GabsPalomo / Lagomorphs-coyotes-Kansas

Repository for the Bayesian model in the paper Mesopredators have differing influences on prey habitat use and diet activity in a multipredator landscape.

Mesopredators have differing influences on prey habitat use and diel activity in a multipredator landscape.

ABSTRACT

Resource distribution, habitat structure, and predators greatly influence when and where prey species’ use the landscape. The risky places hypothesis establishes that prey will exhibit a proactive response to predators presence based on habitat characteristics whereas the risky times hypothesis predicts that prey will behave reactively by increasing vigilance regardless of environmental cues. We used camera-trap data with a Bayesian species interaction occupancy model with local- and landscape-scale covariates to evaluate black-tail jackrabbit (Lepus californicus) and Eastern cottontail rabbit (Sylvilagus floridanus) habitat use in the presence and absence of coyotes (Canis latrans), American badgers (Taxidea taxus), and swift foxes (Vulpes velox). We analysed the patterns of temporal overlap between predators-prey to establish times of increased activity for all. Jackrabbits decreased their habitat use in areas rich in forage and were more likely to use areas with better visibility when coyotes or swift foxes were present. However, cottontails increased their use of open areas with high visibility when badgers were present but with all other predators absent, suggesting the presence of individual predators do not illicit similar responses. Both lagomorph species were nocturnal but exhibited different peaks of activity compared to predators suggesting fine-scale temporal segregation. Our results provide insights into predator-prey dynamics across different habitat features in a multi-predator landscape.

Description

This README includes a description on the various scripts used for this analysis. However, please note that not all data files are saved in this repository because they may be too large to store here. I have the scripts numbered so it's easier to see the flow of the analysis and where to start.

RDS file with results of the Bayesian model.

The Bayesian model took several weeks to run, but I saved the output in a .rds file. However, because it is too large it cannot be stored here but it can be pulled from here as a zip file and read in the session and name it fit so that all analysis can be done. That way, anyone can corroborate my analysis.

Directories

1. /data This folder includes the following documents:

   • landcover_scales.csv: occupancy covariates for different scales.
   • bob_dh.csv: detection history for 28 days across all sites and years for bobcats.
   • sfox_dh.csv: detection history for 28 days across all sites and years for swift foxes.
   • badger_dh.csv: detection history for 28 days across all sites and years for badgers.
   • coyote_dh.csv: detection history for 28 days across all sites and years for coyotes.
   • BTJR_dh.csv: detection history for 28 days across all sites and years for Black-tailed jackrabbit.
   • ECTR_dh.csv: detection history for 28 days across all sites and years for Eastern cottontail rabbit.
   • days_active.csv:
   • scent.csv
   • doy.csv
   • det_altered.csv
   • fine_scale_habitat.csv

2. /figs This folder contains all figures generated in this project. Stored as .png or .jpeg. NOT available in the repository but the figures are in the manuscript.

3. /functions This folder contains some long functions necessary for some data cleaning and analysis, as well as reading the silhouettes.

4. /tables This folder contains all tables generated in this project. Stored as .png or .jpeg. NOT available in the repository but the tables are in the manuscript.

Files

1. big_pred_model.R

This script contains the multi-species model made by Mason Fidino and Gabriela Palomo.

2. 01_data_org.R

This document organizes the data to have it ready to create the objects that will go in the model.

3. 02_data_indices.R

This is where we format the data (data and indices) to fit into the model.

4. 03_run_model.R

   • Here is the code to run the Bayesian model in JAGS.

   • It also has the marginal occupancy and detection estimates of each species(prey and predator) organized in tables.

   • It sources /functions/functions.R and the 02_data_indices.R file.

5. 04_final_plots.R

This is where I coded the predictions and final graphs. It sources the results of the Bayesian analysis from the document 2022_09_15_fit.rds

All plots coded here are in the manuscript as figures.

6. 05_1_summary_table.Rmd

This script contains the code necessary to create a table with the mean and 95% credible intervals of each of the model parameters. Estimates are in the logit scale and appear in Appendix S2 of the manuscript.