Monitoring an Ecosystem in Crisis: Measuring Seagrass Meadow Loss Using Deep Learning in Mosquito Lagoon, Florida

The ecosystem of Mosquito Lagoon, Florida, has been rapidly deteriorating since the 2010s, with a notable decline in keystone seagrass species. Seagrass is vital for many species in the lagoon, but nutrient overloading, algal blooms, boating, manatee grazing, and other factors have led to its loss. To understand this decline, a deep neural network analyzed Landsat imagery from 2000 to 2020. Results showed significant seagrass loss post-2013, coinciding with the 2011–2013 super algal bloom. Seagrass abundance varied annually, with the model performing best in years with higher seagrass coverage. While the deep learning method successfully identified seagrass, it also revealed that recent seagrass coverage is almost non-existent. This monitoring approach could aid in ecosystem recovery if coupled with appropriate policies for Mosquito Lagoon's restoration.