摘要:
Savannas are ecosystems with a background herbaceous layer and intermittently distributed woody plants. A large body of literature has revealed that the ecological processes within savannas are complex and the spatial pattern and abundance of woody and herbaceous plants through time may be influenced by various factors. This dissertation research consists of three studies demonstrating the feasibility of utilizing remote sensing techniques to better understand the ecological complexity of semisavannas in southern Arizona, USA.;The first study investigated the impacts of recent fire disturbance on structural changes of woody plants at varying scales. Results indicated that while field canopy cover and remote sensing woody cover fraction were strong predictors of woody biomass at local and landscape scales, respectively, fire history can significantly alter the nature of these relationships. This work suggested that simple predictions of woody biomass from field and remote sensing cover measures without considering disturbance will underestimate biomass in mature undisturbed settings, and overestimate biomass in recently disturbed locations in most cases.;The second study investigated the ecological stability by implementing a top-down approach to analyze 21 years (1984--2005) of Landsat satellite data. This work suggested that a stable system needed to receive sufficient precipitation for basic plant growth, to be on coarse-textured and shallow soils that can efficiently store precipitation but inhibit the proliferation of woody plants, and to be on east facing slopes which can support clement microclimate. The perspectives gained from this study will enable us to target fine-scale field studies seeking to address circumstances conferring ecosystem stability; and improve predictions of potential carbon stocks in drylands.;The third study derived the unique temporal and spatial signatures of vegetation [the Normalized Difference Vegetation Index (NDVI)] and brightness (red and near-infrared reflectance) signals measured by the Moderate Resolution Imaging Spectroradiometer (MODIS) in areas invaded by a South African tufted perennial grass Eragrostis lehmanniana. Results indicated that biodiversity, and ecophysiological and morphological characteristics of E. lehmanniana were key factors that permitted the discrimination of invaded sites from native vegetation in terms of temporal and spatial variations of MODIS NDVI and brightness time-series data.
展开