Advancing Coastal Habitat Resiliency Through Landscape-Scale Assessment

ABSTRACT

Coastal areas are among the most biologically rich regions in the world. They provide tremendous benefit to people through ecological services like pollution filtration, carbon sequestration, and storm protection. However, human activity has decreased the ability of ecosystems to provide these services, and in many coastal areas, restoration is not as effective as it could be. Limited success is largely attributed to vague goals and a lack of holistic, science-informed planning. Quantifying coastal habitat health can aid in setting better restoration goals and lead to greater restoration success. The Long Island Sound Study National Estuary Program conducted the first comprehensive look at coastal habitat health for the Long Island Sound Estuary. In this study we also reviewed landscape-scale habitat assessment progress within other coastal programs. Results can be applied to establish ecosystem health-oriented goals that improve habitat function and resilience across coastal areas.

KEYWORDS:

• ecosystem services
• EPA National Estuary Program
• estuarine habitat
• landscape-scale restoration
• strategic coastal management

Introduction

Coastal ecosystems are among the most valuable habitats on earth, providing numerous benefits to people and wildlife (Barbier et al. 2010; Gedan, Silliman, and Bertness 2009). The economic value of the ecosystem services that Long Island Sound (LIS) provides was recently estimated to be between $17 billion and $36.6 billion per year (Kocian et al. 2015). However, estuaries across the globe are under tremendous pressure, which compromises their ability to provide such services (Barbier et al. 2010; Nellemann et al. 2009; Valiela 2006). For example, over the last 130 years, Connecticut lost 5,262 tidal wetland acres (21.3 km²), estimated to be a loss of $61.5–$406.5 million per year in wetland benefits (Basso et al. 2015; Kocian et al. 2015).

In order to maintain the benefits that coastal habitats provide, factors contributing to overall habitat health must be well understood. This fundamental understanding and a way to measure habitat health improvements can engage political and public interest and better inform recovery actions (Boulton 1999; Sherwood et al. 2016). Unfortunately, the success of restoration is hindered due in part to vague goals and a lack of science in the planning and goal setting process (Palmer 2008; Silliman, Grosholz, and Bertness 2009; Duarte et al. 2009). The assessment presented here provides a quantitative understanding of ecosystem condition. This understanding can be used to set science-informed goals and priorities and can lead to better restoration outcomes.

LIS is one of the 28 estuaries designated as nationally significant under the U.S. Environmental Protection Agency (USEPA) National Estuary Program (NEP). The habitat condition assessment presented herein was completed in the Connecticut portion of the LIS NEP (Figure 1). NEPs provide ideal locations to implement landscape-scale assessments because they are dedicated to improving and monitoring overall habitat health. NEPs support development and implementation of locally based Comprehensive Conservation and Management Plans (CCMPs). The LIS CCMP (Long Island Sound Study 2015) identifies specific ecosystem recovery targets and recommends developing an index of current overall habitat condition. This index would establish a baseline condition and make it more feasible to track the effectiveness of restoration toward achieving specific ecosystem recovery targets.

Although the LIS CCMP lays out these recommendations, a gap exists at both the local and national levels. Nationally, a coastal condition report (USEPA 2012) and a coastal wetland assessment are complete (Dahl and Stedman 2013), however neither addresses the suite of coastal habitats that affect overall estuary health. Additionally, few comprehensive health assessments exist at the local NEP level. For example, the EcoHealth Report Card for the LIS includes a health score that is solely reflective of water quality and does not include an assessment of LIS's priority terrestrial or aquatic habitats (Save the Sound 2016).

To address this gap, we first developed a landscape-scale framework for assessing priority coastal (terrestrial and aquatic) habitats of LIS NEP. Second, we completed an assessment of the overall health of the LIS NEP. Third, we reviewed the state of landscape-scale assessments within five other coastal programs – LIS NEP, Narragansett Bay NEP, San Francisco Estuary NEP, Tampa Bay NEP, and the Chesapeake Bay Program. Last, we offer recommendations on how the landscape-scale assessment framework developed here could be applied elsewhere to establish a baseline condition, facilitate strategic goal setting, and help assess progress toward achieving desired outcomes.

Methods

Habitat assessment framework

To develop the habitat assessment framework, we assembled a diverse panel of coastal habitat experts from state, federal, nonprofit, and academic institutions. They focused on priority habitats selected by the LIS NEP (Latimer et al. 2014; Long Island Sound Study 2014). The panel included all Long Island Sound priority habitats for which 3 or more metrics could be established using currently available data (Table 1). In identifying metrics, or indicators of health, the panel used best available information, professional judgment, and peer-reviewed literature (Tables 2–6).

Table 1. LIS priority habitats included in this assessment.

Habitat	Definition
Eelgrass (Zostera marina)	Rooted underwater marine plant growing along the LIS coast. Primarily found in the Eastern Sound with isolated patches in the Central Sound, although once common throughout LIS.
Estuarine Embayments	Semi-enclosed areas along the LIS's edge that have a defined mouth and freshwater inflow. Names typically include the words cove, river, creek, and harbor. Average depths are shallower than the open LIS and restricted flow causes greater sedimentation.
Tidal Wetlands	Transitional zone between land and submerged systems, these areas are dominated by rooted plants flooded by the tide. Locations are determined by geology and topography.
Riverine Migratory Corridors (RMC)	River systems that drain to the LIS often bordered by trees and wetlands. Migratory fish use rivers to spawn. The majority of rivers flowing into LIS are in Connecticut and Westchester and Bronx Counties, NY.
Coastal Forests	All forested areas within the LIS coastal boundary (Figure 1). Defined in terms of location and vegetation. Dominated by species including maple, oak, cedar, pine, and beech. Occur in small patches throughout the LIS area with a few large swaths remaining (> 100 acres (0.4km^2)).

The panel considered three factors in defining habitat health: (1) ecological integrity of the habitat (habitat structure and function), (2) resilience to stress (ability to recover from disturbance), and (3) ability to provide services (e.g. provide habitat, filter pollutants) (Boulton 1999; Folke et al. 2004; Karr 1996). Karr et al. (1986) states, “a biological system … can be considered healthy when its inherent potential is realized, its condition is stable, its capacity for self-repair when perturbed is preserved, and minimal external support for management is needed.” As the field of ecology advances, ecological health is viewed as a range of acceptable conditions rather than a static point, with the goal of shifting ecosystem composition, structure, and function into a range that is more desirable than the current degraded state (Palmer 2008; Duarte et al. 2009). To make the framework and assessment broadly applicable, the panel identified metrics that are capable of highlighting habitat condition, applicable to a wide range of systems, and cost effective to measure and monitor (Boulton 1999; Karr 1991; Niemeijer and Groot 2008).

Habitat assessment

The LIS NEP is divided into three geographic basins for the purposes of management and communication (Figure 1). The basins differ across factors including human density, water quality, current, and sediment size (Long Island Sound Study 2015). Using the framework, priority habitat types were assessed in these three basins with associated watersheds to maintain consistency with other LIS publications and management activities.

Figure 1. LIS basins used for habitat status analysis.

Data pertaining to each metric for each priority habitat were compiled from existing data sources and analyzed using The Nature Conservancy's open source Miradi software indicator monitoring component (Benetech 2008). In order to derive a habitat score, a data point for each metric within each priority habitat was mathematically re-indexed using value ranges unique to that metric. These value ranges are habitat condition “cut-points” and are defined in Tables 2–6 for each metric. The re-indexing generated a value that was assigned to one of the following bins: poor (0–25), fair (25–50), good (50–75), and very good (75–100). This allowed a standardized means of analyzing metrics with varying units and ranges of habitat condition. Each metric was then weighted based on its contribution to the overall habitat condition of each priority habitat type. Classification cut points and weighting were set specifically for the Long Island Sound ecosystem. The ranges and weightings can be adjusted to best suit the local ecosystem where the framework is applied.

Table 2. Eelgrass habitat assessment metrics. Density and Bed Size Metric corresponds to eelgrass surveys in LIS as: VERY LOW = low; LOW = lower half of medium, MEDIUM = upper half of medium, HIGH = high (MacLachlan 2014; Tiner, McGuckin, and MacLachlan 2013).

Metric	Metric Unit & Weight for Scoring	Cut Points	Metric Description
Density and Bed Size^1	Percentage of area within density and bed size categories relative to the total area of eelgrass weight = 33.%	poor: 0 – 35% M or H with < 5% in H; remainder in V or L	Status of a bed is defined by shoot density per m^2 and bed size: Density and bed size categories: (V) VERY LOW: ≤ 24 stems & ≤ 10m^2; (L) LOW: 24-299 stems & 10m^2; (M) MEDIUM: ≥ 300 stems & ≤ 10m^2; (H) HIGH: ≥ 300 stems & ≥ 10m^2.
		fair: 39–59% M or H with < 10% in H; remainder in V or L
		good: 60–75% M or H with ≥ 10% in H; remainder in V or L
		very good: 75–100% M or H with ≥ 10% in H; remainder in V or L
Change in Area in LIS^1	Change in area of eelgrass relative to 2006 (%) weight = 33.%	poor: ≤ –20	Change in area relative to the 2006 baseline of 1,904 acres (7.7km^2). To fall in the very good category overall extent must increase by 5% in each basin. Eelgrass beds naturally fluctuate in size. Ideally, five year averages of annual surveys are evaluated.
		fair: –19 to –6
		good: –5 to 5
		very good: > 5
Epiphytes & Epifauna^2	Degree of colonization of eelgrass blades weight = 33.%	poor: Major, heavy coverage	This indicator is a new addition to eelgrass assessment in LIS. Current epiphyte and epifauna coverage was compared to ideal coverage for LIS.
		fair: Moderate, some coverage
		good: Few, slight coverage
		very good: Very few, little to no coverage

¹ 2006 & 2012 USFWS eelgrass survey data (Tiner, McGuckin, and MacLachlan 2013).

² 2006 & 2012 USFWS eelgrass survey data, MacLachlan field records.

Table 3. Embayment habitat assessment metrics. All data came from a rapid assessment of embayments conducted in late summer, 2011-2013; and are thus directly comparable among all sites (Vaudrey 2015; Vaudrey et al. 2013b).

Metric	Metric Unit &Weight for Scoring	Cut Points	Metric Description
Dissolved Oxygen^1^,^2	Concentration (mg/L) weight = 25%	poor: 0 – 1.99	Average water column value for dawn and midday, collected in late summer. If bottom value is > 4mg/L, samples are from surface (0.5m deep) and bottom (0.5m off bottom). Otherwise, samples collected every 0.5m.
		fair: 2 – 3.49
		good: 3.5 – 4.79
		very good: ≥ 4.8
Chlorophyll a^2	Concentration (µg/L) weight = 25%	poor: 20 – 1000	Midday summer chlorophyll concentration collected from surface (0.5m deep) and bottom (0.5m off bottom). Averaged a minimum of three stations.
		fair: 11 – 19
		good: 5 – 10
		very good: 0 – 5
Macroalgae^3	Summer coverage of hard surfaces by macroalgae indicative of high nutrient loads^4 and reports of bloom frequency (% / bloom frequency) weight = 25%	poor: 71 – 100 / persistent or periodic	Relative coverage of a rack deployed in 3 sites (1 embayment per basin) during late summer, for three weeks.
		fair: 51 – 70 / persistent or periodic
		good: 30 – 50 / episodic
		very good: < 30 / few to none
Eelgrass Loss in Embayments	Area of suitable habitat colonized and interannual stability of colony (%, variability) weight = 25%	poor: 0	Current coverage was compared to predicted coverage. Suitable habitat estimated using a GIS site suitability map based on depth, light, temperature, sediment organic content, and grain size (Tiner, McGuckin, and MacLachlan 2013, Vaudrey et al. 2013a).
		fair: 1 – 49, high
		good: 50 – 79, moderate
		very good: 80 – 100, low

¹ Long Island Sound Study (2017).

² National Coastal Assessment (USEPA 2012).

³ NOAA National Estuarine Eutrophication Assessment (Bricker et al. 2007; Bricker, Ferreira, and Simas 2003).

⁴ Algae indicative of high nutrient inputs include Gracilaria sp., Neosiphonia sp., and green algae.

Table 4. Tidal wetland habitat assessment metrics.

Metric	Metric Unit & Weight for Scoring	Cut Points	Metric Description
Size^1	Average of the Tidal Wetland Units (acres)weight = 25%	poor: < 12	Tidal wetland polygons were extracted from the NWI data and aggregated into Tidal Wetland Units - individual tidal wetland pieces separated by less than 100m (Meiman et al. 2012). From these, an average Tidal Wetland Unit size was derived.
		fair: 13 – 40
		good: 41 – 399
		very good: > 400
Impervious Coverr^2	Tidal Wetland Units in HUC12 basins with > 10% impervious surface cover (%)weight = 25%	poor: ≥ 65	HUC12 basins with > 10% impervious cover were identified and intersected with Tidal Wetland Units; the resulting area of tidal wetlands was expressed as a percentage of total Tidal Wetland Unit area.
		fair: 40 – 64
		good: 16 – 39
		very good: ≤ 15
Open Water^3	Percent open water at low tide (%) weight = 50%	poor: 20 – 40 (> 40 = 0 points)	For Tidal Wetland Units > 10 acres (0.04km^2), 25% were randomly sampled in each LIS basin. Open surface water on these areas was analyzed via photos. Field checks verified surface conditions. Area of open water was calculated as a percent of total area sampled.
		fair: 15 – 20
		good: 10 – 15
		very good: 0 – 10

¹ USFWS National Wetlands Inventory, Tidal Wetlands.

² UConn CLEAR 2010 Impervious Surface Cover for HUC12 Basins; Tidal Wetland Units. HUC12 is the USGS national classification standard for sub watersheds.

³ Tidal Wetlands Units; 2010 Tide Controlled Coastal Infrared Aerial Orthophotography; field surveys.

Table 5. Riverine migratory corridor (RMC) assessment metrics.

Metric	Metric Unit & Weight for Scoring	Cut Points	Metric Description
Connectivity^1	Historic run currently available (%) weight = 60%	poor: ≤ 35	RMC lengths classified as “open” were counted from the downstream terminus to the first impassable obstruction. Total length classified as “open” was divided by the total available RMC length in miles.
		fair: 36 – 55
		good: 56 – 75
		very good: ≥ 75
Impervious Cover^2	RMC within a high impervious cover zone (%)weight = 10%	poor: ≥ 65	RMC lengths within HUC-12 basins with >8% impervious cover were compared to the total RMC length.
		fair: 40 – 64
		good: 16 – 39
		very good: ≤ 15
Water Quality^3	RMC classified as EPA “Not Impaired” (%) weight = 30%	poor: 0 - 40	RMC data corresponding to rivers with EPA Assessed Water Quality data were identified as Assessed RMC miles. These were classified as “Impaired” or “Not Impaired” (i.e., suitable for fish consumption, recreational, or habitat uses). “Not Impaired” RMC miles were expressed as a percentage of the total “Assessed RMC” miles.
		fair: 41 – 70
		good: 71 – 84
		very good: ≥ 85

¹ CT DEEP RMC data layer, updated in 2013 by USFWS, CT DEEP, and TNC.

² UConn CLEAR 2010 Impervious Surface Cover for HUC12 Basins (HUC12 is the USGS classification standard for sub watersheds.); CT DEEP RMC data.

³ USEPA State Water Quality Assessment, Connecticut 305b Assessed Rivers, Lakes, and Estuaries Water Quality for 2012. Drinking water categories excluded.

Table 6. Coastal forest habitat assessment metrics. All metrics used 2010 UConn CLEAR land cover data, 30 m resolution (CLEAR 2012). Fragmentation was determined using the UConn CLEAR Landscape Fragmentation Tool (CLEAR 2007). Only forests within the defined coastal boundary were assessed.

Metric	Metric Unit & Weight for Scoring	Cut-points	Metric description
Size	Average forest block size (acres) weight = 33.%	poor: <50	Deciduous and coniferous forest types were extracted from land cover data, combined, and reprocessed into forest blocks. A forest block includes all forest land cover cells that touch other forest cells on at least one side or corner. Blocks are separated by a minimum of one 30 m by 30 m pixel on all sides and at each vertex. Average block size was then derived.
		fair: 51–150
		good: 151–249
		very good: ≥250
Connectivity	Forest edge that borders developed land (%) weight = 33.%	poor: ≤86–100	Computed forest block perimeters (interior and exterior) connected to developed land cover (developed or turf categories). Values were expressed as a percentage of the total forest block perimeters.
		fair: 51–85
		good: 21–50
		very good: ≤20
Fragmentation	Total forest area in basin designated as core forest area (%) weight = 33.%	poor: <30	UConn CLEAR Landscape Fragmentation Tool (v2) (CLEAR 2007) was run on forest blocks to classify them as patch, buffer, perforated, or core (most to least fragmented). Core forests ≥ 100 acres (0.4 km^2) with a 50 m buffer.
		fair: 31–70
		good: 71–89
		very good: >90

Programmatic review of landscape-scale assessments

A review was conducted of five leading national estuary and coastal programs in order to assess the status and application of landscape-scale assessment frameworks like the one developed in this study. The review consisted of conversations with program staff and a review of published literature and other publically available information for each program, including program websites and CCMPs. Because the purpose was to assess publicly available information, anecdotal and non-published data are not included. Specific focus was given to the main website for each program, as that is the location the public likely would check first for habitat health information.

Results

Habitat assessment framework

The metrics indicating health for each priority habitat type, along with a description, classification cut-points (e.g., the range of values that comprise poor, fair, good, and very good), and metric weighting are presented in Tables 2–6. In the sections below, we provide a description of the metrics selected as indicators of habitat health.

Eelgrass

Eelgrass metrics were selected based on common methods for measuring eelgrass health in LIS and other estuaries (Hauxwell, Cebrián, and Valiela 2003; Tiner, McGuckin, and Herman 2013). Reduction in density and bed size, colonization area, and amount of light reaching the blades can all negatively affect eelgrass habitat (Hauxwell, Cebrián, and Valiela 2003). As such, the metrics selected for the LIS NEP include density (# of stems/m²) and bed size (m²), change in overall area (m²) relative to a previous timeframe, and degree of colonization of epiphytes and epifauna. The density of eelgrass beds in the LIS NEP depends, in part, on light. Eelgrass beds receiving more light have up to 1,200 shoots/m² while beds with low light have <50 shoots/m² (Krause-Jensen et al. 2000). Similarly, epiphyte coverage hinders eelgrass success by blocking light to shoots (Brush and Nixon 2002). In extreme cases, epiphytes in LIS completely cover and weigh down shoots.

Embayments

The metrics selected as indictors of embayment health are broadly recognized as indicators of decline in water quality and embayment health across the country (Bricker et al. 2007; USEPA 2012). Chlorophyll A and macroalgae are early indicators of increased nutrient concentrations. Dissolved oxygen and loss of eelgrass are secondary symptoms of more serious eutrophication problems. In LIS, decline in eelgrass is linked to nitrogen enrichment (Bricker et al. 2007).

Tidal wetlands

The metrics chosen as indicators of tidal wetland health reflect the major symptoms of stressed marshes. Open water on the marsh was assigned the greatest weight. Saltwater intrusion was the leading cause of coastal wetland loss in the last two national assessments of wetland condition (Dahl and Stedman 2013). Open water can lead to loss of vegetated marsh (Kirwan et al. 2010; Tiner, McGuckin, and Herman 2013) indicating submergence and potentially a regime shift (Folke et al. 2004; Roman 2017). Open water ranges and cut-points follow recommendations for New England marshes, with “very good” aligning with the average amount of permanent open water at ≤9% of an unditched New England marsh (Adamowicz and Roman 2005). Impervious cover (IC) surrounding the marsh is linked to marsh impairment (Arnold and Gibbons 1996; Schueler 2003). IC thresholds for rivers, discussed below, also apply to wetlands (Arnold and Gibbons 1996) and were used to guide the development of the IC metric. Size was included as an indicator because coastal systems that are larger in size and less fragmented are better able to recover from disturbance and support more biologically diverse communities than smaller, more fragmented habitats (Meiman et al. 2012; Thrush et al. 2008).

Riverine migratory corridors

For the RMC assessment, connectivity was assigned the greatest weight. Decreased connectivity alters a central RMC function, serving as a thoroughfare for diadromous fish to migrate and spawn (Pringle 2003). The other two metrics, water quality and IC, are common indicators used to assess river health (Schueler, Fraley-McNeal, and Cappiella 2009). Surrounding IC is linked to RMC impairment (Arnold and Gibbons 1996; May et al. 1997; Schueler, Fraley-McNeal, and Cappiella 2009). In Connecticut, the least disturbed streams are in areas with less than 4% IC in the upstream watershed; 8% IC begins to negatively affect rivers; at 12% IC rivers no longer support aquatic life goals (Bellucci, Becker, and Beauchene 2011). Other work corroborates these cut-offs and indicates significant impairment starts at approximately 10% IC (Arnold and Gibbons 1996; May et al. 1997).

Coastal forests

Indicators selected to measure coastal forest health include size, connectivity, and fragmentation. Changes in these metrics can affect overall forest health and composition of species like forest interior birds (Tischendorf and Fahrig 2000; Thompson et al. 2009; Askins, Philbrick, and Sugeno 1987). For example, a steep drop in forest interior bird species diversity occurs in forest blocks smaller than about 125 acres (0.5 km²)(Askins, Philbrick, and Sugeno 1987). In the framework, the connectivity metric indicates the fraction of forest perimeter adjacent to developed land. The fragmentation metric reflects forest block shape and uses the UConn CLEAR Landscape Fragmentation Tool. Fragmentation cut-points were set based on expert insight. In Connecticut, forest fragmentation is typically low when total core forest coverage is ≥70%. Fragmentation starts when cover is <70% (Askins 2014).

Habitat assessment application

This assessment is the first, landscape-scale study of habitat health across multiple priority habitat types in the LIS NEP. Overall, the LIS NEP ranked fair for the habitats assessed (27 out of 100 points). Considering the LIS as a whole, tidal wetlands, eelgrass, and coastal forests are in poor condition. Embayments and RMC are in fair condition (Figure 2).

Figure 2. LIS priority habitat health. Habitat scores are out of 100. Colors denote indicator score using evaluation criteria in Tables 2–6. Red is poor, yellow is fair, light green is good, and dark green is very good. Pie chart areas represent weighting applied to indicators when calculating the basin scores.

Habitat health was also assessed by basin. The three LIS basins are a common way of referring to geographic areas of the LIS NEP. Moving from the western to the eastern basin, population density, nutrient loads, and IC generally decrease and tidal flushing increases (Long Island Sound Study 2015). These factors may contribute to the differences in health across the basins. All priority habitats assessed are in poor condition in both the western and central basins, with the exception of estuarine embayments, which are in fair condition in the western and central basins. Across the five habitats assessed in both western and central basins, eelgrass has the lowest health ranking followed next by wetlands and forests. In the eastern basin, wetlands, rivers, and forest are in fair condition and eelgrass and embayments are in good condition. Across the five habitats assessed in the eastern basin, forests have the lowest health ranking followed next by wetlands. In addition to the biotic and abiotic differences between these basins, better health rankings for embayments may be due to a large focus on water quality improvements.

The assessment can also be looked at by metric. A total of 16 metrics were used in this study. In the western basin, 14 out of 16 metrics fall into the poor category for the habitats they measure. In the central basin, 9 of the 16 metrics fall into the poor category and in the eastern basin, only 2 of the 16 metrics fall into the poor category (Figure 2). The metrics appear to be responding, at least in part, to varying levels of development pressure, which is highest in the western basin. As development pressures lessen moving west to east, more metrics fall in the fair, good, and very good categories.

Habitat assessment review

The majority of other coastal programs that we reviewed track geographic extent for most of their priority habitats. Some programs provide a rank for habitat health based solely on geographic extent. Some programs use multiple indicators, but do not synthesize this information into a habitat health rank. In addition to this study, San Francisco Bay and Chesapeake Bay use a multi-indicator approach and provide health rankings for selected priority habitats. Tampa Bay uses a multi-indicator approach for eelgrass. None of the programs reviewed use multiple indicators to comprehensively rank habitat health across all of their priority habitat types. Across all programs grasslands and freshwater wetlands have the least-developed health assessments, although most programs indicate the importance of these habitats to overall coastal ecosystem health (Table 7).

The status and condition of priority habitats within the programs assessed was largely unclear. Programs either lacked the data needed for comprehensive habitat health assessments or had the data but did not integrate it to produce information on overall habitat health.

Discussion

The main objectives of this study were to (1) develop a habitat assessment framework, (2) apply the framework to assess priority habitat types within the LIS NEP, and (3) review the state of habitat assessments across other coastal programs.

Habitat assessment framework

The assessment framework developed herein captures priority habitat types that support diverse, structurally sound communities, provide important ecosystem services, and have significant economic value. The selected metrics measure the stressors on the landscape that affect the health of each habitat type, and in turn, the overall health of the LIS. While there may be no perfect set of metrics, size, habitat connectivity, natural flow, and structural soundness are four central factors related to coastal habitat health. The selected metrics capture these four factors. Coastal habitats of larger sizes are often better able to withstand disturbance and support more diverse communities (Meiman et al. 2012; Thompson et al. 2009; Reed and Hovel 2006). Habitats that are connected are better able to support species movement, genetic diversity, and species' ability to cope with stress (McGuire et al. 2016; Neraas and Spruell 2001). Natural flow of river and tidal waters provide numerous habitat health-related benefits including sediment transport, temperature control, and nutrient dilution which improves water quality and helps keep epiphytes under control (Mann 1982; Bednarek 2001; Hemminga and Duarte 2000). The final major factor contributing to habitat health and resilience relates to structural soundness of communities. Structurally sound communities exhibit the species distribution and diversity expected in an undisturbed habitat. Structural soundness is influenced by factors like size, connectivity, flow, and water quality. Examples of metrics that indicate problems with structure are excessive epifauna on eelgrass, nuisance macroalgae blooms in any of the aquatic habitats, invasive species replacing natural communities, fragmentation, and open water on tidal marshes.

Habitat assessment application and landscape-scale assessments within other coastal programs

Present day stressors on coastal habitats are complex, interconnected, and nuanced (Silliman, Grosholz, and Bertness 2009). Additionally, restoration is not as effective as it could be. This is due in part to vague, non-scientific goals, and attachment to a static, prior condition (Duarte et al. 2009; Palmer 2008; Greening et al. 2014). A holistic approach focused on restoring functional integrity of ecosystems so that they can sustain benefits to society could be more effective (Duarte et al. 2009).

Applying the assessment to the LIS sheds new light on the status of habitat health in the Sound. This assessment creates a starting point, whereby managers can quantitatively discuss, set, and track improvements. Programs can use the habitat assessment framework we developed to (1) set function-based, ecologically informed goals, (2) link on-the-ground projects to goals, (3) use historic data to inform goals, and (4) implement broad landscape-scale assessments to measure progress toward desired outcomes.

Set function-based, ecologically informed goals

Habitat goals in some NEPs are set on size parameters and what was achieved in the past, rather than multiple ecological indicators or science. Programs and policies focus on protecting “resilience and function” of habitats, but these terms are often ambiguous and undefined (Long Island Sound Study 2016). This assessment could help coastal programs understand the ecological impact of current goals, define resiliency and function, and set goals based on what needs to be accomplished to maintain a desired level of habitat function and value. The LIS CCMP provides an example: under the current RMC goal (open 200 miles in 20 years), the RMC connectivity score will be about the same as it is today, remaining in fair and poor categories. Applying this assessment, “open 200 miles in 20 years,” could become, “Within the next 10 years, river connectivity is restored so that 70% of all available RMC miles are open to diadromous fish, a 34% improvement in connectivity from the 2015 baseline. Water quality is improved to fair condition in the western and central basins and to good condition in the eastern basin and there is no net increase in impervious cover in any basin.” In practice, focus should fall on moving habitats toward an acceptable “range” relative to a less disturbed state (Palmer 2008). This assessment provides the language and tools by which to set this range and track progress in a measurable way.

In setting function-based goals, it may be useful to integrate an economic lens. Making a link between habitat quality and economic value can help focus restoration and protection priorities. For example, while coastal forests and embayments currently occupy more total area (Figure 3), the greater per acre value of tidal wetlands yields 42 to 70% of the total economic value of the five assessed habitats. Coastal forests account for 13 to 33% of the total economic value and embayments account for 10 to 21%. Due to the smaller area of eelgrass and RMC, their economic contribution is substantially lower. While ecological links between these habitats suggest that all should be addressed in a comprehensive manner, focus should fall on tidal wetlands as a top priority when considering economic value.

Figure 3. Comparison of habitat quality with economic valuation. Total value of habitats for the low and high per acre estimate (Table 8) illustrate the relative estimated value of habitats in LIS. Area of each type is presented for comparison. Red indicates poor habitat quality and yellow indicates fair habitat quality as presented in Figure 2.

Link goals across habitat types

Linking goals across habitat types enables managers to create policies and programs that are in line with how habitats function within ecosystems, as interconnected pieces rather than independent entities. However, ecosystems and response times are complex (Duarte et al. 2009), so linking goals should not imply a linear relationship.

To fully implement linked goals, quantitative tools are needed. This study provides one possible tool to more holistically address health across habitat types. For example, tidal restrictions like dams can impact rivers, wetlands, and embayments and negatively impact wildlife (Bednarek 2001; Correll 2016). Dams can limit sediment supply, affecting the resilience of sediment-limited salt marshes (Correll 2016). Dams can also cause poor water quality due to lack of flow, which in turn exacerbates thermal stress, hypoxia, and impacts eelgrass health (Bednarek 2001; Hemminga and Duarte 2000). Linking goals can forward an ecosystem services perspective. As discussed above, improving health in one area (dams/connectivity) can affect multiple habitats. And healthier habitats provide more ecosystem service benefits (Barbier et al. 2010).

Approaching the results of the assessment from an ecosystem service angle could help gain community and political support. We developed Table 8 as a starting point for efforts to maintain ecosystem services by improving factors that contribute to habitat health across all priority habitat types.

Table 7. Landscape-level assessments in select coastal programs. white = no habitat extent data found, grey = existing or restored habitat extent tracked; M = multiple indicators tracked; H = multi indicator assessment synthesized into a habitat health rank; N/A = not applicable (ecologically relevant) based LIS habitat definitions.

Table 8. Services, resilience, and value of LIS habitats. Column A includes major contributors to habitat resilience (bold = factors that contribute to habitat resilience in the face of climate change). Column C provides the health status, economic value, current extent, and estimated value to LIS.

	A. Services provided by habitat in LIS area	B. Factors contributing to habitat health and resilience	C. Assessment results for LIS and estimated value.	D. Actions needed to improve or further resilience of the habitat
Eelgrass Beds	feeding grounds, breeding and nursery area, improves water quality by removing nutrients and sediment, flood protection via dampening storm and wave energy, supports recreationally and commercially important finfish and shellfish	High density and area beds provide better habitat. Healthy eelgrass is more resilient to disturbances, recovering more quickly when damaged. Health is improved by good water quality, shading, and cooler temperatures.	Overall, poor. Eastern is good, Central and Western are poor. $9,861 to $28,114 per acre per year ^A 1,640 acres ^B $16.2 - 46.1 million per year	1. Decrease nutrient and sediment inputs from human sources to reduce growth of epiphytes. 2. Encourage development that reduces storm water input. 3. Maintain natural flushing in eelgrass areas to reduce temperatures. 4. Decrease physical disturbance from humans (e.g. dredging, moorings). 5. Plan for migration with sea level rise.
Estuarine Embayments	feeding grounds, breeding and nursery area, can be good areas for submerged aquatic vegetation habitat and tidal wetlands, flood protection via dampening storm energy, commercial and recreational uses	Good water quality and natural levels of tidal and riverine flow support healthy and diverse native communities. Monocultures of nuisance seaweed indicate poor water quality while presence of eelgrass indicates good water quality.	Overall fair. Eastern is good, Central and Western are fair. $2,220 to $4,152 per acre per year ^A 37,313 acres ^C $82.8 - 154.9 million per year	1. Decrease nutrient and sediment inputs from human sources to reduce nuisance seaweed and improve water quality. 2. Protect natural upland areas to intercept nutrients and sediment. 3. Encourage development that reduces storm water input. 4. Maintain natural flushing by reducing tidal and riverine restrictions to keep temperatures low and improve water quality.
Tidal Wetlands	feeding grounds, breeding and nursery area, improves water quality by filtering nutrients and sediment from runoff, flood protection via dampening storm and wave energy, store flood waters as groundwater, supports recreationally and commercially important finfish	Large, contiguous wetlands with good structural integrity provide resilient habitat. A healthy wetland is a mosaic of diverse native plants and animals, with community patterns reflecting tidal influence.	Overall, poor. Eastern is fair, Central and Western are poor. $11,699 to $77,260 per acre per year ^A 14,123 acres $165.2 - $1,091.1 million per year	1. Maintain natural flushing by reducing tidal and riverine restrictions to allow for sediment transport and tidal range (e.g. widen culverts, remove dams). 2. Protect upland buffers to allow for migration inland with sea level rise. 3. Protect upland areas to intercept nutrients. 4. Decrease nutrient inputs from human sources to improve structural integrity of plant communities. 5. Control invasive species (e.g. Phragmites australis).
Riverine Migratory Corridors	spawning and nursery area, moderates streamflow changes, groundwater storage and recharge, sediment transport important for supporting wetland elevation, supports recreationally and commercially important finfish and shellfish, recreational uses	Connectivity is the primary factor; without fish passage, RMC do not function. Good water quality and natural levels of riverine flow are critical. Cooler water temperature is important for many organisms.	Overall fair. Eastern and Western are fair. Central is poor. $248 to $20,537 per acre per year ^D 3,206 acres ^E $0.8 - $65.8 million per year	1. Maintain migration corridors via dam removal or installation of fish ways. 2. Protect and improve vegetated buffers which serve to reduce temperature via shading. 3. Protect natural upland areas to intercept nutrients. 4. When evaluating proposed changes in land-use, consider the impact on water temperature and quality. 5. Maintain natural hydrology by reducing restrictions on river flow or managing flow to allow for sediment transport.
Coastal Forests	complexity of environment supports wildlife diversity, migration corridors, nutrient and carbon storage, improves water quality by filtering nutrients and sediment from groundwater and runoff, wind buffers, recreational uses	Large, contiguous forests with a diversity of native plants and animals provide quality resilient habitat. Connectivity among patches supports wildlife use.	Overall poor. Eastern is fair, Central and Western are poor. $993 to $1,564 per acre per year ^A 130,285 acres $129.4 - $203.8 million per year	1. Protect priority tracts of forest to decrease fragmentation and increase connectivity. 2. Increase connectivity by restoring forests in areas with fragmented patches. 3. Control invasive species to encourage diversity and resilience of native communities.

^A Kocian et al. (2015).

^B Tiner, McGuckin, and MacLachlan (2013).

^C Vaudrey (2015).

^D Kocian et al. (2015) estimate for freshwater does not include valuation of many RMC services and is likely an underestimate of RMC value.

^E Assuming an average river width is 10 m, 806.2 miles of open RMC converts to 3,206 acres.

Use historic data to inform goals and communicate change in health

Historic change can provide important reference points for restoration (Lotze et al. 2006). Tampa Bay and San Francisco Bay provide compelling examples of leveraging an understanding of historic change to advance meaningful improvement on an ecosystem scale. Over 20 years ago, Tampa Bay managers set an ambitious goal to restore seagrass to 1950 acreage levels, ∼38,000 total acres (153.8 km²), requiring restoration of 13,000 acres (52.6 km²). Tampa Bay exceeded their restoration goal, with current seagrass extent at 41,655 total acres (169 km²) (Cicchetti and Greening 2011; Sherwood et al. 2017). Similarly, San Francisco Bay showed impressive restoration gains even under intense urban pressure. Managers used a historic context coupled with an understanding of habitat health to reset assumptions on the scale of wetland restoration needed (Goals Project 1999). Using historic reference points to contextualize restoration goals can advance progress on more ecologically meaningful levels (Goals Project 2015).

Create broader landscape-scale assessments

Landscape-scale assessments could encourage programs in the same region to track and communicate ecosystem health in a similar way. One of the downfalls of restoration projects is that they are often too small to be meaningful. Even so called “landscape-scale” projects are often disjointed efforts by multiple partners who work on their own piece and track progress using their own databases, rather than coordinate (Palmer 2008).

While few of the coastal programs reviewed conduct integrated health assessments, many have the data needed to do it. The indicators presented here allow each program to decide cut-points for their specific location. Thus, outputs can be tailored to local systems and compared cross-programmatically using common terminology. Setting goals across regions can be more economical, bring greater ecological benefit (Neeson et al. 2015), and encourage healthy cooperation and competition between programs in similar ecoregions.

Conclusion

Although large amounts of money and time are spent on restoration efforts, losses may be outpacing gains (USEPA 2017; Palmer 2008). However, success stories do exist. For example, programs in the Chesapeake, Tampa, and San Francisco Bays (USA) show habitat restoration success through health-orientated goal setting. Currently, coastal habitat restoration progress is limited to select locations and habitats. More could be done.

The results of this study and the efforts of other programs highlight the importance of moving away from vague goals, disjointed small-scale projects, and a significant, but limited focus on nutrient inputs. This effort shows that the development of a framework to assess coastal habitat condition is possible for the LIS NEP and for other coastal programs. The framework developed herein enables the quantitative assessment of current priority habitat condition in the LIS NEP and it establishes a baseline. This in turn affords the opportunity to develop science-based, data driven restoration goals. Broader ecosystem health goals can now be used to steer formerly disconnected restoration efforts across the LIS NEP, and progress toward these goals can be tracked. Quantitative, landscape-scale frameworks are timely and can be leveraged to improve restoration success across coastal ecosystems.

Acknowledgments

We thank Susan Adamowicz, Robert Askins, Paul Capotosto, Patrick Comins, Chris Elphick, Charles Frost, Steve Gephard, Sally Harold, Troy Hill, Jeff Horan, Andrew MacLachlan, Victoria O'Neill, Suzanne Paton, Larry Swanson, Roger Wolfe and Harry Yamalis for providing data, field assistance, technical expertise and invaluable discussions that improved this publication. We thank Mark Tedesco and Holly Drinkuth for overall guidance. We thank Holly Greening and Letitia Grenier for their insight. We thank Mark Brouder and Peg Van Patten for critical reviews and John Fisher and Ana Herrera for reference material. Special thanks to David Robinson who supported this effort.

Funding

This project was funded in part by the Interagency Agreement between USEPA and USFWS and by USEPA to the Sponsored Program Services of the University of Connecticut on behalf of Connecticut Sea Grant, and in collaboration with New York Sea Grant (awards LI96144301 and LI96113701).