Final Report: Coastal Wetland Indicators

EPA Grant Number: R828677C003
Subproject: this is subproject number 003 , established and managed by the Center Director under grant R828677
(EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).

Center: EAGLES - Atlantic Coast Environmental Indicators Consortium
Center Director: Paerl, Hans
Title: Coastal Wetland Indicators
Investigators: Morris, James T. , Novakowski, Karyn I. , Gallegos, Charles L. , Montane, Juana M. , Hopkinson, Charles S , Rodriguez, Diana , Herrick, Gabe , Marshall, Helen , Torres, Raymond , Valentine, Vinton
Institution: Marine Biological Laboratory
EPA Project Officer: Packard, Benjamin H
Project Period: March 1, 2001 through February 28, 2005
RFA: Environmental Indicators in the Estuarine Environment Research Program (2000) RFA Text |  Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Water , Ecosystems


The main objectives of this research project were to:  (1) develop a suite of indicators of the condition of coastal wetlands that are based on physical and biological criteria, with an emphasis on higher plant-based pigment indicators; and (2) link these indicators to remote-sensing capabilities.

The Atlantic Coast Environmental Indicators Consortium (ACE INC) Coastal Wetland Indicators project has developed indicators at a range of temporal and spatial scales that are metrics of the status of coastal wetlands.  These scales range spatially from the landscape-scale to the level of an individual leaf and temporally from decades to hours.  At the leaf-scale, we have sought to develop indicators that are based on plant pigment and can be applied to a larger scale using remote sensor data.  The objective here is to develop an indicator that is rapid, nonintrusive and that has high temporal and spatial resolution.  It also should be applicable to a range of stressors and species found in the coastal environment.  At the larger temporal and spatial scales, we have sought to develop indicators based on geomorphological patterns that can be remotely sensed.  These geomorphological indicators are appropriate as metrics of long-term trends and vulnerability.  For example, detection of wetland loss caused by sea-level rise would be an appropriate application for the geomorphological indicators.

Summary/Accomplishments (Outputs/Outcomes):

Pigment Indicators

All plants possess the xanthophyll cycle.  This is a two-step deep oxidation reaction that serves to protect photosystem II (PSII) during stress.  The stress pigment zeaxanthin is an effective nonphotochemical quencher and protects PSII during light, water, temperature, and nutrient stress, and in response to chemical pollutants that impair PSII activity.

The indicator can be expressed as the epoxidation state, which is computed as a ratio of the concentrations of the different forms of the xanthin pigment: 

ES = [zea] + 0.5[anth])/([zea] + [anth] + [viol]). 

These pigments absorb energy within the visible wavelengths (520-580 nm) and are detectable.  This change in the state of the xanthophyll cycle leads to a change in leaf reflectance between 520-580 nm.  The reflectance of this feature, therefore, may be used as an indicator of relative zeaxanthin levels and indirectly as an indicator of plant stress levels.  This can be quantified as follows: 

Stress Index

Where R is the reflectance at the specified wavelength.  We have found that this stress index correlates well with the conversion to zeaxanthin in the xanthophyll cycle (quantified using high performance liquid chromatography analysis) for both species tested.

The epoxidation state can indicate water, temperature, light, pollutant, or nutrient stress.  It does not indicate necessarily the specific cause of stress, but in combination with other measures the causes can be determined.  For example, nutrient stress (deficiency) causes deepoxidation of the xanthophyll cycle, because of a decrease in the plant’s ability to repair normal levels of PSII damage.  A nutrient stress also would be associated with a decline in chlorophyll concentration.  The productivity of salt marsh vegetation is commonly limited by nitrogen availability, so the eutrophication of a marsh would manifest as an increase in epoxidation state.

A change in epoxidation state of the xanthophyll cycle was detected in Spartina alterniflora and S. patens in different nutrient fertilized plots at Plum Island, Massachusetts.  Nutrient additions led to higher epoxidation states (i.e., lower stress levels).  In both species, control plots showed the lowest epoxidation states indicating that both are limited by nutrients under normal in situ conditions.  Xanthophyll cycling occurs in a matter of minutes such that the state of the cycle indicated stress conditions with high temporal resolution.

Macro-Scale Indicators

Land-Classification.  Because of variations in light scattering, reflectance, and absorption, different surfaces show distinctive reflectance signatures throughout the light spectrum.  Leaf reflectance is dependent upon the physical structure of the leaf itself (and thus its scattering/reflecting properties) and the absorbance of light by pigments and biochemicals.  Thus plant communities often have different reflectance properties as they are composed of species that differ in the physical structure of their leaves and biochemical makeup.  Furthermore, plant communities are distinctly different from other land-class types in their reflectance of light.

The land-classification indicator detects physical/structural land surface properties using four wavelength bands (Airborne Data Acquisition and Registration) and uses a neural network to classify vegetated areas, land surface types, and differentiate major coastal wetland plant communities.  A neural network was trained to distinguish six land surface types from four wavelength bands in images taken from fixed wing aircraft.  Work is continuing with hyperspectral data, which we expect will allow the land classes to be refined even further, possibly even providing enough information to compute a stress index of the vegetation (see above).

Land-use classification and change analysis are fundamental management tools that provide information about a wide variety of environmentally important topics that range from the calculation of impervious surface area to changes in forest cover.  Canned GIS packages (e.g., ArchInfo) incorporate algorithms that allow various kinds of classifications to be performed.  The manner in which we have combined neural net classification with elevation data from a Light Detection and Ranging (LIDAR) sensor is novel and provides a means of analyzing an important geomorphological metric that has not been possible before.

Land-use classification has many applications.  In the context of coastal wetlands, a time-series of classified images is used for change detection.  For example, over time is there a change in the area of salt marsh and/or has it moved?  Land-classification can provide geostatistical data (the type described below) and when combined with elevations can provide details about the vulnerability of a coastal wetland to sea-level rise.  The ability to forecast vulnerability can provide an early warning that would prompt a management response to mitigate the threat.  For example, a decision could be made to divert sediment-laden water from a river into the adjacent marshes as is being done now in coastal Louisiana.

Geomorphological Pattern.  The geomorphic configuration of tidal marshes reflects the complex interaction of a variety of factors over time.  Sea-level change, sediment supply, sedimentation and accretion, plant growth, and organic matter preservation interact to create particular landscape configurations within tidal wetland systems.  Tidal channels serve as conduits for sediments, nutrients, detritus, and organisms.  Other hydrographic features, such as pools and pannes, reveal clues about the state of marsh development, condition of the vegetated marsh surface, and habitat space for fish and birds.  These configurations change as the underlying factors change.  When viewing the tidal marsh landscape, there is a range in conditions from maximum marsh development in equilibrium with sea level to highly impacted marshes with extensive ponds and large areas of open-water habitat.  Our work explored the use of geomorphometric measures of the channel and ditch networks and other hydrographic features of tidal marshes as indicators of condition.

Using features mapped from aerial imagery, a variety of metrics can be calculated that describe the physical configuration or form and condition of the marsh channel, ditch network, and other hydrographic features.  The most promising metrics include: 

  • Channel, Ditch, and Total Drainage Densities:  total natural channel length, total ditch length, or total watercourse length per unit area.  These three indicators describe landscape dissection and are a surrogate for habitat edge.
  • Sinuosity mean:  mean of actual natural channel or ditch section lengths divided by straight-line lengths between section ends.  This is an indicator of departure from curvilinear character, variable water velocity, potential for erosion and deposition, and varied edge habitat; larger number indicates more convoluted channel.
  • Mean External Link Length:  mean length of all external or terminal links.  This indicator describes the extent of the network in the marsh; a larger number indicates network extension through headward erosion or some other process.  We recommend interpreting indicator in combination with sinuosity mean.
  • External Link Frequency:  number of external or terminal links per unit area.  This is an indicator of the hydrological capacity of a marsh drainage network; a larger number indicates expansive network.  We recommend interpreting indicator in combination with mean external link length.
  • Marsh Network Area Ratio:  total marsh area to total network area.  This is a measure of marsh area to network area; larger number indicates a more developed tidal wetland.
  • Multi-Fractal Dimensions of Tidal Drainage Network:  negative slope of plot of log of box counts to log of box lengths for each curve segment.  The channel segment fractal dimension provides an indirect measure of sinuosity; channel network fractal dimension gives a composite measure of the drainage density and spatial distribution of channels; and the branching structure fractal dimension gives a measure of channel bifurcation without the need for stream ordering.
  • Marsh Water Area Ratio:  total marsh area to total water area.  This is a measure of marsh area to water area (natural channels, ditches, pools); a larger number indicates a more developed tidal wetland.
  • Pool Density:  number of pools per unit area.  This is a measure of the extent of pools within the marsh area; a larger number indicates more individual pools within the landscape and therefore greater habitat complexity.
  • Pool Surface Area Ratio:  total pool area per unit area.  This is a measure of the proportion of marsh area that is surface water pools and therefore not vegetated.
  • Mean Pool Size:  mean area of individual pools.  A larger size indicates larger pools in marsh area on average and therefore less vegetated marsh and possibly more habitat space.  If natural condition of marsh, however, is small pools, the measure could suggest a nonequilibrium state.  We recommend interpreting indicator in combination with pool density.

Tidal wetlands are threatened by a variety of factors, some of which are natural and some anthropogenic.  The summary by Boesch, et al. (2004) of factors responsible for marsh loss in the Mississippi delta could apply anywhere and range from construction of dikes that prevent flooding (and sediment supply) and dams that trap sediment to sea-level rise.  There also is evidence in some estuaries of a legacy effect from colonial deforestation, after reforestation of watersheds, that now may be resulting in marsh loss caused by a decline in sediment supply.  Our modeling work suggested that marshes are lost quickly when they cross a threshold.  We have hypothesized that as this threshold is approached, there are changes in the geomorphological patterns that can signal when a change is occurring.  Much of our work has been exploratory.  For example, we have examined the relationship between creek length and watershed (or flowshed) area to determine if these properties vary by a constant ratio as they are known to do in terrestrial watersheds.  We found a power function relationship between area and length, as in terrestrial watersheds, which indicates scale invariance between these landscape parameters.  Likewise, length-area analyses showed that there exists an upper limit to the degree to which watersheds greater than approximately 2,000 m2 are dissected by intertidal creek networks.  Data that plotted above this limit may indicate that the landscape is in a dynamic state.  In our study area, there were 24 such creek networks that were among the smaller watersheds in the system.  These observations indicated that the smallest watersheds were likely to be most responsive to estuarine change, and all or part of these 24 networks should be target sites for more refined studies of ecosystem stability.

Tidal creek structure and surface water extent are indicators of coastal wetland stability.  Our findings indicated that tidal creek suspended sediment concentration was related inversely to tidal creek length.  This observation indicated that less extensive and shorter creek networks were the dominant sources of inorganic and nutrient rich organic particulate matter (allocthonous and autochthonous) to the salt marsh platform area.  Hence, creek length and total creek length per unit area are likely indicators of tidal wetland stability.  Surface water extent or pool cover also gives an indication of wetland condition.  In the Plum Island marshes, we found that the large isolated ponds in the interior marsh were related to the next phase of marsh development, namely recovery from the extensive ditching performed since colonial times.

As noted previously, the measures listed above need to be interpreted together as well as individually.  In comparing the metrics examined with results from other studies, there is evidence of regional versus subcoastal versus coastwise applications.  For example, mean pool size appears to apply on a regional level.  Channel and ditch densities, pool density, and pool surface area ratio, however, appear to apply on a broader northeast subcoastal level.

In applying the geomorphological pattern indicators, two considerations are necessary.  First, we found that the selected tidal marsh drainage metrics need to avoid stream ordering of the network.  Stream ordering depends on decision rules that simplify water flow.  Because these networks have bidirectional flow and the watercourses often connect to one another, the ordering tends to oversimplify the network and water flow.  In application, many convoluted marsh areas fall outside of consideration and are not captured in the metrics.  Second, we found that the classic areal accounting unit (e.g., watershed) needs to be discarded.  As with the stream ordering, the watershed delineation requires oversimplification of the marsh surface, often requiring arbitrary selection of the boundary.  In most cases, there are no topographic ridges that encompass a marsh area and direct water flow to an outlet point.  Indeed, in many cases, the interior of the marsh is lower than along the network banks.  Additionally, features that fall within the interior of the marsh (e.g., pools) would be apportioned to different watersheds.  Any metric that describes the condition of interior marsh surface would give meaningless results.  To resolve the accounting unit issue and permit more robust metrics, we favor interfluvial and island areas.

Training and Development

An undergraduate student in the 2002 Marine Biological Laboratory Semester in Environmental Science, Jennifer Franklin, conducted her semester research project investigating geomorphic characteristics of the Plum Island marshes.

In cooperation with Thomas Millette of Mount Holyoke College, we involved 10 undergraduate summer interns and work study students over the period 2003-2005 in mapping surface water features of the Plum Island Estuary marshes from orthophotograph images and multispectral imagery.  We conducted an annual field session with the students each summer to prepare them for mapping pools and pannes and to evaluate results on the ground.

Outreach Activities

We participated in the Workshop on Revisiting the Regional Protocol for Monitoring Tidal Wetland Restoration in the Gulf of Maine, September 29-30, 2004, at the Wells National Estuarine Research Reserve, Wells, Maine.  Our participation focused on examining and revising protocol variables and methods as part of the hydrology and soils/sediments functional group.  We offered information regarding base mapping, elevation data collection, and pond metabolism.  The workshop gave us an opportunity to discuss our research with personnel from a variety of federal, regional, state, and nonprofit organizations active in the Gulf of Maine.

We coorganized (with ACE INC coinvestigator Raymond Torres, Danika van Proosdij, and Sergio Fagherazzi) a Chapman Conference on Salt Marsh Geomorphology:  Physical and Ecological Effects on Landform in Halifax, Nova Scotia, Canada, October 9-13, 2004.  The conference brought together international experts on marsh geomorphology and ecology with a goal of understanding the effects of human activities, climate changes, and sea-level rise on intertidal marshes.

We attended the New England Estuarine Research Society Fall 2004 Meeting held October 21-23, 2004.  We met with a number of people from the U.S. Environmental Protection Agency (EPA) National Health and Environmental Effects Research Laboratory (NHEERL), Atlantic Ecology Division (AED).  We discussed the geomorphometric indicators investigated to date, development of reference conditions, and possible applications of and enhancements to our indicators in support of regional projects.

We served on the organizing committee for the 8th International Conference on Remote Sensing of Marine and Coastal Environments, held May 17-19, 2005, Halifax, Nova Scotia, Canada.  The theme of the conference was to address bridging the gap from research to marine and coastal operations, including the issue of effective integrated response to coastal and marine policy requirements.  Our specific role was to solicit and organize papers related to LIDAR mapping of tidal wetlands.

Chuck Hopkinson and Hans Paerl cochaired a Special Session at Estuarine Research Federation 2005 Biennial Conference in Norfolk, Virginia, entitled:  Identifying, Assessing, and Managing Human and Climatically-Induced Change of Estuarine Ecosystems.

We met with a diversity of stakeholders, nongovernmental organizations, and Commonwealth of Massachusetts and Town officials at the North Shore Audubon Society Headquarters, Beverly, Massachusetts, in December 2005, to explain the implications of our research findings in the Plum Island Sound estuary, especially as they pertain to the effects of land use change, climate change, and sea level rise on coastal wetlands.

We had our direct collaboration with state and federal organizations that manage wetland resources in the Gulf of Maine region.  We had regular contact with personnel from the Massachusetts Office of Coastal Zone Management (MACZM) regarding data and information sharing as well as indicator use for assessing salt marsh responses to human and natural stressors.  They were interested in how they can incorporate our research into their work on rapid assessment of salt marshes.  We also solicited and incorporated MACZM input as we pursued plans to collect LIDAR and digital aerial photography for the Plum Island Estuary study area.  We also shared information with personnel from the Parker River National Wildlife Refuge and the U.S. National Resource Conservation Service.  

We participated in the New England Estuarine Research Society (NEERS) Spring 2005 Meeting held April 28-30, 2005.  We also met with people from U.S. Fish and Wildlife Service (USFWS) Region 5 and U.S. National Park Service (NPS).  We discussed the geomorphometric indicators investigated to date, how the indicators compared with study results performed by these groups, and possible applications of and enhancements to our indicators in support of regional projects.
We were in frequent contact with personnel from EPA NHEERL AED, to share information and to obtain feedback on the indicators that we were investigating.  We also shared information about marsh ponding with personnel from the Parker River National Wildlife Refuge, the USFWS Region 5, and the NPS.

Contributions to State of Knowledge

The indicators that we developed provided new tools for evaluating the condition of coastal wetlands.  The actual products will be indicators that are based on measurements made in the field.  All the indicators investigated, however, have a significant potential for being developed as applications that can be calibrated using remotely sensed data.  Indeed, the geomorphological pattern indicators are derived from interpretation of remote sensing images.  To date:  (1) Progress has been made using pigments and reflected light as indicators of the condition of vegetation.  (2) Neural networks have proven to be effective tools for classifying remote sensor data.  (3) Significant trends in the productivity of coastal wetlands have been observed.  (4) We have documented that we are able to discern interannual changes in the relative elevation of the marsh surface.


Boesch DF, Josselyn MN, Mehta AJ, Morris JT, et al. Scientific assessment of coastal wetland loss, restoration and management in Louisiana. Journal of Coastal Research 1994;20(Special Issue):1-89.

Journal Articles on this Report : 6 Displayed | Download in RIS Format

Other subproject views: All 89 publications 19 publications in selected types All 17 journal articles
Other center views: All 383 publications 99 publications in selected types All 88 journal articles
Type Citation Sub Project Document Sources
Journal Article Cavatorta JR, Johnston M, Hopkinson C, Valentine V. Patterns of sedimentation in a salt marsh-dominated estuary. Biological Bulletin 2003;205(2):239-241. R828677 (Final)
R828677C003 (2003)
R828677C003 (Final)
  • Full-text: Biological Bulletin-Full Text HTML
  • Other: Biological Bulletin-PDF
  • Journal Article Farber S, Costanza R, Childers DL, Erikson J, Gross K, Grove M, Hopkinson CS, Kahn J, Pincetl S, Troy A, Warren P, Wilson M. Linking ecology and economics for ecosystem management. BioScience 2006;56(2):121-133. R828677 (Final)
    R828677C003 (Final)
  • Abstract: BioScience-Abstract
  • Other: Duke University-PDF
  • Journal Article Johnston ME, Cavatorta JR, Hopkinson CS, Valentine V. Importance of metabolism in the development of salt marsh ponds. Biological Bulletin 2003;205(2):248-249. R828677 (Final)
    R828677C003 (2003)
    R828677C003 (Final)
  • Abstract from PubMed
  • Full-text: Biological Bulletin-Full Text HTML
  • Other: Biological Bulletin-PDF
  • Journal Article Morris JT, Sundareshwar PV, Nietch CT, Kjerfve B, Cahoon DR. Responses of coastal wetlands to rising sea level. Ecology 2002;83(10):2869-2877. R828677 (2001)
    R828677 (Final)
    R828677C003 (2003)
    R828677C003 (Final)
    R826944 (2000)
    R826944 (2001)
    R826944 (Final)
  • Full-text: ESA Ecology-Full Text HTML
  • Abstract: ESA Ecology-Abstract
  • Other: South Dakota School of Mines-Full Text PDF
  • Journal Article Morris JT, Porter D, Neet M, Noble PA, Schmidt L, Lapine LA, Jensen JR. Integrating LIDAR elevation data, multi-spectral imagery and neural network modelling for marsh characterization. International Journal of Remote Sensing 2005;26(23):5221-5234. R828677 (Final)
    R828677C003 (2004)
    R828677C003 (Final)
    R829458 (2005)
    R829458C004 (2003)
    R829458C004 (2005)
  • Full-text: Washington University - Taylor & Francis Preview
  • Abstract: Taylor & Francis - Abstract
  • Journal Article Novakowski KI, Torres R, Gardner LR. Geomorphic analysis of tidal creek networks. Water Resources Research 2004;40(5):W05401. R828677 (Final)
    R828677C003 (2004)
    R828677C003 (Final)
  • Abstract: Wiley Online-Abstract
  • Supplemental Keywords:

    coastal wetlands, marsh habitat, higher aquatic plants, photopigments, geomorphology, tidal ecosystems, regional indicators, light detection and ranging, LIDAR, nutrient status, physiology, sea level rise, neural network analysis, wetland management,, RFA, Scientific Discipline, Air, Water, ECOSYSTEMS, Ecosystem Protection/Environmental Exposure & Risk, RESEARCH, estuarine research, Hydrology, Ecosystem/Assessment/Indicators, climate change, Air Pollution Effects, Aquatic Ecosystems, Monitoring, Ecological Monitoring, Atmosphere, Ecological Indicators, environmental monitoring, remote sensing, coastal ecosystem, bioindicator, plant indicator, coastal watershed, estuaries, coastal environments, diagnostic indicators, ecosystem indicators, environmental indicators, coastal ecosystems

    Progress and Final Reports:

    Original Abstract
  • 2001
  • 2002 Progress Report
  • 2003 Progress Report

  • Main Center Abstract and Reports:

    R828677    EAGLES - Atlantic Coast Environmental Indicators Consortium

    Subprojects under this Center: (EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).
    R828677C001 Phytoplankton Community Structure as an Indicator of Coastal Ecosystem Health
    R828677C002 Trophic Indicators of Ecosystem Health in Chesapeake Bay
    R828677C003 Coastal Wetland Indicators
    R828677C004 Environmental Indicators in the Estuarine Environment: Seagrass Photosynthetic Efficiency as an Indicator of Coastal Ecosystem Health