Harnessing multiscale topographic environmental proxies in regional
coral spatial modelling
Abstract
Effective biodiversity conservation requires knowledge of species’
distributions across large areas, yet prevalence data for marine sessile
species is scarce. As marine organism distributions generally depend on
terrain heterogeneity, topographic variables derived from digital
elevation models (DEMs) can be useful proxies in distribution modelling.
However, the suitability of such variables depends on spatial
resolution. Here, we (1) assess three high-resolution bathymetry DEMs
for accuracy, (2) harness DEM-derived topographic variables for regional
coral species distribution models (SDMs), and (3) develop a transferable
framework to produce, select and integrate multi-resolution topographic
variables into marine spatial modelling. We use a case study from three
reef-building Acropora coral species sampled across 23 reefs of
the Great Barrier Reef, Australia. Obtaining three open-source
bathymetric DEMs (15m Allen Coral Atlas (ACA), 30m DeepReef, 100m
DeepReef), we produce eight derived topographic variables generalised to
multiple nested spatial resolutions (15m to 120m) to assess SDM
sensitivity to bathymetry source and resolution. We found that the ACA
and DeepReef DEMs had similar vertical accuracies, each producing
topographic variables relevant to marine SDMs. Slope and vector
ruggedness measure (VRM) explained most of the variance for all three
species. Coral prevalence increased with slope to a moderate steepness
before quickly decreasing with increasing steepness for two species,
while the third species plateaued. The prevalence of all species was
negatively associated with VRM. Topographic variables for coral SDMs
were most relevant at 15–60m resolutions, where the optimal resolution
depended on the variable type and species. Overall, we show that the
finest resolution was unnecessary to achieve high-performing SDMs.
Running SDMs with multi-resolution topographic variables provided
insights into the importance of terrain attributes for distribution
modelling of different species. We provide a transferrable framework to
facilitate the adoption of multiscale SDMs for better-informed
conservation and management planning.