The big picture: appraising the risk of cumulative impacts on seagrass meadows in the Mediterranean Sea

Human well-being and societal resilience are deeply tied to marine coastal ecosystems, yet they are threatened by multiple endogenic and exogenic pressures. This study implements a MAchine leaRnIng-based CumulaTive Impact AssessMEnt (MARITIME) model to understand better the risk of cumulative impacts on seagrass meadows in the Mediterranean Sea and their cascading effects on the capacity to provide ecosystem services for this vital blue carbon ecosystem under a reference state (2017) and future climate change scenarios, which encompass various levels of global warming, from moderate to more extreme. The model employs a Random Forest algorithm to support robust, data-driven analysis of complex environmental interactions. The results indicate a shrinkage of seagrass meadows, particularly at lower depths, and a corresponding reduction in ecosystem services capacity by the years 2050 and 2100. This trend is notably severe across the entire Mediterranean basin in the long-term scenario, with an estimated reduction of approximately 20 % in seagrass distribution under the most severe scenario considered in this study (RCP8.5 projected to the end of the century). Such shrinkage may trigger further impacts, starting with a decline in ecosystem service provision: model projections indicate a potential decrease of approximately 3.5 %–6 % in carbon sequestration and up to 20 % in denitrification potential by the century's end, contingent upon the scenario considered. Furthermore, the loss of seagrass could also lead to the release of buried carbon, potentially transforming Mediterranean coastal areas from carbon sinks into sources of carbon emissions - a phenomenon likened to a ‘carbon bomb.’