Introduction

Sea-level rise is expected to be among the most costly and far-reaching consequences of climate change1,2,3,4,5. Coastal cities and ecosystems are already being transformed by more frequent flooding and the inland movement of oceanic salts into freshwater systems6,7. Coastal wetlands, including marshes, mangrove forests, and freshwater swamps, are among the most threatened ecosystems8,9,10. These wetlands are increasingly recognized for their ecosystem services and their ability to mitigate the impacts of climate change11,12,13. In addition to improving water quality and supporting critical fish and wildlife habitat, coastal wetlands protect coastal communities from storms, support fisheries, and sequester carbon from the atmosphere. Yet for decades, there has been debate regarding the future of these valuable ecosystems in the face of rapidly rising seas9.

Although some studies have indicated that catastrophic wetland loss is imminent8,9,14,15,16,17,18, others have argued that wetlands are resilient ecosystems that can readily adapt to future change19,20,21,22,23,24. The ensuing uncertainty from this debate has hindered decisions regarding how and when to prepare for transformative coastal change. For example, should managers use limited resources to: (1) resist change and help current coastal wetlands adapt where they are now; or (2) acknowledge transformative coastal change and facilitate the retreat and migration of wetlands into new positions in the coastal landscape? To better inform future-focused coastal conservation and restoration decisions, there is a pressing need to advance understanding regarding the potential timing of coastal wetland drowning under rapidly rising seas. Here, we use emerging knowledge regarding coastal wetland drowning thresholds and future relative sea-level rise rates to elucidate when and where coastal wetland drowning could begin across the conterminous United States. We show how the best available data concerning these two critical components can bring clarity to this debate and provide powerful information at large spatial scales regarding the initiation of coastal wetland drowning under accelerated sea-level rise.

In the past decade, there has been growing consensus – from paleoecological records, contemporary instrumental records, and numerical models – regarding critical thresholds for wetland drowning8,16,17,25,26,27,28. Coastal wetland responses to sea-level rise are greatly influenced by biogeomorphic feedbacks that enable wetlands to build surface elevation to avoid drowning and adjust to rising seas20,29,30. These feedbacks involve nonlinear interactions between flooding regimes, root growth, sedimentation, and other factors, which can collectively lead to the vertical accumulation of organic matter and sediments under the right conditions. Although these feedbacks can enable wetlands to gain elevation to adjust alongside small increases in sea level, there are submergence limits16, or drowning thresholds, beyond which coastal wetlands cannot adjust to rising seas and begin to drown8,16,17,25,26,27,31. Wetland drowning begins when sea-level rise rates exceed the upper bounds for plant growth and wetland vertical adjustment, ultimately resulting in plant mortality, wetland edge erosion, wetland fragmentation, and the conversion of wetlands to open water. However, the loss and complete submergence of wetlands does not occur immediately once a drowning threshold has been crossed. Although some wetlands may drown within decades of crossing a threshold, others may persist for longer16,32,33,34, due to factors that can prolong drowning, including tidal range, elevation capital, sediment inputs, landscape position, and the rate of relative sea-level rise.

A recent global synthesis of paleo-stratigraphic and contemporary in situ data from marshes and mangrove forests8 concludes that: (1) a deficit between coastal wetland vertical adjustment and relative sea-level rise is likely at 4 mm yr−1 and highly likely at 7 mm yr−1; and (2) coastal wetland drowning and conversion to open water is likely when these rates are sustained. Regional analyses identify drowning thresholds as 5 mm yr−1 for eastern U.S. marshes25, 6–9 mm yr−1 for Mississippi River delta marshes16, and 7 mm yr−1 for marshes in San Francisco Bay28. Although some of these thresholds are derived from paleoecological records, others are derived from contemporary instrumental records and/or numerical models. Here, we evaluated three sea-level rise thresholds for wetland drowning (4, 7, and 10 mm yr−1), which were selected to bracket the rates identified in these global and regional studies.

Our study area spans the coastal conterminous United States, which includes Washington, D.C., and 22 coastal states along the Pacific Ocean, Gulf of Mexico, and Atlantic Ocean (Fig. 1). Within the study area, we created a grid of 168 1-degree resolution cells for data acquisition and analyses. We examined three alternative sea-level rise scenarios, the Intermediate-Low, Intermediate, and Intermediate-High relative sea-level rise scenarios identified by the most recent U.S. interagency sea-level rise technical report3. These three scenarios correspond to 0.5, 1.0, and 1.5-m global mean sea-level rise levels by 2100, respectively, relative to a 2000 baseline.

Fig. 1: Maps showing when coastal wetland drowning is expected to begin under alternative sensitivity thresholds for wetland drowning and sea-level rise (SLR) scenarios across the conterminous United States.
figure 1

Colors indicate the year in which wetland drowning is expected to begin. This is when rising sea levels are expected to cross thresholds for coastal wetland drowning. The loss and submergence of wetlands do not occur immediately once a drowning threshold has been crossed. Although some wetlands may drown within decades of crossing a threshold, others may persist longer. Panels a–i are combinations of SLR scenarios and sensitivity thresholds. Intermediate-Low corresponds with a 0.5 m Global Mean SLR (GMSLR) by 2100, Intermediate corresponds with a 1.0 m GMSLR by 2100, and Intermediate-High corresponds with a 1.5 GMSLR by 2100. Note that the earliest possible date in our analysis is 2020. However, there are some areas, like Louisiana’s Mississippi River delta, where wetland drowning has already begun7. Base map layers are from publicly available U.S. Geological Survey datasets.

Full size image

For each of the three global mean sea-level rise scenarios, our analyses incorporate decadal regional relative sea-level rise projections3, which account for critical processes that influence local sea-level rise rates, such as: (1) changes in gravitational field and rotation; (2) shifts in oceanographic factors; and (3) vertical land movement (for example, subsidence or uplift) due to sediment compaction, glacial isostatic adjustment, groundwater withdrawals, fossil fuel withdrawals, and other factors. These relative sea-level rise projections are critical in the United States because there is much spatiotemporal variation, with some regions affected by rates that are much higher (for example, the western Gulf of Mexico) or much lower (for example, the Northwest region’s Pacific coast) than global mean sea-level rise rates. We used the three relative sea-level rise scenario projections to determine when the selected drowning thresholds would be crossed. For each grid cell, we used the decadal-scale data associated with the regional relative sea-level rise scenarios3 to identify the decade of coastal wetland drowning initiation, which was defined as the decade in which relative sea-level rise rates are projected to be at rates greater than or equal to 4, 7, or 10 mm yr−1 (see Supplementary Fig. 1 for examples from 4 of the 168 cells). To determine the wetland area that could be affected, the extent of potential drowning initiation within each grid cell was determined using current estuarine wetland coverage data35.

Coastal wetlands in the United States are diverse. Thus, coastal wetland drowning means different things in different regions (Fig. 2a, b). For example, in the Everglades ecosystems of tropical South Florida, coastal wetland drowning often refers to the submergence of mangrove forests and freshwater marshes as they are converted to mudflats and open water36,37. In the Mississippi River delta and Chesapeake Bay, coastal wetland drowning often refers to the submergence of temperate coastal marshes and freshwater forests7,16. In arid parts of California and south Texas, coastal wetland drowning may refer to the submergence of hypersaline, succulent plant-dominated marshes38. Here, our use of the term coastal wetland drowning encompasses all these regional variations of vegetated coastal wetland conversions to tidal flats or open water due to submergence, wetland erosion, and wetland fragmentation.

Fig. 2: Coastal wetlands in the United States are diverse.
figure 2

Thus, coastal wetland drowning means different things in different regions. For example, in the Mississippi River Delta and Chesapeake Bay, coastal wetland drowning can refer to the submergence and erosion of coastal marshes (a). In the Everglades ecosystems of South Florida, coastal wetland drowning can refer to the conversion of mangrove forests to mudflats (b). The most prominent wetland drowning hotspot in the United States is within Louisiana’s Mississippi River delta (c; blue shows land loss and green shows land gain, adapted from Couvillion et al.7). Louisiana has lost 4833 km2 of land area between 1932–20167, due in part to high rates of subsidence and relative sea-level rise. Photograph credits: a Karen McKee, USGS; (b) Michael Osland, USGS. Imagery in panel c from Earthstar Geographics and its licensors, copyright 2024. The state base map layer is from a publicly available U.S. Geological Survey dataset.

Full size image

Results and Discussion

Our results show that there is substantial spatial variation across the conterminous United States in wetland abundance and exposure to high rates of relative sea-level rise, which produces much variability in the potential timing and extent of wetlands crossing thresholds for initiating drowning (Figs. 1, 3, 4; Supplementary Figs. 2-4). We present these results at state-specific (Fig. 3; Supplementary Fig. 2) and conterminous United States scales (Fig. 4). In general, low-lying regions that currently have high areal coverage of coastal wetlands10 are also among the most vulnerable to wetland drowning (Fig. 3; Supplementary Fig. 5). The most prominent wetland drowning hotspot is within Louisiana’s Mississippi River delta. From 1932 to 2016, Louisiana lost 4833 km2 of coastal wetlands due to high rates of subsidence and relative sea-level rise (Fig. 2c)7. Louisiana currently accounts for nearly 30% of the conterminous United States’ coastal wetlands10, and those wetlands are expected to continue to be among the first to drown16 (Figs. 1, 3, 4; Supplementary Fig. 2). At rates exceeding 6 to 9 mm yr-1, considerable conversion of marsh to open water is expected to occur within 50 years in the Mississippi River delta16. A recent analysis of 253 surface elevation change monitoring stations in Louisiana indicated that 87% of wetlands were unable to keep up with rising water levels over a 13-year period of unusually rapid sea-level rise greater than 10 mm yr−1 39.

Fig. 3: The potential state-specific timing and extent of the start of coastal wetland drowning under alternative sea-level rise (SLR) scenarios and sensitivity thresholds for wetland drowning across the conterminous United States.
figure 3

Data are presented for 22 coastal states and Washington D.C. The y-axes reflect the coastal wetland area that could begin to drown. These are wetlands exposed to relative SLR rates that exceed thresholds for initiating coastal wetland drowning. Panels a–i are combinations of SLR scenarios and sensitivity thresholds. For each panel, the SLR scenario and sensitivity threshold rates are shown on the right. Inter.-Low = Intermediate-Low (0.5 m GMSLR by 2100), Inter. = Intermediate (1.0 m GMSLR by 2100), Inter.-High = Intermediate-High (1.5 GMSLR by 2100). The loss and submergence of wetlands do not occur immediately once a drowning threshold has been crossed. Although some wetlands may drown within decades of crossing a threshold, others may persist longer. The earliest possible date in our analysis is 2020. However, there are some areas, like Louisiana’s Mississippi River delta, where wetland drowning has already begun. For percentage-based analyses, which are helpful for states with small wetlands, see Supplementary Fig. 2. GMSLR = global mean SLR. Note that these figures include tidal saline wetlands but they do not include tidal freshwater wetlands because there is not a national land cover classification that identifies tidal freshwater wetlands.

Full size image
Fig. 4: The potential timing and extent of the onset of coastal wetland drowning under alternative sea-level rise (SLR) scenarios and sensitivity thresholds for wetland drowning (symbols within panels) for the conterminous United States.
figure 4

The y-axes reflect the percent of coastal wetlands that could begin to drown on an areal basis. These are wetlands exposed to relative SLR rise rates that exceed thresholds for initiating coastal wetland drowning. The loss and submergence of wetlands do not occur immediately once a drowning threshold has been crossed. Although some wetlands may drown within decades of crossing a threshold, others may persist longer. Panels a–c represent three alternative SLR scenarios. GMSLR = global mean SLR. The earliest possible date in our analysis is 2020. However, there are some areas, like Louisiana’s Mississippi River delta, where wetland drowning has already begun7.

Full size image

In general, wetland drowning is expected to begin earliest along the wetland-rich northwestern Gulf of Mexico, south-Atlantic, and mid-Atlantic coasts, especially within subsidence hotspots, where relative sea-level rise rates already exceed critical thresholds for drowning. In addition to Louisiana, the Texas coast and Chesapeake Bay are also wetland-rich regions10 that also contain subsidence hotspots40,41,42, where wetland drowning is happening now43,44. Although Florida’s coastal wetlands have been comparatively stable over the past thousand years due to minimal subsidence and small rates of relative sea-level rise45,46, the potential extent of future wetland drowning in Florida is very large36,37 due to the state’s low-lying topography and because the Florida coast contains roughly 25% of the conterminous United States’ coastal wetlands10, including the large mangrove-marsh mosaic of the Greater Everglades47. Conversely, the onset of wetland drowning is expected to be later along the Pacific coast, especially along the Washington and Oregon coasts due to the effects of post-glacial rebound, which produces relative sea-level rise rates that are lower than global mean sea-level rise rates48,49 (Supplementary Fig. 1).

Our findings show that the timing and extent of wetland drowning are highly sensitive to relative sea-level rise rates and wetland drowning thresholds (Figs. 1, 3, 4, Supplementary Figs. 1–5). For example, wetland drowning is expected to begin on all three coasts by 2040 under the highest sea-level rise scenario and lowest drowning threshold (Fig. 1g; Fig. 3g; Fig. 4c). Conversely, wetland drowning is delayed under the lowest sea-level rise scenario and highest drowning threshold (Figs. 1c,  3c, 4a). Under the Intermediate-Low scenario, the timing and extent of wetland drowning vary greatly in response to drowning thresholds (compare Fig. 1a–c; Fig. 3a–c; Fig. 4a). However, under the Intermediate-High scenario, wetland drowning begins between 2020–2060 under all three drowning thresholds (compare Fig. 1g-i; Fig. 3g–i; Fig. 4c). These analyses highlight the importance of refining understanding of wetland drowning thresholds and the factors that control these thresholds (for example, soil properties, sediment inputs, tidal range, plant community composition). Our findings also highlight the importance of utilizing the best available relative sea-level rise projections to gauge exposure to inundation levels that exceed these thresholds.

By focusing on wetland drowning thresholds and relative sea-level rise rates that exceed these thresholds, we provide an approach for rapidly gauging when and where rising sea levels could cross tipping points for initiating wetland drowning at large spatial scales. Where data are available, this simple approach can be applied at national or regional scales to quickly quantify and compare differences in the timing and extent of the onset of wetland drowning and thereby help prioritize future scientific and management actions. Our results can be used by scientists to prioritize vulnerable coastal areas where customized local models and improved understanding of local relative sea-level rise rates and local drowning thresholds would be valuable. Local models may be needed to determine the duration of drowning to account for factors that can prolong drowning, including elevation capital, landscape position, plant community, tidal range, sediment inputs, and the rate of relative sea-level rise16,32,33,34,50. Our analyses utilize thresholds and relative sea-level rise rates for the conterminous United States. However, this approach may need to be customized for other parts of the world where thresholds are different or coastal wetland responses to sea-level rise are governed by additional factors.

Collectively, our findings underscore the importance of minimizing sea-level rise acceleration to avoid catastrophic wetland losses and transformative coastal change associated with wetland drowning. Our results also show that wetland drowning under many scenarios could begin within decades. Thus, there is an urgent need to better anticipate and prepare for the social and ecological implications of coastal wetland losses and transformations due to rapidly rising seas.

Methods

Study grid

Our study grid of 168 1-degree resolution cells was created to: (1) match the spatial resolution and registration of the relative sea-level rise data3; and (2) include only cells containing estuarine vegetated wetlands, which were identified using estuarine wetland coverage data (that is, the estuarine emergent, scrub-shrub, and forested wetland classes) from NOAA’s 2016 Coastal Change Analysis Program [C-CAP] 30-m land cover data35. Note that these C-CAP categories include vegetated tidal saline wetlands (that is, the vegetated “estuarine wetland” classes), but they do not include the adjacent tidal freshwater wetlands, which are also vulnerable to the same drowning thresholds. National land cover products like C-CAP and NLCD do not contain a category that identifies tidal freshwater marshes and tidal freshwater forests. Thus, our areal estimates reflect the abundance of tidal saline wetlands within a cell. The inclusion of the tidal freshwater wetlands would increase the total area of vulnerable coastal wetlands, especially in wetland-rich states like Louisiana, Florida, North Carolina, Texas, South Carolina, Georgia, Alabama, and Mississippi.

Global and regional relative sea-level rise scenarios

Our analyses incorporate decadal regional relative sea-level rise projections for the Intermediate-Low, Intermediate, and Intermediate-High scenarios identified by the most recent United States interagency sea-level rise report3. The relationships between these scenarios and the Intergovernmental Panel on Climate Change (IPCC) emissions scenario-based and warming-level-based projections2, including sea-level rise exceedance probabilities under alternative global warming levels can be found in Sweet et al.3. The IPCC’s Sixth Assessment Report on Climate Change projects a global mean sea-level rise by 2100 of 0.56 m (0.44-0.76 m) for the SSP2-4.5 scenario and 0.77 (0.63–1.01 m) for the SSP5-8.5 scenario2, relative to a baseline of 1995–2014. The SSP2-4.5 and SSP5-8.5 scenarios represent Intermediate and Very High future greenhouse gas emissions scenarios, respectively, and those scenarios are based upon processes for which projections can be made with at least medium confidence2. Higher amounts of sea-level rise could be caused by low-confidence processes (for example, ice sheet dynamics) whose quantification is highly uncertain2,3. Those low-confidence processes are not included in the IPCC medium-confidence projections2,3.

On a regional basis, the three sea-level rise scenarios from Sweet et al. 2022 closely align with the regional trajectories that sea-level rise is currently on and expected to maintain for at least the next several decades3. For example, the trajectory along the southeastern U.S. Atlantic coast by 2050 falls between the Intermediate and Intermediate High scenarios (0.36 and 0.43 m rise by 2050, respectively, relative to 2000 sea levels)3. The trajectory along the western Gulf of Mexico coast is even higher, falling between the Intermediate and Intermediate High Scenarios (0.57 and 0.63 m rise by 2050, respectively)3. In contrast, the trajectory along the northwestern U.S. Pacific coast is lower, falling between the Intermediate-Low and Intermediate scenarios (0.15 and 0.18 m rise by 2050, respectively)3.

Data analyses: timing and extent of the start of wetland drowning

We evaluated the following three thresholds for wetland drowning: sea-level rise rates of 4, 7, and 10 mm yr−1. These thresholds were selected to bracket studies indicating that coastal wetland drowning begins when sea-level rates exceed thresholds for vertical adjustment8,16,17,25,26,27,28. The relative sea-level rise projection data begin in 2005 and are available for each decade thereafter from 2020-2150. We used the median projection values for each scenario. To determine the decadal-scale rate of relative sea-level rise, we divided the sea-level increase by the number of years (that is, 10 for most intervals, 15 for 2005-2020). For example, a relative sea-level increase of 30 mm across a decade was converted to a relative sea-level rise rate of 3 mm yr-1. There were some instances where the decadal-scale relative sea-level rise rates would rise above a drowning initiation threshold but then drop below the threshold in the next decade. To determine the decade of wetland drowning initiation, we only used decades where the rate for the following decade was also above the specified threshold. The decade for the onset of drowning was identified by the last year of the first decade (for example, 2030 for the 2020-2030 decade where the 2030-2040 decade was also above the specified threshold). See Supplementary Fig. 1 for examples of the relative sea-level rise data relative to thresholds for four individual cells in Washington (Supplementary Fig. 1a), Maine (Supplementary Fig. 1b), Florida (Supplementary Fig. 1a), and Louisiana (Supplementary Fig. 1d). In Supplementary Figs. 3-4, we provide maps showing when coastal wetland drowning is expected to begin for the low (17th) and high (83rd) percentiles associated with each of the three sea-level rise scenarios. These maps are included to complement the analyses in the main text, which are focused on the median sea-level rise projections.

Data analyses and visualization

We developed maps that show the timing of the onset of wetland drowning under each of the nine threshold-by-sea-level rise combinations (that is, three thresholds by three sea-level rise scenarios). To determine the amount of wetland affected, the extent of potential drowning initiation within each grid cell was determined using estuarine vegetated wetland coverage data (that is, the estuarine emergent, scrub-shrub, and forested wetland classes) from NOAA’s 2016 C-CAP 30-m land cover data35. At the state scale, we developed figures that show the timing, area, and percent of wetlands expected to cross drowning initiation thresholds for each of the 22 coastal states and Washington D.C. At the conterminous United States scale, we developed figures that show the timing and percent of wetland areas expected to cross drowning initiation thresholds.