A Preliminary Test of Estimating Forest Site Quality Using Species Composition in a Southern Appalachian Watershed

Characteristic arborescent communities of mesophytic or xerophytic species have long been recognized as indicative of forest site quality in the Southern Appalachians, where soil moisture availability is the primary environmental variable affecting productivity. But, a workable quantitative system of site classification based on species composition is not available. We devised a prototype expert system by assigning a relative moisture weight to upland forest tree species according to their position of modal occurrence on a soil moisture gradient ranging from xeric to mesic. We classified forest sites by their position on the gradient, which was quantified by an index representing the average moisture weight for all species present. We determined the relationship of the moisture index with upland oak site index on permanent plots dominated by even-aged stands of either mixed oaks (Quercus spp.) or yellow-poplar (Liriodendron tulipifera). Regression analysis indicated the moisture index was significantly (P < 0.001) associated with observed site index and explained 62 percent of its variation. Validation of the model with an independent dataset resulted in a mean absolute error in oak site index of 6.9 feet. Results of this exploratory study suggest that estimation of site index based on species composition has potential for application in mixed upland hardwood stands of the Southern Appalachians. INTRODUCTION The productive capacity of forest stands strongly influences their response to silvicultural treatments (Smith 1962). Site index is the method most used to evaluate site quality in eastern upland hardwood stands (Carmean 1970) and Beck and Trousdell (1973) provide a thorough description of the method, particularly its underlying assumptions and limitations. Estimation of site index in mixed hardwoods is often problematic, however, because suitable sample trees are often lacking, particularly on sites of intermediate or lower quality where oaks (Quercus spp.) and hickories (Carya spp.) predominate (Carmean 1970). Estimated site index may be biased in many stands, therefore, resulting in erroneous classifications of productivity. Replacement of conventional site index estimation based on sample trees with an alternate method is highly desirable for ecosystems where oaks are important. One such method is a procedure reported by Whittaker (1956) for arraying stands on environmental gradients based on composition of the tree stratum. Forest productivity in the Southern Appalachian Mountains is associated primarily with temperature and moisture gradients (Whittaker 1966) and somewhat with fertility. Whittaker (1966) reported that “. . . an index of site moisture conditions based on weighted averages of stand composition . . .” was highly correlated with forest production. He subdivided the topographic-soil moisture gradient within broad elevation zones into four soil moisture classes (mesic, submesic, subxeric, and xeric) and assigned a weight to each class. Each tree species was assigned to a soil moisture class based on its modal frequency of occurrence along the gradient. Whittaker used the weighted average of each species present >1-inch d.b.h. as an index of the soil moisture conditions for a site. The index, which was a means for quantifying the relative position of sites on the moisture gradient, was highly correlated with primary forest production for vegetative communities occupying environments ranging from xeric to mesic in the Great Smoky Mountains National Park (Whittaker 1966). The simplicity of such a site classification system is appealing for a number of reasons: it can be readily applied with data typically collected from sample plots in a systematic inventory of stand conditions, it is easily adapted to other ecosystems with their associated species, it can be extended to other environmental gradients of temperature and nutrients, the system has an ecological basis because it is not based on commercial timber species as is site index, and its underlying basis is easily conveyed to other audiences. This report describes results of a method of forest site classification based on arborescent species composition of upland hardwood stands in the Southern Appalachians. Previous work on the method (McNab and others 2002, 2003) utilized landscape scale datasets to explore possibilities of using species composition for site classification, but our current study is the first assessment using field data appropriate for model development and accuracy testing. This exploratory study investigated the question: Is a measure of species composition correlated with site index? Because methodology for the proposed method of site classification has not been evaluated, the purpose of this investigation was to obtain information on the type of vegetation data to collect. Such information included the strata of vegetation to inventory, e.g., saplings, trees, and the diagnostic value of rare species and ubiquitous species. Other important questions dealing with appropriate species weight values and if an index of moisture regime is more strongly related to growth response of silvicultural treatment or other ecological responses than to site index, will be addressed in future studies. Therefore, the intent of this exploratory study was the initial assessment of a new technique to determine if further evaluation and development is warranted and not to report a new method of site classification for immediate application.