Tropical coastal ecosystems are often characterized by vegetated marine habitats, including mangroves and seagrass beds. Over past decades, these habitats have been impacted by coastal development resulting in vegetation losses with consequences for habitat-dependent species. In Bimini, The Bahamas, development has occurred in proximity to important lemon shark nursery habitats. The NASA/U.S. Geological Survey (USGS) Landsat satellite suite have captured imagery since 1984, allowing for long-term mapping to quantify habitat change. However, mapping is often limited to years with in situ habitat data for training of map classifications. Therefore, we employed the automatic adaptive signature generalisation (AASG) algorithm to map aquatic and terrestrial habitats across a 21-year time-series, requiring only 1 year of in situ habitat data. This resulted in the successful creation of 16 maps from 1999 to 2020 with high overall accuracy for identifying seagrass habitat (82%). In years of terrestrial deforestation, seagrass extent decreased, especially when deforestation was coupled with sediment deposition. However, seagrass extent rebounded within 2–5 years depending on location. Our study highlights the effectiveness of the AASG algorithm in mapping across time as well as the importance of mitigation efforts in reducing impacts on seagrass beds. This is especially important for identified consistent seagrass areas, which are in proximity to proposed future developments.
Cormier et al. (Wed,) studied this question.