Rapid discrimination of wood species from native forest and plantations using near infrared spectroscopy

Abstract

Aim of study: To verify how well near infrared (NIR) spectroscopy is able to discriminate wood specimens from natural and planted forests. This study was carried out using tropical trees from Brazil.

Area of study: Wood specimens coming from Lavras (21°10′S, 44°54′W), Paraopeba (19°16′S, 44°24′W) and Belo Oriente (19°17′S, 42°23′W) cities, Minas Gerais state, southeastern Brazil were insvetigated.

Material and methods: NIR spectra were recorded in the radial surface of wood specimens of four native species (Cedrela sp., Apuleia sp., Aspidosperma sp. and Jacaranda sp.) and two commercial clones (Eucalyptus for bioenergy and pulp & paper).

Main results: The principal component analysis (PCA) of spectral information revealed that it is possible to distinguish wood from planted and native forests. The dispersion of scores in the graphic formed by the first and second principal component formed two groups allowing differentiating very clearly the Eucalyptus clones from the native woods. The partial least squares discriminant analysis (PLS-DA) allowed the prediction of group of species with a high degree of correct classification. The PLS-DA models performed from untreated NIR spectra obtained 86 to 100% accuracy for the natural wood species.

Research highlights: From PLS-DA of treated NIR spectra, no Eucalyptus wood sample was classified as a natural forest species and vice versa. NIR technique associated with multivariate statistics are promising to discriminate wood specimens from native or planted forests and thus identify frauds.