Mapping oak (Quercus spp.) stewardship potential within the Nantahala National Forest.

Abstract

Oaks (Quercus spp.) are foundational to the southern Appalachian Mountains. However,the impacts of fire exclusion with subsequent shifts from xerophytic to mesophytic forest types, known as forest mesophication, are resulting in reduced recruitment of oaks and potential future loss. Furthermore, in seeking suitable sites to address the recruitment issue, it is challenging to consistently match the silvics of oak species with suitable sites across the varied, mountainous landscape of the Southern Appalachian Mountains. This site variability can lead to difficulty in identifying areas with high potential for oak success. Thus, tools including spatially explicit maps can aid in developing species- and site-specific silvicultural strategies. The overarching goal of this project is to explore methods to produce stand-level species-specific site suitability maps within the western extent of the Nantahala National Forest in far west North Carolina. Common stand exam quick plot protocol was followed, and plot data collected from June through August 2023 at 180 sample points within 15 stands totaling 500 acres across the southcentral extent of the Nantahala National Forest, also known as the Nantahala Mountains. At each sample point a nested plot design was used which contained three plots including two fixedradius plots to capture regeneration and midstory metrics and one overstory variable radius plot. Initial regeneration conditions quantified the recruitment issue and supported the importance of collecting regeneration height-class data. Then, using field data and other spatially explicit data sets, two popular site suitability modeling techniques, Maximum Entropy (MaxEnt) and Random Forest, were run for three regionally common oak species - northern red oak (Quercus rubra), white oak (Q. alba), chestnut oak (Q. montana) -- and one “other oak” category containing two species -- black oak (Q. velutina) and scarlet oak (Q. coccinea). The MaxEnt models consistently outperformed the Random Forest models, exhibiting higher receiver operating characteristic (ROC) areas under the curves (AUC), and were used for the final site suitability mapping. The most suitable sites, through interpretation of the jackknife significance charts and variable response curves, aligned with the United States Department of Agriculture (USDA) Forest Service’s published silvics for each species and provided additional spatial-nuance for oak-focused silvicultural site selection. The results show that MaxEnt can be a good choice for modeling site suitability, given the availability of descriptive site characteristic variables and species-presence data, and can be powerful during the silvicultural planning and site identification, selection, and prioritization processes. The maps will assist in the prioritization of sites for silvicultural treatments with the primary objective of successfully regenerating and recruiting oaks.