BeeTool - Bee potential distribution modeling scripts

This repository has all scripts necessary to run create the species distribution models (SDM) for the bee species.

Usage

After cloning this repository on your computer, open the project in RStudio. This will trigger the automatic installation of all dependencies. If the installation of dependencies does not occur automatically, you need to first install the renv package and then execute: renv::restore().

On config.yml file most of the modelling parameters are set. Change them as needed, but default values are the current used.

The preprocessing.R file takes an occurrence file for a given species and performs duplicate record removal. In this context, a duplicate record implies that the occurrence is within a distance equal to or less than the configured resolution parameter. The script also creates an Area of Interest (AOI) for modelling. The AOI is generated by selecting Terrestrial Ecoregions with at least one species presence. After executing the script, the AOI region and de-duplicated occurrence files are created. Open a terminal window to execute the script as follows:

$ Rscript preprocessing.R {occurrence_file.csv}

Tip

Naming the occurrence file with a species code is suggested to ensure clear information in the output files.

The fit_model.R script receives as argument the folder where the AOI and de-duplicated occurrence files were saved. The script will do the modelling based on the parameters in config.yml file, this include which covariates will be considered. As and output of the scripts the following products will be available:

• A Maxent_models.Rds files were the MaxEnt fitted models were saved.
• enmeval_results.csv file withe some model evaluation statistics.
• Raster files with name ENM_prediction_M_raw_*.tif which correspond to the models with minimal AIC.
• Binary raster files under binary_output folder for different cutting levels based on minimal suitability, or 5% or 10% percentiles as threshold suitability.

All the generated data should be reviewed to verify that the results are plausible.

Finally, the future_proj.R script, can be called using the same folder where all the fitted results were stored

$ Rscript future_proj.R output/{sp_code}

to generate a future climate conditions model. As output raster files with future projected model will be generated.

Important

All model outputs should be manually inspected for consistency.

Data Sources

The project utilizes the following datasets:

• WWF Terrestrial Ecoregions of the World: Terrestrial Ecoregions of the World is a biogeographic regionalization of the Earth's terrestrial biodiversity.

  • Dataset Link

• WorldClim version 2.1: WorldClim is a database of high spatial resolution global weather and climate data.

  • Dataset Link

• ECOAB - Bee occurrences dataset: Curated bee occurrences dataset.

Credits and Acknowledgments

The author of this scripts and analysis are (in alphabetical order):

• Juan M Barrios [email protected]
• Carlos Cultid [email protected]
• Oscar Martínez [email protected]
• Remy Vandame [email protected]

Some parts of the code was take from
ModelosDarwinUICN written by Juan M. Barrios [email protected] and Angela P. Cuervo-Robayo [email protected].