Widespread retreat of coastal habitat is likely at warming levels above 1.5°C

The study addresses how rapidly rising relative sea level (RSLR) affects the persistence, vertical adjustment, and spatial extent of key coastal ecosystems—mangroves, tidal marshes, and coral reef island systems. These ecosystems provide critical services (coastal protection, fisheries habitat, blue carbon), and exhibit biogenic feedbacks that can elevate surfaces through sediment trapping and organic production. However, their resilience under higher, projected rates of RSLR is uncertain. Prior contemporary observations have suggested resilience at present-day RSLR in some settings, whereas palaeo records indicate high vulnerability of mangroves and tidal marshes when RSLR exceeds about 6–8 mm yr−1. The authors seek to quantify thresholds at which vertical adjustment fails (leading to elevation deficits, increased inundation, and eventual retreat) and to project global exposure of existing ecosystems to those thresholds under IPCC AR6 sea-level projections across warming scenarios. The work aims to define a safe operating space for coastal ecosystems and highlight implications for conservation, restoration, and climate policy.

The paper synthesizes evidence that coastal wetlands can accrete vertically through mineral deposition and organic matter production, enabling persistence under modest RSLR, but that resilience diminishes as RSLR increases. Palaeo-stratigraphic records from the Holocene show repeated drowning of mangroves and tidal marshes during periods of rapid sea-level rise (for example, meltwater pulse 1A and early Holocene pulses >7–10 mm yr−1), with widespread expansion occurring primarily when RSLR slowed below ~7 mm yr−1. Reef islands largely formed and stabilized during mid- to late-Holocene sea-level stillstands or falls. Contemporary studies using accretion markers and elevation benchmarks (SET-MH) indicate non-linear shallow subsidence that can offset accretion, leading to elevation deficits under higher RSLR. Prior regional assessments have reported marsh retreat at RSLR of ~4–6 mm yr−1 and links between RSLR and vegetation stress (e.g., NDVI decline). For reef islands, recent decadal analyses show general stability at recent GMSL rates, but increasing contraction likelihood above contemporary rates, with case studies (e.g., Solomon Islands) documenting substantial island loss under ~7–10 mm yr−1 RSLR.

The authors integrate three independent lines of evidence:

1. Palaeo-stratigraphic synthesis and GIA-based RSL reconstructions: They review the timing of mangrove, tidal marsh, and reef island advance/retreat since the Last Glacial Maximum, focusing on the past 10 ka. They use glacial isostatic adjustment (GIA) models to estimate historical RSLR rates by region and relate them to stratigraphic indicators of habitat persistence or drowning.
2. Contemporary vertical adjustment analysis using SET-MH networks: They analyze a global dataset of surface elevation table–marker horizon (SET-MH) installations to compare surface accretion, shallow subsidence, and elevation change against local RSLR. For tidal marshes, they build on a network of 477 stations; for mangroves, they perform a new Bayesian analysis on 190 installations. They estimate the cumulative probability that elevation change meets or exceeds local RSLR across RSLR bins, identifying thresholds at which elevation deficits (Δ elevation − RSLR < 0) become likely or very likely.
3. Habitat change and retreat assessment under current RSLR: Using high-resolution global surface water change datasets and tidal wetland maps in the vicinity of tidal marsh SET-MH sites, they quantify trends in conversion to open water and relate them to RSLR, elevation capital (relative position in tidal frame), and measured elevation deficits. For mangroves, canopy cover limits direct surface water detection. For reef islands, they compile morphometric changes (n = 872 islands) from the Indian and Pacific Oceans and evaluate contraction probability versus RSLR.
4. Projections of exposure under future RSLR: They overlay the mapped distributions of global mangroves, tidal marshes, and coral reefs/reef islands with regional RSLR projections for 2080–2100 from IPCC AR6 across warming scenarios (1.5, 2, 3, 4, 5 °C above 1850–1900). They calculate the proportions of existing habitats exposed to RSLR ≥4 mm yr−1 (likely deficit) and ≥7 mm yr−1 (very likely deficit for marshes/mangroves; likely contraction for reef islands). They also analyze potential landward migration capacity under constraints such as population density thresholds (e.g., >20 people km−2) versus a no-barriers scenario.

• Concordance of palaeo and contemporary benchmarks: The probability of vertical adjustment inferred from Holocene palaeo-stratigraphy aligns with SET-MH observations. Mangroves and tidal marshes commonly fail to keep pace as RSLR approaches 7–8 mm yr−1.
• Thresholds for wetlands: Elevation deficits (vertical adjustment < RSLR) are likely at 4 mm yr−1 and very likely at 7 mm yr−1 for both tidal marshes and mangroves, based on Bayesian analyses of 477 marsh and 190 mangrove SET-MH sites.
• Early signals of change: At marsh SET-MH sites, increased surface water presence becomes likely once RSLR exceeds ~2.3 mm yr−1. Marshes are as likely as not (P ≈ 0.5) to be retreating once RSLR exceeds ~5.4 mm yr−1. Relationships between surface water change and both RSLR (r² = 0.16, P < 0.001) and elevation deficit (r² = 0.14, P < 0.001) are significant, with stronger effects in lower elevation marshes (r² = 0.20 vs 0.03 in higher elevation sites), evidencing the buffering role of elevation capital.
• Reef islands: A higher probability of island contraction is observed above contemporary GMSL rates; contraction is likely at RSLR above ~6.2 mm yr−1. Case evidence from the Solomon Islands (RSLR ~7–10 mm yr−1 since 1994) includes complete erosion of 5 of 20 vegetated reef islands and >20% contraction of 6 others (1947–2014).
• Projected exposure (2080–2100): • 1.5 °C: Likely GMSL rise 2.4–6.4 mm yr−1; elevation deficits likely in many mangroves (4–7 mm yr−1), with low global probability (<11%) of ≥7 mm yr−1 except in subsiding regions (e.g., US Gulf Coast, SE Asian deltas). • 2.0 °C: Approximately one-third of global mangroves face ≥7 mm yr−1 and nearly all ≥4 mm yr−1. Tidal marsh and reef exposure to ≥7 mm yr−1 shows comparatively little change relative to 1.5 °C in median projections. • 3.0 °C: Nearly all tropical/subtropical coasts (hosting most mangroves and coral reefs) face ≥7 mm yr−1; probability of elevation deficits in mangroves moves from likely to very likely.
• Quantitative proportions (median; likely ranges in Table 1): • Mangroves exposed to ≥4 mm yr−1 (likely loss): 0.81 at 1.5 °C; 0.99 at 2.0 °C; 1.00 at 3.0–5.0 °C. Exposed to ≥7 mm yr−1 (very likely loss): 0.03 at 1.5 °C; 0.32 at 2.0 °C; 0.98 at 3.0 °C; 1.00 at 4.0–5.0 °C. • Tidal marshes exposed to ≥4 mm yr−1: 0.34 at 1.5 °C; 0.65 at 2.0 °C; ~0.67–0.70 at 3.0–4.0 °C; 0.97 at 5.0 °C. Exposed to ≥7 mm yr−1: 0.03 at 1.5 °C; 0.39 at 2.0 °C; 0.66 at 3.0 °C; 0.97 at 4.0–5.0 °C. • Reef islands (numbers of reefs) likely vulnerable at ≥7 mm yr−1: 0.01 at 1.5 °C; 0.04 at 2.0 °C; 0.99 at 3.0 °C; 1.00 at 4.0–5.0 °C.
• Policy-relevant impact: Increasing warming from 1.5 °C to 2.0 °C would double the area of mapped tidal marsh exposed to ≥4 mm yr−1 RSLR by 2080–2100. At 3 °C warming, nearly all mangroves and coral reef islands and ~40% of mapped tidal marshes are exposed to ≥7 mm yr−1 RSLR.
• Spatial heterogeneity: High-latitude coastlines may experience lower or negative RSLR due to gravitational/elastic effects, reducing proportional losses and enabling potential marsh expansion (e.g., northern Siberia) where landward migration is feasible.

The integrated palaeo, contemporary, and modelling evidence converges on clear RSLR thresholds beyond which biogenic vertical adjustment in mangroves and tidal marshes is unlikely to keep pace, leading to elevation deficits, increased inundation, ecological stress, and eventual retreat. Reef islands also show increasing instability once RSLR exceeds ~6–7 mm yr−1, consistent with both process understanding and observed changes in high-RSLR regions. These thresholds provide mechanistic support for the concept of a narrowing safe operating space for coastal ecosystems under ongoing warming. The projections indicate that moving from 1.5 °C to 2.0 °C disproportionately increases exposure of tidal marshes, and that at 3 °C, widespread and persistent elevation deficits affecting mangroves and reef islands become very likely for most of their geographic range. High-latitude regions may become relative refugia and sites of expansion for tidal marshes due to moderated RSLR and fewer barriers, potentially shifting the global distribution of blue carbon sequestration. The findings underscore the importance of global emissions mitigation to maintain RSLR within bounds that ecosystems can accommodate, and of local management actions (reducing watershed sediment trapping, mitigating pollution, accommodating landward migration) to bolster resilience where possible.

This study establishes robust, multi-line evidence that coastal habitats reliant on biogenic vertical adjustment face likely elevation deficits at RSLR ≈4 mm yr−1 and very likely deficits at ≈7 mm yr−1, with reef islands becoming unstable above ~7 mm yr−1. Under IPCC AR6 projections for 2080–2100, warming pathways above 1.5–2 °C expose rapidly increasing fractions of existing mangroves, tidal marshes, and reef islands to these critical thresholds, with near-ubiquitous exposure at 3–4 °C for mangroves and reefs and substantial exposure for tidal marshes. Achieving the Paris Agreement goals (ideally 1.5 °C) would minimize disruption. Management should prioritize preservation and restoration of wetlands, reduction of local stressors (e.g., watershed damming, coastal development, pollution), and strategic planning to enable landward migration. Future research should refine threshold estimates for reef island stability at higher RSLR, improve monitoring of elevation change and habitat conversion in mangroves (despite canopy constraints), and integrate evolving sediment supply, CO2 fertilization, thermal stress, and ocean acidification effects into predictive frameworks.

• Affiliation of exposure projections to local tide-gauge-informed baselines may incorporate local anomalies (e.g., subsidence from fluid extraction), adding spatial uncertainty.
• Contemporary observations at very high RSLR (>7–10 mm yr−1) are limited for reef islands, constraining precise identification of a “very likely” contraction threshold and necessitating conservative use of 7 mm yr−1 as a likely threshold.
• Mangrove surface water change is difficult to quantify via optical mapping due to canopy cover; interior marsh breakup patches may be under-detected, likely making retreat estimates conservative.
• Extrapolation from palaeo analogues assumes comparable process regimes; however, modern drivers (ocean warming, acidification, altered sediment supply due to dams/land use, CO2 fertilization) can modify vertical accretion and resilience.
• Landward migration potential is constrained by human barriers (e.g., population density, development) and topography; scenarios simplify or generalize these constraints.
• Reef accretion potential under future thermal stress and acidification remains uncertain, with many reefs already exhibiting low accretion relative to current sea-level rise.