EVALUATING THE IMPACT OF STAND COMPOSITION AND COMPETITION ON
WHITE SPRUCE AND ASPEN WOOD ATTRIBUTES IN THE BOREAL
MIXEDWOOD
by
Ryan Jackalin
BSc., University of Victoria 2017

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF SCIENCE IN NATURAL RESOURCES AND
ENVIRONMENTAL STUDIES

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
April 2021
© Ryan Jackalin, 2021

Abstract
I evaluated how intra- and inter-specific competition affects the development of eleven wood
attributes of trembling aspen (Populus tremuloides Michx.) and white spruce (Picea glauca
(Moench) Voss) over 34 years. My analysis was conducted in a mixedwood trial site in Northern
British Columbia, Canada, that included treatments consisting of 0, 1000, 2000, 5000, and 10000
stems per hectare of aspen. Competition was found to negatively influence wood attribute
development in aspen and positively impact spruce (at low levels of competition). Plot level
competition indices were the best predictor of variation in aspen wood attributes, while stand
level competition (population density) best explained the majority of spruce wood attributes.
Maintaining aspen at lower densities in intimate mixture can have a positive effect on spruce
wood quality, while incurring relatively small reductions in spruce volume production and also
retaining the ecological benefits associated with managing for mixed stands.

2

Table of Contents
TABLE OF CONTENTS ............................................................................................................................ 3
LIST OF FIGURES .......................................................................................................................................... 4
LIST OF TABLES ........................................................................................................................................... 5
ACKNOWLEDGEMENTS ................................................................................................................................ 6
1.

INTRODUCTION ............................................................................................................................... 7

1.1
1.2
1.3
1.4
1.5
1.6

BOREAL FOREST INTRODUCTION AND OVERVIEW ............................................................................... 7
MIXEDWOOD STAND DYNAMICS .......................................................................................................... 8
COMBINING ABILITIES OF MIXED STANDS............................................................................................ 8
WOOD ATTRIBUTE OVERVIEW ........................................................................................................... 10
COMPETITION IN THE MIXEDWOOD.................................................................................................... 13
THESIS OBJECTIVES ........................................................................................................................... 14

2. EVALUATING THE IMPACT OF STAND COMPOSITION AND COMPETITION ON
SPRUCE AND ASPEN WOOD ATTRIBUTES IN MIXEDWOOD STANDS .................................. 16
2.1
2.2
2.3
2.4
2.5

ABSTRACT....................................................................................................................................... 16
INTRODUCTION ............................................................................................................................. 17
MATERIALS AND METHODS ....................................................................................................... 20
RESULTS .......................................................................................................................................... 26
DISCUSSION .................................................................................................................................... 37

3. A MIXED EFFECT APPROACH TO EVALUATING THE IMPACTS OF STAND
COMPOSITION, AGE AND NEIGHBOURHOOD COMPETITION ON THE WOOD
ATTRIBUTES OF SPRUCE AND ASPEN GROWN IN MIXEDWOOD STANDS......................... 43
3.1
3.2
3.3
3.4
3.5
3.6

ABSTRACT....................................................................................................................................... 43
INTRODUCTION ............................................................................................................................. 44
MATERIALS AND METHODS ....................................................................................................... 47
RESULTS .......................................................................................................................................... 52
DISCUSSION .................................................................................................................................... 69
CONCLUSION .................................................................................................................................. 74

4.

CONCLUSIONS AND FUTURE WORK ....................................................................................... 75

4.1
4.2

CONCLUSIONS .................................................................................................................................... 75
LIMITATIONS AND FUTURE WORK ...................................................................................................... 78

5.

BIBLIOGRAPHY .............................................................................................................................. 80

6.

APPENDIX ......................................................................................................................................... 86

3

List of Figures

Figure 2.1 A map of Siphon Creek trial site with the treatment map (Richard Kabzems) ........... 20
Figure 2.2 Boxplot showing the microfibril angle variation between aspen and spruce in each
treatment. ...................................................................................................................................... 30
Figure 2.3 Density change over time with separated aspen (top) spruce (bottom) and treatment.
....................................................................................................................................................... 31
Figure 2.4 Microfibril angle change over time with separated (a, left) aspen, and (b, right) spruce
and treatments. .............................................................................................................................. 32
Figure 2.5 Modulus of Elasticity change over time with separated (a, left) aspen, and (b, right)
spruce and treatments. ................................................................................................................... 33
Figure 2.6 A relative improvement heatmap for all 11 wood attributes within spruce ................ 35
Figure 2.7 A relative improvement heatmap for all 11 wood attributes within aspen.................. 36
Figure 3.1 Microfibril angle change over time for aspen samples (top), and spruce samples
(bottom), separated by treatments. ................................................................................................ 51
Figure 3.2 Aspen and MNE (top) with spruce and BAL (bottom)) showing the proportion that
each competitor species, aspen (A), and spruce (S), contributes to the overall index, both (B). 65

4

List of Tables

Table 1.1 A summary of each of the wood attributes that were analyzed in this study ............... 10
Table 2.1Descriptive statistics of sampled trees. Diameter at breast height (DBH) means are
shown with standard deviations in parentheses. ........................................................................... 22
Table 2.2 Summarized results of the multi model inference analysis of the global model (linear
model = Ring + species + treatment + |tree). ................................................................................ 26
Table 2.3. Results of species separated ANOVA of model terms for each species (Full results for
each wood attribute in Chapter 2 Supplementary Appendix). ...................................................... 28
Table 2.4. The results of species separated two-way analysis of variance (ANOVA) of the effects
of treatment (aspen population density) and cambial age on each wood attribute. ...................... 29
Table 3.1 A summary of the results from the evaluation of the neighbourhood level competition
index mixed effect models (a < 0.05)........................................................................................... 54
Table 3.2 Summarized aspen sample tree p-value results for the competition index term of each
mixed effect model iteration (Full model results in Chapter 3 Supplementary Appendix). ......... 55
Table 3.3 Summarized spruce sample tree p-value results for the competition index term of each
mixed effect model iteration ......................................................................................................... 56
Table 3.4 The best neighbourhood competition term was determined by comparing the eight
competition index terms, for each of the eleven wood attributes, for both aspen and spruce
samples. ......................................................................................................................................... 57
Table 3.5 The best neighbourhood competition term was determined by comparing the eight
competition index terms, for each of the eleven wood attributes, for both aspen and spruce
sample, and ranked by AIC value. ................................................................................................ 58
Table 3.6 A summary of the R2 values calculated by the effect size (Rb2) statistic results. ......... 68
Table 6.1. Full results of the MuMIn analysis that forms the basis for table 2.2. Multi model
inference analysis of the global model.......................................................................................... 86
Table 6.2 The results of the analysis of variance of the effects of treatment (aspen population
density), cambial age and species on each wood attribute........................................................... 91
Table 6.3 Aspen results table of complete post-hoc test of species separated linear mixed model
results ............................................................................................................................................ 92
Table 6.4 Spruce results table of complete post-hoc test of species separated linear mixed model
results ............................................................................................................................................ 95

5

Acknowledgements
I would like to begin by acknowledging the people who made completing this project
possible. Thanks to my family: Haley, Barb and Eric Jackalin for always supporting me and at
least feigning interest when I talk to them in great detail about the science of rocks and trees. I’m
forever grateful to my Oma and Opa for teaching me to set goals and follow through. I’m
thankful to have met so many great friends at UNBC who made the class work and late nights
enjoyable. In particular, thanks to Dan Larson (who we tragically lost in 2020), for always being
a friend that everyone could count on. Thanks to Wanda, Tilly, and Larry Corrigall for reminding
me what is important and for keeping things light. Of course, thanks (for infinitely more than just
helping me through this thesis) to my life partner, Elena Corrigall, for being there through all the
ups, downs and everything in between. Finally, many thanks to my supervising committee Dr.
Lisa Wood, Richard Kabzems and Dr. Che Elkin for guiding me through this process and
providing invaluable help along the way.

6

1. Introduction
1.1 Boreal forest introduction and overview
Boreal forests are frequently composed of broadleaf and conifer trees interacting and
competing for limited resources. These forests are among the most productive and diverse forest
ecosystems in North America (Chen and Popadiouk 2002). Aspen (Populus tremuloides Michx.)
and white spruce (Picea glauca (Moench) Voss) are the predominant components of the Western
Canadian boreal mixedwood forest and competition driven succession between these species is
the primary process of forest composition shift (Jiang et al. 2018). The forest industry in western
Canada has largely focused on the utilization of softwood, however, reductions in the predicted
midterm supply, pine beetle infestation, projected future climate change, and a shift towards the
valuation of products such as carbon storage all portend to an increased use of hardwood species
(McCulloch and Kabzems 2009; Mbogga et al. 2010; Dhar et al. 2016). Previous studies have
evaluated the challenges and benefits of growing various mixed stands have primarily focused
on stand productivity, nutrient cycling, over yielding potential, pest and fire mitigation, and
resistance to climate change (Brassard et al. 2008; Girardin and Terrier 2015; Laganière et al.
2015; Kweon and Comeau 2019). Many of the studies focusing on wood quality in mixedwoods
have focused on basal area increment or biomass productivity. However, recently within wood
attributes such as wood density, and modulus of elasticity (MOE), have become a subject of
interest in both aspen and spruce due to the increased economic appeal of aspen products and
softwood value-added forest products. The objective of this thesis is to examine the effect of
competition on the within wood attributes of both aspen and white spruce.

7

1.2 Mixedwood stand dynamics
Historically, aspen has been viewed as a weed, undesirable and inhibiting of preferred
conifer growth (Carleton and Maycock 1981). Aspen is frequently the primary tree species to
grow in post-disturbance landscapes, following destructive fire, beetle infestation or clear-cut
through asexual root suckering (Comeau et al. 2005; Frey et al. 2011; Smith et al. 2011). Aspen
is shade intolerant and displays fast initial growth rates, which allows it to quickly dominate the
forest canopy, compared to the seed-sowed and slow juvenile growth rates of spruce (Lieffers et
al. 1996). Aspen, however, has a short lifespan (50 – 60 years), and reaches the age of
senescence earlier than most conifers (250 – 300 years). Canopy openings from this senescence
allows for establishment or release of more shade tolerant, slow-growing conifer species,
including spruce (Brassard and Chen 2006). Competition is particularly critical for mixedwood
stand development in the stem exclusion or self-thinning phase (20-40 years of stand age) where
resources such as light, soil moisture and nutrients become scarce as the trees expand in size,
leading to logarithmic decreases in stem density (Lieffers et al. 2002; Chen and Popadiouk
2002). Although spruce is a shade tolerant species, studies have shown a strong response in
radial and height growth to increases in light, and poor conifer height growth under dense
hardwood canopies with light levels below 20% of full sunlight (Lieffers and Stadt 1994).

1.3 Combining abilities of mixed stands
Management objectives have been biased towards conifer growth as softwoods offer
superior properties for both pulp and lumber, however, recent studies have demonstrated the
ecological and economic benefits of growing white spruce in conjunction with aspen. As such,
the novel silviculture technique of planting spruce under developing or mature aspen has been

8

shown to effectively create boreal mixedwood establishment (Kabzems et al. 2016). Managing
for mixtures can vary with site fertility, but, can also positively impact the diversification of each
species functional traits, which, maximizes growing space (Pretzsch et al. 2013). Species
diversity can increase ecosystem function through habitat creation and through increased nutrient
cycling leading to productivity and carbon sequestration (Brassard et al. 2008; Laganière et al.
2015; Payne et al. 2019). Competition is the driving force of stand diversity in mixedwood
stands as it strongly influences successional trajectory (Filipescu and Comeau 2007). This
successional trajectory contributes to the landscape patchwork age-class distribution, which in
turn, mitigates fire and insect outbreak (Comeau et al. 2005; Girardin and Terrier 2015).
Since the early 2000’s, utilization of aspen increased and is now the main fibre source for
several oriented strand board (OSB) and veneer mills. Support for mixedwood management
options, including aspen utilization, has the potential to diversify forest silviculture and lower
costs of traditionally intensive silviculture practices while creating more resilient forests in B.C.
Also, the delayed maturation age of spruce relative to aspen allows for multiple entry partial
harvest, thereby allowing constant timber supply, while preserving biodiversity and habitat
continuation (Man and Lieffers 1999).
One of the distinct benefits of mixedwood forests is the potential for overyielding, or the
increase in productivity of a mixed species stand compared to a monoculture (Hector 1998).
Overyielding can occur when managing for space partitioning and size inequality between spruce
and aspen (Hector 1998; Kweon and Comeau 2019). The potential for overyielding in boreal
aspen spruce mixtures has been identified (Man and Lieffers 1999; Kabzems et al. 2007).
However, an increase in volume is only desirable when it corresponds with merchantable timber.
The characterization of wood attributes along with cost effective land use management is at the

9

forefront of the boreal forest industry’s fundamental shift from traditional wood products to
multiple value-added forest products (Chen et al. 2017).

1.4 Wood attribute overview
Mixed species stands have been found to differ from pure stands in terms of growth
(Comeau et al. 2004; Pretzsch et al. 2013), tree shape (Pretzsch 2019), resource use efficiency
(Carr et al. 2020), and resistance to disturbance (Hansen et al. 2016). Much less is known on the
effect of tree mixtures on individual tree wood attribute development, especially in the boreal
mixedwood. Mixedwood stands provide opportunities to apply variation in silviculture
techniques such as planting densities and thinning treatments to manage for desired combinations
of wood properties and yield (Comeau 2021). Wood attributes (Table 1.1) within both soft and
hardwoods have impacts on tree survivability and the grade of economic viability.

Table 1.1 A summary of each of the wood attributes that were analyzed in this study
Wood Attribute
Modulus of
Elasticity
(MOE)

Microfibril
Angle (MFA)

Description

A measure of stress compared
to strain within wood, and
often varies with MFA (Evans
and Ilic 2001)
The orientation of the
crystalline cellulose in the
secondary cell wall (S2) wood
along the fibre axis (Cave
1966).(Varies with age as
juvenile (core) wood has larger
MFA, mature wood has lower
MFA (Barnett and Bonham
2004).

Economic Value

Stiffness of wood increases
resistance to trunk breakage,
xylem explosion and increases
stem stability (Hacke et al.
2001)

Low MFA leads to higher
dimensional stability in lumber
and tensile strength in paper. As
MFA increases, stiffness and
compressive strength decrease
(Aziz 2013).

History

Previous studies have shown
that MOE is the best
representation of wood
stiffness (Antony et al.
2012a; Sattler et al. 2014)
Previous studies have shown
that MFA varies most with
age (De Araujo et al. 2015).

10

Wood Attribute

Density

Description

Economic Value

History

Wood density is defined as the
ratio of the oven-dry weight of
a sample to the weight of a
volume of water equal to the
volume of the sample at a
specified moisture content
(green, air-dry, or oven-dry).

Wood density influences
lumber strength (determining
lumber grade), pulp yield, pulp
quality, timber shrinkage,
stiffness, hardness, heating
value, machinability, and
energy requirement of the
pulping process (Jozsa and
Middleton 1994). Denser wood
is associated with lowered trunk
breakage, xylem implosion and
increased stem stability (Hacke
et al. 2001).
Fibre morphology and cell wall
structure directly influences
fibre flexibility, plasticity and
resistance to processing

Previous study has shown
that there was no effect of
mixed vs pure competition
having an effect on wood
density (De Araujo et al.
2015)

Related to ring width

Ring area

Specific surface
area

Ring width

Tangential
Diameter

Radial
Diameter

Height (rather than diameter)
preferred growth in
competitive environments has
resulted in a lower ring width
in aspen and spruce
(Coopersmith and Hall 1999)

Fibre perimeter divided by
coarseness (See methods
equation 3).

Tensile strength

Ring width represents the
interannual variability in
several factors both
endogenous and exogenous.
Variations in climate, site
conditions, competition, and
disturbance lead to differences
in observed ring widths (Fritts
1976)
Tracheid length parallel, or
tangent, to growth rings.

Lower ring width can indicate
increased density and impact
other wood attributes

Height (rather than diameter)
preferred growth in
competitive environments has
resulted in a lower ring width
in aspen and spruce
(Coopersmith and Hall 1999)

Tracheid length, diameter and
wall thickness not only affect
tensile and tear strength, but
also impact the optical and
printing properties of paper
(Macdonald and Hubert 2002).
Fibre morphology and cell wall
structure directly influences
fibre flexibility, plasticity and
resistance to processing

Longer fibres have been
reported for mixed white
spruce and aspen sites
compared to pure stands (De
Araujo et al. 2015).

Tracheid length that is
perpendicular to growth rings.

Longer fibres have been
reported for mixed white
spruce and aspen sites
compared to pure stands (De
Araujo et al. 2015).

11

Wood Attribute

Coarseness

Description

Economic Value

History

Defined as the weight per unit
length of the wood fibres
(methods equation 3)

Lower Coarseness means
higher tear strength, greater
bonding area, and more fibres
per tonne of pulp, which are
important to papermaking
(Watson and Bradley 2009)
Influences fibre collapsibility
during paper making

Higher coarseness has been
reported for mixed white
spruce and aspen sites
compared to pure stands (De
Araujo et al. 2015).

The thickness of xylem cell
walls.
Wall Thickness

Thickness in the cell wall
may change in response to
long term ecological factors
such as wind, exposure and
climate, thereby changing
MFA and resultant wood
properties (Hein and Lima
2012).

Within angiosperms (aspen) and gymnosperms (spruce), xylem cells are necessary for
structural integrity and for water transport within the tree. Gymnosperm tracheids are
characterized by many bordered pits in their radial walls, and are also much longer than
angiosperm fibres (Barnett and Bonham 2004). The orientation and magnitude of certain
characteristics within the tracheid reflects the influence of external forcing which can include
competition, or environmental conditions such as temperature changes, nutrient availability,
precipitation, increased wind stress or snowpack (Speer 2012).
Traditionally, density has been considered the most important industry wood trait as it
affects wood quality as well as the tear strength and burst of paper (Zobel and Jett 1995).
Studies have shown that ecologically, a denser wood is associated with lowered trunk breakage,
xylem implosion, and increased stem stability (Hacke et al. 2001).
However, it has been shown that the longitudinal modulus of elasticity (MOE) which is a
measure of stress compared to strain within wood, is in fact the best representation of wood
stiffness which is both desirable ecologically and for industry application (Antony et al. 2012b;
Sattler et al. 2014). Microfibril angle (MFA) is defined as the orientation of the crystalline
12

cellulose in the secondary cell wall (S2) wood along the fibre axis (Cave 1966). The MFA is
measured within the S2 layer of the secondary cell wall since the thickness of the S2 layer is
much greater than that of the primary S1 or S3 layers. Furthermore, MFA varies with age as
juvenile (core) wood has a larger MFA where-as mature (outer) wood displays a low MFA
(Barnett and Bonham 2004). As MFA increases in the wood, the stiffness and compressive
strength decrease (Aziz 2013). In general, low MFA leads to higher dimensional stability in
lumber and tensile strength in paper. Thickness in the cell wall may change in response to
exterior forces such as wind and thunderstorm, thereby changing the MFA and the resultant
wood properties (Hein and Lima 2012). Studies have found that Microfibril Angle (MFA) is the
best predictor of MOE, whereby MFA alone accounts for 86% of the variation observed in MOE,
and together with wood density describe 96% of the variation in MOE (Evans and Ilic 2001).
Fibre wall thickness in conjunction with fibre width influences fibre collapsibility during
paper making. Coarseness is defined as the weight per unit length of the wood fibres. Lower
coarseness means higher tensile sheet strength, greater bonding area, and more fibres per tonne
of pulp, all of which are highly prized by papermakers (Watson and Bradley 2009). Ring width
represents the interannual variability in several factors both endogenous and exogenous.
Variations in climate, site conditions, competition, and disturbance lead to differences in
observed ring widths (Fritts 1976). Lower ring width can indicate increased density and other
desirable attributes.

1.5 Competition in the mixedwood
Studies of how competition influences wood attributes within mixedwood systems have
been limited. Competition indices enable the quantitative analysis of relationships between stand

13

composition and wood attributes. Competition indices are numerical expressions that describe
how much each tree is affected by its neighbours. Two main types of competition indices are
often considered: distance-independent and distance-dependent. Distance-independent indices
are typically easier to calculate as they require less data, and have performed similarly to
distance-dependent indices in previous modelling applications (Kahriman et al. 2018). However,
much of the study of competition indices has been confined to models relating exterior tree
growth characteristics or mortality, and there is currently a lack of research concerning the
resolution required for finer wood attributes (Kahriman et al. 2018; Sun et al. 2019). Within
mixed forest plantations in Eastern Quebec, distance independent indices have successfully
quantified the competition effect on basal area increment growth in planted white spruce, while
distance-independent indices worked best for ingrown or non-planted species (BérubéDeschênes 2017). In an analysis of permanent sample plots that included pure and mixed stands
of aspen and white spruce of varying age in Central Alberta, wood density and carbohydrate
content was found to be consistent across sites, while MFA was lower within aspen sampled in
pure stands and fibre characteristics were higher in the pure site for both species (De Araujo et
al. 2015). Previous studies have been limited due to sampling uneven age stands with
inconsistent measures of treatment density. This study builds on previous work by analyzing the
effect of competition on wood attribute development in managed aspen and spruce stands with
homogenous site conditions.

1.6 Thesis Objectives
The purpose of this study was to characterize eleven wood attributes (Table 1.1) of both
white spruce and trembling aspen grown in various levels of competition within a mixedwood

14

trial site where tree ages and other ecological conditions were uniform. The specific objectives
were:
1. To determine if there was a difference in wood attribute response to stand level
competition between species.
2. To determine if aspen and white spruce display similar responses to stand level
competition over time.
3. To analyze if there are any economic benefits observed in wood attributes with
increasing aspen treatment density.
4. To compare spruce and aspen wood attribute response to inter- and intra-specific
competition.
5. To utilize plot data and formulate neighbourhood level competition indices. Then to
determine which neighbourhood level competition indices best improved a model for
wood attributes in both aspen and spruce.
6. To determine the competition resolution required for the wood attributes of both
spruce and aspen.
In addition to the introductory chapter, this thesis contains two chapters written as standalone manuscripts that address the above objectives, as well as a concluding chapter which
synthesizes the conclusions from each of the two manuscripts. The following chapter (Chapter 2)
explored objectives 1 through 3, by applying a mixed effect model approach to tree core data
collected to characterize eleven wood attributes across different competition regimes. Chapter 3
builds on the mixed effect model of Chapter 2, by utilizing plot data from the study site to
increase the competition resolution in order to explore objectives 4 to 6.

15

2. Evaluating the impact of stand composition and competition on
Spruce and Aspen wood attributes in mixedwood stands
2.1 ABSTRACT
In this study, we sampled 50 aspen (Populus tremuloides) and 70 white spruce (Picea
glauca) in northern British Columbia. We analyzed 11 wood attributes (microfibril angle,
density, modulus of elasticity, cell wall thickness, coarseness, ring width, cell population,
specific surface area, ring area, tangential diameter, and radial diameter) over five population
densities (0, 1000, 2000, 5000, and 10,000 stems per hectare (sph) aspen) to evaluate the impact
of competition on wood attributes. Through a mixed effects model framework, two thresholds
emerged from our analysis. First, the largest effect on spruce wood attribute development
occurred when aspen was grown in intimate mixture with spruce compared to spruce grown
without aspen. Second, was the variation observed in the 10,000 sph aspen treatment compared
to the treatments with lesser densities of aspen. Our results indicate that desirable wood attributes
are generally increased when spruce is grown with aspen competitors. Further, our results show
that desirable traits within aspen decrease with increasing levels of competition. Therefore,
maintaining aspen at lower densities in intimate mixture can increase spruce wood quality,
production and also provide the ecological benefits of aspen and their potential positive impacts
on forest resilience, while incurring some reductions in spruce volume.

16

2.2 INTRODUCTION
Mixedwood forests are some of the most productive and diverse forest ecosystems in
North American boreal forests (Chen and Popadiouk 2002), and support a range of ecosystem
services such as reducing the risk of insect/pathogen/fire outbreaks and improving resilience to
climate change (Bergeron et al. 2014). Mixedwood forests can also facilitate forest productivity
(Reyes-Hernández and Comeau 2015), and are a source of lumber and various value added forest
products (McCulloch and Kabzems 2009). Aspen (Populus tremuloides Michx.) and white
spruce (Picea glauca (Moench) Voss) are the primary components of the Western Canadian
boreal mixed forest and competition driven succession between these species is the primary
process of forest composition shift (Jiang et al. 2018). Previous studies have shown that various
levels of competition between species in a mixed stand can influence yield, biomass production,
and exterior growth characteristics (Comeau et al. 2005; Kabzems et al. 2007). These findings
have influenced forest management techniques, but knowledge gaps remain with regard to how
stand composition and neighbourhood competition impacts spruce and aspen wood attributes and
by extension economic value.
The forest industry in western Canada has largely focused on the utilization of conifers,
however, reductions in the predicted midterm timber supply, pine beetle infestation, projected
future climate change, and a shift towards the valuation of ecosystem services such as carbon
storage all portend to an increased use of broadleaf species (McCulloch and Kabzems 2009;
Mbogga et al. 2010; Dhar et al. 2016). Historically, aspen has been viewed as a weed,
undesirable and inhibiting of preferred conifer growth (Carleton and Maycock 1981). Past
management objectives were biased towards conifer growth as they offered superior properties
for both pulp and lumber production, based on industrial scale processes available at the time.

17

Although the majority of current industrial application relies on conifer growth, the benefits of
more holistic forest management plans which incorporate a diversity of tree species are
increasingly recognized (Macdonald et al. 2010). For instance, the silviculture technique of
planting spruce under developing or mature aspen has been shown to effectively establish boreal
mixedwood conditions (Kabzems et al. 2016). Overyielding, or the increase in productivity of a
mixed species stand compared to a monoculture, can occur when managing for space partitioning
and size inequality between spruce and aspen (Hector 1998; Kweon and Comeau 2019). Within a
mixed stand, an aspen density of 2000 – 5000 stems per hectare (sph) and white spruce at 1300
sph allows for overyielding with 21% more volume produced than monocultures of either
(Kabzems et al. 2007). Also, the delayed maturation age of spruce relative to aspen allows for
multiple-entry partial harvest, thereby enabling more consistent timber supply, while preserving
biodiversity and habitat continuation (Man and Lieffers 1999). As a result, there has been
increased interest in managing for intimate mixtures of aspen and spruce.
Developing a more thorough understanding of the complex interactions of the
mixedwood system, and specifically how competition between overstory trees may influence tree
growth and wood properties, is vital to the effective future management of these forests. Within
angiosperms (aspen) and gymnosperms (spruce), xylem cells are important for structural
integrity and for water transport within the tree. Gymnosperm tracheids are characterized by
many bordered pits in their radial walls, and are also, often, longer than angiosperm fibres
(Barnett and Bonham 2004). The orientation and magnitude of certain characteristics within the
tracheid reflects the influence of external forcing which can include competition, or
environmental conditions such as temperature changes, nutrient availability, precipitation,
increased wind stress or snowpack (Speer 2012). Wood density, modulus of elasticity (MOE),

18

microfibril angle (MFA), and other wood attributes (Table 1.1) are potentially responsive to
changes in a trees neighbourhood composition, stand structure, and the type and intensity of
competition that a tree experiences (Pretzsch and Rais 2016).

The primary objective of this study is to characterize the wood attributes of both white
spruce and trembling aspen grown within an established mixedwood research experiment with
variable aspen densities where tree ages and ecological conditions are uniform. The first subobjective is to determine if there is a difference observed in wood attribute variation between
species. The second sub-objective is to determine if there is a difference observed in wood
attribute response to competition between species. The third and final sub-objective is to
determine if spruce and aspen wood attributes display similar responses to competition, and
whether those differences can be translated into discernable economic benefits. To achieve these
sub-objectives, three hypotheses were formulated. The first hypothesis is that spruce and aspen
display differences in their wood attributes which will be tested through the evaluation of a linear
mixed effect model. Spruce has been shown to display physiological differences as well as
variation in basal area increment growth in response to competition, so it is likely that spruce
will also display differences in other wood attributes. Aspen is shade intolerant, and therefore
will likely have a more pronounced reaction to competition than spruce. Therefore, the second
hypothesis is that there is a difference in wood attribute response to competition between species.
Finally, the third hypothesis is that spruce and aspen will display different responses to
competition that will translate to benefits (or detriments) in wood quality. Evaluation of this third
hypothesis will be critical to understanding the role that competition plays on wood quality in the
boreal mixedwood.

19

2.3 MATERIALS AND METHODS
2.3.1 Study Site
The study site is located 45 km northeast of Fort St. John, British Columbia (Figure 2.1)
in the moist warm subzone of the Boreal White and Black Spruce biogeoclimatic zone
(BWBSmw, DeLong et al. 2011).

Figure 2.1 A map of Siphon Creek trial site with the treatment map (Richard Kabzems)

The site is characterized by a mesic to sub hygric moisture regime and rich nutrient
regime (Kabzems et al. 2007). The soils underlying the study area are fine-textured Gray
Luvisols, moderately well drained, with a thick, gray Ae horizon over a more fine-textured Bt
horizon (Lord and Green 1986).

20

The area was selectively logged for conifers in 1968. The remaining aspen trees and
existing vegetation were sheared and windrowed in the early winter of 1984/85. In 1985, the site
was planted with three-year-old bareroot white spruce seedlings at 1480 stems per hectare, which
were sampled for this study at 33 years of age. The study site was established in 1989 as part of
Ministry of Forests Experimental Project 1077, and plots were laid out in 1990. Twenty plots
with varying densities of aspen (0, 5000, and 10,000 stems per hectare) combined with a second
planting of white spruce (0, 500, 700, 900, 1100, 1300 and 1500 stems per hectare) were laid out.
In 1990, the aspen regeneration was manually thinned to target densities of 0, 5000, and 10,000
stems per hectare and all other tree and tall shrub species manually removed, similar to a juvenile
spacing treatment. Repeated manual brushings were applied until 2000 to maintain the 0-aspen
treatment. In the summer of 2000, two additional aspen treatments (1000 and 2000 stems per
hectare) were created by reducing existing densities in previously established plots (Kabzems et
al. 2015).

2.3.2 Data Collection
Samples were obtained in 2018 from trees within the 0, 1000, 2000, 5000, and 10,000 sph
aspen treatment units (Table 2.1). A total of five undamaged spruce trees (i.e., those with no
mechanical damage, forked tops, or other obvious growth issues), were sampled within each
treatment. All sampled spruce trees within a treatment replicate were separated by a minimum
10-meter distance to ensure distribution over the plot. For each sampled spruce tree, one aspen
was deemed to be ‘paired’ with the spruce and sampled. The aspen ‘pair’ was the nearest
healthy, canopy dominant aspen stem within 1-2 meters of the sampled spruce tree.

21

Table 2.1Descriptive statistics of sampled trees. Diameter at breast height (DBH) means are shown with standard
deviations in parentheses.

Treatment
(aspen
sph)
0
1000
2000
s5000
10000

Spruce DBH
(cm)
17.3 (2.3)
16.2 (3.1)
15.0 (5.2)
12.1 (1.5)
11.1 (2.1)

Aspen
DBH (cm)
NA
15.7 (4.2)
14.4 (3.9)
13.7 (2.8)
11.6 (1.8)

Number of
spruce samples
20
5
5
20
20

Number of
aspen samples
0
5
5
20
20

Using an increment borer, a 12 mm core was collected at breast height (1.3m), from each
tree selected. Core position was chosen to minimize effects of slope on the ring patterns and to
facilitate optimal cross dating. The cores were placed in paper tubes for protection and allowed
to air dry prior to preparation. The condition and growth form of each sampled tree was noted
along with any indications of historic abiotic or biotic damage. Height and diameter at breast
height (DBH) were measured for the selected tree and for all neighborhood trees within a 3.99meter radius of the targeted spruce and aspen stems. Competitive trees were deemed to be those
greater than 1.3 m in height. The species and the distance of the competitive tree to the target tree
were also recorded.
2.3.3 Sample Preparation
Tree core samples were air-dried and prepared for analysis in the University of Northern
BC (UNBC) Dendrochronology Laboratory. An important factor controlling density in wood is
resin. Since resin is unevenly distributed within the wood and it has different properties than the
wood itself, it must be removed (Speer 2012). Samples were soaked in acetone for 8 hours,
before being passed through a Soxhlet apparatus for a further 8 hours (Rydval et al. 2014).
Within the Soxhlet apparatus, the acetone boils, and the steam will rise until it hits a condenser,

22

then drips into the core chamber. When the acetone reaches a maximum height, the chamber is
drained and the process is repeated, thereby extracting the resins. Once the samples were
removed of all resins, they were sent to FPInnovations for SilviScan® analysis.
FPInnovations in Vancouver, Canada, further prepared the 12 mm cores collected.
Cores were air-dried and then cut into strips of 2 x 7 mm (tangentially x longitudinally) using a
twin-blade saw. Resulting laths were scanned using SilviScan® technology. Each strip was
scanned for radial and tangential cell dimensions using optical microscopy, wood density using
x-ray densitometry, and microfibril angle (MFA) using x-ray diffractometry. The analysis
performed measured wood properties every 25 µm across the wood lath, with the exception of
microfibril angle, which was measured every 5 mm. Properties measured included ring density,
radial and tangential diameters, and microfibril angle. Coarseness (Cr), cell wall thickness
(CWT), and Specific Surface Area (SSA) values were calculated as a function of other measured
variables through the following equations:
[1]

Cr = R x T x Dw

[2]

CWT =

[3]

SSA = P/Cr

Where:
P = 2 (R + T)
And, R and T are the radial and tangential tracheid diameters, respectively, P is fibre perimeter,
and Dw is the cell wall density (Tong 2019).
2.3.4 Data analysis
The wood attributes analysed included: wood density (kg/m3), radial diameter (µm),
tangential diameter (µm), coarseness (µg/m), cell population (#/mm2), microfibril angle (MFA,
degree), modulus of elasticity (MOE, Gpa), wall thickness (µm), specific surface area (m2/kg),
ring width (mm), and ring area (mm2). For each of the wood properties analysed, a mean value
23

was calculated for each ring in order to determine if significant differences occurred between
trees from varying treatments. Intra-annual differences were observed, but exceed the scope of
this paper, as they did not exhibit clear increasing or decreasing trend over one ring width. The
significance level selected for all levels of the statistical analysis was α = 0.05 (95% confidence
level). Residuals were checked visually with quantile-quantile plots using the ggpubr package
(Kassambara 2020) and were found to not differ from normal at every stage of the analysis. A
comparison of linear and 3rd, 4th, and 5th order polynomials, as well as exponential equations was
used in order to observe the change in a given wood property over time for each species in each
treatment. Akaike’s information criterion (AIC value) was used to evaluate the goodness of fit of
the various line types. All wood attribute variation approximated linear change at approximately
10 years.

A linear mixed model (equation 4), with tree as the random factor, and ring (cambial
age), species, and treatment as the fixed factors, was utilized to investigate differences in wood
attributes. The linear mixed model was created using the lme function within the nlme package
(Pinheiro et al. 2020) in R (version 3.4.3)(R Development Core Team 2017). The dredge
function within the MuMIn package (Barton 2020) for model selection was used to rank the best
model (based on variable inclusion of the model terms) through Akaike’s Information Criterion
(AIC) value. The global model included both aspen and white spruce samples, in order to
determine that there was a difference in wood attribute variation between species. Next, in order
to determine if the observed wood attributes varied within species and between treatments, a
linear mixed model separated by sample species was run (lme: ring, and treatment with tree as
the random effect). An analysis of variance (ANOVA) was then used to determine significant

24

effects of each species separated model. Post-hoc analysis of significant treatment effects were
conducted by the pairwise comparison of contrasts from the lsmeans (Lenth 2016) package with
a Bonferroni adjustment applied to p-values. Although p-values can be the subject of controversy
when used as an indicator of significance in mixed effect models, and in scientific hypothesis
testing in general, the application of the conservative Bonferroni adjustment can mitigate the
potential of Type 1 error. Also, in this instance, the results of the multi model inference agreed
with the same analysis comparing significant model terms based on p-values (Appendix Table
6.2). So, beyond the initial model selection test, p-values were used to represent significance.
The goal of this portion of the study was not to create a predictive model, nor maximize R2 of
each model, but rather answer ecological questions through these statistics.
[4]

Yj=β0+β1xj+β2xj+μ1 +εj,

Where Yj and xj represent fibre property for year j; β0, β1 and β2 are the fixed effects.
μi1 is the random effect of tree and ui2 is the random slope; εij is the error term.
Finally, an “improvement” heatmap was created by normalizing the aspen and spruce
means for each property in each treatment to the mean within the 1000 sph aspen treatment. An
improvement was termed to be either an increase or decrease in the value of each wood property
as per Canadian wood fibre standards (Canadian Wood Fibre Centre et al. 2010). An important
caveat is that these criteria for improvement could change depending on the properties required
for the desired end product. For this study, improvements were defined as the treatments that
displayed lesser values for coarseness, ring width, and MFA relative to the mean. The remaining
wood attributes (density, MOE, ring area, specific surface area, cell population, tangential and
radial diameter) were considered to have positive improvement if they increased between

25

treatments relative to the mean. The years 10-34 were used in this comparison, to omit the
variation observed in early age wood.
2.4 RESULTS
2.4.1 Linear mixed effect models
Table 2.2 shows the results of the multi model inference, where the global model that was
tested included the terms: ring (cambial age), species, and treatment. This method chose the ‘best
model’ (listed first for each wood attribute) based on lowest AIC score, which minimizes
information loss of the model. Displayed in Table 2.2 is the ‘best model’ for each wood attribute
and the next highest-ranking model until the change (delta) in AIC > 2, indicating a significant
difference between models. The results of this analysis show that for all wood attributes, there is
an important difference observed between species, as well as a difference observed between
treatments.
Table 2.2 Summarized results of the multi model inference analysis of the global model (linear model = Ring +
species + treatment + |tree).Loglik is the log likelihood, AIC is Akaike’s information criterion, and delta is the
difference in AIC for best performing

Wood
(Intercept)
attribute
Ring
3.77
Width
3.75

Density

MFA

Ring species treatment df

logLik

AICc

delta weight

-0.0581 NA

+

8

-4169.12

8354.29

0.00

0.73

-0.0581 +

+

9

-4169.09

8356.24

1.95

0.27

2.83

-0.0580 +

NA

5

-4204.51

8419.04

64.75

0.00

393.11

NA +

+

8

0.00

0.45

396.28

NA +

NA

1.00

0.27

392.80

0.0205 +

+

9

31625.60
15803.77

1.94

0.17

395.99

0.0194 +

NA

5

31626.60
15808.29

2.95

0.10

23.15

-0.6538 +

+

9

-8835.62 17689.30

0.00

0.90

31623.65
15803.80
4
31624.66
15808.32

26

Wood
attribute

MOE

Coarseness
Tangential
Diameter
Radial
Diameter

Specific
Surface
Area

Ring Area

Cell
Population

Wall
Thickness

(Intercept)

Ring species treatment df

logLik

AICc

delta weight

20.77

-0.6539 +

NA

5

-8841.83 17693.68

4.38

0.10

8.14

0.2956 +

+

9

-5388.85 10795.75

0.00

0.93

9.45

0.2956 +

NA

5

-5395.53 10801.08

5.33

0.07

542.70

2.3385 +

+

9

0.00

0.99

506.81

2.3359 +

NA

10.05

0.01

40.53

0.0520 +

+

9

-7820.46 15658.97

0.00

0.99

39.29

0.0519 +

NA

5

-7829.07 15668.17

9.19

0.01

28.59

0.1495 +

+

9

-7283.05 14584.16

0.00

1.00

27.27

0.1503 NA

+

8

-7290.68 14597.41

13.25

0.00

263.46

-1.0999 +

+

9

31818.65
15900.29

0.00

0.99

283.86

-1.0987 +

NA

5

31829.21
15909.59

10.56

0.01

491.55 17.8545 NA

+

8

0.00

0.65

460.56 17.8685 +

+

1.25

0.35

98.51 17.8805 +

NA

55.32

0.00

0.00

1.00

25.15

0.00

+
10.2019
1068.51
+
10.2034
905.31

+
NA

36465.47
18223.71
5
36475.52
18232.75

43048.72
21516.34
9
43049.97
21515.96
5
43104.04
21547.01
41263.24
20622.59
5
41288.38
20639.18
9

2.82

0.0083 +

+

9

-907.37

1832.81

0.00

0.97

2.72

0.0083 +

NA

5

-914.87

1839.77

6.96

0.03

27

Next, in order to determine if there were differences in wood attribute variation between
treatments and within species, the same model structure was applied to aspen and spruce samples
separately. There was a significant effect of treatment in all wood attributes except for MFA
(both aspen and spruce), and MOE (aspen) (Table 2.3).
Table 2.3. Results of species separated ANOVA of model terms for each species (Full results for each wood attribute
in Chapter 2 Supplementary Appendix).

Wood
Property
Ring
Width

Aspen Samples

Cambial Age (Ring)
F(1, 1421) = 367.21,
p < 0.001
F(1, 1421) = 113.28,
Density
p < 0.001
Microfibril F(1, 1421) =
Angle
3393.59, p <0.001
Modulus
of
F(1, 1421) =
Elasticity
3930.35, p < 0.001
F(1, 1421) = 39.4, p
Coarseness < 0.001
Tangential F(1, 1421) = 13.71, p
Diameter
< 0.001
Radial
F(1, 1421) = 6.39, p
Diameter
= 0.012
Specific
F(1, 1421) = 109.60,
Surface
p < 0.001
F(1, 1421) = 489.03,
Ring Area p < 0.001
Cell
F(1, 1421) = 17.14, p
Population < 0.001
Wall
F(1, 1421) = 74.01, p
Thickness < 0.001

Spruce Samples

treatment
F(3, 46) = 3.30,
p = 0.0284
F(3, 46) = 3.86,
p = 0.0151
F(3, 46) = 1.04,
p = 0.3822

Cambial Age (Ring)
F(1, 1621) = 488.01,
p < 0.001
F(1, 1621) = 90.97, p
< 0.001
F(1, 1621) =
7768.49, p < 0.001

treatment
F(4, 65) =
24.43, p < 0.001
F(4, 65) = 2.56,
p = 0.046
F(4, 65) = 2.22,
p = 0.076

F(3, 46) = 1.17,
p = 0.3299
F(3, 46) = 6.19,
p = 0.0013
F(3, 46) = 5.67,
p = 0.0022
F(3, 46) = 3.82,
p = 0.016
F(3, 46) = 6.55,
p < 0.001
F(3, 46) = 2.96,
p = 0.042
F(3, 46) = 4.45,
p = 0.0079
F(3,46) = 5.72,
p = 0.0021

F(1, 1621) =
10809.77, p <0.001
F(1, 1621) = 339.61,
p < 0.001
F(1, 1621) = 68.08, p
< 0.001
F(1, 1621) =
3411.57, p < 0.001
F(1, 1621) = 90.64, p
< 0.001
F(1, 1621) =
1110.89, p < 0.001
F(1, 1621) =
1682.25, p < 0.001
F(1, 1621) = 59.91, p
< 0.001

F(4, 65) = 3.42,
p = 0.0135
F(4, 65) =
10.44, p < 0.001
F(4,65) = 9.25,
p < 0.001
F(4,65) = 9.06,
p < 0.001
F(4, 65) = 5.55,
p < 0.001
F(4, 65) =
19.98, p < 0.001
F(4, 65) =
11.64, p < 0.001
F(4, 65) = 4.38,
p = 0.0034

Significant effects were explored through pairwise comparison (summarized in table 2.4,
complete post hoc results in Appendix Tables 6.3 and 6.4). The contrasts that displayed
significant differences between treatments are shown in Table 2.4. From this analysis, two
thresholds became evident. First, for aspen samples, all of the significant treatment effects were

28

driven by differences in the highest aspen density (10,000 sph) contrasted with lower aspen
densities. The second threshold was that for spruce, most of the significant treatment effects were
driven by contrasts between the treatment without aspen (0sph) and treatments where spruce was
grown in mixture with aspen (i.e., > 0sph aspen). This analysis confirmed the hypothesis that
aspen and spruce wood attributes would vary with competition. Also, boxplots show the
variation for each wood attribute separated by species and treatment (Figure 2.3, Figures for each
wood attribute found at the start of each wood attribute analysis chapter in Chapter 2
Supplementary Appendix).
Table 2.4. The results of species separated two-way analysis of variance (ANOVA) of the effects of treatment (aspen population
density) and cambial age on each wood attribute. For significant treatment effects, pairwise post hoc testing (Bonferroni
adjusted) determined contrasts between treatments. For full contrast table, consult appendix (Tables 6.3 and 6.4).

Figure Wood
Property
Ring Width
Density
Microfibril Angle
Modulus of
Elasticity
Coarseness
Tangential
Diameter
Radial Diameter
Specific Surface
Ring Area
Cell Population
Wall Thickness

Species

Treatment p-value

Aspen
Spruce
Aspen
Spruce
Aspen

0.0284
p < 0.001
0.0151
0.0466
0.3822

None
0-2000, 1000-5000, 1000-10000
2000-10000
None
N/A

Aspen
Spruce
Aspen
Spruce
Aspen
Spruce
Aspen
Spruce
Aspen
Spruce
Aspen

0.3299
0.0135
0.0013
p < 0.001
0.0022
p < 0.001
0.016
p < 0.001
p < 0.001
p < 0.001
0.0418

N/A
0-5000, 0-10000
2000-10000, 5000-10000
0-2000, 0-5000, 0-10000
1000-10000, 2000-10000
0-5000, 0-10000
1000-10000
0-5000, 0-10000
2000-10000, 5000-10000
0-2000, 0-10000
None

Aspen
Spruce
Aspen
Spruce

0.0079
p < 0.001
0.0021
0.0034

Spruce

Spruce

0.076

p < 0.001

Post Hoc: Significant Contrasts

N/a

0-2000, 1000-5000, 1000-10000
1000-10000
0-5000, 0-10000
2000-10000, 5000-10000
0k-2k, 0k-10k, 2k-5k, 5k-10k

29

Figure 2.2 Boxplot showing the microfibril angle variation between aspen and spruce in each treatment. This is
figure is shown as an example of the variation between species, for figures of all wood attributes consult Chapter 2
Supplementary Appendix.

2.4.2 Wood attribute variation over time
In order to show wood attribute change over time, we tested the fits of several line types
for each treatment and the results indicate that all spruce attributes exhibit non-linear changes as
the trees develop, with the largest changes in properties being distinguished by differences in
establishing tree (cambial age 0-5 years), and older trees (cambial age 5-30 years) (Figures 2.3,
2.4, 2.5). While developmental changes in most wood attributes were best statistically described

30

by 4th or 5 th order polynomial, most approximated linear changes in the attributes between 5 and
30 years.

Figure 2.3 Density change over time with separated aspen (top) spruce (bottom) and treatment.

31

Figure 2.4 Microfibril angle change over time with separated (a, left) aspen, and (b, right) spruce and treatments.

32

Figure 2.5 Modulus of Elasticity change over time with separated (a, left) aspen, and (b, right) spruce and
treatments.

33

2.4.3 Heatmap
Figures 2.6 and 2.7 show the results of an “improvement” heatmap where a positive
improvement meant a reduction in ring width, coarseness, and microfibril angle, and a
maximization of all other wood attributes. In general, aspen wood quality (as previously defined)
was diminished with increasing levels of treatment and spruce wood quality was improved. The
only common improvement was ring width which was minimized with increasing aspen
treatment density. Aspen wall thickness was improved in the 10,000 sph treatment and displayed
a negative effect in other treatments. Radial diameter and tangential diameter improved with
increasing aspen density. Coarseness was minimized with increasing aspen treatment density.
MOE was most desirable in 10k treatment, while minimized in the other densities of aspen. MFA
showed positive improvement in all densities relative to 1000 sph treatment. Wood density was
negatively associated with increasing aspen population density. Ring width improved with aspen
population density. Within spruce, most desirable wood traits were maximized in spruce with
increasing aspen density. Tangential and radial diameter both did not display improvements as
they decreased with increasing treatment density. Modulus of elasticity and density were
maximized in the 5000 sph treatment. MFA, coarseness and ring width all improved with
increasing aspen population density.

34

Figure 2.6 A relative improvement heatmap for all 11 wood attributes within spruce and their response to increasing levels of aspen
treatment density (relative to control treatment).

35

Figure 2.7 A relative improvement heatmap for all 11 wood attributes within aspen and their response to increasing
levels of aspen treatment density (relative to 1000 sph aspen treatment).

36

2.5 DISCUSSION
Significant differences between spruce and aspen were observed for all wood attributes with
the exception of ring width and ring area. A lack of difference between species for ring width
and ring area is consistent with previous studies that have observed that for both species tree
height to diameter ratio in mixed stands increases with competition (Lanner 1985; Coopersmith
and Hall 1999). Both spruce and aspen exhibit increased slenderness with competition, which
corresponds with reduced ring width and ring area. For all other wood traits, the distinct
differences between species reflects the intrinsic anatomical differences of each species (Pretzsch
and Rais 2016) as well as variations in their growth preferences such as shade tolerance (Man
and Lieffers 1999). For both spruce and aspen, we found that aspen treatment density influenced
most wood properties (Table 2.3), but that the sampled species responded differently to increases
in aspen treatment density. Microfibril angle (MFA) in spruce and aspen, as well as modulus of
elasticity in aspen, was not significantly impacted by aspen density. Our finding that MFA is
insensitive to competition is consistent with previous studies from boreal mixedwood stands in
Western Canada (De Araujo et al. 2015). The age of wood, and whether or not it is juvenile or
mature is the greatest determinant of MFA, with large MFA’s in juvenile, and smaller in mature
wood (Barnett and Bonham 2004). It has been shown that the normal decrease in MFA from pith
to bark can be interrupted by surge in growth rate such as that which may occur following the
thinning and removal of competitor trees (Herman et al. 1999). We did observe a distinct
increase in spruce MFA (Figure 2.5b) following the thinning that occurred in the sample plots in
1990, 8 years into the spruce growth (3-year-old planted spruce in 1985). We also observed high
variation in the MFA in spruce for all samples (Figures 2.2 and 2.4) which could partially
account for a significant treatment effect not being found. Further, MOE in aspen (Figure 2.5),

37

displays greater variation than the spruce (Figure 2.5), and could explain some of the lack of
significance between treatments. Also, since MFA did not exhibit a difference between
treatments, and MFA can explain up to 86% of the variation in MOE it could be expected that
MOE will follow a similar trend. However, within spruce cores, MOE did in fact display a
significant effect of treatment. This could be explained by a dominant effect of wood density on
MOE as it varied between treatments. Approximately ~10% of the variation in MOE can be
explained by variation observed within wood density (Evans and Ilic 2001).
For most of the wood attributes that we measured (ring width, MOE, coarseness, tangential
diameter, radial diameter, specific surface area, ring area, cell population, and wall thickness),
two thresholds consistently emerged from our analysis. First, the largest effect on spruce growth
occurred when aspen was grown in densities greater than zero with the spruce. Ninety percent of
the significant differences within spruce wood properties, and between treatments, were
attributed to the presence of aspen in the stand. Aspen growing in mixture with spruce will
impact the availability of key limiting resources with regard to light, water and nutrients
(Filipescu and Comeau 2007; Kabzems et al. 2015). Past studies have demonstrated that
modifications to spruce’s growing environment influence absolute growth as well as gross
growth characteristics such as height-to-diameter ratio (Kabzems et al. 2015). Our findings build
on this previous work by demonstrating that spruce wood attributes are different for spruce
growing in pure stands (at lower density) compared to spruce grown in mixture.
The second threshold that emerged from our results was the impact of an aspen treatment
density of 10,000 sph, on aspen wood characteristics. Differences between the 10,000 sph
treatment and those with lower densities of aspen accounted for all of the significance observed

38

in aspen wood attributes (Table 2.4). Growth of trembling aspen has been shown to be
significantly affected by inter- and intraspecific competition (Jiang et al. 2018).
An outlier from these two thresholds is wood density. Both aspen and spruce displayed a
significant effect of treatment, but upon post-hoc testing, only aspen wood density exhibited
significant differences between 2000 – 10,000 sph (Table 2.4) treatments. The significant effect
of the overall model term without significant post-hoc treatment contrasts, could be attributed to
the low sample sizes, and an overly conservative result from the Bonferroni correction. This
could further be explored through additional sampling. The finding that wood attributes of both
aspen and spruce are affected by treatment compares to previous studies involving wood density
and mixtures of white spruce and aspen (De Araujo et al. 2015), where no significant differences
between mixed and pure stand wood density were found in either species. A possible
explanation for the overall treatment significance observed in aspen and spruce cores in this
study is that the Siphon Creek research plots provide a higher resolution as they were sampled
from plots with the same ecological conditions, consist of controlled densities, and have the same
age across plots. Since all other environmental variables are held constant, the differences in
density were more readily apparent from the data collected at the research site than the
geographically separate permanent sample plots in previous studies.
Furthermore, our work indicates that all spruce wood attributes exhibit non-linear
changes as the trees develop, with the largest changes in properties being distinguished by the
differences in establishing tree (age 0-5 years), compared to older trees (5-30 years). While
developmental changes in most wood attributes were statistically best described by a third or
fourth order polynomial, most approximated linear changes in the attributes between 5 and 30

39

years. This variation is consistent with previous observations of change in wood attribute
development as trees age (Amarasekara 2002).
In this study we aspired to evaluate if there were discernable economic improvements in
wood attribute value with increasing levels of competition. Our experimental framework does
not allow us to explicitly tease apart the relative impact of increased interspecific competition
compared to overall competition, on spruce wood attributes. However, our results indicate that
desirable spruce wood attributes are generally increased when spruce is grown with aspen
competitors (Figure 2.6). In general, spruce grown in intimate mixture with 5000 sph aspen
density seemed to offer relative improvements in wood properties that are considered the most
important for lumber products. Increased wood density, larger MOE, lower MFA, larger ring
width, and increased wall thickness are the benefits observed in spruce grown in the 5000 sph
treatment which constitute either best or second-best improvements to wood quality, relative to
other treatments. Aspen densities of 5k and 10k are due to juvenile spacing of five-year-old
aspen, while the 1k and 2k aspen have had two juvenile spacing treatments at 5 and 15 years.
Thus, the aspen represents dominant and codominant individuals at the time of spacing
treatment.
Density may be less important for lumber products in the future due to the increased
production of value added engineered wood products, as they are often made stronger and stiffer
by laminating a low-density wood to a higher-density wood (Canadian Wood Fibre Centre et al.
2010). However, density is still important for bioenergy, which values the higher energy content
of dense wood.
MOE has been shown to be the best predictor of wood stiffness, which is desirable both
for tree survivability, and lumber considerations when maximized (as evident in the increase

40

between treatments in Figures 2.5, and the resultant heat map improvement in Figures 2.6 and
2.7).
Further, depending on the intended use, radial and tangential diameter can either be
maximized or minimized. In this case length maximized was viewed as desirable in the
improvement heatmap value as longer fibre results in stronger paper sheets, which resulted in
negative improvement on the heatmap. However, thinner fibre tends to form better quality sheets
of higher tensile strength and greater bonding area (Watson and Bradley 2009), so again, it
depends on intended use. Increased water availability has been linked to formation of wider
tracheids particularly in the radial direction (Vysotskaya and Vaganov 1989; Wilpert 1991).
Since the treatments without aspen have a lower density of trees in general, there is likely to be
more water available, allowing for the discrepancy in radial diameter observed between
treatments (Kerhoulas et al. 2013).
In contrast to spruce wood attributes, our results indicate that most desirable traits (MOE,
density, and MFA) within aspen decrease with increasing levels of aspen competition (Figure
2.7). Lower values for MOE and density increase the risk of trunk breakage, xylem implosion
and decreased stem stability (Hacke et al. 2001). Effectively, in this instance, the increased
competition is weakening the aspen, and accelerating the stand dynamics typical of mixedwood
forests (Bergeron et al. 2014) . However, the majority of the negative associations for the
remaining wood attributes are related to the 10,000 sph aspen treatment, suggesting that, aspen
wood attributes are not significantly affected until densities are increased past a certain threshold.
Previous studies have shown that increases in maximum stand stem density can lead to an
increase in productivity in mixed species forest stands when space partitioning allows for the
favourable expression of the functional traits of each species (Reyes-Hernández and Comeau

41

2015). For both aspen and spruce samples, we found that the increased stockability of the
managed 1000-5000 sph treatments, had very few negative effects on wood attribute values.
Therefore, it is a conclusion of this study that some of the benefits of maintaining aspen in a
stand are realized even at relatively low aspen densities (between 1000 and 5000 sph). This
suggests that maintaining aspen at lower densities in intimate mixture can increase spruce wood
quality and maintain the ecological benefits of aspen and their potential positive impacts of forest
resilience (Macdonald et al. 2010). Changes in measures of productivity for the spruce and aspen
components of a mixture will vary with management choices and desired outcomes (Comeau
2021). Increased competition from aspen may have a negative impact on spruce volume growth
(Cortini et al. 2012), even though intimate mixture stands often exhibit similar total volume
production (spruce and aspen volume growth combined) or potentially increased total yield
(Kweon and Comeau 2019). Forest managers focused on conifer production must therefore
consider the beneficial impacts of a managed aspen component on spruce wood attributes and
increase in overall volume in the context of the concomitant reductions in spruce volume
production.

42

3. A mixed effect approach to evaluating the impacts of stand
composition, age and neighbourhood competition on the wood
attributes of spruce and aspen grown in mixedwood stands
3.1 ABSTRACT
The influence of stand composition, neighbourhood competition, and stand level
competition, on the development of wood attributes of aspen (Populus tremuloides Michx.) and
white spruce (Picea glauca (Moench) Voss) were examined. Data was collected from a
silvicultural trial site in British Columbia, Canada. Eleven wood attributes (wood density, radial
diameter, tangential diameter, coarseness, cell population, microfibril angle (MFA), modulus of
elasticity (MOE), cell wall thickness, specific surface area, ring width, and ring area) were
analysed from 120 sample trees (70 spruce and 50 aspen). Four distance-independent and four
distance-dependent competition indices were tested in order to determine the dominant effect,
and competition resolution required for each wood attribute in each species. Aspen
neighbourhood level (within 3.99m plot radius) competition was best described by a distancedependent index, while spruce was best described by a distance-independent competition index.
Neighbourhood level competition resolution is necessary for determining the wood attributes of
ring area, ring width, radial diameter, cell population, and coarseness in aspen, but relatively
unimportant in spruce. For spruce samples, stand level competition (treatment density) had a
dominant effect on wood density, ring area, ring width, tangential diameter, radial diameter and
cell population. Age was the dominant effect for MFA and MOE variation in both spruce and
aspen. Our models explained up to 75% and 91% of the variation observed in aspen and spruce
wood attributes respectively.

43

3.2 INTRODUCTION
Aspen (Populus tremuloides Michx.) and white spruce (Picea glauca (Moench) Voss) are
the dominant tree species found in the Western Canadian boreal mixed forest and competition
driven succession between these species is the primary process of forest composition shift (Jiang
et al. 2018). Studies have shown that mixedwood competition at various levels can influence
yield, biomass production, and exterior growth characteristics (Comeau et al. 2005; Kabzems et
al. 2007). These findings have influenced forest management techniques, but knowledge gaps
remain with regard to how stand composition and neighbourhood competition impacts spruce
and aspen wood attributes, and by extension, economic value.
Both inter and intra-specific competition play important roles in affecting forest growth,
composition, structure and succession in the boreal mixedwood. Aspen is frequently the primary
tree species to grow in post-disturbance landscapes, following destructive fire, beetle infestation
or clear-cut through asexual root suckering (Comeau et al. 2005; Frey et al. 2011; Smith et al.
2011). Aspen is shade intolerant and displays fast initial growth rates, which allow it to quickly
dominate the forest canopy, compared to the seed-sowed and slow juvenile growth rates of
spruce (Lieffers et al. 1996). Aspen, however, has a short lifespan (50–60 years), and reaches the
age of senescence earlier than most conifers (250-350 years). Canopy openings from this
senescence allow for establishment or release of more shade tolerant, slow-growing conifer
species, including spruce (Brassard and Chen 2006). Competition is particularly critical for
mixedwood stand development in the stem exclusion or self-thinning phase (20-40 years of stand
age) where resources such as light, soil moisture and nutrients become scarce as the trees expand
in size, leading to logarithmic decreases in stand density as trees compete for these resources
(Lieffers et al. 2002; Chen and Popadiouk 2002). Although spruce is a shade tolerant species,

44

studies have shown a strong response in radial and height growth to increases in light, and poor
conifer height growth under dense hardwood canopies with light levels below 20% of full
sunlight (Lieffers and Stadt 1994). Intra-specific competition has been shown to have a stronger
effect on oak and beech growth than inter-specific competition, which shows the importance of
the diversification of each species functional traits in order to maximize growing space (Pretzsch
et al. 2013). Since competition can have an effect on radial and height growth, it is hypothesized
that there will be an effect on the wood attributes of both spruce and aspen as well.
Competition indices enable the quantitative analysis of relationships between stand
composition and wood attributes. Competition indices are numerical expressions that describe
how much each tree is affected by its neighbours, and there are two main types: distanceindependent (Wykoff et al. 1982) and distance-dependent (Martin and Ek 1984). Distanceindependent indices are typically easier to calculate as they require less data, and have performed
similarly compared to distance-dependent indices when the performance of each competition
index and its contribution to a growth model were assessed by the mean square error reduction
(Kahriman et al. 2018). However, much of the study of competition indices has been confined to
models relating exterior tree growth characteristics or mortality, and there is currently a lack of
research concerning the resolution required for finer wood attributes (Kahriman et al. 2018; Sun
et al. 2019). In this study, both distance-dependent and distance-independent competition indices
were used to characterize neighbourhood competition regime.
One of the distinct benefits of managing for competition in a mixedwood forest is
overyielding, or the increase in productivity of a mixed species stand compared to a monoculture
(Hector 1998). Overyielding can occur when managing for space partitioning and size inequality
between spruce and aspen (Hector 1998; Kweon and Comeau 2019). The potential for

45

overyielding in boreal aspen spruce mixtures has been identified (Man and Lieffers 1999;
Kabzems et al. 2007). However, an increase in volume is only desirable when it corresponds
with merchantable timber. The characterization of wood attributes along with cost effective land
use management is at the forefront of the boreal forest industry’s fundamental shift from
traditional wood products to multiple value-added forest products (Chen et al. 2017).
Wood attributes within both broadleaves and conifers have impacts on tree survivability
and the grade of economic viability (Table 1.1). Within angiosperms (aspen) and gymnosperms
(spruce), xylem cells are important for structural integrity and for water transport within the tree.
Gymnosperm tracheids are characterized by many bordered pits in their radial walls, and can be
much longer than angiosperm fibres (Barnett and Bonham 2004). The orientation and magnitude
of certain characteristics within each tracheid reflects the influence of external forcing which can
include competition, or environmental conditions such as temperature changes, nutrient
availability, precipitation, increased wind stress or snowpack (Speer 2012). Therefore,
developing a thorough understanding of the complex interactions of the mixedwood system, and
how, specifically competition between overstory trees may influence tree growth and wood
properties, is vital to the effective future management of these forests.
In this study, the primary objective is to characterize the wood attributes of both white
spruce and trembling aspen grown in varying levels of competition. The first sub-objective is to
compare spruce and aspen wood attribute response to inter- and intra-specific competition. The
second objective is to determine which wood attributes of which species varied with the tested
distance-dependent or distance-independent competition indices. The third and final objective is
to compare the responses of wood properties to neighbourhood and stand level competition. To
achieve these sub-objectives, three hypotheses were formulated. The first hypothesis is that

46

aspen wood attribute response will be most affected by intra specific competition, since aspen is
a shade intolerant species and can grow in close proximity with other aspen. Alternatively,
spruce was planted at prescribed densities, and as such will likely be affected by inter-specific
competition (i.e., competition from in-grown aspen). The variation in shade tolerance, as well as
below ground root behaviours between species leads to the second hypothesis that aspen
neighbourhood level competition will best be described by a distance dependent index and
spruce sample neighbourhood competition will best be described by a distance-independent
index. The final hypothesis is that aspen wood attributes will be most influenced by
neighbourhood competition while spruce will be most affected by stand level competition.

3.3 MATERIALS AND METHODS
3.3.1 Study Site
See section 2.3.1.
3.3.2 Data Collection
See section 2.3.2.
3.3.3 Sample Preparation
See section 2.3.3
3.3.4 Data analysis
The fibre attributes analysed included: wood density (kg/m3), radial diameter (µm),
tangential diameter (µm), coarseness (µg/m), cell population (#/mm2), microfibril angle (MFA,
degree), modulus of elasticity (MOE, Gpa), cell wall thickness (µm), specific surface area
(m2/kg), ring width (mm), and ring area (mm2). For each of the fibre properties analysed, a mean
value was calculated for each ring in order to determine if significant differences occurred

47

between trees from varying treatments. Intra-annual differences were observed, but exceed the
scope of this paper, as they did not exhibit clear increasing or decreasing trend over one ring
width. The data analysis was comprised of three main parts: competition index calculation,
neighbourhood competition model term selection, and mixed effect model evaluation. The null
hypothesis was that wood attribute development is consistent across all trees of the same species,
regardless of their competition index. The significance level selected for all levels of the
statistical analysis was α = 0.05 (95% confidence level).
Based on the plot data for each sample, four competition indices were calculated for each
tree. The first, basal area (denoted BA, Equation 4) is a common distance-independent
competition index that describes the average amount of an area (hectare) occupied by tree stems
(m2). The second competition index, basal area of larger trees (denoted BAL, equation 5) was
calculated in the same way to equation 4, except that only the competitors with a larger diameter
at breast height than the diameter at breast height of the sample were included. This is a
commonly used index where it is assumed that neighbouring trees that are smaller than sample
tree do not place the sample tree at a competitive disadvantage (Wykoff et al. 1982; Sun et al.
2019).
The Martin and Ek index (denoted MNE, Equation 6, (Martin and Ek 1984)) was chosen
as the distance dependent index as it was shown to perform well in similar stand conditions in
previous research (Kahriman et al. 2018). The MNE index places a larger “competitive weight”,
on neighbouring trees that are larger, and closer to the sample tree. The fourth competition index
was another iteration of the Martin and Ek index, with the same size restriction imposed in the
determination of BAL, where only the competitors that had a larger diameter at breast height
than the diameter at breast height of the sample were included (denoted as MNEL, Equation 7).

48

!

"

[4]
BA = (# × & ! ' )
"##
Where d is in cm, to give BA in m2/ha.

(! %! ) "

[5]
BAL = (# × & ! " ' )
"##
Where i is subject tree; j, competitor; d, Diameter at breast height (cm), to give
BAL in m2/ha; ( ' > ), those competitors that are greater in dbh than the dbh of
the sample (Wykoff et al. 1982).
!

[6]

MNE = ∑'0 (!!) × - ( *×,"! )/(!" .!!)

[7]

MNEL = ∑'0 (

"

Where i is subject tree; j, competitor; d, Diameter at breast height (cm);
Lij,distance of subject tree I to competitor j (m) (Martin and Ek 1984).
(!! %!" )
!"

) × - ( *×,"! )/(!" .(!! %!"))

Where i is subject tree; j, competitor; d, Diameter at breast height (cm);
Lij, distance of subject tree i to competitor j (m). ( ' > ), those competitors that
are greater in dbh than the dbh of the sample.
Additionally, two iterations of each of the four competition indices aforementioned (BA,
BAL, MNE, MNEL), were run in order to determine the effect of intra- vs inter-specific
competition. The first iteration included both aspen and spruce competitors combined (denoted
BA, BAL, MNE, MNEL). The second iteration separated aspen and spruce competitors into two
terms (BA.aspen, BA.spruce, BAL.aspen, BAL.spruce, etc) but included both in the model in
order to have the same cumulative competition index value as the combined competitor term.
Therefore, eight indices were tested in total: BA, BA.aspen & BA.spruce, BAL, BAL.aspen &
BAL.spruce, MNE, MNE.aspen & MNE.spruce, MNEL, MNEL.aspen & MNEL.spruce (where
the L denotes a model with the size restriction).
In order to determine which of the competition indices would be used as the
neighbourhood competition term for the mixed effect model for each species, Akaike’s
Information Criterion (AIC value) was used to rank each model for each wood property. For
each of the eleven wood attributes for each species, the best competition index was determined

49

as that with the lowest AIC score for the most wood attributes. The competition index that
described the most wood attributes for each species was then used as the neighbourhood
competition term in the mixed effect model.
A linear mixed effect model (Equation 8), with tree as the random factor, and ring
(cambial age), stand level competition (treatment) and neighbourhood level competition (one of
the eight competition indices) as the fixed factors, was conducted to investigate differences in
fibre attributes.
[8]

Yj=β0+β1xj+β2xj+ (β1 β2)+ μ1 +εj,

Where Yj and xj represent fibre property for year j; β0, β1 and β2 are the fixed effects.
μi1 is the random effect of tree and ui2 is the random slope; εij is the error term.
This portion of the analysis was completed using the lme function within the nlme
statistical package (Pinheiro et al. 2020) in R (version 3.4.3) (R Development Core Team 2017).
The relationships between growth levels and competition have been shown to vary between
young and old stands, resulting in the need to parameterize models that characterize growth as a
function of competition (Filipescu and Comeau 2007). The parameterization applied to this study
was that measurements corresponding with a cambial age less than year 15 were omitted, to
reduce the variation evident in early age wood attributes (Figure 3.1).

50

Figure 3.1 Microfibril angle change over time for aspen samples (top), and spruce samples (bottom), separated by
treatments.This figure is shown as an example of the fluctuation of values between years 0 and 10 in both spruce
and aspen samples and demonstrates that both approximate linear changes after year 15. This trend throughout
wood attributes provided the basis for paramaterization. For the figures for each wood attribute and species,
consult Chapter 2 Supplementry Appendix.

51

The Rb2 statistic was calculated using the Kenward-Roger approach within the r2glmm
(method = “kr”) function in R (Halekoh and Højsgaard 2014). The linear mixed model lmer
function within the lme4 package (Bates et al. 2015) was used for the calculation of effect size,
as the rbeta function does not work with the lme function in the nlme package. This statistic
estimates the R2 value for mixed effect models in the R environment. Tukey’s test for significant
differences (TukeyHSD function in “stats” package (R Development Core Team 2017)) was
used to determine differences between relative contributions of aspen and spruce competitors to
the overall competition index.

3.4 RESULTS
3.4.1 Inter- vs intra specific competition
Table 3.1 summarizes the results of Tables 3.2 and 3.3, which show the results of the mixed
effect model analysis that included cambial age, and neighbourhood level competition terms in
the form of either a combined species “both” term or separated “aspen” and “spruce” terms.
Building on results of the Chapter 2, we reviewed differences between treatment for each wood
attribute. This analysis builds on the previous model by including plot level (3.99m) competition
index terms. Due to the vast amount of information included in model outputs for each wood
attribute, the results of these analyses have been synthesized into several summary tables. Full
statistical results can be found in “Chapter 3 Supplementary Appendix” attached.
In order to test the first hypothesis, that aspen is more affected by intra-specific competition
and that spruce is more affected by inter-specific competition, the number of significant results
for species combined competition indices were tallied and compared to those indices which
separated aspen and spruce (Table 3.1, 3.2 and 3.3). The first result is that, out of the species

52

separated competition terms, the aspen term was more often significant than the spruce term,
indicating that aspen is more susceptible to intra-specific competition, while spruce is more
susceptible to inter-specific competition. Further, comparing the total number of significant
neighbourhood level terms show that aspen wood attributes were determined to be more
sensitive to neighbourhood level competition compared to white spruce. Finally, these results
indicate that both spruce and aspen samples are more affected by competition models that
included a combined “both” term rather than separated aspen and spruce terms. This shows that
defining the overall competitive environment is more important than separating competitor
species.

53

Overall % significant
Number of significant
observations

Overall % significant
Number of significant
observations

Spruce

2

6%

30%
7

52%
12

4

17%

Spruce significance summary
Aspen

13

18

Both

39%

55%

Both

Aspen significance summary
Aspen
Spruce

23

Total

33

total

54

Table 3.1 A summary of the results from the evaluation of the neighbourhood level competition index mixed effect models (a < 0.05) Percentages used for ease of
interpretation (dividing the number of significant observations of either “both” “aspen” or “spruce” by the total number of significant observations of each
species). For full statistical results, consult Table 3.2 and Table 3.3 as well as Chapter 3 supplementary Appendix.

0.121a

0.132

0.061a

0.154

0.134a

MFA

Density

Ring area

Ring width 0.147
Tangential
0.741a
Diameter
Radial
0.194
Diameter
Specific
surface
0.439
area
Cell
0.334
Population
Coarseness 0.965
Wall
0.579
Thickness

0.591a
0.003a
0.191

0.002
0.072e

BAL.
aspen

0.933a

BAL

0.169

0.227c

0.448

BAL.
spruce

0.538
0.673
0.246
0.572
0.358
0.318

0.15

0.528

0.27

0.844

0.672

0.932

0.249

0.013

0.768

0.01

0.091

0.9d <0.001a

0.681

0.189

0.018

0.885

0.012

0.108

0.588

0.971

0.343

0.399

0.404

0.539

<0.001a <0.001e

0.872a,c <0.001a,c <0.001a <0.001a

0.429

0.5144

0.554

BA.
spruce

0.653a

0.17

0.035a

0.992

0.933

MOE

BA.
aspen

BA

Wood
attribute
0.833

MNE.
aspen

0.068
0.167a
0.421

0.005
0.023

0.385a

0.054
0.009

0.194

0.071

0.053
0.006

0.47

0.462a

<0.001d,e

0.0012a,d,e

0.597 0.778c,d,e

0.0213c,e 0.273a,c,e

0.628

MNE

0.958a

MNEL

0.912e

0.699

0.949a

0.598

0.904

0.884

0.373

0.089

0.039a

0.016a

0.022

0.009
0.019

0.054

0.033

0.005
0.212a

0.054

0.003a

0.019a

0.465

0.123a

0.455a

MNEL.
aspen

0.74

0.137c <0.001a,d,e

0.295c <0.001a,d,e

0.559

0.929a 0.0081a,c,d,e

0.972

MNE.
spruce

55

0.372

0.51a

0.843

0.237

0.816

0.534

0.08e

0.138

0.299

0.411

0.372

MNEL.
spruce

Table 3.2 Summarized aspen sample tree p-value results for the competition index term of each mixed effect model iteration (Full model results in Chapter 3
Supplementary Appendix). Each column represents a distinct model iteration (which included age, treatment, one of the competition index terms, and the random
effect of tree). Bolded values correspond with lowest AIC value for each wood attribute. Significant interactions are denoted as follows: age (a), treatment: 1k
(b), treatment: 2k (c), treatment: 5k (d), treatment: 10k (e).

0.523

0.119

0.322

0.44

MOE

MFA

Density

Ring area

Ring width 0.166a
Tangential
0.989a
Diameter
Radial
0.371a,d
Diameter
Specific
surface
0.332
area
Cell
0.556a,d
Population
Coarseness 0.829a
Wall
0.515a
Thickness

BA

Wood
attribute

0.412b
0.848

0.887a
0.4

0.513a

0.006

0.003

0.055a

0.107a

0.345

0.447

0.342

0.992

0.265

0.409a

0.012a

0.002a

0.056a 0.0722

0.372a

0.062

0.056

0.232

0.019a

<0.001

0.581a

<0.001a

0.015

<0.001d,e

0.505

0.727

0.233

0.6

0.08

0.534

0.0012a

0.161

0.031

0.0091

0.766

0.37

0.175

0.994

BAL.
spruce

0.347b

0.591b

BAL.
aspen

0.754

0.243a

0.224

0.17a

0.105

BAL

0.083a

0.148

0.084

0.253a

0.973

BA.
BA.
aspen spruce

0.431a

0.337a

0.488

0.942a 0.315a

0.215

0.135

0.243a 0.639a

0.177
0.188a

0.287a

0.499

0.162

0.821

0.54e

0.569a

0.016

0.9a

0.005a 0.167a

0.556

0.1a

0.08

0.241a

0.581

0.004

0.062a
0.202

0.015

0.876a 0.048a
0.4a

MNE.
spruce

MNE.
aspen

MNE

0.506a

0.395a

0.005

0.588

0.006d,e

0.21

0.046a,c,d,e

0.034c,d,e

0.033a,c

0.228a,b

0.997a,b

MNEL

0.892a

0.644a

0.112

0.882

0.235

0.395

0.988a

0.97a

0.366a

0.804a

0.912a

MNEL.
aspen

0.716

0.896

0.694

0.8

0.37e

0.998

56

0.116a,b,d,e

0.08a,b,d,e

0.514

0.511a

0.9002b

MNEL.
spruce

Table 3.3 Summarized spruce sample tree p-value results for the competition index term of each mixed effect model iteration (Full model results in Chapter 3
Supplementary Appendix). Each column represents a distinct model iteration (which included age, treatment, one of the competition index terms, and the random
effect of tree). Bolded values correspond with lowest AIC value for each wood attribute. Significant interactions are denoted as follows: age (a), treatment: 1k
(b), treatment: 2k (c), treatment: 5k (d), treatment: 10k (e).

There were distinct differences in the type of competition index that best described
neighbourhood level competition for each of aspen and spruce (Table 3.4). Table 3.4 is the
summary of table 3.5, which shows the AIC rankings of each neighbourhood competition index
included in the mixed effect model for each wood property.
Table 3.4 The best neighbourhood competition term was determined by comparing the eight competition index
terms, for each of the eleven wood attributes, for both aspen and spruce samples. Percentages show the number of
times each competition index was chosen as the best term for each wood attribute divided by the total number of
fibre properties (11).

Aspen AIC results
Competition
index

Number of fibre
properties

Percentage

MNE

5

45%

BAL
MNEL

4
2

36%
18%

Spruce AIC results
Number
Competition
Percentage
of fibre
Index
properties
BAL
6
55%
MNE.aspen.spruce

MNE

BAL.aspen.spruce

BA

2
1

18%
9%

1
1

9%
9%

For aspen, no models which separated spruce and aspen competition terms improved
model performance. The distance-dependent Martin and Ek competition index (MNE) yielded
the lowest AIC score for five of the eleven aspen wood attribute models. There was more variety
observed in the spruce samples, where several models displayed low AIC scores (Table 3.4).
However, the distance-independent BAL (basal area where dbh competitor > dbh sample)
yielded the lowest score for six out of eleven wood attributes in spruce so it was chosen as the
best index for spruce. MNE was used as the neighbourhood competition term for the following
iterations of the mixed effect models for aspen, and BAL was used for spruce mixed effect
models in order to separate the effects of age, competition index and stand level competition.

57

Table 3.5 The best neighbourhood competition term was determined by comparing the eight competition index
terms, for each of the eleven wood attributes, for both aspen and spruce sample, and ranked by AIC value.

Aspen
Wood attribute
Ring Width

Model
lme.BAL.a
lme.BAL.aspen.spruce.a
lme.MNEL.a

Spruce
df
AIC
8 1324.418

Model
lme.BAL.s

df
AIC
9 790.0412

9 1325.607 lme.BAL.aspen.spruce.s 10 790.2578
8 1333.212

lme.MNEL.s

9 800.5454

lme.MNEL.aspen.spruce.a 9 1334.793 lme.MNEL.aspen.spruce.s 10 800.9915
lme.MNE.a

8 1341.907

lme.MNE.s

9 801.7944

lme.MNE.aspen.spruce.a 9 1343.009

lme.BA.s

9 802.6489

lme.BA.aspen.spruce.s

10 803.3094

lme.BA.a
lme.BA.aspen.spruce.a

Density

8 1355.271

9 1357.051 lme.MNE.aspen.spruce.s 10 803.7911

lme1.a

5 1749.084

lme1.s

6 1222.7332

lme.BAL.a

8 7167.262

lme.BAL.s

lme.BA.a

8 7168.487 lme.BAL.aspen.spruce.s 10 6421.236

9

6419.25

lme.BAL.aspen.spruce.a

9 7168.962

lme.MNEL.s

9 6421.528

lme.MNE.a

8 7170.278

lme.BA.aspen.spruce.s

10 6423.314

lme.BA.aspen.spruce.a
lme.MNEL.a

9 7170.415 lme.MNEL.aspen.spruce.s 10 6423.454
8

7170.55

lme.MNE.s

9 6424.786

58

Wood attribute

Model

df

AIC

lme.MNEL.aspen.spruce.a 9 7171.448

Model

df

AIC

lme.BA.s

9 6425.188

lme.MNE.aspen.spruce.a 9 7172.005 lme.MNE.aspen.spruce.s 10 6426.195
lme1.a
lme.BAL.a

5 7732.726

lme1.s

6 6854.024

8 1965.562 lme.MNE.aspen.spruce.s 10 2734.74

MFA
lme.BAL.aspen.spruce.a

9 1967.315

lme.BA.s

9 2738.692

lme.MNEL.a

8 1968.172

lme.MNEL.s

9 2739.686

lme.MNE.a

8 1969.939

lme.BA.aspen.spruce.s

10 2740.197

lme.BAL.s

9 2740.267

lme.MNE.s

9 2740.448

lme.MNEL.aspen.spruce.a 9 1970.108
lme.BA.aspen.spruce.a
lme.BA.a

9 1971.504

8 1971.778 lme.BAL.aspen.spruce.s 10 2741.236

lme.MNE.aspen.spruce.a 9 1971.884 lme.MNEL.aspen.spruce.s 10 2741.48
lme1.a

5 3284.046

lme1.s

6

3730.14

lme.MNE.a

8 1941.123 lme.MNE.aspen.spruce.s 10 1901.319

lme.BAL.a

8 1941.355

MOE
lme.BA.aspen.spruce.s

10 1904.728

59

Wood attribute

Model
lme.BA.a

df
AIC
8 1941.355

Model
lme.BA.s

df
AIC
9 1905.025

lme.MNEL.a

8 1941.359

lme.BAL.s

9 1905.146

lme.BAL.aspen.spruce.a

9 1942.479

lme.MNE.s

9 1905.416

lme.MNEL.aspen.spruce.a 9 1942.486

lme.MNEL.s

9

lme.BA.aspen.spruce.a

1905.44

9 1942.981 lme.BAL.aspen.spruce.s 10 1906.62

lme.MNE.aspen.spruce.a 9 1943.122 lme.MNEL.aspen.spruce.s 10 1907.424

Coarseness

lme1.a

5 3021.362

lme1.s

6 2979.759

lme.MNE.a

8 8640.239

lme.MNE.s

9 6512.454

lme.MNE.aspen.spruce.a 9 8642.127 lme.MNE.aspen.spruce.s 10 6512.676
lme.MNEL.a

Tangential
Diameter

8 8642.599

lme.BAL.s

9

6513.03

lme.MNEL.aspen.spruce.a 9 8643.152

lme.MNEL.s

9 6513.109

lme.BA.s

9 6513.795

lme.BAL.a

8 8646.848

lme.BA.a

8 8648.203 lme.BAL.aspen.spruce.s 10 6514.993

lme.BAL.aspen.spruce.a

9 8648.438 lme.MNEL.aspen.spruce.s 10 6515.108

lme.BA.aspen.spruce.a

9 8649.159

lme.BA.aspen.spruce.s

10 6515.11

lme1.a

5 9228.614

lme1.s

6 7200.806

lme.MNE.a

8 3457.947

lme.BAL.s

9 2252.067

60

Wood attribute

Model
lme.MNEL.a

df

AIC

df

AIC

8 3458.519 lme.BAL.aspen.spruce.s 10 2253.732

lme.MNE.aspen.spruce.a 9 3458.556
lme.BAL.a

Model

8 3458.871

lme.MNEL.s

9 2254.194

lme.MNE.s

9 2255.447

lme.MNEL.aspen.spruce.a 9 3459.442 lme.MNE.aspen.spruce.s 10 2255.668

lme.BAL.aspen.spruce.a

9 3460.787

lme.BA.s

lme.BA.a

8 3461.678

lme.BA.aspen.spruce.s

lme.BA.aspen.spruce.a

Radial
Diameter

9

2255.8

10 2255.911

9 3463.021 lme.MNEL.aspen.spruce.s 10 2256.141

lme1.a

5 3970.028

lme1.s

6 2694.796

lme.MNEL.a

8 3105.711

lme.BAL.s

9 2408.213

lme.MNE.a

8

3106.02

lme.MNE.s

9 2410.153

lme.BAL.a

8 3107.104

lme.MNEL.s

9 2410.198

lme.MNEL.aspen.spruce.a 9 3107.551 lme.BAL.aspen.spruce.s 10 2410.212

lme.MNE.aspen.spruce.a 9 3107.958 lme.MNEL.aspen.spruce.s 10 2412.033

lme.BAL.aspen.spruce.a

9 3108.751 lme.MNE.aspen.spruce.s 10 2412.113

61

Wood attribute

Specific Surface
Area

Model

df

lme.BA.a

8 3112.174

lme.BA.aspen.spruce.s

10 2412.577

lme.BA.aspen.spruce.a

9 3113.162

lme.BA.s

9 2417.235

lme1.a

5 3650.971

lme1.s

6 3084.568

lme.MNE.a

8 6874.273

lme.BA.s

9 6324.396

lme.MNEL.aspen.spruce.a 9 6876.251

lme.BAL.s

9 6324.437

lme.MNE.aspen.spruce.a 9 6876.252

lme.MNEL.s

9 6325.056

lme.MNE.s

9 6325.339

lme.MNEL.a

AIC

8 6876.486

Model

df

AIC

lme.BA.a

8 6877.462 lme.MNE.aspen.spruce.s 10 6325.482

lme.BAL.a

8 6877.984

lme.BA.aspen.spruce.s

10 6326.076

lme.BA.aspen.spruce.a

9 6878.564 lme.BAL.aspen.spruce.s 10 6326.437

lme.BAL.aspen.spruce.a

9 6879.325 lme.MNEL.aspen.spruce.s 10 6327.03

lme1.a
lme.BAL.a

5 7448.818

lme1.s

6 6906.066

8 10478.41 lme.BAL.aspen.spruce.s 10 9164.62

Ring Area
lme.BAL.aspen.spruce.a

9 10479.54

lme.BAL.s

9 9165.148

lme.MNEL.a

8 10496.82

lme.MNEL.s

9 9176.824

62

Wood attribute

Model

df

AIC

Model

df

AIC

lme.MNEL.aspen.spruce.a 9 10498.49 lme.MNEL.aspen.spruce.s 10 9176.898
lme.MNE.a

8 10503.15

lme.MNE.aspen.spruce.a 9 10504.79

lme.BA.a
lme.BA.aspen.spruce.a

Cell Population

lme.MNE.s

9 9178.302

lme.BA.aspen.spruce.s

10 9179.621

8 10511.61 lme.MNE.aspen.spruce.s 10 9180.197
9

10513.4

lme.BA.s

9 9180.817

lme1.a

5 11124.87

lme1.s

6 9913.561

lme.MNEL.a

8 9106.109

lme.BAL.s

9 8266.978

lme.MNE.a

8 9106.221

lme.MNEL.s

9 8268.079

lme.BAL.a

8

9106.83

lme.BAL.aspen.spruce.s 10 8268.533

lme.MNE.aspen.spruce.a 9 9107.394 lme.MNEL.aspen.spruce.s 10 8270.064

lme.MNEL.aspen.spruce.a 9 9107.475

lme.BAL.aspen.spruce.a

lme.BA.a
lme.BA.aspen.spruce.a

9 9108.669

lme.MNE.s

9 8270.154

lme.BA.aspen.spruce.s

10 8271.236

8 9112.244 lme.MNE.aspen.spruce.s 10 8271.29
9 9113.212

lme.BA.s

9 8275.697

63

Wood attribute

Model
lme1.a

df
AIC
5 9691.333

Model
lme1.s

df
AIC
6 8940.39

Wall Thickness

lme.MNE.a

8 16.16111

lme.BAL.s

9 -747.3024

lme.MNE.aspen.spruce.a 9 18.15581

lme.MNEL.s

9 -746.8433

lme.BA.s

9 -746.8255

lme.MNE.s

9 -746.3988

lme.MNEL.a

8 18.49016

lme.MNEL.aspen.spruce.a 9

18.8117

lme.BA.a

8 21.12628 lme.MNE.aspen.spruce.s 10 -745.975

lme.BAL.a

8 21.43247 lme.BAL.aspen.spruce.s 10 -745.3287

lme.BA.aspen.spruce.a

9 22.38627

lme.BAL.aspen.spruce.a

9 22.96613 lme.MNEL.aspen.spruce.s 10 -744.9067

lme1.a

5 618.64595

lme.BA.aspen.spruce.s

lme1.s

10 -745.1763

6 -214.5255

3.4.2 Competition index
The competition index terms varied in the relative contributions of each competitor
species across treatments (Figure 3.2). In both spruce and aspen samples, the main contributor to
the overall index value was aspen competitors, as the “A” value (defined as aspen competition)
never differed significantly from the combined “both” indices (competition from both spruce and
aspen). Alternatively, spruce competitors differed from the combined index and the aspen
competitor contribution.
64

Figure 3.2 Aspen and MNE (top) with spruce and BAL (bottom)) showing the proportion that each competitor
species, aspen (A), and spruce (S), contributes to the overall index, both (B).

65

3.4.3 Effect size (Rb2)
The Kenward-Roger approach for calculating the Rb2 (effect size) statistic, was utilized to
determine relative effects of each of the model terms. Overall, the models for each wood
attribute were effective in describing the variation observed in both aspen samples: 23% – 75%,
and spruce samples: 54% – 90% (Table 3.3). The Rb2 approximates an R2 value for mixed effect
models and indicates the fit of the model compared to the data. The terms for each wood attribute
model are represented in the columns for each species. The “model” column shows the overall fit
of the model, and each further column shows the relative contribution of each term to the
model’s performance. The remainder of the variation explained by the model, but not tabulated is
the effect described by the random tree term in the model, but was excluded, in order to focus on
the proportion of the variance explained by fixed effects. The neighbourhood competition model
was chosen for each species based on which performed best for each sample species. For aspen
sample wood attributes, the distance dependent competition index by Martin and Ek (1984)
(denoted MNE) best described neighbourhood level competition variation observed. For spruce
wood attributes, the distance independent competition index of basal area of larger trees (denoted
BAL, where the size of the competitor > size of the sample tree diameter at breast height) best
described neighbourhood level competition variation. The “stand level competition” column
describes the effect attributed to the competitive environment as defined by stems per hectare
aspen (0, 1000, 2000, 5000, or 10000 stems per hectare). This approach to determining effect
size of terms in the model was used rather than global model simplification, as this form of study
answers the question of which model terms are most important for each species and wood
attribute, when the best competition index is included for each species.

66

The model terms for spruce and aspen samples differed in their responses. Each wood
attribute model included four terms: age, neighbourhood level competition (competition index),
stand level competition (treatment measured in aspen stems per hectare), and the random effect
of tree. For each of the eleven wood attributes measured (and for each sample species), one
dominant effect was chosen as the model term that explained the most variation out of the chosen
model terms. Aspen and spruce differed in their dominant effects. In aspen samples, ring area,
ring width, radial diameter, cell population and coarseness variation were all best explained by
the neighbourhood level competition term. In spruce, none of the wood property variation was
explained by the neighbourhood competition term. In both aspen and spruce, age was the
dominant effect for MOE and MFA. However, no other aspen attributes were explained by age,
while specific surface area, coarseness and cell wall thickness were all explained by age in
spruce samples. Stand level competition was the main factor in the variation of density and
tangential diameter in both spruce and aspen. The majority of the wood attribute variation in
spruce was best explained by stand level competition (density, ring area, ring width, tangential
diameter, radial diameter, and cell population) (Table 3.6).

67

0.0159
0.069
0.126

0.342
0.56

0.662

Cell
Population
Coarseness
Wall
Thickness

0.139

0.681

Specific
surface area

0.1

0.129
0.147

0.073

0.145

0.363

0.139

0.102
0.136

0.155

0.079

0.129

0.074

0.025

0.044
0.087
0.129
0.072
0.138

Stand Level
Competition

0.005
0.102
0.006
0.194
0.267

Age

0.206
0.127
0.115
0.045
0.076
0.00000
0.237
8

0.747
0.639
0.611
0.481
0.602

model

MOE
MFA
Density
Ring area
Ring width
Tangential
Diameter
Radial
Diameter

Wood
attribute

Neighbourhood
Competition
(MNE)

Aspen Samples

0.729

0.79
0.836

0.734

0.598

0.732

0.905
0.873
0.591
0.54
0.811

model

0.255

0.276
0.387

0.26

0.091

0.223

0.559
0.479
0.064
0.005
0.302

Age

0.013

0.121
0.012

0.013

0.131

0.052

0.0042
0.0125
0.093
0.207
0.186

Neighbourhood
Competition
(BAL)

Spruce Samples

68

0.131

0.441
0.295

0.15

0.364

0.351

0.108
0.327
0.163
0.455
0.441

Stand Level
Competition

Table 3.6 A summary of the R2 values calculated by the effect size (Rb2) statistic results. Complete results can be found in Chapter 3 Supplementary Appendix
(Page 192).

3.5 DISCUSSION
Our analysis indicated that aspen is more sensitive to neighbourhood level competition than
spruce (Table 3.1). Out of the 88 iterations (11 wood properties x 8 competition index models)
for each species, 38% of the time, neighbourhood competition terms significantly explained the
variation in aspen wood attributes and 26% of the time in spruce. Our results correspond with the
trees life-history characteristics as aspen is shade intolerant and reacts negatively to competition
(Farmer 1963), compared to spruce which (under aspen canopy) exhibits linear height increment
growth in light levels up to 40%, with negligible gain in height increment exhibited between
40% and 100% (Lieffers and Stadt 1994; Groot 1999).
Aspen was found to be more affected by competition from other aspen than from spruce,
based on the observed significance in the models with separated species terms (BA.aspen &
BA.spruce, BAL.aspen & BAL.spruce, etc.) (Table 3.4). Intra-specific competition is often more
detrimental to non-shade tolerant species due to increased leaf area index and canopy closure
from competition from the same species (Pretzsch and Biber 2016). Conversely, in accordance
with previous studies documenting the negative effect of broadleaf competition on white spruce
growth, sampled spruce were more affected by inter specific competition (Bérubé-Deschênes
2017). However, our results suggest that competition from both spruce and aspen needs to be
considered when assessing how wood properties are impacted by competition at the
neighbourhood stand level. This finding could be a reflection of the uneven number of aspen
competitors (~70%) in this study. In this case, the sheer amount of aspen in each treatment drives
the competition index value in the combined “both” term (Figure 3.2). Therefore, since the
competition index that included a combined term for aspen and spruce preformed best for each

69

wood attribute model, our results indicate that differentiating the competitor species is less
important than defining the overall competitive environment.
As hypothesized, there was a difference between best neighbourhood level competition
index for aspen and spruce (Table 3.4). For aspen, the distance dependent MNE index best
described the majority of wood attribute variation with neighbourhood competition environment
(Table 6.1). Aspen typically develop large crowns, are shade intolerant and as such, require more
growing space than a shade tolerant species. Since broadleaves are often more susceptible to
competition, issues related to developing large crown radii can result in a larger importance
placed on proximity to a competing tree (Pretzsch 2019). Alternatively, for spruce samples, the
distance-independent BAL (basal area where DBHcompetitor > DBHsample) index best described the
majority of wood attribute variation. BAL has also been found to effectively describe the
neighbourhood competition effect on basal area increment growth of white spruce (BérubéDeschênes 2017) and other conifers (Kahriman et al. 2018). This could be explained by the fact
that shade tolerant species have been found to be less sensitive to competition (Canham et al.
2006). Interestingly, the best models for both spruce and aspen included a size term (MNE size
ratio term, and BAL size cut-off) further demonstrating that plant size can influence the intensity
of competition (Canham et al. 2004).
Through the calculation of the Rb2 statistic, each modeled wood attribute was determined
to be most affected by one of the three model terms: (i) age, (ii) neighbourhood level competition
(MNE/BAL), or (iii) stand level competition (treatment) (Table 3.6). For aspen samples, the
dominant model term for the tested model structure for each wood attribute was one of the two
competition terms whereas the majority of spruce wood attributes were dominated by the age and
stand level terms.

70

Within aspen samples, the neighborhood competition effect (MNE index) was the
dominant effect in ring width, ring area, radial diameter, cell population, and coarseness (Table
3.6). Ring width (and ring area which is directly related) was negatively impacted by increased
competition levels in aspen samples. This is consistent with the exterior growth characteristics
observed in trees in competitive environments, whereby height, rather than diameter growth is
preferred (Coopersmith and Hall 1999; Carr et al. 2020). Aspen coarseness values have been
shown to be statistically different and larger in pure stands than mixed stands (De Araujo et al.
2015). Our results expand on this finding as they indicate that stands with an increased
proportion of aspen leads to higher values of coarseness. Coarseness is an important
consideration for paper manufacturing but varies in desirability depending on end use. If a paper
with improved density, strength and optical properties is the desired end product, then low
coarseness is desirable as it generates higher fibre collapsibility (low aspen treatment densities).
If high porosity paper is the desired end product, then higher coarseness values are necessary
(high aspen treatment densities) (Seth and Kingsland 1990).
Stand level competition (treatment) was the dominant effect for density, tangential
diameter, specific surface area and cell wall thickness in aspen samples. Density and cell wall
thickness often vary together as density is the proportion of material in the cell which depends on
the ratio of cell wall thickness and cell diameter (Lundgren 2004; Carrillo et al. 2015). Thickness
in the cell wall has been shown to respond to external factors, such as exposure and climate
(Hein and Lima 2012). A possible explanation for the change in density is that variation in
competition regime can influence the sensitivity to these factors, thereby changing the cell wall
thickness and density. The effect of stand composition on wood density in aspen was
documented by De Araujo et al. (2015), but no difference between density in mixed and pure

71

stands were found. The differences observed in our study could be a result of comparing various
managed levels of treatments in homogenous growing conditions and even aged stands rather
than unmanaged uneven sites.
The dominant model terms within spruce wood attributes were split between age and
stand level competition (Table 3.6). The lack of neighbourhood level competition effect could be
due to spruce’s shade tolerance and relative (to aspen in this study) insensitivity to competition.
The variation observed in MOE, MFA, specific surface area, coarseness and wall thickness were
best described by age in spruce samples. It was surprising that cell wall thickness was more
affected by age and not competition as thickness in the cell wall has shown to change with
growth rates from fertilization (Lundgren 2004). However, it has been shown that age of conifers
play an important role in the timings and duration of xylem formation: cell production in the
cambial zone, cell expansion and cell wall thickening (Zeng et al. 2018). A possible explanation
for why cell wall thickness was best described by age in this study, is that some of the strongest
variation in growth rates between treatments were observed (and removed from our
parameterization) early in stand development (< 15 years). This variation stabilized over time to
similar values regardless of treatment.
Stand level competition was the dominant model term for density, ring area, ring width,
tangential diameter, radial diameter and cell population in spruce. Ring width and ring area can
vary with competition as discussed previously with the preferential height to diameter growth in
increasing competition regimes. Water availability changes in microsite competitive
environments, and has been linked to the formation of wider tracheids, which could explain the
strong competition signal in radial diameter (Vysotskaya and Vaganov 1989; Wilpert 1991;
Kerhoulas et al. 2013).

72

The main similarity between aspen and spruce was that MOE and MFA were best
described by the age term for both species. MOE and MFA are two terms that often vary together
and are typically associated with the age of the tree (Barnett and Bonham 2004). Up to 86% of
the variation in MOE can be explained by variation in MFA (Evans and Ilic 2001). Microfibril
angle is the orientation of the crystalline cellulose in the secondary cell wall (S2) wood along the
fibre axis, and this orientation is driven by growth rate of the tree (Cave 1966). Typically, high
MFA values result from fast initial growth rates as the tree develops resulting in the inversely
proportional low MOE, allowing the stem to bend without breaking. As the tree ages, it must
support the crown and weight of branches, and stiffens, resulting in a low MFA and high MOE
(Barnett and Bonham 2004). In this study, while age was the most important determinant for
MFA in both spruce and aspen, neighbourhood level, and stand level competition for both spruce
and aspen respectively were next best terms (Table 3.6). In spruce samples, one of the best
performing models across wood attributes, was in fact the model for MFA, which explained 87%
of the variation, and age together with stand level competition explained 80%, with around 6%
of the variation attributed to the random effect of tree. This competition effect would likely be
pronounced if the entire age of the tree was used (rather the < 15-year cut-off) as there is a large
variation in the initial values of MFA in the early development (age 0-5) of spruce. These
variations corresponding with lower MFA’s with increasing levels of treatment, indicating faster
growth rates with increasingly competitive environments (Figure 3.1). Thus, while age was the
most important factor in the model that described MFA change over time, stand level
competition is also an important consideration for a comprehensive model.

73

3.6 CONCLUSION
We examined eleven wood attributes of aspen and white spruce sampled at a Northern
British Columbia trial site and developed individual tree models to elucidate the effects of
competition on wood attribute development. Our results demonstrated that differentiating the
competition by species was less important than defining the overall competitive environment for
both aspen and spruce wood attributes. We tested four neighbourhood competition indices and
found that aspen wood attribute variation was best described by a distance dependent index,
while spruce was best described by a distance independent index. Both of the neighbourhood
competition indices included a size term, indicating that the size of the competitors was
important for both species wood attribute development. Moreover, we developed a mixed effect
model to describe each wood attribute development that included cambial age, neighbourhood
competition, and stand level competition. Our results indicated that defining the neighbourhood
level competition regime was most important for aspen but was relatively unimportant in spruce
where cambial age and treatment were most important. Two of the most important wood
attributes, microfibril angle and modulus of elasticity were best described by age and not
competition in both spruce and aspen. These preliminary models lay the foundation for further
studies that can make recommendations to forest managers based on desired end product. Since
this study was conducted with samples from a managed trial site, the next step would be to
sample natural even aged stands and compare wood attribute development across a number of
sites. This would require careful consideration in order to find adequately comparable sites from
an age and ecological perspective.

74

4. Conclusions and Future work
4.1 Conclusions
The key finding of my thesis is that wood attributes of white spruce and trembling aspen
vary with competition and differ in their response to competition between species.
Characterizing the effect that competition has on aspen and spruce wood attribute development is
essential in order to effectively balance the benefits and shortcomings of managing for pure or
mixed stands of aspen and white spruce. In this concluding chapter, I synthesize the main
findings from each of the previous chapters, while I discuss the implications for management
techniques, and recommend future research directions.
In Chapter 2, Evaluating the impact of stand composition and competition on Spruce and
Aspen wood attributes in mixedwood stands, the objectives were threefold; (i) to determine if
there was a difference in wood attribute response to stand level competition between species, (ii)
to determine if aspen and white spruce display similar responses to stand level competition over
time, and (iii) to analyze if there are any economic benefits observed in wood attributes with
increasing aspen treatment density. First, my results indicated that aspen treatment density had a
significant effect on most wood attributes in aspen and spruce samples and that each sampled
species showed differences in their response. Second, my results indicated that all spruce wood
attributes exhibit non-linear change as the trees develop (cambial age 0-5 years), and
approximate linear change in older trees (cambial age 5-30 years). Finally, I developed a
framework for visualizing change in industrial desirability of wood attribute as they vary with
increasing levels of competition.

75

For most of the wood attributes that I measured (ring width, density, MOE, coarseness,
tangential diameter, radial diameter, specific surface area, ring area, cell population, and wall
thickness), two thresholds consistently emerged from the analysis.
First, the largest effect on spruce growth occurred when aspen was grown in densities
greater than zero with the spruce (i.e., treatments of > 0 sph). Ninety percent of the significant
differences within spruce wood properties, and between treatments, were attributed to the
presence of aspen in the stand. This result was expected as aspen growing in mixture with spruce
has been shown to impact the availability of key limiting resources with regards to light, water
and nutrients (Filipescu and Comeau 2007; Kabzems et al. 2015). Additionally, past studies have
demonstrated that modifications to spruce’s growing environment influence absolute growth as
well as gross growth characteristics such as height-to-diameter ratio (Kabzems et al. 2015).
The second threshold that emerged from my results was the impact of an aspen treatment
density of 10,000 sph, on aspen wood characteristics. Differences between the 10,000 sph
treatment and those with lower densities of aspen accounted for all of the significance observed
in aspen wood attributes.
The final objective of Chapter 2 was to determine potential industrial desirability of each
wood attribute as a function of competition for both spruce and aspen. As the desirability of
certain wood attributes depend entirely on the end product, this section serves only as a general
framework and not a guide for determining economic suitability with changing competition. The
criteria for desirability were based on the typical uses of each wood attribute as defined by
natural resource Canada, I found that spruce wood attributes are generally increased when spruce
is grown in mixture with aspen. Increased wood density, larger MOE, lower MFA, larger ring
width, and increased wall thickness are the benefits observed in spruce grown in the 5000 sph

76

treatment which constitute either best or second-best improvements to wood quality, relative to
other treatments. Managing for higher MOE can be desirable since MOE is the best predictor of
wood stiffness. Although density may be less important for lumber products in the future due to
the increased production of value added engineered wood products, density is still important for
bioenergy, which values the higher energy content of dense wood (Canadian Wood Fibre Centre
et al. 2010). These results indicate that there are multiple benefits to spruce wood attribute
development up to 5000 sph treatment of aspen.
In contrast to spruce wood attributes, our results indicate that most desirable traits (MOE,
density, and MFA) within aspen decrease with increasing levels of aspen competition (Figure
2.8). Lower values for MOE and density increase the risk of trunk breakage, xylem implosion
and decreased stem stability (Hacke et al. 2001). Effectively, in this instance, the increased
competition is weakening the aspen, and accelerating the stand dynamics typical of mixedwood
forests (Bergeron et al. 2014). However, the majority of the negative associations for the
remaining wood attributes are related to the 10,000 sph aspen treatment, suggesting that, aspen
wood attributes are not significantly affected until densities are increased past a certain threshold.
In Chapter 3, A mixed effect approach to evaluating the impacts of stand composition,
age and neighbourhood competition on the wood attributes of spruce and aspen grown in
mixedwood stands, the mixed effect model of Chapter 2 was expanded by the inclusion of 3.99m
plot data that incorporated information about each competitor and formed the basis for
neighbourhood level competition indices. The objectives were again, threefold; (i) to compare
spruce and aspen wood attribute response to inter- and intra- specific competition, (ii) to
determine which neighbourhood level competition indices best improved a model for wood
attributes in both aspen and spruce, and finally, (iii) To determine the competition resolution

77

required for modelling wood attributes of both spruce and aspen. First, my results indicate that
aspen is more sensitive to neighbourhood competition than spruce, which is similar to previous
findings. The results also suggest that defining the overall competitive environment is more
important for both spruce and aspen, rather than determining inter- vs intra-specific competition,
which is also consistent with the literature (Bérubé-Deschênes 2017). Aspen wood attribute
variation attributed to neighbourhood level competition was best described by the distance
dependent index MNE, while spruce was best described by the distance independent BAL index.
Finally, our results indicate that for aspen samples, most of the variation in wood attributes were
best explained by one of the two competition terms (neighbourhood or stand level) in the model,
while spruce sample variation was best described by age and stand level competition terms.
These preliminary models lay the foundation for further studies that can make recommendations
to forest managers based on desired end product.

4.2 Limitations and future work
In this section, I outline the limitations that apply to both of Chapter 2 and Chapter 3 as
well as provide suggestions for improvements and future work.
The first limitation is that the conclusions of this study are based off a single managed
trial site. Although the single sampling site offered a unique opportunity to compare various
levels of competition in analogous site conditions, my analysis was restricted to one site, so my
results may not accurately depict the effect competition has on wood attribute variation within
aspen and white spruce throughout the boreal mixedwood forest. Further, since the sample site is
a managed trial site that was maintained to target densities and not a natural stand, the within
wood trends observed in this study may not mimic those found in natural stands. Our study also

78

was limited in sample size due to budget restrictions associated with sending 120 samples for the
in-depth Silviscan analysis. Finally, although our methods were such that we sampled trees that
were greater than ten meters away from each other, aspen is a clonal species, and as such, the
sampling process may have captured samples that originate from a single individual. While this
is unlikely due to spatial distribution of samples, it would also likely not have a detrimental
impact to the study as each sampled tree was grown in a unique competition environment.
However, there could still be a small amount of variation attributed to the potential clonality of
aspen.
The previous limitations could all be improved by selecting even-aged stands of similar
site conditions and applying the same study framework. Multiple sites would increase the
predictive nature of the study as well as remove the potential clonality influence of aspen and the
results could then be applied to broader regions with confidence. Also, studying natural stands
would increase the applicability of the study to forest managers. However, finding natural evenaged stands with similar site conditions is both time consuming and difficult, and as such has
been a limitation of previous research on the topic as well.

79

5. Bibliography
Amarasekara, H. 2002. Effects of crown size on wood characteristics of Corsican pine in relation
to definitions of juvenile wood, crown formed wood and core wood. Forestry 75(1): 51–
61. doi:10.1093/forestry/75.1.51.
Antony, F., Schimleck, L.R., Jordan, L., Daniels, R.F., and Clark, A. 2012a. Modeling the effect
of initial planting density on within tree variation of stiffness in loblolly pine. Annals of
Forest Science 69(5): 641–650. doi:10.1007/s13595-011-0180-1.
Antony, F., Schimleck, L.R., Jordan, L., Daniels, R.F., and Clark, A. 2012b. Modeling the effect
of initial planting density on within tree variation of stiffness in loblolly pine. Annals of
Forest Science 69(5): 641–650. doi:10.1007/s13595-011-0180-1.
Aziz, T.A.T. and F.A. 2013. Cellulose Microfibril Angle in Wood and Its Dynamic Mechanical
Significance. Cellulose - Fundamental Aspects. doi:10.5772/51105.
Barnett, J.R., and Bonham, V.A. 2004. Cellulose microfibril angle in the cell wall of wood
fibres. Biological Reviews 79(2): 461–472. doi:10.1017/S1464793103006377.
Barton, K. 2020. Package “MuMIn.” Available from https://cran.rproject.org/web/packages/MuMIn/MuMIn.pdf.
Bates, D., Mächler, M., Bolker, B., and Walker, S. 2015. Fitting Linear Mixed-Effects Models
Using lme4. Journal of Statistical Software 67(1): 1–48. doi:10.18637/jss.v067.i01.
Bergeron, Y., Chen, H.Y.H., Kenkel, N.C., Leduc, A.L., and Macdonald, S.E. 2014. Boreal
mixedwood stand dynamics: ecological processes underlying multiple pathways. The
Forestry Chronicle 90(02): 202–213. doi:10.5558/tfc2014-039.
Bérubé-Deschênes, A. 2017. Quantifying competition in white spruce (Picea glauca) plantations.
Annals of Forest Science: 15.
Brandt, J.P., Flannigan, M.D., Maynard, D.G., Thompson, I.D., and Volney, W.J.A. 2013. An
introduction to Canada’s boreal zone: ecosystem processes, health, sustainability, and
environmental issues. Environmental Reviews 21(4): 207–226. doi:10.1139/er-20130040.
Brassard, B.W., and Chen, H.Y.H. 2006. Stand Structural Dynamics of North American Boreal
Forests. Critical Reviews in Plant Sciences 25(2): 115–137.
doi:10.1080/07352680500348857.
Brassard, B.W., Chen, H.Y.H., Wang, J.R., and Duinker, P.N. 2008. Effects of time since standreplacing fire and overstory composition on live-tree structural diversity in the boreal
forest of central Canada. Can. J. For. Res. 38(1): 52–62. doi:10.1139/X07-125.
Canadian Wood Fibre Centre, Canada, and Natural Resources Canada. 2010. The principal
attributes of Canadian wood fibre. Natural Resources Canada, Ottawa. Available from
http://publications.gc.ca/site/eng/381584/publication.html [accessed 2 April 2020].
Canham, C.D., LePage, P.T., and Coates, K.D. 2004. A neighborhood analysis of canopy tree
competition: effects of shading versus crowding. Canadian Journal of Forest Research
34(4): 778–787. Canadian Science Publishing NRC Research Press, Ottawa, Canada.
Canham, C.D., Papaik, M.J., Uriarte, M., McWilliams, W.H., Jenkins, J.C., and Twery, M.J.
2006. Neighborhood analyses of canopy tree competition along environmental gradients
in New England forests. Ecol Appl 16(2): 540–554. doi:10.1890/10510761(2006)016[0540:naoctc]2.0.co;2.
Carleton, T.J., and Maycock, P.F. 1981. Understorey–canopy affinities in boreal forest
vegetation. Canadian Journal of Botany 59(9): 1709–1716. doi:10.1139/b81-229.
80

Carr, S., Larocque, G.R., Luckai, N., and Bell, F.W. 2020. Effect of competition on individual
white spruce production in young boreal mixedwood forests. Can. J. For. Res. 50(8):
726–735. doi:10.1139/cjfr-2019-0395.
Carrillo, I., Aguayo, M.G., Valenzuela, S., Mendonça, R.T., and Elissetche, J.P. 2015. Variations
in wood anatomy and fiber biometry of Eucalyptus globulus geonotypes with different
wood density. Wood Research 60: 10.
Cave, I.D. 1966. Theory of X-ray measurement of microfibril angle in wood. Forest Products
Journal 16(10): 37–42.
Chen, H.Y.H., and Popadiouk, R.V. 2002. Dynamics of North American boreal mixedwoods.
Environmental Reviews; Ottawa 10(3): 137–166.
Chen, S., Shahi, C., Chen, H.Y.H., and McLaren, B. 2017. Economic analysis of forest
management alternatives: Compositional objectives, rotation ages, and harvest methods
in boreal forests. Forest Policy and Economics 85: 124–134.
doi:10.1016/j.forpol.2017.09.006.
Comeau, P.G. 2021. Effects of Thinning on Dynamics and Drought Resistance of Aspen-White
Spruce Mixtures: Results From Two Study Sites in Saskatchewan. Front. For. Glob.
Change 3: 621752. doi:10.3389/ffgc.2020.621752.
Comeau, P.G., Filipescu, C.N., Kabzems, R., and DeLong, C. 2004. Early growth of white
spruce underplanted beneath spaced and unspaced aspen stands in northeastern British
Columbia. Canadian Journal of Forest Research 34(11): 2277–2283. doi:10.1139/x04105.
Comeau, P.G., Kabzems, R., McClarnon, J., and Heineman, J.L. 2005. Implications of selected
approaches for regenerating and managing western boreal mixedwoods. The Forestry
Chronicle 81(4): 559–574. doi:10.5558/tfc81559-4.
Coopersmith, D., and Hall, E. 1999. Experimental Project 1077 – The Siphon Creek Mixedwood
Trial: BC Ministry of Forests: 10.
Cortini, F., Comeau, P.G., and Bokalo, M. 2012. Trembling aspen competition and climate
effects on white spruce growth in boreal mixtures of Western Canada. Forest Ecology
and Management 277: 67–73. doi:10.1016/j.foreco.2012.04.022.
De Araujo, F., Hart, J., and Mansfield, S. 2015. Variation in Trembling Aspen and White Spruce
Wood Quality Grown in Mixed and Single Species Stands in the Boreal Mixedwood
Forest. Forests 6(12): 1628–1648. doi:10.3390/f6051628.
DeLong, C.S., Banner, A., Mackenzie, W.H., Kaytor, B., and Rogers, B.J. 2011. A field guide to
ecosystem identification for the boreal white and black spruce zone of British Columbia.
Ministry of Forests and Range, Forest Science Program, Victoria, B.C.
Dhar, A., Parrott, L., and Hawkins, C. 2016. Aftermath of Mountain Pine Beetle Outbreak in
British Columbia: Stand Dynamics, Management Response and Ecosystem Resilience.
Forests 7(12): 171. doi:10.3390/f7080171.
Evans, R., and Ilic, J. 2001. Rapid prediction of wood stiffness from microfibril angle and
density. Forest Products Journal 51: 53–57.
Farmer, R.E. 1963. Effect of Light Intensity on Growth of Populus tremuloides Cuttings Under
Two Temperature Regimes. Ecology 44(2): 409–411. Ecological Society of America.
doi:10.2307/1932193.
Filipescu, C.N., and Comeau, P.G. 2007. Competitive interactions between aspen and white
spruce vary with stand age in boreal mixedwoods. Forest Ecology and Management
247(1–3): 175–184. doi:10.1016/j.foreco.2007.04.038.

81

Frey, B.R., Lieffers, V.J., Landhäusser, S.M., Comeau, P.G., and Greenway, K.J. 2011. An
analysis of sucker regeneration of trembling aspen. Canadian Journal of Forest Research.
NRC Research Press Ottawa, Canada. doi:10.1139/x03-053.
Fritts, H.C. 1976. Tree rings and climate. Academic Press, London ; New York.
Girardin, M.P., and Terrier, A. 2015. Mitigating risks of future wildfires by management of the
forest composition: an analysis of the offsetting potential through boreal Canada.
Climatic Change 130(4): 587–601. doi:10.1007/s10584-015-1373-7.
Groot, A. 1999. Effects of shelter and competition on the early growth of planted white spruce
(Picea glauca). Canadian Journal of Forest Research. NRC Research Press Ottawa,
Canada. doi:10.1139/x99-064.
Hacke, U.G., Sperry, J.S., Pockman, W.T., Davis, S.D., and McCulloh, K.A. 2001. Trends in
wood density and structure are linked to prevention of xylem implosion by negative
pressure. Oecologia 126(4): 457–461. doi:10.1007/s004420100628.
Halekoh, U., and Højsgaard, S. 2014. A Kenward-Roger Approximation and Parametric
Bootstrap Methods for Tests in Linear Mixed Models - The R Package pbkrtest. J. Stat.
Soft. 59(9). doi:10.18637/jss.v059.i09.
Hansen, W.D., Chapin, F.S., Naughton, H.T., Rupp, T.S., and Verbyla, D. 2016. Forestlandscape structure mediates effects of a spruce bark beetle (Dendroctonus rufipennis)
outbreak on subsequent likelihood of burning in Alaskan boreal forest. Forest Ecology
and Management 369: 38–46. doi:10.1016/j.foreco.2016.03.036.
Hector, A. 1998. The Effect of Diversity on Productivity: Detecting the Role of Species
Complementarity. Oikos 82(3): 597–599. [Nordic Society Oikos, Wiley].
doi:10.2307/3546380.
Hein, and Lima, J. 2012. Relationships between microfibril angle, modulus of elasticity and
compressive strength in Eucalyptus wood. Maderas. Ciencia y tecnología 14(3): 267–
274. doi:10.4067/S0718-221X2012005000002.
Herman, M., Dutilleul, P., and Avella-Shaw, T. 1999. Growth Rate Effects on Intra-Ring and
Inter-Ring Trajectories of Microfibril Angle in Norway Spruce (Picea Abies). IAWA J
20(1): 3–21. doi:10.1163/22941932-90001542.
Jiang, X., Huang, J.-G., Cheng, J., Dawson, A., Stadt, K.J., Comeau, P.G., and Chen, H.Y.H.
2018. Interspecific variation in growth responses to tree size, competition and climate of
western Canadian boreal mixed forests. Science of The Total Environment 631–632:
1070–1078. doi:10.1016/j.scitotenv.2018.03.099.
Jozsa, L.A., and Middleton, G.R. 1994. A discussion of wood quality attributes and their
practical implications. : 51.
Kabzems, R., Bokalo, M., Comeau, P., and MacIsaac, D. 2015. Managed Mixtures of Aspen and
White Spruce 21 to 25 Years after Establishment. Forests 7(12): 5.
doi:10.3390/f7010005.
Kabzems, R., Comeau, P.G., Filipescu, C.N., Rogers, B., and Linnell Nemec, A.F. 2016.
Creating boreal mixedwoods by planting spruce under aspen: successful establishment in
uncertain future climates. Canadian Journal of Forest Research 46(10): 1217–1223.
doi:10.1139/cjfr-2015-0440.
Kabzems, R., Nemec, A.L., and Farnden, C. 2007. Growing trembling aspen and white spruce
intimate mixtures. BC Journal of Ecosystems and Management 8(1): 15.

82

Kahriman, A., Şahin, A., Sönmez, T., and Yavuz, M. 2018. A novel approach to selecting a
competition index: the effect of competition on individual-tree diameter growth of
Calabrian pine. Can. J. For. Res. 48(10): 1217–1226. doi:10.1139/cjfr-2018-0092.
Kassambara, A. 2020. ggpubr: “ggplot2” Based Publication Ready Plots. Available from
https://CRAN.R-project.org/package=ggpubr [accessed 22 April 2021].
Kerhoulas, L.P., Kolb, T.E., and Koch, G.W. 2013. Tree size, stand density, and the source of
water used across seasons by ponderosa pine in northern Arizona. Forest Ecology and
Management: 10.
Kweon, D., and Comeau, P.G. 2019. Factors influencing overyielding in young boreal
mixedwood stands in western Canada. Forest Ecology and Management 432: 546–557.
doi:10.1016/j.foreco.2018.09.053.
Laganière, J., Cavard, X., Brassard, B.W., Paré, D., Bergeron, Y., and Chen, H.Y.H. 2015. The
influence of boreal tree species mixtures on ecosystem carbon storage and fluxes. Forest
Ecology and Management 354: 119–129. doi:10.1016/j.foreco.2015.06.029.
Lanner, R.M. 1985. On the insensitivity of height growth to spacing. doi:10.1016/03781127(85)90030-1.
Lenth, R.V. 2016. Least-Squares Means: The R Package lsmeans. Journal of Statistical Software
69(1): 1–33. doi:10.18637/jss.v069.i01.
Lieffers, V.J., Pinno, B.D., and Stadt, K.J. 2002. Light dynamics and free-to-grow standards in
aspen-dominated mixedwood forests. The Forestry Chronicle 78(1): 137–145.
doi:10.5558/tfc78137-1.
Lieffers, V.J., and Stadt, K.J. 1994. Growth of understory Piceaglauca, Calamagrostiscanadensis,
and Epilobiumangustifolium in relation to overstory light transmission. Canadian Journal
of Forest Research. NRC Research Press Ottawa, Canada. doi:10.1139/x94-157.
Lieffers, V.J., Stewart, J.D., Macmillan, R.B., Macpherson, D., and Branter, K. 1996. Seminatural and intensive silvicultural systems for the boreal mixedwood forest. The Forestry
Chronicle 72(3): 286–292. doi:10.5558/tfc72286-3.
Lord, T., and Green, A. 1986. Soils of the Fort St. John ~ Dawson Creek area, British Columbia.
Available from
http://www.env.gov.bc.ca/esd/distdata/ecosystems/Soils_Reports/bc42_report.pdf
[accessed 12 November 2018].
Lundgren, C. 2004. Cell wall thickness and tangential and radial cell diameter of fertilized and
irrigated Norway spruce. Silva Fennica 38(1). doi:10.14214/sf.438.
Macdonald, E., and Hubert, J. 2002. A review of the effects of silviculture on timber quality of
Sitka spruce. Forestry (Lond) 75(2): 107–138. Oxford Academic.
doi:10.1093/forestry/75.2.107.
Macdonald, E., Lecomte, Nicolas, Bergeron, Y., Brais, S., Chen, H., Comeau, P., Drapeau, P.,
Lieffers, V., Quideau, S., Spence, J., and Work, T. 2010. Ecological implications of
changing the composition of boreal mixedwood forests: a state of knowledge report.
Sustainable Forest Management Network, Edmonton, Alta. Available from
https://www.deslibris.ca/ID/226979 [accessed 24 February 2021].
Man, R., and Lieffers, V.J. 1999. Are mixtures of aspen and white spruce more productive than
single species stands? The Forestry Chronicle 75(3): 505–513. doi:10.5558/tfc75505-3.
Martin, G.L., and Ek, A.R. 1984. A Comparison of Competition Measures and Growth Models
for Predicting Plantation Red Pine Diameter and Height Growth. Forest Science 30(3):
731–743. doi:10.1093/forestscience/30.3.731.

83

Mbogga, M.S., Wang, X., and Hamann, A. 2010. Bioclimate envelope model predictions for
natural resource management: dealing with uncertainty. Journal of Applied Ecology
47(4): 731–740. doi:10.1111/j.1365-2664.2010.01830.x.
McCulloch, L., and Kabzems, R. 2009. British Columbia’s Northeastern Forests: Aspen
Complex Stand Establishment Decision Aid. Journal of Ecosystems and Management
10(2). Available from http://jem-online.org/index.php/jem/article/view/419 [accessed 18
October 2018].
Payne, N.J., Allan Cameron, D., Leblanc, J.-D., and Morrison, I.K. 2019. Carbon storage and net
primary productivity in Canadian boreal mixedwood stands. J. For. Res. 30(5): 1667–
1678. doi:10.1007/s11676-019-00886-0.
Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., Heisterkamp, S., Van Willigen, B., and Ranke, J.
2020. nlme: Linear and Nonlinear Mixed Effects Models. Available from
https://CRAN.R-project.org/package=nlme [accessed 7 January 2021].
Pretzsch, H. 2019. The Effect of Tree Crown Allometry on Community Dynamics in MixedSpecies Stands versus Monocultures. A Review and Perspectives for Modeling and
Silvicultural Regulation. Forests 10(9): 810. doi:10.3390/f10090810.
Pretzsch, H., and Biber, P. 2016. Tree species mixing can increase maximum stand density.
Canadian Journal of Forest Research 46(10): 1179–1194. NRC Research Press.
doi:10.1139/cjfr-2015-0413.
Pretzsch, H., Bielak, K., Block, J., Bruchwald, A., Dieler, J., Ehrhart, H.-P., Kohnle, U., Nagel,
J., Spellmann, H., Zasada, M., and Zingg, A. 2013. Productivity of mixed versus pure
stands of oak (Quercus petraea (Matt.) Liebl. and Quercus robur L.) and European beech
(Fagus sylvatica L.) along an ecological gradient. Eur J Forest Res 132(2): 263–280.
doi:10.1007/s10342-012-0673-y.
Pretzsch, H., and Rais, A. 2016. Wood quality in complex forests versus even-aged
monocultures: review and perspectives. Wood Sci Technol 50(4): 845–880.
doi:10.1007/s00226-016-0827-z.
Reyes-Hernández, V.J., and Comeau, P.G. 2015. The Influence of Stocking and Stand
Composition on Productivity of Boreal Trembling Aspen-White Spruce Stands. Forests
6(12): 4573–4587. Multidisciplinary Digital Publishing Institute. doi:10.3390/f6124387.
Rydval, M., Larsson, L.-Å., McGlynn, L., Gunnarson, B.E., Loader, N.J., Young, G.H.F., and
Wilson, R. 2014. Blue intensity for dendroclimatology: Should we have the blues?
Experiments from Scotland. Dendrochronologia 32(3): 191–204.
doi:10.1016/j.dendro.2014.04.003.
Sattler, D.F., Comeau, P.G., and Achim, A. 2014. Within-tree patterns of wood stiffness for
white spruce (Picea glauca) and trembling aspen (Populus tremuloides) (1). Canadian
Journal of Forest Research 44(2): 162-.
Seth, R.S., and Kingsland, M.A. 1990. The reinforcing properties of softwood kraft pulps. Pulp
&amp; Paper Canada 91(7): 68–75.
Smith, E.A., O’Loughlin, D., Buck, J.R., and St. Clair, S.B. 2011. The influences of conifer
succession, physiographic conditions and herbivory on quaking aspen regeneration after
fire. Forest Ecology and Management 262(3): 325–330.
doi:10.1016/j.foreco.2011.03.038.
Speer, J.H. 2012. Fundamentals of Tree-Ring Research. University of Arizona Press: 522.

84

Sun, S., Cao, Q.V., and Cao, T. 2019. Evaluation of distance-independent competition indices in
predicting tree survival and diameter growth. Canadian Journal of Forest Research 49(5):
440–447. NRC Research Press. doi:10.1139/cjfr-2018-0344.
Vysotskaya, L.G., and Vaganov, E.A. 1989. Components of the Variability of Radial Cell Size in
Tree Rings of Conifers. IAWA Journal 10(4): 417–426. Brill. doi:10.1163/2294193290001134.
Watson, P., and Bradley, M. 2009. Canadian pulp fibre morphology: Superiority and
considerations for end use potential. The Forestry Chronicle 85(3): 401–408.
doi:10.5558/tfc85401-3.
Wilpert, K. von. 1991, April 27. Intraannual variation of radial tracheid diameters as monitor of
site specific water stress. Available from
https://eurekamag.com/research/002/416/002416079.php [accessed 24 March 2020].
Wykoff, W., Crookston, N.L., Stage, A.R., and Intermountain Forest and Range Experiment
Station (Ogden, U. 1982. User’s guide to the stand prognosis model. U.S. Dept. of
Agriculture, Forest Service, Intermountain Forest and Range Experiment Station, Ogden,
Utah : Available from https://www.biodiversitylibrary.org/item/191561.
Zeng, Q., Rossi, S., and Yang, B. 2018. Effects of Age and Size on Xylem Phenology in Two
Conifers of Northwestern China. Front. Plant Sci. 8. Frontiers.
doi:10.3389/fpls.2017.02264.
Zobel, B.J., and Jett, J.B. 1995. The Importance of Wood Density (Specific Gravity) and Its
Component Parts. In Genetics of Wood Production. Edited by B.J. Zobel and J.B. Jett.
Springer Berlin Heidelberg, Berlin, Heidelberg. pp. 78–97. doi:10.1007/978-3-64279514-5_4.

85

6. Appendix
Table 6.1. Full results of the MuMIn analysis that forms the basis for table 2.2. Multi model
inference analysis of the global model (linear model = Ring + species + treatment + |tree) where
Loglik is the log likelihood, AIC is Akaike’s information criterion, and delta is the difference in
AIC for best performing model and next best.
Wood
(Intercept)
Ring species treatment df
logLik
AICc
delta
attribute
Ring
Width

Density

weight

3.77

-0.058 NA

+

8

-4169.122

8354.3

0.00

7.26E-01

3.75

-0.058 +

+

9

-4169.091

8356.2

1.95

2.74E-01

2.83

-0.058 +

NA

5

-4204.512

8419.0

64.75

6.31E-15

3.00

-0.058 NA

NA

4

-4208.129

8424.3

69.98

4.63E-16

2.87

NA +

+

8

-4542.668

9101.4

747.09

4.29E163

3.04

NA NA

+

7

-4544.708

9103.5

749.16

1.52E163

1.95

NA +

NA

4

-4576.558

9161.1

806.84

4.55E176

2.20

NA NA

NA

3

-4584.712

9175.4

821.14

3.57E179

393.11

NA +

+

8

-15803.8 31623.7

0.00

4.52E-01

396.28

NA +

NA

4

-15808.32 31624.7

1.00

2.73E-01

392.80

0.020 +

+

9

-15803.77 31625.6

1.94

1.71E-01

395.99

0.019 +

NA

5

-15808.29 31626.6

2.95

1.03E-01

367.72

NA NA

+

7

-15816.98 31648.0

24.34

2.34E-06

367.28

0.035 NA

+

8

-15816.88 31649.8

26.15

9.47E-07

381.02

NA NA

NA

3

-15823.37 31652.8

29.10

2.17E-07

86

(Intercept)
Density

MFA

MOE

Ring species treatment df

logLik

AICc

delta

weight

380.54

0.035 NA

NA

4

-15823.27 31654.6

30.90

8.82E-08

23.15

-0.654 +

+

9

-8835.623 17689.3

0.00

8.99E-01

20.77

-0.654 +

NA

5

-8841.831 17693.7

4.38

1.01E-01

27.00

-0.655 NA

+

8

-8851.43 17718.9

29.60

3.35E-07

23.41

-0.656 NA

NA

4

-8863.718 17735.5

46.14

8.58E-11

13.26

NA +

+

8

20945.9 3256.61 0.00E+00
10464.935

10.81

NA +

NA

4

18.80

NA NA

+

14.43

NA NA

NA

3

21028.4 3339.06 0.00E+00
10511.176

20951.8 3262.45 0.00E+00
10471.872
7
21007.7 3318.42 0.00E+00
10496.844

8.14

0.296 +

+

9

-5388.845 10795.8

0.00

9.35E-01

9.45

0.296 +

NA

5

-5395.529 10801.1

5.33

6.51E-02

5.67

0.296 NA

+

8

-5416.273 10848.6

52.84

3.13E-12

7.79

0.296 NA

NA

4

-5430.869 10869.8

74.00

7.96E-17

12.61

NA +

+

8

-7696.901 15409.9 4614.10 0.00E+00

13.95

NA +

NA

4

-7703.76 15415.5 4619.78 0.00E+00

9.37

NA NA

+

7

-7740.604 15495.2 4699.50 0.00E+00

11.84

NA NA

NA

3

-7756.135 15518.3 4722.53 0.00E+00

87

Wood
attribute
Coarseness

Tangential
Diameter

Radial
Diameter

(Intercept)

Ring species treatment df

logLik

AICc

delta

weight

542.70

2.339 +

+

9

-18223.71 36465.5

0.00

9.93E-01

506.81

2.336 +

NA

5

-18232.75 36475.5

10.05

6.52E-03

578.11

NA +

+

8

-18318.71 36653.5

188.00

1.49E-41

542.37

NA +

NA

4

-18327.53 36663.1

197.59

1.23E-43

301.70

2.370 NA

+

8

-18323.99 36664.0

198.56

7.60E-44

372.13

2.370 NA

NA

4

-18328.99 36666.0

200.52

2.84E-44

331.38

NA NA

+

7

-18421.52 36857.1

391.60

9.19E-86

404.57

NA NA

NA

3

-18426.56 36859.1

393.66

3.27E-86

40.53

0.052 +

+

9

-7820.458 15659.0

0.00

9.90E-01

39.29

0.052 +

NA

5

-7829.074 15668.2

9.19

9.99E-03

41.32

NA +

+

8

-7854.339 15724.7

65.75

5.22E-15

40.08

NA +

NA

4

-7862.818 15733.7

74.68

6.03E-17

26.73

0.053 NA

+

8

-7985.163 15986.4

327.40

7.98E-72

31.45

0.053 NA

NA

4

-7991.461 15990.9

331.96

8.15E-73

27.39

NA NA

+

7

-8020.34 16054.7

395.74

1.15E-86

32.17

NA NA

NA

3

-8026.665 16059.3

400.37

1.14E-87

+

9

-7283.053 14584.2

0.00

9.99E-01

28.59

0.150 +

88

Wood
attribute

(Intercept)

Radial
Diameter

27.27

0.150 NA

+

26.28

0.150 +

25.90

Specific
Surface

Ring Area

Ring species treatment df

logLik

AICc

delta

weight

8

-7290.684 14597.4

13.25

1.33E-03

NA

5

-7296.345 14602.7

18.54

9.38E-05

0.150 NA

NA

4

-7298.178 14604.4

20.20

4.09E-05

30.85

NA +

+

8

-7640.389 15296.8

712.66

1.77E155

28.55

NA +

NA

4

-7653.621 15315.3

731.09

1.76E159

29.15

NA NA

+

7

-7652.896 15319.8

735.66

1.79E160

27.95

NA NA

NA

3

-7658.331 15322.7

738.51

4.31E161

263.46

-1.100 +

+

9

-15900.29 31818.7

0.00

9.95E-01

283.86

-1.099 +

NA

5

-15909.59 31829.2

10.56

5.06E-03

246.80

NA +

+

8

-15991.13 31998.3

179.65

9.69E-40

267.14

NA +

NA

4

-16000.2 32008.4

189.77

6.17E-42

343.51

-1.116 NA

NA

4

-16005.9 32019.8

201.17

2.06E-44

371.47
328.24

-1.115 NA
NA NA

+
NA

8
3

-16002.03 32020.1
-16099.32 32204.7

201.45
386.00

1.79E-44
1.51E-84

357.51

NA NA

+

7

-16095.37 32204.8

386.13

1.41E-84

491.55 17.855 NA

+

8

-21516.34 43048.7

0.00

6.52E-01

460.56 17.869 +

+

9

-21515.96 43050.0

1.25

3.48E-01

98.51 17.880 +

NA

5

-21547.01 43104.0

55.32

6.32E-13

177.16 17.841 NA

NA

4

-21551.84 43111.7

62.97

1.38E-14

89

Wood
attribute
Ring area

Cell
Population

Wall
Thickness

(Intercept)

Ring species treatment df

logLik

AICc

delta

weight

715.12

NA NA

+

7

-22107.41 44228.9 1180.15

3.54E257

730.88

NA +

+

8

-22107.31 44230.7 1181.95

1.43E257

370.68

NA +

NA

4

-22139.14 44286.3 1237.58

1.19E269

421.30

NA NA

NA

3

-22141.25 44288.5 1239.79

3.96E270

+

+

9

-20622.59 41263.2

0.00 1.00E+00

+

NA

5

-20639.18 41288.4

25.15

3.46E-06

NA

NA

4

-20742.49 41493.0

229.75

1.29E-50

NA

+

8

-20740.99 41498.0

234.80

1.03E-51

10.202
1068.51
10.203
1363.40
10.284
1457.18
10.281
905.31

750.85

NA +

+

8

-20982.28 41980.6

717.37

1.68E156

913.18

NA +

NA

4

-20998.87 42005.8

742.52

5.80E162

1222.60

NA NA

NA

3

-21108.26 42222.5

959.30

4.91E209

1328.40

NA NA

+

7

-21106.55 42227.1

963.89

4.94E210

2.82

0.008 +

+

9

-907.3739

1832.8

0.00

9.70E-01

2.72
2.94
2.84
2.05

0.008
NA
NA
0.008

+
+
+
NA

NA
+
NA
+

5
8
4
8

-914.874
-975.1697
-982.4607
-985.5927

1839.8
1966.4
1972.9
1987.2

6.96
133.58
140.13
154.43

2.99E-02
9.56E-30
3.62E-31
2.84E-34

2.28

0.008 NA

NA

4

-990.9849

1990.0

157.18

7.18E-35

2.15

NA NA

+

7

1055.7946

2125.6

292.82

2.52E-64

2.40

NA NA

NA

3

1061.2184

2128.4

295.64

6.16E-65

90

Table 6.2 The results of the analysis of variance of the effects of treatment (aspen population density), cambial age
and species on each wood attribute. This test was conducted to corroborate results from the MuMIn anlaysis
(Figure 6.1) and justify the use of p-values for hypothesis testing.

Wood
Property
Ring
Width

Cambial Age (Ring)
F(1, 3043) = 861.13,
p < 0.001
F(1, 3043) = .288,
Density
p = 0.59
Microfibril F(1, 3043) = 5899,
Angle
p < 0.001
Modulus
of
F(1, 3043) = 10896,
Elasticity
p < 0.001
F(1, 3043) = 226.07,
Coarseness p < 0.001
Tangential F(1, 3043) = 112.97,
Diameter
p < 0.001
Radial
F(1, 3043) = 811.08,
Diameter
p < 0.001
Specific
F(1, 3043) = 219.69,
Surface
p < 0.001
F(1, 3043) =
Ring Area 1434.54, p < 0.001
Cell
F(1, 3043) = 891.21,
Population p < 0.001
Wall
F(1, 3043) = 159.32,
Thickness p < 0.001

treatment

species

F(4, 114) = 27.31, p < 0.001

F(1, 114) = 0.062, p = 0.802

F(4, 114) = 4.18, p = 0.003

F(1, 114) = 29.2, p < 0.001

F(4, 114) = 8.87, p < 0.001

F(1, 114) = 36.13, p < 0.001

F(4, 114) = 13.03, p < 0.001

F(1, 114) = 69.42, p < 0.001

F(4, 114) = 13.88, p < 0.001

F(1, 114) = 514.82, p < 0.001

F(4, 114) = 51.68, p < 0.001

F(1, 114) = 1731, p < 0.001

F(4, 114) = 4.52, p = 0.002

F(1, 114) = 16.24, p < 0.001

F(4, 114) = 10.95, p < 0.001

F(1, 114) = 531.97, p < 0.001

F(4, 114) = 24.29, p < 0.001

F(1, 114) = 0.76, p = 0.39

F(4, 114) = 5.55, p < 0.001

F(1, 114) = 742.48, p < 0.001

F(4, 114) = 10.38, p < 0.001

F(1, 114) = 319.99, p < 0.001

91

Table 6.3 Aspen results table of complete post-hoc test of species separated linear mixed model results (summarized
in Figure 2.3). Lsmeans pairwise comparison to establish contrasts, p-value adjusted by Bonferroni method.

Wood Property
Ring Width

contrast
1000 2000
1000 5000
1000 10000
2000 5000
2000 10000

estimate

SE

df

t.ratio

p.value

0.078

0.274

46

0.2845

0.7773

0.2788

0.2171

46

1.2841

0.2055

0.5488

0.2172

46

2.5266

0.015

0.2008

0.2163

46

0.9286

0.3579

0.4708

0.2164

46

2.1759

0.0347

0.27

0.1374

46

1.965

0.0555

37.4002

16.649

46

2.2464

0.0295

19.7303

13.1705

46

1.4981

0.1409

-0.0497

13.1719

46

-0.0038

0.997

2000 5000

-17.6699

13.1564

46

-1.3431

0.1858

2000 10000

-37.4499

13.1581

46

-2.8462

0.0066

5000 10000

-19.78

8.3309

46

-2.3743

0.0218

5000 10000
Density

1000 2000
1000 5000
1000 10000

MFA

No significant effect of treatment

MOE

No significant effect of treatment

Coarseness

1000 2000
1000 5000
1000 10000
2000 5000

29.609

43.7801

46

0.6763

0.5022

-19.2998

34.6432

46

-0.5571

0.5802

-89.1754

34.6489

46

-2.5737

0.0133

-48.9089

34.589

46

-1.414

0.1641

92

Wood Property

Tangential Diameter

Radial Diameter

Specific Surface Area

contrast

estimate

SE

df

t.ratio

p.value

2000 10000

118.7844

34.5955

46

-3.4335

0.0013

5000 10000

-69.8755

21.915

46

-3.1885

0.0026

0.4259

1.3173

46

0.3233

0.7479

-1.3026

1.0423

46

-1.2498

0.2177

-2.9991

1.0425

46

-2.877

0.0061

-1.7285

1.0408

46

-1.6608

0.1036

-3.425

1.041

46

-3.2902

0.0019

-1.6965

0.6593

46

-2.573

0.0134

-1.4275

0.9353

46

-1.5263

0.1338

-1.3656

0.7403

46

-1.8446

0.0715

-2.3451

0.7405

46

-3.1669

0.0027

2000 5000

0.0619

0.7388

46

0.0838

0.9336

2000 10000

-0.9176

0.7389

46

-1.2417

0.2206

5000 10000

-0.9795

0.4684

46

-2.0913

0.0421

-23.3402

14.5948

46

-1.5992

0.1166

-7.2824

11.5478

46

-0.6306

0.5314

18.6077

11.5494

46

1.6111

0.114

16.0579

11.5316

46

1.3925

0.1705

1000 2000
1000 5000
1000 10000
2000 5000
2000 10000
5000 10000
1000 2000
1000 5000
1000 10000

1000 2000
1000 5000
1000 10000
2000 5000

93

Wood Property

Ring Area

Cell Population

Wall Thickness

contrast
2000 10000
5000 10000
1000 2000
1000 5000
1000 10000
2000 5000
2000 10000
5000 10000
1000 2000
1000 5000
1000 10000
2000 5000
2000 10000
5000 10000
1000 2000
1000 5000
1000 10000
2000 5000
2000 10000

estimate

SE

df

t.ratio

p.value

41.9479

11.5335

46

3.637

0.0007

25.89

7.3048

46

3.5442

0.0009

23.9787 115.4942

46

0.2076

0.8364

115.3324

91.3691

46

1.2623

0.2132

217.5375

91.3803

46

2.3806

0.0215

91.3537

91.2624

46

1.001

0.3221

193.5588

91.2752

46

2.1206

0.0394

102.2051

57.7961

46

1.7684

0.0836

34.0495

54.8124

46

0.6212

0.5375

62.0499

43.3708

46

1.4307

0.1593

130.7563

43.3775

46

3.0144

0.0042

28.0004

43.3068

46

0.6466

0.5211

96.7068

43.3144

46

2.2327

0.0305

68.7064

27.4356

46

2.5043

0.0159

0.2374

0.1708

46

1.3903

0.1711

0.0415

0.1351

46

0.3074

0.7599

-0.2315

0.1351

46

-1.7131

0.0934

-0.1959

0.1349

46

-1.4518

0.1534

-0.4689

0.1349

46

-3.4748

0.0011

94

Wood Property

contrast

estimate

SE

df

t.ratio

p.value

5000 10000

-0.273

0.0855

46

-3.1944

0.0025

Table 6.4 Spruce results table of complete post-hoc test of species separated linear mixed model results
(summarized in Figure 2.3). Lsmeans pairwise comparison to establish contrasts, p-value adjusted by Bonferroni
method.

Wood Property
Ring Width

Density

contrast

estimate

SE

df

t.ratio

p.value

0 - 1000
0 - 2000

0.2499
0.6628

0.1967
0.1975

65
65

1.2707
3.3569

0.2084
0.0013

0 - 5000

0.9481

0.1246

65

7.6098

< 0.001

0 - 10000

1.1053

0.1247

65

8.8669

< 0.001

1000 - 2000

0.4129

0.2492

65

1.6568

0.1024

1000 - 5000

0.6982

0.1966

65

3.5511

0.0007

1000 10000

0.8554

0.1967

65

4.3497

< 0.001

2000 - 5000

0.2853

0.1974

65

1.445

0.1533

2000 10000

0.4425

0.1975

65

2.2409

0.0285

5000 10000

0.1572

0.1246

65

1.2618

0.2115

0 - 1000

9.7875

9.4269

65

1.0382

0.303

0 - 2000

11.8164

9.4583

65

1.2493

0.216

0 - 5000

-11.8251

5.9704

65

-1.9806

0.0519

0 - 10000
1000 - 2000

-1.7975
2.0289

5.973
11.9437

65
65

-0.3009
0.1699

0.7644
0.8656

1000 - 5000

-21.6126

9.4252

65

-2.2931

0.0251

1000 10000

-11.585

9.4269

65

-1.2289

0.2235

2000 - 5000

-23.6415

9.4567

65

-2.5

0.015
95

Wood Property

contrast

estimate

SE

df

t.ratio

p.value

2000 10000

-13.6139

9.4583

65

-1.4394

0.1548

5000 10000

10.0277

5.9703

65

1.6796

0.0978

No significant effect of treatment

MFA
MOE

Coarseness

0 - 1000

-0.1738

0.7592

65

-0.2289

0.8197

0 - 2000

-1.2728

0.7596

65

-1.6756

0.0986

0 - 5000

-1.4354

0.4803

65

-2.9887

0.004

0 - 10000

-1.4623

0.4803

65

-3.0447

0.0034

1000 - 2000

-1.099

0.9606

65

-1.1441

0.2568

1000 - 5000

-1.2616

0.7592

65

-1.6617

0.1014

1000 10000

-1.2886

0.7592

65

-1.6972

0.0944

2000 - 5000

-0.1626

0.7595

65

-0.2141

0.8312

2000 10000

-0.1896

0.7596

65

-0.2496

0.8037

5000 10000

-0.027

0.4803

65

-0.0562

0.9554

0 - 1000

25.5604

11.3826

65

2.2456

0.0281

0 - 2000
0 - 5000

43.1356
28.6305

11.4279
7.211

65
65

3.7746
3.9704

< 0.001
< 0.001

0 - 10000

44.2045

7.2147

65

6.127

< 0.001

1000 - 2000

17.5752

14.426

65

1.2183

0.2275

1000 - 5000

3.07

11.38

65

0.2698

0.7882

18.6441

11.3825

65

1.638

0.1063

-14.5052

11.4256

65

-1.2695

0.2088

1.0689

11.4279

65

0.0935

0.9258

1000 10000
2000 - 5000
2000 10000

96

Wood Property

Tangential
Diameter

Radial Diameter

Specific Surface
Area

contrast

estimate

SE

df

t.ratio

p.value

5000 10000

15.574

7.2108

65

2.1598

0.0345

0 - 1000

0.1889

0.4386

65

0.4307

0.6681

0 - 2000

1.2584

0.4425

65

2.8438

0.006

0 - 5000

1.3478

0.2785

65

4.84

< 0.001

0 - 10000
1000 - 2000

1.4168
1.0695

0.2788
0.5572

65
65

5.0822
1.9194

< 0.001
0.0593

1000 - 5000

1.1588

0.4384

65

2.6434

0.0103

1000 10000

1.2279

0.4386

65

2.7996

0.0067

2000 - 5000

0.0894

0.4423

65

0.202

0.8405

0.1584

0.4425

65

0.358

0.7215

0.0691

0.2784

65

0.2481

0.8049

1.3694

0.7861

65

1.742

0.0862

0 - 2000

1.6556

0.7868

65

2.1041

0.0392

0 - 5000

2.2055

0.4974

65

4.4343

< 0.001

0 - 10000

2.8468

0.4974

65

5.7229

< 0.001

1000 - 2000
1000 - 5000
1000 10000
2000 - 5000
2000 10000
5000 10000

0.2863
0.8361

0.9948
0.7861

65
65

0.2878
1.0637

0.7744
0.2914

1.4774

0.7861

65

1.8794

0.0647

0.5499

0.7868

65

0.6989

0.4871

1.1911

0.7868

65

1.5138

0.1349

0.6412

0.4974

65

1.2893

0.2019

0 - 1000

-23.4501

10.2854

65

-2.2799

0.0259

0 - 2000
0 - 5000

-33.5114
-10.694

10.3201
6.5142

65
65

-3.2472
-1.6416

0.0018
0.1055

2000 10000
5000 10000
0 - 1000

97

Wood Property

Ring Area

Cell Population

contrast
0 - 10000
1000 - 2000
1000 - 5000
1000 10000
2000 - 5000
2000 10000
5000 10000
0 - 1000
0 - 2000
0 - 5000
0 - 10000
1000 - 2000
1000 - 5000
1000 10000
2000 - 5000
2000 10000
5000 10000
0 - 1000

estimate
-26.1929
-10.0614
12.7561

SE
6.5171
13.0316
10.2834

df
65
65
65

t.ratio
-4.0191
-0.7721
1.2405

p.value
0.0002
0.4429
0.2193

-2.7429

10.2853

65

-0.2667

0.7906

22.8174

10.3183

65

2.2114

0.0305

7.3185

10.3201

65

0.7092

0.4808

-15.499

6.5141

65

-2.3793

0.0203

112.182 84.8913
282.1284 84.9941
382.9638 53.7168
424.3377 53.7254
169.9464 107.4439
270.7818 84.8856

65
65
65
65
65
65

1.3215
3.3194
7.1293
7.8983
1.5817
3.19

0.191
0.0015
< 0.001
< 0.001
0.1186
0.0022

312.1557

84.8913

65

3.6771

0.0005

100.8355

84.9889

65

1.1865

0.2398

142.2093

84.9941

65

1.6732

0.0991

41.3739

53.7168

65

0.7702

0.444

-62.6329
0 - 2000
133.2715
0 - 5000
165.5025
0 - 10000
195.9866
1000 - 2000 -70.6385
1000 - 5000
102.8695
1000 10000
133.3537
2000 - 5000
-32.231
2000 -62.7152
10000

49.7296

65

-1.2595

0.2124

49.8366

65

-2.6742

0.0095

31.4798

65

-5.2574

< 0.001

31.4887

65

-6.224

< 0.001

62.97

65

-1.1218

0.2661

49.7236

65

-2.0688

0.0425

49.7295

65

-2.6816

0.0093

49.8311

65

-0.6468

0.52

49.8365

65

-1.2584

0.2127

31.4797

65

-0.9684

0.3364

5000 10000

-30.4842

98

Wood Property
Wall Thickness

contrast
0 - 1000
0 - 2000
0 - 5000
0 - 10000
1000 - 2000
1000 - 5000
1000 10000
2000 - 5000
2000 10000
5000 10000

estimate
0.1152
0.1805
0.0476
0.135
0.0653
-0.0676

SE
0.0611
0.0614
0.0387
0.0388
0.0775
0.0611

df
65
65
65
65
65
65

t.ratio
1.8845
2.9405
1.2281
3.4837
0.8429
-1.1067

p.value
0.064
0.0045
0.2238
0.0009
0.4024
0.2725

0.0198

0.0611

65

0.3239

0.7471

-0.133

0.0614

65

-2.1661

0.034

-0.0455

0.0614

65

-0.7414

0.4611

0.0874

0.0387

65

2.2575

0.0273

99