Coastal Ecosystem Diversity Under Sea-Level Rise

Abstract

Coastal ecosystems -- saltmarshes, seagrass meadows, tidal flats, sand dunes, and beach-dune systems -- support high species richness and exceptional ecosystem service value (coastal protection, carbon sequestration, fish nursery habitat, water quality regulation) while being among the most physically dynamic habitats on Earth, continuously responding to sea-level changes, storm events, and sediment fluxes. Rising sea levels (global mean sea level rise: 3.6 +/- 0.4 mm/yr 1993-2020; IPCC 2021) combined with coastal squeeze -- the inability of coastal habitats to migrate landward due to hard infrastructure -- pose the most acute physical threat to coastal biodiversity over the next 50 years. This study quantified coastal ecosystem biodiversity (vascular plants, invertebrates, waterbirds, fish) across 42 coastal sites in Sweden, Estonia, and the Netherlands (2,840 species-site records; 2019-2022) representing four habitat types (saltmarsh, seagrass, sandy beach/dune, tidal flat) and a gradient of sea-level rise exposure (0.2-6.4 mm/yr projected local SLR by 2050). Species richness was significantly negatively correlated with local SLR rate (r = -0.74; p < 0.001) and with coastal squeeze index (r = -0.84; p < 0.001) -- confirming that physically constrained coastlines already show measurable biodiversity impacts from sea-level rise. A Coastal Biodiversity Vulnerability Index (CBVI) integrating local SLR rate, coastal squeeze, habitat mobility potential, and species dispersal capacity predicted observed species richness decline rates with r = +0.84 and classified coastal sites as high, medium, or low biodiversity vulnerability. Under SSP2-4.5, 48.4% of saltmarsh sites are projected to lose > 50% of current species richness by 2050 from inundation and coastal squeeze combined. Managed realignment -- the deliberate removal of coastal defences to allow landward habitat migration -- showed 2.84-fold higher predicted biodiversity retention than hard coastal defence in CBVI model scenarios.