Modelling the potential distribution of African Wormwood (Artemisia afra) using machine learning algorithm-based approach (MaxEnt) in Sekhukhune District, South Africa

Published Jun 23, 2025 on Dryad. https://doi.org/10.5061/dryad.dz08kps71

Data files

Jun 23, 2025 version files 10.15 KB

Occurrence_data.zip

1.19 KB
README.md

8.96 KB

Abstract

Artemisia afra Jacq. Ex Willd, commonly known as African wormwood, is a native medicinal plant that has been unsustainable harvested primarily for its leaves due to its medicinal properties. The unsustainable harvesting of this plant underscores the urgent need for conservation and management practices. This study, therefore, used the MaxEnt model of the potential distribution of A. afra. Location: Sekhukhune District Municipality, South Africa. We used 105 sampled records and 27 environmental variables to model the potential spatial distribution of A. afra using the MaxEnt modelling approach. The predictions were performed using current climatic and topographic conditions. A significant portion of the area, 54.46%, is highly suitable for the distribution of A. afra, with various suitability degrees. Precipitation contributed 33.6% to the suitability predictions, followed by NDVI, soil, and distance from rivers with 27.1%, 8.1%, and 5.7%, respectively. Artemisia afra is predicted to be persistent in mountainous areas and along riverbanks.

README

Title: Modelling the Potential Distribution of African Wormwood (Artemisia afra) Using a Machine Learning Algorithm-Based Approach (MaxEnt) in Sekhukhune District, South Africa

DOI: https://doi.org/10.5061/dryad.dz08kps71

Description of the Data and File Structure

This dataset contains georeferenced field occurrence records and raster-based environmental predictors used to model the potential distribution of Artemisia afra using the MaxEnt algorithm. Field data were collected in the Greater Sekhukhune District Municipality, Limpopo, South Africa. Environmental variables were sourced from publicly available datasets and processed to meet the requirements of MaxEnt.

File: `Occurrence_Data.zip`

Contains presence-only occurrence records for Artemisia afra used in model development and validation.

Artemisia afra train.csv
- 70% of field-verified occurrence points (training set)
- Columns:
  - Latitude: Decimal degrees
  - Longitude: Decimal degrees
  - Species: Always listed as Artemisia afra
Artemisia afra test.csv
- 30% of field-verified occurrence points (testing set)
- Same column structure as the training set

File: `ENVIRONMENTAL_LAYERS.zip`

Contains raster files of environmental predictors used in the MaxEnt modelling.

Format: .asc (ESRI ASCII Raster)
Projection: WGS 1984 UTM Zone 35S
Resolution: 30 arc seconds (climatic layers), 30 meters (topographic/land layers)

Variable Name	Units	Resolution	Source
Bio1: Annual Mean Temperature	°C	30 arc seconds	WorldClim (https://www.worldclim.org/)
Bio2: Mean Diurnal Range	°C	30 arc seconds	WorldClim
Bio3: Isothermality (BIO2/BIO7 × 100)	%	30 arc seconds	WorldClim
Bio4: Temperature Seasonality	%	30 arc seconds	WorldClim
Bio5: Max Temperature of Warmest Month	°C	30 arc seconds	WorldClim
Bio6: Min Temperature of Coldest Month	°C	30 arc seconds	WorldClim
Bio7: Temperature Annual Range	°C	30 arc seconds	WorldClim
Bio8: Mean Temperature of Wettest Qtr	°C	30 arc seconds	WorldClim
Bio9: Mean Temperature of Driest Qtr	°C	30 arc seconds	WorldClim
Bio10: Mean Temp of Warmest Qtr	°C	30 arc seconds	WorldClim
Bio11: Mean Temp of Coldest Qtr	°C	30 arc seconds	WorldClim
Bio12: Annual Precipitation	mm	30 arc seconds	WorldClim
Bio13: Precipitation of Wettest Period	mm	30 arc seconds	WorldClim
Bio14: Precipitation of Driest Period	mm	30 arc seconds	WorldClim
Bio15: Precipitation Seasonality (CV)	%	30 arc seconds	WorldClim
Bio16: Precipitation of Wettest Qtr	mm	30 arc seconds	WorldClim
Bio17: Precipitation of Driest Qtr	mm	30 arc seconds	WorldClim
Bio18: Precipitation of Warmest Qtr	mm	30 arc seconds	WorldClim
Bio19: Precipitation of Coldest Qtr	mm	30 arc seconds	WorldClim
Elevation	meters	30 m	SRTM DEM (https://code.earthengine.google.com/)
Slope	degrees	30 m	Derived from DEM
Aspect	degrees	30 m	Derived from DEM
Topographic Wetness Index	index	30 m	Derived from DEM
Soil	-	1:1 million	SOTER Database (https://www.isric.org/explore/soter)
Land Cover	-	30 m	https://egis.environment.gov.za/gis_data_downloads
NDVI (Normalized Difference Vegetation Index)	-	30 m	Sentinel-2 MSI (https://code.earthengine.google.com/)
Distance from Rivers	meters	1:50,000	https://www.dws.gov.za/iwqs/gis_data/river/

Additional Files and Subfolders:

.prj and .xml files: Associated with selected raster layers for projection and metadata
Selected variables: Contains the final 13 predictor layers used in model construction
maxent.cache: Model-generated internal cache files created by MaxEnt during runs

Code/Software

1. MaxEnt (Maximum Entropy Modelling Software)

Version: 3.4.1
URL: https://biodiversityinformatics.amnh.org/open_source/maxent/
Usage: Input: training/testing CSVs + environmental layers in .asc format. Output: suitability maps, AUC, response curves, jackknife plots.

2. ArcGIS 10.8 (Esri)

Used for raster pre-processing (projection, resampling, masking, format conversion to .asc)

3. Microsoft Excel / LibreOffice Calc

Used for CSV inspection and editing

Access Information

Other publicly accessible locations of the data:

The dataset is exclusively hosted on Dryad: https://doi.org/10.5061/dryad.dz08kps71

Data was derived from the following sources:

Source Name	Description	License
WorldClim v2.1	Bioclimatic variables	Open for research/non-commercial use
SRTM (CGIAR-CSI)	Elevation, slope, aspect	Public domain
MODIS / Sentinel-2	NDVI	NASA/Public use
SANLC (DFFE)	Land cover	Open access South Africa
SoilGrids / SOTER	Soil classification	ODbL license
HydroSHEDS / DWS	River data and proximity	Free for scientific use

This README ensures complete transparency and usability for future users and reviewers.