https://doi.org/10.5061/dryad.wdbrv15vk

This data accompanies the article ‘Harnessing multiscale topographic environmental proxies in regional coral spatial modelling’ by Annie Guillaume, Renata Ferrari, Oliver Selmoni, Véronique Mocellin, Hugo Denis, Melissa Naugle, Emily Howells, Line Bay, and Stéphane Joost. Here, the accuracy of two open-source digital elevation models (DEMs) are assessed, along with the relevance of their derived topographic variables for use in marine ecological modelling. We provide scripts and relevant csv files to illustrate our methods with a case study based on three common Acropora coral species across the Great Barrier Reef, where we produce high performing distribution maps using purely topographic variables derived from three open-access bathymetry models.

The environmental variables used in this study are described in the Environmental_variable_Description.docx. There we have provided the name of the variable, the abbreviation used in the datafiles and scripts, the description of the variable, the reference, and the parameters used to calculate the variable. More information on these variables are provided below.

Description of the data and file structure

The following files are used as inputs for the scripts, outlined at the end of the ReadMe

DEM raster files to produce variables (GBR_bathymetryDEM_multiscale.zip)

This is a very large zip file, containing rasters of the bathymetric DEMs  (as geotiffs - projected tiff files; .TIFF) from one of three DEM sources: Allen Coral Atlas (ACA15), DeepReef at 30m (gbr030), and DeepReef at 100m (gbr100). DEMs were generalised to different spatial resolutions: ACA from 15m to 30m, 60m and 120m. DeepReef 30m from 30m to 60m and 120m, and DeepReef 100 was kept at 100m. From these eight DEMS (3 sources at multiple resolutions), we extracted eight variables (described in Environmental_variable_Description.docx), such that we had 64 variables + 8 DEM values = 72 predictor variables in total.

These GeoTiffs can be imported into R using the terra package, or into any GIS software (e.g., QGIS, SAGAGIS, GeoDa, ArchGIS, etc). Upon opening, the GeoTiff will appear as a raster format, which can be manipulated as required (e.g., change projection, crop to new size, etc). These were cropped using the shapefile (described below).

Note that this folder does not need to be extracted for running the MultiscaleMarineSDM_script.R script, as the values that are used in the analyses have been extracted from the rasters (using the extract function of the Terra package in R) and are available in the csv files described below.

Shapefiles (GBRMPA_extent_max100m_WGS84.shp)

The shapefile ’GBRMPA_extent_max100m_WGS84.shp’ is provided to crop the DEMs to the study  region of the Great Barrier Reef. These files are provided for use in the MultiscaleMarineSDM_producevariables_script.R , but can also be used if the User wishes to download their own variables from other repositories. 

A shapefile is a file format for storing information about a vector alongside information about its projection on a map. Note that shapefiles require accessory files with the filenames ending in: .cpg, . dbf, .prj, .shp, .shx. Please download all the files together to ensure that they can be read by the software. The file ending in .shp is the one that needs to be provided if used in R or other GIS software (e.g., QGIS, SAGAGIS, GeoDa, ArchGIS, etc). Upon opening, the shapefile will appear as a vector format, which can be manipulated as required (e.g., change projection, crop to new size, etc).

CSV data files

In the following files, values are provided for the environmental variables, which have been extracted using the MultiscaleMarineSDMscript2*.R*. These values appear in the following CSV files with an abbreviated column name. The nomenclature follows the rule: “DEM source”“resolution”“variable”: 1) the DEM source is either: Allen Coral Atlas (ACA15), DeepReef at 30m (gbr030), and DeepReef at 100m (gbr100); 2) the resolution is what the variable is generalised to: 015, 030, 060, 100, 120; and, 3) the variables, where the acronym corresponds to the variables listed in Environmental_variableDescription.docx. 

1. File: MultiscaleMarineSDM_presence_2368coral.csv

Coral coordinates: Coordinates collected within the framework of the RRAP (Reef Restoration and Adaptation Program)
Column descriptions:

• ID - Unique identifier for each colony
• species - Species name (n=3; A. hyacinthus, A. spathulata, A. kenti)
• ID.reef - Reef sampled (n=23)
• X;Y - X,Y coordinates in UTM55 indexd on GDA2020 (epsg:7855)
• decimalLatitude; decimalLongitude - Lat, Lon in WGS84 (epsg:4326)
• collect.YEAR; collect.MONTH; collect.DAY; collect.TIME; collect.DATE - Date and time of collection
• depth.raw.m; depth.correct.m.below: Tide_adjust - Depth as raw, LAT corrected, tide adjustment applied
• DEMaccuracy - Boolean values to filter: 1 = used in DEM accuracy measurements ; 0 = not used in DEM accuracy measurements
• SDManalysis - Boolean values to filter: 1 = used in SDM analyses ; 0 = not used in DSDM analyses
• ACA15_015_DEM, gbr030… etc until the end - 72 predictor variable values, extracted from variable rasters using terra::extract(). These columns are named as described above

2. File: MultiscaleMarineSDM_variables_15000randompoints.csv

DEM and variable assessment points
Coordinates of 15’0000 random background points, biased to coral sampled areas, only across 23 reefs
Column description:

• X;Y - X,Y coordinates in UTM55 indexd on GDA2020 (epsg:7855)
• ACA15_015_DEM, gbr030… etc until the end - 72 predictor variable values, extracted from variable rasters using terra::extract(). These columns are named as described above

3. File: MultiscaleMarineSDM_variables_10000maxent.csv 

MaxEnt Background points
Coordinates of 10’0000 random background points, biased to coral sampled areas, only across 23 reefs
Column description:

• X;Y - X,Y coordinates in UTM55 indexd on GDA2020 (epsg:7855)
• ACA15_015_DEM, gbr030… etc until the end - 72 predictor variable values, extracted from variable rasters using terra::extract(). These columns are named as described above

Sharing/Access information

Some of the data are publicly available elsewhere.

Links to other publicly accessible locations of the data, available from the Australian Institute for Marine Sciences (AIMS) server:

• The full RRAP coral colony metadata file: used to obtain colony GPS and depth data; will be available online on the AIMS archive from November 2024.\
  Reference: https://doi.org/10.25845/8DJ4-0V58
• Coordinates (X,Y, depth) records maintained by AIMS\
  Reference: Australian Institute of Marine Science (AIMS) (2024) Metadata associated with ethanol-preserved coral samples collected for the Reef Restoration and Adaptation Program’s Genetic Basis of Coral Traits project. Australian Institute of Marine Science (AIMS).\
  https://researchdata.edu.au/metadata-associated-ethanol-traits-project/2832117?fl
• Multiscale DEMs and topographic variables: Variables produced using the Produce_variables script (below) are available online on the AIMS archive.\
  Reference: Australian Institute of Marine Science (AIMS). (2024) Multiscale bathymetric digital elevation models and derived topographic variables extending across the Great Barrier Reef. https://doi.org/10.25845/ggyv-pq15

Environmental data was derived from the following sources (also available from the zip folder provided: GBR_multiscaleDEM.zip):

• Allen Coral Atlas at allencoralatlas.org\
  Reference: Allen Coral Atlas (2022) Imagery, maps and monitoring of the world’s tropical coral reefs. https://doi.org/10.5281/zenodo.3833242 Beaman, R.J. (2010)

  DeepReef30m at pid.geoscience.gov.au/dataset/ga/115066\
  Reference: Beaman, R.J. (2017) High-resolution depth model for the Great Barrier Reef - 30 m dataset. Geoscience Australia. https://doi.org/10.4225/25/5a207b36022d2

• DeepReef100m at deepreef.org/2010/07/06/gbr-bathy\
  Reference: Project 3D-GBR: A high-resolution depth model for the Great Barrier Reef and Coral Sea. Marine and Tropical Sciences Research Facility (MTSRF) Project 2.5i.1a Final Report. Cairns, Australia. https://doi.org/10.4225/25/5a207b36022d2

Code/Software

We have provided the code necessary to replicate the results in the published article. These scripts require R (v4.1.0, 2021) to be installed, with the required packages are noted at the top of each script. If the user wishes to produce their own variables (Produce_variables), note that this also requires MatLab (MathWorks: MATLAB version R2019a) to be installed. As MatLab is a paid software, an alternative (but untested) R script is provided. Alternatively, the products of the script are provided, so that the Main script can be run.

Produce_variables:\
An R script (MultiscaleMarineSDM_producevariables_script.R) outlining the acquisition, processing and generalisation of open-source bathymetric DEMs (available at the links above). These DEMs are cropped using the ‘GBRMPA_extent_max100m_WGS84.shp’, which is a subset of the GBR marine protected area shapefile, to max depth of 100m. Note that it requires the use of MatLab for generalisation (code: MultiscaleMarineSDM_producevariables_GaussianPyramid_MATLAB.mlx), where alternative code in R is provided, but not tested. This script also outlines how to derive topographic variables used in downstream analyses.

Main script:\
An R script (MultiscaleMarineSDM_script_2.R) to follow analyses of the article, including bathymetry DEM vertical assessment (comparing in-situ collected depth measurements with bathymetry estimated depths), derived variable correlations, and MaxEnt species distribution modelling analyses (parameter optimisation, full model, prediction rasters). Note that the final prediction raster across the GBR (section 7) will only run if the full 72 variables have been produced, as these files are not yet uploaded.

Version Changes

13-apr-2025: Script was updated to include author and publication metadata.