Performance of generic, species-group and species-specific volumetric equations in a Forest Management Area in Brasil Novo, Pará, Brazil

Abstract

Accurate estimation of timber volume is crucial for successful forest management. In tropical forests, high species diversity and the variability of tree biometric attributes increase the difficulty of developing highly accurate generic and/or specific models. One potential solution is the development of species-specific or species-group volumetric models based on similar structural characteristics. This study aimed to fit multispecific volumetric models and to compare the performance of the same models when applied to groups of species with similar characteristics and, finally, to individual species, in order to predict the tree volume of Hymenaea courbaril L. (Jatobá) and Piptadenia suaveolens Miq. (Timborana) in a Forest Management Area in the Brazilian Amazon. Ten volumetric models were tested, with and without the inclusion of commercial height. The dataset was divided into two subsets: 70% for parameter estimation and 30% for validation of model generalisation. The best model was selected using statistical criteria, such as the adjusted coefficient of determination, residual standard error, Akaike information criterion, and the prediction residual error sum of squares. Residual normality, homoscedasticity and independence were assessed using diagnostic plots and statistical tests, while multicollinearity was evaluated using the variance inflation factor. The logarithmic Husch and Schumacher–Hall models showed the best fit and satisfied the assumptions of linear regression, and thus are recommended for predicting the individual tree volume for these species. However, the Husch model is more practical and introduces less bias, as it relies solely on diameter as the predictor variable.