FINE-SCALE DISTURBANCE AND STAND DYNAMICS
IN MATURE, SPRUCE-SUBALPINE FIR FORESTS OF
CENTRAL BRITISH COLUMBIA
by
R. Douglas Thompson
B.Sc., University of Northern British Columbia, 2001

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

THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA
September 2005

© R. Douglas Thompson, 2005

1^1

Library and
Archives Canada

Bibliothèque et
Archives Canada

Published Heritage
Branch

Direction du
Patrimoine de l'édition

395 W ellington Street
Ottawa ON K 1A 0N 4
Canada

395, rue W ellington
Ottawa ON K 1A 0N 4
Canada
Your file Votre référence
ISBN: 978-0-494-28390-5
Our file Notre référence
ISBN: 978-0-494-28390-5

NOTICE:
The author has granted a non­
exclusive license allowing Library
and Archives Canada to reproduce,
publish, archive, preserve, conserve,
communicate to the public by
telecommunication or on the Internet,
loan, distribute and sell theses
worldwide, for commercial or non­
commercial purposes, in microform,
paper, electronic and/or any other
formats.

AVIS:
L'auteur a accordé une licence non exclusive
permettant à la Bibliothèque et Archives
Canada de reproduire, publier, archiver,
sauvegarder, conserver, transmettre au public
par télécommunication ou par l'Internet, prêter,
distribuer et vendre des thèses partout dans
le monde, à des fins commerciales ou autres,
sur support microforme, papier, électronique
et/ou autres formats.

The author retains copyright
ownership and moral rights in
this thesis. Neither the thesis
nor substantial extracts from it
may be printed or otherwise
reproduced without the author's
permission.

L'auteur conserve la propriété du droit d'auteur
et des droits moraux qui protège cette thèse.
Ni la thèse ni des extraits substantiels de
celle-ci ne doivent être imprimés ou autrement
reproduits sans son autorisation.

In compliance with the Canadian
Privacy Act some supporting
forms may have been removed
from this thesis.

Conformément à la loi canadienne
sur la protection de la vie privée,
quelques formulaires secondaires
ont été enlevés de cette thèse.

While these forms may be included
in the document page count,
their removal does not represent
any loss of content from the
thesis.

Bien que ces formulaires
aient inclus dans la pagination,
il n'y aura aucun contenu manquant.

Canada

ABSTRACT
Fine-scale disturbance and stand dynamics within mature, spruce-subalpine fir forests
were compared and contrasted across the central interior of British Columbia. The central
interior is ideal for the study of fine-scale disturbance because sub-boreal and subalpine
ecosystems span the dry interior plateau and grade eastward into the progressively wetter,
cooler, high-elevation ecosystems of the Rocky Mountain Foothills. Three biogeoclimatic
variants where selected across this climatic gradient: the dry-warm Sub-Boreal Spruce
(SBSdw3), the wet-cool Sub-Boreal Spruce (SBSwkl), and the wet-cool Engelmann SpruceSubalpine Fir (ESSFwk2). Traditional and newly developed dendroecological techniques
were used to identify the effects of past intermediate- and coarse-scale disturbance events,
and to infer regional climate fluctuations, in order to separate these from the effects of finescale disturbance. Periods of canopy mortality occurred in the ESSFwkl variant during the
1710s and 1830s, in the SBSwkl variant during the 1830s and 1870s, and in the SBSdw3
variant during the 1870s. Based on historical documentation and dendroecological evidence,
these periods of canopy mortality were likely caused by spruce beetle, and there is strong
evidence of notable shifts in species composition. A western balsam bark beetle outbreak
occurred in the ESSFwk2 variant during the 1920s and 1930s, and tree ring analyse provides
strong evidence that species dominance shifted from subalpine fir to spruce within these
stands. Moran’s I analysis of canopy ascension dates suggests that fine-scale disturbance is
random in space and time across all ecological variants. Canopy openings caused by
disturbance averaged -2.5 m in diameter, which is equivalent to the average distances
between nearest neighbours. Decadal disturbance rates did not differ by variant (p = 0.117)
or by species nested within variant (p = 0.418). Although the decadal disturbance rates did
not differ by variant, mean rates within SBSdw3, SBSwkl, ESSFwk2 variants were 7.4, 6.4
ii

and 5.0 %, respectively. The results of this research indicate that fine-scale disturbances are
important ecological processes in sub-boreal and subalpine forest ecosystems. Within these
ecosystems, however, there is considerable overlap between disturbance types of varying
spatial and temporal scales. Intermediate- and coarse-scale disturbance events, such as those
caused by insects (e.g., spruce beetle and western balsam bark beetle), have been more
influential in causing distinct changes in stand structure and composition than fine-scale
disturbances. Consideration of disturbance events at multiple scales, however, shows that
fine-scale disturbance does exhibit continuous stand level maintenance and renewal.

Ill

TABLE OF CONTENTS
ABSTRACT..................................................................................................................................... ii
TABLE OF CONTENTS............................................................................................................... vi
LIST OF TABLES.........................................................................................................................vii
LIST OF FIGURES................................................................................................................. viii
ACKNOWLEDGEMENT............................................................................................................. x
PREFACE....................................................................................................................................... I
CHAPTER I - ECOLOGY, FOREST DYNAMICS AND DISTURBANCE AGENTS OF
THE SUB-BOREAL SPRUCE AND ENGELMANN SPRUCE-SUBALPINE FIR
BIOGEOCLIMATIC ZONES
Introduction..........................................................................................................................6
Classification of British Columbia’s forested ecosystems...............................................7
General description of the Sub-Boreal Spruce Z o n e...................................................... 7
General description of the Engelmann Spruce-Subalpine Fir Z o n e .............................. 9
Stand dynamics within the SBS and ESSF.................................................................... 11
SBS and ESSF as fire driven ecosystems...................................................................... 12
SBS and ESSF as patch driven ecosystems.................................................................. 13
Fine-scale processes within the SBS and ESSF.......................................................... 15
Decay fungi......................................................................................................... 16
Tomentosus root disease.................................................................................... 17
Western balsam bark beetle............................................................................... 18
Conclusions...................................................................................................................... IS
Literature eited................................................................................................................. 20
CHAPTER 2 - DETECTING STAND AND LANDSCAPE-LEVEL DYNAMICS WITHIN
MATURE SPRUCE-SUBALPINE FIR FORESTS OF CENTRAL BRITISH COLUMBIA:
DENDROECOLOGICAL CONSIDERATIONS I
Abstract............................................................................................................................ 27
Introduction........................................................................................................................29
Materials and Methods......................................................................................................31
Field Sampling......................................................................................................31
Dendroecological analyses...................................................................................33
Results............................................................................................................................... 36
Diameter and age structure at breast height........................................................36
Ring-width chronologies...................................................................................... 38
Discussion........................................................................................................................ 42
Literature cited.................................................................................................................. 47

IV

CHAPTER 3 - USING TREE RING GROWTH RATE CRITERIA TO ASSIGN DATES OF
CANOPY DISTURBANCE IN MATURE SPRUCE-SUBALPINE FIR FORESTS OF
CENTRAL BRITISH COLUMBIA: DENDROECOLOGICAL CONSIDERATIONS II
Abstract............................................................................................................................. 62
Introduction........................................................................................................................64
Materials and Methods......................................................................................................66
Field sam pling......................................................................................................66
Dendroecological materials and procedures...................................................... 68
Early growth rates and gap-origin criteria......................................................... 68
Percent change in radial growth and release criteria..................................... 70
Results................................................................................................................................ 72
Early growth rates and gap-origin criteria............................................... 72
Percent change in radial growth and release criteria..................................... 73
Discussion........................................................................................................................ 74
Literature cited................................................................................................................. 79
CHAPTER 4 - A NEW DENDROECOLOGICAL METHOD TO SEPARATE RESPONSES
TO CANOPY TREE MORTALITY FROM CLIMATE
Abstract............................................................................................................................. 88
Introduction...................................................................................................................... 89
Methods..............................................................................................................................96
Calibrating percent divergence............................................................................96
Application of the divergence method................................................................98
Analyses of the divergence method.................................................................... 99
Results............................................................................................................................ ICO
Calibrating percent divergence........................................................................ 100
Application and analyses of the divergence method........................................101
Discussion...................................................................................................................... 102
Literature cited............................................................................................................... 108
CHAPTER 5 - FINE-SCALE SPATIO-TEMPORAL DISTURBANCE PATTERNS IN
MATURE, SPRUCE-SUBALPINE FIR FORESTS FOUND IN THE CENTRAL
INTERIOR OF BRITISH COLUMBIA
Abstract.......................................................................................................................... 120
Introduction.................................................................................................................. 121
Methods............................................................................................................................ 123
Field sampling................................................................................................... 123
Dendroecological materials and procedures................................................... 125
Interpretation of radial growth patterns.......................................................... 126
Disturbance chronologies and frequency........................................................ 127
Point pattern analysis....................................................................................... 128
Spatio-temporal analysis of canopy ascension................................................. 129
Results............................................................................................................................ 132
Non-spatial structure of live and dead trees.................................................... 132

spatial patterns of live and dead trees............................................................. 134
Canopy ascension type by ecological variant and species............................ 134
Disturbance chronologies and canopy ascension by decade........................... 135
Spatial-temporal patterns of canopy disturbance............................................. 136
Discussion...................................................................................................................... 136
Literature cited............................................................................................................... 144
CHAPTER 6 - SYNOPSIS: SUMMARY OF FINDINGS, MANAGEMENT
IMPLICATIONS, FUTURE RESEARCH AND CONCLUSIONS
Disturbance in forested ecosystems...............................................................................162
Ecological sum m ary.......................................................................................................163
Disturbance histories........................................................................................ 163
Canopy disturbance rates................................................................................ 164
Canopy opening size......................................................................................... 165
Stand composition, canopy replacement and future stand composition
165
Management implications............................................................................................. 166
Future research............................................................................................................... 170
Conclusions.................................................................................................................... 172
Literature cited............................................................................................................... 173

VI

LIST OF TABLES
Table 2.1. Summary statistics for the 30 species- and plot-specific chronologies.

51

Table 2.2. Mean summary statistics for the species- and site-specific chronologies.

52

Table 2.3. Summary statistics for the principle component analyses (PCA).

53

Table 2.4. Summary statistics of the regionally- and species-specificchronologies.

54

Table 3.1. Nested ANOVA results of early growth rates.

82

Table 3.2. Nested ANOVA results of percent change in radial growthvalues.

83

Table 4.1. Nested ANOVA results of percent divergence values.

112

Table 4.2. Nonlinear regression statistics.

113

Table 4.3. Conceptual framework identifying the two error types during the
identification of release.

114

Table 5.1. Structural attributes of the canopy layer (live trees >15 cm diameter
at breast height, dbh).

150

Table 5.2. Structural attributes of dead trees >15 cm diameter at breast height (dbh).

151

Table 5.3. Proportion of canopy ascension types grouped by ecological variant
and by species.

152

Table 5.4. Summary statistics of mutual and partial independence tests between
species, ecological variant and ascension type using contingency analyses.

153

Table 5.5. Summary statistics for the untransformed decadal percentages of canopy
ascension grouped by ecological variant and species.

154

Table 5.6. Nested ANOVA results of percent decadal canopy ascension over the
long-term and the short-term.

155

Vll

LIST OF FIGURES
Figure 1.1. Locations of the Sub-Boreal Spruce and the Engelmann SpruceSubalpine Fir biogeoclimatic zones and study areas.

26

Figure 2.1. Diameter distributions of living spruce and subalpine fir for the
15 study plots.

55

Figure 2.2. Age-class distributions of living spruce and subalpine fir for the
15 study plots.

56

Figure 2.3. Diameter distributions of dead spruce, subalpine fir and unidentifiable
species for the 15 study plots.

57

Figure 2.4. Species-specific, standard ring-width chronologies for the 15 study plots.

58

Figure 2.5. Percentage of cubic-spline chronologies showing marker rings in each year.

59

Figure 2.6. Species-specific, standard ring-width chronologies for the 15 study plots.

60

Figure 2.7. Regional- and species-specific chronologies.

61

Figure 3.1. Frequency distributions of radial growth rates for gap-origin and suppressed
saplings.

84

Figure 3.2. Frequency distribution of the number of spruce and subalpine fir trees
originating in a canopy gap within each decade for each plot.

85

Figure 3.3. Frequency distribution of percent release for gap-fillers.

86

Figure 3.4. Frequency distributions of the number of spruce and subalpine fir
releasing from suppression within each decade for each plot.

87

Figure 4.1. Pictorial outline of the divergence method.

115

Figure 4.2. Gap-filler percent divergence values versus percent release values.

116

Figure 4.3. Frequency distribution of gap-filler percent divergence values.

117

Figure 4.4. Regional chronologies compared with frequency of release events identified
and removed by the divergence method per decade.

118

Figure 4.5. Relationship between the proportion of release events removed and the
mean decadal ring-width indices.

119

Figure 5.1. Graph of L(d) versus distance for live spruce >15 cm DBH.

156

V lll

Figure 5.2. Graph of L(d) versus distance for

live subalpine fir >15 cm DBH.

157

Figure 5.3. Graph of L(d) versus distance for

dead spruce >15 cm DBH.

158

Figure 5.4. Graph of L(d) versus distance for

dad subalpine fir >15 cm DBH.

159

Figure 5.5. Disturbance chronologies for each plot showing the percentage of
cored trees ascending to the canopy in each decade.

160

Figure 5.6. Correlograms of the Moran’s I coefficients (left) and standard normal
deviates (right) for canopy ascension dates by distance lag.

161

IX

ACKNOWLEDGEMENT
Many people gave their time to this project and supported me over the last three years, and to
them I am indebted. For financial contributions, I would like to thank the John Prince
Research Forest, Canadian Forest Products, British Columbia Forest Science Program,
Canadian National Science and Engineering Research Council and the University of
Northern British Columbia. I wish to thank my research assistant, Orrin Quinn. Thanks to
my external reviewer, Roberta Parish. To Lori, Mike and Staffan - 1 often showed up at your
office without notice and you were always willing to help - thank you. Kathy, I came to
your office and exclaimed, “I wanted to do my Masters.” You replied with great enthusiasm,
“I was waiting for you to say that.” You have been extremely positive and a mentor from
day one.
To my sister Tania, and to my Mom and Dad who phoned every Sunday at six.
Finally, to Donna - we worked through this together. We laughed, cried and supported one
another when it was most important. This is only the beginning. Thank you.

PREFACE
“To have the ultimate, even if idealistic, objective of fusing the shattered fragments into the
original unity is of great scientific and practical importance; practical because so many
problems in nature are problems of the ecosystem rather than of soil, animals or plants, and
scientific because it is our primary business to understand.” Alex. S. Watt, 1947.
Forest management within British Columbia has traditionally been biased towards
sustaining fibre yields. Arguably, this bias still remains as the dominant management
framework today. Present forest health concerns, however, have forest managers re­
evaluating how forested ecosystems should be managed. Although in its youth, a paradigm
shift is occurring in British Columbia, and this is changing the way we approach forest
management. The new approach, although in debate as to its appropriateness across a wide
range of ecological and tenure systems, is based on two premises. First, successful
conservation and maintenance of healthy forest ecosystems requires knowledge pertaining to
historical patterns and processes resulting from naturally occurring disturbance events (e.g.,
the spatial and temporal distribution of forest patches caused by wildfire). Second,
ecological values inherent in forest ecosystems are best maintained by using forest practices
that mimic these natural disturbances. Together, these premises define forest ecosystem
management (Attwill 1994, Rogers 1996, Bergeron et al. 1999).
This thesis adds to the knowledge base that is essential for examining the application
of forest ecosystem management. The objective of this thesis was to quantify and
characterize fine-scale disturbance patterns and processes as they vary across three distinct
sub-boreal and subalpine eeosystems of eentral British Columbia. In British Columbia, the
central interior is ideal for the study of fine-scale disturbance because sub-boreal and
subalpine ecosystems span the dry interior plateau and grade eastward into the progressively
wetter, cooler, high-elevation ecosystems of the Rocky Mountain Foothills (Meidinger and

Pojar 1991). My research goals were to: (1) develop and modify existing dendroecological
techniques to date fine-scale disturbance; (2) quantify present forest composition and
structure as it relates to unique disturbance events; and (3) determine the spatial and temporal
occurrence of fine-scale disturbance within and between three climatically distinct, mature
spruce-subalpine fir forest ecosystems.
In Chapter 1 ,1 review the general forest ecology, current understanding of stand
dynamics and disturbance ecology set in the Sub-Boreal Spruce and Engelmann Spruce Subalpine Fir biogeoclimatic zones. Although coarse-scale disturbance events, such as
stand-replacing fires, play a significant role in shaping these ecosystems, fine-scale
disturbance events can have an impact on the forest structure and dynamics at the stand level.
In Chapter 2 ,1 develop a set of stand and landscape-level standardized chronologies
to identify past growing conditions attributed to disturbances and climatic events at scales
greater than the sampling unit (50-m x 50-m plots). To characterize fine-scale disturbance
regimes in any region, historical legacies created by past intermediate- and coarse-scale
disturbance events and the regional climate influence must be documented. In the first
standardization method, crossdated series were transformed by dividing the observed ringwidth by the mean series ring-width (i.e., horizontal-line standardization). Standardizing
series by their mean preserves long-term growth trends that can be used to identify
intermediate- and coarse-scale canopy removal. In the second standardization method,
crossdated series were transformed by fitting a cubic-smoothing-spline of 50 % frequency
response of 50 years. Standardizing series using a cubic-spline preserves short-term growth
trends that can be used to infer yearly climatic fluctuations. These two chronologies were
used to aid in: (1) the reconstruction of individual stand histories; and (2) the explanation of

differences in stand structure, composition and forest dynamics across the three sub-boreal
and subalpine ecosystems.
In Chapter 3 ,1 develop methods to use radial-growth-rate patterns in present canopy
trees to determine dates of ascension towards the canopy. Two growth-rate criteria were
calibrated for the sub-boreal and subalpine ecosystems, including: (1) early growth rates that
indicate gap-origin of sample trees; or (2) release criteria that indicate that a sample tree
released from suppression due to overhead mortality. The release criteria were calibrated
using a sample of gap-maker/gap-filler pairs with known dates of death and response.
Despite a conservative release threshold, analyses revealed that a substantial number of
release events, attributed to climate, were being counted.
In Chapter 4, in response to the problem identified with the release threshold defined
in Chapter 3 ,1 develop a new dendroecological method to separate responses to overhead
tree mortality from climate. Divergence from the regional year-to-year signal was used to
assess an individual tree’s growing state at fine-scales (<0.25 ha). The year when an
individual series diverges from the regional trend, to some degree for an extended length of
time, was used to separate responses to overhead tree mortality from climate.
In Chapter 5 ,1 summarize the fine-scale disturbance regimes of the three climatically
distinct, mature spruce-subalpine fir ecosystems. This study combines the results and
methodologies developed in Chapters 2, 3 and 4 to develop fine-scale disturbance
chronologies for each plot. Basic stand composition and attribute data were sampled. Spatial
autocorrelation analysis of final ascension dates and point-pattem analysis of living and dead
trees were used to determine differences in the canopy level disturbance structures across the
three climatically distinct ecosystems.

In Chapter 6, the results of the individual studies are integrated, drawing final
conclusions regarding the fine-scale disturbance regimes within the mature, spruce-subalpine
fir forests types studied within this thesis. Furthermore, in Chapter 6 ,1 discuss the possible
implications for forest management and suggest future research directions within the mature,
spruce-subalpine fir forests found across central British Columbia.

Literature cited
Attiwill, P. 1994. The disturbance of forest ecosystems: the ecological basis for conservation
management. For. Ecol. Man. 63:247-300.
Bergeron, Y., Harvey, B., Leduc, B. and Gauthier, S. 1999. Forest management guidelines
based on natural disturbance dynamics: Stand and forest level considerations.
For. Chron. 75:49-54.
Meidinger, D. and Pojar, J. 1991. Ecosystems of British Columbia. Crown Publications Inc.
Special Report Series No. 6. Victoria, BC.
Rogers, P. 1996. Disturbance ecology and forest management: A review of the literature.
USDA. For. Serv. Gen. Tech. Rep. INT-GTR-336. Ogden, UT, USA.
Watt, A., S. 1947. Pattern and process in the plant community. J. Ecol. 35:1-22.

CHAPTER 1 - ECOLOGY, FOREST DYNAMICS AND DISTURBANCE AGENTS
OF THE SUB-BOREAL SPRUCE AND ENGELMANN SPRUCE-SUBALPINE FIR
BIOGEOCLIMATIC ZONES

Introduction
Within the central interior of British Columbia there is a climatic gradient that spans
the dry interior, sub-boreal plateau to progressively wetter, cooler, high-elevation, subalpine
eeosystems of the Rocky Mountain Foothills (Meidinger and Pojar 1991). Within these
forested eeosystems fire has historically been thought of as the most important disturbance
agent (Johnson 1992, Andison 1996, DeLong and Tanner 1996, Hawkes et al. 1997, DeLong
and Kessler 2000). A review of the ecology, forest dynamics and disturbance literature
within the sub-boreal and subalpine ecosystems suggests that fire may not always be the
principal agent of disturbance controlling stand structure and composition (Hawkes et al.
1997, Parish et al. 1999, Antos and Parish 2002a, Antos and Parish 2002b). For example.
Parish et al. (1999) found that stand composition of a mature spruce-subalpine fir forest in
southeastern British Columbia, was the result of multiple disturbance types of differing
intensities acting at multiple spatial scales. This overlapping of disturbance types supports
White’s (1979) hypothesis that a gradient between major and minor disturbance events,
rather than a uniquely definable set of major catastrophes, can exist. The objectives of this
chapter are to: (1) review the general ecology, stand dynamics and disturbance regimes
within the sub-boreal and subalpine forest ecosystems; and (2) review the most common
agents of disturbance influencing forest ecology in the central interior of British Columbia.

Classification of British Columbia’s forested ecosystems
The terrestrial ecosystems of British Columbia have been hierarchically classified
using the biogeoclimatic ecosystem classification system (Meidinger and Pojar 1991). This
system integrates climate, soil, and vegetation data, and attempts to provide a universal
framework for resource management and scientific research (Meidinger and Pojar 1991).
The biogeoclimatic subzone is the basic unit of classification, resulting from zonal (climatic)
categorization, and subzones represent groups of ecosystems under the influence of the same
regional climate (DeLong et al. 1993). Subzones can be grouped to form biogeoclimatic
zones and subdivided to define ecological variants. The variant reflects a further
differentiation in regional climate and is generally recognized as an area that is slightly drier,
wetter, snowier, warmer, or colder than other areas within the same subzone (Meidinger and
Pojar 1991). Similar climax vegetation can occur over a range of subzones and variants, and
in extreme cases among biogeoclimatic zones. As a consequence, similar climax-forest
communities may represent ecosystems from different regional climates and with different
soils (Meidinger and Pojar 1991).
The Sub-Boreal Spruce (SBS) and Engelmann Spruce-Subalpine Fir (ESSF) are two
of the dominant biogeoclimatic zones of central and east-central British Columbia. In
contrast to true boreal ecosystems, the sub-boreal and subalpine climates are slightly less
continental, thus slightly warmer in January and cooler in July (Meidinger and Pojar 1991).

General description of the Suh-Boreal Spruce Zone
The SBS biogeoclimatic zone is located within central British Columbia. It is
bordered by the Interior Cedar Hemlock (ICH), Sub-Boreal Pine-Spruce (SBPS), Interior
Douglas-fir (IDF) and Engelmann Spruce-Subalpine Fir (ESSF) zones. It occurs from valley
7

bottoms up to 1100-1300 m elevation and between 52 and 57 °N latitude and 122 and 128
°W longitude (Meidinger and Pojar 1991).
Due to its moderate boreal climate, the SBS is affected by seasonal extremes in
temperature and by relatively moderate levels of precipitation. The region is broadly
continental with long, cold, snowy winters and short, warm, moist summers (DeLong et al.
1993). Mean annual temperature of the SBS ranges from 1.7-5 °C and average temperatures
are below 0 °C for 4-5 months of the year, and above 10 °C for 2-5 months (Meidinger and
Pojar 1991). Mean annual precipitation, recorded at long-term stations, ranges from 440-900
mm, of which 25-50 % is snow (Meidinger and Pojar 1991). Upland soils are primarily from
the luvisolic, podzolic, and brunisolic soil orders found on abundant morainal deposits and
imperfectly to poorly drained sites typically will display gleyed soil horizons (Meidinger and
Pojar 1991, Agric. and Agri-Food Can. Publ. 1998).
The SBS zone is dominated by upland coniferous forest types. Hybrid white spruce
(Picea engelmannii x glauca Voss x Parry) and subalpine fir (Abies lasiocarpa Nutt.) are the
dominant climax tree species. Lodgepole pine (Pinus conforta var. latifoUa Dough), a serai
species in the SBS zone, is common in mature forests in the drier parts of the zone and both
lodgepole pine and trembling aspen (Populus tremuloides Michx.) pioneer the extensive serai
stands (Meidinger and Pojar 1991). Paper birch (Betula papyrifera Marsh), another pioneer
species, will establish on moist, rich sites. Douglas-fir (Pseudotsuga menziesii Franco), a
long-lived species, will occur abundantly on dry, warm, rich sites as a consistent, although
small, component of mesic forest types.
Ten subzones are recognized within the SBS zone, largely due to variability in
climate, especially precipitation (Meidinger and Pojar 1991). Within this study, I will

consider two biogeoclimatic variants (Figure 1.1): the Stuart Dry Warm Sub-Boreal Spruce
(SBSdw3) and the Willow Wet Cool Sub-Boreal Spruce (SBSwkl) (Meidinger and Pojar
1991). The former variant falls within DeLong's (2002) Moist Interior - Plateau Natural
Disturbance Unit (NDU) with fire and mountain pine beetle (Dendroctonus ponderosae
Hopkins) as the key stand-replacement disturbance agents. The disturbanee cycle assigned to
this NDU is 100 years (DeLong 2002), supported by the work of Andison (1996) and
DeLong (1998). The latter variant falls within the McGregor Plateau NDU with fire as the
key stand-replacement disturbance agent (DeLong 2002). The stand-replacementdisturbance cycle assigned to this unit by DeLong (2002) is 220 years.

General description of the Engelmann Spruce-Subalpine Fir Zone
The ESSF biogeoclimatic zone is one of the most extensive forest zones in British
Columbia, covering 13.3 million ha or 14 % of BC’s land area (M l et al. 1996). It is the
uppermost forested area in elevation, the widest ranging biogeoclimatic zone in British
Columbia, and found in Alberta, the Pacific Northwest and Rocky Mountain states. In
British Columbia, the ESSF zone oecurs at elevations of 1200-2100 m in the south-west,
from 1500-2300 m in the south-east and from 900-1700 m in the northern part of the zone
(Meidinger and Pojar 1991). In east-central British Columbia it is found on the McGregor
Plateau below the Alpine Tundra (AT) zone and above the SBS and ICH zones. The
MeGregor Plateau is at the eastern edge of the interior plateau, between the northern and
southern portions of the Rocky Mountain trench (DeLong et al. 1994).
At elevations >1000 m, the climate is cold and precipitation is variable with
windward slopes being wetter than lee slopes (DeLong et al. 1994). The ESSF zone has a
cold, moist, snowy continental elimate. In the southern regions of British Columbia, mean

annual temperatures range from -2 to 2 °C (Meidinger and Pojar 1991). Mean monthly
temperatures are below 0 °C for 5-7 months and above 10 °C for 0-2 months (Meidinger and
Pojar 1991). Although precipitation is highly variable, windward slopes can receive up to
2200 mm annual precipitation. Most of the precipitation (50-70 %) falls as snow and
maximum snow pack ranges from 1-4 m (Meidinger and Pojar 1991). Due to the short cool
growing seasons, and long cold winters, the ESSF zone is one of the most severe climates for
forest growth and regeneration (full et al. 1996).
Within east-central British Columbia, at the lower elevation range, soils are
predominantly brunisols, lu vi sols and podzols (see Agric. and Agri-Food Can. Publ. 1998).
Humo-ferric podzols have formed on parent materials consisting predominantly of medium
textured morainal and colluvial deposits (DeLong et al. 1994).
The ESSF zone includes continuous forest at its lower and middle elevations, and
subalpine parkland at its upper elevations (Meidinger and Pojar 1991). At lower elevations
the ESSF zone is dominated by Engelmann spruce {Picea engelmannii Parry ex Engelm.) and
subalpine fir {Abies lasiocarpa). Spruce, which is the longer-lived species, usually
dominates the canopy of mature stands despite its relatively limited recruitment in the
understory (Veblen 1986, Meidinger and Pojar 1991, Veblen et al. 1991, Veblen et al. 1994,
Kneeshaw and Burton 1997, Lewis and Lindgren 1999, Lewis and Lindgren 2002).
Fifteen forested subzones are currently recognized in the ESSF zone (Meidinger and
Pojar 1991), largely due to the broad latitudinal and elevation range, and to the variability in
climate, especially precipitation. Within this study, I will consider one biogeoclimatic
variant (Figure 1.1), the Misinchinka Wet Cool Engelmann Spruce-Subalpine Fir (ESSFwk2)
(Meidinger and Pojar 1991). This variant falls within DeLong's (2002) Wet Mountain NDU

10

with stand-replacement disturbance events occurring at irregular intervals with as much as
1000 years between such events on any site.

Stand dynamics within the SBS and ESSF
There are recognizable and seemingly equivalent patterns and processes that appear
after stand-replacing disturbances. Models of stand dynamics, therefore, are fundamental in
understanding stand genesis, maintenance, and renewal over long periods of time. The
progressive change in vegetation has lead to the development of two competing models of
forest succession. The first, where one species invades after another, until one species or a
group of species invades, predominates and replaces itself rather than being replaced, is
known as relay floristics (Clements 1916, Oliver and Larson 1996). This process of
replacement creates a stable endpoint to succession known as the “steady-state”,
“equilibrium” or “climax” vegetation association (Clements 1916, Oliver and Larson 1996).
The second, where species that predominate later have been present since, or almost
immediately after, the stand replacing disturbance event is known as initial floristics (Egler
1954, Drury and Nisbet 1973).
Due to the stochastic nature of most disturbance types, variability exists in the
potential stand development for a given area. Elements of both initial and relay floristics,
thus, are characteristic of stand development. The invasion patterns after disturbance,
however, predominantly follow the initial floristics pattern (Oliver and Larson 1996). As a
direct outcome of the latter model, Oliver and Larson (1996) proposed four distinct
successional phases that characterize the progression of forest succession. These phases are
(1) stand initiation, (2) stem exclusion, (3) stem reinitiation and (4) old-growth. This standdynamics model has been applied to SBS and ESSF biogeoclimatic zones, and comparable
11

forest types (Aplet et al. 1988, Kneeshaw and Burton 1997, Lewis and Lindgren 1999,
Newbery 2001, Antos and Parish 2002a, Antos and Parish 2002b). Stand dynamics and
disturbance ecology studies within the SBS and ESSF biogeoclimatic zones, however,
suggest that the theoretical phases of the traditional models are not ecologically strict (Antos
and Parish 2002a, Antos and Parish 2002b).
The relative importance of coarse-scale (e.g., wildfire) versus fine-scale processes on
stand structure and dynamics are currently under review (Kimmins 1999, McCarthy 2001).
Stands initiated by fire are subsequently affected by various partial disturbances (Agee 1993,
Antos and Parish 2002a). Under certain circumstances, however, legacies from the initial
stand-replacing events can continue to dominate stand structure and dynamics; even with
long periods between stand-replacement (Antos and Parish 2002a). In contrast, some subboreal and subalpine forest types are strongly influenced by single and multiple-tree
disturbance events (Veblen 1986, Lertzman and Krebs 1991, Lertzman 1992, Kneeshaw and
Burton 1997, Lewis and Lindgren 2000, Newbery 2001). This range from coarse- to finescale disturbance types, coexisting on the landscape, supports White’s (1979) hypotheses
that: (1) there is a gradient from major to minor disturbance events rather than a uniquely
definable set of major catastrophes; and (2) some disturbances are initiated or promoted by
the biotic component of the ecosystem.

SBS and ESSF as fire driven ecosystems
The role of fire as a natural process vital to the health and renewal of boreal, subboreal and sub-alpine forests of western North America has long been acknowledged
(Johnson and Rowe 1977, Hawkes 1980, Romme 1980, Heinselman 1981, Pyne 1984).
More recently, the SBS and ESSF biogeoclimatic zones of central and east-central British
12

Columbia have received greater attention with respect to stand-replacing fires, which have
historically been thought of as the most important natural disturbance agent (Johnson 1992,
Andison 1996, DeLong and Tanner 1996, Oliver and Larson 1996, Hawkes et al. 1997,
DeLong 1998). Although the SBS and ESSF forests, in most cases, do originate by fire,
recent studies have shown that the time between stand-replacing events are enormously
varied (Hawkes et al. 1997, Sanborn et al. 2001). In these forest types, fire may be so
infrequent that the development of true old-growth forests may be possible. Hawkes et al.
(1997) report that fire return intervals in very wet and cool SBS subzones near the Rocky
Mountains likely range from 1200-6250 years. DeLong and Tanner (1996) reported return
intervals ranging from 227-345 years in slightly drier SBS forests in the foothills of the
Rocky Mountains.
As the role of fire diminishes, other biotic patch disturbance agents could assume a
dominant role in renewal of boreal, sub-boreal and sub-alpine forests. A number of the biotic
patch disturbance agents important within the SBS and ESSF are discussed below.

SBS and ESSF as patch driven ecosystems
At sporadic intervals a limited number of insect species have the ability to initiate
patch disturbances (Lewis and Lindgren 2000). Among these insects are the primary bark
beetles and a number of defoliating insects. The most common patch-initiating agent in
mature, spruce-subalpine fir forests of central British Columbia is the spruce beetle
{Dendroctonus rufipennis Kirby).
The spruce beetle can cause severe mortality at regular intervals (Humphreys and
Safranyik 1993). Outbreaks often follow severe windthrow events (Safranyik 1985). During
outbreaks, this insect can cause sustained mortality over a number of years, sometimes
13

leading to drastic species conversion and change of stand age structure (Schmid and Frye
1977). The effect is a release of suppressed trees similar to that observed following selective
harvesting of overstorey spruce (Veblen et al. 1991, Lindgren and Lewis 1997). These
effects are most pronounced in stands with a homogeneous, spruce-dominated canopy. In
heterogeneous stands, only the larger spruce trees are successfully attacked, resulting in a less
patch-like disturbance event. Due to the influence of aggregation pheromones, which are
species-specific chemicals produced by the bark beetles to aggregate, attacks on single trees
can often spill over to adjacent trees (Borden 1982). In heterogeneous stands, the likelihood
of such spill over is lower because the large susceptible spruce trees are generally spaced
further apart, and the impact of outbreaks is generally much less severe (Lewis and Lindgren
2000). From an ecological point of view, the long-term impacts of a moderate outbreak on a
stand may be small. From the point of view of forest management, however, even moderate
spruce beetle-caused mortality leads to considerable loss in timber volume, given that a
relatively small proportion of large diameter spruce would contribute a large portion of the
harvestable volume in the stand.
The effect of spruce beetle on stand structure is such that heterogeneity tends to
increase following outbreak (Veblen et al. 1991) and susceptibility of the stand decreases. In
the absence of stand-replacing fire or spruce beetle outbreak, decay fungi may maintain stand
heterogeneity. In stands where root disease or other mortality agents act selectively on
subalpine fir, spruce dominance may increase over time. This leads to significant impact
when spmce beetle populations build up (Lindgren and Lewis 1997, Lewis and Lindgren
1999), which may have a patch initiation-like effect on the stand.

14

Fine-scale processes within the SBS and ESSF
Fine-scale disturbance agents influence forest dynamics through single and smallgroup mortality, and the pattern and frequency of mortality differs with disturbance agent.
With increasingly greater time periods between stand replacing events, fine-scale
disturbances, caused by biotic and abiotic agents, maintain and renew the forest structure in
stands with long fire-retum intervals (Lewis and Lindgren 1999, Newbery 2001).
Fine-scale disturbanee patterns and processes have been eharaeterized within tropical
forests (Denslow 1987, Lawton and Putz 1988), temperate North-American forests (Frelich
and Graumlich 1984, Runkle 1984, Lorimer et al. 1988, Lorimer and Frelich 1989, Payette et
al. 1990, Abrams et al. 1995), eastern sub-boreal forests (Frelich and Reich 1995), western
coastal forests (Canham 1988, Lertzman et al. 1996, Brett and Klinka 1998) and western
subalpine forests (Lertzman and Krebs 1990, Lertzman 1992, Lundquist and Beatty 2002).
Although there is an emerging body of literature suggesting that fine-scale disturbances are
an important element of boreal and sub-boreal forest structure (McCarthy 2001), little is
known about fine-scale disturbance regimes in the SBS and ESSF biogeoclimatic zones
found in central and east-central British Columbia. Further, much of the work on fine-scale
disturbance has been dedicated to describing gap structure and tree regeneration dynamics
within and around canopy gaps (White and Mackenzie 1985, Yamamoto 1995, Kneeshaw
and Bergeron 1998), and not the agents of gap formation. There is increasing evidence that
insects, disease, and abiotic agents collectively play a significant role in gap formation
(Dickman and Cook 1989, Hennon 1995, Lewis and Lindgren 1999), but these various gapforming agents do not necessarily result in identical gap structure and function. The
following sections review three of the most common fine-scale disturbance agents within the
central interior of British Columbia.
15

Decay fungi
Many decay fungi can directly or indirectly cause tree mortality. Normally, decay
fungi weaken the roots and/or the bole to such an extent that breakage occurs. Root and butt
rots are some of the most important gap-making agents in sub-boreal and related forest types
(Veblen 1986, Worrall and Harrington 1988, Holah et al. 1993, Hennon 1995, Lewis and
Lindgren 1999), and therefore, are major contributors to processes that maintain
heterogeneity of stand structure and composition.
Due to decay resistance of sapwood of living trees, decay fungi are often limited to
the heartwood, or areas of dead or injured sapwood. Access to the heartwood by decay fungi
occurs through natural openings, such as branch stubs (true heart rots), and through wounds
(wound-entry heart rots). It takes years from the time of infection for extensive decay to
develop. Therefore, the prevalence of decay fungi usually increases with stand age, becoming
most evident only in older forest stands (Wagener and Davis 1954, Pothier et al. 2004).
Wound-entry heart rot fungi are numerous and usually have broad host ranges that
include many members of particular tree families (e.g., Pinaceae). Many of these fungi are
saprotrophs with a limited ability to decay wood in living trees. For example, Fomitopsis
pinicola (Sw.:Fr) P. Karst is one of the most important decay fungi of dead coniferous wood,
but in boreal ecosystems, it is also known as a wound-entry heart rot (Etheridge 1973).
Decay fungi generally cause breakage of individual trees, although broken trees may
knock over or substantially damage other trees, such that a gap will expand over time
(Hennon 1995). These gaps contribute to structural and species heterogeneity typical of oldgrowth forests in sub-boreal and subalpine forests.

16

Tomentosus root disease
Inonotus tomentosus (Fr.:Fr) S. Teng. is the dominant root pathogen in sprucedominated boreal and sub-boreal ecosystems of B.C. This fungus attacks a broad range of
conifer hosts, but is most frequently found on spruce and pine species. Hardwood species are
thought to be immune to /. tomentosus (Whitney 1962, Whitney and Bohaychuck 1976).
Inonotus tomentosus causes gradual dysfunction of the root system; eventually the
tree dies standing or is windthrown. The disease occurs in small clumps (2-3 trees) that
gradually coalesce into larger areas of infected trees. Therefore the effect on stand structure
depends on the incidence of root disease. At low levels it causes occasional gaps, but at
higher levels (e.g., > 2 0 % spruce infected) 7. tomentosus creates greater structural and
compositional diversity in the forest by causing small openings. Over time, 7. tomentosus can
cause a species shift to sub-alpine fir or other resistant species, or a more open, brushy stand
structure (Whitney 1962). Root diseases are often considered to be diseases of the site. Once
established in an area, they persist from one generation to the next and result in different
processes occurring on infected sites than would occur without root disease. VanderKamp
(1991) has termed this condition a root disease climax.
Tomentosus root disease also interacts with other agents of disturbance. It is thought
to predispose trees to attack by root collar weevils (Cerezke 1994) and other agents that are
more successful on stressed trees. Recent work has indicated that at endemic population
levels, spruce beetles are more suceessful when they attack trees with low to moderate levels
of root disease. At outbreak levels, there was no difference in attack success between healthy
and infected trees (Lewis and Lindgren 2002). Therefore, tomentosus root disease can be an
important stand structural and compositional modifier. It also influences population

17

dynamics of spruce beetle, another late-successional modifier that has varying effeets on the
forest depending on population size.

Western balsam bark beetle
Western balsam bark beetle, Dryocoetes confusus Swaine is an example of an insect
that can cause either fine-scale disturbanees, or intermediate- to coarse-seale (patch)
disturbances. Western balsam bark beetle is widespread in British Columbia wherever
subalpine fir grows (Garbutt 1992). Mortality caused by this beetle can be widespread and
severe, but is generally seattered over large areas and outbreaks are sustained for many years
(Stoek 1991). Harder (1998) observed that attacked trees were found disproportionately in
larger diameter classes. Bleiker et al. (2003) found that suceessfully attacked trees had
redueed growth rates in the last 1 0 years relative to unattaeked trees, indicating that this
insect prefers declining trees. Subalpine fir is subject to many mortality agents, and it has
been speculated that species-specific mortality agents prevent this species from displacing
spruce in many areas (Lewis and Lindgren 1999). The western balsam bark beetle appears to
contribute to the turnover rate, but rarely becomes the driving force.

Conclusions
Within the mature, spruce-subalpine fir forests of central British Columbia the effects
of fine- to coarse-scale disturbance agents can coincide at several temporal and spatial scales.
As the time between stand-replacing events lengthens, the influence of fine-scale disturbanee
on stand structure and dynamics should increase. Despite the simplicity of the hypothesis,
little is known about the fine-scale components of the disturbance regimes within the subboreal and subalpine ecosystems of central British Columbia. To characterize the fine-scale
18

disturbance regimes in any region, the legacy of past intermediate- and coarse-scale
disturbance events, and regional climate fluctuations, must be documented and partitioned as
a source of variation controlling stand dynamics. The following chapters are intended to
stand alone as separate manuscripts; however, together they are intended to assess the above
hypothesis by quantifying and characterizing fine-scale disturbance patterns and processes as
they vary across distinct sub-boreal and subalpine ecosystems of central British Columbia.

19

Literature Cited

Abrams, M., Oewig, D. andDemeo, T. 1995. Dendroecological analysis of successional
dynamics for presettlement origin white-pine-mixed-oak forest in the southern
Appalachians, USA. J. Ecol. 83:123-133.
Agee, J. 1993. Fire ecology of Pacific Northwest forests. Island Press, Washington, DC
Agriculture and Agri-Food Canada, Research Branch. 1998. The Canadian system of soil
classification / Soil Classification Working Group. Imprint Ottawa: NRC Research
Press, Canada.
Andison, D. 1996. Managing for landscape patterns in the sub-boreal spruce forests of
British Columbia. PhD. Dissertation. Faculty of Forestry, University of British
Columbia, Vancouver, BC.
Antos, J. and Parish, R. 2002a. Dynamics of an old-growth, fire-initiated, subalpine forest in
southern interior British Columbia: tree size, age, and spatial structure. Can. J. For.
Res. 32:1935-1946
Antos, J. and Parish, R. 2002b. Structure and dynamics of a nearly steady state subalpine
forest in south-central British Columbia, Canada. Oecol. 130:126-135.
Aplet., G., Laven, R. and Smith, F. 1988. Patterns of community dynamics in Colorado
Fngelmann spruce-subalpine fir forests. Fcol. 69:312-319.
Bleiker, K., Lindgren, S. and Maclauchlan, L. 2003. Characteristics of subalpine fir
susceptible to attack by western balsam bark beetle (Coleoptera: Scolytidae). Can. J.
For. Res. 33:1538-1543.
Borden, J.H. 1982. Aggregation pheromones. Pp. 74-139 in J.B. Mitton and K.B. Sturgeon
(eds.). Bark beetles in North American conifers. University of Texas Press, Austin,
TX, USA.
Brett, R. and Klinka, K. 1998. A transition from gap to tree-island regeneration patterns
in the subalpine forest of south-coastal British Columbia. Can. J. For. Res. 28:18251831.
Canham, C. 1988. Growth and canopy architecture of shade tolerant trees: response to
canopy gaps. Fcol. 69:786-795.
Cerezke, H. F. 1994. Warren rootcollar weevil, Hylobius warreni Wood (Coleoptera:
Curculionidae), in Canada: Ecology, behaviour, damage relationships, and
management. Can. Fntomol. 126:1383-1446
Clements, F. F. 1916. Plant succession: an analysis of the development of vegetation.
20

Carnegie Institution of Washington. Washington, DC.
DeLong, C. 1998. Natural disturbance rate and patch size distribution of forests in
northern British Columbia: Implications for forest management. Northwest Soi 72:3548.
DeLong, C. 2002. Natural Disturbance Units of the Prince George Forest Region:
Guidance for Sustainable Forest Management.
ftp://prgftp.env.gov.bc.ca/pub/outgoing/srm/planning/landscape level Accessed
September 1, 2005
DeLong, S.C. and Kessler, W.B. 2000. Ecological characteristics of mature forest
remnants left by wildfire. For. Ecol. Man.. 131:93-106.
DeLong, C. and Tanner, D. 1996. Managing the pattern of forest harvest: lessons form
wildfire. Bio. Con. 5:1191-1205.
DeLong, C., Tanner, D. and lull, M. 1993. A field guide for site identification and
interpretation for the southwest portion of the Prince George Forest Region. British
Columbia Ministry of Forests, B.C. Land Management Handbook 24.
DeLong, C., Tanner, D., and lull, M. 1994. A field guide for site identification and
interpretation for the Northern Rockies portion of the Prince George Forest Region.
British Columbia Ministry of Forests, B.C. Land Management Handbook 29.
Denslow, J.S. 1987. Tropical rain forest gaps and tree species diversity. Pages 307-323 in
The ecology of natural disturbance and patch dynamics. S. Pickett and P. White,
editors. Academic Press, New York, NY, USA.
Dickman, A. and Cook, S. 1989. Fire and fungus in a mountain hemlock forest. Can. J. Bot.
67:2005-2016.
Drury, W. and Nisbet, I. 1973. Succession. J. For. 50:12-16.
Egler, F. E. 1954. Vegetation science concepts I. Initial floristic composition a factor
in old field vegetation development. Vegetatio, 4:412-417.
Etheridge, D.E. 1973. Wound parasites causing tree decay in BC. Can. For. Serv., Pac. For.
Res. Cent., BC. Forest Pest Leaflet No. 62. Victoria, BC.
Frelich, L. and Graumlich, L. 1994. Age-class distribution and spatial patterns in oldgrowth hemlock-hardwood forests. Can. J. For. Res. 24:1939-1947.
Frelich, L. and Reich, P. 1995. Spatial patterns and succession in a Minnesota southernboreal forest. Ecol. Mon. 65:325-346.

21

Garbutt, R. 1992. Western balsam bark beetle. Can. For. Serv., Pac. For. Res. Cent. BC.
Forest Pest Leaflet No. 64. Victoria, BC.
Harder, L. 1998. Impact of the western balsam bark beetle, Dryocoetes confusus Swaine, at
the Sicamous Creek research site, and the potential for semiochemical based
management in alternative silviculture systems. M.P.M. Thesis, Simon Fraser
University, Burnaby, BC
Hawkes, B. 1980. Fire history in Kananaskis Provincial Park - Mean fire return intervals, pp.
42-45 in Proceedings of the fire history workshop, Oct. 20-24, 1980, Tucson,
Arizona. USDA, For. Serv. Gen. Tech. Rep. RM-81. Tucson, AZ, USA.
Hawkes, B., Vasbinder, W. and DeLong, C. 1997. Retrospective Fire Study Final Report:
Fire regimes in the SBSvk and ESSFwk2/wc3 biogeoclimatic units of northern British
Columbia. McGregor Model Forest Association, Prince George, BC.
Heinselman, M. 1981. Fire intensity and frequency as factors in the distribution and
structure of northern ecosystems, pp.7-57 in Fire Regimes and Ecosystem Properties,
December 11-15, 1978, Honolulu, Hawaii. USDA Forest Service Gen. Tech. Rep.
WO-26. Washington, DC.
Hennon, P.E. 1995. Are heart-rot fungi major factors of disturbance in gap-dynamic
forests? Northwest Sci. 69:284-293.
Holah, J.C., Wilson, M.V. and Hansen, E.M. 1993. Effects of a native forest pathogen,
Phellinus weirii, on Douglas-fir forest composition in western Oregon. Can. J.
For. Res. 23:2473-2480.
Humphreys, N., and Safranyik, L. 1993. Spruce beetle. Nat. Res. Can., Pac. For. Cent., Forest
Pest Leaflet No. 13. Victoria, BC.
Johnson, E.A. 1992. Fire and vegetation dynamics: studies from the North American
Boreal forest. Cambridge studies in ecology. Cambridge Univ. Press, Cambridge,
UK.
Johnson, E. and Rowe, J. 1977. Fire and vegetation change in the western sub-arctic.
Canadian Department of Indian Affairs and Northern Development, ALUR 75:76-61.
Jull, M., DeLong, C., Eastham, A., Sagar, R. Stevenson, S. and DeLong, R. 1996. Testing
silviculture systems for the ESSF: Early results of the Lucille Mountain Project.
Ministry of Forests Research Program, Forest Research Note, Note # PG-01. Victoria,
BC.
Kimmins, J. P. 1999. Biodiversity, Beauty and the “Beast”: Are beautiful forests sustainable,
and is “small” always ecologically desirable? For. Chron. 75:955-960.

22

Kneeshaw, D.D. and Bergeron Y. 1998. Canopy gap characteristics and tree replacement in
the southeastern boreal forest. Ecol. 79:783-794.
Kneeshaw, D. and Burton, P. 1997. Canopy and age structures of some old sub-boreal
Picea stands in British Columbia. J. Veg. Sci. 8:615-626.
Lawton, R. and Putz, F. 1988. Natural disturbance and gap formation in a wind exposed
tropical cloud forest. Ecol. 69:764-777.
Lertzman, K. 1992. Patterns of gap-phase replacement in a subalpine, old-growth forest.
Ecol. 72:657-669.
Lertzman, K. and Krebs, C. 1991. Gap-phase structure of a subalpine old-growth forest.
Can. J. For. Res. 21:1730-1741.
Lertzman, K.P., Sutherland, G.D., Inselberg, A. and Sunders, S.C. 1996. Canopy gaps and
the landscape mosaic in a coastal temperate rain forest. Ecol. 77: 1254-1270.
Lewis, K. and Lindgren, S. 1999. Influence of decay fungi on species composition and
size class structure in mature Picea glauca x engelmannii and Abies lasiocarpa in
sub-boreal forests of central British Columbia. For. Ecol. Man. 123:135-143.
Lewis, K. and Lindgren, S. 2000. A conceptual model of biotic disturbance ecology in
the central interior of BC: How forest management can turn Dr. Jekyll into Mr. Hyde.
For. Chron. 76:433-443.
Lewis, K. and Lindgren, S. 2002. Relationships between spruce beetle and tomentosus
root disease: two natural disturbance agents of spruce. Can. J. For. Res. 32:31-37.
Lindgren, S. and Lewis K. 1997. The natural role of spruce beetle and root pathogens in a
sub-boreal spruce forest in central British Columbia: a retrospective study. Pages 122130 in J.C. Grégoire, A.M. Liebhold, F.M. Stephen, K.R. Day and S.M. Salom, eds.
Proceedings: integrating cultural tactics into management of bark beetle and
reforestation pests. USDA Forest Service General Report NE-236. USA.
Lorimer, C. and Frelich, L. 1989. A methodology for estimating canopy disturbance
frequency and intensity in dense temperate forests. Can. J. For. Res. 19:651-663.
Lorimer, C., Frelich, L. and Nordheim, E. 1988. Estimating gap origin probabilities for
canopy trees. Ecol. 69:778-785.
Lundquist, J. and Beatty, J. 2002. A method for characterizing and mimicking forest
canopy gaps caused by different disturbances. For. Sci. 48:582-594.
McCarthy, J. 2001. Gap dynamics of forest trees: A review with particular attention to boreal
forests. Environ. Rev. 9:159.

23

Meidinger, D. and Pojar, J. 1991. Ecosystems of British Columbia. Crown Publications Inc.
Special Report Series No. 6 . Victoria, BC.
Newbery, J. E. 2001. Small scale disturbance and stand dynamics in Inonotus tomentosus
infected and uninfected old-growth and partial cut, wet, sub-boreal forests in
British Columbia. M.Sc. Natural Resource Management, University of Northern
British Columbia, Prince George, BC.
Oliver, C. and Larson, B. 1996. Forest Stand Dynamics. John Wiley & Sons, Inc. New York,
NY, USA.
Parish, R. Antos, J. A. and Fortin, M.-J. 1999. Stand development in an old-growth
subalpine forest in southern interior British Columbia. Can. J. For. Res. 29:13471356.
Payette, S., Filion, L. and Delwaide, A. 1990. Disturbance regimes of a cold temperate
forest as deduced from tree-ring patterns: the Tantare Ecological Reserve,
Quebec. Can. J. For. Res. 20:1228-1241.
Pyne, S. 1984. Introduction to wildland fire. John Wiley, New York, NY, USA.
Pothier, D., Raulier, F. and Riopel, M. 2004 Ageing and decline of trembling aspen stands in
Quebec. Can. J. For. Res. 34:1251:1258.
Romme, W. 1980. Fire frequency in subalpine forests of Yellowstone National Park, pp. 2730 in Proceedings of the fire history workshop, Oct. 20-24, 1980, Tucson, Arizona,
USDA, For. Serv. Gen. Tech. Rep. RM-81. Fort Collins, CO, USA.
Runkle, J. 1984. Patterns of disturbance in some old-growth forests of the eastern United
States. Fcol. 63:1533-1546.
Safranyik, L. 1985. Infestation incidence and mortality in white spruce stands by
Dendroctonus rufipennis Kirby (Coleoptera, Scolytidae) in central British Columbia.
Z. Fntomol. 99:86-93.
Sanborn, P., Geertsema, M., Jull, A. and Hawkes, B. 2001. Soil evidence for Holocene
fires in an inland temperate rainforest: east-central B.C. In Canadian Quaternary
Association Meetings, 2001: Program and Abstracts. Government of the Yukon,
Heritage Branch, Whitehorse, Y.T. Occasional Papers in the Earth Sciences 1.
Schmid, J.M., and Frye, R.H. 1977. Spruce beetle in the Rockies. U.S. Dept, of Agric., For.
Serv., Rocky Mtn. For. Range Fxp. Stn., Gen. Tech. Rep. RM-49, Fort Collins, CO.,
USA.
Stock, A.J. 1991. The western balsam bark beetle, Dryocoetes confusus Swaine: impact and
semiochemical-based management. Ph.D. Thesis, Simon Fraser University, Burnaby,
B.C.
24

VanderKamp, B.J. 1991. Pathogens as agents of diversity in forested landscapes. For. Chron.
67:353-354.
Veblen. T. 1986. Treefalls and the co-existence of conifers in sub-alpine forests of the
central Rockies. Ecol. 67:644-649.
Veblen, T. Hadley, K, Marion, S. and Rebertus, A. 1991. The Response of Subalpine
Forests to Spruce Beetle Outbreak in Colorado. J. Ecol. 72:213-231.
Veblen, T., Hadley, K., Nel, E., Kitzberger, T, Reid, M. and Villalba, R. 1994.
Disturbance regime and disturbance interactions in a Rocky mountain subalpine
forest. J. Ecol. 82:125-135.
Wagener, W. and Davidson, R. 1954. Heartrots in living trees. Bot. Rev. 20:61-134.
White, P. S. 1979. Pattern, process, and natural disturbance in vegetation. Bot. Rev.
45:229-299.
White, P.S. and MacKenzie, M.D. 1985. Natural disturbance and gap phase dynamics in
southern Appalachian spruce-fir forests. Can. J. For. Res. 15:233-240.
Whitney, R.D. 1962. Studies in Forest Pathology XXIV: Polyporus tomentosus Fr. as a major
factor in stand-opening disease of white spruce. Can. J. Bot. 40:1631-1658.
Whitney, R.D. and Bohaychuck, W.P. 1976. Pathogenicity of Polyporus tomentosus and P.
tomentosus var. circinatus on seedlings of 11 conifer species. Can. J. For. Res. 6:129131.
Worrall, J.J. and Harrington, T.C. 1988. Etiology of canopy gaps in spruce-fir forests at
Crawford Notch, New Hampshire. Can. J. For. Res. 18:1463-1496.
Yamamoto, S. 1995. Gap characteristics and gap regeneration in subalpine old-growth
coniferous forests, central Japan. Ecol. Res. 10:31-39.

25

^

#

Study Area

City/town
Road/Rail

B iogeoclim atic variant

^SBSdk

[

SBS d w 3
SBS me 2

g g SBS mk 1
SBS wk 1
m a SBSvk

R,.!:.

r

ESSFwk 2

amESSFrnvl

I 2: E S S F m v 2
f
ESSFm v3
m i l C H v k 2w k 2
[

Alpine tu n d ra

100
mm K ilom eters

Figure 1.1. Locations of the Sub-Boreal Spruce and the Engelmann Spruce-Subalpine Fir
biogeoclimatic zones and study areas.

26

CHAPTER 2 - DETECTING STAND AND LANDSCAPE-LEVEL DYNAMICS
WITHIN MATURE SPRUCE-SUBALPINE FIR FORESTS OF CENTRAL BRITISH
COLUMBIA: DENDROECOLOGICAL CONSIDERATIONS I

Abstract
This study used dendroecological techniques to examine the influence of past
intermediate- and coarse-scale disturbance events, and regional climatic variation, on tree
growth and stand development. As part of a larger project intended to characterize fine-scale
disturbance in central British Columbia, three forested ecosystems were selected: dry subboreal, mesic sub-boreal and wet subalpine. Within each ecosystem, five 0.25-ha plots were
located in mature spruce-subalpine fir stands (n = 15). Within each plot, increment cores
were collected at breast height from all living and dead trees >15 cm diameter at breast height
(dbh), species was tallied and dbh recorded. Cores were mounted, sanded, aged and ringwidth series measured to the nearest 0 . 0 1 mm. Two standardization techniques were used to
identify disturbance and infer climatic variation at scales greater than the individual plot. In
the first standardization method, used to highlight long-term growth trends (e.g., spruce
beetle outbreaks or stand-replacing fire), all series were transformed by dividing the observed
ring-widths by each series’ mean ring-width, and then averaged by species and by plot (n =
30). In the second standardization method, used to highlight short-term growth trends (e.g.,
climate), all series were transformed by fitting a cubic-smoothing-spline of 50 % frequency
response of 50 years, and then averaged by species and by plot (n = 30). Periods of canopy
mortality occurred in the ESSFwkl variant during the 1710s and 1830s, in the SBSwkl
variant during the 1830s and 1870s, and in the SBSdw3 variant during the 1870s. Based on
historical documentation and dendroecological evidence, these periods of canopy mortality
27

were likely caused by spruce beetle. There is strong evidence that these events caused abrupt
shifts in species composition. A western balsam bark beetle outbreak occurred in the
ESSFwk2 variant during the 1920s and 1930s. Dendroecological evidence provides strong
evidence that species dominance shifted abruptly from subalpine fir to spruce within these
stands. The results of this study clearly demonstrate that dendroecological techniques can be
used to date and examine the effects of past intermediate- and coarse-scale disturbance events
and regional climate fluctuations. Furthermore, these events can have a significant influence
on tree-growth and stand development, and therefore should be taken into consideration
before estimating fine-scale disturbance frequencies.

28

Introduction^
The sub-boreal and subalpine forest ecosystems of central British Columbia have
been receiving greater attention with regard to their natural disturbance regimes (Andison
1996, DeLong and Tanner 1996, Hawkes et al. 1997, DeLong 1998, Newbery 2001, Sanborn
et al. 2001). Historically, stand-replacing fire was considered the primary disturbance agent
controlling stand and landscape-level dynamics (Johnson 1992, Andison 1996, DeLong and
Tanner 1996, Hawkes et al. 1997, DeLong and Kessler 2000). Although in most cases these
forests do originate following fire, recent studies have shown that the times between standreplacing events are enormously varied. For example, DeLong and Tanner (1996) reported
fire-retum intervals ranging from 227-345 years in drier forest types in the foothills of the
Rocky Mountain. Hawkes et al. (1997) report fire-retum intervals in extremely wet and cool
forest types in the central Canadian Rocky Mountains ranging from 1200-6250 years.
Two major problems occur when defining the disturbance regime of a forested
ecosystem exclusively by the disturbance agents that cause stand-replacement over coarse
spatial scales (e.g., coarse-scale fire in boreal ecosystems). First, disturbance events are not
strictly coarse-scaled, stressing and/or causing the death of the majority of canopy trees over
large contiguous areas (White 1979). Disturbance events can be fine-scaled, causing single
and small group mortality. Furthermore, the fine-scale components of a disturbance regime
ean be stand-maintaining as opposed to stand-replacing (Lewis and Lindgren 1999). Standreplacing disturbance events typically promote stand-initiation, even-aged stand structure and
shifts to early-serai species composition (Oliver and Larson 1996). Conversely, stand-

' A version o f this document will be submitted for publication under multiple authors. All data used in this
study were collected by R. D. Thompson and a number of field and lab assistants. In this thesis I use 'we' and
‘our’ to reflect the involvement of others in various aspects o f this work.

29

maintaining disturbance events typically cause gap-formation, uneven-aged stand structure
and the maintenance of late-seral species composition or slow shifts in species composition
(Lewis and Lindgren 1999). The second problem occurs because a gradient between fineand coarse-scale disturbance agents can exist, leading to intermediates between the fine- and
coarse-scaled disturbance processes (Antos and Parish 2002). With increasing periods
between stand-replacing events, fine- to intermediate-scale disturbance events are
increasingly important in forest dynamics (Lewis and Lindgren 2000). These disturbances,
caused by a collection of biotic and abiotic agents, and interactions between agents, can
maintain and renew forest structure in stands with long fire-retum intervals (Lewis and
Lindgren 2000). Due to our traditional management and scientific focus on coarse-scale
disturbance, however, little is known about these fine-scale disturbance processes. This is
particularly true in wet temperate forests with long intervals between stand-replacing events.
In order to advance our understanding of the disturbance regimes within the central interior
of British Columbia, we must not focus solely on coarse-scale processes, but include the
characterization of fine-scale processes and determine their relative importance in
maintaining ecosystem function and health.
Tree-ring patterns contain a wealth of information and dendroecological analyses
offer the possibility of examining the effects of past climate, disturbance and competition on
tree growth and stand development, provided that the effects of these factors can be
disentangled (Parish et al. 1999). Within sub-boreal and subalpine forest ecosystems the
effects of fine- to coarse-scale disturbance agents can coincide at several temporal and spatial
scales (Parish et al. 1999, Lewis and Lindgren 2000). Furthermore, forested ecosystems in
central British Columbia have been shown to be sensitive to climatic variation, and attempts
have been made to reconstruct past climate to understand the specific growth-climate
30

relationships (Zhang et al. 1999). Therefore, we expect that the influence of disturbance and
climate together produces a diversity of tree-ring growth responses. Furthermore, these
signals can be complex and often inseparable (Parish et al. 1999).
To characterize the fine-scale disturbance regimes in any region, the legacy of past
intermediate- and coarse-scale disturbance events, and regional climate fluctuations, must be
documented. Our objectives were to develop size- and age-class distributions, and a set of
stand- and regional-level chronologies, to diagnose past intermediate- and coarse-scale
disturbance events and climatic fluctuations that have significantly influenced stand
development. Subsequent chapters use the diagnoses of these events to aid in the
reconstruction of individual stand histories and to compare and contrast stand structure,
composition and dynamics in the mature spruce-subalpine fir forests of central British
Columbia.

Materials and Methods
Field Sampling
We selected three biogeoclimatic variants to represent the climatic gradient that spans
the dry-warm interior plateau to the wet-cold foothills of the Canadian Rockies (Chapter 1,
Figure 1.1). The Stuart Dry Warm Sub-Boreal Spruce (SBSdwS), Willow Wet Cool SubBoreal Spruce (SBSwkl) and Misinchinka Wet Cool Engelmann Spruce-Subalpine Fir
(ESSFwk2) biogeoclimatic variants are located in two of the most common biogeoclimatic
zones located in central British Columbia. The SBSdw3, SBSwkl and ESSFwk2
biogeoclimatic variants will be referred to as the dry, mesic and wet variants, respectively.
Within each variant, we selected five forest stands. Stands were selected using
several criteria. First, forest stands were beyond the stem exclusion stage as defined by
31

Oliver and Larson (1996). This criterion was used to select stands in which trees were
establishing in the understorey and, in the absence of stand-replacing disturbances, were
advancing toward the true old-growth stage of development (Oliver and Larson 1996).
Second, forest soils had intermediate soil moisture (mesic) and nutrient (mesotrophic)
regimes (Meidinger and Pojar 1991). This criterion was intended to minimize variation in
tree species growth characteristics and disturbance processes as a result of highly divergent
edaphic influences of soil moisture and nutrients. Third, forest stands had a mixture of
spruce and subalpine fir as the dominant component of the canopy, with basal area >80 %,
for all trees over 15 cm diameter at breast height (dbh, measured 1.3 m above the ground). A
common tree composition and density minimized variation in stand susceptibility to
disturbance. Finally, forest stands had no evidence of industrial forest activities. This final
criterion ensured that stand dynamics resulted from natural disturbances, which may include
past traditional uses and practices of First Nations. Consequently, stands are not
representative of the average of the biogeoclimatic variant, but rather, they are typical of
stands satisfying the criteria within each variant.
Within each stand, one 0.25-ha plot was located by randomly placing the south­
western plot comer, which was defined as the plot origin (0, 0 m), for a total of 15 plots.
From the plot origin, plot boundaries were defined by a 50-m x 50-m closed traverse along
each cardinal bearing. At each plot comer and traverse midpoint (i.e., 25 m), stakes were
placed and numbered one to eight.
For all living and dead trees with dbh >15 cm, we mapped the x and y coordinates
relative to the plot origin. All coordinates were calculated using two azimuths taken from
any two adjacent stakes with a 25-m base and related back to the origin using trigonometry.
Coordinates were used for revisiting each plot and spatial analyses in subsequent chapters.
32

Trees were numbered, species tallied and dbh measured. For all dead and downed material,
with dbh >15 cm, dbh was recorded based on an estimate of breast height measured from the
root collar and species was tallied if it could be determined based on bark characteristics.
Diameter distributions were plotted for living spruce and subalpine fir, and for dead spruce,
subalpine fir and the unidentifiable trees, separately for each plot. Differences in the
diameter distributions were analyzed using Kruskal-Wallis single-factor analysis of variance
by ranks (Zar 1996, SPSS Inc. 1999).

Dendroecological analyses
Increment cores were taken at breast height from each live and dead tree. Cores were
taken at breast height because they tend to have less stem decay than cores taken lower on the
stem enabling maximum length in ring-width series. Further, a large number of the dead
trees in the study areas exhibited butt-rot which was largely avoided by coring at breast
height. We only cored dead trees if the outer most tree ring corresponded to the last ring
formed prior to tree death. We assumed that the outer ring was the last ring formed prior to
tree death if the bark was intact and/or the sapwood was present and firm. Following the
methods of Storaunet and Rolstad (2002), cross-sectional discs were collected from felled
snags and logs if intact sapwood, on at least one portion of the tree, could be found and used
to determine the date of death by cross-dating. Tree-ring series from disc samples were used
to develop chronologies.
Cores were prepared following standard dendrochronology techniques (Stokes and
Smiley 1968). Discs were frozen after field sampling and prepared and measured based on
the methods of Herman et al. (1972). Individual ring-width series were measured to the
nearest 0.01 mm using WinDENDRO (Regent Instruments Inc. 2000). For discs and cores
33

that had periods of extremely narrow ring-widths (individual rings <0.5 mm wide) the ringwidth series were measured to the nearest 0.001 mm using the Velmex “TA” System
(Velmex, Inc. 1992) in conjunction with MeasureJ2X (VoorTech Consulting 2004).
Ring-width series were cross-dated by matching the ring-width patterns among
samples from live and dead trees. Cross-dating assures the proper placement of calendar
years to each growth ring and identifies missing, partial, locally absent and false rings (Fritts
1976). We used the computer program COFECHA (Holmes 1983) and visually inspected the
samples to detect measurement and cross-dating errors. Once the ring-width series were
dated, frequency distributions of age at breast height were plotted for living spruce and
subalpine fir, separately for each plot. Disc samples were not used to develop age-class
distributions because they were not necessarily taken from breast height. A few core samples
did not include the pith; therefore, the number of missed rings was calculated as the estimated
distance to the pith divided by the mean ring-width of the first five complete rings.
Standardization is a common technique used to estimate and remove systematic
variation in individual ring-width series that is not associated with an environmental signal of
interest (Cook and Peters 1981, Cook 1987). A signal is defined here as the information in
tree rings which is relevant to the study of a particular problem (Cook 1987). Standardization
involves fitting the observed ring-width series with a predictive function and computing an
index by dividing the observed value by the predicted value. This reduces variance among
series and transforms individual ring-widths into dimensionless index values.
Standardization permits computation of average ring-width chronologies without the average
being dominated solely by faster growing trees with large ring widths (Veblen et al. 1991b).
Two standardization techniques were used to identify unique disturbance and climatic
events at scales greater than the individual plot. In the first, horizontal-line standardization,
34

all cross-dated series were transformed by dividing the observed ring width by the mean
series ring width. Veblen et al. (1991b) found that standardizing series by their mean
preserved long-term growth trends and could be used to identify canopy disturbance. Using
the program ARSTAN (Cook and Holmes 1984) individual series were standardized and
averaged by species and by plot (n = 30). These chronologies were used to; (1) date partial
canopy removal severe enough to register within a majority of the existing canopy trees; and
(2) reveal tree growth patterns that indicate stand-replacement events. Dating events
associated with stand-replacement and partial canopy removal will be used to assist in the
differentiation between these typically coarse- and intermediate-scaled events from more
fine-scaled processes.
The goal of the second standardization was to develop a set of chronologies from
which we could interpret short-term growth trends. Included in this category would be
defoliator cycles, inter-annual and short-term intra-annual variations in climate. Fine- to
coarse-scale disturbance events are common in the spruce-subalpine fir forest types of central
British Columbia, and plots were selected to sample disturbance effects rather than to
highlight the year-to-year climate variation. We evaluated numerous methods of
standardization and found the strongest estimates of the short-term population signal using
stochastic opposed to deterministic methods (Cook, 1987). Using the program ARTS AN
(Cook 1985) ring-width series were standardized by fitting a cubic-smoothing spline of 50 %
frequency response of 50 years and averaged by species and by plot (n = 30).
Marker-ring analysis, within- and between-species correlations, and principal
component analysis (PCA) were used to assess individual cubic-spline chronologies within
each ecological variant. Wide and narrow marker rings that are temporarily synchronous
among chronologies suggest a common influence on tree growth (Daniels and Watson 2003).
35

We defined marker rings as years when the normalized ring-width indices of the standard
chronologies were more than one standard deviation from the chronology mean. The degree
of commonality of these positive and negative marker rings was assessed by examining the
percentage of chronologies showing marker rings in each year. Years when greater than 60
% of the chronologies had a marker ring of the same sign were considered common (Daniels
and Watson 2003).
Within-species correlation matrices were calculated for each of spruce and subalpine
fir in each ecological variant (2 species x 3 ecological variants = 6 matrices) (SPSS Inc.
1999). Each pair-wise matrix correlates the five species-specific chronologies against one
another resulting in 10 unique r values per matrix. Between-species correlation matrices
were calculated for each ecological variant (SPSS Inc. 1999). Each pair-wise matrix
correlates each spruce chronology against each subalpine fir chronology resulting in 25
unique r values per matrix.
Separate PCA, using varimax rotation, were run for each ecological variant (SPSS
Inc. 1999, Tabachnick and Eidell 2001). PCA analyses were conducted over the common
pair-wise intervals for the 1 0 chronologies within each ecological variant. If species respond
differently to climatic variation then the PCA should group the chronologies accordingly.

Results
Diameter and age structure at breast height
Diameter distributions of living spruce and subalpine fir varied between plots within
and between each ecological variant (Figure 2.1). Within the dry variant, plots 1 and 2 were
dominated by subalpine fir in the smallest size-classes and plots 3 and 4 were dominated by
spruce in the smallest size-classes. Plots 1 and 2 are at a latter stage of stand development
36

based on the ingress of subalpine fir into smaller size-classes, and the flatter distribution and
lower density of spruce relative to subalpine fir. Plot 5 represents an intermediate stage of
stand development between the two pairs, respectively (Figure 2.1). All plots in the mesic
variant were similar to plots 1 and 2 of the dry variant, and all plots in the wet variant were
similar to plots 3 and 4 of the dry variant (Figure 2.1). In the mesic variant, spruce had
flatter distributions (Figure 2.1) and consistently larger maximum diameters compared with
subalpine fir (%^ = 6.818, df = 1, p < 0.01, Figure 2.1). In contrast, subalpine fir was more
abundant in smaller size-classes compared with spruce within the mesic variant (Figure 2.1).
Within the wet variant, the diameter distributions of both spruce and subalpine fir formed a
re verse-J; however, spruce had larger maximum diameters than subalpine fir

= 6.818, df =

1, p < 0.01, Figure 2.1).
Within the dry variant, plots are primarily single cohort stands (Figure 2.2). Plots 1
and 2 have dominant subalpine fir cohorts at approximately 130 years breast height age, with
residual spruce and fir up to 270 years. Plots 3, 4 and 5 all have dominant spruce cohorts
with approximately 110-130 years breast height age, with a few residuals up to 210 years.
Together, these observations provide further evidence that plots 1 and 2 within the dry
variant are at a latter stage of stand development than plots 3, 4 and 5. Within the mesic
variant, plots range from 50 to 250 years at breast height. Plots 1 and 3 have a bimodal
distribution of breast height ages with subalpine fir around 110 years and spruce around 230
years. Plots 2, 4 and 5 have multimodal distributions of breast height age with old spruce and
one or two younger influxes composed of spruce and/or subalpine fir. Within the wet
variant, most trees are <170 years old at breast height, with a few residuals up to 300 years
old. In plot 3, most spruce attained breast height 130 to 150 years ago and in the remaining
plots spruce and subalpine fir attained breast height 50 to 150 years ago (Figure 2.2).
37

The general form of the diameter distributions of dead spruce, subalpine fir and
unidentifiable trees varied in plots within and among each ecological variant (Figure 2.3).
Within the dry variant, the dead tree diameters from plots 1 and 2 are greater than plots 3, 4
and 5

= 107.8, df = 1, p < 0.001). Plots 1 and 2 were dominated by subalpine fir in the

smallest size-classes and plots 3 and 4 were dominated by spruce in the smallest classes.
Within plot 5, subalpine fir dominated the smallest size class and spruce dominated in the
larger size-classes. Similar to the live tree diameter distributions, plot 5 represents an
intermediate stage of stand development between plots 1 and 2, and plots 3 and 4.
Furthermore, plots 1 and 2 are indicative of a latter stage of stand development based on the
greater abundance of unidentifiable trees in all size-classes and the abundance of spruce and
subalpine fir larger size-classes. All plots in the mesic variant were similar to plots 1 and 2
of the dry variant. All plots in the wet variant, although similar to plots 3 and 4 of the dry
variant with regard to their size-class distributions of dead spruce and subalpine fir, had a
larger component of trees without bark, indicating a later stage of stand development.

Ring-width chronologies
All chronologies have a maximum sampling depth >20 trees and 90 % of the
chronologies have starting dates >200 years before present (Table 2.1). Furthermore, 93 %
of the chronologies have mean between-tree correlations >0.40 (Table 2.1). The maximum
sampling depths, chronology start dates and mean correlations for all 30 chronologies, using
horizontal standardization, suggest that adequate replication was attained to detect radial
growth patterns indicative of stand-replacement and/or partial canopy removal.
Of the five plots in the dry variant, the chronologies from plots 1 to 2 included a
period in which spruce showed increased growth, possibly the result of a sand-replacing
38

event (Figure 2.4). Based on the limited sampling depths during these periods, however, it is
difficult to place accurate dates on stand-replacement. Similarly, all of the chronologies in
the mesic variant indicate an initial period of rapid growth (Figure 2.4). In contrast to the dry
and mesic variants, the majority of the chronologies in the wet variant do not indicate an
initial period of rapid growth (Figure 2.4).
Within the dry variant, the chronologies of plots 1 and 2 have a minor sustained
increase in index values after the 1870s (Figure 2.4). Within the mesic variant, all plot
chronologies also show a similar pattern during the 1830s and 1870s (Figure 2.4). Within the
wet variant, there are three periods of increased index values: the first period of increase in
the 1710s; the second increase in the 1830s; and the third in the early 1930s (Figure 2.4).
Across all ecological variants, the chronologies using cubic-spline standardization
included 21 to 54 cores and spanned at least 159 years (Table 2.2). In all variants, the mean
chronology length was consistently longer for spruce compared with subalpine fir (Table
2.2). The mean sensitivity and standard deviation values provide a measure of the year-toyear variability in the chronologies (Fritts 1976). The average mean sensitivities, standard
deviations and first-order autocorrelations for the spruce chronologies within the dry, mesic
and wet variants are similar to the shorter subalpine fir chronologies (Table 2.2). The
expressed population signal (EPS) is a measure, based on a finite number of samples, of how
well a chronology approximates the theoretical population (Cook and Kariukstis 1990). This
is related to the strength of the common signal and sample replication. Mean correlation
between trees values (rbO also serve as a preliminary and conservative estimate of the
common signal in the chronologies. The number of trees required to attain an EPS of 0.85
has been suggested as desirable for approximating a common signal within the theoretical
population (Briffa 1995, Briffa and Jones 1990). The mean correlations for spruce
39

chronologies within the dry, mesic and wet variants (0.44, 0.47 and 0.45, respectively)
indicate that, on average, seven or eight samples are required to yield the suggested EPS
(Table 2.2). Correlations for subalpine fir within dry, mesic and wet variants (0.42, 0.44 and
0.37, respectively) indicate that, on average, eight to thirteen samples are required to yield the
suggested EPS (Table 2.2). The greater number of samples required for subalpine fir within
the wet variant, implied by the lower rtt, may suggest that the subalpine fir populations
respond differently and more variably to climate and past disturbance (Table 2.2). In all
ecological variants, there are a number of years where greater than 60 % of the chronologies
have common marker rings suggesting that the chronologies possibly have a common climate
related signal within each variant (Figure 2.5).
Eighty-five percent of the correlations over the maximum pairwise intervals exceeded
0.4 and 98 % of the correlations were significant (p < 0.05). Except for a few pairs, all
correlations were strong with a mean of 0.62 (n = 60). Seventy-nine percent of the
correlations over the maximum pairwise intervals exceeded 0.4 and 8 8 % of the correlations
were significant (p < 0.05). Again, except for a few pairs, all correlations were quite high
with a mean of 0.49 (n = 75).
Within the dry variant, two components where identified by the PCA and accounted
for 82 % of the variance within the 10 chronologies (Table 2.3). These components were
retained and subjected to an orthogonal varimax rotation. The loadings on the two
components grouped the spruce and subalpine fir of plots 3, 4 and 5 and the spruce and
subalpine fir of plots 1 and 2 , suggesting distinct differences in group responses to cubicspline standardization and variation in stand dynamics. Within the mesic variant, two
components accounted for 81 % of the variance within the 10 chronologies (Table 2.3). The
loadings identified two groups with the subalpine fir of plot 3 separate from all other
40

chronologies. Within the wet variant, two components accounted for 84 % of the variance
within the 10 chronologies (Table 2.3). The loadings identified two groups with all
chronologies loading together except for the subalpine fir chronologies of plots 1, 2 and 4.
The above analyses, along with an examination of the co-located chronologies (Figure
2.6), suggest common ring-width variability. Based on the spatial scale of these
chronologies, it is justifiable to assume that the common variability between the two species
is controlled by a common climatic factor at a regional scale. Disturbance histories,
however, have resulted in important differences between species and between plots,
especially within the dry variant. Subsequent analysis of short-term variation in the
chronology dataset, therefore, should include mean regional chronologies for each species.
In addition, based on these analyses, the dry variant requires further separation into two
groups based on the differences in stand development. Plots 1 and 2 and plots 3, 4 and 5 will
be combined into groups representing late and early-successional stages, respectively.
Eight final cubic-spline chronologies have been built for spruce and subalpine fir
within the central interior of British Columbia (Table 2.4 and Figure 2.7). ARSTAN first
order autocorrelations (Cook 1985) within the final chronologies were strong, suggesting that
ring-width values are highly autocorrelated with the previous ring-width (Table 2.4).
Therefore, post-hoc analyses of extended periods of below and above average growth should
not include 2-year periods. Consequently, below average growth was defined as index values
<1 for >3 consecutive years. There were no significant differences in the duration of below
average growth across the ecological variants (F3 ,4 = 5.71, p = 0.063) and there were no
significant differences in the duration of below average growth between spruce and subalpine
fir nested within ecological variant (F4 ,9 o = 0.441, p = 0.779). Mean duration of below
average growth was four years and ranged from 3-16 years. The duration of above average
41

growth, defined as index values >1 for >3 consecutive years, were not significantly different
by ecological variant (F3 4 = 0.115, p = 0.947) or by species nested within ecological variant
(F4 9 5 = 1.604, p = 0.179). Mean duration of above average growth was five years and ranged
from 3-13 years.

Discussion
The results of this study clearly demonstrate that dendroecological techniques can be
used to date and examine the legacy of intermediate- and coarse-scale disturbance events and
regional climate fluctuations within the mature spruce and subalpine fir forest types of central
British Columbia. Furthermore, the two standardization procedures used within this study
suggest that intermediate- and coarse-scale disturbances and variation in climate have a
substantial influence on tree-growth and stand development, and should be taken into
consideration before characterizing fine-scaled disturbance regimes in these forests. Without
doing so, the above factors of tree-growth, that have the potential to cause growth patterns
that are similar to fine-scale disturbance events, may go undetected. Additionally, estimated
rates of fine-scale disturbance may be inflated due to intermediate- and coarse-scaled events
not accounted for.
Standardization with a horizontal line passing through the mean ring-width value is
meant to preserve long-term growth trends that characterize stand dynamics as a whole. At
the beginning of each chronology, the above-average radial increment of spruce is indicative
of open conditions and provides evidence for stand-replacement. Below-average growth
followed by a dramatic and sustained increase in growth would provide evidence for partial
canopy removal (Veblen et al. 1991a). With few exceptions, all plots within the dry and
mesic variants indicate open conditions during the earliest periods of their chronologies
42

(Figure 2.4). The lack of an initial period of increased growth in spruce in plots 2 and 5
within the wet variant suggests that these plots are older than the earliest periods represented
in the chronologies. Small sample depths, however, limit the strength of interpretation of
these early periods. Veblen et al. (1991a) found that the pattern of release caused by spruce
beetle outbreaks was usually evident with as few as five samples, but it is desirable to have at
least 15-20 samples to assure adequate replication. If this guideline for sample size is
transferable to the detection of growth patterns after stand-replacement, it becomes difficult
to state with any veracity whether the periods of rapid early growth indicate standreplacement or a host-specific disturbance agent that is represented by gap-origin samples
alone. For example, plots 4 and 5 within the dry variant and plots 2 and 5 within the wet
variant show below average growth in the earliest periods of their chronologies, followed by
release and a prolonged period of sustained above average growth. These periods, however,
are represented by fewer than five samples (Figure 2.4).
Without detailed historical records, stand-replacement and partial canopy removal can
be difficult to differentiate. A number of non-dendrochronological methods and historical
records were, therefore, used to make the distinction between these two disturbance types. In
addition, shifts in age at breast height and diameter distributions for each species are both
useful tools used to detect past beetle outbreaks and to distinguish their effects from coarsescale wildfire (Veblen et al. 1991a). For example, within the wet and mesic variants, and
plots 1 and 2 of the dry variant, bimodal or multimodal distributions of breast height ages
indicate pulses in which spruce and subalpine fir attained breast height (Figure 2.2).
Insects have played an important role in opening the canopy of mature, spruce
subalpine fir stands (Schmid and Frye 1977, Holsten 1990). In central British Columbia,
observations made by McLean (1849) and the dendroecological evidence provided by
43

Lindgren and Lewis (1997) and Zhang et al. (1999) suggest that a spruce beetle outbreak
occurred in the 1830s. No written historical records support spruce beetle outbreak in the
other periods. Newbery (2001), however, studying the influence of Inonotus tomentosus (the
causal agent of tomentosus root disease) on stand structure and dynamics within sites similar
to the plots within the mesic variant of this study, found a slight increase in the frequency of
trees that originated in gaps and releases from suppression during the 1870s through to the
1890s. A western balsam bark beetle outbreak substantially reduced the proportion of
subalpine fir within all wet variant plots during the 1920s and 1930s, and beetle galleries are
still abundant within these plots (R. D. Thompson, UNBC, unpublished data). A similar
disturbance, caused by western balsam bark beetle, was identified during the 1930s in the
Caribou Plateau in southeastern British Columbia (Koot and Hodge 1992, Parish et al. 1999).
The bark beetle outbreak identified within our study resulted in the recruitment of live spruce
and subalpine fir into smaller diameter-classes (Figure 2.1), a large number of dead subalpine
fir in the smaller diameter-classes (Figure 2.3) and a pulse of spruce and subalpine fir
recruitment after the 1920s and 1930s (Figure 2.2). This beetle outbreak provided
opportunities for spruce and subalpine fir to achieve breast height. In recent periods, a
similar pattern was evident within plots 1, 2 and 5 of the dry variant and within plots 1 ,3 ,4
and 5 of the mesic variant. Within these plots, a large number of dead subalpine fir in
smaller diameter-classes (Figure 2.3) and abundant diagnostic signs (i.e., beetle galleries)
suggests that western balsam bark beetle is currently reducing the proportion of subalpine fir
within the canopy (R. D. Thompson, UNBC, unpublished data). Based on this evidence
alone, however, it is too early to suggest that spruce is recruiting into the canopy.
The second standardization procedure, which identified short-term variations in
growth that are primarily used to study climate-related growth patterns, suggests that spruce
44

and subalpine fir exhibit moderate dendroclimatic utility. Compared with other coniferous
species in the interior of British Columbia, mean sensitivity and standard deviation values for
spruce and subalpine fir tend to be lower (Zhang et al. 1999, Watson and Luckman 2001,
Watson and Luckman 2002, Daniels and Watson 2003, Wilson and Luckman 2003). In
addition, the number of samples required for spruce and subalpine fir to attain the
recommended EPS is slightly greater, reflected by the lower correlation between trees (Zhang
et al. 1999, Watson and Luckman 2001, Watson and Luckman 2002, Daniels and Watson
2003, Wilson and Luckman 2003). PCA analyses suggest that variation among the
chronologies is related to disturbance histories and not necessarily to differences in each
species’ response to climate. This also suggests that a generalization of the regional climate
signal is possible; however, separate species/regionally-specific chronologies would be
necessary because of partial canopy removal caused by the host-specific disturbance agents
identified above (i.e., spruce beetle and western balsam bark beetle).
A major objective of this study, however, was not to reconstruct climate, but to
develop chronologies that identify periods where below and above average growth rates may
appear similar to patterns caused by disturbance. The duration of climate-related growth
patterns are key factors used to distinguish disturbance events from climatic variation.
Within the central interior of British Columbia the mean duration of below and above
average growth were 4 and 5 years and ranged from 3-16 and 3-13 years, respectively. In
rare cases, below average growth followed by above average growth (or simply the return to
average growth) could result in regionally synchronous patterns in the majority of canopy
trees that appear as growth releases, but are not related to disturbance. Within the dry variant
chronologies, one such event occurred during the 1960s (Figure 2.7) and is of concern
regarding future estimation of fine-scale disturbance frequencies. A comparison of the dry
45

variant chronologies in Figures 2.4 and 2.7 suggests that the pattern is linked to climate
because no sustained release patterns were apparent in the horizontal line standardized
chronologies for any of the dry variant plots during the same decade.
In conclusion, an accurate identification of growth patterns related to intermediateand coarse-scale disturbance and climate, and their overall effects on stand development,
enable greater confidence in the identification of growth patterns associated with fine-scale
disturbance. Further, dendroecological techniques contribute to characterizing natural
disturbance. These techniques, however, need to be matched by non-dendroecological
methods such as the examination of diameter distributions for living and dead samples, shifts
in species composition through time, spatial analyses of disturbance, and the quantification of
unique disturbance agents that brought about the diverse histories within each stand.

46

Literature Cited
Andison, D. 1996. Managing for landscape patterns in the sub-boreal spruce forests of British
Columbia. PhD. Dissertation. Faculty of Forestry, University of British Columbia,
Vancouver, BC.
Antos, J. and Parish, R. 2002. Structure and dynamics of a nearly steady state subalpine
forest in south-central British Columbia, Canada. Oecol. 130:126-135.
Briffa, K. 1995. Interpreting high-respolution proxy climate data - the example of
dendroclimatology. Pp. 77-93 in Analysis of Climate Variability: Applications of
Statistical Techniques, eds. H von Storch and A. Navarra,. Spiinger-Verlag, Berlin.
Briffa, K. and Jones, P. 1990. Basic chronology statistics and assessment. In Methods of
dendrochronology: applications in the environmental sciences. Eds. E. Cook and L.
Kairiukstis. 137-152. Kluwer Academic publishers, Dordrecht, Netherlands.
Cook, E. 1985. A time series analysis approach to tree-ring standardization. Unpublished
PhD. Dissertation University of Arizona, Tucson, AZ. USA.
Cook, E. 1987. Decomposition of tree-ring series for environmental studies. Tree-Ring Bull.
47:37-59.
Cook, E. and Holmes, R. 1984. Users manual for program ARSTND. Laboratory of tree-ring
research. University of Arizona, Tucson, AZ. USA.
Cook E. and Kairiukstis, L. 1990. Methods of dendrochronology: applications in the
environmental sciences. Kluwer Academic publishers, Dordrecht, Netherlands.
Cook, J. E. and Peters, K. 1981. The smoothing spline: A new approach to standardizing
forest interior tree-ring width series for dendroclimatic studies. Tree-ring Bull. 41: 4553.
Daniels, L. and Watson, E. 2003. Climate-fire-vegetation interactions in the Cariboo forests:
A dendrochronological analysis. Forest Innovation and Investment, Forest Research
Program.
DeLong, S.C. 1998. Natural disturbance rate and patch size distribution of forests in northern
B.C.: implications for forest management. Northwest Soi. 72:35-48.
DeLong, S.C. and Kessler, W.B. 2000. Ecological characteristics of mature forest
remnants left by wildfire. For. Ecol. Man. 131:93-106.
DeLong, C. and Tanner, D. 1996. Managing the pattern of forest harvest: lessons form
wildfire. Biodiversity and Conservation 5:1191-1205.
Fritts, C. 1976. Tree Rings and Climate. Academic Press, London UK.
47

Hawkes, B., Vasbinder, W. and DeLong, C. 1997. Retrospective Fire Study Final Report:
Fire regimes in the SBSvk and ESSFwk2/wc3 biogeoclimatic units of northern British
Columbia. McGregor Model Forest Association, Prince George, BC.
Herman, F., Smith, C. and Firth, J. 1972. Ring count of decayed wood made easier through
freezing. J. For. 70:743.
Holmes, R. 1983. Computer-assisted quality control in tree-ring dating and measuring. TreeRing Bull. 43:69-78.
Holsten, E.H. 1990. Spruce beetle activity in Alaska: 1920-1989. U.S. Dept. Agric., For.
Serv., Tech Rep. RlO-90-18. Anchorage, AK, USA.
Johnson, E.A. 1992. Fire and vegetation dynamics: studies from the North American boreal
forest. Cambridge studies in ecology. Cambridge Univ. Press, Cambridge, UK.
Koot, P. and Hodge, J. 1992. History pf population fluctuations and infestations of important
forest insects in the Kamloops Forest Region. Forestry Canada FIDS Report 92-11.
Canadian Forestry Service, Victoria.
Lewis, K. and Lindgren, S. 1999. Influence of decay fungi on species composition and size
class structure in mature Picea glauca x engelmannii and Abies lasiocarpa in subboreal forests of central British Columbia. For. Ecol. Man. 123:135-143.
Lewis, K. and Lindgren, S. 2000. A conceptual model of biotic disturbance ecology in
the central interior of BC: How forest management can turn Dr. Jekyll into Mr. Hyde.
For. Chron. 76:433-443.
Lindgren, S. and Lewis K. 1997. The natural role of spruce beetle and root pathogens in a
sub-boreal spruce forest in central British Columbia: a retrospective study. Pages 122130 in I.e . Grégoire, A.M. Liebhold, F.M. Stephen, K.R. Day and S.M. Salom, eds.
Proceedings: integrating cultural tactics into management of bark beetle and
reforestation pests. USDA Forest Service General Report NE-236.
McLean, J. 1849. Notes of a twenty-five years’ service in the Hudson’s Bay Territory, p.
286, Vol. 1. Bentley, London.
Meidinger, D. and Pojar, J. 1991. Ecosystems of British Columbia. Crown Publications Inc.
Special Report Series No. 6 . Victoria, BC.
Newbery, J. E. 2001. Small scale disturbance and stand dynamics in Inonotus tomentosus
infected and uninfected old-growth and partial cut, wet, sub-boreal forests in British
Columbia. M.Sc. Natural Resource Management, University of Northern British
Columbia, Prince George, BC.
Oliver, C. and Larson, B. 1996. Forest Stand Dynamics. John Wiley & Sons, Inc. New York,
48

NY, USA.
Parish, R., Antos, J. and Hebda, R. 1999. Tree-ring patterns in an old-growth, subalpine
forest in southern interior British Columbia. Pp. 231-248 in Tree-ring analysis. Eds.
R. Wimmer and R. E. Vetter. CAB international. Wallingford, UK.
Regent Instruments. 2003. WinDendro: Density V 2003c. Regent Instruments Inc.
Chicoutimi, Qc, Cananda.
Schmid, J. M. and Frye, R. H. 1977. Spruce beetle in the Rockies. U.S. Dept, of Agric., For.
Serv., Rocky Mtn. For. Range Exp. Stn., Gen. Tech. Rep. RM-49, Fort Collins, CO.,
USA.
Sanborn, P., Geertsema, M., lull, A. and Hawkes, B. 2001. Soil evidence for Holocene fires
in an inland temperate rainforest: east-central B.C. Abst. Canadian Quaternary Assoc,
conf., Aug. 20-24, Whitehorse, Yukon.
SPSS Inc. 1999. SYSTAT9 Statistics I and II. SPSS Inc, Chicago, IE, USA.
Stokes, M. and Smiley, T. 1968. An introduction to tree-ring dating. University of Chicago
Press, Chicago, IE. USA.
Storaunet, K. and Rolstad, J. 2002. Time since death and fall of Norway spruce logs in oldgrowth and selectively cut boreal forests. Can. J. For. Res. 32:1801-1812.
Tabaehnick, B. and Fidell, E. 2001.Using Multivariate Statistics - 4‘^ ed. Allyn and Bacon,
Needham Heights, MA, USA.
Veblen, T. Hadley, K, Marion, S. and Rebertus, A. 1991a. The response of subalpine
forests to spruce beetle outbreak in Colorado. Ecol. 72:213-231.
Veblen, T. Hadley, K, Reid, M. and Rebertus, A. 1991b. Methods of detecting past spruce
beetle outbreaks in Rocky Mountain subalpine forests. Can. J. For. Res. 21:242-254.
Velmex, Inc. 1992. The Velmex “TA” System for research and non-contact measurement
analysis. Bloomfield, NY, USA.
VoorTech Consulting. 2004. MeasureJ2X. Holdemess, NH, USA.
Watson, E. and Euckman, B. 2001. Dendroclimatic reconstruction of precipitation for sites in
the southern Canadian Rockies. Holoc. 11:203-213.
Watson, E. and Euckman, B. 2002. The dendroclimatic signal in Douglas-fir and ponderosa
pine tree-ring chronologies from the southern Canadian Cordillera. Can. J. Eor. Res.
32:1858-1874.
Wilson, R. and Euckman, B. 2003. Dendroclimatic reconstruction of maximum summer
49

temperatures from upper treeline sites in Interior British Columbia, Canada. The
Holoc. 13:851-861.
White, P. S. 1979. Pattern, process, and natural disturbance in vegetation. Bot. Rev. 45:229299.
White, P, and Pickett, S. 1985. Natural Disturbance and Patch Dynamics: An Introduction,
pp. 3-13 in The ecology of natural disturbance and patch dynamics. Eds. S. T. A.
Pickett and P. S. White. Academic Press, Inc. San Diego, CA, USA.
Zar, J. H. 1996. Biostatistical Analysis. Edited by S. Fisher. Prentice Hall, Inc. Upper Saddle
River, NJ, USA.
Zhang, Q., Alfaro, R. and Hebda, R. 1999. Dendroecological studies of tree growth, climate
and spruce beetle outbreaks in central British Columbia, Canada. For. Ecol. Man.
121:215-225.

50

Table 2.1. Summary statistics for the 30 species- and plot-specific chronologies.
Chronologies were standardized using a horizontal line.

Variant

Species

Plot

N

Dry

Spruce

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

39
27
88
86
58
67
86
26
31
59
38
41
48
44
28
52
77
85
67
87
67
68
76
62
69
59
51
40
47
58

Subalpine fir

Mesic

Spruce

Subalpine fir

Wet

Spruce

Subalpine fir

Chronology
starting date
1724
1710
1862
1805
1869
1777
1771
1799
1811
1795
1748
1769
1706
1766
1761
1769
1763
1791
1741
1784
1623
1633
1614
1658
1649
1732
1690
1741
1712
1649

Mean
correlation
0.49
0.40
0.79
0.80
0.81
0.48
0.55
0.72
0.77
0 .6 8

0.53
0.49
0.58
0.53
0.55
0.47
0.67
0.51
0.62
0.55
0 .6 6

0.62
0.60
0.61
0.54
0.33
0.39
0.68
0.50
0.42

51

Table 2.2. Mean summary statistics for the species- and site-specific chronologies. Chronologies were standardized using a cubic
spline. Mean length is the mean length of all cores in a chronology. Mean sens., Std. dev., and AC(1) are mean sensitivity, standard
deviation and first order autocorrelation, respectively. Mean ryt is the mean correlation between trees. Trees EPS 0.85 is the number of
trees required to attain an expressed population signal (EPS) value of 0.85.

Variant

Species

No.
Cores

Start

Dry

Spruce

54

Mesic

Subalpine
fir
Spruce

Wet

Subalpine
fir
Spruce
Subalpine
fir

LA
K)

Mean
Length
(yrs)
185

Mean
sens.

Std. dev.

AC(1)

Mean ryt

1818

Range in
Length
(.yrs)
134 - 230

Trees
EPS 0.85

0.14

0.18

0.53

0.44

8

29

1843

115-209

160

0.13

0.19

0.59

0.42

8

2 1

1770

206-251

233

0.15

0 .2 1

0.59

0.47

7

50

1821

136 - 229

182

0.18

0.26

0.59

0.44

8

53

1775

157-388

228

0.15

0 .2 0

0.49

0.45

8

2 0

1844

95 - 225

159

0.15

0.23

0.55

0.34

13

Table 2.3. Summary statistics for the principal component analyses (PCA). Components
were subjected to a varimax rotation. Bold values identify the highest loading for each
chronology.

Variant
Dry

Species
spruce

subalpine fir

Mesic

spruce

subalpine fir

Wet

spruce

subalpine fir

Plot
1
2
3
4
5
1
2
3
4
5
% of variance
1
2
3
4
5
1
2
3
4
5
% of variance
1
2
3
4
5
1
2
3
4
5
% of variance

PCI
0.22
0.34
0.85
0.85
0.72
0.26
0.26
0.86
0.86
0.81
65.4
0.83
0.92
0.83
0.91
0.91
0.65
0.69
-0.17
0.80
0.75
67.0
0.87
0.87
0.92
0.89
0.89
0 .2 0

0.63
0.70
0.37
0.69
76.6

PC2
0.91
0.87
0.31
0.37
0.48
0.87
0.88
0.05
0.23
0.40
16.5
-0.04
0.14
0 .1 2
0 .1 1

0.08
0.61
0.58
0.88
0.48
0.54
13.5
0.37
0.41
0.27
0.38
0.35
0.85
0.65
0.58
0.68
0.59
7.5

53

Table 2.4. Summary statistics of the regionally- and species-specific chronologies.
Chronologies were standardized using a cubic spline. Mean sens., Std. dev., and AC(1) are
mean sensitivity, standard deviation and first order autocorrelation, respectively. Mean rbt is
the mean correlation between trees.

Variant

Species

N

Start

Dry (early)

spruce
subalpine fir
spruce
subalpine fir
spruce
subalpine fir
spruce
subalpine fir

72
65
50
47
44
93
95
80

1876
1873
1788
1804
1772
1774
1790
1828

Dry (late)
Mesic
Wet

Mean
sens.
0.12
0.12
0.14
0.14
0.13
0.19
0.15
0.12

Std.
dev.
0.15
0.15
0.19
0.19
0.17
0.25
0.20
0.15

AC (1)
0.50
0.58
0.54
0.59
0.55
0.59
0.50
0.47

Mean
Tbt
0.50
0.37
0.41
0.49
0.41
0.44
0.53
0.36

54

Dry

Wet

Mesic

140

80
P lo t 1
n = 86

120
100

P lo t 1
n = 108

P lo t 1
n = 91

60
40

20

10 20

30

40

50

60

70

80

90 100

140
P lo t 2
n = 77

120

10 20

30

40

lu
50

60

70

80

10
160

120

P lot 2
140
n = 100
120

100

100

90 100

140

20

30

40

50

60

70

80

90 100

P lo t 2
n = 103

100

10

90 100
160

20

30

40

50

60

70

80

90 100

140
P lo t 3
n = 135

140
I " 120

10

20

30

40

50

60

70

80

90 100

160
P lo t 3
n = 112

120
100

S 100

P lo t 3
ti = 89

140
120
too

90 100
160

10 20

30

40

50

60

70

80

90 100

P lo t 4
n = 143

140

10
160

120

P lo t 4
n = 93

100

120

P lo t 4

140
120

100

100

90 too

10

20

30

40

50

60

70

80

90 100

140
P lo t 5
n = 94

Pot 5

120
100

30

40

50

60

70

80

90 100

20

30

40

50

60

70

80

90 100
P lo t 5
n = 110

120
100

80

80

60

60

40

20

10
140

40

20

1

0

1

20
1

10 20

30

0
40

50

60

70

80

90 100

10

20

30

40

50

60

70

80

90 100

D iam eter a t b reast h eig h t (cm )

Figure 2.1. Diameter distributions of living spruce (black) and subalpine fir (grey) for the 15
study plots. Plots are grouped as dry (left), mesic (middle) and wet (right) variants.
Midpoints of each diameter-class are indicated on the x-axis.
55

Dry

Wet

Mesic
70

80

100
P lo t 1
n = 86

Plot 1
n = 91

60

P lo t 1
n = 108

60
50
40

40
30
20

20

10

0

0
10

50

90

10

130 170 210 250 290

50

90

130 170 210 250 290
60

120

Plot 2

P lo t 2
n = 77

100
80

40

60

30

40

20

20

10

0
10

50

1 1 , rt i J ,
90 130 170 210 250 290

300

0
50

90

10

130 170 210 250 290

100

50

90

130 170 210 250 290

12 0

P lo t s
n = 135

250

P lo t 2
n = 103

50

P lo ts
n = 112

P lo t s
n = 89

too

200
150
c

100

250 290

10

50

90

130 170 210 250 290

10

200
Pb t4
n = 143

150

P lo t 4

P lot 4
tl = 93

100

10

50

90

160
140

120

50

130 170 210 250 290

90

130 170 210 250 290

10

50

90

130 170 210 250 290

120

P lo t 5
n = 94

P lot 5
tl = 110

P lot 5
n = 111

100

100

UMl
250 290

10

50

90

130 170 210 250 290

10

50

90

130 170 210 250 290

B reast h eig h t age (years)

Figure 2.2. Distributions of breast height age of living spruce (black) and subalpine fir (grey)
for the 15 study plots. Plots are grouped as dry (left), mesic (middle) and wet (right)
variants. Midpoints of each age-class are indicated on the x-axis.

56

Dry

Wet

Mesic

120

120

100

Plot 1
n = 84

P lo t 1
n = 107

too

P lot 1
n = 113

too

40

80

90 100

10

20

30

40

50

60

70

80

90 100

80

100
P lot 2
n = 143

80

P lot 2
n = 65

60

60
40
40

20
0

10

20

30

40

50

,1

I.

60

70

80

j

0

ll

ll

1

1

1

1

20

30

40

50

60

70

90 100

10

80

90 100

P lo t 3
n = 89

140
120

5 100

Plot 3
40

80

•t

60

20

(§ 40

10
0

10

20

30

40

50

60

70

90 100

10 20

30

.111

n

.

40

50

60

70

80

90 100

Plo ts
n = 129

160
140
100
80
60
40
20
0

20
0

1
1

120

30

6

P lot 2
n = 142

180

50

160
J

20

10
200
180
160
140
120
100
80
60
40
20
0

10

20

30

40

50

60

70

80

90 100

100

I

10

I 111 . 1 1 .

1
1

IH

20

30

40

50

60

70

80

90 100

160
Plot 4
n = 78

P lot 4
n = 76

80

P lo t 4
n = 120

140
120

60

100

40

60

80

40

20

10

20

30

40

50

0
60

70

80

90 100

180
160
140

■

■

II(II 11

20

30

40

20

I

0

10

20

30

40

50

60

70

80

90 100

60
P lo t 5
n = 109

10

50

■
60

70

80

90 100

120
P lo t 5

P lo t s
n = 122

100

50
40

120
100

30
20

40

10
0
100

10

20

30

40

50

60

70

80

90 100

10

D iam eter at breast heig h t (cm )

Figure 2.3. Diameter distributions of dead spruce (black), subalpine fir (grey) and
unidentifiable species (dark grey) for the 15 study plots. Plots are grouped as dry (left),
mesic (middle) and wet (right) variants. Midpoints of each diameter-class are indicated on
the x-axis.

57

Dry

Mesic

Wet

Plo t 1

Plo t 1

Plot 1

Plo t 2

Plo t 2

Plo t 2

P lots

P lo ts

Plot S

50
0

h w ,

II
Plo t 4

I

Plol 4

^ 1"'

G
100
50

Plo t 4

0
100
50

-

0

0

3

4

P lots

2-

I
0
100

0
100
50

50
0
1 700

P lo ts

Plo ts

3
2

1

0

1750

1850
Y ear

1900

2000

1700

1750

1850
Y ear

1900

1950

2000

1600

1800
Y ear

Figure 2.4. Species-specific, standard ring-width chronologies for the 15 study plots. Black and grey lines represent spruce and
subalpine fir, respectively. Chronologies were standardized using a horizontal line. The sample depth curves represent the number of
cores used to calculate the yearly ring-width indices.

Ul
00

M esic

«
100

W et

100
1600

1700

1800

1900

2000

Year

Figure 2.5. Percentage of cubic-spline chronologies showing marker rings in each year.
Black and grey bars represent positive (wide) and negative (narrow) marker rings,
respectively.

59

Mesic

Dry
2.0

Wet

2.0

P lo t 1

P lo t 1

0.5

P lo t 1

0.5

0.0

0.0

2.0

2.0

P lo t 2

2.5

P lo t 2

0.5

0.5

0.0

0.0

2.0

2.5

2.0

0.5

0.0
2.0

P lo ts

P lo ts

P lo ts

2.0

1

I

È
P<

0.5

0.5

0.0

0.0

2.5

2.0

P lo t 4

0.0
2.0

P lo t 4

2.0

P lo t

0.5
0.5

0.5

0.0

0.0

0.0

2.0

2.0

2.0

P lo ts

P lo ts

0.5

P lo ts

0.5

0.0
1 750

0.5

0.0
1800

1850

1900

1950

2000

1 750

0.0
1800

1850

1900

1950

2000

1600

1700

1 800

1900

Year

Figure 2.6. Species-specific, standard ring-width chronologies for the 15 study plots. Black and grey lines represent spruce and
subalpine fir, respectively. Chronologies were standardized using a cubic-spline.

g

2000

2

Dry (early)

1

0
100

1750

1800

1850

1900

2000

1950

2

Dry (late)

1

100

1750

1800

1850

1900

2000

1950

2

Mesic

1

0
100

1750

1800

1850

1900

2000

1950

2

Wet

1

0
100

1750

1800

1850

1900

1950

2000

Year

Figure 2.7. Regional- and species-specific chronologies. Black and grey lines represent
spruce and subalpine fir, respectively. Ring width series were standardized using a cubic
spline of 50% frequency response of 50 years. The sample depth curves represent the
number of cores used to calculate the yearly ring-width indices.

61

CHAPTER 3 - USING TREE RING GROWTH RATE CRITERIA TO ASSIGN
DATES OF CANOPY DISTURBANCE IN MATURE SPRUCE-SUBALPINE FIR
FORESTS OF CENTRAL BRITISH COLUMBIA: DENDROECOLOGICAL
CONSIDERATIONS II

Abstract
Year of gap-origin and/or release from suppression, corresponding to developmental
canopy openings caused by the death of single and multiple trees, can be used to study
disturbance regimes by estimating disturbance frequency and intensity. This chapter
summarizes the development of two radial-growth-rate criteria used to assign dates of canopy
disturbance in mature spruce-subalpine fir forests of central British Columbia. Three
ecological variants were selected: dry sub-boreal, mesic sub-boreal and wet subalpine.
Within each variant, five 0.25-ha plots were located in mature spruce-subalpine fir stands (n
= 15). Within each plot, two sampling methods were used. First, each plot was subdivided
into 16 evenly spaced grid points. At each point cores where extracted from the closest gapmaker/gap-filler pair. Year of death in each gap-maker was estimated by crossdating. If a
pattern of release in the gap-filler was visually apparent ± 1 0 years of the gap-maker year of
death then the gap-filler was classified as a release tree, and percent change in radial growth
was calculated (n = 117). Second, from the 16 grid points, four were randomly selected and
3.99m radius subplots were established for a total of 20 plots within each variant. In each
subplot, saplings (height >1.3 m and dbh <15 cm) were classed as either shaded or receiving
direct sunlight in a canopy gap. Disc or core samples were collected from each sapling 30
cm above the point of germination. Early growth rates were calculated as the average width
of five consecutive annual rings, beginning at the first annual ring two cm from the pith (n =
62

422). From these data growth rate criteria were developed. The criteria are as follows: (1)
for spruce and subalpine fir, gap-origin criteria for the dry, mesic and wet variants were set at
1.05, 0.9 and 0.8 mm/year, respectively, and (2) for spruce and subalpine fir across all
ecological variants the release criterion was a 60 % increase in radial growth rates, when
comparing adjacent 15-year periods.

63

Introduction^
Several dendroecological methods have been developed to reconstruct disturbance
histories within forests of complex age structure. These methods can be classified as direct
or indirect. Direct methods involve the cross-dating of dead samples to determine the year of
death and require that samples have intact sapwood. Indirect methods involve: (1) predictive
models that estimate the year of death; or (2 ) ring-width growth rate criteria that identifies the
date a living tree had released from suppression, or the date a tree had established in a canopy
opening caused by the death of a neighbouring tree.
The direct method of cross-dating involves the statistical comparison of overlapping
segments of dead ring-width series against live series (Holmes 1983, Grissino-Mayer et al.
1994). Daniels et al. (1997) found that cross-dating could be used to date tree rings and
assign the year of death for Thuja plicata Donn ex D. Don logs and snags. Storaunet and
Rolstad (2002) found that Picea abies (L.) Karst logs, with intact sapwood on at least one
portion of the length or circumference, could be used to estimate the number of years since
death by cross-dating. As time passes, however, the ability to successfully cross-date
samples to estimate the year of death diminishes. A sample can still be precisely dated, but
the accuracy of that date for estimating year of death decreases as the log decays and the
outer rings are lost.
The first group of indirect methods - using predictive models to estimate time since
death - is based on the reality that snags and logs inevitably decay over time. Storaunet
(2004), using Picea abies snags, developed a time-since-death model that predicted year of

' A version o f this document will be submitted for publication under multiple authors. All data used in this
study were collected by R. D. Thompson and a number of field and lab assistants. In this thesis I use 'we' and
‘our’ to reflect the involvement o f others in various aspects o f this work.

64

death based on levels of snag decomposition. Newbery et al. (2004), using Picea
engelmannii x glauca Voss x Parry dind Abies lasiocarpa (Hook.) Nutt, snags and logs from
70-year-old permanent sample plots, developed time-since-death models based on species,
position (i.e., standing and down) and levels of decomposition. Due to the 70-year limitation
of the data, as well as the increasing error variance in the decomposition variables, Newbery
et al. (2004) suggested that the models be restricted to 70 years before present, making it
difficult to reconstruct long disturbance histories.
The second group of indirect methods - involving the development of growth rate
criteria - uses absolute growth and relative changes in growth in live canopy trees to date
canopy disturbance. Lorimer et al. (1988) and Lorimer and Frelich (1989) developed two
indirect methods to date canopy disturbance. First, they used rapid early growth-rates to
indicate origin of a tree in a gap. Gap-origin criteria are used to estimate the probability of
gap-origin for current canopy trees, and are developed by integrating the early growth rates
of currently suppressed and open grown saplings (Lorimer et al. 1988, Lorimer and Frelich
1989). Second, the authors used abrupt and sustained increases in growth to indicate release
from suppression as a response to the death of a dominant neighbouring tree. Ring-width
series taken from recent gap-maker trees can grade between patterns of slow decline resulting
from root disease (Lewis 1997) to a more sudden cessation resulting from windthrow, bole
breakage or bark beetle attack. In the former case, gap-filling trees may show complex or
ambiguous growth patterns (Lorimer and Frelich 1989). In the latter case, abrupt and
sustained increases in radial growth within the ring-width series of gap-filling trees are
expected and indicate that a tree has released from suppression due to the mortality of a
dominant neighbouring tree (Lorimer and Frelich 1989, Newbery 2001). To identify
disturbance-caused release from suppression, and to integrate these events into release
65

criteria, mortality dates of recent gap-maker trees can be estimated by cross-dating and
compared with ring-width series from gap-filling trees (Newbery 2001).
Daniels et al. (1997) point out a number of differences that can occur between direct
and indirect methods. First, trees do not necessarily die and fall immediately. Whether a tree
remains standing or falls down after it dies, and how long it has been in each position, will
influence its exposure to physical weathering and susceptibility to decay organisms
(Newbery et al. 2004). Therefore, the decay rates of snags and downed trees will differ.
Second, the outermost rings naturally erode following death, or ring formation can often
cease prior to tree death, making it impossible to date mortality accurately. Finally, there can
be lengthy delays in tree establishment or release after any disturbance event (Cherubini et al.
1996, Mast and Veblen 1994). Considering the above phenomena, it is impossible for one
method alone to capture all disturbance events. Ideally, a collection of direct and indirect
methods should be used to describe disturbance processes over lengthy temporal scales.
The objectives of the following study were to: (1) develop and calibrate a suite of
dendroecological tools to quantify the effects of fine-scale disturbance events (i.e., single and
multiple tree mortalities) across three distinct sub-boreal and subalpine ecosystems of central
British Columbia; and (2) to test for differences between spruce and subalpine-fir, two
common coniferous species found across central British Columbia.

Materials and Methods
Field sampling
We selected three biogeoclimatic variants to represent the climatic gradient that spans
the dry-warm interior plateau to the wet-cold foothills of the Canadian Rockies (Chapter 1,
Figure 1.1). The Stuart Dry Warm Sub-Boreal Spruce (SBSdw3), Willow Wet Cool Sub66

Boreal Spruce (SBSwkl) and Misinchinka Wet Cool Engelmann Spruce-Subalpine Fir
(ESSFwk2) biogeoclimatic variants are located in two of the most common biogeoclimatic
zones located in central British Columbia. The SBSdwS, SBSwkl and ESSFwk2
biogeoclimatic variants will be referred to as the dry, mesic and wet variants, respectively.
Within each variant, we selected five forest stands. Stands were selected using
several criteria. First, forest stands were beyond the stem exclusion stage as defined by
Oliver and Larson (1996). This criterion was used to select stands in which trees were
establishing in the understorey and, in the absence of stand-replacing disturbances, were
advancing toward the true old-growth stage of development (Oliver and Larson 1996).
Second, forest soils had intermediate soil moisture (mesic) and nutrient (mesotrophic)
regimes (Meidinger and Pojar 1991). This criterion was intended to minimize variation in
tree species growth characteristics and disturbance processes as a result of highly divergent
edaphic influences of soil moisture and nutrients. Third, forest stands had a mixture of
spruce and subalpine fir as the dominant component of the canopy, with basal area >80 %,
for all trees over 15 cm diameter at breast height (dbh, measured 1.3 m above the ground). A
common tree composition and density minimized variation in stand susceptibility to
disturbance. Finally, forest stands had no evidence of industrial forest activities. This final
criterion ensured that stand dynamics resulted from natural disturbances, which may include
past traditional uses and practices of First Nations. Consequently, stands are not
representative of the average of the biogeoclimatic variant, but rather, they are typical of
stands satisfying the criteria within each variant.
Within each stand, one 0.25-ha (50-m x 50-m) plot was randomly placed, for a total
of 15 plots. All living trees equal to or greater than 15 cm diameter at breast height (dbh, 1.3
m) were numbered, species tallied and dbh measured.
67

Dendroecological materials and procedures
From each 0.25-ha plot, increment cores were taken at breast height from each live
tree. Cores were taken at breast height because they tend to have less stem decay enabling
maximum ring-width series. Cores were prepared following the standard dendrochronology
techniques of Stokes and Smiley (1968). Individual ring-width series were measured to the
nearest 0.01 mm using WinDENDRO (Regent Instruments Inc. 2000). For all core samples
that had periods of extremely small ring-widths (<0.5 mm) the series were measured to the
nearest 0.001 mm using the Velmex “TA” System (Velmex, Inc. 1992) in conjunction with
MeasureJ2X (2004). The computer program COFECHA (Holmes 1983) was used to assist in
the detection of measurement and cross-dating errors. Once the ring-width series were dated,
the growth rate criteria outlined in the following sections were applied to each ring-width
series. Decadal frequency distributions of gap-origin and release from suppression were
plotted for spruce and subalpine fir separately for each plot and used to visually assess for
possible sources of error in the application of the criteria.

Early growth rates and gap-origin criteria
Each plot was subdivided into 16 evenly spaced grid points using a lO-m x lO-m grid,
excluding all boundary points. From the 16 grid points, four were randomly selected. At
each selected grid point a 3.99-m ( 1/200'^ ha) radius subplot was established for a total of 20
plots within eaeh variant. Within each subplot, saplings (height >1.3 m and dbh <15 cm)
were classed as either shaded (i.e., suppressed by canopy trees) or receiving direct sunlight in
a gap (i.e., gap-origin). Gaps were defined as openings in the canopy that have resulted from
the death of one or more trees, where evidence of the dead tree still exists, and in which
68

regeneration is less than half the average height of the living canopy trees at the canopy
margin (Worrall and Harrington 1988). Discs were taken for smaller saplings 30 cm above
the point of germination. For large saplings cores were taken 30 cm above the point of
germination. The early growth rate was calculated for each sapling as the average width of
five consecutive annual rings beginning at the first annual ring two cm from the pith
(Lorimer et a l 1988). For core samples that did not include the pith, the distance to the pith
was estimated using the geometry of an arc defined by the inner incomplete ring. If the
estimated distance was < two cm, the sample was kept, and the early growth rate was
calculated as the first five rings. Early growth rates (mm/year) of current saplings in the four
cm dbh class were measured as the average width of the last five rings (Lorimer et al. 1988).
We assumed that all sapling growth rates were comparable because the measured rings all
came from a narrow size-class (i.e., ~2-6 cm). Saplings with damaged or bowed stems were
excluded from analyses.
Difference in the early growth rates between sapling class, species and variant were
analyzed by nested ANOVA, with species nested within variant (Zar 1996, SPSS Inc. 1999).
Explanatory variables in the nested ANOVA were treated as random effects. The number of
criteria used to define “gap-origin” was based on differences in the explanatory variables.
The frequency distributions of gap-origin and suppressed sapling early growth rates were
plotted using 0.5 mm/year growth rate classes.
The probability that a canopy tree was growing in a gap, when in the four cm dbh
class, given an observed early growth rate, is dependent on the degree of overlap in the
growth rate frequency distributions of the suppressed vs. gap-origin saplings. Likewise, the
probability that a tree was initially suppressed is determined partly by the percentage of gap
saplings exceeding a stated growth threshold, as well as by the relative abundance of the two
69

sapling types and the differential probability of both surviving to maturity. Equation [1]
below states that the probability that a sample tree was growing in a gap, when in the four cm
dbh class, given an observed growth rate, X, equal to or greater than a specified threshold
growth rate, x, is numerically equal to the ratio of the proportion of gap saplings exceeding
growth rate x to the proportion of both suppressed and gap saplings exceeding the threshold
(Lorimer and Frelich 1989):

where:
Gxi

=

proportion of gap saplings in size class i exceeding growth rate x;

Sxi

=

proportion of suppressed (understory) saplings of size class i exceeding
growth rate x;

Q’g

=

overall proportion of trees in the stand that were growing in gaps when four
cm dbh; and

Q’s

=

overall proportion of trees in the stand that were suppressed when four cm
dbh.

Sxi and Gxi are calculated directly from the contemporary sapling growth rate distributions.
Q’s and Q’g are historical estimates obtained from radial patterns and growth rates of mature
trees (Lorimer et al. 1988). The formula is then used in an iterative process until Pxi= 95%.

Percent change in radial growth and release criteria
Within each plot, from the 16 grid points defined in the previous section, one core or
disc was taken from the closest gap-maker. These ring-width series were cross-dated against
species- and plot-specific master chronologies, developed in Chapter 2, using the computer

70

program COFECHA (Holmes 1983) to estimate year of death. The year of death was then
compared to possible years of release identified from the ring-width series of the two closest
potential gap-filler trees taken from intermediate and suppressed crown classes (Newbery
2001). If the year of release occurred within 10 years of the year of death then the tree was
classified as a gap-filler tree and the percent change in radial growth was calculated from the
ring-width series as;
[2]

%CRG -

*100

where:
%CRG =

percent change in radial growth;

GRB„

the average growth increment based on n years of growth before the date of

=

mortality in the gap forming tree; and
GRAm

=

the average growth increment based on m years of growth after the date of
mortality in the gap forming tree.

The values of n and m were selected using empirical data resulting from the
species/regionally-specific cubic-spline chronologies developed in Chapter 2 and the findings
of Lorimer et al. (1988) and Lorimer and Frelich (1989). The value n, the number of years
prior to the date of mortality, was used to avoid classifying recovery from extended periods
of climate-related below average growth as gap-related release from suppression. The 95*^
percentile, based on the frequency distribution of below average growth (i.e., ring-width
indices less than zero for > three years), was used as a first estimate of n. The value of m, the
number of years after the date of mortality, was used to avoid classifying extended periods of
climate-related above average growth as gap-related release from suppression. The 95*-

71

percentile, based on the frequency distribution of above average growth (i.e., ring-width
indices greater than zero for > three years), was used as a first estimate of m. In addition,
existing canopy trees can fill in gaps through lateral expansion given relatively high initial
stem densities (Lorimer and Frelich 1989). Therefore, a highly conservative value of m was
required to eliminate the classification of responses to crown thinning (i.e., trees in the
canopy responding to the death of other trees in the canopy) as release, despite a possibility
of missing true release events.
To assess the need for multiple release criteria, differences in the percent change in
radial growth by ecological variant and by species were analyzed using nested ANOVA, with
species nested within variant (Zar 1996, SPSS Inc. 1999). Explanatory variables in the
nested ANOVA were treated as random effects. The frequency distributions of percent
change in radial growth were plotted using 50 % classes. A relatively high percent release
value was chosen (25*-percentile) because adjacent mortality usually causes weaker growth
increases than overhead releases (Lorimer and Frelich 1989, Newbery 2001).

Results
Early growth rates and gap-origin criteria
Early growth rates of saplings were positively skewed (Figure 3.1, n = 422). To meet
the ANOVA assumptions of normality and homogeneity of variance, the data were
transformed using the Box-Cox transformation (Sokal and Rohlf 1995). Although the
distributions of early growth rates overlapped between saplings that were initially suppressed
and saplings that originated in gaps (Figure 3.1), the nested ANOVA detected a significant
difference between the two sapling classes (Table 3.1). Although there was no significant
difference between ecological variant at conventional alpha levels, the results suggest that a
72

proportion of the variation in early growths rates can be explained by ecological variant
(Table 3.1). No significant difference between the early growth rates of spruce and subalpine
fir was found (Table 3.1). Based on these results, criteria were developed for each variant to
identify spruce and subalpine fir canopy samples that showed rapid early growth rates
(Figure 3.1). Using the formula to determine rapid early growth rates with 95 % confidence
levels, the threshold values for the dry, mesic and wet variants were set at 1.05, 0.9 and 0.8
mm/year, respectively.
Based on the gap-origin criteria, within the dry variant, plots 1 and 2 show a period of
establishment after 1870, and plots 3, 4 and 5 show a period of establishment after 1880
(Figure 3.2). Within the mesic variant, the gap-origin frequencies show a wide range in the
number of trees establishing in each decade and minor synchronicity in establishment dates
across all plots (Figure 3.2). Within the wettest sites, the gap-origin frequencies also show a
wide range in the number of trees establishing in canopy openings during each decade
(Figure 3.2). Furthermore, two common periods of establishment are evident after 1850 and
1930 (Figure 3.2).

Percent change in radial growth and release criteria
Based on the species/regionally-specific chronologies developed in Chapter 2, the
mean duration of below average growth was four years, ranging from 3-16 years, and the
95*-percentile was 12 years. A conservative “15-year growth increment” before the date of
mortality in the gap forming tree (GRAm) was selected. The mean duration of above average
growth was five years, ranging from 3-13 years, and the 95^‘'-percentile was 10 years. A
conservative “15-year growth increment” after the date of mortality in the gap forming tree
(GRBn) was selected.

73

The distribution of percent change in radial growth was positively skewed (Figure
3.3). To meet the ANOVA assumptions of normality and homogeneity of variance, the data
were transformed using the Box-Cox transformation (Sokal and Rohlf 1995). The nested
ANOVA detected no significant differences in the percent change in radial growth values
between ecological variant or between spruce and species nested within ecological variant
(Table 3.2). Based on these results, one criterion for release was developed based on the
pooled data (Figure 3.3, n = 121). Fifty percent of the observations for percent change in
radial growth fell between 60 and 230 % with the median being 116 %. Based on the 25*percentile, a 60 % increase in radial growth was chosen to identify release from suppression
caused by canopy disturbance.
Based on the release criterion, within the driest sites, plots 1 and 2 show an increased
number of release events during the 1940s, and all plots show an increase in the number of
release events during the 1960s (Figure 3.4). Within the mesic sites, three common periods
of release occurred during the 1870s, 1930s and 1960s (Figure 3.4). Within the wettest sites,
one common period of release occurred during the 1930s into the 1940s (Figure 3.4)

Discussion
The results of this study clearly demonstrate that radial-growth-rate criteria can be
used to date fine-scale (i.e., tree-level) disturbance. Furthermore, in comparison with
previous work, early growth rates and release from suppression can show a wide range in
variation by species or by region of study, and therefore suggests that radial-growth-rate
criteria must be calibrated to account for variation within the specific ecosystems studied.
In regard to gap-origin criterion, the results of this study suggest that grouping species
into one gap-origin criteria for each ecological variant is conservative and parsimonious.
74

Using formula [1] to determine the gap-origin criteria, with 95% confidence levels, the
threshold values for the dry, mesic and wet variants were set at 1.05, 0.9 and 0.8 mm/year,
respectively. Antos and Parish (2002a) and Parish et al. (1999), using early growth rates to
identify gap-origin, did not separate species and defined initial rapid growth as ring-widths
>1.0 mm/year within the first 25 years and sustained for 10 years thereafter for Picea
engelmannii Parry ex Engelm. and Abies lasiocarpa within the subalpine forests of southern
British Columbia. In contrast, Newbery (2001) adapted the methods of Lorimer et al. (1988)
and Lorimer and Frelich (1989) for Picea glauca x engelmannii dmd Abies lasiocarpa in the
sub-boreal forests of central British Columbia and found significant differences in the early
growth rates of the two species, and set the gap-origin criteria at >1.7 mm/year and >1.5
mm/year, respectively. Although his results suggested a statistically significant difference
between the early growth rates of spruce and subalpine fir, the difference was not large,
suggesting that the early growth rates are generally the same when establishing in canopy
gaps.
Wright et al. (1998) showed that in continental subalpine environments the radial
growth of spruce and subalpine fir converged at high light levels. Wright et al. (1998) also
showed that radial growth rates of subalpine fir and spruce were similar at light levels < 2 0 %
full sun. Furthermore, Coates (2000) found very little difference between spruce and
subalpine fir in mean seedling diameter five years after planting within undisturbed forest,
and extending up to medium gap sizes (300 to 1000 m^). Within our present study, the
canopy openings are caused by the death of single and multiple trees. It is reasonable to
assume that the spruce and subalpine fir saplings we sampled come from canopy openings
well within the range investigated by Coates (2000). Grouping species into one gap-origin

75

criteria for each ecological variant is practical for the ecological variants selected within our
study.
In regard to release criteria, when comparing the results of our study with previous
silvicultural and historical reconstructions of disturbance, pooling data to develop one release
criterion for all species across all ecological variants is appropriate. Based on these pooled
data, a release event was defined as a >60 % increase in mean radial growth, when adjacent
groups of 15 years are compared.
Most silvicultural studies show that typical release from suppression falls within a
range of 40 to 100 %. The release criterion selected for this study is well within this range.
The response of individual hardwoods to heavy commercial type thinning (including Acer
rubrum L., Acer saccharum Marsh., Prunus serotina Ehrh., Liriodendron tulipifera L., and
Quercus rubra L.) have indicated growth releases between 70 and 90 % (Stone 1977, Ellis
1979, Erdmann et al. 1985, Lamson et al. 1990, Smith and Miller 1991, in Singer and
Lorimer 1997). Ellis (1979) reported that A. saccharum in an 85-year-old stand given full
crown release had a mean diameter growth rate increase of 94 %. Jones and Thomas (2004)
found that A. saccharum, within the 10 to 25 cm diameter class, averaged a 50 % increase in
diameter growth four years after harvest. Singer and Lorimer (1997) found that basal area
growth rates in mature northern hardwoods can be increased on average by 70 to 100 %
using heavy commercial type thinning.
Similar results have been found in coniferous forests dominated by Picea spp. and/or
Abies spp. Balvinder et al. (2001), for example, found an average diameter increase of 37 %
in response to glyphosate treatments in a 16-year-old Picea glauca (Moench) Voss
plantation. Biring et al. (1999) reported an 85 % increase in growth in Picea glauca and
Abies balsamea (L.) Miller over 12 years following glyphosate treatment. Sutton (1995)
76

found a 50 to 100 % increase in diameter growth in a 30-year-old Picea glauca stand
following fertilization, weed control and irrigation.
Historically, the selected release criteria used to quantify fine-scale disturbance
frequencies and intensities fall inline with the range of release responses reported within the
silvicultural studies. Cherubini et al. (1996) defined release from suppression as a sudden
increase of at least 40 % over the previous four years. Lorimer and Frelich (1989) and
Abrams and Orwig (1996) defined a major release as an average growth increase >100 %,
and a moderate release as an average growth increase between 50 to 100 %. Nowacki and
Abrams (1997) used a 25 % increase in radial growth as a cut-off for the detection of minor
releases, such as those created by single tree removal. Parish et al. (1999) defined a release
as a 100 % increase in ring-width, sustained for at least 10 years. Veblen (1986) used a 150
% increase criterion for Picea glauca x engelmannii and Abies lasiocarpa, but required
releases to be sustained only five years. For the same species, Newbery (2001) used a 65 %
increase in growth based on a 15-year slow growth-rate criteria before release followed by a
15-year sustained release criteria. These release criteria are intended to be strict enough to
screen out most cases of climatic, thinning, and side-lighting responses, but not so strict that
valid releases would often be excluded and disturbance frequencies and intensities
underestimated (Lorimer and Frelich 1989). Based on the above evidence regarding release
caused by silvicultural research treatments and release criteria used in previous studies, a 60
% growth increase (which corresponds to the 25^'^-percentile in the percent change in radial
growth change data distribution) based on a “15-year sustained release criteria” and a “15year slow growth before release criteria” is reasonable and conservative.
Although the gap-origin and release criteria developed within this study are highly
conservative. Chapter 2 demonstrates that inferred climate events can result in growth
77

patterns that are similar to those caused by fine-scale disturbance. Together with the results
of this study, this is clearly the case in a number of the plots (Figure 3.4). Within the central
interior of British Columbia the mean durations of growth decrease and increase were four
and five years. The maximum growth decrease and increase, however, were 16 and 13 years,
respectively. Extended periods of growth decrease followed by growth increase have created
pseudo-release patterns. This is visible as a majority of trees showing release from
suppression in one time period across all plots. Based on an assessment of both release
frequencies (Figure 3.4) and the species/regionally-specific cubic-spline chronologies
developed in Chapter 2 (Figures 2.2 and 2.4), we can conclude that these events are climate
driven, and not due to some intermediate- or coarse-scale disturbance event. For example,
during the 1960s within the driest sites, one such climate event occurred and is of concern
regarding future estimation of fine-scale disturbance frequencies. A comparison of the dry
variant chronologies of Chapter 2 (Figures 2.2 and 2.4) suggests that the pattern is likely
linked to climate because no abrupt and sustained release patterns occurred during the same
decade in any of the SBSdw3 chronologies developed using horizontal-line standardization.
Lorimer and Frelich (1989) found that the identification of canopy disturbance, based
on rapid early growth and release from suppression are accurate because the levels of
uncertainty or ambiguity associated with these patterns are low. The results of this study
clearly demonstrate that when these methods are applied to data sets taken from mature,
spruce-subalpine fir dominated sub-boreal and subalpine forest types, the levels of
uncertainty and ambiguity associated with growth responses to developmental canopy gaps
can be minimized, but not removed completely. Multiple dendroecological methods, in
addition to the radial growth rate criteria developed within this study, are absolutely
necessary to validate the identification of fine-scale disturbance.
78

Literature cited
Abrams, M. and Orwig, D. 1996. A 300-year history of disturbance and canopy
recruitment for co-occuring white pine and hemlock on the Allegheny Plateau, USA.
J. Ecol. 84:353-363.
Antos, J. and Parish, R. 2002. Structure and dynamics of a nearly steady-state subalpine
forest in south-central British Columbia, Canada. Oeeologia. 130:126-135.
Balvinder, S. Yearsley, H. and Hays-Byl, W. 2001. Ten year responses of white spruce
and associate vegetation after Glyphosate treatment at Tsiloch River. Extension Note
55. British Columbia Ministry of Forests. Victoria, BC.
Biring, B., Hays-Byl, W. and Hoyles, H. 1999. Twelve-year conifer and vegetation
responses to discing and glyphosate treatments on a BWBSmw backlog site. British
Columbia Ministry of Forests, Research Branch. Working Paper 43. Victoria, BC.
Cherubini, P., Piusii, P. and Schweingruber, F. 1996. Spatial growth dynamics and
disturbances in a subalpine spruce forest in the Alps: a dendroecological
reconstruction. Can. J. For. Res. 26:991-1001.
Coates, D. 2000. Conifer seedling response to northern temperate forest gaps. For. Ecol.
Man. 127:249-269.
Daniels, L., Dobry, J., Klinka, K. and Feller, M. 1997. Determining year of death of logs
and snags of Thuja plicata in southwestern coastal British Columbia. Can. J. For. Res.
27:1132-1141.
Ellis, R. 1979. Response of crop trees to sugar maple, white ash, and black cherry to release
and fertilization. Can. J. For. Res. 9: 179-188.
Erdmann, G., Peterson, R. and Oberg, R. 1985. Crown releasing red maple poles to shorten
high-quality sawlog rotations. Can. J. For. Res. 15:694-700.
Grissino-Mayer, H., Holmes, R. and Fritts, H. 1994. International tree-ring data bank
library, version 1.1. Laboratory of tree-ring research. University of Arizona, Tucson,
AZ., USA.
Holmes, R. 1983. Computer-assisted quality control in tree-ring dating and measuring.
Tree-Ring Bull. 43:69-78.
Jones, T. and Thomas, S. 2004. The time course of diameter increment responses to
selection harvests in Acer saccharum. Can. J. For. Res. 34:1525-1533.
Lamson, N., Smith, H., Perkey, A. and Brock, S. 1990. Crown release increases growth
of crop trees. USDA For. Serv. Res. Pap. NE-635. USA.

79

Lewis, K. 1997. Growth reductions in spruce infected by Inonotus tomentosus in central
British Columbia. Can. J. For. Res. 27:1669-1974.
Lorimer, C. and Frelich, L. 1989. A methodology for estimating canopy disturbance
frequency and intensity in dense temperate forests. Can. J. For. Res. 19:651-663.
Lorimer, C., Frelich, L. and Nordheim, E. 1988. Estimating gap origin probabilities for
canopy trees. Ecol. 69:778-785.
Mast, J. and Veblen, T. 1994. A dendroecological method of studying tree mortality
patterns. Phys. Geogr. 15:529-544.
Meidinger, D. and Pojar, J. 1991. Ecosystems of British Columbia. Crown Publications Inc.
Special Report Series No. 6 . Victoria, BC.
Newbery, J. E. 2001. Small scale disturbance and stand dynamics in Inonotus tomentosus
infected and uninfected old-growth and partial cut, wet, sub-boreal forests in British
Columbia. M.Sc. Natural Resource Management, University of Northern British
Columbia, Prince George, BC.
Newbery, J. E., Lewis, K. and Walters, M. 2004. Estimating time since death of Picea
glauca X engelmannii and Abies lasiocarpa in wet cool sub-boreal spruce forest in
east-central British Columbia. Can. J. For. Res. 34:931-938.
Nowacki, G. and Abrams, M. 1997. Radial-growth averaging criteria for reconstructing
disturbance histories from presettlement-origin oaks. Ecol. Monog. 67:225-249.
Oliver, C. and Larson, B. 1996. Forest Stand Dynamics. John Wiley & Sons, Inc. New York,
NY, USA.
Parish, R., Antos, J. and Hebda, R. 1999. Tree-ring patterns in an old-growth, subalpine
forest in southern interior British Columbia. Pp. 231-248 in Tree-ring analysis. Eds.
R. Wimmer and R. E. Vetter. CAB international. Wallingford, UK.
Pojar, J. Trowbridge, R. and Coates, D. 1984. Ecosystem classification and interpretation
of the sub-boreal spruce zone. Price Rupert Forest Region, British Columbia. Land
Manage. Rep. No. 17., British Columbia Ministry of Forests, Victoria, BC
Regent Instruments. 2003. WinDendro: Density V 2003c. Regent Instruments Inc.
Chicoutimi, Qc, Cananda.
Singer and Lorimer, 1997. Crown release as a potential old-growth restoration approach in
northern hardwoods. Can. J. For. Res. 27:1222-1232.
Smith, H. and Miller, G. 1991. Releasing 75 to 80 year-old Appalachian hardwood
sawtimber trees: 5-year dbh response, pp. 402-413 in Proceedings of the 8 * Central

80

Hardwood Forest Conference, University Park, Penn. Edited by L.H. McCormick and
K.W. Gottschalk. USDA For. Serv. Gen. Tech. Rep. NE-148. Radnor, PA, USA.
Sokal, R. R. and Rohlf, F. J. 1995. Biometry: The principles and practices of statistics in
biological research. 3’^‘*Edition. W.H. Freeman and Company. New York, NY, USA.
SPSS Inc. 1999. SYSTAT9 Statistics I and II. SPSS Inc. Chicago, IL, USA.
Stokes, M. and Smiley, T. 1968. An introduction to tree-ring dating. University of Chicago
Press, Chicago, IL, USA.
Stone, D. 1977. Fertilizing and thinning northern hardwoods in the Lake States. USDA
For. Serv. Res. Pap. NC-141. USA.
Storaunet, K. 2004. Models to predict time since death of Picea abies snags. Scand. J.
For. Res. 19:250-260.
Storaunet, K. and Rolstad, J. 2002. Time since death and fall of Norway spruce logs in
old-growth and selectively cut boreal forests. Can. J. For. Res. 32:1801-1812.
Sutton, R. 1995. White spruce establishment: initial fertilization, weed control, and
irrigation evaluated after three decades. New For. 9:123-133
Tabachnick, B. and Fidell, L. 2001.Using Multivariate Statistics - 4* ed. Allyn and Bacon,
MA, USA.
Veblen, T. 1986. Treefalls and the coexistence of conifers in subalpine forests of the
central Rockies. Ecol. 67:644-649.
Velmex, Inc. 1992. The Velmex “TA” System for research and non-contact measurement
analysis. Bloomfield, NY, USA.
VoorTech Consulting. 2004. MeasureJ2X. Holdemess, NH, USA.
Worrall, J.J. and Harrington, T.C. 1988. Etiology of canopy gaps in spruce-fir forests at
Crawford Notch, New Hampshire. Can. J. For. Res. 18:1463-1496.
Wright, E., Coates, D., Canham, C. and Bartemucci, P. 1998. Species variability in
growth response to light across climatic regions in northwestern British Columbia.
Can. J. For. Res. 28:871-886.
Zar, J. H. 1996. Biostatistical Analysis. Edited by S. Fisher. Prentice Hall, Inc. Upper Saddle
River, NJ, USA.

81

Table 3.1. Nested ANOVA results of early growth rates between sapling class, ecological
variant and between species nested within ecological variant.

SS

df

MS

F

P

Sapling class

73.981

1

73.981

220.839

<0 . 0 0 1

Variant

5.522

2

2.761

8.242

0.058

Species (Variant)

1.006

3

0.335

1.530

0.206

Error

90.745

415

0.219

Source

82

Table 3.2. Nested ANOVA results of percent change in radial growth values between
ecological variant and between species nested within ecological variant.

Source

SS

df

MS

F

P

Variant

33.939

2

16.970

1.298

0.393

Species (Variant)

39.221

3

13.074

0.837

0.476

Error

1796.902

115

15.625

83

d ry v a ria n t g a p o rig in

60 -

d ry v a ria n t su p p re ssed

50 40
30
20
10 H

I«
.

0

0.0

0.5

1.0

1.5

I

2.0

2.5

3.0

3.5

4.0

80
m e sic v a ria n t gap o rig in

_

&
S'
g

Ph

m e sic v a ria n t su p p ressed

60

40

20

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

100
w e t v a ria n t gap o rigin

80 -

w e t v a ria n t su p p ressed

60 40 20 0

0.0

i L1
0.5

1.0

—

1.5

2.0

2.5

I—

3.0

3.5

4.0

Radial growth (mm/year)

Figure 3.1. Frequency distributions of radial growth rates for gap-origin (black) and
suppressed (grey) saplings, pooled by species. The dashed lines indicate the early growth
rate (EGR) criteria for each ecological variant. Sample sizes for the dry, mesic and wet
variants are 171, 71 and 180, respectively.

84

Dry
P lo tl
n = 86

1700

1700

ll li I

1750

1800

1850

1900

1950

Plot 2
n = 100

Plot 2
n = 77

1650

P lot 1
n = 108

n = 91

1650

g,

Wet

Mesic

1750

1800

1850

1900

1950

[650

1700

650

1700

1750

1800

1 850

1750

1800

1850

1900

1950

1750

1800

1850

1900

1950

1750

1800

1750

1800

1850

1900

1950

P lo t 2
n = 103

1750

1800

1850

1900

1950

1650

1700

00

ce
Plot 3
n=135

Î

P lot 3

Plot 3
n = 112

t

1

1650

1700

1750

1800

1850

1900

1950

1650

P lo t 4
n = 143

1700

1750

1800

1850

1900

1950

1650

1700

P lo t 4

P lot 4
n = 93

10
5
0

1650

1700

1750

1800

1850

1900

1800

1950

1850

1900

1950

1650

1700

40 ----------------------------------------------------------------------

30

1650

P lo t 5
n = 94

1700

P lo t 5
n = 110

P lo t s
n = 111

1750

1800

1850

1900

1950

1650

1700

1750

1800

1850

1900

J
1950

1650

1700

Decade

Figure 3.2. Frequency distribution of the number of spruce (black) and subalpine fir (grey)
trees originating in a canopy gap within each decade for each plot. Plots are grouped as dry
(left), mesic (middle) and wet (right) variants.

85

30

25 -

I
I

I

I
20 -

I

I
I

10 -

5 -

100

200

300

400

500

600

700

800

Percent change in radial growth (%CRG)

Figure 3.3. Frequency distribution of percent release for gap-fillers, pooled by species and
ecological variant. Dotted line indicates the release criteria (25*-Percentile = 60%). n = 117.

86

Wet

Mesic
P lo t 1
n = 108

P lo tl
n = 91

P lot 1
n = 86

, M ,

1650

1700

1750

1800

1850

1900

1650

1950

1700

1750

1800

1850

1900

1950

1650

1700

M

1750

1800

1850

1900

1750

1800

1850

1900

1750

1850

1900

1750

1850

1900

40 -

P lot 2
n = 100

Plot 2
n = 77

P lo t 2
n = 103

&

I

1650

1700

1750

1800

1850

1900

1650

1950

1700

1750

1800

1850

1900

1950

1650

1700

I
Plo t 3
n = 135

P ot 3
n = 89

P lot 3
n = 112

»

1
1650

%

1700

1750

1800

1850

1900

1950

1650

1700

1750

1800

1850

1900

1950

1650

1700

25

Plot 4
n = 143

20

P lo t 4
n = 88

P lot 4
n = 93

15

10

5

0

1650

1700

1750

1800

1850

1900

P lo t 5
n = 94

15

1650

1950

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

20 T

1700

1750

1800

1850

1900

1950

50

40

1650

1700

1950

14

12 P lo t s
n = no

P lo t s
n = 111

30

J
1

6

2

.............................1i
1650

1700

1750

1800

1850

l

l

1900

1950

” ---------—
1650

1700

JL , h
1750

1800

1850

1900

J m

Pk.
1950

1650

1700

1750

1800

1850

1900

l ,
1950

Decade

Figure 3.4. Frequency distributions of the number of spruce (black) and subalpine fir (grey)
releasing from suppression within each decade for each plot. Plots are grouped as dry (left),
mesic (middle) and wet (right) variants.

87

CHAPTER 4 - A NEW DENDROECOLOGICAL METHOD TO SEPARATE
RESPONSES TO CANOPY TREE MORTALITY FROM CLIMATE

Abstract
As part of a larger project describing fine-scale disturbance within mature, spruce-fir
forests within central British Columbia, release from suppression in response to canopy tree
mortality was defined as >60 % change in radial-growth rates with respect to subsequent 15year periods. This release criterion was calibrated using a sub-sample of gap-maker/gapfiller pairs with known years of mortality and response. Despite the conservative release
criterion, analyses revealed that a substantial number of release events, attributed to climate,
were being counted. In response, a new dendroecological method was developed using: (1)
the species/regionally-specific cubic-spline chronologies, developed to emphasize regional
growth trends for spruce and subalpine fir; and (2 ) gap-filler ring-width series, collected for
spruce and subalpine fir to identify increases in radial growth as a response to the death of
their corresponding gap-makers. The regional chronology was subtracted from each
standardized gap-filler series producing a residual ring-width series for each gap-filler. One
was added to each value of the residual series to avoid negative or erroneously large values
when calculating percent divergence over time. Gap-filler percent divergence values were
calculated as the percent change in residuals, centred on the year of death estimated for each
gap-filler. The percent divergence criterion was set at a >15 % increase in residuals, with
respect to subsequent 15-year periods.

88

Introduction^
Description and quantification of regional disturbance regimes is often based on the
reconstruction of stand disturbance histories. Many techniques are available for
reconstructing stand histories and each one is particularly suited for analyzing specific
disturbance types. For example, in wet temperate regions with extended periods between
coarse-scaled fires, the ages of the oldest trees may not be indicative of stand-initiation
(Hawkes et al. 1997). In regions where uneven-aged stands with closed canopies and shade
tolerant species are common, ages may also bear no meaningful relationship to the fine-scale
disturbance history because disturbance is not a prerequisite for germination (Lorimer and
Frelich 1989).
One of the fundamental dendroecological approaches for identifying disturbance
events within uneven-aged stand types is the identification of release from suppression
caused by the mortality of canopy trees. It is generally accepted that radial growth rates of
previously suppressed trees will increase following removal of the canopy initiated by
mortality, and that the release pattern is due to reduced levels of competition and increased
resource availability (Lorimer and Frelich 1989, Abrams et el. 1995, Nowacki and Abrams
1997, Jones and Thomas 2004). Release is commonly defined as an event where the percent
change in radial growth in a tree-ring series exceeds a given minimum threshold, such as 25,
50 or 100 %, which in some cases must be maintained for a certain length of time such as 5,
10 or 15 years (Lorimer and Frelich 1989, Payette et al. 1990, Nowacki and Abrams 1997,
Black and Abrams 2004).

' A version of this document will be submitted for publication under multiple authors. All data used in this
study were collected by R. D. Thompson and a number of field and lab assistants. In this thesis I use ‘w e’ and
‘our’ to reflect the involvement of others in various aspects o f this work.

89

Lorimer and Frelich (1989) suggested that the identification of canopy disturbance
based on release, measured as the percent change in radial growth, should be fairly accurate
because uncertainty and ambiguity associated with the pattern is low. Ring-width series of
gap-making trees, however, show a range of ring-width patterns prior to death, including
slow decline resulting from root disease (Lewis 1997), to abrupt death of healthy trees with
wide rings, resulting from windthrow or bark beetle outbreaks (Hard et al. 1983, Holsten
1984, Payette et al. 1990, Lewis 1997, Lewis and Lindgren 2002, Bleiker et al. 2003). When
canopy trees decline before death, gap-filling trees may show complex or ambiguous growth
patterns. When healthy canopy trees die suddenly, a more abrupt and sustained release
within the ring-width series of gap-filling trees may result. Furthermore, the magnitude of
release should differ considerably between these two types of disturbance, assuming that the
magnitude in release corresponds to the magnitude of canopy removal. As a result, release
criteria historically have been based on the magnitude of release expected after a specific
disturbance type or they have been set with respect to goals of analysis, site, and species.
Veblen et al. (1991a), investigating the effects of major spruce beetle (Dendroctonus
rufipennis Kirby) outbreaks, defined release as >250 % increase in mean radial growth rate
when adjacent groups of 10 years are compared. In contrast, Nowacki and Abrams (1997)
investigating relatively fine-scale disturbance histories of presettlement-origin oaks, defined
release as >25 % increase in mean radial growth rate when adjacent groups of 10 years are
compared. The application of different release criteria to one data set will result in
substantial differences in the number of disturbance events identified (Rubino and McCarthy
2004). Varying the threshold used to identify a disturbance event will therefore have a large
impact on characterization of historical disturbance regimes.

90

Disturbance is not the only factor influencing patterns of radial growth. According to
the linear-aggregate model of radial growth, a ring width during a particular year is the result
of a number of biotic and abiotic factors such as tree age, climate, endogenous and
exogenous disturbances, anthropogenic influences on the environment, and measurement
error (Fritts and Swetnam 1989). Release criteria are applied to individual ring-width series
to partition the factors, particularly climate, from endogenous and exogenous disturbances
(Lorimer and Frelich 1989). Despite the mathematical simplicity of identifying release,
climatic variation can cause radial growth patterns that are identical to the release caused by
mortality of neighbouring canopy trees, especially following extended periods of drought.
As a result, an individual cannot be certain that a release, measured as a percent change in
radial growth, is caused by disturbance. Lorimer and Frelich (1989) suggest that
distinguishing release resulting from disturbance versus release from other factors is feasible
primarily because the effects of the other factors generally do not persist as long as release
from suppression caused by inter-tree competition. To account for variation in ring-widths
caused by climate, yet permit the identification of past disturbance, the differentiation of
climatic- versus disturbance-caused variation in radial growth is of utmost importance.
Release criteria have been used to describe the disturbance regimes for a diverse set
of forest types, including eastern boreal forests (Bergeron and Charron 1994); temperate
hardwood forests (Lorimer and Frelich 1989, Payette et al. 1990, Frelich and Graumlich
1994, Abrams et al. 1995); temperate coniferous forests of western North America (Zhang et
al. 1999, Newbery 2001); and coniferous montane or subalpine forests (Veblen 1986, Veblen
et al. 1991a, Veblen et al. 1991b, Veblen et al. 1994, Parish et al. 1999, Eisenhart and
Veblen 2000). Despite the wide range of forest types examined, most of the disturbance
histories quantified have focused on forests with closed canopies. This is not surprising, as
91

stand dynamics are known to have a profound effect on annual radial growth rates, including
release from suppression, in closed canopy forests (Cook and Peters 1981, Phipps 1982,
Rubino and McCarthy 2004). The development of dendroecological techniques, therefore,
has been concentrated on the resource dynamics that occur when a closed canopy exerts
strong influence over the growth of sub-canopy trees.
Understorey vegetation is often limited in closed-canopy forests. Further, understory
establishment and recruitment into the canopy can be largely due to above- and belowground resource dynamics resulting from recently dead or dying trees. For the most part,
release from suppression results from changes in a tree’s environment, such as increases in
light or water availability following disturbance that removes or negatively impacts
surrounding competitors (White and Pickett 1985, Rubino and McCarthy 2004).
Within central British Columbia, few disturbance histories have been reconstructed
within the sub-boreal and subalpine forest ecosystems, which have relatively open canopies.
It is unclear if the traditional release criteria of the literature are appropriate for use in these
open-canopy forest types with unique stand structure, composition and light regimes.
Although the light levels within sub-boreal and subalpine forested ecosystems are poorly
quantified, there are a number of reasons to assume that light levels in these forest types are
greater and less variable than in typical closed canopy stands. Furthermore, the light regimes
of an open-canopy forest should result in differences in how understory trees establish, grow
and eventually ascend into the canopy. Claveau et al. (2002) found that at <10 % full
sunlight, eastern Abies and Picea species carried a higher live crown ratio, thus had larger
crowns per unit of height than western species of the same genera. They hypothesized that
the differences could be related to canopy tree composition and structure, where eastern
species were under a canopy composed of deciduous species producing a closed canopy
92

forest, and western species were under a canopy composed of coniferous species producing
an open canopy. This evidence suggests that the western genera are under less competition
for light due to the open coniferous canopy.
Kneeshaw and Burton (1997) found that the canopies of most stands, within moist
cool sub-boreal forests, were considerably open in their horizontal structure, with estimated
canopy closure as low as 30.5 %. Coates (2000) found that seedlings of Tsuga heterophylla
(Raf.) Sarg., Thuja plicata Donn. ex D. Don, Abies lasiocarpa (Hook.) Nutt., Pinus contorta
var latifolia Engelm., Picea engelmannii x glauca Voss x Parry planted just off the canopy
gap edge under canopy trees all performed poorly regardless of the cardinal direction from
the gaps or gap size. This suggests that the assumed light advantage experienced off the
north end of high latitude gaps, compared to other cardinal directions, is not necessarily
beneficial for tree growth. Furthermore, their results suggest that below-ground resource
dynamics may have an important influence on growth of sub-canopy trees within opencanopy forests. As canopy conditions become increasingly broken, light becomes less
limiting.
Wright et al. (1998) found that saplings of Abies amabilis Dougl. ex Forbes
Abies lasiocarpa, Tsuga heterophylla, Tsuga mertensiana (Bong.) Carr., Thuja plicata, Pinus
contorta var latifolia, Picea glauca (Moench) Voss, Picea engelmannii Parry ex Fngelm.,
Picea engelmannii x glauca, Picea mariana (Mill.) BSP, Betula papyrifera Marsh., Populus
tremuloides Michx. and Populus balsamifera ssp. trichocarpa Torr. and Gray all had
measurable radial and height growth rates even at the lowest light levels sampled in their
study. Understory light levels in mature temperate and subalpine forests in the region rarely
drop below five % full light (F.F. Wright and K.D. Coates, unpublished data, in Wright et al.
1998), unlike the much lower light levels (1-2 % of full sun) often found in broadleaf forests
93

of eastern North America or the tropics (Canham et al. 1990, Clark et al. 1996). Therefore,
the radial growth of these coniferous species in high-latitude sub-boreal and subalpine forests
appear to have fundamentally different responses to light than deciduous and coniferous
species in broadleaf forests and lower latitude temperate forests (Pacala et al. 1994, Wright et
al. 1998), where growth rates of most species sharply increase at very low light levels and
can reach 50 % of maximum growth at light levels <10 % of full sun and near maximum
growth at 20 % full sun (Fetcher et al. 1993).
The above literature suggests that understory light levels within sub-boreal and
subalpine forests of central British Columbia can be abundant and relatively homogenous.
Further, the importance of light as a limiting factor of growth within these ecosystems
diminishes. As sub-canopy light levels increase, we expect that the importance of other
growth factors, defined within the linear aggregate model, will increase. Most importantly,
climate as a limiting factor may explain a greater proportion of variation in tree growth,
introducing a new complexity when attempting to partition the effects of disturbance from
climate. Further, release responses to the average fine-scale disturbance may diminish in
magnitude with greater canopy openness, lowering the release threshold required to detect
fine-scale disturbance and increasing the chance of interpreting climatic variation as evidence
of overhead mortality.
With these important differences between the resource dynamics of open- versus
closed-canopy forest types, there is an inherent problem associated with the application of
traditional dendroecological techniques to forests with open canopies, despite being
calibrated to account for the difference in forest composition and structure, or changes in the
disturbance regimes that preside over the landscape (i.e., fine- vs. intermediate- to coarsescale disturbance). That is, when we apply the same dendroecological techniques across
94

many regions, the level of uncertainty and ambiguity associated with the pattern of release
from suppression is not constant. As a result, the error associated with the estimation of
disturbance frequencies will change.
In previous work (Chapter 3), release from suppression caused by mortality of canopy
trees was defined as >60 % increase in mean radial growth, when adjacent groups of 15 years
are compared. This release criterion was applied to live Abies lasiocarpa and Picea
engelmannii x glauca ring-width series from canopy trees >15 cm diameter at breast height
(dbh, measured 1.3 m above the ground) sampled from sub-boreal and subalpine forests of
central British Columbia. Numerous analyses revealed that the regional effect of (inferred)
climate produced radial growth patterns identical to disturbance-caused release. Extended
below-average growth rates, followed by a return to the average growth rate, triggered
growth patterns identical to disturbance-caused release from suppression. Further, based on a
comparison of decadal release frequencies (Chapter 3) and chronologies developed using
horizontal-line standardization (Chapter 2), it was argued that the patterns were driven by
climate rather than a regionally synchronous disturbance event. The horizontal-line
chronologies did not reveal the dramatic and sustained release patterns that would be
expected with synchronicity of release from suppression caused by major canopy
disturbance.
Rozas (2001) found that standardized chronologies, developed from stands with either
extremely low levels of past disturbance (control) or relatively high levels of past disturbance
(affected), responded differently to standardization. By subtracting the affected chronologies
from the control chronologies, the residual chronology could be used to date periods of
coarse-scale disturbance. This suggests the individual standardized ring-width series, used to
develop the chronologies within the disturbed stands, maintained a dendroecological signal
95

associated with canopy-removal. We hypothesize that standardized ring-width series that
diverge from a co-located species/regionally-specific chronology - based on the same
detrending method - could also be used to identify stand dynamics at fine-spatial scales (i.e.,
individual trees). The objectives of this study were to: (1) develop a new dendroecological
method, based on divergence between chronologies, to separate fine-scale disturbances from
regionally synchronous events; and (2) discuss how the divergence method adds to the
traditional release method.

Methods
As part of a larger research project intended to characterize the fine-scale disturbance
regimes in spruce-subalpine fir dominated forests within central British Columbia three
biogeoclimatic variants were selected. The Stuart Dry Warm Sub-Boreal Spruce (SBSdw3),
Willow Wet Cool Sub-Boreal Spruce (SBSwkl) and Misinchinka Wet Cool Engelmann
Spruce-Subalpine Fir (ESSFwk2) biogeoclimatic variants are located in two of the most
common biogeoclimatic zones located in central British Columbia. The SBSdw3, SBSwkl
and ESSFwk2 variants will be referred to as the dry, mesic and wet variants, respectively.
Within each variant, five spruce-subalpine fir stands were selected and within each stand one
0.25-ha (50-m x 50-m) plot was randomly placed, for a total of 15 plots.

Calibrating percent divergence
The following method uses (1) the species/regionally specific cubic-spline
chronologies, developed in Chapter 2, to emphasize regional radial growth trends for spruce
and subalpine fir, and (2) the gap-filler ring-width series, collected in Chapter 3 for spruce
and subalpine fir from each variant, to identify release events as responses to the death of
96

their corresponding gap-makers. Examples of these data are presented in Figures 4.1a - c.
We assumed that the gap-filler ring-width series, in response to the death of their
corresponding gap-maker, would diverge to some degree from the regional trend for a given
length of time. Gap-filler ring-width series were compared against their co-located regional
chronology to identify periods where radial growth diverged from the regional growth trend.
Using the same standardization method used to develop the chronologies, each gap-filler
series was fit with a cubic-spline of 50 % frequency response of 50 years (Figure 4.1c) and
ring-width indices were calculated by division (Figure 4. Id). The regional chronology was
then subtracted from each standardized gap-filler series (Figure 4.le) producing a residual
ring-width series. One was added to each value of the residual series to avoid negative or
erroneously large values when calculating percent divergence over time. Percent divergence
was calculated for the residual series based on the year of death estimated for each gap-filler
as.

[1] Percent Divergence =

^ M B A,^ - M P g , /
MT>5i5

*100

j

where:
MDBis

=

15-year residual mean before the gap-maker year of mortality; and

MDAi5

=

15-year residual mean after and including the gap-maker year of mortality.

Fifteen-year periods before and after the gap-maker year of mortality were used to maintain
consistency with the original release threshold developed in Chapter 3. As the residual series
is the result of the regional chronology being subtracted from each standardized series,
erroneous residuals can result when subtracting the regional chronology in years where it is
represented by too few samples (n < 5). To avoid this, calculation of the residual series

97

began when the regional chronology sampling depth was > five, and the calculation of
percent divergence began 15 years thereafter (Figure 4.1b, e, and f).
Gap-filler percent divergence values were plotted against the original gap-filler
percent release values, and correlated (SPSS Inc. 1999, Tabachnick and Fidell 2001).
Differences in the levels of percent divergence by species and by ecological variant were
tested using a nested ANOVA with species nested within variant (Zar 1996, SPSS Inc. 1999).
Explanatory variables in the nested ANOVA were treated as random effects. To maintain
consistency with the original release threshold calibration methods, the number of unique
percent divergence criteria was based on any differences found in the ANOVA. Further, the
25*-percentile of the gap-filler percent divergence values was used as the divergence
threshold.

Application of the divergence method
Within each plot, all living trees >15 cm dbh were cored at breast height. Cores were
prepared, measured and cross-dated following standard dendrochronology techniques (Stokes
and Smiley 1968, Holmes 1983). Once the ring-width series were dated, yearly percent
divergence values were calculated for each canopy tree (Figure 4. If).
A comparison of the yearly percent divergence and yearly percent release identifies
four possible outcomes (Figures 4. If and 4.1g). First, the release and divergence criteria are
both met. Second, the release and divergence criteria are both not met. Third, release is met
while divergence is not. Fourth, divergence is met while release is not. Physiologically, we
assume that if a canopy disturbance had occurred in the past and an understory tree was there
to ascend towards the canopy, the understory tree would have shown a release pattern.
Furthermore, the release pattern would likely have met the release threshold defined in
98

Chapter 3. To be confident, however, that a release event was indeed caused by fine-scale
disturbance, and not some other regionally synchronous event (e.g., climate), the release
event was only counted if the divergence threshold was also met. It is unlikely that the
calculations would result in the two criteria being met during the same year; therefore, the
release and divergence criteria had to be met ± five years of one another. The amalgamated
“release and divergence” criterion will now be referred to as the divergence method.

Analyses of the divergence method
To compare the number of disturbance events identified by the divergence method
against the number of events originally identified by the release method (Chapter 3), decadal
frequency distributions of release (Chapter 3, Figure 3.4) and divergence were calculated for
each species, within each plot. Plot level frequency distributions of divergence were not
plotted, but were used in the following methods. The number of release events was
subtracted from the number of divergence events, giving the total number of release events
removed per decade. The resulting value was negative if release events were removed.
These events were determined to be caused by a regionally synchronous event. The number
of release events and the number of release events removed per decade were pooled by
variant.
To assess the relationship visually between the species/regionally-specific
chronologies and the frequency of release events identified and the frequency of release
events removed per decade, the chronologies were plotted with the decadal frequency of
release events identified and removed per decade.
To assess the relationship between the species/regionally-specific chronologies and
the proportion of release events removed at the plot level, the proportion of release events
99

removed per decade was plotted against the decadal ring-width index. The decadal ringwidth index was calculated by averaging the ring-width indices for each decade.
The proportion of release events removed per decade was calculated separately for spruce
and subalpine fir, by dividing the number of release events removed per decade by the
sample size of each species within each plot.
The divergence method was based on the assumption that a negligible proportion of
fine-scale disturbance events would be removed during any period where the
species/regionally-specific chronology showed average growth rates. Conversely, if the
regional chronology showed an extended period of below average growth, followed by a
return to average growth, it was assumed that a substantial proportion of release events would
be removed. With extended and/or more severe periods of below average growth we
expected that the proportion of release events removed would increase exponentially because
release patterns would become common to a majority of the sample trees. Further, the
pattern of release would be maintained as ring-width variation within the species/regionallyspecific chronology. To test this hypothesis, nonlinear regression was used to fit a twoparameter negative-exponential curve to the scatter plot of the proportion of release events
removed per decade versus the mean decadal ring-width index (SPSS Inc. 1999, Zar 1996).

Results
Calibrating percent divergence
Gap-filler percent release and percent divergence values were positively correlated
(Figure 4.2). Thirty of the original 117 gap-filler samples had divergence values < zero,
despite all showing positive release (Figure 4.2). Based on the uncertainty that these samples
represent the population of gap-filling trees, we excluded these values from further analyses.
100

Seventy-two of the 87 gap-fillers with positive divergence values had release values > 60 %
(Figure 4.2).
After the negative percent divergence values were removed, the distribution of
percent divergence was positively skewed (Figure 4.3). To meet the ANOVA assumptions of
normality and homogeneity of variance, the data were transformed using the Box-Cox
transformation (Sokal and Rohlf 1995). ANOVA results detected no significant differences
in percent divergence between ecological variant (Table 4.1) or between species nested
within variant (Table 4.1). Based on these results, one unique percent divergence threshold
was selected based on the pooled data for spruce and subalpine fir across all three ecological
variants (Figure 4.3). Fifty percent of the observations for percent divergence fell between
15 % and 42 % with the median being 25 %. Based on the 25*-percentile, a 15 % divergence
from the regional chronology was chosen to represent a disturbance at a fine-spatial scale
(i.e., the individual tree) (Figure 4.3).

Application and analyses of the divergence method
Despite the coarse temporal scale of the number of release events removed in eaeh
decade, qualitative analysis of Figure 4.4 suggests a moderately consistent pattern. The
largest frequency of release events removed in each decade coincides with decades of growth
suppression (Figure 4.4). One exception occurred within the dry (early) variant where
growth suppression was followed by increased growth rates and a period of positive growth,
resulting in consecutive decades of release removal (Figure 4.4). It is also apparent that some
original release events were moved into adjacent decades because both divergence from the
regional chronology and percent release >60 % could be met ± five years of eaeh other
(Figure 4.4). The association between the proportion of release events removed in each
101

decade and the mean decadal growth index was not extremely strong, however, the
regression is significant, suggesting that if the regional chronology showed an extended
period of below average growth, followed by the return to average growth, a greater
proportion of release events would be removed (Table 4.2 and Figure 4.5).

Discussion
Objective criteria are needed for identifying radial growth patterns that represent
radial growth release in a gap following the death of a canopy tree. This study has outlined a
dendroecological method to do so. The results of this study demonstrate that the divergence
method is better than traditional release methods with regard to its ability to separate
responses to fine-scale disturbance from regionally synchronous events (e.g., climate) within
sub-boreal and subalpine forests of central British Columbia.
Black and Abrams (2004) argue that even if the effects of age, size, or climate are
quantified and taken into account, the utility of radial growth criteria (i.e., release criteria) is
still not optimized unless a single, standardized method is developed. Numerous release
criteria have been employed, each specific to investigators’ research goals, selected sites,
species, and disturbance regimes (Black and Abrams 2004, Rubino and McCarthy 2004).
These authors argue that without standardized criteria, direct comparisons among these
studies are complicated. We suggest, however, that it is not the lack of a standardized
method that diminishes the potential power of dendroecology to quantify disturbance patterns
across diverse landscapes or regions, but the lack of a standardized way of assessing the
varying levels of error associated with each new growth rate criteria. This study, in
combination with previous work (Chapters 2 and 3), suggests that the application of
standardized growth rate criteria to different data sets, from distinct regions, results in
102

considerable changes to the counting errors associated with the identification and
quantification of disturbance.
Two types of error can occur when applying release criteria to assess canopy
disturbance, and an analogy can be drawn from the relationships between hypotheses and
decisions made in statistical hypothesis testing (Table 4.3). A type I error is committed when
a canopy release is identified but there is no coincident death of a canopy tree. In this case, a
growth release may occur but is caused by climate variation or a disturbance other than the
formation of a canopy gap. A type II error is committed when release occurs following the
creation of a canopy gap, but the release is not detected. In this case, the magnitude of the
release may be too small to be detected. To extend the analogy further, the power associated
with a given growth rate criterion can be defined as the ability of the criterion to detect a
growth release when a canopy tree dies and creates a gap. With this in mind, by selecting
“liberal” release criteria, disturbances would be identified more often than they actually
occur, decreasing type II errors and increasing the power of the test, but also increasing the
number of type I errors. In contrast, by selecting “conservative” release criteria, disturbances
would be identified less often than they actually occur, decreasing type I errors, but
decreasing the power of the test and increasing the number of type II errors.
In Chapter 3, the release threshold was set at a percent change in radial growth >60
%. With this release threshold, there was a high likelihood that a release event would be
identified when an overhead mortality event truly did occur. Despite the power associated
with the release threshold, however, many type I errors occurred because both climate and
disturbance result in similar ring-width patterns that are difficult to differentiate using
historical reconstruction techniques. One option to decrease the frequency of type I errors

103

would be to increase the magnitude of release required to identify overhead mortality. This
solution, however, would be at the cost of increased type II errors and decreased power.
A second option to decrease the frequency of type I errors would be to increase the
“adjacent groups of 15 years compared” to adjacent groups of 20 or 25 years. To do so,
requires the recalibration of the original gap-maker/gap-filler pairs. This would have
decreased the release threshold below 60 %, because an increased number of years used to
calculate growth rates before and after any given year has a higher likelihood of bringing
before and after growth rates closer to the mean ring-width for the entire period used in
release calculation. The before and after mean growth rates for each canopy tree sampled,
however, should also have the same likelihood of being closer to the mean ring-width for the
entire period used in release calculation. The mean growth rate after the year of disturbance
is, therefore, less likely to be much greater than the mean growth rate before disturbance.
Theoretically, there is a counter-balancing effect by increasing the number of years used to
calculated mean growth rates before and after a given date. Therefore, no substantial shift in
error types from the origin levels should occur.
The above options maintain unacceptable levels of error and justify the addition of the
divergence criteria to the original release criteria. Although this is highly conservative, by
combining the two criteria, the frequency of type I errors caused by regional-scale variation
(e.g., climate) have been decreased considerably. The cost of increased type II errors and
reduced power associated with the reduced frequency of type I errors, however, remains
uncertain. Figures 4.4 and 4.5 both show a small number or proportion of release events
being removed in a number of decades where it is not expected because we assumed that a
negligible proportion of true disturbance events would be removed during any period where
the regional chronology showed an average or above average ring-width indices. Although
104

the divergence and release thresholds had to be met ± five years of each other to count as a
canopy disturbance event, and it was possible for some of the original release events to be
moved into adjacent decades, it does not account for all of the unexpected release events
removed during periods where the regional chronology showed average or above average
ring-width indices.
The most plausible explanation would simply be that the release threshold was met
while divergence was not. This is illustrated where a small or negative percent divergence
values are possible over a wide range of percent change in radial growth values (Figure 4.2).
It is possible that these removed release events are disturbance related. It is difficult,
however, to say whether the response was caused by overhead mortality or a response to the
death of an adjacent canopy tree. In addition to the release of understory trees by the death of
an overhead tree, growth increases in existing canopy trees do occur in response to the death
of adjacent canopy trees (Frelich and Martin 1988, Lorimer and Frelich 1989, Pedersen and
Howard 2004). Lorimer and Frelich (1989) suggest that the frequency of disturbance in a
given decade can be overestimated by the addition of the amount c/N, where c is the number
of sample trees showing a canopy thinning response in that decade and N is the total number
of sample trees in the plot. If these responses are the result of adjacent canopy tree
mortalities, then it is preferable to eliminate such responses. These thinning responses could
also be removed by using age, size, growth and/or distance criteria in the year a tree showed
release. For example, Newbery (2001) found that only trees >20 cm dbh caused release
when they died, and understory trees were typically < five m from the dead canopy tree, they
were at least five cm dbh less than the dead tree and they were of vigorous growth.
Coincidentally, the divergence criterion also removes intermediate- to coarse-scale
disturbance events that are regionally synchronous. Depending on the research question, this
105

may or may not be considered a type II error. With regard to the larger project objectives the characterization of fine-scale disturbance regimes in central British Columbia - this is not
an error because the goal was to estimate the frequency of fine-scale disturbance. In fact, it
can be considered the partitioning of intermediate- to coarse-scale, regionally synchronous
disturbance events, from fine-scale disturbance events. In Figure 4.4, both the mesic (ca.
1870s) and wet (ca. 1930s) variant plots have a period where the relative slow growth is not
climate driven, but a product of standardization using a cubic-spline that is not flexible
enough to conform tightly to the common, dramatic and sustained patterns of release
expected within individual ring-width series after a disturbance event that removes much of
the canopy (Veblen et al. 1991a, 1991b). The interpretation of this phenomenon is
justifiable, and illustrated for a spruce ring-width series selected from the mesic variant
(Figures 4.1c and d). In this example, the spline diverges prior to the 1870s abrupt and
sustained release. This results in ring-width indices consistently less than one just before the
release. If this trend is consistent within most canopy trees across an entire region, then the
regional trend will express negative ring-width indices followed by increasing growth rates.
Furthermore, when calculating the residual series, series will not show a divergence pattern
from the regional trend, thus partitioning regionally synchronous disturbance events from
fine-scale events.
The final criticism of the divergence method is not necessarily related to error type,
but is concerned with the inherent limitations placed on the maximum disturbance
chronology length. The new method truncates the disturbance chronology length because the
calculation of residual series begins when the sampling depth of the regional chronology is >
five, and the calculation of percent divergence can only begin 15 years thereafter.
Substantial stem decay at coring height, loss of older trees through natural mortality, and the
106

destruction of young or mature stands as a result of intermediate- and coarse-scale
disturbance can all reduce the apparent intensity of past disturbance (Lorimer and Frelich
1989). Furthermore, Lorimer and Frelich (1989) suggested that the loss of evidence from
earlier disturbance should be addressed by restricting disturbance chronology length to a
period of time where the evidence loss rate is fairly small. Average residence time of canopy
trees is a common value used to determine reasonable chronology length. In the absence of
stand-replacing fire, residence time within the sub-boreal and subalpine forests within this
study are often over 200 years (Chapter 2, Figure 2.2). The limitations placed on the
disturbance chronology lengths by the divergence calculation, however, ranged from 100 to
200 years (Figure 4.4). Therefore, the calculation of divergence places a reasonable limit on
chronology length.
In conclusion, the divergence method is not without error. By combining the new
divergence criteria with the original release criteria, however, the frequency of type I errors,
caused by regional-scale variation (e.g., climate), have been decreased considerably, without
a great cost of increased type II errors and reduced power. Furthermore, the results show that
standardizing release criteria should not necessarily be the solution to direct comparisons of
disturbance regimes across different regions. It is clear that, despite their final objective use,
all growth-rate criteria will continue to maintain a level of error in the identification of
disturbance. Whether the criteria are calibrated to account for the effects of age, size, or
climate, or whether a completely new technique is developed to identify disturbance, all
criteria generally have a minimized, yet misunderstood and accepted level of uncertainty.
Without a standardized way of assessing the varying levels of error associated with each new
growth rate criteria, direct comparisons of disturbance regimes across different regions will
continue to be a complex issue.
107

Literature Cited
Abrams, M., Orwig, D. and Demeo, T. 1995. Dendroecological analysis of successional
dynamics for presettlement origin white-pine-mixed-oak forest in the southern
Appalachians, USA. J. Ecol. 83:123-133.
Bergeron, Y. and Charron, D. 1994. Postfire stand dynamics in a southern boreal forest
(Quebec): A dendroecological approach. Ecosci. 1:173-184.
Black, B. A. and Abrams, M. D. 2004. Development and application of boundary-line
release criteria. Dendrochrono. 22:31-42.
Bleiker, K. P., Lindgren, B. S., and Maclauchlan, L. E. 2003. Characteristies of subalpine
fir susceptibility to attack by western balsam bark beetle (Coleoptera: Scolytidae).
Can. J. For. Res. 33:1538-1543.
Canham, C. D., Denslow, J. S., Platt, W. J., Runkle, J. R., Spies, T. A. and White, P. S.
1990. Light regimes beneath closed canopies and tree-fall gaps in temperate and
tropical forests. Can. J. For. Res. 20:620-631.
Clark, D. B., Clark, D. A., Rich, P. M., Weiss, S. and Oberbauer, S.F. 1996. Landscapescale evaluation of understory light and canopy structure: methods and application
In a neotropical low-land rain forest. Can. J. For. Res. 26:747-757.
Claveau, Y., Messier, C., Comeau, P.O. and Coates, D. 2002. Growth and crown
morphological responses of boreal eonifer seedlings and saplings with contrasting
shade tolerance to a gradient of light and height. Can. J. For. Res. 32:458-468.
Coates, K. D. 2000. Conifer seedling response to northern temperate forest gaps. For.
Fcol. Man. 127:249-269.
Cook, J. F. and Holmes, R. 1984. Users manual for program ARSTAN. Laboratory of
tree-ring research. University of Arizona, Tucson, AZ., USA.
Cook, J. F. and Peters, K. 1981. The smoothing spline: A new approach to standardizing
forest interior tree-ring width series for dendroelimatie studies. Tree-Ring Bull. 41:
45-53.
Cook, E. 1987. Decomposition of tree-ring series for environmental studies. Tree-Ring Bull.
47:37-59.
Eisenhart, K. S. Veblen, T. T. 2000. Dendroecological detection of spruce beetle
outbreaks in northwestern Colorado. Can. J. For. Res. 30:1788-1798.
Fletcher, N., Oberbauer, S. F. and Chazdon, R. L. 1993. Physiological ecology of plants. Pp.
128-141 in La Selva: ecology and natural history of a neotropical rain forest. Edited

108

by L. C. McDade, K. S. Bawa, H. A. Hespenheide and G. S. Hartshorn. University of
Chicago Press, Chicago, IL., USA.
Frelich, L. and Graumlich, L. 1994. Age-class distribution and spatial patterns in an oldgrowth hemlock-hardwood forests. Can. J. For. Res. 24:1939-1947.
Frelich, L. E. and Martin, G. L. 1988. Effects of crown expansion into gaps on evaluation
of disturbance intensity in Northern Hardwood Forests. For. Sci. 34:530-536.
Fritts, H. C. and Swetnam, T. W. 1989. Dendroecology: A new tool for evaluating
variations in past and present forest environments. Adv. Ecol. Res. 19:111-188.
Hard, J. S., Werner, R. A. and Holsten, E. H. 1983. Susceptibility of white spruce to
attack by spruce beetle during early years of an outbreak in Alaska. Can. J. For.
Res. 13:678-684.
Hawkes, B., Vasbinder, W. and Delong, C. 1997. Retrospective Fire Study Final Report:
Fire regimes in the SBSvk and ESSFwk2/wc3 biogeoclimatic units of northern
British Columbia. McGregor Model Forest Association. Prince George, BC.
Holmes, R. 1983. Computer assisted quality control in tree-ring dating and measuring.
Tree-Ring Bull. 43: 69-78.
Holsten, E. H. 1984. Factors of susceptibility in spruce beetle attack on white spruce in
Alaska. J. Entomol. Soc. Brit. Col. 81: 39-45.
Jones, T. A. and Thomas, S. C. 2004. The time course of diameter increment responses to
selection harvests in Acer sacchamm. Can. J. For. Res. 34: 1525:1533.
Kneeshaw, D. and Burton, P. 1997. Canopy and age structures of some old sub-boreal
Picea stands in British Columbia. J. Veg. Sci. 8:615-626
Lewis, K. J. 1997. Growth reduction in spruce infected by Inonotus tomentosus in central
British Columbia. Can. J. For. Res. 27:1669-1674.
Lewis, K. and Lindgren, S. 2002. Relationships between spruce beetle and tomentosus
root disease: two natural disturbance agents of spruce. Can. J. For. Res. 32:31-37.
Lorimer, C. and Frelich, L. 1989. A methodology for estimating canopy disturbance
frequency and intensity in dense temperate forests. Can. J. For. Res. 19:651-663.
Lorimer, C., Frelich, L. and Nordheim, E. 1988. Estimating gap origin probabilities for
canopy trees. Ecol. 69:778-785.
Meidinger, D. and Pojar, J. 1991. Ecosystems of British Columbia. Crown Publications Inc.
Special Report Series No. 6. Victoria, BC.

109

Newbery, J. E. 2001. Small scale disturbance and stand dynamics in Inonotus tomentosus
infected and uninfected old-growth and partial eut, wet, sub-boreal forests in British
Columbia. M.Sc. Natural Resource Management, University of Northern British
Columbia, Prince George, BC.
Nowacki, G. J. and Abrams, M. D. 1997. Radial-growth averaging criteria for
reconstructing disturbance histories from presettlement-ori gin oaks. Ecol. Monogr.
67:225-249.
Parish, R., Antos, J. and Hebda, R. 1999. Tree-ring patterns in an old-growth, subalpine
forest in southern interior British Columbia. Pp. 231-248 in Tree-ring analysis. Eds.
R. Wimmer and R. E. Vetter. CAB international. Wallingford, UK.
Pacala, S. W., Canham, C. D., Silander, J. A., Jr. and Kode, R. K. 1994. Sapling growth as a
function of resources in a north temperate forest. Can. J. For. Res. 24: 2172-2183.
Payette, S., Filion, L. and Delwaide, A. 1990. Disturbance regimes of a cold temperate
forest as deduced from tree-ring patterns: the Tantare Ecological Reserve, Quebec.
Can. J. For. Res. 20:1228-1241.
Pedersen, B. S. and Howard, J. L. 2004. The influence of canopy gaps on overstory tree
and forest growth rates in a mature mixed, age-species forest. For. Ecol. Man.
196:351-366.
Phillips, R. L. 1982. Comments on interpretation of climatic information from tree-rings,
eastern North America. Tree-Ring Bull. 42:11-22.
Pojar, J. Trowbridge, R. and Coates, D. 1984. Ecosystem classification and interpretation
of the sub-boreal spruce zone. Prince Rupert Forest Region, British Columbia.
Land Manage. Rep. No. 17., British Columbia Ministry of Forests, Victoria, BC.
Rozas, V. 2001. Detecting the impact of climate and disturbances on tree-rings of Fagus
sylvatica L. and Quercus robur L. in a lowland forest Cantabria, Northern Spain. Ann,
For. Sci. 58:237-251.
Rubino, D. L. and McCarthy, B. C. 2004. Comparative analysis of dendroecological
methods used to assess disturbance events. Dendrochrono. 21:97-115.
Sokal, R. R. and Rohlf, F. J. 1995. Biometry: The principles and practices of statistics in
biological research. 3"^^ Edition. W.H. Freeman and Company. New York, NY, USA.
SPSS Inc. 1999. SYSTAT 9 Statistics I and II. SPSS Inc. Chicago, IL, USA.
Tabachnick, B. and Fidell, L. 2001.Using Multivariate Statistics - 4‘*^ed. Allyn and Bacon.
Boston, MA, USA.
Veblen. T. 1986. Treefalls and the co-existence of conifers in sub-alpine forests of the
110

central Rockies. Ecol. 67:644-649.
Veblen, T., Hadley, K., Nel, E., Kitzberger, T, Reid, M. and Villalba, R. 1994.
Disturbance regime and disturbance interactions in a Rocky mountain subalpine
forest. J. Ecol. 82:125-135.
Veblen, T., Hadley, K. S., Reid, M. S. and Rebertus, A. J. 1991a. Methods of detecting
past spruce beetle outbreaks in Rocky Mountain subalpine forests. Can. J. For.
Res. 21:242-254.
Veblen, T. Hadley, K, Marion, S. and Rebertus, A. 1991b. The Response of Subalpine
Forests to Spruce Beetle Outbreak in Colorado. Ecol. 72:213-231.
White, P, and Pickett, S. 1985. Natural Disturbance and Patch Dynamics: An Introduction,
pp. 3-13 in The ecology of natural disturbance and patch dynamics. Eds. S. T. A.
Pickett and P. S. White. Academic Press, Inc. San Diego, CA,USA.
Wright, E. P., Coates, K, D., Canham, C. D. and Bartemucci, P. 1998. Species variability
in growth response to light across climatic regions in northwestern British Columbia.
Can. J. For. Res. 28:871-886.
Zhang, Q., Alfaro, R. and Hebda, R. 1999. Dendroecological studies of tree growth,
climate and spruce beetle outbreaks in central British Columbia, Canada. For.
Ecol. Man. 121:215-225.
Zar, J. H. 1996. Biostatistical Analysis. Edited by S. Fisher. Prentice Hall, Inc. Upper Saddle
River, NJ, USA.

I ll

Table 4.1. Nested ANOVA results of percent divergence values between ecological variant
and between species nested within ecological variant.

Source

SS

df

MS

F

P

Variant

2.766

2

1.383

0.182

0.842

Species (Variant)

22.806

3

7.602

1.762

0.161

Error

349.410

81

4.314

112

Table 4.2. Nonlinear regression statistics for proportion of release events removed in each
decade against the mean decadal indices for each regional chronology using two parameter
negative-exponential function where f(x) = 11.053 x

Source

SS

df

MS

F

p

Regression

1.013

2

0.507

101.4

<0.001

Residual

0.793

152

0.005

Total

1.806

156

Corrected

1.083

155

Raw R^

0.561

Adjusted R^

0.268

113

Table 4.3. Conceptual framework identifying the two error types during the identification of
release.

Overhead mortality event

Overhead mortality event

did not truly happen

did truly happen

Release not identified

No error

Error type II

Release identified

Error type I

No error (power)

114

4

1

0

D
2 -

î l

"V

....V........................

2

#0
(S

1

0

40 20-

-20 -40
300

-100
1800

1900

2000

Year

Figure 4.1. Pictorial outline of the divergence method used to identify responses to overhead
canopy mortality events. In graph C, the individual ring-width series is fit with a cubicspline of 50 % frequency response of 50 years. Dashed lines in graph F and G represent the
15 % divergence and 60 % release thresholds, respectively.

115

200

r = 0.498
p < 0 .0 0 1

n= 117
150

o
o

100

-

§

;

50 <» •

s
0

Ph

•
-50

-100

200

400

600

800

1000

Percent release (%)
Figure 4.2. Gap-filler percent divergence values versus percent release values. The dashed
vertical line indicates the 60 % release threshold.

116

n = 87

Percent divergence (%)
Figure 4.3. Frequency distribution of gap-filler percent divergence values. The dashed line
indicates the divergence threshold (25*-percentile = 15%).

117

I
}■

2

Dry
(late)

Dry
(early)
1

0
40

80 T
60 40 -

20 -

20 -|i—

-20 -

-20
1750

1800

1850

1900

1950

-40 -60 1750

1800

1850

1900

1950

Mesic

r ÎIf PI

Decade

Decade

Figure 4.4. Regional chronologies (upper plot) compared with frequency of release events
original identified and the frequency of release events removed by the divergence method
(lower plot) in each decade. Black and grey lines and bars represent spruce and fir,
respectively. Dashed vertical lines indicate 15 years after the species/regionally specific
chronology is represented by > five samples. Within the lower plots, four bars are presented
per decade. The first two bars represent release events identified and the second two bars
represent the number of releases removed by the divergence method.

118

0.4

Adjusted R = 0.27
n = 157

I

0
1
2
%

##

I

0.0

I

-

0.2
0.7

0.8

0.9

1.0

1.1

1.2

Mean decadal ring-width index

Figure 4.5. Relationship between the proportion of release events removed and the mean
regional decadal ring-width indices. The proportion of release events removed in each decade
was regressed against the mean decadal indices for each regional chronology using two
parameter negative-exponential function where f(x) = 11.053 x
^

119

CHAPTER 5 - FINE-SCALE SPATIO-TEMPORAL DISTURBANCE PATTERNS IN
MATURE, SPRUCE-SUBALPINE FIR FORESTS FOUND IN THE CENTRAL
INTERIOR OF BRITISH COLUMBIA

Abstract
Within central British Columbia fire is viewed as the primary disturbance influencing
forest dynamics. Although these forests do originate by fire, times between stand-replacing
events are enormously varied. With increasing time between stand-replacement, fine-scale
disturbance can assume an important role in forest dynamics. To characterize fine-scale
disturbance regimes across central British Columbia three ecosystems were selected: dry subboreal, mesic sub-boreal and wet subalpine. Within each ecosystem, five 0.25-ha plots were
located in mature, spruce-subalpine fir forests. Growth rate criteria were used to date
ascensions towards the canopy caused by the creation of canopy gaps resulting from the
death of canopy trees. Moran’s I analysis of ascension dates suggests that disturbance is
random in space and time. Canopy openings caused by disturbance averaged ~2.5 m in
diameter. Decadal disturbance rates did not differ by variant (p = 0.117) or by species nested
within variant (p = 0.418). Although the decadal disturbance rates did not differ by variant,
mean rates within dry, mesic and wet ecosystems were 7.4, 6.4 and 5.0 %, respectively. The
consideration of disturbance at multiple scales suggests that fine-scale disturbance exhibits
continuous stand level maintenance and renewal; however, notable compositional changes in
species dominance tend to be abrupt and due to intermediate- and coarse-scale disturbances.

120

Introduction^
The vital role of intense, coarse-scale stand-replacing fire in boreal, sub-boreal and
sub-alpine forests of western North America has long been acknowledged (Johnson and
Rowe 1977, Hawkes 1980, Romme 1980, Heinselman 1981, Pyne 1984). Recently, subboreal and subalpine forest ecosystems within the central interior of British Columbia have
been receiving greater attention with regard to similar catastrophic events (Andison 1996,
DeLong and Tanner 1996, Hawkes et al. 1997, DeLong 2000). As a result, historical fire
regimes in central British Columbia are seen as largely responsible for the patterns we see
today on the landscape. In regions where fire plays a critical role in mediating forest
dynamics, stand-replacing fires generally come at intervals less than tree longevity (Johnson
1992). If so, fine-scale disturbance processes caused by the fall of single or multiple trees
should have little influence on long-term forest dynamics because several tree generations do
not pass between intense disturbance events.
Sub-boreal and subalpine forest stands, in most cases, do originate by fire. Recent
studies, however, suggest that the times between these intense, coarse-scale stand-replacing
events are enormously varied within central British Columbia (Hawkes et al. 1997, Sanborn
et al. 2001). Within this study, we assume that much of this variation is linked to the climatic
gradient that spans from the dry-warm interior plateau to the wet-cold foothills of the
Canadian Rockies. In progressively wetter ecosystems, fire may become so infrequent that
the development of true old-growth forests may be the norm (see Oliver and Larson 1996).
Hawkes et al. (1997) report that fire-retum intervals in very wet and cool sub-boreal and

* A version o f this document will be submitted for publication under multiple authors. All data used in this
study were collected by R. D. Thompson and a number of field and lab assistants. In this thesis I use ‘w e’ and
‘our’ to reflect the involvement of others in various aspects o f this work.

121

subalpine forests near the Rocky Mountains likely range from 1200 to 6250 years. DeLong
and Tanner (1996) reported return intervals ranging from 227 to 345 years in slightly drier
suh-horeal forests in the foothills of the Rocky Mountains. With increasingly greater time
periods between stand-replacing events, fine-scale disturbance events can assume an
important role in long-term forest dynamics. These disturbances, caused by a collection of
biotic and abiotic agents, may maintain and renew the forest structure in stands with long
fire-retum intervals (Lewis and Lindgren 2000, Newbery 2001).
An extensive body of literature describes fine-scale disturbance patterns and
processes in numerous forest types, including tropical (Denslow 1987, Lawton and Putz
1988), temperate North-American (Runkle 1984, Lorimer et al. 1988, Lorimer and Frelich
1989, Payette et al. 1990, Frelich and Graumlich 1994, Abrams et al. 1995), eastern boreal
and sub-boreal (Bergeron and Charron 1994, Frelich and Reich 1995, Kneeshaw and
Bergeron 1998), western coastal (Canham 1988, Lertzman et al. 1996, Brett and Klinka
1998, Daniels 2003) and coastal subalpine forests (Lertzman and Krebs 1990, Lertzman
1992, Lundquist and Beatty 2002).
An emerging body of literature suggests that fine-scale disturbances within central
British Columbia can play an important role in stand-level forest dynamics (Kneeshaw and
Burton 1997, Lewis and Lindgren 1999, Newbery 2001, Antos and Parish 2002a, Antos and
Parish 2002b). Much of the work, though, has been dedicated to describing gap size or
shape, and its influence on tree regeneration dynamics (Kneeshaw and Bergeron 1997,
Wright et al. 1998a, Wright et al. 1998b, Eastham and lull 1999, Coates 2000).
Across the climatic gradient from the dry-warm interior plateau to the wet-cold
foothills of the Canadian Rockies, we expect that fine-scale disturbance events vary greatly
in spatial scale and frequency. One common feature across the gradient is the existence of
122

spruce-subalpine fir forest types. As part of the ecosystem classification system used in
British Columbia, Meidinger and Pojar (1991) suggest that similar climax vegetation can
occur over a range of ecological systems. Similar climax forest communities can represent
ecosystems from completely different regional climates and disturbance regimes, or similar
regional climates and disturbance regimes can lead to differences in climax communities.
Although forest ecosystems may be similar in their constituent species, considerable
differences in the fine-scale disturbance types are expected. The objectives of this study
were to describe and compare the spatial and temporal patterns of fine-scale disturbance in
mature, spruce-subalpine fir forests found across the central interior of British Columbia.
These objectives were addressed by developing canopy tree stem-maps, analyzing species
structure and composition, and the assessing the spatial-temporal patterns of disturbance.

Methods
Field sampling
We selected three biogeoclimatic variants to represent the climatic gradient that spans
the dry-warm interior plateau to the wet-cold foothills of the Canadian Rockies (Chapter 1,
Figure 1.1). The Stuart Dry Warm Sub-Boreal Spruce (SBSdw3), Willow Wet Cool SubBoreal Spruce (SBSwkl) and Misinchinka Wet Cool Engelmann Spruce-Subalpine Fir
(ESSFwk2) biogeoclimatic variants are located in two of the most common biogeoclimatic
zones located in central British Columbia. The SBSdw3, SBSwkl and ESSFwk2
biogeoclimatic variants will be referred to as the dry, mesic and wet variants, respectively.
Within each variant, we selected five forest stands. Stands were selected using
several criteria. First, forest stands were beyond the stem exclusion stage as defined by
Oliver and Larson (1996). This criterion was used to select stands in which trees were
123

establishing in the understorey and, in the absence of stand-replacing disturbances, were
advancing toward the true old-growth stage of development (Oliver and Larson 1996).
Second, forest soils had intermediate soil moisture (mesic) and nutrient (mesotrophic)
regimes (Meidinger and Pojar 1991). This criterion was intended to minimize variation in
tree species growth characteristics and disturbance processes as a result of highly divergent
edaphic influences of soil moisture and nutrients. Third, forest stands had a mixture of
spruce and subalpine fir as the dominant component of the canopy, with basal area >80 %,
for all trees over 15 cm diameter at breast height (dbh, measured 1.3 m above the ground). A
common tree composition and density minimized variation in stand susceptibility to
disturbance. Finally, forest stands had no evidence of industrial forest activities. This final
criterion ensured that stand dynamics resulted from natural disturbances, which may include
past traditional uses and practices of First Nations. Consequently, stands are not
representative of the average of the biogeoclimatic variant, but rather, they are typical of
stands satisfying the criteria within each variant.
Within each stand, one 0.25-ha plot was located by randomly placing the south­
western plot comer, which was defined as the plot origin (0, 0 m), for a total of 15 plots.
From the plot origin, plot boundaries were defined by a 50-m x 50-m closed traverse along
each cardinal bearing. At each plot comer and traverse midpoint (i.e., 25 m), stakes were
placed and numbered one to eight.
For all living and dead trees with dbh >15 cm, we mapped the x and y coordinates
relative to the plot origin. All coordinates were calculated using two azimuths taken from
any two adjacent stakes with a 25-m base and related back to the origin using trigonometry.
Stems were numbered, species tallied and dbh measured. For all downed material >15 cm

124

dbh at the time of death, dbh was recorded based on an estimate of breast height measured
from the root collar and species was determined based on bark characteristics.
Differences in the percentages of live and dead spruce between the three ecological
variants were analyzed using the Kruskal-Wallis single-factor analysis of variance by ranks
(Zar 1996, SPSS Inc. 1999). This non-parametric test was used based on previous analyses
in Chapter 2, which suggested that any transformation of percent live or dead spruce would
not work equally across all three ecological variants. The critical values of Kruskal-Wallis H
for small sample sizes were used (Zar 1996). Differences in mean stand age between the
three ecological variants were also analyzed using the Kruskal-Wallis test (Zar 1996, SPSS
Inc. 1999).

Dendroecological materials and procedures
Increment cores were taken at breast height from each canopy tree because they tend
to have less stem decay enabling maximum ring-width series. Cores were then prepared
following the standard dendrochronology techniques of Stokes and Smiley (1968).
Individual ring-width series were measured to the nearest 0.01 mm using WinDENDRO
(Regent Instruments Inc. 2000). For all core samples that had periods of extremely small
ring-widths (<0.5 mm) the ring-width series were measured to the nearest 0.001 mm using
the Velmex “TA” System (Velmex, Inc. 1996) in conjunction with MeasureJ2X (2004).
Ring-width series were cross-dated using the computer program COFECHA (Holmes 1983)
and visual inspected to assist in the detection of measurement and cross-dating errors.

125

Interpretation of radial growth patterns
Once the ring-width series were dated, ascensions towards the canopy, resulting from
overhead canopy mortality, were identified within each series using three radial growth rate
criteria (Chapters 3 and 4). First, high rates of early growth (>1.1, 0.9 and 0.8 mm/year
within the dry, mesic and wet, respectively) were used to indicate that a tree was growing in
an opening created by disturbance at the time it reached the sampling height (1.3 m).
Second, a >60 % increase in mean radial growth, when adjacent 15-year groups are
compared, was used to identify release from suppression. Previous work (Chapter 3)
indicated that this release threshold identified most fine-scale disturbance events that were
represented within the data set by release from suppression. The release criterion alone,
however, was too liberal, as revealed by the substantial number of release events attributed to
regionally synchronous events. In response, a third radial growth rate threshold was developed
(Chapter 4) to separate release responses to overhead mortality from regionally synchronous
events. The new method assumes that individual standardized ring-width series that diverge
from a regional standardized chronology, using the same detrending method, should identify
stand dynamics at spatial scales much finer than the individual plot (0.25 ha). In Chapter 2, a
cubic-smoothing-spline (50 % frequency response of 50 years) was used to develop
species/regionally specific chronologies and to standardize individual canopy trees (Cook
and Peters 1981). The regional chronology was subtracted from eaeh standardized series
producing a residual ring-width series. One was added to each value of the residual series to
avoid negative or erroneously large values when calculating percent divergence over time.
Yearly percent divergence values were calculated for each residual series as.

126

[1] % Divergence ■

MD,

*100

where MDi and MD 2 equal the preceding and subsequent 15-year residual means,
respectively. To be confident that a release event was indeed caused by fine-scale
disturbance, and not some other regionally synchronous event (e.g., climate), the release
event was only counted if the divergence criterion was met ± five years of the release
criterion being met.
Ascensions toward the canopy were then tallied by ecological variant and species.
The canopy ascensions, by species within each ecological variant, were correlated with the
number of cores sampled of that species within each ecological variant (Zar 1996, SPSS Inc.
1999). The canopy ascensions were then tabulated by ascension type (i.e., gap-origin and
release from suppression), species and ecological variant. The relationship between
ascension type, species and ecological variant was analyzed using mutual and partial
independence tests (Zar 1996).

Disturbance chronologies and frequency
Disturbance chronologies were defined in this study as the percentage of cored trees
that ascended towards the canopy during each decade (Frelich and Graumlich 1994, Frelich
2002). Chronologies were constructed for spruce and subalpine-fir separately and then
combined to summarize plot level disturbance chronologies. The divergence method
truncates the disturbance chronology length because the calculation of residual series begins
when the sampling depth of the regional chronology is > five, and the calculation of percent
divergence can only begin 15 years thereafter.

127

Differences in the decadal canopy ascension rates for each species and ecological
variant were analyzed by nested ANOVA with species nested within ecological variant (Zar
1996, SPSS Inc. 1999). Explanatory variables in the nested ANOVA were treated as random
effects. Past host-specific disturbance events can substantially decrease the relative
abundance of one species over others, making it difficult to compare canopy ascension by
species. Therefore, decadal ascension rates for each species were standardized by dividing
the number of ascensions by the present number of that species sampled within the plot. To
assess for temporal variation in the canopy ascension rates between spruce and subalpine fir,
two separate nested ANOVA were conducted, including: (1) long-term analysis including the
full disturbance chronologies and (2) a short-term analysis including only the last 50 years of
each chronology. A 50-year period was selected for the short-term analysis based on
empirical data from Chapter 2, which suggests a common stem-reinitiation period for all
study plots in the past 50 years.

Point pattern analysis
All living and dead stems >15 cm dbh were stem-mapped to assess the univariate
spatial pattern within each plot. The function L(d), a transformation of Ripley’s K(d), was
used to assess the degree to which point patterns deviated from randomness. Ripley’s K(d), a
second-order statistic, is based on the distances between all pairs of points, counting the
number of “neighbours” within a certain distance, d, of each point, with d taking on a range
of values. This analysis can be considered an examination of the cumulative frequency
distribution of the plant-to-all-plant techniques (Dale 1999).
First, the distance matrix, dÿ, between all pairs of points within each plot is tabulated.
K(d) is then defined as:
128

"

"

&

{d)

1

if

d..

<

d

Where A is the area of the plot, d is the distance interval considered and n is the number of
individuals. Second, to linearize the plot of K(d) against d, as well as to stabilize the
variances, L(d) is used to transform K(d) (Diggle 1983):

[ 3] L{ d) = ^ K {d )l 7 T - d

Ripley analysis leads to a graph of L(d) versus the distance d. The expected value of
L(d) is zero for trees conforming to a Poisson distribution. Therefore, positive values of L(d)
indicate spatial clustering and negative values indicate over-dispersion.
Separate analyses within each plot were completed for live and dead spruce and
subalpine fir at one m distance classes up to 25 m using the Ripley module within the
program ADS in ADE-4 (Pélissier and Goreaud 1999). Departures from randomness were
analyzed by comparing the computed L(d) with a corresponding confidence envelope
generated via 1000 Monte Carlo simulations of random patterns at each distance class and
then ranked. The 10^** and 990^^ values corresponded to a 0.01 significance level (Diggle
1983).

Spatial-temporal analysis of canopy ascension
The spatial patterns of canopy ascension were used to characterize gap size and the
spatial-temporal distribution of fine-scale disturbances within each 0.25-ha plot. This
129

approach does not require previous knowledge of canopy tree locations. Rather, it assumes
that canopy ascension dates coincide with death of canopy trees in the immediate vicinity
(Newbery 2001). All spatial-temporal analyses were done using the program RookCase
v.0.9.6 (Sawada 1999).
Measures of spatial autocorrelation test whether an observed value at one location is
independent of values at neighbouring localities (Sokal and Oden 1978a). In this study,
Moran’s I was used to test for spatial independence in a canopy ascension date at one
location relative to adjacent locations. Each tree can only have one year of ascension
towards the canopy in this type of correlation analysis, therefore, only the most recent
ascension was used for those trees that showed a response to disturbance in more than one
decade (Frelich and Graumlich 1994). Adjacency depends on inter-tree distance and the lag
distance set for the analysis (Sokal and Oden 1978a, Sokal and Oden 1978b, Legendre and
Fortin 1989, Frelich et al. 1993, Frelich and Reich 1995). RookCase vO.9.6 (Sawada 1999)
was used to calculate nearest-neighbour distances for each plot. A preliminary ANOVA was
used to analyze the nearest-neighbour distances between plots in each ecological variant, and
found no significant differences in mean nearest-neighbour distance between trees in each
ecological variant (F = 2.64 < Fo.o5 (i),2 ,i2 = 3.89, p = 0.112). Therefore, the grand mean of
nearest-neighbour distance was approximately 2.5 m, allowing for spatial analysis at scales
<2.5 m. Therefore, cumulative 2.5-m distance lags (i.e., 0-2.5, 0-5 ..., 0-50 m) were
employed. For each distance lag, d, a pair of trees, i and j, were considered neighbours if the
distance between them was < d. Those not within the distance lag were ignored. The
Moran’s 1 coefficients for each distance lag were then calculated as:

130

Where 1(d) is the Moran’s coefficient for distance lag d, N is the total number of mapped tree
locations, xj (i = 1, . . N) are the ascension dates at each of the N locations, Wÿ = 1 if points i
and j are neighbours within radius d (0 if otherwise), and W = ^ w» (RookCase vO.9.6
Sawada 1999).
The assumption of randomization was used as the null hypothesis (Upton and
Fingleton 1985), whereby the observed dates of canopy ascension are distributed randomly
among all locations. Under this hypothesis the Moran’s I coefficient is asymptotically
normal with a usual range of -1 to 1 and expected value of E[I(d)] = - (N - 1)

~ 0 (Cliff and

Ord 1981). The interpretation of the Moran’s I coefficient is similar to interpretations for the
standard correlation coefficient (Sokal and Oden 1978a); therefore, the statistic represents the
strength of spatial association for canopy ascension dates for each defined distance lag.
Values of 1(d) > E[l(d)] are considered positively spatially autocorrelated and indicate that
canopy ascension dates are similar at distances < d. Values of 1(d) < E[I(d)] are considered
negatively spatially autocorrelated and indicate that canopy ascension dates are dissimilar at
distances < d.
Significant deviations from the expected value in each distance lag were tested for by
calculating the variance of 1(d) under the null hypothesis. Standard normal deviates, z(d),
were then computed for each distance class as (Duncan and Stewart 1991; RookCase vO.9.6,
Sawada 1999):

131

[5]
yjvarUid)]

Correlograms for each 0.25-ha plot were constructed for both 1(d) and z(d) versus d.
Significance tests for spatial autocorrelation coefficients for each distance lag, however, do
not provide a test of significance for the data set as a whole because the k tests are not
independent (Duncan and Stewart 1991). Therefore, significance testing was done first at the
global level by determining if at least one p-value for the standard normal deviates was
significant at the global level. This was done by using a Bonferroni correction method for
multiple tests (e.g. 0.05/k of distance lags (20) = 0.0025) (Sokal and Oden 1978a). If the
global test has at least one significant standard normal deviate then the point at which the
correlogram becomes non-significant locally (a = 0.05) can be interpreted as the average
patch size diameter created by disturbance (Frelich et al. 1993, Frelich and Reich 1995).
When global correlograms included no significant results, the average patch diameter can be
interpreted as less than the distance lag.

Results
Non-spatial structure of live and dead trees
The study plots varied considerably within and between each ecological variant. For
example, across the three variants the percentage of spruce within the canopy ranged from 10
to 80 % (Table 5.1) and a slight difference was found in the percentages of spruce between
the three ecological variants (%^= 6.540, df = 2, p = 0.038). Within ecological variant, x ±
sd of percent spruce were 48.0 ± 25.1, 34.2 ± 3.7 and 67.6 ± 7.8 % for the dry, mesic and wet

132

variants, respectively. Subalpine fir, for the most part, made up the remainder of stand
composition (Table 5.1).
Across the three ecological variants the densities of trees >15 cm dbh ranged from
308 to 572 trees/ha (Table 5.1), while the within variant density x ± sd were 428 ± 120, 405
± 39, and 398 ± 43 trees/ha for the dry, mesic and wet variants, respectively. The diameter
means across all three ecological variants ranged from 26.3 to 35.0 cm (Table 5.1) and the
within variant diameter x ± sd were 28.7 ± 0.8, 33.0 ±1.7 and 27.0 ± 0.9 cm for the dry,
mesic and wet variants, respectively.
Mean tree age at breast height across all plots ranged from 100 to 150 years (Table
5.1), and there was a significant difference in mean ages between the three ecological
variants (%^= 8.060, df = 2, p = 0.018). Within each variant, the x ± sd of age at breast
height were 131 ± 18, 149 ± 5 and 116 ± 18 years for the dry, mesic and wet variants,
respectively.
Across the three ecological variants the percentage of dead spruce ranged widely
from 2.5 to 67.7 % (Table 5.2) and there was a significant difference in the percentage of
dead spruce between the three ecological variants (%^ = 9.420, df = 2, p = 0.009). Within
variant x ± sd for dead spruce percentages were 39.0 ± 17.9, 43.6 ± 26.0 and 5.4 ± 3.8% for
the dry, mesic and wet variants, respectively. A large percentage of the dead tree
compositions, up to 47.7 %, were composed of an unidentifiable component (Table 5.2). The
within variant x ± sd of percent unidentifiable trees were 19.7 ± 17.6, 16.1 ± 10.8 and 40.1 ±
5.5 % for the dry, mesic and wet variants, respectively.
The density of dead trees ranged from 204 to 572 trees/ha across all ecological
variants (Table 5.2), while within variant density x ± sd were 419 ± 101, 276 ± 50 and 500 ±

133

44 trees/ha for the dry, mesic and wet variants, respectively. Mean diameters of dead-trees
within each plot were quite variable and ranged from 22.8 to 40.5 cm (Table 5.2), while the
within variant dead-tree diameter x ± sd were 28.7 ± 6.4, 37.2 ± 3.7 and 29.1 ± 2.2 cm for
the dry, mesic and wet variants, respectively.

Spatial patterns of live and dead trees
Eleven of the 60 L(d) plots are represented by too few samples (n < 20) for spatial
analyses of this kind, however, these plots were retained to illustrate the limitations of point
pattern analyses in these forest types at fine spatial scales (Figures 5.1, 5.2, 5.3 and 5.4).
Both live spruce and subalpine fir (>15 cm dbh) show a random spatial pattern at scales <25
m (Figures 5.1 and 5.2). When aggregation of either species did occur, it was variable in
magnitude and scale (Figures 5.1 and 5.2). For example, spruce in plots 3, 4 and 5 in the dry
variant were aggregated at spatial scales < five m to >25 m (Figure 5.1). Dead spruce and
subalpine fir trees (>15 dbh) show random spatial pattern at scales <25 m (Figures 5.3 and
5.4).

Canopy ascension type by ecological variant and species
Of the 491 canopy trees sampled in the dry variant, there were 590 ascensions
towards the canopy identified, meaning that present canopy trees often ascend multiple times
before reaching the canopy. Within the mesic variant, 721 ascensions towards the canopy
were identified in 501 canopy trees and in the wet variant, 495 ascensions towards the
canopy were identified within 426 canopy trees (Table 5.3). There was a strong,
methodological relationship between the number of cores sampled and the number of

134

ascensions towards the canopy (r = 0.886, p = 0.003, n = 6). Species, ecological variant and
ascension type were not mutually independent (Table 5.4). Furthermore, analyses of partial
independencies revealed that species, variant and ascension type were not independent of one
another. Therefore, the rejection of the mutual independence indicates a strong interaction
between all three variables (Table 5.4).

Disturbance chronologies and canopy ascension by decade
The distributions of decadal canopy ascension percentages were not normally
distributed as revealed by a number of the standard deviations being greater than the mean
(Table 5.5). To meet the ANOVA assumptions of normality and homogeneity of variance,
the data were transformed using the arcsine transformation (Sokal and Rohlf 1995). For the
long-term analysis of the full disturbance chronologies, the nested ANOVA found no
significant differences in decadal canopy ascension percentages for species nested within
variant, or between variants (Table 5.6). For the short-term analysis of the full disturbance
chronologies, the nested ANOVA found no significant differences in decadal canopy
ascension percentages for species nested within variant, or between variants (Table 5.6).
Although the decadal disturbance rates did not differ by variant, mean recruitment into the
canopy, based on the full disturbance chronologies, was 7.4, 6.4 and 5.0 % per decade within
the dry, mesic and wet variants, respectively (Table 5.5). Many of the plots had decades with
no canopy recruitment (Figure 5.5) and maximum decadal canopy recruitment among the
three ecological variants were 43.9, 27.5 and 24.7 % per decade for the dry, mesic and wet
variants, respectively (Table 5.5).

135

Spatial-temporal patterns of canopy disturbance
The previous point-pattem analysis suggests that the spatial pattern of canopy trees at
scales <25 m is random, indicating the null hypothesis that observed dates of canopy
ascension are distributed randomly among all locations is correct. For correlograms with 20
distance lags, the global significance test, using Bonferroni correction, must have at least one
Moran’s I coefficient with p < 0.0025. Therefore, the global and local significant levels of z
are 1.96 and 2.94, respectively. Few plots in any of the three ecological variants had
significant Moran’s I coefficients at the global level (Figure 5.6). Plot 3 within the dry
variant was globally significant, however, the significant Moran’s I coefficients ranged from
r-values of 0.029 to 0.060, which suggests weak similarities in ascension dates. A similar
pattern occurred for plot 5 within the wet variant where the correlogram was globally
significant with r-values only ranging from 0.003 to 0.107 (Figure 5.6). Spatial-temporal
aggregation of ascension dates can occur at distances <2.5 m, however, it occurred only once
within plot 4 of the wet variant (Figure 5.6). The above results indicate that the average
diameter of a canopy gap caused by fine-scale disturbances is approximately equal to the
distances between nearest neighbours (=2.5 m).

Discussion
A central model of plant ecology, the steady-state plant community, has been
extensively used to describe the development and maintenance of spruce-subalpine fir forests
types throughout the central and northern Rocky Mountains of North America. The steadystate plant community can be defined as the stable endpoint of succession (Clements 1916,
Oliver and Larson 1996). Previous interpretations of age- and size-class distributions, and
136

recruitment and mortality rates, have lead some authors to suggest that spruce and subalpine
fir can coexist as a temporally and structurally stable community (Costing and Reed 1952,
Veblen 1986). Other authors have suggested that increasing time between stand-replacing
events will inhibit coexistence between these species, and eventually lead to the replacement
of spruce by subalpine fir, especially if the two coniferous species are disproportionately
abundant within the understory and have similar mortality rates (Day 1972, Peet 1981,
Newbery 2001). Within the central interior of British Columbia both species are relatively
long lived, often approaching and exceeding 200 years of age. Therefore, it is not surprising
that the two species coexist to some extent within late successional stages of forest
development (Chapter 2, Figure 2.2). Observations made within this study, together with
previous work in British Columbia, have identified a range of variation in the age, size,
spatial structure and recruitment rates of trees in old spruce-subalpine fir forests (Kneeshaw
and Burton 1997, Lewis and Lindgren 1999, Parish et al. 1999a, Parish et al. 1999b, Antos
and Parish 2002a, Antos and Parish 2002b). The results of this study suggest that stand
dynamics in mature spruce-subalpine fir forests of central British Columbia are not
necessarily described properly by the steady-state community concept, but are more a
product of punctuated disturbance events acting at multiple spatial and temporal scales. This
is consistent with the findings of Parish et al. (1999b) where the composition of a mature
spruce-subalpine fir forest in southeastern British Columbia was the result of multiple scales
of disturbance of differing intensities, together with recruitment during interdisturbance
periods. Fine-scale disturbance events may bring about continuous stand-level maintenance
and renewal over extended periods of time in the absence of intermediate- to coarse-scale
disturbance events; however, empirical data presented within this study and previous work

137

(Chapter 2) suggest that change in species composition tend to be abrupt and due to
intermediate- and coarse-scale disturbance events.
The accuracy of historical reconstruction is greatly affected by the overlap between
disturbance types of varying spatial and temporal scales. Lorimer and Frelich (1989) and
Frelich and Graumlich (1994) point out that recent disturbances can remove members of an
older cohort, replacing them with younger cohorts. Therefore, the potential for one species to
ascend towards the canopy over another is not always discernable. Within plots 1 and 2 of
the dry variant, subalpine fir was consistently the most frequent species releasing and
ascending towards the canopy (Table 5.3). Within plots 3, 4 and 5 of the dry variant, the
most recent coarse-scale disturbance event in the latter part of the 1800s, which caused
43.9% of the sample trees to ascend towards the canopy (Table 5.5), allowed spruce to
establish and dominate over subalpine fir (Figure 5.5). In regions where fire acts largely as a
stand-replacing disturbance agent that allows few, if any, trees to survive, spruce and/or pine
tends to dominate seedling establishment (Veblen et al. 1991). The total number of spruce
entering the canopy, and the large proportion of gap-origin events compared to the small
proportion of release events, supports this (Table 5.3). Within the mesic variant, subalpine
fir was consistently the most frequent species releasing and ascending towards the canopy
(Table 5.3). Within the wet variant, western balsam bark beetle during the 1920s and 30s
substantially reduced the proportion of subalpine fir (Chapter 2, Figure 2.4). In response,
spruce was the most frequent species ascending into the canopy (Table 5.3). This refutes
evidence that increasing periods of time between stand-replacing events will eventually lead
to the replacement of spruce by subalpine fir.
If the number of canopy ascensions is related to new disturbances that remove
members of an older cohort, then ascensions alone are insufficient to provide insight for the
138

potential of one species to ascend into the canopy over another. Although it has been
suggested that increasing periods of time between stand-replacing events will lead to the
eventual replacement of spruce by subalpine fir, there is little evidence in any of our plots
that suggests this is actually occurring as a result of fine-scale disturbance. Neither the long­
term nor the short-term analyses support the idea of replacement, given that the decadal
canopy disturbance rates for species nested within ecological variant did not differ (Table

5.6).
The mean decadal canopy disturbance rates of 7.4, 6.4 and 5.0 % within the dry,
mesic and wet variant plots, respectively, suggest that fine-scale disturbance events do
gradually maintain and renew the current forest structure in the absence of intermediate- or
coarse-scale disturbance events. Although the differences between the ecological variants
were not statistically significant, the lack of significance could be explained by lack of
statistical power, which can be attributed to the high variation in the estimates of canopy
disturbance frequencies within the dry variant (Table 5.5). Within the dry variant, the mean
and maximum decadal disturbance rates were 7.4 and 43.9 %, respectively. This relatively
higher rate of disturbance can be explained by the intermediate- and coarse-scale disturbance
events that we were unable to partition completely from fine-scale events. The stand-level
rates of fine-scale canopy disturbance in the absence of stand-replacing events tend to
decrease, from the dry-warm interior plateau to the wet-cold foothills of the Canadian
Rockies (Table 5.5).
The estimates of decadal canopy disturbance rates are consistent with previous
estimates of canopy disturbance rates. Within spruce-subalpine fir forests of the mesic
variant, located less than 100 km south of the mesic variant plots in this study, Newbery
(2001) estimated canopy disturbance rates ranging from 5 to 8 % per decade. Within the
139

northern hardwood forests of Upper Michigan, Lorimer and Frelich (1989) estimated canopy
disturbance rates ranging from 5.7 to 6.9 % per decade. Within the hemlock-hardwood
forests of Upper Michigan, Frelich and Graumlich (1994) estimated canopy disturbance rates
ranging from 2 to 12 % per decade. Runkle (1985) points out that the rates of forest
disturbance average -1 0 % per decade and range - five to 20 % per decade, with little
variation between forest types, despite wide differences in vegetation and agents of
disturbance. The methods within our study were specifically designed to partition the
majority of the intermediate- and coarse-scale disturbance events from fine-scale events. The
end goal was to describe the fine-scale disturbance regimes. Therefore, it is not surprising
that the canopy disturbance rates in this study are at the low end of the range of disturbance
rates outlined by Runkle (1985), as his values describe all disturbance types combined.
Disturbance rates for specific agents of disturbance, or for particular components of a
disturbance regime, are expected to be relatively low (Runkle 1985).
A decreasing rate of fine-scale disturbance from the driest to wettest ecosystems, at
first, seems counterintuitive. Preliminary etiological analyses conducted for the three
ecosystems within this study, suggest that from the dry to wet variant, a number of fine-scale
disturbance agents tend to decrease in abundance, and some agents become completely
absent (R.D. Thompson, UNBC, unpublished data). For example, the incidence of Inonotus
tomentosus (Fr.) Teng., the causal agent of tomentosus root disease, is greatest and most
variable within the dry variant. Within the mesic variant, incidence of this disturbance agent
decreases, and within the wet variant, it is completely absent. Lewis et al. (2004) found that
incidence of tomentosus root disease was negatively correlated with direct measurements of
soil moisture. Whitney (1976) also found that disease incidence varied inversely with soil
moisture regimes as determined indirectly using a classification system. The relative soil
140

moisture content of zonal sites selected within the wet variant are greater than the soil
moisture content of zonal sites within the dry variant (Meidinger and Pojar, 1991), and may
partially explain the declining incidence of this disturbance agent. From dry to wet
ecosystems, the climate may become inhospitable to some agents of disturbance. If Runkle’s
(1985) hypothesis - rates of disturbance for specific agents of disturbance will be less than
estimates for an entire regional disturbance regime - is correct, then a decrease in the number
of different fine-scale disturbance agents from the dry-warm interior plateau to the wet-cold
foothills of the Canadian Rockies may partially explain the decreasing rates of stand-level
fine-scale disturbance.
Within this study, point-pattem analysis of live and dead spruce and subalpine fir
stems (>15 dbh) resulted in random spatial patterns at scales <25 m across all ecological
variants. Parish et al. (1999b) found that mortality was random within canopies of subalpine,
spruce-subalpine fir forest types, which in turn leads to random patterns of release. Based on
our Moran’s analysis, canopy ascension dates were spatio-temporally random, and average
canopy openings were =2.5 m in diameter across all ecological variants. Canopy opening
diameters, however, varied enormously from small (<2.5 m diameter) to fairly large (27.5 m
within the dry variant and >50 m within the wet variant) openings created within the canopy
(Figure 5.6). The random spatial pattern of canopy ascension could be due to the etiology of
the fine- and intermediate-scale disturbance agents occurring within these forest types, which
are primarily biotic, and may not result in gap-contagion. These results are similar to those
of Newbery (2001), who found that gap sizes averaged less than seven m in diameter, but
varied from small (<7 m) to large (=28 m diameter) canopy openings within sites similar to
our mesic variant plots. Frelich and Graumlich (1994) found a lack of spatial autocorrelation
of tree ascension dates for the smallest lag distance defined (10 m). This meant that the
141

average canopy opening diameter was less than 10 m within the hemlock-hardwood forests
of Upper Michigan. The difference in average canopy opening diameter between our
research and the results of Newbery (2001) and Frelich and Graumlich (1994) can be
explained by the lag distance specified in the spatial autocorrelation analyses (2.5 m versus 7
m, respectively) and how samples are collected for the Moran’s 1 analyses. The results of
Newbery (2001) and Frelich and Graumlich (1994) were based on the spatial analysis of
ascension dates sampled from evenly spaced grid points (7 and 10 m, respectively). The
estimates of canopy opening size, within the present study, were based on an irregular lattice
of stem-mapped points. This allowed spatial analysis at nearest neighbour distances (2.5 m)
which is much smaller than previous work. The results suggest that canopy openings on
average tend to be small, spatially independent and caused by the death of single trees.
Due to the variability in the distribution of canopy opening size, however, a number
of disturbance processes are working together through time. First, the relatively random
spatio-temporal pattern of small openings is indicative of structures resulting from a suite of
fine-scale disturbance agents. Second, the larger patch structures, identified through pointpattem analyses, are more likely indicative of the wide-spread insect outbreaks that were
suggested in Chapter 2. Together, this provides further evidence for the overlap between
different types of disturbances of differing intensities, together with fine-scale recmitment
between intermediate-scale disturbance events.
In conclusion, the results of this study suggest that intermediate-scale disturbance
events have played a larger mediating role over distinct changes in stand stmcture than that
of fine-scale disturbance events. Further, this study supports the proposal of Wilson and
Agnew (1992) that the punctuated equilibrium concept be extended to the spatial and
temporal processes in plant ecology. Frelich and Graumlich (1994) suggest that for this
142

concept to work within a forest ecology framework, a landscape of given size must exist in a
state of “punctuated quasi equilibrium” whereby disturbances with fine spatial extent relative
to the landscape maintain a quasi equilibrium during the intervals between larger
disturbances with a longer rotation period that punctuate the more stable periods. As we
move across the central interior of British Columbia, regardless of the climatic gradient,
examples of nested disturbance types are undoubtedly responsible for the patterns we see
today on the landscape. Therefore, it is feasible to extend this concept to even finer spatial
scales such as the individual stand composed of spruce and subalpine fir. To have a
disturbance regime of hierarchically nested disturbance types on a given forest landscape,
catastrophic disturbances such as stand-replacing fire must occur rarely so that several tree
generations may pass between disturbances. The consideration of disturbance events at
multiple scales shows that fine-scale disturbance exhibits continuous stand level maintenance
and renewal; furthermore, that changes in stand composition can be multidirectional and tend
to be abrupt, due to intermediate-scale disturbances such as those caused by insect outbreaks.

143

Literature cited
Abrams, M., Orwig, D. and Demeo, T. 1995. Dendroecological analysis of successional
dynamics for presettlement origin white-pine-mixed-oak forest in the southern
Appalachians, USA. J. Ecol. 83:123-133.
Andison, D. 1996. Managing for landscape patterns in the suh-horeal spruce forests of British
Columbia. PhD. Dissertation. Faculty of Forestry, University of British Columbia,
Vancouver, BC.
Antos, J. and Parish, R. 2002a. Structure and dynamics of a nearly steady-state subalpine
forest in south-central British Columbia, Canada. Oecol. 130:126-135.
Antos, J. and Parish, R. 2002b. Dynamics of an old-growth, fire-initiated, subalpine forest in
southern interior British Columbia: tree size, age, and spatial structure. Can. J. For.
Res. 32:1935-1946.
Bergeron, Y. and Charron, D. 1994. Postfire stand dynamics in a southern boreal forest
(Quebec): A dendroecological approach. Fcosci. 1:173-184.
Brett, R. and Klinka, K. 1998. A transition from gap to tree-island regeneration patterns in
the subalpine forest of south-coastal British Columbia. Can. J. For. Res. 28:18251831.
Canham, C. 1988. Growth and canopy architecture of shade tolerant trees: response to
canopy gaps. Fcol. 69:786-795.
Clements, F. F. 1916. Plant succession: an analysis of the development of vegetation.
Carnegie Institution of Washington. Washington, DC.
Cliff, A. D. and Ord, J. K. 1981. Spatial processes: Models and applications. Pion Limited.
London, England.
Coates, K. D. 2000. Conifer seedling response to northern temperate forest gaps. For.
Fcol. Man. 127:249-269.
Cook, J. F. and Peters, K. 1981. The smoothing spline: A new approach to standardizing
forest interior tree-ring width series for dendroclimatic studies. Tree-Ring Bull.
41:45-53.
Dale, M. R. T. 1999. Univariate Spatial pattern analysis in plant ecology. Cambridge Studies
in Ecology Series. Cambridge University Press, Cambridge, UK.

Daniels, L. D. 2003. Western redcedar population dynamics in old-growth firests:
Contrasting ecological paradigms using tree rings. For. Chron. 79:517:530.

144

Day, R. J. 1972. Stand structure, succession, and use of southern Alberta’s Rocky Mountain
forest. Ecol. 53:472-478.
DeLong, S.C. 2000. Natural disturbance rate and patch size distribution of forests in
northern B.C.: implications for forest management. Northwest Sci. 72:35-48.
DeLong, C. and Tanner, D. 1996. Managing the pattern of forest harvest: lessons form
wildfire. Bio. Cons. 5:1191-1205.
Denslow, J.S. 1987. Tropical rain forest gaps and tree speeies diversity. Pages 307-323 in
The ecology of natural disturbance and patch dynamics. S. Pickett and P. White,
editors. Academic Press, New York, NY, USA.
Diggle, P. J. 1983. Statistical analysis of spatial point patterns. Academic Press, New
York, NY, USA.
Duncan, R. P. and Stewart, G. H. 1991. The temporal and spatial analysis of tree age
distributions. Can. J. For. Res. 21:1703-1710.
Eastham, A. M. and Jull, Michael J. 1999. Factors affecting natural regeneration of Abies
lasiocarpa and Picea engelmannii in a subalpine silvicultural systems trial. Can.
J. For. Res. 29:1847-1855.
Frelich, L. E. 2002. Forest dynamics and disturbance regimes: Studies from temperate
evergreen-deciduous forests. Cambridge Studies in Ecology Series. Cambridge
University Press, Cambridge, UK.
Frelich, L.E., Calcote, R.R., Davis, M B. and Pator, J. 1993. Patch formation and
maintenance in an old growth hemlock-hardwood forest. Ecol. 74:513-527.
Frelich, L. and Graumlich, L. 1994. Age-class distribution and spatial patterns in an oldgrowth hemlock-hardwood forests. Can. J. For. Res. 24:1939-1947.
Frelich, L. and Reich, P. 1995. Spatial patterns and succession in a Minnesota southernboreal forest. Ecol. Mon. 65:325-346.
Hawkes, B. 1980. Fire history in Kananaskis Provincial Park - Mean fire return intervals,
pp. 42-45 in Proceedings of the fire history workshop, Oct. 20-24, 1980, Tucson,
Arizona, USDA, For. Serv. Gen. Tech. Rep. RM-81, Tucson, AZ, USA.
Flawkes, B., Vasbinder, W. and DeLong, C. 1997. Retrospective Fire Study Final Report:
Fire regimes in the SBSvk and ESSFwk2/wc3 biogeoclimatic units of northern British
Columbia. McGregor Model Forest Association, Prince George, BC.
Heinselman, M. 1981. Fire intensity and frequency as factors in the distribution and
structure of northern ecosystems, pp.7-57 in Fire Regimes and Ecosystem Properties,

145

December 11-15, 1978, Honolulu, Hawaii, USDA Forest Service Gen. Tech. Rep.
WO-26, Washington, DC.
Holmes, R. 1983. Computer assited quality control in tree-ring dating and measurement.
Tree-Ring Bull. 44:69-75.
Johnson, E.A. 1992. Fire and vegetation dynamics: studies from the North American
boreal forest. Cambridge studies in ecology. Cambridge Univ. Press, Cambridge
UK.
Johnson, E. and Rowe, J. 1977. Fire and vegetation change in the western sub-arctic.
Canadian Department of Indian Affairs and Northern Development. ALUR. 75:7661.
Kneeshaw, D.D. and Bergeron Y. 1998. Canopy gap characteristics and tree replacement in
the southeastern boreal forest. Ecol. 79:783-794.
Lawton, R. and Putz, F. 1988. Natural disturbance and gap formation in a wind exposed
tropical cloud forest. Ecol. 69:764-777.
Legendre, P. and Fortin, M. 1989. Spatial pattern and ecological analysis. Vegetatio.
80:107-138.
Lertzman, K. 1992. Patterns of gap-phase replacement in a subalpine, old-growth forest.
Ecol. 72:657-669.
Lertzman, K. and Krebs, C. 1991. Gap-phase structure of a subalpine old-growth forest.
Can. J. For. Res. 21:1730-1741.
Lertzman, K.P., Sutherland, G.D., Inselberg, A. and Sunders, S.C. 1996. Canopy gaps and
the landscape mosaic in a coastal temperate rain forest. Ecol. 77: 1254-1270.
Lewis, K. and Lindgren, S. 1999. Influence of decay fungi on species composition and
size class structure in mature Picea glauca x engelmannii and Abies lasiocarpa in
sub-boreal forests of central British Columbia. For. Ecol. Man. 123:135-143.
Lewis, K. and Lindgren, S. 2000. A conceptual model of biotic disturbance ecology in
the central interior of BC: How forest management can turn Dr. Jekyll into Mr. Hyde.
For. Chron. 76:433-443.
Lewis, K.J., Trummer, L.M. and Thompson, R.D. 2004. Incidence of tomentosus root disease
relative to spruce density and slope position in south-central Alaska. For. Ecol. Man.
194:159-167.
Lorimer, C. and Frelich, L. 1989. A methodology for estimating canopy disturbance
frequency and intensity in dense temperate forests. Can. J. For. Res. 19:651-663.

146

Lorimer, C., Frelich, L. and Nordheim, E. 1988. Estimating gap origin probabilities for
canopy trees. Ecol. 69:778-785.
Lundquist, J. and Beatty, J. 2002. A method for characterizing and mimicking forest
canopy gaps caused by different disturbances. For. Sci. 48:582-594.
Meidinger, D. and Pojar, J. 1991. Ecosystems of British Columbia. Crown Publications Inc.
Special Report Series No. 6. Victoria, BC.
Newbery, T. 2001. Small scale disturbance and stand dynamics in Inonotus tomentosus
infected and uninfected old-growth and partial cut, wet, sub-boreal forests in
British Columbia. M.Sc. Natural Resource Management, University of Northern
British Columbia, BC.
Oliver, C. and Larson, B. 1996. Eorest Stand Dynamics. John Wiley & Sons, Inc. New York,
NY, USA.
Oosting, H. J. and Reed, J. F. 1952. Virgin spruce-fir of the Medicine Bow Mountains,
Wyoming. Ecol. Mon. 22:69-91.
Payette, S., Filion, L. and Delwaide, A. 1990. Disturbance regimes of a cold temperate
forest as deduced from tree-ring patterns: the Tantare Ecological Reserve, Quebec.
Can. J. For. Res. 20:1228-1241
Parish, R., Antos, J. and Hebda, R. 1999a. Tree-ring patterns in an old-growth, subalpine
forest in southern interior British Columbia. Pp. 231-248 in Tree-ring analysis. Eds.
R. Wimmer and R. E. Vetter. CAB international. Wallingford, UK.
Parish, R. Antos, J. A. and Fortin, M.-J. 1999b. Stand development in an old-growth
subalpine forest in southern interior British Columbia. Can. J. For. Res. 29:13471356.
Peet, R. K. 1981. Forest vegetation of the Colorado Front Range: composition and
dynamics. Vegetatio. 45:3-75.
Pélissier, R. and Goreaud, F. 1999. Notice d’ADS - Module Ripley. Documentation d’ADS
in ADE, http://pbil.univ-lvonl.fr/ADE-4/ADE-4F.html Accessed September 1, 2005.
Pojar, J. Trowbridge, R. and Coates, D. 1984. Ecosystem classification and interpretation
of the sub-boreal spruce zone. Price Rupert Forest Region, British Columbia. Land
Manage. Rep. No. 17., British Columbia Ministry of Forests, Victoria, BC
Pyne, S. 1984. Introduction to wildland fire. John Wiley, New York, NY, USA.
Regent Instruments. 2003. WinDendro: Density V 2003c. Regent Instruments Inc.
Chicoutimi, Qc, Cananda.

147

Romme, W. 1980. Fire frequency in subalpine forests of Yellowstone National Park, pp. 2730 in Proceedings of the fire history workshop, Oct. 20-24,1980, Tucson, Arizona,
USDA, For. Serv. Gen. Tech. Rep. RM-81. Fort Collins, CO, USA.
Runkle, J. 1984. Patterns of disturbance in some old-growth forests of the eastern United
States. Ecol. 63:1533-1546.
Runkle, J. 1985. Disturbance Regimes in Temperate Forests. Chapter 2 in The ecology of
natural disturbance and patch dynamics. Eds. S. T. A. Pickett and P. S. White.
Academic Press, Inc. New York, NY, USA.
Sanborn, P., Geertsema, M., Jull, A. and Hawkes, B. 2001. Soil evidence for Holocene
fires in an inland temperate rainforest: east-central B.C. In Canadian Quaternary
Association Meetings, 2001: Program and Abstracts. Government of the Yukon,
Heritage Branch, Whitehorse, Y.T. Occasional Papers in the Earth Sciences 1.
Sawada, M. 1999. ROOKCASE: An excel 97/2000 Visual Basic (VB) Add-in for exploring
global and local spatial autocorrelation. Bull. Ecol. Soc. Am. 80:231-234.
Sokal, R.R. and Oden, N.L. 1978a. Spatial autocorrelation in biology: 1 methodology.
Biol. J. Linn. Soc. 10:199-228.
Sokal, R.R. and Oden, N.L. 1978b. Spatial autocorrelation in biology: 2. Some biological
implications and four applications of evolutionary and ecological interest. Biological
J. Linn. Soc. 10:199-228.
Sokal, R. R. and Rohlf, F. J. 1995. Biometry: The principles and practices of statistics in
biological research. 3"^ Edition. W.H. Freeman and Company. New York, NY, USA.
Stokes, M. and Smiley, T. 1968. An introduction to tree-ring dating. University of Chicago
Press, Chicago, IE, USA.
Tabachnick, B. and Fidell, L. 2001.Using Multivariate Statistics - 4* ed. Allyn and Bacon,
Needham Heights, MA, USA.
Upton, C. J. G. and Fingleton, B. 1985. Spatial data analysis by example. Vol. 1. Point
pattern and quantitative data. John Wiley & Sons, New York, NY, USA.
Veblen, T. 1986. Treefalls and the coexistence of conifers in subalpine forests of the
central Rockies. Ecol. 67:644-649.
Veblen, T. Hadley, K, Marion, S. and Rebertus, A. 1991. The Response of Subalpine
Forests to Spruce Beetle Outbreak in Colorado. Ecol. 72:213-231.
Velmex, Inc. 1992. The Velmex “TA” System for research and non-contact measurement
analysis. Bloomfield, NY, USA.

148

VoorTech Consulting. 2004. MeasureJ2X. Holdemess, NH, USA.
Whitney, R.D. 1976. Root rot of spruce and balsam fir in northwestern Ontario. Part I.
Damage and implications for forest management. Dept. Environ. Rept. O-X-241. Can.
For. Serv. Sault Ste. Marie, Ontario.
Wilson, J. B. and Agnew, A. D. Q. 1992. Positive feedback switches on plant
communities. Adv. Ecol. Res. 23:263-336.
Wright, E. P., Coates, K, D., Canham, C. D. and Bartemucci, P. 1998a. Species
Variability in growth response to light across climatic regions in northwestern
British Columbia. Can. J. For. Res. 28:871-886.
Wright, E. F., Coates, K, D. and Bartemucci, P. 1998b. Regeneration from seed of six
tree species in the interior cedar-hemlock forests of British Columbia as affected by
substrate and canopy gap position. Can. J. For. Res. 28:1352-1364.
Zar, J. H. 1996. Biostatistical Analysis. Edited by S. Fisher. Prentice Hall, Inc. Upper Saddle
River, NJ, USA.

149

Table 5.1. Structural attributes of the canopy layer (live trees >15 cm diameter at breast height, dbh).
Variant
Plot
Approx. stand initiation
Stand Density (trees/ha)

M esic

Dry

W et

1

2

3

4

5

1

2

3

4

5

1

2

3

4

5

1760

1760

1840

1840

1800

1740

1760

1700

1750

1760

<1650

<1650

<1650

<1650

<1650

344

308

540

572

376

364

400

448

372

444

432

412

356

352

440

35

10

70

66

59

35

30

38

38

31

60

65

80

70

63

64

90

21

15

37

60

70

63

61

69

40

33

19

30

37

1

0

9

19'

4

4

0

0

1

0

0

2

1

0

0

30.8

29.2

29.4

28.7

30.2

35.0

31.6

34.3

31.0

33.0

26.4

26.8

27.3

26.3

28.4

13.2

10.4

11.3

9.6

10.7

13.4

13.2

14.9

13.1

14.2

11.1

11.9

12.3

12.2

13.9

151.0

150.7

114.1

122.6

117.4

152.7

139.8

151.6

151.0

151.4

103.0

105.3

147.2

109.2

119.3

50.3

43.0

33.1

29.1

18.6

54.4

56.6

59.0

67.3

58.9

59.1

49.1

72.4

46.1

56.4

Stand Composition (%)
Spruce
Subalpine fir
Other
Stand strncture
dbh (cm, mean)
dbh (cm, s.d.)
Age (mean)
Age (s.d.)

^ Plot 4 within the dry variant, due to the random nature of all plot locations, had a small patch of Pinus contorta located on the plot
boundary.

LA

o

Table 5.2. Structural attributes of dead trees >15 cm diameter at breast height (dbh).
Variant
Plot
Density (trees/ha)

Dry

M esic

W et

1

2

3

4

5

1

2

3

4

5

1

2

3

4

5

428

572

356

304

436

336

260

204

312

272

452

568

516

480

488

15

29

58

54

39

12

68

47

37

54

4

3

12

3

7

51

27

28

24

47

61

29

43

35

29

49

63

44

60

57

1

0

9

14

5

1

2

0

1

0

0

0

0

0

0

33

44

4

8

9

26

2

10

27

16

48

35

44

38

37

35.4

35.7

22.8

23.5

26.2

32.9

39.6

33.4

39.6

40.5

32.2

28.1

26.5

28.5

30.0

12.6

13.6

7.1

9.7

10.9

11.5

12.0

11.9

15.1

15.5

9.7

7.8

9.7

8.9

10.1

Composition (%)
Spruce
Subalpine fir
Other
Unidentifiable
Structural attributes
dbh (cm, mean)
dbh (cm, s.d.)

Table 5.3. Proportion of canopy ascension types grouped by variant and by species.

Variant

Dry

Mesic

Wet

Number

Total

Type of ascension

of core

number of

(proportion of total)

samples

ascensions

Release

Gap origin

Spruce

282

337

0.36

0.64

Subalpine fir

209

253

0.52

0.48

Pooled

491

590

0.43

0.57

Spruce

173

297

0.59

0.41

Subalpine fir

328

424

0.63

0.37

Pooled

501

721

0.62

0.38

Spruce

334

294

0.28

0.72

Subalpine fir

161

132

0.30

0.70

Pooled

495

426

0.29

0.71

Species

152

Table 5.4. Summary statistics of mutual and partial independence tests between species,
variant and ascension type using contingency analyses. Column 1 represents the variable(s)
being tested for independence. Column 2 represents the variables against which column 1 is
tested against in the case of partial independence.

Independence test

Test grouping
Column 1

df

P

245.2

7

<0.001

105.5

6

<0.001

188.4

6

<0.001

140.7

5

<0.001

Column 2

Species
Mutual

Variant
Ascension
Species

Variant

Partial
Ascension
Variant

Species

Partial
Ascension
Ascension

Species

Partial
Variant

153

Table 5.5. Summary statistics for the untransformed decadal percentages of canopy ascension
grouped by variant and species.

Standard
Variant

Species

Maximum (%)

Mean (%)
deviation

Dry

Mesic

Wet

Spruce

7.8

10.6

47.3

Subalpine fir

7.2

9.6

46.4

Pooled

7.4

8.7

43.9

Spruce

7.5

7.2

31.3

Subalpine fir

5.9

6.2

28.6

Pooled

6.4

5.4

27.5

Spruce

5.1

5.7

24.6

Subalpine fir

4.8

6.7

32.4

Pooled

5.0

5.4

24.7

154

Table 5.6. Nested ANOVA results of percent decadal canopy ascension over the long-term
and the short-term (i.e., last 50 years of each disturbance chronology).

Source

SS

df

MS

F

P

Variant

805.665

2

402.833

4.761

0.117

Species (Variant)

253.835

3

84.612

0.947

0.418

Error

47527.119

532

89.337

Variant

501.215

2

250.608

1.952

0.286

Short-term Species (Variant)

385.211

3

128.404

1.331

0.267

Error

13753.120

144

96.508

Test

Long-term

155

Dry

M esic

W et

P lo t 1

P lo t 1

n= 3 0

n= 32

n= 65

P lo t 2

P lo t 2

P lo t 2

n = 30

n= 67

P lo ts

P lo ts

P lo t 3

%
/

n=95

n=42

n=7I

P lo t 4

P lo t 4

P lo t 4

n = 35

n=62

P lo ts

P lo ts

P lo ts

n= 55

n= 3 4

n=69

Distance (m)

Figure 5.1. Graph of L(d) versus distance for live spruce >15 cm DBH. The solid line shows
L(d), and the dotted grey lines show the 99% confidence envelope for random distribution.
Positive values of L(d) indicate clustering. Plots are grouped as dry (left), mesic (middle)
and wet (right) variants.

156

M esic

W et

Pot

P lo t 1

P lo tl

n=55

n=55

n=4S

P ot 2

P lo t 2

P lo t 2

P lo ts

P lo ts

P lo ts

%

n=28

P lo t 4

n= 17

P lo t 4

P lo t 4

/ -----

n=22

n= 2 6

P lo ts

n=S 5

d=41

D is ta n c e (m )

Figure 5.2. Graph of L(d) versus distance for live subalpine fir >15 cm DBH. The solid line
shows L(d), and the dotted grey lines show the 99% confidence envelope for random
distribution. Positive values of L(d) indicate clustering. Plots are grouped as dry (left), mesic
(middle) and wet (right) variants.

157

Dry

M esic

P lo t 1

P lo t!

n= 1 6

n = 10

W et

P lo t 2

n=41

n = 44

11=4

P lo t 3

P lo t 5,

P lo t 3

n=53

n = 30

n = 15

P lo t 4

P lo t 4

n=41

n=29

P lo ts

P lo ts

P lo t 4

n=37

D i s t a n c e (m )

Figure 5.3. Graph of L(d) versus distance for dead spruce >15 cm DBH. The solid line
shows L(d), and the dotted grey lines show the 99% confidence envelope for random
distribution. Positive values of L(d) indicate clustering. Plots are grouped as dry (left), mesic
(middle) and wet (right) variants.

158

Dry

M esic

W et

P lo t 1

P lo t 1

P lo t 1

n=55

n=51

n=56

P lo t 2

P lo t 2

P lo t 2

n = 19

P lo t 3

2.
“J

11=26

n= 22

n=57

P lo t 4

P lo t 4

P lo t 4

n=28

n = 72

P lo ts

P lo ts

n = 20

n= 6 9

\/
n= 18

n=51

D is ta n c e (m )

Figure 5.4. Graph of L(d) versus distance for dead subalpine fir >15 cm DBH. The solid line
shows L(d), and the dotted grey lines show the 99% confidence envelope for random
distribution. Positive values of L(d) indicate clustering. Plots are grouped as dry (left),
mesic (middle) and wet (right) variants.

159

Dry

1750

1800

1850

M esic

W et

P lo t 1

P lo t 1

P lo t 1

P lo t 2

P lo t 2

P lo t 2

1900
30
1

1

P lo t 3

25

i

ir '
,,,'Æm ,^h- JuJLLk.
20

15

1

1

1

1
1

1

10 •

10 ■

1850

1900

1850

1900

0

1950

P lo ts

P lo ts

1800

5 .

P lo t 4

P lo t 4

1800

P lo t 3

1

1950

1850

1900

D ecade

Figure 5.5. Disturbance chronologies for each plot showing the percentage of cored trees
ascending to the canopy in each decade. Each graph represents one plot with the plot
identification at top right. Black and grey bars represent spruce and subalpine fir,
respectively. Vertical dashed lines indicate chronology length. Plots are grouped as dry
(left), mesic (middle) and wet (right) variants.

160

6

0.5
D ry

D ry

- p lo t 1

0 .4 -

■ p lo t 2

4

” p lo ts

0.2 -

- p lo t 4

2

- p lo t 5

0

0.0

-

■2

0.2 -

-0.3 -

•4

-0 .4 -

•6

-0.5

10

0

20

30

40

0

50

10

20

30

40

50

10

20

30

40

50

10

20

30

40

50

6

0.5
0 .4 -

M esic

M e sic
4

2

§-

0

00
-

■2

0.2 -

-0 .3 -

•4

-0 .4 -

■6

-0.5

10

0

20

30

40

0

50

6

0.5
0 .4 -

W et

W et
4

2

0

0.0

-

■2

0,2 -

-0 .3 -

■4

-0 .4 -

6

-0.5

0

10

20

30

Distance (m)

40

50

0

Distance (m)

Figure 5.6. Correlograms of the Moran’s I coefficients (left) and standard normal deviates
(right) for canopy ascension dates by distance lag. Horizontal long and short-dashed lines
(right) represent the global and local significance levels, respectively.

161

CHAPTER 6 - SYNOPSIS: SUMMARY OF FINDINGS, MANAGEMENT
IMPLICATIONS, FUTURE RESEARCH AND CONCLUSIONS

Disturbance in forested ecosystems
Natural disturbance ecology encapsulates the interrelationships between the biotic and
abiotic components of an ecosystem, attempting to characterize the patterns and processes
influencing mortality and regeneration. In most forested ecosystems, disturbance is a key
driver of ecosystem function, affecting landscape pattern (DeLong and Tanner 1996, Hawkes
et al. 1997, DeLong 1998), stand structure (Oliver and Larson 1996, Kneeshaw and Burton
1997, Lewis and Lindgren 1999) and biodiversity (Bunnell 1995).
White and Pickett (1985) proposed the most widely used definition of disturbance as
any relatively discrete event in time that disrupts ecosystem, community, or population
structure and changes resources, substrate availability, or the physical environment. As part
of their definition, disturbances are regarded as discrete events in time. 1 suggest, however,
that it is their persistence on the landscape that sustains the patterns and processes so
important in maintaining biodiversity. Therefore, stand- or landscape-level disturbances are
not only characterized by their spatial structure, but include the temporal patterns of mortality
for the dominant individuals over extended periods of time. Collectively, the various spatial
and temporal components are referred to as a plant community’s “disturbance regime”
(Runkle 1985).
Historically, the disturbance regime has been defined in terms of the major
catastrophic events that originate from the physical environment, and was regarded as an
exogenous agent of vegetation change (White 1979). This paradigm has lead to a strong
research and management focus on catastrophic wildfire as the dominant agent of disturbance
162

within boreal ecosystems of Canada. These ecosystems generally have short fire return
intervals that maintain the landscape with low diversity in age-class structure, simply because
fire return intervals are shorter than tree life spans (Veblen 1986; Frelich and Reich 1995).
As the coarse-scale stand dynamics of sub-boreal ecosystems are similar to boreal
ecosystems, a focus on stand-replacing fire has been adopted for the sub-boreal and, in many
cases, the subalpine ecosystems of central British Columbia. Although this focus has been
informative, an understanding of what occurs in the natural absence of fire is limited.
In order to contribute to the knowledge and understanding of forest dynamics during
the intervals between stand replacing events, I have attempted to quantify and characterize
fine-scale disturbance as it varies within and between three distinct sub-boreal and subalpine
ecosystems of central British Columbia. The specific research questions were:
1. What are the spatial and temporal patterns of fine-scale disturbance across three
climatically distinct, mature spruce-subalpine fir forest types and do they differ;
2. What role has the combined effects of fine-scale disturbance played in forging the
present forest composition and structure; and
3. Does fine-scale disturbance lead to considerable changes in future forest
composition?
The remainder of this chapter summarizes the findings of this research, discusses potential
forest management applications and discusses future research needs.

Ecological summary
Disturbance histories
Mortality of canopy trees was indicated indirectly through canopy ascension
information as far back as 1700 within the SBSdwB, SBSwkl and ESSFwk2 variants. The
163

oldest individuals in these stands were over 290 years in the SBSdwS variant, well over 300
in the SBSwkl variant and approaching 400 years of age in the ESSFwk2 variant, indirectly
suggesting that stand-replacing events can remain absent from these stands for extensive
periods of time. In the absence of stand-replacing events, a number of disturbance agents
have acted to modify stand structure. In the mesic variant plots and in the oldest plots (I and
2) of the dry variant, canopy mortality during the 1870s removed substantial proportions of
the spruce basal area, which provided opportunity for canopy ascension. Within the
ESSFwk2 variant there were two disturbance events. The first occurred in the 1830s,
probably spruce beetle, and the second in the early 1930s has been identified as western
balsam bark beetle. These periods were present within all ESSFwk2 variant plots and
represent regionally synchronous outbreaks. The 1930s outbreak of western balsam bark
beetle removed a substantial proportion of subalpine fir, which released both spruce and
subalpine fir.

Canopy disturbance rates
The average canopy disturbance rates were 7.4, 6.4 and 5.0 % per decade within the
SBSdw3, SBSwkl and ESSFwk2 variants, respectively. These disturbance rates were
developed based on the species pooled because there were no significant differences between
the canopy disturbance rates for spruce and subalpine fir. From the driest to the wettest
variants, two changes occurred with regard to the fine-scale component of their respective
disturbance regimes. First, maximum decadal canopy disturbance rates decreased and
second, variation in the disturbance rates decreased.

164

Canopy opening size
Canopy opening diameter averaged ~2.5 m, corresponding to a gap area of ~ five m^.
Note that this gap size estimate was limited to the lag distance specified for the spatial
analysis (2.5 m), corresponding to the nearest-neighbour distances measured across all three
ecological variants. Larger gaps, as a result of multiple tree mortalities, were present in all
ecological variants; however, these events occurred less frequently than single tree gaps.
Furthermore, results from the point pattern and spatial-temporal analyses do not imply
strictly that the single tree mortalities - causing average sized canopy openings - are
independent events randomly dispersed through time. Lewis and Lindgren (2000), in their
conceptual model of disturbance ecology within sub-boreal and subalpine ecosystems,
suggest that decay fungi are important agents of disturbance, modifying the forest matrix by
causing individual or group mortalities over extended periods of time. Normally, decay fungi
do weaken the roots and/or the bole to such an extent that breakage is inevitable. Wind
storms, however, can aggregate such breakage in time, resulting in many small, spatially
distinct openings in the forest matrix that appear as a single event. The spatial pattern
analysis of dead stems supports this because the co-ordinates were generally random in
nature and spatial contagion rarely occurred. It was also found that regionally synchronous
bark beetle outbreaks cause similar spatial-temporal patterns throughout the forest matrix:
however, mortality is host specific and the disturbance event extended over larger spatial
areas.

Stand composition, canopy replacement and future stand composition
The results of this study indicate that increasing periods of time between standreplacing events do not necessarily lead to the replacement of spruce by subalpine fir. In
165

many plots, subalpine fir was exceedingly abundant in smaller diameter classes and younger
breast height age classes, suggesting that subalpine fir would eventually displace spruce.
Size and age distributions, however, are static representations of stand dynamics. Due to the
relatively shorter fire-retum intervals within increasingly dry/warm ecosystems, shifting
much of the landscape back successionally, the species shift from spruce to subalpine fir is
observed more often than not. Furthermore, the abundance in subalpine fir was rarely due to
a slow and methodical replacement of spruce caused by fine-scale disturbance. The complete
opposite occurred within the ESSFwk2 variant during the 1930s western balsam bark beetle
outbreak that substantially reduced the proportion of subalpine fir within all ESSFwk2
variant plots. These results suggest a common theme across all mature, spruce-subalpine fir
forest types found within the central interior of British Columbia. That is, future canopy
composition is not unidirectional or convergent, always leading to subalpine fir. On the
contrary, forest composition is quite dependent on the “punctuated” disturbance types
discussed in Chapter 5. That is, individual stands can exist in a state of “punctuated quasi
equilibrium” whereby relatively small-scale disturbances will maintain and renew forest
structure during the intervals between intermediate- scale disturbances with a longer rotation
period that punctuate the more stable periods.

Management implications
It has been suggested that successful conservation and maintenance of overall forest
health will require an understanding of disturbance patterns and processes (Pickett and
Thompson 1978, White 1979, White and Pickett 1985, Bergeron et al. 1999, Lewis and
Lindgren 2000). It has also been suggested that ecological values inherent in forest

166

ecosystems can best be maintained by using forest practices that mimie natural disturbanee
regimes (Bunnell 1995, Hawkes et al. 1997, Bergeron et al. 1999).
The disturbance chronologies presented in this research indicated that decadal canopy
disturbance was 7.4, 6.0 and 5.0 % within the SBSdw3, SBSwkl and ESSFwk2 variants,
respeetively. In that order, this would correspond to approximately 22 to 42, 22 to 33 and 18
to 22 mortalities per hectare per decade based on the current tree densities in each ecological
variant. Furthermore, again in that order, the average plot diameters of dead canopy trees
ranged from 22 to 38, 33 to 41 and 27 to 32 cm dbh. Species composition of dead trees (>15
em dbh) within the SBSdw3, SBSwkl and ESSFwk2 variants ranged anywhere from 15 to
58, 12 to 68 and four to 12 % spruce, respectively.
To aehieve the above ranges, partial-cut systems conducted every 10 years, would
remove an extremely wide range of possible densities, diameters and portions of each
constituent speeies (e.g., percent spruce). Furthermore, harvests would take on a variety of
canopy opening sizes, with the average diameter =2.5 m. This suggests that a collection of
single tree and small group selection systems would be the most appropriate to maintain
natural forest structure. Unlike “beetle proofing” a stand to minimize the future chance of
economic loss due to bark beetle infestations, these partial cuts would create uncertainty
around future intermediate-scale disturbance events. Beetle proofing is the application of
silvicultural treatments (e.g., stand thinning) meant to promote resistance in the remaining
trees to future beetle outbreaks and to minimize economic loss. Partial cut systems do not
minimize additional mortality due to the continuous low intensity nature of fine-scale
disturbance. As a result, there is an additive effect, where the expected fine-scale disturbance
events continue through time in addition to the removal of stems during each partial cut.
Therefore, it is unclear whether or not the spatial and/or temporal patterns of fine-scale
167

disturbance, as described in this thesis, can be mimicked reasonably by the application of
partial cut silvicultural systems.
An alternative approach is to focus on the imitation of intermediate-scale disturbance
events, such as the insect outbreaks identified within this study. For example, within the
watershed where the ESSFwk2 variant plots were sampled, three mortality events show up in
the majority of the mature tree ring records in the 1710s, 1830s and 1930s, suggesting that
the events are at least synchronous within the watershed. Although not conclusive, because
of our short dendroecological record, let us assume that these events were caused by spruce
beetle and western balsam bark beetle, and the approximate retum-interval would be 200
years with a 100-year lag between bark beetle species.
The application of this disturbance regime in the appropriate ecosystem would
involve harvest of one species (e.g., spruce), from a moderately narrow group of diameter
classes based on host susceptibility, over a 10-year period. This does not mean removing all
of the trees from one large, economically valuable, diameter class (i.e., high-grading). The
proportion removed from the susceptible diameter classes would be aligned with the
proportions historically removed by the disturbance agent (e.g., ~35 % of spruce stems >45
cm and <65 cm dbh). Regeneration would focus on the alternative species (e.g., subalpine
fir). The area would then be left unharvested until the end of the 100-year period, when the
process would begin again, although the focus would shift to the removal of subalpine fir and
the regeneration of spruce. Essentially, this silvicultural practice would be thinning from
above and would place small, spatially random, openings into the forest matrix over coarse
spatial scales.
In contrast to the application of partial cutting silvicultural systems, which involves
an entry every 10 years in any given area and requires continual activity over an entire
168

management area such as an entire watershed, imitating the temporal interchange between
spruce beetle and western balsam bark beetle over a 200 year rotation period would only
require one 10-year entry every 100 years.
Much like bark beetle proofing stands to minimize the future chance of economic loss
due to bark beetle infestations, this alternative silvicultural system would minimize the
chance for future intermediate-scale disturbance events. Conversely, such a silvicultural
system would not minimize further mortality due to the continuous low intensity nature of
fine-scale disturbance within any given stand of trees. Therefore, the continuous long-term
inputs of such things as wildlife trees and coarse woody debris would be maintained as a
result.
DeLong (2002) has developed a Natural Disturbance Unit (NDU) model for the entire
Prince George Region. In comparison to British Columbia’s five natural disturbance types
(Ministry of Forests British Columbia 1995), for the Prince George region alone there are
nine NDU. The SBSdw3 variant plots fall within the Moist Interior-Plateau NDU with fire
and mountain pine beetle (Dendroctonus ponderosae) as the key stand-replacement
disturbance agents (DeLong 2002). The disturbance cycle assigned to this NDU is 100 years
based on work by Andison (1996) and DeLong (1998). The SBSwkl variant plots fall within
the McGregor Plateau NDU with fire as the key stand-replacement disturbance agent
(DeLong 2002). The stand replacement disturbance cycle assigned to this unit is 220 years.
The ESSFwk2 variant plots fall within Wet Mountain NDU with stand-replacement
disturbance events occurring at irregular intervals with as much as 1000 years between such
events on any site (DeLong 2002).
Confronting forest ecosystem management are other management philosophies and
approaches such as integrated resources management, conservation, and zoning. Within the
169

wet mountain NDU, typical old forest characteristics historically dominated the landscape.
Therefore, it seems reasonable to apply the alternative silvicultural approach that attempts to
mimie bark beetle outbreaks over large spatial areas. In contrast, the disturbance cycle
within the SBSdw3 is 100 years. Therefore, very little eonnectively exist between the rarely
occurring mature spruce-subalpine fir forest types. Few ecological benefits would therefore
be realized by micro-managing old remnant patches that would be left behind naturally by
wildfire. Delong (2002) suggests that maintaining mature forest patches could be
accomplished by rotating old forest reserves. Within the McGregor Plateau NDU, old forest
charaeteristies were historically dispersed unevenly across the landscape (i.e., rare in some
areas and abundant in others). It seems reasonable to apply some combination of the two
extremes discussed above. Therefore, in some watersheds, silvicultural systems that attempt
to mimic the complex interchange between fine- and intermediate-seale disturbanee events
could be used. In other watersheds, larger patches with reserves could be used to mimie
catastrophic wildfire. Over extended periods of time (i.e., a disturbanee cycle ~ 220 years)
the silvicultural systems used in each watershed could be rotated.

Future research
Finally, the results of this study also provide tools to monitor the application of new
silvicultural systems in the spruce-subalpine forest types within the SBSdw3, SBSwkl and
ESSFwk2 variants. First, it is unclear whether or not the spatial and temporal patterns of
fine-seale disturbanee can be mimicked reasonably by silvicultural applications. Long-term
silviculture trials are required in these forests types to compare and contrast standard,
benchmark indicators of mature, spruce-subalpine forests. Will the application of
silvicultural systems intended to mimic natural disturbanee at fine spatial scales maintain
170

healthy forest ecosystems at the stand and landscape levels? Second, an alternative to partial
cut silviculture systems was to mimic regionally synchronous disturbance, such as bark
beetle infestations, across coarser spatial and longer temporal scales. It would be unrealistic
to set up a 200-year long silvicultural trial for monitoring purposes. Long-term forest
modelling, however, could provide evidence for the feasibility and ecological
appropriateness of such silvicultural systems by comparing predicted states to the current
states quantified in this thesis. Third, although the interplay between fine-scale disturbance
and intermediate-scale, patch modifying events were important factors of forest dynamics at
the stand level, stand replacing fire undoubtedly plays an increasingly important role moving
west from the high elevation, wet/cold environments of the subalpine forest to the dry/warm
plateau of central British Columbia. At the landscape level, it is expected that the proportion
of the land occupied by mature, spruce-subalpine fir forests would decrease and change in
spatial distribution on the landscape. What are the historical spatial and temporal
distributions of spruce-fir forests on across the central interior of British Columbia? Fourth,
it will be important to compare and contrast the silvicultural systems that mimic small- or
intermediate-scale events. Snag density and snag class distribution, coarse woody debris
(CWD) abundance and CWD size-class distribution, road density and stream crossings, are
examples of important variables that should be compared and contrasted at various spatial
and temporal scales. Does either system provide a better solution to disturbance based forest
management? Is there some other system more appropriate (e.g., clear-cut with permanent
reserves)? Is some other form of forest management (e.g., integrated resources management,
conservation or zoning) more appropriate? Finally, many conceptual models have been
proposed to explain the natural succession of spruce-subalpine fir forest types in western
North America. Some explain succession in the absence of disturbance and others maintain
171

that disturbance is paramount in predicting future stand structure and dynamics. This study
has presented further evidence that disturbance is the key driver of change and that steady
state communities are actually the minority in many sub-boreal and subalpine ecosystems.
Can all spruce-subalpine forest types through western North America really be explained by
one succinct successional model? Or, does the stochastic nature of multiple disturbance
types working at numerous spatial and temporal scales lead to individualistic succession
making it virtually impossible to predict future states?

Conclusions
This research has provided further evidence that fine-scale disturbances are important
ecological processes in sub-boreal and subalpine forest ecosystems. Within these
ecosystems, however, there is considerable overlap between disturbance types of varying
spatial and temporal scales. For example, intermediate-scale disturbance events have played
a large role in distinct successional changes in stand structure; more so than that of fine-scale
disturbance events. Consideration of disturbance events at multiple scales, however, shows
that fine-scale disturbance does contribute to continuous stand level maintenance and
renewal; furthermore, that shifts in species dominance tend to be abrupt and due to intermediatescale disturbances such as those caused by insect outbreaks.

172

Literature cited
Andison, D. 1996. Managing for landscape patterns in the sub-boreal spruce forests of
British Columbia. PhD. Dissertation. Faculty of Forestry, University of British
Columbia, Vancouver, BC.
Bergeron, Y., Harvey, B., Ledue, B. and Gauthier, S. 1999. Forest management guidelines
based on natural disturbance dynamics: Stand- and forest level considerations. For.
Chron. 75:49-54.
Bunnell, F. 1995. Forest-dwelling vertebrate faunas and natural fire regimes in British
Columbia. Cons. Bio. 9:636-644.
DeLong, C. 1998. Natural disturbance rate and patch size distribution of forests in
northern British Columbia: Implications for forest management. Northwest Sci.
72:35-48.
DeLong, C. 2002. Natural Disturbance Units of the Prince George Forest Region: Guidance
for Sustainable Forest Management.
ftp://prgftp.env.gov.bc.ca/pub/outgoing/srm/planning/landscape level September 1,
2005
DeLong, C. and Tanner, D. 1996. Managing the pattern of forest harvest: lessons form
wildfire. Biod. and Con. 5:1191-1205.
Frelich, L. and Reich, P. 1995. Spatial patterns and succession in a Minnesota southernboreal forest. Ecol. Mon. 65:325-346.
Hawkes, B., Vasbinder, W. and DeLong, C. 1997. Retrospective Fire Study Final Report:
Fire regimes in the SBSvk and ESSFwk2/wc3 biogeoclimatic units of northern
British Columbia. McGregor Model Forest Association, Prince George, BC.
Kneeshaw, D. and Burton, P. 1997. Canopy and age structures of some old sub-boreal
Picea stands in British Columbia. J. Yeg. Sci. 8:615-626
Lewis, K. and Lindgren, S. 1999. Influence of decay fungi on species composition and
size class structure in mature Picea glauca x engelmannii and Abies lasiocarpa in
sub-boreal forests of central British Columbia. For. Ecol. Man. 123:135-143.
Lewis, K. and Lindgren, S. 2000. A conceptual model of biotic disturbance ecology in the
central interior of BC: How forest management can turn Dr. Jekyll into Mr. Hyde.
For. Chron. 76:433-443.
Ministry of Forests. 1995. Biodiversity Guidebook.
http://www.for.gov.be.ca/tasb/legsregs/fpc/fpcguide/biodiv/biotoc.htm#begin
Accessed September 1, 2005.

173

Oliver, C. and Larson, B. 1996. Forest Stand Dynamics. John Wiley & Sons, Inc. New York,
NY, USA.
Pickett, S. and Thompson, J. 1978. Patch dynamics and the design of natural reserves.
Biology and Conservation. 13:27-37.
Runkle, J. 1985. Disturbance Regimes in Temperate Forests. Chapter 2 in The ecology of
natural disturbanee and patch dynamics. Eds. S. T. A. Pickett and P. S. White.
Academic Press, Inc. New York, NY, USA.
Veblen. T. 1986. Treefalls and the eo-existence of conifers in sub-alpine forests of the
central Rockies. Ecol. 67:644-649.
White, P. S. 1979. Pattern, process, and natural disturbance in vegetation. Bot. Rev.
45:229-299.
White, P, and Pickett, S. 1985. Natural Disturbance and Patch Dynamics: An Introduction,
pp. 3-13 in The ecology of natural disturbance and patch dynamics. Eds. S. T. A.
Pickett and P. S. White. Academic Press, Inc. New York, NY, USA.

174