This work formed Chapter 4 of my PhD thesis, titled: "Performance of camera trap-based density estimators for unmarked populations".
The aim was to evaluate three commonly used unmarked abundance estimators (REM, REST, CTDS) under a common set of simulations, spanning a wide range of animal population and movement scenarios.
The original papers describing these estimators are as follows:
- Random Encounter Model (REM): https://doi.org/10.1111/j.1365-2664.2008.01473.x
- Random Encounter and Staying Time (REST): https://doi.org/10.1111/1365-2664.13059
- Camera Trap Distance Sampling (CTDS): https://doi.org/10.1111/2041-210X.12790
The directory should contain the following folders: R to contain the R scripts and Data to store simulated data. The latter contains subfolders MovementSims and Detections to store the simulated movement and detection data, and Estimates to store the model estimates.
The simulation workflow comprises three main steps:
-
Simulate movement trajectories for a population of
$N$ number of unique inviduals exhibiting either solitary or group-living movement behaviour on a X by X unit landscape. -
Generate detection data from the simulated movement trajectories with a grid of
$J$ pie-shaped detectors with detection distance$d$ units and radius$r$ units. Detection probability decreases with distance from the detector. - Get model estimates by applying an estimator (REM, REST or CTDS) to appropriately formatted input data.
Each step can be run from the command line as shown in the image below:
