This repository is part of a Master's Thesis to test the influence of variation in walking speed and environmental context on desert ant foraging.

This repository provides an AgentPy model for foraging movement of desert ants. The main purpose of the model is to study the influence of variation in walking speed and environmental context on path characterstics of desert ant foraging paths. Additionally a set of tools to analyse the model as well as observed and simulated movement data are offered within this repository.

Python Packages

• python
• notebook
• matplotlib
• agentpy
• salem
• shapely
• rasterio
• nbformat
• geopandas
• pandas
• seaborn

R Packages

• R
• ggpubr
• ggplot2
• trajr
• sf
• tools
• ggstatsplot
• rstudio
• stats
• graphics
• grDevices
• utils
• datasets
• methods

A quick introduction of the minimal setup you need to get a hello world up & running.

For installation of Python on your OS see here. Python v3.9.12 was used for the thesis.

pip install notebook==6.4.8
pip install matplotlib==3.5.1
pip install agentpy==0.1.5
pip install salem==0.3.9
pip install shapely==2.0.1
pip install rasterio==1.3.5.post1
pip install nbformat==5.3.0
pip install geopandas==0.12.2
pip install seaborn==0.11.2

For installation of R on your OS see here. From within R, the packages can be installed using

packages.install(c("ggpubr", "ggplot2", "trajr", "sf", "tools", "ggstatsplot", "rstudio", "stats", "graphics", "grDevices", "utils", "datasets", "methods"))

All necessary code to run data analysis and to simulate the model is provided within multiple jupyter notebooks. Documentation and comments can be found within the notebooks. An overview of what the notebooks are meant for is given here:

• Model Submission notebook: This notebook includes the model code. (It does not run the model, just defines it)
• Model Trajectory Visualization notebook: This notebook includes runs the calibrated and uncalibrated model versions and plots the reuslting trajectories in the notebook. Plots are also saved to files into the model_outputs folder.
• Model Validation notebook: This notebook runs the model validation. It splits the observation data into two sets and uses one to estimate parameters. Then it runs the calibrated and uncalibrated model versions, computes errors compared to the other observation set. Plots are printed in the notebook and saved int the validation_data folder.
• Sobol Sensitivity Analysis notebook: This notebook runs the Sobol' sensitivity analysis. Parameter ranges and sample size are defined and the model is run (may take several hours for large sample sizes). From the model results Sobol' indices are computed and visualized. Plots are printed in the notebook and saved int the sensitivity_results folder.
• Trajectory Analysis notebook: This notebook is basically just a wrapper for running the trajectory processing and analysis in R. It executes the R scripts and visualizes the figures in the notebook (does not work on Safari).

This presented model code in Model_Submission.ipynb is an adaption of the code by Thierry Hoinville, 2018, as part of the following work:

Hoinville T, Wehner R., Optimal multiguidance integration in insect navigation Proceedings of the National Academy of Sciences, 115 (11), 2824-2829 (2018). DOI: 10.1073/pnas.1721668115

The tracking data used here is part of the Ant Ontogeny Dataset from the following work:

Lars Haalck et al., CATER: Combined Animal Tracking & Environment Reconstruction. Science Advances, 9 (16), eadg2094 (2023). DOI:10.1126/sciadv.adg2094