Multiple climate change-driven tipping points for coastal systems

As the climate evolves over the next century, the interaction of accelerating sea level rise (SLR) and storms, combined with confining development and infrastructure, will place greater stresses on physical, ecological, and human systems along the ocean-land margin. Many of these valued coastal systems could reach “tipping points,” at which hazard exposure substantially increases and threatens the present-day form, function, and viability of communities, infrastructure, and ecosystems. Determining the timing and nature of these tipping points is essential for effective climate adaptation planning. Here we present a multidisciplinary case study from Santa Barbara, California (USA), to identify potential climate change-related tipping points for various coastal systems. This study integrates numerical and statistical models of the climate, ocean water levels, beach and cliff evolution, and two soft sediment ecosystems, sandy beaches and tidal wetlands. We find that tipping points for beaches and wetlands could be reached with just 0.25 m or less of SLR (~ 2050), with > 50% subsequent habitat loss that would degrade overall biodiversity and ecosystem function. In contrast, the largest projected changes in socioeconomic exposure to flooding for five communities in this region are not anticipated until SLR exceeds 0.75 m for daily flooding and 1.5 m for storm-driven flooding (~ 2100 or later). These changes are less acute relative to community totals and do not qualify as tipping points given the adaptive capacity of communities. Nonetheless, the natural and human built systems are interconnected such that the loss of natural system function could negatively impact the quality of life of residents and disrupt the local economy, resulting in indirect socioeconomic impacts long before built infrastructure is directly impacted by flooding.


Study approach.
We performed a multidisciplinary climate vulnerability study by integrating twenty-first century projected changes in climate forcing with the associated response of coastal watersheds, beaches, and tidal wetlands, and the exposure of the urbanized environment to flooding and erosion. A Mediterranean climate exists in the Santa Barbara region, characterized by warm, dry summers, and sporadically wet, cool winters 42 , with atmospheric rivers and quasi-periodic El Niño events strongly influencing local hydrology 43 and coastal hazards 44,45 . This is an optimal location to perform such a study, as projected SLR in this region is substantial 30 and it contains a representative swath of coastal environments along the 95-km study area (Fig. 1), including coastal bluffs and cliffs, dunes, beaches, river mouths, and tidal wetlands, with development ranging from rural to urbanized. Further, the influence of coastal hazards (e.g., erosion and flooding) on the ability of ecosystems to adapt is dependent on the variability and degree of urbanization.
The foundational study was designed to broadly support local government officials tasked with developing plans to adapt to climate-driven threats to natural and human communities 46 . Downscaled twenty-first century climate projections that included rainfall, winds, and SLR 47 drove a coastal hazards model, the Coastal Storm Modeling System (CoSMoS) 48,49 , as well as an analysis of local impacts to watersheds 50 , tidal wetlands 46 , sandy beach ecosystems 51 , and socioeconomic exposure 52 . Each of the research components used a consistent set of atmospheric forcing and SLR scenarios provided by downscaled climate projections.
For this study, potential tipping points were defined by identifying the SLR scenarios at which subsequent projected changes to system metrics (e.g. ecosystem habitat-rich zones, amount of developed land in floodhazard zones) exceed 50% compared to current conditions (i.e. zero SLR and no storm scenarios). We assume that a negative change of at least 50% of the total metric represents a substantial change in state at which point further degradation is likely 53,54 . Further, we assume that projected changes at a specific SLR increment cannot be reversed without human intervention, and therefore a potential tipping point has been reached. We also highlight SLR scenarios where the relative increase in projected change between two SLR scenarios exceeds 50%, less as a potential tipping point of system function and more as a moment at which the system experiences a considerable change that may mobilize communities to implement interventions to reduce future losses. We demonstrate that the timing of the projected changes for total system metrics and between SLR scenarios for the different components of this coastal system are out of phase, which can have implications for natural resource management and climate adaptation planning.

Results
The rate of twenty-first century SLR is dependent on surface temperature increases 29 56 . Moreover, the years 2015-2020 were the 6 warmest in the 140-year instrumental record [56][57][58] . Thus, by 2029 the Paris Agreement target of 1.5 °C will likely be reached under either RCP4.5 or 8.5, and possibly sooner 59 , which is widely considered a critical tipping point for climate impacts to natural and human systems 3 . As recent temperatures are rising at a rate equal or higher than either RCP4.5 or RCP8.5 scenarios, it seems prudent to consider future SLR that accords with the higher RCP8.5 scenario. Downscaled wind, surface pressure and temperature from eight RCP8.5 GCMs were employed to drive short period sea level variability, superimposed upon three long term SLR scenarios (Fig. 2). These scenarios are consistent with recent SLR guidance from the State of California 30,32 . Mid-century SLR projections (2050) for the Santa Barbara region range from a median of 15 cm (low range) to 60 cm (high range), and 45 cm to 170 cm by 2100. An extreme scenario that considers rapid warming coupled with accelerated ice sheet decay projects as much as ~ 3 m of SLR by the end of the century for Santa Barbara 31,32,47 .
Beaches and cliffs provide the first lines of defense for SLR and storm-driven hazards that can result in the more frequent exposure and degradation of in situ and adjacent ecological and human systems. For the Santa Barbara study area, the relative amounts of beach erosion and cliff retreat are projected to increase most significantly beyond 0.50 m (55%) and 0.25 m (95%) of SLR, respectively (Fig. 3A). Coastal erosion rates would be expected to accelerate as the rate of SLR increases under the higher scenarios, but this is not projected for the Santa Barbara region as coastal defenses and other urban infrastructure limit landward retreat in many locations. www.nature.com/scientificreports/ The future width of the ecologically important upper intertidal zone of sandy beaches in the Santa Barbara region, whose landward boundaries included coastal bluffs, dunes, and armoring as well as intensively groomed www.nature.com/scientificreports/  www.nature.com/scientificreports/ beaches, was assessed for SLR scenarios ranging from 0 to 2.0 m in 0.5 m increments (Fig. 3B). Using CoSMoS projections of the mean high water (MHW) shoreline and maximum water levels along the beaches, projected future upper intertidal zone habitat sharply decreased immediately with the transition from 0 to 0.5 m SLR, averaging 74% habitat loss across all beach types, including 87% and 99% for the bluff-backed and armored beaches, respectively, that represent the majority of beaches in the study region, with a resulting loss of ecosystem functions and services 46 . When SLR reaches 1.0 m, < 10% of the upper tidal zone habitat space remains, and when SLR reaches 1.5 m SLR, < 5% remains. Analyzing the loss of the upper intertidal zone when considering SLR combined with annual storms yielded identical tipping point patterns (Supplemental Table S1). With the most severe ecosystem habitat loss observed after the first SLR scenario (i.e., between 0 and 0.5 m), these results indicate that a tipping point for sandy beach ecosystems may have already been reached. Tidal marsh habitat zones in Carpinteria salt marsh were tracked for SLR scenarios based on the present-day relationship between habitat distribution, elevation and flooding frequency (Fig. 3C). Although mid-marsh, high marsh, and transition-upland zones currently account for > 90% of the tidal marsh habitat areas, with just 0.36 m of SLR relative to the marsh surface, those zones are projected to comprise only 46% of the tidal marsh, with the consequent loss of ecological function, including nursery and foraging habitat and biodiversity of rare, threatened and endangered species 46,[60][61][62] . With increasing SLR, the mudflats and sub-tidal habitats become dominant, including a more than tripling of the lower mud flat area when transitioning from 0.18 to 0.36 m of SLR, and the high mudflat zone between 0.36 and 0.61 m of SLR. During the SLR transition from 0.18 to 0.61 m, areas of the mid-marsh, high marsh, transition, and upland zones decrease most acutely by 54%. Upon reaching 1 m of SLR, mud flats and sub-tidal regions are projected to account for 85% of the tidal marsh, and 97% at 2 m. In short, by ~ 0.6 m of SLR, consistent with sandy beach ecosystems by 0.5 m of SLR, vegetated tidal marsh habitat will have already severely degraded, with the largest change occurring after 0.18 m of SLR. Since the habitat changes are relative to the marsh surface, the timing of the tipping point depends on the rate of accretion and SLR scenario. Assuming a marsh accretion rate of 4 mm/year 63 and the high range scenario of SLR, mudflat would comprise 56% of habitat by 2050 and > 80% by the end of the twenty-first century 46 , representing a major shift in habitat type.
The potential for socioeconomic tipping points was based on projected changes in the number of residents and employees and the amount of developed land and parcel values in areas that are estimated to have daily/ permanent coastal flooding (i.e. inundation) and storm-driven flooding (i.e. annual and 100-year return interval events) relative to CoSMoS SLR scenarios (0 to 2.0 m). Changes in flood-hazard exposure were estimated for five communities near the sandy beach and tidal wetland ecosystem study sites, including Goleta, Isla Vista, Santa Barbara, Montecito, and Carpinteria (Fig. 1). Exposure to daily flooding (i.e. no storms considered) for each societal metric relative to study area totals is far below the 50% tipping point threshold, instead only directly affecting 2.4% of total parcel values, 2.7% of total residents, 2.9% of total employees, and 7.4% of total developed land even for the largest SLR scenario of 2.0 m (Fig. 3D-G). Although total system exposure is not projected to surpass the 50% threshold observed in the ecological systems, there are certain SLR scenarios that could result in considerable, relative increases between SLR scenarios that may galvanize communities to mitigate further losses. For example, the largest relative increases of societal exposure to inundation occurs across all four metrics when transitioning from 0.75 m to 1.0 m of SLR ( Fig. 3D-G, Supplemental Table S1), with an increase of 1,712 residents (175% increase), 972 employees (254% increase), $283 million dollars of parcel value (160%) and 1.5 km 2 of developed land (144% increase). Peak increases in coastal flood hazard exposure associated with SLR inundation are preceded by the largest increases in the relative amounts of beach erosion and cliff retreat (Fig. 3A).
Even without considering future SLR, the Santa Barbara region is already at risk from coastal storm, wavedriven flooding hazards ( Fig. 3D-G). The five communities in our study area collectively have 2,302 residents and $345 M in property value in projected flood areas for coastal storms with an average annual return interval, increasing to 3,232 residents and $581 M in property values for storms with a 100-year return interval. These storm-related exposure values rise considerably when taking into account the role of SLR in increasing potential flood areas. Nevertheless, similar to the non-storm scenarios, total system exposure is not expected to surpass the 50% threshold for any societal metric, instead only directly affecting 3-7% of residents and parcel values, 3-8% of employees, and 8-13% of developed land for the largest SLR scenario of 2.0 m (range noting annual and 100-year storm estimates). There are, however, considerable relative increases in flood hazard exposure from 100year storms when transitioning from 1.5 to 2.0 m of SLR ( Fig. 3D-G, Supplemental Table S1), with an increase of 6,542 residents (130% increase), 6,532 employees (171% increase), $2B of parcel value (156%) and 2.6 km 2 of developed land (66% increase). This analysis of the physical and societal hazards related to SLR demonstrates that the largest changes in exposure for the different styles of projected future flooding are asynchronous (e.g., when transitioning from 0.75-1.0 m of SLR for daily flooding/inundation compared to 1.5-2.0 m of SLR for storm-driven flooding). Further, when exposure changes are calculated uniformly by comparing scenarios based on jurisdictional and habit area totals and not relative changes, the human system values do not occur until much higher SLR scenarios, are far smaller, and do not qualify as tipping points.

Discussion
The results described herein indicate that degradation of coastal ecosystems in the Santa Barbara region is on the leading edge of climate impacts, consistent with observations world-wide 5-8,10-12 . Sandy beach ecosystems have possibly reached a tipping point, and tidal marshes will reach that point perhaps as early as ~ 0.25 m of SLR. This timing (~ mid-century or sooner) is similar to the temperature tipping point of the terrestrial biosphere projected to be within just a few decades 64 , although the precise timing of this transition is difficult to determine due to uncertainty related to future climate warming, response of SLR to warming, SLR scenario resolution, and future sediment supply to beaches and marshes. Greenhouse gas emissions are currently tracking the RCP8. 5  www.nature.com/scientificreports/ most closely 59 , so warming is likely to rapidly exceed 1.5 °C and therefore SLR could reach 0.25 m by mid-century under either a middle or high SLR scenario, and just a few decades later for the low SLR scenario (Fig. 2). Even if greenhouse gas emissions are reduced to a net zero level, the current concentration of C0 2 in the atmosphere has already committed oceans to an additional ~ 1.7 m of global mean SLR 65 , which would match the higher end SLR scenario projected for Santa Barbara by 2100. Given that 250 million people currently live within 1 m of present-day high tide across the world 66 , a SLR tipping point may have already been reached globally and locally. Sandy beach ecosystems are especially vulnerable and appear on the verge or have already exceeded a tipping point, especially for armored beaches (e.g. refs. 67,68), where 99% of the existing upper beach zone habitat is projected to be lost with just 0.5 m of SLR (range of all beach types = 51-99%, mean = 74%). The majority of beaches are projected to decline in overall width with increasing SLR. Importantly, the loss of beach width will not be evenly distributed across intertidal zones. Upper beach zones are projected to experience the greatest declines in width and losses with SLR. Although often narrow in width, these upper intertidal zones are vital components of biodiversity and ecosystem function 51,69 . The sandy upper intertidal zone is associated with the distributions of key beach organisms, biodiversity and ecosystem functions, including the accumulation of macrophyte wrack and the wrack-associated invertebrate community 67,69 . This often narrow zone supports ~ 45% of total intertidal invertebrate biodiversity, provides prey resources for birds and fish, and plays a vital role in detrital processing and nutrient cycling 51,68-73 . These key upper beach zones are already scarce and/or ephemeral for many beaches in the study region with consequent loss of biodiversity and ecosystem function (e.g. refs. 67,74,75 ). When seawater regularly reaches the bluff toe, armoring structure or beach limit, drift macrophyte wrack and the rich intertidal biodiversity, ecosystem functions and prey resources that macrophyte supports, as well as critical habitat for fish and wildlife are eliminated from the beach ecosystem. The upper intertidal transition, the zone where native coastal strand vegetation develops that can enhance resilience by trapping and storing sand and building dune topography 76 , also is eliminated.
While continued cliff retreat might create more space for sandy beach habitats, SLR-driven erosion will remove the sand in front of cliffs. Across Southern California, one-to two-thirds of beaches, and the habitats therein, are expected be drowned and lost due to SLR this century, the loss accelerated by curbs on landward migration due to cliffs and/or urban resistance 77 . Existing coastal armoring already restricts the migration potential of up to 57% of Southern California county beaches, including 12% in Santa Barbara County 78 , and impacts intertidal biodiversity and function 67,68 . There has been a greater than five-fold increase in armoring over the last 50 years across this region, and as the coastal hazards associated with rising seas increase in this urbanized setting, as in others worldwide, more structures are likely to be constructed that protect human populations but limit habitat migration.
Urbanization is a major factor in the observed habitat squeeze of sandy beach and tidal marsh ecosystems and related short-term tipping points as engineered tidal marsh shorelines, armored beaches, as well as bluffs prevent the upland transgression of tidal marshes and landward migration of sandy beaches (Fig. 4). In the absence of urbanization, these coastal habitats are resilient and capable of responding to high rates of SLR, as evidenced by their ability to survive during the > 100 m of SLR of the late Quaternary, including several centuries during Meltwater Pulse 1A where rates approximated 5 cm/year 79 .
However, tidal marshes have more recently been in decline across the coast of California due to urbanization and land management practices, with a 48% loss of habitat across southern California since 1850, including a 62% decline in Santa Barbara County 80 . Accelerating sea levels world-wide and limits to landward mobility will further increase tidal inundation of marshes leading to changes in key physical and biological properties known to structure marsh plant communities and habitats 81 . As demonstrated in this study, major changes are forthcoming or already underway in Carpinteria salt marsh, and this regionally scarce ecosystem may pass a tipping point with less than 0.25 m of SLR with the significant loss of high salt-marsh and transition habitats and the functions these habitats provide. Similar losses of existing vegetated marsh by the end of the century have been projected for other tidal marshes region-wide 82 and are consistent with global studies linking early warning signals to marsh collapse 83 . The precise timing of habitat evolution will depend on the rate of SLR and the accretion rate of the marsh surface, but ultimately, with limited landward accommodation space, this vegetated marsh will be converted to mudflat. While specific tipping point dates are tidal marsh-specific and depend on local hydrology, sediment supply, and topography, urbanized tidal marshes are the norm along the California coast and other populated settings. This study, therefore, may serve as a proxy for the response of similar tidal marsh systems world-wide. Mitigating and adapting to the anticipated vulnerability of this ecosystem to climate-related impacts are a key planning and management priority for local, state, and federal agencies [84][85][86] .
If the sole trigger for communities across the region to implement adaptation options is based on substantial increases in the physical exposure of developed land and built systems to SLR and storms, then communities might delay adaptation strategies until after 2050 as the largest absolute and relative changes in exposure are not projected until SLR exceeds 0.75 m for daily flooding (i.e. inundation) and 1.50 m for storm-driven flooding, and therefore not projected to occur until toward the end of the twenty-first century or later. Further, when looking at socioeconomic exposure, the population, employees, property values, and developed land directly exposed to flooding in the study area jurisdictions represents just 2-13% of totals even under the most extreme SLR scenario considered here (i.e. 2 m), and the largest changes between scenarios are all 5% or less. However, societal exposure to other coastal hazards, such as tsunamis 87 and groundwater hazards 88 , will also increase due to SLR but were not considered here.
Although projected changes in flood-hazard exposure may not represent high percentages of the total number or amount of residents, employees, property values and developed land in the communities, these values do not consider the interconnection of the human-natural system. For instance, the study region is highly dependent on tourism, driven largely by visitors enjoying the beaches and wetlands along the coast. For example, an analysis of visitor profiles in the study area identified several activities that relate to these coastal systems, including www.nature.com/scientificreports/ going to the beach (52-68% of visitors, depending on type of visitor), going to parks (19-25%), and water-based recreational activities (2-8%). This report also estimates that the region had an average of 28,884 daily visitors, which supported 13,482 jobs and resulted in $1.9 billion in annual revenue 89 . While there is not a direct 1:1 relationship between beach loss and loss of tourism, it has been shown to lead to economic impacts in other regions and is something for which coastal communities could prepare. In addition, valuable public infrastructure and public services, such as the Santa Barbara airport, the railroad, Highway 101, the El Estero Water Resource Center, along with stormwater drains, sewage pump stations, and harbors are located in these coastal strips; even intermittent coastal storm-related flooding can cause severe interruptions of these systems, resulting in further economic impacts that can compound over time. Understanding the timing of different tipping points (or at least the point when the largest changes in exposure are projected to occur), including losses to natural systems, upon which economic drivers such as coastal tourism are dependent, could therefore be important for robust adaptation planning 15 . Admittedly, assuming an irreversible change in our tipping point definition does not consider the adaptive capacity of each system to respond to SLR. Adaptive capacity may be limited when beach and tidal marsh habitat www.nature.com/scientificreports/ areas are squeezed between rising seas and hardened urban landscapes. In such cases, endemic plant and animal populations dependent on those habitats may be completely eliminated. Other biota (e.g. deeper water fish assemblages) might use the flooded habitat, however, overall ecological composition, biodiversity, and functions will change. Conversely, while the socioeconomic impacts of more frequent coastal flooding may be severe, human populations have the ability to migrate to safer, inland settings, although social inequities can play a major role in migration potential for underserved communities. The complexity of adaptive capacity, human response, and societal value placed on different coastal systems makes a universal definition for a community-scale tipping point difficult to quantify in a way that directly supports management action across the board. What is clear, however, is that regardless of the chosen metric, individual coastal systems will continue to evolve at different rates and reach critical thresholds at different times due to climate change, prompting the need to adopt a multitiered, multi-disciplinary approach to address the range of physical, biological, and human impacts associated with climate change in coastal systems. The most substantial changes and/or tipping points identified here are based on projections of future coastal hazard exposure and ecosystem response assuming no interventions, i.e., they are not inevitable. Measures that could counter local SLR and storm impacts include ecosystem restoration, removing barriers to inland transgression 90 , increasing sediment supply (e.g. dam removal), and removing shoreline armoring and reducing mechanized beach grooming 76 . Ultimately, the pace of management action will depend upon societal values and the availability of community resources (e.g. financial resources and political will) to mitigate potential losses, absorb losses, and/or implement adaptation measures within the coastal zone. However, if communities only look at singular triggers to make adaptation planning decisions -either for natural (e.g. limited marsh conversion) or human systems (e.g. number of businesses impacted) -they are missing the interconnectedness of natural-human systems. The full range of vulnerability, and their associated tipping points, must be analyzed in tandem. If the prevention of significant ecosystem changes is a priority, then swift action is essential as those tipping points are imminent. Similarly, beyond the intrinsic value of having thriving coastal ecosystems, like many coastal towns, a healthy coastal zone is also critical to this region's economy.
In summary, defining a tipping point that universally captures the point of significant degradation across physical, biological, and human systems is challenging. Cascading interactions between SLR, coastal hazards, and human response amplify impacts but are complex and difficult to quantify; humans could choose to adapt and stay in place or move away from a hazardous region. Moreover, in a tourism-driven economy such as Santa Barbara's, focusing only on flooding to infrastructure and property, and not preparing for the impacts to the coastal ecology and beach loss, ignores an important socio-natural interconnection on which the region's economy depends; natural system degradation will likely negatively impact coastal tourism, which in turn then makes local businesses and communities vulnerable well before direct flood exposure to physical assets. Nevertheless, to better inform resource decisions, this study identifies climate-adaptation planning that considers tipping points for multiple components of a coastal system, including for natural ecosystems, as opposed to the more common singular focus on human components.

Materials and methods
Climate modeling. Ten global climate models (GCMs) were selected from the Coupled Model Intercomparison Project Phase 5 (CMIP5) archive 55,91 based on their realism in representing California's historical climate 47 . The Local Constructed Analogs (LOCA) statistical technique was used to downscale each GCM to 6 km resolution for daily temperature and precipitation from 1950-2100 for Representative Concentration Pathway (RCP) scenarios RCP4.5 and RCP8.5 (ref. 92 ).
Sea level rise projections for the 21st Century were derived from the National Research Council 85 and integrated with short period fluctuations due to tides, meteorological conditions, and short period climate variability (e.g. El Niño) to produce hourly coastal water levels for the study area. Model inputs for the multiple-linear regression model were based on water level observations at Santa Barbara Harbor and historical NCEP meteorological reanalysis data, with the variables including daily climate model data, surface pressure, wind stress, and both local sea surface temperature and central Pacific Ocean sea surface temperature to assess El Niño variability 93 . Coastal hazards. Driven by the wind and pressure fields from the native resolution and downscaled GFDL-ESM2M (RCP4.5 scenario) GCM from above, the Coastal Storm Modeling System (CoSMoS) 48,49,94,95 was applied to the study area by considering 40 SLR (i.e. 0, 0.25, 0.50, 0.75, 1.00, 1.25, 1.50, 1.75, 2.00 and 5.00 m) and storm scenarios (i.e. daily conditions and annual, 20-year and 100-year storms). Twenty-first century wind fields were fed into a global and nested Eastern North Pacific WAVEWATCH III wave model 96 to establish wave conditions at the continental shelf edge 97 , and then dynamically downscaled using SWAN 98 to transform the waves to the nearshore. To establish nearshore water level boundary conditions, SWAN was coupled with DELFT3D-FLOW to capture the tides, seasonal water level anomalies, river discharge and storm surge 95,99,100 . High-resolution hydrodynamic grids (O ~ 10 m) were used to model total water levels and overland flooding for complex shorelines, including protected embayments, while cross-shore, 1-D XBeach models 101 were spaced every 100 m alongshore on the open coast to predict wave set-up and swash. Projected flood levels for each scenario were interpolated onto a 2-m grid and differenced from a Digital Elevation Model (DEM) 102 to provide the flood extent and depth. Storm scenarios for the full computational application of CoSMoS were established via a proxy approach 103 , which also served to provide a twenty-first century, continuous time series of total water levels to drive the coastal change models for sandy beaches (CoSMoS-COAST) 77 and cliffs 104 along the established XBeach transects. The long-term coastal change projections were used to evolve the DEM for the future flooding scenarios 105 . All the model projections are freely available for viewing 106  www.nature.com/scientificreports/ Sandy beach ecosystems. Standard elevational metrics were related with ecological components and habitat zones of beaches to identify the vulnerability of sandy beach ecosystems to SLR. The current and future state of the ecologically critical, upper tidal zone of each beach, which ranged from bluff-backed, dune-backed, armored, and groomed, was measured and modeled using the total water level datum 108,109 as a proxy for the dynamic landward extent of the upper intertidal zone, equivalent to the daily High Strand line (HTS) 51 . Ecological research on area beaches 51,67,69-71 combined with beach surveys and coastal processes studies enabled a predictive framework to be established between SLR and changes in the upper beach zones 84,110 . The total water level projections from CoSMoS 48,49,95 for 0.50, 1.00, 1.50, 2.00 and 5.00 m of SLR combined with background conditions (i.e. daily) and annual storm scenarios were used to establish the landward extent of the upper beach zone. The seaward limit of the upper beach zone was determined by the projected MHW elevation, as determined by shoreline modeling driven by a CoSMoS-generated twenty-first century water level time series 77 . Upper beach zone landward migration and width was restricted by the presence of non-erodible structures, such as revetments, sea walls, roads or parking lots.
Tidal wetlands. This study examined the wetland evolution of Carpinteria salt marsh, a 93 ha tidal wetland ( Fig. 1). This wetland represents an urbanized marsh system, surrounded by urban and residential development that restricts potential upland migration, and an engineered inlet that maintains the tidal connection to the ocean. The system contains a wide range of marsh species and habitats, including salt tolerant pickleweed (Salicornia pacifica) dominating the regularly flooded middle tidal marsh, and various succulent, grass, and perennial and annual herb species that occur in the high marsh and upland transition zones, including rare and endangered species 46,62 . Existing habitats were delineated using multispectral aerial imagery as open water subtidal, high and low mudflat, coastal salt marsh, (mid and high), transition, and undeveloped upland. Using Santa Barbara Harbor tide data from 2006-2014 (ref. 111 ) and elevation surveys from both aerial Lidar and in situ Real Time Kinematic Global Positioning System (RTK GPS), habitat was linked to elevation and flooding frequency 46,62 . This relationship was then applied to assess future marsh habitat evolution using SLR ranging from 0 to 2.5 m. The potential timing of habitat evolution was evaluated assuming a vertical accretion rate of 4 mm/year 63