Biomonitoring disturbed urbanised estuaries with fine-resolution species distribution models driven by seabed photo data

Species distribution modelling has been utilised to great success in terrestrial systems, often to interrogate the influence of climatic or environmental disturbance on a species range. In marine systems, ecologists are... [ view full abstract ]

Species distribution modelling has been utilised to great success in terrestrial systems, often to interrogate the influence of climatic or environmental disturbance on a species range. In marine systems, ecologists are confined by a paucity of biological data, and a highly dynamic, hetereogeneous system. Habitat maps have often failed to represent this inherent variability due to small sets of low-detail point-data abstracted across broad spatial areas. Species distribution models predict distribution across space based on point data, but respond to a suite of environmental factors. This study used data from images collected with the Catlin Seaview Survey and classified with SCRIPPS CoralNET machine learning software to build fine-scale species distribution models for key marine habitats in Sydney Harbour, Australia. Model predictions were informed by a range of environmental variables including human use and seabed backscatter collected with a Reson 7125 multi-beam echo-sounder. Although Sydney Harbour is relatively iconic, and central to Australia's largest city, this study will provide a uniquely detailed assessment of the distribution and condition of rocky reef habitat. The outcomes from this project are an example of improvements in our ability to assess and monitor the distribution and condition of marine habitats at a fine scale, given recent technological developments.