Multidecadal mapping of arid intertidal ecosystems reveals a dynamic mosaic of habitats north-western Australia

Intertidal areas are dynamic environments and represent an important ecotone, with flora and fauna adapted to variation in tidal inundation. In arid coastlines there is complexity in tidal inundation, due to high temperatures and evaporative conditions that strongly influence the distribution of biota. The Pilbara coastline in north-west Australia is one of seven arid mangrove regions that also comprises extensive saltmarsh, cyanobacterial mats, and salt flats. Spatial extent and temporal mapping of these habitats has not previously been conducted. In this study, we undertake eight time-point models of 10 habitat classes over 400km of the Pilbara coastline. Using a combination of Landsat 5, 7, 8 and 9 satellite scenes and high-resolution aerial images we develop a spatial-extent of predicted habitats observing the dynamic nature of the habitats, particularly for the cyanobacterial mat habitat. We find that cyanobacterial mats comprise the largest spatial extent of the intertidal habitat classes covering on average 34% of the intertidal region, double that of the mangrove area (16%). Across the time-points, pixels transitioning between high intertidal salt flat, cyanobacterial mat, and areas with bright salt were the largest combination of classes, and the second largest class overall (behind coastal vegetated habitats). This research comprehensively models the extent of the habitats that comprise the high intertidal zone of the Pilbara Coast that is a globally significant area of hypersaline habitat providing an important baseline and insights into the impacts of climate on arid intertidal habitats. • Arid intertidal salt flats are dynamic environments with transitional habitats • High Intertidal Salt Flat, Cyanobacterial Mat, Bright Salt largest combination class • Cyanobacterial mats have the largest spatial extent of the intertidal habitats (34%) • Mangroves comprise (16%) of the intertidal area • 8 time-point habitat models were generated with good accuracy for 400 km of coast