Assessing the importance of acoustic backscatter and bottom trawling impact as predictor variables for mapping deep sea coral reefs

Seamounts host some of the most biodiverse and vulnerable deep-sea ecosystems, as epitomised by the elevated diversity and biomass of fauna associated with extensive reefs of the frame-work forming coral Solenosmilia variabilis. In places, these vulnerable marine ecosystems (VMEs) have been substantially impacted by commercial bottom trawl fishing and face increasing threats from climate variability. Understanding both the current and historical distribution of coral reefs enables naturalness and recovery to be considered in conservation planning. Our study on the Tasmanian seamounts, an area characterised by extensive deep-sea coral reefs and historical trawling impacts, employed fine-scale Species Distribution Modelling (SDM) using acoustically derived bathymetry and backscatter data resolved to 15 m resolution to capture fine-scale habitat heterogeneity. We used a two-model methodology to predict VME habitat distributions before and after trawling impacts across a broad region (>480 km²) containing multiple seamounts with peak depths ranging from 720–2,073 m and varied substrates, as indicated by a spectrum backscatter intensity values. Bathymetry, relative slope position, backscatter and trawling history were most influential among a suite of predictor variables. Comparison of before and after models showed a reduction in both total area and patch sizes of coral habitat consistent with areas classified as historically impacted by trawling. The reduction in extent of coral VME area was dependent on the probability threshold of habitat suitability used in the model, decreasing by 20.44%–26.07% following trawling across a range of thresholds from 0.5–0.85. Our study highlights the benefits to future spatially based biodiversity management initiatives that will stem from using high-resolution bathymetry and backscatter data, and information on anthropogenic impacts, in SDM predictions.