Marsh migration and beyond: A scalable framework to assess tidal wetland resilience and support strategic management

Tidal wetlands are critical but highly threatened ecosystems that provide vital services. Efficient stewardship of tidal wetlands requires robust comparative assessments of different marshes to understand their resilience to stressors, particularly in the face of relative sea level rise. Existing assessment frameworks aim to address tidal marsh resilience, but many are either too localized or too general, and few directly translate resilience evaluations to recommendations for management strategies. In response to the deficiencies in existing frameworks, we identified a set of metrics that influence overall marsh resilience that can be assessed at any spatial scale. We then developed a new comprehensive assessment framework to rank relative marsh resilience using these metrics, which are nested within three categories. We represent resilience as the sum of results across the three metric categories: current condition, adaptive capacity, and vulnerability. Users of this framework can add scores from each category to generate a total resilience score to compare across marshes or take the score from each category and refer to recommended management actions we developed based on expert elicitation for each combination of category results. We then applied the framework across the contiguous United States using publicly available data, and summarized results at multiple spatial scales, from regions to coastal states to National Estuarine Research Reserves to finer scale marsh units, to demonstrate the framework’s value across these scales. Our national analysis allowed for comparison of tidal marsh resilience across geographies, which is valuable for determining where to prioritize management actions for desired future marsh conditions. In combination, the assessment framework and recommended management actions function as a broadly applicable decision-support tool that will enable resource managers to evaluate tidal marshes and select appropriate strategies for conservation, restoration, and other stewardship goals.


Introduction
Tidal marshes are critical components of coastal ecosystems.Wetlands protect the physical characteristics of coasts by buffering energy from storms and waves [1] while also providing habitat for nekton that support a broad suite of fish and wildlife, including wetland birds and economically significant fisheries species [2,3].Increasingly, these essential coastal features are under threat.Tidal marshes are facing anthropogenic stressors such as coastal squeeze, which is the reduction of habitat when sea level rise causes the shoreward boundary of a marsh to move inland while anthropogenic structures prevent equivalent migration of the landward edge, as human populations expand along coasts [4][5][6].They also face impacts from climate change, including higher storm intensity and storm surge that cause increased flooding which may erode marsh borders.While increased sediment deposition during storms can offset erosion in some cases, in many places sedimentation rates may not keep up with the pace of relative sea level rise (RSLR) [7,8].Additionally, accelerated RSLR has the potential to reduce available habitat [9], change the composition of associated plant and animal communities [10,11], and reduce the ability of tidal marshes to act as carbon sinks [12].
Strategic planning for the future of tidal marshes is critical in the face of RSLR.Wetlands have been managed for human activities for centuries, most commonly diked and drained to support farming or impounded for waterfowl hunting [13,14].Contemporary marsh conservation and restoration strategies more often focus on goals such as maintaining habitat values for the benefit of ecologically and economically significant species [15], protecting coastal infrastructure, mitigating storm damage [16], and sequestering greenhouse gas emissions by fostering conditions that enable wetlands to act as a carbon sink [17].Many options exist to achieve these goals, and the strategy for different areas depends on the ultimate desired outcome.Managers may choose to prioritize protection of existing marshes through conservation efforts [13], or seek to improve the quality of existing wetlands through restoration or enhancement [18][19][20], or create new marshes where altered shorelines enable space for additional marsh habitats [21].
In order to conduct effective tidal marsh strategic planning in the face of accelerating RSLR, resource managers need a standardized approach to understand how particular marshes will respond to rapidly rising sea levels [22].Not all marshes will respond equally to increasing RSLR; the current condition of tidal wetlands determines their vulnerability to outcomes such as excessive submergence when rising sea levels overtake a marsh's accretion rate [23,24].Moreover, marshes vary in the challenges they face, in addition to or associated with RSLR, across local, regional, and national scales [25,26].For example, additional challenges may include nutrient loading, invasive species, or physical barriers like dams and dikes.
In addition to considering marsh vulnerability, management approaches to these challenges will vary depending on management goals.Some coastal managers may prioritize protection of particular plant or animal species [27,28].Others may determine that a particular tidal wetland is too degraded or faces too many challenges to be worth saving [29].The cost to implement these strategies can differ considerably, so managers responsible for deciding where to allocate funds need science-based prioritization via a standardized assessment tool to evaluate and compare the resilience of marsh systems.This tool should be designed to provide management recommendations based on available data, and the outcomes of the marsh assessment should be linked to management actions.
A variety of resources are available to inform tidal marsh management decisions, yet no previously existing tool addresses all assessment needs.For example, the marsh resilience to sea level rise (MARS) multi-metric index focuses on future response by marshes to a single stressor, sea level rise, and does not evaluate either current condition or adaptation potential [30].Similarly, the Unvegetated to Vegetated Ratio (UVVR) is a vulnerability metric developed as an indicator of sediment budget and vertical growth potential to predict future marsh condition, but does not account for availability or extent of migration pathways [31].Several sophisticated marsh modeling techniques have been developed (e.g.MEM, WARMER, SLAMM) that are not always accessible for managers to employ on a local scale due to complexity or absence of required data inputs [32][33][34][35].Conversely, many other studies are focused on a local or regional scale with metrics that are not applicable across geographic areas [36][37][38].Many evaluation tools focus on marsh assessment and fail to provide recommended approaches to managers [15,37,[39][40][41].
Our goal was to create an assessment framework that can be applied to tidal marshes at any spatial scale, from local to national or international, thereby enabling an apples-to-apples approach to comparing relative resilience of different marshes to RSLR in order to inform management decisions.We created an assessment framework wherein total marsh resilience is represented as the sum of scores across three metric categories: current condition, vulnerability, and adaptive capacity, and scores for any marsh within a study area represent resilience relative to other marshes included in the same analysis.We then linked marsh resilience scores directly to suggested actions to enable resource managers to target restoration and conservation projects strategically in key areas where project success and ecological impact will be maximized.Our new approach can thus be applied as a spatial planning tool at small or large geographic scales.
After developing the new framework, we applied it to 1,984 individual marsh units across the entire coastal contiguous United States (US).We included each of the 25 National Estuarine Research Reserves (NERRs) within this area, which collect data consistently and serve as an ideal platform for geographic syntheses [30,[42][43][44].In addition, we summarize results for 8 hydrologic units within a single tidal marsh complex on the US Gulf Coast, where states in the southern US border the Gulf of Mexico, to examine smaller-scale patterns and associated recommendations for management.Overall, this framework functions as a decision-support tool that links relative resilience results to a variety of management options and enables cost-effective strategic planning decisions.

Methods
Our objective was to create a tool that can be used by resource managers and coastal decisionmakers to determine the best management options for tidal marshes at any spatial scale.We start by describing how the framework was developed, and then detail its scoring components and how the resilience scores link to management.Finally, to demonstrate use of this tool, we applied it at multiple geographic scales with publicly available national datasets and report on results within the contiguous US from regions, states, NERRs, and the Gulf Islands National Seashore, which includes sections of the Mississippi and Florida coastlines.

Development of the framework
We considered site resilience using the definition adopted by the Nature Conservancy in their North Atlantic Landscape Conservation Cooperative resilience assessment model: "capacity of a physical site to maintain species diversity and ecological function even as the composition and proportion of habitats change in response to climate change" [38].Guided by this definition, we then generated a comprehensive set of resilience metrics through a literature review and an expert elicitation process via a stakeholder survey completed by 34 staff members of the NERR system.During this process, we chose to focus on three critical components of resilience: current condition, adaptive capacity, and vulnerability.Available options for strategic spatial planning vary between categories, so we then solicited expert opinions on the most useful resilience metrics separately under each category from NERR staff during a session at the 2018 National Estuarine Research Reserve Association annual meeting.Next, we simplified the outcomes of the meeting to apply to all geographies and reviewed results with key members of the NERRs.We ultimately selected metrics that can be assessed in the US and internationally using publicly available data for vegetated tidal habitats, excluding mangrove systems.The list of metrics we provide can be expanded by users of this framework in areas where additional data are available.We intend the set of metrics within the framework's three resilience categories to be flexible to the needs and resources of users.The resilience metrics outlined below are to be measured at the finest scale of interest and then aggregated at the marsh unit scale, defined as the smallest scale at which data for all metrics are available.

Rationale behind categories and metrics
Current condition.The current condition of a marsh represents its relative starting point for ecosystem function before taking into account the effects of future RSLR (Table 1).The metrics selected for this category are based on measures of a marsh's area compared to the length of its edge, the proportion of vegetated to unvegetated edge, and the proportion of surrounding impervious, agricultural, and natural land cover.Marsh configuration affects erosion exposure and geomorphology because a greater proportion of edges to core area, especially unvegetated edges, increase exposure to erosion [31,42,[45][46][47].Land cover and use is measured at the marsh unit scale within a small buffer to account for upland drivers of potential

Marsh Configuration
Area to edge ratio + more core and less edge represents less exposure to erosion through wave action [46,55] Unvegetated edge ratio -more unvegetated edge represents more risk to tidal currents and RSLR [31,47,56] Land Cover/Use in watershed % impervious -more impervious can negatively impact runoff/habitat [48,49] % natural + more natural areas can absorb runoff/protect habitat [52][53][54] % agricultural -more agriculture can negatively impact runoff/habitat [50,51] Vulnerability Erosion Soil erodibility -high erosion potential represents greater risk of loss of sediment from marsh [45,57] Tides Tidal range + a larger tidal range represents more resilience to RSLR, since plant elevational distribution is broader [30,32,37,58,59] % marsh below MHHW -more marsh below MHHW represents more risk of converting high marsh to low marsh with RSLR [24,60] % marsh below MTL -more marsh below MTL represents more area at risk of excessive inundation in the face of RSLR [45,56,61] Adaptive Capacity Shoreline Hardened shoreline -more hardened shoreline represents more impediments to migration and decreased sediment supply [4,62,63] Shoreline complexity + the more complex the shoreline the more opportunity for diverse habitats to support greater biodiversity [38,64,65] Migration Ability Marsh migration space + more migration space in low-lying uplands adjacent to marsh represents a larger potential future area for marsh to move into [9,35,36,66] Wetland "connectedness" + marshes that are more connected to each other in the future are able to support species migrations [67] Metrics are hierarchically nested within groups and categories.The ± column indicates whether a certain metric has a positive or negative influence on tidal marsh resilience to relative sea level rise (RSLR).MHHW refers to Mean Higher High Water, and MTL refers to Mean Tide Level. https://doi.org/10.1371/journal.pone.0293177.t001 environmental stressors like runoff and pollution.Impervious and agricultural cover contribute runoff and pollutants that degrade marshes [48][49][50][51], while increased natural cover mitigates those effects [52][53][54].
Vulnerability to accelerating RSLR.These metrics evaluate how current marsh area will respond to increased stress from rapid RSLR (Table 1).Higher vulnerability is associated with lower overall resilience, and these metrics are calculated based on erodibility and elevation relative to tides.Marshes with higher erosion potential are less likely to keep up with RSLR, unless sediment inputs exceed erosion rates [45,57].Soil erosion could also increase with more frequent inundation [68].Tidal metrics indicate how rapidly marshes will be affected by rising water levels [32], as marshes are more vulnerable to sea level rise when more of their vegetation is lower in the tidal frame, especially when that frame is not large [37,58,59].A higher percentage of marsh below the Mean Tide Line (MTL) means more marsh consists of mudflat and higher percentage below the elevation of the highest tides translates to more low marsh elevation and associated vegetation [24].Areas at risk of conversion from mud flat to open water, or low marsh to mud flat, in the face of accelerating RSLR may face marsh loss due to excessive submergence [45] or a loss of area to support low marsh plants that are limited by inundation and salinity ranges [60,61].RSLR comes from a combination of changes in the eustatic sea level driven by global climate change and changes in local isostatic sea level rise resulting from land subsidence in some areas, for instance from diking or groundwater overdraft [69].Whether absolute water levels have risen, or the land has dropped, the two tidal marsh metrics account for more marsh area below the high tide line in the future.Metrics related to the elevational distribution of vegetation across a marsh [30] have been identified as robust indicators of observed marsh persistence [56].
Adaptive capacity.Exposure to wave energy, erosion, and inundation stress does not directly determine future condition, as tidal marshes may be able to expand laterally in response to RSLR.Adaptive capacity metrics address this point, as marshes with barriers to migration have lower resilience to RSLR [35].These metrics are associated with shoreline features to indicate how easily the borders of a marsh can expand geographically and the ability for marshes to continue supporting ecological function (Table 1).The more sinuous, or complex, a shoreline, the more protection it affords a marsh [65] and more habitat niches exist to support a greater diversity of plants and animals [38,64].Hardened shorelines-which include rocky shores and built structures-impede migration [4,62,63].The final metric group within this category, migration ability, addresses differences in land form and elevation that may help or hinder marsh area expansion [9,36,66].The more connected a landscape is, the more capacity it has to support species that re-distribute in response to changing environmental conditions [67].

Resilience scoring and links to management
After identifying the fundamental metrics and categories that contribute to tidal marsh resilience, we created a scoring system to sum results across categories.To generate scores, users of this framework first gather data associated with each metric for each marsh unit within an explicit spatial scale.Marsh units should be set by framework users based on the resolution of available data.Users then calculate decile breaks based on the spread of data for marshes within the chosen sample, which can be applied as cut points to create 10 equal bins within the raw data for each metric.Users then assign a rank score to each marsh unit relative to other marshes in the sample for each metric based on the corresponding bin (1-10) that contains the marsh units' data value for a given metric.Next, users multiply metrics with a negative impact (as illustrated in Table 1) by negative 1, creating final scores for each metric.
These scores represent the status of each marsh unit relative to the other marsh units assessed for each metric.
Users can add the individual metric scores together to create category resilience scores.All metrics are given equal weight when added, except hardened shoreline is weighted 1.5 times, and migration space is weighted 2 times to reflect the high impact of these metrics on tidal marsh resilience.Users should then calculate decile breaks by category for the summed metric scores and create 10 equal bins for each category based on the breaks.Users can assign rank scores to each marsh unit based on the corresponding bin (1-10) that contains the summed value for each marsh unit in each category.Users should assign a negative value to the category resilience scores for vulnerability to represent the negative impact of vulnerability on overall resilience.These scores represent the status of each marsh unit relative to the other marsh units assessed in each category.
Total resilience scores convert category resilience scores into a single metric that enables comparisons between marshes.To calculate total scores, users sum the three final scores for current condition, vulnerability, and adaptive capacity.Next, users calculate decile breaks for the metric scores summed across the categories and create 10 equal bins based on the breaks.Users then assign a rank to each marsh unit based on the corresponding bin (1-10) containing the summed value.These scores represent the overall resilience to RSLR of each marsh unit relative to the other marsh units assessed in a given analysis.
To inform strategic planning, users can link category resilience scores to a series of management options.Users assign results for each marsh unit an ordinal category, either "low" or "high" based on their value.Category resilience scores less than 6 (greater than -6 for scores in the vulnerability category) are "low" while scores greater than or equal to 6 (less than or equal to -6 for scores in the vulnerability category) are "high".All three categories are considered together to recommend best conservation or restoration actions, so users can generate a total management category using a concatenation of ordinal results within the three resilience categories.Finally, users are able to associate each total management category with a suite of management strategies that were developed during stakeholder meetings regarding restoration, conservation, and land use by policy professionals in New Hampshire (Fig 1).We simplified highly detailed locally relevant language from these meetings to be broadly applicable across a wide range of tidal marsh systems, then reviewed the final list of management options with key members of the NERR research, stewardship, and geographic information systems community.

Application of the framework to US marshes
We calculated a total resilience score for each subwatershed hydrologic unit in the coastal contiguous US, and then summarized results across geographies to demonstrate the assessment framework's value in comparing marshes at various spatial scales.We selected subwatersheds as our marsh unit for this analysis because that is the finest scale hydraulic unit in the Watershed Boundary Dataset, a data source from the US Geological Survey that provides polygon boundaries for summarizing surface water drainages within the entire US [70].In the dataset, subwatershed hydraulic units are represented as 12 digit codes (HUC-12s), and are typically between 40 to 162 square kilometers [71].We adopted these boundaries for our marsh units and used HUC-12 codes as Marsh Unit Codes (MUCs) for unique marsh unit identifiers.We provide additional details for metric calculation within each MUC in the Supporting Information (S1 File).
We compiled spatial data for current condition metrics from the National Oceanic and Atmospheric Administration (NOAA) Coastal Change Analysis Program (C-CAP) 2010 Best management options to maximize resilience to relative sea level rise for each marsh type.Green text indicates a positive condition, red reflects a negative one.From a restoration or adaptation practitioner's perspective the more resilience categories that are shown in red for each marsh type, the more costly a project will likely be to implement.https://doi.org/10.1371/journal.pone.0293177.g001 Regional Land Cover Data for the coastal United States [72].We used the C-CAP 30 meter land cover data to identify estuarine wetland classes within MUCs and created a classification scheme to characterize core wetlands and edges.Using these data, we calculated a core-to-edge ratio [46], and a unvegetated to vegetated edge ratio inspired by the UVVR metric in development by [31] at the time of this analysis, for each 30 meter segment and then aggregated results for each MUC.We also used C-CAP to summarize development intensity for land cover metrics of percent impervious cover, natural cover, and agricultural cover with a 150 m buffer on each MUC.We selected this buffer size to match the land cover buffer used for tidal marsh monitoring in US National Wildlife Refuges [73].
We calculated soil erodibility using Esri's USA SSURGO-Erodibility Factor imagery layer, which is 30 meter resolution, [74] and then aggregated and summarized by MUC.This layer draws upon on the US Department of Agriculture's gNATSGO database [75] which is a compilation of data from walking surveys.We extracted tide datum data from NOAA's VDatum tool [76] and interpolated data across gaps when any existed.Tidal range is the height difference between MHHW and Mean Lower Low Water (MLLW) measured in meters.The marsh elevations relative to tidal datums were determined using NOAA's sea level rise lidar-based digital elevation model (DEM) [77] and tidal datums from NOAA's VDatum grids.The lidar and tidal datum data were combined to create high resolution classified rasters depicting different tidal zonations.To measure percent of marsh below MHHW and below MTL, building on an earlier approach [30], we calculated the area below each level and divided values by the total marsh area in each marsh unit, then aggregated results by MUC.We used MHHW rather than MHW (Mean High Water), the tidal height used by an earlier assessment [30], to capture the maximum extent of tidal marsh transitional area.The underlying DEMs used are, with a few exceptions, gridded at a 3 meter horizontal resolution with a 0.328 feet (10 cm) or less root mean square error (RMSE) for vertical accuracy.NOAA's VDatum model adds additional error (RMSE) to the base data that ranges from several centimeters to tens of centimeters depending on location.We further acknowledge that lidar data are often positively biased in marshes [78] but we did not attempt to remove this bias.
We used the NOAA Environmental Sensitivity Index (ESI) database [79] to calculate percent hardened shoreline and shoreline complexity for each marsh unit.We calculated migration space using NOAA's sea level rise inundation data, and generated scores for each MUC based on projected inundation scenarios between 0.3 and 1.8 meters.Following an existing approach [36], within every MUC and for each foot of inundation above MHHW we divided the area of potential future marsh by the current area of marsh to generate a ratio of future to present "potential" marsh area.We then averaged the six scenarios for each MUC and generated a new percentile rank and score to create an average migration ratio metric.We computed wetland connectedness using MUC level data analyzed at a projected 1.2 meter sea level rise scenario, then used a region grouping process to group all connected marsh units under current and future scenarios.Within each MUC group, the number of unique future marsh units were subtracted from the number of unique current MUCs, and divided by the number of unique current MUCs.We used NOAA ESI shoreline vector data to characterize sinuosity within MUCs via a python script publicly available through Esri [80] using Esri ArcGIS software.
We followed the scoring framework outlined above to generate resilience scores by category and then calculated total scores to represent overall resilience for each marsh.We then summarized total resilience scores by averaging results across various scales (region/state/NERR) to demonstrate insights gained from large to medium scale application of this tool.Finally, we outline two case studies.In the first, we examined results from our national analysis on a marsh complex scale to represent the use of this framework within a single management area.
We report metric and resilience scores for the 8 marsh units that make up the Gulf Islands National Seashore, a region on the Gulf Coast of the US that is preserved by the federal government for public recreation, in this example.This National Seashore includes islands in Florida and Mississippi, on either side of Mobile Bay, Alabama.In the second case study, to further demonstrate the how this framework can be downscaled for local decision making, we summarized its application by a conservation group in New Hampshire, with details provided in S3 File.

Results
Here we provide illustrative results from implementation of the tidal marsh assessment framework within the contiguous US.Full results for each MUC are available in a geodatabase on NOAA's Digital Coast website (S1 File) and summaries for each spatial scale detailed below are included in S2 File.On a national scale, 22% of the 1984 total MUCs meet the criteria for high priority for conservation as they are of good current condition, low vulnerability to RSLR, and have high adaptation so are likely to self-sustain into the future (Table 2).Of the marsh units we assessed, 17.5% are likely to not persist as they are of low current condition, are highly vulnerable to RSLR, and have low adaptation potential (Table 2).

Regional results
Results of our analysis show that the Southeastern region of the US has the highest average total resilience score (7.04 ± 2. 65

Results by state
Consistent with the regional analysis, the least resilient states were in the Northeast.New York, Connecticut, and Rhode Island scored lowest in average total resilience (1.95 ± 1.70,  The soil erodibility metric, in the vulnerability category, was the most negative factor for several states (CT, ME, MI, NH, OR, TX, WA), followed by percent impervious surface (CA, FL, MA, NJ, NY, RI, see results in S2 File).The most common positive factor was mean marsh tidal range (CA, CT, GA, MA, ME, NH, NJ, NY, OR, PA, RI, WA) followed by tidal wetland connectedness (AL, DE, FL, MD, SC, see full results in S2

Results by NERR
We evaluated resilience for all 25 NERRs in the coastal contiguous US (Fig 3).The NERR with the lowest current condition is Padilla Bay in Washington State and in contrast to the regional trend for the West Coast, this reserve is also the most vulnerable.Hudson River, New York ranks lowest in adaptive capacity and overall resilience (S2 File and Fig 3), which aligns with the Northeast regional trend.The highest current condition NERRs were in the Southeast region (North-Inlet Winyah Bay, South Carolina and Sapelo Island, Georgia) as were the least vulnerable (Rookery Bay, Florida) and best adaptive capacity and overall resilience (Apalachicola Bay, Florida, S2 The most common metrics reducing resilience were related to land cover, with the top two being percent impervious surface (7 reserves) and percent agriculture (6 reserves).Percent of marsh below MHHW was also a common negative factor (6 reserves).The most common positive factors were marsh core-to-edge ratio (12 reserves) and mean tidal range (8 reserves, S2

Case study: Gulf Islands National Seashore
The eight marsh units in the Gulf Islands National Seashore span almost 2,500 square kilometers of Mississippi and Florida coastline on either side of Mobile Bay, Alabama (Fig 4).We present detailed results from these MUCs as a demonstration of the applicability of our assessment framework.The average resilience score for the entire National Seashore is 6.62 (± 3.07), driven by moderate average scores in each of the three metric categories (6.88 ±3.44 current condition, -4.00 ± 2.20 vulnerability, 5.88 ±1.88 adaptive capacity).Individual metric scores vary significantly between the MUCs in Mississippi to the west and Florida to the east (Fig 5).The average total score for marsh units in the western portion of the National Seashore (marsh units A-E) is 8.60 (± 1.14) while the average total resilience score for marsh units in the eastern portion (marsh units F-H) is 3.33 (± 2.08).While these two sides of the National Seashore have relatively similar average adaptive capacity scores (6.60 ± 1.14 for Mississippi and 4.67 ± 2.52  The two categories that have a positive influence on tidal marsh resilience to relative sea level rise-current condition and adaptive capacity-range on a positive scale from 1 (low) to 10 (high).Collectively, metrics in the vulnerability category have a negative influence on resilience, so this category is scored on a negative scale from -1 (low vulnerability) to -10 (high vulnerability).This figure serves as an example of detailed results generated by applying our tidal marsh assessment framework.https://doi.org/10.1371/journal.pone.0293177.g005 for Florida), their average current condition scores differ considerably (9.20 ± 0.84 for Mississippi and 3.00 ± 2.00 for Florida).

Discussion
We combined resilience categories, each composed of multiple resilience metrics, with management recommendations in a framework that can be used to evaluate and compare tidal wetlands worldwide.The framework can be used to examine overall marsh resilience, or a single or subset of metrics.Many of the resilience metrics can be mitigated by management strategies (e.g., percent impervious surface) while others cannot (e.g., tidal range).Understanding the contribution of individual metrics to overall resilience is valuable when developing strategic priorities aimed at addressing relevant mechanisms.Robust datasets from authoritative sources (e.g., C-CAP) facilitated the assessment of marshes across a broad geographic scale, the entire coastal contiguous US.As these datasets are updated, the marsh assessments can be updated as well.
There are several limitations to our study.We did not conduct validation on the framework as a whole but this would be a valuable project in the future, following an established approach [56].Future efforts could also include consideration of mangroves, another critical coastal wetland system that is susceptible to RSLR.Our current analysis is also limited in that we do not include a metric to account for sediment supply, which is a factor in vertical migration potential [81].However, if future users of this framework have sediment supply data available for their area of interest, that metric could be easily incorporated.Evaluating sediment supply in combination with tidal range metrics is likely most valuable within regions, because variability in geomorphology complicates generalizations across regions [56].The results of our analysis at the national scale lack the precision afforded by methods based on field data collection [82] but are a valuable extension of previous efforts to characterize tidal marsh resilience across geographies that relied on interpolating results between far fewer data points [30,83,84].

Trends in tidal marsh resilience categories
Our development of the three categories in this framework improves our ability to evaluate and compare relative resilience across marshes.
Current condition.We developed a straightforward method to characterize the current condition of any tidal marsh by selecting assessment metrics that are applicable across geographies.Previous studies have developed methods to evaluate marsh condition on a local or state-wide scale (see [14,[85][86][87] for examples), but methods and metrics that are highly relevant locally do not always translate to broad-scale comparisons between marshes.For example, percent cover of brackish border species, a vegetation metric used to indicate human disturbed marshes in the Northeast, would not be an indicator for degraded marsh in other regions with naturally lower salinities [83].Current condition scores from the broadly applicable metrics in our framework agreed with a sampling method to assess the current condition of small marshes ultimately determining that Rhode Island tidal marshes are degraded [36].This conclusion, based on data collected in the field, aligns with our findings, based on remote sensing, that RI had the lowest current condition of any state.Our assessment also found that the Northeast ranked low in current condition overall, which agrees with an analysis that found that marshes in this region were already experiencing declines in response to RSLR [88].
Vulnerability.We adopted a robust approach to represent marsh vulnerability that is easier to apply than complex models though our results in this category vary from modeled results to different degrees.Many methods have been used to evaluate vulnerability, particularly in the last decade when accelerated RSLR has become a growing concern.For example, the West Coast was considered highly vulnerable due to low accretion rates and limited migration space [84], while we found that the West Coast is the least vulnerable region relative to others within the contiguous US based on erodibility, tidal range, and area above mean tide levels.An earlier investigation found that the mid-Atlantic is highly vulnerable relative to the Mississippi River and Texas Gulf [89], which aligns with our results that the mid-Atlantic is more vulnerable than the Gulf of Mexico.Their conclusion resulted from differentials in sediment supply between regions.As previously mentioned, our current analysis did not account for sediment supply, however our vulnerability assessment also reached the same conclusion as another study that assessed sediment supply and predicted the Southeast to be less vulnerable to RSLR than the Northeast [60].
Adaptive capacity.Our framework accounts for previously overlooked factors that contribute to future marsh conditions.Until lately, adaptive capacity has received the least attention in assessing tidal marsh resilience though there have been several recent broad scale assessments [9,66,[90][91][92].Our approach is similar to others that utilize remote sensing techniques as a robust measure of migration space; for example, low resilience in the Mid-Atlantic was linked to limited available area for landward retreat in addition to limited vertical accretion potential [41].We found that the Mid-Atlantic scores about average for migration ratio compared to other regions and that no one metric stands out as contributing to, or detracting from, resiliency in this region.We rank the West Coast overall above the Northeast and Mid-Atlantic in adaptive capacity driven in part by a lower percent armored shoreline and a higher average migration ratio.Our results for the Northwest align with an assessment that determined that while that region may be highly vulnerable, it is not the most vulnerable region within the contiguous US [41].Another study characterized the Gulf Coast as highly vulnerable to RSLR, but identified significant areas available for marsh migration [66].We found the Gulf Coast to be one of the more resilient regions, driven in part by higher scores in current condition metrics and variable, but about average, migration ratio scores.

Linking assessment to strategic planning
We provide a framework for applying metrics that can be used as a screening tool to select the most appropriate management practices for a specific site.Coastal managers need an approach that integrates robust metrics [56] to focus on increasing resilience [13], which our study achieves at multiple spatial scales.The decision support table associated with our resilience scoring can be used with most data sources that characterize tidal marsh status within the three metric categories, so Fig 1 could be applied in areas with different data sources.This approach allows managers to identify the most cost-effective projects to achieve strategic planning goals.For example, marshes with low adaptive capacity scores could be good candidates for restoration projects that remove barriers to migration.If the same marsh has a high vulnerability to sea level rise, experimental techniques like thin layer placement (TLP) may also be appropriate.TLP raises the level of the marsh plateau, decreasing its vulnerability to sea level rise.A strategy like this may be suited to restoring Marsh G in our case study example, which has a moderate current condition score.Furthermore, a non-profit or government agency aiming to improve resiliency of tidal marshes in the Southeast and Gulf Coast regions may link scores from our analysis to the management recommendations in Table 1 to conclude that investments should be directed towards protections for existing marshes in Georgia, removal of barriers to migration in some areas in Louisiana, and creation of new higher marsh plains in Alabama.
Comparing resilience categories can be enhanced by looking more closely at individual metrics most relevant to the management option under consideration.For example, the portion of marsh below MHHW could be used to identify marshes with high-high-high composite resilience scores in that are within an elevation range optimal for TLP.A marsh that is almost all below MHHW, such as the average marsh unit within the South Slough NERR in Oregon, might need a large amount of additional sediment to keep up with sea level rise, making TLP less appropriate than thick-layer sediment addition.Another site that is almost all above MHHW, such as the Weeks Bay NERR in Alabama, may only need a small amount of sediment to reduce vulnerability and therefore TLP may be more feasible.The management recommendations we provide serve as a broadly applicable starting point for planning purposes, though individual sites will need further evaluation and consideration.

Scalability of new framework
Our framework can be applied at the estuary or watershed level across tidal marshes on a national and international scale.While the scores generated through this framework should not be used as absolute ranks to compare resilience to RSLR across studies, they are a valuable tool for comparing marshes analyzed within a study.By applying the framework to tidal marshes throughout the nation, we found that the most commonly recommended management option linked to resilience scores in the contiguous US is to maintain current marsh extent through conservation actions (Figs 1 and 2).This is primarily driven by the Gulf of Mexico and Southeast, which were the regions with the highest average total resilience scores (Fig 2).These areas are characterized by higher core to edge ratio, higher percent natural watershed, and lower erodibility than other regions we assessed.The second most commonly recommended management category was to limit protection of existing vegetation because of low overall resilience, with The Mid-Atlantic and Northeast contributing the majority of total marsh units with this condition (Table 2).Marshes in these regions are candidates for investments in major restoration and marsh facilitation projects including removal of barriers to migration, and they may also offer opportunities to test experimental restoration methods.
Resilience metrics analyzed within our framework can be used to evaluate how well a specific marsh represents the broader system in which it is located, or to see if the spatial design of a research or monitoring program represents the range of conditions in the study area.For example, in our case study all but one of the marsh units in the Gulf Islands National Seashore score well for the percent of marsh below MHHW, so any marsh unit besides Marsh Unit A can be considered representative of the National Seashore's overall marsh elevation (Fig 5).The value of individual metrics can also vary across scales, for example although tidal range varies between regions within the US, it is fairly consistent within local geographies.In contrast, land cover metrics such as UVVR and hardened shoreline can be locally heterogeneous.In cases where high resolution data are available, on a plot or transect scale, this framework can be applied to compare areas within a single tidal marsh.Additional metrics can also be added within any of the three resilience categories to expand assessments wherever data are available across the area of interest.For example, we used a downscaled application of this framework to support local conservation planning in the US state of New Hampshire, adding metrics to the current condition category in response to stakeholder needs (S3 File).
Multi-component regions or reserves can benefit from considering how resilience scores differ among sites and how that may inform experimental design, infrastructure placement, or priority areas to acquire for conservation.In our primary case study example, marsh units at the western end of the Gulf Islands National Seashore will require less investment than marsh units on the eastern side for increased resilience to RSLR (Fig 4).Interventions that are specific to an area of management concern may benefit more than one marsh unit, so identifying local trends could enhance efficiency of restoration investments.For example, adjacent marsh units on the eastern side of our Gulf Islands National Seashore case study may all be good candidates for actions that increase sediment supply to the marsh (Figs 4 and 5).While our application of this framework at the national level supports large-scale planning and can serve as a screening tool for smaller scales, local and regional planners would benefit from applying the framework within their specific area of interest, with the finest resolution data available for that scale, to maximize the precision of local management recommendations.In combination, individual metrics, total resilience scores, and linked management categories provide critical information for resource managers to proactively plan for tidal marsh response to RSLR at the local and landscape scale.S1.Geospatial metadata for calculating tidal marsh resilience metrics.All geospatial metadata for the application of this tidal marsh resilience framework in the coastal contiguous United States is available on the NOAA InPort NMFS Enterprise Data Management Program site at https://www.fisheries.noaa.gov/inport/item/62985.A copy of this webpage is also included below as a static archive.Results from the application of these methods, titled "Tidal Marsh Resilience to Sea Level Rise," are available for download as a geodatabase on the NOAA Office for Coastal Management Digital Coast website at https://coast.noaa.gov/digitalcoast/data/marshresilience.html and as an Excel file on Github at https://github.com/e-bish/Tidal_Marsh_Resilience.All of the code used to generate these results is available in the Github repository in both Python and in R languages.We also provide an Excel file that additionally replicates the calculations to generate the same results.A story map to visualize the results, titled "Salt Marsh Resilience to Sea Level Rise" can also be found on arcgis.comat (arcgis.com)https://experience.arcgis.com/experience/25037478b4634bb2bf421f443ac47541/page/page_0/?views=view_6%2Cview_19%2Cview_24.The NOAA Coastal Change Analysis Program (C-CAP) produces national standardized land cover and change products for the coastal regions of the U.S. C-CAP products inventory coastal intertidal areas, wetlands, and adjacent uplands with the goal of monitoring changes in these habitats, on a one-to-ve year repeat cycle.The timeframe for this metadata is reported as 2010-Era, but the actual dates of the Landsat imagery used to create the land cover may have been acquired a few years before or after each era.These maps are developed utilizing Landsat Thematic Mapper imagery, and can be used to track changes in the landscape through time.This trend information gives important feedback to managers on the success or failure of management policies and programs and aid in developing a scienti c understanding of the Earth system and its response to natural and human-induced changes.This understanding allows for the prediction of impacts due to these changes and the assessment of their cumulative e ects, helping coastal resource managers make more informed regional decisions.NOAA C-CAP is a contributing member to the Multi-Resolution Land Characteristics consortium and C-CAP products are included as the coastal expression of land cover within the National Land Cover Database.The Watershed Boundary Dataset (WBD) is a seamless, national hydrologic unit dataset.Simply put, hydrologic units represent the area of the landscape that drains to a portion of the stream network.More speci cally, a hydrologic unit de nes the areal extent of surface water drainage to an outlet point on a dendritic stream network or to multiple outlet points where the stream network is not dendritic.A hydrologic unit may represent all or only part of the total drainage area to an outlet point so that multiple hydrologic units may be required to de ne the entire drainage area at a given outlet.Hydrologic unit boundaries in the WBD are determined based on topographic, hydrologic, and other relevant landscape characteristics without regard for administrative, political, or jurisdictional boundaries.The WBD seamlessly represents hydrologic units at six required and two optional hierarchical levels.Soil erodibility was computed using Esri's USA Soils Erodibility Factor image service (https://landscape11.arcgis.com/arcgis/rest/services/USA_Soils_Erodi-bility_Factor/ImageServer,accessed March 2018).Analysis was performed within each marsh unit, not the entire HUC.The average erodibility factor for each marsh unit was weighted by the size of the marsh unit and aggregated by 12-digit HUC.

Contact
Tidal range was computed as the height di erence between Mean Higher High Water (MHHW) and Mean Lower Low Water (MLLW) measured in meters.The tidal datum data were extracted from the VDatum tool and interpolated across data gaps to provide complete coverage within the study area.
Percent of marsh below mean higher high water (MHHW) was computed by intersecting all marsh units below MHHW, dividing by the total marsh area, and aggregating by 12-digit HUC.
Percent of marsh below mean tide level (MTL) was computed by intersecting all marsh units below MTL, dividing by the total marsh area, and aggregating by 12-digit HUC.
Percent hardened shoreline was computed using the Environmental Sensitivity Index (ESI) database.The ESI shoreline data were divided by and associated with 12-digit HUC codes.Within each HUC, all shoreline features that were armored (GENERALIZED_ESI_TYPE LIKE '%Armored%') were divided by the total shoreline length and multiplied by 100.
Environmental Sensitivity Index (ESI) maps provide a concise summary of coastal resources that are at risk if an oil spill occurs nearby.Examples of atrisk resources include biological resources (such as birds and shell sh beds), sensitive shorelines (such as marshes and tidal ats), and human-use resources (such as public beaches and parks).

Process
Date/Time: 2020-09-01 00:00:00 Description: NOAA ESI National Shoreline (2017) Migration space was determined using NOAA's sea level rise inundation data.Within every 12-digit HUC and for each foot of inundation above MHHW, the area of potential future marsh was divided by the current area of marsh to generate a ratio of future to present "potential" marsh area.This operation was performed for 1-6 feet of sea level rise inundation scenarios.For each SLR scenario, the resulting migration ratio values were ranked and scored by a quantile distribution function.The six scenarios were were then averaged, a new quantile rank and score was generated and reported by AVG_migration_ratio.
Wetland connectedness, at this national level analysis, was computed using the marsh unit data (MUCs) and an analysis of projected fragmentation/consolidation under the 4 foot future sea level rise (SLR) scenario.A region grouping process was used to group all connected marsh units under current and future scenarios.Within each 12-digit HUC, the number of unique future marsh units were subtracted from the number of unique current marsh units, and divided by the number of unique current marsh units.This became the unitless raw value for the Wetland_Connectedness metric.
Shoreline sinuosity was assessed using the NOAA ESI shoreline vector data to characterize sinuosity within each 12-digit HUC.A sinuosity index was computed using a Sinuosity python script provided by Esri.High values represent linear shorelines and low values represent sinuous shorelines.Since the the sinuosity script generates values from 0-1, where 1 is a straight feature, we multiplied the sinuosity index by -1 so that values closer to 0 were scored higher in the quantile index.
Quantile scores were generated using a python script developed by NOAA OCM.Each metric was ranked and scored 1-10 based on the quantile position in the ranking.The quantile scores were then inversed if required based on the intent of the metric (negative or positive contributor to resilience).The quantile scores were summed by category (current condition, vulnerability, and adaptive capacity) and all combined, and new quantile rankings and scores were generated.
The numerical quantile scores were used to generate ordinal data describing the degree to which marsh units scored "low" or "high" for each resilience category (current condition, vulnerability, and adaptive capacity).Scores 1-5 were assigned "low" and scores 6-10 were assigned "high."An overall management category was generated using a concatenation of the three resilience categories, in the order just shown.
Vulnerability scores were updated to x an error caused by the erroneous inclusion of the hardened shoreline metric in the original scoring (3/23/2022).

S2. Summary results for each spatial scale of application to US marshes Table A. Abbreviations used for metrics, states, and National Estuarine Research Reserves
Table B. Tidal marsh resilience scores for the coastal contiguous US.Average resilience score (with standard deviation) by metric, metric category, and total marsh units analyzed within the scope of this study.Readers should note that the raw metric data were relativized to create metric resilience scores, so the nation-wide averages presented below are the averages of the distribution of scores between 1 and 10 for each metric.
Table C. Tidal marsh resilience scores for regions within the coastal contiguous US.Average resilience score (with standard deviation) by metric, metric category, and total marsh units analyzed within the scope of this study.
Table D. Tidal marsh resilience scores for states within the coastal contiguous US.Average resilience score (with standard deviation) by metric, metric category, and total marsh units analyzed within the scope of this study.

Table E. Tidal marsh resilience scores for National Estuarine Research Reserves (NERRs)
within the coastal contiguous US.Average resilience score (with standard deviation) is provided by metric, metric category, and total marsh units analyzed within the scope of this study.Scores for NERRs which only contain a single MUC are listed without standard deviations.

S3. Case Study: Parcel Scale Planning in New Hampshire
Downscaling a broadly applicable framework can be helpful to support parcel-level management planning, the scale at which most on the ground restoration and conservation projects are implemented.Marsh Units within our framework can be created at any scale, as demonstrated by a consortium that implemented this framework in New Hampshire.Partners including Great Bay National Estuarine Research Reserve, New Hampshire Coastal Program, University of New Hampshire, and The Nature Conservancy subdivided 13 of the marsh units used in the national analysis into 224 new marsh units using natural breaks in high resolution (< 1 m) tidal wetland land cover data.These land cover data were created by the National Oceanic and Atmospheric Administration's (NOAA's) Office for Coastal Management and can be downloaded via NOAA's Digital Coast (coast.noaa.gov/digitalcoast/data/ccapsalthabitat.html).
The average area of the newly created local marsh units was 26.13 acres.While marsh units in our national scale application of the framework contained emergent marshes in addition to their associated watersheds, the marsh units in this case study contained emergent marsh only.
The users then calculated resilience scores for each of the 224 marsh units using localized data sets for the same 13 metrics outlined in the main text, along with 6 additional metrics for current condition including ditch density, berm presence, and presence of invasive Phragmites australis.The additional metrics were developed with partners to address the locally relevant drivers of resilience given available data.In addition to informing best management options for a specific Marsh Unit (Table 1 in main text), resilience scores can be used to spatially plan the best locations for specific management techniques.For example, a practitioner interested in implementing a thin layer placement project could screen their geography of interest to select areas with high unvegetated to vegetated ratio to further assess in the field.
In this real-world example, a local conservation group was interested where best to locate land protection projects.The users assessed resilience scores for marsh units within unprotected parcels of marsh and migration space in three towns of interest, and then opted to extend the framework to include a conservation score.Marsh units with high current condition, low vulnerability to RSLR, and high adaptation potential (Fig. 1, main text) were scored high for conservation (5).On the other end of the spectrum, those with low current condition, high vulnerability to RSLR, and low adaptation potential scored 0. Those with one marsh resilience category to mitigate receive a score of 3 and those with two categories to mitigate are scored 1.
No marshes received a score of 2 or 4.This prioritization allowed the municipalities to focus where to invest landowner outreach and investment of conservation dollars strategically (Fig. S1).More information about this project can be found at https://greatbay.org/salt-marsh-plan/.

Fig 1 .
Fig 1.Best management options to maximize resilience to relative sea level rise for each marsh type.Green text indicates a positive condition, red reflects a negative one.From a restoration or adaptation practitioner's perspective the more resilience categories that are shown in red for each marsh type, the more costly a project will likely be to implement.
https://doi.org/10.1371/journal.pone.0293177.t002current condition scores (4.44-4.94).The Mid-Atlantic also ranked high in average vulnerability (-6.00 ±2.71) while the West Coast is ranked the least vulnerable on average (-3.45 ±2.95).The biggest challenge, or worst score within any category, varied by region.The Gulf of Mexico is limited by a high unvegetated edge to edge marsh ratio, while the Mid-Atlantic has a high percentage of agricultural area (S2 Table C in S2 File).The Northeast and West Coast are both limited by high erodibility, while the Southeast is limited by percent marsh below MHHW.Conversely, along the West Coast and Northeast large tidal ranges were a positive influence.In the Gulf of Mexico, large unfragmented marsh blocks (a high core to edge ratio) strengthened resilience.Connectedness of tidal wetlands had the most beneficial effect of any factor in both the Southeast and Mid-Atlantic (S2 Table C in S2 File).

Fig 2 .
Fig 2. Regional resilience of marshes as assessed by metrics of current condition, vulnerability to relative sea level rise and marsh adaptation potential.The two categories that have a positive influence on tidal marsh resilience to relative sea level rise-current condition and adaptive capacity -range on a positive scale from 1 (low) to 10 (high).Collectively, metrics in the vulnerability category have a negative influence on resilience, so this category is scored on a negative scale from -1 (low vulnerability) to -10 (high vulnerability).MHHW stands for Mean Higher High Water.Country outline: United States Census Bureau.https://doi.org/10.1371/journal.pone.0293177.g002

Fig 3 .
Fig 3. Average marsh resilience score for each coastal state and each National Estuarine Research Reserve within the contiguous US.Reserves with a standard deviation of zero have their boundaries contained all within one HUC-12 boundary.Country outline: United States Census Bureau.https://doi.org/10.1371/journal.pone.0293177.g003

Fig 4 .
Fig 4. Resilience scores for marsh units of the Gulf Islands National Seashore in Mississippi and Florida.Colored polygons represent MUCs assessed in our national application of the framework.Islands labeled with letters are individual subwatershed boundaries within the Gulf Islands National Seashore.Country outline: United States Census Bureau.World Imagery source: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community.https://doi.org/10.1371/journal.pone.0293177.g004

Fig 5 .
Fig 5. Resilience scores for each marsh unit, resilience category, and composite metric in the Gulf Islands National Seashore.Letters identify marsh units within the National Seashore, as displayed in Fig 4.The two categories that have a positive influence on tidal marsh resilience to relative sea level rise-current condition and adaptive capacity-range on a positive scale from 1 (low) to 10 (high).Collectively, metrics in the vulnerability category have a negative influence on resilience, so this category is scored on a negative scale from -1 (low vulnerability) to -10 (high vulnerability).This figure serves as an example of detailed results generated by applying our tidal marsh assessment framework.
(MUC) were created from the C-CAP 30 meter land cover data.Classes 16-18 (the estuarine wetland classes) were extracted and recoded using a morphometric algorithm to distinguish between di erent components of the marsh units.The following classi cation scheme was established: gridcode1 = core wetland gridcode2 = vegetated edge of wetland gridcode3 = unvegetated edge of wetland Core-to-edge ratio computed from MUCs and aggregated by 12-digit HUC.[Core_Edge_ratio = gridcode1 / ( gridcode2 + gridcode3 )] Unvegetated edge to vegetated edge ratio was computed from MUCs and aggregated by 12-digit HUC.[UnvegVegEdge_ratio = gridcode3 / gridcode2] A simpli ed version of the C-CAP 30 meter land cover was generated to facilitate derivation of land cover metrics.The resulting data layer had the following classi cation scheme: gridcode2 = high intensity developed gridcode3 = medium intensity developed gridcode4 = low intensity developed gridcode5 = open space developed gridcode6 = agricultural classes (pasture/hay, cultivated) gridcode7 = natural cover types (grassland, shrub, forest) Percent impervious cover was computed from the simpli ed version of C-CAP 30 meter land cover data and aggregated by 12-digit HUCs.Analysis was performed within a 150 meter bu er around each marsh unit.[Perc_IC = ((grid-code2 * 0.8503) + (gridcode3 * 0.5768) + (gridcode4 * 0.2929) + (gridcode5 * 0.0941)) / Total_Area * 100] Percent natural cover was computed from the simpli ed version of C-CAP 30 meter land cover data and aggregated by 12-digit HUCs.Analysis was performed within a 150 meter bu er around each marsh unit.[Perc_Natural = gridcode7 / Total_Area * 100] Percent agricultural cover was computed from the simpli ed version of C-CAP 30 meter land cover data and aggregated by 12-digit HUCs.Analysis was per-Process Step 2 CC ID: 989548 formed within a 150 meter bu er around each marsh unit.[Perc_Ag = grid-code6 / Total_Area * 100]

Figure S1 .
Figure S1.The number of local scale marsh units in each national scale marsh unit in New Hampshire and example spatial tax parcel prioritization for land protection based on the total resilience score.Imagery source: Esri, HERE, Gamin, (c) OpenStreetMap contributors, and the GIS user community.

Table 2 . Summary of marsh resilience score results for tidal marsh units in coastal regions of the contiguous US. Marsh Resilience Score Region Total Current Condition-Vulnerability-Adaptive Capacity Gulf of Mexico Mid-Atlantic Northeast Southeast West Coast
SD, Fig 2, S2 Table C in S2 File), while the Northeast has the lowest average total resilience score (4.22 ± 2.62).This is driven by the Southeast having the highest average current condition score (7.08 ±2.45), and average adaptive capacity score (6.98 ± 2.43).Conversely, the Northeast has the highest vulnerability, reflected as the most negative average vulnerability score (-6.48 ± 2.59) and the lowest average adaptive capacity score (4.83 ±2.65).While the Gulf of Mexico also has a high average current condition score (6.9 ± 2.67), the West Coast, Mid-Atlantic, and Northeast have relatively equal low average are a concatenation of results for each resilience category: current condition, vulnerability, and adaptive capacity.Each unique combination of the three marsh resilience categories has a corresponding set of management recommendations, presented in Fig 1.
Table D in S2 File).
Table E in S2 File).