Assessing age- and silt index-independent diameter growth models of individual-tree Southern Appalachian hardwoods
Abstract
Models of forest vegetation dynamics based on characteristics of individual trees are more suitable to predicting growth of multiple species and age classes than those based on stands. The objective of this study was to assess age- and site index-independent relationships between periodic diameter increment and tree and site effects for 11 major hardwood tree species. Model formulations of western conifers were evaluated by multiple regression for significance of three types of variables affecting diameter growth: tree size, competition, and site effects. Many variables were highly correlated with diameter growth but were also intercorrelated. Importance of size or competition variables differed by species and parsimonious models of highly significant (p<0.01) variables explained an average of 60% of the total variation associated with diameter increment. One or two topographic or geographic site-effect variables were significant in models for seven species. but accounted for little (<5%) variation. Detailed evaluation of a yellow-poplar (Liriodendron tulipifera) model indicated prediction errors were not correlated with tree size. Stand age, or site index. Validation testing of the yellow-poplar model with independent data suggested prediction errors were uniformly distributed, but strongly biased. Unexplained bias was associated mostly with tree size. The formulation of diameter growth models for western conifer species is appropriate for eastern hardwoods, but reasons for the unexplained bias should be addressed.