Data from: Survival and causes of mortality of introduced wild pigs in the presence of cougars
Data files
Dec 06, 2024 version files 142.22 MB
-
PigSurvivalwithCougars.zip
142.21 MB
-
README.md
12.03 KB
Abstract
Wild pigs (Sus scrofa) are an introduced species that affect ecosystems through myriad pathways. They are the target of population control efforts, but one outstanding question is the extent to which native predators affect population demography. Cougars (Puma concolor) are a known predator of wild pigs and cooccur with wild pigs in several regions of the Americas. However, investigations of wild pig survival in regions with cougars are lacking. We investigated the survival and causes of mortality for wild pigs in California, USA where cougars could act as a predator. We also explored habitat selection of wild pigs with respect to cougar predation risk. We collared 69 wild pigs, 15 adults, 15 subadults, and 39 juveniles. Captures occurred in winter and spring and survival estimates primarily reflect spring and summer survival. We estimated 6-month survival of adults and juveniles to be 0.923 (95% confidence interval: 0.789 – 1.00) and 0.563 (0.416 – 0.762), respectively, which are similar to estimates from areas without cougars. Only juvenile and one subadult wild pig were killed by cougars, suggesting that cougars are a limited predator of adults. We also investigated feeding locations of 3 GPS-collared cougars to generate a spatial layer of cougar kill occurrence. We used this layer in an integrated step selection function for wild pigs and observed wild pigs selecting areas with low probability of cougar kill occurrence during times when cougars are most active. Combined, these results suggest that cougars are a minor source of mortality for wild pigs during the spring and summer and may be unlikely to provide substantial biological resistance against the spread of wild pigs. Our monitoring occurred when native prey (e.g. ungulate neonates) are most abundant which may influence cougar predation of pigs. It is likely that a preference for native prey, the novelty of wild pigs, wild pig habitat selection, and the potential danger of adult wild pigs reduces cougar predation.
README: Data from: Survival and causes of mortality of introduced wild pigs in the presence of cougars
Metadata for the manuscript 'Pig Survival and causes of mortality of introduced wild pigs in the presence of cougars' published in Biological Invasions by MA Parsons, KC Vercauteren, JA Dellinger, and JK Young.
The data and R code provided here can be used to recreate results from this manuscript. The data files included several raster layers of topographic and landcover variables, capture and survival monitoring data for collared wild pigs, and a probability of cougar kill occurrence raster.
The R scripts conduct all analyses including survival analysis of wild pigs, a resource selection function of cougar kill sites, and a step selection function for wild pigs.
DATA FILES
The datafiles include:
- Raster layers of landcover and topographic data
- Capture and monitoring data for wild pigs
- Cougar kill site locations and data
log_forest_dist.tiff
This raster layer is a 30x30m resolution raster of the log transformed distance to forest cover (m). This layer was derived from the 2021 National Landcover Database (https://www.mrlc.gov/data; classes 41, 42, and 43).
log_grass_dist.tiff
This raster layer is a 30x30m resolution raster of the log transformed distance to herbaceous cover (m). This layer was derived from the 2021 National Landcover Database (classes 71, 72, 73, 74, 81, 82).
log_riparian_dist.tiff
This raster layer is a 30x30m resolution raster of the log transformed distance to riparian cover (m). This layer was derived from the 2021 National Landcover Database (classes 11, 12, 90, 95).
log_shrub_dist.tiff
This raster layer is a 30x30m resolution raster of the log transformed distance to shrub cover (m). This layer was derived from the 2021 National Landcover Database (classes 51, 52).
propforested_25cell.tif
This raster layer is a 30x30m resolution raster of proportion of forest cover in a 5x5 cell moving window. This layer was derived from the 2021 National Landcover Database (classes 41, 42 and 43).
propgrass_25cell.tif
This raster layer is a 30x30m resolution raster of proportion of herbaceous cover in a 5x5 cell moving window. This layer was derived from the 2021 National Landcover Database (classes 71, 72, 73, 74, 81, 82).
propriparian_25cell.tif
This raster layer is a 30x30m resolution raster of proportion of riparian cover in a 5x5 cell moving window. This layer was derived from the 2021 National Landcover Database (classes 11, 12, 90, 95).
propshrub_25cell.tif
This raster layer is a 30x30m resolution raster of proportion of shrub cover in a 5x5 cell moving window. This layer was derived from the 2021 National Landcover Database (classes 51, 52).
is_land.tiff
This raster layer is a 30x30m resolution raster of whether a cell is terrestial land or open water. This layer was derived from the 2021 National Landcover Database (all classes except open water) and was used to mask ocean areas from modeled habitats.
resampled_DEM.tiff
This raster layer is a 30x30m resolution raster of elevation (m). This layer was derived from the USGS 1/3 arc second digital elevation model (https://apps.nationalmap.gov/downloader/)
resampled_TRI.tiff
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 measure is unitless.
resampled_TPI.tiff
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 measure is unitless.
LionPredRisk_logdist.tiff
This raster layer is a 30x30m resolution raster of the probability of cougar kill occurrence based on cougar kill site resource selection fucntion modeling. We derived this layer from observed cougar kill sites, but cannot share the raw data because of it's sensitive nature.
pig_captures_MJ23_update.csv
This csv file contains the capture information for each indiviudal pig and is used in the survival analysis. Each row represents a capture event and columns provide information on the event and the individual pig. There are 26 columns:
- pig_id - the ID of the individual pig
- date - the date of the capture event
- day - the day of the month of the capture event
- month - the month of the capture event
- year - the year of the capture event
- location - a rough location of the capture event within Fort Hunter Liggett, USA. (TA22 = training area 22, stonyres = Lower Stony Reservoir, peanut = Peanut Hill area, riverrd = Along River Road)
- method - type of trap used to capture the animal
- type - whether the individual pig was a new capture (initial) or a recapture
- habitat - a rough habitat category of the trapping location
- drugged - whether the individual pig was chemically immobilized
- drug_administered - what drug was used if chemical immobilization occurred
- reversal_administered - what reversal drug was used
- collar_id - the serial number for the GPS collar deployed on the pig
- ear_tag_L_color - the color of the ear tag affixed to the pigs left ear
- ear_tag_L_number - the number of the ear tag affixed to the pigs left ear
- ear_tag_R_color - the color of the ear tag affixed to the pigs right ear
- ear_tag_R_number - the number of the ear tag affixed to the pigs right ear
- vhf_eartag_ear - whether or not the pig received a VHF eartag
- age_class - the age class of the individual pig assigned by estimated weight (A = adult (>45kg), S = subadult (27-45 kg), J = juvenile (<27 kg))
- nursing/pregnant - whether female pigs were nursing or pregnant
- current/past - whether nursing/pregnancy was current or previous
- est_weight - the estimated weight of the pig (lbs)
- hair_taken - whether a hair sample was collected
- blood - whether a blood sample was collected
- notes - any additional notes associated with the capture
NA indicates columns that were not relevant for a particular capture (e.g., drug information for a pig that was not drugged, collar ID for a pig that was not collared).
PigEartagLocations3.csv
This csv file includes monitoring data for all tagged and collared pigs used in the survival analysis. Each row represents a detection of a pig and provides information on the pigs location and status. There are 7 columns:
- Date - the date the pig was detected
- Pig_ID - the ID of the individual pig
- Training_Area - the training area ("TA", location within Fort Hunter Liggett, CA) that the pig was located
- Signal_type - whether the detected collar/tag signal was in normal operation or mortality mode
- Status - whether the pig was alive (0) or dead (1)
- Cause - If a mortality occurred, the determined cause of mortality
- Notes - Any relevant notes related to the detection. Examples include notes on mortalities or comments on tag malfunctions (e.g., tracked a mortality signal to a live pig)
NA indicates that no notes were associated with the entry.
pig_collar_deployments.csv
This csv file includes information on individual deployments of GPS collars on pigs. Each row represents a different deployment with a unique combination of collar_ID and pig_ID. Individual pigs may have been collared more than one and individual collars may have been used on more than one pig during the project. There are 12 columns:
- deployment_id - a unique identifier for each collar deployment
- collar_id - the five digit unique ID for the deployed GPS collar
- pig_id - the alphanumeric code identifying the pig the collar was deployed on
- start_date - the mdy date when the collar was deployed
- start_day - the day of the month the collar was deployed
- start_month - the month (1-12) the collar was deployed
- start_year - the four digit year the collar was deployed
- end_deployment - the mdy date when the deployment ended
- end_day - the day of the month the deployment ended
- end_month - the month (1-12) the deployment ended
- end_year - the four digit year the deployment ended
- end_cause - the reason the deployment ended
Files in the RawData/PigLocations folder
Each of these files contains the raw GPS collar location data for an individual collar on an individual pig.
The naming convention of the files is:
- "collar" followed by the GPS collars ID number, this matches collar_id in the pig capture data.
- the alphanumeric pig ID. This matches the pig_id column in the pig capture data
- the mdy date that the collar was deployed followed by the mdy date that the collar was removed from the pig.
These files contain a large number of columns related to the technical details of each GPS location. The primary columns of interest are:
- CollarID - the five digit collar ID that matches the collar_id column in the pig capture data.
- UTC_Date - the date the GPS location was taken in UTC.
- UTC_Time - the time the GPS location was taken in UTC.
- Latitude [°] - the degrees latitude of the GPS location in WGS84.
- Longitude [°] - the degrees longitude of the GPS location in WGS84.
- DOP - a measure of accuracy for the GPS location with higher values indicating lower precision.
N/As indicate missing data which includes schedule fixes that were not successfully recorded/transfered (i.e., missed fixes) and missing data from successful fixes.
Cougar GPS collar data
We are unable to share these raw data because of their sensitive nature.
cougar_cluster_investigations
We are unable to share these raw data because of their sensitive nature.
CODE FILES
- Estimate pig survival rates
- Estimate cougar home range areas
- Estimate RSF of cougar kills for cougar predation risk
- Estimate SSF of pig GPS collars to understand if they avoid areas of high cougar risk.
01_Pig survival_to share.R
This file uses pig capture history data to estimate survival rates using Kaplan-Meier esitmation method.
Then uses cox proportional hazard models to estimate if survival varies by age class.
Then creates plots of survival curves, estimated rates +/- 95% CIs
Then runs chi squared test to see if causes of mortality vary by age class.
02_HomeRanges_NewCollars.R
This file uses cougar GPS collar data and creates home ranges.
After inital processing and data cleaning, uses AMT to estiamte 95% autocorrelated kernel density estimates.
And 100% minimum convex polygons for three collared cougars: SF9, SF1, and SM6.
Currently only writes shapefiles for MCPs since that is what is used in cougar RSF.
Note: We cannot provide the raw data to run this code, but provide the code as an illustration of what we did.
03_RSF_multiple-animals.R
This file estimates an resource selection function of cougar kill sites.
It then uses this RSF to create a layers of predicted probability of cougar kill occurrence.
Uses cluster investigation data, cougar MCPs, and GIS data.
Outputs include the parameter estimates for kill site RSF and the cougar risk layers.
Note: We cannot provide the raw data to run this code, but provide the code as an illustration of what we did.
We provide the produced cougar kill occurrene layer which is used in the pig step selection function.
04_SSF_PigCollars.R
This file estimates integrated step selection functions for wild pigs.
It first processes and cleans pig GPS collar data and adds sunrise/sunset times to identify day and night locations.
Then reads in GIS data, including the cougar risk layer.
Finally it estimates an integrated step selection function for pigs both with and without cougar risk included.
The final section compares models with and without cougar risk to understand if it improves model performance.
There is also code for creating plots of pig 50% core area kernel density estimate utilization distributions plotted over cougar risk.
Methods
These data include capture records and telemtry tracking data for wild pigs that were monitored with GPS collars and VHF ear tags. Raw location data are provided for the GPS collars. VHF ear tag data provides the date and status of observed pigs throughout the monitoring period. Location data for cougars are not provided due to the sensitive nature of these data.