Investigating the Effects of Forestry Practices on the Dynamics of the Viaphytic Fungi-Trees Ecosystem

Core course Agent-Based Modelling, MSc Computational Science (UvA/VU, 2024)

This repository contains the code for the course project assignment in Agent-Based Modelling (ABM). The foundation for our framework is a model proposed by D. C. Thomas, R. Vandegrift and B. A. Roy (2020)[1], investigating the dynamics between viaphytic fungi and trees, which the fungi can use as propagation vectors. We expand this model by creating a mutualist feedback cycle in which both agent types deposit resources, which in turn affect the other agent type's behaviour.

A particular focus of our research is the exploration of forestry techniques consisting of harvesting and planting, which can optimise timber yield while preserving the tree and fungi populations. Inspiration for the implementation of forestry in the ABM was taken from research by Zenith Arnejo et al.(2023)[2].

Contents

The experiments with the model are structured in several Jupyter notebooks:

• runmodel.ipynb contains general purpose procedures for running the model with predefined parameters and analysing model results saved as .parquet files;
• SA_paramspace.ipynb contains a global and local sensitivity analysis of several model parameters, visualisations of specific metrics in the sampled parameter space and additional simulations for determining the tree volume distribution;
• planting_experiments.ipynb contains the setup for more specific simulations at a more narrowed-down range of planting and harvesting parameters;
• phase_space.ipynb contains analysis procedures for analysing the temporal dynamics of model outputs at sampled input parameters.

The main Python modules imported in the notebooks partly follow Mesa conventions and include:

• model.py - model-related procedures;
• agent.py - agent-related procedures;
• visualisation.py - procedures for visualising results;
• sensitivity_analysis.py - procedures for global and local sensitivity analysis.

Requirements

To use the model, please install the dependencies by running pip install -r requirements.text.

The packages used in the model include:

• NumPy
• SciPy
• Matplotlib
• Seaborn
• Mesa
• Pandas
• SALib
• pyarrow

References

[1] Thomas, Daniel C., Roo Vandegrift, and Bitty A. Roy. "An agent-based model of the foraging ascomycete hypothesis." Fungal Ecology 47 (2020): 100963.
[2] Arnejo, Zenith, Leonardo Barua, Paul Joseph Ramirez, Cristino Tiburan Jr, and Nathaniel Bantayan. "An Agent-Based Model of a Sustainable Forest Operation in a Theoretical Lowland Dipterocarp Forest Modeled after Mount Makiling Forest Reserve, Philippines." Forests 14, no. 2 (2023): 428.