Ecosystem Services Profiles for Communities Benefitting from Estuarine Habitats along the Massachusetts Coast, USA

Abstract

The Massachusetts Bays National Estuary Partnership is one of 28 programs in the United States Environmental Protection Agency’s National Estuary Program (NEP) charged with developing and implementing comprehensive plans for protecting and restoring the biological integrity and beneficial uses of their estuarine systems. The Partnership has recently updated their comprehensive management plan to include restoration targets for coastal habitats, and as part of this effort, the program explored how to better demonstrate that recovery of ecological integrity of degraded ecosystems also provides ecosystem services that humans want and need. An essential step was to identify key stakeholders and understand the benefits important to them. The primary objective of the study presented here was to evaluate variability in beneficial uses of estuarine habitats across coastal communities in Massachusetts Bays. We applied a text mining approach to extract ecosystem services concepts from over 1400 community planning documents. We leveraged a Final Ecosystem Goods and Services (FEGS) classification framework and related scoping tool to identify and prioritize the suite of natural resource users and ecosystem services those users care about, based on the relative frequency of mentions in documents. Top beneficiaries included residents, experiencers and viewers, property owners, educators and students, and commercial or recreational fishers. Beneficiaries had a surprising degree of shared interests, with top ecosystem services of broad relevance including for naturalness, fish and shellfish, water movement and navigability, water quality and quantity, aesthetic viewscapes, availability of land for development, flood mitigation, and birds. Community-level priorities that emerged were primarily related to regional differences, the local job industry, and local demographics. Identifying priority ecosystem services from community planning documents provides a starting point for setting locally-relevant restoration goals, designing projects that reflect what stakeholders care about, and supporting post-restoration monitoring in terms of accruing relevant benefits to local communities.

1. Introduction

Ecosystem restoration aims to recover the ecological integrity and biodiversity of degraded ecosystems while providing the ecosystem services that humans want and need (Gann et al. 2019). Inclusion of ecosystem services into ecosystem restoration planning and monitoring has the potential for motivating action, sustaining interest, and attracting broad support when the project links those benefits of restoration to stakeholder and community interests (Alexander et al. 2016; DeAngelis et al. 2020).

Layering multiple social and economic concerns onto an already complex ecological problem can seem intractable (Link and Browman 2017). Although ecosystem services benefits of restoration are widely acknowledged as primary motivators for restoration (Martin 2017), habitat, biodiversity, or ecosystem function are still often used as surrogates for inferred ecosystem services benefits (Hughes et al. 2016, Diefenderfer et al. 2022). In recent years, a number of frameworks have been published to help guide incorporation of ecosystem services into restoration planning, implementation, and monitoring (e.g., Tazik et al. 2013; NESP 2014; Wainger et al. 2020; Jackson et al. 2022). Impacts of ecological restoration on human welfare can be characterized by the supply and accessibility of benefits from the environment, and the demand for them by users (Pouso et al. 2020). As such, an essential step to effectively designing restoration projects and characterizing restoration outcomes is identifying key stakeholders and understanding what benefits are important to them (Schuster and Doerr 2015). This forms the foundation for setting socio-economic goals of restoration, comparing tradeoffs among project options, and monitoring outcomes in terms of metrics that resonate with the public.

A “beneficiary-focused” perspective can help ensure that potential interested or impacted parties, or stakeholders, are not overlooked and that a full suite of potential benefits of restoration are considered in project design and monitoring (Landers and Nahlik 2013). Biological components of nature that are directly used, enjoyed, or appreciated by people are known as final ecosystem goods and services, or FEGS (Boyd and Banzhaf 2007; DeWitt et al. 2020). A final ecosystem goods and services approach emphasizes identification of the following:

• The beneficiary - Who is benefitting from the restoration?

• The biophysical attributes of ecosystems - What do beneficiaries use or care about?

• The environmental context – What ecosystems are providing the service and where?

The answers to these questions can help to reveal new stakeholders to target for inclusion in planning and outreach; can help to identify priority habitats for restoration to achieve desired beneficial uses; and can help to identify indicators for monitoring that will resonate with a broad range of stakeholders in communicating restoration outcomes. A final ecosystem goods and services approach has been used to identify and prioritize potential co-benefits of restoration (Rossi et al. 2021), expand communication of restoration benefits to a broader set of stakeholders (Hernandez et al. 2022), compare and choose restoration project options to achieve desired benefits (Sharpe et al. 2022), identify measurable and meaningful metrics for monitoring restoration success (Angradi et al. 2019; Santavy et al. 2020; US EPA 2020), characterize economic trends through ecosystem accounting (Warnell et al. 2020), and quantify historic levels of ecosystem services to establish baselines for restoration target-setting (Branoff et al. 2023).

Increasingly in the United States of America (USA), federal, state, and local environmental management programs are adopting goals and implementing activities that link ecological protection and restoration with ecosystem services benefits to communities (Harwell 2020). For example, the National Estuary Programs, established under a revision to the United States Clean Water Act, are mandated to protect the biological, physical, and chemical integrity of the estuary in support of water quality, fish and wildlife, recreational activities, and other designated uses (US Code 2000). Federal estuary management programs have universally identified final ecosystem goods and services concepts in their comprehensive management plans (Yee et al. 2019), typically to provide context for management programs to stakeholders (Martin 2014, Guo and Kildow 2015).

The Massachusetts Bays National Estuary Partnership (MassBays) has recently worked to update their comprehensive management plan (2019) to include restoration targets for coastal habitats, based in part on examining historical trends in area of lost habitat. The MassBays vision includes not just healthy and resilient estuaries, but sustainable ecosystems that support life and the communities dependent on them. As such, a key question in target setting is not only “What kind of ecological future do we want?” but “What kind of socio-economic future do we want?”. An understanding of potential natural resource benefits of ecosystem restoration can help to communicate potential benefits of restoration in ways that motivate local implementation and provide a means for measuring progress in ways that take local concerns into account.

Individual communities, land and environmental management programs, and public or private organizations invest heavily in developing planning and reporting documents to communicate their goals, activities, and progress to a variety of audiences. Development of these documents often involves stakeholder engagement efforts to understand local values and concerns (Reed et al. 2008), or assessments of natural resource benefits to each local community (Elmendorf 2000), providing a potential wealth of information to understand local priorities on ecosystem services and the people who benefit. A key challenge is finding a consistent and efficient way to extract information from documents, given the inconsistencies in structure, language, and purpose. Text-based content analysis is a well-established technique to identify keywords across documents representing a common theme (Hsieh and Shannon 2005), and has been used to evaluate environmental management priorities (Altaweel et al. 2019), community well-being goals (Fulford et al. 2017), sustainable development goals (Li et al. 2023), trends in fisheries research (Fytilakos 2021), and wildlife observations in social media (Edwards et al. 2022). Comparative analysis relies on the development of a ‘corpora’ or linguistics database that allows written language to be tagged to common themes (Cushing 2017).

The primary objective of this study was to apply content analysis to evaluate variability in the beneficial uses of estuarine habitats among coastal communities in Massachusetts Bays. We leveraged a final ecosystem goods and services classification framework, the National Ecosystem Services Classification System Plus (NESCS Plus; Newcomer-Johnson et al. 2020, US EPA 2022b) as a systematic way to categorize document language describing users or beneficiaries of natural resources, the ecosystem services attributes they use or care about, and the coastal habitats providing those services. The FEGS Scoping Tool was used to provide a structured process for identifying priority user groups and ecosystem services profiles for each community, based on the relative frequency of mentions in community planning documents (Sharpe 2021, US EPA 2022a). We evaluated the degree to which communities are similar in their beneficiary and ecosystem services profiles, and the degree to which similarities can be scaled-up from local to regional to MassBays-wide. Finally, we evaluated the degree to which community-scale social, economic, and ecological characteristics explained variability in the relative importance of different user groups and ecosystem services to each community. Community-level ecosystem services assessments ultimately are intended to support identification of stakeholders, and to set the stage for defining locally-relevant restoration goals and targets, for designing and implementing projects that reflect what stakeholders care about, and for monitoring the effectiveness of restoration in terms of accruing benefits to local communities.

2. Methods

2.1. Study Area

The MassBays planning area encompasses 1700 kilometers of Massachusetts coastline and includes 43 separate estuarine embayments and 25 inter-embayment assessment areas that intersect more than 55 coastal cities and towns (Figure 1; MassBays 2017). Mass Bays NEP is dedicated to protecting, restoring, and enhancing living estuarine resources, including approximately 13,700 hectares of salt marsh, 5,700 hectares of tidal flats, 4,000 hectares of seagrass, 300 hectares of rocky intertidal shorelines, 4,400 hectares of dunes and sand beaches, 250 km of herring runs or fishways, and 27,900 hectares of benthic shellfish habitat.

Figure 1.

Massachusetts Bay assessment areas. Numbers and colors indicate assessment areas which clustered together based on priority beneficiaries and priority FEGS derived from document analysis. Acronyms as follows: BCH – Beach; BRK – Brook; CRK – Creek; CST – Coast; HRB – Harbor; ILD(S) – Island(s); PD – Pond; PNT – Point; RVR – River; SND – Sound.

The MassBays comprehensive management plan has already recognized the importance of tying restoration of ecological habitats and water quality to desired beneficial uses by coastal communities and sustainable delivery of ecosystem services (MassBays 2019). In general, fish runs and benthic shellfish habitats support recreational and commercial fishing, as well as filtering of nutrients and pollution discharged up the watershed. Beaches, salt marshes, tidal flats, and rocky outcrops provide habitat and feeding grounds for native and migratory bird fauna. Rocky shores help to stabilize shorelines, provide haul out areas for seals, and help to maintain clean water through flushing by wave action. Beaches are popular for recreational uses, residential and commercial development, as well as providing protective barriers against powerful storms. Seagrass supports commercially important species of fish. Tidal flats support biodiversity, commercially important fish and shellfish, and are important for storm prevention and flood control. Salt marshes contribute to nutrient cycling, water quality improvement, and protect against storms and sea level rise. Along with seagrass, salt marshes also provide highly effective carbon capture.

Environmental management operates at different scales however, and restoration goals, design, and monitoring ultimately need to accommodate differing local perspectives and priorities (Alexander et al. 2016; Williams and Hoffman 2020). To characterize the local concerns of communities within each assessment area in terms of how they use or interact with these coastal ecosystems, we used the FEGS Scoping Tool framework (US EPA 2022a) to structure a text mining approach to identify and prioritize the ecosystem services of highest relevance for each type of coastal ecosystem in each assessment area community.

2.2. FEGS Scoping Tool Steps

Identifying FEGS for a given ecosystem type begins with identifying which beneficiary groups directly interact with that ecosystem, then considers what specific components of nature are directly used, enjoyed, or appreciated by each beneficiary group (DeWitt et al. 2020). The FEGS Scoping Tool provides a step-by-step approach for identifying and prioritizing ecosystem services (Sharpe 2021, US EPA 2022a) and has been used in a number of case studies, including to identify benefits of wetland restoration (Hernandez et al. 2022) and upstream benefits of management practices to improve estuarine water quality (Rossi et al. 2022). First, key stakeholders are identified along with criteria to prioritize them. Second, a beneficiary profile for each stakeholder group is developed based on how they use ecosystems. Third, relevant ecosystem services attributes are identified for each beneficiary type. Lastly, a final score for each ecosystem services attribute is calculated based on their total importance across beneficiary roles, weighted by the relative importance of different beneficiary roles to stakeholders.

As an alternative to using the FEGS Scoping Tool in workshops with each local community, an analysis of publicly available community planning documents was used to represent stakeholders (Figure 2). Mentions of beneficiary types and ecosystem services attributes in each document were used to infer the relative importance of beneficiaries in each assessment area, and the relative importance of types of ecosystem services to each beneficiary. Text-based document analysis is a well-established technique (Fulford et al. 2017), including for identification of ecosystem services to characterize community goals and identify relevant indicators (Angradi et al. 2019, Williams and Hoffman 2020, Rossi et al. 2022). A document analysis can be useful as a starting point prior to further stakeholder engagement that leverages existing material and reduces overall burden on community participants and is a practical alternative when a large number of communities are being compared.

Figure 2.

Diagram illustrating how steps of the FEGS Scoping Tool Framework were used to structure document analysis to generate beneficiary and ecosystem services attribute profiles for each assessment area community.

The FEGS Scoping Tool uses the National Ecosystem Services Classification System Plus (NESCS Plus; Newcomer-Johnson et al. 2020, US EPA 2022b) to categorize types of beneficiaries and types of ecosystem services attributes. Classification systems help provide structure for comprehensively identifying ecosystem services and the types of users who benefit from them, so that less obvious attributes or beneficiaries are not overlooked. NESCS Plus defines a classification system for FEGS based on three components: 1) the environment providing the service; 2) the ecological attributes defining the service; and 3) a beneficiary who uses or cares about the service. Each component is composed of classes and subclasses (Table 1). Based on our initial scoping of documents, we further broke down the NESCS Plus category of ‘fauna’ into the type of taxa (e.g., fish, birds), distinguished open space for aesthetic value from open land for development, and added intermediate regulating services (e.g., buffering, filtration), to provide an additional layer of detail and make sure important concepts commonly mentioned in documents were not overlooked.

Table 1.

Classes and subclasses (in parentheses) from NESCS Plus used to characterize ecosystem services terms from planning documents by the three components of a FEGS: the ecosystem providing the service, the types of people who are benefitting, and the ecosystem services attributes they care about.

Ecosystem	Beneficiary	Ecosystem Services Attributes
Where is the benefit provided?	Who is benefitting?	What ecosystem attributes do they use or care about?
Beaches & Dunes Benthic Habitats Anadromous/ Diadromous Fish Habitat Rocky Shore Salt Marsh Seagrass/Eelgrass Tidal Flats Nonspecific Coastal Habitats	Agricultural Users (Livestock Grazers, Processors, Aquaculturists, Farmers, Foresters) Commercial/Industrial Users (Food Extractors/Fisheries, Timber/Fiber/Ornamental Extractors, Processors, Pharmaceutical/Supplement Suppliers, Fur/Hide Trappers, Private Energy Generators, Private Water/Dam Operators, Property Owners) Government/Residents (Property Owners, Military/Coast Guard, Public Energy Generators, Public Water/Dam Operators) Transporters (Goods, People) Subsistence Users (Timber/Fiber, Building Material, Food/Medicinal) Recreational Users (Experiencers/Viewers, Food Pickers/Gatherers, Hunters, Fishermen, Waders/Swimmers/Divers, Boaters) Inspirational Users (Spiritual/Ceremonial Participants, Artists) Learners (Educators/Students, Researchers) People Who Care (Existence Value, Option/Bequest) Humanity & Public Wellbeing	Fauna including Mammals, Birds, Insects/Invertebrates, Fish/Shellfish (Bait, Community, Edible, Fur/Hide/Trophy, Medicinal, Keystone, Charismatic, Rare, Commercial Important, Pest/Invasive, Pollinating) Flora (Community, Edible, Medicinal, Keystone, Charismatic, Rare, Commercially Important, Spiritually/Commercially Important, Pest/Invasive) Fungi (Community, Edible, Medicinal, Keystone, Charismatic, Rare, Commercially Important, Spiritually/Commercially Important, Pest/Invasive) Soil & Substrate (Quality, Quantity) Water (Quantity, Quality, Movement) Atmosphere/Weather (Air Quality, Wind Strength/Speed, Precipitation, Sunlight, Temperature) Natural Components (Fuel Quantity/Quality, Fiber Quantity/Quality, Mineral/Chemical Quantity/Quality, Ornamental or for Consumption) Mixed/Composite Attributes (Aesthetic Sounds/Scents, Aesthetic Views, Naturalness, Aesthetic Open Space, Open Land for Development) Regulating/Buffering Services (Climate/Carbon Sequestration, Air Quality/Atmospheric, Water Quality including Nutrient Reduction, Soil/Sediment) Mitigation of Extreme Events (Flooding, Fire, Weather, Earthquakes/Landslides)

To review documents, we developed a set of keywords to describe each class or subclass (Yee et al. 2019). To ensure community documents were reviewed consistently, we used an automated process with a script written in R (www.r-project.org) to search sentences in each document for a list of keywords associated with each FEGS category (Table 1). Keywords are essentially synonymous words defining each subclass (See Supplementary Material S1), and could be paired with companion words (‘and’) or exclusion words (‘but not’) to minimize false hits (see Yee et al. 2019, for additional details).

To be considered a FEGS, a sentence must contain each of the three components: an ecosystem, a beneficiary, and the ecosystem services attribute they use or care about. For example, the phrase “mud flats provide opportunities for recreational bird observation” would be categorized as “Tidal Flats”, “Experiencers/Viewers”, ”Charismatic Birds”; the phrase “commercial development on barrier and coastal beaches” would be categorized as “Beach”, ”Commercial/Industrial Property Owners”, “Open Land for Development”; the phrase “salt marshes protect upland areas in town from the full brunt of high energy coastal winds” would be categorized as “Salt Marsh”, “Government/Residential”, “Mitigation of Extreme Weather”. Phrases could be binned into more than one FEGS combination, if for example commercial development also mentioned “Construction”, or extreme weather specifically mentioned “Wind”. Recognizing that relevant information may be contained in the enclosing paragraph, we also searched the prior two or following two sentences if one of the components was missing. This generated a list of relevant FEGS for each document, regardless of how many times they appeared in a document. Despite efforts to optimize accuracy of the automated search process, some of these FEGS inevitably will be ‘false hits’ that arise from combinations of words by happenstance occurring in the same sentence. More common FEGS, e.g., those occurring in more than a single document, are more robust to being ‘true’ FEGS, rather than artifacts of the automated search protocol. Although false hits cannot be eliminated entirely in an automated analysis, they are uncommon and ultimately have little influence on the overall prioritization for each assessment area (e.g., top 20).

The list of FEGS was used to generate priority beneficiary and ecosystem services attribute scores for each document, following the steps of the FEGS Scoping Tool (Figure 2).

Step 1: Community Planning Documents

MassBays assessment areas are named by the major natural features, i.e., embayments, rivers, beaches, that they encompass (Figure 1). We further identified which coastal cities or towns intersected each assessment area, with many cities or towns intersecting more than one assessment area, and most assessment areas intersecting more than one city or town (State of Massachusetts 2021). We then conducted an internet search to obtain as many publicly available community documents as possible for each assessment area, with each document representing a ‘stakeholder’ in the first step of the FEGS Scoping Tool framework. The search terms consisted of the named natural features or towns intersecting each assessment area in combination with types of community documents likely to mention natural coastal resources: adaptation plans or strategies; brochures or guides; climate, resilience, or vulnerability plans or assessments; community, town, development, or economic plans or reports; recovery or improvement plans; environmental impact reports or statements; feasibility reports or studies; conservation or restoration projects or plans;open space or recreation plans; mitigation plans; habitat assessments; landscape or resource inventories; harbor plans; sustainability or stewardship plans; estuarine, stormwater, or wildlife management plans; watershed plans or reports; water quality assessments or reports; and similar. Document authors included local governments, non-profit organizations, federal agencies, among others, and could include agencies or contractors external to the local area that solicited local feedback. Documents must be fairly recent (2000–2021) and be targeted to specific communities or regions to be included, such that documents, for example, targeting the entire state or coastline of Massachusetts were excluded. In particular, the MassBays comprehensive management plan itself, and related MassBays-wide implementation or assessment documents, were not included in our analysis as they could not help us discern differences among local communities.

For each document, we used the automated keyword search to identify which of the eight possible ecosystem categories (Table 1) were mentioned at least once as a component of a FEGS. Documents that did not mention any of the ecosystems in combination with at least one type of beneficiary and one type of ecosystem services attribute were dropped. Documents retained for each assessment area were assumed to be weighted of equal priority, such that assessment area beneficiary or ecosystem services attribute prioritization scores could be calculated as the average across documents.

Step 2: Beneficiary Profiles

For each ecosystem category and document, a beneficiary profile was generated from the list of beneficiary subclasses (Table 1) mentioned as a component of a FEGS identified by the automated keyword search for that document. The relative importance of each beneficiary i (B_i) was calculated as a fraction of the total number of identified beneficiaries for that document, such that relative importance across all beneficiaries sums to 1 for a given document and ecosystem.

Step 3: Ecosystem Services Attributes of Relevance to Each Beneficiary

Next, we identified the ecosystem services attributes (Table 1) relevant to each identified beneficiary subclass in the beneficiary profile for each document and ecosystem category. Any ecosystem services attribute mentioned in combination with that beneficiary subclass and the given ecosystem, identified as FEGS through the automated document search, were assumed to be relevant. The relative importance of each ecosystem services attribute j to each beneficiary i (ES_ij) was calculated as a fraction of the total number of identified ecosystem services attributes for that beneficiary, such that the relative importance across all ecosystem services attributes sums to 1 for a given beneficiary, ecosystem, and document.

Step 4: Priority Ecosystem Services Attributes

The total importance score of each ecosystem services attribute j was then calculated as the sum of relative ecosystem services importance scores ES_ij for each beneficiary i (Step 3) weighted by the relative importance of each beneficiary type B_i (Step 2), such that ecosystem services attributes of importance to multiple beneficiaries or of importance to a high scoring beneficiary would receive higher overall priority scores:

$$ESj=\sum _i{B}_i\times {ES}_{ji}$$

The relative importance across all ecosystem services attributes (ES_j), or equivalently across the full matrix of weighted ecosystem services components for each beneficiary (B_i×ES_ji), sums to 1 for a given document and ecosystem.

2.3. Statistical Analysis

We used Distance-based Redundancy Analysis (DbRDA) to evaluate whether the frequency by which the seven estuarine ecosystems, plus an eighth category representing generic mentions of “coastal habitat” (Table 1), were mentioned in documents differed significantly among assessment areas. DbRDA, also known as permutational multivariate analysis of variance (PERMANOVA), is a constrained multivariate ordination method based on dissimilarity measures that partitions variation in response to one or more factors (McArdle & Anderson 2001; Anderson 2017). DbRDA is often used to analyze ecological community data describing the relative abundances or presence/absence of different species at different sites. Analogous for our study, we examined the presence/absence of each of the eight ecosystem categories in each document. We then used DbRDA, with a Jaccard dissimilarity measure for presence/absence data (Anderson et al. 2006), to test whether these ecosystem profiles differed significantly among the 68 assessment areas. We conducted Kruskal-Wallis tests as a post-hoc analysis to evaluate which specific ecosystems were significantly different among assessment areas, using Dunn’s correction for multiple tests. For DbRDA and post-hoc analyses, documents were treated as replicates. DbRDA analysis were implemented in R using “dbrda” in the library package. Kruskal-Wallis tests were conducted using “kruskal.test” in the R library “stats”.

Next, we evaluated whether the FEGS Scoping Tool steps, i.e., the beneficiary profile (Step 2), the relative importance of ecosystem services attributes to each beneficiary (Step 3), and the final ecosystem services attribute profile (Step 4), differed among either assessment areas or by ecosystem. First, we used DbRDA to evaluate whether beneficiary profiles, i.e., the relative importances of beneficiary subclasses, differed among assessment areas or by ecosystem. Additionally, we examined the interaction (Assessment Area × Ecosystem) to evaluate whether the use of ecosystems by beneficiaries differed among assessment areas. Second, we evaluated whether the relative importances of ecosystem services attributes to each beneficiary differed significantly among beneficiary subclasses using DbRDA. Specifically, we evaluated whether the ecosystem services attribute profiles for each beneficiary subclass differed across beneficiaries, across assessment areas, and by ecosystem (Beneficiary × Ecosystem × Assessment Area). Third, we used DbRDA to evaluate whether the final ecosystem services profile, i.e., the summed relative importance of attributes across all beneficiaries, differed among assessment areas or ecosystems. Again, we examined the interaction (Assessment Area × Ecosystem) to evaluate whether the prioritized list of ecosystem services attributes for each ecosystem differed among assessment areas. In all steps, Bray-Curtis dissimilarities were used as the distance measure in DbRDA, as is common for relative frequency community data (Anderson et al. 2017). Kruskal-Wallis tests were used as a post-hoc analysis to identify which specific beneficiary subclasses or specific ecosystem services attributes were significantly different, using Dunn’s correction for multiple tests.

We evaluated which assessment areas were most similar in their use of different ecosystems (beneficiary profile by ecosystem) and the ecosystem services attributes most important to them (ecosystem services profile by ecosystem) using hierarchical cluster analysis with Bray-Curtis dissimilarity. Similarity profile analysis (Clarke et al. 2008) was used to determine the number of significant clusters. Cluster analysis was run on assessment area scores, calculated as the mean across documents within an assessment area for each ecosystem. Clusters were also used to aid in interpretation of patterns of beneficiary profiles and ecosystem services attributes profiles across assessment areas.

Finally, we evaluated whether environmental and socio-economic characteristics of coastal communities explained significant variability in differences among assessment areas in how ecosystems are used and the most important ecosystem services attributes. We hypothesized that frequencies at which each different ecosystem was mentioned in documents would be related to i) availability of that habitat in assessment areas, ii) whether communities border an estuarine embayment, or iii) regional differences. To test this, we used stepwise multiple linear regression to evaluate relationships between the fraction of documents in each assessment area mentioning each of the seven ecosystem types versus data on the corresponding habitat coverages of beach, eelgrass, fish runs, rocky shore, salt marsh, shellfish habitat, or tidal flats. Habitat coverages for each assessment area were obtained from MassBays (2017) and percent habitat area per total assessment area were calculated in ArcMap 10.8.1. Regional differences were evaluated as a categorical factor, with assessment areas assigned to regions by MassBay NEP (2017) as Upper North Shore, Lower North Shore, Boston Metro, South Shore, or Cape Cod. Mass Bays NEP also assigns assessment areas as estuarine or inter-estuarine depending on whether the area encompasses an estuarine embayment. Regressions were conducted using “lm” in the R library “stats”. We used a forward-step approach, retaining variables which resulted in a smaller AIC value and were significant at p<0.05.

We hypothesized that beneficiary use of and preference for ecosystem services would depend on the regional and ecological differences described above, but additionally the iv) frequencies of different industries in the area, particularly those dependent on natural resources, and v) cultural and social indicators, such as age or income or urban/rural. As previously described, percent habitat coverages for each assessment area, regions, and estuarine vs. inter-estuarine were obtained from MassBay NEP (2017). Industries in each assessment area were measured as the percent of the working population in each category, obtained from the US Census (2019) for each census tract: Agriculture/Fishing, Construction, Manufacturing, Wholesale Trade, Retail Trade, Transportation/Utilities, Information, Finance, Professional, Education/Health, Arts and Recreation, Other Services, and Public Administration. Cultural and social characteristics of each assessment area included percent population below age 5, percent population above age 65, percent minority, percent less than high school educated, and fraction of the population working. Median income of each assessment area was calculated relative to the median income across Massachusetts ($81,215). Population density per assessment area was used as measures of rural to urban gradient. Population characteristics for each census tract were obtained from the US Census (2019) and calculated in ArcMap 10.8.1 for each assessment area as weighted means based on the relative overlap of census tracts with assessment area boundaries. Step-wise DbRDA was used to identify variables explaining significant variability in beneficiary scores or final ecosystem services attribute scores for each assessment area and ecosystem, calculated as the mean score across documents. We used a forward-step approach, retaining variables which resulted in a smaller AIC value and were significant at p<0.05. For variables with strong collinearity (Pearson correlation coefficient >0.65), we restricted our analysis to the one explaining the highest variability. Post-hoc Pearson correlations between each explanatory variable in the selected model and the relative importance of each beneficiary or ecosystem services attribute were used to help interpret model results.

3. Results

Our document analysis, structured by use of the NESCS Plus Classification System and FEGS Scoping Tool, identified significant variability in the frequency with which ecosystems, beneficial uses, and ecosystem services attributes are mentioned in community planning documents for each of the 68 MassBays assessment areas.

3.1. Documents

In total, 1,588 planning documents were obtained from the document search as relevant to natural resource use and management by coastal towns and communities of the 68 MassBays assessment areas (Figure 1; Supplementary Material S2). From the full set of identified documents, 162 documents were excluded from further analysis due to lack of mention of any keywords associated with any of the seven focal habitats or generic coastal habitats (Table 1). Retained planning documents included economic development or revitalization plans (14.9%), restoration or conservation planning or reports (13.0%), climate adaptation plans and vulnerability assessments (11.6%), natural resource inventories or assessments (9.3%), environmental or wildlife management plans (8.3%), community and activity guides (6.9%), water quality and stormwater management reports or plans (6.6%), watershed reports and plans (5.7%), harbor management reports and plans (5.5%), recreational or open space plans (5.3%), environmental impact assessments or mitigation plans (4.4%), feasibility reports (3.2%), and city or town reports (3.0%). A single assessment area was assigned as few as 24 and as many as 288 documents. Approximately 40% of documents were uniquely assigned to a single assessment area, and 71% crossed boundaries of 3 or fewer assessment areas.

The automated document analysis identified 11,325 unique FEGS, each of which was a unique combination of a beneficiary, an ecosystem services attribute they use or care about, and the coastal ecosystem providing that service. Of these, 8,115 were mentioned in more than one document and 2,316 were ‘common’, being mentioned in atleast 10% of documents. Overall, a median of three documents mentioned each FEGS.

Documents mentioned as few as one single FEGS, and as many as 1,271 unique FEGS, with a median of 57 unique FEGS per document. The most common FEGS, identified in more than 300 documents, included aesthetic viewscapes for experiencers/viewers provided by beaches; naturalness of beaches, fish habitat, salt marsh, and coastal habitats in general to experiencers/viewers, general government/residential, and people who care about existence value; flora of salt marsh to people who care; fish and shellfish from fish habitat to government/residents in general; general regulating services of salt marsh and coastal habitats in general to people who care; water movement along beaches to recreational boaters; water movement in fish habitat related to water plant or dam operations and government/residential in general; water quality, water quantity, and flood mitigation of coastal habitats in general to people who care.

3.2. Ecosystem Frequency in Documents

Of the 1,426 reviewed documents, references to beaches, fish habitat, salt marsh, and coastal habitats in general were most common, being mentioned in more than 60% of documents (Figure 3). Tidal flats and benthic habitats were the next most commonly mentioned ecosystems, appearing in 37% and 31% of documents, respectively. Rocky shore and seagrass were least common, each being mentioned in less than 20% of documents.

Figure 3.

Proportion of documents mentioning each of the 7 focal habitats or generic references to coastal habitats in the context of ecosystem services. Error bars indicate the minimum and maximum proportion of documents mentioning each habitat for a single assessment area. Symbols indicate whether frequencies of habitat mentions were significantly different across assessment areas by Kruskal-Wallis test for each habitat type: ***, p≤0.001; **, 0.001<p≤0.01; *, 0.01<p≤0.05; @, 0.05<p≤0.1; NS, p>0.1.

Mentions of the eight coastal ecosystem categories in community planning documents differed significantly among assessment areas (DbRDA, F-value = 3.52, p=0.001). For some individual assessment areas, upwards of 100% or as few as 60% of documents mentioned beaches or salt marsh (Figure 3). Mentions of rocky shore were most variable, with some assessment areas having more than 60% of assigned documents mention rocky shore, whereas for other assessment areas less than 2% did. Seagrass was consistently infrequently mentioned across assessment areas (Figure 3).

3.3. Community Beneficiary Profiles

Beneficiary profiles describe the frequency with which different types of users were mentioned in association with each coastal ecosystem. The ways people are described to use each coastal ecosystem differed from community to community, and differed among ecosystems ((DbRDA; Ecosystem × Assessment Area F-value = 1.24; p<0.001). Identified users of beaches and salt marsh were most consistent, or least variable, among assessment areas (Figure 4). Identified users of fish habitat were also fairly consistent across assessment areas, with the greatest variability among assessment areas in how frequently dam/water plant operations were mentioned. Beaches and fish habitat had the greatest diversity of beneficiary subclasses, mentioned on average in at least 2–10% of documents (Figure 4). Benthic habitats, rocky shore, and seagrass by contrast tended to have fewer different types of beneficiary subclasses mentioned in documents, such that only a few user types were fairly common (>10%).

Figure 4.

Mean relative frequency of beneficiary subclasses among assessment areas for each habitat type. Bars indicate minimum and maximum (shown up to ≥ 0.2) relative frequency for any single assessment area. Beneficiaries are in order of highest to lowest mean frequency across all habitat types. Beneficiary subclasses with a mean relative frequency less than 0.01 for each habitat not shown.

People who care (existence value) and general references to government/residents were on average the most commonly identified beneficiary subclasses for all 7 ecosystem types (Figure 4; Supplementary Material S3). Experiencers and viewers also ranked in the top three for beaches, salt marshes, and tidal flats. Researchers were identified as top users of benthic habitats and seagrass. Dam/water plant operations were identified as a top beneficiary of fish habitat. Property owners were identified as one of the top beneficiaries of rocky shores. Commercial food extraction and fisheries were identified among the top users benefitting from seagrass and tidal flats. Benefits of coastal habitats to educators and students were also commonly mentioned for all 7 ecosystem types.

Assessment areas were clustered into groups based in part on similarities in beneficiary profiles for each ecosystem type. Similarity profile analysis identified 42 significant clusters, with the majority of the 68 assessment areas being assigned to its own unique cluster (Figure 5). Therefore, in order to be able to interpret broader patterns among assessment areas, we used the hierarchy of partitioning to further combine assessment areas into 12 groups, the fewest number such that no group contained a single assessment area. In general, assessment areas were most similar to other assessment areas within their same region, with Cape Cod assessment areas having beneficiary profiles for each coastal habitat most different than the more northern regions, and Boston Metro area more similar to South Shore than Lower or Upper North Shore (Figure 5; Supplementary Material S4).

Figure 5.

Twelve groups of assessment areas based on similarity in relative frequency of beneficiaries and final relative importance score of ecosystem services attributes as determined by cluster analysis. Labels indicate larger scale regions that clusters were approximately associated with.

Clusters showed a lot of similarities in their top 10 identified beneficiaries, but often varied in the ecosystem with which that beneficiary was most strongly associated (Table 2). For each cluster, people who care (existence value), government/residential in general, researchers, and experiencers/viewers were in the top 5 identified beneficiaries. Relative importance of experiencers/viewers in association with beach was high in the Cape Cod area (Clusters 2, 5), but higher in association with rocky shore in the southern part of the Upper North Shore (Cluster 6). Researchers were most commonly mentioned in association with benthic habitats or seagrass for most assessment areas, although for other assessment areas researchers were mentioned at similar frequencies for all ecosystems. Commercial food extractors/fisheries, educators/students, public water plant and dam operations, recreation in general, recreational boaters, and residential property owners were also commonly mentioned across all assessment areas, usually ranking in the top 10 identified beneficiaries, or top 5 for some clusters. Commercial and industrial related uses, people who care for option/bequest value, public property owners, recreational fishermen, and transportation uses were also common (top 20) across most assessment areas.

Table 2.

Mean relative frequency of beneficiary subclasses among assessment areas within each cluster (see Figs. 1, 4). Only beneficiaries in the top twenty for at least one cluster are shown. Shading indicates for each cluster, the beneficiaries in the top five (dark grey, bold), top ten (medium grey), top twenty (light grey), or lower (white), on average across all ecosystem types. Superscripts indicate the habitat for which relative frequency of that beneficiary was highest, if ecosystems were significantly different within that cluster (Kruskal-Wallis; Dunn’s corrected alpha <0.0001). Bch=Beach; Ben=Benthic Habitat; Fsh=Fish Habitat; Rck=Rocky Shore; SaM=Salt Marsh; SeG=Seagrass; TiF=Tidal Flat; NCo=Non-specific coastal habitats.

	1		2		3		4		5		6		7		8		9		10		11		12
Agriculture in General	0.01	NCo	0.02	Rck	0.01	Ben	0.01		0.01		0.01	SaM	0.02	SaM	0.01		0.01	Rck	0.01	SeG	0.01		0.01	SaM
Aquaculturists	0.01		0.02	TiF	0.01		0.02		0.02	TiF	0.01	SeG	0.00	SeG	0.00		0.00		0.01	TiF	0.02		0.00	Fsh
Artists	0.01	Fsh	0.01	Bch	0.01	SaM	0.01		0.01	Bch	0.02	Rck	0.01	Rck	0.01		0.01	Rck	0.00		0.01		0.00	Bch
Commercial & Industrial in General	0.02		0.01	Bch	0.02	Bch	0.02	Ben	0.02	Bch	0.02	SaM	0.02	Bch	0.02		0.02		0.02		0.02		0.02
Commercial Food Extractors & Fisheries	0.04	TiF	0.04	SeG	0.04		0.04	TiF	0.04	Ben	0.04	SeG	0.04	SeG	0.03		0.03	SeG	0.05		0.05		0.04
Commercial/Industrial Processors	0.01	SaM	0.01		0.01	NCo	0.01		0.01		0.01	Fsh	0.01	SaM	0.01		0.01		0.02		0.01		0.02
Commercial/Industrial Property Owners	0.02	Ben	0.02		0.02	Bch	0.03		0.03		0.04	Rck	0.03	Rck	0.03		0.03		0.04		0.02		0.03	SaM
Educators/Students	0.04	TiF	0.05		0.05	SaM	0.05		0.05	Ben	0.05	Ben	0.04	SeG	0.04		0.04	Ben	0.05		0.05		0.05
Experiencers/Viewers	0.10	SeG	0.07	Bch	0.07		0.08		0.07	Bch	0.09	Rck	0.08		0.12		0.11		0.05		0.07		0.08
Farmers	0.03	Fsh	0.02	SaM	0.02	SaM	0.01	Rck	0.02	SeG	0.02	SaM	0.03	Rck	0.02		0.02	Rck	0.02	SaM	0.03		0.01	SaM
Food Pickers/Gatherers	0.01		0.02	TiF	0.02		0.03	TiF	0.03	TiF	0.02	NCo	0.02		0.02		0.01	SeG	0.02	TiF	0.02		0.01
Government/Residential in General	0.13	Rck	0.14	Fsh	0.15		0.12		0.12	SeG	0.12	TiF	0.11		0.11		0.12	TiF	0.13		0.11		0.14	Rck
Humanity & Public Health	0.01	Bch	0.02	Bch	0.02		0.02	Bch	0.02	NCo	0.01	SeG	0.02		0.01		0.01	Fsh	0.02	NCo	0.02		0.02
Learning	0.02	Bch	0.02	SaM	0.02	NCo	0.02	NCo	0.02	NCo	0.02	TiF	0.03	TiF	0.03		0.02	TiF	0.01		0.02		0.01
Nonspecific Commercial Transportation	0.02	Ben	0.02	TiF	0.03		0.02		0.02	TiF	0.03	Rck	0.03	Fsh	0.02		0.02	TiF	0.02		0.02		0.03	TiF
Nonspecific Inspirational	0.02	Bch	0.01	Rck	0.01	Bch	0.03	Ben	0.02	Ben	0.02	SaM	0.02	NCo	0.02		0.02	NCo	0.01		0.02		0.02
People Who Care (Existence)	0.13	Rck	0.11		0.10		0.10		0.10		0.12	SeG	0.12	SeG	0.12		0.13	Ben	0.11		0.14		0.12	SeG
People Who Care (Option/Bequest)	0.03	SaM	0.02	NCo	0.02	NCo	0.02	Rck	0.02	NCo	0.03	NCo	0.04	NCo	0.03		0.03	NCo	0.02		0.02		0.02	NCo
Public Property Owners	0.02	NCo	0.02	NCo	0.02	NCo	0.02		0.02	NCo	0.03	NCo	0.02		0.02		0.03	SeG	0.03		0.03		0.03	NCo
Public Water Plant Operators	0.05	SeG	0.05	SeG	0.04		0.04	SeG	0.05	SeG	0.03	Fsh	0.04	Fsh	0.03		0.03		0.06	Ben	0.04		0.05	Fsh
Recreational Boaters	0.03	SeG	0.03	TiF	0.03		0.04		0.04	Fsh	0.04	Ben	0.04	Fsh	0.05		0.04		0.04		0.04		0.04
Recreational Fishermen	0.02	SeG	0.02	TiF	0.03		0.03	TiF	0.03	TiF	0.03	Bch	0.03	Rck	0.03		0.02	Fsh	0.02		0.02		0.01	TiF
Recreational in General	0.04	NCo	0.04	TiF	0.04	NCo	0.03		0.03	NCo	0.04	Bch	0.04	Rck	0.04		0.05		0.04		0.04		0.04	Bch
Researchers	0.06	Ben	0.07		0.07		0.07	Ben	0.07	SeG	0.05	Ben	0.05	SeG	0.05		0.05	Ben	0.05		0.06		0.06	Ben
Residential Property Owners	0.05	Rck	0.04	Rck	0.04		0.05		0.05	Rck	0.04	SaM	0.04	Ben	0.04		0.05	SaM	0.05		0.04		0.05	SaM
Transporters of People	0.01	Rck	0.02		0.01		0.02		0.02		0.01	SaM	0.02	Bch	0.02		0.02		0.02		0.02		0.03
Waders & Swimmers & Divers	0.02	Ben	0.01		0.02		0.01		0.02		0.02	Rck	0.01		0.02		0.02	Bch	0.01	Bch	0.02		0.02	Bch

Differences between clusters were most apparent in the bottom half of the top 20 users, with aquaculture, agriculture or farmers, and waders/divers/swimmers mentioned more commonly in some assessment areas than others and in association with different ecosystems (Table 2). Waders, swimmers, and divers, for example, had varying relative importance in association with benthic habitats, rocky shores, or beaches, depending on the cluster (Supplementary Material S4). Aquaculture was commonly associated with tidal flats in Cape Cod areas (Clusters 2, 5), but with seagrass in the Upper North Shore (Clusters 6, 7).

3.4. Ecosystem Services Profiles for Each Beneficiary

For each beneficiary subclass, we identified the relative importance of different ecosystem services attributes to that beneficiary based on the relative frequency with which different ecosystem services subclasses were mentioned in association with that beneficiary across documents. In general, composite characteristics related to site appeal and aesthetics were of highest relative importance (>20%) to the majority of beneficiary subclasses (Figure 6), particularly for artists and other inspirational uses, experiencers/viewers or other general recreational users, property owners, learning, and bequest value. Major exceptions included high relative importance (>40%) of fish and shellfish to commercial food extractors/fishermen, recreational fishermen, food pickers/gatherers, and aquaculturists, as well as water quality, quantity, or navigability to water plant and dam operations, recreational boaters, and transportation. Most beneficiaries were identified in documents to some degree (5–10% relative importance) in association with regulating services or mitigation of extreme events, which was highest for public property owners (12% relative importance). Presence of flora and fauna, other than fish or shellfish, were also commonly identified as relatively important to many types of beneficiaries, with high relative values (>15%) for people who care about existence or bequest value, experiencers and viewers, educators and students, and learning, among others. The remaining ecosystem services attributes were among the least mentioned in association with any beneficiary, with some exceptions such as minor importance (>5%) of natural materials to artists or industrial processors, weather and atmospheric conditions to property owners and recreation in general, soil quality or quantity to agricultural users, and availability of open land for development by property owners, transportation, or agricultural users.

Figure 6.

Mean relative importance of ecosystem services attributes (grouped by classes) among all assessment areas and ecosystems for each of the top beneficiary subclasses in Figure 3.

The relative importance of ecosystem service attributes differed among beneficiaries, and the degree of importance depended on both the ecosystem and the assessment area (DbRDA, Beneficiary × Ecosystem × Assessment Area, F=2.11, p<0.001). The relative importance of aesthetic viewscapes to experiencers/viewers or people who care, for example, tended to be higher for rocky shores and tended to be higher in the North Shore (Supplementary Materials S3, S4). The relative importance of fish and shellfish to people who care, recreation, learning, or other uses was most commonly associated as a benefit of fish habitat and tended to be higher in Cape Cod. Charismatic, commercial, rare, or other flora were more likely to be assigned a high relative importance to various user groups as a benefit of seagrass or salt marsh. Water movement and navigability in fish habitat had higher relative importance to residential or recreational users in Boston metro area and South Shore. Flora and sediment/substrate of seagrass and benthic habitats was more likely to be associated with recreational fishermen and food pickers/gatherers in Cape Cod and North Shore assessment areas, or with waders/swimmers in North Shore and South Shore. The relative importance of water movement to public water/dam operations was highest in association with fish habitat, particularly in the North Shore area. The relative importance of flood mitigation to residents tended to be higher in South Shore areas, and was most commonly associated generically as a benefit of coastal habitats rather than a specific ecosystem type.

3.5. Community Ecosystem Services Profiles

To generate an overall prioritization of ecosystem services attributes across all beneficiaries, the relative importance of each ecosystem services attribute was summed across all beneficiary types, weighted by the relative importance score of each beneficiary type. For example, naturalness was of high relative importance to many of the most commonly mentioned beneficiary types, including people who care about existence value, experiencers/viewers, and government/residential in general, and thus had the highest overall ecosystem services score on average across all assessment areas (Figure 7; Supplementary Material S3). Water movement/navigability, quantity, and quality were on average in the top ten scoring ecosystem services attributes, due in part to relevance to government/residents in general, recreational boaters, and water plant/dam operations. Fish and shellfish in general, and in particular edible fish and shellfish, scored high overall and were relevant for a number of common beneficiaries, including government and residents in general, people who care about existence value, researchers, commercial fisheries, and recreational fishermen.

Figure 7.

Contributions of beneficiary weights to mean relative ecosystem services attributes priority scores across all assessment areas. Only top ecosystem services attributes with mean score >0.01 are shown. Beneficiaries with a contribution weight less than 0.01 grouped as “All Other”.

Final ecosystem services attribute profiles, which describe the overall relative importance of ecosystem services attributes to beneficiaries mentioned in association with each coastal ecosystem, differed significantly among assessment areas and ecosystems (DbRDA; Ecosystem × Assessment Area F-value = 1.24; p<0.001). Relevant ecosystem services attributes of beaches, fish habitat, and salt marsh were most consistent, or least variable, among assessment areas (Figure 8). By contrast, ecosystem services attributes that scored the highest in association with rocky shore, seagrass, or tidal flats, were far more variable among assessment areas. Naturalness was on average the most relevant (score > 20%) ecosystem services attribute for beaches and rocky shore, and to a slightly lesser extent (score > 10%) for salt marsh, seagrass, and tidal flats. Naturalness was also relatively important (score >10%) for benthic habitat and fish habitat, however fish & shellfish and water movement were often considered equally or even more important. Fish & shellfish, especially edible fish and shellfish, were of moderate relevance (score >5%) for beaches, rocky shore, seagrass, and tidal flats as well. Other fauna, such as birds or charismatic fauna, were also of some relevance (score >1%), but only scored in the top five for rocky shore and tidal flats. Aesthetic views were on average in the top five for beaches and seagrass. Presence of flora, including charismatic or commercially important flora, were in the top five ecosystem services attributes on average for salt marsh and seagrass.

Figure 8.

Mean ecosystem services attribute relative priority scores among assessment areas for each habitat type. Bars indicate minimum and maximum (shown up to ≥ 0.27) scores for any single assessment area. Attributes are in order of highest to lowest mean score across habitat types. Attributes with a mean relative frequency less than 0.01 for each habitat not shown.

For all assessment area clusters (Figure 5; Supplementary Material S4), the top five most relevant ecosystem services attributes included fish and shellfish, naturalness, and water movement or navigability (Table 3). For most clusters, these were scored of particular high relevance in association with tidal flats. Aesthetic viewscapes, edible fish and shellfish, water quality or water quality regulation, and water quantity were in the top ten most relevant ecosystem services attributes for all clusters. Aesthetic views were ranked of particularly high importance for assessment areas in the Upper and Lower North Shore (Clusters 6–9). Open land for aesthetics or development, birds, charismatic flora, flood mitigation, soil and sediment regulating services, along with general references to ecosystem components, regulating services, and water, were in the top twenty for assessment areas across all clusters. The relative importance of flood mitigation was among the most variable across clusters, scoring in the top five in only the upper South Shore (Cluster 10). Availability of open land for development was of primary importance in the North Shore and Boston Metro Area. Charismatic fish/shellfish, charismatic flora, and commercially important flora were generally scored with higher relevance for assessment areas in the Cape Cod area.

Table 3.

Mean relative priority scores of FEGS attributes among assessment areas within each cluster (see Figs. 1, 4). Only beneficiaries in the top twenty for at least one cluster are shown. Shading indicates for each cluster, the FEGS in the top five (dark grey, bold), top ten (medium grey), top twenty (light grey), or lower (white), on average across all ecosystem types. Superscripts indicate the ecosystem for which relative score of that FEGS was highest, if ecosystems were significantly different within that cluster (Kruskal-Wallis; Dunn’s corrected alpha <0.0001). Bch=Beach; Ben=Benthic Habitat; Fsh=Fish Habitat; Rck=Rocky Shore; SaM=Salt Marsh; SeG=Seagrass; TiF=Tidal Flat; NCo=Non-specific coastal habitats.

	1		2		3		4		5		6		7		8		9		10		11		12
Aesthetic Open Space	0.03	TiF	0.03	SaM	0.02	TiF	0.03	SaM	0.02	TiF	0.02	TiF	0.02	NCo	0.02		0.03		0.03		0.03		0.03	TiF
Aesthetic Viewscapes	0.05	TiF	0.03	SaM	0.03	TiF	0.04		0.03	SeG	0.06	TiF	0.05	TiF	0.08		0.05	TiF	0.03		0.03		0.02	TiF
Birds	0.02	SaM	0.04	TiF	0.03	TiF	0.03	SaM	0.02	TiF	0.02	TiF	0.02	TiF	0.02	TiF	0.01	TiF	0.01	TiF	0.02	SaM	0.02	TiF
Charismatic Birds	0.01	SaM	0.01	SaM	0.01	SaM	0.00	SaM	0.00	TiF	0.01	TiF	0.01	TiF	0.01		0.01	TiF	0.00		0.01		0.00
Charismatic Fish/Shellfish	0.01	SeG	0.01		0.01	SeG	0.01		0.01	SeG	0.01	NCo	0.01		0.01		0.01		0.01		0.01		0.01
Charismatic Flora	0.02	SaM	0.01	SaM	0.02	SaM	0.02	SaM	0.02	SaM	0.02	SaM	0.02	SaM	0.02		0.01	SaM	0.01		0.01		0.01	SaM
Commercially Important Fish/Shellfish	0.01	TiF	0.01	TiF	0.01	TiF	0.01	TiF	0.01	TiF	0.02	TiF	0.01		0.00		0.01	SeG	0.01	TiF	0.02		0.01
Commercially Important Flora	0.01	SaM	0.02	SaM	0.02	SaM	0.02		0.02	SaM	0.01	SaM	0.01	SaM	0.01		0.01	SaM	0.01		0.01		0.01	SaM
Edible Fish/Shellfish	0.06	TiF	0.05	SeG	0.07		0.08	TiF	0.07	TiF	0.08	TiF	0.07		0.06		0.06		0.04		0.07		0.05	SeG
Fish & Shellfish	0.08	TiF	0.11	TiF	0.08		0.09	TiF	0.10	TiF	0.06	TiF	0.06	TiF	0.06		0.05	TiF	0.07		0.09		0.06	TiF
Fish/Shellfish Community	0.00	TiF	0.00		0.00		0.00	SeG	0.01	TiF	0.00		0.00		0.00		0.00	SeG	0.01		0.01		0.01	TiF
Flood Mitigation	0.02	NCo	0.03	SeG	0.03	NCo	0.03		0.03	NCo	0.03	NCo	0.03	NCo	0.02		0.04		0.06		0.03		0.04	NCo
Flora Community	0.01	TiF	0.01	SeG	0.01		0.01		0.01	SeG	0.01	TiF	0.01	SaM	0.01		0.01		0.01		0.01		0.01	SeG
Mitigation of Extreme Weather Events	0.00	SaM	0.00	NCo	0.01		0.01		0.01	TiF	0.01	NCo	0.01	SeG	0.00		0.01	SeG	0.01		0.01		0.01
Multiple Ecosystem Components	0.03	NCo	0.02	TiF	0.03	NCo	0.03	TiF	0.03	NCo	0.02	SeG	0.02	NCo	0.02		0.03	NCo	0.03	TiF	0.03		0.03	NCo
Naturalness	0.21	SeG	0.19	TiF	0.17	TiF	0.18	TiF	0.17	TiF	0.22	TiF	0.21	TiF	0.22		0.22	TiF	0.20		0.18		0.20	TiF
Open Land for Development	0.02	SeG	0.02		0.02		0.02	SeG	0.03	SeG	0.04	SeG	0.02		0.03		0.04	SaM	0.02		0.02		0.04	TiF
Ornamental Natural Materials	0.01	TiF	0.00	TiF	0.00		0.00		0.00		0.01	SaM	0.01		0.01		0.01	TiF	0.01	SaM	0.01		0.01	TiF
Other Flora	0.02	SaM	0.02	SaM	0.02	SaM	0.02		0.02		0.02	SeG	0.02	SeG	0.02		0.02	SaM	0.03		0.02		0.02	SeG
Rare Birds	0.01	TiF	0.00	TiF	0.00	SaM	0.00	SaM	0.00	SaM	0.00	TiF	0.01	TiF	0.00		0.00	SeG	0.01		0.01		0.00	NCo
Regulating Services in General	0.02	NCo	0.02		0.02	SaM	0.02		0.03	TiF	0.02	TiF	0.03	TiF	0.02		0.03	TiF	0.04		0.03		0.03
Soil	0.00	TiF	0.00	TiF	0.00	TiF	0.00	TiF	0.00	TiF	0.01	TiF	0.01	TiF	0.02	TiF	0.01	TiF	0.00	TiF	0.01	TiF	0.00	TiF
Soil & Sediment Regulation	0.01	TiF	0.02	TiF	0.01	TiF	0.01		0.01	TiF	0.01	TiF	0.01	TiF	0.01		0.01	TiF	0.01		0.01		0.02
Soil Quality	0.00	TiF	0.00	TiF	0.02	TiF	0.00		0.00	TiF	0.00	TiF	0.00	TiF	0.00		0.00		0.01		0.00		0.00
Water	0.02	TiF	0.04	TiF	0.02	TiF	0.03	TiF	0.02	TiF	0.02	TiF	0.02	TiF	0.02	TiF	0.02	TiF	0.02	TiF	0.02	TiF	0.02	TiF
Water Movement/Navigability	0.07	TiF	0.06	TiF	0.08	TiF	0.09		0.08	TiF	0.10	TiF	0.10	TiF	0.11		0.09	TiF	0.10		0.09		0.11	TiF
Water Quality	0.04	NCo	0.05	NCo	0.05	SeG	0.04	TiF	0.04	NCo	0.03	TiF	0.03	SeG	0.03		0.04	TiF	0.03	SeG	0.04		0.04	TiF
Water Quality Regulation	0.06	TiF	0.05	TiF	0.05		0.04		0.05	TiF	0.04	TiF	0.03		0.04		0.04		0.05	TiF	0.04		0.05
Water Quantity	0.06	NCo	0.05	NCo	0.05		0.05		0.06	SeG	0.03	NCo	0.05		0.03		0.04		0.04		0.04		0.05

3.6. Socio-Ecological Factors Explaining Community Differences

We evaluated whether the amount of coastal habitat, neighboring an estuarine embayment, or regional differences in coastal communities could explain significant differences in the frequency at which coastal ecosystems were mentioned in documents (Figure 3; Supplementary Materials S5). For almost all ecosystems, regional differences significantly explained the frequency by which different ecosystems were mentioned (Table 4; Supplementary Materials S4). For example, mentions of rocky shore and tidal flats tended to be more common in Upper North Shore areas. After accounting for regional differences, assessment areas with greater percent coverage of a particular ecosystem (Supplementary Material S6) were more likely to mention that ecosystem in planning documents (Table 4). The fraction of documents mentioning rocky shore, salt marsh, tidal flats, and to a lesser degree shellfish or fish habitat, were significantly related to the relative cover of each habitat in the area. Mentions of beach were not significantly related to beach cover, but were more likely in communities that adjoined estuarine embayments (Table 4). Seagrass mentions were not significantly different across assessment areas (Figure 3), by regions, or in relation to availability of eelgrass or estuarine habitat (Table 4).

Table 4.

F-values for top ecological or socio-economic variables, identified from step-wise regression or step-wise DbRDA, explaining differences among assessment areas in fraction of documents mentioning each ecosystem, relative importance of beneficiaries, or relative importance of ecosystem services attributes.

	Beaches	Benthic Habitat	Fish Habitat	Rocky Shore			Salt Marsh		Seagrass		Tidal Flat
Ecosystem Frequencies
Region	4.3**	11.7***	14.7***	58.5***			8.5***		2.3NS		24.8***
Estuarine	21.3***
Habitat Cover		4.7*	6.1*	22.4***			47.9***				8.9**
R 2	0.38	0.45	0.51	0.81			0.57		0.13		0.64
Beneficiary Profile
Region	6.5***	7.6***	13.5***	2.8**			5.1***		4.7***		5***
% Less than English	5.4***	3.4**					3.2**				2.4*
Median Income							2.2*
% Over 65				3.9*					3.3**		3.5**
% Agriculture/Fishing				5.1**					2.9**		4.3**
% Construction				3.2*
% Finance							2.4*
% Manufacturing	4.7***	4.1***	7.1***	3.4*			3.7**		3.8**		5.2***
% Retail Trade	2.1*
% Transportation/Utilities									2.4*		2.8**
Habitat Cover	3.1**						4.3***		2.4*
Estuarine				4.4**			3.5***
% Variation Explained	44.3	38.2	52.0	51.5			47.9		45.6		40.9
Ecosystem Services Profile
Region	6***	7.1***	11.6***	4.1***		11***		3.8***		6***
% Less than English	2.3*									2.4*
% Over 65		3.6**		5.3***				2.6*		3.7**
% Agriculture/Fishing				3.1*				3.3**		5.3**
% Information				2.4*
% Manufacturing	5.6***	2.7*	7.6***			3.5**		5.9***		5.1***
% Public Administration	2.3*	3.5*		3.5**
% Retail				4**
% Transportation/Utilities										4.1***
Habitat Cover						2.4*
Estuarine	3.2**	2.4*	2.5*	2.6*
% Variation Explained	51.0	47.3	57.0		47.4		47.6		36.1		44.9

P-values given as:

^***, shaded, p≤0.001;

^**, shaded, 0.001<p≤0.01;

^*, 0.01<p≤0.05; NS, p>0.05.

Additionally, we evaluated whether any socio-economic or ecological characteristics of coastal communities could explain significant differences in the frequency with which different beneficiaries (beneficiary profiles) were mentioned in documents (Table 2). Percent of the population speaking less than fluent English was highly correlated with population density, percent population with less than high school education, and percent minority (Pearson correlation coefficient, r>0.65, p<0.001; Supplementary Material S6). Therefore, to simplify interpretation and reduce collinearity, we restricted stepwise regressions to only consider percent less than English as a surrogate metric representing dense minority population areas. Similarly, percent population over 65 was highly negatively correlated with percent population under age 5 and fraction of population working, so we restricted stepwise regressions to consider percent over 65 as a representative metric of older (i.e., retirement) communities.

Community characteristics explained more than 38% of variability in beneficiary profiles among assessment areas, and in particular more than 50% of variability in identified users of fish habitat or rocky shore (Table 4). Regional differences significantly explained variation in relative importance of beneficiaries associated with each type of ecosystem. For example, mentions of aquaculture or food pickers/gatherers tended to be highest in the South Shore and Cape Cod (Supplementary Materials S4). Mentions of artists, inspirational and recreational users tend to be highest for assessment areas in the North Shore. Transportation, shipping, water, energy, commercial, and industrial uses tended to be highest in the Boston Metro area.

Significant variability in beneficiary profiles for most ecosystems was also explained by whether communities had substantial manufacturing (Table 4), with manufacturing communities more likely to mention people who care and commercial or agricultural processing, and less likely to mention aquaculture, food pickers/gatherers, military/coast guard, or researchers, among others (Supplementary Materials S7). Percent of population employed in agriculture or fishing also explained significant variation in beneficial uses of rocky shore, seagrass, and tidal flat, with farming/fishing communities more likely to mention benefits to farmers, researchers, inspirational users, and residential property owners. Dense minority population areas, represented by percent speaking less than English (Table 4), were more likely to mention people who care or residential property owners, and less likely to mention commercial or recreational fishermen or food pickers/gatherers. A high percentage of people over 65 was associated with how benefits were identified for seagrass and tidal flats, and to a lesser extent rocky shore (Table 4), with older communities less likely to mention general commercial/industrial benefits, and more likely to mention specific benefits of tidal flats like aquaculture, educators/students, or food pickers/gatherers.

Variability in the relative importance of beneficiaries to assessment areas also depended on the amount of beach or salt marsh habitat in each assessment area, or whether assessment areas were estuarine or inter-estuarine (Table 4). For example, communities were more likely to mention general commercial or industrial benefits of beaches in assessment areas with a higher percent cover of beach habitat (Supplementary Materials S7). Agricultural, food gatherer, spiritual/ceremonial, or humanity benefits of salt marsh had greater relative importance in places with a higher percentage of salt marsh habitats, and general government/residential benefits of salt marsh were less likely to be mentioned in inter-estuarine assessment areas.

Finally, we evaluated whether any socio-economic or ecological characteristics of coastal communities could explain significant differences in the final relative importance of ecosystem services attributes mentioned in documents (Table 3). Community characteristics explained more than 36% of variability in beneficiary profiles among assessment areas, and in particular more than 50% of variability for beaches and fish habitat (Table 4). Regional differences significantly explained variation in relative importance of ecosystem services attributes associated with each type of ecosystem. For example, aesthetic viewscapes and natural site appeal were more likely to be mentioned in the North Shore, especially in association with rocky shore, salt marsh, or tidal flats (Supplementary Material S4). Fish/shellfish tended to score higher relative importance for assessment areas in Cape Cod, while birds and other fauna, as well as flora, tended to have lowest relative importance in the areas surrounding the Boston Metro area, including in the Lower North Shore and South Shore. Water quality and quantity also tended to score relatively higher in Cape Cod, but water movement and navigability tended to have greater importance in the North Shore and Boston Metro area, particularly in association with fish habitats.

Significant variability in the relative importance of ecosystem services attributes of most ecosystems was also explained by whether communities had substantial manufacturing (Table 4; Supplementary Material S6), with manufacturing communities tending to score lower relative importance for fish and shellfish, but more likely to mention aesthetic viewscapes (Supplementary Materials S7). Farming/fishing communities, quantified as the percent population employed in agriculture/fishing, generally scored a higher relative importance to fish and shellfish from tidal flats and water navigability through seagrass. Communities with a high population percentage in transportation/utilities were generally more likely to score higher importance to climate regulation, mitigation of extreme weather, and wind. Population over 65 also explained significant variability in ecosystem services attributes associated with benthic habitats, rocky shore, or tidal flats, with flora and fauna communities tending to score higher relative importance in older communities.

4. Discussion

4.1. Beneficial Uses of Coastal Habitats to Coastal Communities

Overall, the use of the NESCS Plus Classification System and FEGS Scoping Tool, in combination with an automated keyword analysis, provided us a powerful, consistent, and structured way to identify ecosystems, beneficial uses, and ecosystem services attributes of greatest relevance for each assessment area community. Differences in the relative importance of ecosystem services attributes to communities was driven in large part by the profile of natural resource users, or beneficiaries, in each community, which ultimately determines the relative importance of ecosystem services attributes to the people in that community. Stakeholders will inevitably differ in how they experience, use, and value ecosystem services (Diaz et al. 2010, Chan et al. 2012), and barriers to resource planning can arise if the different values of stakeholders are not adequately considered (Martin-Lopez et al. 2012).

MassBays assessment areas were most consistent in how they identified the relative importance of different users of beach and salt marsh, with people who care, general references to government and residents, and experiencers/viewers among the most commonly mentioned beneficiaries. Although experiential users of ecosystem services, such as wildlife or scenic viewers, are broadly common (Darvill and Lindo 2016), high scores for government, residents, and people who care may to some degree reflect broad generalizations in document language, such as ‘residents of our community’, or ‘future generations’, that could not be categorized to more specific user groups. The top beneficiaries were fairly consistent across assessment areas, although mentions in particular of aquaculture, agriculture or farming, and wading or swimming, were more common in some areas than others, likely reflecting local scale differences in opportunities and access to ecosystems (Syrbe and Walz 2012). Furthermore, commonly mentioned beneficiaries were often associated with different ecosystems by different assessment areas, such as the importance of beach versus rocky shore to experiencers/viewers or the importance of seagrass versus tidal flats to aquaculture.

Although the relative importance of attributes to users differed, our document analysis highlights a surprising degree of shared interests, even among seemingly disparate user groups. Shared interests may point toward opportunities for consensus or coalitions (Diaz et al. 2010, Simpson et al.2016). Beneficiaries were identified throughout documents in association with a wide variety of ecosystem services attributes, beyond the more obvious associations such as fish and shellfish to commercial or recreational fishermen, aesthetic appeal to artists, and water navigability to boaters. Composite ecosystem elements conferring site appeal, such as naturalness and aesthetic viewscapes, water, fish and other fauna, were universally associated as important to some degree for almost all beneficiaries. Other less common attributes, such as natural materials, had significant importance to only certain user groups, such as artists or industrial processors.

In the FEGS Scoping Tool framework, ecosystem services attributes can attain high priority scores by either being of moderate importance to many different kinds of beneficiaries, or of substantial importance to a few high-ranking beneficiaries (Sharpe et al. 2020). In our analysis, the top most important ecosystem services attributes were generally of broad importance to many different kinds of users, and included naturalness, water movement and navigability, water quality and quantity, aesthetic viewscapes, availability of land for development, flood mitigation, and birds. Flora was also commonly mentioned for salt marsh and seagrass. Of the top attributes, edible or commercial fish and shellfish had the fewest identified users, with more than 50% of their final score being attributed to the needs of commercial or recreational fishermen. However, general references to fish and shellfish were of broad importance to many different kinds of beneficiaries, but we note may include different species or size classes than the kinds of fish relevant for edible or commercial uses, such that different metrics might be needed to meaningfully communicate assessment and monitoring (Ringold et al. 2022).

The relative ranking of ecosystem services attributes was fairly consistent across assessment areas for beaches, fish habitat, and salt marsh, with naturalness being consistently important for beaches and salt marsh. In our study, naturalness broadly captured document language related to composite ecosystem attributes that reflect good ecosystem condition, structure, and function, which can have indirect benefits to many different user groups by setting the supportive foundation for fauna, flora, and other valued ecosystem services attributes (van Oudenhoven 2012). Naturalness may also reflect the degree to which habitats are unmodified by human activities, which may have more direct value to experiential users (e.g., hikers, residents, or educators) who wish to feel connected with nature (Sanna and Eja 2017). For fish habitat, fish, shellfish, and water movement were consistently scored as most important. Similar to naturalness, water movement may be interpreted as a desire for a hydrologically functioning river system (Boon 2000), in addition to having aesthetic value for experiential users or more direct relevance to operation of dams, boats, and ships (Pflüger et al. 2010). For many assessment areas, relatively fewer ecosystem services attributes were mentioned in association with tidal flats, and as a result commonly mentioned ecosystem services attributes such as naturalness, fish and shellfish, and water movement tended to have a higher relative importance compared to other habitats where a greater diversity of attributes were mentioned. Although the importance of tidal flats as a component of coastal ecosystems is generally acknowledged, ecosystem services benefits are broadly overlooked (Kim 2013).

4.2. Community Characteristics Inform Ecosystem Services Priorities

Regional differences explained a large proportion of the differences among assessment areas in the relative frequency with which different coastal ecosystems were mentioned, the relative importance of different beneficiary groups, and the relative importance of different ecosystem services attributes. Regional identities, driven by shared customs, shared histories of landuse planning or political decisions, or regional promotion and marketing, can exert a strong influence on individual perceptions and priorities (Cook et al. 2012).

After accounting for regional differences, a significant portion of variability in beneficiary and ecosystem services attributes profiles was explained by characteristics of individual communities, in particular the industry profile of each community and whether there was a high proportion of the population employed in manufacturing, agriculture, or fishing industries. These results are consistent with other studies that have found individual ecosystem services preferences to be significantly different for stakeholders more directly dependent on natural resources for their livelihoods (e.g., fishermen, agriculture), which in particular tended to have a stronger preference for provisioning services, than stakeholders with more diverse professional activities (Cabellero-Serrano et al. 2017, Tauro et al. 2018, Lefeuvre et al. 2022). Additional strong predictors of ecosystem services attribute use included the age distribution of the population (e.g., retirement communities) and whether populations were dense urban minority populations. Urban-rural gradients in ecosystem services preferences have previously been identified for many kinds of ecosystems (Martin-Lopez et al. 2012, Williams et al. 2017, Lapointe et al. 2019), with older rural populations general expressing higher preference for provisioning services than younger urban populations, which generally expressed greater preference for cultural or regulating (e.g., air quality, aesthetics) ecosystem services.

The amount of coastal habitat cover in each assessment area also significantly impacted how coastal communities talked about coastal habitats. Beach and fish habitat had the greatest diversity of beneficiaries associated with them, whereas fewer unique users were associated with benthic habitats, rocky shore, or seagrass. Benefits of salt marsh, fish habitat, rocky shore, and tidal flats were more likely to be mentioned in assessment areas where these habitats were present in substantial amounts. Presence of conspicuous biodiversity can make people more appreciative of green space benefits (Gunnarsson et al. 2016), with visible green and blue spaces conferring higher property values (Mithal and Byahut 2016) and mental health benefits (Velarde et al. 2007, Nutsford et al. 2016). This can raise a challenge for ecosystem management, however, as “invisible” habitats may be of high ecological importance (Barañano et al. 2022). Indeed, mentions of seagrass in community documents were universally uncommon across all assessment areas, even in assessment areas with larger amounts of seagrass. Yet seagrass is a high priority habitat for the estuary management program (MassBays 2019), making outreach on the benefits of seagrass to local communities all the more important. By contrast, mentions of beaches, one of the potentially most conspicuous natural coastal habitats to people, were widely common in documents regardless of the particular amount of beach in the assessment area.

4.3. Implications for Restoration Planning and Implementation

The MassBays comprehensive management plan has recognized the importance of tying restoration of ecological habitats and water quality to desired beneficial uses by coastal communities and sustainable delivery of ecosystem services (MassBays 2019), including recreational and commercial fishing and harvesting, water quality regulation (including nutrient reduction), habitat for bird fauna, shoreline stabilization, native fauna, recreational uses, residential and commercial development, storm protection, and flood control. Our document analysis identified additional ecosystem services attributes that may be of high relevance to communities, in particular i) aesthetic aspects of site appeal including naturalness and viewscapes, ii) fish and shellfish biodiversity, including charismatic and rare species, not just recreationally or commercially important species, iii) flora biodiversity, including charismatic and commercially important species, iv) quality of sediments and soils, v) water movement and navigability, for both commercial transportation and recreational boaters, as well as for public water/dam operations, and vi) availability of natural materials, such as shells, for recreational, artistic, and commercial uses.

Almost universally across documents, top users of coastal habitats were identified as people who care about existence value and general references to government or residents. This may reflect the true nature of how communities perceive natural resources in terms of existence value, that is, the people or town or residents benefit from the satisfaction of knowing these ecosystem services exist (Davidson 2013). Commonly shared values, like existence value, can indeed be overlooked in restoration planning, leading to undesirable outcomes (Chan et al. 2012). Alternatively, this result could reflect lack of specificity in written language in community documents, where terminology such as “the town”, “the people”, and “the community” are common. Beneficiary-focused ecosystem services approaches, like the FEGS framework, can help to reduce ambiguity by directly linking ecosystem services attributes to the users who care about them (DeWitt et al. 2020). A highly relevant attribute, such as water quality, can mean very different things and require different metrics for quantification or monetary valuation, such as for existence value, commercial uses, aesthetic views, or swimming. Attributes such as ‘naturalness’ may be particularly challenging to quantify, and may comprise multidimensional components, such as greenness, biodiversity, visual complexity, and lack of man-made objects (reviewed in Reyes-Riveros et al. 2021). Identifying metrics that closely represent how beneficiaries use natural ecosystems helps to ensure that stakeholders and the public are engaged in decision processes, and benefits of environmental management decisions are clearly communicated (Wainger and Mazzotta 2011).

Discussions about historic losses of coastal habitats and their benefits are helping to support restoration target setting for MassBays to restore seagrass, prevent further losses of salt marsh and tidal flats, and improve water quality, which are essential to maintain and restore valuable ecosystem services (MassBays 2023). Priority beneficiaries and ecosystem services attributes identified from community planning documents were presented to the MassBays NEP Science Advisory Committee to provide a starting point for discussion of essential ecosystem services to consider when setting restoration targets, planning restoration projects, and identifying metrics for monitoring of restoration outcomes that resonate most broadly with stakeholders. Because the document analysis is linked to individual assessment area communities, unique differences among communities can be extracted to help support implementation of local restoration projects or communicate potential benefits of restoration in ways that resonate with each local community. Quantification of historic losses or gains of relevant ecosystem services attributes in individual assessment areas can further help support implementation of local restoration projects to restore past losses or protect recent gains, or to provide a framework for comparing future alternative restoration options (Branoff et al. 2023).

4.4. Limitations and Opportunities

Although automated content analysis is a well-developed technique for extracting information from written documents (Cushing 2017, Li et al. 2023), interpretation ultimately depends on careful development of word lists to optimize recall and precision (Ball 1994). Recall is based on the concern ‘you don’t know what you’re missing’, whereas precision evaluates whether extracted information is actually tagged correctly. The struggle is in developing a corpora of keywords that is not so precise that it is inflexible to linguistic or typographic variations, but not so flexible that it frequently mis-classifies the document’s intent. We addressed these concerns by developing keyword lists through iterative comparisons of automated searches with manual reads. Linguistic and typographic differences among documents, including parsing differences arising as portable document format (pdf) documents are converted to text, make it virtually impossible to achieve perfect line-by-line matching. Therefore, we aimed for overall consistency of concepts mentioned in a document and relative frequencies across documents, rather than the frequency of mentions (e.g., sentence counts) within a document, which can be susceptible to linguistic and typographic choices (Ball 1994).

Use of a classification system was essential in helping to provide a unifying language, improve recall and precision, simplify and structure keyword coding, and rigorously define the scope of the analysis (Finisdore et al. 2018). Although we used NESCS Plus to identify and classify ecosystem services, the document analysis could easily be adapted to other ecosystem services classification systems, such as Common International Classification of Ecosystem Services (CICES; Haines-Young and Potschin, 2013, 2018), Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES; Christie et al. 2019, IBPES 2018), or the Millennium Ecosystem Assessment (MEA 2005), by reclassifying the components based on a crosswalk of categories across the different frameworks (e.g., United Nations 2021).

NESCS Plus has advantages over other classification frameworks in that it explicitly distinguishes intermediate from final ecosystem goods and services, which reduces the likelihood of overlapping categories or double counting, such as the intermediate regulating service of water purification and the final service of freshwater supply for drinking (Newcomer-Johnson et al. 2020). As such, NESCS Plus does not explicitly include categories for regulating services. In our study, however, we found language characteristic of intermediate regulating services to be fairly common in documents, and thus added this category so that important concepts would not be overlooked. It could be inferred that these regulating services (e.g., contaminant filtration) contribute to one or more final ecosystem services (e.g., water quality, edible fauna), but without additional language, could not be positively classified as such.

Another advantage of NESCS Plus is the potential for interoperability with other related FEGS tools (Newcomer-Johnson et al. 2020), illustrated explicitly in this study by its use in combination with the FEGS Scoping Tool framework (Sharpe 2021). The initial document-based beneficiary profiles and priority ecosystem services attributes provide an initial assessment that could be further developed using the FEGS Scoping Tool to modify importance scores based on local knowledge, explicitly link beneficiaries and attributes to public and private stakeholder organizations, or overlay weights to different stakeholders based on factors such as their proximity to restoration activities or likelihood to be impacted by decisions (Sharpe et al. 2020; Hernandez et al. 2022). Once categories of priority ecosystem services attributes are identified, a five-step process that leverages NESCS Plus to delineate ecosystems, specify beneficiaries, and specify attributes, has been developed to help identify FEGS metrics and data to measure and monitor progress, with examples for many different kinds of ecosystems including estuaries (Ringold et al. 2020). NESCS Plus categories can also be leveraged within the EcoService Models Library (ESML; US EPA 2022c) to identify models to translate attributes of ecological condition into measures of ecosystem services, or to model and compare projected outcomes of alternative management options. NESCS Plus has also recently been used to facilitate natural capital accounting for developing supply and use tables, to give a more complete picture of a local area’s environmental-economic trends (Warnell et al. 2020).

5. Conclusions

Our study reinforces that communities are not one size fits all. In decision making in general, the most important initial step includes understanding the objectives, or “what really matters”, about a decision at hand (Gregory et al. 2012). Though demonstrated here for estuarine management, the keyword hierarchy and FEGS text mining approach have broad applicability and transferability to other environmental management scenarios. A beneficiary-centric approach to ecosystem services, in particular by leveraging classification systems, can help ensure a broad suite of ecosystem services are considered, and that stakeholders are not overlooked when planning restoration projects, monitoring outcomes, and communicating benefits (DeWitt et al. 2020). The FEGS Scoping Tool, in particular, provides a structured approach for identifying and prioritizing ecosystem services attributes, based on the foundational idea that stakeholders have different needs, values, and concerns (Sharpe et al. 2020). As such restoration planning, monitoring, and environmental outreach can only be partially generalized when considering ecosystem services benefits, and ultimately should be tailored to local needs.

To some degree, community perceptions of ecosystem services can be generalized by regional differences and socio-ecological characteristics. Social and demographic characteristics, local dependency on natural resources, and local ecological characteristics have previously been used to classify communities nationally based on their local community priorities in ways that ultimately can have impacts on the sustainable well-being of communities (Fulford et al. 2015), including economic well-being, social cohesion, and health outcomes, and resilience to natural disasters (Summers et al. 2022). Classification approaches to generalize community needs may be invaluable when decisions are large-scale, making it impractical to engage with all local communities, or to serve as a starting point, or ‘strawman’, to jump-start local discussions.

A foundational component of ecosystem restoration is clarifying social and well-being benefits to people, including through delivery of ecosystem services, alongside ecological recovery goals (Gann et al. 2019; Pouso et al. 2020; Jackson et al. 2022). Coastal communities, and a diversity of stakeholder groups, clearly have a stake in maintaining, managing, and restoring coastal habitats, as indicated by community planning documents. Connecting ecosystem restoration to essential ecosystem services local communities care about, and ultimately community well-being outcomes, can bring to light shared values among disparate stakeholders, confer a sense of community ownership, and greatly improves the chances that restoration projects will be accepted by the public and ultimately be successful (Alexander et al. 2016; DeAngelis et al. 2020).