Modeling Forest Structure and Vascular Plant Diversity in Piedmont Forests

When the interacting stressors of climate change and land cover/land use change (LCLUC) overwhelm ecosystem resilience to environmental and climatic variability, forest ecosystems are at increased risk of regime shifts and hyperdynamism in process rates. To meet the growing range of novel biotic and environmental stressors on human-impacted ecosystems, the maintenance of taxonomic diversity and functional redundancy in metacommunities has been proposed as a risk spreading measure ensuring that species critical to landscape ecosystem functioning are available for recruitment as local systems respond to novel conditions. This research is the first in a multi-part study to establish a dynamic, predictive model of the spatio-temporal dynamics of vascular plant diversity in North Carolina Piedmont mixed forests using remotely sensed data inputs. While remote sensing technologies are optimally suited to monitor LCLUC over large areas, direct approaches to the remote measurement of plant diversity remain a challenge.

This study tests the efficacy of predicting indices of vascular plant diversity using remotely derived measures of forest structural heterogeneity from aerial LiDAR and high spatial resolution broadband optical imagery in addition to derived topo-environmental variables. Diversity distribution modelling of this sort is predicated upon the idea that environmental filtering of dispersing species help define fine-scale (permeable) environmental envelopes within which biotic structural and compositional factors drive competitive interactions that, in addition to background stochasticity, determine fine-scale alpha diversity. Results reveal that over a range of Piedmont forest communities, increasing structural complexity is positively correlated with measures of plant diversity, though the nature of this relationship varies by environmental conditions and community type. The diversity distribution model is parameterized and cross-validated using three high quality vegetation survey datasets, including Duke Forest Korstian permanent plots, Forest Inventory Analysis (FIA), and the scale transgressive, nested module Carolina Vegetation Survey (CVS).