DECOMPOSITION AND CARBON LOSS IN LODGEPOLE PINE
(PINUS CONTORTA VAR. LAT/FOL/A) WOOD FOLLOWING ATTACK BY
MOUNTAIN PINE BEETLE (DENDROCTONUS PONDEROSAE)
by

Benita Kaytor
B.Sc., University of Northern British Columbia, 2003

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

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
March 2016
© Benita Kaytor, 2016

ABSTRACT

Mountain pine beetle (MPB)-killed wood remaining on the landscape is predicted to release
significant amounts of carbon to the atmosphere as it decays. A lack of field-based wood
decomposition data for validating simulation models reduces certainty in such predictions.
Using a chronosequence approach, I quantified decomposition of MPB-killed wood to
improve decay rate parameters. Changes in carbon density over time and climatic variability
showed distinct patterns for bole position categories. Snag carbon density was similar to that
of live lodgepole pine, and did not change considerably with time or climatic influences.
Decay in suspended boles increased with summer precipitation, but declined with increasing
summer temperature, suggesting decay in elevated boles is moisture-limited on warmer sites.
Down boles decayed four times faster with increasing proximity to the soil than suspended
boles, but did not clearly reflect climatic influences. Position of dead boles appears more
important for wood decay than previously thought.

ii

Figure B 1. Field data sheet. ..... ................... .. .. ............ ................... ............. ..... .. ... ...................... ....... . I 00
Figure C 1. Distribution of carbon density in wood samples clasified by categories of live
lodgepole pine, and MPS-killed lodgepole pine snags (Stand), suspended (Susp), and down
CWD ........... .. .. .... .. ......... ......... ... ........................... .......... .................. ...... .......... ......... .... ... .... .... ... ........ 103
Figure DI . Carbon density (g C cm-3) of snags (blue triangles) and live lodgepole pine (green
circles) for reference, versus predictor variables as follows: a) years dead and b) BEC subzones.
Data labels indicate the years dead for each snag. Continued on next page .... ......... ............. .. ....... .... I 04
Figure D2 . Carbon density (g C cm-3) of suspended CWD (red diamonds) and live lodgepole
pine (green circles) for reference, versus predictor variables as follows: a) years dead, b) years
down, c) sample height off the ground, d) BEC subzones, e) MAP, and f) MAT. Data labels
indicate the years dead for the bole. Continued on next page ......... .... .... .. .... ......... ........ ....... ..... ........ 106
Figure D3 . Carbon density (g C cm-3) of down CWD (purple squares) and live lodgepole pine
(green circles) for reference, versus predictor variables as follows: a) years dead, b) years down,
c) sample height off the ground, d) BEC subzones, e) MAP, and f) MAT. Data labels indicate
the years dead for the bole. Continued on next page ........... .. ...... ..... ........................ ......... .... .. ........... I 08
Figure El . Median and quartiles of the years between tree death and tree fall for suspended
(Susp) and Down (in contact with the ground) CWD .... ...... ...... ..... ............... .. .. ......... .. .. ... ...... ... .. ...... 110
Figure E2. Mean within-bole heterogeneity of specific wood density (g cm- 3) with maximum
(+) and minimum(-) for each suspended and down bole. Down boles exhibited the greatest
variability as demonstrated by the two lowest wood density observations near 60 years dead,
which are from boles that also have maximum wood density observations within the range of
live trees. Live trees and snags are included for reference only ..... .... .. ........ .. .................. ...... ........... . 111

VII

ACKNOWLEDGEMENTS

Thanks first and foremost to my supervisors, Drs. Phil Burton and Art Fredeen for your guidance,
patience and support. This project promised to be challenging from the start, and indeed it was. You
warned me, but supported my decision, and helped push me through when 1 wasn' t sure it could be
done. Thanks to Craig DeLong for taking the role as committee member, and to Dr. Dale Seip for
being my external examiner. Other individuals for whose help I am sincerely grateful include: Doug
Thompson, Ping Bai, John Pousette, David Dunn, and Drs. Kathy Lewis, Ian Hartley, Carolyn Smyth,
Werner Kurz, Tony Trofymow, Chris Johnson, Stefanie LaZerte.
At no fault of their own, I had a record number of field assistants pass through my company during
field work. Coralie Lenne, an intern of forestry studies from France, worked with me far more than
her fair share. We improved each other' s proficiency in French (me) and English (Coralie) while
cutting up logs in the Chilcotin, and suffered the record-bad season of bugs together. Thank you for
your willing dedication to my project Coralie, but even more so for your friendship. Craig and Mike
DeLong, Bruce Rogers, and Hardy Griesbauer all helped during a joint-effort multi-tasking field trip.
That was an all-star team and an efficient machine, thank you guys! Kallie Smith, Laura Kennedy,
Chris Konchalski, Kimberly Kaytor, and Alyson Watt also helped at different times. Alyson, thank
you so much for helping me get those last few sites sampled, and my sincerest apologies for putting
your dogs ' lives in danger on more than one occasion!
Chief Zach Parker and the Council of the Ulkatcho First Nation not only permitted me to collect
samples on their territory in the Chilcotin, but insisted on providing a place for us to stay while we
worked there. I am still moved by your gracious hospitality. Thank you. I look forward to sharing
the data and results with you in person, and hope that this work can contribute to the well-being of
your forests and community.
Thanks to Kate Hrinkevich, Laura Kennedy, Mette Fillll0en and Barry Hagen, who helped with the
dendrochronology, and to Lynda Kaytor and John Hagen who helped me measure wood densities.
My partner, John, was the rock I needed to persevere through what became an excessive period of
time as a grad student. I am so very grateful for the support you gave in a multitude of ways - thank
you, John! My family contributed in the roles of field assistant, lab assistant, and GIS specialist, and
perpetually cheered me on - that meant so much to me, thank you! The unwavering support and
encouragement of my friends and family were huge in the final phases of this thesis. To my dear
fellow grad students who shared this process with me - I am so thankful for our friendships . The
highlights of graduate studies would not exist without comradery, and for this, I thank you.
The work could not have been done without the financial support of the Natural Sciences Engineering
Research Council of Canada (NSERC), Mitacs, the Canadian Forest Service (CFS), and the
University of Northern British Columbia. Mitacs facilitated the collaboration with CFS, and it is
through this partnership that my results will contribute to the process of calculating Canada' s forest
carbon fluxes.
Vlll

INTRODUCTION

Context

Between the 1990s and early 2010s, lodgepole pine (Pinus contorta var. latifolia) forests in
British Columbia (BC) succumbed to a mountain pine beetle (Dendroctonus ponderosae) (MPB)
attack exceeding the magnitude of any previously documented outbreak 1 (Wilson and Tkacz
1996, Taylor et al. 2006, Wagner et al. 2006). In 2013 , the BC provincial government's estimate
of the affected outbreak area was 18.3 million ha (MFLNRO 2013), approximately equivalent to
one fifth of BC's land base. The extent ofMPB-induced tree mortality spurred forest research
activities on an array of topics, such as MPB ecology (Carroll and Safranyik 2003), climate
change and MPB range expansion (Cudmore et al. 2010), interactions of wildfire and MPB
disturbance (Hawkes et al. 2003 , Axelson et al. 2009), economic impacts of MPB (Pousette and
Hawkins 2006), fall rates of MPB-killed snags (Lewis and Thompson 2009), and shelf life of
MPB wood (Woodward 2010), for example. In spite of widespread efforts to salvage MPBkilled trees and minimize wood quality losses, much of this wood will remain on the landscape
to degrade by natural processes (e.g. , fungal decomposition, fire) (Pedersen 2003, Wagner et al.
2006, Cornwell et al. 2009).
Broadly speaking, forests are considered a major sink for atmospheric carbon dioxide (CO 2) at
the global scale because of their capacity to sequester carbon via photosynthesis into long-lived

'Based on a maximum observation period of approximately 160 years, this is the most extensive MPB outbreak on
record. Field and aerial surveys for MPB attacks date back to the early 1900s in BC; dendrochronological crossdating records extend back no further than the 1870s in BC (Heath and Alfaro 1990, Taylor et al. , 2006), and the
early 1800s in the US (Thorne 1935, Cole and Amman 1980, Mitchell et al. 1983, Wilson and Tkacz 1996). No
evidence of documented First Nations ' knowledge of MPB outbreak history was found .

I

wood biomass. As such, living forests play a major role in regulating climate conditions by
offsetting the "greenhouse" effects resulting from rising atmospheric CO2 (Grace 2005 , IPCC
2007). The capacity of forests to fulfil this buffering role, however, is responsive to climate.
Due to rising concentrations of anthropogenic CO 2 and other greenhouse gases [GHGs] in the
troposphere, more solar energy is being trapped and reradiated back to Earth, contributing to
warmer surface temperatures (IPCC 2013). Under warmer surface conditions, many forest
disturbance agents including insect outbreaks and fire may occur with greater frequency,
severity, and/or distribution (Carroll et al. 2006, Haughian et al. 2012). Forest disturbances
result in a greater abundance of dead organic matter on the landscape, and a subsequent release
of CO 2 as this material decays. Therefore, forests may actually exacerbate global GHG
emissions by succumbing to large-scale disturbances, resulting in a surplus of decomposing
wood (Volney and Fleming 2000). In this context, large MPB outbreaks have the potential to
surpass local, regional, and even provincial scale impacts by rendering Canada' s forests a net
carbon source, rather than a sink (Kurz et al. 2008, Hicke et al. 2012).
This thesis focuses on the loss of carbon from dead lodgepole pine wood following MPB attack,
with the goal of improving the understanding of decay rates and therefore, implications for
atmospheric CO2 contributions from forest disturbances. In the following background section I
review leading methodologies for estimating forest carbon fluxes and wood decay, with
emphasis on the approach used by Canadian Forest Service researchers for forest carbon
accounting. I then briefly review a few of the leading factors affecting wood decomposition in
natural landscapes. In the methods section I describe the field, laboratory, and statistical
procedures used for all phases of the project. These include the methods for a precursory
assessment of wood density as a surrogate for carbon density, as well as the main investigation
2

of carbon loss from decaying MPB-killed lodgepole pine wood. My findings for the preliminary
work and focal research are presented in the results section. In the discussion, I talk about the
results and implications of the project in the context of existing literature, with special attention
on how this research applies to the Canadian Forest Service's national forest carbon budget
model. I also share some reflections about forest ecosystem integrity, and thoughts for future
research. In the conclusions I highlight the main points of the research.
Background

Metrics offorest carbon fluxes
Various approaches have been employed in efforts to quantify ecosystem-level forest carbon
fluxes under both undisturbed and disturbed conditions. Regardless of the methodology or
ecosystem state, carbon flux quantification involves certain basic metrics which are either
measured directly or inferred over seasonal or annual time periods. These include 1) the total
amount of ecosystem photosynthesis (gross primary productivity - GPP), 2) carbon released by
vegetation during respiration (autotrophic respiration - Ra), 3) net carbon sequestered by
vegetation (net primary productivity- NPP, defined as GPP - Ra), 4) the amount of carbon lost
through decomposition or herbivory (heterotrophic respiration - Rh), and 5) the net carbon
sequestered by an ecosystem (net ecosystem productivity - NEP, defined as NPP - Rh) (Chapin
et al. 2006, Hicke et al. 2012). These terms are generally expressed in units of g C m- 2 yr- 1,
requiring specification of both an area and a time period with the affiliated carbon quantity.
Positive NEP values indicate that a system has accumulated carbon (i.e., was a net sink), whereas

3

negative NEP values indicate that a system has released more carbon than was sequestered (i.e. ,
was a net source) over a given period2 .
Disturbances such as MPB interrupt the net carbon storage capacity of forests in two ways. First,
photosynthetic carbon sequestration (GPP) is arrested, which occurs immediately when trees die.
The second is a much slower response: the volume of dead wood going through the process of
decomposition (Rh) increases relative to live biomass and pre-outbreak levels of dead organic
matter (DOM), thereby amplifying the amount of carbon being released back to the atmosphere.
This long-term gradual response of coarse woody debris (CWD) represents the fate of the
majority of carbon stored in above-ground biomass in conifer forests (Bois et al. 2009).

Approaches for quantifying forest ecosystem carbon fluxes
Measuring biosphere exchanges of gases (e.g. , CO 2) has an established history in the fields of
plant physiology and ecology (Heinicke and Childers 193 7, Thomas and Hill 193 7). Early work
involved closed-chamber gas exchange rate measurements which were only capable of capturing
at the scale of single leaf, plant, or small soil units over short observation periods (Heinicke and
Childers 193 7, Musgrave and Moss 1961 , Schulze and Koch 1971 , Field et al. 1982). Many
biases and assumptions limited the extent to which these observations could be interpreted,
making ecosystem-scale extrapolations fundamentall y problematic (Rochette et al. 1992, Lund et
al. 1999).
Quantifying carbon stocks and stock changes later became an independent way to measure
ecosystem carbon fluxes, as well as a means to verify extrapolated flux estimates from fine2

Similar to NEP, net ecosystem exchange (NEE) is the net flux expressed relative to the atmosphere, rather than the
ecosystem. Therefore, the sign is reversed: positive NEE indicates a net increase in atmospheric carbon, negative
NEE implies a net loss of atmospheric carbon and gain to the ecosystem (Hicke et al. 2012).

4

scaled gas exchange measurements. Carbon stocks are quantified by assessing soil carbon
concentration, soil organic layer thickness or volume, living biomass volume, and annual inputs
of litter and CWD, for example. Jenny' s (1950) work established this approach by making the
assumption that, for stable forest ecosystems, " ... the annual rate of addition of organic carbon is
equal to the annual rate of loss," (NEP = 0). The extension of Jenny' s assumption is that for
systems not in equilibrium, the changes in carbon stocks represent the net ecosystem carbon flux
(NEP -t 0). Measuring carbon stocks has persisted as an important tool not only for verifying gas
exchange measurements, but for estimating landscape-scale carbon fluxes directly (Laiho and
Prescott 1999, Matsuzaki 2011 ).
Subsequent improvements in closed-cell gas exchange technology, together with advances in
micrometeorological instrumentation (Baldocchi 2003), enabled the evolution of openchambered gas sensors (Bowen ratio I energy balance system) (Spittlehouse and Black 1980,
Held et al. 1990), and three-dimensional open-air sensors (eddy-covariance technique)
(Swinbank 1951 , Mooney 1972, Wesely et al. 1983 ). This technology bypassed the spatial and
temporal limitations of the closed-cell systems. Eddy covariance instruments measure CO 2,
H 2 0 , and energy fluxes from ecosystems, permitting the estimation of forest-scale carbon flux
variables such as Rh and NEP (Baldocchi et al. 1988). As part of the ' Fluxnet Global Network'
(2013), eddy covariance systems are now stationed at several hundred sites worldwide, and are
recognized as integral tools for stand and landscape-scale estimates of ecosystem gas exchange
(Baldocchi 2003 ).
With the growing importance of understanding global-scale carbon fluxes as a driver of climate
change, other methods have emerged for performing CO2-flux measurements over even larger
5

areas. Remote sensing has been shown to effectively infer variables such as atmospheric CO 2
concentrations, biomass per unit area, land cover/use and - given adequate record sequences land cover/use changes (Grace 2005), and gross primary productivity (Running et al. 2004).
Atmospheric inversion models use a hybrid of remotely-sensed atmospheric CO2 concentration
data, land surface information, and computer simulation modelling to extrapolate CO 2 fluxes
over a very broad scale (Dolman et al. 2005). Light Detection and Ranging (LiDAR) has also
been reported effective for assessing three-dimensional canopy structure for upscaling eddy
covariance measurements of gross ecosystem productivity to landscape scales (Hilker et al.
2008a). Integration of multiple data sources and comparison of ecosystem carbon flux estimates
among independent methodologies has yielded validation and support for these technologies.
Computer-based simulation models represent a well-developed approach for estimating
ecosystem carbon flux parameters such as NEP. Models generally contain three categories of
components: 1) pools (stocks) which accumulate and store carbon, 2) transfer mechanisms that
move carbon between pools, and 3) factors that regulate transfer rates and therefore, the size of
the pools (Chen et al. 2000a). A major distinguishing factor among models is the
methodological approach used to simulate sequestration of carbon in the modelling framework.
There are two primary categories of models. In process-based models, carbon sequestration is
computed from empirically derived relationships between physiological parameters such as
photosynthesis, and environmental factors such as atmospheric and growing condition variables
(Waring et al. 2014). By responding to external control factors, process-based models inform
how changes in environmental conditions may mechanistically influence changes in forest
carbon stocks and fluxes. In contrast, inventory-based models simulate carbon sequestration as a
function of growth and yield relationships, which are derived from historic growth data for
6

specific forest age classes and stand types (Kurz et al. 2009). This distinction makes inventorybased models suited to predicting how forest growth and carbon balances may be affected by
management practices and natural disturbances (Kurz et al. 2009) under static environmental
conditions.
Canada's carbon budget model and carbon flux predictions for MPB disturbance

Of the numerous methodologies that exist for quantifying forest carbon fluxes, inventory-based
computer simulation modelling is the approach used by Canadian Federal researchers for
national-level carbon accounting and reporting. The Carbon Budget Model of the Canadian
Forest Sector (CBM-CFS3 , or CBM) was initiated in the early 1990s in accordance with the
Intergovernmental Panel on Climate Change (IPCC) standards and the Kyoto Protocol mandate
(Kurz et al. 1992, Kurz and Apps 2006) to assess the impacts of disturbance, climate change 3,
and forest policy options (Kurz et al. 1992) on the carbon balance of managed Canadian forests.
The CBM-CFS3 is therefore parameterized specifically to estimate national-scale forest carbon
stocks and fluxes in Canada, and it is the primary source of information used by the Canadian
government for national-level carbon accounting.
Like other carbon flux simulation models, the CBM-CFS3 is based on a structure incorporating
carbon pools, carbon transferring processes, and transfer regulating mechanisms. As of 2009,
the model included 21 carbon pools, and numerous sub-models to represent above- and belowground annual tree growth, litter fall , biomass turnover and decay. The model incorporates

3

Although assessing impacts of climate change was specified as one of the mandates for the CBM-CFS, an
inventory-based modelling approach does not have the capacity to respond to climate change factors (Kurz, pers.
comm .).
7

explicit detail for the disturbance processes of harvesting, mortality from insect and fire and
associated forest fuel consumption (Kurz et al. 2008).
In 2008, the Canadian carbon modelling team published results applying the CBM-CFS3
specifically to the MPB disturbance in BC. Based on model output, Kurz et al. (2008) predicted
that BC's MPB-killed stands would release an average of 36 g C m- 2 y{ 1 between 2000 and 2020
due to the surplus of decaying wood, accelerated harvesting, and forest fires on MPB-affected
landscapes. Although this prediction may be plausible, external validation of CWD decay rates
is an outstanding need (W. Kurz, personal communication, 2014), and the sign, magnitude and
timing of net carbon emissions following MPB attack - specifically decay-related carbon
emissions - remain topics of debate. Eddy covariance (fixed and roving tower) measurements
made since 2007 in several locations within the MPB-affected area in central BC show that
recovery ofNEP following beetle attack has occurred rapidly (Brown et al. 2010; 2012). The
observed increase in NEP was due to an increase in GPP. Furthermore, NEP and foliar gasexchange measurements indicated a significant increase in photosynthetic capacity, largely
because of understory tree recovery (Bowler et al. 2012). Evidence of a significant increase in Rh
has not emerged from the eddy covariance measurements, indicating that decomposition of the
additional dead organic matter is slow, contrary to the CBM-CFS3 predictions (Brown et al.
2010; 2012).
Limitation of decomposition (R,J estimates in the CBM-CFS3

Scientific models gain internal validation (reliability) when their limitations are clearly
delineated through a process of sensitivity, uncertainty, and error analyses. Model estimates
gamer broader acceptance when external validation occurs through comparison with results from
8

empirical and independent methods. Many components of the CBM-CFS3 have been extensively
calibrated, tested and verified (White et al, 2008, Smyth et al. 2010, Shaw et al. 2014), and have
established credibility with the Canadian federal government and the IPCC community
(Environment Canada 2011). However, despite the model's wide usage in Canada, one of its
most important components - the decomposition rates for dead trees - lacks adequate
substantiation by field-based observations, particularly in the context of tree mortality associated
with the recent MPB epidemic in BC.
Original decomposition rates used in the CBM were based on personally-communicated
estimates by M. Harmon and A. Trofymow, who independently have extensive experience
studying litter turnover processes in forests . Harmon and Trofymow developed decay rate
estimates for the Moderate Temperate ecoclimatic province of southern Ontario 4 (Kurz et al.
1992). These original estimates then served as reference decay rates for all DOM pools 5 in the
CBM. Regionally-specific decay rates for the other ten Canadian ecoclimatic provinces and their
respective DOM carbon pools were derived as a function of the reference decay rates and local
mean annual temperatures (Kurz et al. 1992).
In 2009 and 2010 the CBM team calibrated the DOM sub-model with data from the Canadian
Intersite Decomposition Experiment (CIDET) - a project designed to evaluate litter decay over a
12-year period at 16 forested sites across Canada (Trofymow et al. 2002, Smyth et al. 2010).
These improvements included adjusting decay rates and expanding other decay parameters to

4

Hannon and Trofymow have extensive knowledge of decomposition processes in ecosystems of the Pacific
Northwest and BC's west coast, respectively. Harmon has also studied wood decomposition in northwestern Russia.
However, I was unable to find studies by either scientist in ecosystems of southern Ontario, for which the reference
decay rates were developed, and upon which all other CBM decay rates were subsequently based.
5
See Table Al in Appendix A for descriptions of the CBM-CFS3 DOM pools.

9

more accurately represent DOM carbon pools and processes. Even though the CIDET project
verified CBM decay rates, it only assessed litter material classified in the very fast and fast pools
(litter, fine woody debris (FWD), roots, and stumps). To date, the medium decay rate - which
represents decomposition of merchantable-sized stem wood (i.e., CWD and snags) and the
largest above ground carbon pool - has had little verification due to a shortage of suitable fieldbased data at appropriate spatial scales for national-level modelling (W. Kurz personal
communication, 2014).
Kurz et al. (1992) and others (Yin 1999, Zhou et al. 2007, Smyth et al. 2010) have stated this
knowledge gap exists regarding decomposition rates, largely because of the challenges and
limitations of quantifying the heterogeneous processes of wood decay (Lambert et al. 1980,
Graham and Cromack 1982). In spite of the challenges, however, the topic has been addressed
in a number of comprehensive textbooks (Scheffer 1973, Rayner and Boddy 1988, Zabel and
Morrell 1992), journal articles (Rayner and Todd 1979, Harmon et al. 1986, Yin 1999,
Cornelissen et al. 2012, Jomura et al. 2008, Shorohova and Kapitsa 2014 ), and technical reports
(Bull 1983, Maser and Trappe 1984, Lewis et al. 2006) over recent decades. I will review this
literature in the following section, with emphasis on lodgepole pine when possible.

Methodologies for quantifying wood decay
Early efforts to quantify decomposition rates resulted in what was known as the equilibrium
principle (Jenny 1950). In other words, if climatic and environmental conditions are stable,
accumulation of dead biomass material in a forest will eventually reach a point when the rate of
decomposition matches the rate of input, thereby resulting in a relatively constant volume of
organic material on the forest floor. With this assumption, the rate of decomposition could be
10

evaluated by measuring the annual input of litter material. This conceptual model for estimating
decomposition rates has been used by many (Olson 1963 , Minderman 1968), and is frequently
cited in recent decomposition studies (Johnson and Greene 1991).
Real-time methods for quantifying wood decay include closed-cell chambers either attached to or
completely enclosing CWD. Instantaneous rates of CO 2 evolution are measured as the air within
the chamber is circulated through a gas analyzer system. Similar to Boddy' s (1983a) enclosed
chamber for branch respiration measurements, Jomura et al. (2008) developed a large-volume
chamber for measuring instantaneous CO 2 release from decomposing logs at forested field sites
in Japan. Closed-cell chamber methods have been used both independently and in conjunction
with other methodologies as a means of verifying components of whole-ecosystem CO 2 fluxes
(Emmel et al. 2014).
Like real-time chamber measurements, long-term monitoring approaches also use real-time
observations, but mass or volume losses due to decomposition are determined by re-measuring
sample specimens that have been exposed to field conditions over specific periods of time (e.g. ,
years or decades), rather than measuring instantaneous rates. The moisture-corrected difference
between consecutive measurements signifies the loss of mass or volume over time. Two

country-wide long-term litter decomposition studies have been conducted in North America: one
over a IO-year period in the US (LIDET 1995, Gholz et al. 2000), and another over a 12-year
period in Canada (CID ET, Trofymow et al. 2002). These studies have yielded some of the most
spatially broad datasets for litter decomposition. More so than for litter, long-term studies
evaluating wood decay pose a greater challenge because of the inherently slow nature of the
process. In spite of the logistical challenges, however, several longer-term monitoring
11

experiments have successfully executed wood decay rate estimates over, for example, 20 years
(Herrmann and Prescott 2008), and even 65 years (Stone et al. 1998).
The more common method for estimating decay rates is the chronosequence approach, which has
been used in numerous studies in recent decades (e.g., Graham and Cromack 1982, Means et al.
1985, Johnson and Greene 1991 , Busse 1994, Brown et al. 1998, Beets et al. 2008, Ryan et al.
2008, Angers et al. 2010). Also known as a retrospective approach, chronosequence studies rely
on dendrochronological analysis (tree ring dating) to determine the year of death and/or fall of
woody debris (Means et al. 1985). Losses of mass or density are calculated based on an assumed
maximum at the time of tree death (Busse 1994). The decay rate is determined by dividing the
change in wood mass or density by the number of years since tree death (e.g., g cm-3 y{ 1).
Although there are more potential sources of error associated with the chronosequence method
(Means et al. 1985), it is frequently employed because of the logistically simpler data collection
process that requires only a single visit to field sites while still enabling multi-year or multidecadal observations (Walker et al. 2010).
In 2000, M. Harmon and colleagues introduced the decomposition-vector method for evaluating
wood decay in a simultaneous comparison with the chronosequence approach. The vector

method hybridizes chronosequence and long-term monitoring methods; it relies on
dendrochronological dating to establish years-since-death, combined with short-term re-sampling
- in this case 3 years - to capture a specific period (i.e. , 3 years) of the decay process following
tree death. In this way, the short-term decay rate (over the 3-year period) can be compared with
the overall decay rate from the time of tree death to ascertain changes in the decay rate over over
time (Harmon et al. 2000).
12

Factors affecting wood decay
Decomposition of wood is influenced by an array of environmental variables that interact within
and across spatial and temporal scales (Lambert et al. 1980, Harmon et al. 1986). This
complexity has been described by Laiho and Prescott (2004) simply as derivatives of two key
factors: substrate quality and environment. However, the intricate network of confounding
"derivative" influences is extensive, and untangling them remains an important yet daunting
proposition for researchers (Maser and Trappe 1984, Yin 1999). Given these layers of
complexity, the factors affecting wood density described here are not intended to be
comprehensive, but rather a brief highlighting of some of the leading factors and corresponding
relationships that have been identified in the literature.
Temporal

Among the many complex influences challenging the quantification of decay rates is time itself
(Graham and Cromack 1982; Maser and Trappe 1984, Yin 1999). The sheer time required for
decomposition to progress dwarfs typical short-term study designs by exceeding convenient
observation periods (Harmon et al. 1986, Johnson and Greene 1991 ). Beyond this technical
challenge, the time for complete decomposition extends well beyond human lifespans, which

inherently limits our capacity to comprehend the suite of processes occurring from the time a tree
dies to when it is entirely assimilated into other forms .
Another aspect of the temporal dimension of wood decay concerns the variability of CWD inputs
over time. Forests with mixed age-class structures and small-scale/single-tree mortality events
are likely to have an even distribution of dead wood across all decay classes at all times (Graham
13

and Cromack 1982). In contrast, landscapes having even-aged forests are more likely to undergo
phases of low and high CWD inputs as stands uniformly mature and succumb to disturbances
(Sturtevant et al. 1997, Page and Jenkins 2007). Thus, even-aged stands may have an entire
forest cohort proceeding through decay classes at approximately the same time. Wholeecosystem decay rates for these two scenarios would therefore be very different: the first having
a relatively stable rate of decay, the second having periods of high and low decay rates.
At the individual tree level, the onset of decomposition following tree death can be rapid or
extremely slow, dictated largely by bole-specific conditions of resource quality and microclimate
suitability (Harmon et al 1986, Rayner and Boddy 1988, Yin 1999, Laiho and Prescott 2004).
These micro-scale conditions ultimately determine the onset and rate of wood decay, but are, in
tum, controlled by factors that act at coarser spatial scales, which will be considered next.
Climate

Optimal conditions for microbial decomposition occur when wood remains moist at moderate
temperatures. Toward the limits of optimal temperature and moisture ranges, microbial
processes decline, and at extremes of hot or cold, dry or wet, decay activity can be inhibited
entirely (Zabel and Morrell 1992). Climate therefore, by regulating temperature and moisture
inputs, plays a major regulatory role in wood decomposition.
At the microclimate scale, most decay fungi thrive between 20 and 30 °C (Deverall 1965,
Scheffer 1973, Rayner and Todd 1979, Boddy 1983b), although Zabel and Morell (1992)
reported optimal temperatures ranging from 15-45 °C. For some decay fungi, survival thresholds
can be below O °C, or as high as 60 °C (Deverall 1965). Chen et al. (2000b) found that CO2
14

release from decomposing lodgepole pine roots peaked at 30 °C, which was about ten degrees
lower than the peak respiration temperature for dead roots of other conifers in the same region of
Oregon.
Many decay fungi require a minimum wood moisture content around 30 % of dry wood weight,
(Kaarik 1974, Rayner and Todd 1979) and exhibit peak growth within a range of 40-80 %
moisture (Scheffer 1973, Rayner and Todd 1979). This range varies, however; Kaarik (1974)
reported growth of decay fungi in moisture conditions up to 160 %, and Boddy (1983a) showed
peak CO 2 release from decomposing branches having moisture contents of nearly 200 %. Peak
respiration rates for dead woody lodgepole pine roots were reported between 125 and 225 %
moisture by Chen et al. (2000b).
Boddy' s work ( 1983a) illustrated the interdependence of temperature and moisture and their
combined effects on decomposition rates. When temperature was held steady at 15 °C with
increasing moisture, CO2 release from decaying branches increased to a maximum then declined.
A similar pattern was observed at 25 °C, but with a much greater CO2 release, suggesting that at
higher temperatures and higher moisture contents, the maximum respiration rate also increases.
Respiration rates of pine CWD in a temperate broad-leaved forest in Japan exhibited an
exponential response to temperature, and a quadratic response to moisture (Jomura et al. 2008).
The decline ofrespiration at higher moisture contents observed by Boddy (1983a) was attributed
to constricted gas exchange as voids in the substrate became filled with water. The internal
microclimate of wood can therefore affect gas exchange potential (Boddy 1983a), fungal growth
rates, and fungal species assemblages (Boddy 1983b, Son et al. 2011) which I discuss in more

15

detail later on. Thus, temperature and moisture affect wood decay rates indirectly, as well as
directly.
If one imagines zooming out from the micro-scale to a broader perspective, microclimate

conditions themselves are dictated by many factors, such as host tree species (Alban and Pastor
1993) and wood density (Boddy 1983a), and at an even broader scale, by regional climate (Jenny
et al. 1949). Yin's (1999) integrative approach for estimating decay of woody debris in field
conditions led to the conclusion that, in addition to tree species, latitude and elevation, January
and July temperatures and precipitation were key parameters defining decay rates. Accordingly,
latitude and seasonality influence wood decay rates by determining the form of precipitation
(e.g., snow versus rain), annual temperature cycles or lack thereof (Jenny 1950), and hence, the
extent of exposure to ideal decay conditions. Coastal and tropical regions where precipitation
falls as rain year-round have a greater potential for rapid decay, compared to areas that have subzero winter temperatures and/or hot dry summers (Jenny et al. 1949). At higher latitudes, depth
and duration of snow cover can influence the rate of decay for wood at or below the snow
surface; though growth of mesophilic 6 fungi may be supressed, other cold-tolerant fungal species
can remain active in the cold moist conditions under the shelter of snow (Kfilirik 1974).
Over time, the collective effects of physical, chemical, and biological processes (which I will
discuss in more detail below) render wood tissue more prone to mechanical weathering. Hence,
repeated wetting and drying, freezing and thawing events (Zhou et al. 2007) that are relatively
negligible factors in the early stages of wood decay may be more significant players later on,
following initial chemical breakdown by microbes.
6

Mesophilic: thriving at moderate temperatures.

16

Physical

The physical size, position, and extent of fragmentation of CWD are other factors affecting the
microclimate and substrate conditions that dictate decay potential. Compared to smaller pieces
of CWD, large logs have low surface-area-to-volume ratios, which can limit the penetration of
decay agents from external sources (Harmon et al. 1986). Therefore, the rate at which dead
wood becomes colonized and degraded by decay organisms may to some extent depend on bole
size, although efforts to demonstrate this size dependency have yielded inconclusive results
(Harmon et al. 1986).
Other size-related issues include the dependence of volume on moisture content. The diameter
of wood samples is often used to determine sample volume and density, which can then be used
to indicate the extent of decomposition. However, the moisture content of wood affects the
volume; disproportionate shrinkage in wood occurs when the internal humidity drops below the
fibre saturation point (30 %) (Barton and Brownell 1981 , Harmon et al. 1986), resulting not only
in loss of volume, but also in checking (cracking) (Lewis et al. 2006).
If dead trees remain standing, they may persist as snags without substantial decay (Johnson and

Greene 1991) for considerable lengths of time. Trees with pre-existing root rot or butt rot at the
time of death are more likely to fall sooner and follow an accelerated path of degradation
(Angers et al. 2010). If otherwise uninfected, snags may persist standing for many years to
decades because they are less accessible to soil-borne decomposers, intercept less precipitation
by being vertical, and retain less moisture, e.g., because of gravitational flow down the stem and
exposure to canopy air movement and evaporation (Bull 1983, Fahey 1983). By remaining in an
upright position snags ultimately avoid exposure to many decay agents (Shorohova and Kapitsa
17

2014); however, they are more prone to checking (longitudinal cracking) than fallen boles
(Lewis et al. 2006). Checking happens more readily in snags because of the drier atmosphere
coupled with exposure to temperature extremes. Checking effectively increases bole surface
area, and provides a direct pathway for airborne fungal spores to become established internally.
Fallen logs are more readily accessible to agents of decay once they are in contact with the
ground (Harmon et al. 1986, Busse 1994); they retain more moisture from precipitation and
upward wicking, as well as by being within the relatively protected and humid boundary layer of
vegetation and snow cover. The process of falling itself can accelerate the rate of decay when
stem breakage occurs (Angers et al. 2010). Snags with pre-existing rot (e.g. , from roots, wounds,
insect damage, checking, etc.) are more likely to fracture or shatter upon impact (Lewis and
Hartley 2006), thus increasing the surface area exposed to moisture and decomposers.
Fragmentation, by falling or other causes (e.g. , lightning, checking, woodpeckers, mammal
excavation, etc.), exposes interior surfaces that would otherwise take much longer for decay
pathogens to penetrate.
Chemical

The chemical composition of wood adds another layer of complexity to the challenge of
quantifying decomposition. Trees are a heterogeneous combination of chemical compounds that
range from volatile, to soluble in certain media, to relatively inert - properties that change over
the course of degradation. All trees manufacture cellulose, hemicellulose, and lignin: organic
compounds that make up the cell walls of woody tissue. This trio is responsible for the structural
support in stem wood and accounts for nearly 100 % of wood mass (Barton and Brownell 1981 ).
The relative amounts of these compounds in lodgepole pine are similar to other softwood tree
18

species; cellulose is most abundant with lesser amounts ofhemicellulose and lignin (43, 32, and
25 % respectively, approximate averages) (Lieu et al. 1979, Kim et al. 1989, Koch 1996).
Carbon is the backbone of all three structural compounds, which explains why wood contains
approximately 50 % carbon by mass. Other elemental constituents of cellulose, hemicellulose,
and lignin vary, as do their molecular size, shape, and structural complexity (Barton and
Brownell 1981 ). These differences result in differential susceptibility to break-down by
microbial enzymes. Not only are cellulose and hemicellulose more abundant in wood relative to
lignin, they are also the preferred nutrient source for many decay fungi (Harmon et al. 1986).
Other compounds occurring in wood are referred to as extractives and have lower molecular
weights than cellulose, hemicellulose, and lignin. Extractives account for less than 4 % of wood
mass in lodgepole pine (Kim et al. 1989), and include a range of organic and inorganic
compounds (Koch 1996) including resins that are toxic to decay fungi and insects (Barton and
Brownell 1981 ). Extractives can be volatile, water-soluble, or relatively inert (i.e. remain as ash
after combustion) (Koch 1996). Though they occur in micro amounts compared to the three cell
wall components, extractives largely define species-specific wood traits, including the
susceptibility to decay (Barton and Brownell 1981 , Harmon et al. 1986, Cornwell et al. 2008).
Extractives are therefore key to the tree ' s defense system against pathogens. The abundance and
distribution of extractives in wood are much less predictable than the three major compounds,
adding to the complexity of determining specific decay rates.

19

Biological

A multitude of biological factors influence wood decomposition, from microbial and invertebrate
communities, to species-specific wood traits and complex forest stand attributes. Fungi are the
primary microbial agents that decompose wood (Scheffer 1973, Kaarik 1974, Dolenko et al.
1981) in terrestrial environments (Harmon et al. 1986). Decay fungi specialize in breaking down
the compounds of cell walls, degrading the structural integrity of dead wood and causing unique
fragmentation patterns. The two main fungal groups responsible for wood decay are brown and
white rot fungi (Kirk and Highley 1973). White rot fungi produce enzymes that metabolize both
lignin and cellulose, whereas brown rot fungi are only capable of breaking down cellulose
(Harmon et al. 1986). Other fungi such as the sap-staining symbiotic partners of the MPB infest
and occupy fresh wood, but do not affect cell wall components the way decay fungi do. Rather,
they extract nutrients from cell contents without affecting the structural integrity of the wood
(Rayner and Boddy 1988). Though they do not break down cell walls directly, sap-stain fungi
can facilitate the entrance of decay fungi into the sapwood, predisposing the tree to early onset of
decay processes (Lewis et al. 2006).
External environmental factors can influence the diversity and abundance of fungal communities.
Higher year-round precipitation in the Prince George region compared to other drier regions of
BC was associated with greater diversity of decay fungi (Son et al. 2011 ). Succession of
complementary fungal communities can accelerate decomposition, but if multiple competitive
fungal species co-occur, their demand for the same resources can slow the process
(Hattenschwiler et al. 2005).

20

The role of insects in wood decomposition can be considerable, but is much less predictable than
that of decay fungi. Primary bark beetles such as the mountain pine beetle may be initially
instrumental in excavating the cambium and inoculating sapwood with blue stain fungi, causing
tree death (Carroll and Safranyik 2003), and secondary bark beetles may opportunistically
follow, mining the cambium while it is relatively fresh. But with or without the influence of
primary and secondary bark beetle attacks, dead wood can remain unaffected by wood-boring
insects indefinitely (Harmon et al. 1986). If present, however, local populations of termites,
carpenter ants, or other insect decomposers (i.e., saproxylic insects) that excavate dead wood can
dramatically accelerate its breakdown (Maser and Trappe 1984 ).
As described earlier, species-specific traits of dead wood such as the chemical constitution and
wood density can influence the predisposition of wood to decay (Cornwell et al. 2008, Weedon
et al. 2009). But in addition to these factors, overall forest stand characteristics in conjunction
with the relative abundance of a host tree' s component parts will influence the volume of dead
wood available to decomposers at any given time, and therefore, the overall rate of
decomposition for the ecosystem. Stand attributes such as age, density, canopy closure, slope,
aspect, and elevation affect the volume of dead wood and the microclimate conditions
determining decay rates of DOM (Maser and Trappe 1984). Needles, twigs, fine roots, and bark
degrade much more quickly than bole wood (Trofymow et al. 2002) because of higher surface
area to volume ratios and lower lignin and cellulose contents, but these constituents make up a
smaller portion of the tree ' s biomass, and therefore, a smaller source pool of degradable material.
For lodgepole pine in BC, Comeau and Kimmins (1989) reported that the bole constitutes about
70 % of total tree biomass, foliage and branches account for 6 %, and belowground about 24 %.
Other studies conducted in western Canada (Johnstone 1970) and northwest US (Little and
21

Shainsky 1992, Litton et al. 2003) report similar biomass allometry for lodgepole pine. Hence,
in addition to the physical and chemical properties of wood described earlier, the relative
abundance of tree parts and the total abundance of woody debris on the landscape will influence
the overall rate of decomposition for the stand, and ultimately the CO 2 respired to the
atmosphere.
Research objective
In an effort to alleviate some of the uncertainty about wood decomposition, the objective of this
research was to estimate wood decomposition rates for the case of recent MPB outbreaks in BC.
Specifically, my goal was to quantify the carbon released from MPB-killed lodgepole pine wood
over the first 40-50 years after tree death, and to characterize the influence of selected
environmental attributes on the rate of carbon loss.

Scope of research
Because of the extreme heterogeneity of environmental conditions that influence rates of wood
decay, I restricted my research to lodgepole pine boles killed by MPB in the central interior of
British Columbia. Limiting the scope of research in this way was valuable for several reasons.
First, boles make up the majority of biomass in a tree (Comeau and Kimmins 1989), and
therefore, account for the bulk of above-ground carbon storage in forest ecosystems. Second, as
dead bole wood is also the carbon pool least characterized in the CBM-CFS3 , it was of particular
value to focus on this component of forest carbon. And third, trees that have succumbed to the
same mode of mortality are more likely to follow similar decomposition trajectories (e.g., MPB
galleries and blue stain fungi act as initial pathways for other decay organisms (Lewis et al.

22

2006). Therefore, focussing my research efforts on MPB-killed boles was a way to eliminate
some of the potential noise that could be introduced by including trees that died by other means.

Research question, hypotheses, and methodological rationale
This research set out to answer the question: What are the rates of wood decay as related to a)
time since tree death and fall, b) climatic moisture and temperature regimes, and c) bole position
(standing, fallen but suspended, or fallen and down on the ground)?
With the support of my committee, I defined the following hypotheses regarding the influences
of the parameters of interest. I hypothesized that:
Hl : The rate of decay is low to negligible for snags.
H2: Suspended CWD is distinctly different from snags because of the higher catch surface upon
which precipitation lands and potentially remains - especially in the case of snow.
H3: Down CWD differs from suspended CWD because, by being in direct contact with the
ground, down CWD is more susceptible to colonization by soil-borne agents of decay, as well as
up-wicking of soil moisture. Down CWD is also within the boundary layer of ground surface
conditions such as vegetation cover in summer, and snow cover in winter.
H4: With respect to climate factors, wood decay rates are faster on sites with higher climate
moisture and temperature because of more optimal micro-conditions for biological activity.
Likewise, I expected sites with drier and colder climatic conditions to have slower
decomposition.

23

To address the research question, I estimated long-term density loss in dead wood specimens at
various stages of decay. I collected samples from MPB-killed boles on sites reflecting a gradient
of climatic variability (i.e., moisture and temperature), and assessed the wood density of snags,
suspended, and down logs. Year of death and year of fall (for fallen boles) were evaluated using
dendrochronological methods. In order to expedite lab work and reduce sample processing
costs, I conducted a preliminary assessment of a subset of samples to validate the use of wood
density as a surrogate for carbon density. This step allowed me to use simpler laboratory
techniques to extrapolate carbon density for the remainder of the sample inventory.

Significance of research
This research is important for several reasons. It will: a) enable better understanding of
ecosystem carbon fluxes by delineating one key CO 2 source, which may influence other forest
carbon-flux components; and b) provide regionally specific field data upon which to base
parameters for the modelling of a major CO 2 source over time. As indicated by Canada' s carbon
accounting team, the disturbance matrix component of the CBM-CFS3 could be improved if
more detailed wood decomposition data were available (W. Kurz and C. Smyth personal
communication 2014, 2015). The dataset generated by this research therefore provides a
verification tool for the decay rates used by the CBM-CFS3 to estimate Canada' s forest carbon
fluxes following high-mortality forest disturbance events. Considering the potential impact of
the CBM-CFS3 for managing carbon emissions from local to international scales, improving
Canada's ability to predict forest-based CO2 contributions to the atmosphere is a high priority
(Johnston et al. 2010).

24

In addition to direct linkages to the CBM-CFS3 , this research has significance for local forest
managers in government, industry, and First Nations groups. Given the relevance of carbonrelated issues to global climate change, local salvage harvesting following major forest
disturbance events will play an important role in the carbon balance of forested ecosystems.
Knowledge of forest carbon dynamics can contribute to responsible forest management decisions
for long-term viability of ecosystems and a timber harvesting industry. I therefore want to
ensure that these findings reach decision makers at the local level.

METHODS

Study area

The study area was located in the sub-boreal central interior of British Columbia between 53 .32°
and 53.98° N latitude, and 122.41 ° and 125.11 ° W longitude (Figure 1). Elevations ranged from
698 to 1252 m ASL. Based on climate normals ( 1960-1990), mean annual precipitation (MAP)
ranged from 300 to 730 mm, and mean annual temperature (MAT) from 1.0 to 3.2 °C (Climate
WNA; Wang et al. 2012) (Table 1). Stands selected for sampling were either pure lodgepole pine
or pine-leading mixed conifer forests , all with extensive pine mortality due to MPB attack.
Within this region, sites were chosen to reflect a range of climatic moisture in the Sub-Boreal
Spruce and Sub-Boreal Pine - Spruce biogeoclimatic zones (Meidinger et al. 1991). By
sampling across this climatic gradient, I hoped to capture the range of variability in wood decay
that may result from different climate conditions. Based on Biogeoclimatic Ecosystem
Classification (BEC), only sites defined as circum-mesic (01 site series, DeLong 2003) were
used, thereby restricting site-specific moisture influences (i.e., soil and/or topographic features) .
25

Mesic sites, referred to as ' zonal sites' in BEC, are areas with average soil and site conditions
that best reflect the regional climate (Pojar et al. 1987). Four biogeoclimatic (BGC) subzones
are represented in this thesis: 1) very dry cold (xc), 2) dry cool (dk), 3) moist cold (me), and 4)
wet cool (wk) (DeLong 2003). Two sites were sampled from each of these four subzones for a
total of eight sites (Table 1).

' .
- -·/
-- -,

·~

,'

~

""

.....
~

BK003

•

,,,.........

BEM22

<:I"

.:::>

BEM26 ~
.A • ·.,
BEM24
'

-.

.,

0

' ' .

.&
•

0

'~

..

./)-...

N

\

British
Columbia

-,,: Hagensborg "'
4

l

0

0

-

.. .

"\:::,

•

25

50

•

...\\

·0

., "'\!J

,

BK304 .-··

if°'

BK296

!: :··~. ~
1

100 Kilometers
~

..

Figure 1. Study area and site locations in the central interior of BC.

26

A

Table 1. Study sites and affiliated climate conditions.
Site

BGC unit

Latitude

Longitude

Elevation (m)

MAP(mm)

MAT (°C)

BK304

SBPSxc

52.0906

-124.8890

1252

463

1.1

BK296

SBPSxc

52.0831

-124.2033

947

295

2.6

BK003

SBSdk

53.6512

-125.1063

909

500

2.1

BEM22

SBSdk

53.5483

-125.0206

920

494

1.9

BEM26

SBSmc3

53.4630

-124.5266

1061

544

1.3

BEM24

SBSmc3

53.3203

-124.4353

1146

544

1.0

BK004

SBSwk

53.9781

-122.4056

698

698

3.2

BK007

SBSwkl

53 .7370

-122.3594

850

732

2.9

Field sampling

Field work was conducted between August and September 2011 . At each site, I selected dead
lodgepole pine boles that had evidence of MPB attack; signs included pitch tubes and beetle exit
holes in the bark, and linear J-shaped galleries under the bark. Blue stain in the sapwood also
aided in confirming MPB-induced mortality. With the help of a field assistant, I collected
wedges from snags and live lodgepole pine trees, and disc samples from suspended and down
boles. Live lodgepole pine, if found at a site, were sampled to provide a baseline for maximum
wood and carbon density.
I used a census-based approach for sampling MPB-killed CWD at each field site. All CWD that
met the criteria were included until the target number of boles had been sampled, or site
conditions were no longer suitable, whichever came first. For every two fallen boles (suspended
27

or down CWD) sampled, I sampled one MPB-killed snag 7. The objective was to sample a
similar number of boles in each of the position categories (snags, suspended, and down) for a
total of approximately 24 boles per site.
The threshold for classifying CWD as suspended or down was determined on the basis of
whether the majority of the bole was greater than 10 cm off the ground (suspended), or not
(down). Samples were later individually classified by the same criteria as either suspended or
down. Coarse woody debris (suspended or down) was only sampled if the following conditions
were met:
•

Dimensions 2: 7.5 cm diameter at breast height (DBH, defined as 1.3 m from the base of
the tree), and 2: 2 m in unbroken length

•

At an angle of ::S 45 ° from the ground and not self-supporting 8

•

Sufficient evidence to confirm species as lodgepole pine, and MPB impacted

•

Presence of an adjacent live 'witness' tree having either a scar inflicted from the fall
event, or a curve at the base (i.e. , with reaction wood) from being pushed over during the
fall event.

Before cutting a log for samples, the bole, and selected sample locations along the bole were
characterized by decay class (Maser et al. 1979) and several other micro-site factors (see
Appendix B, Table B 1). The height of the bole from the ground was measured at each sample
location, including negative values for boles imbedded in the forest floor. These values were

7

Any standing dead tree > I .3 m tall and self-supporting, as defined by Thomas et al. (I 979) and BC Ministry of
Forests (I 998).
8
A self-supporting bole is only supported by its own roots. A non-self-supporting bole therefore is either resting on
the ground, or is held up off the ground by something other than its own roots (e.g., leaning or resting on other
standing trees, other logs, its own branches, etc.).

28

later used to determine the classification of sample position as either suspended or down, as well
as characterize the sample position numerically.
The primary sample collection method involved cutting a cross-sectional slice (i.e., cookie) from
the main stem of CWD. Samples collected from the witness trees captured either the fall-scar or
reaction wood which was then used to date the year of fall of the sample tree (Maser and Trappe
1984, Harmon et al. 1986). I took two cookies from every suspended or down bole to
characterize the variability of decay within individual stems. A third cookie was taken to date
the year of death of the tree.
Collection of samples from CWD with advanced decay (i.e. , decay class 3 or greater, Maser and
Trappe 1984, BC Ministry of Forests 1998) was limited by the extent to which I could clearly
differentiate the degraded CWD wood from the surrounding organic soil and detrital material,
and retain the sample intact for volumetric measuring. For such cases, a section of the log was
cut and removed adjacent to the sample in order to observe and measure it without disrupting the
log' s morphology. If there was evidence of physical fragmentation and loss of volume, undecayed diameters were inferred, based on the horizontal diameter and the most intact parts of
the cross-section that remained. This was not a mathematically-derived correction for volume
losses as developed by Fraver et al. (2013); rather, it was a subjective method that allowed
estimates to be made based on the condition of individual logs. Highly decayed samples were
shrink-wrapped and transported to freezer storage as quickly as possible.

29

Snags and live trees were sampled by cutting a wedge from the trunk at 1.3 m from the base of
the tree 9 . For snags, a core was taken at the same height for dating the tree's year of death.
In the field, maximum and minimum diameters of all cookies were measured and recorded for
volumetric calculations. Diameter measurements excluded bark to ensure consistency whether
bark was present or not. Cookie thickness was measured at approximately eight points around
the perimeter to ensure that the most representative average value was reported. The field data
sheet and a list of field parameters are included in Appendix B (Figure B 1 and Table B 1).
Laboratory analyses

Determination of wood density
Samples were transported to the Enhanced Forestry Laboratory at UNBC in Prince George, BC,
and stored at either room temperature (22 °C), in a cooler at 4 °C in burlap sacks, or wrapped in
plastic and frozen at -20 °C. All samples with advanced decay were frozen until processing. In
the lab, I removed the bark, took photos and weighed each sample.
Basic wood density is defined as oven-dry mass divided by green (i.e., field) volume
(Williamson and Weimann 2010). I chose to use field volumes for my density measurements
because they would be most representative of the tree' s original live dimensions. Two methods
were used for measuring cookies: 1) field-volumes were calculated from the geometric
measurements recorded in the field (excluding bark), and 2) lab-based volumes were measured
by the immersion method described by Graham and Cromack (1982), Williamson and Weimann
(2010), and others.

9

This was the best way I could acquire samples from snags without being qualified to fall trees.

30

Field volumes were calculated using the equation of a cylinder:
(1)

where Vis the cookie volume (cm\ dmax and dmin are the largest and smallest diameters (cm) of
the disc, respectively, and lavg is the average length (cm) or thickness of the cookie.
For lab-based volumes, samples were vacuum sealed in plastic and immersed in a 50 L water
tank on an electronic balance. Based on bag weight and density, I calculated the bag volume to
subtract from the total displacement volume, thereby reducing measurement error and inflated
sample volumes.
The immersion method was used to verify volumes for irregularly-shaped samples including
wedges, which were not measurable in the field by geometric means. The volume of oven-dried
samples was also measured in order to calculate the amount of shrinkage that occurred from field
to oven-dry weight. I then used this shrinkage factor to extrapolate field volumes for wedges,
which had lost moisture and presumably volume during storage in the lower-humidity
conditions 1°.
All samples were dried at 101 °C until constant weight was achieved, as recommended by
Williamson and Wiemann (2010). Samples were cooled in large plastic bags with desiccant to
10

Shrinkage was determined by calculating the percent volume loss that occurred for decay class I and 2 samples
between field and oven-dried moisture content. The shrinkage that occurred from indoor storage to oven-dry
conditions was also measured and subtracted from the field-to-oven shrinkage. The remaining shrinkage factor (4. 1
% ± 0.7) represented the volume loss for wedges between field and low-humidity storage conditions. The volume
of wedges stored in low-humidity conditions was therefore expanded by the shrinkage correction factor to
extrapolate field volumes.

31

minimize reabsorption of moisture. Oven-dry weights were used with field-based (i.e., green)
volumes to calculate basic wood density (Husch et al. 2003 , Williamson and Wiemann 2010).
Because of the dependence of wood volume on moisture content (MC), it is important to
document the MC at the time of volume measurement. Samples were not readily weighed at the
time of collection, so field MC was determined from samples that were stored in the freezer, as
these had not lost moisture and volume due to the drier conditions of other storage methods.
Samples covering a full range of decay classes had been stored in the freezer, so these provided a
means to extrapolate the field moisture content of the other samples.

Carbon concentration analysis
A subsample of 4 7 cookies covering a range of decay classes was used to evaluate carbon
concentration. The first step of grinding samples was completed at the Enhanced Forestry Lab at
UNBC in Prince George; this involved grinding each sample in its entirety to a coarse powder
using a Micron hammer mill. Complete samples were ground to ensure the carbon analysis
would be representative of the full cross section of the bole at the sampling point. All
subsequent sample preparation steps and analyses were completed at the Pacific Forestry Centre
in Victoria, BC. Using a Humboldt splitter, I divided the ground material of each sample

repeatedly until it was reduced to an amount appropriate for further processing (approximately
100 mL). These subsamples were ground with a Retsch cyclone mill and finally with a Retsch
ball grinder to obtain an even-textured fine grind of less than 10 µm.
Approximately 5 mg of each sample were used for carbon concentration analysis in a Costech
elemental analyzer; an additional amount of the final ground material was concurrently dried at
32

103 °C overnight to determine the moisture correction factor for each sample. Resultant carbon
concentrations were converted to carbon densities by multiplying the concentration data by the
respective basic wood density of each sample.

Dendrochronology
All samples for dating the year of death and year of fall of sampled trees were processed and
analyzed in the Dendrochronology Lab at UNBC. Increment cores were mounted and sanded
using standard dendrochronology techniques (Stokes and Smiley I 968). Cookies that were sound
were sanded directly without any mounting. Those with evidence of deterioration were mounted
on plywood before sanding. Fall-scar and reaction wood samples were sanded and evaluated for
the year of fall of all suspended and down boles.
To determine the year of death, annual ring widths were measured to the nearest 0.001 mm using
a Velmex System (Velmex, Inc. 1992) and MeasureJ2X software (VoorTech Consulting 2004).
Each sample was cross-dated against a master chronology that had been established previously
for the same study area (Thompson 2005). Cross-dating was done using a combination of the
COFECHA (Holmes 1983) software program and visual cross-dating for verification.
The year of tree fall was determined from fall-scar wedges by counting from the outermost ring
(2011) back to the ring in which the damage occurred. For reaction wood samples, the year of
tree fall was evident as the ring in which tree growth changed from normal concentric rings to
darkened disproportionately thicker growth on only one side of the sample cross-section (i.e. , the
altered growth of a conifer as it rights itself from being pushed over).

33

Statistical analyses

Wood density as a means to quantify carbon density
To test if basic wood density can be used as an effective predictor of carbon density, I used
simple linear regression. Because carbon density and wood density were correlated by the
common unit of sample volume, I ran a second linear regression to evaluate the relationship
between C concentration and wood density, which were completely independent parameters.
None of the data strayed too far from normal distributions, so assumptions of normality were
acceptable.

Carbon density loss from decaying MPB-killed wood
To evaluate if carbon density differed among the position categories of dead standing (snags),
suspended, or down boles, I used analysis of variance (ANOVA), followed by Sheffe's test to
determine where differences occurred. Samples from live lodgepole pine trees were included as
a category in the ANOV A as a baseline for carbon density from which other categories would
diverge. To meet the assumption of normality, carbon density data were square-transformed
prior to running the ANOV A. Sheffe's test can handle uneven numbers of observations between
groups (Tabachnick and Fidell 2001) which was the case with the carbon density data (Appendix
C, Table Cl). The preliminary ANOVA assessment supported differentiating among the snag,
suspended CWD, and down CWD categories, so the remainder of statistical analyses were
conducted on these groups independently.
I used linear mixed effects models to account for the hierarchical structure of my study design,
which consisted of samples nested in boles, and boles nested within sites. In mixed effects
models, the fixed portion of the model includes the parameters of interest (i.e., the fixed effects);
34

the random effects are variables that are not of interest per se, but may account for variance
introduced by sampling different groups within a larger population. The error associated with
the random effects is not estimated with that of the fixed effects; rather, it is estimated as a
separate component of the overall model error (Rabe-Hesketh and Skrondal 2008). By
partitioning the variance among the fixed and random factors, mixed models can explain more of
the residual error of a hierarchical study design than a simple regression model (Rabe-Hesketh
and Skrondal 2008). The fixed portion of my models consisted of carbon density regressed on
the predictor variables (i.e., the fixed effects at level 1); these observations were nested in the
random effects of boles (level 2), which were nested within sites (level 3). The formula
describing the partitioning of variance for a three-level nested random-intercept model is:
(2)

Yijk =

(1)
/3 + ujk(2) + uk(3) + Eijk

where i = the level 1 observations,}= the level 2 groups (i.e., boles), k = the level 3 groups (i.e.,
sites) (StataCorp. 2009a). Applied to my study, Yijk is a carbon density observation nested in a
bole and site, f3 is the fixed effects intercept, ui~) is the level 2 random intercept for boles nested
in sites, uf) is the level 3 random intercept for sites, and E(/~ is the level 1 error term associated
with individual observations.
To evaluate the influence of the fixed effects, I constructed a set of candidate models for each
tree position category that reflected my hypotheses, which were based on empirical observations
and evidence from the literature. Candidate models consisted of years since tree death or tree
35

fall as the main predictor variable, with climate parameters and height off the ground as
additional explanatory variables (see Table C2 in Appendix C for a summary of all potential
predictor variables).
I used four different potential climate variables to evaluate which, if any, best explained the
variation in the data; these were 1) MAP and MAT, 2) mean winter (January, February, and
December precipitation and temperature, 3) mean summer (June, July, and August) precipitation
and temperature, and 4) BEC subzones. The four types of climate variables were chosen for the
following reasons: MAP and MAT are commonly used climate predictors and are therefore
useful for comparing with other studies, and potentially for extending results to other locations.
Because of the reported importance of January and July climate parameters as significant
predictors of wood decay by Yin ( 1999), I included average winter or summer precipitation and
temperature data as more specific alternatives to MAP and MAT. Both the mean annual and
seasonal climate data were interpolated using the online version of Climate WN A (Wang et al.
2012). BEC subzones provide another means of describing climatic differences between sites by
incorporating climate and other ecological factors into ecosystem-specific classification units
(Meidinger et al. 1991 ); I therefore used BEC subzones as a fourth and simpler option that
grouped sites into one of four climatic categories (very dry cold, dry cool, moist cold, and wet
cool).
To evaluate which of the candidate models best described changes in carbon density, I used
Akaike's Information Criterion (AIC) for small datasets (Al Cc; Burnham and Anderson 2002).
Information theoretic approaches such as AIC follow the principle of parsimony: models are
ranked to balance the benefits of improving the explained variance against the costs of including
36

more parameters in a model (Gotelli and Ellison 2004). Models that explain more of the
variance with fewer parameters have a lower AIC score. Therefore, the best model out of a
given set of candidates is identified by the lowest AIC score. If a candidate model ranks within 2
AIC units of the best model, it is often considered to explain a comparable proportion of the
observed variance as the top-ranked model (Burnham and Anderson 2002). In some cases,
parameter coefficients of the top-ranking models can be averaged to yield a single, best-fitting
model. However, averaging may not always be the preferred route to describe the system being
modelled; in the case of complex systems, multiple models of similar explanatory value but
differing parameters may be more realistic. For this reason, I reported all models ranking within
2 AI Cc units of the top model.
To enable the comparison of the relative influence of predictors on carbon density, I plotted
standardized versions of the coefficients from the top-ranking models. This was done by scaling
and centering the predictor variables, which accounted for the different units with which they
were quantified. To maintain interpretability, I reported only the raw unstandardized
coefficients, which therefore related directly to the units of both carbon density and each
respective predictor variable.
I checked model validity by evaluating the assumptions of linear mixed effects regression
models. The residuals were normally distributed, but there were some issues of multicollinearity
for the predictor variables. In these cases, I ran the model again without the variable causing the
collinear issue (Tabachnick and Fidell 2001). Influential points were inspected and in most cases
determined to be within the reasonable bounds of the dataset. Any dubious outliers were
omitted. All analyses were performed using Stata 11 (StataCorp 2009b). Figures were generated
37

using Stata (StataCorp 2009b), Microsoft Excel 2010 (Version 14.0.7166.5000), and Microsoft
PowerPoint (Version 14.0.7166.5000).

RESULTS

Wood density as a means to quantify carbon density
Across the range of observed wood decomposition, wood density was a very good predictor of
carbon (C) density (r2 = 1.00, p < 0.001 , n=47). Carbon density decreased by 0.4996 g cm-3 for
every 1 g cm-3 decrease of wood density (Table 2, Figure 2). This relationship is described by
the equation:

de = 0.4996 X dw + 0.0043
where de is the density of carbon (g C cm-3 sample), 0.4996 is the slope, dw is the basic density of
wood (g wood cm-3 sample), and 0.0043 is they-intercept.
Table 2. Linear regression coefficients for carbon density versus wood density, and carbon
concentration versus wood density.
(n=47)

coefficient

standard error

p-value

95 % confidence interval

0.00339

<0.001

0.492 - 0.506

0.0102

<0.001

-0.0776 - -0.0365

carbon density versus wood density:
Wood density

0.4996

carbon concentration versus wood density:
Wood density

-0.05707

2

Wood density was also a good predictor of carbon concentration (r = 0.41 , p <0.001 , n=47). For
every 1 g cm-3 of wood density lost (Figure 3, right to left), carbon concentration increased by
0.057 g C t of wood (Table 2). The concentration of carbon in wood samples was slightly
1

38

0.3

0.25
:g'
0
0

3:

~

E
u

0.2

u

..

tlO

';: 0.15
"iii
C

QI

"Cl
C

0

...

6

Live

0

Snags

<> Suspended

0.1

..D
111

a Down

u

0.05

- - y = 0.4996x + 0.0044

R2 = 0.9979

0

0.0

0.1

0.3

0.2

0.4

0.5

0.6

Wood density (g cm·3)

Figure 2. The relationship between carbon density and wood density for all categories of
lodgepole pine samples, with a best-fitting trend line.
higher at lower wood density values (i.e. , higher carbon concentration with increasing decay),
although there were fewer data representing this end of the spectrum (Figure 3).
The second regression demonstrated the relative stability of carbon concentration within the
range of observed wood densities. Basic wood density ranged from 0.48 g cm·3 down to 0.11 g
cm·3, which represented a loss of about 76 % of the initial wood density. In contrast, over this
broad range of wood density, carbon concentration remained relatively constant, within a margin
of 7.5 % (from 0.50 to 0.54 g C g· 1 wood) (see Table Cl in Appendix C for wood and carbon
densities of the full dataset, with carbon concentrations for comparison).

39

0.6
DD D

:;- 0.5

D

0

0

...3:

b.o 0.4
u

..e!!
C

.....
0

~ 0.3

A Live

C
QI
u

Snags

C

0

C

0

<> Suspended

u"' 0.1

D

8 0.2

...

.c

Down

-- y

0

0.0

0.1

0.3

0.2

0.4

=-0.0571x + 0.5341
R2 =0.41
0.5

0.6

Wood density (g cm·3 )

Figure 3. Relationship between carbon concentration and wood density for all categories of
lodgepole pine.

Carbon density loss from decaying MPH-killed wood

Carbon density by position categories
Mean carbon density of samples from suspended and down MPB-killed CWD was significantly
lower than that of live lodgepole pine or upright MPB-killed snags (Table 3). Suspended and
down CWD were also significantly different from each other, with the latter having the lowest
average carbon density. Carbon density of snags did not differ from that of live lodgepole pine
(Table 3), although the range of observed carbon density was larger for snags (Figure 4). Live
lodgepole pine trees, which were sampled to provide a baseline starting point of carbon density
for decayed boles, had a mean carbon density of0.247 (± 0. 014) g C cm-3 of sample wood
(Table 3).

40

Table 3. Scheffe's test comparing mean carbon density for position categories, including live
wood as a baseline maximum. Different superscripted lowercase letters indicate significant
differences.
Mean C density
Tree category

n

(g C · cm-3)

St. err.

95 %CI

Live

10

0.247a

0.004

0.237 - 0.257

Snags

45

0.225 a

0.003

0.218 - 0.231

Suspended

116

0.196b

0.003

0.191-0.202

Down

69

0.161 C

0.007

0.148 - 0.175

0.300

0.250

-

•
T
!
•

•

•
•

"°E' 0.200

T

u

u

~

>

•

~ 0.150
C
CII

,::,

•

C

0

..D

:0 0.100

u

0 .050

0.000

•

n = 10

n -49

n=ll3

n -

Live

Snags

Suspended

Down

/;R

Figure 4. Mean carbon density± standard error for dead-tree position categories, with live-tree
carbon density as a baseline for comparison. Minimum and maximum observations for each
category are indicated with dots.

41

The variance of carbon density was greater within tree position categories than it was among
them (Table 4), and the within-group variability increased with each category as evidenced by
the maxima and minima for snags, suspended, and down boles, respectively (Figure 4, see also
Appendix C, Figure Cl for the variability in raw data by position categories).
Table 4. ANOV A comparison of carbon density by position categories.

ss

MS

df

F

Prob > F

Among groups

0.147

0.049

3

35.13

<0.0001

Within groups

0.330

0.001

236

Total

0.478

0.002

Best candidate models and predictors of carbon loss
Snags

For snags, the ' null ' model ranked highest, which indicated that none of the candidate models
explained an appreciable amount of the observed variance of carbon density compared to a
model with no explanatory predictors (Table 5). The lack of correlation between carbon density
and the predictor variables considered suggests that none of these factors strongly influence the
loss of carbon in snags (Appendix D, Figure Dl). In other words, the carbon density of snags
remains relatively constant over time, regardless of the variability in mean annual or seasonal
precipitation and temperature.

42

Table 5. Candidate models and AICc ranking for snags. The null model ranked highest, and no
other models ranked within :S 2 ~AI Cc scores of the null model. Random effects are italicized.
Candidate models for snags (n=43): *

Model#

K

AI Cc

~AI Cc

Null model: site

01

3

-195.4

0

yrsdead+site

11

4

-193 .0

2.4

yrsdead+MAP+MAT+site

21

6

-190.1

5.3

yrsdead+PPTwt+Twt+site

22

6

-191.8

3.6

yrsdead+ PPTsm+Tsm+ site

23

6

-189.9

5.5

yrsdead+BEC4+site

24

5

-193.1

2.3

yrsdead+MAP+MA T+yrsdead*MAP+site

41

7

-187.6

7.8

Model parameter definitions: yrsdead = years dead, MAP = mean annual precipitation (mm), MAT = mean annual
temperature (0 C), PPTwt = mean winter precipitation (mm), Twt = mean winter temperature (0 C), PPTsm = mean
summer precipitation (mm), Tsm = mean summer temperature (°C), BEC4 = biogeoclimatic ecosystem
classification with 4 subzone categories.

The random effect of site was included in all models 11 for snags, which accounted for a very
small portion of the observed variance. In all cases, the within-site variance was greater than the
between-site variance. Regression coefficients for the null model are provided in Table Dl ,
Appendix D.
Suspended CWD

For suspended CWD, sample height off the ground, mean summer precipitation and temperature,
and years dead were all good predictors of carbon density. None of the candidate models
containing years down, mean annual or mean winter climate variables, or BEC subzones ranked
highly (Table 6). The top two models contained the same suite of predictors, with the secondbest model including the addition of an interaction term for years dead and mean summer
11

Samples were nested directly in sites as the random effect for the snag category because only one sample was
taken per snag.

43

precipitation (Table 7). The benefit of adding the interaction term in the second-best model did
not outweigh the cost of adding another parameter.
Table 6. Candidate models and AICc ranking for suspended CWD. Top-ranked models have
~AI Cc scores within :S 2 of the best model (highlighted in bold). Random effects are italicized.
Model#

K

AI Cc

~AI Cc

Null model: site+bole

01

4

-416.2

22.8

yrsdead+site+bole

11

5

-424.6

14.3

yrsdown+s ite+bole

12

5

-423.7

15.2

yrsdead+selev+site+bole

21

6

-430.9

8

yrsdead+selev+MAP+MA T+site+bole

22

8

-436.4

2.6

yrsdead+selev+PPTwt+site+bo/et

23

7

-434.6

4.3

yrsdead+selev+ PPTsm+ Tsm+site+bole

24

8

-438.9

0

yrsdead+selev+BEC4+site+bo/e

25

7

-430.4

8.6

yrsdown+selev+site+bole

31

6

-428 .7

10.3

yrsdown+selev+MAP+MA T+site+bole

32

8

-433.4

5.6

yrsdown+selev+PPTwt+Twt+site+bole

33

8

-434.3

4.6

yrsdown+selev+PPTsm+Tsm+site+bole

34

8

-434.7

4.2

yrsdown+selev+BEC4+site+bole

35

7

-427 .7

11.3

yrsdead+selev+MAP+MAT +yrsdead *MAP+site+bole

41

9

-436.0

3.0

yrsdead+selev+PPTsm+Tsm+yrsdead*PPTsm+site+bo/e

42

9

-437.2

1.7

Candidate models for suspended CWD (n=99): •

Model parameter definitions are the same as in Table 5, with the addition of: yrsdown = years down, and selev =
sample elevation (cm).
t Winter temperature dropped from candidate model for multicollinearity issues.

44

Table 7. Regression coefficients of top-ranked linear mixed models for MPB-killed lodgepole
pine CWD in a suspended position. Continued next page.
model 1

Coeff.

St. Err.

95 % C. I.

years dead

-0.000551

0.000183

-0.000910 - -0.000192

sample elevation

0.000442

0.000121

0.000205 - 0.000680

precipitation(s)

-0.000369

0.0000902

-0.000546 - -0.000192

temperature(s)

0.00764

0.00318

0.00141 - 0.0139

(intercept)

0.138

0.0394

0.0608 - 0.215

4.86 X 10-I O

4.05 X 10-09

3.97 X 10-l?S - 0.00595

boles (nested in sites)

0.0150

0.00353

0.00943 - 0.0237

st. dev. (residual)

0.0196

0.00229

0.0156 - 0.0247

fixed effects*:

random effects:
site

Definitions and units : sample elevation (cm), precipitationcs) = mean summer precipitation (mm), temperature(s) =
mean summer temperature ( 0 C).

In both top-ranking models for suspended CWD, sample elevation had the greatest influence on
carbon density, followed by mean summer precipitation, years dead, and mean summer
temperature (Figure 5). Sample elevation and mean summer temperature were positively
correlated with carbon density (Table 7). Summer precipitation and years dead both had
negative associations with carbon density. The interaction term for years dead and summer
precipitation was generally positive, meaning higher precipitation over time yielded greater loss
of carbon density, but the pattern was not consistently positive for all cases (Figure 5).
With all other parameters in the best-ranked model held steady, carbon density was 0.000442 ±
0.000238 g cm-3 greater for every centimeter of sample height off the ground (Table 7).
45

Table 7 continued. Regression coefficients of top-ranked linear mixed models for MPB-killed
lodgepole pine CWD in a suspended position.
Coe ff.

St. Err.

95 % C. I.

years dead

-0.00113

0.000716

-0.00254 - 0.000269

sample elevation

0.000453

0.000121

0.000214 - 0.000691

precipitatioD(s)

-0.000473

0.000153

-0.000772 - -0.000174

temperature(s)

0.00734

0.00319

0.00110 - 0.0136

5.09 X 10"06

6.04 X 10"06

-6.74 X 10·06 - 0.0000169

0.154

0.0434

0.0686 - 0.239

6.48 X 10-I O

6.90 X 10·09

5.57 X 10" 19 - 0.753

boles (nested in sites)

0.0148

0.00393

0.00881 - 0.0249

st. dev. (residual)

0.0196

0.00293

0.0146 - 0.0263

model 2

fixed effects*:

years dead * preciPcs)
(intercept)
random effects:
site

Definitions and units: sample elevation (cm), precipitationc,i = mean summer precipitation (mm), temperature(,)=
mean summer temperature ( 0 C).

In other words, carbon density was 0.18 ± 0.096 % closer to that of live lodgepole pine with
every centimeter of sample elevation off the ground. Likewise, for every millimeter increase in
mean summer precipitation, carbon density decreased by -0.000369 ± 0.000177 g C cm·3 (-0.15
± 0.071 % of live lodgepole pine carbon density).

Suspended boles lost carbon density at a rate

of -0.000551 ± 0.000359 g C cm·3 per year. This is equivalent to an average loss of -0.22 ±
0.074 % carbon density per year, as represented by these data for the 70-year period following
tree death. Carbon density was 0.00764 ± 0.00624 g C cm·3 (3.09 ± 2.52 %) higher for every
increase of 1 °C (Table 7).

46

0.025
0 .02

-u
0 .015
*"'I

~

0 .01

.!!!

0 .005

ti=Q)

0

...."'C
u

0
u

H

-o
-0 .005
Q)
.!::!
~

ro

-0.01

""O

~ -0.015
ti

-0 .02

-0.025

H

•

years dead

sample elevation
(cm)

precipitatiOIJcs)
(mm)

temperature(,)

Model #2: 0

years dead

sample elevation
(cm)

precipitatiOIJcs)
(mm)

temperature(,)
(OC)

Model # 1:

(°C)

years dead *
precipitatiOIJcs)

Figure 5. Standardized coefficients (with 95 % C.I.) of the top-ranked models for carbon density
in suspended CWD. The further a standardized coefficient is from zero, the better predictor it is
of carbon density. Sample elevation had the greatest relative effect on carbon density, followed
by summer precipitation, years dead, and summer temperature. The confidence interval for the
interaction term in model #2 included zero and was therefore not a strong predictor.
The range of mean summer temperature across the study area was only 3.3 °C (11 to 14.3 °C), so
the relative effect of temperature on carbon density was small compared to other parameters,
despite the impression given by a high percentage rate of change per °C (Figure 5 illustrates the
relative influence of the predictor variables on carbon density in the top candidate models for
suspended CWD).

47

Down CWD

For down CWD, four candidate models ranked within 2 scores of b.AICc = 0 (Table 8), and each
of these four models included years down as a leading predictor variable. Sample elevation and
years down were the only predictors of carbon density in the top-rated model. Sample elevation
and years down therefore explained more variance together than the combination of predictors in
the subordinate-ranking models. Years down, sample elevation, and mean winter precipitation
and temperature were explanatory variables in the second and third ranking models. Model #3
also included an interaction term between years down and winter precipitation, but this term did
not improve the amount of explained variance beyond the cost of an additional parameter. The
fourth-ranked model had only years down as the sole predictor variable. Years dead, mean
annual and mean summer climate variables, and BEC subzones were all excluded from the bestranked models (Table 8), suggesting that none of these predictors explained appreciable amounts
of the observed variance in carbon density for down CWD.
Sample elevation was the best predictor of carbon density in down CWD for all models in which
it occurred. Mean winter temperature was the second-best predictor, followed by years down
(Figure 6). Sample elevation and mean winter temperature were both positively affiliated with
carbon density. Years down and mean winter precipitation were negatively associated with
carbon density, although precipitation did not consistently have a negative association. The
interaction term of years down and winter precipitation was generally positive, meaning the
negative effect of one amplified the negative effect of the other, but the pattern was inconsistent.

48

Table 8. Candidate models and AICc ranking for down CWD. Top-ranked models have ~AICc
scores within< 2 of the best model (highlighted in bold). Random effects are italicized.
Model#

K

AI Cc

~AI Cc

Null model : site+bole

01

4

-I 78.7

16.9

yrsdead+site+bole

II

5

-185.0

10.6

yrsdown+site+bole

12

5

-193.9

1.7

yrsdead+selev+site+bole

21

6

-I 86.5

9.1

yrsdead+selev+MAP+MAT+site+bole

22

8

-181.3

14.3

yrsdead+selev+ PPTwt+site+bolet

23

7

-184.0

11.6

yrsdead+selev+ PPTsm+ Tsm+s ite+bole

24

8

-181.2

14.4

yrsdead+selev+BEC4+site+bole

25

7

-I 83.9

11.7

yrsdown+selev+site+bole

31

6

-195.6

0

yrsdown+selev+MAP+MA T+site+bole

32

8

-190.6

5

yrsdown+selev+PPTwt+Twt+site+bole

33

8

-194.9

0.7

yrsdown+selev+ PPTsm+Tsm+s ite+bole

34

8

-I 90.3

5.3

yrsdown+selev+BEC4+site+bole

35

7

-193. I

2.5

yrsdead+selev+MAP+MA T+yrsdead *MAP+site+bole

41

9

-I 79.1

16.5

yrsdown+selev+PPTwt+Twt+yrsdown *PPTwt+site+bole

42

9

-194.8

0.8

Candidate models·
Down boles (n=56):

Model parameter definitions are the same as in Table 5, with the addition of: yrsdown = years down, and selev =
sample elevation (cm).
t Winter temperature dropped from candidate model for multicollinearity issues .
In the top-ranked model for down CWD, carbon density was 0.00182 ± 0.00170 g cm-3 higher
for every centimeter of elevation off the forest floor (Table 9). This translated to a loss of -0. 74
± 0.65 % carbon density for every centimeter that a bole was closer to or sinking into the duff.

Down CWD lost carbon density at a rate of -0.00298 ± 0.0012 g C cm-3 yr- 1 since tree fall; this
49

was equivalent to a loss of -1.21 ± 0.27 % carbon density per year. In the second-ranked model,
higher mean winter temperatures were affiliated with higher carbon density by a margin of
0.0422 ± 0.0298 g cm-3, or 17.1 ± 12.1 % for every 1 °C increase in temperature (Table 9).
0 .12

-u

0 .1

*- 0 .08
VI
C)

-:; 0 .06

....C:

-~
u 0 .04

!f

ti=

Cl)

0

u

0 .02

-0
Cl)

.!:!

"E

0

tHt

ro

-g-0.0 2
ro

t;;

-0.04
-0.06

Model # 1: •

years down

sample elevation
(cm)

Model #2: 0

years down

sample elevation
(cm)

precipitatiO°(w)
(mm)

temperature(w)
(OC)

Model #3: •

years down

sample elevation
(cm)

precipitatiO°(w)
(mm)

temperature(w)
(°C)

e

years down

Model #4 :

years dead *
precipitatiO°(w)

Figure 6. Standardized coefficients (with 95 % C.I.) of the four top-ranked models for
carbon density in down CWD. Sample elevation had the greatest relative influence on
carbon density, followed by years down, and mean winter temperature. Mean winter
precipitation included zero and was therefore a less reliable predictor, as was the interaction
term in model #3.
The effect of mean winter temperature appeared large, but the range across the study area was
only 2.5 °C (-8.9 to -6.4 °C); therefore, the coefficient affiliated with temperature described the
variance in only 2.5 units, whereas the coefficients of other predictors expressed the variance
over broader ranges (e.g., 23 cm for sample elevation, 32 for years down). Once standardized,
however, the effect of temperature relative to the other predictors was not large (Figure 6).
50

Table 9. Regression coefficients of top-ranked linear mixed models for MPB-killed lodgepole
pine CWD resting on the ground. Continued next page.

Model #1

Coeff.

St. Err.

95 % C. I.

years down

-0.00298

0.000622

-0.00420 - -0.00176

sample elevation

0.00182

0.000869

0.000121 - 0.00353

(intercept)

0.193

0.0112

0.171-0.215

site

0.0200

0.00858

0.00859 - 0.0464

boles (nested in sites)

0.0121

0.0152

0.00103 -0.142

st. dev. (residual)

0.0323

0.00612

0.0223 - 0.0468

years down

-0.00198

0.000737

-0.00342 - -0.000537

sample elevation

0.00176

0.000824

0.000143 - 0.00337

precipitation(w)

-0.000161

0.000124

-0.000405 - 0.0000826

temperature(w)

0.0422

0.0152

0.0124 - 0.0720

(intercept)

0.523

0.122

0.283 - 0.763

1.08 X 10-ll

1.38 X 10-l l

1.50 X 10-23 - 0.0783

boles (nested in sites)

0.0191

0.0124

0.00537 - 0.0682

st. dev. (residual)

0.0306

0.00799

0.0183 - 0.0540

fixed effects*:

random effects:

Model#2
fixed effects*:

random effects:
site

*Definitions and units : sample elevation (cm), precipitation(w) = mean winter precipitation (mm), temperature(w) =
mean winter temperature (°C).

51

Table 9 continued. Regression coefficients of top-ranked linear mixed models for MPB-killed
lodgepole pine CWD resting on the ground.
Coe ff.

St. Err.

95 % C. I.

years down

-0.00469

0.00175

-0.00812 - -0.00126

sample elevation

0.00190

0.000810

0.000312 - 0.00349

precipitation(w)

-0.000345

0.000163

-0.000665 - -0.0000246

temperature(w)

0.0342

0.0153

0.00418 - 0.0643

0.0000185

0.000011

-3.03 X 10-06 - 0.0000401

0.490

0.119

0.257 - 0. 724

6.95 X 10-ll

6.33 X 10-I O

1.22 X 10-IS - 0.00397

boles (nested in sites)

0.0166

0.0103

0.00494 - 0.0560

st. dev. (residual)

0.0309

0.00578

0.0214 - 0.0446

years down

-0.00303

0.000640

-0.00428 - -0.00177

(intercept)

0.193

0.0117

0.170 - 0.216

site

0.0217

0.00894

0.00969 - 0.0487

boles (nested in sites)

0.00989

0.0193

0.000215 - 0.0455

st. dev. (residual)

0.0342

0.00634

0.0237 - 0.0491

Model#3
fixed effects*:

years down*precip.(w)
(intercept)

random effects:
site

Model#4
fixed effects:

random effects:

Definitions and units: sample elevation (cm), precipitation(w) = mean winter precipitation (mm), temperature(w) =
mean winter temperature (0 C).

52

DISCUSSION

Wood density as a means to quantify carbon density
The strong relationship between wood density and carbon density supports the use of wood
density as a reliable measure of carbon loss from decaying lodgepole pine wood. For the range
of wood densities represented by the data (i.e., up to 80 % of total wood density lost), carbon
concentrations remained relatively stable. This stability of carbon concentration across the
observed spectrum of wood decomposition affirms the predictive capacity of wood density, and
supports its use in lieu of carbon content analysis.
The slight increase in carbon concentration at lower wood densities is consistent with the
findings for other temperate and boreal conifer species. Preston et al. ( 1998, 2006) and
Hishinuma et al. (2015) reported increasing carbon concentration with higher decay classes for
Douglas-fir (Pseudotsuga menziesii (Mirb.) menziesii Franco), western hemlock (Tsuga

heterophylla (Raf.) Sarg), and western red cedar (Thuja plicata (Donn.)) on Vancouver Island,
BC, and in Japan. Similarly, Norway spruce (Picea abies L. Karst), Scots pine (Pinus sy lvestris
L.), and birch (Betula pendula Roth and B. pubescens Ehrh) in Sweden (Sandstrom et al. 2007)
and Estonia (Koster et al. 2015) showed the same trend. Sakai et al. (2012) found that dead
wood of Japanese cedar (Cryptomerajaponica) D. Don. increased in carbon concentration with
advanced decay, but Japanese cypress (Chamaecyperus obtusa Sieb. Et Zucc.) Endl. did not.
The trend of increasing carbon concentration with advanced decay is attributable to the cell wall
chemistry of wood, as well as to the fungal groups that dominate different phases of
decomposition. Lodgepole pine wood is composed of approximately 43 % cellulose, 32 %
53

hemicellulose, and 25 % lignin (Koch 1996), and the corresponding carbon content in each of
these is approximately 44 % (McGinnis and Shafizadeh 1991 ), 40 % (Whistler and Chen 1991 ),
and 64 % (Chen 1991 , Karrnanov et al. 2014), respectively. Therefore, decomposition processes
that favour one or two of these components can lead to a shift in the overall carbon
concentration, as was observed in this study.
Of the two main fungal groups, white-rot fungi are capable of metabolizing lignin, cellulose, and
hemicellulose indiscriminately. Brown rots are not able to break down carbon-rich lignin, and
by selectively consuming cellulose and hemicellulose, brown rot fungi leave a lignin-rich
residual that is consequently higher in carbon concentration (Preston et al. 2006). White rots are
often the primary colonizers of dead wood (Schilling et al. 2015), and as a result, carbon
concentrations have been shown to remain more or less stable in early phases of decay (Preston
et al. 1998). Higher brown rot activity typically becomes more apparent in later phases of
decomposition, bringing about a shift to higher lignin and carbon concentrations. In contrast,
van der Val et al. (2007) and references therein suggested the opposite pattern: initial
colonization by fungi that can exploit easily-accessible cellulose and hemicellulose, followed by
later dominance of lignin-degrading fungi.
My results are consistent with a pattern of white-rot dominance in early stages of decay, and
higher brown-rot activity in later stages. This trend suggests that the sequence of fungal
colonization for MPB-killed lodgepole pine is generally typical of other temperate and boreal
softwood species (Preston et al. 1998 and 2006, Sandstrom et al. 2007), and that the presence of
blue-stain fungi introduced by MPB does not cause a divergence from this degradation pathway.

54

Carbon density loss from decaying MPB-killed wood
For each of the position categories ofMPB-killed lodgepole pine, a unique set of predictor
variables described the loss of carbon density, except in the case of snags, for which carbon
density was not well described by any of the predictor combinations considered. For both
suspended and down CWD, sample elevation was the best predictor of carbon density, but this
was the only factor in common between the two groups (Figure 7). Elements of time and climate
were important predictors for decay in suspended and down CWD, but different iterations of
these influences and their magnitudes revealed distinct conditions operating for the two fallen
position categories.
0.12
0 .1

*0 .08
U'l
u

~

....C
VI

.!!:!

0.06

u

ii:QJ 0 .04
0

u

"C

QJ

.!::!

....

"C
n:,
"C
C

0.02
0

n:,

ti

-0 .02

-0 .04

•

Suspended CWD

years dead

sample elevation
(cm)

precipitatiOil(s)
(mm)

temperaturecs)
(°C)

DownCWD

years down

sample elevation
(cm)

precipitatiOil(w)
(mm)

temperature(w)
(°C)

•

Figure 7. Relative influences of best predictor variables (with 95 % C.I.) for suspended and
down CWD. Coefficients are standardized for comparability. Note that the predictors of carbon
density loss are different for the two position categories with the exception of sample elevation
(e.g. , years dead versus years down, and summer versus winter climate variables).

55

Suspended and down CWD

Effect ofposition

With all other model predictors held steady, the decline in carbon density was about four times
greater for down CWD than for suspended CWD (0.74 % and 0.18 %, respectively) for every
centimeter closer that a bole was to the soil, or deeper that it was embedded into the duff (Figure
8). This observation supports the hypothesis that the position of dead boles is an important
factor controlling the influence of agents of decay.

M

• 27

. 6

•s

L()

-

N

M

• 12
• 15

E

--

(.) "!

•s •s

•s
30

. 7

()

C)

. 30

Z,.LD

·-en .......

. 60

C

<I>
"O
C

0 ......

..c
._

cc

(.)

. 36

• 32

LO

q

• 52

d l 25

0

-20

0

20

40

sample height (cm)

60

80

100

Figure 8. Carbon density (g C cm-3) loss with decreasing sample height off ground (right to left)
for suspended CWD (> 10 cm height, red diamonds) and down CWD (~ 10 cm height, purple
squares). Data labels indicate the number of years since tree death.

56

Many studies of wood decomposition have incorporated the distinction between snags and fallen
boles, but few, to my knowledge, have explicitly considered the effect of CWD height from the
ground as a determinant of decay. Some have anecdotally commented on elevated CWD without
quantifying any effects. Such was the case for Busse (1994), who conducted a study similar to
mine evaluating wood density in insect-killed lodgepole pine snags and CWD in Central Oregon.
In spite of lumping suspended and down CWD, Busse observed a remarkably concise
distribution of wood density over time since tree fall. However, the first and most notable result
Busse reported was the observation that decay class I boles that were elevated off the ground
showed no sign of decomposition. Busse (1994) also reported that of21 decay-class-I boles, 19
were in an elevated position, and all boles in higher decay classes were in contact with the
ground. Johnson and Greene (1991) similarly identified ' layers' of CWD which they associated
with distinct mortality assemblages (due to fire and self-thinning); bottom layers were from an
old fire disturbance assemblage, top layers consisted of the more recent self-thinning
assemblage, and middle layers were a combination of the two. The authors did not elaborate to
define the physical proximity of these respective layers to the soil surface, but a likely
assumption is that the top layers were further from ground and the bottom layers rested on the
ground. Despite the description of CWD layers, Johnson and Greene (1991) did not characterize
this feature quantitatively or discuss the potential influence the layering may have had on their
results.
Shorohova and Kapitsa (2014) characterized decay rates for a variety of dead wood positions,
among other factors, and found that leaning logs took longer to decompose than fallen logs. Of
the species considered, pine was the most resistant to decay. Van der Wal et al. (2007) found
that woody fragments buried in soil to a depth of 3 cm exhibited higher fungal growth than
57

samples incubated on the soil surface. My findings of accelerated carbon density loss for
samples closer to and embedded in the forest floor concur with the results reported by van der
Wal et al. (2007) and Shorohova and Kapitsa (2014). I chose 10 cm above the duff as an
arbitrary threshold to differentiate suspended and down CWD, but visual inspection of the data
after the fact suggests that 20 or 25 cm above the soil may be a more biologically meaningful
break point to distinguish these groups (Figure 8). The appearance of a break point at 20 or 25
cm from the forest floor may correspond to microhabitat factors such as average depth of
persistent snow cover during non-freezing weather and boundary layer effects of vegetation
cover in the study area.
Temporal effects

Time was a stronger predictor of decay in down CWD than it was for suspended CWD. Of the
two time factors evaluated, years since fall was a good predictor of carbon density loss in CWD
resting within 10 cm of the forest floor, but years since death did not rank highly in any of the
candidate models. By contrast, the time since fall was not at all correlated with decay in
suspended CWD, and years since death was only a minor predictor of decay (relative to the
effect of years since fall for down CWD). These findings for down CWD were consistent with
the trends reported by Storaunet and Rolstad (2002) and DeLong et al. (2005), who evaluated the
effectiveness of decay classes to estimate time since death and time since fall for Picea abies (L.)
Karst. (Storaunet and Rolstad 2002), and Picea glauca (Moench) Voss x Picea engelmannii
Parry ex Engelm. and Abies lasiocarpa (Hook.) Nutt. (DeLong et al. 2005). Both studies showed
that CWD decay classes were more correlated with time since fall than with time since death.
DeLong et al. (2005) suggested that the lack of explanatory power between decay class and time
58

since death was due to the variability in standing time for snags before they fall. This is
potentially the case for the suspended boles in my dataset. The fact that decay in suspended
CWD was not correlated with time since fall, and was only weakly related to time since death
suggests that for this dataset, suspended CWD may be as unlike down CWD as they are unlike
snags.
Climatic factors

Relative to sample elevation, the coefficients for climate variables were smaller but important for
carbon density loss in suspended and down boles, and each position category was affiliated with
different season-specific variables. Suspended CWD exposed to higher summer precipitation
were more decayed than their drier counterparts. Carbon densities were lower under wetter and
colder summer conditions, and higher under drier and warmer conditions. This was contrary to
Johnson and Greene ' s (1991) speculation that wetter cooler conditions limited the potential for
microbial activity, resulting in the slower decomposition that they observed in older stands.
Johnson and Greene ( 1991) attributed the wetter colder conditions to the forest structure of the
older stands, which were more dense with limited light penetration through the forest canopy. It
is possible that the conditions at my sites were less influenced by forest structure, and that for
suspended CWD, moisture was a limiting factor for decay. This would explain the positive
relationship with precipitation during the growing season.
Other studies have shown general patterns similar to mine with precipitation and temperature
being good predictors of decomposition. Trofymow et al. (2002) found that for the Canadian
Intersite Decomposition Experiment, temperature and summer precipitation were among the best
explanatory variables for litter decomposition (others being chemical traits of the decomposing
59

substrates). Weggler et al. (2012) reported that precipitation, temperature and evapotranspiration
explained changes in wood density for decay class 1 spruce CWD in Switzerland; precipitation
was also an important predictor of wood density in decay classes 2 and 3, but not in decay class
4, which had evapotranspiration alone as the best predictor. Similar to what Weggler et al.
(2012) reported for decay class 4, my data for carbon density in down CWD are not clearly
correlated with precipitation. Rather, my data show a positive trend with mean winter
temperature; e.g., samples exposed to colder winter temperatures were more decayed. The lack
of a clear influence of precipitation for down CWD may indicate that, with the closer proximity
to the forest floor, the moisture content of CWD - and therefore, the potential for microbial
activity - is regulated more by soil moisture than by precipitation. This would support my
hypothesis that grounded CWD are subject to up-wicking of soil moisture, easier colonization by
soil-borne agents of decay, as well as boundary layer effects of vegetation cover, temperature
moderation, and snow cover.
The pattern of higher carbon density at warmer seasonal temperatures for both suspended and
down CWD was not expected. It is possible that for suspended boles, warmer summer
temperatures facilitate drying of the substrate, and thus suppress the potential for microbial
activity. This would correspond with the notion that moisture is a limiting factor for
decomposition in MPB-affected forests, and explain why higher temperatures are affiliated with
less decay in wood. Boddy (1983b) identified optimal temperature ranges for activity of
decomposition fungi, but it is unlikely that the influence of temperature on fungal activity in
central BC would be on the right-hand side of such a bell-shaped temperature response curve,
especially for winter temperatures in the case of down CWD. One explanation for the apparent
trend of decreasing carbon density with decreasing temperatures may have to do with the
60

combination of temporal and geographic dynamics of the MPB outbreak in BC. The sites
associated with the 1980s MPB outbreak are in the southwest portion of the study area, which is
also the driest and coldest. Consequently, these sites have the oldest MPB attack, and few
sample boles from the same climate regime represent the more recent, less decayed cohort of
MPB-killed trees. On the wetter sites in the northeast part of the study area, the majority of
MPB-killed trees died more recently as the 1990-201 Os outbreak swept northeastward. The
effect of this temporal sequence across the geographic area likely explains some of the apparent
trend of more advanced decay on colder sites. This was a risk I tried to avoid during field
sampling by seeking a temporally diverse array of beetle-killed trees at each site; however, as is
often the case in studies of natural ecosystems, conditions are not necessarily conducive to ideal
scientific methods.

Snags
The evidence presented here supports the hypothesis that carbon density in MPB-killed snags
does not change considerably as long as boles remain upright. The persistence of carbon density
in snags was consistent with the findings of similar studies evaluating wood density. For insectkilled snags and live lodgepole pine in central Oregon, wood density was indistinguishable (0.43
± 0.04 and 0.44 ± 0.03 , respectively) (Busse 1994). Lodgepole pine snags that died by non-

insect causes (fire or self-thinning) in south-western Alberta also showed the same pattern of
stable mass density for up to thirty years after tree death (Johnson and Greene 1991). In my
dataset, the range of carbon density in sampled snags was more or less stable for a timespan of
approximately thirty years post-mortality (Appendix D, Figure Dl).

61

Although not significantly different, the broader range of snag carbon densities compared to live
lodgepole pine suggests that minor degradation and/or changes in wood properties do occur, at
least for some snags (Appendix C, Figure C 1). The lack of a trend in carbon density over time or
with the climate variables considered demonstrates that the capacity for decay in snags is limited,
and, that the decay that does occur is unrelated to these parameters. Lewis et al. (2006) found
that drying, blue stain, and checking were the main factors affecting wood quality in the first few
years following MPB attack, but degradation did not otherwise progress, except at the base of the
tree where moisture enabled continued microbial activity. The fact that carbon density of snags
does not show a pattern of decline over time confirms that the characteristics of snag wood at 1.3
m height (where I collected samples) are independent of the factors that dictate the likelihood of
falling.
Although there is much evidence supporting the hypothesis that lodgepole pine snags can remain
more or less sound for extended periods after mortality, it is important to recognize that the
individuals that persist are not representative of the whole population. Presumably, other snags
that have succumbed to more rapid degradation pathways are no longer standing. Thus, snag fall
rates are an integral component factoring into wood decomposition, which I discuss in more
detail later on.
Curve-fitting implications
Patterns of DOM decay over time have been described statistically with linear, negative
exponential (single or multiple), and sigmoidal (also called negative/reverse sigmoidal, or lagtime) models (Figure 9), as well as logarithmic variations. Of these, the exponential model
(Figure 9 B) is the most commonly used for wood (Olson 1963, Harmon et al. 1986), although
62

evidence suggests that models incorporating a time lag (i.e. , negative sigmoidal, Figure 9 C) may
represent the full process of decay more accurately (Boyce 1923, Buchanan and Englerth 1940,
Means et al. 1992, Busse 1994, Laiho and Prescott 2004 ).

A

1.0

0.5

OJ)

.!:

C:
·ce 1.0

5...

"'"'cc
E
ceC: 0.5

B

.......

.......

.......

.......

_

--- -- -- -- ----

...._ ...... slow

"oi)

·;:

,_0
0

C:

0

·;:;

....... 1.0
(.)

cc

---

C

0.5

---Time~

Figure 9. Three models for estimating wood decomposition (e.g., loss of mass or density) over
time: the linear model (A) corresponds to a constant rate of decay, regardless of the amount of
substrate. Single and multiple exponential models (B; solid and dashed lines, respectively)
represent constant rate(s) of decay relative to the amount of substrate available. The lag-time or
sigmoidal model (C) incorporates an initial period of minimal decay to reflect the time during
which decay agents become established in dead wood; this lag period transitions to an
exponential rate as optimal decay conditions evolve. Panels B and C adapted from Harmon et al.
(1986).

63

Despite efforts to compare the fit of different models, many cases appear ambiguous with
multiple models fitting a single dataset equally well. Graham and Cromack (1982) applied
linear, exponential, and logarithmic models to their chronosequence data for spruce and hemlock
decomposition in the Olympic Peninsula (Washington state, USA), and found that all three
models fit the data equally well. Similarly, Busse (1994) concluded that a single exponential
decay model fit chronosequence data for lodgepole pine in central Oregon very well, and,
recognizing the variability of decay rates over time, tested a double-exponential model to
accommodate distinctly different DOM components, but this did not improve the fit. In their
comparison of a single exponential and a summation-exponential model (equivalent to multiple
exponential) for Douglas-fir decomposition in the Western Cascades of Oregon, Means et al.
( 1985) found no conclusive improvement in fit with one model over the other.
Much of the uncertainty about wood decay rates may be attributed to contrasting interpretations
of the evidence of decay patterns over time. Though many studies have reported good fits with
the single exponential model, it assumes that the decay rate remains constant through time - an
assumption that is too simplistic for scenarios of major forest disturbances (Harmon et al. 2000)
and changing climate regimes. Laiho and Prescott (2004) pointed out that many datasets that
have been fit with exponential models would actually be better suited to a sigmoidal model,
incorporating a time lag before rapid onset of decay.
In contrast to the ambiguity in older literature about best-fitting decay curves, a number of more
recent studies decisively indicate that in some species, and in Pinus in particular, decomposition
follows a sigmoidal decay curve. The vector-based 12 observations made by Harmon et al. (2000)
12

The vector-method is described in more detail in the Background section.

64

yielded decay rates differing by an order of magnitude for different periods of decomposition.
They also found that the short-term decay rates of the vector method were higher than the overall
rates generated by the chronosequence method. By being lower than the vector-produced rates,
the chronosequence data suggest that earlier phases of the decomposition process were slower.
Harmon et al. (2000) therefore concluded that for Pinus sylvestris in northwestern Russia and
other species exhibiting similar patterns, the negative exponential model does not accurately
represent the variation of decay rates over time. Using a similarly novel approach, Freschet et al.
(2012) also found that a sigmoidal curve best described the loss of wood density for Pinus
sylvestris stems in northern Sweden. Laiho and Prescott (2004) pointed out that Busse' s (1994)

results for Pinus contorta appear better suited to a sigmoidal curve than the exponential curve
that was reported.
Although a comparison of different decay curves for the data in this thesis would be pertinent in
the context of current wood decomposition literature, it was beyond the scope of my research to
explore the subject statistically. Visually, the amount of noise in my data suggests that linear
versus exponential or sigmoidal comparisons may be inconclusive (Appendix D, Figure Dla,
Figure D2a, and Figure D3a). Conceptually, the rates of density loss for the three position
categories may be approximated to a stick-model version of a sigmoidal curve, as suggested by
Yatskov et al. (2003) for the different phases of decay. The negligible, slow, and faster rates of
density loss for snags, suspended, and down boles, respectively, would correspond to the initial
lag, onset, and more rapid phases of decay in the first half of a sigmoidal curve. Boddy (1983a)
evaluated linear and exponential models as part of her Ph.D. thesis and found that both models
described wood decay data equally well (Boddy, unpublished 1980), as many others have found.
Evaluation of different decay curves for this dataset therefore remains a project for future work.
65

In the meantime, following the rule of parsimony (Burnham and Anderson 2002) holds merit: the
simple linear models used here reveal some interesting patterns that warrant further
investigation.
For CWD decomposition data to clearly demonstrate the suitability of a sigmoidal (or
exponential) curve, sufficient observations must be made within the first 5 years of treefall
(Fahey 1983, Freschet et al. 2011), and/or tree death. This early lag period is often associated
with limited decomposition as wood becomes colonized by decay fungi and acquires optimal
moisture conditions (Laiho and Prescott 2004). Therefore, focusing sampling efforts on the early
years following tree death and fall would aid in the determination of a best-fitting curve
(Freschet et al. 2012). At the time of sampling, my primary objective was to acquire as broad a
range of decay conditions as possible, and the resultant minimum years since tree death in my
data are 3, 5, and 7 years for snags, suspended, and down CWD, respectively (Appendix C,
Table C2). Hence, the number of observations within the first 5 years of tree death is not large.
The addition of samples from early decay phases would likely make curve-fitting a more
conclusive proposition for my dataset.
To appreciate the importance of model fitting, consider the application of a simulation model for
forest carbon emissions using an exponential decay curve. In a simulation of the first 20 years
following a major forest disturbance, the decay process would initiate immediately with a high
decay rate, and tail off gradually. By contrast, the same 20-year simulation using a sigmoidal
decay curve would initiate with an early phase of low/slow carbon release, followed by a gradual
increase before the maximum decay rate would be achieved. The resulting prediction of carbon

66

emissions for the same 20-year period would be lower for the sigmoidal decay curve, compared
to the simulation using an exponential decay curve.
In the case of the CBM-CFS3 , predictions of forest carbon emissions are used to determine
Canada' s carbon budget, which emphasizes the need to use the best possible approximation of
reality. Therefore, the importance of curve-fitting is clear in the context of over- or underestimating carbon fluxes from wood decay over time.
MPB-killed wood density and the CBM-CFS3

Until recently, the CBM-CFS3 used exponential decay curves to simulate the loss of carbon from
all DOM in forests of all types and tree species across Canada. The four groups of carbon pools
that the CBM uses to categorize DOM include the very fast (foliar litter and fine roots), fast
(woody litter, branches, small stem wood, and roots), medium (large stem wood and bark), and
slow (humified organic matter) carbon pools (described in Appendix A, Table Al) (Smyth et al.
2010). Although exponential decay equations may adequately represent the very fast and fast
carbon pools, they overestimate the initial decomposition phase for the medium pool. With the
growing body of evidence supporting sigmoidal decay patterns for CWD (Laiho and Prescott
1999, Harmon et al. 2000, Yatskov et al. 2003 , Freschet et al. 2012), it has become clear that
exponential decay curves are not appropriate, especially considering that tree stem wood
accounts for the bulk of above-ground forest carbon (Johnstone 1970). In light of this evidence,
the CBM carbon accounting team initiated the transition to a sigmoidal decay curve for snags
and CWD in 2014-2015 (C. Smyth personal communication, 2015). While this potentially
represents a substantial refinement of decay parameterization for stem biomass in the medium
pool, the opportunity to integrate field-based decay rate observations remains.
67

Because my evaluation of carbon density loss in MPB-killed lodgepole pine was conducted
using linear regression models, the resultant decay rates are not directly comparable to the rates
(exponential or sigmoidal) used in the CBM-CFS3 . For the CBM simulation ofMPB impacts on
forest carbon balances (Kurz et al. 2008), the model parameters included an annual turnover rate
(i.e., transition from one carbon pool to another) of snag carbon to CWD carbon, in conjunction
with an annual release of snag carbon to the atmosphere. To simulate annual decay for fallen
CWD, portions of the CWD carbon were transferred to the slow decay pool and released to the
atmosphere. In order to relate my carbon density observations to the parameters used in the
CBM simulation for MPB, further work is necessary beyond the scope of this thesis. This will
involve converting my data to be applicable to biomass metrics for broader spatial applications.
An important factor addressed by the CBM but unaccounted for in my study is the transfer of

carbon to other more recalcitrant biomass pools, in addition to the assumed fate as CO 2 ; I
discuss this component of decomposition processes later in the thesis.
In earlier iterations of the the CBM, including the simulation for MPB, temperature was the only
climatic variable included to moderate decomposition (Kurz et al. 2008). Since then, results
from the Canadian Intersite Decomposition Experiment have been used to improve the predictive
capacity of the CBM (Smyth et al. 2010, 2011). Specifically, Smyth et al. (2010) reported that
inclusion of summer and winter precipitation would reduce the error of model estimates; this
moisture influence was further developed as a modifier for water stress effects under dry
conditions (Smyth et al. 2011 ). My observations of the influence of summer precipitation for
suspended CWD support the theory that decomposition is limited under low moisture conditions.

68

Bole positions and fall rates

The distinction between suspended and down CWD has had little recognition in wood
decomposition research. However, researchers in other fields of study have distinguished
between elevated logs and completely downed logs, and the ecosystem functions associated with
these two classes. Elevated CWD affect wildlife habitat value, serving as elevated runways for
intermediate-sized mammals (Delong et al 2005), and obstructing mobility of large mammals
(particularly ungulates) in cases of extreme blow-down events (Cichowski et al. 2008), for
example. Fire scientists and forest managers recognize the importance of classifying different
patterns and quantities of fuel loading because of the resultant effects on fire behaviour (Page
and Jenkins 2007).
Tree fall rates were outside the scope of this research; however, some observations may shed
light on factors associated with fall rates in MPB-killed stands in central BC. For example, the
range of carbon densities in snags could help parameterize the extent of potential decay for snags
in simulation modelling. The number of years that boles stood prior to falling (as determined for
all fallen CWD) may be an indicator of fall rates for these and other MPB-affected forests. Of
the 100 suspended and 58 down CWD samples assessed in this study, half had fallen within 13 to
14 years of tree death, and all boles had fallen by 50 to 75 years. (Appendix E, Figure El). In
simple terms this equates to annual fall rates of 3.5 to 3.9 % for the first half of all snags, which
are comparable to snag fall rates reported by DeLong et al. (2008) in wet spruce-fir forests not
noticeably affected by insect outbreaks. Fall rates in long-term MPB monitoring plots in central
BC were in the same range (C. DeLong and B. Rogers (MFLNRO), unpublished). In the current
dataset, snags persisting beyond 13 or 14 years fell at slower rates of 0.81 to 1.35 % per year.
69

The slower fall rates for the persistent snags were consistent with those reported by Lewis and
Hartley (2006) for MPB-killed pine in central BC. In contrast, fall rates have been higher (7 %
per year) on a nutrient-poor site with small lodgepole pine in east-central BC (D. Seip personal
communication, 2015).
Another observation from my dataset suggests that boles falling sooner are more likely to stay
suspended, and boles that remain standing longer are more likely to fall directly to the ground
(Appendix E, Figure El). This may indicate physical changes in persistent snags prior to falling,
e.g., loss of branches with which they might otherwise become hung up on other trees during
fall, or be supported off the ground in the absence of catching other trees on the way down.
Trees that remain standing longer may also succumb to extensive heart rot protected by a shell of
undecayed sapwood (case-hardening). When these trees do fall, they are more likely to break at
impact points along the stem and hence result in more ground contact.
The pattern of earlier-falling boles remaining suspended may reflect demographics of the stand
and the episodic nature of disturbances in lodgepole pine forests . When mass mortalities occur
as in an MPB outbreak, heavy wind events are likely to result in many boles falling at once (i.e.,
blowdown) into a complex matrix of suspended CWD. Higher stand densities may also
correspond to more CWD in a suspended position because of the greater likelihood of falling
snags intersecting other boles on the fall trajectory. In contrast, boles that persist longer as snags
may be the result of single-mortality events or legacies from earlier outbreak events.

70

Reflections for future consideration

Metrics of wood decomposition
Many approaches exist for estimating carbon losses from decaying wood, including various
forms and combinations of quantifying real-time CO2 emissions and monitoring changes in wood
characteristics over long periods (LID ET 1995, Jomura et al. 2008, Brown et al. 2010).
Quantifying the carbon remaining in coarse woody debris (CWD) is one method for estimating
carbon losses; however, this approach requires several labour-intensive steps as well as costly
elemental analyses (Laiho and Prescott 1999, Preston et al. 2006, 2009). Evaluating changes in
the density of dead wood is simpler and less costly, so this approach has been used extensively
alone and in conjunction with other parameters as a means of characterizing wood decay
(Graham and Cromack 1982, Johnson and Greene 1991 , Busse 1994, Paletto and Tosi 2010,
Sakai et al. 2012, Fraver et al. 2013). Fewer studies, however, have demonstrated the
relationship between wood density and carbon content per unit volume (i.e. , carbon density) in
decayed wood samples (Sollins et al. 1987). Expressing decay in terms of decayed mass per unit
green volume or green mass is an important distinction because it provides a way to compensate
for biases introduced by volume and/or mass losses in samples of advanced decay (Holub et al.
2001 , Fraver et al. 2013). In particular, carbon concentrations reported in many decomposition
studies only reflect the proportion of carbon relative to the concurrent composition of other
elemental constituents; these decayed proportions do not necessarily represent the chemical
profile of green wood, and may therefore be biased.

71

Significance of position categories and management implications
My sampling strategy was to sample one snag for every two fallen boles; it can be assumed that
the selection of suspended or down CWD was random, based on the availability of CWD in
these positions. Of the 15 5 samples from fallen boles, 99 were suspended and 56 were on the
ground. If the proportional representation of suspended to down CWD in my dataset reflects the
abundance of CWD in these respective positions on the landscape, then the slower decay rates of
suspended CWD should be taken into consideration when determining stand- or landscape-scale
carbon emissions from DOM.
The relative abundance and distinct decay patterns of suspended versus down CWD have
implications not only for forest carbon fluxes in unmanaged (i.e., not harvested) forests, which
were the focus of this thesis, but also for the management of CWD on harvested sites. Timber
harvesting is known to render forests net carbon sources for at least 8 years following cutting
(Fredeen et al. 2007), due to the removal of living biomass, fragmentation and accelerated decay
of residual dead woody material, and increased respiration of disturbed soil organic matter.
Therefore, efforts to retain snags and a higher proportion of suspended CWD (relative to down
CWD) on harvested sites may reduce the rate at which carbon is released from harvesting
disturbances. Encouraging harvesters to incorporate structural CWD complexity into harvesting
operations will aid in achieving this objective. Implementation should include avoidance of
crushing or fragmenting existing CWD, and leaving or creating suspended logs in and around
wildlife tree patches and throughout the cut area. Such a management strategy would enable the
carbon-fixing capacity of cut blocks to recover somewhat with the growth of vegetation and
planted trees before the progression of carbon release from decomposition gets underway.
72

This recommendation aligns with existing CWD management guidelines in BC for biodiversity
value on harvested sites. Work by Lloyd (2004) pointed to the importance of retaining snags
and longer and larger diameter pieces of CWD in elevated and leaning positions during logging
activity to improve habitat availability for wildlife, invertebrates, and vegetation whose
persistence depends on these structures. As stated in the Forest Practices Code Biodiversity
Guidebook, ' the more that managed forests resemble the forests that were established from
natural disturbances, the greater the probability that all native species and ecological processes
will be maintained ' (Parminter 1995). The BC Chief Forester further emphasized that leaving
snags, stubs (snags cut off at several metres height), and elevated logs of longer length (2 10 m)
and larger diameters (2 20 cm) on harvested sites will perpetuate diversity of CWD structures
over the course of secondary forest development (Snetsinger 2010). The slower decomposition
and higher abundance of suspended CWD that I observed in unmanaged forests thus identifies
another reason to manage for CWD complexity during salvage or regular harvesting operations.
Maintaining an array of CWD positions and age classes throughout the harvesting and regrowth
phases of managed landscapes will not only fulfill distinct ecological functions for biodiversity,
but will also moderate the release of carbon from wood decomposition.
Permanent sample plots established to monitor the long-term ecological effects of MPB in
unharvested forests are a complementary source of data to this thesis and to further specify CWD
management objectives. Observations from the Beetle Ecosystem Monitoring project (BEM, C.
DeLong and B. Rogers (MFLNRO), unpublished) quantify relative amounts of CWD by position
categories, residence times of suspended boles, and snag fall rates, which are important pieces in
the bigger picture of wood decay over time. The long-term BEM data together with my thesis
data will potentially provide a valuable quantitative basis with which to a) calibrate wood decay
73

simulations for calculating forest carbon fluxes, and b) refine objectives for CWD management
on harvested sites.

Other fates of carbon from decomposing wood
One common assumption in many decomposition studies, and more alarmingly in the context of
carbon simulation models, is the notion that all carbon lost from decaying wood is released
entirely as CO 2 to the atmosphere. Some authors have pointed out quite importantly that there
are alternative fates for the carbon stored in decomposing wood. Holub et al. (2001) and authors
cited therein suggest CWD mass losses are likely due to a combination of leaching, physical
removal by animals and invertebrates, fungal biomass, vegetation uptake, and fragmentation, in
addition to the commonly emphasized heterotrophic respiration. Decayed organic material can
make up a substantial and long-lasting portion of the forest floor in coniferous forests, especially
on wetter sites or in wetter climates (Sanborn et al. 2011 ).
Fragmented CWD will decay more quickly than entire logs because of the increase in exposed
surface area. Lambert et al. (1980) found that decay rates of fragmented Abies balsamea boles
were faster than those of unbroken boles. Graham and Cromack (1982) acknowledged that they
may have underestimated decay rates by not accounting for fragmentation. I observed multiple
stem breakages in a minor portion of my sampled boles, and deliberately avoided sampling
within 1 m of these fragmented zones. Hence, my results may underestimate decay rates by the
exclusion of fragmented wood in my sample inventory. Alternatively, my results may
overestimate decay rates if it is assumed that all carbon losses exited exclusively in the form of
CO2. Relative to other softwood species, pine wood tends to be more resistant to decay, and
thus, less susceptible to fragmentation during falling (Angers et al. 2010). Therefore, this

74

particular degradation pathway is likely to be variable and dependent on many factors, including
species-specific wood traits (Weedon et al. 2009).

Moisture content, volume determination, and implications for decay rate estimates
Although wood density is a simple metric for characterizing wood decay, it is influenced by a
number of factors that must be accounted for in order to avoid measurement errors, as pointed
out by Harmon et al. (2000), Laiho and Prescott (2004), and Williamson and Weimann (2010).
Because the volume of wood changes depending on its moisture content (Heygreen and Bowyer
1996), measurements of volume are specific to the moisture content of wood at the time of
measurement. Moisture contents should therefore be reported with wood densities for data to be
comparable among studies.
In this study, samples were dried at 102 °C as recommended by Williamson and Weimann
(2010) after being stored for several months, so the pre-oven weights did not reflect field MCs in
most cases. As a means of detecting the field MCs associated with field-based volumes, I
compared the percent of moisture lost from freezer-stored samples to that of the air-dried
samples. Freezer-stored samples did not change in volume or MC because they were wrapped in
plastic, and the water present in the wood was immobilized.
A second factor complicating the use of wood density is the loss of mass and volume occurring
with increased decay. As pointed out by Fraver et al. (2013), estimates of wood density for
chronosequence studies are prone to be biased if loss of volume is not accounted for in highly
decayed samples. Similarly, elemental concentrations of decayed samples do not reflect the
overall change relative to undecayed wood, because the original mass has been reduced (Holub
75

et al. 2001). Solutions to these issues include implementing a ' collapse ratio ' to extrapolate
green-wood volume from the horizontal diameter of decayed logs (Fraver et al. 2013), or
volume-adjusted concentrations, which express elemental concentrations of decayed wood
relative to the mass ofundecayed wood (Holub et al. 2001). Few of the sample logs in this study
were decayed to the point of losing volume through either fragmentation or collapsing. For those
that were, the original shape of each log was apparent enough that measurements for green-wood
volume were determined with confidence. By using a volume-based concentration of carbon
(i.e. , carbon density) as the primary indicator of wood decay, the issue of mass loss pointed out
by Holub et al. (2001) is avoided.
In retrospect, several methodological factors employed during this research caused undue
complexity for other portions of the study; therefore, the following points may be useful
considerations for future work:
•

Oven drying temperature is an important factor: it should be > 100 °C to ensure 100 % of
moisture is evaporated, but at the cost oflosing some C via volatiles (- 2 % of mass)
(Williamson and Weimann 2010). Freeze drying is an alternative to prevent loss of
volatile C compounds, but this option was limited by the size of the freeze dryer at
UNBC, which could not accommodate a full-size cookie, not to mention many of them.

•

Field volume is recommended, as was used here; however, field measurements should
include the associated weight of samples in the field to ensure proper determination of
moisture content. Measurement of field volume is also important to avoid the
complication of shrinkage from drier storage conditions.

76

•

Wedge samples for snags were not ideal because of the inability to measure volumes in
the field, and because of the potential biases introduced by having different proportions
of heartwood and sapwood compared to disc samples. I address the potential errors
associated with wedges in the next section. In general, to avoid a sequence of
complications, a consistent sampling strategy (e.g. , cookies) is advisable for all boles.

Sources of error
A weakness of this study was the lack of adequate sample size for the live and standing dead
categories. This limitation in the dataset was the result of not being qualified to fall trees in
order to collect full disc/cookie samples (the preferred sampling method used for suspended and
down logs). Wedge samples from live and dead standing trees were the compromise in order to
obtain some limited data from these categories. Potential sources of error due to this
compromise include different proportional amounts of heartwood and sapwood compared to
cookies, and differences in wood density, extractive content, and susceptibility to decay.
Heartwood and sapwood of lodgepole pine do not differ significantly in lignin content (Campbell
et al. 1990), so the disproportionate amount of heartwood to sapwood in wedges may not affect
density measurements relative to those of full discs based on lignin. However, Koch ( 1996)
reported that chemical composition of lodgepole pine varied by early wood and Iatewood, as
well as by sapwood and heartwood. Extractives, which generally reduce the susceptibility of
wood to decay, are found in higher concentrations in heartwood (Harmon et al. 1986). Fungal
activity is therefore likely to be quite different, and sapwood may succumb to decay before
heartwood. Woo et al. (2005) demonstrated that within 8 months of attack, physical and
chemical properties of MPB-killed lodgepole pine heartwood and sapwood were different from
77

those of green, un-attacked pine wood (although these results were based on samples from a
single tree in each category). Wedges from old snags may therefore have lower wood density
values than would be found in equivalent disc samples, due to the higher proportion of
extractive-poor sapwood in wedges compared to disc-shaped samples.
The effect of bole diameter on decay rates was not included as an explanatory variable in this
study (see Appendix C, Table Cl for summary statistics). Many authors include bole diameter
as a covariate when evaluating the effects of other predictors (Cornelissen et al. 2012), and many
studies identify diameter as an influential factor in wood decomposition. The decision to not
include diameter was based on two factors. First, my preliminary diagnostic evaluation of the
data suggested that a relationship did not exist between carbon density and sample diameter; an
observation that has been common (Harmon et al. 1986). Secondly, during the process of
constructing candidate models for AICc evaluation, it was important to limit the predictor
variables to a maximum of five per model in order to not exceed the allowable degrees of
freedom relative to the number of observations. It was also important to keep the number of
candidate models for each position category below twenty, as recommended by Burnham and
Anderson (2002). Given these circumstances, I chose to exclude sample diameter as a covariate.
My rationale for taking two samples from fallen boles was to capture the within-bole
heterogeneity of decay, and indeed, the extent of decay within individual logs spanned the
extremes from completely sound wood to advanced decay (Appendix E, Figure E2). Hence, the
wood densities of all samples plotted together over time had a much broader range of variability
than observed by Busse (1994). Another factor that may partially explain the range of wood
densities in my dataset is the inadvertent inclusion of branch nodes, scars, or reaction wood in

78

CWD cookies. Johnson and Greene ( 1991) pointed out that these features will skew density
measurements, presumably in the higher direction. I generally tried to avoid sampling wood that
was not clear-grained, knowing it may lead to the overestimation of sound wood, but I was also
conscious that to entirely exclude these features could potentially skew the dataset to lower
carbon densities. In the interest of representing the overall population, therefore, I did not
rigorously exclude wood with branch nodes or abnormal growth if it happened to occur at an
otherwise suitable sampling point along the bole.

CONCLUSIONS

Of all the aboveground biomass in forested ecosystems, tree boles constitute the largest reservoir
of forest carbon. Net forest carbon balances rest heavily on the relative amounts of living and
dead bole wood, and the demographic distribution of deadwood across all stages of decay. The
outbreak of MPB in central BC and the prediction of large carbon emissions from affected
forests highlight a lack of knowledge about regionally-specific wood decay rates. This
knowledge gap was the motivation for this project.
The objective of my research was to quantify the rate of carbon density loss from decaying
MPB-killed lodgepole pine bole wood, and characterize the decay process in terms of basic
physical and climatic attributes. In order to use carbon density as the primary metric of wood
decay, I first validated the assumption that carbon density can be readily determined from wood
density. Wood density was a very good predictor of carbon density (r 2 = 1.00); this relationship
was further verified by an observed low variability in carbon concentration across the range of a
76 % reduction in wood density.

79

The evidence presented here strongly suggests that standing, suspended, and down boles are
differentiated by rates of decay, as well as by differences in the importance and magnitude of
seasonal climate variables. These observations supported my first three hypotheses of limited
decomposition in snags, and progressively faster decay in suspended and down CWD. Bole
height off the ground was the single best predictor of carbon density for both suspended and
down CWD. Excluding the influence of other predictors, carbon density declined four times
faster with proximity to the soil in down CWD than it did for suspended CWD. Elevation of
CWD from the ground has not been acknowledged or quantified in most wood decay literature,
but the importance of this factor in the current dataset suggests that it is a key indicator of
decomposition processes, and should be considered in future work.
The pattern of lower carbon densities with increasing summer precipitation for suspended CWD
was consistent with the expectation of my fourth hypothesis. A rather surprising outcome
counter to my expectation was a positive influence of summer temperature on carbon density in
suspended CWD, which may correspond to a drying effect under warmer summer conditions,
thus limiting the capacity for microbial activity. The correlation of carbon density and climate
variables affirms recent research that also identified seasonal precipitation as a valuable predictor
of decay (Smyth et al. 2010).
Completely downed CWD lost carbon density more rapidly than suspended CWD, and this loss
was affiliated with years since fall. Years since death and summer climate variables were
completely excluded as useful explanatory variables for down CWD. Winter temperature had a
greater effect on carbon density than did precipitation, and this relationship was positive, as
summer temperature was for suspended CWD. This was counter to my expectation of more
80

decay with warmer temperatures. For down CWD, the demographic distribution across the study
area is mostly older CWD at colder sites, and mostly younger CWD at warmer sites, which could
produce the same trend. Because of the confounded nature of these data, drawing conclusions of
biological significance would be tenuous.
Carbon density in snags did not correlate well with any of the predictor variables considered,
which suggests that as long as boles remain upright, they are largely unaffected by climate
variability. This was consistent with what others have found (Johnson and Greene 1991 ), and
suggests that the factors influencing bole survival are largely related to decay at the base of the
tree, but not higher in the bole (Mitchel and Preisler 1998).
This thesis contributes a small but important piece of knowledge toward a better understanding
of the broader forest carbon flux picture. The future work of integrating these observations into
the decay component of the CBM-CFS3 and other forest carbon models is expected to improve
the accuracy of model predictions of forest carbon fluxes following insect outbreaks such as the
MPB. The accuracy of forest carbon flux predictions is vital for implementing the best possible
management of forests, not only for resource value but also for ecosystem integrity and
resilience.

81

REFERENCES

Alban, D. and Pastor, J. 1993 . Decomposition of aspen, spruce, and pine boles on two sites in
Minnesota. Canadian Journal of Forest Research 23: 1744-1749.
Angers, V.A. , Drapeau, P. , and Bergeron, Y. 2010. Snag degradation pathways of four North
American boreal tree species. Forest Ecology and Management 259: 246-256.
Axelson J.N., Alfaro, R.I. , and Hawkes, B.C. 2009. Influence of fire and mountain pine beetle on
the dynamics of lodgepole pine stands in British Columbia, Canada. Forest Ecology and
Management 257: 1874-1882.
Baldocchi, D. 2003. Assessing the eddy covariance technique for evaluating carbon dioxide
exchange rates of ecosystems: past, present and future . Global Change Biology 9: 479492.
Baldocchi, D., Hicks, B., and Meyers., T. 1988. Measuring biosphere-atmosphere exchanges of
biologically related gases with micrometeorological methods. Ecology 69: 1331-1340.
Barton, G.M. and Brownell, H.H. 1981. The chemistry of wood. In Canadian Woods: Their
Properties and Uses. Edited by E.J. Mullins and T.S. McKnight. University of Toronto
Press in cooperation with the Canadian Forest Service, Environment Canada, and the
Canadian Government Publishing Centre, Supply and Services Canada. pp. 97-127.
BC Ministry of Forests. 1998. Field Manual for Describing Terrestrial Ecosystems. BC Ministry
of Forests. Land Management Handbook Number 25.
Beets, P.N. , Hood, I.A. , Kimberley, M.O., Oliver, G.R. , Pearce, S.H., and Gardner, J.F. 2008.
Coarse woody debris decay rates for seven indigenous tree species in the central North
Island of New Zealand. Forest Ecology and Management 256: 548-557.
Boddy, L. 1983a. Carbon dioxide release from decomposing wood: effect of water content and
temperature. Soil Biology and Biochemistry 15(5): 501-510.
Boddy, L. 1983b. Effects of temperature and water potential on growth rate of wood-rotting
basidiomycetes. Transactions of the British Mycological Society 80(1): 141-149.
Bois, C.H., Janzen, D.T., Sanborn, P.T., and Fredeen, A.L. 2009. Contrasting total carbon stocks
between ecological site series in a sub-boreal spruce research forest in central British
Columbia. Canadian Journal of Forest Research 39(5): 897-907.
Bowler, R. , Fredeen, A.L. , Brown, M.G. , and Black, T.A. 2012. Residual vegetation importance
to net CO2 uptake in pine-dominated stands following mountain pine beetle attack in
central British Columbia, Canada. Forest Ecology and Management 269: 82-91.
Boyce, J. 1923 . The deterioration of felled western yellow pine on insect control projects. US
Department of Agriculture Department Bulletin No . 1140.

82

Brown, M. , Black, T.A. , Nesic, Z., Foord, V.N. , Spittlehouse, D.L., Fredeen, A.L. , Grant, N.J.,
Burton, P.J., and Trofymow, J.A. 2010. Impact of mountain pine beetle on the net
ecosystem production of lodgepole pine stands in British Columbia. Agricultural and
Forest Meteorology 150: 254-264.
Brown, M. , Black, T.A. , Nesic, Z. , Fredeen, A.L. , Foord, V.N., Spittlehouse, D.L., Bowler, R. ,
Burton, P.J. , Trofymow, J.A. , Grant, N.J., Lessard, D. 2012. The carbon balance of two
lodgepole pine stands recovering from mountain pine beetle attack in British Columbia.
Agricultural and Forest Meteorology 153: 82-93.
Brown, S., Mo, J. , McPherson, J.K. , and Bell, D.T. 1996. Decomposition of woody debris in
western Australian forests. Canadian Journal of Forest Research 26: 954-966.
Buchanan, T.S. , and Englerth, G.H. 1940. Decay and other volume losses in wind thrown timber
on the Olympic peninsula, Wash. US Department of Agriculture Technical Bulliten No.
733.
Bull, E. 1983 . Longevity of snags and their use by woodpeckers. In Snag Habitat Management
Symposium, Flagstaff, AZ, 7-9 June 1983. pp. 64-67.
Burnham, K.P. and Anderson, D.R. 2002. Model Selection and Multimodel Inference: A
Practical Information-Theoretic Approach, 2nd Ed. Springer Science+Business Media,
Inc. New York, NY. pp. 487.
Busse, M.D. 1994. Downed bole-wood decomposition in lodgpole pine forests of central
Oregon. Soil Science Society of America Journal 58 : 221-227.
Campbell, A.G. , Kim, W.J. , and Koch, P. 1990. Chemical variation in lodgepole pine with
sapwood/heartwood, stem height, and variety. Wood and Fibre Science 22(1 ): 22-30.
Carroll, A.L. , Regniere, J. , Logan, J.A. , Taylor, S.W. , Bentz, B., and Powell, J.A. 2006. Impacts
of climate change on range expansion by the mountain pine beetle. Natural Resources
Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria, BC. Mountain Pine
Beetle Working Paper 2006-14.
Carroll, A.L. and Safranyik, L. 2003 . The bionomics of the mountain pine beetle in Lodgepole
pine forests: establishing a context. In Mountain Pine Beetle Symposium: Challenges and
Solutions, Kelowna, B.C., 30-31 October 2003 . Edited by T.L. Shore, J.E. Brooks, and
J.E. Stone. Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre,
Information Report BC-X-399, Victoria, B.C. pp. 21-32.
CIDET. 1995. The Canadian Intersite Decomposition Experiment (CIDET): Long-term Rates of
Leaf Litter and Wood Decay. Pacific Forestry Centre. pp. 87-90.

83

Chapin, F.S., Woodwell, G.M. , Randerson, J.T. , Rastetter, E.B. , Lovett, G.M., Baldocchi, D.D.,
Clark, D.A., Harmon, M.E., Schimel, D.S. , Valentini , R. , Wirth, C., Aber, J.D. , Cole, 1.1. ,
Goulden, M.L. , Harden, J.W. , Heimann, M., Howarth, R.W., Matson, P.A., McGuire,
A.D., Melillo, J.M., Mooney, H.A. , Neff, J.C. , Houghton, R.A. , Pace, M.L. , Ryan, M.G.,
Running, S.W., Sala, O.E., Schlesinger, W.H. , and Schulze, E.D. 2006. Reconciling
carbon-cycle concepts, terminology, and methods. Ecosystems 9: 1041-1050.
Chen, C.L. 1991. Lignins: occurrence in woody tissues, isolation, reactions and structure. In
Wood Structure and Composition. Edited by M. Lewin and LS. Goldstein. Marcel
Dekker, Inc. New York, NY. pp. 238.
Chen, W., Chen, J., Liu, J. , and Cihlar, J. 2000a. Approaches for reducing uncertainties in
regional forest carbon balance. Global Biogeochemical Cycles 14(3):827-838.
Chen, H. , Harmon, M.E. , Griffiths, R.P. , and Hicks, W. 2000b. Effects of temperature and
moisture on carbon respired from decomposing woody roots. Forest Ecology and
Management 13 8: 51-64.
Cichowski, D., Williston, P., and Haeussler, S. 2008. The response of caribou terrestrial forage
lichens to mountain pine beetles and forest harvesting in the East Ootsa and Entiako
areas: Annual Report - 2007/08 - Year 7. Morice-Lakes Innovative Forest Practices
Agreement, Bulkley Valley Center for Natural Resources Research and Management, and
BC Ministry of Environment. pp. 46.
Cole, W.E. , and Amman, G.D. 1980. Mountain pine beetle dynamics in lodgepole pine forests
Part I: Course of an infestation. United States Department of Agriculture, Forest Service.
General Technical Report INT-89.
Comeau, P.G. and Kimmins, J.P. 1989. Above- and below-ground biomass and production of
lodgepole pine on sites with differing soil moisture regimes. Canadian Journal of Forest
Research 19:447-454.
Cornelissen, J.H.C., Sass-Klaassen, U. , Poorter, L. , van Geffen, K. , van Logtestijn, R.S.P. , van
Hal, J. , Goudzwaard, L. , Sterck, F.J., Klaassen, R.K.W.M., Freschet, G.T., van der Wal,
A. , Eshuis, H. , Zuo, 1, de Boer, W. , Lamers, T., Weemstra, M. , Cretin, V. , Martin, R. ,
den Ouden, J. , Berg, M.P. , Aerts, R. , Mohren, G.M.J., and Hefting, M.M. 2012. Controls
on coarse wood decay in temperate tree species: birth of the LOGLIFE Experiment.
Ambio 41 (supplement 3): 231-245 .
Cornwell, W.K. , Cornelissen, J.H.C. , Amatangelo, K. , Dorrepaal, E., Eviner, V.T. , Godoy, 0 .,
Hobbie, S.E., Hoorens, B., Kurokawa, H. , Perez-Harguindeguy, N. , Quested, H.M. ,
Santiago, L.S. , Wardle, D.A., Wright, I.J. , Aerts, R. , Allison, S.D., Van Bodegom, P.,
Brovkin, V., Chatain, A. , Callaghan, T.V. , Diaz, S., Garnier, E. , Gurvich, D.E., Kazakou,
E., Klein, J.A. , Read, J. , Reich, P.B., Soudzilovskaia, N.A. , Vaieretti, M.V. , and
Westoby, M. 2008. Plant species traits are the predominant control on litter
decomposition rates within biomes worldwide. Ecology Letters 11: 1065-1071.
84

Cornwell, W.K. , Cornelissen, J.H.C., Allison, S.D., Bauhus, J. , Eggleton, P., Preston, C.M. ,
Scarff, F., Weedon, J.T. Wirth, C., and Zanne, A. 2009. Plant traits and wood fates across
the globe: rotted, burned, or consumed? Global Change Biology 15(10): 2431-2449.
Cudmore, T.J. , Bjorklund, N. , Carroll, AL. , and Lindgren, B.S. 2010. Climate change and range
expansion of an aggressive bark beetle: evidence of higher beetle reproduction in nai"ve
host tree populations. Journal of Applied Ecology 47: 1036-1043.
DeLong, C. 2003. A field guide to site identification and interpretation for the southeast portion
of the Prince George Forest Region. Research Branch, BC Ministry of Forests, Victoria,
BC. Land Management Handbook No. 51.
DeLong, S.D., Daniels, L.D. , Heemskerk, B., and Storaunet, K.O. 2005. Temporal development
of decaying log habitats in wet spruce-fir stands in east-central British Columbia.
Canadian Journal of Forest Research 35: 2841-2850.
DeLong, S.C., Sutherland, G.D. , Daniels, L.D., Heemskerk, B.H., and Storaunet, K.O. 2008.
Temporal dynamics of snags and development of snag habitats in wet spruce-fir stands in
east-central British Columbia. Forest Ecology and Management 255: 3613-3620.
Deverall, B.F. 1965. The physical environment of fungal growth. 1. Temperature. In The Fungi.
I. The Fungal Cell. Edited by G.C. Ainsworth and A.S. Sussman. Academic Press, New
York. pp. 543-550.
Dolenko, A.J. , Shields, J.K. , King, F.W., Roff, J.W. , and Ostaff, D. 1981. Wood protection. In
Canadian Woods: Their Properties and Uses. Edited by E.J. Mullins and T.S. McKnight.
3rd ed. University of Toronto Press, Toronto, ON. pp. 176-223.
Dolman, A.J.H., Ronda, R., Miglietta, F., and Ciais, P. 2005. Regional measurement and
modelling of carbon balances. In The Carbon Balance of Forest Biomes. Edited by H.
Griffiths and P.G. Jarvis. Taylor & Francis Group, Scarborough, ON. pp. 93-108.
Emmel, C., Paul-Limoges, E. , Bowler, R. , Black, T.A. , and Christen, A. 2014. Vertical
distribution of carbon dioxide sources and sinks in a recovering mountain pine beetleattacked lodgepole pine stand. Agricultural and Forest Meteorology 195-196: 108-122.
Environment Canada. 2011. Greenhouse gas sources and sinks in Canada - Executive Summary.
Her Majesty the Queen in Right of Canada, represented by the Minister of the
Environment. National Inventory Report: 1990-2009.
Fahey, T. 1983. Nutrient dynamics of aboveground detritus in lodgepole pine (Pinus contorta
ssp. latifolia) ecosystems, southeastern Wyoming. Ecological Monographs 53(1 ): 51-72.
Field, C., Berry, J.A. , and Mooney, H.A. 1982. A portable system for measuring carbon dioxide
and water vapour exchange of leaves. Plant, cell and Environment 5: 179-186.
Fluxnet Global Network. 2013. http://fluxnet.ornl.gov/ [Accessed October 8, 2014]
85

Fraver, S., Milo, A.M. , Bradford, J.B. , D' Amato, A.W., Kenefic, L. , Palik, B.J., Woodall, C.W.,
and Brissette, J. 2013. Woody debris volume depletion through decay: implications for
biomass and carbon accounting. Ecosystems 16: 1262- 1272.
Fredeen, A.L. , Waughtal, J.D. , and Pypker, T.G. 2007. When do replanted sub-boreal clearcuts
become net sinks of CO2 ? Forest Ecology and Management 239(1-3) : 210-216.
Freshet, G.T. , Weedon, J.T., Aerts, R. , van Hal, J.R., and Cornelissen, J.H.C. 2012. Interspecific
differences in wood decay rates: insights from a new short-term method to study longterm wood decomposition. Journal of Ecology 100: 161-170.
Gholz, H.L. , Wedin, D.A., Smitherman, S.M., Harmon, M.E., and Parton, W.J. 2000. Long-term
dynamics of pine and hardwood litter in contrasting environments: toward a global model
of decomposition. Global Change Biology 6: 751-765.
Gotelli, N.J. , and Ellison, A.M. 2004. A Primer in Ecological Statistics. Sinaeur Associates, Inc.
Publishers, Sunderland, Massachusetts.
Grace, J. 2005. Role of forest biomes in the global carbon balance. In : The Carbon Balance of
Forest Biomes. Edited by H. Griffiths and P.G. Jarvis. Taylor & Francis Group,
Scarborough, ON. pp. 19-45.
Graham, R.L. and Cromack, K. , Jr. 1982. Mass, nutrient content, and decay rate of dead boles in
rain forests of Olympic National Park. Canadian Journal of Forest Research 12: 511-521.
Harmon, M.E., Franklin, J.F., Swanson, F.J. , Sollins, P. Gregory, S.V., Lattin, J.D. , Anderson,
N.H. , Cline, S.P. , Aumen, N .G., Sedell, J.R., Lienkaemper, G.W. , Cromack, K., Jr. , and
Cummins, K. W. 1986. Ecology of coarse woody debris in temperate ecosystems.
Advances in Ecological Research 15:133-302.
Harmon, M.E. , Krankina, O.N., and Sexton, J. 2000. Decomposition vectors: a new approach to
estimating woody detritus decomposition dynamics. Canadian Journal of Forest Research
30: 76-84.
Hattenschwiler, S., Tiunov, A.V ., and Scheu, S. 2005. Biodiversity and litter decomposition in
terrestrial ecosystems. Annual Review of Ecology, Evolution, and Systematics 36: 191218.
Haughian, S.R., Burton, P.J., Taylor, S.W., Curry, C.L. 2012. Expected effects of climate change
on forest disturbance regimes in British Columbia. Journal of Ecosystem Management,
13(1): 1-24.

86

Hawkes, B., Taylor, S.W. , Stockdale, C., Shore, T.L. , Alfaro, R.I. , Campbell, R. , and Vera, P.
2003. Impact of mountain pine beetle on stand dynamics in British Columbia. In
Mountain Pine Beetle Symposium: Challenges and Solutions, Kelowna, B.C., 30-31
October 2003. Edited by T.L. Shore, J.E. Brooks, and J.E. Stone. Natural Resources
Canada, Canadian Forest Service, Pacific Forestry Centre, Information Report BC-X399. pp. 177-199.
Heath, R. and Alfaro, R. 1990. Growth response of a Douglas-fir/lodgepole pine stand after
thinning of lodgepole pine by the mountain pine beetle. Journal of the Entomological
Society of British Columbia 87: 16-21.
Heinicke, A.J. and Childers, N .F. 193 7. The daily rate of photosynthesis, during the growing
season of 1935, of a young apple tree of bearing age. Cornell University Agricultural
Experiment Station. Memoir 201 .
Held, A.A. , Steduto, P., Orgaz, F., Matista, A. , Hsiao, T.C. 1990. Bowen ratio/energy balance
technique for estimating crop net CO2 assimilation, and comparison with a canopy
chamber. Theoretical and Applied Climatology 42(4): 203-213.
Heygreen, J.G. and Bowyer, J.L. 1996. Forest Products and Wood Science, an Introduction. 3rd
Ed. Iowa State University Press, Ames, Iowa. pp. 192.
Hicke, J.A. , Allen, C.D., Desai, A.R. , Dietze, M.C., Hall, R.F. , Hogg, E.H. , Kashian, D.M.,
Moore, D., Raffa, K.F., Sturrock, R.N. , and Vogelmann, J. 2012. Effects of biotic
disturbances on forest carbon cycling in the United States and Canada. Global Change
Biology 18: 7-34.
Hilker, T. , Coops, N.C. , Hall, F.G., Black, T.A. , Chen, B., Krishnan, P. , Wulder, M.A. , Sellers,
P.J. , Middleton, E.M. , and Huemmrich, K.F.2008a. A modeling approach for upscaling
gross ecosystem production to the landscape scale using remote sensing data. Journal of
Geophysical Research 113: G03006. doi:10.1029/2007JG000666.
Hishinuma, T. , Osono, T. , Fukasawa, Y. , Azuma, J.I. , and Takeda, H. 2015 . Application of 13 C
NMR spectroscopy to characterize organic chemical components of decomposing coarse
woody debris from different climatic regions. Annals of Forest Research 58(1): 3-13.
Holmes, R. 1983 . Computer-assisted quality control in tree-ring dating and measuring. TreeRing Bulletin 43: 69-78.
Holub, S.M. , Spears, J.D.H, Lajtha, K. 2001. A reanalysis of nutrient dynamics in coniferous
coarse woody debris. Canadian Journal of Forest Research 31: 1894-1902.
Husch, B., Beers, T.W. , and Kershaw, J.A. Jr. 2003. Forest Mensuration. John Wiley & Sons,
Inc. Hoboken, NJ. pp. 152.

87

IPCC (Intergovernmental Panel on Climate Change). 2007. Climate Change 2007: the Physical
Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the
Intergovermental Panel on Climate Change. Edited by S. Solomon, D. Qin, M. Manning,
Z. Chen, M. Marquis, K.B. Averyt, M. Tignor, and H.L. Miller. Cambridge University
Press, UK.
IPCC (Intergovernmental Panel on Climate Change). 2013. Climate Change 2013: The Physical
Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change. Edited by T.F. Stocker, D. Qin, G.-K.
Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M.
Midgley. Cambridge University Press, New York, NY.
Jenny, H. 1950. Causes of the high nitrogen and organic matter content of certain tropical forest
soils. Soil Science 69(1 ): 63-70.
Jenny, H. , Gessel, S.P., and Bingham, F.T. 1949. Comparative study of decomposition rates of
organic matter in temperate and tropical regions. Soil Science 68 : 419-432.
Johnson, E.A. and Greene, D.F. 1991. A method for studying dead bole dynamics in Pinus
contorta var. latifolia - Picea engelmannii forests . Journal of Vegetation Science 2: 523530.
Johnston, M. , Williamson, T. , Munson, A. , Ogden, A. , Moroni, M. , Parsons, R., Price, D., and
Stadt, J. 2010. Climate Change and Forest Management in Canada: Impacts, Adaptive
Capacity and Adaptation Options. A state of knowledge report. Sustainable Forest
Management Network.
Johnstone, W.D. 1970. Component dry weights of 100-year-old lodgepole pine trees. Canadian
Forestry Service, Department of Fisheries and Forestry. Information Report A-X-31.
Jomura, M. , Kominami, Y. , Dannoura, M. , and Kanazawa, Y. 2008. Spatial variation in
respiration from coarse woody debris in a temperate secondary broad-leaved forest in
Japan. Forest Ecology and Management 255:149-155.
Kaarik, A.A. 1974. Decomposition of wood. In Biology of Plant Litter Decomposition, Volume
1. Edited by C.H. Dickinson. Academic Press Inc. New York, NY. pp. 129.
Karmanov, A.P. , Borisenkov, M.F. , and Kocheva, L.S. 2014. Chemical structure and antioxidant
properties of lignins from conifer, broadleaf, and herbaceous plants. Chemistry of Natural
Compounds 50(4): 702-705.
Kim, W.J. , Campbell, A.G., and Koch, P. 1989. Chemical variation in lodgepole pine with
latitude, elevation, and diameter class. Forest Products Journal 39(3): 7-12.
Kirk, T.K. and Highley, T.L. 1973. Quantitative changes in structural components of conifer
woods during decay by white-and brown-rot fungi. Phytopathology 63(11): 1338-1342.

88

Koch, P. 1996. Chemical constituents. In Lodgepole Pine in North America, Volume 2, Part III:
Characterization of tree parts. Forest Products Society, Madison, WI. pp. 621-638.
Koster, K., Metslaid, M. , Engelhart, J., and Koster, E. 2015 . Dead wood basic density, and the
concentration of carbon and nitrogen for main tree species in managed hemiboreal
forests. Forest Ecology and Management 354: 35-42.
Kurz, W.A. and Apps, M.J. 2006. Developing Canada' s national forest carbon monitoring,
accounting and reporting system to meet the reporting requirements of the Kyoto
Protocol. Mitigation and Adaptation Strategies for Global Change 11: 33-34.
Kurz, W.A. , Apps, M.J. , Webb, T.M. , and McNamee, P.J. 1992. The carbon budget of the
Canadian forest sector: Phase 1. Forestry Canada, Northwest Region, Northern Forestry
Centre. Information Report NOR-X-326.
Kurz, W.A. , Dymond, C.C., Stinson, G., Rampley, G.J. , Neilson, E.T., Carroll, A.L. , Ebata, T.,
and Safranyik, L. 2008. Mountain pine beetle and forest carbon feedback to climate
change. Nature 452: 987-990.
Kurz, W.A. , Dymond, C.C. , White, T.M. , Stinson, G. , Shaw, C.H. , Rampley, G.J. , Smyth, C.,
Simpson, B.N., Neilson, E.T. , Trofymow, J.A. , Metsaranta, J. , and Apps, M .J. 2009.
CBM-CFS3: A model of carbon-dynamics in forestry and land-use change implementing
IPCC standards. Ecological Modelling 220: 480-504.
Laiho, R. and Prescott, C.E. 1999. The Contribution of coarse woody debris to carbon, nitrogen,
and phosphorus cycles in three Rocky Mountain coniferous forests. Canadian Journal of
Forest Research 29: 1592-1603 .
Laiho, R. and Prescott, C.E. 2004. Decay and nutrient dynamics of coarse woody debris in
northern coniferous forests: a synthesis. Canadian Journal of Forest Research 34: 763777.
Lambert, R. , Lang, G.E., and Reiners, W.A. 1980. Loss of mass and chemical change in
decaying boles of a subalpine balsam fir forest. Ecology 61(6): 1460-1473.
Lewis, K. and Hartley, I.D. 2006. Rate of deterioration, degrade, and fall of trees killed by
mountain pine beetle. Journal of Ecosystems Management 7(2):11-19.
Lewis, K. and Thompson, D. 2009. Change in wood quality and fall rate of trees up to ten years
after death from mountain pine beetle. Natural Resources Canada, Canadian Forest
Service Mountain pine beetle Working Paper 2008-30.
Lewis, K. , Thompson, D. , Hartley, I., and Pasca, S. 2006. Wood decay and degradation in
standing lodgepole pine (Pinus contorta var. latifolia Engelm.) killed by mountain pine
beetle (Dendroctonus ponderosae Hopkins: Coleoptera). Natural Resources Canada,
Canadian Forest Service, Pacific Forestry Centre. Mountain pine beetle initiative working
paper 2006-11. pp. 18.
89

LIDET (Long-Term Intersite Decomposition Experiment Team). 1995. Meeting the challenge of
long-term, broad-scale ecological experiments. Forest Research Laboratory, Oregon State
University, Paper 2964.
Lieu, P.J., Kelsey, R.G. , and Shafizadeh, F. 1979. Some chemical characteristics of green and
dead lodgepole pine and western white pine. US Department of Agriculture, Forest
Service, Intermountain Forest and Range Experiment Station. Research Note INT-256.
Little, S.N. and Shainsky, L.J. 1992. Distribution of biomass and nutrients in lodgepole
pine/bitterbrush ecosystems in central Oregon. U.S. Department of Agriculture, Forest
Service, Pacific Northwest Research Station. Research Paper PNW-RP-454.
Litton, C.M., Ryan, M.G. , Tinker, D.B., and Knight, D.H. 2003 . Belowground and aboveground
biomass in young postfire lodgepole pine forests of contrasting tree density. Canadian
Journal of Forest Research 33 : 351-363.
Lloyd, R. 2004. Results of operational trials to manage coarse woody debris in the Northern
Interior. Morice & Lakes Innovative Forest Practices Agreement, Ecosystem
Management Summary No. 35.
Lund, C.P., Riley, W.J., Pierce, L.L. , and Field, C.B. 1999. The effects of chamber
pressurization on soil-surface CO 2 flux and the implications for NEE measurements
under elevated CO2 . Global Change Biology 5: 269-281.
Maser, C. and Trappe, J.M. 1984. The seen and unseen world of the fallen tree. USDA Forest
Service, Pacific Northwest Forest and Range Experiment Station, General Technical
Report PNW-164, pp. 56.
Maser, C., Anderson, R.G., Cromack, K. , Williams, J.T. , and Martin, R.E. 1979. Dead and
down woody material. In Wildlife habitats in managed forests: the Blue Mountains of
Oregon and Washington. Edited by J.W. Thomas. US Department of Agriculture, Forest
Service. Washington, DC. Agriculture Handbook No. 553. pp. 78-95.
Matsuzaki, E. 2011. Carbon stocks of western red cedar and western hemlock stands in Canada' s
inland temperate rainforests. M.Sc. thesis, Natural Resources and Environmental Studies,
Forestry, University of Northern British Columbia, Prince George, BC.
McGinnis, G.D. , Shafizadeh, F. 1991. Cellulose. In Wood Structure and Composition. Edited by
M. Lewin and I.S. Goldstein. Marcel Dekker, Inc. New York, NY. pp. 146.
Means, J.E., Cromack, K. , Jr. , and MacMillan, P.C. 1985. Comparison of decomposition models
using wood density of Douglas-fir logs. Canadian Journal of Forest Research 15: 10921098.
Means, J.E., MacMillan, P.C., and Cromack, K. , Jr. 1992. Biomass and nutrient content of
Douglas-fir logs and other detrital pools in an old-growth forest, Oregon, U.S.A.
Canadian Journal of Forest Research 22: 1536-1546.

90

Meidinger, D., Pojar, J., and Harper, W.I. 1991. Sub-Boreal Spruce zone. In The Ecosystems of
British Columbia. Edited by D. Meidinger and J. Pojar. Special Report Series 6. BC
Ministry of Forests, Victoria, BC pp. 209-222.
MFLNRO (Ministry of Forests, Lands and Natural Resource Operations). 2013. Facts About
B. C.' s Mountain Pine Beetle [online].
http://www.for.gov.bc.ca/hfp/mountain pine beetle/Updated-BeetleFacts April2013.pdf (accessed 24 April 2014].
Minderman, G. 1968. Addition, decomposition and accumulation of organic matter in forests.
Journal of Ecology 56(2): 355-362.
Mitchell, R.G. , and Preisler, H.K. 1998. Fall rate of lodgepole pine killed by the mountain pine
beetle in central Oregon. Western Journal of Applied Forestry 13(1): 23-26.
Mitchell, R.G. , Martin, R.E., and Stuart, J. 1983. Catfaces on lodgepole pine - fire scars or strip
kills by the mountain pine beetle? Journal of Forestry 81(9): 598-613 .
Mooney, H.A. 1972. Carbon dioxide exchange of plants in natural environments. The Botanical
Reivew 38(3): 455-469.
Musgrave, R.B. and Moss, D.N. 1961. Photosynthesis under field conditions. I. A portable
closed system for determining net assimilation and respiration of com. Crop Science 1:
37-41.
Olson, J.S. 1963. Energy storage and the balance of producers and decomposers in ecological
systems. Ecology 44(2): 322-331.
Page, W.G. and Jenkins, M.J. 2007. Mountain pine beetle-induced changes to selected lodgepole
pine fuel complexes within the Intermountain Region. Forest Science 53( 4): 507-518.
Paletto, A., and Tosi, V. 2010. Deadwood density variation with decay class in seven tree species
of the Italian Alps. Scandinavian Journal of Forest Research 25: 164-173.
Parminter, J. (co-author and co-editor). 1995. Biodiversity Guidebook- Forest Practices Code of
British Columbia. BC Ministry of Forests and BC Environment, Victoria, B.C.
Pedersen, L. 2003. How serious is the mountain pine beetle problem? From a timber supply
perspective. In Mountain Pine Beetle Symposium: Challenges and Solutions, Kelowna,
B.C., 30-31 October 2003 . Edited by T.L. Shore, J.E. Brooks, and J.E. Stone. Natural
Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Information Report
BC-X-399, Victoria, BC. pp. 10-18.
Pojar, J., Klinka, K. and Meidinger, D.V. 1987. Biogeoclimatic ecosystem classification in
British Columbia. Forest Ecology and Management 22: 119-154.

91

Pousette, J. and Hawkins, C. 2006. An assessment of critical assumptions supporting the timber
supply modelling for mountain-pine-beetle-induced allowable cut uplift in the Prince
George Timber Supply Area. BC Journal of Ecosystems and Management, 7(2): 93-104.
Preston, C.M. , Trofymow, J.A. , Niu, J. , and Fyfe, C.A. 1998. 13 CPMAS-NMR spectroscopy and
chemical analysis of coarse woody debris in coastal forests of Vancouver Island. Forest
Ecology and Management 111 : 51-68.
13

Preston, C.M., Trofymow, J.A. , and Flanagan, L.B. 2006. Decomposition, 8 C, and the "lignin
paradox" . Canadian Journal of Soil Science 86: 235-245.
Preston, C.M. , Nault, J.R. , and Trofymow, J.A. , 2009. Chemical changes during 6 years of
decomposition of 11 litters in some Canadian forest sites. Part 2. 13 C abundance, solidstate 13 C NMR spectroscopy and the meaning of "lignin". Ecosystems 12: 1078-1102.
Rabe-Hesketh, S., and Skrondal, A. 2008. Multilevel and Longitudinal Modelling Using Stata.
2nd Ed. A Stata Press Publication, StataCorp LP, College Station, TX.
Rayner, A.D.M and Boddy, L. 1988. Fungal Decomposition of Wood: Its Biology and Ecology.
John Wiley & Sons, Toronto. pp. 587.
Rayner, A.D .M . and Todd, N.K. 1979. Population and community structure and dynamics of
fungi in decaying wood. Advances in Botanical Research 7: 333-420.
Rochette, P. , Gregorich, E.G., and Desjardin, R.L. 1992. Comparison of static and dynamic
closed chambers for measurement of soil respiration under field conditions. Canadian
Journal of Soil Science 72: 605-609.
Running, S.W., Nemani, R.R. , Heinsch, F.A. , Zhao, M.S. , Reeves, M. , and Hashimoto, H. 2004.
A continuous satellite-derived measure of global terrestrial primary production.
BioScience 54: 547-560.
Ryan, M.G. , Kashian, D.M. , Smithwick, E.A.H. , Romme, W.H. , Turner, M.G. , and Tinker, D.B.
2008. Carbon cycling at the landscape scale: the effect of changes in climate and fire
frequency on age distribution, stand structure, and net ecosystem production. USDA
Forest Service, JFSP Project 03-1-1-06 BLM.
Sakai, Y. , Ugawa, S., Ishizuka, S., Takahashi, M., and Takenaka, C. 2012. Wood density and
carbon and nitrogen concentrations in deadwood of Chamaecyparis obtusa and
Cryptomeriajaponica. Soil Science and Plant Nutrition 58(4): 526-537.
Sanborn, P. , Lamontagne, L. , and Hendershot, W. 2011. Podzolic soils of Canada: genesis,
distribution, and classification. Canadian Journal of Soil Science 91: 843-880.

92

Sandstrom, F., Petersson, H. , Kruys, N. , and Stahl, G. 2007. Biomass conversion factors (density
and carbon concentration) by decay classes for dead wood of Pinus sylvestris, Picea
abies and Betula spp. in boreal forests of Sweden. Forest Ecology and Management
243(1): 19-27.
Scheffer, T.C. 1973. Microbial degradation and the causal organisms. In Wood Deterioration and
Its Prevention by Preservative Treatment, Volume 1: Degradation and Protection of
Wood. Edited by D.D. Nicholas. Syracuse University Press, Syracuse, NY. pp. 31-106.
Schilling, J.S. , Ayres, A. , Kaffenberger, J.T. , and Powers, J.S. 2015. Initial white rot type
dominance of wood decomposition and its functional consequences in a regenerating
tropical dry forest. Soil Biology and Biochemistry 88: 58-68.
Schulze, E.D. and Koch, W. 1971 . Measurement of primary production with cuvettes. In
Productivity of Forest Ecosystems (Ecology and conservation, 4.), Proceedings of the
Brussels Symposium, 1969. Unesco. pp. 141-157.
Shaw, C.H., Hilger, A.B. , Metsaranta, J. , Kurz, W.A. , Russo, G. , Eichel, F. , Stinson, G., Smyth,
C., and Filiatrault, M. 2014. Evaluation of simulated estimates of forest ecosystem carbon
stocks using ground plot data from Canada' s National Forest Inventory. Ecological
Modelling 24-72: 323-347.
Shorohova, E. and Kapitsa, E. 2014. Influence of the substrate and ecosystem attributes on the
decomposition rates of coarse woody debris in European boreal forests . Forest Ecology
and Management 3 15 : 173-184.
Smyth, C.E., Trofymow, J.A. , Kurz, W.A. , and CIDET Working Group. 2010. Decreasing
uncertainty in CBM-CFS3 estimates of forest soil carbon sources and sinks through use
of long-term data from the Canadian Intersite Decomposition Experiment. Natural
Resources Canada, Canadian Forest Service, Pacific Forestry Centre. Information Report
BC-X-422.
Smyth, C.E., Kurz, W.A. , and Trofymow, J.A. 2011. Including the effects of water stress on
decomposition in the Carbon Budget Model of the Canadian Forest Sector CBM-CFS3.
Ecological Modelling 222( 5): 1080-1091.
Snetsinger, J. 2010. Chief Forester' s guidance on coarse woody debris management.
Government of British Columbia.
Sollins, P., Cline, S.P., Verhoeven, T. , Sachs, D., and Spycher, G. 1987. Patterns oflog decay in
old-growth Douglas-fir forests. Canadian Journal of Forest Research 17: 1585-1595.
Son, E. , Kim, J. , Lim, Y.W., Au-Yeung, T.T., Yang, C.Y.H., and Breuil, C. 2011. Diversity and
decay ability of basidiomycetes isolated from lodgepole pines killed by the mountain pine
beetle. Canadian Journal of Microbiology 57: 33-41.

93

Spittlehouse, D.L. and Black, T.A. 1980. Evaluation of the Bowen ratio/energy balance method
for determining forest evapotranspiration. Atmosphere-Ocean 18(2): 98-116.
StataCorp. 2009a. Stata Longitudinal-data/Panel-data Reference Manual, Release 11. Stata Press,
College Station, TX.
StataCorp. 2009b. Stata: Release 11. Statistical Software. StataCorp LP. College Station, TX.
Stokes, M. and Smiley, T. 1968. An Introduction to Tree-ring Dating. University of Chicago
Press, Chicago, IL.
Stone, J.N. , MacKinnon, A. , Parminter, J.V. , and Lertzman, K.P. 1998. Coarse woody debris
decomposition documented over 65 years on southern Vancouver Island. Canadian
Journal of Forest Research 28: 788-793.
Storaunet, K.O. , and Rolstad, J. 2002. Time since death and fall of Norway spruce logs in oldgrowth and selectively cut boreal forest. Canadian Journal of Forest Research 32(10):
1801-1812.
Sturtevant, B.R., Bissonette, J.A. , Long, J.N ., and Roberts, D.W. 1997. Coarse woody debris as a
function of age, stand structure, and disturbance in boreal Newfoundland. Ecological
Applications 7(2): 702-712.
Swinbank, W.C. 1951. The measurement of vertical transfer of heat and water vapor by eddies in
the lower atmosphere. Journal of Meteorology 8(3): 135-145.
Tabachnick, B.G. and Fidell, L.S. 2001. Using Multivariate Statistics. Allyn and Bacon,
Toronto. pp. 82-85.
Taylor, S.W., Carroll, A.L. , Alfaro, R.I. , and Safranyik, L. 2006. Forest, climate and mountain
pine beetle outbreak dynamics in western Canada. In the Mountain Pine Beetle A
Synthesis of Biology, Management and Impacts on Lodgepole Pine. Edited by L.
Safranyik and B. Wilson. Natural Resources Canada, Canadian Forest Service, Pacific
Forestry Centre, Victoria, BC. pp. 67-94.
Thomas, J.W. , Anderson, R.G. , Maser, C. , and Bull, E. 1979. Snags. In Wildlife Habitats in
Managed Forests: the Blue Mountains of Oregon and Washington. Edited by J.W.
Thomas. US Department of Agriculture, Forest Service. Washington, DC. Agriculture
Handbook No. 553 . pp. 60-77.
Thomas, M.D., and Hill, G.R. 1937. The continuous measurement of photosynthesis, respiration,
and transpiration of alfalfa and wheat growing under field conditions. Plant Physiology
12: 285-307.

94

Thompson, R.D. 2005 . Fine-scale disturbance and stand dynamics in mature, spruce-subalpine
fir forests of central British Columbia. M.Sc. thesis, Natural Resources and
Environmental Studies, Forestry. University of Northern British Columbia, Prince
George, BC.
Thome, G. 1935. Nemic parasites and associates of the mountain pine beetle (Dendroctonus
monticolae) in Utah. Journal of Agricultural Research 51(2): 131-144.
Trofymow, J.A. , Moore, T.R. , Titus, B., Prescott, C., Morrision, I. , Siltanen, M. , Smith, S.,
Fyles, J. , Wein, R. , Camire, C. , Duschene, L. , Kozak, L. , Kranabetter, M ., Visser, S.
2002. Rates of litter decomposition over 6 years in Canadian forests : influence of litter
quality and climate. Canadian Journal of forest Research 32: 789-804.
van der Wal, A. , de Boer, W. , Smant, W ., and van Veen, J.A. 2007. Initial decay of woody
fragments in soil is influenced by size, vertical position, and nitrogen availability and soil
origin. Plant Soil 301 : 189-201.
Velmex, Inc.1992. The Velmex "TA" system for research and non-contact measurement
analysis. Bloomfield, NY.
Volney, W.J .A. and Fleming, R.A. 2000. Climate change and impacts of boreal forest insects.
Agriculture, Ecosystems and Environment 82 : 283-294.
VoorTech Consulting. 2004. MeasureJ2X. Holderness, NH USA.
Wagner, W.L. , Wilson, B., Peter, B., Wang, S., and Stennes, B. 2006. Economics in the
management of mountain pine beetle in lodgepole pine in British Columbia: a synthesis.
In the Mountain Pine Beetle A Synthesis of Biology, Management and Impacts on
Lodgepole Pine. Edited by L. Safranyik and B. Wilson. Natural Resources Canada,
Canadian Forest Service, Pacific Forestry Centre, Victoria, BC. pp. 277-299.
Walker, L.R. , Wardle, D.A. , Bardgett, R.D. , and Clarkson, B.R. 2010. The use of
chronosequences in studies of ecological succession and soil development. Journal of
Ecology 98(4): 725-736.
Wang, T., Hamann, A. , Spittlehouse, D.L. , and Murdock, T. 2012. ClimateBC - high-resolution
spatial climate data for Western North America. Journal of Applied Meteorology and
Climatology 51: 16-29.
Waring, R.H., Coops, N.C., Mathys, A. , Hilker, T., and Latta, G. 2014. Process-based modeling
to assess the effects of recent climatic variation on site productivity and forest function
across western North America. Forests 5: 518-534.
Weedon, J.T. , Cornwell, W.K. , Cornelissen, J.H.C., Zanne, A.E., Wirth, C., and Coomes, D.A.
2009. Global meta-analysis of wood decomposition rates: a role for trait variation among
tree species? Ecology Letters 12: 45-56.
95

Weggler, K. , Dobbertin, M. , Jlingling, E. , Kaufmann, E. , and Thu.rig, E. 2012. Dead wood
volume to dead wood carbon: The issue of conversion factors. European Journal of Forest
Research 131 : 1423-1438.
Wesely, M.L. , Cook, D.R., and Hart, R.L. 1983. Fluxes of gases and particles above a deciduous
forest in wintertime. Boundary-Layer meteorology 27(3): 237-255.
Whistler, R.L. and Chen, C.C. 1991. Hemicelluloses. In Wood Structure and Composition.
Edited by M. Lewin and I.S. Goldstein. Marcel Dekker, Inc. New York, NY. pp. 288.
White, T. , Luckai, N. , Larocque, G.R. , Kurz, W.A. , and Smyth, C. 2008. A practical approach
for assessing the sensitivity of the Carbon Budget Model of the Canadian Forest Sector
(CBM-CFS3). Ecological Modelling 219: 373-382.
Williamson, G.B. , and Wiemann, M.C. 2010. Measuring wood specific gravity .. . Correctly.
American Journal of Botany 97(3): 519-524.
Wilson, J.L. and Tkacz, B.M. 1996. Historical perspectives on forest insects and pathogens in the
southwest: implications for restoration of ponderosa pine and mixed conifer forests . In
Conference on Adaptive Ecosystem Restoration Management: Restoration of Cordilleran
Conifer Landscapes of North America, Flagstaff, Arizona, 6-8 June 1996. Technical
Report RM-GTR-278 . United States Department of Agriculture, Forest Service, Rocky
Mountain Forest and Range Experiment Station, Fort Collins, CO. pp. 26-31.
Woo, K.L., Watson, P., and Mansfield, S.D. 2005 . The effects of mountain pine beetle attack on
lodgepole pine wood morphology and chemistry: Implications for wood and fiber quality.
Wood and Fibre Science 37(1):112-126.
Woodward, B. 2010. Log quality characteristics of the mountain pine beetle killed lodgepole
pine trees from southern interior British Columbia for board processing. M.Sc. thesis,
Natural Resources and Environmental Studies, University of Northern British Columbia,
Prince George, B.C.
Yatskov, M. , Harmon, M.E., and Krankina, O.N. 2003 . A chronosequence of wood
decomposition in the boreal forests of Russia. Canadian Journal of Forest Research 33:
1211-1226.
Yin, X. 1999. The decay of forest woody debris: Numerical modeling and implications based on
some 300 data cases from North America. Oecologia 121(1): 81-98.
Zabel, R.A. and Morell, J.J. 1992. Factors affecting the growth and survival of fungi in wood
(fungal ecology). In Wood Microbiology: Decay and Its Prevention. Academic Press,
Inc. , London, UK. pp. 90-115 .
Zhou, L. , Dai, L. , Gu, H., and Zhong, L. 2007. Review on the decomposition and influence
factors of coarse woody debris in forest ecosystem. Journal of Forestry Research 18(1):
48-54.

96

APPENDIX A. CARBON POOLS OF THE CBM-CFS3

Table Al. Carbon pools of the CBM-CFS3 *
CBM-CFS3 pool
Description
Merchantable + bark
Live stemwood of merchantable size plus bark
(SWorHW)
Other wood + bark
Live branches, stumps and small trees including bark
(SW orHW)
Foliage (SW or HW) Live foliage
Fine roots (SW or
Live roots, approximately < 5mm diameter
HW)
Coarse roots (SW or Live roots, approximately 2: 5mm diameter
HW)
Snag stems DOM
Dead standing stemwood of merchantable size including bark
(SWorHW)
Snag branches DOM Dead branches, stumps and small trees including bark
(SW orHW)
Medium DOM
Coarse woody debris on the ground
Aboveground fast
Fine and small woody debris plus dead coarse roots in the forest floor,
DOM
approximately 2: 5 and < 75mm diameter
Aboveground very
The L horizon comprised of foliar litter plus dead fine roots,
fast DOM
approximately <5mm diameter
Aboveground slow
F, H and O horizons
DOM
Belowground fast
Dead coarse roots in the mineral soil, approximately 2:5 diameter
DOM
Belowground very
Dead fine roots in the mineral soil, approximately <5mm diameter
fast DOM
Belowground slow
Humified organic matter in the mineral soil
DOM

.From Kurz et al. 2009

97

APPENDIX 8. PARAMETER DESCRIPTIONS

T abl e B 1 p·1eld and l aboratory paramet ers: um s an d defim1"f ions.
Variable
units
definition
Site-specific field parameters
Biogeoclimatic unit
BGCunit
o
c
mean annual temperature
MAT
mm
mean annual precipitation
MAP
mean winter precipitation
PPTwt
mm
SiteElev
m
site elevation
crew initials
Fieldcrew
notes related to site description
SiteNotes
Date
date of sample collection
site name
Site ID
Bole-specific field parameters:
BoleID
tree label
sampleID
sample label
S=standing, DE=down but elevated, DG=down on ground
Position
Angle
degrees if Position=DE, angle of bole relative to ground surface from tree
base: (+) if slope is upward toward tip, (-) if slope is downward
toward tip
DBH
diameter at breast height (1.3 m from tree base)
cm
DBHat
m
distance from tree base for DBH if not possible at 1.3 m
%
estimated percentage of bark cover on whole bole
Bbark
Falltype
RP = root plate, RS= root snap/stub, WS = wind snap
if Falltype = WS, distance from base of tree to snap height
WSheight
m
(measured at tree pith)
Orientation
degrees if Positon = DG or DE, compass azimuth from base to tip
Breaks
number of breaks in bole, including WS ; breaks above 7.5 cm
diameter not counted
m
Break 1. ..
distance from tree base to first break (measured at tree pith); same
for second break (Break2), Break3 , etc.
photos of stump, any rot fungi on bole, and down the length of the
BolePhoto
bole
Branch
branch order applying to most recent living crown (5=needles on,
4=needles off, fine branches present, 3=tertiary branch divisions
present, 2=secondary branch divisions present, 1=only primary
branches present, O(?)=no branches present)
Checking
minor = cracks < 2mm wide, major = cracks 2'.: 2mm wide, or yes/no
for presence/absence
ElevMax
cm
greatest height from forest floor along entire bole
TreeNotes
comments pertaining to entire bole

98

Table B 1 continued.
units
definition
Variable
Sample-specific field and laboratory parameters:
C = Cookie, W = wedge
sampletype
distance from base of tree at which sample was collected
m
from base
intact length of bole segment from which sample was collected (if
seglength
m
breaks are present in stem), no samples collected above 7.5 cm
diameter
canopy cover at sample point: DC = dark conifer (Sw), LC = light
cover
conifer (Pl), DD= dark deciduous (Alder), LD = light deciduous (At,
Ep), Open= no direct canopy cover (sample point in a gap), and
combinations . ..
straight-up photo above sample point: canopy cover above sample
samplephoto
%
log surface (top-down view) covered by vegetation at sample point
veg
vegetation species covering log at sample point (7-letter species codes,
vegSp
etc.)
%
percentage of bark present around perimeter of cookie (before cutting)
bark
distance from organic soil surface to bottom of log; (+) or (-) for
Elev
cm
elevated or embedded logs
decay class according to DeLong et al.: 1-5, similar to Maser's
DecayClass
thickness of sample cookie, averaged around circumference of disk
thick
cm
diameter of cookie across largest cross-section (or smallest); bold font
Diaml
cm
indicates lab-verified measurement
Diam2
cm
diameter of cookie across smallest cross-section (or largest); bold font
indicates lab-verified measurement
3
wood density (specific gravity)
Density
g/cm
YrsDead
years years from tree death to time of sampling
YrsDown
years years from tree fall to time of sampling
%
estimated heartwood rot (based on discolouration, not hardness)
HeartRot
estimated sapwood rot(based on discolouration, not hardness or
%
Sap Rot
bluestain)

99

F
i
e
l
dc
r
ew

S
i
t
eID

e
l
ev(m
)

Sub
zon
e
/SS

o
ldo
rn
ew

l
ayou
t

w
e
a
th
e
r

GPSw
ayp
tn
am
e

d
im
en
s
ion
s

Pho
tono. D
e
s
c
r
ip
t
ion

p
ag
e

1no
t
e
s

D
a
t
e

sk
e
t
ch

~ ~ ID_____Jsp.............t:c::el~ks.......t:~~::;~s:.... 1::i;knt............~jphotos

~~ ~~ ID[frombas~~:~er...

-·j
±
·-j

t:
;
:
:
n
g
t
h.
.
.
.
.
.
.
.·
t
:
:~k
·
·
-

1

.i

j
#b
r
e
ak
s

j
RP
/RS
/WS@

b
a
rk

l
ed
e
s
c
r:0b
r.
S
4
3-2
1 bo
n
s
/B
/W
I
~~ ~ .:~:ss thicknessIdiamht/wd sro-t:lnsH/-Bl/W
/b
lu
e

:~:ss

±e
l
ev
(
cm
),
~~
;~
:
..

B
o
l
eID.
.
.
.
.
. __........• S
c
a
r_
ID..-..·
·
·
..-~/DG
/DE
_
±·
~r
i
en
t_
_
_
_"
j
pho
to
s
J
S
p

L
a
u
r
a
'
sID:

ch
e
ck

e
l
ev
(
cm
)

~~
:~~
:
ID!
f
romb
a
s~~
:
;
e
r.. __ ·
-t:
;
:
:
n
g
t
h
·
·
~:~k
·
·
·
..

DBH

I

%

th
i
ckn
e
s
s
I
d
i
amh
t
/wd j
ro
t
:H
-ln
s
/B
/W

____ ,
,

c
c
ing __ ____.
__h
__e
__
_k
.
.
.
.
.
.
.
.
.
.
...
.
.
.
,
,
,

,
,
,
,
,
,

'Sln
s
/B
/W
/b
lu
e
L
au
r
a
'
sID:

b
r.5
4
3
-2
1 bo
l
ed
e
s
c
r:0-

I
d
i
amh
t
/wd ro
t
:H
-ln
s
/B
/W
th
i
ckn
e
s
s
~~ ~~ ID[tl"CJl11bas~~:~er.......---···t :;::ngth_____....t :~k-- _......: ±elev(cm)I
d
e
c
ayc
l
as
s
1
-2
34
5
S-ln
s
/B
/W
/b
lu
e
j
pho
to

v
eg

o/c ±e
l
ev
(
cm
)d
e
c
ayc
l
a
s
s

i
s
egl
eng
th

b
a
rk

%

1
-2
-34
5

!
s
c
a
rID

!
s
/DG
/DE±

o
r
i
en
t

•pho
to
s

ch
e
ck
ing

\#b
r
e
ak
s

!
RP
/RS
/WS@

b
a
rk

%

j
c
ov
e
r

;
pho
to

v
eg

e
c
ayc
l
a
s
s
% ±e
l
ev
(
cm
)d

t
:H
-ln
s
/B
/W
I
th
i
ckn
e
s
s
I
d
i
amh
t
/wd ro

f
romb
a
s
e
(m
)
C
/W
/
co !

i
s
egl
eng
th

b
a
rk

%

1
-2
-3
-4-5

S-ln
s
/B
/W
/b
lu
e

s
amp
l
eID

!
pho
to

v
eg

e
c
ayc
l
a
s
s
% ±e
l
ev
(
cm
)d

t
:H
-ln
s
/B
/W
l
th
i
ckn
e
s
sj
d
i
amh
t
/wd ro

i
s
egl
eng
th

b
a
rk

%

1
-2
34
-5

S-ln
s
/B
/W
/b
lu
e

!
s
c
a
rID

!
s
/DG
/DE±

o
r
i
en
t

•pho
to
s

ch
e
ck
ing

\#b
r
e
ak
s

i
RP
/RS
/WS@

b
a
rk

%

!
cov
e
r

s
amp
l
eID

:C
ov
e
r

f
romb
a
s
e
(m
)
C
/W
/
co !
B
o
l
eID
DBH

l
sp

s
amp
l
eID

C
/W
/
co
Bo
l
eID
DBH

!
cov
e
r

I
f
romb
a
s
e
(m
)
l
sp

;

bo
l
ed
e
s
c
r:0b
r.5
4
3
2
-1 I

!
pho
to

v
eg

% ±e
l
ev
(
cm
)d
e
c
ayc
l
a
s
s

i
s
egl
eng
th

b
a
rk

%

s
amp
l
eID

!
pho
to

v
eg

o/c ±e
l
ev
(
cm
)d
e
c
ayc
l
a
s
s

!
s
egl
eng
th

b
a
rk

%

C
/W
/
co

;

:
cov
e
r

I
f
romb
a
s
e
(m
)

S-ln
s
/B
/W
/b
lu
e

b
r
.5
4
3
2
1 I
bo
l
ed
e
s
c
r:0-

f
romb
a
s
e
(m
)
C
/W
/
co !

s
amp
l
eID

t
:H
-ln
s
/B
/W
l
th
i
ckn
e
s
s
l
d
i
amh
t
/wd ro

1
-2
3
4
5
1
-2
-3
-4
-5

F
igu
r
eB1
.F
i
e
ldd
a
t
as
h
e
e
t
.
100

L
au
r
a
'
sID:

L
a
u
r
a'
sID:

t
/wd ro
t
:H
-ln
s
/B
/W
l
th
i
ckn
e
s
s
l
d
i
amh
S-ln
s
/B
/W
/b
lu
e
t
/wd ro
t
:H
-ln
s
/B
/W
l
th
i
ckn
e
s
sj
d
i
amh
S-ln
s
/B
/W
/b
lu
e

APPENDIX C. DATA SUMMARIES

Table Cl. Summary statistics for wood density, carbon density and carbon concentration.
Sample size (n), mean(± st. dev.) and range (minimum and maximum values).
C
Basic wood density*
C density+
C concentration
ategory
(g. cm-3)
(g C . cm-3)
(g C. g-1 wood)

Live

Snags

Suspended

Down

All

(n = 10)

(n = 10)

(n = 6)

0.485 (± 0.023)

0.247 (± 0.014)

0.509 (± 0.007)

0.450 - 0.518

0.228 - 0.267

0.500 - 0.521

(n = 49)

(n = 49)

(n= 8)

0.441 (± 0.045)

0.224 (± 0.023)

0.508 (± 0.006)

0.298 - 0.551

0.153 - 0.279

0.501-0.516

(n= 113)

(n = 113)

(n = 17)

0.385 (± 0.060)

0.196 (± 0.030)

0.509 (± 0.005)

0.237 - 0.545

0.123 - 0.277

0.502 - 0.524

(n = 68)

(n = 68)

(n = 16)

0.314 (± 0.110)

0.161 (± 0.055)

0.519 (± 0.011)

0.032 - 0.472

0.020 - 0.240

0.504 - 0.541

(n = 240)

(n = 240)

(n = 47)

0.380 (± 0.089)

0.194 (± 0.045)

0.512 (± 0.009)

0.032 - 0.551

0.020 - 0.279

0.500 - 0.541

Wood densities reported represent the full dataset, including the subset analyzed for carbon concentration.
tcarbon densities reported include the samples analyzed for carbon concentration, as well as the remainder of the
samples whose carbon densities were derived from the outcome of the preliminary analysis.

101

Table C2. Summary data for all predictor variables considered for AICc models. Data shown are sample size (n), mean (±sd),
minimum and maximum . Not all predictors considered were included in AICc candidate models.

Categ.

sample
diameter•
(cm)
(n =4)

MAP
(mm)
(n = 10)

MAT
(OC)
(n = 10)

winter
prec1p.
(mm)
(n = I 0)

winter
temp.
(OC)
(n = 10)

summer
precip.
(mm)
(n = 10)

summer
temp.
(OC)
(n = 10)

17.7 (± 1.5)

575 (± 143)

2.4 (± 0.8)

149 (± 47) -8.2 (± 0.6)

98 (± 29)

12.4 (± 1.1)

16.2 - 19.1

330-814

1.7 - 3.8

73 -228

-8.8--7.3

62- 148

11.1 - 14.3

(n = 43)

(n = 49)

(n = 49)

(n = 49)

(n = 49)

(n = 49)

(n = 49)

(n = 49)

13 (± 13)

22.6 (± 9)

586 (± 174)

2.7 (± 0.6)

141(±61) -7.3(±0.5) 108(±34)

12.6 (± 0.9)

3-65

10.8-47.7

330-914

1.7 - 3.8

65 -262

-8.7 - -6.4

62- 177

11.1 - 14.3

(n = 103) (n = 100) (n = 107) (n = 113)

(n = 112)

(n=l13)

(n = 113)

(n = 113)

(n = 113)

(n=ll3)

(n = 113)

24 (± 16) 16(±13)

8 (± 8)

35 (± 26)

16.4 (± 6.1)

596(±181)

2.7 (± 0.6)

148(±65) -7.4(±0.5) 108(±34)

12.6 (± 1)

5-69

0-65

0-32

11 - 190

7.3 -32.8

333 - 914

1.6-3.8

65 -268

-8.4 - -6.4

65 - 177

11-14.3

(n = 67)

(n = 58)

(n = 59)

(n = 66)

(n = 67)

(n = 68)

(n = 68)

(n = 68)

(n = 68)

(n = 68)

(n = 68)

32 (± 23) 20 (± 20)

11 (± 9)

0 (± 6)

18.5 (± 8.6)

555 (± 176)

2.8 (± 0.6)

134 (± 60) -7.5 (± 0.6)

99 (± 34)

12.8 (± 0.9)

7-87

0-32

-19 - 10

8.4- 48.9

329-914

1.6-4

67 - 262

-8.9- -6.4

61 - 177

11 - 14.5

(n = 223) (n = 159) (n = 167) (n = 180)

(n= 180)

(n = 240)

(n = 240)

(n = 240)

(n = 240)

(n = 240)

(n = 240)

23 (± 19) 17(± 16)

9 (± 8)

23 (± 30)

17.1 (± 7.2)

582 (± 177)

2.7 (± 0.6)

143(±62) -7.4(±0.6) 105(±34)

12.7 (± 0.9)

3-87

0-32

-19 - 200

7.3 - 48.9

329-914

1.6-4

65 - 268

11 - 14.5

years
dead

years
standing

years
down

sample
elevation
(cm)

Live •

Snags •

Susp.

Down

All

0- 73

0-73

For live trees and snags, data shown for sample diameter are the measured diameters at breast height (DBH).
102

-8.9- -6.4

61 - 177

Ct')

••
- I•
--

•...

LO

••

N

Ct')

E
(.)

I••

-

0~
0)

~
·en
C

Q) LO

"C~
C

I

...0

..c

I

CCI

0

~

•

I

Live

Stand

Position categories

Susp

Down

Figure C 1. Distribution of carbon density (g C cm-3) in wood samples classified by categories of
live lodgepole pine, and MPB-killed lodgepole pine snags (Stand), suspended (Susp ), and down
CWD.

103

APPENDIX D. (ARBON DENSITY AND PREDICTOR RELATIONSHIPS

Table DI. Regression coefficients for best linear mixed model for carbon density in MPB-killed
lodgepole pine snags. Carbon density did not vary appreciably with any of the predictor
variables considered, therefore, parameter relationships were not evident in any of the candidate
models and the null model was best-ranked.
M odel #1 (null model)

Coeff.

St. Err.

95 % C. I.

0.224

0.00352

0.217-0.231

Site

4.64 X 10-0S

3.72 X 10-0?

7.11 X 10-I S - 0.0303

st. dev. (residual)

0.0231

0.00249

0.0187 - 0.0285

fixed effects
(intercept)

Random effects

ti.S~· .

~

...

• ...
"' • ...
•
I
- · ... "
"!

i'>
E
U N

t

•

()

s

C:

.

...

•

•

•
•

•

!~
""

....

I[
. ..

.t. 13

·~
I~

·~
lta
•

•••

AJO

.,

.t. 23

47

.,

.t. 36

~~
B

~

a)

0

0

20

years dead

40

60

b)
'ery dry cold

3

Dry cool

BEC subzone

Moist cold

Wet coo

Figure DI. Carbon density (g C cm- ) of snags (blue triangles) and live lodgepole pine (green
circles) for reference, versus predictor variables as follows: a) years dead and b) BEC subzones.
Data labels indicate the years dead for each snag. Continued on next page.

104

..,

"l

"'<'!

lr

E
UN

& 13

-"'
CJ

lll~o

.,

.

s

..,

..

••

•

6 28

•
• •• *32 a&24as• I •to
& 53

&7
&23
67

6§

It

i fO

u

67

..

• &9 • 15 u·

.,

47

& 13

•n

67

&7

.,,"

& 23

67
&36
&7

&36
&7

!~

••

6 38

C:

C:

~'"'.

B

~

200

400

600

..,

• 1 11

.,

& 13

E

~~

CJ

s
~"'
·en"'":

1000

1.5

A 28

• 1
&23

2

2.5

MAT(deg C)

3.5

4

• 1

It " "
.,

u

&13

..

... •to

Ai

.,

• ss • 53 •

,0 7

•
& 15 •
6 28

U'
&32

e

67

& 23

....t. 7

67

61

.,

& 36

&36
&7

.,,"

3

..,

..

•
6• 28

• •
~ ·

! •·11,s

it'

"'<'!

800

MAP(mm)

"l

"'

d)

c)

0

C:
C:

~'"'.

B

~

50

150

100

winter precipttation (mm)

..,

"l

"'

f)

e)

0

"'<'!

. .. , g

E

& 13

•s

~~

CJ

•

, 1~,

• 2•. ,,

• lfis ~r:el•
j

200

250

•
6• 28
&23

ti

.~.ro
61

· ~ 15

It

••

6 28

4 23

.. , 67

..,

• 36
• 1

.,,"

..

• • 2• t g .. ,

• 13 & 7

•n

• 36

i~

-8.5

winter temperature (deg C)

• & 24
•an t;I • •
•to

67

s

-7

.,

..
.,
&7

•& 9

-7.5

-8

C:
C:

.& '"'.
iii

"

~

g)

0

50

100

150

summer precipitation (mm)

h)

200

11

12

13

summer temperature (deg C)

14

15

Figure Dl continued: Carbon density (g C cm-3) of snags (blue triangles) and live lodgepole pine
(green circles) for reference, versus predictor variables as follows : c) MAP, d) MAT, e) mean
winter precipitation, f) mean winter temperature, g) mean summer precipitation, and h) mean
summer temperature. Data labels indicate the years dead for each snag.

105

•

•

•

•
b)

a)
40

20

0

80

60

years dead

0

+6

10

20

years down

•

•

ID

'°

-·
E
UN

u

30

.9

r

l>"'

~"-:

B

~

d)
0

40

20

60

80

sample height (cm)

100 'ery dry cold

Dry cool

•
• ••

Moist cold

BEC subzone

Wet coo

••

+6

_Jiff--~ ~ ~ - F
~;---:
2, 45

.,__..

~

+ 45

IIO

+ 52

+30

f)
200

400

600

MAP(mm)

800

1000

1.5

2

2.5

3

MAT (degrees C)

3.5

Figure D2. Carbon density (g C cm-3) of suspended CWD (red diamonds) and live lodgepole
pine (green circles) for reference, versus predictor variables as follows: a) years dead, b) years
down, c) sample height off the ground, d) BEC subzones, e) MAP, and f) MAT. Data labels
indicate the years dead for the bole. Continued on next page.

106

4

•so

h)

g)

0

50

100

.,,

•

150

200

•

-7.5

250

mean winter precipitation (mm)

-7

~ .5

mean winter temperature (degrees C )

••

~rr-::~~~~1-~~~--.~:i3~3·~2~·_Jt
• 45

· ~ ~ 30
• 30

•so

. ,2

• •

j)

i)
50

100

150

mean summer precipi1ation (mm)

200

11

12

13

14

mean summer temperature (degrees C )

15

Figure D2 continued: Carbon density (g C cm-3) of suspended CWD (red diamonds) and live
lodgepole pine (green circles) for reference, versus predictor variables as follows: g) mean winter
precipitation, and h) mean winter temperature, i) mean summer precipitation, and j) mean
summer temperature. Data labels indicate the years dead for the bole.

107

~

.•
I

• •

-

• •
20

0

•

40

•s

•s

927

.,.

..•

• •s

60

•n

,.,. ,,
• 02

. ,2

• 20

932

b)

80

•2•20
••o l ~i.50

928

• 20

• 49 45 • 2e
• 36

• 02

• 36

a)

years dead

1 82

•

• e1 • 3s

0

10

20

years down

30

..,
IH

925

c)
-15

-5

-10

"!

,,
••

•

II)

0

5

sample height (cm)

.,

• 1. ,s

• 21

m
• 2a

¥la

• 2s

-

W78

•
•

d)
10

'ery dry cold

Dry cool

•
• ••

••

e I
932

.,

8 15

:;o

• 02

··®

, ao

.w,•s

••• ••s
.,2l :l§

l!ll

• 50
• 50

I ii

• 02

••s
••s

. 36

932

• ,s

Moist cold

BEC subzone

l io

•

•

..•••,

Wet coo

W20

• 21
8 28

•• •
Ii,
l 3?m z,

• 50
. 50

•2•
• 2s

.,. ..,

• 02

925

f)
200

400

600
MAP (mm)

800

1000

1.5

2

2.5

3

MAT (degrees C)

3.5

4

Figure D3. Carbon density (g C cm-3) of down CWD (purple squares) and live lodgepole pine
(green circles) for reference, versus predictor variables as follows: a) years dead, b) years down,
c) sample height off the ground, d) BEC subzones, e) MAP, and f) MAT. Data labels indicate
the years dead for the bole. Continued on next page.

108

•

•

·~
• 2a

••

•

•

•• • •

•a

• 1a
• 1s

9 39

• 1a 1s

, 11

•11111

••

•

uc·s•

• 32
. .. 52

• 21

• so
• so

••a ••5

• s2

ff

. ,o

•• so
so
• 36
• 25

• a1

. 32

• 35
• 32

• s2

h)
50

100

150

••
. . ,5 •
l!t ,. •• I"' .,.32
.::eml!2 , ,o
927

•

• 2a

•

9 78

1~

9 20

9 28

9 48

•'1i2a

200

.,...,

••

• •

IH

••5

••

•

-8

-7.5

-7

mean winter temperature (degrees C )

••
•

•

.,
.,

• • 1 15

• 32

-6.5

•

•

•a

• s2

924
• 33 w21

9 7

5

-8.5

-9

250

mean winter precipitation (mm)

• 10

Ill>

•• so
so

IH

••s

8 15

9 20

9 10

9 48

• 52

• •25

• 32

. 36

. 35

9 87

•• so
so

9 28

9 28

• 52
9 87

9 25

j)
50

100

150

mean summer precipitation (mm)

200

11

12

13

14

mean summer temperature (degrees C)

15

Figure D3 continued: Carbon density (g C cm-3) of down CWD (purple squares) and live
lodgepole pine (green circles) for reference, versus predictor variables as follows: g) mean winter
precipitation, and h) mean winter temperature, i) mean summer precipitation, and j) mean
summer temperature. Data labels indicate the years dead for the bole.

109

APPENDIX E. TREE FALL-TIMES AND WITHIN-BOLE HETEROGENEITY

0

co

0

•

(0

C)

C

:c;
C

.So
(/) -.:::(/)
._
ro

Cl>

>-

0
N

0

Down

Susp

Figure El. Median and quartiles of the years from tree death to tree fall for suspended (Susp)
and Down CWD.

I I0

0.6

I"~•; . ~· . •
• ~· • ...
&

I

0.5 I

I ;•

I
{0.4
u

~

>
-~o.3
C

QI

"C
"C

0
0

3:0.2

A

&

•- t

A+

'*'
•• •~tt.+ ++
._ - • • ++m; +-' +A
*• * +
A

&

• Live
& Snags

+

-

:.:ii.

+

+

&

•• •

+

+

+

••
• ii •

+

- MinWDensity

~

+

• Elevated
• Down
+ MaxWDensity

+

+

• +

• ••

•

ii
+

•
+

•

-

0.1

0

10

20

30

40

so
Years dead

60

70

80

90

100

Figure E2. Mean within-bole heterogeneity of specific wood density (g cm-3) with maximum(+)
and minimum (-) for each suspended and down bole. Down boles exhibited the greatest
variability as demonstrated by the two lowest wood density observations near 60 years dead,
which are from boles that also have maximum wood density observations within the range of
live trees. Live trees and snags are included for reference only.

111