Alternative models for estimating woody plant biomass

Abstract

A number of untransformed regression models were compared to the log-log form of the allometric function for estimating biomass. Models were evaluated using two tree species, Betulapapyrifera Marsh, and Pinusresinosa Ait., and one tall shrub, Ilexverticillata (L.) Gray, with total aboveground biomass as the dependent variable. Using goodness of fit as the criterion, both weighted linear and weighted nonlinear models proved to be acceptable alternatives to the transformed allometric equation. Weighted models retain the advantage of the log-log form, i.e., compatibility with the homogeneity of variance assumption, but avoid the transformational bias.