Derived variables and coordinates to assess the ecological relevance of multiscale bathymetry for coral species distribution modelling across the Great Barrier Reef

With constant improvements in the accuracy and availability of open-source digital elevation models (DEMs) comes a need to properly understand the relevance of their derived topographic variables in ecological research, particularly in marine environments. Here, we provide data and scripts used to investigate the ecological relevance of two open-source bathymetric models for deriving topographic variables to perform species distribution modelling of coral across the Great Barrier Reef (GBR), Australia. We illustrate our methods with a case study based on three common Acropora coral species (A. hyacinthus, A. spathulata, A. kenti) across 23 reefs of the GBR, where we produce high performing distribution maps using purely topographic variables derived from three open-access bathymetry models. 

A script is provided to first download and process bathymetry DEMs from the Allen Coral Atlas and the DeepReef projects, before performing a multiscale generalisation to acquire these DEMs at  15m, 30m, 60m, 100m, 120m resolutions.  From these DEMs, we provide code for deriving 14 topographic variables used in analyses. We provide the csv files containing values of the DEM and topographic variables at random assessment points across the 23 reefs,  as well as at the coral colony sample sites. The script contains all analyses, including DEM vertical depth assessment, variable correlations, and MaxEnt species distribution models.

Methods
Georeferenced coral occurrence data and depth measurements of three Acropora species were recorded within the context of coral adaptation research. Data was collected over three years (2021-2023) at 23 reefs spanning the extent of the Great Barrier Reef as part of the Reef Restoration and Adaptation Program (RRAP), targeting Acropora hyacinthus, Acropora spathulata and Acropora tenuis colonies. Here, coordinates, depth and health chart data of 2368 colonies are provided, alongside indications of which colony was used in specific analyses. Georeferenced coordinates were collected using a GPS towed behind divers during sampling campaigns, where timestamps on colony photos were used to find the associated coordinates. Colony depth was recorded using dive computers and corrected to the lowest astronomical tide using a custom R script. Details of the tide station along with the collection date’s tide levels are provided. The coordinates and corrected depths were used to investigate the accuracy of open-source bathymetric digital elevation models (DEMs) and to assess the ecological relevance of multiscale bathymetric topographic variables for regional species distribution modeling (SDM). Columns are provided to note the colonies used for DEM accuracy assessments, as well as those used in SDM analyses.

Multiscale bathymetric digital elevation models (DEMs) and derived topographic variables used to characterise the seascape environment spanning the Great Barrier Reef Marine Park Authority (GBRMPA) region. This data was collected to investigate the accuracy of open-source bathymetric DEMs and to assess the ecological relevance of multiscale bathymetric topographic variables for regional species distribution modeling (SDM) of coral species. The base DEMs were downloaded from two publicly available online sources: the Allen Coral Atlas (ACA; Allen Coral Atlas, 2022; www.allencoralatlas.org) at approximately 10m resolution and the DeepReef Explorer (www.deepreef.org) at approximately 30m (Beaman, 2017) and 100m (Beaman, 2010) resolutions. After processing, the base DEMs were generalised using a Gaussian pyramid algorithm: ACA from 15m to 30m, 60m, 120m, DeepReef from 30m to 60m and 120m, as well as DeepReef at 100m. From each DEM, eight topographic variables were derived: slope (SLO), eastness (EAST), northness (NORTH), vertical curvature (VCU), horizontal curvature (HCU), vector ruggedness measure (VRM), sky view factor (SVF), bathymetric position index (BPI).  Variables are available as SAGA sdat files upon request. Variable values at all spatial resolutions from each bathymetry source are available as csv files: i) 15000 random points used to assess correlation and scatterplots, ii) 10000 random points used in MaxEnt SDM analyses, where points avoided areas that also contained presence data.