Proposed theoretical and experimental work

Theoretical work

Over the years there have been a number of attempts to mathematically model plant cell walls (usually fibres or tracheids) from cell wall constituents e.g. \citet{mark1967cell,Astley_1998,Yamamoto_2002} and \citet{Kojima_2004}. However very few efforts have used these techniques to investigate the formation of growth stresses \citep{Archer_1987,Yamamoto_1998,Guitard_1999}).
Currently the most advanced model for how growth stresses develop within the cell wall was presented by \citet{Alm_ras_2005} using the unified hypothesis \cite{okuyama1986,Okuyama_1994,yamamoto1991,Yamamoto_1998,Yamamoto1992} utilising both the lignin swelling and cellulose contraction hypotheses. For details see Section \ref{section:GS}.
Proposed model of the cell: Modelling of a generic single cell with variable cell wall parameters to investigate the required geometry and constituents to create maximum longitudinal and tangential extension and contraction via the lignin swelling hypothesis. The single cell model should have the capacity to put limits on the magnitude of stress generation by lignin swelling under different constituent and geometric makeups.
Because the proposed experiments (see Section 1.2) induce tension wood in species both with and without G-layers an experimental upper limit of the stress generation by lignin swelling can be measured and compared to the theoretical data derived from the model.
It is expected that the base model and parameters will be similar to those utilised to describe lignin swelling by \citet{Alm_ras_2005} and \citet{Yamamoto_1998}. Cell wall layer radii, thickness, MFA, moduli of the CMF bundles and matrix will all be included. Additional variables will be included as necessary. It is intended to add the standard deviation of the MFA within the cell wall layers, as in \citet{Harrington_1998}, pore size (or conversely fibril aggregate size) \citep{Fahl_n_2005,Chang_2014,BIORESBioRes_07_1_0521_Salmen_OSSR_Struc_Organis_Wood_Polymers,Kim_2011} and cell wall constituents \citep{Baba_2009,Donaldson_2001} and layer properties/geometries \cite{Bergander_2002,grozdits1984differentiation,Alm_ras_2005,Yamamoto_1998,Chang_2014,Yamamoto1995} to form a model, conceptually similar to the qualitative architecture presented by \citet{Mellerowicz_2011}, \citet{Salm_n_2009} and others. Boundary conditions will be initially derived from those presented by \citet{Alm_ras_2005} and further modified for increased realism and/or usability of later models.
One of the major differences between the model presented here and those in previous literature is the inclusion of intertwining cellulose fibril aggregates. Recently \citet{Chang_2014} measured the pore size and shape within tension wood and opposite wood of poplar during cell wall maturation. With this recent advancement, reasonable assumptions around how regularly fibrils interact with each other can be made. It is thought that these pores occur between joining fibrils connecting the macrofibrils into the larger structure that is the forming cell wall. If the deposition of lignin into the pores forcing the fibrils apart is the mechanism by which growth stresses develop the quantity and geometry of pores are important parameters to investigate as they will largely affect the ability of the mechanism to cause stress.
The model is limited to a single unconstrained cell, which differs significantly from a cell within a tree, due to the constraining effects of the surrounding wood. Boundary conditions will be used to minimize this problem because modeling a significantly large volume of wood is not feasible. The minimum resolution which will be considered is the fibril aggregate, these are thought to be the smallest structural components responsible for growth stresses under the lignin swelling hypothesis.

Experimental work

Currently neither lignin swelling or cellulose contraction (described in Section \ref{section:GS}) have any direct experimental evidence. The tension which cellulose is under on the stem periphery has been directly measured using x-ray diffraction showing a strain reduction of 0.2% in cellulose when the stress is released \citep{Clair_2006}.
Experimental evidence of the G-layer providing contraction within tension wood has been presented by \citet{Goswami_2008}. Longitudinal extension and tangential contraction were observed when the G-layer was enzymatically removed from poplar tension wood samples. The S2 layer was reported to have a high MFA (36 degrees) as has been reported previously for other G-layer producing species \citep{M_ller_2006}. \citet{Goswami_2008} suggested lateral swelling of the G-layer caused the contraction.
The primary goal of the set of experiments which will be presented within this chapter is to attempt to identify which cell wall constituents contribute to stress generation. In order to evaluate stress generation mechanisms a number of experimental techniques have been identified.
Basic cell wall anatomy and geometry needs to be investigated for the NZDFI species involved in this project. Where possible literature values will be used to approximate values for model parametrization.
The following properties are required:
The cell wall anatomy of different wood types (tension, normal and opposite) needs to be investigated for the various NZDFI species (principally E. bosistoana ). The anatomy study will consist of investigating which species produce a G-layer (microscopy with staining \citep{Qiu_2008}), cell wall architecture (atomic force and electron microscopy), and the MFA and MFA standard deviation in all three wood types (X-ray diffraction). Within tension wood the G-layer (in G-layer producing species) will be removed in order to determine the secondary cell wall properties of tension wood (enzymatic removal).
Note: growth stresses for a large number of samples will be collected during the breeding work, however because of the time consuming nature of the experimental works presented here only a small number of specimens will be tested as needed.
In order to produce the three types of wood required two different growth manipulation techniques are suggested:
Technique one: Young stems (less than three month old growth from coppice) will be restrained to a loop, similar to \citet{jacobs1945l} and allowed to grow for approximately three months, with regular adjustments of the restraints to make sure the cambium is not damaged.
Technique two: Straight one year old stems (from coppice, and seedlings of a mixture of E. camadulensis, E. tricarpa and E. quadrangulata will be bent and restrained and allowed to grow for a further three months, with regular adjustment of the restraints to avoid cambium damage. Normal wood samples can be collected from these stems higher up where they are straight.
The following experiment is proposed in order to investigate the proportion of the stem reorientation that is due to the G-layer. During growth tension wood production is induced by forcing curvature into the living stem, as described above. By introducing an enzyme treatment to the plant while it is still transpiring it should be possible to degrade the G-layer and reverse any straightening that was caused by the G-layer. With the G-layer removed the remaining stress can be released.
Due to the expense, time, equipment and expertise required to test all of the properties needed to parameterise the model presented in section 1.1 a number of properties will be taken from literature and assumed to be a good approximation for the current work. It is likely to be the case that a number of these properties are not from Eucalyptus species.

Breeding

Growth stresses cause a number of issues for harvesting and milling timber, tree breeding programs can be used in order to select for genetics which reduce these effects. There is no reason to expect breeding for growth stress differs significantly from breeding trees for any other trait. Over the last few decades many advances have been made in experimental and statistical techniques which rapidly improve the time and accuracy of tree breeding.
It is suspected that the most efficient way to minimise the issues growth stresses cause during the production of timber is through appropriate genetic selection. Eucalyptus species, in particular E. bosistoana are showing promise within the NZDFI trials to produce high value naturally durable structural timber. To make this product profitable, growth stresses need to be reduced to minimise the effects discussed in Section xxxx. Within the NZDFI project there are a number of additional concerns for breeders (such as durability, form and growth rate). Using conventional breeding methods discussed below, growth stresses will be minimised within the NZDFI genetics. Currently two trials have been established or will soon be established, these include:
All trials at Harewood are set out as randomised individual tree trials. Principally this work will be concerned with E. bosistoana of which there are two trials. One with 8 replicates of 20 families, planted in 2012 and coppiced in 2013, due to be harvested in spring/summer 2015. The other E. bosistoana trial has 10 replicates of 20 families, was planted in 2010 and harvested for the first time in 2012, the plants were then coppiced and harvested again in December 2014. Four families representing the highest and lowest growth stress generating genetics were coppiced for a second time and will be due for harvest in 2016. Preliminary results from the 2012 and 2014 harvests show heritability of growth strain generated family rankings. The same data was collected from E. argophloia plants planted in 2010 measured and coppiced in 2012, with final measurements completed in 2014.
Due to the success of the previous trials NZDFI is currently setting up a  10,000 plant trial with 200 families of E. bosistoana and 336 seedlings over 13 families of E. argophloia . The trials are set up as alpha lattices and harvest is expected in late 2016 or early 2017.
Note: the term family is used here to mean the same mother, but not necessarily the same father. If seeds are collected at different times even from the same tree variability exists due to the possibly of different fathers. Also some Eucalyptus self propagate, but it is unknown what proportion of seeds are self propagated within the NZDFI genetic material. Any effects of self propagation are ignored.
Within the structure of the breeding program there is the ability to statistically check within family rankings from young stems and more mature stems, although the trees are grown under different environmental conditions.
In order to select for low growth stress producing families experimental tests need to be undertaken on each plant. The main test to be used to determine the extent of growth stress within the breeding population is the split test (also known as the Pairing Test) as described in \citet{Chauhan_2010}. The test involves taking a straight 500mm long stem section and cutting along the pith to create a radial split. The diameter of the stem is taken before testing and the width of the opening measured immediately after splitting. Once the opening is measured the stem is cut across the grain to give two samples (one from each side of the split). Density is measured by measuring the mass (using balances) and volume using the displacement method on each of the pieces. Acoustics are also taken using Wood-Spec to calculate the dynamic modulus, and hence the stress can be derived \citep{Chauhan_2010}. Due to the size of the Woodville trial it may be the case that fewer tests are carried out. Decisions on the essential tests will be made near the time of harvesting.
It is expected that typical statistical techniques used in breeding will be applicable. The two objectives to be achieved for the breeding trials are to estimate genetic parameters, (in particular heritability of growth strain) and to predict breeding values for families. Mixed model methods are commonly used within tree and animal breeding for these purposes \citep{White_2007}.
The breeding work undertaken in the thesis will be limited to the investigation of growth strain in the NZDFI Woodville and Harewood E. bosistoana trials. The assumption is made that families which rank well at young ages will rank well at a commercially viable harvest size.