Codes in support of coral community change

Note, some parts of this codebase are designed to run on a HPC. In particular, certain sections of R/fit_models.R and R/process_models.R make use of either furrr::map or parallel::mclapply.

The repo is structured as:

\
|- R
|  |- main.R
|  |- process_data.R
|  |- fit_models.R
|  |- process_models.R
|  |- summary_plots.R
|- .gitignore
|- LICENCE
|- Makefile
|- README.md
|- analysis.slurm

To be able to actually run the code, the structure needs to be:

\
|- data
|  |- primary
|  |  |- comp2021_Oct2025.RData
|  |  |- video_codes.csv
|  |  |- group_cover_transect.RData
|  |  |- comp21_broad_groups_lookup.csv
|  |- processed
|  |- modelled
|- R
|  |- main.R
|  |- process_data.R
|  |- fit_models.R
|  |- process_models.R
|  |- summary_plots.R
|- outputs
|- .gitignore
|- LICENCE
|- Makefile
|- README.md
|- analysis.slurm
|- ltmp_dashboard1.sif

To run the full codebase on a HPC:

make slurm_R

The main output from this codebase is a heatmap that summarises the contrasts of each coral group between early 1990s and early 2020s conditional on sector and shelf of the Great Barrier Reef. This heatmap is created in the /output/ directory.

Interactively running codes

After cloning the repo locally, you can step through the code. The intended order of R scripts is:

• main.R
  • this is actually the main entry point and can be used to run the entire codebase
  • important it loads the required R packages and defines/creates the paths
• process_data.R
  • this script reads in the primary data sources and performs some data wrangling tasks
  • processed data are stored in /data/processed
• fit_models.R
  • this script prepares the data for INLA modelling
  • this script fits the INLA models in parellel using furrr
  • Note, it assumes there are at least 20 CPU cores available, if running outside a HPC environment, you are advised to reduced this
  • individual models are stored in /data/modelled
• process_models.R
  • this script extracts posteriors from the INLA models
  • collated data are stored in /data/modelled/
• summary_plots.R
  • this script generates plots of estimated cover conditional on COMP2021, half-decade, sector and shelf
  • this script performs various contrasts on the posteriors
  • this script collates the contrasts and generates a summary heat map
  • figures are output to /output/

The scripts assume that the working directory is /R.