Estimation of basic wood density of eucalyptus using non-destructive methods
Abstract
The traditional estimation of basic wood density is time- and resource-intensive, limiting broader sampling in eucalyptus plantations. This study evaluated the accuracy of non-destructive methods and developed predictive models using Linear Mixed Models (LMM). Five eucalyptus clones, aged 6.6 to 7.2 years, from three sites were analyzed. Discs collected at various heights were used for basic density determination by destructive methods. Among the non-destructive techniques, NIR spectroscopy showed the highest accuracy, with no significant difference between observed and predicted values, followed by Pilodyn, which differed by only 1 kg m^−3. The Increment Borer method did not produce reliable models. Genetic material and growth site significantly influenced the accuracy of the methods, emphasizing the need to include these variables in predictive models. The adjusted LMMs confirmed that integrating genetic and site factors enhances estimation robustness. NIR emerged as the most reliable method, while Pilodyn presents a practical field alternative. Future work should expand datasets to incorporate more genetic variations, growth conditions, and tree ages to improve model calibration.