Evaluating Habitat Vulnerability and Sustainability of Urban Seagrass Resources to Sea Level Rise

The seagrass resource provides essential ecosystem functions for many marine species. This research evaluated the vulnerability and sustainability of the seagrass resource in an urbanized area to the effects of sea level rise. The assessment required analysis of information regarding the biogeography of the seagrass resource, and developing a method to model the spatial extent of the suitable habitat for seagrass, and applying the model to predict the implications of simulated sea level rise scenarios on the seagrass resource.;Examining the biogeography of the seagrass resource required the development of a seagrass monitoring and assessment field survey and a comprehensive seagrass resource map (SGRM). The mesoscale field survey was designed and conducted in St. Joseph Sound (STJS) and Clearwater Harbor North (CLWN), Pinellas County, Florida from 2006-2010 to determine the seagrass species composition and spatial distribution for the resource. The seagrass species found in the study area consisted of Syringodium filiforme Kutzing ( Syringodium), Thalassia testudinum Banks ex Konig (Thalassia), and Halodule wrightii Ascherson (Halodule). These seagrass species occurred in monospecific and mixed beds in all combinations throughout the study area. Spatially, Thalassia was the dominant nearshore in STJS and Halodule in CLWN. Syringodium was most frequently found in STJS in the mid to deep depths.;The SGRM was mapped from satellite remote sensing imagery with training information from the mesoscale field survey data. Landsat 5 Thematic Mapper (TM) and Earth Observing-1 Hyperion (HYP) were processed to map the seagrass resource in the study area in a nearshore shallow coastal area of Pinellas County, FL, USA. A maximum likelihood classification (MLC) was used to classify both TM and HYP imagery into three classes (seagrass estimated coverage) of the seagrass resource. The overall accuracy for the TM MLC map was 91% (kappa = 0.85) and the HYP was 95% (kappa = 0.92). Due to areas of cloud cover in the HYP image, it was necessary to composite the classification values from the TM MLC to accurately define these areas. The validation accuracy (n=72) of the composite seagrass resource map was 81% which was much more rigorous than the previous accuracy estimates. These results support the application of remote sensing methods to analyze the spatial extent of the seagrass resource.;The development of a spatial habitat suitability model (HSM) for the seagrass resource provided a management tool to better understand the relationship between seagrass, water quality, and other environmental factors. The motivation to develop the spatial HSM was to provide a spatial modeling tool to simulate changes in the water quality environment and evaluate the potential impact on the seagrass resource. High resolution bathymetry and field survey water quality data were used to fit general additive models (GAM) to the STJS (Adjusted R2= 0.72, n=134) and CLWN (Adjusted R2= 0.75, n=138) seagrass resource. The final GAMs included water quality variables including salinity, chlorophyll-a concentration, total suspended solids, turbidity, and light. The only significant variable was the light metric in STJS (p-value= 0.001) and CLWN (p-value= 0.006). The light metric was the logarithmic light attenuation calculated from the water quality field survey transmittance (660nm) data and the high resolution bathymetry. The overall accuracy (OA) of the predictive GAM rasters was higher in CLWN (95%, kappa =0.88) than in STJS (82%, kappa = 0.40). The increased prediction error in STJS was spatially correlated with the areas of lower density seagrass along the deep edge of the bed. While there may be a plethora of factors contributing to the decreased density of the seagrass, this may indicate these seagrass were already living at the edge of the suitable habitat. (Abstract shortened by UMI.).