Pixel based and object based vegetation community classification methods were performed using 30 meter spatial resolution Landsat satellite imagery of the Arthur R. Marshall Loxahatchee National Wildlife Refuge (Refuge), a remnant of the northern Everglades. Supervised classification procedures using maximum likelihood and parallelepiped algorithms were used to produce thematic maps with the following vegetation communities : wet prairie, sawgrass, cattail, tree island, brush, aquatic/open water. Spectral data, as well as NDVI, texture and principal component data were used to produce vegetation community classification maps. The accuracy levels of the thematic maps produced were calculated and compared to one another. The pixel based approach using the parallelepiped classification algorithm on the spectral and NDVI dataset had the highest accuracy level. A generalized form of this classification using only three vegetation communities (all wet prairie, tree island/brush and aquatic/open water) was compared to a previously published classification which used 1987 SPOT imagery in order to extract information on possible vegetation community transitions that are occurring within the Refuge. Results of the study indicate that 30 meter spatial resolution may be useful for understanding broad vegetation community trends but not species level trends. Pixel based procedures provide a more accurate classification than object based procedures for this landscape when using 30 meter imagery. Lastly, since 1987 there may be a trend of tree island/brush communities replacing wet prairie communities in the northern part of the Refuge and a transition to wet prairie communities in place of tree island/brush communities in the southern portion of the Refuge. / by Dorianne M. Barone. / Thesis (M.S.)--Florida Atlantic University, 2008. / Includes bibliography. / Electronic reproduction. Boca Raton, FL : 2008 Mode of access: World Wide Web.