Projects

Habitat suitability index for submerged aquatic vegetation of the Mississippi Coast

End Date: 2-01-12

Abstract

The overall goal of this project is to develop successful Habitat Suitability Index (HSI) for Submerged Aquatic Vegetation (SAV) via a decision-tree algorithm approach that utilizes landscape properties. Our specific objectives are:

  1. To create SAV beds distribution maps along the major tidal areas of the Mississippi coast to characterize the attributes (parameters) of landscape properties of the areas that support particular types of SAV beds;
  2. To weigh and priotize the parameters to build a decision algorithm;
  3. To create SAV habitat suitability maps of two chosen river basins by applying the HSI values calculated from the decision tree algorithm;
  4. To assess the accuracy of the HSI by creating contingency tables that compare the habitat suitability mapping results and the field data; and
  5. To disseminate usages of the algorithm and index value computation to end users.  We propose to develop a HSI for SAV beds/types through a decision tree approach that utilizes parameters of shore vegetation and landscape features. 

The index will be developed from the past SAV survey data by the PIs; additional field investigation along the Mississippi coastal waters; topographic maps; and Geographic Information System (GIS).  After we develop a tree-based algorithm for index, its validation will be assessed using a separate set of field data. After validation, the user manuals for the algorithm/index will be disseminated to potential end users. 

The economic value of the global coastal seagrass/SAV beds is estimated as 11 times higher than that of the equivalent acreage of coral reefs and 15 times higher than that of salt marshes, however, seagrass/SAV habitats receive the least media, funding, research attention among the major coastal ecosystems.

There are several water quality/environmental models for seagrass/SAV habitat requirements; these models were developed based on long-term monitoring data.  Application of those models by resource managers also requires extensive/consistent water quality monitoring data, hence, limiting their usages to the areas with well monitored habitats. In addition, the conventional habitat models are often developed from classical regression methods, which makes it difficult to apply them in areas with complex landscape features.

This project will identify and prioritize parameters of landscape factors important for locating and managing SAV resources.  Application of the index will not be restricted to the well-protected and monitored areas because the index will use geographic, topographic, and shore vegetation parameters.  Using GIS, the resultant HSI can be used to visualize potential SAV bed locations and to predict how coastal landscape alteration would affect the distribution and abundance of the beds. The algorithm and the resultant HSI will provide a useful tool for management and research by extrapolating knowledge from sampled to the unsurveyed areas and help identify the location of the fisheries habitats and what are the current limiting factors that need to be considered in SAV restoration.

Objectives

The Overall Goal of this project is to develop successful Submerged Aquatic Vegetation Habitat Suitability Index (SAV-HSI) via a decision-tree algorithm approach that utilizes landscape properties (geographic/topographic/shore vegetation information).

 

Our specific objectives are: To create SAV beds distribution maps along the major tidal estuarine and stream areas of the Mississippi coast to characterize the attributes (parameters) of landscape properties of the areas that support particular types of SAV beds. 

  1. To weigh and priotize the parameters to build a decision algorithm. 
  2. To create SAV habitat suitability maps of two chosen river basins by applying the HSI values calculated from the decision tree algorithm. 
  3. To assess the accuracy of the HSI by creating contingency tables that compare the habitat suitability mapping results and the field data.
  4. To disseminate usages of the algorithm and index value computation to end users.

Methodology

We propose to develop a HSI for SAV beds/types through a decision tree approach that utilizes parameters of shore vegetation and landscape features.  The index will be developed from the past SAV survey data by the PIs; additional field investigation along the Mississippi coastal waters; topographic maps; and Geographic Information System (GIS). We will field investigate total of five tidal estuarine/riverine areas of the Mississippi coast for the SAV type/distribution, shore vegetation, salinity and sediment types at selected locations.  A portion of the data will be used as learning samples (training information) to develop an index algorithm; the rest will be used for post classification accuracy assessment (to create contingency tables). The field data will be integrated into a GIS system. Using the ArcGIS 9.3 Desktop suite, the following will be calculated and characterized: mean slope and elevation for the surveyed basin area, channel/basin shape, shore aspects and meandering direction, and the extent and the relative location of the each shore vegetation type within the landscape. We will use Classification and Regression Tree (CART) method using the surveyed information as the “learning sample”. The landscape variables will be used as predictive variables. For simpler adaptation of the process, we will use categorical data types for the variable attributes instead of continuous data values. We will use a set of logical if-then conditions for predicting or classifying cases.  Top-down induction method will be used to split the learning sample. The splitting will continue until it reaches the user specified restrictions (i.e. if all objects from learning sample-input parameters- have the same class). The dependable variables will include (1) a unitless continuous index (HSI), for example, index value ranging 0-1, with larger values indicating higher possibility to support SAV beds; and (2) categorical predictor index that indicates the SAV types.  After we develop a tree-based algorithm for index, its validation will be assessed using a separate set of field data. After validation, the user manuals for the algorithm/index will be disseminated to potential end users.

Rationale

Submerged Aquatic Vegetation (SAV) beds provide numerous ecosystem services, which include: nursery for juvenile stages of finfish and shellfish, an important food-source to aquatic organisms and wading birds, sediment stabilization and buffering wave energy, and nutrient uptake and sequestration that mitigate eutrophication. The economic value of the global coastal seagrass/SAV beds is estimated as 11 times higher than that of the equivalent acreage of coral reefs and 15 times higher than that of salt marshes, however, seagrass/SAV habitats receive the least media, funding, research attention among the major coastal ecosystems. There are several water quality/environmental models for seagrass/SAV habitat requirements; these models were developed based on long-term monitoring data.  Application of those models by resource managers also requires extensive/consistent water quality monitoring data, hence, limiting their usages to the areas with well monitored habitats. In addition, the conventional habitat models are often developed from classical regression methods, which makes it difficult to apply them in areas with complex landscape features. This project will identify and priotize parameters of landscape factors important for locating and managing SAV resources.  Application of the index will not be restricted to the well-protected and monitored areas because the index will use geographic, topographic, and shore vegetation parameters.  Using GIS, the resultant HSI can be used to visualize potential SAV bed locations and to predict how coastal landscape alteration would affect the distribution and abundance of the beds. The algorithm and the resultant HSI will provide a useful tool for management and research by extrapolating knowledge from sampled to the unsurveyed areas and help identify the location of the fisheries habitats and what are the current limiting factors that need to be considered in SAV restoration.

This project is closely related to the first 2010-2011 MASGC Focus Area (Healthy Coastal Ecosystems).  Through the proposed work, we will (1) identify and priotize important geographic/topographic parameters for locating SAV beds; and (2) provide a tool (an algorithm for Habitat Suitability Index) to assess how changes in the coastal landscape structures would affect potential SAV habitat. Our project will support integrated research, education, and outreach by involving local volunteers as field guides and assistance, developing a HSI for the state coastal areas, and distributing an education tool that demonstrates how coastal landscape changes will affect the essential fisheries habitat, and providing instruction to the end users on applications of the index. 

For More Information Contact: the MASGC Research Coordinator, Loretta Leist (Loretta.leist@usm.edu). 
Please reference the project number R/CEH-31.