Assessing the Role of Invasive Species in Successional Plant Communities Using Hyperspectral Remote Sensing

Abstract

I AbstractBiodiversity has been correlated with ecosystem stability and function. Invasive species can decrease biodiversity and cause ecological and economical damage. These species are especially prevalent in abandoned agricultural fields, and can disrupt secondary successional dynamics. Although field surveys are used to assess the impacts of these species, remote sensing can be more efficient, especially at large spatial extents. I examined the ability to use ground-level hyperspectral remote sensing to study invasive species and their effects on community properties at the Blandy Experimental Farm (BEF) in north-central Virginia. Their effects on ecosystem properties could not be assessed using ground-level remote sensing in this system; thus, I used leaf and soil measurements to assess their effects on secondary succession.I found that remote sensing can be used to differentiate among plant communities.The most influential species to community discriminability are considered invasive, suggesting that these species can strongly influence species compositions and other community properties. The most influential wavelengths for discrimination were distributed throughout the spectral profile and corresponded with plant physiological and structural elements.Thus, spectral differences across species were large enough to be used in aggregate to differentiate plant communities. Additionally, these differences were large enough to differentiate individual species, but discriminability varied by species. The two thistle species that are similar phylogenetically and structurally were readily distinguished amongst each other. However, the shrubby buckthorn was difficult to distinguish from the oriental bittersweet vine despite phylogenetic distance and differences in structure. This was likely due to physical overlap in the field and thus the difficulty in obtaining pure signatures for discrimination. Discriminability also differed by the spectral region examined. Buckthorn and oriental bittersweet were least discriminable in the 550-599 nm and 650-699 nm regions, due 2 to greater intraspecific variability of spectral data in these regions.Additionally, remote sensing can be used to estimate higher order community properties like species diversity, and thus assess the effects of invasive species on plant diversity.However, correlations between species diversity and spectral diversity varied by the spectral transformation technique used and the spectral region examined. There was a strong positive correlation between the two in the visible region when band depth was used and in the near-infrared region when first derivatives were used. There were no strong correlations in the red edge, due to high intraspecific variability in chlorophyll content.Lastly, I assessed the effects of these exotic invasive species on ecosystem properties, specifically secondary succession, using leaf and soil data. Differences in soil characteristics were larger across fields and stages than across species. However, there were species-level ...