Abstract

HERNANDEZ, M. A.. Developing NIR spectroscopy models for a breeding programme of Eucalyptus grandis for timber production. Quebracho (Santiago del Estero) [online]. 2019, vol.27, n.2, pp.76-76.  Epub Mar 01, 2019. ISSN 0328-0543.

In order to screen genotypes of Eucalyptus grandis, near infrared spectroscopy models were developed to estimate physical and mechanical wood properties for a breeding programme aimed at producing solid wood products. For this purpose, 60 twelve-year-old trees from three progeny trials were felled to obtain wood samples. One hundred samples, which represented the species wood variability, were used to make the physical and mechanical measurements. The physical variables measured were basic and air-dried density, while mechanical variables included the flexural properties modulus of elasticity and modulus of rupture as well as wood hardness. Modulus of elasticity and modulus of rupture were measured through a universal testing machine in static bending, whereas wood hardness was determined on tangential and radial faces by means of Janka hardness test. Partial least squares regression was the technique used to predict wood properties. The NIR spectroscopy models yielded similar errors for calibration and validation sets. These models had coefficient of determination (R2.) ranging from 0.74 to 0.93, while their ratio of performance to deviation (RPD) fluctuated from 2.0 to 3.7. The most reliable models were those developed for basic and air-dried wood density, because their RPD values were the highest; in contrast, the models for modulus of elasticity, modulus of rupture and hardness had lower RPD than the models for wood density, making them suitable for rough screening. Near infrared spectroscopy estimated physical and mechanical wood properties accurately, enabling the breeding programme of Eucalyptus grandis to ease and speed up measurements.

Keywords : hardness; near infrared spectroscopy; static bending test; tree selection; wood quality.

        · abstract in Spanish     · text in English     · English ( pdf )