Assessing sapwood depth and wood properties in Eucalyptus and Corymbia spp. using visual methods and near infrared spectroscopy (NIR)

Abstract

Accurate measurement of sapwood depth (DS) is essential for calculating volumetric water use of individual trees and stands. Various methods are available to measure DS but their accuracy is rarely cross-validated. We sampled 15 Eucalyptus and 1 Corymbia species along a gradient of aridity and obtained reference values of DS in fresh wood cores using light microscopy, which represents our reference method. We compared this method to the simpler and widely used macroscopic method: visual assessment of natural or induced colour change from sapwood to heartwood. In a third method, estimation of DS was based on species-specific models that rely on wood properties measured using near infrared spectroscopy (NIR). Microscopy allowed clear identification of DS based on the presence of blocked vessels. Measurement of DS using microscopic methods was possible for 78 of a total of 80 cores and ranged from 3.6 mm (E. loxophleba) to 43.8 mm (E. viminalis). Macroscopic assessment clearly differentiated sapwood and heartwood in 60 cores. Results from microscopic and macroscopic methods agreed closely ( 50% deviation between estimates), macroscopic measurement across all species agreed well with microscopic assessment of DS (R2 = 0.92). Models developed for differentiation between sapwood and heartwood using NIR spectroscopy were very robust (high coefficient of determination) for four species, but DS could only be predicted well for one (E. obliqua) of the four species. Even after elimination of apparent false estimates, prediction of DS by NIR across species was not as strong as for macroscopic assessment (R2 = 0.88). DS can accurately be measured using microscopy if vessel occlusion is clearly visible. Although slightly overestimated, DS from macroscopic assessment was generally similar to that measured by microscopy. NIR spectroscopy was unable to predict DS with acceptable accuracy for the majority of species. Further improvements in the prediction of DS using NIR will require more intensive model calibration and validation, and may not be applicable to all species.