Characterisation of wood quality of Eucalyptus nitens plantations and predictive models of density and stiffness with site and tree characteristics

Abstract Wood density and stiffness are among the most important wood quality traits for timber products and the prediction of these traits in trees planted across large geographic areas is of increasing interest for the forest industry. This study was conducted across 65 stands of E. nitens fibre-grown plantations, the most widely planted hardwood species in Tasmania, Australia, and aimed to (i) identify the main sources of variability in density (basic density, ρ b ) and stiffness (dynamic modulus of elasticity, E d ) of E. nitens fibre-grown trees, (ii) identify site, climatic, environmental and geomorphological factors across the E. nitens estate and tree characteristics at the site level that influence ρ b and E d of the trees, and based on this information (iii) develop models to predict the two wood properties at the tree and site level. An extensive dataset (n = 1880) of tree ρ b and E d measures was obtained through direct tree wood sampling for basic density and non-destructive acoustic assessments for stiffness. Large variation in tree density and stiffness was found across the study sites and among trees, with basic density ranging from 0.36 g cm−3 to 0.65 g cm−3 and stiffness from 9.2 GPa to 24.7 GPa. Candidate models to predict the two properties were developed on a training dataset and validated on a separate part of the dataset. The best tree-level models describing the variation of ρ b and E d included both tree and site factors. The best model describing ρ b variation on trees included tree diameter, maximum temperature of the warmest period and precipitation of the wettest period as explanatory variables, while the best E d model for trees included tree slenderness, height and basic density and the climatic variables, elevation and site index. The predictive models explained 31% of the total variability of ρ b and 59% of the total variability of E d across the validation sites. At the site level, the best models for density and stiffness included forest structure, climatic, and geomorphological variables, as well as plot location. The model for site density explained 73% of the variability of average density, while the model for site stiffness explained 35% of the variability of average stiffness. These findings demonstrate that different site and tree factors influence the two commercially important wood quality traits, and these can be used in combination to satisfactorily predict tree basic density and stiffness across different environments both at the tree and site level. This knowledge can be employed to support the forest and wood processing industry to expand the possible uses of E. nitens fibre-grown trees, to better manage the harvest for specific products, and to develop models for planning next generation plantings.