Ecological Modeling of Plant Communities on the Ecological Sanctuary of the Long Branch Environmental Education Center

The effects of habitat heterogeneity on plant species have been widely studied throughout the history of ecology. The importance of understanding the direct causes of species success has increased in the face of dwindling wildlife areas due to habitat fragmentation as well as the changing global climate. If the parameters that affect specie's distributions can be identified, then statistical analysis can be performed which can quantitatively evaluate the specific affects of these parameters on plant population composition. A conceptual model can be constructed in order to indicate where strong relationships exist between the characteristics of the environment and plant community composition. Once significant relationships are discovered between species composition and environment, these relationships can be used to predict community composition and other characteristics of the community such as resilience to disturbance. This conceptual model can be used for the creation of environmental education packets that can inform individuals about the requirements for diverse habitats as well as the individual requirements of plant communities and their distributions.

The study area consists of the Long Branch Environmental Education Center and its associated wildlife conservation lands. The area consists of 1,588 acres and is located in the Newfound around the Sandy Mush Basin in Buncombe and Haywood Counties of western North Carolina. The site is located within the Blue Ridge Province in North Carolina in close proximity to the Great Smoky Mountain National Park and Mount Mitchell. The study site will be stratified with respect to aspect and elevation. Dividing the land horizontally into 1000-ft. increments and vertically into aspects of North, South, East, and West will create a grid work of sampling areas. This will result in a series of possible sampling locations, which will be randomly sampled. Three of the properties of the Long Branch Education Center will be divided into one of the possible sampling sites. Big Sandy Mush Bald will represent areas recently disturbed, as it was logged 20 years ago. The Willow Creek property and the property behind the education center will represent relatively undisturbed sites as they have had at least 50 years since logging operation occurred.

Six randomly placed 100m^2 plots will be established in each sample site in early March. Three plots in both of the disturbance regimes. The initial plots will be located using topographic maps and random number generation for location selection. After selection, plots will be located and marked using a topographic map. Direct selection of the plot will be determined randomly with generated compass and distance values. These plots will be surveyed for all woody plants that will be identified and counted. Diameter at breast height measurements will be taken for all trees over 7cm in diameter. A variety of environmental gradients will be measured at each plot. Light levels will be recorded by evaluating canopy cover photographs and determining percent open sky. Aspect will be devised using a compass and recorded as one of the four aspect categories. Slope curvature will be assigned by ranking each plot as convex, side slope, or concave. Slope steepness will be evaluated using an inclinometer. Relative location will be measured using topographic maps to evaluate the distance between nose and hollow for each plot (Stevenson and Mills, 1999). Relative landform index will be calculated by averaging measures of the angle to the horizon in four directions (McNab, 1992).

Ten to twenty nested quadrants of 1 m^2 will be randomly selected within the 100m^2 plots (Small and McCarthy 2002). These plots will be selected by placing a grid pattern over the larger plot and randomly selecting points of intersection along the grid for quadrant locations. The herbaceous vegetation will be surveyed within these quadrants in the spring (April-May) and again in summer (late June-early August). Species will be identified and catalogued. Cover for each species will be estimated using standard cover classes. Soil samples will be collected during spring and summer sampling using a hand held soil core and a hand shovel. Soil will be analyzed for nutrient and moisture content. Soil texture, organic matter, and pH will also be measured from soil analysis. Light may also be evaluated at the quadrant level using photosensitive paper or electronic light meters in order to determine light gradients within each plot.

Species' richness and species' abundance will be calculated for each quadrant and each plot. Richness will be examined in relation to each environmental variable through multinomial logistic regression (Guisan and Harrel 2000). Species' abundance will be analyzed using detrended correspondence analysis (DCA) for both woody vegetation and herbaceous vegetation. This analysis will reveal the major gradient in the environment to which the species are responding. Each environmental variable will be regressed against the first DCA axis in order to determine the environmental gradient reflected by the axis. Due to the problems with correspondence analysis (Austin 2002), the data will also be evaluated using general additive models (GAM). Each species score from the DCA will be compared to the corresponding score from the GAM using statistical tests, which compare distribution. Cluster analysis will be used together with ordination results to identify plant communities.

Disturbance will be evaluated by creating similarity indices between regimes and within regimes. The difference in similarity of plots that are disturbed with one another and with undisturbed plots will be evaluated. Plots that are less similar between the two levels of disturbance will be analyzed for community's characteristics determined from the modeling of the site's environmental variables. These community characters will then be evaluated for effects on disturbance resilience and recovery from logging.

The results from these analyses will reveal the overall plant response on the Long Branch Environmental Education Center land. These statistical tests will be evaluated to create a conceptual model that will allow an individual to relate environmental variables to a specific community or species of plant. These relationships will be used to design a program that will allow individuals to identify environmental parameters and relate them to specific plant communities that exist within the parameters. The model will allow average individuals to use scientific deduction to inform themselves about plant community responses. The model will also convey the processes involved in distribution and success of biologically diverse ecosystems.

Works Cited

Guisan, A. and F.E. Harrell. 2000. Ordinal response models in ecology. Journal of Vegetation Science 11:617-626.

Small, C.J. and B.J. McCarthy. 2002. Spatial and temporal variability of herbaceous vegetation in an eastern deciduous forest. Plant Ecology 164: 37-48.

Stephenson, S.L. and H.H. Mills. 1999. Contrasting vegetation of noses and hollows in the Valley and Ridge province, southwestern Virginia. Journal of the Torrey Botanical Society 126(3): 197-212.