Estimativa de biomassa para colmos de Dendrocalamus giganteus wall. ex munro (poaceae)

Abstract

The present study aimed to evaluate and compare regression models for estimating the biomass of Dendrocalamus giganteus culms using biometric variables that are easily measured in the field. This species exhibits rapid growth and high biomass production potential, making it promising for sustainable forest management and justifying the need for studies focused on accurate biomass quantification. Six regression models were tested and grouped into three categories: non-log-transformed linear models, log-transformed linear models, and nonlinear models. The log-transformed models were obtained through the linearization of nonlinear models by applying the natural logarithm. Data collection involved measurements of diameter at breast height, total height, and culm dimensions, with data obtained from two forest fragments in which 13 Dendrocalamus giganteus culms were harvested, presenting diameters ranging from 6.05 to 17.28 cm and total heights between 10.3 and 28.0 m. In addition, basic density was determined under laboratory conditions, and volume was estimated using the Smalian method, followed by conversion to dry biomass. The predictor variables, in both original and log-transformed forms, were correlated with biomass using Pearson’s correlation coefficient. Linear models were fitted using the ordinary least squares method, while nonlinear models were adjusted using the iterative Levenberg–Marquardt method. Model performance was evaluated based on statistical assumptions and goodness-of-fit metrics. The results indicated positive and significant correlations between biometric variables and biomass, with superior performance observed for the log-transformed models. The log-transformed Spurr model was the only model that fully met the statistical assumptions and presented all significant parameters. It is concluded that logarithmic transformation of variables improves model fitting and predictive performance in estimating the biomass of Dendrocalamus giganteus culms, highlighting the log-transformed Spurr model as a reliable tool for forest inventories and sustainable bamboo management.