This repository contains examples of the pycoalescence package described here in support of Thompson et al (2019). Primarily, the binder jupyter notebook can be run through a browser to demonstrate the simulation process used in the paper for both spatially explicit neutral models and calculations of the effective connectivity parameter. Examples of map files can be found here.
-
results: folder contains the outputs of simulations used for all analyses.
-
code_examples: folder contains scripts and jupyter notebooks used for generating the landscapes found here, performing the dispersal simulations on HPC systems and performing coalescence simulations on HPC systems.
-
figures: folder containing all figures present in the main text.
-
example_simulation.ipynb: run simulations through MyBinder as an example of how models were performed.
-
figure_generation.R: generate the figures shown in the main text using simulation results stored in csv files in results folder.
-
plot_colours.R: controls the colours and labels for the plots.
-
preston.R: contains functions for generating the analytical solutions from the Preston function.
Maps are split into three types:
- Clustered maps (also called "contrived maps") involve near-circular islands of habitat see ContrivedMapGeneration.ipynb for examples.
- Random maps were generated using random noise (see "fragment_generation.py")
- Real maps were extracted from data (see "fragment_generation.py")
Coalescence simulations were performed using the pycoalescence package (detailed here) and run on HPC systems at Imperial College London. Examples of code can be found in example_simulation.ipynb and in code_examples/coalescence_simulations.
MyBinder is used to generate an interactive jupyter notebook containing the code for running the example simulations and analyses.
The docker image used to generate the binder notebook is available
here or installable using
docker pull thompsonsed/pycoalescence-circleci-0.0.1