Abstract Detail



Ecology

Richards, Jennifer [1], Gann, Daniel [2], Harris, Brittany [3], Redwine, Jed [4].

Using WorldView 2 satellite data to map patterns of vegetation change in Northeast Shark River Slough, Everglades National Park .

In response to on-going restoration affecting water flow, vegetation in Everglades National Park’s (ENP) Northeast Shark River Slough (NESRS) is expected to change.  Two restoration projects were completed in 2012/13.  In order to monitor vegetation responses, we used bi-seasonal WorldView-2 (WV2) satellite data from 2010 wet/2013 dry seasons and 2016 wet/2017 dry seasons to map vegetation in a 146 km2 region of interest (ROI) in NESRS.  WV2 data was processed in R using a random forest classifier trained by spectral reflectance patterns of > 35,000 training samples.  Twenty-two vegetation classes were mapped.  The resulting 2010 and 2016 maps were then cross-tabulated to establish type and location of vegetation changes. NESRS is dominated by sawgrass vegetation classes, which covered 71% of the ROI in 2010 and 73% in 2016.  Short-graminoid marsh classes represent 13% (2010) and 11% (2016) of the landscape.  The northern region has prominent halos of woody vegetation centered on culverts that historically delivered water below the Tamiami Trail.  Woody vegetation increased from 2010 to 2016, reflected in an increase from 2% to 3% total cover in shrub/tall-graminoid vegetation.  Marl prairie is dominant in the southeastern corner of the ROI.  The central and southcentral/southwestern areas of the ROI have sawgrass and short-graminoid marshes that reflect historical flow in the southwesterly orientation of tree islands.  These general landscape features for the ROI were similar in both maps, but vegetation densities changed greatly.  Regular sawgrass decreased from 43 to 37% cover, while sparse sawgrass increased from 23 to 38% cover.  These changes were especially apparent in the western region of the ROI.  Even when net change estimates fall within the margin of error (e.g., net woody increase of 1%), their spatially explicit losses and gains across the landscape are larger and are spatially aggregated, indicating that they are real.  Vegetation patterns in the 2010 map reflected the shapes of fires mapped by ENP from 2001 through 2009, while vegetation changes between 2010 and 2016 paralleled fire outlines from this interval.  This suggests that fire has interacted with increasing hydroperiods to reduce vegetation density in sawgrass communities and points to the possibility that fire can be used to systematically thin marsh vegetation over large areas.  An increase in sparse sawgrass, however, was also seen on the eastern edge of NESRS.  This region had no major fires between 2010 and 2016, so change in this area could reflect increased hydrology from restoration modifications.


1 - Florida International University, Department Of Biological Sciences, Miami, FL, 33199, United States
2 - Florida International University, GIS/RS Center, 11200 SW 8th St., Miami, FL, 33199, United States
3 - Florida International University, Dept. of Biological Sciences, 11200 SW 8th St., Miami, FL, 33199, United States
4 - Everglades National Park, South Florida Natural Resources Center, 950 N. Krome Avenue , FIU, Dept. of Biological Sciences, Homestead, FL , FL, 33030, United States

Keywords:
remotely sensed data
Cladium
vegetation mapping
mapping wetlands
Restoration 
sawgrass marsh
marl prairie.

Presentation Type: Oral Paper
Number: 0007
Abstract ID:300
Candidate for Awards:None


Copyright © 2000-2019, Botanical Society of America. All rights reserved