Preliminary Evaluation of Methods for Classifying Forest Site Productivity Based on Species Composition in Western North Carolina
The species indicator approach to forest site classification was evaluated for 210 relatively undisturbed plots established by the USDA Forest Service Forest Inventory and Analysis uni (FIA) in western North Carolina. Plots were classified by low, medium, and high levels of productivity based on 10-year individual tree basal area increment data standardized for initial stocking. Chi-square analysis of contingency tables indicated that productivity classes were not independent (P < 0.05) of the frequencies of occurrence for 4 of 27 common tree species. Multiple logistic regression of a binary variable formed by the high productivity class compared to the combined low and medium classes resulted in a model consisting of elevation and seven significant (P < 0.05) species that produced a classification accuracy of 85 percent; a similar model based on the low productivity class resulted in classification accuracy of 70 percent. A multinomial logistic regression model indicated that elevation and six species were significantly (P < 0.05) associated with the three productivity classes, but overall classification accuracy dropped to 61 percent, mainly due to the poor predictability of low productivity classes. Chestnut oak (Quercus prinus) and serviceberry (Amelanchier spp.) were the most consistent indicator species. Results of this exploratory study suggest that using indicator species for site classification shows promise in hardwood stands by avoiding problems associated Lvith conventional methods based on site index.