Mangrove ecosystems provide vital contributions to human well-being, which are increasingly threatened by climate change. The mangrove ecosystem services of carbon sequestration and disaster risk reduction are especially important for climate change mitigation and adaptation, respectively^1,2. However, these ecosystems are subject to both slow-onset climate-related changes, such as sea-level rise, and changes in the intensity and frequency of extreme events, such as tropical cyclones. Both can have potentially large impacts on ecosystem service provision by mangroves; however, threats to mangroves from tropical cyclone frequency changes and sea-level rise have been studied separately^3,4,5,6, and often without considering ecosystem services, or restoration and conservation priorities^7,8. Thus, an assessment of risk that accounts for changes in ecosystem disturbance through tropical cyclones and simultaneous sea-level rise for both mangroves and their services is urgently needed.

Mangroves in the world’s tropical cyclone basins are adapted to their historical disturbance regimes^3,9, yet novel disturbance regimes induced by climate change, with greater frequency and intensity of extreme events than historically, can substantially alter ecosystem composition, structure, and function^10,11. Investigating tropical cyclone risk to mangroves under climate change., Mo et al. ³ project substantial increases in tropical cyclone frequency at the country-level in North and Central America under 2 °C warming, while Kropf et al. ⁶ find widespread risk for most mangrove ecoregions affected by tropical cyclones under Shared Socio-economic Pathway (SSP) 585 until 2050. Tropical cyclone frequency, alongside temperature and precipitation, plays a critical role in limiting mangrove canopy height¹² and affects forest composition and functional traits¹³, which raises the concern that climate-driven changes in tropical cyclones could lead to substantially altered mangrove ecosystems. Further, increased tropical cyclone frequency could prevent mangroves from recovering fully in between events, potentially decreasing mangrove area¹⁴ and, in extreme cases, inducing ecological regime shifts by converting mangrove forest to mudflats¹⁵. These potential changes in mangrove forest structure and function could alter the biodiversity that depends on these systems, ecosystem services (e.g., carbon sequestration, coastal protection, fisheries), and the role of mangroves in climate mitigation and adaptation.

Under climate change, the impact of tropical cyclones on mangroves is further compounded by sea-level rise. Biogenic feedbacks can allow mangroves to adjust vertically to sea-level rise, however, high rates of sea-level rise may lead to elevation deficits beyond mangrove adaptive capacity^4,5. Saintilan et al. ⁴ project widespread loss of coastal ecosystems, including mangroves, due to rates of sea-level rise exceeding adaptive capacity at warming levels of 1.5–4.0 °C. While sea-level rise constitutes a chronic stressor and tropical cyclones are an acute disturbance, both can cause mangrove mortality and subsequently peat collapse¹⁶. Thus, the capacity of mangroves to adapt to sea-level rise may be compromised by tropical cyclone-induced mortality, and vice versa, leading to peat collapse and surface elevation loss¹⁵. Hence, interactions with sea-level rise play a key role in the risks of a changing tropical cyclone disturbance regime, necessitating studies that consider both.

This combined risk from changes in tropical cyclones and sea-level rise is relevant to conservation and restoration planning in ways that have not been fully considered to date. Mangroves play a critical role in climate mitigation and adaptation^1,17, and millions of people in coastal areas around the world depend on mangroves for coastal protection, nutrition, and income^18,19. Consequently, mangrove protection and restoration are receiving increasing attention from conservation actors and policy^8,20. While these initiatives recognize the relevance of sea-level rise and altered tropical cyclones under climate change, they do not include comprehensive future risk assessments for the mangrove ecosystem services these initiatives desire to increase. Structural changes, degradation, or loss of mangrove forests triggered by a novel disturbance regime could have cascading impacts, resulting in the loss of the vital services provided by these ecosystems. For instance, storm-induced damages can compromise the protective function of mangrove forests for future events and increase risk for coastal communities and assets⁹. Mangrove carbon stocks can be negatively affected by post-storm disturbance processes such as peat collapse¹⁷. Since mangroves serve as habitats for many fish and invertebrate species, fishery benefits are threatened by mangrove loss^18,19. To support ecosystem management decisions that are adaptive to climate change, understanding potential losses of ecosystem services induced by novel disturbance regimes is crucial^11,21.

Here, we assess the risk of a disturbance regime shift for mangroves²² under climate-modified tropical cyclone activity^23,24 and sea-level rise^4,25. We combine our spatially explicit risk index with the global mangrove conservation priority areas by Dabalà et al. ⁷, which rank mangrove areas based on biodiversity, coastal protection for people and property, carbon sequestration, and fishery benefits. We then analyze where and to what extent the ecosystem services of coastal protection for people and property¹, fishery benefits¹⁹, and carbon sequestration^4,26 are at risk. Across SSPs, we find about half of global mangrove extents at risk by the end of the century, with areas of high ecosystem service provision at proportionally higher levels of risk.

Changes in tropical cyclone disturbance for global mangrove extents

We consider a disturbance regime to be “the spatial and temporal dynamics of disturbance across a landscape over a long timeâ€�¹¹. The IPCC defines risk as a combination of hazard, exposure, and vulnerability²⁷. Since novel disturbance regimes are characterized by a departure from baseline conditions¹¹, we define tropical cyclone disturbance regime shift risk (from here on risk) for mangroves (Fig. 1A), as a combination of shifts in the frequency with which mangroves experience a given tropical cyclone windspeed intensity (Fig. 1B), and the rate of relative sea-level rise (Fig. 1C) until the end of the century. Based on previous research¹⁵, we assume sea-level rise to increase vulnerability to tropical cyclones, and vice versa, which we translate into in the five risk categories (low, medium, high, very high, severe) reflected in the risk matrix (Fig. 1A). A low to medium risk rank reflects rates of sea-level rise that mangroves are likely able to adapt to in combination with an increased frequency of low tropical cyclone windspeed intensities. A high or very high-risk rank reflects a combination of high to very high rates of sea-level rise, combined with changes in high tropical cyclone windspeeds. Finally, a severe risk rank indicates a combination of sea-level rise rates that mangroves are highly unlikely to adapt to with increased frequency in very high windspeeds. For further clarification on how the risk index is constructed, please see Supplementary Fig. 1.

A Risk index at global mangrove locations for SSP245, 370, and 585 until 2100. The legend is the risk matrix to the right showing relative sea-level rise (RSLR) and tropical cyclone (TC) risk factors and the resulting combined risk rank. Risk ranks 1–5 correspond to low, medium, high, very high, and severe risk. RSLR1–3 correspond to the median rates of regional relative sea-level rise of 0–4 mm y^-1, 4–7 mm y^-1, and 7 mm y^-1, while TC1-3 refer to the windspeed categories (1) 33–49 m s^−1, (2) 50–70 m s^−1, and (3) >70 m s^−1. For further clarification on how the risk index is constructed, please see Supplementary Fig. 1. B Change in tropical cyclone (TC) frequency per intensity category for SSP245, 370, and 585 from the reference period 1995–2014 until 2100. Points are shown where the frequency at least doubles per given intensity category or if an area is newly affected by windspeeds of that intensity at least once in 250 years. C Median relative sea-level rise (RSLR) at mangrove locations for SSP245, 370, and 585 until 2100.

Full size image

The probabilistic tropical cyclone data used in this study are downscaled from a statistical-dynamical tropical cyclone model^23,24, which is forced by CMIP6 global climate model outputs across various SSPs for the present day and future climate. The tropical cyclone simulations over the historical period establish a baseline that realistically characterizes the current tropical cyclone disturbance regime, overcoming the limitations of the historical record of observations²⁸. Meanwhile, tropical cyclone data for the three SSPs represents future disturbance regimes by the end of the century. Our tropical cyclone windspeed intensity categories correspond to (1) 33–49 m s^-1, (2) 50–70 m s^-1, (3) >70 m s^-1, which are equivalent to Saffir-Simpson Hurricane Wind Scale categories 1–2, 3–4, and 5, respectively²⁹. We consider mangroves to experience a novel disturbance regime if they are newly affected by windspeed intensities of a given intensity category, or if the relative change in frequency within the category exceeds 100% (Fig. 1B). Only windspeed intensities with return periods of up to 250 years are considered ecologically relevant and included⁶.

We define our sea-level rise categories as (1) 0–4 mm y^-1, (2) 4–7 mm y^-1, and (3) 7 mm y^-1 of median regional rate of sea-level rise until the end of the century (Fig. 1C). Previous research has investigated potential thresholds in the vertical adjustment capacity of mangroves, and found deficits are likely at sea-level rise rates of 4 mm y^-1, and highly likely at 7 mm y^-14.

Applying this methodology, we find widespread changes in tropical cyclone frequency across all levels of windspeed intensity and high rates of sea-level rise, which put mangroves at varying levels of risk in all tropical cyclone basins (Fig. 1). Globally, depending on the SSP, we find 40–56% of mangrove area will be at high to severe risk (Table 1). A region of particular concern is Southeast Asia, where, depending on the SSP, 52–78% of current mangrove extents are at high to severe risk (Fig. 2, Supplementary Figs. 7 and 8). In Southeast Africa, Central America, and Southwest Pacific, 48–86%, 51–71%, and 17–31% of current mangrove extents, respectively, are at high to severe risk depending on the SSP (Fig. 2, Supplementary Fig. 7 and 8). Hotspots of severe risk in Southeast Asia are in the Philippines, Vietnam, Malaysia, and parts of Indonesia (Fig. 1A and Fig. 2A). Areas in the Southwest Pacific, especially northern Australia, New Caledonia and Fiji, are also at severe risk (Fig. 1A and Fig. 2A). In contrast, severe risk levels are less widespread in Central America and Southeast Africa but do occur at the east coast of Florida and the Bahamas, as well as in northern Mozambique and southern Tanzania (Fig. 1A and Fig. 2A).

A shows the risk score for mangroves in the four hotspot areas. B shows the proportions of ecosystem services (people and assets protected, fishing intensity, and carbon storage), as well as the areal extent of mangroves, at risk across the 5 risk scores: low, medium, high, very high, and severe risk. Please see Supplementary Figs. 7 and 8 for ecosystem services at risk under SSPs 245 and 585, and Supplementary Figs. 9–20 for more highly resolved hotspot maps under SSPs 245, 370 and 585.

Full size image

Global mangrove ecosystem services and biodiversity at risk

To investigate risks to ecosystem service provision, we combine our risk index with the mangrove conservation priority areas by Dabalà et al. ⁷, which rank global mangrove area based on mangrove species biodiversity, coastal protection for people and property, carbon sequestration, and fishery benefits. We then investigate the distribution of risk levels for mangrove areas ranked as high conservation priorities areas⁷, i.e., the top 10% and top 25% highest ranking areas in terms of their biodiversity, coastal protection for people and property, carbon sequestration and fishery benefits. Across all SSPs, these high-priority mangrove areas are proportionally at higher risk than mangrove areas overall (Table 1). For example, under SSP370, 13% of global mangrove area will be at severe risk by the end of the century, while for the top ranking 25% it is 2% more, and for the top ranking 10% the proportion at severe risk is almost double than that of mangroves in general (Table 1).

Zooming in on ecosystem service provision across regions, we observe distinct patterns of the areal extent and ecosystem services affected by varying degrees of risk (Fig. 2). We project Southeast Asia to be most affected, both in terms of area and ecosystem services. Under SSP370, mangroves providing 98% and 97% of  coastal protection to people and properties, respectively, are projected to be at high to severe risk, while 62% of carbon and 78% fishery benefits are highly to severely threatened under SSP370 (Fig. 2, Panel B). In Central America, especially coastal protection for assets is at high to severe risk (75% for SSP370), while in Southeast Africa about half of all services are at high to severe levels of risk. The Southwest Pacific is the least affected, with carbon sequestration being the most threatened service in the region (28% for SSP370).

Further, we observe heterogeneity in terms of proportions of services at risk compared to the locations’ mangrove areal extent at risk. In all regions, the proportions of carbon sequestration and to a lesser degree also fishery benefits at a given level of risk correspond closely to the proportion of overall mangrove area (Fig. 2B). However, mangroves providing coastal protection, both for people and property tend to be at higher levels of risk in all regions than mangrove area in general, except in Central America. This heterogeneous distribution of some ecosystem services at risk in relation to mangrove area underscores the importance of mapping risk to services of interest explicitly.

Unchartered territory

Under climate change, disturbance regimes of different degrees of novelty will emerge¹¹. We project that the frequency changes across different tropical cyclone windspeed intensities (Fig. 3) will lead global mangrove areas to be newly or more frequently affected by high (categories 3 and 4, middle row of maps in Fig. 3) and very high (category 5, bottom row of maps in Fig. 3) windspeeds. These areas include South Asia, Southeast Asia, Southeast Africa, Central America and the Caribbean. Therefore, some locations will experience future disturbance regimes that resemble the current regimes of other areas, e.g., Central American mangroves receiving high intensity winds more frequently in the future, similarly to Southeast Asian mangroves currently (Fig. 3, also see Supplementary Figs. 2 and 3). Other regions may become newly affected by tropical cyclones altogether (shown as red points in Fig. 3), e.g., mangroves at higher latitudes due to a poleward shift in tropical cyclones. Even at the lowest emission pathway, many areas across the globe, including in Central America, Southeast Africa, East Asia, and Oceania, will become newly affected by very high intensity tropical cyclone winds (Fig. 3). These projected regimes are locally novel but may be similar to historical regimes in other geographies.

The columns of maps from left to right refer to the SSPs 245, 370, and 585, while the rows show different windspeed intensities from top to bottom, corresponding to categories 1–2, 3–4, and 5 on the Saffir-Simpson scale. For points which have previously been affected by the given windspeed intensity at least once in 250 years the relative change in frequency is shown on a continuous color scale (blue to yellow). Points which are projected to be newly affected by a given windspeed intensity at least once in 250 years by 2100 are shown in a separate color (red).

Full size image

We also project that some areas will experience a non-analog future, i.e., a globally novel disturbance regime, with tropical cyclones windspeed intensities at frequencies that are currently unprecedented in the baseline. Fig 4 shows mangrove extents which are projected to experience novel tropical cyclone conditions under SSP245, 370, and 585 by the end of the century, meaning that the projected future frequency exceeds the current global maximum frequency per given intensity category. Here, Southeast Asia and especially the Philippines clearly stand out, irrespective of intensity category or SSP. Historically, the global maximum frequency of tropical cyclone windspeeds at any mangrove centroid above 70 m s^-1 is 0.02 (a return period of approximately 50 years). In the east of the Philippines, however, this frequency is projected to be exceeded by up to 254% in some mangrove areas, leading to a frequency of 0.08 (a return period of 12.5 years) (Fig. 4).

The color bar indicates the projected frequency by 2100 as a percentage of the historical global maximum frequency per intensity category, with dark colors indicating projected frequencies of up to 200% of the historical global maximum frequency and lighter colors indicating projected frequencies up 350% of the historical global maximum. Only mangrove centroids with future TC frequencies higher than the historical global maximum per intensity category are shown. The historical maximum frequencies per category are 0.45 for TC1, 0.11 for TC2 and 0.02 for TC3, which corresponds to return periods of 2.22 years, 9.43 years, and 41.92 years, respectively.

Full size image

We present the first global, spatially explicit probabilistic projection of climate-modified changes in tropical cyclone disturbance regimes in combination with sea-level rise affecting mangroves. This risk index incorporates changes in two stressors, tropical cyclones and sea-level rise, and allows connecting the risk perspective with the spatial distribution of mangroves and the services they provide. Our analysis highlights widespread risk to mangrove areas and ecosystem services under climate change globally and identifies Southeast Asia as a region of especially high concern.

Uncertainties and limitations

An improved understanding of the climate-induced changes in the tropical cyclone disturbance regime of mangrove ecosystems is vital in anticipating potential ecological responses of mangrove forests to climate change. However, the findings presented here are subject to several uncertainties and limitations. First, the choice of tropical cyclone model, as well as the global climate model (GCM) data used to drive it influences the subsequent risk assessment^28,30. For instance, using GCMs with high climate sensitivity has been shown to result in higher TC risk increases under climate change³⁰. We account for this by using median TC frequency changes based on three GCMs of varying climate sensitivities in our results. Second, the choice of the relative change in frequency threshold beyond which to consider mangroves at risk is a modeling choice which affects the risk score assigned to the mangrove area and its associated ecosystem services. Our chosen threshold of a doubling in frequency by 2100 offers a balance between considering every deviation from the baseline novel and only considering the most extreme deviations as relevant (for more detail on the threshold choice, please consult the “Methodsâ€� section). Third, a static distribution of mangrove extents and ecosystem services based on current conditions is assumed. Future research could integrate models for potential future mangrove distribution under other projected climatic changes, such as temperature and precipitation³¹ with the risk approach presented here.

To fully grasp novel ecological outcomes, it is not only necessary to analyze changes in the disturbance regime, but also investigate ecological responses and recovery mechanisms¹¹. The global approach taken in this study does not allow including a variety of factors, which may greatly affect mangrove vulnerability and recovery such as mangrove forest structural properties, geomorphology, and anthropogenic disturbance³². While these factors could be included in studies aiming to mechanistically model the processes leading to ecological shifts, our approach is more suited to investigate characteristics of the changing disturbance regime and its spatial relation to mangroves and their services.

Risk hotspots for ecosystem services

The overrepresentation of service-providing mangroves in high-risk categories points to considerable societal impacts that could occur across regions due to disturbance regime shifts under climate change. In Central America our findings indicate high risk for several ecosystem services but especially coastal protection for assets. Considering the magnitude of economic benefits provided through risk reduction by mangroves in the region^1,33, these ecosystem functions and recovery should be closely monitored³². In Southeast Africa all assessed ecosystem service types are at risk which is alarming given the region’s high dependence on coastal ecosystem services for their livelihoods, nutrition, and coastal protection¹⁸. We project risk to current mangrove carbon stocks in the Southwest Pacific, including Australia, where previous research indicates both losses and gains in carbon accumulation may occur depending on the climate change scenario and the level of coastal squeeze¹⁷. Our risk index can complement such research in the prioritization and monitoring of blue carbon projects in the region. Southeast Asia is the area of highest concern, both in terms of widespread severe physical risk and the share of ecosystem services affected.

According to our projections, a historically and globally novel disturbance regime may emerge in the Philippines, with tropical cyclone frequencies which are unprecedented in the baseline. Historically, the Philippines have been affected by a large share of all recorded tropical cyclone events and are exposed to storms that closely mirror the global distribution of tropical cyclones¹. The country also has total mangrove extents of more than 260,000 ha (approximately 2% of global extents)^1,34. Considering the frequency increases we find especially in high intensity tropical cyclone windspeeds in this area, it is uncertain in what state mangroves will be able to persist in the Philippines under unprecedented tropical cyclone frequencies across various levels of intensity. Given the relationship between tropical cyclone frequency and mangrove forest structural attributes, such as canopy height and aboveground biomass¹², as well as community dynamics¹³, we may at least expect considerable structural changes under this unprecedented disturbance regime. Mangrove restoration may support recovery, however, restored mangrove forests are estimated to require 55 years to reach functional equivalency to their natural counterparts³⁵. Consequently, the nature and degree of future ecosystem service provision in these areas may also be subject to considerable change.

Outlook and management implications

We project severe and widespread risk for mangrove ecosystems, especially in places currently providing high levels of ecosystem services. These estimates hint at large potential socioeconomic impacts resulting from climate-modified disturbance regime. Indeed, more than 775 million people are currently estimated to be highly dependent on coastal ecosystems¹⁸, and the number of people depending on protection by coastal ecosystems is expected to increase further under climate change³⁶. Financially, loss of mangrove protective functions could contribute to accelerating losses under climate change, since even under current conditions mangroves provide flood protection benefits beyond US$ 65 billion annually¹. Neither the loss of mangroves themselves, nor the value derived from their contributions to people is currently accounted for in Loss & Damage discussions³⁷. Thus, there is a pressing need to further develop conceptual and modeling frameworks, as well as funding mechanisms, to capture the loss of ecosystems and their services.

Developing effective ecosystem management that is adaptive to changing disturbance regimes is imperative^11,21. Our findings support this call to action, underscoring the need to adapt conservation and restoration planning to sustain ecosystem service provision under climate change and ecological transformation. In support of this goal, we map shifts in disturbance regimes and develop a risk index that can be used to identify hotspots for potential climate-induced ecological shifts. Indices such as ours can complement empirical observations of mangrove distributions and recovery failure to support conservation actors in prioritizing areas for monitoring and restoration. Taking a forward-looking perspective, our risk index can inform planning over longer time horizons as climate change occurs. For instance, our results point to the importance of considering future climate risk in the planning of blue carbon projects, especially to accurately quantify and mitigate reversal risks for nature-based blue carbon projects¹⁷. In some cases, it may also unveil untapped potential for blue carbon by identifying areas not at risk of a shifting tropical cyclone regime, but where moderate rates of sea-level rise could allow for high rates of vertical accretion and carbon capture⁴. Considering the higher tropical cyclone frequency and associated severity of potential impacts, however, supporting ecosystem recovery through active restoration post-event may become necessary to support mangrove forest persistence and ecosystem service provision in some areas.

Finally, this study underscores the valuable contributions of the hazard and risk modeling communities in studying emerging disturbance regimes under different levels of climate change and highlights opportunities for interdisciplinary research on the topic. Specifically, we point to investigating further dimensions of disturbance novelty such as potential changes in seasonality or duration of disturbance. The risk framework presented here can be extended to include other climate-driven hazards relevant for mangrove ecosystems, such as drought and marine heatwaves, or applied to other sets of ecosystems and disturbances. Additionally, future research could integrate ecological responses and recovery through vulnerability functions calibrated based on vegetation characteristics and hazard intensity.

Our spatially explicit risk index, which combines climate change-driven changes in tropical cyclones with sea-level rise outlines an emerging novel disturbance regime for mangroves under climate change by the end of the century. Further, our analysis highlights the importance of explicitly including ecosystem services in risk assessments, which are crucial to inform ecosystem-based climate mitigation and adaptation planning.

Tropical cyclone data

We use probabilistic tropical cyclone (TC) tracks from the MIT model^23,24, downscaled from 3 different CMIP6 models of varying climate sensitivity, namely ECEarth6, CESM2, and MIROC6, and under three Shared Socioeconomic Pathways (SSPs), SSP245, SSP370, and SSP585. TC tracks were generated for the historical period (1995–2014) and by the end of the century (2081–2100). The MIT model simulates TCs using three steps: genesis, track, and intensity modeling. TC genesis is the first step, where random seeding is used to initiate warm-core vortices in space and time within a synthetic, time-evolving, large-scale atmosphere-ocean environment derived from global reanalyses or climate models, as in this study. TC frequency is a direct outcome of this random seeding process, as only a fraction of seeds in favorable conditions (e.g., warm sea surface temperatures, low vertical wind shear) survive to develop into TCs, following a natural selection algorithm²³. For each simulated year, 500 synthetic TC tracks are generated, and TC frequency of each individual track ((F_i)) is calculated based on the fraction of initial seeds that develop into TCs under the given boundary conditions. This well-established approach^23,24 has been demonstrated to successfully reproduce key characteristics of the observed historical TC climatology^23,38. Previous research indicates modeling approaches using highly resolved CMIP models such as those used here generally project higher increases in very intense TCs than those with coarser resolutions^38,39,40.

Using the CLIMADA platform⁴¹, we generate two-dimensional wind fields by applying a parametric wind model⁴² to each TC track at 150 arcseconds resolution (150 arcseconds corresponds to 4.63 km at the equator and 3.54 km at 40°). In detail, we compute the gridded 1-min average sustained winds at 10 meters above ground. We divide tropical cyclone windspeed intensities into three categories:

• TC1: 33–49 m s^-1

• TC2: 50–70 m s^-1

• TC3: >70 m s^-1

The expected impacts for these windspeed intensities range from defoliation and occasional stem breakage to extensive canopy loss and finally widespread mortality and peat collapse⁹. For each intensity category (cin) [TC1, TC2, TC3], at each grid location (x), we obtain the average annual frequency (f_x^c) by summing the frequency of all individual events (F_i) whose intensity falls within (c) (see Eq. 1).

where (delta _c_ileft(xright),c) = 1 if (c_ileft(xright)==c), else (delta _c_ileft(xright),c) = 0.

For each point in the wind field, we calculate (f_x^c) for the historical baseline ((f_hist)) and the projections ((f_fut)). Next, we calculate the absolute ((f_fut-f_hist)) and relative change in frequency (((f_fut-f_hist)/f_hist)) at the different windspeed intensities. The findings presented in the main text are based on median values of the three TC track sets based on the global climate models mentioned above. Visualizations of the return periods for the historical reference period and the three SSPs are provided in Supplementary Figs. 2 and 3, while relative changes in TC frequency per climate model can be found in the Supplementary Figs. 4–6.

Sea-level rise projections

As sea-level rise data we use the medium-confidence regional sea-level rise projections from the IPCC AR6²⁵ (https://doi.org/10.5281/zenodo.6382554), for SSP245, SSP370, and SSP585. Rates of sea-level rise are reported for different levels of probability; we use the mean of medians across all IPCC AR6 workflows. We calculate the median rate of regional relative sea-level rise (RSLR) from 2020 until 2100 at each grid point (minimum 1 degree resolution, higher at coastline gauges). The median rates are then grouped into three intensity categories (0–4 mm y^-1, 4–7 mm y^-1, >7 mm y^-1), based on mangrove vertical adjustment capability reported by Saintilan et al.⁴. Vertical deficits are likely at RSLR rates beyond 4 mm y^-1, and highly likely beyond 7 mm y^-14:

• RSLR1: 0–4 mm y^-1

• RSLR2: 4–7 mm y^-1

• RSLR3: >7 mm y^-1

Mangrove data

We use the global mangrove extents polygons by the Global Mangrove Watch (GMW)²². The polygons are reprojected using an equal area projection (Eckert IV) to calculate their area and centroid coordinates. The resulting centroids are then matched with the tropical cyclone and regional relative sea-level rise data using the nearest-neighbor method.

Mangrove priority areas for conservation

Information on mangrove biodiversity and ecosystem services used was generated in a previous study by Dabalà et al. ⁷ (available from https://doi.org/10.5281/zenodo.8272951). Dabalà et al. base their assessment on three different ecosystem services, namely coastal protection for people and assets, carbon storage, and fisheries benefits. The number of people and value of properties protected is based on a study by Menéndez et al. ¹. Carbon storage is estimated based on aboveground biomass⁴³ and soil carbon²⁶. Fishing intensity data is based on zu Ermgassen et al. ¹⁹. Dabalà et al. use a combination of mangrove biodiversity data, which they intersect with the GMW global mangrove extents²², including mangrove species data from the IUCN Red List of Threatened species⁴⁴, mangrove biophysical typology⁴⁵, and marine ecoregions data⁴⁶. Dabalà et al. have generated a variety of optimizations; we use the priority areas optimizing for biodiversity and ecosystem services (equally weighted) regardless of existing protected areas. The priority area planning units are hexagonal polygons at a resolution of 20 km along the coastline based on the GMW mangrove distribution data⁷. We assign the mangrove centroids used in the TC and RSLR risk analysis to the corresponding priority area hexagon using the nearest-neighbor method.

Mangrove TC & RSLR risk index

We compile the three intensity categories for TC and RSLR into a 3 × 3 risk matrix, assigning risk scores ranging from 1 to 5, where higher scores indicate greater risk. (yielding the risk categories low, medium, high, very high, and severe). For each mangrove centroid, we calculate the risk score based on the TC and RSLR category at the given location. Here, we only consider TC windspeeds with return periods of less than 250 years to be of ecological relevance⁶. The TC risk criterion is considered fulfilled if a centroid is newly affected by windspeeds of the given intensity or if the frequency is projected to at least double (((f_fut-f_hist)/f_histge 1)). If this criterion is met across multiple TC intensity categories at the same centroid, the highest possible risk rank is assigned. Since we focus primarily on the shift in TC risk, we do not include areas only affected by sea-level rise in the risk index. For an illustration of how the risk index is constructed, consult Supplementary Fig. 1.

The relative change in frequency threshold beyond which to consider a centroid at risk is a modeling choice, which affects the final risk classification, and has implications for the mangrove area and ecosystem services reported to be at risk. Novelty of a disturbance regime emerges across a spectrum and can be at varying degrees of departure from the historical range of variability¹¹. One could consider any deviation from the baseline frequency novel; however, we take a more conservative approach in setting the threshold at a 100% increase, i.e., doubling, beyond the historical baseline. Please see Supplementary Table 1 for an overview of the share of mangrove centroids above a 50%, 100%, and 200% increase in frequency. We chose the 100% threshold, since a doubling in frequency represents a clear deviation from the baseline, while also maintaining variation in terms of the share of mangrove centroids meeting this criterion across different intensity categories and SSPs.

Ecosystem service risk analysis

The results presented in Table 1 are based on the priority rank assigned to each individual mangrove centroid and the corresponding risk rank per centroid. The matching between risk score and priority unit is achieved by assigning the mangrove centroids containing the risk score information to the priority area planning units using the nearest-neighbor method, as described above. Here, we compare the risk index of the top-ranked 10% and 25% (i.e., 10th and 25th percentile priority rank, which corresponds to a rank of equal to or higher than 2 or 8, respectively) mangrove centroids to the risk index of all mangrove centroids globally. Since Dabalà et al. ⁷ report the absolute values of ecosystem services provided by mangroves per hexagonal planning unit, the proportions of ecosystem services at various levels of risk shown in Fig. 2B are also calculated at the level of the planning units. To obtain the risk index at the scale of the planning units rather than the individual centroids, we calculate an area-weighted mean risk index based on the areal extent and risk index values of all mangrove centroids contained in each planning unit. Despite some loss in the resolution of the risk index, we follow this approach to avoid misrepresenting the total amount of ecosystem services provided in a given area. The area-weighted risk score (R_weighted) per priority area planning unit (P) is calculated by dividing the sum of the products of each risk score (r) of mangrove centroids contained in the given planning unit and its associated area (A) with the total area of the given planning unit (Eq. 2).

This spatial overlap approach assumes a static provision of ecosystem services and does not take into account potential human responses to changes in extreme events or fluctuations in ecosystem service provision. However, it does allow identifying where areas with currently high ecosystem service provision coincide with future risk under climate change.

Non-analogous tropical cyclone frequencies

The mangroves affected by non-analogous tropical cyclone frequencies under SSP245, 370, and 585 (Fig. 4) are calculated as follows: First, we filter for all centroids where the projected future frequency ((f_fut)) exceeds the global historical maximum frequency ((f_max )) within each windspeed intensity category. Next, we calculate the change ratio ((C_nonanalog)) between the projected future frequency ((f_fut)) for the given mangrove centroid and the global historical maximum frequency ((f_max )) per windspeed intensity category: