DISTURBANCE HISTORY AND ITS INFLUENCE ON DROUGHT TOLERANCE
OF INTERIOR DOUGLAS-FIR (Pseudotsuga menziesii var. glauca (Beissn.) Franco)
IN THE CARIBOO-CHILCOTIN REGION OF BRITISH COLUMBIA, CANADA
by
Neil P. Thompson
B.Sc., University of Maine, 2013

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
IN
NATURAL RESOURCE AND ENVIRONMENTAL STUDIES

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
August 2017
© Neil P. Thompson, 2017

Abstract:
Disturbance—the death of trees due to external forces such as wildfire or windstorms—
drives forest stand dynamics and shapes ecosystems. Natural disturbances arising from the
interaction of climate, topography, and established tree species have often occurred with
some regularity, resulting in apparently perpetual renewal of particular forest structures.
Minor changes in climate can have profound impacts on these disturbance regimes,
breaking historically observed cycles and introducing novel stand conditions. Long-term
historical baselines are critical to understanding such changes. Observational records are
often inadequate, especially in western North America, where 400 year-old stands are
common but reliable data are generally unavailable prior to the 20th century. I use tree ring
analysis to investigate the history of Douglas-fir beetle and western spruce budworm
infestations, and the influence of partial disturbances on the drought tolerance of surviving
trees, developing baseline understanding of disturbance interactions in interior British
Columbia. No evidence is found of any outbreaks of western spruce budworm or Douglasfir beetle that exceed the magnitude of outbreaks in the early 21st century, suggesting that
recent outbreaks represent historically high levels of insect activity. Both natural and
anthropogenic partial disturbances are demonstrated to positively affect the drought
tolerance of surviving trees in old-growth remnants and younger managed stands,
respectively. Access by roots to areas with uninterrupted precipitation throughfall appears
to be the driving force behind observed drought resistance, and will likely become more
important under climate scenarios where droughts are prolonged and intensified by warmer
temperatures. A major growth release ca. 1800 brought previously suppressed trees from a
number of age classes into preeminence within their respective stands across the study
ii

area, forming the basis of the structure we now consider to be representative of old forests
in the region. I propose that mountain pine beetle is the leading cause of this regionally
synchronous growth release, based on the number of sites affected and the survival of
sapling-sized Douglas-fir. Silvicultural prescriptions designed to provide open growing
space to residual trees may help reverse overstocking resulting from wildfire exclusion and
enhance resilience of stands within the timber harvesting area to increased temperatures.

iii

Table of Contents
Abstract: .................................................................................................................................ii
Table of Contents .................................................................................................................. iv
List of Tables: ......................................................................................................................vii
List of Figures: ....................................................................................................................... x
Acknowledgements:...........................................................................................................xxii
1.

Introduction .................................................................................................................... 1
1.1.

Preface ..................................................................................................................... 1

1.2.

Study objectives and rationale: ............................................................................. 17

1.3.

Organization of dissertation .................................................................................. 18

1.3.1.

A note on maps: ............................................................................................. 18

2. Adjacency to a harvest corridor in partially harvested stands increases drought
resistance of interior Douglas-fir (Pseudotsuga menziesii var. glauca (Beissn.) Franco)
near the grassland interface in central British Columbia ..................................................... 19
2.1.

Abstract: ................................................................................................................ 19

2.2.

Introduction ........................................................................................................... 20

2.3.

Methods ................................................................................................................. 24

2.3.1.

Site Selection and Sample Collection ............................................................ 24

2.3.2.

Laboratory Preparation .................................................................................. 30

2.3.3.

Statistical Analysis ......................................................................................... 31

2.4.

Results ................................................................................................................... 34

2.4.1.

Drought conditions in event years ................................................................. 34

2.4.2.

Stand Conditions and Tree Characteristics .................................................... 34

2.4.3.

Postharvest Growth ........................................................................................ 39

2.4.4.

Climate Correlations ...................................................................................... 41

2.4.5.

Drought Resistance and Resilience................................................................ 43

2.5.

Discussion ............................................................................................................. 47

2.6.

Conclusion ............................................................................................................ 54

3. Drought resistance of old-growth interior Douglas-fir (Pseudotsuga menziesii var.
glauca (Beissn.) Franco) in the context of disturbance history in the Cariboo-Chilcotin
region of British Columbia .................................................................................................. 55
3.1.

Abstract ................................................................................................................. 55

3.2.

Introduction ........................................................................................................... 56

3.3.

Methods ................................................................................................................. 62
iv

3.3.1.

Site Selection and Field Methods .................................................................. 62

3.3.2.

Laboratory Preparation .................................................................................. 65

3.3.3.

Statistical Analysis ......................................................................................... 65

3.4.

Results ................................................................................................................... 74

3.5.

Discussion ............................................................................................................. 89

3.6.

Conclusion ............................................................................................................ 95

4. Douglas-fir beetle (Dendroctonus pseudotsugae) outbreak history inferred from tree
rings, wood scars and resin pockets in the Cariboo-Chilcotin region of British Columbia 97
4.1.

Abstract ................................................................................................................. 97

4.2.

Introduction ........................................................................................................... 98

4.3.

Methods ............................................................................................................... 102

4.3.1.

Field and Lab Methods ................................................................................ 102

4.3.2.

Selection of ringwidth chronologies for growth release reconstruction ...... 114

4.3.3.

Analysis: growth releases ............................................................................ 115

4.3.4.

Analysis: scars and resin pockets................................................................. 116

4.4.

Results ................................................................................................................. 118

4.5.

Discussion ........................................................................................................... 144

4.6.

Conclusion .......................................................................................................... 148

5. Reconsidering western spruce budworm outbreak history inferred from the annual
growth of host and non-host tree species in central British Columbia .............................. 149
5.1.

Abstract ............................................................................................................... 149

5.2.

Introduction ......................................................................................................... 150

5.3.

Methods ............................................................................................................... 158

5.3.1.

Site Selection and Field Methods ................................................................ 158

5.3.2.

Laboratory Preparation ................................................................................ 162

5.3.3.

Outbreak Reconstruction ............................................................................. 162

5.4.

Results ................................................................................................................. 166

5.5.

Discussion ........................................................................................................... 195

5.5.1.

Regionally synchronous periods of differential growth............................... 195

5.5.2.

Overview of total record of trees meeting outbreak criteria ........................ 204

5.5.3.

Limitations of host/non-host defoliation reconstruction method in this study
204

5.5.4.

A note on regeneration as it relates to western spruce budworm................. 205

5.6.
6.

Conclusion .......................................................................................................... 205

Concluding synthesis ................................................................................................. 208
v

6.1.

Abstract ............................................................................................................... 208

6.2.

Summary of chapter results ................................................................................. 209

6.3.

Ecological implications and silvicultural applications ........................................ 213

6.4.

Suggested topics for future research: .................................................................. 221

7.

Literature Cited .......................................................................................................... 223

8.

Appendix A: R Scripts Used in Analyses .................................................................. 239
8.1.

Introduction ......................................................................................................... 239

8.2.

Ring width to basal area increments (Chapters 2, 3, and 4) ................................ 240

8.3.

Climate correlations and response functions (Chapters 2 and 5) ........................ 242

8.4.

Generation of .tre files (Chapter 5) ..................................................................... 245

9.

Appendix B: AIC Calculation and Top Ranked Models for Chapter 3 ..................... 247

10.

Appendix C: ANOVA output tables from Chapter 3 ........................................... 271

vi

List of Tables:
Table 2.1: Mean 2015 stand characteristics measured in the field and 1975 diameter inside
bark calculated from tree rings. BA/ha value includes the basal area of subject tree
(accounting for 8m2/ha). LLB = lowest live branch (single), LLW = lowest live whorl of
branches (3 or more), DBH = diameter at breast height (1.3m measured on the high side of
the tree), BA/ha = basal area per hectare measured with 8m2/ha prism. DIB = diameter
inside bark. Classes that do not share a letter with others at the same site are significantly
different (Bonferroni post-hoc test, α=0.05); significant differences are in bold. ............... 35
Table 2.2: Live crown ratios on the basis of lowest live branch (LCR(B)), lowest live whorl
of three or more branches (LCR(W)), and sapwood basal area (m2) among three
competition classes at three sites (Bonferroni post-hoc test, α=0.05); significant differences
are in bold. ........................................................................................................................... 37
Table 2.3: Pearson correlation coefficients between monthly mean temperature and total
precipitation and total ring, earlywood, and latewood widths each detrended with a 10-year
cubic smoothing spline. Only significant (p < 0.05) correlation values are shown. Lower
and uppercase headings indicate months in previous and current year respectively. .......... 42
Table 2.4: Drought resistance values (drought/5-year pre-drought mean) among three
competition classes at three sites for the years 1988/89, 1995, and 2010 (Bonferroni posthoc test, α=0.05). ................................................................................................................. 43
Table 2.5: Drought resilience values (5-year post-drought mean/5-year pre-drought mean)
among three competition classes at three sites for the years 1988/89, 1995, and 2010
(Bonferroni post-hoc test, α=0.05)....................................................................................... 44
Table 2.6: Baseline drought resistance values (drought/1971–1975 mean) among three
competition classes at three sites for the years 1988/89, 1995, and 2010 (Bonferroni posthoc test, α=0.05). ................................................................................................................. 46
Table 3.1: Description of variables used in multinomial logistic regression models used to
explain the drought resilience of subject trees. .................................................................... 72

vii

Table 3.2: Candidate variable sets tested using multinomial logistic regression used to
explain the drought resilience of subject trees ..................................................................... 73
Table 3.3: Identified drought years and regionally narrow years (negative index values (10year spline) in Jasper and Banff chronologies), low reconstructed PDSI (negative values at
Cook’s reconstruction point #30) in the current and previous year, and below-average
previous June through current June total precipitation (pJJ). Gray cells indicate no data. . 75
Table 3.4: Site codes, elevation (meters asl), basal area per hectare in trees over 15cm dbh
(BAPH; m2 per hectare), trees per hectare in trees under 15cm dbh (TPH) and locations
(decimal degrees). Inventory data not collected at sites with grayed-out cells. .................. 76
Table 3.5: Drought resistance by age class. Kruskall-Wallis rank-sum test (95% CI; age
classes that do not share a letter within a given year are significantly different in pairwise
multiple comparisons. Pairwise comparisons not conducted for 1938 when overall model
was insignificant. No trees under 50 in 2010). .................................................................... 84
Table 3.6: Drought resistance in trees with and without ongoing release. Two-way
ANOVA; full model outputs in Appendix C. Significant variables in bold; α=0.05. .......... 85
Table 3.7: Top-ranked multinomial logistic regression models explaining drought resilience
scores for each drought year ................................................................................................ 86
Table 3.8: Coefficients of top-ranked models for each drought year. Bold values indicate
statistical significance in the original mlogit models (95% CI). Coefficients multiplied by
the value at the top of each column for concise display. Gray boxes indicate variables not
included in the top-ranking model. “1 vs. 2” indicates the probability of a tree being in the
middle 33% vs. the lowest 33%; “1 vs. 3” indicates the probability of a tree being in the
highest 33% vs. lowest 33%. Complete outputs can be found in Appendix D. ................... 88
Table 4.1: Estimated mortality within 500m buffer around increment core sites from 2001
to 2013. DFB = Douglas-fir beetle; MPB = mountain pine beetle. “% Total” columns
indicate the estimated percentage of the total number of hypothetical trees (n = 7,854)
affected. Rows highlighted in gray indicate the sites selected for further analysis. .......... 121
viii

Table 4.2: Summary tabulations of the preceding table: estimated mortality within the
500m buffer around increment core sites from 2001 to 2013. DFB = Douglas-fir beetle;
MPB = mountain pine beetle. “% Total” columns indicate the estimated percentage of the
total number of hypothetical trees (n = 7,854) affected. .................................................... 122
Table 4.3: Estimated mortality of trees within the 500m buffer around cookie collection
sites. DFB = Douglas-fir beetle; MPB = mountain pine beetle. “% Total” columns indicate
the estimated percentage of the total number of hypothetical trees (n = 7,854) affected. . 123
Table 5.1: Site attributes; 21st cent. = total years of defoliation; Med/Sev = years in which
defoliation was recorded as medium or severe. MAP = mean annual precipitation; BAPH =
basal area per hectare (m2) in trees >15cm DBH; TPH is trees per hectare <15cm DBH. 170
Table 5.2: Significant Pearson’s correlation coefficients (p<0.05) between detrended ring
width indices and monthly precipitation totals. Months in lowercase belong to the
preceding calendar year; months in uppercase belong to the current calendar year. ......... 171
Table 5.3: Significant Pearson’s correlation coefficients (p<0.05) between detrended ring
width indices and monthly average temperature. Months in lowercase belong to the
preceding calendar year; months in uppercase belong to the current calendar year. ......... 172
Table 5.4: Mean annual, seasonal, and monthly precipitation total at Williams Lake and
Kamloops weather stations from 1937 to 2012, with pairwise correlations and significance
score for each variable. Sample size varies due to missing data values excluded from the
analysis. pJJ=total precipitation from the previous June to the current June. JJA=total
precipitation for the months of June, July and August. All values are in mm. Significant
correlations marked by bolded variables names (P<0.05). ................................................ 192

ix

List of Figures:
Figure 1.1: Study area and distribution of Douglas-fir leading stands in British Columbia.
Douglas-fir leading stands in designated mule deer winter range (MDWR) highlighted in
black within the study area. MDWRs outside of the study area are not shown. ................... 6
Figure 1.2: Distribution of human and lightning initiated fires from 2000 to 2013,
highlighting affected stands of interior Douglas-fir
(https://catalogue.data.gov.bc.ca/dataset/fire-perimeters-historical accessed 2016) ........... 11
Figure 1.3: Distribution of Douglas-fir beetle from 2000 to 2013 according to provincial
fixed-wing aerial overview surveys and more detailed regional helicopter surveys.
Polygons recorded as “trace” severity class (<1% mortality) excluded from the map to
reduce visual clutter. Helicopter survey data accessed 2015:
http://www2.gov.bc.ca/gov/content/environment/research-monitoringreporting/monitoring/aerial-overview-surveys/methods/standards-for-detailed-surveys
fixed-wing survey data accessed 2014:
https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ ............................ 12
Figure 1.4: Western spruce budworm defoliation identified in provincial aerial overview
surveys from 2000 to 2013. Survey data accessed 2014:
https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ ............................ 13
Figure 1.5: Distribution of mountain pine beetle in the study area from 2000 to 2013. Areas
displayed in red are not Douglas-fir leading stands. Shades of red indicate the maximum
severity recorded at a location in any single year, not the cumulative impact. Severe
records indicate >30% mortality, moderate 11–29%, and light 1–10%. Data accessed 2014:
https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ ............................ 14
Figure 1.6: Tree PYP02 (ca. 1641) in 2013 surrounded by fallen lodgepole pine and dense
pine regeneration, with younger Douglas-fir at the edge of the frame to left and right, and
in the background in front of dense sapling pine. ................................................................ 15

x

Figure 2.1: Aerial image of Vert Lake site showing the regular pattern of harvesting trails
radiating from clearings where wood was collected and processed. ................................... 26
Figure 2.2: Conceptual schematic of interior, edge and open growing trees following
harvest (trees not drawn to scale). Labeled trees show the relative position of the three
competition classes within the stand. ................................................................................... 27
Figure 2.3: Location of study sites. Douglas-fir dominated stands are those where Douglasfir is classified as the leading species in the Vegetation Resources Inventory; the grassland
benchmark describes historical grassland areas that may now be sparsely treed. ............... 28
Figure 2.4: Typical examples (left to right) of Open, Edge and Interior class trees indicated
by white triangles. ................................................................................................................ 30
Figure 2.5: Examples of cores from Open (COL59A) and Interior (COL58A) classes at
Colpitt Lake site, harvested 1975 (black arrow). Drought years 1988/89 marked by black
rectangles labeled “D” above each core, and magnified for each sample at lower right.
Drought years 1995 and 2010 marked by black rectangles beneath each core. Resistance
equals D/Y, resilience equals Z/Y, and baseline resistance equals D/X, where X Y and Z are
average basal area increment over the 5-year periods indicated and D is the average basal
area increment of 1988 and 1989. Hollow triangles indicate the color change at the
heartwood/sapwood boundary. Pencil marks indicate decade years, from the left: 1940,
1950, 1960, 1970, 1980, 1990 (also in magnification), 2000, and 2010. ............................ 33
Figure 2.6: Mean annual basal area increment plotted by competition class among all sites
since 1950. Vertical dashed line indicates the first recorded defoliation by western spruce
budworm in 2001. Black bars above the X axis indicate the drought years under study
(1988/89, 1995, and 2010). Two sites (Colpitt and Sting Lakes) were harvested in 1975,
one (Vert Lake) harvested in 1979. ...................................................................................... 36
Figure 2.7: Live volume (m3 based on 12.5cm utilization) in areas designated for mule deer
winter range conservation by crown closure (%) in 40-year age classes. Total area
(hectares) in each age class is noted beneath X axis............................................................ 38

xi

Figure 2.8: Basal area and mean annual basal area increment at each site (cm2). Raw data
are displayed below; differences among the classes are significant for a given period if
values do not share a letter (Bonferroni post-hoc test, α=0.05). .......................................... 40
Figure 2.9: Example of crown shyness where the crown of the edge tree (left) has been
limited by an interior tree (right) of similar height. ............................................................. 50
Figure 2.10: Regeneration in a skid trail at Vert Lake ......................................................... 51
Figure 3.1: Tree PUD18, ca. 1721, surrounded by dead lodgepole pine and regeneration of
pine and Douglas-fir, with mature Douglas-fir in the background. ..................................... 59
Figure 3.2: Study sites in the context of Douglas-fir leading stands and permanent old
growth management areas (OGMAs). ................................................................................. 64
Figure 3.3: Increment cores BLU 17A ("17"; release) and BLU 15A ("15"; no release).
Drought resistance equals Y/X; drought resilience equals Z/X. Core 17 shows an example
of a false ring in 1931, followed by continued growth that accounts for more than half of
the ring width and a high drought resistance score. Black dots indicate decade years, from
the left: 1910, 1920, and 1930. ............................................................................................ 67
Figure 3.4: Detrended ring width index (10-year cubic smoothing spline) calculated from
all trees in the study. Dotted line indicates 10th percentile (0.7978) used to define drought
events. Black boxes indicate drought years selected for analysis........................................ 74
Figure 3.5: Percentage of the total sample reaching coring height in a given decade
(“Initiation”; grey) and percentage of trees present at the end of each decade showing
growth release >50% in each decade (black). ..................................................................... 77
Figure 3.6: Mean basal area increments by 50-year age class across the entire study area
since 1615, smoothed with a 50-year spline prior to averaging to display long-term growth
trends. ................................................................................................................................... 78

xii

Figure 3.7: Overall diameter distributions in 1800 and 1840, with 1840 trees separated into
accretion and ingrowth, respectively those that had and had not reached breast height by
1800. .................................................................................................................................... 79
Figure 3.8: Core GAR14A from 1740 to 1840 with basal area increment. Single pencil
marks indicate decade years, double mark indicates 1750, and triple mark identifies 1800.
Radial growth from 1800 to 1840 was 6.5cm, suggesting approximately 13cm diameter
growth at an average rate of 3.25mm per year. The widest ring (1820) is 2.9mm,
suggesting a diameter growth of 5.8mm on a tree 26cm DBH. Basal area increment
continued to increase until 1917, where a period of suppression marks the beginning of a
slow decline. The crown was completely isolated when sampled in 2014, with live limbs
beginning at 3–4m height. ................................................................................................... 81
Figure 3.9: Tree GAR14 (ca. 1609) in 2013. Underexposure of original photograph
corrected in GIMP 2............................................................................................................. 82
Figure 3.10: Tree JKY10 (ca. 1723), 106cm DBH (outside bark), apparently isolated since
1800 after being released at the age of 101 ca. 1800 (white triangle at left). Core JKY10C,
at left: 1769–1860; at right: 1861–2013. Single dots at decade years, double-dots at
1850/1950, triple dots at 1800/1900, quadruple dots at 2000. Raw and 50-year splined
basal area increment (BAI) graphed as solid and dashed lines respectively. ...................... 83
Figure 3.11: Site JKY in 2013, with an average 105m2/ha and 12,692 trees per hectare
under 15cm. Tree at center dates to 1555; tree behind and to the left dates to 1475, both
have charred bark but no open wound. ................................................................................ 90
Figure 4.1: Sampled cull pile locations for Douglas-fir beetle outbreak reconstruction
overlaid by a combination of Federal Forest Insect and Disease Survey and Provincial
Aerial Overview Survey records of beetle-caused mortality identified by insect surveys
from 1957–2013, Provincial helicopter-based mortality surveys from 2007–2014, and
forest stands dominated by Douglas-fir. “Trace” severity (<1% mortality) not shown on
map to reduce clutter. Five ringwidth chronologies labeled with three-letter codes. ........ 103

xiii

Figure 4.2: Top: sampled cull pile at site H; photograph corrected for underexposure and
white balance in Microsoft Office. Bottom: Cross sections collected at site K. Each site
filled a midsize pickup truck. ............................................................................................. 104
Figure 4.3: Partially healed injury in the year prior to death of tree D09, photographed
before (top) and after (bottom) sanding of the bark to reveal resin saturation of the apparent
gallery. Photographs cropped but not manipulated with regards to color or contrast. ...... 108
Figure 4.4: Injury to tree D09 in the year prior to being killed by DFB. Photograph cropped
but not manipulated with regards to color or contrast. White arrows mark the bounds of the
apparent gallery in the bark. .............................................................................................. 109
Figure 4.5: Progression of scar formation: A: Scar with extensive traumatic resin duct
formation (top triangle); apparent resin-filled entry tunnel (middle triangle), and frass
(bottom triangle) from a tree at Esler Canyon (an early reconnaissance site not large
enough to fully sample). B: Deceased Douglas-fir beetle in current-year gallery, area
impregnated with resin, tree D10. C: 2013 scar with apparent gallery tree M23; 2014 ring
beginning to close over wound. D: 1990 scar, mirrored in the bark tree K20. .................. 110
Figure 4.6: A typical scar in the process of formation (A), a typical resin pocket (B), and an
empty beetle exoskeleton stuck in a current-year gallery adjacent to a 2010 resin pocket (C
and D). The key distinction between a scar and a pocket is that the scar is healing over an
open wound (e.g. the arrow at A) while the wood has simply grown around the resin
pocket without interruption (e.g. the arrow at B). ............................................................. 111
Figure 4.7: Mountain pine beetle (black arrow) arrested within the bark of lodgepole pine
by defensive resins. A resin pocket would be identified in a cross-section above this point
while a scar would be found at the level of the failed gallery. Underexposure of original
image corrected in GIMP 2.0. ............................................................................................ 112
Figure 4.8: Six of eight scars in tree M23 in 2013; note the relative widths of the scar and
the resin-impregnated area in the phloem, the heavy production of resin ducts radiating
from all scars in 2013 (circled in E) and 2014 especially in C, and the discoloration of the

xiv

sapwood especially in A and F. Measurements of B were made on the opposite side of the
cookie; pencil slash is a reminder to not count the same scar twice. ................................. 113
Figure 4.9: Rate of attack expected within the surface area represented by cross-sections of
various diameters based on the rates described by Johnson & Pettinger (1961) at a distance
of 0–1.2m and 1.5–4.5m from the sampled tree, assuming cookie thickness of 6cm and two
samples per tree. ................................................................................................................ 117
Figure 4.10: Estimated number of trees killed by Douglas-fir beetle recorded by aerial
surveys since 1957. “Trace,” “Light,” “Moderate,” “Severe,” and “Very Severe” are codes
used in fixed-wing aerial surveys; “AoS Spot” refers to spot attacks identified in the fixedwing aerial survey, and “Helicopter” accounts for all trees tallied in helicopter-based
surveys (helicopter data 2007-2013).................................................................................. 120
Figure 4.11: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site BLU. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.
........................................................................................................................................... 125
Figure 4.12: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site EPI. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.
........................................................................................................................................... 127
Figure 4.13: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site GAR. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.
........................................................................................................................................... 128
Figure 4.14: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site SUG. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.
........................................................................................................................................... 129
Figure 4.15: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site WIL. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.
........................................................................................................................................... 130

xv

Figure 4.16: Scars, resin pockets, and growth releases identified in cross sections at Site D.
Total count of scars (red) labeled where the bar exceeds the standard Y axis. .................. 131
Figure 4.17: Scars, resin pockets, and growth releases identified in cross sections at Site H.
Total count of scars (red) labeled where the bar exceeds the standard Y axis. .................. 132
Figure 4.18: Scars, resin pockets, and growth releases identified in cross sections at Site K.
Total count of scars (red) labeled where the bar exceeds the standard Y axis. .................. 133
Figure 4.19: Scars, resin pockets, and growth releases identified in cross sections at Site M.
Total count of scars (red) labeled where the bar exceeds the standard Y axis. .................. 134
Figure 4.20: Scars, resin pockets, and growth releases identified in cross sections at Site P
........................................................................................................................................... 135
Figure 4.21: Scars, resin pockets, and growth releases identified in cross sections at Site W.
Total count of scars (red) and resin pockets (black) labeled where the bar exceeds the
standard Y axis. .................................................................................................................. 136
Figure 4.22: Scars and resin pockets at all sites displayed as a ratio to the total number of
Douglas-fir >12.7cm diameter, and the percentage of Douglas-fir (of any diameter) with
ongoing growth release. Ratio labeled for 2013 where bar exceeded the Y axis. ............. 137
Figure 4.23: Sampled Douglas-fir trees recording initiation of growth release (above 0%
line) and ongoing release (below 0% line) as a percentage of total sample size in each year,
displayed by site, with 11-year running total of initiations (total percentage, subject year
plus preceding and following five-year periods). .............................................................. 138
Figure 4.24: Years where one or more trees recorded multiple scars or resin pockets within
a single year. Gray dashed line indicates the percentage accounted for by a single tree. .. 139
Figure 4.25: Mountain pine beetle, wildfire, and drought histories inferred from tree ring
analysis and annual insect surveys compared to scars attributed to Douglas-fir beetle
attack. “Growth Release” is coded gray when two or three trees and black for four or more
trees were recording growth release in a given year. “DFB Mortality Estimate” (see Figure
xvi

4.11) is coded gray when >5,000 and black when >50,000 trees were estimated to have
been killed. Cook & Krusic’s (2004) PDSI reconstructions are coded gray when between 0
and -0.5, and black when below -0.5. Daniels & Watson JJA = reconstructed June, July and
August precipitation total at their site near Williams Lake; Daniels & Watson pJJ =
reconstructed previous-July through current-June precipitation total at their site near
Williams Lake. All records under the “Drought” heading except Cook’s PDSI
reconstruction highlight only prolonged periods of low precipitation reconstruction values.
Daniels & Watson 2003 fire records coded gray when only one tree at one site recorded a
fire and black when two or more trees were affected at a single site or trees at two or more
sites were affected. Harvey 2017 indicates years in which >25% of trees at the Churn
Creek Protected Area study site recorded a wildfire. Hawkes et al. 2004 coded black if 10
or more trees recorded a fire scar. Axelson et al. 2010 coded black if two or more sites
recorded a fire and gray if fire scars were found at only one site. Erickson 1992 coded gray
for localized infestations affecting <10,000 trees and black for larger infestations. Alfaro et
al. 2010 coded black if two or more scars were attributed to beetle attacks in that year. . 143
Figure 5.1: Present range of ponderosa pine, lodgepole pine, and Douglas-fir in southern
portion of the study area and adjacent regions, overlaid by Chasm Provincial Park and the
regional centers of Lillooet and Kamloops. BC VRI data used under Open Government
License 2.0. ........................................................................................................................ 153
Figure 5.2: Cumulative years of defoliation near the northern extent of surveyed
defoliation in the Fraser River valley, with contiguous Douglas-fir forests in the upland
region to the west. Contour interval is 25m, with thick contours at 100m. Aerial Overview
Survey and Vegetation Resources Inventory data used under BC Open Government
License 2.0. ........................................................................................................................ 157
Figure 5.3: Map of study sites overlying the cumulative defoliation history from 2001–
2013 and the range of Douglas-fir leading stands according to the British Columbia
Vegetation Resources Inventory (2016). 20th century defoliation at Quesnel Lake and Riske
Creek indicated in brown. The Chilcotin region is the area west of the Fraser River; the
Cariboo is east of the Fraser. .............................................................................................. 160
xvii

Figure 5.4: Annual progression of surveyed western spruce budworm defoliation, 2001–
2008. .................................................................................................................................. 167
Figure 5.5: Annual progression of surveyed western spruce budworm defoliation, 2009–
2013; total years of defoliation in the 21st century; maximum severity of defoliation in the
21st century overlaid by site locations; stands with Douglas-fir listed as the leading,
secondary, and tertiary species........................................................................................... 168
Figure 5.6: Reconstructed outbreaks based on the percentage of trees at each site meeting
the defined outbreak criteria. Light gray exceeds the 15% threshold; medium gray 50%,
and black 75%. 21st=total years of surveyed defoliation in the 21st century outbreak.
Vertical dashed line at 2001 indicates the beginning of the 21st century outbreak. ........... 174
Figure 5.7: Mean detrended chronology values for all Douglas-fir in the study and
ponderosa pine near Kamloops, and the Douglas-fir chronology less the ponderosa pine
chronology (host minus non-host) from 1640 to 2011. Areas shaded by gray bars indicate
the periods of regionally synchronous reconstructed outbreaks ca. 1744–1752, 1770–1786,
1795–1808, 1854–1861, 1870–1882, 1899–1909, and 1938–1948, the sporadic period of
the 1980s and 1990s, and the known outbreak period 2001–2011. ................................... 175
Figure 5.8: Mean detrended chronology values for all Douglas-fir in the study and
ponderosa pine, and the Douglas-fir chronology less the ponderosa pine chronology (h-Nh
= host minus non-host) for eight periods of regionally synchronous reconstructed outbreak
activity ca. 1744–1752, 1770–1786, 1795–1808, 1854–1861, 1870–1882, 1899–1909 and
1938–1940, the sporadic period of the 1980s and 1990s, and the known outbreak period of
2001–2011.......................................................................................................................... 176
Figure 5.9: Maximum percentage of trees meeting outbreak criteria in a given year at each
site in the period 2001–2011, overlaying the cumulative years of defoliation in the 21st
century outbreak and the range of Douglas-fir leading stands .......................................... 177
Figure 5.10: Forest Insect and Disease Survey records from 1985 to 1996 showing data for
two-year cycle budworm, mountain pine beetle, pine needle cast, and Douglas-fir beetle,

xviii

overlaid by site locations displayed by the maximum number of trees meeting outbreak
criteria during this period. .................................................................................................. 178
Figure 5.11: Maximum percentage of trees meeting outbreak criteria in a given year at each
site in the period 1938–1948, overlaying the cumulative years of defoliation in the 21st
century outbreak and the range of Douglas-fir leading stands .......................................... 179
Figure 5.12: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1899–1909, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands ................................... 180
Figure 5.13: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1870–1882, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands.................................... 181
Figure 5.14: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1854–1861, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands.................................... 182
Figure 5.15: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1795–1808, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands.................................... 183
Figure 5.16: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1770–1786, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands.................................... 184
Figure 5.17: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1744–1752, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands.................................... 185
Figure 5.18: Percent of non-host sample initiating growth release and percent with ongoing
release (growth increase >50%) calculated using 20-year periods. ................................... 186

xix

Figure 5.19: Six outputs of the R package TRADER showing growth trends in ponderosa
pine, displaying the annual basal area increment in black (left Y axis) and the percent
growth change in blue (right Y axis), growth releases marked by dashed vertical lines. .. 187
Figure 5.20: Previous June through current June precipitation totals (pJJ) at Kamloops
from 1897–2012 and Williams Lake from 1937–2012. Mean lines at 313mm for Kamloops
and 548mm for Williams Lake. Periods of below-average precipitation for each site
marked in gray. .................................................................................................................. 188
Figure 5.21: Cook's reconstruction of Palmer’s Drought Severity Index at grid points 30
(90km northwest of Williams Lake) and 42 (76km south-southeast of Kamloops) for the
years 1640–1940. Years of reconstructed outbreaks marked in gray above the 0 line, years
where PDSI reconstruction is higher (more moist) at the point nearer Kamloops marked in
gray below the 0 line. ......................................................................................................... 190
Figure 5.22: Archived ponderosa pine core PP14 from the UNBC Tree Ring Lab showing
growth reduction in the late 1920s/early 1930s with growth release beginning ca. 1935 . 191
Figure 5.23: Percent of total ponderosa pine (PIPO, beneath X axis) and Douglas-fir
(PSME, above X axis) with ongoing growth release exceeding light (50%), moderate
(100%), and severe (200%) thresholds on the basis of 20-year average basal area increase.
Black bars indicate growth differentials meeting outbreak receonstruction criteria on a
regionally consistent basis. ................................................................................................ 193
Figure 5.24: Percentage of sites with reconstructed outbreaks exceeding 15, 50, and 75%
thresholds, and ponderosa pine trees displaying growth increase exceeding 50, 100, and
200% thresholds on the basis of five-year averages. ......................................................... 194
Figure 6.1: Tree PYP01, ca. 1619, surrounded by several age classes of lodgepole pine
regeneration and Douglas-fir of at least three cohorts. ...................................................... 218
Figure 6.2: Continuity of Douglas-fir from the coast to the interior ‘dry belt’ via the Bute
Inlet. Stands containing Douglas-fir (BC Vegetation Resources Inventory; primary,
secondary, or tertiary species) overlaying the mean annual precipitation (mm; “MAP”)
xx

from 1961 to 1990 (www.climatewna.com accessed 2017), showing the ‘bridge’ of
Douglas-fir from the high-precipitation coastal region to the dry interior areas described in
this dissertation (study area corresponds to the northeast quadrant of this map................ 222

xxi

Acknowledgements:
I would like to thank Drs. Kathy Lewis and Lisa Poirier for taking me on as an M.Sc.
student and supporting my transfer to the doctoral program. It has been a pleasure to have
two co-supervisors who work so well together, consistently providing suggestions in the
same direction. Funding for this work was provided through Dr. Lewis’ NSERC Discovery
Grant, Dr. Poirier’s research fund, and the UNBC Research Project Award program.
Thanks also to my committee members: Dr. Mark Dale for his mathematical expertise, Dr.
Chris Johnson for his wide-ranging advice on statistics and writing, and Dr. Jodi Axelson
for indulging a somewhat overzealous replication of elements of her own work.
I’ve had the pleasure of working with a number of excellent field and lab assistants:
Pamod Hemakumara, Martin Janik, Matthew Waldie, Brendan Carswell, David Powe,
Trevor de Zeeuw, Evan Ball, Milo Bookout and Marion Simon all performed work critical
to the success of the studies presented here. Kirsten Thompson (née Kilde) allowed me to
pass a day’s fieldwork off as a date, married me anyway, and has put in many more days in
the field and lab since. I’m sure she has a lot to do with any sanity I may have retained
through graduate school.
Among many excellent professors at the University of Maine, Dr. Alan White has my
deepest gratitude for his patient instruction and encouragement in the field of disturbance
ecology. I hope I have followed Dr. Robert Seymour’s first commandment of tree ring
research, to remember and pursue the practical values of the science and return a product
of genuine usefulness. I blame Ben Rice for teaching me how to conduct an efficient field
sampling campaign, and will be forwarding any complaints about sample storage and
disposal to his circular file.
Thanks of course to my father, who buys red ink by the barrel and edited many early drafts,
and to my mother for her consistent encouragement and example.

xxii

1. Introduction
1.1. Preface
“After due reproofs of the late impolitic waste and universal sloth amongst us, we should
now turn our indignation into prayers, and address ourselves to our better-natured
countrymen, that such woods, as do yet remain entire, might be carefully preserved, and
such as are destroyed, sedulously repaired: It is what all persons who are owners of land
might contribute to, and with infinite delight, as well as profit, who are touched with that
laudable ambition of imitating their illustrious ancestors, and of worthily serving their
generation.”
-- J. Evelyn in the preface to the 1678 print edition of Silva
Forest disturbances shape ecological communities by killing or damaging
established trees, making growing space and other resources available to survivors or new
individuals (Oliver & Larson, 1996). In many regions, interactions between climate,
topography, and established tree species have resulted in repeated patterns of natural
disturbance and renewal that are taken as normative and used as a benchmark for managed
forests (Cooper, 1913; Pickett & White, 1985). This perspective is represented by Aldo
Leopold’s reflections on land health and conservation, which he defined as “the capacity of
land for self-renewal” and “our effort to understand and preserve this capacity”
respectively (Leopold, 1949). Under this paradigm, natural disturbances are seen as a part
of the environment with which species regularly interact, and an understanding of this
ecological dynamic between forests and disturbances is considered to be the keystone of
both active conservation in production forestry and stewardship of parks and preserves
(Smith et al., 1997).
A question facing ecologists and foresters around the world is whether the range of
disturbances observed today are a continuation of the patterns that shaped the structure and
composition of ecosystems in the recent past, or if they are a novel response to human
1

influence and the beginning of an unknown trajectory (Romme & Despain, 1989; Turner,
2010). Historical knowledge of disturbance type, extent, rate and severity, as well as the
predisposing factors and outcomes of numerous events, is prerequisite to approaching this
question on a regional basis: that baseline is the theme of this dissertation.
A broad range of natural disturbances affect the succession of forest communities
over time, but insect outbreaks are often among the most common and influential (Dale et
al., 2001; Bentz et al., 2010). Trees may be predisposed to insect attack by local
environmental conditions, climate change, stand composition and density, host genetics,
other biotic and abiotic disturbance events (Manion, 1981), and these factors may interact
positively and negatively at different spatial scales (Raffa et al. 2008). Mechanisms of
predisposition include influences on the lifecycle of the pest, the defensive capacity of the
host, and the triggering of systemic defensive responses in the tree (Franceschi et al., 2005;
Eyles et al., 2010), while accumulation of susceptible trees on the landscape can promote
insect population buildup that leads to density-dependent attack behaviors more
threatening to healthy trees (Burke & Carroll, 2017).
Interaction between individual disturbances and their respective predisposing
factors has been a subject of recent study. Increased insect activity has been linked to
observed changes in local and global climate (Allen et al., 2010), though some have
suggested that competition between trees has thus far exerted a stronger influence on tree
growth and mortality than climate change (Zhang et al., 2015). Co-occurrence of several
predisposing factors can have a synergistic influence on forest insect populations (Powers
et al., 1999), while some forms of inter-tree competition have been shown to exacerbate
climatic stresses more than other forms of competition (Mölder & Leuschner, 2014).
2

Some of these trajectories from predisposing conditions to actual widespread
mortality can be interrupted by silvicultural treatments that disturb the stand in a way that
removes or dulls the effect of a predisposing factor (Smith et al., 1997). This has been
done by thinning to increase the growth rate and efficiency of individual trees, thereby
increasing their defensive capacity (Mitchell et al., 1983) and by removing overtopped
trees that are likely to attract pests to the stand in their decline (Williamson & Price, 1971).
Other studies of disturbance interactions reveal no preventative measures, but do suggest
assessment practices to accurately predict disturbance hazard and impact under specific
conditions (Hood et al., 2007; Day & Pérez, 2013).
Natural disturbances that kill only a fraction of the trees in a stand have been
described as thinning agents that confer definite benefits to surviving trees and the overall
stand, such as Weaver’s “Fire—Nature’s Thinning Agent in Ponderosa Pine Stands”
(1947). Weaver’s (1947) argument emphasized a point that had been hinted at before, and
has been rigorously tested since: that regular fire in the dry forests of western North
America is necessary not only for stand development and the perpetuation of ecosystem
services that residents have come to rely on, but also as a preventative measure against
catastrophic fire (Hardy & Arno, 1996; Agee & Skinner 2010). Without using the language
of modern disturbance ecology, Weaver distilled its key point: disturbances interact with
one another over the course of decades and centuries, and those interactions shape forest
ecosystems.
This principle is not limited to wildfire. Each disturbance consumes its own ‘fuel’
and reduces the immediate availability of that fuel in the stand and on the landscape (Raffa
3

et al., 2008). If the disturbance is excluded and its fuel preserved, the next event may
consume the accumulation in surprising ways. The effects of various disturbances on
predisposing conditions and ‘fuel loads’ for other disturbance agents over time are still
being actively investigated, but the net effects are already understood to be constraining
management options in some regions, especially western Canada (Dhar et al., 2013).
The Province of British Columbia (BC), Canada has a 25 million hectare timber
harvesting landbase, and the majority of that area has been affected by insect pests in the
21st century. Mountain pine beetle (Dendroctonus ponderosae), spruce beetle (D.
rufipennis), and western balsam bark beetle (Dryocetes confusus) have infested
approximately 18 million, 1 million and 2.7 million hectares of their respective host trees
and caused substantial reductions in projected allowable annual cuts in the coming decades
(BC MoF, 2012). The severity of these outbreaks has been attributed to warm temperatures
associated with climate change (Bentz et al., 2010) and overstocking of mature host stands
that has resulted from long-term fire suppression (Whitehead et al., 2001; Shore et al.,
2006).
Douglas-fir forests offer one of the remaining sources of accessible and
merchantable live trees in central BC, having weathered a decade of insect outbreaks with
relatively low mortality. These forests are often overstocked, making density-related
stagnation and competition stress a serious concern (Day, 1998a). Mature Douglas-fir
stands also provide critical mule deer winter habitat, and conserving that habitat is the
primary objective of management over several hundred thousand hectares of public land
(Armleder et al., 1994; Day, 1998b; Figure 1.1). Designated mule deer wintering ranges
are established to provide critical thermal and forage habitat, and in the study area are
4

legally protected under the Cariboo-Chilcotin Land Use Plan (CCLUP;
https://www.for.gov.bc.ca/tasb/slrp/plan129.html accessed 2017). Douglas-fir leading
stands cover 4.58 million hectares of the British Columbian landscape; this dissertation
considers 1 million hectares near the northern portion of that range, and 373 000 hectares
within that area are designated mule deer winter range.
The interior Douglas-fir ecosystem type under study (Steen & Coupé, 1997) is
bounded by a moisture limitation at the lower elevation dry bunchgrass zone and a
temperature limitation at the higher elevation sub-boreal pine-spruce zone. Stands near the
drier margin will be at higher risk in a warmer climate (Griesbauer & Green, 2010) and
many are expected to convert to bunchgrass over the course of the 21st century, while the
cooler margin is expected to advance to higher elevations and latitudes beginning as early
as 2025 (Hamann & Wang, 2006). These changes may jeopardize the mule deer winter
range strategy within its current boundaries, forcing an adaptive approach as the effects of
climate change are realized. Drought stress can kill trees directly and is a key predisposing
factor for Douglas-fir beetle, but a drought does not impact all stands or trees equally as
the stand structure and occupancy of the growing space strongly affect the
microenvironment that individual trees interact with.

5

Figure 1.1: Study area and distribution of Douglas-fir leading stands in British Columbia. Douglas-fir leading stands in designated
mule deer winter range (MDWR) highlighted in black within the study area. MDWRs outside of the study area are not shown.
6

Douglas-fir is commercially important throughout its range as raw material for log
homes, timber frame homes, glulam beams, utility poles, plywood, and other visual grade
and structural products. Standing volume is approximately 91 million cubic meters on 1
million hectares of Douglas-fir leading stands in the study area (BC VRI Rank 1 Polygon
layer, accessed 2015). This inventory of raw material is in principle worth $6.3 billion
(CAD) at a rate of $70/m3 (BC Interior Log Market Report, 2016) and substantially more
once manufactured. Salvage logging of lodgepole pine (Pinus contorta) stands killed by
mountain pine beetle has been the focus of most harvesting efforts in the study area over
the past decade, but attention is returning to Douglas-fir as the merchantability of dead
pine degrades over time (Lewis et al., 2006.). Harvesting of Douglas-fir within mule deer
winter ranges is permitted under area-specific guidelines, and is often required to prevent
stagnation and maintain the desired stand conditions in the absence of natural wildfire
(Dawson et al., 2007).
The major insect disturbance agents of Douglas-fir in the study area are Douglas-fir
beetle (Dendroctonus pseudotsugae; DFB) and western spruce budworm (Choristoneura
freemani = C. occidentalis; WSB), both native to British Columbia (Maclauchlan &
Buxton, 2016). Douglas-fir beetle has historically been found at the northern limits of its
host in scattered stands near Fort St. James in British Columbia and Jasper National Park
in Alberta (Paulson, 1995). Drought stress, windthrow, and wildfire are known
predisposing factors that have been involved in positive feedback loops with one another
(Johnson & Belluschi, 1969). Western spruce budworm was recorded in a small infestation
near Riske Creek in 1975 (Harris et al., 1985) but otherwise no observational records of
infestation in the study area exist prior to the outbreak that began in 2000. Adjacent areas
7

near Bella Coola to the west and Quesnel Lake to the east were defoliated by western
spruce budworm in 1970 and 1987/88 respectively (Harris et al., 1985; Forest Insect and
Disease Survey data accessed June 2016).
Tree-ring reconstructions of wildfire in the 18th and 19th centuries identified a
mixed-severity regime with an average return interval of 20–26 years depending on
location, with forests adjacent to grasslands burning more frequently (Daniels & Watson,
2003; Harvey, 2017). That regime was interrupted ca. 1920 by fire suppression efforts and
the proliferation of cattle grazing (which consumes fine grass fuels) on forested range,
though both lightning and human-initiated fires burned extensively in 1931. The Ministry
of Forests, Lands and Natural Resources Operations classifies the Douglas-fir type in this
region as an “ecosystem with frequent stand-maintaining fires” (record accessed January
2017: https://catalogue.data.gov.bc.ca/dataset/biogeoclimatic-attribute-catalogue) and
stands adjacent to low-elevation grassland areas are known to have burned more frequently
than higher elevation areas (Harvey, 2017). Exclusion of wildfire in the latter 20th century
facilitated a substantial increase in stand density, proliferation of an understory horizon of
Douglas-fir, and incursion of trees into grassland areas (Strang & Parminter, 1980; Daniels
& Watson, 2003; Wong & Iverson, 2004).
Provincial survey records from the 21st century (2000-2015) indicate that wildfire
and insect disturbances affecting Douglas-fir have been of similar magnitude in the study
area, though their spatial patterns have been distinct. Wildfire has affected 51 000 hectares
containing 4.3 million m3 of Douglas-fir inventory volume from 2000–2015
(https://www.for.gov.bc.ca/hts/vridata/accessed June 2016; Figure 1.2). Douglas-fir beetle
surveys (https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ accessed
8

June 2016) have recorded an estimated 3.2 million trees killed from 2000–2015,
amounting to approximately 6.2 million m3 loss assuming a merchantable volume of 1.88
m3 per tree (personal communication, Ken Day, University of British Columbia, 2015;
Figure 1.3). At a rate of $70/m3 these wildfires and beetle outbreaks have affected $301
million and $431 million worth of raw material respectively. Western spruce budworm has
defoliated approximately 650 000 hectares of Douglas-fir leading forests since 2000
(https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ accessed June
2016; Figure 1.4).
Mountain pine beetle has recently had a devastating effect on the lodgepole pine
that commonly grew in mixture with Douglas-fir in the study area (Figure 1.5). In three
years of fieldwork for this dissertation only a handful of surviving pine greater than 20cm
diameter were found, though standing and fallen snags killed in the 2000–2010 outbreak
were abundant in many stands (Figure 1.6). This outbreak is discussed thoroughly
elsewhere (Shore et al., 2006; de la Giroday et al., 2012; Hrinkevich, 2012), but bears
mentioning here as many Douglas-fir stands in the study area included a component of
lodgepole pine. Previous mountain pine beetle outbreaks have influenced the growth of
Douglas-fir in the study area, with more than half of Douglas-fir sampled in mixed stands
recording growth release following the previous outbreak in the 1970s (Hawkes et al.,
2004). Reconstructed mountain pine beetle outbreaks in the 1890s/early 1900s,
1930s/1940s, and 1970s are likely to have affected the growth of Douglas-fir throughout
the study area (Alfaro et al., 2004; Hawkes et al., 2004). The relatively short lifespan of
lodgepole pine precludes positive identification of outbreaks in the study area prior to
1890, but it is probable that undescribed infestations have influenced the growth of trees in
9

mixed stands.

10

Figure 1.2: Distribution of human and lightning initiated fires from 2000 to 2013,
highlighting affected stands of interior Douglas-fir
(https://catalogue.data.gov.bc.ca/dataset/fire-perimeters-historical accessed 2016)

11

Figure 1.3: Distribution of Douglas-fir beetle from 2000 to 2013 according to provincial
fixed-wing aerial overview surveys and more detailed regional helicopter surveys.
Polygons recorded as “trace” severity class (<1% mortality) excluded from the map to
reduce visual clutter. Helicopter survey data accessed 2015:
http://www2.gov.bc.ca/gov/content/environment/research-monitoringreporting/monitoring/aerial-overview-surveys/methods/standards-for-detailed-surveys
fixed-wing survey data accessed 2014:
https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/

12

Figure 1.4: Western spruce budworm defoliation identified in provincial aerial overview
surveys from 2000 to 2013. Survey data accessed 2014:
https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/

13

Figure 1.5: Distribution of mountain pine beetle in the study area from 2000 to 2013. Areas
displayed in red are not Douglas-fir leading stands. Shades of red indicate the maximum
severity recorded at a location in any single year, not the cumulative impact. Severe
records indicate >30% mortality, moderate 11–29%, and light 1–10%. Data accessed 2014:
https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/

14

Figure 1.6: Tree PYP02 (ca. 1641) in 2013 surrounded by fallen lodgepole pine and dense
pine regeneration, with younger Douglas-fir at the edge of the frame to left and right, and
in the background in front of dense sapling pine.
15

Mean annual temperatures recorded at weather stations in the study have risen
significantly in the 1950–2001 period, with temperatures at Quesnel Airport increasing at a
rate of 0.34°C per decade and those at Tatlayoko Lake increasing at 0.22°C per decade,
implying a 1.1–1.7°C increase (Dawson et al., 2008). Temperatures are projected to
continue rising in the 21st century, while no significant change in precipitation has been
observed and relatively minor changes are projected (Dawson et al., 2008). This increase
in temperature is expected to lead to more severe and widespread outbreaks of many pest
species, including those affecting Douglas-fir (Rudinsky & Vite, 1956; Woods et al., 2010;
Murdock et al., 2013). Increased temperature without a substantial increase in precipitation
is expected to lead to more frequent and severe drought stress, a predisposing factor for
many disturbance agents including Douglas-fir beetle. Understanding the baseline
relationships between natural disturbance agents and predisposing conditions under
historical climatic conditions will provide context necessary to describe change over time
and identify predisposing factors that may be monitored or managed.

16

1.2. Study objectives and rationale:
The objectives are 1) to delineate historical relationships between disturbance events
and subsequent drought tolerance in surviving trees, 2) identify those disturbances that are
attributable or partially attributable to Douglas-fir beetle, building a historical timeline of
outbreak activity directly comparable to those developed for wildfire history, 3) to develop
new methods of tree ring analysis that provide the greatest possible temporal resolution for
Douglas-fir beetle reconstructions, and 4) to expand and refine western spruce budworm
outbreak history reconstructions in the study area. The aim of this dissertation is to
describe disturbance histories and disturbance/drought interactions as they have been
historically, which may suggest ways of directly reducing the impact of projected climate
change on the defined objectives of mule deer winter range conservation and timber
production.

17

1.3. Organization of dissertation
This dissertation has an introduction, four data chapters, a concluding synthesis
chapter, and two appendices providing computer code and statistical outputs referenced in
the text. The first two data chapters are conceptually related, as each explores changes in
drought tolerance apparent after disturbance: the first is based on known anthropogenic
disturbances while the second considers natural disturbances inferred from the growth
response of surviving trees over the past three centuries. The last two data chapters
contribute to the development of timelines of insect outbreaks over the past three to four
centuries. The third data chapter builds a chronology of Douglas-fir beetle outbreaks based
on scars left by failed attacks, while the fourth critically explores the reconstruction of
western spruce budworm outbreaks based on analysis of growth rates of tree species that
are and are not hosts of this common forest pest.
1.3.1. A note on maps:
All maps of the study area are shown with the grassland benchmark area
(https://catalogue.data.gov.bc.ca/dataset/grassland-benchmark-for-the-cariboo-region
accessed January 2017) obscuring stands which are classified in the Vegetation Resources
Inventory as Douglas-fir leading, but in actuality are sparsely treed and best defined by
their herbaceous component, especially in a historical context. All GIS data are used under
the BC Open Government License v2.0; all maps were composed entirely by the author.

18

2. Adjacency to a harvest corridor in partially harvested stands increases drought
resistance of interior Douglas-fir (Pseudotsuga menziesii var. glauca (Beissn.)
Franco) near the grassland interface in central British Columbia
2.1. Abstract:
Drought tolerance of trees may be modified by competition, but most studies
quantifying the relationship do not consider the effect of stem clustering. Interior Douglasfir forests near the grassland interface in central British Columbia are managed under a
clumpy retention system to facilitate the conservation of mule deer winter habitat through
timber harvesting. Climate change projections indicate continued increases in temperature,
potentially stressing trees and jeopardizing winter range conservation. Trees incidentally
placed in different states of competition by mechanical harvesting in the 1970s are sampled
to provide a 40-year comparison of clustered and open growing trees. Tree-ring analysis is
used to assess the reduction in growth rate during drought years and resumption in
subsequent years. Following harvest, a clear separation of growth patterns is evident
between trees that were fully released, partially released, and not directly released.
Partially released trees show intermediate growth but no difference is found in the longterm climate/growth relationship compared to fully released trees. Trees growing within
the matrix between skid trails have stronger correlations with several monthly climatic
variables and greater growth reduction during drought events. These results suggest
silvicultural practices recommended for the winter ranges are likely to provide concurrent
benefits in drought tolerance.

Chapter 2 has been submitted for publication with the following authorship: Thompson N.T, Lewis K.J., Poirier L.M.

19

2.2. Introduction
Climate change projections for central British Columbia indicate a rise in
temperature over the coming century without a commensurate increase in precipitation,
leading to more frequent and severe drought stress in forest stands (Dawson et al., 2008).
Mortality resulting directly from moisture deficit is expected to increase under such
climate change (Allen et al., 2010), while disturbances from insect pests and pathogens are
expected to become more frequent (Currano et al., 2008; Woods et al., 2010). Stand
thinning has been recommended as a means of reducing these risks (Mitchell et al., 1983;
Sohn et al., 2013) while maintaining productivity in the short- to mid-term (Elkin et al.,
2015; Calev et al., 2016). Thinning reduces canopy interception of rainfall and inter-tree
competition for soil moisture, resulting in greater moisture availability to residual trees
(Stogsdili et al., 1992; Bréda et al., 1995). Trees in thinned stands have maintained higher
transpiration rates during droughts compared to controls (Lagergren et al., 2008) and have
shown less growth reduction in drought years (Laurent et al., 2003).
The spatial distribution of trees following thinning may be uniform, random, or
clustered depending on the objectives, stand conditions, and harvesting equipment used.
Uniform spacing has been recommended for centuries as a means of maximizing yield, but
a random pattern often emerges as compromises are made to retain the best growing stock
(Evelyn, 1662). Clumpy distributions are common in naturally established forests
(Hamilton, 1984; Harrod et al., 1999) and this pattern is sometimes encouraged to support
specific wildlife habitats (Long & Smith, 2000; Dawson et al., 2007). Clumpy distributions
also result from row thinning in plantations (Makinen et al., 2005) and the practice of
harvesting trees from corridors at more or less equal intervals, referred to as systematic
20

thinning by Makinen (2005) in Finland and partial harvesting by Fuller et al. (2004) in
Maine.
Experiments comparing clustered tree distributions with uniform or random spatial
patterns have found lower growth rates in clumped vs. regularly spaced red pine (Pinus
resinosa; Stiell, 1982), slash and loblolly pine (Pinus elliotii, Pinus taeda; Baldwin et al.,
1989), and Norway spruce (Picea abies; Makinen, 2005) after treatment. While uniform or
random spacing may be implied by the absence of modifiers such as row thinning or
systematic thinning, authors discussing the forest health benefits of thinning often do not
specify whether the post-treatment stand had a uniform, random, or clustered arrangement
of trees (Williamson & Price, 1971; Mitchell et al., 1983; Giuggiola et al., 2013; Gebhardt
et al., 2014). Similar oversights are found in tree-ring evaluations of the climate/growth
relationship in stands thinned to various residual densities (Kohler et al., 2010; Sohn et al.,
2013). This gap in the literature is problematic as clumped distributions are common in
regions where row or systematic thinning is employed and in areas where clumped
distributions are encouraged to meet wildlife habitat objectives.
Conservation of mule deer (Odocoileus hemionus (Rafinesque)) winter range is
legally mandated on select public lands in the interior Douglas-fir (Pseudotsuga menziesii
var. glauca (Beissn.) Franco) forests of British Columbia (Dawson et al., 2007). Mule deer
in this region migrate to relatively low elevation areas and preferentially utilize mature
stands of Douglas-fir with moderate crown closure, as interlocking crowns intercept snow
and provide browse through crown abrasion (Armleder et al., 1986). A system of
harvesting with clumpy retention in these forest types has been under development since
1983 to conserve such winter range (Armleder, 1999; Day et al., 2003). These designated
21

ranges cover 373 000 hectares and include some of the driest forested areas in the region,
receiving a mean annual precipitation of 435mm per year from 1961–1990
(http://www.climatewna.com/ accessed 2016).
Studies from other regions suggest that thinning may mitigate the increased drought
stress introduced by climate change (Elkin et al., 2015; Calev et al., 2016), but there is no
information available on the net effect of clustered distributions on drought response in a
dry environment. The percentage of gross rainfall lost to interception has been found to
progressively decline with increased thinning intensity in established plantations in mesic
and xeric systems (Teklehaimanot et al., 1991; Molina & del Campo, 2012). Strip thinning
of a Japanese cypress plantation (1 265mm annual precipitation) reduced total interception
from 28.7% to 20.8% of gross precipitation (Sun et al., 2015). These studies are based on
relatively young plantations managed for wood production. The Douglas-fir forests under
study here are typically older, were not originally established on a regular spacing, and are
now managed to facilitate canopy interception of snow within clumps.
Long-term (>30 years) silvicultural trials are not available to assess the drought
tolerance of clustered Douglas-fir in the study area, but mechanical harvesting in the latter
half of the 20th century provides regular spatial patterns that are in some cases analogous to
those recommended for mule deer conservation. Presence of these conditions in the same
stand provides an opportunity to compare trees growing in different competitive situations,
addressing the following research questions:

22

1. What was the average postharvest growth response among trees in three
competition classes: open growing, skid trail edge, and matrix interior?
2. Which monthly temperature and precipitation variables have long-term
correlations to radial growth and are there differences in climate
correlations among these three competition classes?
3. How do drought resistance and resilience (Lloret et al., 2011) vary among
these three classes?
4. How do live crown ratio and sapwood basal area vary among the
competition classes four decades after harvest?
The first question follows from the existing literature on clumpy spatial
arrangements, which suggests a reduction in growth and yield at the tree and stand level,
though only individual tree metrics are applicable in this study (Stiell, 1982). The second
looks for differences in the long-term climate/growth relationship among classes, while the
third considers the response to short-term drought events. The final question addresses
indicators of crown vigor, also following from literature describing reduced canopies in
clumped trees (Stiell, 1982).

23

2.3. Methods
2.3.1. Site Selection and Sample Collection
Candidate sites were first screened using the British Columbia Vegetation Resources
Inventory dataset (VRI; https://www.for.gov.bc.ca/hts/vridata/, accessed June 2015) to
identify stands between 80 and 130 years old with intermediate crown closure (30–70%).
Historical wildfire perimeters were overlaid to ensure that potential sites were not affected
by recent fires, either natural or human-caused (records extending to 1917 in the study
area; https://catalogue.data.gov.bc.ca/dataset/fire-perimeters-historical). Inventory
polygons meeting these criteria were exported from ArcMap® to Google Earth®, where
aerial imagery (dated 2005) was examined to confirm the presence of regularly spaced skid
trails (Figure 2.1). Candidate sites were assessed in the field to determine whether there
were sufficient numbers of trees in all three competition classes (Figure 2.2): Open (fully
released from crown competition during the harvest), Edge (on the edge of skid trails) and
Interior (the unharvested matrix between skid trails).
Most remotely identified sites had two of the classes in abundance but lacked the
third; three sites contained all classes and were deemed appropriate for sampling (Figure
2.3). Both the Colpitt and Sting Lake sites showed visually apparent growth releases in
1976 consistent with a 1975 harvest, while Vert Lake was released in 1980, suggesting a
harvest in 1979. These dates are corroborated by harvest records in the VRI database for
Colpitt and Sting Lake sites, but there are no records for the Vert Lake site, likely due to
incomplete digitization of datasets. Patterns of cut stumps at all sites suggest that large
sawtimber was targeted for removal and the smaller stems constituting the population now
under study were only cut if necessary for removal of the primary product (as in
24

Maclauchlan and Brooks, 2009).
Vegetation resources inventory (VRI) data were clipped to the designated mule deer
winter range boundaries (http://www.env.gov.bc.ca/wld/frpa/uwr/approved_uwr.html
accessed 2015) and total merchantable volumes were calculated by age class and crown
closure class to describe the stands surrounding the study sites. VRI age estimates were
consolidated to six 40-year age classes and one class including all trees greater than 240
years. Percent crown closure values were reduced to five equal classes. Volume estimates
were the sum of the inventory volume that was calculated with a merchantability threshold
of 12.5cm top diameter.

25

Figure 2.1: Aerial image of Vert Lake site showing the regular pattern of harvesting trails radiating from clearings where wood was
collected and processed.

26

Figure 2.2: Conceptual schematic of interior, edge and open growing trees following
harvest (trees not drawn to scale). Labeled trees show the relative position of the three
competition classes within the stand.

27

Figure 2.3: Location of study sites. Douglas-fir dominated stands are those where Douglas-fir is classified as the leading species in
the Vegetation Resources Inventory; the grassland benchmark describes historical grassland areas that may now be sparsely treed.
28

Subject trees were selected to represent three levels of post-harvest competition:
between harvest trails, with no potential competitors removed (“Interior”), on the edge of
harvest trails with potential competitors removed on only one side (“Edge”) and within the
harvested area with all potential competitors removed (“Open”; Figure 2.2). Only
dominant and codominant trees were considered as subject trees; no overtopped trees were
sampled and saplings (less than ½ the height of the subject trees) were not considered to be
competitors. Subject trees were selected systematically along transect lines established on
a 50m grid with a random azimuth as a starting point. Subject trees were selected as the
closest trees meeting the above criteria every 20m, enforcing a minimum 20m distance
between two trees of the same competition class. Twenty or more subject trees from each
competition class were selected at each of the three sites, and two increment cores were
removed from opposite sides of each tree at breast height (1.3m). A variable-radius forest
inventory plot was centered on each subject tree, using an 8m2/ha glass prism to determine
plot level basal area.

29

Figure 2.4: Typical examples (left to right) of Open, Edge and Interior class trees indicated
by white triangles.
2.3.2. Laboratory Preparation
Increment cores were dried, glued to wooden mounts, and processed according to
standard methods, progressing from 120 to 600 grit sanding belts (Stokes & Smiley, 1968).
The sapwood boundary was delineated based on visual assessment of color change, which
is readily apparent in Douglas-fir (Figure 2.5). Sapwood width, heartwood width, and
distance from last measurable ring to estimated pith location were measured to the nearest
millimeter using calipers and a ruler. Sapwood basal area was calculated by subtracting the
area of the heartwood from the total basal area of the tree inside bark on each core, and a
mean of the two measurements was produced for each tree.
30

Ring widths and earlywood/latewood boundary locations were measured in
Windendro® 2012. Crossdating of series was accomplished by scanning in batches of 20 or
more cores and visually comparing the graphed ring width measurements in Windendro®,
beginning with the known date of the final ring and utilizing regionally established marker
years 2010, 1988/89/90 and 1959 to confirm later dates. When ring width was less than
0.3mm, both ring width and latewood width values were checked using an Olympus® SZ61
stereo microscope and Velmex® measuring slide. Correct assignment of calendar years to
ring width measurements was confirmed using the dendrochronological program
COFECHA (Holmes, 1983). Ring widths were converted to basal area increments using
the package dplR (version 1.6.4) in the open source statistical program R (version 3.2.2.).
2.3.3. Statistical Analysis
Growth response to thinning was quantified on the basis of 10-year increments to
provide a dataset statistically analogous to remeasurements in a standard thinning study.
Basal area (inside bark) was calculated for the years 1955, 1965, 1975, 1985, 1995, 2005,
and 2015, while mean annual basal area increment (BAI) was calculated on the basis of
10-year means in the periods 1956–1965, 1966–1975, 1976–1985, 1986–1995, 1996–2005,
and 2006–2015. Two time steps in the 1950s and 1960s were included to test the
assumption that the population had similar growth rate prior to the modification of the
competitive environment in the 1970s, and the 1975 basal area was used to check that trees
were initially of a similar size. Oneway ANOVA with a Bonferroni post-hoc test (α=0.05)
was used to assess the significance of differences in 10-year basal area and basal area
increment among competition classes at each site. Data were tested for normality in Stata
12.1® and transformed as noted to meet the assumption of normal distribution.
31

Drought events were defined as years in which the total precipitation from the
beginning of the previous June to the end of the current June was in the 30th percentile
(1936-2012, (http://www.ec.gc.ca/dccha-ahccd/ accessed 2017) and where study-wide
mean ring width index was in the 10th percentile (1936-2012). This index was a biweight
mean chronology generated using a 10-year smoothing spline on raw ring widths to
minimize the influence of stand dynamics on annual variability, and inter-annual
autocorrelation was reduced by adding the residuals from an autoregressive model to each
series prior to averaging to create the mean index (Bunn et al., 2017). Three years meeting
these criteria were identified after the harvest (1988, 1995, and 2010). Precipitation was
above the 30th percentile in 1989, but the ring width index was the lowest in the series, so
that year was averaged with 1988 for analysis (following Lloret et al., 2011). Drought
resistance was quantified as the ratio of the basal area increment in the event year (mean of
two years in 1988/1989) to the mean basal area increment in the 5 years preceding the
event (Figure 2.5; Lloret et al., 2011). Drought resilience was quantified as the ratio of the
5-year mean basal area increment after the event to the 5-year mean basal area increment
before. An additional metric particular to this study was the drought resistance relative to
pre-harvest growth (hereafter “baseline resistance”): the ratio of the basal area increment in
the event year(s) to the 5-year mean basal area increment in the period 1970–1974, prior to
harvesting in all stands. Oneway ANOVA (Bonferroni post-hoc test, α=0.05) was used to
assess the significance of differences in drought resistance and resilience, with variables
transformed as noted to meet the assumption of normal distribution in the population under
comparison.

32




)LJXUH([DPSOHVRIFRUHVIURP2SHQ &2/$ DQG,QWHULRU &2/$ FODVVHVDW
&ROSLWW/DNHVLWHKDUYHVWHG EODFNDUURZ 'URXJKW\HDUVPDUNHGE\EODFN
UHFWDQJOHVODEHOHG³'´DERYHHDFKFRUHDQGPDJQLILHGIRUHDFKVDPSOHDWORZHUULJKW
'URXJKW\HDUVDQGPDUNHGE\EODFNUHFWDQJOHVEHQHDWKHDFKFRUH5HVLVWDQFH
HTXDOV'<UHVLOLHQFHHTXDOV=<DQGEDVHOLQHUHVLVWDQFHHTXDOV';ZKHUH;<DQG=DUH
DYHUDJHEDVDODUHDLQFUHPHQWRYHUWKH\HDUSHULRGVLQGLFDWHGDQG'LVWKHDYHUDJHEDVDO
DUHDLQFUHPHQWRIDQG+ROORZWULDQJOHVLQGLFDWHWKHFRORUFKDQJHDWWKH
KHDUWZRRGVDSZRRGERXQGDU\3HQFLOPDUNVLQGLFDWHGHFDGH\HDUVIURPWKHOHIW
 DOVRLQPDJQLILFDWLRQ DQG
7RWDOULQJZLGWKHDUO\ZRRGZLGWKDQGODWHZRRGZLGWKLQGLFHVZHUHFDOFXODWHG
XVLQJD\HDUVPRRWKLQJVSOLQHWRUHFRYHUDQQXDOJURZWKYDULDWLRQZLWKPLQLPDO
LQWHUIHUHQFHIURPWKHJURZWKUHVSRQVHWRKDUYHVWLQJ3HDUVRQFRUUHODWLRQFRHIILFLHQWV
EHWZHHQWKHVHLQGLFHVDQGWKHPRQWKO\FOLPDWLFYDULDEOHVPHDVXUHGDWWKH:LOOLDPV/DNH
$LUSRUW SUHYLRXV-XQHWRFXUUHQW6HSWHPEHUKWWSZZZHFJFFDGFFKDDKFFG ZHUH
FDOFXODWHGIRUWKH±SHULRGXVLQJWKHSDFNDJHWUHHFOLP YHUVLRQ  =DQJ 
%LRQGL LQWKHRSHQVRXUFHVWDWLVWLFDOSURJUDP5 YHUVLRQVFULSWLQ$SSHQGL[
$ (DFKULQJZLGWKFRPSRQHQWZDVGHWUHQGHGDQGDQDO\]HGVHSDUDWHO\DVWKHVH
FRPSRQHQWVRIJURZWKKDYHSUHYLRXVO\EHHQIRXQGWRUHVSRQGWRGLVWLQFWFOLPDWLFYDULDEOHV
/HERXUJHRLV:DWVRQ /XFNPDQ 
/LYHFURZQUDWLRVRIVDPSOHWUHHVZHUHFDOFXODWHGRQWKHEDVLVRIWKHORZHVWOLYH
EUDQFK QRQHSLFRUPLF DQGWKHORZHVWOLYHZKRUORIWKUHHRUPRUHEUDQFKHVE\WKHIRUPXOD
7RWDO+HLJKW±/RZHVW%UDQFK+HLJKW 7RWDO+HLJKW 





2.4. Results
2.4.1. Drought conditions in event years
Average precipitation from the beginning of the previous June to the end of the
current June (pJJ) was 548mm at the Williams Lake Airport from 1937 to 2012. 1988 had
the lowest pJJ on record, measuring 394mm. 1989, 1995, and 2010 measured 510mm,
485mm, and 462mm respectively. Pairwise correlation between pJJ and the mean ring
width index among all trees at all sites from 1937 to 2012 was significant (n=76, r = 0.58,
p < 0.0001).
2.4.2. Stand Conditions and Tree Characteristics
Differences in height to lowest live branch and lowest live whorl were never
significantly different among the Open and Edge classes at any site, and ranged from 7.5–
9.7 and 10.4–12.5m respectively among all sites (Table 2.1). Interior trees had significantly
higher crown bases by both single-branch and three-branch whorl metrics at all sites. Mean
height of residual trees in 2015 varied by as much as 1.9m among competition classes at
individual sites, and differences were statistically significant at two of the three sites. The
1975 diameter inside bark calculated from the tree rings was not significantly different
among the classes at the Colpitt and Sting Lake sites, but a significant difference between
the Open and Edge trees was identified at Vert Lake.

34

Table 2.1: Mean 2015 stand characteristics measured in the field and 1975 diameter inside
bark calculated from tree rings. BA/ha value includes the basal area of subject tree
(accounting for 8m2/ha). LLB = lowest live branch (single), LLW = lowest live whorl of
branches (3 or more), DBH = diameter at breast height (1.3m measured on the high side of
the tree), BA/ha = basal area per hectare measured with 8m2/ha prism. DIB = diameter
inside bark. Classes that do not share a letter with others at the same site are significantly
different (Bonferroni post-hoc test, α=0.05); significant differences are in bold.
Height
(m)
Open
23.6 ns
Colpitt Edge
23.6 ns
Interior 23.1 ns
Open
21.9 a
Sting
Edge
21.3 ab
Interior 20.1 b
Open
22.0 a
Vert
Edge
20.2 b
Interior 20.1 ab

LLB
(m)
8.7 a
9.7 a
14.1 b
7.5 a
8.7 a
12.4 b
8.5 a
9.4 a
11.5 b

LLW
(m)
10.8 a
12.5 a
16.4 b
10.4 a
11.7 a
14.6 b
10.5 a
11.3 a
13.4 b

35

DBH
(cm)
39.6 a
34.9 b
29.6 c
40.4 a
34.5 b
27.2 c
37.0 a
31.5 b
28.8 b

BA/ha
(m2)
17.6 A
29.2 B
56.4 C
16 A
33.2 B
48.8 C
17.2 A
28.7 B
46.4 C

1975
DIB
(cm)
13.4 ns
13.5 ns
15.4 ns
16.5 ns
15.3 ns
15.2 ns
17.5 a
14 b
16.4 ab

Age
(2015)
118 ns
121 ns
121 ns
130 a
124 b
152 ab
123 ns
118 ns
122 ns

Annual BAI since 1950 is summarized among all sites by competition class; an
increase in growth rate is evident ca. 1975 in both Edge and Open classes (Figure 2.6).

Basal Area Increment (cm2)

25

20

Interior
Edge
Open

15

10

5

0
1950

1960

1970

1980

1990

2000

2010

Figure 2.6: Mean annual basal area increment plotted by competition class among all sites
since 1950. Vertical dashed line indicates the first recorded defoliation by western spruce
budworm in 2001. Black bars above the X axis indicate the drought years under study
(1988/89, 1995, and 2010). Two sites (Colpitt and Sting Lakes) were harvested in 1975,
one (Vert Lake) harvested in 1979.

36

Live crown ratios were significantly higher in the Open and Edge classes compared
to the Interior class in all cases. Values in the Open class were consistently higher than the
Edge class and these differences were significant in 50% of the comparisons (Table 2.2).
Sapwood basal area followed a clear high-low gradient from Open to Edge to Interior, with
significant differences among all classes at all sites.
Table 2.2: Live crown ratios on the basis of lowest live branch (LCR(B)), lowest live whorl
of three or more branches (LCR(W)), and sapwood basal area (m2) among three
competition classes at three sites (Bonferroni post-hoc test, α=0.05); significant differences
are in bold.

Colpitt

Sting

Vert

Open
Edge
Interior
Open
Edge
Interior
Open
Edge
Interior

LCR(B)
0.63 a
0.59 a
0.39 b
0.65 a
0.59 a
0.38 b
0.61 a
0.54 b
0.43 c

LCR(W)
0.54 a
0.48 a
0.29 b
0.53 a
0.45 b
0.27 c
0.52 a
0.44 b
0.33 c

37

Sapwood
BA (cm2)
319 a
247 b
129 c
290 a
221 b
105 c
293 a
226 b
121 c

Of the 373 000 hectares identified for conservation as mule deer winter range in the
Cariboo-Chilcotin region, 309 000 are listed as Douglas-fir leading stands in the BC VRI
Rank 1 polygon layer (accessed 2015). Total inventory volume in this area is
approximately 33 million m3, with the majority of both volume and area in mature stands
(>80 years old) with intermediate crown closure (41–60%; Figure 2.7). Allowing for a
difference of 20 years between age at coring height and true age (Wong & Lertzman,
2001), all three sites currently fall in the age class (121–160 years) that contains the
highest volume on the landscape while occupying the second greatest area. At the time of
harvest, the stands would have been in the age class (81–120 years) that presently contains
the second highest volume while occupying the greatest area in the winter ranges.

Volme (million m3)

12
10
8
6

Crown Closure (%)

80+
61-80
41-60
21-40
1-20

4
2
0

1-40

7 417

41-80

54 239

81-120
90 644

121-160
78 747

161-200
36 847

201-240
25 036

240+

16 194

Projected Age (2015)
Area in Age Class (ha)
Figure 2.7: Live volume (m3 based on 12.5cm utilization) in areas designated for mule deer
winter range conservation by crown closure (%) in 40-year age classes. Total area
(hectares) in each age class is noted beneath X axis.

38

2.4.3. Postharvest Growth
Differences in mean basal area and mean annual BAI among classes in the two 10year time steps prior to the harvest were not statistically significant at Colpitt and Sting
Lakes (Figure 2.8).Vert Lake showed a significant difference in BA and BAI prior to the
harvest between the Open and Edge categories only, with the Open class being the larger
and faster growing (Figure 2.8).
Mean annual basal area increment in the first postharvest period (1976–1985) was
significantly higher in the Open class at all sites, with the Edge trees intermediate between
the Open and Interior classes (Figure 2.8). Differences in BAI between the Edge and
Interior classes were significant at Colpitt and Sting Lakes in the first period after harvest,
but not Vert Lake. All classes were significantly different from one another in the second
and third decades; by the final period differences in BAI among the Edge and Open classes
were statistically insignificant at all sites.
Mean basal area was significantly different among all classes at the Sting Lake site
by 1995 and Colpitt Lake by 2005 (Figure 2.8). Edge trees at Vert Lake did not achieve a
significantly higher basal area than the Interior trees by the end of the study period, though
the Open class had a significantly higher basal area since 1965, a decade prior to the
harvest, and maintained that position throughout the postharvest period.

39

Figure 2.8: Basal area and mean annual basal area increment at each site (cm2). Raw data
are displayed below; differences among the classes are significant for a given period if
values do not share a letter (Bonferroni post-hoc test, α=0.05).
40

2.4.4. Climate Correlations
Several significant correlations were found between detrended total
ring/latewood/earlywood widths and monthly weather record (Table 2.3). Total
precipitation in the November prior to the growing season had a positive correlation with
both earlywood and latewood width among most of the classes and sites tested, though the
correlation values were lower for the total ring width series. Previous-December
precipitation was correlated with earlywood and latewood width in the Interior competition
class at all sites, and also with the latewood width in the Open class at Sting Lake.
Precipitation in the current April had a negative correlation with latewood width in most
classes at all sites. June precipitation and temperature respectively had strong positive and
negative correlations with latewood width in nearly all classes at all sites, while
temperature in July was often negatively correlated with latewood width. Previous-summer
temperature and precipitation variables were inconsistent between sites; the precipitation
of the previous August was the most common variable with a statistically significant
correlation in this analysis while the mean temperature of the previous July had a
significant correlation with earlywood width in all classes at Sting Lake and the Edge class
at Vert Lake. Correlation values between latewood width and June temperature and
precipitation were consistently higher in the Interior class than the Open class, while the
Edge was usually intermediate between the two.

41

Edge Interior
Edge Interior Open
Edge Interior Open
Edge Interior Open
Edge Interior Open
Edge Interior Open
Open

Sting Lake
Sting Lake
Vert Lake

Average Monthly Temperature (°C)

Colpitt Lake

Vert Lake

Total Monthly Precipitation (mm)

Colpitt Lake

Table 2.3: Pearson correlation coefficients between monthly mean temperature and total
precipitation and total ring, earlywood, and latewood widths each detrended with a 10-year
cubic smoothing spline. Only significant (p < 0.05) correlation values are shown. Lower
and uppercase headings indicate months in previous and current year respectively.
jun jul aug sep oct nov dec JAN FEB MAR APR
0.25
0.20 0.25 0.23
Total
0.34
0.50 0.38
-0.27
Early
0.21
0.41 0.35
-0.42
Late
0.25
Total
0.30
Early 0.36
-0.44
Late
Total
Early
Late
0.32
0.29 0.22
-0.25
Total 0.14
0.47
0.51 0.36
Early
0.30
0.41 0.31
0.20 -0.38
Late
0.25 0.23
Total 0.15 0.18 0.35
0.47
0.45
Early
0.41
-0.41
Late
0.23
Total 0.16 0.22 0.34
0.44
0.40
Early 0.35
0.39 0.42
Late
0.29
0.29 0.25
-0.26
Total
0.57 0.37
Early
0.44 0.39
-0.37
Late
0.31
0.26
0.20
-0.26
Total 0.19
0.34
0.40
Early 0.45
0.42
0.23
-0.41
Late
0.30
0.27 0.22
-0.25
Total 0.16
0.41
Early 0.39
0.45
-0.38
Late
Total
-0.29
0.20
Early
-0.25
Late
Total
0.31
0.27
Early
Late
Total
Early
Late
-0.20
Total
-0.26
Early
0.23
Late
-0.18
Total
-0.29
Early
Late
-0.20
Total
-0.34
Early
Late
Total
Early
Late
-0.21
Total
-0.27 -0.31
Early
Late
Total
Early
Late

42

MAY JUN JUL AUG SEP
0.28
0.19
0.46
0.28

-0.25
-0.43
-0.42
-0.28
-0.51
-0.46

0.49
0.23
0.30
0.31

-0.50
-0.37

0.58
0.29

-0.51
-0.37

0.53
0.34
0.49
0.36
0.27
0.59
0.34
0.25
0.54
0.31

-0.16
-0.28

-0.33

0.24

-0.48
-0.31
-0.39
0.23

0.19

0.46
-0.26 -0.22
-0.32
-0.32 -0.27
-0.22
-0.27 -0.25

-0.39
-0.39
-0.43
-0.38
0.34

-0.21

0.46

-0.24
-0.29
-0.28
-0.40
-0.24
-0.34
-0.21
-0.33
-0.32
-0.31
-0.33
-0.42
-0.29
-0.29
-0.39
-0.28
-0.34

0.37

-0.30

-0.32
-0.23
-0.34
-0.24
-0.27
-0.39
-0.22

-0.36 -0.38

0.46

0.43

-0.19
-0.23
-0.18

-0.19

-0.18

0.34

2.4.5. Drought Resistance and Resilience
Drought resistance values were similar between the Open and Edge classes at all
sites in all years. In all cases where statistically significant differences were found, the
Interior class underperformed either the Open or the Edge class, and in three cases (all
years at Colpitt Lake) drought resistance in the Interior class was significantly lower than
both other classes.
Table 2.4: Drought resistance values (drought/5-year pre-drought mean) among three
competition classes at three sites for the years 1988/89, 1995, and 2010 (Bonferroni posthoc test, α=0.05).

Drought Resistance
Open
Colpitt Edge
Interior
Transformation
df
F
Prob > F
Open
Sting Edge
Interior
Transformation
df
F
Prob > F
Open
Vert Edge
Interior
Transformation
df
F
Prob > F

1988/89
0.56 a
0.52 a
0.34 b
Log
2, 57
13.95
<0.01
0.46 a
0.39 ab
0.32 b
Log
2, 57
3.55
<0.01
0.60 a
0.53 ab
0.40 b
1/sqrt
2, 59
4.25
0.01

43

1995
0.64 a
0.72 a
0.52 b
Log
2, 56
8.90
<0.01
0.35 ns
0.34 ns
0.34 ns
Sqrt
2, 57
0.21
0.81
0.50 ns
0.54 ns
0.46 ns
Inverse
2, 59
1.48
0.24

2010
0.94 a
0.88 a
0.71 b
Log
2, 57
9.42
<0.01
0.39 ns
0.38 ns
0.33 ns
None
2, 57
1.98
0.15
0.45 ab
0.47 a
0.40 b
Sqrt
2, 59
4.18
0.02

Only two statistically significant differences were found among the drought
resilience values, and the highest value did not belong to the Open class in either case
(Table 2.5). In one case the Open class significantly underperformed both other classes,
and in the other the Open class was not significantly different from the interior. Both
significant comparisons occurred in the 1988/89 period, when all values exceeded 1.0
indicating a positive resilience (increasing growth trend) among all classes.

Table 2.5: Drought resilience values (5-year post-drought mean/5-year pre-drought mean)
among three competition classes at three sites for the years 1988/89, 1995, and 2010
(Bonferroni post-hoc test, α=0.05).

Drought Resilience
Open
Colpitt Edge
Interior
Transformation
df
F
Prob > F
Open
Sting Edge
Interior
Transformation
df
F
Prob > F
Open
Vert Edge
Interior
Transformation
df
F
Prob > F

1988/89
1.31 ns
1.37 ns
1.21 ns
Log
2, 57
0.90
0.41
1.16 a
1.31 b
1.54 b
Log
2, 57
4.12
0.02
1.22 a
1.44 b
1.29 ab
Log
2, 59
3.68
0.03

44

1995
1.00 ns
1.04 ns
0.90 ns
1/sqrt
2, 56
2.68
0.08
0.70 ns
0.74 ns
0.77 ns
1/sqrt
2, 57
1.00
0.37
0.83 ns
0.84 ns
0.73 ns
Inverse
2, 59
2.14
0.13

2010
1.17 ns
1.12 ns
1.09 ns
1/sqrt
2, 57
0.81
0.45
0.91 ns
0.98 ns
1.01 ns
1/sqrt
2, 57
0.17
0.85
0.86 ns
1.00 ns
0.90 ns
Log
2, 59
2.81
0.07

Significant differences in baseline resistance were identified between one or more
competition classes at all sites in all the years considered (

45

Table 2.6). Values in the Interior class were significantly lower than the other two
classes in all cases. Baseline resistance was significantly higher in the Open class than the
Edge class in two cases; differences between these classes were insignificant in the
remaining seven tests. Baseline resistance values exceeded 1.0 (indicating higher growth in
the drought year(s) compared to the pre-harvest average) in eight of nine cases for both
Open and Edge classes, while the Interior class exeeded 1.0 in one out of nine cases.

46

Table 2.6: Baseline drought resistance values (drought/1971–1975 mean) among three
competition classes at three sites for the years 1988/89, 1995, and 2010 (Bonferroni posthoc test, α=0.05).

Baseline Drought Resistance
1988/89
Open
4.37 a
Colpitt Edge
2.64 a
Interior
0.63 b
Transformation 1/sqrt
df 2, 57
F 31.39
Prob > F <0.01
Open
5.00 a
Sting Edge
1.95 b
Interior
0.54 c
Transformation
Log
df 2, 57
F 26.23
Prob > F <0.01
Open
2.49 a
Vert Edge
2.08 a
Interior
0.66 b
Transformation
Log
df 2, 59
F 17.34
Prob > F <0.01

47

1995
6.46 a
4.26 a
1.10 b
Log
2, 56
43.81
<0.01
3.91 a
1.96 b
0.78 c
Log
2, 57
20.85
<0.01
2.19 a
3.08 a
0.95 b
Log
2, 59
15.96
<0.01

2010
5.29 a
3.41 a
0.94 b
Log
2, 57
24.85
<0.01
1.74 a
0.93 a
0.39 b
Log
2, 57
13.44
<0.01
0.96 a
1.18 a
0.41 b
Log
2, 59
14.00
<0.01

2.5. Discussion
Trees growing in the open since the 1970s had higher growth, greater sapwood
basal area, and larger crowns, but the Open class had no clear advantage over the Edge
class in terms of drought resistance or resilience. Interior trees had poorer drought
resistance, as well as the lowest growth, least sapwood basal area, and shortest crowns.
Earlywood and latewood widths in the Interior classes showed a correlation with
precipitation in the month of December that was less pronounced in the other classes, and
correlation values with current-June precipitation were generally higher in the Interior
class. These findings suggest a lower sensitivity to drought stress in both the Open and
Edge classes following harvest, with few distinctions between the two. Negative
correlations between both earlywood and latewood widths and temperatures in June/July
suggest that the rise in summer temperatures indicated by climate change projections will
have a negative influence on Douglas-fir growth.
Increased throughfall of precipitation in the trails offers an explanation for the lack
of difference between the Edge and Open class in drought resistance. In conventional
thinning trials, gaps created between canopies permit rainfall to reach the forest floor
directly rather than striking and potentially adhering to the canopy, leading to increased
moisture availability in the soil (Stogsdili et al., 1992; Bréda et al., 1995). The same
phenomenon has been observed in harvesting trails established to facilitate a thinning
treatment, and partially linked to the increased growth of trees adjacent to the trails
compared to the interior of the thinned stand (Wallentin & Nilsson, 2011). When trees are
harvested exclusively from skid trails in a strip thinning, those corridors experience
increased throughfall and available moisture (Sun et al., 2015). Trees retained within the
48

trails would have access to this resource in all directions and those on the edges have
partial access, while those in the interior receive no such benefit from the harvest and some
face competition from their better-supplied neighbors on the edge of trails.
Inter-tree competition offers the simplest explanation for the differences found in
live crown ratio, sapwood basal area, and growth rate, though increased soil moisture in
the trails likely contributes to growth as well. Mechanical abrasion between crowns
commonly leaves gaps between tree canopies (Franco, 1986; Meng et al., 2006), an effect
known as crown shyness that appears to limit the crown dimensions of both Edge and
Interior trees (Figure 2.9). Crown recession, the process by which lower branches die
(reflected in live crown ratios), is also attributed to inter-tree competition (Maguire &
Hann, 1990). Open growing trees have little to no crown abrasion or shading from
competitors, while Edge trees face that competition on one side and Interior trees are
entirely surrounded. Greater crown size and conductive structure implies a greater demand
for moisture, potentially leading to a greater relative loss in growth when that demand
cannot be met. Dominant Douglas-fir with fully exposed crowns have been found to
experience increased water stress, with negative consequences for oleoresin exudation
pressures critical to initial defense against Douglas-fir beetle (Rudinsky, 1966). The
moderate demand of the smaller-crowned, partly-shaded Edge trees may explain their
occasionally superior drought resistance and resilience.
Increased radial growth rate following release from competition also implies
increased capacity to set aside starch reserves that may carry trees through brief periods of
stress, especially defoliation (Waring, 1987; Loescher et al., 1990). The mechanisms of
mobilizing starch reserves are poorly understood, but the general equation for converting
49

starch to glucose includes an input of water (Ray & Behera, 2011). In a study of sudden
aspen decline in Colorado, substantial loss of hydraulic conductance was observed in
affected trees, while little difference was found in carbohydrate reserves (Anderegg et al.,
2012). If the process of mobilizing starch reserves is moisture-limited at any step, then the
starch reserves may not play a substantial role in drought resistance. If starch reserves are
proportional to stem growth (Mitchell et al., 1983), higher resistance might be expected in
the Open class compared to the Edge class on account of their significantly higher growth
rate. This trend is not apparent in the results of this study, providing no support for or
evidence against the hypothetical relationship between starch reserves and drought
tolerance.

50

Figure 2.9: Example of crown shyness where the crown of the edge tree (left) has been
limited by an interior tree (right) of similar height.

The growth reduction apparent among all classes starting in 2002 (Figure 2.6) is
likely related to the defoliation by western spruce budworm (Choristoneura freemani
(Razowski) = C. occidentalis (Freeman)) that began in 2001 and continued intermittently
at the sites through 2012. Western spruce budworm is a defoliating insect native to western
North America that causes growth reduction by consuming buds and needles of Douglasfir, true firs and white spruce (Nealis, 2016). Budworm defoliation was mapped at site
Colpitt for eight years, beginning in 2002, and sites Sting and Vert for nine years beginning
in 2001. The negative effect of defoliation on growth is expected to continue for several
years after defoliation has ended while the tree recovers its foliage (Alfaro et al., 2002).
51

Regeneration in the skid trails was not quantified in this study, but its presence offers
another explanation for the growth reduction in the overstory. Regeneration has become
substantial in many of the trails, potentially intercepting rain and snow and competing for
soil moisture and likely causing growth reduction in combination with western spruce
budworm defoliation (Figure 2.10; Sterba et al., 1993; Dolph et al., 1995).

Figure 2.10: Regeneration in a skid trail at Vert Lake

52

The mechanical harvests that form the basis of this study differ from mule deer
conservation practices (Dawson et al., 2007) in several potentially important ways. First,
mule deer guidelines suggest the creation of discrete clumps of 4–10 trees, while these
harvests created more or less open corridors several hundred meters in length, with a
largely unmanipulated matrix between them. “Clumpy retention” silviculture would leave
far fewer trees in the interior of clumps than were found in these stands, more trees on the
edge of clumps, and fewer in complete isolation (Dawson et al., 2007). Second, thinning
from below is recommended during the first pass in a young stand of Douglas-fir in the
winter ranges, and in subsequent harvests where a substantial pole-size (12–35cm
diameter) component remains (Dawson et al., 2007). This would control the most severe
inter-tree competition, while in the study stands there was no such treatment within the
matrix. Finally, the single-tree selection harvesting recommended for mature stands
includes a basal area target that varies by habitat class, ranging from 16–29 m2/ha
depending on the moisture regime of the stand (Dawson et al., 2007). Inventory on a perhectare basis is a notoriously unreliable indicator of the competition experienced by trees
growing in clumped arrangements due to the variability inherent in the structure
(Hamilton, 1984), but imposing the recommended stand-level maximum basal area may
further reduce competition stress in clumps maintained for mule deer conservation.
The methods of this study limit inference to individual codominant and dominant
trees following the removal of older cohorts from an uneven aged stand. While young
regeneration and saplings of intermediate age were present at all sites, there were no
substantially larger or older trees remaining within 50m of any of the subject trees. The
applicability of these conclusions to other age classes and structures remains to be tested.
53

Coordinates of individual subject trees were not recorded, preventing analysis of
spatial autocorrelation. Distances between different trees within the same competition class
ranged from 20m to approximately 50m, with the minimum distance enforced to avoid
sampling trees directly interacting through root grafts, mycorrhizal networks, or
competition. Spatial autocorrelation may be positive or negative depending on distance,
such that growth patterns may be more similar between trees at short distances, more
different at intermediate distances, and more similar again at greater distances (Dale &
Fortin, 2014). These correlations may reflect differences in the underlying soil, the
influence of topography on the water table, or the spatial legacy of the primary-growth
forest that was partially harvested. Conclusions of this analysis would benefit from further
investigation in stem-mapped plots to assess autocorrelation at a range of distances,
perhaps building on work by LeMay et al. (2009), who measured spatial patterns of
mortality in the same interior Douglas-fir ecosystem type. This would better inform the
design of the inferential statistical test, as effective sample size may be less than the
nominal value if spatial autocorrelation is positive (Dale & Fortin, 2014). As it could not
be determined whether spatial autocorrelation would affect statistical tests positively,
negatively, or at all, no adjustments were made. If positive autocorrelation was affecting
the results, less variation would be seen in the sample compared to the overall population,
while negative autocorrelation would lead to an overestimate of variability.

54

2.6. Conclusion
Reduction of inter-tree competition through complete release supports larger
crowns and conductive structures, leading to higher basal area growth rates but providing
no commensurate increase in drought tolerance relative to partially released trees.
Increased precipitation throughfall in the harvest corridors, similar to that noted in thinning
trials, offers the most parsimonious explanation for the similarity in drought resistance
between fully and partially released trees, though a variety of interacting factors are
certainly at play. These benefits are likely to be realized operationally in stands managed
for mule deer conservation as recommended practices include more gaps than were created
in the harvests of the 1970s. Long-term negative correlations between summer temperature
and growth indicate a higher risk of drought stress under projected climate change, but
these risks may be mitigated in the short to mid-term by the creation of canopy gaps and
either complete or partial release from competition.

55

3. Drought resistance of old-growth interior Douglas-fir (Pseudotsuga menziesii var.
glauca (Beissn.) Franco) in the context of disturbance history in the CaribooChilcotin region of British Columbia
3.1. Abstract
Old-growth forest reserves in managed landscapes are threatened by changes to the
moisture regime under which they developed, and that threat may be magnified by
exclusion of density-regulating disturbances. In this study I investigate the influence of
gaps created by natural disturbances on the drought tolerance of surviving interior
Douglas-fir during seven droughts in the presettlement era (1717–1869) and seven in the
modern period. I use tree ring analysis to describe the relationship between drought
resistance (relative growth reduction) and age, and growth response to a disturbance.
Multinomial logistic regression models explaining drought resilience (return to pre-drought
growth rate) are compared in an information theoretic approach, assessing the relative
influence of five site-level and five tree-level factors. Regional growth release rates varied
from 5–25% per decade in the presettlement era, with the exception of 1800–1809 when
38% of trees recorded release. Drought resistance is higher in young trees and those with
ongoing growth release. Adjacency to canopy gaps reduces competition, explaining growth
release, and canopy gaps are known to facilitate increased precipitation throughfall,
explaining drought resistance. Density-regulating disturbance of any kind is expected to
enhance drought tolerance in structurally complex stands of interior Douglas-fir.

56

3.2. Introduction
Conservation of primary-growth and complex late-successional forests is a
cornerstone of ecologically oriented landscape-scale forest management in production
forests around the world (Seymour & Hunter, 1992; Frelich & Reich, 2003). These
reserves are functionally, aesthetically, and culturally valued. They provide required habitat
(Müller et al., 2010), serve as a benchmark of ecosystem composition and function for
managed forests (Fraver et al., 2007), and they are a preferred setting for activities ranging
from the educational to the spiritual, recreational, and sporting (Ananda & Herath, 2003).
Climate change threatens old-growth forest structures by altering the moisture regime
under which the stands developed (Noss, 2001). In dry ecosystems of western North
America, the direct effects of climate change on moisture availability are potentially
compounded by decades of fire exclusion (Cocke et al., 2005). Absence of densityregulating disturbance for a prolonged period has led to higher stand densities with
increased transpiration demand and canopy interception of rainfall, intensifying the stress
of reduced moisture availability (Pypker et al., 2005). Increased drought stress in these
forests has increased both direct mortality and predisposition of stands to disturbance by
native insects and catastrophic fires (Franceschi et al., 2005; Guarín & Taylor, 2005).
Thinning has been found to increase drought tolerance of managed stands in a number
of ecosystems (Sohn et al., 2013; Elkin et al., 2015; Calev et al., 2016), with increased
precipitation throughfall and decreased competition for available moisture identified as
underlying mechanisms (Stogsdili et al., 1992; Mazza et al., 2011; Nanko et al., 2016).
Most studies have focused on relatively young, even-aged stands, often plantations. The
simplicity of these stands supports straightforward and repeatable experimental design, but
57

it is difficult to translate the conclusions to older and more structurally complex forests.
Precipitation throughfall in late-successional forests with complex vertical and
horizontal structure has been found to be directly related to the quantity of plant material
overhead (Nadkarni & Sumera, 2004). Disturbance can reduce the amount of overhead
plant material in the short term, leading to increased throughfall (Zirlewagen & von
Wilpert, 2001) which is known to have a substantial effect on the overall water balance of
resultant stands (Bouten et al., 1992). Exclusion or suppression of natural disturbances
may have a substantial effect on the water balance of forest stands by reducing the
frequency of patches and gaps in the canopy, with measurable effects on the ability of trees
to tolerate short-term droughts.
Old-growth interior Douglas-fir (Pseudotsuga menziesii var. glauca) forests without
evidence of direct anthropogenic disturbance are scattered throughout the CaribooChilcotin region in central British Columbia. Permanent old-growth management areas
(OGMAs) limit harvesting on over 200,000 hectares of Douglas-fir forest in the region
(Figure 3.2; https://catalogue.data.gov.bc.ca/dataset/old-growth-management-areas-legalcurrent accessed January 2017), while hundreds of other old-growth stands have been
avoided by chance or choice. In addition to OGMAs, where harvesting is limited to
specific situations related to forest health and wildlife habitat, Community Areas of Special
Concern are protected as socially essential no-harvest areas (Ministry of Agriculture and
Lands, 2011).
Unharvested stands tend to be both horizontally and vertically complex, with diverse
canopy strata punctured by gaps and patches initiated by windthrow, root disease, bark
beetles or wildfire. The range of common natural disturbances includes factors that act
58

from above, mostly killing larger trees, and from below, (Smith et al., 1997) primarily
killing smaller trees: Douglas-fir beetle (Dendroctonus pseudotsugae) preferentially
attacks trees over 40cm diameter (Kano, 2006), while wildfire more readily kills smaller
trees with thinner bark. Wildfire frequency has decreased since the 1920s due to
firefighting efforts and consumption of fine fuels by cattle (Daniels & Watson, 2003;
Harvey, 2017), and the general exclusion of this disturbance has led to increases in average
stand density and proliferation of a dense understory (Wong & Iverson, 2004).
Most Douglas-fir leading stands in the region historically included a component of
lodgepole pine (Pinus contorta) that was almost completely destroyed by the mountain
pine beetle (Dendroctonus ponderosae) outbreak of the 2000s (Figure 1.5), leaving stands
of pure or nearly-pure Douglas-fir (Figure 3.1). Previous mountain pine beetle outbreaks
ca. 1890–1900, 1930–1940, and 1970–1985 caused extensive damage to the lodgepole
pine component of mixed stands, in some cases causing substantial growth release in more
than half of the Douglas-fir (Hawkes et al., 2004). Lodgepole pine may have been largely
removed from many stands for the near future, but the past has been defined by a twospecies system, with each conifer species affected by a native bark beetle on a more or less
regular basis. These insect disturbance regimes are distinct, with mountain pine beetle
tending to affect extensive swaths of timber periodically while Douglas-fir beetle typically
affects small groups or patches of trees, but rarely all mature trees within a stand
(Erickson, 1992).

59

Figure 3.1: Tree PUD18, ca. 1721, surrounded by dead lodgepole pine and regeneration of
pine and Douglas-fir, with mature Douglas-fir in the background.
60

Precipitation at the nearby Williams Lake Airport averaged 435mm per year from
1961–1990 (http://www.climatewna.com/) and has not significantly changed in recent
years, while temperatures have increased and are projected to continue increasing over the
next century (Dawson et al., 2008). Outbreaks of Douglas-fir beetle have been severe in
the 21st century, and western spruce budworm (Choristoneura freemani = C. occidentalis)
is expected to increase its range, frequency and severity in a warmer climate (Murdock et
al., 2013; Marciniak, 2015). Droughts lasting one or two years are common, and are
considered to be predisposing factors for both Douglas-fir beetle (Powers et al., 1999) and
mountain pine beetle (Shore et al., 2004) in addition to wildfire (Daniels & Watson, 2003;
Harvey, 2017) and potentially western spruce budworm (Flower et al., 2016).
Susceptibility to Douglas-fir beetle is largely dependent on the tree’s vigor and ability
to produce and mobilize defensive compounds. Pressure of toxic oleoresin is a major
determinant of success in rejecting the initial attack, while secondary resins produced after
the attack can prevent brood development (Rudinsky, 1966). Oleoresin pressure and
productive capacity are closely related to water balance, which is found to fluctuate
substantially both within the day and throughout the season. Trees on a southern aspect and
those with fully exposed crowns record higher daily variation, and pressure levels are
typically lowest midday and in the afternoon when beetle flights most commonly occur.
Even a brief period of moisture deficit within the tree is expected to substantially increase
vulnerability to an attack, making competition for moisture and its overall availability
particularly important to survival.
Stand thinning has been found to reduce mortality to Douglas-fir beetle in the long
term (Williamson & Price, 1971), an effect probably attributable to improved water
61

economy and removal of weak and dying trees responsible for spillover attacks. Prevention
of natural wildfire over the past century has led to the opposite effect, allowing stands to
reach densities at which stand stagnation is a major concern due to competition from dense
populations of seedlings and saplings (Day, 1997).
In this study I investigate the influence of openings created by natural disturbances on
drought resistance and resilience in interior Douglas-fir using tree ring analysis (Lloret et
al., 2011). Radial growth is used to estimate trees’ response to drought, but can be a proxy
for other allocations of photosynthate, including the generation of carbohydrate reserves
and the ability to create and mobilize chemical defenses against bark beetles (Waring,
1987; Franceschi et al., 2005). The specific research questions are:
1. Does adjacency to gaps created by natural partial disturbances, inferred by a sudden
increase in basal area increment, correspond to higher drought resistance of
surviving trees?
2. Does age at the time of disturbance affect the drought resistance of trees in unevenaged stands?
3. Is the return to pre-drought growth rate (drought resilience) more closely related to
site-level factors (solar radiation, elevation, annual heat-moisture index, northing,
easting) or tree-level factors (age, resistance to current drought, basal area, average
latewood proportion, prior growth rate)?

62

3.3. Methods
3.3.1. Site Selection and Field Methods
Study sites were identified using aerial imagery in Google Earth® (dated 2005,
accessed 2013 and 2014) to locate stands that had not been previously logged and were
likely to contain old trees. Lack of visual evidence of skid trails was the primary criterion
for selection of study sites; the size of the stand and presence of trees with large crowns
was also considered. A number of candidate sites were identified throughout the range of
Douglas-fir forests in the region. The largest stands were visited first and sampling took
place immediately if the sites were found to be appropriate. Presence of cut stumps or lack
of trees >250 years old was grounds for dismissing the site and moving to an alternate. A
total of 46 sites were sampled (Figure 3.2).
Approximately 20 sample trees were selected along a transect perpendicular to the
prevailing slope at each site, with two trees selected at points every 20m (paced distance).
Sample trees were generally >40cm DBH and evidently older trees were preferentially
sampled. Charred bark and absence of branches on the lower bole were generally reliable
indicators of old age. Two 5.1mm increment cores were taken at breast height from
opposite sides of each tree, perpendicular to the slope to avoid reaction wood. Trees
obviously damaged by fire (i.e. an open catface), lightning, or wind breakage were avoided
whenever possible, though minor spike tops (up to 3–4” diameter) and charred bark were
acceptable.
Diameter of each subject tree was measured at 1.3m and a photograph showing the
crown and bole was taken for reference. Forest inventory was recorded in 3–5 fixed-radius
plots along each sampling transect. Trees greater than or equal to 15cm DBH were
63

measured in 1/40th ha plots while smaller trees were measured in 1/250th ha plots. All data
were scaled to a per-hectare basis.

64

Figure 3.2: Study sites in the context of Douglas-fir leading stands and permanent old growth management areas (OGMAs).
65

3.3.2. Laboratory Preparation
Increment cores were air dried, glued to individual wooden mounts, and sanded with
a progression from 240 to 400 to 600 grit sanding belts (Stokes & Smiley, 1968).
Automotive polishing paper (1,200 and 2,000 grit) was used where necessary to discern
miniscule rings. Visual crossdating was performed on all cores under an Olympus® SZ-61
microscope using a combination of the list method and the memorization method (Speer,
2010). The list method was used to identify marker years in the first several sites to be
crossdated until a number of regionally consistent marker years were identified then used
for visual crossdating of all samples.
Crossdated increment cores were scanned at 1,200 DPI using an Epson 1640XL
flatbed scanner. Ring width and latewood width measurements were made using the
computer program Windendro® (Regent Instruments 2014). Batches of 8–20 cores were
scanned and processed together so that crossdating could be examined using the inprogram ring width index display. Crossdating was verified at the site level using the
program COFECHA (Holmes, 1983). Cores or sections that were excessively damaged or
otherwise impossible to crossdate reliably were discarded from the study, leaving 1,693
cores from 907 trees at 46 sites (Figure 3.2).
3.3.3. Statistical Analysis
I defined a drought event as a year in which the mean ring width index calculated
among all cores in the study (detrended with a 10-year cubic smoothing spline) was in the
10th percentile. Previous work (Lloret et al., 2011) used Cook’s tree-ring reconstruction of
Palmer’s Drought Severity Index (PDSI; Cook & Krusic, 2004) in addition to narrow ring
66

width to identify drought years. I did not use PDSI directly because visual inspection of the
plotted PDSI reconstruction and the detrended ring width indices revealed a lag in which a
low reconstructed PDSI value sometimes preceded a narrow ring by one year while in
other cases both the index value and PDSI value were remarkably low in the same year.
Dendroclimatic work in the region identified the total precipitation from the previous June
through current May, not PDSI, as having the greatest correlation value with Douglas-fir
ring width (Watson & Luckman, 2002). Previous-June through current-June (pJJ)
precipitation total provided the highest pairwise correlation with detrended ring width
indices in my work (r = 0.58, n = 76, P < 0.0001), and narrow rings often occurred in years
with below-average pJJ values.
While low-growth events are attributed to drought and referred to throughout this
chapter by the year in which a sharp growth reduction occurred, it is acknowledged that
this reduction may be attributable or partly attributable to low precipitation in the previous
summer or an unidentified non-drought factor such as a late frost. One event (1988/89)
lasted two years and all values were calculated based on the mean of the two years, but the
event is referred to exclusively by the first year. Longer-term droughts are understood to
have occurred in the region (Hart et al., 2010), but this analysis is focused only on those
which lasted one or two years.
Drought resistance was calculated by dividing the basal area increment in the drought
year by the mean basal area increment over the five previous years (Lloret et al., 2011).
Drought resilience was calculated by dividing the five year mean basal area increment after
the event by the five-year mean basal area increment before (Figure 3.3).

67

Figure 3.3: Increment cores BLU 17A ("17"; release) and BLU 15A ("15"; no release).
Drought resistance equals Y/X; drought resilience equals Z/X. Core 17 shows an example
of a false ring in 1931, followed by continued growth that accounts for more than half of
the ring width and a high drought resistance score. Black dots indicate decade years, from
the left: 1910, 1920, and 1930.
Growth releases were identified using the package TRADER in the open-source
statistical program R (Version 3.3.1; script in Appendix A). Growth release criteria used by
Hadley & Veblen (1993; 250% increase in mean ring width in subsequent vs. previous five
years)to identify Douglas-fir beetle mortality were tested and modified by averaging the
growth over a longer period (20 years) to eliminate false positives that commonly occurred
following ~10 year periods of growth suppression corresponding to known regional
droughts (Wolfe et al., 2001; Hart et al., 2010). Ring widths were converted to basal area
increments to reflect as accurately as possible the actual growth of the tree (as in Lloret et
al., 2011). The criteria ultimately used to define growth release were:

68



Period length: 20 years before and after each year



Minimum rate of basal area increment increase: 50%



Minimum time elapsed before another release can be identified: 15 years



Minimum duration: one year

These criteria were further tested against the data from Chapter 2, where the date of
disturbance was known. For those data, growth release was identified in 80% of trees
located on the edge of skid trails and 38% of the trees growing in the matrix between skid
trails. Individual-tree plots of the interior trees showing release were reviewed, and there
was in fact an increase over that period of time in these cores, so the criteria were accepted.
Growth releases were consolidated to a per-decade rate and the percentage of the trees
recording release in each decade was calculated by dividing the number of releases by the
number of trees present at the end of each decade. Initiation dates at coring height were
estimated based on the oldest ring among the paired cores for each tree. Initiation dates at
coring height were consolidated to a per-decade basis for comparison against release dates
on the basis of the percentage of the total sample initiated in any given decade.
Each tree was assigned to an age class (at coring height) at the time of each drought to
compare resistance values by age. The age classes were 5–49, 50–149, 150–249, and 250+,
roughly representing young, mature, old, and very old trees. As the data could not be
transformed to meet the assumption of normality, the nonparametric Kruskall-Wallis ranksum test (CI 95%; Stata® 12.1) was used to assess the differences between age classes and
to perform a post-hoc test when the fit of the overall model was significant.
As a point of reference for individual-tree growth trends, mean basal area increments
across the entire study area were plotted for 50-year age classes based on the current age
69

(at coring height) of specimens: 200–249 (n=175), 250–299 (n=169), 300–349 (n=132),
350–399 (n=101), and 400+ (n=84). These data were smoothed with a 50-year spline prior
to averaging to visualize long-term trends at the expense of annual and decadal variability
(see Figure 3.10). To further describe trends apparent in the data, diameter at breast height
in the years 1800 and 1840 was calculated by doubling the sum of accumulated ring widths
at each date. The resulting diameter distributions were graphed by 5cm DBH classes, with
trees in 1840 separated into ingrowth (trees not present in 1800) and accretion (trees
present in 1800).
All trees over the age of 50 were classified as having an ongoing growth release or
not. Drought resistance values were compared in a two-way ANOVA considering
release/no release and site code as explanatory variables, including an interaction term.
Resistance values were transformed as noted in the output table and residuals were tested
for heteroscedasticity to confirm that residuals had equal variance across the predicted
range. Conformity of the residuals to the normal distribution was visually assessed using
histograms.
Multinomial logistic regression was used to assess the resilience of individual trees
(mlogit, Stata 14). I placed each tree into one of three equally populated categories based
on resilience scores. Low resilience trees were those with the lowest 33.3% of values, mid
resilient trees included those in the middle third of resilience scores, and high resilience
trees included those with the top third of scores. These categorical scores were regressed
against a set of independent variables that described site and growth factors for each tree
(Table 3.1).
ClimateWNA (http://www.climatewna.com/, accessed 2016) was used to generate
70

annual heat-moisture (AHM) indices for each site. Elevation data (1:50,000 scale)
(http://ftp.geogratis.gc.ca/pub/nrcan_rncan/elevation/geobase_cded_dnec/50k_dem/
accessed 2016) were used to calculate total solar radiation at each study site using the Area
Solar Radiation tool in ArcGIS. Elevation data were assigned to sites, and the projected
easting and northing values were calculated in ArcGIS.
Age and basal area at the time of each drought, and average percent latewood over the
life of the tree, were calculated from Windendro® measurements. Multinomial logistic
regression in Stata 14 was used to assess the influence of each of these variables (Table
3.1) on the probability of a given tree falling into one of the higher resilience categories vs.
the lowest category. Tolerance scores with a threshold of 0.1 were used to identify
multicollinearity in model structure.
An information theoretic approach was used to identify the most parsimonious
multinomial logistic regression model (Burnham & Anderson, 2004). I used the Akaike
Information Criterion for small sample sizes (AICc) to identify the best model of the set of
models that I tested. Models with AICc differences <2 were considered to be similar and
are reported. In total, nine candidate models were assessed (Table 3.2), and standard errors
were adjusted within the mlogit model to account for clustering at the site level (Rogers,
1993). Three models consisted of the complete model less one variable or conceptuallyrelated pair of variables (northing/easting). Resistance was expected to be a powerful
variable based on previous work (Lloret et al., 2011); a model excluding that variable was
included to screen for changes in coefficient sign. A model excluding elevation was
included to screen for cases of coefficients changing sign as elevation has had an
overwhelming influence on some model fits (Chris Johnson, UNBC, personal
71

communication 2017). A model excluding northing/easting was included to assess the
importance of geographic location. Models using exclusively site-level and tree-level
variables were included to test whether one category or the other best explained the
observed variation alone. Two models including a subset of site and tree level variables
were included to assess the explanatory power of more limited combinations in
comparison to the full model. One model included only elevation to test whether that
variable alone could provide the most parsimonious explanation for the variation observed.
AICc scores were used to evaluate the explanatory ability of each model in the context
of its complexity, accounting for sample size, using the formula
(LL*-2)+(2*k)+((2*k(k+1))/(n-k-1)
where k is the number of parameters in the model, n is the sample size, and LL is
the maximised log likelihood of each model from the mlogit output.

72

Table 3.1: Description of variables used in multinomial logistic regression models used to
explain the drought resilience of subject trees.
Dependent Variable
Resilience = 5-year post-drought/5-year pre-drought growth
ResilCD

1 = lowest 33.3%; 2 = middle 33.3%, 3 = highest 33.3%
Independent Variables

LW
AHM
Solar
Elev
Resist
Prior
Age
BA
Easting
Northing

Mean latewood % as a decimal, over the full life of the tree
Annual Heat-Moisture Index (ClimateWNA)
Total annual solar radiation; watt-hours per square meter
Elevation; meters above sea level
Resistance to the current drought (drought/previous 5-year avg.)
Average growth rate over the past 20 years; square mm
Age at the time of the drought
Basal area at the time of the drought; square mm
Projected easting value in meters (NAD 1983 Albers)
Projected northing value in meters (NAD 1983 Albers)

73

Table 3.2: Candidate variable sets tested using multinomial logistic regression used to
explain the drought resilience of subject trees
Model Title
Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

Variables
LW+AHM+Solar+Elev+Resist+Prior+Age+BA+Easting+Northing
LW+AHM+Solar+Elev+Prior+Age+BA+Easting+Northing
LW+AHM+Solar+Elev+Resist+Prior+Age+BA
LW+AHM+Solar+Resist+Prior+Age+BA+Easting+Northing
Elev
AHM+Solar+Elev+Easting+Northing
LW+Resist+Prior+Age+BA
LW+AHM+Elev+Resist+Prior+BA
AHM+Solar+Elev+Prior+Age+BA+Easting+Northing

74

3.4. Results
Drought years in 2010, 1988/1989 (referred to as 1988), 1959, 1946, 1938, 1931,
1905, 1869, 1842, 1800, 1772, 1760, 1734 and 1717 were selected for analysis (Figure
3.4).
1.5
1.4
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5

1717

1734

1760

1772

1905
1800
1842

1938

1869
1931

2010

1946
1959

1988

1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

Ring Width Index

1.3

Figure 3.4: Detrended ring width index (10-year cubic smoothing spline) calculated from
all trees in the study. Dotted line indicates 10th percentile (0.7978) used to define drought
events. Black boxes indicate drought years selected for analysis.

75

All of the selected drought years had a negative reconstructed PDSI value in either
the preceding or current year (Table 3.3). All years within the period of instrumental record
(1936 to 2012) had below average previous June through current June precipitation totals,
ranging from a 16 to 28% departure from the average. Growth rings in Jasper and Banff
were concurrently narrow in six out of 12 years for which the data overlapped, suggesting
these years were regional rather than local droughts.
Table 3.3: Identified drought years and regionally narrow years (negative index values (10year spline) in Jasper and Banff chronologies), low reconstructed PDSI (negative values at
Cook’s reconstruction point #30) in the current and previous year, and below-average
previous June through current June total precipitation (pJJ). Gray cells indicate no data.

Year
2010
1988
1959
1946
1938
1931
1905
1869
1842
1800
1772
1760
1734
1717

Narrow in
Jasper and
Banff

Y
Y
Y
Y
Y

Y

Low
PDSI
Current
Year

Y
Y
Y
Y
Y

Y
Y

Low
PDSI
Previous
Year
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y

Below
Average
pJJ
Y
Y
Y
Y
Y

Study sites ranged from 645 to 1329m ASL, basal area ranged from 23 to 105 m2
per hectare in trees >15cm DBH, and trees per hectare <15cm DBH ranged from 337 to
81,330 (Table 3.4).

76

Table 3.4: Site codes, elevation (meters asl), basal area per hectare in trees over 15cm dbh
(BAPH; m2 per hectare), trees per hectare in trees under 15cm dbh (TPH) and locations
(decimal degrees). Inventory data not collected at sites with grayed-out cells.
Site
ABR
ACS
ALE
ALK
BCA
BIG
BLU
CAL
CAM
CHI
CTI
DCB
DOG
ENT
EPI
FUZ
GAR
HAN
HAW
JKY
KCR
LEE
LEN
MAB
MAQ
MAY
MCL
MEL
MON
NEE
NOJ
NRA
PTI
PUD
PVA
PYP
REN
RES
RIS
SOA
SUG
TWA
WEX
WHI
WIL
YUN

Elevation
1,133
1,024
959
954
935
1,261
871
1,248
1,087
906
818
947
1,136
910
1,262
1,068
1,239
1,329
1,306
1,136
831
1,031
1,268
945
931
1,127
679
939
1,127
1,070
1,016
970
1,005
967
786
1,107
988
953
1,259
645
658
839
1,013
857
677
996

BAPH >15cm
39
82
23

TPH <15cm
12,818
2,147
2,147

32

443

50
53
53

6,112
5,725
35,361

52
52
51

5,557
1,957
3,220

51
47
29
59
59
105
59
55
87
47
59
43
39
65
47
74
54
40
34
44
54
35
64
28
73

573
337
3,537
2,021
3,199
12,692
5,999
2,526
2,463
19,533
2,442
6,062
2,273
1,452
1,452
1,894
1,137
1,831
1,326
1,200
1,213
1,314
1,011
2,589
1,642

41
52
35
77
48
38

1,642
1,894
1,326
1,515
674
81,330

77

Latitude
51.74389230
52.01175487
52.06831024
51.78347817
52.09017948
51.31380538
52.33392135
52.06929872
51.69793030
51.90004686
52.65916903
51.57270984
51.47823126
51.98366340
51.60443734
52.28706360
51.70913776
51.96520698
51.96565471
51.69171494
52.06426670
51.95623186
52.03817036
51.90773761
52.35246379
51.88406410
52.42191831
52.25422122
51.18898694
52.18793720
52.46709374
52.21898222
52.16269145
52.18467036
52.28627230
52.03219767
52.16901851
52.13741314
51.94922460
52.42216492
52.10276261
52.54848220
52.06140806
52.21711702
52.15969659
51.88829296

Longitude
-123.01188846
-123.29969878
-123.40771210
-122.18762809
-123.35763661
-122.14807292
-122.24453402
-122.61526964
-121.61113394
-121.93915042
-122.37337676
-122.22259727
-122.15729418
-121.87442622
-122.01334020
-122.68007984
-122.43351471
-122.91335345
-122.93040419
-122.20377503
-121.88228808
-123.07392013
-123.10607239
-122.27201389
-122.51813162
-122.19751849
-122.36814577
-122.39560948
-121.19581713
-122.63864732
-122.72308404
-122.50518352
-123.91295910
-123.85250382
-122.18609312
-124.13537884
-123.71319263
-123.74083106
-122.63877948
-122.41362522
-122.02310368
-122.63091856
-123.60929129
-122.32213471
-122.23387497
-123.10229104

Aside from the spikes to 38% and 34% in 1800–1809 and 1930–1939, the
percentage of trees recording growth release in the presettlement era varied between 5 and
25% per decade, averaging 14% from 1600–1609 to 1880–1889. No decade has exceeded
10% of trees recording growth release since the 1930s. Initiation rates (at coring height) of
sample trees varied between 5 and 13% per decade after 1600.

Growth Release and Initiation

40%
35%
30%
25%

20%
15%
10%
5%
2000
1980
1960
1940
1920
1900
1880
1860
1840
1820
1800
1780
1760
1740
1720
1700
1680
1660
1640
1620
1600
1580
1560
1540
1520
1500

0%

Figure 3.5: Percentage of the total sample reaching coring height in a given decade
(“Initiation”; grey) and percentage of trees present at the end of each decade showing
growth release >50% in each decade (black).

78

Mean basal area increments among the four older age classes were synchronized
ca. 1800–1840, and then separated by age class, with younger cohorts continuing to
increase their growth rate and older cohorts remaining steady or falling slightly (Figure
3.6). Growth rate in all cohorts precipitously declined starting ca. 1995.

Mean Basal Area Increment (mm2)

1200

1000

400+
350-399
300-349
250-299
200-249

800

600

400

200

2005
1995
1985
1975
1965
1955
1945
1935
1925
1915
1905
1895
1885
1875
1865
1855
1845
1835
1825
1815
1805
1795
1785
1775
1765
1755
1745
1735
1725
1715
1705
1695
1685
1675
1665
1655
1645
1635
1625
1615

0

Figure 3.6: Mean basal area increments by 50-year age class across the entire study area
since 1615, smoothed with a 50-year spline prior to averaging to display long-term growth
trends.

79

In 1800 a majority of the sampled trees that had reached breast height were less
than 20cm DBH, and 80% were less than 30cm. A shift towards the larger diameter classes
is apparent by 1840 (Figure 3.7), with an average diameter growth of 8.7cm or 2.2mm per
year in the intervening period. Ingrowth accounted for 102 of the 116 trees less than 10cm
DBH in 1840, while several newly initiated trees reached the 20–24.9cm class.
120

1800
100

1840 Ingrowth
1840 Accretion

Count Trees

80
60
40

60-64.9

55-59.9

50-54.9

45-49.9

40-44.9

35-39.9

30-34.9

25-29.9

20-24.9

15-19.9

10-14.9

5-9.9

0

0-4.9

20

DBH Class (cm)
Figure 3.7: Overall diameter distributions in 1800 and 1840, with 1840 trees separated into
accretion and ingrowth, respectively those that had and had not reached breast height by
1800.

80

Visual inspection of cores and plotted series supports the growth trend identified in
the analysis, where small, suppressed trees began growing at rates expected of canopy
dominants beginning ca. 1800 (e.g. Figure 3.8). The length of the suppressed growth in
most cores (more than 100 years in some trees) is too long to be attributed to climatic
variation, while 5–10 year periods of more extreme growth reduction correspond to
regional droughts inferred from tree ring analyses ca. 1770 and 1790 (Wolfe et al., 2001;
Hart et al., 2010), suggesting that competition was responsible for the slow growth and its
removal was the cause of the acceleration. Many of the trees showing release ca. 1800 still
remain without substantial crown competition (Figure 3.9). Tree JKY10 is one example
among hundreds of trees showing this general pattern, having reached 106cm DBH
(outside bark) after 213 years without apparent crown competition (Figure 3.10).

81

1cm

3000
2500
2000
1500
1000

1840

1835

1830

1825

1820

1815

1810

1805

1800

1795

1790

1785

1780

1775

1770

1765

1760

1755

1750

0

1745

500

1740

Basal Area Increment (mm2)

3500

Figure 3.8: Core GAR14A from 1740 to 1840 with basal area increment. Single pencil
marks indicate decade years, double mark indicates 1750, and triple mark identifies 1800.
Radial growth from 1800 to 1840 was 6.5cm, suggesting approximately 13cm diameter
growth at an average rate of 3.25mm per year. The widest ring (1820) is 2.9mm,
suggesting a diameter growth of 5.8mm on a tree 26cm DBH. Basal area increment
continued to increase until 1917, where a period of suppression marks the beginning of a
slow decline. The crown was completely isolated when sampled in 2014, with live limbs
beginning at 3–4m height.

82

Figure 3.9: Tree GAR14 (ca. 1609) in 2013. Underexposure of original photograph
corrected in GIMP 2.
83




)LJXUH7UHH-.< FD FP'%+ RXWVLGHEDUN DSSDUHQWO\LVRODWHGVLQFH
DIWHUEHLQJUHOHDVHGDWWKHDJHRIFD ZKLWHWULDQJOHDWOHIW &RUH-.<&
DWOHIW±DWULJKW±6LQJOHGRWVDWGHFDGH\HDUVGRXEOHGRWVDW
WULSOHGRWVDWTXDGUXSOHGRWVDW5DZDQG\HDUVSOLQHG
EDVDODUHDLQFUHPHQW %$, JUDSKHGDVVROLGDQGGDVKHGOLQHVUHVSHFWLYHO\




The youngest age class (5–49 years) had significantly higher drought resistance in
nine of 14 years tested (Table 3.5). The oldest age class (250+ years) had significantly
lower drought resistance vs. the next youngest class (150–249 years) in 1931, 1959, and
1988. In 1800 and 1959, the second-youngest age class (50–149 years) had significantly
higher resistance than the next older class (150–249 years).
Table 3.5: Drought resistance by age class. Kruskall-Wallis rank-sum test (95% CI; age
classes that do not share a letter within a given year are significantly different in pairwise
multiple comparisons. Pairwise comparisons not conducted for 1938 when overall model
was insignificant. No trees under 50 in 2010).
Age Class
1717
1734
1760
1772
1800
1842
1869
5–49
0.70a 0.78a 0.88a
0.84a 1.07a
0.77a 0.65a
50–149
0.59b 0.66b 0.75b 0.66b 0.92ac
0.65b 0.52b
150–249
0.57ab 0.71ab 0.69b 0.60b 0.79b
0.62b 0.50b
250+
0.56ab 0.64ab 0.63ab 0.41ab 0.76abc 0.67ab 0.56b
n
302
372
452
501
599
734
784
Chi-squared 8.88
15.25 38.77 35.96
20.04
25.69 38.92
d.f.
3
3
3
3
3
3
3
P
0.03098 0.00162 0.00010 0.00010 0.00017 0.00010 0.00010

Age Class
5–49
50–149
150–249
250+
n
Chi-squared
d.f.
P

1905
0.71a
0.62b
0.64ab
0.61b
845
12.62
3
0.0552

1931
1938
1946
1.00a 0.92ns 0.92a
0.70b 0.89ns 0.71b
0.67b 0.86ns 0.69b
0.57c 0.84ns 0.68b
872
874
879
67.46
5.53
34.76
3
3
3
0.0001 0.1371 0.0001

85

1959
1988
2010
0.63abc 0.60a
0.62a
0.65a 0.68a
0.56b
0.63a 0.64ab
0.43c
0.57b 0.62b
884
886
878
89.35
9.60
8.49
3
3
3
0.0001 0.0223 0.0143

Trees with an ongoing release had higher drought resistance than trees without an
ongoing growth release in every year except 1869 (Table 3.6). Site was a significant
explanatory variable in every year that a test could be conducted, while growth release
status was significant in seven of the 10 years that tests could be conducted. The
interaction of site and growth release status was significant in 1842, 1931, and 1946
(p<0.05; Appendix C). The data for the years 1760, 1959, and 1988 did not meet the
assumption of normality after transformation. The ANOVA and plot of residual vs. fitted
values for 1760, 1959, and 1988 are presented in Appendix C for information.
Table 3.6: Drought resistance in trees with and without ongoing release. Two-way
ANOVA; full model outputs in Appendix C. Significant variables in bold; α=0.05.

Ongoing Release
No Release
Transform
Release F
Release Prob > F
Site F
Site Prob > F

1717
0.72
0.56
Sqrt
6.86
0.010
7.35
<0.001

1734
0.72
0.66
Sqrt
19.95
<0.001
8.02
<0.001

1760
0.97
0.68

Ongoing Release
No Release
Transform
Release F
Release Prob > F
Site F
Site Prob > F

1905
1931
1938
1946
0.65
0.69
0.88
0.75
0.62
0.64
0.85
0.68
Box-Cox Box-Cox Box-Cox Box-Cox
17.02
10.91
26.22
19.92
<0.001
0.001 <0.001 <0.001
15.62
17.55
12.09
11.43
<0.001 <0.001 <0.001 <0.001

86

1772
1800
1842
1869
0.78
0.99
0.72
0.51
0.60
0.78
0.63
0.52
Box-Cox Box-Cox Box-Cox Box-Cox
0.52
12.86
2.56
0.50
0.47
0.110
0.478
<0.001
9.21
6.35
14.19
13.03
<0.001 <0.001 <0.001 <0.001
1959
0.88
0.52

1988
0.64
0.56

AICc differences <2 in two models in 1931, two in 1869, and three in 1717 led to
17 top-ranked models from the 13 drought years considered (Table 3.7). Among these, 15
were the full model or its close derivative (seven “Full,” five “Less elevation,” three “Less
easting and northing”). “Mixed 1” (% latewood, annual heat-moisture, elevation,
resistance, prior growth and basal area) was among the three top-ranked models for 1717
and was the exclusive top-ranked model for 1772. All output tables and AICC calculations
may be found in Appendix B.
Table 3.7: Top-ranked multinomial logistic regression models explaining drought resilience
scores for each drought year
Year
1988
1959
1946
1938
1931A
1931B
1905
1869A
1869B
1842
1800
1772
1760
1734
1717A
1717B
1717C

Top-Ranked Model
Full model
Full model less elevation
Full model less elevation
Full model
Full model
Full model less elevation
Full model less easting and northing
Full model
Full model less easting and northing
Full model
Full model less elevation
Mixed 1
Full model less elevation
Full model
Full model
Full model less easting and northing
Mixed 1

87

Resistance to the current drought had an exclusively positive influence on the
probability of higher drought resilience, and was significant in 13 of 17 top-ranked models
in the low-resilience vs. mid-resilience categories and 16 of 17 in the low-resilience vs.
high-resilience comparisons (Table 3.8). Prior growth rate negatively influenced drought
resilience in eight top-ranked models in the low-resilience vs. mid-resilience comparisons
and ten in low-resilience vs. high-resilience comparisons. Age at the time of the drought
provided several significant negative coefficients: one in the low-resilience vs. midresilience comparisons and five in the low-resilience vs. high-resilience comparisons. Sitelevel variables and other tree-level variables were occasionally significant, with
inconsistent coefficient signs between years.

88

Table 3.8: Coefficients of top-ranked models for each drought year. Bold values indicate
statistical significance in the original mlogit models (95% CI). Coefficients multiplied by
the value at the top of each column for concise display. Gray boxes indicate variables not
included in the top-ranking model. “1 vs. 2” indicates the probability of a tree being in the
middle 33% vs. the lowest 33%; “1 vs. 3” indicates the probability of a tree being in the
highest 33% vs. lowest 33%. Complete outputs can be found in Appendix D.
1 vs. 2
Year
1988
1959
1946
1938
1931A
1931B
1905
1869A
1869B
1842
1800
1772
1760
1734
1717A
1717B
1717C

1
LW
0.029
-0.057
0.024
-0.030
-0.012
-0.009
-0.080
-0.001
-0.016
0.020
-0.038
-0.030
-0.094
-0.096
0.056
0.025
0.007

1
AHM
0.053
0.007
0.066
-0.057
0.090
0.065
-0.051
-0.074
-0.002
-0.102
-0.074
-0.057
0.041
-0.197
-0.168
-0.065
-0.051

100,000
Solar
-0.017
-0.017
0.009
-0.072
0.015
0.018
0.119
0.016
0.003
0.069
0.120

1 vs. 3
Year
1988
1959
1946
1938
1931A
1931B
1905
1869A
1869B
1842
1800
1772
1760
1734
1717A
1717B
1717C

1

1
AHM
0.020
0.052
0.084
-0.120
0.045
0.061
-0.058
0.101
0.097
0.128
-0.042
-0.095
0.015
-0.307
-0.032
0.001
0.016

100,000
Solar
-0.082
-0.039
0.032
-0.096
0.061
0.060
0.131
0.005
0.004
0.069
0.173

LW
0.095
-0.021
0.071
-0.049
-0.001
-0.003
-0.052
0.010
0.010
0.034
-0.055
-0.035
-0.067
-0.159
0.081
0.064
0.069

100
Elev
0.356

-0.259
0.098
-0.198
-0.258
-0.015
-0.454
-0.047

-0.027
-0.029
0.091
0.090

-0.213
-0.333
-0.141
-0.100
100
Elev
1.932

-4.057
-0.722
-5.186
0.405
0.316
-0.197
-0.450

-0.103
-0.091
-0.048
-0.045

-6.162
2.905
2.216
2.872

1
Resist
2.189
3.138
0.604
2.417
1.012
1.021
3.612
0.474
0.602
3.921
1.945
3.758
3.276
1.633
5.449
6.326
6.489

100
Prior
-0.050
-0.031
-0.004
-0.034
-0.086
-0.085
-0.064
-0.067
-0.079
-0.035
-0.108
0.022
-0.197
-0.076
-0.066
-0.065
-0.047

100
Age
0.173
-0.074
0.031
-0.298
0.171
0.186
-0.045
-0.076
-0.106
0.009
-0.397

1
Resist
4.817
5.934
0.691
6.106
2.183
2.180
6.908
1.917
1.883
9.251
3.820
7.725
7.408
7.163
11.447
12.606
12.702

100
Prior
-0.066
-0.107
-0.033
-0.107
-0.150
-0.151
-0.159
-0.163
-0.164
-0.118
-0.257
-0.118
-0.401
-0.470
0.010
-0.010
0.008

100
Age
-0.047
-0.444
0.220
-0.393
0.250
0.240
-0.708
-0.725
-0.730
-0.155
-0.527

89

-0.757
-0.612
-0.860
-0.762

-1.549
-1.676
-0.461
-0.329

10,000
BA
-0.008
-0.023
0.005
-0.006
0.006
0.004
0.006
0.024
0.035
-0.035
0.042
-0.099
0.128
0.042
0.176
0.172
0.062

10,000
Easting
0.035
-0.042
-0.083
-0.012
0.084
0.057

10,000
Northing
0.037
0.092
-0.060
-0.082
0.017
-0.028

-0.090

-0.157

-0.005
-0.069

-0.115
-0.068

0.137
-0.296
-0.159

-0.198
-0.109
-0.048

10,000
BA
0.016
-0.036
-0.036
-0.016
-0.055
-0.054
0.018
0.042
0.042
-0.026
0.033
-0.104
0.263
0.226
-0.171
-0.172
-0.238

10,000
Easting
-0.043
-0.025
-0.137
0.029
0.028
0.046

10,000
Northing
-0.156
0.161
-0.115
-0.130
-0.089
-0.058

0.003

0.007

0.306
-0.038

0.066
-0.172

0.145
-0.367
-0.072

-0.363
-0.321
0.189

3.5. Discussion
Drought resistance is higher in trees with ongoing growth release and young trees in
general, and higher resistance to drought increases the probability of superior growth
resilience. Previous work has shown that exclusion of wildfire from this ecosystem has led
to stand densities that exceed historical norms, especially in the understory (Wong &
Iverson, 2004). Stand inventory data collected in this study support the conclusion that
stands are now largely overstocked, often with extraordinary densities in the understory
(Table 3.4, Figure 3.11). This work indicates that besides regulation of the understory and
overall stand density, a side-effect of the historical natural disturbance regime of wildfire
and bark beetle outbreaks was the alleviation of drought stress in surviving trees.
Limiting the analysis of drought resilience to mature trees (aged >50 years) was
intended to avoid confounding the analysis with young trees just initiating within gaps
rather than being adjacent to gaps. Age had a negative effect on the probability of higher
drought resilience whenever the variable was significant, so the exclusion of trees <50
years old may have suppressed a positive influence of youth on resilience. This study has
no bearing on the management of even-aged stands in the region and conclusions about
young trees are limited to those growing within uneven-aged stands: only one of the 46
sites considered in this analysis was even-aged (WEX, coring height ca. 1800), and that
site contributed very few of the samples in the analysis.

90

Figure 3.11: Site JKY in 2013, with an average 105m2/ha and 12,692 trees per hectare
under 15cm. Tree at center dates to 1555; tree behind and to the left dates to 1475, both
have charred bark but no open wound.
91

The mechanism underlying increased drought tolerance in trees adjacent to gaps
created by partial disturbances is likely a combination of increased precipitation
throughfall supplementing the available moisture in the soil (Bouten et al., 1992; Nadkarni
& Sumera, 2004), decreased competition in the rooting zone (Casper & Jackson, 1997),
and perhaps increased carbohydrate reserve capacity in faster-growing trees (Waring,
1987). The presence of false rings in some trees, indicating continued growth after a
temporary shut-down, is a common pattern in tree rings collected in regions driven by latesummer rains where growth is restarted when moisture becomes available (Griffin et al.,
2011). In this context, false rings observed in some samples with ongoing growth release
(e.g. Figure 3.3) may have resulted from the reception of moisture later in the season. Latesummer thunderstorms are common in the study area, but the rainfall is often brief and
light. It is possible that the effect of these brief summer storms on soil moisture is heavily
influenced by canopy interception (Bouten et al., 1992), providing a benefit to trees
adjacent to gaps that is unavailable to their interior neighbors.
On average, 14% of trees recorded growth release per decade between 1600–1609 and
1880–1889 (inclusive), varying from 5 to 25% in the presettlement era except for a peak at
38% in 1800–1809. Rates of initiation varied between 5 and 13% per decade after 1600;
the drop-off in initiation rates after 1860 is meaningless as older trees were targeted for
sampling. This variation in regional growth release and initiation rate is consistent with
previous work describing a mixed-severity fire regime with an average 20-year return
interval (Daniels & Watson, 2003; Axelson et al., 2010) and periodic mountain pine beetle
outbreaks with a 30–40 year return interval (Alfaro et al., 2010). Only one site, WEX in
the Chilcotin region, appears to be even-aged with no surviving remnants of the previous
92

cohort, meeting the definition of a high-severity disturbance used by Heyerdahl et al.
(2012). Harvey (2017) also reconstructed high-severity disturbances establishing evenaged stands in two of 27 plots in Douglas-fir forests adjacent to grasslands in the CaribooChilcotin, suggesting that these events were a part of the historical ecosystem.
The net effect of a mixed-severity regime is reasonably comparable to the
recommendations for management on mule deer winter ranges (Dawson et al., 2007),
which constitute the primary management objective in many of the drier managed forests
within the study area. These recommendations (Dawson et al., 2007) include a low basal
area target in the low-snowpack (drier) areas and intermediate basal area in moderatesnowpack stands. Pole-size trees (12.5–37.5cm) are to be thinned from below. A clumpy
distribution of mature trees is encouraged to intercept snow and provide browse through
crown abrasion. Some, but not all trees over a certain maximum size are to be retained.
The most substantial departure from the historical regime is protection of the youngest
regeneration horizon during harvest, as even a light ground fire under historic disturbance
regimes would kill a substantial proportion of the regeneration.
A mountain pine beetle outbreak severely affecting much of the Chilcotin region
(Erickson, 1992), corroborated by scars and growth releases attributed to beetle mortality
by Alfaro et al. (2010), offers a partial explanation for the widespread growth release in the
1930–1939 decade. The year 1931 was also known to be dry and prone to wildfire,
including human-initiated wildfires (https://catalogue.data.gov.bc.ca/dataset/fireperimeters-historical accessed 2017). The growth releases in Douglas-fir during the 1930s
likely have several causes, including mountain pine beetle and natural wildfires (Alfaro et
al., 2010), accidental wildfires may also have played a role.
93

The effective sample depth at almost all the study sites discussed in Alfaro et al.
(2010) is exhausted by 1800, but one lodgepole pine series and two Douglas-fir series
show growth releases starting ca. 1800 that the authors tentatively attribute to mountain
pine beetle. A number of the stands described in the fire history developed by Daniels and
Watson (2003) include large cohorts established ca. 1800, but without much evidence of
extraordinary fire activity. Several tree ring reconstructions in western Canada for the
period 1790–1799 indicate dry conditions, with low streamflow in rivers in Alberta and
Saskatchewan to the east (Case & MacDonald, 1995, 2003) and the Chilko River to the
west (Hart et al., 2010), increased sand-dune mobilization in southern Saskatchewan
(Wolfe et al., 2001), and depressed growth in Douglas-fir in Banff and Jasper National
Parks and the trees considered in this study. This prolonged moisture deficit likely
predisposed stands to Douglas-fir beetle (Jantz & Rudinsky, 1966), mountain pine beetle
(Alfaro et al., 2010), and wildfire (Harvey, 2017; Daniels & Watson 2003), while extreme
drought in 1800 may have caused direct mortality in addition to the markedly narrow rings
observed in nearly all samples. A combination of all four factors may be responsible for
the anomalous peak in growth release in the 1800–1809 decade, but all occurred before the
first known European contact in the region (Simon Fraser’s 1808 expedition).
This extraordinary growth release event in the beginning of the 19th century appears to
have had a substantial influence on the long-term growth patterns and stand development
in the region. An inflection in average growth rate occurs ca. 1795 in all four established
age classes shown in Figure 3.6, and is followed by several decades of similar average
growth among the previously distinct groups. A sharp release ca. 1800 and rapid
subsequent growth is visually apparent on many cores, and a major shift in diameter
94

distributions occurred between 1800 and 1840. Though not conclusive, these patterns
suggest an event that brought a majority of surviving trees into positions of preeminence
within their respective stands. The rapid initial growth of the cohort initiated after 1800 is a
further indication of available growing space, further supported by age structure and
growth rates described at several nearby sites by Daniels and Watson (2003) and Hawkes
et al. (2004).
The scale of the series of events starting ca. 1790 and their apparently persistent
effects may be applicable to landscape management in the area, as one intense and
widespread combination of disturbances shaped forest structure for centuries, establishing
the dominance of the large “old veteran” trees that are now considered to be characteristic
of old-growth forests in the region. Long-term fire exclusion and its effects on forest
composition and structure are not compatible with the mixed-severity disturbance regime
identified in this study (see also Harvey, 2017, Heyerdahl et al., 2013, Wong & Iverson,
2004). Preparation of forests for the warmer and effectively drier decades projected under
climate change scenarios (Dawson et al., 2008) will require re-establishment of more
open-grown forests and a reduction in understory density to enable drought resilience in
residual trees.

95

3.6. Conclusion
Results of this study indicate that adjacency to canopy gaps created by partial
disturbance provides significant increases in drought resistance in terms of radial growth,
which implies concurrent availability of photosynthate for production of chemicals used to
defend against bark beetle attack (Waring, 1987). The historical disturbance regime was
such that canopy gaps, inferred by growth release in surviving trees, were more common
prior to the exclusion of wildfire ca. 1920 and the subsequent proliferation of a dense
understory (Wong & Iverson, 2004). Changes to the moisture regime driven by
anthropogenic climate change and development of a dense understory due to wildfire
exclusion will likely make hands-off preservation a less viable strategy over time.
Increases in summer temperature are likely to put overstocked stands at a high risk of
Douglas-fir beetle attack, which, if severe, would remove many of the older trees from the
stand. This would reduce the density of the overstory substantially, but in the absence of
low-intensity fire to reduce the density of the lower strata, the result would be a dense
young stand. Overstocking is also expected to increase the risk of stand-replacing crown
fires as accumulated seedlings and saplings provide a route for flames to reach the main
canopy and a fire can then spread easily between closely-packed canopies (Agee &
Skinner, 2005). Furthermore, understory seedlings and saplings in multi-storied stands are
often heavily utilized by western spruce budworm (Maclauchlan & Brooks, 2009),
potentially increasing the pest population affecting older trees.
Application of the silvicultural principles recommended for mule deer winter range
(Dawson et al., 2007), particularly the thinning of young trees and basal area targets for
larger trees, would provide the kind of structural conditions expected to increase drought
96

resistance based on the results of this study. Within Community Areas of Special Concern
and other areas where harvesting is not a suitable solution, controlled burning could be
used to manage understory density and create some canopy gaps, with antiaggregation
pheromones used to discourage subsequent utilization of dead or damaged trees by
Douglas-fir beetle (Ross & Wallin, 2008). Outside of mule deer winter ranges and special
management areas, increased harvest levels in the next planned entry, inspired by the
general scale and effect of the disturbances of the 1790s, may help bring production forests
smoothly into a warmer and drier 21st century.

97

4. Douglas-fir beetle (Dendroctonus pseudotsugae) outbreak history inferred from
tree rings, wood scars and resin pockets in the Cariboo-Chilcotin region of British
Columbia
4.1. Abstract
Douglas-fir beetle is the primary tree-killing bark beetle attacking mature Douglasfir in central British Columbia, but its contribution to historical stand dynamics is
unknown. I used two methods of tree ring analysis to develop a historical timeline of beetle
activity. I selected tree-ring data from sites that had low lodgepole pine mortality in the 21st
century mountain pine beetle outbreak with the goal of creating a chronology of growth
release events driven primarily by Douglas-fir beetle. This approach may have been
compromised by shifting tree species range margins over the past several centuries. I
examined cross-sections of Douglas-fir for scars and resin pockets attributable to failed
Douglas-fir beetle attacks, creating a timeline from 1695 to 2014 and providing further
evidence for the attribution of these features to Douglas-fir beetle. Abundance of scars and
resin pockets was less than would be expected in trees exhibiting growth release due to
pheromone mediated mass-attack killing an adjacent competitor, suggesting that Douglasfir beetle has played a secondary role compared to wildfire and mountain pine beetle
outbreaks in the stand dynamics of the region. Douglas-fir beetle activity appears to have
passed a historical threshold in 2016 and should be monitored for changes in attack
patterns.

98

4.2. Introduction
Outbreaks of Douglas-fir beetle in the first decade of the 21st century appear to
have exceeded the magnitude of previously recorded infestations in central British
Columbia, raising concerns about unprecedented attack behavior by this native insect.
Douglas-fir beetle is a bark-boring insect that attacks Douglas-fir and western larch in
pheromone-mediated mass-attacks that are often fatal to the host tree (Walters, 1956).
Walters (1956) estimated that a fully occupied Douglas-fir could contain 11,000 attacking
beetles, while Belluschi et al. (1965) estimated that up to 3,000 beetles could be rejected
by the host’s defenses without causing mortality. Eggs are laid beneath the bark and hatch
in the same season; the larvae may mature in the same season or overwinter and pupate in
the spring (Walters 1956). Maturation time is negatively correlated to temperature (Vité and
Rudinsky, 1957), and spring flights are cued by warming temperatures following a cold period
(Atkins, 1960). Douglas-fir beetle outbreaks in British Columbia have typically affected trees
in clumps or small patches dispersed on the landscape (Erickson, 1992).

Aerial Overview Surveys have identified areas affected by Douglas-fir beetle since
1957, but survey standards changed in 2004 with the addition of Trace (<1% mortality) and
Very Severe (>50% mortality) mortality categories. This change in survey standards
introduces uncertainty into observations of trends in Douglas-fir beetle impacts. This
survey record is also substantially shorter than the lifespan of the host tree species,
providing an incomplete picture of the ecology of the host/pest system, which may be
driven by stand dynamics or predisposing factors occurring at longer timescales. A longer
baseline of Douglas-fir beetle activity, similar to those reconstructed for other bark beetle
species using tree ring analysis (Axelson et al., 2010; Sherriff et al., 2011; Hrinkevich &

99

Lewis, 2011), is desirable to put current events in historical context at timescales more
relevant to the long-lived host tree.
Daily, seasonal, annual, decadal, and longer-term climatic trends all have an
influence on the physiology of both bark beetles and their host trees, suggesting a cause for
changing outbreak behavior as average temperatures in the region have trended upwards
over the latter half of the 20th century (Dawson et al., 2008). Increased temperatures in
spring and summer are expected to accelerate the maturation of Douglas-fir beetle larvae
(Vité & Rudinsky, 1957), increasing the proportion of beetles completing development
before winter and intensifying the synchronized spring attack in the following year.
Increased drought stress resulting from observed and projected increases in temperature
(Dawson et al., 2008) and global circulation anomalies leading to prolonged droughts
(Mann et al., 2017) are expected to reduce the defensive capacity of the host by increasing
water deficit and thereby reducing the pressure of oleoresin that provides the first line of
defense against Douglas-fir beetle (Rudinsky, 1966).
While the impacts of Douglas-fir beetle are typically dispersed across the landscape,
they are not minor in an economic or ecological sense. Millions of dollars’ worth of
standing timber has been affected in almost every year since 2000, and unrecovered
volume supports extensive recruitment of snags and downed woody debris, providing
habitats and resources to animals and understory plants (Hunter, 1990). These impacts are
economically and ecologically significant, but relatively dispersed compared to the effects
of spruce beetle or mountain pine beetle (Raffa et al., 2008), which have more extensive
dendrochronological records (Sheriff et al., 2011; Hrinkevich & Lewis, 2011). I use two
methods of dendrochronological reconstruction in this study to account for the typically
100

dispersed effects of Douglas-fir beetle.
The first method of Douglas-fir beetle outbreak reconstruction follows the wellestablished method of growth-release detection, where synchronous growth releases are
attributed to mortality of adjacent trees when some evidence of the insect’s presence can be
established (Hrinkevich & Lewis, 2011). This approach is limited in this situation by the
powerful influences of wildfire and mountain pine beetle on stand dynamics (Alfaro et al.,
2004) and the typically patchy nature of Douglas-fir beetle (Negrón et al., 2001). Tree ring
series from Chapter 3 are selected from sites that are predominantly Douglas-fir with
current and historical evidence of minimal impact from the most recent mountain pine
beetle outbreak.
The second line of evidence is the dating of resin pockets and scars left at the location
of failed Douglas-fir beetle attacks. Similar scars have been observed in a number of
conifer species (Johnson & Shea, 1963; Belluschi et al., 1965), and have been used to
support the reconstruction of mountain pine beetle infestations in lodgepole pine in the
Chilcotin region of British Columbia (Hawkes et al., 2004), spruce beetle near Hudson
Bay, Canada (Caccianiga et al., 2008), and red oak borer (Enaphalodes rufulus) in
Missouri, USA (Muzika & Guyette, 2004). Resin pockets found in Douglas-fir have been
attributed to Douglas-fir beetle since at least 1965, when Belluschi et al. (1965) marked the
location of each entry wound in a number of attacked trees and waited several years before
felling the trees and examining the wood at the attacked locations. They found both resin
pockets and scars at the site of the observed attacks, and dated earlier scars to the year of a
known spot outbreak in an adjacent patch.
Douglas-fir beetle is only one disturbance agent among many others that individually,
101

or in combination have shaped the study area over time. Historical wildfires (reconstructed
by Daniels & Watson, 2003; Hawkes et al., 2004; Axelson et al., 2010; Harvey, 2017),
droughts (reconstructed by Case & MacDonald, 1995; Cook & Krusic, 2004; Wolfe et al.,
2001; Case & Macdonald, 2003; Hart et al., 2010), and mountain pine beetle outbreaks
(reconstructed by Alfaro et al., 2010 with early surveys summarized by Erickson, 1992),
are all potentially involved in the stand dynamics of the old-growth interior Douglas-fir
forests under study. The interpretation of both growth release and scar/resin pocket
frequency data to address the following research questions must be considered in the
context of other disturbances potentially affecting the sampled stands.
1. Can the evidence of Douglas-fir beetle attacks account for observed patterns of
growth release, especially the widespread growth release described in Chapter
3 following the regional drought at the beginning of the 19th century?
2. Did historical Douglas-fir beetle outbreaks tend to follow periods of drought,
or precede periods of growth release?
3. Did historical Douglas-fir beetle outbreaks typically affect sites across the
region?
.

102

4.3. Methods
4.3.1. Field and Lab Methods
Local foresters provided locations of Douglas-fir beetle salvage logging operations
that they knew to be inactive and accessible. Sites were visited to confirm the presence of
cull piles (piled ends of logs that were removed prior to hauling due to undesirable features
that would affect processing into wood products) that had not yet been burnt and that were
large enough to provide 25 sample trees, without being so large as to pose a safety risk. Six
sites that were a minimum of 4 km apart were selected (Figure 4.1).
Cutoffs from the lower end of the first log were abundant at all selected sites, so
sampling was focused entirely on sections with evidence that they had come from the
bottom section of the tree (i.e. falling cuts or paint) to maintain consistency throughout the
sample. Twenty-five cull sections were identified for sampling and two cross sections 5–
10cm thick were recovered from each: one at approximately the breast height of the tree or
the highest point available on shorter samples, and the other just above the falling cut or
lowest section. Nearly all sampled trees had visible evidence of successful bark beetle
galleries and a few contained live beetles.

103

Figure 4.1: Sampled cull pile locations for Douglas-fir beetle outbreak reconstruction overlaid by a combination of Federal Forest
Insect and Disease Survey and Provincial Aerial Overview Survey records of beetle-caused mortality identified by insect surveys
from 1957–2013, Provincial helicopter-based mortality surveys from 2007–2014, and forest stands dominated by Douglas-fir.
“Trace” severity (<1% mortality) not shown on map to reduce clutter. Five ringwidth chronologies labeled with three-letter codes.
104

Figure 4.2: Top: sampled cull pile at site H; photograph corrected for underexposure and
white balance in Microsoft Office. Bottom: Cross sections collected at site K. Each site
filled a midsize pickup truck.
105

Cross sections were flattened with a power plane to remove gouges from the
chainsaw teeth prior to sanding then were sanded using a progression of grits to120 at
which point scars and resin pockets could be reliably identified. Further polishing to 600
grit was focused on the locations around identified scars to avoid the time and expense of
sanding the entire cookie. 1200 and 2000 grit wet/dry automotive polishing sandpapers
were used when the location of the scar within the annual ring was not easily discernable.
Cross sections were sanded on each side as resin pockets at the site of Douglas-fir beetle
attacks have been demonstrated to be less than 2cm in length in some cases (Belluschi et
al., 1965).
Each section was visually crossdated using marker years known from Chapter 3 and
the year of death was identified to ensure each cross section came from a recently killed
tree. Growth rings with scars or resin pockets were traced to the dated path, and the
calendar year of each scar was recorded. Position within the ring was classified into early
earlywood (first 33% of earlywood), mid earlywood (second 33%), late earlywood (third
33%), and latewood. Distance to pith was measured to the nearest millimeter using a ruler.
Each scar was classified as either a pocket (open, filled with resin) or a true scar (closed,
not filled with resin). Scars appear to be formed where the cambium is interrupted by the
failed gallery, while resin pockets are found above and below failed galleries and where
attacking beetles were rejected before establishing a gallery (Belluschi et al., 1965; Figure
4.7).
Resin pockets bear resemblance to the phenolic lesions formed in conifers inoculated
with blue stain fungi associated with bark beetles (diagrammed in Wong et al., 1977).
Douglas-fir beetle also carries pathogenic blue stain fungi and the phenolic lesion response
106

is probably common among related species, but no studies of lesion formation following
inoculation could be found in the published literature. With other host species, there is
rapid expansion of lesions in the second week following inoculation (Raffa & Smalley,
1988). The formation of necrotic lesions (resin pockets) is a common adaptive response in
conifers that occurs in advance of the hyphae of the blue stain fungus (Berryman, 1972).
The formation of a necrotic lesion follows both simple wounding and inoculation with blue
stain fungi, but the development is faster in the inoculated wounds (Wong et al., 1977).
Further lines of evidence suggesting Douglas-fir beetle as the agent responsible for
the scars and resin pockets include observations of brood galleries in the phloem with
adjacent scars in the xylem, and alignment of the scar dates with dates of recorded red and
green attack in adjacent stands ( Figure 4.3 and Figure 4.4). Beetles or parts of beetles
were also found and many samples contained partially healed scars in the year preceding
death, in some cases with identifiable frass (Figure 4.5). Scars were distinguished from
resin pockets by the presence of an open wound subsequently healed over (Figure 4.6).
Every cross section with one or more scars or resin pockets was scanned at 1200 DPI
and ring width measurement was performed in the program Windendro (Regent
Instruments 2014). Visual crossdating was confirmed using the program COFECHA
(Holmes, 1983). Previously crossdated samples from nearby sites were used to corroborate
the assignment of calendar years in samples where too few trees were present within the
site to build a master chronology. Measurement paths were initiated from the estimated
location of pith on trees left hollow by rot so as to provide an estimate of the basal area at
particular dates. Ringwidth measurements around rotten sections were allowed to deviate
from the master series used for crossdating in order to provide the best possible estimate of
107

the size of the tree at a particular date despite the rot obscuring the rings.

108

1 cm

Figure 4.3: Partially healed injury in the year prior to death of tree D09, photographed
before (top) and after (bottom) sanding of the bark to reveal resin saturation of the apparent
gallery. Photographs cropped but not manipulated with regards to color or contrast.
109

1 cm
Figure 4.4: Injury to tree D09 in the year prior to being killed by DFB. Photograph cropped
but not manipulated with regards to color or contrast. White arrows mark the bounds of the
apparent gallery in the bark.
110

A

B

C

D

Figure 4.5: Progression of scar formation: A: Scar with extensive traumatic resin duct
formation (top triangle); apparent resin-filled entry tunnel (middle triangle), and frass
(bottom triangle) from a tree at Esler Canyon (an early reconnaissance site not large
enough to fully sample). B: Deceased Douglas-fir beetle in current-year gallery, area
impregnated with resin, tree D10. C: 2013 scar with apparent gallery tree M23; 2014 ring
beginning to close over wound. D: 1990 scar, mirrored in the bark tree K20.
111



$

%

&

'


)LJXUH$W\SLFDOVFDULQWKHSURFHVVRIIRUPDWLRQ $ DW\SLFDOUHVLQSRFNHW % DQGDQ
HPSW\EHHWOHH[RVNHOHWRQVWXFNLQDFXUUHQW\HDUJDOOHU\DGMDFHQWWRDUHVLQSRFNHW &
DQG' 7KHNH\GLVWLQFWLRQEHWZHHQDVFDUDQGDSRFNHWLVWKDWWKHVFDULVKHDOLQJRYHUDQ
RSHQZRXQG HJWKHDUURZDW$ ZKLOHWKHZRRGKDVVLPSO\JURZQDURXQGWKHUHVLQ
SRFNHWZLWKRXWLQWHUUXSWLRQ HJWKHDUURZDW% 





Figure 4.7: Mountain pine beetle (black arrow) arrested within the bark of lodgepole pine
by defensive resins. A resin pocket would be identified in a cross-section above this point
while a scar would be found at the level of the failed gallery. Underexposure of original
image corrected in GIMP 2.0.
113

A

B

C

D

E

F

Figure 4.8: Six of eight scars in tree M23 in 2013; note the relative widths of the scar and
the resin-impregnated area in the phloem, the heavy production of resin ducts radiating
from all scars in 2013 (circled in E) and 2014 especially in C, and the discoloration of the
sapwood especially in A and F. Measurements of B were made on the opposite side of the
cookie; pencil slash is a reminder to not count the same scar twice.
114

4.3.2. Selection of ringwidth chronologies for growth release reconstruction
I subsampled the dataset previously used in Chapter 3 by selecting sites with minimal
mortality attributed to mountain pine beetle in the 21st century outbreak. A 500-meter
radius buffer (78.54 ha) was generated around the locations of all sites using ArcGIS.
Aerial overview survey records for mountain pine beetle from the years 2001 to 2013 were
merged into a pair of individual layers. The site buffer layer and merged aerial overview
surveys were overlaid using the intersect tool in ArcMap Desktop 10.2, resulting in a
dataset retaining all the records of severity within each buffered area.
British Columbia standards for aerial overview surveys were used to assign mortality
coefficients to the severity codes, such that Trace = 1%, Light = 10%, Moderate = 29%,
Severe = 40% and Very Severe = 60% mortality of Douglas-fir
(https://www.for.gov.bc.ca/hts/risc/pubs/teveg/foresthealth/assets/aerial.pdf accessed
2017). These mortality coefficients were multiplied by the area affected within the buffer,
and extended to the tree level by multiplying by a hypothetical 100 potentially susceptible
trees per hectare (coefficient*area*100 trees/ha). A pivot table was generated to calculate
the sum of the total affected trees at the site level. The process was repeated for Douglasfir beetle outbreak surveys over the same period for the increment core sites from Chapter
3 and the six sites sampled for this chapter.
Estimates of annual tree mortality attributed to Douglas-fir beetle were made by
assessing overview survey data from 1957 to 2013 based on severity codes and area
affected. Responsibility for comprehensive forest health surveys passed from the federal
government to the provincial government in 1999 following a brief period without data in
some regions of British Columbia. The provincial Aerial Overview Survey added two new
115

severity classes in 2004: “trace” indicating 1% or less current mortality and very severe
indicating >50% current mortality. The addition of the “trace” category can exaggerate
change over time as far more area is recorded under attack compared to older surveys. The
same severity coefficients as above were multiplied by the total area of each polygon, and
raw count data from helicopter-based surveys were added where available
(https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ accessed 2017).
4.3.3. Analysis: growth releases
Ring width data for the selected sites were exported to the R package TRADER
(Altman et al., 2014) for identification of growth releases, and a release event was defined
as a year in which the 20-year average basal area increment following the subject year
exceeded the 20-year average preceding the subject year by at least 50% (adapted from
Nowacki & Abrams, 1997). The selection and validation of these parameters is described
in Chapter 3. I used two metrics of growth release: initiation of growth release events,
defined as the year in which the maximum increase in growth rate occurred, and ongoing
release, defined as years in which the rate of increase was >50%. The latter metric is less
conservative as it may include years prior to the actual event due to the length of the
periods considered before and after each subject year, the rapid initial growth in young
trees, and ongoing growth increase many years after the event that opened the growing
space. The value of the metric is that it quantifies the duration of the period where growth
was increasing. The same growth release metrics calculated for the increment core data
were computed from the ring width series measured from the cross sections.

116

4.3.4. Analysis: scars and resin pockets
Using cross-sections collected from cull piles at the six harvested sites, only scars and
resin pockets more than 6.35cm from the pith were considered for analysis. This radius
corresponds to the minimum diameter considered in Douglas-fir beetle risk assessment
(Hood et al., 2007) and earlier studies of host-finding behavior (Rudinsky, 1966). While
smaller trees may be attacked, the 12.7cm diameter (6.35cm radius) was used to avoid
counting scars created by other, normally secondary bark beetle species that favor saplings
(Walters & McMullen, 1956; McMullen & Atkins, 1962). Every effort was made to avoid
counting small sealed fire scars as Douglas-fir beetle scars, and several dozen scars that
bore some resemblance to those described by Smith et al. (2016) were discarded from the
final analysis. Sealed fire scars may be formed by heat damage to the cambium that does
not cause an open wound, and I considered a scar to be suspect and discarded it if I found
an abundance of traumatic resin ducts after the year of the scar, a misshapen appearance
unlike the examples photographed here, or saturation of the wood beneath the scar with
resin (Smith et al., 2016).
Sampling for scars and resin pockets is expected to be most effective in cases where
the subject tree is showing a growth response to the death of a competitor in a pheromonemediated mass attack: a tree that is close enough to show a growth response to the death of
its competitor will fall within the zone of attraction created by the female beetle as she
processes the resins of the host tree. Johnson & Pettinger (1961) calculated the average
density of Douglas-fir beetle attacks within a radius of an attractive source (a windthrown
or felled tree), and found all trees within 1.2 meters of the source were attacked, at an
average rate of 33.3 attacks per square meter. At a range of 1.5–4.5 meters not all trees
117

were attacked, but the rate of attack among all trees averaged 5.6 attacks per square meter.
At a minimum diameter of 12.7cm and a thickness of 6cm, the smallest samples
considered in this study have a bark surface area of 239.3cm2, accounting for 479.7cm2
with two samples from each tree. If the average rate of attack is 5.6 per square meter, there
will be one attack per 0.178m2, or 1,780cm2, while a rate of 33.3 attacks per square meter
corresponds to one per 300 cm2. At these rates of attack, one sampled tree in four should
record an attack if it was within 1.5–4.5 meters of an attractive source, and evidence of
attack should be left in every tree within 1.2 meters. Extrapolating that rate of attack to a
range of diameters, any tree 48cm diameter, represented by two cross-sections each 6cm
thick, would be expected to reliably record evidence of that attack (Figure 4.9).

Attacks Expected per Sampled Tree

7
6
0-1.2m

5

1.5-4.5m

4
3
2
1
0

12

14

16

18

20

22

24 26 28 30 32 34 36 38
Diameter at Time of Attack (cm)

40

42

44

46

48

50

Figure 4.9: Rate of attack expected within the surface area represented by cross-sections of
various diameters based on the rates described by Johnson & Pettinger (1961) at a distance
of 0–1.2m and 1.5–4.5m from the sampled tree, assuming cookie thickness of 6cm and two
samples per tree.

118

The rate of attack in smaller trees within the 1.5–4.5m radius of an attractive tree
suggests that multiple trees would be required to record the event, though the odds are
>50% in trees over 24cm diameter. This estimate serves as a minimum, as other work has
found that logs with virgin female beetles attract many times more beetles than freshly cut
logs without females (maximum 105 (fresh-cut log) vs. 3,259 (log with virgin females);
Jantz & Rudinsky, 1966). In a pheromone-mediated mass attack (vs. the windthrown trees
studied by Johnson & Pettinger (1961)) that attracted ten times more beetles, a higher rate
of attack would be expected in adjacent trees, and these would be more likely to preserve
scars or resin pockets. Trees in the immediate vicinity of a windthrown individual are
likely to record the event if they are >24cm diameter, while trees close enough to record
growth release due to the death of a competitor in a mass-attack organized by pheromone
cues will almost certainly contain scars or resin pockets.
I produced a simple timeline at the site level of the total number of scars and resin
pockets per year. I also aggregated all six sites by dividing the number of scars or resin
pockets by the number of trees >12.7cm diameter in each year to account for changing
sample depth over time. Three methods of aggregating data among the six sites were used
to compare against previously published records of wildfire, drought, and mountain pine
beetle: years where multiple scars or resin pockets were found within a single tree, years in
which ten or more scars or resin pockets were found in the complete dataset, and years in
which three or more sites each had one or more scars or resin pockets.
4.4. Results
Surveys conducted by the Canadian Forest Insect and Disease Survey and British
Columbian Aerial Overview Survey and detailed helicopter surveys suggest periods of
119

Douglas-fir beetle infestation from 1961–1969, 1983–1985, 1989–1996, and 2000–2016
(Figure 4.10). Within the latter period, 2000–2009 and 2014–2016 are most likely distinct
events due to the reduction in mortality from 2010–2013. Mortality estimated from the
2016 survey is markedly higher than any preceding year, or the cumulative total of any of
the periods outlined in the 20th century. Estimated mortality in the “severe” class
constituted a small portion of the total in most years in the 21st century, while this class
was predominant in most years in the 20th century.
Every site recorded some level of mountain pine beetle activity, while fixed-wing
aircraft surveys identified Douglas-fir beetle mortality at 30 sites and detailed helicopter
surveys found patches killed in the vicinity of an additional nine sites (Table 4.2). Five
sites with the lowest recorded mountain pine beetle activity (BLU, EPI, GAR, SUG, WIL)
were selected for further analysis.
Mountain pine beetle accounted for more mortality than Douglas-fir beetle around all
the locations sampled specifically for this study, though site P had very low mountain pine
beetle activity compared to the others (Table 4.3).

120

1,000,000

600,000

Trees Killed

800,000

Helicopter
AoS Spot
Very Severe
Severe
Moderate
Light
Trace

400,000

0

1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015

200,000

Figure 4.10: Estimated number of trees killed by Douglas-fir beetle recorded by aerial
surveys since 1957. “Trace,” “Light,” “Moderate,” “Severe,” and “Very Severe” are codes
used in fixed-wing aerial surveys; “AoS Spot” refers to spot attacks identified in the fixedwing aerial survey, and “Helicopter” accounts for all trees tallied in helicopter-based
surveys (helicopter data 2007-2013).

121

Table 4.1: Estimated mortality within 500m buffer around increment core sites from 2001
to 2013. DFB = Douglas-fir beetle; MPB = mountain pine beetle. “% Total” columns
indicate the estimated percentage of the total number of hypothetical trees (n = 7,854)
affected. Rows highlighted in gray indicate the sites selected for further analysis.
Code
ABR
ACS
ALE
ALK
BCA
BIG
BLU
CAL
CAM
CHI
CTI
DCB
DOG
ENT
EPI
FUZ
GAR
HAN
HAW
JKY
KCR
LEE
LEN
MAB
MAQ
MAY
MCL
MEL
MON
NEE
NOJ
NRA
PTI
PUD
PVA
PYP
REN
RES
RIS
SOA
SUG
TWA
WEX
WHI
WIL
YUN

Elevation
(m)

1,133
1,024
959
954
935
1,261
871
1,248
1,087
906
818
947
1,136
910
1,262
1,068
1,239
1,329
1,306
1,136
831
1,031
1,268
945
931
1,127
679
939
1,127
1,070
1,016
970
1,005
967
786
1,107
988
953
1,259
645
658
839
1,013
857
677
996

DFB
Fixed
0
597
1330
9
0
73
25
809
769
94
0
394
149
161
442
0
0
2
9
981
1706
3329
432
2797
0
4
0
0
0
0
0
0
0
0
4033
0
1422
347
13
0
901
0
589
42
53
32

DFB
Heli
0
20
62
109
10
12
25
0
0
62
0
155
13
88
89
24
52
0
4
16
29
37
75
133
0
18
19
3
0
0
0
31
36
6
201
0
127
141
23
20
32
5
75
54
181
16

122

Total
DFB
0
617
1392
118
10
85
50
809
769
156
0
549
162
249
531
24
52
2
13
997
1735
3366
507
2930
0
22
19
3
0
0
0
31
36
6
4234
0
1549
488
36
20
933
5
664
96
234
48

Total
MPB
2465
9838
5542
1723
860
1208
432
8088
2397
1780
10403
783
1156
2991
120
1371
113
5172
8430
1917
3557
2468
5790
1305
4887
2289
2242
5843
4485
2900
3379
3323
4782
3136
872
7106
2717
5398
1927
1305
30
5380
4152
4682
524
2151

DFB %
Total
0.0%
7.9%
17.7%
1.5%
0.1%
1.1%
0.6%
10.3%
9.8%
2.0%
0.0%
7.0%
2.1%
3.2%
6.8%
0.3%
0.7%
0.0%
0.2%
12.7%
22.1%
42.9%
6.5%
37.3%
0.0%
0.3%
0.2%
0.0%
0.0%
0.0%
0.0%
0.4%
0.5%
0.1%
53.9%
0.0%
19.7%
6.2%
0.5%
0.3%
11.9%
0.1%
8.4%
1.2%
3.0%
0.6%

MPB %
Total
31.4%
125.3%
70.6%
21.9%
10.9%
15.4%
5.5%
103.0%
30.5%
22.7%
132.5%
10.0%
14.7%
38.1%
1.5%
17.5%
1.4%
65.8%
107.3%
24.4%
45.3%
31.4%
73.7%
16.6%
62.2%
29.1%
28.5%
74.4%
57.1%
36.9%
43.0%
42.3%
60.9%
39.9%
11.1%
90.5%
34.6%
68.7%
24.5%
16.6%
0.4%
68.5%
52.9%
59.6%
6.7%
27.4%

Table 4.2: Summary tabulations of the preceding table: estimated mortality within the
500m buffer around increment core sites from 2001 to 2013. DFB = Douglas-fir beetle;
MPB = mountain pine beetle. “% Total” columns indicate the estimated percentage of the
total number of hypothetical trees (n = 7,854) affected.

Count
Average
% Total
min
max

Elevation
(m)
1,005
645
1,329

DFB
Fixed
30
468
6.0%
0
4033

DFB
Heli
36
44
0.6%
0
201

123

Total
DFB
39
512
6.5%
0
4234

Total
MPB
46
3335
42.5%
30
10403

DFB %
Total

MPB %
Total

7%

42%

0.0%
53.9%

0.4%
132.5%

Table 4.3: Estimated mortality of trees within the 500m buffer around cookie collection
sites. DFB = Douglas-fir beetle; MPB = mountain pine beetle. “% Total” columns indicate
the estimated percentage of the total number of hypothetical trees (n = 7,854) affected.
Site Code

Elevation
(m)

DFB
Fixed

DFB
Heli

Total
DFB

Total
MPB

DFB %
Total

MPB %
Total

D

1,303

71

25

96

1799

1.2%

22.9%

H

1,328

0

19

19

6935

0.2%

88.3%

K
M
P
W
Count
Average
% Total
min
max

967
1,000
897
1,252

72
831
63
0
4
173
2.2%
0
831

16
39
41
119
6
43
0.5%
16
119

88
870
104
119
6
216
2.8%
19
870

2674
3345
311
1370
6
2739
34.9%
311
6935

1.1%
11.1%
1.3%
1.5%

34.0%
42.6%
4.0%
17.4%

0.2%
11.1%

4.0%
88.3%

1,125
897
1,328

124

The first growth release at site BLU was identified in 1760, during the initiation
phase of the population under study, at a time when 100% of trees were accelerating their
growth rate (Figure 4.11). Initiation of a growth release was identified in one tree in 1777,
one in 1801 and three in 1809, at a time when 75% of trees were in accelerated growth
phases. Two trees initiated growth release in 1847 and 1859, while less than 20% of trees
were growing faster. Several trees initiated growth release between 1870 and 1910, while
30% of trees were accelerating their growth rate. Growth releases were identified in one
tree in a number of years from 1924–1945, and the proportion of trees continuing to
increase in growth rate increased over that period. Site BLU was known to have been
harvested in the 1960s, and a number of the trees record a growth release at that time.

125

Severe(200%)
Initiation%

-40%
-60%
-80%
-100%

15
10
5

1990

1910

1890

1870

1850

1830

1810

1790

Ongoing Release

-20%

1770

1750

0%

1970

Sample Depth

20%

1950

40%

Moderate(100%)

1930

60%

Light (50%)

Initiating Release

80%

20

Sample Depth

100%

0
-5
-10
-15
-20

Figure 4.11: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site BLU. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.

126

Growth release at site EPI occurred sporadically in no more than one tree per year
from 1736 to 1796, except 1786 when two trees recorded release (Figure 4.12). Despite the
continued initiation of new trees, only a small fraction experienced an accelerated growth
rate prior to 1790. A total of 88% of trees had accelerated growth rates from 1800–1805,
and nine of the sixteen trees present at the time recorded a growth release in that period,
and an additional four had recorded a release by 1911. A general increase in growth
occurred from 1840–1860, and eight trees recorded growth release during that time.
Growth rates again accelerated in the 1930s, peaking at 55% of established trees in 1938
and 1939, while 12 of the 20 trees established by that time recorded growth release
between 1936 and 1941.

127

100%

40%

Moderate(100%)
Severe(200%)
Initiation%
Sample Depth

Sample Depth

60%

Initiating Release

80%

20

Light (50%)

-40%
-60%
-80%
-100%

2000

1980

1960

1940

1920

1900

1880

1860

1840

1820

1800

1780

Ongoing Release

-20%

1760

1720

0%

1740

20%

15

10
5
0
-5
-10
-15
-20

Figure 4.12: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site EPI. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.

128

Initiation of trees at site GAR was relatively abrupt, beginning around 1600 and
completed by 1706 (Figure 4.13). Peaks in growth release and increasing growth rates
occurred ca. 1650 and 1690 during this period of initiation. Every tree sampled at this site
had accelerated growth from 1809–1811, while 18 of 20 trees recorded growth release
between 1800 and 1810 and an additional tree recorded a release in 1812. Ongoing growth
acceleration was found only in a small proportion of trees following this event, while two
trees recorded initiation of growth release in 1844, 1861, 1867, 1874, and 1876.
100%

40%

Moderate(100%)
Severe(200%)
Initiation%
Sample Depth

Sample Depth

60%

Initiating Release

80%

20

Light (50%)

-20%
-40%
-60%
-80%

-100%

Ongoing Release

0%

1610
1630
1650
1670
1690
1710
1730
1750
1770
1790
1810
1830
1850
1870
1890
1910
1930
1950
1970
1990

20%

15
10
5
0
-5
-10
-15

-20

Figure 4.13: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site GAR. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.

129

Site SUG records several periods of increasing growth corresponding to events
meeting criteria defining initiation of growth release, beginning ca. 1660, 1680, 1740,
1770, 1800, 1880, 1900, 1940, and 1980 (Figure 4.14). The proportion of trees recording
accelerated growth during these events generally decreased with time, as did the number of
trees recording initiations.
100%

40%

Moderate(100%)
Severe(200%)
Initiation%
Sample Depth

Sample Depth

60%

Light (50%)

Initiating Release

80%

20

-20%
-40%
-60%
-80%
-100%

Ongoing Release

0%

1600
1620
1640
1660
1680
1700
1720
1740
1760
1780
1800
1820
1840
1860
1880
1900
1920
1940
1960
1980
2000

20%

15
10
5
0

-5
-10
-15
-20

Figure 4.14: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site SUG. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.

130

Periods of growth release while stand initiation was ongoing at site WIL occurred
beginning around 1675 and 1740, while growth increase affecting more than 50% of the
trees at the site occurred ca. 1770 and 1810 after all sampled trees were established (Figure
4.15). Growth acceleration in a fraction of the trees at this site was ongoing from 1840
through 1900, and a number of trees recorded the initiation of growth release during that
period. Only a few trees recorded growth release or accelerating growth during the 20th
century.

20%

-40%
-60%

-80%

1980

1960

1880

1860

1840

1820

1800

1780

1760

1740

1720

1700

1680

1660

1640

-20%

Ongoing Release

0%

Sample Depth

Sample Depth

Initiation%

2000

Severe(200%)

1940

40%

Moderate(100%)

1920

60%

15

Light (50%)

1900

80%

20
Initiating Release

100%

10

5
0
-5
-10
-15

-100%

-20

Figure 4.15: Initiation of growth release and ongoing acceleration of growth in Douglas-fir
at site WIL. “Light,” “Medium,” and “Severe” refer to 50, 100, and 200% rate of increase.

131

The first resin pocket at site D was dated in 1718, followed by two in 1746 and one
in 1747 (Figure 4.16). Increased frequency of resin pockets was observed in the early 20th
century, and scar frequency peaked in 1931. Resin pockets were relatively frequent in the
1967–1974 and again in 1987–1992. Some scarring and resin pockets dated to the first
decade of the 21st century, and many scars were found in 2013 and 2014 before the stand
was salvage-logged. Growth releases were recorded throughout the period where new trees
were reaching sampling height, with peaks in 1760, 1800–1812, 1860, the 1930s/early
1940s, and the late 1970s/early 1980s.
10

18

8
6

15

4

10

2

5

0

0

-2

-5

-4

-10

Trees >12.7cm DBH

Resin Pockets and Scars

25

Growth Releases

Scars (D)
Pockets (D)
Growth Releases (D)
Trees >12.7cm

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

20

Figure 4.16: Scars, resin pockets, and growth releases identified in cross sections at Site D.
Total count of scars (red) labeled where the bar exceeds the standard Y axis.

132

Site H recorded the first resin pocket in 1716, but did not record a year with more
than one scar or resin pocket until 1909 (Figure 4.17). Six scars and one resin pocket were
found in 1920, and one scar in 1921. Four resin pockets were found in 1976. One pocket
and four scars were counted in 1999, three scars in 2004, and two in 2010. The final years
of the series included 11 scars and one pocket in 2013 and one scar in the process of
healing in 2014. Growth releases peaked after 1800, in the 1890s, and the 1930s/early
1940s.

10

11

8
6

15

4

10

2

5

0

0

-2

-5

-4

-10

Trees >12.7cm DBH

Resin Pockets and Scars

25

Growth Releases

Scars (H)
Pockets (H)
Growth Releases (H)
Trees >12.7cm

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

20

Figure 4.17: Scars, resin pockets, and growth releases identified in cross sections at Site H.
Total count of scars (red) labeled where the bar exceeds the standard Y axis.

133

The first resin pocket at site K was recorded in 1732 (Figure 4.18). Three scars
were counted in 1797, two scars and two resin pockets in 1798, two resin pockets in 1799,
and six in 1801. Peaks during the 19th century ocurred in 1818, 1824, 1827/1828, 1839,
1841, 1843, 1849/1850, 1868, 1873/1874, 1884, and 1898. Three resin pockets were tallied
in 1935, 1947, and 1952, two in 1964, five in 1965, six resin pockets plus one scar in 1969,
and two scars and one resin pocket in 1971. A peak in scarring occurred in 1990 with a
total of 15, followed by five resin pockets in 1991 and five again in 1997. Six scars and
three pockets were counted in 2009, followed by 11 scars in 2013 and four scars and one
pocket in 2014 before the stand was harvested. Releases peaked in the 1740s, around 1810,
and in the 1930s/early 1940s.
10

15

25

8
6

15

4

10

2

5

0

0

-2

-5

-4

-10

Trees >12.7cm DBH

Resin Pockets and Scars

11

Growth Releases

Scars (K)
Pockets (K)
Growth Releases (K)
Trees >12.7cm

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

20

Figure 4.18: Scars, resin pockets, and growth releases identified in cross sections at Site K.
Total count of scars (red) labeled where the bar exceeds the standard Y axis.

134

Site M stands out for recording a large number of scars in 1770, two years after the
first scar and resin pocket were recorded at the site in 1768 (Figure 4.19). Two resin
pockets were recorded in 1836, and one scar and one resin pocket were found in 1854;
throughout the rest of the 19th century no more than one scar or resin pocket was counted
in any single year. Four resin pockets were counted in 1969, two in 1972, and one each in
1971/1973. Scarring in the early 1990s peaked with 22 scars in 1993. Sporadic activity
persisted until 2013, when 26 scars were recorded, and 2014, which contained five scars.
Growth releases peaked in the 1770s, following 1800, the 1930s/1940s, and around 1980.
15

25

6

15

4

10

2

5

0

0

-2

-5

-4

-10

Trees >12.7cm DBH

8

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

Resin Pockets and Scars

Scars (M)
Pockets (M)
Growth Releases (M)
Trees >12.7cm

Growth Releases

10

20

Figure 4.19: Scars, resin pockets, and growth releases identified in cross sections at Site M.
Total count of scars (red) labeled where the bar exceeds the standard Y axis.

135

The first scar at site P was counted in 1706 and the second in 1760; two resin
pockets and three scars were found in 1794 followed by one scar in 1795 (Figure 4.20).
Four scars were tallied in 1802, followed by one each in 1803, 1804, and 1806. Eight scars
were located in 1866. Five pockets were found in 1919, four in 1938 and three in 1941.
Scars or resin pockets were found in almost every year from 1966–1988, with peaks in
1967, 1975, 1980, 1986, and 1988. Later peaks in frequency occurred 1992–1994 and
2001/2002, but only a few scars and resin pockets were found in the final years of the
series. Growth releases were frequent around 1860, the early 20th century, and in the 1930s.
10

25

6

15

4

10

2

5

0

0

-2

-5

-4

-10

Trees >12.7cm DBH

Resin Pockets and Scars

8

Growth Releases

Scars (P)
Pockets (P)
Growth Releases (P)
Trees >12.7cm

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

20

Figure 4.20: Scars, resin pockets, and growth releases identified in cross sections at Site P

136

Site W contains the earliest recorded resin pockets: a pair in 1695 (Figure 4.21). An
additional four were tallied in 1750 and 16 scars were found in 1800. The most resin
pockets were found in 1992 and these 12 were followed by several years in which one scar
or resin pocket was identified. Five scars were dated to 2010, but only two in 2013 and
none in 2014. Eight growth releases were identified following 1800, another five around
1860, and 16 between 1920 and 1940.
16

12

8
6

15

4

10

2

5

0

0

-2

-5

-4

-10

Trees >12.7cm DBH

Resin Pockets and Scars

Scars (W)
Pockets (W)
Growth Releases (D)
Trees >12.7cm

25

Growth Releases

10

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

20

Figure 4.21: Scars, resin pockets, and growth releases identified in cross sections at Site W.
Total count of scars (red) and resin pockets (black) labeled where the bar exceeds the
standard Y axis.

137


3HDNVLQVFDUDQGUHVLQSRFNHWIUHTXHQF\DFURVVWKHVL[VLWHVFRPELQHGRFFXUUHGLQ
WKHV±WKHVDQGV±WKHVDQG± )LJXUH
 6FDUVDQGUHVLQSRFNHWVZHUHIUHTXHQWIURPWKHVWKURXJKZLWKSHDNVLQ
±DQGDJDLQLQWKHVWDQGV¶ILQDO\HDUV





6FDU5DWLR
3RFNHW5DWLR
,QFUHDVLQJ*URZWK



,QFUHDVLQJ*URZWK5DWH!



5DWLR6FDUV7UHHV!FP























































)LJXUH6FDUVDQGUHVLQSRFNHWVDWDOOVLWHVGLVSOD\HGDVDUDWLRWRWKHWRWDOQXPEHURI
'RXJODVILU!FPGLDPHWHUDQGWKHSHUFHQWDJHRI'RXJODVILU RIDQ\GLDPHWHU ZLWK
RQJRLQJJURZWKUHOHDVH5DWLRODEHOHGIRUZKHUHEDUH[FHHGHGWKH<D[LV










Periods of increased growth release across the six sites occurred 1720–1760, 1770–
1780, 1800–1820, 1840–1880, and 1920–1940, and particularly high rates of growth
release occurred in 1749, 1760, 1772, 1800, 1859/1860, and 1938 (Figure 4.23). The total
percentage of the sample initiating growth release within a moving 11-year window
reflects the percentage of trees increasing their rate of growth, though slightly fewer trees
are meeting the release initiation criteria at any given time.

Initiation of Release

60%

40%

Site D
Site H
Site K
Site M
Site P
Site W
11-year Total

20%

0%

Ongoing Release

-20%

-60%

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990

-40%

Figure 4.23: Sampled Douglas-fir trees recording initiation of growth release (above 0%
line) and ongoing release (below 0% line) as a percentage of total sample size in each year,
displayed by site, with 11-year running total of initiations (total percentage, subject year
plus preceding and following five-year periods).
139

Multiple scars or resin pockets were recorded within a single tree in a given year
114 times between 1695 and 2014 (Figure 4.24).
0.1

0.08
0.07

0.09
0.08

One Tree

0.07

0.06

0.06

0.05

0.05

0.04

0.04

0.03

0.03

0.02

0.02

0.01

0.01

0

0

1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

Proportion of Total Sample (trees >12.7cm)

0.09

0.1

Site D
Site H
Site K
Site M
Site P
Site W

Figure 4.24: Years where one or more trees recorded multiple scars or resin pockets within
a single year. Gray dashed line indicates the proportion accounted for by a single tree.

140

All drought metrics (Cook & Krusic, 2004; Case & Macdonald, 2003; Daniels &
Watson, 2003; Hart et al., 2010) indicate that the last decade of the 18th century was drier
than average (Figure 4.25), consistent with what has previously been regarded as a
prolonged regional drought (Wolfe et al., 2001). Multiple fire scars in Douglas-fir were
dated by Daniels & Watson (2003) to 1791, 1797, and 1800, and one scar was dated to
1802 (the two fire chronologies constructed from lodgepole pine do not extend to this
period). Evidence of Douglas-fir beetle presence from 1790 through 1802 includes several
years with multiple scars or resin pockets within a given year, one year with 10 or more
scars or resin pockets (1800), and one year in which three or more sites recorded one or
more scars or resin pockets (1794; Figure 4.25). An increased frequency of years with
multiple growth releases began at the turn of the 19th century and continued over the next
two decades.
Multiple sources indicate dry conditions in the study area from the mid–1830s
through the late 1840s; three or more sites recorded scars or resin pockets in four of the
years within that period and multiple scars within the same tree were found in four of these
years (Figure 4.25). Multiple Douglas-fir fire scars were found in three years, and two
years had one fire scar (Daniels & Watson, 2003). Hawkes et al. (2004) found fire scars in
ten or more lodgepole pine during one year within this period, and Axelson et al. (2010)
recorded fire scars in two of their three lodgepole pine sites. Growth release was recorded
in four or more trees during two years within this period, and two or three trees showed
growth release in two other years.
Dry conditions in the 1920–1940 period are indicated by all precipitation
reconstructions, though the pJJ reconstructions presented by Daniels & Watson (2003)
141

suggest a longer period of low precipitation beginning in the late 1890s (Figure 4.25).
Cook & Krusic’s (2004) PDSI reconstruction point 30 indicates several years of negative
PDSI in the 1890s and late in the decade of 1900, but does not support a continuous period
of drought. Erickson (1992) records a widespread outbreak of mountain pine beetle in the
Chilcotin region from 1930–1935, also described by Alfaro et al. (2010) on the basis of
scars left by failed attacks and growth releases in surviving trees. Growth releases occurred
frequently in the 1930s, and 10 or more fire scars were found in 1922 and 1926 (Hawkes et
al., 2004). Evidence for Douglas-fir beetle presence in this period includes three or more
sites with at least one scar or resin pocket in 1920, 1921, 1923, 1926, 1931, 1935, 1937,
and 1938, and multiple scars or resin pockets were found within a single tree in 1919,
1920, 1921, 1923, 1924, 1931, 1935, 1936, 1937, and 1938.
No consistency is found among the reconstructions of drought after 1950 (Figure
4.25), and the frequency of scars attributed to wildfire drops precipitously over the course
of the 20th century as human settlements expanded and wildfire fighting became more
effective (Daniels & Watson, 2003). A severe outbreak of mountain pine beetle was
surveyed in the late 1970s and 1980s (Erickson, 1992) and was described by both growth
release and scars in lodgepole pine (Alfaro et al., 2010). Scars or resin pockets in Douglasfir identified in this study were found at multiple sites in the majority of years in the latter
half of the 20th century, while the number of years with 10 or more scars is more limited.
Aerial surveys suggest that Douglas-fir beetle mortality peaked in 1966, a year when three
or more sites were affected. Multiple scars or resin pockets were found within one tree in
1967, and every year from 1969 through 1974.
The surveyed Douglas-fir beetle mortality in the late 1980s and early 1990s
142

coincides with a number of scars and resin pockets, including multiple scars and resin
pockets within a single tree in every year from 1983 through 1994 (except 1989), and 10 or
more scars or resin pockets in 1986, 1987, 1988, 1990, 1992, and 1993. Three or more sites
were affected in every year from 1983 through 1993 except from 1986 and 1989. Lower
streamflow in the Chilko River was reconstructed from 1979 through 1988 (Hart et al.,
2010), and 1988 and 1989 are two of the driest years in the instrumental record (Chapter
2). Mountain pine beetle populations were rising at the time, but did not reach epidemic
levels until the 21st century (Shore et al., 2004).

143

1700

1800

1900

2000

1700

1800

1900

2000

Growth Release
Multiple Scars One T ree

DFB

10+ Scars or Pockets
3+ Sites Affected
DFB Mortality Estimate
Cook & Krusic 2004 PDSI
Daniels & Watson 2003 JJA

Drought Daniels & Watson 2003 pJJ
Case & MacDonald 2003

Hart, Smith and Clague 2010
Daniels & Watson 2003 Fire

Fire

Harvey 2017 Churn Creek >25%
Hawkes et al. 2004 Fire
Axelson et al. 2010 Fire

MPB

Erickson 1992 MBP Survey
Alfaro et al. 2010 MPB Scars

Figure 4.25: Mountain pine beetle, wildfire, and drought histories inferred from tree ring analysis and annual insect surveys
compared to scars attributed to Douglas-fir beetle attack. “Growth Release” is coded gray when two or three trees and black for four
or more trees were recording growth release in a given year. “DFB Mortality Estimate” (see Figure 4.10) is coded gray when >5,000
and black when >50,000 trees were estimated to have been killed. Cook & Krusic’s (2004) PDSI reconstructions are coded gray
when between 0 and -0.5, and black when below -0.5. Daniels & Watson JJA = reconstructed June, July and August precipitation
total at their site near Williams Lake; Daniels & Watson pJJ = reconstructed previous-July through current-June precipitation total at
their site near Williams Lake. All records under the “Drought” heading except Cook’s PDSI reconstruction highlight only prolonged
periods of low precipitation reconstruction values. Daniels & Watson 2003 fire records coded gray when only one tree at one site
recorded a fire and black when two or more trees were affected at a single site or trees at two or more sites were affected. Harvey
2017 indicates years in which >25% of trees at the Churn Creek Protected Area study site recorded a wildfire. Hawkes et al. 2004
coded black if 10 or more trees recorded a fire scar. Axelson et al. 2010 coded black if two or more sites recorded a fire and gray if
fire scars were found at only one site. Erickson 1992 coded gray for localized infestations affecting <10,000 trees and black for
larger infestations. Alfaro et al. 2010 coded black if two or more scars were attributed to beetle attacks in that year.
144

4.5. Discussion
This study corroborates previous work by Belluschi et al. (1965) attributing the scars
and resin pockets found in cross-sections of Douglas-fir to failed attacks by the Douglas-fir
beetle. In several cases intact or partial Douglas-fir beetles were found within galleries
sealed off by defensive resins and there were many examples of apparently failed massattack in years where local beetle populations are known to have been high. The longevity
of interior Douglas-fir allowed the development of a chronology of failed attacks into the
18th century (not counting the pair of resin pockets dated to 1695), exceeding a similar
record established for mountain pine beetle using scars in lodgepole pine in the region
(Hawkes et al., 2004) and on par with a chronology of resin pockets attributed to spruce
beetle in white spruce near Hudson Bay, Canada (Caccianiga et al., 2008).
The prolonged regional drought at the end of the 18th century has previously been
established as a critical moment in the history of stands across this million-hectare study
area due to widespread growth release implying mortality of competitors between 1790
and 1800 (Chapter 3). Evidence for the regional presence of Douglas-fir beetle populations
during this period is strong, though the frequency of scars and resin pockets is substantially
less than the frequency of trees experiencing growth release at the beginning of the 19th
century. Growth release in a survivor implies proximity to a tree that has died. In the case
of trees killed by beetles, the survivor would be within the zone of attraction generated by
beetle pheromones, suggesting that the survivor should also show signs of unsuccessful
attacks. The difference between number of unsuccessfully attacked trees and number of
trees experiencing growth release suggests another cause of mortality was driving the

145

widespread growth release in the early 1800s. This logic extends throughout the period
under study, where growth releases affecting substantial fractions of trees at a majority of
sites do not correspond to any peaks in scar or resin pocket frequency. The long-term
perspective provided by this work suggests that Douglas-fir beetle is more responsive to,
than responsible for, the mortality patterns affecting the interior Douglas-fir forest type at
landscape scales in central British Columbia.
The frequency of years with multiple scars or resin pockets on a single tree appears to
have been increasing over the past century, as does the number of years where three or
more sites were affected. There are several explanations for this phenomenon which may
interact additively or synergistically. First, there has been an increase in temperature over
the past 50 years with a statistically significant trend (Dawson et al., 2008), which may
affect the development time of the beetle. An increase of two degrees Celsius, depending
on the starting temperature, may allow the beetle to progress from egg to flight-ready adult
a week to a month faster (Vité & Rudinsky, 1957). The observed increase in average
temperature has been of this magnitude from 1950–2001, with Quesnel Airport increasing
at a rate of 0.34°C and Tatlayoko Lake increasing at 0.22°C per decade, implying a 1.1–
1.7°C increase over the latter half of the 20th century.
The second explanation is a demographic wave in the host population, resulting from
disturbances that more or less synchronized the development of stands across this
landscape, causing a large proportion of forested areas to become vulnerable over the past
several decades. A synchronous growth release evident in the 1930s has been attributed to
the widespread mountain pine beetle outbreak affecting the region at the time and to a
number of natural and accidental wildfires (Alfaro et al., 2010). A disturbance or series of
146

disturbances at the end of the 18th century had a similar effect on tree growth (Chapter 3),
and all six sites show a growth response following both of these periods. Ingrowth and
crown expansion would have filled many of the gaps created by the 1930s disturbances by
the latter half of the 20th century, and most mature survivors would have been of a
vulnerable size after decades or centuries of free growth. In the sampled stands, very few
trees recorded growth release after 1950. The ready supply of mature trees facing heavy
competition from below (Wong & Iverson, 2004) would facilitate the steady growth of
endemic Douglas-fir beetle populations, which had outbreaks in the late 1980s and early
1990s, mid to late 2000s, and from 2013 to present.
Finally, the transition from a fire-driven disturbance regime to a mountain pine beetledriven disturbance regime described by Axelson et al. (2010) in lodgepole pine forests in
the Chilcotin region may also be occurring in the region’s Douglas-fir forests. Fire
frequencies are known to have declined since the early 1900s throughout the region
(Daniels & Watson, 2003, Harvey, 2017), and inter-tree competition has increased as a
result (Wong & Iverson, 2004). An abundance of mature trees that might otherwise have
died in natural wildfires may be fueling increased Douglas-fir beetle outbreak intensity in
the same way that mountain pine beetle has been facilitated by abundant mature and overmature host stands.
The sites studied were selected for their lack of mountain pine beetle infestation in the
current outbreak due to the lack of available lodgepole pine in the stands. Elimination of
other disturbance agents as the origin of the growth release, leaving only mountain pine
beetle, suggests that disturbance may have fundamentally altered the species composition
of low-elevation stands. Stands in the Chilcotin and Fraser River valleys are presently
147

dominated by the Interior Douglas-fir very dry-mild ecosystem subzone, which typically
has very little lodgepole pine except on glaciofluvial terraces (Steen & Coupé, 1997).
Lower temperatures during the Little Ice Age (Bradley & Jonest, 1993) could have
permitted the Interior Douglas-fir dry-cool ecosystem subzone (Steen & Coupé, 1997) to
extend further downslope. This subzone commonly includes lodgepole pine and was
heavily impacted by the mountain pine beetle outbreaks of the 1930s, 1970s, and 2000s
(Alfaro et al., 2010; Figure 1.5). An outbreak coming at a time of climatic change may
have led to a regeneration failure in lodgepole pine at low elevations, allowing the current
Douglas-fir dominated stand structure to become established.
Average summer temperatures in North America and the Northern Hemisphere in
general are estimated to have been 0.5–1.5°C cooler during the 16th and 17th centuries
compared to the 1860–1959 baseline (Bradley & Jonest, 1993). Temperatures rose to this
baseline between 1760 and 1800 (Bradley & Jonest, 1993), coincident with the widespread
growth release of Douglas-fir. Tree species range expansions have been documented at
alpine treelines in North America, Europe, and Asia following recovery from local Little
Ice Age temperature depressions correlating to the rise in temperature (Lescop-sinclair &
Payette, 1995; Bogaert et al., 2010; Kharuk et al., 2010), coincident with growth
accelerations in trees within Little Ice Age treelines. Ecosystem change in the valleys of
the Chilcotin and Fraser Rivers may parallel these alpine dynamics, though only fineresolution sediment core analysis could confirm the species shift (e.g. Morris et al., 2017).

148

4.6. Conclusion
The long-term historical baseline of Douglas-fir beetle activity established in this study
is consistent with Walters (1956), Johnson & Belluschi (1969), and McGregor (1975), who
identified DFB outbreaks following droughts, windthrow, and other predisposing events
but not driving mortality or stand dynamics with the same regularity or magnitude as
mountain pine beetle or wildfire. There is no evidence in these data that any historical
outbreak exceeded the magnitude of the 21st century outbreaks. This permits the use of
2000–2015 aerial survey data as a baseline for comparison of 2016 data, which were all
collected using consistent survey methods. The magnitude of mortality estimated from
2016 survey data exceeds that of any other year in the 21st century by a wide margin, and
suggests that the Douglas-fir beetle population is higher than it has been at any point in
recent history. Projections of continued regional warming (Dawson et al., 2008), increased
stagnation of weather systems due to atmospheric circulation anomalies (Mann et al.,
2017), and accumulation of Douglas-fir regeneration providing competition from below
(Wong & Iverson, 2004) suggest that outbreaks of DFB in the future may exceed historical
limits due to changes in stand structure and climate

149

5. Reconsidering western spruce budworm outbreak history inferred from the
annual growth of host and non-host tree species in central British Columbia
5.1. Abstract
Western spruce budworm is a common pest of Douglas-fir in western North
America, and its historical influence in southern British Columbia is well documented.
Outbreak history at the northernmost extent of its present range in British Columbia is less
clear. The discrepancy among studies describing outbreak history may stem from the
dendrochronological approach used, where the mean ringwidth index of a population of
non-host ponderosa pine is subtracted from the ringwidth index of individual host
Douglas-fir to compensate for the influence of regional climate. These methods are applied
to a collection of Douglas-fir ringwidth chronologies from 45 sites distributed throughout
and beyond the range of the recent 800,000-hectare outbreak, using a ponderosa pine
chronology from an adjacent region. Periods of relatively low growth in Douglas-fir are
identified matching those previously identified as outbreaks, but with a synchrony and
spatial distribution that suggests abiotic mechanisms. The growth differential of Douglasfir relative to ponderosa pine that could be interpreted as widespread outbreaks in the 20th
century and before may be due to differences in precipitation between regions favoring the
ponderosa pine population.

150

5.2. Introduction
Western spruce budworm (WSB; Choristoneura freemani = C. occidentalis) is a
defoliating insect native to western North America, where periodic outbreaks affect
populations of Douglas-fir (Pseudotsuga menziesii), true firs (Abies spp.), western larch
(Larix occidentalis) and white spruce (Picea glauca) at landscape scales (Nealis, 2016).
Adult western spruce budworm moths emerge in July and August and disperse after
depositing a portion of their eggs on needles remaining at their original site (Nealis, 2016).
They then fly to seek new host material before laying the remainder of their eggs on
needles at the new site. The eggs are laid and hatch in the same year. Attraction of the
freshly-hatched larvae to light leads to overcrowding on the branch tips where the eggs
were laid, causing many larvae to disperse on silken threads (Wellington & Henson, 1947).
This dispersal may cause larvae to fall to lower branches or lower canopy horizons, or the
larvae may be carried by wind above the canopy to new locations (Wellington & Henson,
1947). When the larvae reach an appropriately rough surface after their initial dispersal,
they spin hibernaculae, moult, and overwinter (Nealis, 2016). Emerging second-instar
larvae disperse again, moving to branch tips and potentially falling to lower levels or being
dispersed on silken threads during windy conditions (Wellington & Henson, 1947). After
this second dispersal, the larvae feed by mining into old needles or the swelling buds
(Nealis, 2016).
Outbreaks of WSB are driven by temperature trends that promote synchronous
budburst (accumulation of degree days after a certain day length) and larval emergence
(degree days without regard to day length) (Thomson et al., 1984; Thomson & Benton,
2007). The larvae must emerge early enough to enter the buds as they begin to swell, thus
151

accessing the foliage while it is still succulent and lacking chemical defenses, without
being too early, which could expose the larvae to cold temperatures and starvation (Nealis,
2016). The ideal timing of WSB larval emergence in southern British Columbia has been
calculated at 18 days prior to budburst (Thomson et al., 1984). The range of WSB is
expected to expand to higher elevations and latitudes under a warmer climate scenario as
this temperature shift favors optimal phenological synchrony between larval emergence
and budburst in these areas (Marciniak, 2015). A widespread outbreak at the northernmost
known extent of the species range in central British Columbia may represent an example of
substantial range expansion in the 21st century. This region is known as the CaribooChilcotin (alternatively the Cariboo forest district) and is located in the Interior Plateau of
British Columbia (Figure 5.3).
Western spruce budworm outbreaks are common throughout southern British
Columbia in the 20th century, affecting Douglas-fir forests most extensively but also
impacting true first, larch, and spruce (Maclauchlan et al., 2006) . Severe outbreaks
extended to approximately 51° North in the vicinity of Lillooet during the 20th century
(Erickson, 1987). North of Lillooet, the Cariboo-Chilcotin region has had no recorded
infestations except for several patches of defoliation recorded in 1975 near Riske Creek
and in 1987 near Quesnel Lake (Erickson, 1992; Figure 5.3). This northern region is
ecologically distinct from interior Douglas-fir (Pseudotsuga menziesii var. glauca) forests
in southern British Columbia and the Rocky Mountains in the United States. Ponderosa
pine (Pinus ponderosa) is absent while lodgepole pine (Pinus contorta) is abundant, and
outbreaks of the mountain pine beetle (Dendroctonus ponderosae) are especially common
and severe (Alfaro et al., 2010).
152

Ponderosa pine is the typical associate of Douglas-fir on dry aspects in Rocky
Mountain forests in Idaho and adjacent British Columbia, while pure ponderosa pine
commonly occupies a belt between mixed Douglas-fir/ponderosa pine forests and
grasslands where the retention of winter snowpack is insufficient for the survival of
Douglas-fir seedlings (Daubenmire & Daubenmire, 1968). Only a small population of
ponderosa pine is found in the Cariboo-Chilcotin, in the lowest elevation areas in the
Fraser River valley at the southern limit of the study area (Figure 5.1).
Lodgepole pine in the Cariboo-Chilcotin region is less drought resistant than Douglasfir (Brix, 1979). Douglas-fir is found in pure or nearly pure stands in the dry valley
bottoms and slopes adjacent to grasslands, while lodgepole pine becomes more abundant
with elevation and becomes the dominant species in the uplands, where it is often mixed
with spruce. Lodgepole pine in the Cariboo-Chilcotin has one of the most extensive
records of mountain pine beetle infestation in the province (Alfaro et al., 2010) and the
influence of these outbreaks on the growth of Douglas-fir is substantial where the two
species co-occur (Hawkes et al., 2004). Mature lodgepole pine was almost completely
eliminated from mixed stands by the mountain pine beetle outbreak of the 2000s, and had
previously experienced widespread mortality in the outbreaks of the 1980s and 1930s
(Erickson, 1992).

153

Figure 5.1: Present range of ponderosa pine, lodgepole pine, and Douglas-fir in southern
portion of the study area and adjacent regions, overlaid by Chasm Provincial Park and the
regional centers of Lillooet and Kamloops. BC VRI data used under Open Government
License 2.0.

154

The extent of the 800,000-hectare WSB outbreak in the Cariboo-Chilcotin region
(2001-2015) is without precedent in the written record, but the outbreak was highly
variable spatially and temporally. At many locations along the northern and western edges
of the outbreak, defoliation was limited to dry valleys and slopes despite an abundance of
available host in upland areas. Unfavorable climatic factors may explain the apparent lack
of defoliation, for example insufficient degree days for maturation or poor phenological
synchrony between budburst and larval emergence preventing feeding. While upland
Douglas-fir stands may presently be climatically suboptimal for WSB, they are projected
to be at higher risk of infestation by the 2020s (Murdock et al., 2013).
Knowledge of historical outbreak patterns can be used to determine risk of expansion
of insect infestations into novel areas due to climate change, forest management practices
or other influences on forest and insect dynamics (Ryerson et al., 2003; Flower et al.,
2014). Tree growth is affected by defoliation (Alfaro et al., 1982), therefore tree ring
analysis can be used to develop long term outbreak chronologies that go beyond the
limited extent of historical survey records. Swetnam et al. (1985) developed an outbreak
reconstruction method for WSB that uses a non-host species as a climate control, which
has since been applied in the reconstruction of several spruce budworm species, insect
defoliators in hardwoods and softwoods, and fungal diseases of conifers (Case &
MacDonald, 2003; Zhang & Alfaro, 2003; Huang et al., 2008; Flower et al., 2014; Welsh
et al., 2014; Robson et al., 2015).
An important assumption underlying the use of a non-host as a climate control is that
both the host species and the non-host respond similarly to climate. A study of the potential
use of Douglas-fir and ponderosa pine in tree-ring reconstructions of historical climate in
155

the southern Cariboo-Chilcotin found that the growth trends of both species growing in the
same stands were similar (Watson & Luckman, 2002). They also concluded that there was
little evidence of differential growth due to insect infestations in the 20th century.
Comparison of host and non-host species growing in the same stand should result in an
overestimate of the severity of any outbreak as the defoliation of the host gives adjacent
non-host trees a competitive advantage (Swetnam et al., 1985).
A more recent dendrochronological study of WSB outbreaks in the Cariboo-Chilcotin
identified a number of widespread and severe events prior to the 21st century, notably ca.
1900–1910 and 1935–1950, both prior to the beginning of aerial forest health surveys in
1957 (Axelson et al., 2015). This study used a tree ring chronology developed from
populations of ponderosa pine near Kamloops to account for climatic influences at 11
Douglas-fir sites in the central, southern, eastern and western portions of the 21st century
outbreak. Periods were identified as budworm outbreaks when the detrended ringwidth
index of individual Douglas-fir trees was substantially lower than the mean detrended
index of the ponderosa pine populations around Kamloops for at least eight years.
Axelson et al. (2015) demonstrated that the Cariboo-Chilcotin region had a history of
WSB outbreaks prior to the large scale outbreak starting in 2001. The outbreak history in
areas further north and upward in elevation from their study was unknown but of great
interest considering the potential for range expansion and increases in severity of outbreaks
due to climate change.

156

The objectives of my study were to:
1. Reconstruct outbreak history of western spruce budworm at and beyond the
edge of the 21st century infestation
2. Examine climate factors that contributed to historical outbreaks
3. Determine whether the timing of historical outbreaks in the western Chilcotin
region is consistent with immigration, as appears to have been the case in the
21st century outbreak, or irruption from an endemic population.
Results of preliminary analyses of data collected for the first objective raised
questions about the outbreak reconstruction approach, which led to the final research
objective:
4. Critically reassess regional western spruce budworm outbreak reconstructions
by identifying other potential causes of differential growth between the host
and non-host populations used to describe outbreak history.

157

Figure 5.2: Cumulative years of defoliation near the northern extent of surveyed defoliation in the Fraser River valley, with
contiguous Douglas-fir forests in the upland region to the west. Contour interval is 25m, with thick contours at 100m. Aerial
Overview Survey and Vegetation Resources Inventory data used under BC Open Government License 2.0.
158

5.3. Methods
5.3.1. Site Selection and Field Methods
As a point of reference for dendrochronological reconstructions, the annual
progression of surveyed western spruce budworm defoliation from 2001 to 2013
(https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ accessed 2013)
was mapped and overlaid by the study site locations. Defoliation records (total years of
defoliation, years of medium/severe rated defoliation) and mean annual precipitation data
(www.climatewna.com accessed 2016) were attached to each site using a spatial join in
ArcGIS.
Sites used for reconstruction of historical outbreaks were selected in the context of
their cumulative defoliation history in the current outbreak. Annual defoliation records
(2000–2013) were retrieved from the BC Ministry of Forests website
(https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/ accessed 2013)
and internal boundaries were dissolved using standard tools in ESRI ArcGIS ® software,
yielding annual boundaries of affected area. The 14 resulting data layers were overlaid
using the “Count Overlapping Polygons” plug-in tool
(https://blogs.esri.com/esri/arcgis/2012/11/26/spaghetti_meatballs_one_to_many/) in
ArcGIS®, generating a single data layer describing the total number of years that each area
had been affected. A final overlay was performed to generate a data layer that identified
areas with 6 or more years of surveyed defoliation from 2000–2013, of which at least one
year was coded as medium or severe in the survey data. The resulting data layers were
used to classify areas as severely affected (6 or more including one medium or severe
record), moderately affected (3+ years at any level), lightly affected (1–2 years at any
159

level), and without any record of defoliation. These layers were exported to Google Earth®
to inform the site selection process.
Aerial imagery in Google Earth® (dated 2005, accessed 2013 and 2014) was used to
identify stands that had not been previously logged and were likely to contain old trees.
Three main criteria guided the sampling design:
1. A number of well-distributed sites covering all four categories of defoliation
defined above: heavy, moderate, light, and no record of defoliation.
2. Heavy sampling along the Chilcotin River valley, which suffered severe defoliation
later than the central area. This delay could be due to immigration rather than an
irruption from a ubiquitous endemic population, which may or may not be a regular
feature of outbreaks in the region.
3. Collection of several sites that escaped 21st century defoliation in the higher
elevation area west of the northernmost “arm” of the current outbreak in the Fraser
River valley and several adjacent sites in the heavily defoliated lower elevation
areas for comparison.
A number of suitable candidate sites were identified for each of the categories above.
The largest stands were visited first and were sampled immediately if no cut stumps were
evident and trees over 250 years old were present. A total of 45 sites were sampled in the
2013 and 2014 field seasons (Figure 5.3).

160

Figure 5.3: Map of study sites overlying the cumulative defoliation history from 2001–2013 and the range of Douglas-fir leading
stands according to the British Columbia Vegetation Resources Inventory (2016). 20th century defoliation at Quesnel Lake and Riske
Creek indicated in brown. The Chilcotin region is the area west of the Fraser River; the Cariboo is east of the Fraser.
161

At each site a transect line was established perpendicular to the prevailing slope.
Every 20m along the transect line, two trees sampled up to a total of 20 trees. Two 5.1mm
increment cores were taken at breast height from opposite sides of each tree, perpendicular
to the slope to avoid reaction wood (Speer, 2010). Acceptable trees had no obvious signs
of damage to the trunk and no dead branches below approximately 2 meters (an indication
of older trees). A photo of each tree was taken for later reference. Tree and stand attributes
were measured in 3–5 fixed radius plots along each transect and included diameter at
breast height, species, and stand density.
Ponderosa pine chronologies collected near Kamloops, British Columbia, were
provided by Lori Daniels of the University of British Columbia Tree Ring Lab in
Vancouver. This chronology provides a sample depth of four or more trees from 1677–
2011. Kamloops is 82km from the southernmost site MON near Monkian Lake, and 302
km from the westernmost site PYP near Pyper Lake.

162

5.3.2. Laboratory Preparation
Increment cores were glued to individual wooden mounts and sanded with
progressively finer grits to 600 (Stokes & Smiley, 1968). All cores were visually
crossdated using a combination of the list method and the memorization method (Speer,
2010). The list method was used to identify marker years in the first several sites to be
crossdated until a number of regionally consistent marker years were identified.
Visually crossdated increment cores were scanned in batches of 8–20 at 1200 DPI
using an Epson® 1640XL flatbed scanner. Ringwidth measurements were made using the
computer program Windendro® (Regent Instruments 2014) and crossdating was confirmed
using the program COFECHA (Holmes, 1983). Cores or sections that were excessively
damaged or otherwise impossible to crossdate reliably were discarded from the study,
leaving 1,651 series from 886 trees at 45 sites.
5.3.3. Outbreak Reconstruction
All ring width series, both host and non-host, were standardized using a doubledetrending process where a modified negative exponential curve was fit to each series,
followed by a 50-year cubic smoothing spline (Axelson et al., 2015). The
dendrochronological program OUTBREAK (Holmes & Swetnam, 1996) was used to
reconstruct historical outbreak periods by correcting detrended individual host-tree series,
consisting of the average of two cores from each tree, with the detrended non-host mean
chronology to remove the regional climatic trend from the host series. Program default
settings were used to define outbreak thresholds: minimum 8 years below-average growth
in the corrected series (allowing up to two years of positive deviation) and at least one year

163

of growth reduction below -1.28 standard deviations from the corrected series mean (as in
Axelson et al., 2015). Reconstructed outbreaks were classified based on the percentage of
trees recording an outbreak within each site, using the thresholds of 15, 50 and 75% to
define light, medium, and severe outbreaks respectively (Axelson et al., 2015). Maps were
generated for each regionally-synchronous outbreak period, displaying the maximum
percentage of trees meeting outbreak criteria at each site overlaying the cumulative years
of defoliation in the 21st century and the range of Douglas-fir leading stands.
As there were a large number of sites involved in the study, scripts were developed
utilizing the package dplR within the statistical package R (Version 3.2.2.; Bunn et al.,
2017) to create the detrended input files for the program OUTBREAK.
Pearson correlation coefficients between detrended site-level chronologies and
monthly average temperature and monthly total precipitation measured at the Williams
Lake Airport (previous June to current August; http://www.ec.gc.ca/dccha-ahccd/) were
calculated for the 1937–2012 period using the package treeclim (version 1.0.16) (Zang &
Biondi, 2015) in the open source statistical program R (version 3.2.2; script in Appendix
A).
To identify possible alternative explanations for patterns observed from the
OUTBREAK results, detrended host and non-host chronologies were plotted for
comparison, and periods of regionally synchronous reconstructed outbreaks were isolated
from the plotted series for visual assessment. This led to a close re-examination of the nonhost ring width data provided from the Kamloops region, and of available ponderosa pine
samples previously collected near Kamloops and stored at the UNBC Tree Ring
Laboratory to confirm apparent growth trends. For initial visual analysis, the average
164

detrended ring width chronology from the ponderosa pine series provided by Lori Daniels
and a combined Douglas-fir chronology (all 45 sites) were plotted together.
Individual ring width series making up the ponderosa pine chronology were
converted to annual basal area increments in the R package dplR, and growth releases were
identified using the R package TRADER (Nowacki & Abrams, 1997; Altman et al., 2014).
Growth releases were defined as years in which the 20-year mean basal area increment
following a given year exceeded the 20-year mean preceding the subject year by 50% or
more. Releases were plotted by both the date of initiation (year of maximum increase) and
the dates for which a growth release was ongoing (remaining above 50% rate of increase).
Precipitation totals recorded from the previous June to current June (pJJ) were plotted
for the Kamloops weather station from 1897 to 2012 and for the Williams Lake weather
station from 1937 to 2012. Adjusted and homogenized monthly climate data for Kamloops
(station elevation 345m) and Williams Lake (station elevation 940m) were retrieved from
Environment and Climate Change Canada (http://www.ec.gc.ca/dccha-ahccd/ accessed
2017) and pairwise correlations were calculated in Stata 12.1® between monthly
precipitation totals, seasonal and annual precipitation totals, the sum of previous-June
through current-June, and June, July and August precipitation at each weather station.
Cook’s Palmer Drought Severity Index for the period 1640–1990 for the reconstruction
grid points 30 (122.5W 52.5N, 90km north-northwest of Williams Lake) and 42 (120W
50N, 76km south-southeast of Kamloops) were plotted as a point of reference for periods
where weather station data were not available for both regions. Reconstructed PDSI values
at point 42 were subtracted from those at point 30 to identify periods in which
reconstructed drought severity was higher at one grid point than the other.
165

Forest Insect and Disease Survey data for Douglas-fir beetle infestations
(https://www.for.gov.bc.ca/ftp/HFP/external/!publish/Aerial_Overview/FIDS%20data/
accessed 2017) were mapped to confirm the presence of aerial survey crews in the vicinity
of study sites during a period when budworm defoliation was reconstructed.

166

5.4. Results
The most recent outbreak was first recorded at four locations in the central
Cariboo-Chilcotin in 2001, and defoliation progressed outward from this region in
subsequent years (Figure 5.4). Populations reached Alexis Creek in the central Chilcotin in
2004 and became severe in that region in 2005. The outbreak reached its maximum
western extent in 2007 and subsequently retreated, though defoliation remained extensive
through 2011 (Figure 5.5). Sites in the plateau country immediately east of the Fraser River
received the most cumulative years of defoliation, while the highest cumulative medium or
severe defoliation occurred at site WIL near Williams Lake with a total of six years (Table
5.1). Basal area per hectare in trees >15cm DBH ranged from 23 m2/ha at site ALE to 105
m2/ha at site JKY, while the density of trees <15cm DBH ranged from 337 trees per hectare
at site FUZ to 81,330 at site YUN. Lodgepole pine and trembling aspen seedlings were
present at some sites, but are not included in the totals, which consider only Douglas-fir.

167

Figure 5.4: Annual progression of surveyed western spruce budworm defoliation, 2001–
2008.

168

Figure 5.5: Annual progression of surveyed western spruce budworm defoliation, 2009–
2013; total years of defoliation in the 21st century; maximum severity of defoliation in the
21st century overlaid by site locations; stands with Douglas-fir listed as the leading,
secondary, and tertiary species.
169

Precipitation in the previous summer had a greater number of positive correlations
with ring width than current-summer precipitation, though the correlation with currentJune precipitation was uniform across all sites but MON, the site closest to the Kamloops
region (Table 5.2). All significant precipitation correlations were positive, except for those
in February. The Kamloops ponderosa pine chronology had similar correlations with
precipitation compared to the rest of the dataset: positive in the previous June, July and
August, negative in February, and positive in the current June.
Monthly temperature/ring width correlations were less consistent, but when
significant, were negative in both previous and current-summer months (Table 5.3).
Growth at several sites showed positive correlations with the temperature of the preceding
November and current February, while ring width at site GAR had a negative correlation
with the temperature in March.
Only a few correlations for either temperature or precipitation were significant in
March, April and May, and none were significant in the current August (Table 5.2, Table
5.3).

170

Table 5.1: Site attributes; 21st cent. = total years of defoliation; Med/Sev = years in which
defoliation was recorded as medium or severe. MAP = mean annual precipitation; BAPH =
basal area per hectare (m2) in trees >15cm DBH; TPH is trees per hectare <15cm DBH.
Code
ABR
ACS
ALE
ALK
BCA
BIG
BLU
CAL
CAM
CHI
CTI
DCB
DOG
ENT
EPI
FUZ
GAR
HAN
HAW
JKY
KCR
LEE
LEN
MAB
MAQ
MAY
MCL
MEL
MON
NEE
NOJ
NRA
PTI
PUD
PVA
PYP
REN
RES
RIS
SOA
SUG
TWA
WHI
WIL
YUN

21st Cent.
0
4
6
10
6
8
7
7
7
7
0
10
9
7
9
0
6
3
5
9
8
6
2
9
0
9
7
1
6
1
0
0
1
0
9
1
2
2
9
7
9
0
7
9
6

Med/Sev
0
2
2
1
4
1
3
2
2
3
0
2
1
2
1
0
2
1
1
0
1
2
0
1
0
0
1
0
2
0
0
0
0
0
1
0
0
0
4
2
6
0
2
6
5

Elevation (m)
1,133
1,024
959
954
935
1,261
871
1,248
1,087
906
818
947
1,136
910
1,262
1,068
1,239
1,329
1,306
1,136
831
1,031
1,268
945
931
1,127
679
939
1,127
1,070
1,016
970
1,005
967
786
1,107
988
953
1,259
645
658
839
857
677
996

Latitude
51.7438923
52.0117549
52.0683102
51.7834782
52.0901795
51.3138054
52.3339213
52.0692987
51.6979303
51.9000469
52.6591690
51.5727098
51.4782313
51.9836634
51.6044373
52.2870636
51.7091378
51.9652070
51.9656547
51.6917149
52.0642667
51.9562319
52.0381704
51.9077376
52.3524638
51.8840641
52.4219183
52.2542212
51.1889869
52.1879372
52.4670937
52.2189822
52.1626914
52.1846704
52.2862723
52.0321977
52.1690185
52.1374131
51.9492246
52.4221649
52.1027626
52.5484822
52.2171170
52.1596966
51.8882930

171

Longitude
-123.0118885
-123.2996988
-123.4077121
-122.1876281
-123.3576366
-122.1480729
-122.2445340
-122.6152696
-121.6111339
-121.9391504
-122.3733768
-122.2225973
-122.1572942
-121.8744262
-122.0133402
-122.6800798
-122.4335147
-122.9133535
-122.9304042
-122.2037750
-121.8822881
-123.0739201
-123.1060724
-122.2720139
-122.5181316
-122.1975185
-122.3681458
-122.3956095
-121.1958171
-122.6386473
-122.7230840
-122.5051835
-123.9129591
-123.8525038
-122.1860931
-124.1353788
-123.7131926
-123.7408311
-122.6387795
-122.4136252
-122.0231037
-122.6309186
-122.3221347
-122.2338750
-123.1022910

MAP BAPH TPH
356
39 12,818
341
82 2,147
334
23 2,147
466
325
32 443
430
480
50 6,112
477
53 5,725
480
53 35,361
494
497
52 5,557
412
52 1,957
440
51 3,220
473
474
51 573
467
47 337
381
29 3,537
459
59 2,021
457
59 3,199
509
105 12,692
484
59 5,999
375
55 2,526
431
87 2,463
452
47 19,533
502
59 2,442
507
43 6,062
458
39 2,273
459
357
47 1,452
487
74 1,894
484
54 1,137
465
40 1,831
307
34 1,326
317
44 1,200
440
54 1,213
375
35 1,314
328
64 1,011
323
28 2,589
404
73 1,642
423
415
41 1,642
461
52 1,894
456
77 1,515
404
48 674
363
38 81,330

Table 5.2: Significant Pearson’s correlation coefficients (p<0.05) between detrended ring
width indices and monthly precipitation totals. Months in lowercase belong to the
preceding calendar year; months in uppercase belong to the current calendar year.
Site
ABR
ACS
ALE
ALK
BCA
BIG
BLU
CAL
CAM
CHI
CTI
DCB
DOG
ENT
EPI
FUZ
GAR
HAN
HAW
JKY
KCR
LEE
LEN
MAB
MAQ
MAY
MCL
MEL
MON
NEE
NOJ
NRA
PTI
PUD
PVA
PYP
REN
RES
RIS
SOA
SUG
TWA
WHI
WIL
YUN
PIPO

Total Monthly Precipitation
jun jul aug sep oct nov dec JAN FEB MAR APR MAY JUN JUL AUG
0.22 0.29 0.31
0.36
0.22 0.25 0.33
-0.25
0.44
0.30
-0.23 0.20
0.32
0.26 0.21 0.31
0.37 0.27
0.32
0.36
0.28 0.26
-0.19
0.34
0.22 0.28
0.31 0.26
0.31
0.25
0.24
0.37
0.28
0.36
0.24
0.31 0.30
0.38
0.31
0.33 0.27
0.26
0.34
0.23 0.34
0.37
0.24 0.37 0.34
0.41
0.30 0.18 0.32
0.42 0.30
0.23
0.37
0.34
0.41 0.23 0.28
0.29 0.17
0.18 0.38
0.30
0.32
0.24 0.43 0.31
0.30 0.29
0.34 0.29
-0.25
0.28 0.24
0.30 0.30
0.23
0.32
0.26
0.43 0.39
0.29 0.28
-0.21 0.20
0.44
0.24 0.33 0.30
-0.25
0.33 0.27
0.21 0.24 0.40
0.24
0.38
0.29 0.23
0.27
0.36 0.34 0.31
0.20
0.33 0.25
0.33
0.27
0.23 0.26
0.27 0.22
0.37 0.26 0.30
0.26 0.21
0.31 0.24 0.23
-0.23
0.28 0.21 0.26
0.23 0.25
0.33
0.28 0.30
0.23 0.22 0.29
0.25 0.21
0.22
0.39
0.23
0.27 0.26
0.28
0.29
-0.23
0.29
0.38
0.31 0.33
0.22
0.30
0.32
0.27
0.26 0.24
0.27
0.34 0.24
0.25
0.27 0.21
0.24
0.33 0.25
0.27 0.20 0.30
0.26
0.33 0.25
0.29
0.27
0.30 0.23
0.33 0.24 0.32
0.24
0.39 0.22
0.28 0.32
-0.28
0.26
0.39 0.36 0.33
0.32 0.26
0.30
0.26
0.24
0.30 0.25
0.21 0.39
-0.27
0.31
0.20 0.22 0.39
-0.27
0.31

172

Table 5.3: Significant Pearson’s correlation coefficients (p<0.05) between detrended ring
width indices and monthly average temperature. Months in lowercase belong to the
preceding calendar year; months in uppercase belong to the current calendar year.
Site
jun
ABR
ACS
ALE
ALK -0.23
BCA
BIG
BLU
CAL -0.28
CAM -0.24
CHI -0.30
CTI
DCB
DOG
ENT -0.31
EPI
FUZ -0.32
GAR
HAN
HAW
JKY
KCR
LEE
LEN
MAB
MAQ -0.28
MAY
MCL -0.29
MEL -0.29
MON
NEE -0.28
NOJ -0.29
NRA -0.26
PTI
PUD
PVA
PYP
REN
RES
RIS -0.25
SOA -0.32
SUG -0.26
TWA -0.22
WHI -0.31
WIL -0.27
YUN
PIPO

Mean Monthly Temperature
jul
aug sep oct nov dec JAN FEB MAR APR MAY JUN
-0.31 -0.14
-0.27
-0.26
-0.23
-0.26
-0.28
-0.29
0.25
-0.23
-0.34
0.24
-0.24
-0.19
-0.23
-0.24
-0.35
-0.21
-0.29
-0.24
-0.38

0.34

-0.33
-0.24

0.21

-0.32

0.26

0.21

-0.30
-0.26

0.30

0.23

-0.23
-0.23

0.22

-0.30

0.29
-0.23

-0.29
-0.22
-0.16
-0.26

-0.19
-0.18
-0.24
-0.18

-0.23

-0.29
-0.26 -0.23

0.27

-0.31
-0.32

-0.33
-0.36

0.23

-0.27

JUL AUG

-0.20
-0.26
-0.20

0.23
0.27

0.25
0.22

0.23

-0.28
-0.28

0.24
0.25

173

-0.26 -0.24
-0.23 -0.24
-0.24 -0.17
-0.24
-0.31 -0.23
-0.31 -0.32
-0.17
-0.23
-0.28
-0.20

My WSB reconstructions using the program OUTBREAK are very similar to those
of Axelson et al. (2015) (Figure 5.6). All sites meet outbreak criteria several times over the
course of the tree ring record. Most sites meet outbreak criteria around the turn of the 20th
century and from 1938-1949, in agreement with Axelson et al. (2015) (Figure 5.6).
Reconstructions for the 21st century often do not match the surveyed values: sites ALE,
CAM, ENT, EPI, GAR, JKY, KCR, MON, PVA, SOA, and WHI all survived five or more
years of surveyed defoliation, but never pass the threshold of 15% of trees meeting
outbreak criteria after 2000. Sites MAQ and TWA, which received no surveyed defoliation,
record several years above the 15% threshold in the 21st century and site LEN, which
received two years of surveyed defoliation, exceeds the 50% threshold for a decade.
Detrended ring width indices for all Douglas-fir in the study area and those of the
ponderosa pine non-host are very similar from one year to the next, with differences in
magnitude (Figure 5.7). In addition to 1899–1909 and 1938–1940, generally synchronous
periods of reconstructed outbreaks were identified ca. 1744–1752, 1770–1786, 1795–1808,
1854–1861, and 1870–1882 (Figure 5.8).

174

Site
ABR
ACS
ALE
ALK
BCA
BIG
BLU
CAL
CAM
CHI
CT I
DCB
DOG
ENT
EPI
FUZ
GAR
HAN
HAW
JKY
KCR
LEE
LEN
MAB
MAQ
MAY
MCL
MEL
MON
NEE
NOJ
NRA
PT I
PUD
PVA
PYP
REN
RES
RIS
SOA
SUG
T WA
WHI
WIL
YUN

21st 1650

1700

1750

1800

1850

1900

1950

2000

1700

1750

1800

1850

1900

1950

2000

0
4
6
10
6
8
7
7
7
7
0
10
9
7
9
0
6
3
5
9
8
6
2
9
0
9
7
1
6
1
0
0
1
0
9
1
2
2
9
7
9
0
7
9
6
1650

Figure 5.6: Reconstructed outbreaks based on the percentage of trees at each site meeting
the defined outbreak criteria. Light gray exceeds the 15% threshold; medium gray 50%,
and black 75%. 21st=total years of surveyed defoliation in the 21st century outbreak.
Vertical dashed line at 2001 indicates the beginning of the 21st century outbreak.
175



'RXJODVILU

3RQGHURVD3LQH

K1K

5LQJ:LGWK,QGH[























































































)LJXUH0HDQGHWUHQGHGFKURQRORJ\YDOXHVIRUDOO'RXJODVILULQWKHVWXG\DQGSRQGHURVDSLQHQHDU.DPORRSVDQGWKH'RXJODV
ILUFKURQRORJ\OHVVWKHSRQGHURVDSLQHFKURQRORJ\ KRVWPLQXVQRQKRVW IURPWR$UHDVVKDGHGE\JUD\EDUVLQGLFDWHWKH
SHULRGVRIUHJLRQDOO\V\QFKURQRXVUHFRQVWUXFWHGRXWEUHDNVFD±±±±±±
DQG±WKHVSRUDGLFSHULRGRIWKHVDQGVDQGWKHNQRZQRXWEUHDNSHULRG±




Douglas-fir
Ponderosa Pine
h-Nh

Ring Width Index

1.5

1

1870
1882

1899
1909

1938
1948

1975
1999

2001
2011

1942

1995
1990
1985
1980
1975

2010
2005
2000

1854
1861

1909
1904
1899

1795
1808

1861
1856

1780
1775
1770
1753
1748
1743

0

1770
1786

1805
1800
1795

1744
1752

1879
1874

0.5

-0.5

Figure 5.8: Mean detrended chronology values for all Douglas-fir in the study and
ponderosa pine, and the Douglas-fir chronology less the ponderosa pine chronology (h-Nh
= host minus non-host) for eight periods of regionally synchronous reconstructed outbreak
activity ca. 1744–1752, 1770–1786, 1795–1808, 1854–1861, 1870–1882, 1899–1909 and
1938–1940, the sporadic period of the 1980s and 1990s, and the known outbreak period of
2001–2011.

177

The period of the modern outbreak beginning in 2001 includes five sites in the
Chilcotin River valley exceeding the 50% threshold and 17 exceeding the 15% threshold
throughout the study area (Figure 5.9). A number of sites in heavily defoliated areas fail to
meet any of the outbreak thresholds, while two in the upland area northwest of Williams
Lake which has no recorded defoliation, exceed the 15% threshold.

Figure 5.9: Maximum percentage of trees meeting outbreak criteria in a given year at each
site in the period 2001–2011, overlaying the cumulative years of defoliation in the 21st
century outbreak and the range of Douglas-fir leading stands

178

Forest Insect and Disease surveys conducted during the period of sporadic outbreak
reconstruction in the 1989s and 1990s identified widespread infestations of mountain pine
beetle and pine needle cast in lodgepole pine and Douglas-fir beetle in Douglas-fir. No
records were made of western spruce budworm defoliation during this period, but two-year
cycle budworm (Choristoneura biennis) was mapped northeast of the study area around
Quesnel Lake (Figure 5.10).

Figure 5.10: Forest Insect and Disease Survey records from 1985 to 1996 showing data for
two-year cycle budworm, mountain pine beetle, pine needle cast, and Douglas-fir beetle,
overlaid by site locations displayed by the maximum number of trees meeting outbreak
criteria during this period.

179

A majority of trees met outbreak criteria at all but two sites in the 1938–1948
period (Figure 5.11), and 21 sites exceeded the 75% threshold. Five sites northwest of
Williams Lake with no recorded defoliation in the 21st century exceeded the 50%
threshold.

Figure 5.11: Maximum percentage of trees meeting outbreak criteria in a given year at each
site in the period 1938–1948, overlaying the cumulative years of defoliation in the 21st
century outbreak and the range of Douglas-fir leading stands

180

All but two sites had more than 75% of trees meeting the outbreak criteria in the
1899–1909 period, while the remainder surpassed the 50% threshold (Figure 5.12).

Figure 5.12: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1899–1909, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands

181

For the 1870–1882 period, three sites that received no recorded defoliation in the
21st century exceeded the 50% threshold in the region northwest of Williams Lake, while
sites NOJ and TWA exceeded the 75% threshold (Figure 5.13).

Figure 5.13: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1870–1882, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands

182

Every tree at site NEE met the outbreak criteria in the 1854–1861 period, while this
site received only one year of light infestation during the 21st century event (Figure 5.14).
Three upland sites northwest of Williams Lake and site ABR in the Chilcotin exceeded the
50% threshold in this same period, but had no defoliation recorded in the 21st century.

Figure 5.14: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1854–1861, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands

183

The 1795–1808 period includes a number of sites exceeding the 75% threshold in
the plateau country immediately east of the Fraser River, and four exceeding the 50%
threshold northwest of Williams Lake that received no defoliation recorded in the 21st
century (Figure 5.15).

Figure 5.15: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1795–1808, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands

184

In the 1770–1786 period, more than 75% of trees met outbreak criteria at two sites
northwest of Williams Lake that had no defoliation recorded during the 21st century
outbreak (Figure 5.16). A number of sites in the Chilcotin River valley exceeded the 50
and 75% thresholds, including sites RES and PUD with two and zero years of defoliation
recorded in the 21st century.

Figure 5.16: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1770–1786, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands

185

Ten sites exceeded the 75% threshold in the 1744-1752 period, including two sites
that had only one year of recorded defoliation in the 21st century (Figure 5.17).

Figure 5.17: Maximum percentage of trees meeting outbreak criteria in a given year at
each site in the period 1744–1752, overlaying the cumulative years of defoliation in the
21st century outbreak and the range of Douglas-fir leading stands

186

Peaks in growth release initiations in the non-host chronology occurred ca. 1760,
1800/1801, 1898/1899, 1938/39 and 1976–1990 (Figure 5.18). Peaks in ongoing growth
releases correspond to peaks in initiation dates, which is expected as the method of
calculating releases identifies the year with the maximum rate of increase, rather than the
first year exceeding the threshold (Figure 5.19). Many of these release events appear to
represent substantial changes in position within the stand, e.g. ascension to a canopy
position from the understory (1899 in “A” or 1731 in “B” and “C” in Figure 5.19) or the
death of a competitor (1977 in “D,” 1897 in “E,” or 1976 in “F” in Figure 5.19).

60%

40%

35

Light (50%)
Moderate(100%)
Severe(200%)
Initiation%
Sample Depth

Sample Depth

80%

Initiating Release

100%

20%

25

15
5

0%

-60%

-80%
-100%

Ongoing Release

-40%

-5
-15
-25

1650
1660
1670
1680
1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000

-20%

-35

Figure 5.18: Percent of non-host sample initiating growth release and percent with ongoing
release (growth increase >50%) calculated using 20-year periods.

187

A

B

C

D

E

F

Figure 5.19: Six outputs of the R package TRADER showing growth trends in ponderosa
pine, displaying the annual basal area increment in black (left Y axis) and the percent
growth change in blue (right Y axis), growth releases marked by dashed vertical lines.
188


3URORQJHGSHULRGVRIEHORZDYHUDJHWRWDOSUHFLSLWDWLRQIURPWKHSUHYLRXV-XQH
WKURXJKFXUUHQW-XQHRFFXUUHGDWWKH.DPORRSVZHDWKHUVWDWLRQLQ±±
±±DQG±DQGDWWKH:LOOLDPV/DNHZHDWKHUVWDWLRQLQ






S--3UHFLSLWDWLRQ 0LOOLPHWHUV 


± H[FHSWLQJDQG ±DQG± )LJXUH 

:LOOLDPV/DNH
.DPORRSV





































)LJXUH3UHYLRXV-XQHWKURXJKFXUUHQW-XQHSUHFLSLWDWLRQWRWDOV S-- DW.DPORRSV
IURP±DQG:LOOLDPV/DNHIURP±0HDQOLQHVDWPPIRU.DPORRSV
DQGPPIRU:LOOLDPV/DNH3HULRGVRIEHORZDYHUDJHSUHFLSLWDWLRQIRUHDFKVLWH
PDUNHGLQJUD\










PDSI was negative both of at Cook’s reconstruction points in the periods 1661–
1664, 1717–1722, 1756–1760, 1792–1800, 1839–1843, 1869–1873, 1918–1933, and
1967–1973 (Figure 5.21; Cook & Krusic, 2004). Reconstructed PDSI generally was lower
at point 30 (closer to Williams Lake) than point 42 (closer to Kamloops) for nine periods
of the length of the record, and higher for six periods (Figure 5.21).

190







3RLQW :LOOLDPV/DNH

3RLQW .DPORRSV




3'6,














































)LJXUH&RRN VUHFRQVWUXFWLRQRI3DOPHU¶V'URXJKW6HYHULW\,QGH[DWJULGSRLQWV NPQRUWKZHVWRI:LOOLDPV/DNH DQG
NPVRXWKVRXWKHDVWRI.DPORRSV IRUWKH\HDUV±<HDUVRIUHFRQVWUXFWHGRXWEUHDNVPDUNHGLQJUD\DERYHWKHOLQH
\HDUVZKHUH3'6,UHFRQVWUXFWLRQLVKLJKHU PRUHPRLVW DWWKHSRLQWQHDUHU.DPORRSVPDUNHGLQJUD\EHORZWKHOLQH








Inspection of archived ponderosa pine cores collected near Kamloops consistently
revealed a low-growth event corresponding to the period of low precipitation reported at
the Kamloops weather station from 1929–1933. As an example, Figure 5.22 shows a core
from a tree that was a small sapling at the time of the drought. Growth reduction in this
specimen began prior to the drought event, perhaps due to its suppressed or intermediate
status in the canopy.

Figure 5.22: Archived ponderosa pine core PP14 from the UNBC Tree Ring Lab showing
growth reduction in the late 1920s/early 1930s with growth release beginning ca. 1935

192

Pairwise correlations between precipitation totals recorded at Williams Lake and
Kamloops weather stations are significant at the annual level and in summer and autumn,
but not in winter or spring, or the months of December, January, February, and March
(Table 5.4). All precipitation means were lower in Kamloops than Williams Lake, but the
differences were more pronounced in the winter months: mean annual precipitation at
Kamloops is slightly less than half that of Williams Lake, but precipitation in winter at
Kamloops is only 15% of that recorded in Williams Lake.
Table 5.4: Mean annual, seasonal, and monthly precipitation total at Williams Lake and
Kamloops weather stations from 1937 to 2012, with pairwise correlations and significance
score for each variable. Sample size varies due to missing data values excluded from the
analysis. pJJ=total precipitation from the previous June to the current June. JJA=total
precipitation for the months of June, July and August. All values are in mm. Significant
correlations marked by bolded variables names (P<0.05).
Williams
Pairwise
Variable
Lake
Kamloops Correlation P Value
489
225
0.42
0.0003
Annual
Winter
119
18
-0.15
0.19
Spring
87
50
0.2
0.078
166
96
0.55
<0.0001
Summer
117
60
0.54
<0.0001
Autumn
548
328
0.4
<0.0001
pJJ
166
96
0.56
<0.0001
JJA
January
44
5
0.12
0.27
February
27
6
-0.12
0.27
March
24
9
0.04
0.74
23
15
0.31
0.005
April
40
25
0.38
0.0008
May
62
39
0.61
<0.0001
June
53
29
0.43
0.0001
July
51
28
0.61
<0.0001
August
39
26
0.70
<0.0001
September
38
20
0.42
0.0001
October
40
14
0.26
0.024
November
December
47
8
-0.08
0.47
193

n
74
75
74
75
75
75
75
75
75
74
75
74
75
75
75
75
75
75
75

Rates of basal area increment increase calculated by comparing the 20-year means
preceding and following each year are generally similar between the ponderosa pine and
Douglas-fir populations under analysis, except for the period 1890–1905, where a peak
occurs in the ponderosa pine but not the Douglas-fir, and from 1840 through 1880 where
elevated rates of growth increase are found in Douglas-fir but not ponderosa pine (Figure

40%

1000

20%

-20%
-40%
-60%
-80%

-100%

15

250

5

Ongoing Release PIPO 1650
1660
1670
1680
1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
Sample Depth

0%

750

60%

25

500

80%

35

Sample Depth

100%

Ongoing Release PSME

5.23).

Light (50%)
Moderate(100%)
Severe(200%)
Light (50%)
Moderate(100%)
Severe(200%)

-5
-15
-25

-35

Figure 5.23: Percent of total ponderosa pine (PIPO, beneath X axis) and Douglas-fir
(PSME, above X axis) with ongoing growth release exceeding light (50%), moderate
(100%), and severe (200%) thresholds on the basis of 20-year average basal area increase.
Black bars indicate growth differentials meeting outbreak receonstruction criteria on a
regionally consistent basis.

194

Peaks in reconstructed western spruce budworm outbreaks tend to coincide with
peaks of growth increase in the non-host chronology (Figure 5.24). Reconstructed
outbreaks beginning ca. 1899, 1938, and 1980 respectively correspond to peaks in growth
increase affecting >35% of trees in the non-host chronology. All of these periods include a
number of trees increasing at a rate of over 100% and 10–30% of trees exceeding a 200%
rate of increase on the basis of 20-year averages before and after the subject year.

80%
60%
40%
20%
0%

-20%

1650
1660
1670
1680
1690
1700
1710
1720
1730
1740
1750
1760
1770
1780
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010

Percent of Sites Recording WSB Outbreak (Positive)
Percent of Ponderosa Recording Increase (Negative)

100%

-40%
Release: 50-99%
Release: 100-199%
Release >200%
Budworm: 15-49.9%
Budworm: 50-74.9%
Budworm: >75%

-60%
-80%

-100%

Figure 5.24: Percentage of sites with reconstructed outbreaks exceeding 15, 50, and 75%
thresholds, and ponderosa pine trees displaying growth increase exceeding 50, 100, and
200% thresholds on the basis of five-year averages.

195

5.5. Discussion
A majority of the periods when outbreak criteria were met in the host/non-host
analysis coincide with a difference in measured or reconstructed precipitation between the
locations of the host and non-host populations, and several align with growth increases in
the non-host population that are not found in the host populations. Growth increases in the
non-host population may be a release from competition due to the death of neighboring
trees. This can lead to faster recovery after drought (Erickson & Waring, 2014), leading to
higher ring width index values. Finally, the spatial pattern of several reconstructed
outbreaks is different than what was observed in the recent outbreak. Sites that received
little or no defoliation in the 21st century were reconstructed as being more heavily affected
than sites that were recently defoliated. It is possible that endemic populations of WSB
were not near those sites, or some unknown factors resulted in little or no defoliation
during the recent outbreak. It is also possible that these sites, being upslope and/or further
north than others, have not experienced WSB outbreaks due to cooler temperatures and
asynchronous larva and bud development, and that the reconstructed outbreaks are actually
due to some other factor that caused growth differentials between the host and non-host.
Seven periods experienced regionally synchronous growth differentials such that the
average detrended growth of the ponderosa pine was greater than the detrended growth of
individual Douglas-fir. Each period is discussed below, considering possible confounding
factors and other evidence regarding the cause of differential growth between Douglas-fir
in the Williams Lake region and ponderosa pine in the Kamloops region.
5.5.1. Regionally synchronous periods of differential growth
Aerial surveys of the 21st century outbreak (2001–2013) observed medium or severe
196

defoliation at 28 of 45 sites and several years of light defoliation at two sites, one or two
years of light defoliation at seven sites, and no defoliation of any severity at eight sites.
This distribution is not reflected in the outputs of the program OUTBREAK: 50% of trees
met outbreak criteria at only five sites (implying at least 50% defoliation of those trees;
Alfaro et al., 1982), all clustered in the Chilcotin River valley. Each of these sites began
meeting outbreak criteria 2–3 years before defoliation surveys identified any infestation.
Nine sites that are known to have been severely impacted by the 21st century outbreak did
not exceed the 15% threshold in any year between 2001 and 2011, while three sites with no
record of defoliation did exceed the lowest threshold. This is the only period when a falsenegative result is known to have occurred, where defoliation took place but was not
identified in the tree rings.
This difference in surveyed vs. reconstructed defoliation appears to have been caused
by a depression in the growth of the non-host trees, apparent in both individual and
aggregate plotted chronologies, coincident with the defoliation of the host trees. This
growth depression in the non-host may be due to precipitation in Kamloops returning to,
and sometimes falling below, the historical average after higher levels from the mid 1970s
to the late 1990s. The decades of above-average precipitation may have allowed stands to
increase in density beyond the limits of the normal climate, and when precipitation levels
returned to the historical average, the abnormally dense stands suffered a reduction in
growth beyond what would result from drought alone.
While confirmation of this would require further study, a sharp reduction in growth
rate ca. 2000 is apparent in many of the ponderosa pine samples archived at the UNBC
Tree Ring Laboratory, and the basal area increment of the ponderosa pine chronology is
197

generally trending downwards in the final decade (Figure 5.9). This suppression would
account for the very poor relationship between the reconstructed values and the observed
defoliation because the index value of the non-host would be both lower than average and
an inaccurate representation of the climate.
Reconstructed outbreaks 1980s and 1990s were found in the greatest proportion of
trees at host sites in northern and western portions of the study area. If substantial
defoliation was occurring in more than half of the trees, the Aerial Overview Survey crews
that recorded Douglas-fir beetle, mountain pine beetle, pine needle cast and two-year cycle
budworm in the area would probably have observed it. Western spruce budworm was one
of eleven pathogens considered to be “major forest insects and diseases” of national
relevance in the 1990s (Hall et al., 1994). According to the BC Forest Health Aerial
Overview Survey Standards
(https://www.for.gov.bc.ca/hts/risc/pubs/teveg/foresthealth/assets/aerial.pdf accessed 2017)
the optimal timing of survey flights for Douglas-fir beetle is mid-June through late-August,
while western spruce budworm survey flights should be made from late-June to midAugust, an almost complete overlap. This suggests that these periods of suppression are
not attributable to defoliation by the western spruce budworm, and that the reconstruction
method has resulted in several false-positives.
Precipitation in Kamloops is substantially above the long-term average in the 1980s
and 1990s, while precipitation in Williams Lake remained close to its historical average.
This could account for the differential growth patterns meeting outbreak criteria at some
sites, but not all the variation within the study area.
Both ponderosa pine and Douglas-fir populations record a peak in growth increase ca.
198

1980, but a much greater proportion of the ponderosa pine are affected. A mountain pine
beetle outbreak was active in the 1980s and 1990s in the study area, and some patches of
pine were also killed in the vicinity of Kamloops (Erickson, 1987). It is possible that this
resulted in growth release in surviving trees making up the non-host chronology, while
simultaneous outbreaks to the north caused growth release in some Douglas-fir growing in
mixed stands but not others (Heath & Alfaro, 1990). The impact of this infestation on
Douglas-fir would depend very much on the amount of mature lodgepole pine present in
the stand at the time. Many sites in the western portion of the study area were heavily
affected by the mountain pine beetle outbreak of the 1930s (Erickson, 1992) and would
have only been 45 years old in 1980, too young and small to have a major impact on the
growth of Douglas-fir even if they were killed by the mountain pine beetle (Heath &
Alfaro, 1990). Stand dynamics may be at the root of the difference in the number of trees
meeting outbreak criteria among sites: Douglas-fir in the mixed stands heavily affected by
the mountain pine beetle in the 1980s and 1990s would have been in a state of growth
release at the same time as non-host, reducing the differential between them.
Relatively low growth in Douglas-fir from 1938 to 1948 was recorded across the study
area, potentially affecting 100,000 hectares more than the budworm outbreak of the 21st
century. It comes at a time when detrended ringwidth indices of both Douglas-fir and
ponderosa pine chronologies are well above average, but the ponderosa pine index is
higher.
Forest health reports from the Cariboo region recorded two-year cycle budworm
(Choristoneura biennis) defoliating spruce immediately east of the study area in 10
different years from 1926 to 1946 (Erickson, 1992). Forest tent caterpillar (Malacosoma
199

disstria) was recorded as defoliating trembling aspen (Populus tremuloides) north of
Williams Lake in 1937, and in many locations within the study area in 1941, 1942, and
1943 (Erickson, 1992), during the period identified as an outbreak of WSB (Axelson et al.,
2015; Harvey, 2017). Records of western spruce budworm defoliation in Douglas-fir near
Lillooet in the 1940s (Erickson, 1987) indicate that the pest and its host were of concern in
the interior of the province. It seems unlikely that a severe landscape-scale outbreak of
western spruce budworm would go unnoticed by trained surveyors who would have
travelled through the affected area on their way to observe defoliation in trembling aspen.
While minor defoliation in some areas cannot be ruled out, the landscape-scale event is
most likely a false-positive in the tree ring reconstruction of budworm defoliation.
The prolonged difference in precipitation trends between Williams Lake and Kamloops
during this period could account for the differential growth between Douglas-fir and
ponderosa pine, resulting in the false-positive. Williams Lake is below the long-term
average in all years during this period except 1942, while Kamloops falls below average
only in 1943 and 1944.
Growth suppression in Douglas-fir from 1899–1909 occurs before any written records
from insect survey crews, so budworm defoliation is a possible cause of the growth
reduction. As reconstructed, the magnitude of the event is several times that of the 21st
century outbreak, and the extent would be at least 100,000 hectares greater than the
recorded infestation. The near-perfect synchronicity of the suppression period, if caused by
budworm, would require an immediate irruption of endemic populations in every affected
area. This scale and timing are incongruent with the temperature/elevation gradient which
is the basis of projections of future range expansion under climate change (Murdock et al.,
200

2013), and the spread pattern apparent in modern survey records. The temporal
synchronicity and severity of the event suggests that some factor other than budworm was
at play.
A peak in the proportion of trees showing ongoing growth release in the ponderosa
pine chronology ca. 1900 is not matched by a peak in the Douglas-fir population, which is
unusual for these chronologies. The 1890s are known to be a period of more frequent
groundfire in the Cariboo-Chilcotin region, and the decade is suspected to have been a time
of mountain pine beetle outbreaks (Hawkes et al., 2004), but the disturbance history of the
ponderosa stand used as the non-host chronology is unknown. Changes in basal area
increment in ponderosa pine ca. 1900 are substantial, and appear to represent ascension to
the main canopy or release from very strong competition. This may have been a defining
event for the stand dynamics of the non-host site, securing the position of many trees that
are now canopy dominants. A disturbance of this magnitude affects the hydrology around
surviving trees, especially in dry ponderosa pine stands (Erickson & Waring, 2014), and
can weaken the climate/growth relationship (Laurent et al., 2003).
The indices of ponderosa pine and Douglas-fir closely parallel one another during this
period, suggesting that they are responding to the same weather systems, and the difference
in magnitude is what would be expected during a growth release in the ponderosa stand.
The observed growth release in the ponderosa pine, and the magnitude of that release,
explains the differential between the growth rates of the two species. Defoliation cannot be
ruled out as survey records are not available, but the landscape-level growth differential
pattern could not have resulted exclusively from budworm defoliation. Flower et al. (2014)
identified outbreaks in Washington, Idaho, and Montana from the 1890s through the 1920s,
201

with initiation and maximum defoliation dates varying by location. Campbell et al. (2006)
also reconstructed outbreaks from the 1890s through the 1910s at many of their sites in the
Kamloops region. These outbreaks may have extended to the Chilcotin region, but the
synchronicity of initiation and cessation of the reconstructed outbreaks in this chapter are
inconsistent with both of these previous studies, suggesting an abiotic component to the
tree ring signal.
The ring width indices of both species trend upwards in the 1870–1882 period
following a period of suppression in the early 1870s. Cook’s reconstruction of Palmer’s
Drought Severity Index is negative at both points 42 and 30 from 1869–1873
(corresponding to the suppression apparent in both ringwidth indices), but values were
lower (indicating drier conditions) at the point north of Williams Lake than the point south
of Kamloops from 1875–1881. The upland sites northwest of Williams Lake exceed the
higher outbreak criteria in this period but based on the reconstructed PDSI, these sites
would have been experiencing unusually dry conditions in the 1870s. The severity of the
suppression exceeds the magnitude of the 21st century outbreak in marginal sites in the
Chilcotin region as well as the upland regions in the north.
The presence of budworm cannot be ruled out, but the difference in reconstructed PDSI
between the two regions suggests that moisture was the growth limiting factor in this
period. Campbell et al. (2006) identified budworm defoliation at only a few of their sites
near Kamloops during this period, while Flower et al. (2014) reconstructed WSB
defoliation at approximately half of their sites in Oregon, Idaho and Montana.
The years 1854–1861 include a depression in the ring width indices of both species that
is parallel but of greater magnitude in the Douglas-fir chronology. Most sites only
202

exceeded the 15% threshold of trees meeting outbreak criteria, except for a cluster meeting
the 50 and 75% thresholds in the Chilcotin and northwest of Williams Lake, including sites
with little or no recent defoliation. Cook’s reconstructed PDSI values vary during this
period, and the climate data available do not suggest any cause for differential growth.
However, the pattern of reconstructed outbreak severity is inconsistent with the local
elevation gradients that seem to have limited the extent of current outbreak.
It is possible that budworm defoliation did occur concurrently with other factors that
limited the growth of Douglas-fir at the upland sites, though the nearest WSB
reconstruction (Campbell et al. 2006) indicates only three of 17 sites experienced
defoliation during this period.
Both Douglas-fir and ponderosa pine show sharp growth reduction from 1796–1800
during the 1795–1808 period, most likely caused by the regional drought of the late 18th
century (Wolfe et al., 2001), but the Douglas-fir average detrended index is much lower
than the ponderosa pine index. Cook’s reconstructed PDSI is lower (drier) at point 30 (near
Williams Lake) than point 42 (near Kamloops) from 1801–1803 and 1806–1814,
overlapping with much of the period of differential growth.
The footprint of this differential pattern is again substantially larger and more
synchronous than the 21st century budworm outbreak, suggesting an abiotic mechanism.
The relatively low (dry) PDSI values at the point nearer Williams Lake provide a plausible
explanation for at least a portion of the growth pattern, though defoliation cannot be ruled
out for this period. Campbell et al. (2006) identified defoliation signatures around this
time, but there is substantial variation in the beginning and end dates between their sites.
This is also the case in my reconstruction: at some sites, such as KCR, defoliation is
203

reconstructed as beginning the 1770s and continuing after 1800. This period is not
identified as a regional outbreak in the northwestern United States by Flower et al. (2014)
Both ponderosa pine and Douglas-fir ring width indices were trending strongly
upwards on a parallel trajectory in 1770–1786, but the ponderosa pine exceeded 1.0 in
several years that the Douglas-fir did not. This period also coincided with a period when
PDSI values indicate greater moisture deficit in the Williams Lake region, potentially
causing the growth differential to meet outbreak criteria. This period is generally quiescent
in reconstructions by Flower et al. (2014) and Campbell et al. (2006).
The final period, 1744–1752, includes a number of years in which the ponderosa pine
ringwidth index was above 1.0 while the Douglas-fir followed a trajectory below 1.0. More
than half of the ponderosa pine trees were in a state of growth release at the time, likely
due to the initiation of new trees or growth response of surviving trees to the disturbances
that opened growing space for that recruitment. The period also overlaps with a differential
in Cook’s PDSI reconstruction where values were lower (drier) at Williams Lake compared
to Kamloops. All three potential confounding factors were present, with the additional
influence of young seedlings that may not respond to climate as their established peers
would. The spatial pattern of the reconstructed outbreaks was skewed towards the western
sites, where five exceeded the 75% threshold and five exceeded 50% in addition to the
isolated site ABR, which was also >50%. Several sites in the central area also exceeded
both of the higher thresholds, but four sites did not meet the 15% threshold.
Defoliation is recorded by Campbell et al. (2006) at most sites following 1750, but
only two in the 1740s. Flower et al. (2014) identified widespread WSB defoliation
beginning in the 1750s and peaking in the 1760s, but less in the 1740s, though some sites
204

do break with the general trend and record extensive defoliation during this period.
5.5.2. Overview of total record of trees meeting outbreak criteria
Aerial surveys record the 14 sites as collectively defoliated for 231 years between
2001 and 2013, a rate of 39.5% (231/585). For the period 1640-2011, OUTBREAK
identified a similar percentage of site-years exceeding the threshold of 15% of trees
impacted by WSB (43%). It is unlikely that the same rate of attack has occurred over time
without recorded cases after the initiation of the Forest Insect and Disease Survey.
Differences in moisture deficit and disturbance histories may have caused at least some of
the growth differentials that met OUTBREAK criteria in the host-nonhost procedure.
5.5.3. Limitations of host/non-host defoliation reconstruction method in this study
The use of pure host and non-host stands in geographically separate locations (as in
this study), or the use of host and non-host from the same mixed stand both can result in
problems. Use of mixed stands in outbreak reconstruction has been discouraged from the
earliest development of the host/non-host technique, as defoliation of the host may allow
the non-host a competitive advantage, leading to an overestimation of outbreak severity
based on the difference between the chronologies (Swetnam et al., 1985). Swetnam et al.
(1985) also emphasized the importance of selecting host and non-host stands that would be
likely to experience the same weather systems and respond in a similar way due to shared
topographic situation and underlying soils. Recognizing that host species may not always
be available under these conditions, the authors suggest using undefoliated host trees and
limiting analyses to the present outbreak where these trees are known to have escaped
defoliation. My study explored the alternative approach of using a non-host growing
205

outside of the immediate study area, but differences in topographic position, soils, and
weather systems have the potential to affect results.
5.5.4. A note on regeneration as it relates to western spruce budworm
Regeneration surveyed in old-growth stands in this study was often found at extremely
high densities, similar to stands described by Wong & Iverson (2004). This overstocking, a
historically unprecedented outcome of long-term wildfire exclusion, provides a substantial
food source for western spruce budworm outbreaks (Maclauchlan & Brooks, 2009). If
historical insect survey records (Erickson, 1992) are taken at face value, then 21st century
outbreak may represent the range expansion that has been anticipated under the regional
warming trend that is already underway, and continued warming will support more
outbreaks of similar or greater magnitude (Dawson et al., 2008; Murdock et al., 2013). The
abundance of host material in the understory, where larvae falling from the main canopy
can find a second chance at feeding, may increase the difficulty of mapping and managing
future outbreaks.
5.6. Conclusion
No firm support for any extensive western spruce budworm outbreaks prior to the
events of the 21st century was found in this study. The results of the earlier reconstructions
of western spruce budworm outbreaks in the region (Axelson et al., 2015; Harvey, 2017)
were broadly replicated using the same host/non-host approach with independent datasets,
but the synchronous persistence of these same differential growth patterns, including
improbable locations based on budworm biology and spread dynamics, raises the
possibility that growth differences in the host-nonhost were actually due to differences in

206

climate and spatio-temporal variation in disturbance history
Brief and localized events such as the one surveyed at Riske Creek in 1975 (Erickson,
1992) were not identified using the host/non-host method using standard outbreak
detection criteria, as differential growth must last at least eight years. The poor description
of the 21st century outbreak by the host/non-host method, despite the cumulative
defoliation record of eight or more years at twelve sites, allows for the possibility of
outbreaks not described in this study or others. Mid-20th century insect surveys identified
small patches of defoliation in species of lower commercial value within the study area,
suggesting that had there been major outbreaks between 1930 and 2000, they would have
been detected.
These findings are different than those of Flower et al. (2014) and Campbell et al.
(2006) who reconstructed historical outbreaks Oregon, Idaho and Montana and southern
interior British Columbia respectively, using the same host-nonhost approach analyzing
ponderosa pine from sites near to their Douglas-fir and grand fir sample sites. The defining
feature of my reconstructed outbreaks in the 20th century is regional synchronicity in dates
of initiation and cessation, while both of these studies record wide variation in these dates.
The variation described in these studies is consistent with the spatio-temporal variation of
the 21st century outbreak in the Cariboo-Chilcotin region.
Modeling of budburst and larval emergence dates over the past 100 years indicates that
higher effective latitudes (a combination of latitude and elevation) have been experiencing
optimal phenological synchrony recently (Marciniak, 2015). The overall effect of the
observed warming trend in the study area (Dawson et al., 2008) may act on other stages of
the WSB lifecycle as well, including higher temperatures during autumn and spring
207

dispersal stages facilitating increased movement within and between stands (Wellington &
Henson, 1947). Anomalously warm temperatures (average June and July daily temperature
of approximately 18 °C and above) combined with low precipitation have preceded several
outbreaks southern British Columbia (Thomson et al., 1984). Warmer average
temperatures in the study area since 1950 (Dawson et al., 2008) suggest that these warm
summer conditions are more likely to occur. Forest insect and disease surveys (Erickson,
1992) show that WSB was present in the study area at low levels as early as 1975;
warming temperatures may have created conditions favorable to the expansion of these
endemic populations.
Based on my conclusion that it is unlikely that there were major historical
outbreaks over the time period and spatial scale studied, the 800,000 hectare outbreak in
the first decade of the 21st century reflects a range expansion that is probably attributable to
the regional temperature increase observed in the late 20th century (Dawson et al., 2008).
Continued warming is expected to cause further expansion of western spruce budworm
range (Murdock et al., 2013), meaning that more naïve populations will likely experience
defoliation in the near future.

208

6. Concluding synthesis
6.1. Abstract
In this dissertation I consider the history of disturbances in the Cariboo-Chilcotin
region of central British Columbia and the effect of canopy gaps created by disturbances
on the drought tolerance of surviving interior Douglas-fir. Adjacency to canopy gaps
created by natural disturbances or harvesting increased drought resistance of Douglas-fir.
My reconstruction of the history of Douglas-fir beetle attacks indicated that the rate of
attack has increased in recent decades, but other disturbance agents have been the primary
drivers of stand dynamics in the region. My effort to expand western spruce budworm
outbreak reconstruction to and beyond the margins of the 21st century outbreak revealed
spatial patterns that did not appear to be caused by defoliation, but may be attributable to
regional differences in precipitation driving differential growth. Widespread disturbance
during the drought at the end of the 18th century was found to have had a strong influence
on stand dynamics, bringing many trees into the position of canopy dominance they still
maintain. In this synthesis I review major results and discuss the implications of the work
as a whole.

209

6.2. Summary of chapter results
Chapter 2: Three sampled stands of interior Douglas-fir were partially harvested in the
1970s, leaving the sampled population of residual trees averaging 90 years old and 15cm
DBH at the time. Trees that received growing space on all sides grew significantly faster
following harvest compared to trees that gained access to growing space on one side, and
trees that received no new growing space grew more slowly than either open growing or
edge trees. This difference in radial growth is reflected in higher live crown ratios and
sapwood basal areas in trees with more growing space, but the relationship does not hold
for drought tolerance. Growth reduction during drought, measured relative to the growth
rate over the preceding five years, was not significantly different between fully and
partially-released trees in most years, though growth reduction in trees with no access to
open gaps was more severe than either. The only regular and statistically significant
difference between the fully-released trees and partially-released trees with regards to
drought tolerance was found when the growth rate during drought was compared to the
growth rate prior to harvest: higher growth rate in the open growing trees translated into
relatively high growth rate during drought.
Among several interacting mechanisms, the key factor driving drought tolerance in this
situation appears to be precipitation throughfall, as reduced canopy interception in open
areas allows more rainfall to reach the soil to be taken up by roots. Strong negative
correlations between ringwidth and the mean temperature of the current June suggest
warmer temperatures projected under anthropogenic climate change scenarios will have a
substantial negative effect on tree growth, though the overall detriment to the tree may be
mitigated by access to open gaps.
210

Chapter 3: Trees experiencing a growth release attributed to the death of a
neighboring competitor showed higher drought resistance (less relative reduction in growth
rate during drought) compared to trees not experiencing growth release, and the
relationship was statistically significant in the majority of drought years tested. Drought
resilience, or the return to pre-drought growth rate, was positively related to the resistance,
and the relationship was almost always statistically significant. As in the previous chapter,
precipitation throughfall is the most likely driver of increased drought resistance, with
canopy gaps providing areas of higher moisture during the growing season. Resilience was
often negatively influenced by both age and basal area of trees at the time of drought,
though the relationship was not statistically significant in a majority of cases. Drought
resistance of young trees (<50 years old) was found to be significantly higher than that of
older trees in many of the drought years tested. Few significant differences were found
between older age classes (50–149 vs. 150–249 and 250+), but in the three cases where a
difference was significant, the oldest age class (250+) had lower resistance.
The average rate of trees showing growth release in a given decade was 14% in the era
prior to European settlement in the late 19th century, while 38% of trees showed growth
release in the first decade of the 19th century. This event brought a large number of
previously suppressed individuals into positions of canopy dominance that many have
retained to the present day. One of the 46 sites appears to have been even-aged, with all 20
trees apparently initiated in the late 18th or early 19th centuries. The rest were uneven-aged,
with the oldest trees initiated between the 14th and 18th centuries. Density of understory
regeneration and basal area of overstory trees in the old-growth remnants sampled was
generally high, in some cases far beyond historical levels and silvicultural
211

recommendations, and may present risk of stand stagnation or catastrophic fire due to the
accumulation of ladder fuels.

Chapter 4: Failed Douglas-fir beetle attack was quantified temporally based on evidence
of attack scars and resin pockets in the annual rings. Scars and resin pockets attributed to
failed attacks were identified from the end of the 17th century to the death of the sampled
Douglas-fir trees in 2014, indicating a regular occurrence that appears to have increased
over the 20th century. Douglas-fir beetle does not appear to have been a major driver of
stand dynamics compared to other common disturbances, especially wildfire and mountain
pine beetle. Douglas-fir beetle outbreaks may become more important in the context of
rising temperatures, increased stand density, and recent wildfire damage.

Chapter 5: My effort to expand upon previous reconstructions of western spruce
budworm outbreak history, which were based on the comparison of growth rates between
potential host Douglas-fir trees and a non-host ponderosa pine population in an adjacent
region, turned into an investigation of confounding factors that became apparent due to the
expansion of the study to and beyond the margins of the current outbreak. Growth
differentials between the host and non-host tree populations appear to be influenced by
differences in climate between the regions where the populations were sampled. The
annually surveyed 21st century outbreak was poorly described by the same host/non-host
comparison due to an unexplained suppression in the growth rate of the ponderosa pine.
This suppression may be due to a complex interaction involving prolonged above-average
precipitation in the non-host region followed by a return to normal conditions after
212

extensive ingrowth of trees.
Failure of the host/non-host analysis to accurately describe the known event allows for
the possibility that previous outbreaks were also undescribed. However, this study does not
contradict earlier work suggesting no history of major outbreaks (Watson & Luckman,
2002). If the historical record from insect surveyors is taken at face value since the 1930s,
then the 21st century outbreak represents substantial range expansion that is likely
attributable to observed warming trends that are expected to continue under anthropogenic
climate change.

213

6.3. Ecological implications and silvicultural applications
Adjacency to canopy gaps created by disturbances conferred definite benefits in
drought tolerance to surviving trees, as measured by radial growth during the drought,
whether the canopy gaps were created by natural or anthropogenic disturbances. The
dominant mechanism supporting this resistance appears to be access by the roots to areas
where precipitation falls directly to the ground without being intercepted by the forest
canopy and lost to evaporation. Open space on all four sides does not appear to be a
requirement. While Douglas-fir beetle does not seem to have been responsible for the
majority of historical canopy gaps, the presence of gaps created by other disturbances
likely reduced stand vulnerability to Douglas-fir beetle by increasing the moisture
available to potential host trees.
These results all favor the refinement and widespread application of clump-and-gap
type silviculture within mule deer winter range (Dawson et al., 2007), balancing needs for
gaps allowing precipitation throughfall and clumps preventing that throughfall for the
benefit of mule deer mobility and browse. Erring towards smaller clumps (4–6 trees) vs.
larger clumps (7–10 trees) would likely favor drought tolerance, especially in combination
with thorough application of the spacing and stocking recommendations already
established (Dawson et al., 2007). Projections of ecosystem response to climate change
predict rapid expansion of the bunchgrass zone into what is presently mule deer winter
range. These results suggest widespread application of established density management
practices to maintain habitat quality to the greatest extent possible while studies are
conducted and decisions are made about adapting range boundaries to match the realized
effects of climate change in the 2020s and 2030s (Hamann & Wang, 2006).
214

Precedent for almost any system of silviculture may be found in the natural history of
old-growth stands in the Cariboo-Chilcotin region, including the conservative policy of
infrequent removals roughly equaling net growth over the period since the last entry. It is
understood that exclusion of natural fires over the past century, both through firefighting
efforts and consumption of fine fuels by free range cattle, has resulted in overstocked
conditions and growth stagnation across much of the study area (Wong & Iverson, 2004).
In the context of overstocking and a warming climate, stand-level growth in recent decades
may not be a reliable measure of the productive capacity of the land. Density objectives set
relative to current stocking and stand-level growth rates may provide a misleading target.
Definition of density objectives and structural goals for each size class under specific
moisture regimes (e.g. Dawson et al., 2007), is likely the more sound approach, and some
results of this study may inform goals for stands not covered by recommendations for
conservation of mule deer winter range.
The series of disturbances affecting stands in the late 18th century, which resulted in
nearly three times the average decadal rate of growth release, provides a precedent for
more substantial harvests in stands outside of mule deer winter ranges that already have
some history of harvest. The silvicultural objective would be the rapid growth of small but
established trees that was seen between 1800 and 1840, combined with the increased
drought tolerance observed after harvest in the 1970s. Accretion on established trees
provides the greater benefit to the mid-term timber supply, but where that is impractical,
establishment and density management of new regeneration for the long-term timber
supply would be a reasonable objective. The utility of this approach is its immediate
correction of the overstocking attributed to wildfire exclusion (Wong & Iverson, 2004) in
215

operations that are generally profitable due to the timber volume extracted. Correction of
overstocking would reduce the risk of catastrophic fire and limit the stress of projected
climate change in the short to mid-term, mitigating two of the major predisposing factors
for Douglas-fir beetle attack.
The events of 1790–1800 allowed a large cohort of previously suppressed Douglas-fir
to reach a position of preeminence across the region. The question remains as to why so
many suppressed trees were released at this time. Such an extensive wildfire should have
left more scarred trees than have been found in previous studies (Daniels & Watson, 2003;
Harvey, 2017), and suppressed trees would not be expected to survive a severe fire that
killed so many mature Douglas-fir. Windthrow is a possibility, but the region is not known
to be prone to hurricanes or tornadoes, and windthrow should have led to more extensive
populations of Douglas-fir beetle than are apparent in the scar/resin pocket data. The only
serious candidate remaining is mountain pine beetle, which is considered to be responsible
for growth release on a similar scale in the 1930s and is known to have affected nearly all
mature pine in the area in the 2000s. A complication in this hypothesis is that the
geographical footprint of the growth release pattern is greater than could have resulted
from the modern distribution of lodgepole pine.
The modern distribution of lodgepole pine follows a gradient from low density in the
relatively warm and dry valley bottoms to higher density in the uplands, where Douglas-fir
is limited by temperature (Griesbauer & Green, 2010) and is ultimately relegated to small
inclusions on advantageous aspects among forests of pine and spruce. If this trend held
through the Little Ice Age, the effect of lower temperatures in that era could have been that
those forests which are now pure or nearly pure Douglas-fir were then more reminiscent of
216

stands found at the upland margins of Douglas-fir’s realized niche, such as those near
Pyper Lake (site PYP; Figure 6.1) or Punti Lake (site PUD; Figure 3.1) in the Chilcotin
region. The disturbance at the end of the 18th century would have come at the end of the
Little Ice Age, and may have been the trigger for a substantial compositional shift with the
study area.
There is some support for a species shift in a sediment core sampled in the study area
Hansen (1955), which found an increase in Douglas-fir pollen and a reduction in pine
pollen in the top 2cm. However there was limited resolution of the dating, and a sample
size of only one core in this study.
If lodgepole pine was more dominant in the study area in the late 1700s, then sites at
lower elevations presently dominated by Douglas-fir may have resembled sites at the
upland margins of the range of Douglas-fir following the mountain pine beetle outbreak of
the 2000s (Figure 6.1). Even a light ground fire at any time in the 10–20 years following
the mountain pine beetle outbreak could wipe out all lodgepole pine regeneration in the
pictured stand (Figure 6.1) before the young trees could form a substantial reserve of
serotinous cones. Such a fire would spare most of the mature Douglas-fir, bringing that
species into the majority and creating an ideal substrate for seed germination, allowing
Douglas-fir to thoroughly occupy the growing space. Given Douglas-fir’s longevity,
resistance to wildfire, and intermediate shade tolerance, the shade-intolerant lodgepole
pine would be slow to re-establish even if it were able to grow under the new climatic
conditions. With a mean plot-level fire return interval of 16–36 years, depending on the
location (Harvey, 2017), the odds of a fire at an inopportune time for the regenerating
lodgepole pine are fair.
217

Another implication of this hypothesis is that the extensive history of mountain pine
beetle in the Chilcotin region (Alfaro et al., 2010) reflects the ebb and flow of the
lodgepole pine forest as a viable competitor with interior Douglas-fir according to changes
in climate at the margin of its range. In this context, mountain pine beetle would not be
displaying cyclic behavior in the way the word is normally used in disturbance ecology
(Pickett & White, 1985), but behaving as a consequence of climate change and a facilitator
of species migration. This is a plausible scenario as the stands in question are in close
proximity to their climatically-limited range margins on several fronts. If this dynamic is
supported by future pollen core analyses, it would provide useful information on the future
trajectory of stands under anthropogenic climate change, especially if developed in
conjunction with an accurate estimate of local temperature change since the Little Ice Age.
The combination of temperature change and range margin change reconstructions would
support more precise projections of anthropogenic climate change scenarios onto the
landscape to identify locations where planners might favor Douglas-fir over lodgepole
pine.

218

Figure 6.1: Tree PYP01, ca. 1619, surrounded by several age classes of lodgepole pine
regeneration and Douglas-fir of at least three cohorts.
219

The history of Douglas-fir beetle attacks apparent in resin pockets and scars
indicates that at some points in 18th and 19th centuries, there were more widespread
infestations than were recorded during 20th century aerial surveys. There is nothing in this
250-year record that suggests an event equaling the magnitude of the 21st century
outbreaks. Douglas-fir mortality estimated from the 2016 records is more than twice that
recorded in fixed-wing surveys in any of the preceding 15 years. Based on the combination
of long-term tree ring evidence and more recent aerial surveys, I conclude that the 2016
mortality is unusual. The unusually high beetle population warrants close monitoring for
atypical attack behavior, focusing especially on mass attacks in healthy trees.
The history of western spruce budworm in the region remains unclear. Based on my
results, evidence of differential weather patterns that could lead to false positives in
outbreak reconstruction, and the lack of recorded evidence of past budworm outbreaks, it is
most likely that the 21st century outbreak is anomalous in its extent and severity. If that is
the case, the stand density allowed to build over a century of fire exclusion, especially in
the understory (Wong & Iverson, 2004), warrants re-evaluation as a food source potentially
intensifying future budworm outbreaks in an interaction with a warmer climate.
Interior Douglas-fir forests in the dry-belt region of central British Columbia have
survived the insect outbreaks of the early 21st century mostly intact, and this work suggests
several practical steps that may be taken to maintain them despite changes to the climate
that are already being realized. Projections based on anthropogenic climate change
scenarios forecast conversion of many grassland-margin forests to bunchgrass ecosystems
over the course of the 21st century, beginning as early as 2025 (Hamann & Wang, 2006).
The pace of this conversion is not a foregone conclusion. Characteristics of the
220

environment around individual trees can modulate the individual tree’s response to changes
brought about by climate or disturbance, to a degree that silvicultural intervention can
prolong the existence of essential habitat features. Most prescriptions would require
enough trees to be removed that the interventions would be commercially viable timber
harvests. These harvests will provide time for careful study of the realized effects of
climate change on the ground, allowing confident long-term planning to meet mule deer
winter range and Douglas-fir timber production objectives on the landscape.

221

6.4. Suggested topics for future research:
In addition to the pollen-core analysis addressing the hypothesis of dynamic range
margins driving disturbance patterns over recent centuries, three other topics of future
research are suggested:
1. Mitigation of drought stress by the creation of persistent canopy gaps during
juvenile spacing or early commercial thinning. This dissertation is focused on
primary-growth stands and primary-growth stands with large sawtimber
removed, and may not be transferable to even-aged stands managed from
initiation.
2. Extent of Douglas-fir beetle mass-attacks in healthy trees under increased
Douglas-fir beetle population pressure, and comparison of brood production
between these healthy host trees and dead, diseased, or fire-damaged trees. This
study finds that recent mortality is most likely without precedent within the past
three centuries, and suggests close attention in the context of climate warming.
3. Population genetics of Douglas-fir extending along river valleys from the dry
interior Chilcotin region to the Pacific Ocean via the Bute Inlet. The traits
selectively retained along this narrow ‘bridge’ may add to the basic
understanding of the effect of climate on the processes of evolution and
migration of tree species (Figure 6.2).

222

Figure 6.2: Continuity of Douglas-fir from the coast to the interior ‘dry belt’ via the Bute Inlet. Stands containing Douglas-fir (BC
Vegetation Resources Inventory; primary, secondary, or tertiary species) overlaying the mean annual precipitation (mm; “MAP”)
from 1961 to 1990 (www.climatewna.com accessed 2017), showing the ‘bridge’ of Douglas-fir from the high-precipitation coastal
region to the dry interior areas described in this dissertation (study area corresponds to the northeast quadrant of this map
223

7. Literature Cited
Agee JK, Skinner CN. 2005. Basic principles of forest fuel reduction treatments. Forest
Ecology and Management 211: 83–96.
Alfaro R, Campbell E, Hawkes B. 2010. Historical frequency, intensity and extent of
mountain pine beetle disturbance in British Columbia. Victoria, BC: Pacific Forestry
Centre, Canadian Forest Service, Natural Resources Canada. Mountain Pine Beetle
working paper 2009-30. 52p.
Alfaro RI, Campbell R, Vera P, Hawkes B, Shore TL. 2004. Dendroecological
reconstruction of mountain pine beetle outbreaks in the Chilcotin Plateau of British
Columbia. Pages 245-256 in Mountain pine beetle symposium: Challenges and solutions,
October 30-31, 2003, Kelowna, British Columbia, Canada. Shore TL, Brooks JE, Stone
JE, editors. 2004. Victoria, BC: Pacific Forestry Centre, Canadian Forest Service, Natural
Resources Canada. Information Report BC-X-399. 298p.
Allen CD, Macalady AK, Chenchouni H, Bachelet D, McDowell N, Vennetier M
Kitzberger T, Rigling A, Breshears DD, Hogg EH, Gonzalez P, Fensham R, Zhang Z,
Castro J, Demidova N, Lim JH, Allard G, Running SW, Semerci A, Cobb N. 2010. A
global overview of drought and heat-induced tree mortality reveals emerging climate
change risks for forests. Forest Ecology and Management 259: 660–684.
Altman J, Fibich P, Dolezal J, Aakala T. 2014. TRADER: A package for tree ring
analysis of disturbance events in R. Dendrochronologia 32: 107–112.
Ananda J, Herath G. 2003. Incorporating stakeholder values into regional forest
planning: A value function approach. Ecological Economics 45: 75–90.
Anderegg WRL, Berry JA, Smith DD, Sperry JS, Anderegg LDL, Field CB. 2012. The
roles of hydraulic and carbon stress in a widespread climate-induced forest die-off.
Proceedings of the National Academy of Sciences 109: 233–237.
Armleder, H. 1999. Clumpy spacing: juvenile spacing Douglas-fir into clumps to imitate
natural stand structure. Victoria, BC: British Columbia Ministry of Forests. Extension Note
32. 6p.
Armleder HM, Dawson RJ, Thomson RN. 1986. Handbook for timber and mule deer
management co-ordination on winter ranges in the Cariboo Forest Region. Victoria, BC:
British Columbia Ministry of Forests. Handbook number 13. 108p.
Armleder HM, Waterhouse MJ, Keisker DG, Dawson RJ. 1994. Winter habitat use by
mule deer in the central interior of British Columbia. Canadian Journal of Zoology 72:
1721–1725.

224

Atkins MD. 1960. A Study of the Flight of the Douglas-Fir Beetle Dendroctonus
pseudotsugae Hopk.(Coleoptera: Scolytidae): II. Flight Movements1. The Canadian
Entomologist, 92(12), 941-954.
Axelson JN, Alfaro RI, Hawkes BC. 2010. Changes in stand structure in uneven-aged
lodgepole pine stands impacted by mountain pine beetle epidemics and fires in central
British Columbia. The Forestry Chronicle 86: 87–99.
Axelson JN, Smith DJ, Daniels LD, Alfaro RI. 2015. Multicentury reconstruction of
western spruce budworm outbreaks in central British Columbia, Canada. Forest Ecology
and Management 335: 235–248.
Baldwin VCJ, Feduccia DP, Haywood JD. 1989. Postthinning growth and yield of rowthinned and selectively thinned loblolly and slash pine plantations. Canadian Journal of
Forest Research 19: 247–256.
Belluschi P, Johnson N, Heikkenen H. 1965. Douglas-fir defects caused by the Douglasfir beetle. Journal of Forestry 63: 252–256.
Bentz BJ, Régnière J, Fettig CJ, Hansen EM, Hayes JL, Hicke JA, Kelsey RG,
Negrón JF, Seybold SJ. 2010. Climate change and bark beetles of the western United
States and Canada: Direct and indirect effects. BioScience 60: 602–613.
Berryman AA. 1972. Resistance of conifers to invasion by bark beetle-fungus
associations. BioScience 22: 598–602.
Van Bogaert R, Jonasson C, De Dapper M, Callaghan T V. 2010. Range Expansion of
Thermophilic Aspen (Populus tremula L.) in the Swedish Subarctic. Arctic, Antarctic, and
Alpine Research 42: 362–375.
Bouten W, Heimovaara TJ, Tiktak A. 1992. Spatial patterns of throughfall and soil-water
dynamics in a Douglas-fir stand. Water Resources Research 28: 3227–3233.
Bradley RS, Jonest PD. 1993. ‘Little Ice Age’ summer temperature variations: their
nature and relevance to recent global warming trends. The Holocene 3: 367–376.
Bréda N, Granier A, Aussenac G. 1995. Effects of thinning on soil and tree water
relations, transpiration and growth in an oak forest (Quercus petraea (Matt.) Liebl.). Tree
Physiology 15: 295–306.
Brix H. 1979. Effects of plant water stress on photosynthesis and survival of four conifers.
Canadian Journal of Forest Research 9: 160–165.
Bunn A, Korpela M, Biondi F, Campelo F, Merian P, Qeadan F, Zang C, 2017. dplR:
Dendrochronology Program Libraryin R. R package version 1.6.6. https://r-forge.rproject.org/projects/dplr/
225

Burnham KP, Anderson DR. 2004. Multimodel inference understanding AIC and BIC in
model selection. Sociological Methods and Research 33: 261-304.
Burke JL, Carroll AL. 2017. Breeding matters: Natal experience influences population
state-dependent host acceptance by an eruptive insect herbivore. PLoS ONE 12: e0172448
Caccianiga M, Payette S, Filion L. 2008. Biotic disturbance in expanding subarctic
forests along the eastern coast of Hudson Bay. New Phytologist 178: 823–834.
Calev A, Zoref C, Tzukerman M, Moshe Y, Zangy E, Osem Y. 2016. High-intensity
thinning treatments in mature Pinus halepensis plantations experiencing prolonged
drought. European Journal of Forest Research 135: 1–13.
Case R, Macdonald GM. 2003. Tree ring reconstructions of streamflow for three
Canadian prairie rivers. Journal of the American Water Resources Association 39: 703–
716.
Case RA, MacDonald GM. 2003. Dendrochronological analysis of the response of
tamarack (Larix laricina) to climate and larch sawfly (Pristiphora erichsonii) infestations in
central Saskatchewan. Ecoscience, 10(3), 380-388.
Case RA, MacDonald GM. 1995. A dendroclimatic reconstruction of annual precipitation
on the western Canadian prairies since A.D. 1505 from Pinus flexilis James. Quaternary
Research 44: 267–275.
Casper BB, Jackson RB. 1997. Plant competition underground. Annual Review of
Ecology and Systematics 28: 545–570.
Cocke AE, Fulé PZ, Crouse JE. 2005. Forest change on a steep mountain gradient after
extended fire exclusion: San Francisco Peaks, Arizona, USA. Journal of Applied Ecology
42: 814–823.
Cook ER. 1987. The decomposition of tree-ring series for environmental studies. TreeRing Bulletin 47: 37–59.
Cook ER, Krusic PJ. 2004. The North American drought atlas. Palisades, NY: LamontDoherty Earth Observatory and the National Science Foundation.
http://iridl.ldeo.columbia.edu/SOURCES/.LDEO/.TRL/.NADA2004/.pdsi-atlas.html
Cooper WS. 1913. The climax forest of Isle Royale, Lake Superior, and its development.
I. Botanical Gazette 55: 1–44.
Currano ED, Wilf P, Wing SL, Labandeira CC, Lovelock EC, Royer DL. 2008.
Sharply increased insect herbivory during the Paleocene-Eocene Thermal Maximum.
Proceedings of the National Academy of Sciences 105: 1960–1964.

226

Dale MRT, Fortin M-J. 2014. Spatial analysis: a guide for ecologists. Cambridge
University Press.
Dale VH, Joyce LA, McNulty S, Neilson RP, Ayres MP, Flannigan MD, Hanson PJ,
Irland LC, Lugo AE, Peterson CJ, Simberloff D, Swanson FJ, Stocks BJ, Wotton BM.
2001. Climate change and forest disturbances. BioScience 51: 723–734.
Daniels, LD, Watson E. 2003. Climate–fire–vegetation interactions in the Cariboo forests:
A dendrochronological analysis. Vancouver, BC: Forest Innovation and Investment, Forest
Research Program. 56p.
Daubenmire R, Daubenmire JB. 1968. Forest vegetation of eastern Washington and
northern Idaho. Pullman, WA: Washington State University, College of Agriculture,
Washington Agricultural Experiment Station. Technical Bulletin 60. 104p.
Dawson R, Armleder H, Bings B, Peel D. 2007. Management Strategy for Mule Deer
Winter Ranges in the Cariboo-Chilcotin Part 1a: Management Plan for Shallow and
Moderate Snowpack Zones. Land Management Handbook Number 60. British Columbia
Ministry of Forests and Range Forest Sciences Program. 58p.
Dawson, R., Werner, A.T., Murdock, T.Q., 2008: Preliminary Analysis of Climate
Change in the Cariboo-Chilcotin Area of British Columbia. Pacific Climate Impacts
Consortium, University of Victoria, Victoria BC, 49 pp.Day JK. 1997. Interior Douglas-fir
and selection management at the UBC/Alex Fraser Research Forest. Pages 542-552 in
Proceedings of the IUFRO interdisciplinary uneven-aged management symposium,
September 1997. Emmingham, WH compiler. 1997. Corvallis, OR: International Union of
Forest Research Organizations. 713p.
Day JK. 1998a. Stocking standards for uneven-aged interior Douglas-fir. Pages 37-52 in
Managing the dry Douglas-fir forests of the southern interior: workshop proceedings April
29-30, 1997 Kamloops, British Columbia, Canada. Vyse A, Hollstedt DC, Huggard DJ,
editors. 1998. Victoria, BC: Ministry of Forests. Working Paper No. 34. 299p.
Day JK. 1998b. Selection management of interior Douglas-fir for mule deer winter range.
MSc Thesis, University of British Columbia.
Day JK, Pérez DM. 2013. Reducing uncertainty and risk through forest management
planning in British Columbia. Forest Ecology and Management 300: 117–124.
Day JK, Weckerle G, Mitchell JL, Gray RW, Trethewey C, Atkins BA. 2003.
Commercial thinning in mule deer winter range: Improving habitat through forest
management. Alex Fraser Research Forest, BC: University of British Columbia. Technical
Report, Forestry Innovation Investment Contract R2003-0216. 45p.

227

Dhar A, Coates KD, Rogers B, Hardy K. 2013. Impact of mountain pine beetle on midterm timber supply in sub boreal spruce zone of British Columbia. In Proceedings of the
16th International Boreal Forest Research Association (IBFRA) Conference, Edmonton,
AB, Canada, 7–10 October 2013. 28p.
Dolph KL, Mori SR, Oliver WW. 1995. Long-term response of old-growth stands to
varying levels of partial cutting in the Eastside Pine type. Western Journal of Applied
Forestry 10: 101–108.
Elkin C, Giuggiola A, Rigling A, Bugmann H. 2015. Short- and long-term efficacy of
forest thinning to mitigate drought impacts in mountain forests in the European Alps.
Ecological Applications 25: 1083–1098.
Erickson RD. 1987. Maps of major forest insect infestations: Kamloops Forest Region
1912-1986. Victoria, BC: Pacific Forestry Centre, Canadian Forestry Service. FIDS Report
87-08. 68p.
Erickson R. 1992. History of important forest insects in the Cariboo Forest Region 19131991. Victoria, BC: Pacific Forestry Centre, Canadian Forestry Service, Natural Resources
Canada. FIDS Report 92-14. 48p.
Erickson CC, Waring KM. 2014. Old Pinus ponderosa growth responses to restoration
treatments, climate and drought in a southwestern US landscape. Applied Vegetation
Science 17: 97–108.
Evelyn J. 1662. Silva, or a discourse of forest-trees and the propagation of timber in his
majesty’s dominions. London, UK: The Royal Society. 330p.
Eyles A, Bonello P, Ganley R, Mohammed C. 2010. Induced resistance to pests and
pathogens in trees. New Phytologist 185: 893–908.
Fauss DL, Pierce WR. 1969. Stand conditions and spruce budworm damage in a western
Montana forest. Journal of Forestry 67: 322–325.
Flower A, Gavin DG, Heyerdahl EK, Parsons RA, Cohn GM. 2014. Drought-triggered
western spruce budworm outbreaks in the interior Pacific Northwest: A multi-century
dendrochronological record. Forest Ecology and Management 324: 16–27.
Franceschi VR, Krokene P, Christiansen E, Krekling T. 2005. Anatomical and chemical
defenses of conifer bark against bark beetles and other pests. New Phytologist 167: 353–
376.
Franco M. 1986. The influence of neighbours on the growth of modular organisms with an
example from trees. Philosophical Transactions of the Royal Society of London B:
Biological Sciences 313: 209–225.

228

Fraver S, Seymour RS, Speer JH, White AS. 2007. Dendrochronological reconstruction
of spruce budworm outbreaks in northern Maine, USA. Canadian Journal of Forest
Research 37: 523–529.
Frelich LE, Reich PB. 2003. Perspectives on development of definitions and values
related to old-growth forests. Environmental Reviews 11: S9–S22.
Fuller AK, Harrison DJ, Lachowski HJ. 2004. Stand scale effects of partial harvesting
and clearcutting on small mammals and forest structure. Forest Ecology and Management
191: 373–386.
Gebhardt T, Häberle K-H, Matyssek R, Schulz C, Ammer C. 2014. The more, the
better? Water relations of Norway spruce stands after progressive thinning. Agricultural
and Forest Meteorology 197: 235–243.
de la Giroday HMC, Carroll AL, Aukema BH. 2012. Breach of the northern Rocky
Mountain geoclimatic barrier: Initiation of range expansion by the mountain pine beetle.
Journal of Biogeography 39: 1112–1123.
Giuggiola A, Bugmann H, Zingg A, Dobbertin M, Rigling A. 2013. Reduction of stand
density increases drought resistance in xeric Scots pine forests. Forest Ecology and
Management 310: 827–835.
Greenbank DO, Schaefer GW, Rainey RC. 1980. Spruce Budworm (Lepidoptera:
Tortricidae) Moth Flight and Dispersal: New Understanding From Canopy Observations,
Radar, and Aircraft. Memoirs of the Entomological Society of Canada 112: 1–49.
Griesbauer HP, Green DS. 2010. Regional and ecological patterns in interior Douglas-fir
climate–growth relationships in British Columbia, Canada. Canadian Journal of Forest
Research 40: 308–321.
Griffin D, Meko DM, Touchan R, Leavitt SW, Woodhouse CA. 2011. Latewood
chronology development for summer-moisture reconstruction in the US southwest. TreeRing Research 67: 87–101.
Guarín A, Taylor AH. 2005. Drought triggered tree mortality in mixed conifer forests in
Yosemite National Park, California, USA. Forest Ecology and Management 218: 229–244.
Hadley K, Veblen T. 1993. Stand response to western spruce budworm and Douglas-fir
bark beetle outbreaks, Colorado Front Range. Canadian Journal of Forest Research 23:
479–491.
Hall JP, Bowers WW, Crew GC, Magasi LP, Hurley JE, Lachance D, Howse GM,
Applejohn MJ, Syme PD, Myren DT, Nystrom KL, Hopkin AA, Meating JH, Power
JM, Cerezke HF, Brandt JP, Van Sickle GA, Wood CS. 1994. Forest insect and disease
conditions in Canada: 1992. Forest Insect and Disease Survey. Ottawa, Canada.

229

Hamann A, Wang T. 2006. Potential effects of climate change on ecosystem and tree
species distribution in British Columbia. Ecology 87: 2773–2786.
Hamilton FC. 1984. Contagious distributions in even aged forest stands: Dynamics of
spatial pattern and stand structure. MSc Thesis, Oregon State University.
Hansen H. 1955. Postglacial forests in south-central and central British Columbia.
American Journal of Science 253: 640–658.
Hardy CC, Arno SF. 1996. The use of fire in forest restoration. Ogden, UT: Intermountain Research Station, Forest Service, U.S. Department of Agriculture. General
Technical Report INT-GTR-341. 86p.
Harris JWE, Alfaro RI, Dawson AF, Brown RG. 1985. The western spruce budworm in
British Columbia 1909-1983. Victoria, BC: Pacific Forest Research Centre, Canadian
Forestry Service, Natural Resources Canada. Information Report BC-X-257. 32p.
Harrod RJ, McRae BH, Hartl WE. 1999. Historical stand reconstruction in ponderosa
pine forests to guide silvicultural prescriptions. Forest Ecology and Management 114:
433–446.
Hart SJ, Smith DJ, Clague JJ. 2010. A multi-species dendroclimatic reconstruction of
Chilko river streamflow, British Columbia, Canada. Hydrological Processes 24: 2752–
2761.
Harvey JE. 2017. Disturbance dynamics in west central British Columbia: Multi-century
relationships of fire, western spruce budworm outbreaks and climate. PhD Dissertation,
University of Victoria.
Hawkes B, Taylor S, Stockdale C, Shore T, Alfaro R, Campbell R, Vera P. 2004.
Impact of mountain pine beetle on stand dynamics in British Columbia. Pages 177-199 in
Mountain pine beetle symposium: Challenges and solutions, October 30-31, 2003,
Kelowna, British Columbia, Canada. Shore TL, Brooks JE, Stone JE, editors. 2004.
Victoria, BC: Pacific Forestry Centre, Canadian Forest Service, Natural Resources Canada.
Information Report BC-X-399. 298p.
Heath R, Alfaro RI. 1990. Growth response in a Douglas-fir/lodgepole pine stand after
thinning of lodgepole pine by the mountain pine beetle: A case study. Journal of
Entomological Society of British Columbia 87: 16–21.
Heyerdahl EK, Lertzman K, Wong CM. 2012. Mixed-severity fire regimes in dry forests
of southern interior British Columbia, Canada. Canadian Journal of Forest Research 42:
88–98.
Holmes RL. 1983. Computer-assisted quality control in tree-ring dating and measurement.
Tree-ring Bulletin 43: 69–78.

230

Holmes RL, Swetnam, TW, 1996. Dendroecology program library: program
OUTBREAK user’s manual. Laboratory of Tree-Ring Research, University of Arizona,
Tucson.
Hood S, Bentz B, Gibson K, Ryan K, DeNitto G. 2007. Assessing post-fire Douglas-fir
mortality and Douglas-fir beetle attacks in the northern Rocky Mountains. Fort Collins,
CO: Rocky Mountain Research Station, Forest Service, U.S. Department of Agriculture.
General Technical Report RMRS-GTR-199. 31p.
Hrinkevich K. 2012. Northern range limit mountain pine beetle (Dendroctonus
ponderosae) outbreak dynamics and climate interactions in mixed sub-boreal pine forests
of British Columbia. PhD Dissertation, University of Northern British Columbia.
Hrinkevich K, Lewis KJ. 2011. Northern range limit mountain pine beetle outbreak
dynamics in mixed sub-boreal pine forests of British Columbia. Ecosphere 2: 1–16.
Huang, JG, Tardif J, Denneler B, Bergeron Y, Berninger F, 2008. Tree-ring evidence
extends the historic northern range limit of severe defoliation by insects in the aspen stands
of western Quebec, Canada. Canadian Journal of Forest Research, 38(9), 2535-2544.
Hunter MLJ. 1990. Wildlife, forests, and forestry: Principles of managing forests for
biological diversity. Englewood Cliffs, NJ: Prentice Hall, Inc. 370p.
Jantz O, Rudinsky J. 1966. Studies of the olfactory behavior of the Douglas-fir beetle,
Dendroctonus pseudotsugae Hopkins. Corvallis, OR: Agricultural Experimental Station,
Oregon State University. Technical Bulletin 94. 38p.
Johnson N, Belluschi P. 1969. Host-finding behavior of the Douglas-fir beetle. Journal of
Forestry 67: 290–295.
Johnson N, Pettinger L. 1961. Douglas fir beetle attacks in living trees as influenced by
the presence of fresh windthrow. Centralia, WA: Forest Research Center, Timberland
Division, Weyerhaeuser Company. Forest Research Note 37. 8p.
Johnson N, Shea K. 1963. White fir defects associated with attacks by the fir engraver.
Centralia, WA: Forest Research Center, Timberland Division, Weyerhaeuser Company.
Forest Research Note 54. 8p.
Kano T. 2006. Characteristics of trees infested by Douglas-fir beetles in Kootenay Lake
Forest District. MSc Thesis, University of British Columbia.
Kharuk VI, Im ST, Dvinskaya ML, Ranson KJ. 2010. Climate-induced mountain treeline evolution in southern Siberia. Scandinavian Journal of Forest Research 25: 446–454.
Kohler M, Sohn J, Nägele G, Bauhus J. 2010. Can drought tolerance of Norway spruce
(Picea abies (L.) Karst.) be increased through thinning? European Journal of Forest
Research 129: 1109–1118.

231

Lagergren F, Lankreijer H, Kučera J, Cienciala E, Mölder M, Lindroth A. 2008.
Thinning effects on pine-spruce forest transpiration in central Sweden. Forest Ecology and
Management 255: 2312–2323.
Laurent M, Antoine N, Joël G. 2003. Effects of different thinning intensities on drought
response in Norway spruce (Picea abies (L.) Karst.). Forest Ecology and Management
183: 47–60.
Lebourgeois F. 2000. Climatic signals in earlywood, latewood and total ring width of
Corsican pine from western France. Annals of Forest Science 57: 155–164.
Lejeune R, McMullen L, Atkins M. 1961. The influence of logging on Douglas fir beetle
populations. The Forestry Chronicle 37: 308–314.
LeMay V, Pommerening A, Marshall P. 2017. Spatio-temporal structure of multi-storied,
multi-aged interior Douglas for (Pseudotsuga menziesii var. glauca) stands. Journal of
Ecology 97: 1062-1074
Leopold A. 1949. A Sand County Almanac. New York, NY: Oxford University Press, Inc.
226p.
Lescop-sinclair K, Payette S. 1995. Recent Advance of the Arctic Treeline Along the
Eastern Coast of Hudson Bay. Journal of Ecology 83: 929–936.
Lewis K, Thompson D, Hartley I, Pasca S. 2006. Wood decay and degradation in
standing lodgepole pine (Pinus contorta var. latfolia Engelman.) killed by mountain pine
beetle (Dendroctonus ponderosa Hopkins: Coleoptera). Victoria, BC: Natural Resources
Canada, Canadian Forest Service, Pacific Forestry Centre. Mountain Pine Beetle Initiative
working paper 2006-11. 18p.
Lloret F, Keeling EG, Sala A. 2011. Components of tree resilience: Effects of successive
low-growth episodes in old ponderosa pine forests. Oikos 120: 1909–1920.
Loescher WH, Mccamant T, Keller JD. 1990. Carbohydrate reserves, translocation, and
storage in woody plant roots. HortScience 25: 274–281.
Long J, Smith F. 2000. Restructuring the forest: Goshawks and the restoration of
southwestern ponderosa pine. Journal of Forestry 98: 25–30.
Maclauchlan L, Buxton K, 2016. 2016 Overview of Forest Health Conditions in Southern
British Columbia. Ministry of Forests, Lands and Natural Resource Operations.
Maclauchlan L, Brooks J. 2009. Influence of past forestry practices on western spruce
budworm defoliation and associated impacts in southern British Columbia. Journal of
Ecosystems and Management 10: 37–49.

232

Maclauchlan L, Brooks J, Hodge J. 2006. Analysis of historic western spruce budworm
defoliation in south central British Columbia. Forest Ecology and Management 226: 351–
356.
Maguire DA, Hann DW. 1990. Constructing models for direct prediction of 5-year crown
recession in southwestern Oregon Douglas-fir. Canadian Journal of Forest Research 20:
1044–1052.
Makinen H, Isomäki A, Hongisto T. 2005. Effect of half-systematic and systematic
thinning on the increment of Scots pine and Norway spruce in Finland. Forestry 79: 103–
121.
Manion PD. 1981. Tree disease concepts. Englewood Cliffs, NJ: Prentice-Hall, Inc. 399p.
Mann ME, Rahmstorf S, Kornhuber K, Steinman BA, Miller SK, Coumou D. 2017.
Influence of anthropogenic climate change on planetary wave resonance and extreme
weather events. Scientific Reports 7: 45242.
Marciniak A. 2015. Shifting disturbances in a warming environment: The western spruce
budworm and Douglas-fir beetle in British Columbia. MSc Thesis, University of British
Columbia.
Mazza G, Amorini E, Cutini A, Manetti MC. 2011. The influence of thinning on rainfall
interception by Pinus pinea L. in Mediterranean coastal stands (Castel Fusano-Rome).
Annals of Forest Science 68: 1323–1332.
McGregor MD. 1962. Evaluation of the Douglas-fir beetle infestation: North Fork
Clearwater River drainage, northern Idaho, 1970-73. USDA Forest Service, Northern
Region. Report no. 74-7.
McMullen L, Atkins M. 1962. The life history and habits of Scolytus unispinosus
LeConte (Coleoptera: Scolytidae) in the interior of British Columbia. The Canadian
Entomologist 94: 269–275.
Meng SX, Rudnicki M, Lieffers VJ, Reid DEB, Silins U. 2006. Preventing crown
collisions increases the crown cover and leaf area of maturing lodgepole pine. Journal of
Ecology 94: 681–686.
Ministry of Agriculture and Lands. 2011. Land use objectives for the Cariboo-Chilcotin
land use plan (CCLUP) area. Ministry of Agriculture and Lands Integrated Land
Management Bureau Ministerial Order. 20p.
Ministry of Forests Lands and Natural Resource Operations. 2012. Beyond the beetle:
A mid-term timber supply action plan. 19p.
Mitchell R, Waring R, Pitman G. 1983. Thinning lodgepole pine increases tree vigor and
resistance to mountain pine beetle. Forest Science 29: 204–211.

233

Mölder I, Leuschner C. 2014. European beech grows better and is less drought sensitive
in mixed than in pure stands: Tree neighbourhood effects on radial increment. Trees 28:
777–792.
Molina AJ, del Campo AD. 2012. The effects of experimental thinning on throughfall and
stemflow: A contribution towards hydrology-oriented silviculture in Aleppo pine
plantations. Forest Ecology and Management 269: 206–213.
Morris JL, Higuera PE, Haberle S, Whitlock C. 2017. Modern pollen from small
hollows reflects Athrotaxis cupressoides density across a wildfire gradient in subalpine
forests of the Central Plateau, Tasmania, Australia. The Holocene: 95968361770222.
Müller J, Noss RF, Bussler H, Brandl R. 2010. Learning from a ‘benign neglect strategy’
in a national park: Response of saproxylic beetles to dead wood accumulation. Biological
Conservation 143: 2559–2569.
Murdock T, Taylor S, Flower A, Mehlenbacher A, Montenegro A, Zwiers FW, Alfaro
RI, Spittlehouse DL. 2013. Pest outbreak distribution and forest management impacts in a
changing climate in British Columbia. Environmental Science & Policy 26: 75–89.
Muzika R, Guyette R. 2004. A dendrochronological analysis of red oak borer abundance
Pages 102-105 in Upland oak ecology symposium: history, current conditions, and
sustainability. Spetich MA editor. 2004. Asheville, NC: Southern Research Station, Forest
Service, U.S. Department of Agriculture. General Technical Report SRS–73. 311p.
Nadkarni NM, Sumera MM. 2004. Old-growth forest canopy structure and its
relationship to throughfall interception. Forest Science 50: 290–298.
Nanko K, Onda Y, Kato H, Gomi T. 2016. Immediate change in throughfall spatial
distribution and canopy water balance after heavy thinning in a dense mature Japanese
cypress plantation. Ecohydrology 9: 300–314.
Nealis VG. 2016. Comparative ecology of conifer-feeding spruce budworms (Lepidoptera:
Tortricidae). The Canadian Entomologist 128: S33–S57.
Noss RF. 2001. Beyond Kyoto: Forest management in a time of rapid climate change.
Conservation Biology 15: 578–590.
Nowacki GJ, Abrams MD. 1997. Radial-growth averaging criteria for reconstructing
disturbance histories from presettlement-origin oaks. Ecological Monographs 67: 225–
249.
Oliver CD, Larson BC. 1996. Forest stand dynamics: Update edition. New York, NY:
John Wiley and Sons. 544p.

234

Paulson L. 1995. Monitoring and dynamics of a Douglas-fir beetle outbreak in Jasper
National Park, Alberta. Journal of the Entomological Society of British Columbia 92: 17–
24.
Pickett S, White P, editors. 1985. The ecology of natural disturbance and patch
dynamics. San Diego, CA: Academic Press Ltd. 472p.
Powers J, Sollins P, Harmon M, Jones J. 1999. Plant-pest interactions in time and space:
a Douglas-fir bark beetle outbreak as a case study. Landscape Ecology 14: 105–120.
Pypker TG, Bond BJ, Link TE, Marks D, Unsworth MH. 2005. The importance of
canopy structure in controlling the interception loss of rainfall: Examples from a young
and an old-growth Douglas-fir forest. Agricultural and Forest Meteorology 130: 113–129.
R Core Team, 2017. R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Raffa KF, Aukema BH, Bentz BJ, Carroll AL, Hicke J a., Turner MG, Romme WH.
2008. Cross-scale Drivers of Natural Disturbances Prone to Anthropogenic Amplification:
The Dynamics of Bark Beetle Eruptions. BioScience 58: 501.
Raffa KF, Smalley EB. 1988. Response of red and jack pines to inoculation with
microbial associates of the pine engraver, Ips pini (Coleoptera: Scolytidae). Canadian
Journal of Forest Research 18: 581–586.
Ray RC, Behera S. 2011. Bio-fuel agro-industrial production system in sweet sorghum.
Chapter 2 in Sorghum: Cultivation, Varieties and Uses. Pereira TD, editor. 2011.
Hauppauge , NY: Nova Science Publishers. 181p.
Robson JRM, Conciatori F, Tardif JC, Knowles K, 2015. Tree-ring response of jack
pine and scots pine to budworm defoliation in central Canada. Forest Ecology and
Management, 347, 83-95.
Rogers, W. H. 1993. sg17: Regression standard errors in clustered samples. Stata
Technical Bulletin 13: 19–23. Reprinted in Stata Technical Bulletin Reprints, vol. 3, pp.
88–94. College Station, TX: Stata Press.
Romme WH, Despain DG. 1989. Historical perspective on the yellowstone fires of 1988.
BioScience 39: 695–699.
Ross DW, Daterman GE. 1997. Using pheromone-baited traps to control the amount and
distribution of tree mortality during outbreaks of the Douglas-fir beetle. Forest Science 43:
65–70.
Ross D, Wallin K. 2008. High release rate 3-methylcyclohex-2-en-1-one dispensers
prevent Douglas-fir beetle (Coleoptera: Curculionidae) infestation of live Douglas-fir.
Journal of Economic Entomology 101: 1826–1830.
235

Rudinsky J. 1966. Host selection and invasion by the Douglas-fir beetle, Dendroctonus
pseudotsugae Hopkins, in coastal Douglas-fir forests. The Canadian Entomologist 98: 98–
111.
Rudinsky J, Vite J. 1956. Effects of temperature upon the activity and the behavior of the
Douglas-fir beetle. Forest Science 2: 258-267.
Sauchyn D, Beaudoin A. 1998. Recent environmental change in the southwestern
Canadian Plains. The Canadian Geographer 42: 337–353.
Seymour R, Hunter M. 1992. New forestry in eastern spruce-fir stands: Principles and
applications to Maine. Orono, ME: Maine Agricultural and Forest Experiment Station,
University of Maine. Miscellaneous Publication 716. 41p.
Sherriff R, Berg E, Miller A. 2011. Climate variability and spruce beetle (Dendroctonus
rufipennis) outbreaks in south-central and southwest Alaska. Ecology 92: 1459–1470.
Shore TL, Brooks JE, Stone JE, editors. 2004. Mountain pine beetle symposium:
Challenges and solutions, October 30-31, 2003, Kelowna, British Columbia, Canada.
Victoria, BC: Pacific Forestry Centre, Canadian Forest Service, Natural Resources Canada.
Information Report BC-X-399. 298p.
Shore TL, Safranyik L, Hawkes BC, Taylor SW. 2006. Effects of the mountain pine
beetle on lodgepole pine stand structure and dynamics. Pages 95-116 in The mountain pine
beetle: A synthesis of biology, management, and impacts on lodgepole pine. Safranyik L,
Wilson B, editors. 2006. Victoria, BC: Pacific Forestry Centre, Canadian Forest Service,
Natural Resources Canada. 304p.
Smith KT, Arbellay E, Falk DA, Sutherland KE. 2010. Macroanatomy and
compartmentalization of recent fire scars in three North American conifers. Canadian
Journal of Forest Research 46: 535-542
Smith DM, Larson BC, Kelty MJ, Ashton PMS. 1997. The Practice of silviculture:
Applied forest ecology, 9th ed. Hoboken, NJ: John Wiley and Sons, Inc. 560p.
Sohn JA, Gebhardt T, Ammer C, Bauhus J, Häberle K-H, Matyssek R, Grams TEE.
2013. Mitigation of drought by thinning: Short-term and long-term effects on growth and
physiological performance of Norway spruce (Picea abies). Forest Ecology and
Management 308: 188–197.
Speer JH. 2010. Fundamentals of tree-ring research. Tucson, AZ: University of Arizona
Press. 368p.
Stahle DW, Cook ER, Cleaveland MK, Therrell MD, Meko DM, Grissino-Mayer HD,
Watson E, Luckman BH. 2000. Tree-ring data document 16th century megadrought over
North America. Eos, Transactions American Geophysical Union 81: 121.

236

Steen, O., and R. Coupé. 1997. A field guide to forest site identification and
interpretation for the Cariboo Forest Region. Ministry of Forests, Research Program. URL
http://www.for.gov.bc.ca/hfd/pubs/docs/lmh/lmh39.htm
Sterba H, Andrae F, Pambudhi F. 1993. Crown efficiency of oak standards as affected by
mistletoe and coppice removal. Forest Ecology and Management 62: 39–49.
Stiell WM. 1982. Growth of clumped vs. equally spaced trees. Forestry Chronicle 58: 23–
25.
Stogsdili WR, Wittwer RF, Hennessey TC, Dougherty PM. 1992. Water use in thinned
loblolly pine plantations. Forest Ecology and Management 50: 233–245.
Stokes M, Smiley T. 1968. An introduction to tree-ring dating. Tucson, AZ: University of
Arizona Press. 73p.
Strang RM, Parminter J V. 1980. Conifer encroachment on the Chilcotin grasslands of
British Columbia. The Forestry Chronicle 56: 13–18.
Sun X, Onda Y, Chiara S, Kato H, Gomi T. 2015. The effect of strip thinning on spatial
and temporal variability of throughfall in a Japanese cypress plantation. Hydrological
Processes 29: 5058–5070.
Swetnam TW, Thompson MA, Sutherland EK. 1985. Using dendrochronology to
measure radial growth of defoliated trees. Washington, D.C.: Cooperative State Research
Service, Forest Service, U. S. Department of Agriculture. Agriculture Handbook No. 639.
39p.
Teklehaimanot Z, Jarvis PG, Ledger DC. 1991. Rainfall interception and boundary layer
conductance in relation to tree spacing. Journal of Hydrology 123: 261–278.
Thomson A, Benton R. 2007. A 90-year sea warming trend explains outbreak patterns of
western spruce budworm on Vancouver Island. The Forestry Chronicle 83: 867–869.
Timber Pricing Branch. 2016. B.C. Interior Log Market Report for the 3 month period
January 1, 2016 to March 31, 2016. Ministry of Forests, Lands and Natural Resource
Operations.
Turner MG. 2010. Disturbance and landscape dynamics in a changing world. Ecology 91:
2833–2849.
Veblen TT, Hadley KS, Reid MS, Rebertus AJ. 1991. Methods of detecting past spruce
beetle outbreaks in Rocky Mountain subalpine forests. Canadian Journal of Forest
Research 21: 242–254.
Vité J, Rudinsky J. 1957. Contribution toward a study of Douglas fir beetle development.
Forest Science 3: 156-167.
237

Wallentin C, Nilsson U. 2011. Initial effect of thinning on stand gross stem-volume
production in a 33-year-old Norway spruce (Picea abies (L.) Karst.) stand in Southern
Sweden. Scandinavian Journal of Forest Research 26: 21–35.
Walters J. 1956. Biology and Control of the Douglas-Fir Beetle in the Interior of British
Columbia. Publication no. 975 March. Canadian Department of Agriculture.

Walters J, McMullen L. 1956. Life history and habits of Pseudohylesinus nebulosus
(LeConte)(Coleoptera: Scolytidae) in the interior of British Columbia. The Canadian
Entomologist 88: 197-202.
Waring RH. 1987. Characteristics of trees predisposed to die. Bioscience 37: 569–574.
Watson E, Luckman BH. 2002. The dendroclimatic signal in Douglas-fir and ponderosa
pine tree-ring chronologies from the southern Canadian Cordillera. Canadian Journal of
Forest Research 32: 1858–1874.
Weaver H. 1947. Fire-Nature’s thinning agent in ponderosa pine stands. Journal of
Forestry 45: 437–444.
Wellington WG, & Henson WR 1947. Notes on the effects of physical factors on the
spruce budworm, Choristoneura fumiferana (Clem.). The Canadian Entomologist, 79(910): 168–170.
Welsh C, Lewis KJ, Woods AJ, 2014. Regional outbreak dynamics of Dothistroma needle
blight linked to weather patterns in British Columbia, Canada. Canadian journal of forest
research, 44(3), 212-219.
Whitehead R, Martin P, Powelson A. 2001. Reducing stand and landscape susceptibility
to mountain pine beetle. Victoria, BC: Forest Practices Branch, British Columbia Ministry
of Forests. 11p.
Williamson R, Price F. 1971. Initial thinning effects in 70-to 150-year-old Douglas-fir-western Oregon and Washington. Portland, OR: Pacific Northwest Research Station, Forest
Service, U.S. Department of Agriculture. USDA Forest Service Research Paper PNW-117.
21p.
Wolfe SA, Huntley DJ, David PP, Ollerhead J, Sauchyn DJ, MacDonald GM. 2001.
Late 18th century drought-induced sand dune activity, Great Sand Hills, Saskatchewan.
Canadian Journal of Earth Sciences 38: 105–117.
Wong BL, Berryman AA. 1977. Host resistance to the fir engraver beetle. 3. Lesion
development and containment of infection by resistant Abies grandis inoculated with
Trichosporium symbioticum. Canadian Journal of Botany 55: 2358–2365.
Wong C, Iverson K. 2004. Range of natural variability: Applying the concept to forest
management in central British Columbia. Journal of Ecosystems and Management 4: 1–14.
238

Wong CM, Lertzman KP. 2001. Errors in estimating tree age: Implications for studies of
stand dynamics. Canadian Journal of Forest Research 31: 1262–1271.
Woods A, Heppner D, Kope H, Burleigh J, Maclauchlan L. 2010. Forest health and
climate change: A British Columbia perspective. The Forestry Chronicle 86: 412–422.
Zhang QB, Alfaro RI, 2003. Spatial synchrony of the two‐year cycle budworm outbreaks
in central British Columbia, Canada. Oikos, 102(1), 146-154.
Zang C, Biondi F. 2015. treeclim: An R package for the numerical calibration of proxyclimate relationships. Ecography 38: 431–436.
Zhang J, Huang S, He F. 2015. Half-century evidence from western Canada shows forest
dynamics are primarily driven by competition followed by climate. Proceedings of the
National Academy of Sciences 112: 4009-4014.
Zirlewagen D, von Wilpert K. 2001. Modeling water and ion fluxes in a highly
structured, mixed- species stand. Forest Ecology and Management 143: 27–37.

239

8. Appendix A: R Scripts Used in Analyses
8.1. Introduction
The following scripts are those used for the analyses in this dissertation, in some
cases modified and noted to be used as examples with data stored in the UNBC Tree Ring
Lab folder called “R Scripts for Dendro.” If these folders are not available to you and you
wish to acquire the example data, write to both nthompso@unbc.ca and
neil.thompson@maine.edu, both of which should be monitored for the foreseeable future.
These scripts “should” work with any properly formatted ringwidth measurement files in
Tucson or Heidelberg format (use the program TRiCYCLE available at
http://www.tridas.org/software.php to convert between formats). Substituting
[format=c(“tucson”)] for [format=c(“heidelberg”)] or vice versa will allow you to use the
other format.

240

8.2. Ring width to basal area increments (Chapters 2, 3, and 4)
# This script converts raw ring width measurements from the Tucson format to basal area
increments in mm2
# Last edited by Neil Thompson on January 21, 2016
# Please copy this script and examples to your own directory so you do not overwrite the
original
# Developed using R 3.2.2.
# Example dataset provided is a subsample of one site from Neil Thompson's Ph.D.
dissertation work
# You may need to use the latest version of dplR to use the tree averaging command. The
latest version is available here https://r-forge.r-project.org/R/?group_id=1105
# Load dplR (https://cran.r-project.org/web/packages/dplR/index.html)
library(dplR)
# Set working directory
setwd(choose.dir(default = "", caption = "Select folder"))
# To confirm success
getwd()
# It won't take you to the directory the first time, so you will have to navigate manually. It
will be automatic after the first one.
# Read the raw ring width data file (Tucson format) in as the tag "dro"
dro <- read.rwl (file.choose (new = FALSE), format = c("tucson"))
# Read the distance to pith data file, measured in millimeters. Two labeled columns
("series" and "d2pith" in mm) must be present and match the above format
d2p <- read.csv(file.choose (new = FALSE))
# Calculate the annual BAI of each series
droB <- bai.in(rwl = dro, d2pith = d2p)
# If no d2pith file is available, then substitute "d2pith = NULL"
# To average the cores within a tree (example COL01B)
droB.ids <- read.ids(droB, stc = c(3, 2, 1))
# where 3 is the number of characters defining site (COL)
# 2 is the number of characters defining the tree (01)
# 1 is the character defining core within tree (B)
View(droB.ids)
# Confirm that there are the correct number of trees, cores, sites
241

# Calculate the tree level mean across cores, retaining years with missing values
droB.treeMean <- treeMean(droB, droB.ids, na.rm=TRUE)
# Tree names are dropped but order is retained, will have to add in later
# Write a CSV file of the output
write.csv(droB.treeMean, file = "droB.csv")
# Convert mm2 to cm2 if desired
# Manually check your output: calculate the basal area increment of a single ring using a
ruler and the area formula (pi*r2)
# Calculate the total basal area in Excel by summing each annual increment.
# Reverse the above area calculation to calculate the diameter [2*(sqrt(area/pi))]. This
better look something like your field DBH measurements.

242

8.3. Climate correlations and response functions (Chapters 2 and 5)
# COPY THE EXAMPLE FOLDER TO YOUR OWN DIRECTORY so that you do not
overwrite or clutter the original
# This script will take you through climate correlations and response functions using the
latest R package as of January 2016
# Prepared by Neil Thompson, 12 January 2016 for use in the UNBC Dendrochronology
lab
# Download R and packages dplR and treeclim
# http://cran.us.r-project.org/
# http://www.r-bloggers.com/installing-r-packages/
# https://cran.r-project.org/web/packages/treeclim/index.html
# https://cran.r-project.org/web/packages/dplR/index.html
# Materials (examples provided in this folder):
# 1. Crossdated ring width measurements in Tucson format
# 2. Monthly temperature data in column format (see example; columns are Year, jan, feb,
mar, etc.)
# 3. Monthly precipitation data in column format (see example; columns are Year, jan, feb,
mar, etc.)
# Monthly data for BC are available here: http://www.ec.gc.ca/dccha-ahccd/ but will
require some curation to meet the format (no heading labels or extra columns)
# To use these scripts, copy them to the R command console in the order that they appear
# My notes are indicated by #hashtags and are ignored by the program; they may be copied
with commands
# The following line is a command, having no hashtag
2+2
# The following line stores the mathematical argument "2+2" as "y" using the < and - (<-)
characters
y <- 2+2
y
# This method is used to store complex equations and datasets as simple characters
throughout the script
##############Data input and prep for Treeclim############
# Copy and paste everything from here to ########Correlation###### to the command
console and follow the prompts
# Set the working directory interactively (the version of this folder that you copied to your
own directory)
243

setwd(choose.dir(default = "", caption = "Select folder"))
# Clear the workspace
rm(list=ls())
# Load dplR and bootRes
library(dplR)
library(treeclim)
# Load temperature and precipitation tables interactively
temp <- read.table (file.choose (new = FALSE))
prec <- read.table (file.choose (new = FALSE))
# Load site interactively
X1<- read.rwl (file.choose (new = FALSE), format = c("tucson"))
# Detrending, 50 year cubic smoothing spline (to change spline length to X, set"nyrs=X"
below)
X2 <- detrend(X1, y.name = names(X1), make.plot = TRUE, method = c("Spline"),
nyrs=50, f = 0.5, verbose = FALSE, return.info = FALSE)
# Build a residual chronology using a biweight mean
X3 <- chron(X2, prefix="xx", biweight = TRUE, prewhiten = TRUE)

##################Correlation#######################
X <- dcc (chrono=X3, climate = list(temp, prec), selection = -6:9, method = "correlation",
moving = FALSE, var_names = c("temp", "prec"))
plot(X)
# Save a plot ###########WILL OVERWRITE EXISTING##############
jpeg("Correlation.jpg", width=12, height=8, units = "in", res =300)
plot(X)
dev.off()
XX <- dcc (chrono=X3, climate = list(temp, prec), selection = -6:9, method =
"correlation", moving = TRUE, win_size=35, win_offset = 5, var_names = c("temp",
"prec"))
plot(XX)
# Save a plot ###########WILL OVERWRITE EXISTING##############
jpeg("Moving Correlation.jpg", width=12, height=8, units = "in", res =300)
244

plot(XX)
dev.off()
XXStat <- dcc (chrono=X3, climate = list(temp, prec), selection = -6:9, method =
"correlation", moving = TRUE, win_size=35, win_offset = 5, var_names = c("temp",
"prec"), ci = 0.05, boot = "stationary")
plot(XXStat)

#####################Response Function##############
X1 <- dcc (chrono=X3, climate = list(temp, prec), selection = -6:9, method = "response",
moving = FALSE, var_names = c("temp", "prec"))
plot(X1)
# Save a plot ###########WILL OVERWRITE EXISTING##############
jpeg("Response Function.jpg", width=12, height=8, units = "in", res =300)
plot(X1)
dev.off()
XX1 <- dcc (chrono=X3, climate = list(temp, prec), selection = -6:9, method = "response",
moving = TRUE, win_size=35, win_offset = 5, var_names = c("temp", "prec"))
plot(XX1)
# Save a plot ###########WILL OVERWRITE EXISTING##############
jpeg("Moving Response.jpg", width=12, height=8, units = "in", res =300)
plot(XX1)
dev.off()

245

8.4. Generation of .tre files (Chapter 5)
This script repeats for each of the 45 sites, but only two iterations are shown for brevity.
# Load dplR (https://cran.r-project.org/web/packages/dplR/index.html)
library(dplR)
# Set working directory
setwd(choose.dir(default = "", caption = "Select folder"))
# To confirm success
getwd()
# It won't take you to the directory the first time, so you will have to nagivage manually. It
will be automatic after the first one.

#####ABR##############################################################
# Read the raw ring width data file (Heidelberg format) in as the tag "X1"
X1 <- read.rwl ("ABR(01).fh", format = c("heidelberg"))
# Detrending, modnegexp
X1a <- detrend(X1, y.name = names(X1), make.plot = TRUE, method = c("ModNegExp"),
verbose = FALSE, return.info = FALSE)
# Detrending, 150 year cubic smoothing spline (to change spline length to X, set"nyrs=X"
below)
X2 <- detrend(X1a, y.name = names(X1a), make.plot = TRUE, method = c("Spline"),
nyrs=50, f = 0.5, verbose = FALSE, return.info = FALSE)
# To average the cores within a tree (example COL01B)
X2.ids <- read.ids(X2, stc = c(3, 2, 1))
# where 3 is the number of characters defining site (COL)
# 2 is the number of characters defining the tree (01)
# 1 is the character defining core within tree (B)
View(X2.ids)
# Confirm that there are the correct number of trees, cores, sites
# Calculate the tree level mean across cores, retaining years with missing values
X2.M <- treeMean(X2, X2.ids, na.rm=TRUE)
# Tree names are dropped but order is retained, will have to add in later
246

# Write a tre file of the output
write.tucson(X2.M, fname = "ABR.tre")
#######END########
#####ACS##############################################################
# Read the raw ring width data file (Heidelberg format) in as the tag "X1"
X1 <- read.rwl ("ACS(01).fh", format = c("heidelberg"))
# Detrending, modnegexp
X1a <- detrend(X1, y.name = names(X1), make.plot = TRUE, method = c("ModNegExp"),
verbose = FALSE, return.info = FALSE)
# Detrending, 150 year cubic smoothing spline (to change spline length to X, set"nyrs=X"
below)
X2 <- detrend(X1a, y.name = names(X1a), make.plot = TRUE, method = c("Spline"),
nyrs=50, f = 0.5, verbose = FALSE, return.info = FALSE)

# To average the cores within a tree (example COL01B)
X2.ids <- read.ids(X2, stc = c(3, 2, 1))
# where 3 is the number of characters defining site (COL)
# 2 is the number of characters defining the tree (01)
# 1 is the character defining core within tree (B)
View(X2.ids)
# Confirm that there are the correct number of trees, cores, sites
# Calculate the tree level mean across cores, retaining years with missing values
X2.M <- treeMean(X2, X2.ids, na.rm=TRUE)
# Tree names are dropped but order is retained, will have to add in later
# Write a tre file of the output
write.tucson(X2.M, fname = "ACS.tre")
#######END########

247

9. Appendix B: AIC Calculation and Top Ranked Models for Chapter 3
Calculation tables of Akaike’s Information Criterion for small sample sizes (AICc)
from the log likelihood and sample size in each drought year. Models with AICc
differences (“AIC DIFF”) <2 were considered to be ties and are full outputs of the
multinomial logistic regression are reported in the tables following the AICc calculation.
n=907

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

n=904

minAICc=

1988

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-863.72
-909.06
-870.31
-868.58
-964.33
-924.29
-904.74
-879.44
-912.74

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

AICC

1767.43
1854.13
1772.61
1773.15
1932.66
1868.58
1829.47
1782.88
1857.48

minAICc=

1959

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-827.50
-895.96
-839.66
-827.62
-981.36
-952.62
-847.31
-843.89
-898.13

1768.4

1695.00
1827.91
1711.32
1691.24
1966.72
1925.25
1714.62
1711.78
1828.25
248

AIC
DIFF w

1768.4
0.0 0.87
1854.9 86.5 <0.01
1773.2
4.8 0.08
1773.9
5.5 0.05
1932.7 164.3 <0.01
1868.8 100.4 <0.01
1829.7 61.3 <0.01
1783.2 14.8 <0.01
1858.1 89.7 <0.01

exp(1/2delta)
1.0000
0.0000
0.0887
0.0627
0.0000
0.0000
0.0000
0.0006
0.0000

1692.0

AICC

AIC
DIFF w

1696.0
1828.7
1711.9
1692.0
1966.7
1925.5
1714.9
1712.1
1828.9

3.9
136.7
19.9
0.0
274.7
233.5
22.9
20.1
136.8

0.12
<0.01
<0.01
0.87
<0.01
<0.01
<0.01
<0.01
<0.01

exp(1/2delta)
0.1397
0.0000
0.0000
1.0000
0.0000
0.0000
0.0000
0.0000
0.0000

n=896

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

n=893

minAICc=

1946

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-924.59
-926.48
-932.89
-924.88
-983.28
-941.68
-965.66
-934.68
-930.17

1938

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

1886.5

AICC

AIC
DIFF w

exp(1/2delta)

1889.18
1888.95
1897.79
1885.76
1970.55
1903.36
1951.33
1893.35
1892.35

1890.1
1889.7
1898.4
1886.5
1970.6
1903.6
1951.6
1893.7
1893.0

3.6
3.2
11.9
0.0
84.0
17.1
65.0
7.2
6.4

0.1651
0.2023
0.0026
1.0000
0.0000
0.0002
0.0000
0.0277
0.0401

minAICc=

1785.8

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-872.43
-943.77
-881.46
-878.35
-974.48
-964.16
-908.38
-893.08
-944.57

1784.87
1923.55
1794.93
1792.70
1952.97
1948.32
1836.77
1810.16
1921.14

249

0.11
0.18
<0.01
0.87
<0.01
<0.01
<0.01
0.02
0.03

AICC

AIC
DIFF w

1785.8
1924.3
1795.5
1793.5
1953.0
1948.6
1837.0
1810.5
1921.8

0.0
138.5
9.7
7.6
167.1
162.7
51.2
24.7
135.9

0.97
<0.01
0.01
0.02
<0.01
<0.01
<0.01
<0.01
<0.01

exp(1/2delta)
1.0000
0.0000
0.0078
0.0218
0.0000
0.0000
0.0000
0.0000
0.0000

n=890

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

n=860

minAICc=

1931

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-896.32
-911.52
-903.94
-897.61
-977.49
-963.51
-908.96
-906.91
-911.73

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

AICC

1832.65
1859.04
1839.87
1831.21
1958.98
1947.02
1837.92
1837.82
1855.47

minAICc=

1905

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-791.34
-858.38
-792.32
-801.37
-928.94
-903.14
-838.42
-815.36
-862.46

1832.0

1622.68
1752.76
1616.65
1638.75
1861.89
1826.29
1696.84
1654.71
1756.93

250

AIC
DIFF w

1833.6
1.6 0.29
1859.8 27.8 <0.01
1840.5
8.5 0.01
1832.0
0.0 0.64
1959.0 127.0 <0.01
1947.3 115.3 <0.01
1838.2
6.2 0.03
1838.2
6.2 0.03
1856.1 24.1 <0.01

exp(1/2delta)
0.4456
0.0000
0.0143
1.0000
0.0000
0.0000
0.0458
0.0456
0.0000

1617.3

AICC

AIC
DIFF w

1623.7
1753.6
1617.3
1639.6
1861.9
1826.5
1697.1
1655.1
1757.6

6.4
136.3
0.0
22.3
244.6
209.3
79.8
37.8
140.3

0.04
<0.01
0.96
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01

exp(1/2delta)
0.0410
0.0000
1.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000

n=794

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

n=737

minAICc=

1869

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-789.47
-796.29
-793.84
-793.49
-928.94
-843.13
-808.50
-800.26
-796.62

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

AICC

1618.94
1628.58
1619.68
1622.97
1861.89
1706.27
1637.00
1624.52
1625.24

minAICc=

1842

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-632.26
-704.43
-669.24
-640.55
-792.18
-731.76
-691.11
-671.40
-707.07

1620.0

1304.52
1444.85
1370.49
1317.11
1588.35
1483.52
1402.21
1366.81
1446.14

251

AIC
DIFF w

1620.0
0.0 0.47
1629.5
9.4 0.01
1620.4
0.4 0.80
1623.9
3.8 0.14
1861.9 241.9 <0.01
1706.5 86.5 <0.01
1637.3 17.3 <0.01
1624.9
4.9 0.08
1625.9
5.9 0.05

exp(1/2delta)
1.0000
0.0089
0.8374
0.1473
0.0000
0.0000
0.0002
0.0865
0.0519

1305.7

AICC

AIC
DIFF w

1305.7
1445.8
1371.2
1318.1
1588.4
1483.8
1402.5
1367.2
1446.9

0.0
140.1
65.5
12.4
282.7
178.1
96.8
61.5
141.2

1.00
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01

exp(1/2delta)
1.0000
0.0000
0.0000
0.0021
0.0000
0.0000
0.0000
0.0000
0.0000

n=588

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

n=475

minAICc=

1800

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-529.63
-585.78
-537.19
-530.31
-642.87
-621.54
-551.48
-549.00
-586.00

1772

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

1097.8

AICC

AIC
DIFF w

1099.26
1207.55
1106.38
1096.62
1289.75
1263.07
1122.96
1121.99
1204.00

1100.7
1208.8
1107.3
1097.8
1289.8
1263.5
1123.3
1122.5
1204.9

2.9
110.9
9.5
0.0
191.9
165.6
25.5
24.7
107.1

minAICc=

845.6

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-407.38
-466.93
-408.07
-407.96
-507.84
-490.51
-414.18
-410.49
-467.14

854.76
969.86
848.14
851.92
1019.68
1001.02
848.36
844.97
966.28

252

0.19
<0.01
0.01
0.81
<0.01
<0.01
<0.01
<0.01
<0.01

AICC

AIC
DIFF w

856.6
971.4
849.3
853.4
1019.7
1001.5
848.8
845.6
967.5

11.0
125.7
3.7
7.8
174.1
155.8
3.2
0.0
121.8

<0.01
<0.01
0.13
0.02
<0.01
<0.01
0.16
0.81
<0.01

exp(1/2delta)
0.2319
0.0000
0.0086
1.0000
0.0000
0.0000
0.0000
0.0000
0.0000

exp(1/2delta)
0.0042
0.0000
0.1591
0.0205
0.0000
0.0000
0.2032
1.0000
0.0000

n=438

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

n=354

minAICc=

1760

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-417.31
-417.31
-399.91
-370.90
-479.51
-453.71
-405.30
-407.32
-421.40

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

AICC

874.62
870.62
831.82
777.80
963.02
927.42
830.60
838.64
874.80

minAICc=

1734

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-286.14
-320.21
-300.78
-290.24
-382.77
-354.57
-307.16
-306.36
-323.07

779.4

876.6 97.2 <0.01
872.3 92.8 <0.01
833.1 53.7 <0.01
779.4
0.0 0.81
963.0 183.6 <0.01
927.9 148.5 <0.01
831.1 51.7 <0.01
839.4 59.9 <0.01
876.1 96.7 <0.01

253

exp(1/2delta)
0.0000
0.0000
0.0000
1.0000
0.0000
0.0000
0.0000
0.0000
0.0000

614.8

AICC

612.28
676.42
633.56
616.49
769.54
729.14
634.31
636.71
678.15

AIC
DIFF w

AIC
DIFF w

614.8
0.0 0.87
678.5 63.7 <0.01
635.2 20.4 <0.01
618.5
3.7 0.13
769.6 154.8 <0.01
729.8 115.0 <0.01
635.0 20.2 <0.01
637.6 22.8 <0.01
679.8 65.0 <0.01

exp(1/2delta)
1.0000
0.0000
0.0000
0.1547
0.0000
0.0000
0.0000
0.0000
0.0000

n=292

minAICc=

1717

Full
Less resistance
Less coordinates
Less elevation
Elevation
Site level
Tree level
Mixed 1
Mixed 2

K

Log
Likelihood AIC

20
18
16
18
2
10
10
12
16

-215.42
-262.02
-219.83
-219.04
-311.30
-275.15
-227.58
-224.51
-266.23

473.6

AICC

470.84
560.03
471.66
474.09
626.60
570.30
475.16
473.03
564.45

254

AIC
DIFF w

473.9
0.3 0.27
562.5 88.9 <0.01
473.6
0.0 0.31
476.6
2.9 0.07
626.6 153.0 <0.01
571.1 97.4 <0.01
475.9
2.3 0.10
474.1
0.5 0.24
566.4 92.8 <0.01

exp(1/2delta)
0.8634
0.0000
1.0000
0.2288
0.0000
0.0000
0.3157
0.7776
0.0000

resilcd1988

Robust
Coef.

Std. Err.

z

P>z

[95% Conf.

Interval]

1

(base outcome)

2
lwmean
ahm
solar
elev
resist1988
prior1988
age1988
ba1988
easting
northing
_cons

0.029419
0.053303
0.000000
0.003558
2.188537
-0.000497
0.001735
-0.000001
0.000003
0.000004
-13.904640

0.03091510
0.03672830
0.00000039
0.00125330
0.79328470
0.00023660
0.00189490
0.00000244
0.00000581
0.00000854
14.21691000

0.95
1.45
-0.45
2.84
2.76
-2.10
0.92
-0.33
0.60
0.44
-0.98

0.341
0.147
0.652
0.005
0.006
0.036
0.360
0.739
0.550
0.663
0.328

-0.0311736
-0.0186831
-0.0000009
0.0011011
0.6337273
-0.0009604
-0.0019794
-0.0000056
-0.0000079
-0.0000130
-41.7692800

0.0900116
0.1252893
0.0000006
0.0060138
3.7433460
-0.0000329
0.0054484
0.0000040
0.0000149
0.0000205
13.9600000

3
lwmean
ahm
solar
elev
resist1988
prior1988
age1988
ba1988
easting
northing
_cons

0.094984
0.019592
-0.000001
0.001932
4.817027
-0.000658
-0.000469
0.000002
-0.000004
-0.000016
10.989060

0.03359190
0.07805370
0.00000062
0.00232020
1.08506300
0.00027130
0.00227540
0.00000300
0.00000794
0.00001490
24.26692000

2.83
0.25
-1.32
0.83
4.44
-2.43
-0.21
0.52
-0.54
-1.04
0.45

0.005
0.802
0.187
0.405
<0.0001
0.015
0.837
0.605
0.592
0.296
0.651

0.0291456
-0.1333905
-0.0000020
-0.0026160
2.6903420
-0.0011900
-0.0049282
-0.0000043
-0.0000198
-0.0000448
-36.5732400

0.1608232
0.1725743
0.0000004
0.0064790
6.9437110
-0.0001267
0.0039910
0.0000074
0.0000113
0.0000136
58.5513500

255

resilcd1959

Coef.

Robust
Std. Err.

z

P>z

[95% Conf.

Interval]

1

(base outcome)

2
lwmean
ahm
solar
resist1959
prior1959
age1959
ba1959
easting
northing
_cons

-0.056653
0.007371
0.000000
3.138056
-0.000313
-0.000742
-0.000002
-0.000004
0.000009
-1.192390

0.02541890
0.03312190
0.00000043
0.59780630
0.00021970
0.00120850
0.00000180
0.00000370
0.00000414
6.87437400

-2.23
0.22
-0.39
5.25
-1.42
-0.61
-1.30
-1.13
2.23
-0.17

0.026
0.824
0.694
<0.0001
0.154
0.539
0.192
0.260
0.026
0.862

-0.1064731
-0.0575471
-0.0000010
1.9663770
-0.0007434
-0.0031109
-0.0000059
-0.0000114
0.0000011
-14.6659200

-0.0068330
0.0722885
0.0000007
4.3097350
0.0001177
0.0016264
0.0000012
0.0000031
0.0000173
12.2811400

3
lwmean
ahm
solar
resist1959
prior1959
age1959
ba1959
easting
northing
_cons

-0.020750
0.052262
0.000000
5.933811
-0.001067
-0.004443
-0.000004
-0.000002
0.000016
-10.675210

0.03386960
0.04395850
0.00000059
0.77739240
0.00034140
0.00223280
0.00000209
0.00000545
0.00000579
10.70599000

-0.61
1.19
-0.66
7.63
-3.12
-1.99
-1.72
-0.45
2.77
-1.00

0.540
0.234
0.512
<0.0001
0.002
0.047
0.085
0.652
0.006
0.319

-0.0871330
-0.0338948
-0.0000015
4.4101500
-0.0017357
-0.0088190
-0.0000077
-0.0000131
0.0000047
-31.6585700

0.0456335
0.1384195
0.0000008
7.4574720
-0.0003974
-0.0000667
0.0000005
0.0000082
0.0000274
10.3081500

256

resilcd1946
1

lwmean
ahm
solar
resist1946
prior1946
age1946
ba1946
easting
northing
_cons

lwmean
ahm
solar
resist1946
prior1946
age1946
ba1946
easting
northing
_cons

2

3

Robust
Std. Err.

z

0.023936
0.066024
0.000000
0.603534
-0.000044
0.000312
0.000000
-0.000008
-0.000006
11.820400

0.03457120
0.02958000
0.00000034
0.41210240
0.00024730
0.00188680
0.00000218
0.00000298
0.00000370
5.36835100

0.69
2.23
0.27
1.46
-0.18
0.17
0.21
-2.77
-1.63
2.20

0.071153
0.084156
0.000000
0.691093
-0.000333
0.002199
-0.000004
-0.000014
-0.000012
21.054490

0.03432550
0.03686400
0.00000044
0.49570910
0.00022240
0.00195790
0.00000222
0.00000386
0.00000517
7.47548000

2.07
2.28
0.72
1.39
-1.50
1.12
-1.61
-3.55
-2.22
2.82

Coef.

P>z

[95% Conf.

Interval]

0.489
0.026
0.788
0.143
0.859
0.869
0.833
0.006
0.104
0.028

-0.0438225
0.0080481
-0.0000006
-0.2041714
-0.0005285
-0.0033860
-0.0000038
-0.0000141
-0.0000133
1.2986300

0.0916939
0.1239996
0.0000008
1.4112400
0.0004407
0.0040102
0.0000047
-0.0000024
0.0000012
22.3421800

0.038
0.022
0.474
0.163
0.134
0.261
0.107
<0.0001
0.027
0.005

0.0038764
0.0119040
-0.0000006
-0.2804790
-0.0007690
-0.0016384
-0.0000079
-0.0000213
-0.0000216
6.4028200

0.1384301
0.1564080
0.0000012
1.6626650
0.0001029
0.0060364
0.0000008
-0.0000061
-0.0000013
35.7061600

(base outcome)

257

resilcd1938

Coef.

Robust
Std. Err.

z

P>z

[95% Conf.

Interval]

1

(base outcome)

2
lwmean
ahm
solar
elev
resist1938
prior1938
age1938
ba1938
easting
northing
_cons

-0.030118
-0.056953
-0.000001
-0.002594
2.416910
-0.000337
-0.002982
-0.000001
-0.000001
-0.000008
13.006570

0.02701030
0.05371610
0.00000035
0.00156390
0.66871360
0.00032530
0.00189560
0.00000196
0.00000575
0.00000842
15.10658000

-1.12
-1.06
-2.07
-1.66
3.61
-1.04
-1.57
-0.31
-0.21
-0.97
0.86

0.265
0.289
0.038
0.097
<0.0001
0.300
0.116
0.755
0.832
0.332
0.389

-0.0830571
-0.1622350
-0.0000014
-0.0056586
1.1062550
-0.0009743
-0.0066972
-0.0000045
-0.0000125
-0.0000247
-16.6017800

0.0228211
0.0483282
0.0000000
0.0004716
3.7275650
0.0003007
0.0007335
0.0000032
0.0000100
0.0000083
42.6149200

3
lwmean
ahm
solar
elev
resist1938
prior1938
age1938
ba1938
easting
northing
_cons

-0.048869
-0.120137
-0.000001
-0.004057
6.106275
-0.001065
-0.003928
-0.000002
0.000003
-0.000013
13.590990

0.03268240
0.05914320
0.00000052
0.00194910
0.74073510
0.00042090
0.00255640
0.00000297
0.00000679
0.00001130
18.67833000

-1.50
-2.03
-1.86
-2.08
8.24
-2.53
-1.54
-0.55
0.43
-1.16
0.73

0.135
0.042
0.063
0.037
<0.0001
0.011
0.124
0.581
0.668
0.247
0.467

-0.1129254
-0.2360551
-0.0000020
-0.0078775
4.6544610
-0.0018904
-0.0089382
-0.0000075
-0.0000104
-0.0000351
-23.0178700

0.0151871
-0.0042180
0.0000001
-0.0002370
7.5580890
-0.0002405
0.0010826
0.0000042
0.0000162
0.0000091
50.1998400

258

A
resilcd1931

Coef.

Robust
Std. Err.

z

P>z

[95% Conf.

Interval]

1

(base outcome)

2
lwmean
ahm
solar
elev
resist1931
prior1931
age1931
ba1931
easting
northing
_cons

-0.012383
0.089744
0.000000
0.000976
1.011700
-0.000862
0.001713
0.000001
0.000008
0.000002
-15.594400

0.02634530
0.05002070
0.00000045
0.00148320
0.46614560
0.00023720
0.00190230
0.00000221
0.00000486
0.00000798
13.21523000

-0.47
1.79
0.33
0.66
2.17
-3.64
0.90
0.27
1.73
0.22
-1.18

0.638
0.073
0.742
0.510
0.030
<0.0001
0.368
0.784
0.083
0.830
0.238

-0.0640190
-0.0082946
-0.0000007
-0.0019307
0.0980716
-0.0013273
-0.0020158
-0.0000037
-0.0000011
-0.0000139
-41.4957700

0.0392527
0.1877829
0.0000010
0.0038835
1.9253290
-0.0003974
0.0054412
0.0000049
0.0000180
0.0000174
10.3069600

-0.001115
0.044860
0.000001
-0.000722
2.182757
-0.001502
0.002503
-0.000006
0.000003
-0.000009
2.236028

0.03133570
0.05945640
0.00000059
0.00204500
0.53656200
0.00040940
0.00227400
0.00000246
0.00000607
0.00001040
18.06747000

-0.04
0.75
1.03
-0.35
4.07
-3.67
1.10
-2.26
0.46
-0.86
0.12

0.972
0.451
0.305
0.724
<0.0001
<0.0001
0.271
0.024
0.646
0.390
0.902

-0.0625314
-0.0716724
-0.0000006
-0.0047302
1.1311140
-0.0023049
-0.0019544
-0.0000104
-0.0000091
-0.0000292
-33.1755600

0.0603023
0.1613923
0.0000018
0.0032861
3.2343990
-0.0007000
0.0069594
-0.0000007
0.0000147
0.0000114
37.6476200

3

lwmean
ahm
solar
elev
resist1931
prior1931
age1931
ba1931
easting
northing
_cons

259

B
resilcd1931

Coef.

Robust
Std. Err.

z

P>z

[95% Conf.

Interval]

1

(base outcome)

2
lwmean
ahm
solar
resist1931
prior1931
age1931
ba1931
easting
northing
_cons

-0.008932
0.065470
0.000000
1.020697
-0.000846
0.001858
0.000000
0.000006
-0.000003
-7.095796

0.02674990
0.05078140
0.00000046
0.46718520
0.00023360
0.00184020
0.00000219
0.00000466
0.00000502
9.13467700

-0.33
1.29
0.40
2.18
-3.62
1.01
0.17
1.23
-0.56
-0.78

0.738
0.197
0.693
0.029
<0.0001
0.313
0.866
0.219
0.574
0.437

-0.0613611
-0.0340598
-0.0000007
0.1050306
-0.0013034
-0.0017491
-0.0000039
-0.0000034
-0.0000127
-24.9994300

0.0434964
0.1649995
0.0000011
1.9363630
-0.0003876
0.0054643
0.0000047
0.0000149
0.0000070
10.8078400

3
lwmean
ahm
solar
resist1931
prior1931
age1931
ba1931
easting
northing
_cons

-0.002931
0.061290
0.000001
2.179971
-0.001513
0.002395
-0.000005
0.000005
-0.000006
-3.619915

0.03279180
0.04865020
0.00000060
0.54427840
0.00040750
0.00218900
0.00000241
0.00000463
0.00000505
9.06618800

-0.09
1.26
0.99
4.01
-3.71
1.09
-2.23
0.99
-1.15
-0.40

0.929
0.208
0.321
<0.0001
<0.0001
0.274
0.026
0.320
0.252
0.690

-0.0672014
-0.0340631
-0.0000006
1.1132050
-0.0023115
-0.0018954
-0.0000101
-0.0000045
-0.0000157
-21.3893200

0.0613400
0.1566421
0.0000018
3.2467370
-0.0007143
0.0066854
-0.0000007
0.0000137
0.0000041
14.1494900

260

resilcd1905

Coef.

Robust
Std. Err.

z

P>z

[95% Conf.

Interval]

1

(base outcome)

2
lwmean
ahm
solar
elev
resist1905
prior1905
age1905
ba1905
_cons

-0.079674
-0.051085
0.000001
-0.001980
3.612302
-0.000638
-0.000451
0.000001
2.837433

0.02784780
0.03274720
0.00000045
0.00078470
0.69772640
0.00026700
0.00227290
0.00000249
1.98128300

-2.86
-1.56
2.63
-2.52
5.18
-2.39
-0.20
0.23
1.43

0.004
0.119
0.008
0.012
<0.0001
0.017
0.843
0.820
0.152

-0.1342549
-0.1152686
0.0000003
-0.0035177
2.2447830
-0.0011614
-0.0049055
-0.0000043
-1.0458100

-0.0250935
0.0130981
0.0000021
-0.0004418
4.9798210
-0.0001148
0.0040041
0.0000054
6.7206760

3
lwmean
ahm
solar
elev
resist1905
prior1905
age1905
ba1905
_cons

-0.051762
-0.058098
0.000001
-0.005186
6.907991
-0.001586
-0.007084
0.000002
4.889696

0.03751540
0.04932430
0.00000080
0.00131280
1.19821700
0.00039300
0.00292880
0.00000330
2.84491700

-1.38
-1.18
1.63
-3.95
5.77
-4.04
-2.42
0.55
1.72

0.168
0.239
0.102
<0.0001
<0.0001
<0.0001
0.016
0.580
0.086

-0.1252913
-0.1547723
-0.0000003
-0.0077586
4.5595290
-0.0023566
-0.0128238
-0.0000046
-0.6862394

0.0217664
0.0385755
0.0000029
-0.0026124
9.2564520
-0.0008160
-0.0013432
0.0000083
10.4656300

261

A
resilcd1869
1

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

2
lwmean
ahm
solar
elev
resist1869
prior1869
age1869
ba1869
easting
northing
_cons

-0.001019
-0.073960
0.000000
-0.002582
0.474111
-0.000673
-0.000756
0.000002
-0.000009
-0.000016
28.431750

0.03927100
0.06315590
0.00000043
0.00199900
0.61583050
0.00030950
0.00272770
0.00000335
0.00000714
0.00001030
19.28659000

-0.03
-1.17
0.37
-1.29
0.77
-2.18
-0.28
0.71
-1.26
-1.53
1.47

0.979
0.242
0.714
0.196
0.441
0.030
0.782
0.477
0.206
0.126
0.140

-0.0779889
-0.1977433
-0.0000007
-0.0065001
-0.7328950
-0.0012799
-0.0061023
-0.0000042
-0.0000230
-0.0000358
-9.3692750

0.0759508
0.0498233
0.0000010
0.0013360
1.6811160
-0.0000668
0.0045899
0.0000089
0.0000050
0.0000044
66.2327700

3
lwmean
ahm
solar
elev
resist1869
prior1869
age1869
ba1869
easting
northing
_cons

0.009676
0.100954
0.000000
0.000405
1.917497
-0.001625
-0.007247
0.000004
0.000000
0.000001
-3.869999

0.04153380
0.08113240
0.00000045
0.00253060
0.90732760
0.00043010
0.00347900
0.00000435
0.00000828
0.00001530
25.39381000

0.23
1.24
0.11
0.16
2.11
-3.78
-2.08
0.96
0.03
0.04
-0.15

0.816
0.213
0.914
0.873
0.035
<0.0001
0.037
0.338
0.975
0.965
0.879

-0.0717285
-0.0580621
-0.0000008
-0.0045546
0.1391673
-0.0024680
-0.0140654
-0.0000044
-0.0000160
-0.0000294
-53.6409600

0.0910808
0.2599709
0.0000009
0.0053652
3.6958260
-0.0007820
-0.0004278
0.0000127
0.0000165
0.0000307
45.9009600

262

B
resilcd1869
1

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

2
lwmean
ahm
elev
solar
resist1869
prior1869
age1869
ba1869
_cons

-0.015949
-0.001669
-0.000145
0.000000
0.602097
-0.000787
-0.001063
0.000003
0.817651

0.03948870
0.03811590
0.00092450
0.00000044
0.64607140
0.00033760
0.00283270
0.00000356
2.57532900

-0.40
-0.04
-0.16
0.07
0.93
-2.33
-0.38
0.97
0.32

0.686
0.965
0.875
0.947
0.351
0.020
0.707
0.331
0.751

-0.0933454
-0.0763750
-0.0019574
-0.0000008
-0.6641794
-0.0014490
-0.0066150
-0.0000035
-4.2299000

0.0614474
0.0730367
0.0016665
0.0000009
1.8683740
-0.0001258
0.0044890
0.0000104
5.8652020

3
lwmean
ahm
elev
solar
resist1869
prior1869
age1869
ba1869
_cons

0.010486
0.097426
0.000316
0.000000
1.883131
-0.001636
-0.007298
0.000004
-2.799774

0.04059950
0.04403420
0.00103380
0.00000046
0.85742340
0.00045220
0.00356520
0.00000441
2.50191000

0.26
2.21
0.31
0.10
2.20
-3.62
-2.05
0.96
-1.12

0.796
0.027
0.760
0.923
0.028
<0.0001
0.041
0.338
0.263

-0.0690874
0.0111206
-0.0017103
-0.0000009
0.2026118
-0.0025225
-0.0142857
-0.0000044
-7.7034280

0.0900597
0.1837314
0.0023420
0.0000009
3.5636500
-0.0007499
-0.0003105
0.0000129
2.1038800

263

resilcd1842
1

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

2
lwmean
ahm
solar
elev
resist1842
prior1842
age1842
ba1842
easting
northing
_cons

0.019515
-0.101725
0.000001
-0.004536
3.921190
-0.000349
0.000093
-0.000004
-0.000001
-0.000012
14.389920

0.03772090
0.09004310
0.00000050
0.00287140
0.96965120
0.00033240
0.00361660
0.00000471
0.00000985
0.00001370
26.71764000

0.52
-1.13
1.36
-1.58
4.04
-1.05
0.03
-0.74
-0.05
-0.84
0.54

0.605
0.259
0.172
0.114
<0.0001
0.294
0.979
0.458
0.957
0.40
0.59

-0.0544167
-0.2782060
-0.0000003
-0.0101634
2.0207090
-0.0010002
-0.0069951
-0.0000127
-0.0000198
-0.0000383
-37.9756800

0.0934464
0.0747566
0.0000017
0.0010922
5.8216720
0.0003027
0.0071818
0.0000057
0.0000188
0.0000153
66.7555300

3
lwmean
ahm
solar
elev
resist1842
prior1842
age1842
ba1842
easting
northing

0.033931
0.128021
0.000001
-0.000197
9.250511
-0.001181
-0.001547
-0.000003
0.000031
0.000007

0.04281680
0.14322250
0.00000066
0.00393660
1.30960400
0.00049230
0.00426570
0.00000554
0.00001650
0.00001610

0.79
0.89
1.05
-0.05
7.06
-2.40
-0.36
-0.47
1.85
0.41

0.428
0.371
0.293
0.96
<0.0001
0.016
0.717
0.635
0.064
0.681

0.1178500
0.4087321
0.0000020
0.0075183
11.8172900
-0.0002164
0.0068133
0.0000082
0.0000630
0.0000382

_cons

-53.179050

39.67630000

-1.34

0.18

-0.0499887
-0.1526899
-0.0000006
-0.0079128
6.6837340
-0.0021463
-0.0099079
-0.0000135
-0.0000018
-0.0000250
130.9432000

264

24.5850700

resilcd1800
1

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

2
lwmean
ahm
solar
resist1800
prior1800
age1800
ba1800
easting
northing
_cons

-0.037723
-0.074213
0.000001
1.945396
-0.001080
-0.003965
0.000004
-0.000007
-0.000007
15.131330

0.03763700
0.03862110
0.00000054
0.42645530
0.00039170
0.00253790
0.00000436
0.00000512
0.00000643
10.44804000

-1.00
-1.92
2.21
4.56
-2.76
-1.56
0.96
-1.34
-1.05
1.45

0.316
0.055
0.027
<0.0001
0.006
0.118
0.338
0.179
0.292
0.148

-0.1114905
-0.1499088
0.0000001
1.1095590
-0.0018478
-0.0089395
-0.0000044
-0.0000169
-0.0000194
-5.3464570

0.0360437
0.0014832
0.0000023
2.7812330
-0.0003125
0.0010090
0.0000127
0.0000032
0.0000058
35.6091200

3
lwmean
ahm
solar
resist1800
prior1800
age1800
ba1800
easting
northing
_cons

-0.054968
-0.041501
0.000002
3.820334
-0.002565
-0.005268
0.000003
-0.000004
-0.000017
17.218410

0.05378680
0.05284630
0.00000066
0.54304600
0.00056750
0.00389990
0.00000996
0.00000706
0.00000968
14.25989000

-1.02
-0.79
2.63
7.04
-4.52
-1.35
0.33
-0.54
-1.78
1.21

0.307
0.432
0.008
<0.0001
<0.0001
0.177
0.742
0.588
0.076
0.227

-0.1603883
-0.1450781
0.0000004
2.7559830
-0.0036772
-0.0129111
-0.0000162
-0.0000177
-0.0000362
-10.7304600

0.0504522
0.0620758
0.0000030
4.8846840
-0.0014527
0.0023762
0.0000228
0.0000100
0.0000018
45.1672800

265

resilcd1772
1
2
lwmean
ahm
elev
resist1772
prior1772
ba1772
_cons
3

lwmean
ahm
elev
resist1772
prior1772
ba1772
_cons

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

-0.030333
-0.057367
-0.000467
3.758332
0.000219
-0.000010
1.138637

0.03530760
0.03599210
0.00103410
0.77101950
0.00047710
0.00000466
2.06593500

-0.86
-1.59
-0.45
4.87
0.46
-2.13
0.55

0.390
0.111
0.652
<0.0001
0.646
0.034
0.582

-0.0995351
-0.1279099
-0.0024939
2.2471610
-0.0007160
-0.0000190
-2.9105210

0.0388683
0.0131765
0.0015599
5.2695020
0.0011542
-0.0000008
5.1877950

-0.034600
-0.094993
-0.000450
7.724760
-0.001184
-0.000010
-0.164125

0.04672490
0.05429260
0.00154890
1.16056100
0.00087650
0.00000793
3.60135100

-0.74
-1.75
-0.29
6.66
-1.35
-1.31
-0.05

0.459
0.080
0.771
<0.0001
0.177
0.189
0.964

-0.1261789
-0.2014041
-0.0034859
5.4501030
-0.0029016
-0.0000259
-7.2226430

0.0569795
0.0114189
0.0025857
9.9994180
0.0005343
0.0000051
6.8943920

266

resilcd1760
1

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

2
lwmean
ahm
solar
resist1760
prior1760
age1760
ba1760
easting
northing
_cons

-0.093879
0.040846
0.000000
3.276225
-0.001968
-0.007572
0.000013
0.000014
-0.000020
-0.852170

0.04237650
0.02584590
0.00000043
0.78583800
0.00057390
0.00369560
0.00000705
0.00000314
0.00000592
5.62941600

-2.22
1.58
-0.62
4.17
-3.43
-2.05
1.82
4.36
-3.35
-0.15

0.027
0.114
0.533
<0.0001
0.001
0.040
0.069
<0.0001
0.001
0.880

-0.1769357
-0.0098114
-0.0000011
1.7360110
-0.0030928
-0.0148151
-0.0000010
0.0000075
-0.0000314
-11.8856200

-0.0108227
0.0915026
0.0000006
4.8164390
-0.0008430
-0.0003286
0.0000267
0.0000198
-0.0000082
10.1812800

3
lwmean
ahm
solar
resist1760
prior1760
age1760
ba1760
easting
northing
_cons

-0.067336
0.015304
-0.000001
7.407947
-0.004005
-0.015487
0.000026
0.000015
-0.000036
9.504638

0.06409600
0.05951870
0.00000084
1.03346500
0.00067240
0.00507140
0.00000919
0.00000621
0.00001010
13.08287000

-1.05
0.26
-1.23
7.17
-5.96
-3.05
2.86
2.33
-3.60
0.73

0.293
0.797
0.220
<0.0001
<0.0001
0.002
0.004
0.020
<0.0001
0.468

-0.1929619
-0.1013508
-0.0000027
5.3823940
-0.0053229
-0.0254268
0.0000083
0.0000023
-0.0000561
-16.1373200

0.0582899
0.1319584
0.0000006
9.4335010
-0.0026873
-0.0055472
0.0000443
0.0000267
-0.0000165
35.1466000

267

resilcd1734
1

lwmean
ahm
elev
solar
resist1734
prior1734
age1734
ba1734
easting
northing
_cons

2

3
lwmean
ahm
elev
solar
resist1734
prior1734
age1734
ba1734
easting
northing
_cons

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

-0.096108
-0.196514
-0.002129
0.000000
1.632822
-0.000762
-0.006116
0.000004
-0.000030
-0.000011
55.870580

0.04631140
0.09787100
0.00267750
0.00000066
0.74054210
0.00069580
0.00340190
0.00000759
0.00001250
0.00001160
27.92960000

-2.08
-2.01
-0.80
-0.44
2.20
-1.09
-1.80
0.55
-2.38
-0.94
2.00

0.038
0.045
0.427
0.661
0.027
0.274
0.072
0.58
0.017
0.347
0.045

-0.1868769
-0.3883372
-0.0073765
-0.0000016
0.1813863
-0.0021255
-0.0127831
-0.0000107
-0.0000540
-0.0000335
1.1295670

-0.0053395
-0.0046900
0.0031190
0.0000010
3.0842580
0.0006021
0.0005522
0.0000191
-0.0000052
0.0000118
110.6116000

-0.158666
-0.307386
-0.006162
-0.000001
7.162789
-0.004705
-0.016758
0.000023
-0.000037
-0.000032
88.211330

0.06342550
0.13068980
0.00399390
0.00000098
1.21212600
0.00111480
0.00719910
0.00001500
0.00001670
0.00001980
39.66616000

-2.5
-2.35
-1.54
-0.93
5.91
-4.22
-2.33
1.51
-2.20
-1.62
2.22

0.012
0.019
0.123
0.353
<0.0001
<0.0001
0.020
0.132
0.028
0.105
0.026

-0.2829776
-0.5635336
-0.0139898
-0.0000028
4.7870660
-0.0068894
-0.0308684
-0.0000068
-0.0000695
-0.0000709
10.4670900

-0.0343541
-0.0512390
0.0016662
0.0000010
9.5385120
-0.0025196
-0.0026485
0.0000520
-0.0000040
0.0000067
165.9556000

268

A
resilcd1717
1

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

2
lwmean
ahm
solar
elev
resist1717
prior1717
age1717
ba1717
easting
northing
_cons

0.055846
-0.168046
0.000001
-0.003329
5.448682
-0.000656
-0.008596
0.000018
-0.000016
-0.000005
27.324910

0.04449780
0.07428980
0.00000075
0.00239880
1.07584500
0.00065940
0.00488160
0.00001210
0.00000989
0.00001420
25.49037000

1.26
-2.26
1.21
-1.39
5.06
-0.99
-1.76
1.45
-1.61
-0.34
1.07

0.209
0.024
0.228
0.165
<0.0001
0.320
0.078
0.148
0.108
0.736
0.284

-0.0313687
-0.3136514
-0.0000006
-0.0080301
3.3400640
-0.0019480
-0.0181640
-0.0000062
-0.0000353
-0.0000327
-22.6352900

0.1430597
-0.0224409
0.0000024
0.0013729
7.5573000
0.0006369
0.0009716
0.0000413
0.0000035
0.0000231
77.2851200

3
lwmean
ahm
solar
elev
resist1717
prior1717
age1717
ba1717
easting
northing
_cons

0.081418
-0.031985
0.000000
0.002905
11.446810
0.000096
-0.004609
-0.000017
-0.000007
0.000019
-15.993710

0.05579420
0.13609200
0.00000087
0.00366090
2.06430900
0.00124900
0.00740820
0.00001980
0.00001180
0.00001810
32.96529000

1.46
-0.24
-0.55
0.79
5.55
0.08
-0.62
-0.86
-0.61
1.05
-0.49

0.144
0.814
0.58
0.428
<0.0001
0.939
0.534
0.389
0.541
0.295
0.628

-0.0279369
-0.2987205
-0.0000022
-0.0042705
7.4008410
-0.0023518
-0.0191285
-0.0000559
-0.0000302
-0.0000165
-80.6045000

0.1907722
0.2347502
0.0000012
0.0100800
15.4927800
0.0025442
0.0099111
0.0000218
0.0000159
0.0000544
48.6170700

269

B
resilcd1717
1

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

2
lwmean
ahm
solar
elev
resist1717
prior1717
age1717
ba1717
_cons

0.024911
-0.065385
0.000001
-0.001412
6.325952
-0.000653
-0.007621
0.000017
-0.984830

0.05018340
0.04671330
0.00000063
0.00088040
1.34646300
0.00071220
0.00533410
0.00001590
2.75549300

0.50
-1.40
1.44
-1.60
4.70
-0.92
-1.43
1.08
-0.36

0.620
0.162
0.151
0.109
<0.0001
0.359
0.153
0.281
0.721

-0.0734469
-0.1569417
-0.0000003
-0.0031370
3.6869320
-0.0020489
-0.0180760
-0.0000141
-6.3854970

0.1232683
0.0261712
0.0000021
0.0003141
8.9649710
0.0007428
0.0028331
0.0000484
4.4158370

3
lwmean
ahm
solar
elev
resist1717
prior1717
age1717
ba1717
_cons

0.063862
0.001412
0.000000
0.002216
12.605840
-0.000096
-0.003295
-0.000017
-10.798790

0.06055520
0.07763640
0.00000073
0.00170240
2.16859100
0.00121180
0.00738280
0.00002100
3.54756600

1.05
0.02
-0.62
1.30
5.81
-0.08
-0.45
-0.82
-3.04

0.292
0.985
0.534
0.193
<0.0001
0.937
0.655
0.412
0.002

-0.0548242
-0.1507524
-0.0000019
-0.0011207
8.3554770
-0.0024715
-0.0177646
-0.0000583
-17.7518900

0.1825480
0.1535767
0.0000010
0.0055527
16.8562000
0.0022786
0.0111755
0.0000239
-3.8456850

270

C
resilcd1717
1
2
lwmean
ahm
elev
resist1717
prior1717
ba1717
_cons
3

6

lwmean
ahm
elev
resist1717
prior1717
ba1717
_cons

Coef.

Robust
Std. Err.

(base out

come)

z

P>z

[95% Conf.

Interval]

0.007133
-0.050639
-0.001004
6.489108
-0.000472
0.000006
-0.910793

0.05077830
0.05192840
0.00093700
1.26563700
0.00063100
0.00000973
2.66641000

0.14
-0.98
-1.07
5.13
-0.75
0.64
-0.34

0.888
0.329
0.284
<0.0001
0.455
0.524
0.733

-0.0923912
-0.1524163
-0.0028407
4.0085060
-0.0017085
-0.0000129
-6.1368600

0.1066562
0.0511392
0.0008323
8.9697110
0.0007649
0.0000253
4.3152730

0.069485
0.016067
0.002872
12.702030
0.000080
-0.000024
-12.810780

0.06571590
0.07481370
0.00160280
2.12768200
0.00115980
0.00001320
3.81384800

1.06
0.21
1.79
5.97
0.07
-1.80
-3.36

0.290
0.830
0.073
<0.0001
0.945
0.072
0.001

-0.0593161
-0.1305647
-0.0002697
8.5318530
-0.0021934
-0.0000497
-20.2857800

0.1982854
0.1626994
0.0060133
16.8722100
0.0023528
0.0000022
-5.3357720

271

10. Appendix C: ANOVA output tables from Chapter 3
The following output tables include the output of two-way ANOVA models.
Drought resistance is the dependent variable, transformed as noted in the summary output
in Chapter 3. Explanatory variables are the growth release status (binary; biyear) and site
(sitecd). Outputs proceed from the most recent year backwards. Years 1959 and 1760,
which did not meet assumptions of normality and were excluded from the main results
table, include scatterplots of residual vs. fitted values for information.

272

273

274

275

276

277

278

279

280