Introduced wild pigs affect the foraging ecology of a native predator as both prey and scavenger
Data files
Nov 11, 2025 version files 696.89 MB
-
Parsons-et-al_Pigs-affect-cougar-foraging.zip
696.86 MB
-
README.md
29.69 KB
Abstract
Introduced species can disrupt trophic interactions by acting as novel predators, prey, or scavengers. Predicting impacts of these disruptions can be integral to conservation of native species and maintenance of ecological function, but it is challenging, especially for species involved in multiple trophic interactions. Wild pigs (Sus scrofa) are a widely introduced ungulate that can function as prey, predator, and scavenger, and disrupt ecosystems through many pathways. One underexplored avenue is how wild pigs alter predator-prey dynamics of mammalian carnivores. We explored how wild pigs affect cougar (Puma concolor) foraging ecology as both prey for cougars and scavengers of cougar kills. We monitored cougars in an area with wild pigs to evaluate impacts on cougar habitat selection, diet, and feeding behavior. We also monitored wild pigs scavenging at cougar kills to understand if scavenging by pigs affects cougar feeding behavior or kill rates. Habitat selection patterns indicated cougars selected for areas based on both landscape factors and prey density, but selection was stronger for native black-tailed deer (Odocoileus hemionus columbianus) than for pigs. We also documented that wild pigs, especially juveniles, were an important secondary prey item for cougars. Finally, we observed that wild pigs scavenged large prey items killed by cougars, which led to reduced feeding time by cougars at individual kills. However, wild pigs did not scavenge frequently enough to affect cougar kill rates. Our research highlights multiple impacts of wild pigs on cougar ecology, both as prey and scavengers, to illustrate how wild pigs disrupt ecosystems. Areas with a higher density of pigs or a lower density of native prey would likely see stronger impacts of wild pigs on predator-prey interactions, and management of wild pigs should account for these interactions, especially when native predators or prey are species of conservation concern.
Metadata for the manuscript 'Introduced wild pigs affect the foraging ecology of a native predator' published in Wildlife Biology by MA Parsons, JA Dellinger, KC Vercauteren, JV Lombardi, and JK Young.
The data and R code provided (within the main folder: Parsons-et-al_Pigs-affect-cougar-foraging.zip) here can be used to recreate all results from this manuscript. The data files included several raster layers of topographic and landcover variables, photo data from a camera grid, cougar kill site information, photo data from camera-monitored cougar kills, and cougar scat analysis data.
The R scripts conduct all analyses, including estimating the density of deer and pigs, analyzing cougar diet based on kill sites and scat, conducting a step selection analysis of cougar GPS collar data, conducting a resource selection function of cougar kill sites, and analyzing the effects of pig scavenging on cougar foraging behavior. The GPS collar data and kill site locations are not publicly available because of the sensitive nature of these data. Please contact Mitchell Parsons at parsonsmitch953@gmail.com regarding this data.
USAGE NOTES
This repository contains four main file types: csv, TIFF/tif, shp, and R files.
- csv: csv files are text files containing data in a row-by-column structure. Each row represents an observation, and each column represents a measurement collected for that observation. These files can be opened and viewed in a text editor or spreadsheet software or loaded into software like Program R for data formatting and analysis.
- Tiff/tif: TIFF files are a common format for image data. In the case of this repository, all TIFF files are spatially referenced raster images containing data on landscape characteristics (e.g., vegetation, topography). These files can be opened, viewed, and analyzed in Program R (as done here) or any GIS software.
- shp: Shapefiles are a common format for vector-based geographic information system (GIS) data, such as points, lines, and polygon.s The shapefiles in this repository include polygons of study area boundaries and polygons of cougar home ranges. They can be opened and used in Program R, Python, or any GIS software. A shapefile consists of multiple file types beyond the .shp (specifically, .cpg, .dbf, .prj, .sbn, and .sbx). The user only interacts directly with the .shp file, but the other files need to be in the same directory.
- R: R files are text files that contain R language code that can be executed in Program R or R Studio. This code is used to load and process data, conduct statistical analyses, and visualize results. Lines that start with "#" are comments that provide context for the code but are not executed.
DATA FILES
The data files include:
- Raster layers of landcover and topographic data
- Camera station and photo data from a camera grid used to monitor deer and pigs
- Cougar kill site locations and data
- Cougar scat data
- Camera station and photo data from cameras deployed at cougar kill sites to monitor scavengers
GIS Data
Vegetation
This raster layer is a 30x30m resolution raster of the log-transformed distance to forest cover. This layer was derived from the 2021 National Landcover Database (https://www.mrlc.gov/data; classes 41, 42, and 43).
This raster layer is a 30x30m resolution raster of the log-transformed distance to herbaceous cover. This layer was derived from the 2021 National Landcover Database (classes 71, 72, 73, 74, 81, 82).
This raster layer is a 30x30m resolution raster of the log-transformed distance to riparian cover. This layer was derived from the 2021 National Landcover Database (classes 11, 12, 90, 95).
This raster layer is a 30x30m resolution raster of the log-transformed distance to shrub cover. This layer was derived from the 2021 National Landcover Database (classes 51, 52).
This raster layer is a 30x30m resolution raster of the log-transformed distance to agricultural cover. This layer was derived from the 2021 National Landcover Database.
This raster layer is a 30x30m resolution raster of the proportion of forest cover within a 17x17 cell moving window (510m) derived from the 2021 National Landcover Database.
This raster layer is a 30x30m resolution raster of the proportion of herbaceous cover within a 17x17 cell moving window (510m) derived from the 2021 National Landcover Database.
This raster layer is a 30x30m resolution raster of the proportion of riparian cover within a 17x17 cell moving window (510m) derived from the 2021 National Landcover Database.
This raster layer is a 30x30m resolution raster of the proportion of shrub cover within a 17x17 cell moving window (510m) derived from the 2021 National Landcover Database.
This raster layer is a 30x30m resolution raster that is 1/0, indicating whether a cell occurs on land or over the ocean. This was used for mapping and masking other data to avoid calculating animal density at sea. This was derived from the 2021 National Landcover Database by summing all terrestrial landcover types and reclassifying any terrestrial landcover as 1.
These rasters are 30mx30m resolution rasters of the average seasonal NDVI value for four seasons: summer (May-July), fall (August-October), winter (November-January), and spring (February-April). These data are derived from MODIS 16-day data at the 250m scale.
Topographic
This raster layer is a 30x30m resolution raster of elevation. This layer was derived from the USGS 1/3 arc-second digital elevation model (https://apps.nationalmap.gov/downloader/)
This raster layer is a 30x30m resolution raster of slope calculated from resampled_DEM.tiff using the terra package in R.
This raster layer is a 30x30m resolution raster of the terrain ruggedness index calculated from resampled_DEM.tiff using the terra package in R.
This raster layer is a 30x30m resolution raster of the topographic position index calculated from resampled_DEM.tiff using the terra package in R.
This raster layer is a 30x30m resolution raster of the terrain ruggedness index averaged over a 17x17 cell moving window (510 m).
This raster layer is a 30x30m resolution raster of the topographic position index averaged over a 17x17 cell moving window (510m).
Animal Density Layers
These TIFF files are 30x30m rasters of predicted deer and pig density for the study area based on the selected random encounter and staying time models developed in the manuscript. There are three files for each species:
- average.tif is the estimated species density ignoring season-specific covariates in the model.
- _Summer.tifis the estimated species density incorporating summer-specific covariates in the model.
- _WWinter.tifs the estimated species density incorporating winter-specific covariates in the model.
These tiff files are 30x30m rasters of predicted deer and pig density averaged over a 17x17cell moving window (510m). These are derived from the DeerDensityPred*.tif and PigDensityPred*.tif files. There are three files for each species:
- *average_289cell.tif is the estimated species density ignoring season-specific covariates in the model.
- *_summer_289cell.tifis the estimated species density incorporating summer-specific covariates in the model.
- *winter_289cell.tif is the estimated species density incorporating winter-specific covariates in the model.
Cougar Home Ranges
*_HomeRange_MCP.shp are the estimated minimum convex polygon home ranges for each cougar used in the resource selection function of cougar kill sites. The first three alpha-numeric characters (e.g., SF1) are the cougar ID.
Other GIS Layers
This raster layer is a 30mx30m resolution raster that defines the subset of areas wwithinthe study area that are closed to hunting.
These three shape files are the boundary and rivers of the Fort Hunter Liggett and are used for mapping and figure making.
Camera Data
This csv file contains the camera station locations and deployment dates. Each row represents a camera deployment a,nd columns provide information on the deployment. There are 26 columns:
- Station ID - the name of the camera location; each location was sampled multiple times
- CSY - the camera-season-year combination of the deployment
- CSY2 - the camera-season-year combination plus a unique identifier when there were multiple deployments at a location within a season. This occurred when cameras were moved or replaced because of technical issues. This is a unique identifier for each deployment.
- UTM_X - the UTM easting of the camera deployment
- UTM_Y - the UTM northing of the camera deployment
- Habitat - a broad habitat category for the camera location
- Distance - the distance of the 100% detection zone in meters. All detection zones used a30-degreee field of view
- Set - was this location part of the "even" set of cameras or the "odd" set of cameras during our camera rotations
- Season - season of the camera deployment
- Year - year of the camera deployment
- Set Date - the date the camera was deployed
- Pull - the date the camera was removed from the location
- Deployment_days - the total length of the deployment
- Camera Type - The type of camera that was used: Browning (Brown), Bushnell (Bush), Reconyx. Old (O), New (N), UCD (from project partner)
- Problem_From - The start date of a problem with the camera that prevented detection (e.g., card filled, batteries died, animal damaged/displaced). Max of 4 problems for a single deployment. NAs indicate a lack of a problem period
- Problem_to - The end date of a problem with the camera. NAs indicate a lack of a problem period
- Problem_days - The number of days the problem persisted
This csv file contains the information on visits to cameras by deer and pigs and is used to run the Random Encounter Staying Time model. Each row represents a visit by a single animal to the camera. There are 10 columns:
- CSY - the camera-season-year combination of the camera deployment where the visit occurred
- CameraID - the station ID of the camera where the visit occurred
- Date - the date of the visit
- Time - the time the visit began
- season - the season of the visit
- year - the year of the visit
- Species - the species for each visit. Either deer or pig
- Start - the start time of the visit in seconds (origin 1/1/1970)
- Stop - the stop time of the visit in seconds (origin 1/1/1970)
- visit_length - the duration of the visit in seconds
These csv files contain the photo data from cameras used to monitor pig scavenging behavior at cougar kills. Each file contains data from the field season(s) indicated in the file name. Each row in these files represents data from a single photo. There are 15 columns.
- File - the image file name
- Relative path - the folder where the photo was stored. Also corresponds to the kill site ID.
- Folder - the parent folder containing all photos
- Date - the date the photo was captured
- Time - the time the photo was captured
- ImageQuality - a software-produced column indicated the quality of the photo
- DeleteFlag - indicates where a checkbox to have the photo deleted was checked.
- Species - the species of wildlife identified in the photo. NAs indicate no wildlife was documented.
- Species2 - the second species of wildlife identified in the photo is >1 species was present. NAs indicate no second species was present.
- SpeciesCount - the number of individuals of the first species
- Species2Count - the number of individuals of the second species
- IfOther - a text column that was used if the "Species" column was Other to identify uncommon species. NAs indicate that the "Species" column was not "Other"
- IfOther2 - a text column that was used if the "Species2" column was other to identify uncommon species. NAs indicate that the "Species2" column was not "Other"
- Comment - a text column that includes user-entered comments for the photo. NAs indicate that no comment was recorded
- Pig Behavior - a column that identifies the behavior of pigs if pigs were present. NAs indicate no pigs present
Cougar data and kill site information
We are unable to share these raw data because of their sensitive nature. We uploaded a censored version of this data that excludes GPS locations (cougar_clusters_noGPS.csv)
This csv file includes information on cougar kill site investigations. Each row represents a cluster of cougar GPS points that was investigated because it met a criterion to be a possible kill site. Each column provides information on the cluster. There are 62 columns.
- cluster_id - the name of the GPS cluster investigated
- cougar_id - the ID of the individual cougar associated with the cluster
- type - what type of cluster was investigated. The majority are "current" clusters indicated that they were investigated soon after formation. Other options include April (clusters investigated long after formation), day (presumed daybed locations), Den (confirmed den locations), and trapping (area where we deployed bait to trap animals). Only 'current' clusters were used in the analysis.
- collar_serial - the serial number of the cougar's GPS collar
- n_points - the number of points in the GPS cluster
- radius - the radius of the GPS cluster
- duration_h - the duration between the first and last point at the GPS cluster
- form_date - the date of the first point at the GPS cluster
- centroid_lat - the latitude of the centroid of the GPS cluster
- centroid lon - the longitude of the centroid of the GPS cluster
- inv_date - the date we investigated the GPS cluster
- personnel - initials of the investigators
- time_200m - the time of day we reached 200m from the cluster centroid
- habitat - broad habitat category that the cluster was in
- habitat2 - broad secondary habitat category that the cluster was in. NAs indicate that there was only one dominant habitat type (habitat column)
- canopy - whether the site had open, moderate, or closed canopy cover
- carcass_found - whether or not a carcass was found at the site
- time_found - the time of day the carcass was found. NA indicates that no carcass was found
- carcass_lat - the latitude the carcass was located at. NA indicates that no carcass was found
- carcass_lon - the latitude the carcass was located. NA indicates that no carcass was found
- species - the species of the located carcass. NA indicates that no carcass was found
- how_id - what features were used to identify the species of the carcass. NA indicates that no carcass was found
- sex - the sex of the carcass. F = female, M = male, UNK = unknown. NA indicates that no carcass was found
- how_sex - what features were used to determine the sex of the carcass. NA indicates that no carcass was found
- age - the age category of the animal that was killed. Neonate (<6 mo), Fawn/Calf (6 - 18 mo), yearling (18 - 30 mo), adult (>30 mo). NA indicates that no carcass was found
- how_age - what features were used to determine the age of the carcass. NA indicates that no carcass was found
- est_age - the estimated age of the carcass in months/years. NA indicates that no carcass was found
- cached - whether or not the carcass was cached currently, previously, or not at all. NA indicates that no carcass was found
- n_prey - the number of prey items at the cluster. NA indicates that no carcass was found
- n_piles - the number of separate piles of remains and evidence. NA indicates that no carcass was found
- condition - the condition of the carcass (intact, dismantled, dispersed). NA indicates that no carcass was found
- parts - a description of what parts of the carcass were located. NA indicates that no carcass was found
- bite_wounds - whether bite wounds were located. NA indicates that no carcass was found
- hemorrhaging - whether hemorrhaging was located and if so the location of hemorrhaging. NA indicates that no carcass was found
- fed_on - whether the carcass had been fed on. NA indicates that no carcass was found
- per_rem - the rough percentage of biomass remaining on the carcass. NA indicates that no carcass was found
- feeding_use - a description of the observed feeding sign on the carcass. NA indicates that no carcass was found
- marrow - the color and texture of the bone marrow from long bones. NA indicates that no carcass was found or marrow was not assessed
- drag_trail - whether a drag trail was located. NA indicates that no carcass was found
- length - the length of the drag trail. NA indicates that no drag trail was found
- kill_site - whether a kill site was located. NA indicates that no carcass was found
- kill_lat - the latitude of the kill site. NA indicates that no kill site was found
- kill_lon - the longitude of the kill site. NA indicates that no kill was found
- evidence_kill_site - the evidence used to identify the kill site. NA indicates that no kill site was found
- consistent_date - whether the condition of the carcass is consistent with the cluster formation date. NA indicates that no carcass was found
- confidence_cougar_kill - confidence level in whether a cougar killed or scavenged the prey. positive, probable, unknown, or scavenge. NA indicates that no carcass was found
- other_sign - sign of other species that was observed at the carcass. NA indicates that no carcass was found
- scavenged - whether or not the carcass was scavenged by other species. Y, N, or Likely. NA indicates that no carcass was found
- scavened_by - what other species scavenged the carcass. NA indicates that no evidence of scavenging was found
- kill_2 - whether there was a second kill at the cluster. NA indicates that no carcass was found
- species - the species of the second prey item. NA indicates that no second carcass was found
- cached - whether the second prey item was cached. NA indicates that no second carcass was found
- notes - any notes about the cluster investigation. NA indicates that no notes were recorded
- cache_photo - whether photos were taken of the cache
- carcass_photo - whether photos were taken of the carcass
- bite_photo - whether photos were taken of the bite wounds
- hem_photo - whether photos were taken of the hemorrhaging
- drag_photo - whether photos were taken of the drag trail
- kill_photo - whether photos were taken of the kill site
- coug_sign_photo - whether photos were taken of cougar sign
- scav_sign_photo - whether photos were taken of scavenger sign
- time_leave - the time of day we left the GPS cluster
We are unable to share these raw data because they include sensitive GPS locations. We have uploaded a sensored version that does not include coordinates.
This csv file contains summary information for each cougar GPS cluster. Each row represents a cluster and there are 12 columns.
- AID - the animal ID of the cougar that generated the cluster
- clus_ID - the animal specific cluster ID
- clus_start - the date and time of the first GPS location at the cluster
- clus_end - the date and time of the last GPS location at the cluster
- open - a Y/N indicator of whether the cluster was still active. All should be N except for those that I failed to update.
- dur - the duration in hours of the cluster. Calculated as clus_end minus clus_start
- locs - the total number of GPS locations associated with the cluster
- radius - the radius in meters of all points included in the cluster
- inv date - the date (MDY format) that the cluster was investigated by a researcher. NAs indicate a cluster that was not investigated
- found - a Y/N indicator of whether a prey item was located at the cluster. NAs indicate a cluster that was not investigated
- species - the species of prey item found at the cluster. NAs indicate a cluster that was not investigated or was investigated but no prey was found
- notes - any notes recorded about the cluster. NAs indicate no notes were recorded
This csv file contains information on GPS collars that were deployed on cougars. Each row represents a separate deployment. There are 9 columns.
- deployment_id - A unique numeric ID of the collar deployment
- collar_id - the GPS collar deployed for that deployment
- cougar_id - the cougar animal ID that the collar was deployed on
- start_date - the date (MDY format) the deployment began
- start_day - the numeric day the deployment began
- start_month - the numeric month the deployment began
- start_year - the numeric year the deployment began
- end_date - the date the deployment ended. NA indicates the collar was active at the end of the study period
- end_cause - the reason the deployment ended. NA indicates the collar was active at the end of the study period
This csv files contains information on when collared female cougars were known to have kittens with them. Each row represents a separate litter and there are 4 columns.
- cougar_id - the animal ID of the adult female who had kittens
- kitten_start - our best estimate at when kittens were born based on GPS collar data. Estimated based on evidence of denning behavior
- kitten_end - our best estimate at when the female no longer had dependent juvenlies. This estimate is based on camera and track evidence at kill sites and kill rates
- min_num - the minimum number of kittens known to be present based on photo data.
This csv file contains data collected from disecting cougar scats collected in the field. Each row represents a separate scat and there are 15 columns.
- Date - the date the scat was disected
- Obs - the initials of the person disecting the scat
- Sample_ID - the ID of the scat
- Species_1 - the species ID of the most common species found in the scat
- Sp1_% - the volumetric percentage of the most common species found in the scat
- Species_2 - the species ID of the second most common species found in the scat. NAs indicate that this many species were not found in the scat
- Sp2_% - the volumetric percentage of the second most common species found in the scat. NAs indicate that this many species were not found in the scat
- Species_3 - the species ID of the third most common species found in the scat. NAs indicate that this many species were not found in the scat
- Sp3_% - the volumetric percentage of the third most common species found in the scat. NAs indicate that this many species were not found in the scat
- Species_4 - the species ID of the fourth most common species found in the scat. NAs indicate that this many species were not found in the scat
- Sp4_% - the volumetric percentage of the fourth most common species found in the scat. NAs indicate that this many species were not found in the scat
- Notes - any notes that were recorded by the observer
- Season - the season the scat was deposited by the cougar. Only available for freshly collected scats. NAs indicate season of deposition was unknown
- Cluster - a Y/N indicator of whether or not the scat was collected at an investigated GPS cluster
- Method - the method of scat collection. Options are kill (collected at a cluster), dog (collected by scat detection dog teams), day (collected at an investigated day bed), and opportunistic (collected opportunitistically while doing fieldwork)
The species codes used in this data are:
- BONE - bone fragments unidentifiable to species
- ODHE - black-tailed deer
- SUSC - wild pig
- PUCO - cougar
- CALA - coyote
- LECA - black-tailed jackrabbit
- BONE (sm) - bone fragments from a small animal
- SYBA - cottontail rabbit
- PRLO - raccoon
- CECA - elk
- BONE (lg) - bone fragments from a large animal
- BIRD - unidentifiable bird remains
- UNK - remains of an unknown species
- Small Mammal - unknown small mammal
- SCGR - grey squirrel
- PLANT - plant matter
- OTBE - California ground squirrel
- MEGA - wild turkey
- CACA - beaver
- BONE (rodent) - bone fragments from an unidentified rodent
CODE FILES
- Extracting camera covariates. Also used later to make layers of predicted deer and pig density
- Run the random encounter staying time model for deer and pigs. Separate file for each species
- Estimate home ranges for collared cougars
- Analyze cougar kill site data
- Analyze cougar scat data
- Run a step selection function from cougar GPS collar data and evaluate error propogation on the analysis
- Run a resource selection function of the cougar kill site data and evaluate error propagation of the analysis
- Evaluate the effects of pig scavenging on cougar foraging behavior
-
01_CameraCovariates.R
This file uses the camera station data and GIS data to extract covariates for each camera location
It creates and saves a new file (CameraCovariates_NDVI_289_log.csv) that is needed by the random encounter staying time model
Then, starting on Line 174, this file creates rasters of predicted deer and pig density needed for the deer body condition models
The second half of this script cannot be run until the REST model has been run and output saved.
- 02_RESTAnalysis_NoVS_Randomeffect_deer_20240405_logunscale.R
- 03_RESTAnalysis_NoVS_Randomeffect_pig_20240405_logunscale.R
These files use the camera station data, camera covariate data (created from previous script), and deer and pig visit data.
It processes data and runs the random encounter staying time model for the species.
It then runs a cross validation of the model to test model performance
Note that these models can take several hours to run depending on the number of iterations, chains, and computer specs
The beginning of this code is also used to estimate deer and pig temporal overlap
These models require JAGS to run https://mcmc-jags.sourceforge.io/
- 04_HomeRanges_NewCollars.R
This file generate minimum convex polygon home range estimates for GPS collared cougars. The data required for this analysis is not provided because of its sensitive nature. Please contact the corresponding author with requests.
- 05_Kill_Summary_correctedages.R
This file uses the cougar cluster data to summarize cougar diet by sex and season. It then uses generalized linear models to evaluate if cougar foraging behavior differs by season, sex, or the presence of kittens with females.
- 06_Scatanalysis.R
This file uses the cougar scat data to summarize cougar diet by season and evaluate if diet composition changes between seasons or based on the method of scat collection.
- 07_LionSSF.R
This file prepares GPS collar data and runs a step selection function of cougar movements to evaluate if habitat selection is affected by landscape covariates, prey density, or both. The data required for this analysis is not provided because of its sensitive nature. Please contact the corresponding author with requests.
- 07-1_ErrorPropogation_SSF.R
This file repeates the step selection analysis over many iterations by pulling in different possible values from the deer and pig density models. This is intended to evaluate if our results were sensitive to error propogation from the density models. Note that this code can take >12 hours to run at the default settings.
- 08_RSF_multiple-animals_withprey.R
This file prepares the kill site data and runs a resource selection function of kill sites to evaluate if locations are related to landscape covariates, prey density, or both. The data required for this analysis is not provided because of its sensitive nature. Please contact the corresponding author with requests.
- 08-1_ErrorPropogation. R
This file repeates the resource selection analysis over many iterations by pulling in different possible values from the deer and pig density models. This is intended to evaluate if our results were sensitive to error propogation from the density models. Note that this code can take >12 hours to run at the default settings.
- 09_Pig Scavenging and Kill Rates_correctedages.R
This file uses the cougar cluster data and camera data from cougar kill sites to evaluate if scavenging by pigs affected cougar foraging at individual kill sites and over longer time intervals. It uses camera data to identify how frequently and how quickly pigs located cougar kills and the behavior displayed at that kill. It then uses regression models to evaluate if cougar feeding duration was effected by pig scavenging at individual kills or if long-term cougar kill rates were affected by the long-term prevalence of pig scavenging.
