The Cyanobacterial Mats of the Exmouth Gulf, Western Australia: Mapping Report

High intertidal cyanobacterial mats and salt flats are features of intertidal zones of the arid regions of the world. Their ecology and functions have not been fully explored but likely they are habitat to unique microbial communities, they are used by fauna when flooded and contribute to outwelling of materials to the coastal zone that supports production. Globally, studies on the spatial and temporal dynamics of cyanobacterial mats in coastal salt flats is limited. However, these environments are increasingly being recognised as potential sources for mineral commodities and industry development, as well as for conservation. This study, to our knowledge, is the first comprehensive assessment of cyanobacterial mats and the high intertidal salt flats in the Exmouth Gulf. Here, we develop a technique to construct a representative habitat model of the cyanobacterial mat area in the high intertidal zones of the Exmouth Gulf across 5 timepoints (single-time-points) using Landsat 8 and Random Forest remote sensing modelling techniques. In doing so, we found that cyanobacterial mats comprised 11% of the high intertidal zone within the Exmouth Gulf (22,327 ha), and that mangroves comprised 7% of study extent. We observed that the cyanobacterial mats are dynamic in space and time, with their area ranging from 9-20% (20,279 ha - 42, 653 ha) of the study extent over the individual time-point habitat layers predicted from the 5 time-points analysed. Within the National Directory of Important Wetlands extent the combined Cyanobacterial Mat and Low Density Cyanobacterial Mat classes comprised 17% of this area.

Our results provide insight into the temporal dynamics of the spatial extent of high intertidal habitats which is important for their management. The extent of different ecosystems may fluctuate even more than we observed because our observations were limited by the availability of high resolution imagery, and the absence of ground validation data. Through model uncertainty assessment we identified habitat classes and regions where the model performed well, such as the Cyanobacterial Mat class which had a class accuracy of 89%. The north region around Urala Creek was identified as an area where highest model error occurred, and this is likely due to the highly dynamic nature of this section of coast, as well the dense co-occurrence of habitat classes. The co-location or location of different habitats to create a spatially heterogeneous environment is known to be important for marine fauna. The Low Density Cyanobacterial Mat had the greatest model confusion, with the model often predicting this habitat as Bright Salt or within the High Intertidal Salt Flat classes. All habitat classes are likely integral to the ecological functions and processes of this system that supports marine, intertidal, and terrestrial fauna. However this is a reflection of the transitional and dynamic nature of this habitat, and provides further evidence for the need for more detailed temporal studies and ground validation of interannual changes in the extent of cyanobacterial mats.