Near-infrared spectroscopy and hyperspectral imaging can aid in the prediction and mapping of polyploid acacia hybrid wood properties in tree improvement programs

Abstract

Abstract Acacia, including Acacia hybrids, are some of the most important species grown as part of the Vietnamese wood industry. Rapid methods to identify the variations in the wood properties of Acacia hybrids however, are a currently lacking and creating limits for their breeding programs. In this study, nine Acacia hybrid clones, including those that were diploid, triploid, and tetraploid were evaluated using near-infrared spectroscopy (NIR) and hyperspectral imaging (HSI). The standard normal variate (SNV) and second derivative (SP2D) were applied to compare the performances of NIR and HSI using partial least square regression. The HSI images were acquired at wavelengths from 1033 to 2230 nm and the SNV and SP2D described the variations in the wood properties. The NIR predicted the wood physical properties better than HSI, while they provided similar predictions for the mechanical properties. The mapping results showed low densities around the pith area and high densities near the bark. They also revealed that the air-dry moisture content changed at different positions within a disk and was dependent on its position within the tree. Overall, NIR and HSI were found to be potential wood property prediction tools, suitable for use in tree improvement programs.