NIR for combined selection in hardwoods for both growth and wood properties

Abstract

Near infrared (NIR) spectroscopy provides an inexpensive and rapid alternative to traditional laboratory methods for estimation of wood properties. Its advantages are well-known as an appropriate approach for tree improvement programs where assessment of a large number of trees is usually required and elite individual trees need to be retained for use as future parents. Hence, tree breeders are beginning to realise the potential of this technology for the selection of species, taxa, provenances, families and individual trees with superior wood quality. Forest products are consumed worldwide for multiple purposes to support daily living. Many of these products are sourced from planted forests. In north-eastern Australia large areas of hardwood plantations have been established in the last 15 years; however, due to poor tree species selection, significant areas have failed. Research to support hardwood plantation development in north eastern Australia included the establishment of taxa trials across multiple sites in Queensland and northern New South Wales. Previous work involving this trial network focused primarily on the growth and survival of the various taxa and species included, and identified a number of promising species, including Corymbia citriodora subsp. variegata (CCV) and Eucalyptus pellita. In this thesis, the initial focus was to evaluate the most promising hardwood taxa across a subset of well-replicated and linked trials, to identify the taxa most suitable for solid wood and pulpwood production across the diverse environments of north-eastern Australia. The second focus was to assess open-pollinated families within different provenances (or seed sources) in two key studied species (CCV and E. pellita) to investigate the genetic control of key wood properties, and relationships among wood and growth traits. Here near infrared spectroscopy was used to rapidly and cost effectively assess the wood properties of these taxa across multiple trial sites and methods developed to deal with unbalanced representation of taxa/species across and within trials. This thesis aims to address the following research questions: Question 1: What are the best taxa for both growth and wood property traits for deployment across the diverse environments in tropical and subtropical regions of Queensland and northern New South Wales? Question 2: How do genetic parameters vary in growth and wood property traits of CCV across three progeny trials in southeast Queensland, Australia? Question 3: How do genetic parameters vary in growth and wood property traits of E. pellita across two progeny trials in Vietnam? Question 4: What selection strategies will maximise genetic gains in all key traits in both CCV and E. pellita? In order to address these questions, growth data (diameter at breast height over bark [DBH] and tree height [H]) and wood samples of 20 taxa have been collected from a large number of trials across a wide range of regions in north eastern Australia and Vietnam. DBH and H were used to calculate conical volume (V, dm3 or m3). All wood samples were prepared and scanned to acquire NIR spectral data that were then used to estimate Kraft pulp yield (KPY, %), basic density (DEN, kg/m3), modulus of elasticity (MOE, GPa) and microfibril angle (MfA, degree). Statistical methods studied included: univariate, bivariate and across-trials mixed models for breeding value prediction and restricted maximum likelihood (REML) approximation of genetic parameter estimation; hierarchical clustering to facilitate grouping of trials into classes based on similarity in the performance of individual taxa; and, principal component analysis to analyse the pattern of performance across sites. Finally, a deterministic simulation was constructed to demonstrate the impact of different weightings for each trait on predicted genetic gains in both CCV and E. pellita. After the initial introduction chapter, the second chapter of the thesis provides information on the productivity and wood properties of examined species/taxa, thence a more robust basis for taxa selection across geographical locations in north-eastern Australia during the first 10 years of age can be better understood. The third and fourth chapters investigate a range of genetic parameter estimates based on open-pollinated progeny trials established in southeast Queensland for CCV and in Vietnam for E. pellita. The fifth chapter constructs the deterministic simulation to iteratively estimate genetic gain for various combinations of relative selection weighting in both CCV for production of solid wood in southeast Queensland and E. pellita for the production of pulpwood on short rotations in Vietnam. This chapter indicates selection weights can be applied to select trees as parents of the next generation of breeding and to select trees for retention in seedling seed orchards, and will lead to genetically improved CCV and E. pellita seed for afforestation in the future. Jointly, this thesis focuses on hardwood forest tree improvements for both growth and wood quality traits.