recorded with each observation; not relevant to this analysis but can be useful
- horses: number of horses recorded with each observation; not relevant to this analysis but can be useful
- n_humans: number of humans recorded with each observation
- dataset: which data provider supplied information for this camera
- elevation: elevation above sea level in metres
- slope: slope of terrain in degrees
- terrain_ruggedness: Terrain Ruggedness Index (TRI)
- aspect_radians: Aspect of slope in radians
- snow_annual: Average annual fractional snow cover (% covered)
- barren: binary variable indicating whether the camera site belongs to the “barren / recently burned” landcover class
- conif_open: binary variable indicating whether the camera site belongs to the “open coniferous or deciduous forest” landcover class
- herbaceous: binary variable indicating whether the camera site belongs to the “herbaceous / wetland” landcover class
- shrub: binary variable indicating whether the camera site belongs to the “shrub” landcover class
- dtown_m: distance from nearest town in metres
- droad_m: distance from nearest road in metres
- trail_01_formal: binary variable indicating whether or not the site was located within 50 m of a formal trail
- trail_01_informal: binary variable indicating whether or not the site was located within 50 m of an informal trail
- road_01_paved: binary variable indicating whether or not the site was located within 50 m of a paved road
- road_01_unpaved: binary variable indicating whether or not the site was located within 50 m of an unpaved road
- disturbance_bv: binary variable indicating whether or not this location is within the Bow River Valley; not relevant for this analysis
- disturbance_bc: binary variable indicating whether or not this site is located in the backcountry area of Banff National Park; not relevant for this analysis
- dwater_m: distance from nearest waterbody in metres
- humans_per_day: average intensity of human use (detections/day) at the site where each detection was recorded
- biker_01: binary variable indicating whether the detection included a biker
- skier_01: binary variable indicating whether the detection included a skier
- vehicle_01: binary variable indicating whether the detection included an off-highway (motorized) vehicle
- site_name: the name given to the camera site where each detection was recorded
human_camera_effort_SAMPLE.csv
: CSV file containing all information about camera effort (times at which cameras were active) as well as associated covariates relevant to the “human use model”.- DateStart: time at which the camera was turned on, in days
- DateLastWorking: time at which the camera was turned off or lost battery, in days
- elevation: elevation above sea level in metres
- slope: slope of terrain in degrees
- terrain_ruggedness: Terrain Ruggedness Index (TRI)
- aspect_radians: Aspect of slope in radians
- snow_annual: Average annual fractional snow cover (% covered)
- barren: binary variable indicating whether the camera site belongs to the “barren / recently burned” landcover class
- conif_open: binary variable indicating whether the camera site belongs to the “open coniferous or deciduous forest” landcover class
- herbaceous: binary variable indicating whether the camera site belongs to the “herbaceous / wetland” landcover class
- shrub: binary variable indicating whether the camera site belongs to the “shrub” landcover class
- dtown_m: distance from nearest town in metres
- droad_m: distance from nearest road in metres
- trail_01_formal: binary variable indicating whether or not the site was located within 50 m of a formal trail
- trail_01_informal: binary variable indicating whether or not the site was located within 50 m of an informal trail
- road_01_paved: binary variable indicating whether or not the site was located within 50 m of a paved road
- road_01_unpaved: binary variable indicating whether or not the site was located within 50 m of an unpaved road
- disturbance_bv: binary variable indicating whether or not this location is within the Bow River Valley; not relevant for this analysis
- disturbance_bc: binary variable indicating whether or not this site is located in the backcountry area of Banff National Park; not relevant for this analysis
- dwater_m: distance from nearest waterbody in metres
- site_name: the name given to the camera site where each detection was recorded
grizzly_detections_SAMPLE.csv
: CSV file containing all detections of grizzly bears from a subset of cameras in our study area.- t: time of observation (in days)
- rec_type: species observed (here, always equal to “Grizzly Bear”)
- dogsonleash: number of dogs on leash recorded with each observation; not relevant to this analysis but can be useful
- dogsoffleash: number of dogs off leash recorded with each observation; not relevant to this analysis but can be useful
- horses: number of horses recorded with each observation; not relevant to this analysis but can be useful
- n_humans: number of humans recorded with each observation
- dataset: which data provider supplied information for this camera
- elevation: elevation above sea level in metres
- slope: slope of terrain in degrees
- terrain_ruggedness: Terrain Ruggedness Index (TRI)
- aspect_radians: Aspect of slope in radians
- snow_annual: Average annual fractional snow cover (% covered)
- barren: binary variable indicating whether the camera site belongs to the “barren / recently burned” landcover class
- conif_open: binary variable indicating whether the camera site belongs to the “open coniferous or deciduous forest” landcover class
- herbaceous: binary variable indicating whether the camera site belongs to the “herbaceous / wetland” landcover class
- shrub: binary variable indicating whether the camera site belongs to the “shrub” landcover class
- dtown_m: distance from nearest town in metres
- droad_m: distance from nearest road in metres
- trail_01_formal: binary variable indicating whether or not the site was located within 50 m of a formal trail
- trail_01_informal: binary variable indicating whether or not the site was located within 50 m of an informal trail
- road_01_paved: binary variable indicating whether or not the site was located within 50 m of a paved road
- road_01_unpaved: binary variable indicating whether or not the site was located within 50 m of an unpaved road
- disturbance_bv: binary variable indicating whether or not this location is within the Bow River Valley; not relevant for this analysis
- disturbance_bc: binary variable indicating whether or not this site is located in the backcountry area of Banff National Park; not relevant for this analysis
- dwater_m: distance from nearest waterbody in metres
- humans_per_day: average intensity of human use (detections/day) at the site where each detection was recorded
- biker_01: binary variable indicating whether the detection included a biker
- skier_01: binary variable indicating whether the detection included a skier
- vehicle_01: binary variable indicating whether the detection included an off-highway (motorized) vehicle
- site_name: the name given to the camera site where each detection was recorded
grizzly_camera_effort_SAMPLE.csv
: CSV file containing all information about camera effort (times at which cameras were active) as well as associated covariates relevant to the “wildlife models”.- DateStart: time at which the camera was turned on, in days
- DateLastWorking: time at which the camera was turned off or lost battery, in days
- elevation: elevation above sea level in metres
- slope: slope of terrain in degrees
- terrain_ruggedness: Terrain Ruggedness Index (TRI)
- aspect_radians: Aspect of slope in radians
- snow_annual: Average annual fractional snow cover (% covered)
- barren: binary variable indicating whether the camera site belongs to the “barren / recently burned” landcover class
- conif_open: binary variable indicating whether the camera site belongs to the “open coniferous or deciduous forest” landcover class
- herbaceous: binary variable indicating whether the camera site belongs to the “herbaceous / wetland” landcover class
- shrub: binary variable indicating whether the camera site belongs to the “shrub” landcover class
- dtown_m: distance from nearest town in metres
- droad_m: distance from nearest road in metres
- trail_01_formal: binary variable indicating whether or not the site was located within 50 m of a formal trail
- trail_01_informal: binary variable indicating whether or not the site was located within 50 m of an informal trail
- road_01_paved: binary variable indicating whether or not the site was located within 50 m of a paved road
- road_01_unpaved: binary variable indicating whether or not the site was located within 50 m of an unpaved road
- disturbance_bv: binary variable indicating whether or not this location is within the Bow River Valley; not relevant for this analysis
- disturbance_bc: binary variable indicating whether or not this site is located in the backcountry area of Banff National Park; not relevant for this analysis
- dwater_m: distance from nearest waterbody in metres
- site_name: the name given to the camera site where each detection was recorded
site_data_SAMPLE.csv
: CSV file containing relevant information about each camera site included in this subset of the study area.- dataset: which data provider supplied information for this camera
- easting: UTM easting, in metres, of the site. All locations jittered (by equal values, so distance calculations remain accurate) so exact camera locations cannot be determined.
- northing: UTM northing, in metres, of the site. All locations jittered (by equal values, so distance calculations remain accurate) so exact camera locations cannot be determined.
- uptime_days: total number of days in which each camera was on
- tot_humans: total number of detections of human recreationalists on each site
- tot_bikes: total number of detections of recreationalists on bikes at each site
- tot_hikers: total number of detections of hikers at each site
- tot_vehicles: total number of detections of recreationalists with (motorized) off-highway vehicles at each site
- tot_horses: total number of detections of horses (with recreationalists) at each site
- tot_skiers: total number of detections of skiers at each site
- tot_grizzlies: total number of detections of grizzly bears at each site
- tot_wolves: total number of detections of wolves at each site
- tot_offleash: total number of detections of off-leash dogs (with recreationalists) at each site
- humans_per_day: number of detections per day of human recreationalists on each site
- bikes_per_day: number of detections per day of recreationalists on bikes at each site
- hikers_per_day: number of detections per day of hikers at each site
- vehicles_per_day: number of detections per day of recreationalists with (motorized) off-highway vehicles at each site
- horses_per_day: number of detections per day of horses (with recreationalists) at each site
- skiers_per_day: number of detections per day of skiers at each site
- grizzlies_per_day: number of detections per day of grizzly bears at each site
- wolves_per_day: number of detections per day of wolves at each site
- offleash_per_day: number of detections per day of off-leash dogs (with recreationalists) at each site
- elevation: elevation above sea level in metres
- slope: slope of terrain in degrees
- terrain_ruggedness: Terrain Ruggedness Index (TRI)
- aspect_radians: Aspect of slope in radians
- dtown_m: distance from nearest town in metres
- dpavedroad_m: distance from nearest paved road in metres
- snow_annual: Average annual fractional snow cover (% covered)
- barren: binary variable indicating whether the camera site belongs to the “barren / recently burned” landcover class
- conif_open: binary variable indicating whether the camera site belongs to the “open coniferous or deciduous forest” landcover class
- herbaceous: binary variable indicating whether the camera site belongs to the “herbaceous / wetland” landcover class
- shrub: binary variable indicating whether the camera site belongs to the “shrub” landcover class
- dwater_m: distance from nearest waterbody in metres
- in_study: whether or not each site should be included in the analysis (some sites were out of bounds for certain covariate layers and were removed as a result)
- trail_01_formal: binary variable indicating whether or not the site was located within 50 m of a formal trail
- trail_01_informal: binary variable indicating whether or not the site was located within 50 m of an informal trail
- road_01_paved: binary variable indicating whether or not the site was located within 50 m of a paved road
- road_01_unpaved: binary variable indicating whether or not the site was located within 50 m of an unpaved road
- droad_m: distance from nearest road in metres
- disturbance_bv: binary variable indicating whether or not this location is within the Bow River Valley; not relevant for this analysis
- disturbance_bc: binary variable indicating whether or not this site is located in the backcountry area of Banff National Park; not relevant for this analysis
- site_name: the name given to the camera site where each detection was recorded
humantraffic.tif
: Model-predicted human use intensity across the entire study area, exactly as generated in our manuscript. We note that given a value of the “alpha” parameter from the wildlife models, one can derive disturbance indices from this raster for any location in the study area.specialcovs_human_20231031.R
: C++ likelihood function for the “human use model” that is fit to human recreationalist camera detection data. This function is implicitly used by the R script via the TMB software library.specialcovs_20231016.R
: C++ likelihood function for the “wildlife model” that is fit to grizzly bear camera detection data. This function is implicitly used by the R script via the TMB software library.FUNCTIONS.R
: R script containing functions necessary to run the scripts below. Each function is documented here.fit_human_model.R
: Code used to fit the model for human recreational use intensity in a subset of the study area.fit_wildlife_model.R
: Code used to fit the models for grizzly bear detection intensity in a subset of the study area, using the human traffic raster generated from the entire study area.
Code/software
All analyses included in this Dryad repository can be run using R, which is free and open-source. The scripts require the following R software packages, which are all free and can be downloaded using the install.packages
R function:
- TMB (Template Model Builder)
- tmbstan
- rstan
- tidyverse
- terra
- data.table
Access information
Other publicly accessible locations of the data:
- N/A