Spatial Associations Between Infestations of Mountain Pine Beetle and Landscape
Features in the Peace River Region of British Columbia
Honey-Marie C. de la Giroday
B.Sc, University of Northern British Columbia, 2004

Thesis Submitted in Partial Fulfillment Of
The Requirements for the Degree Of
Master of Science
In
Natural Resources and Environmental Studies
(Biology)

The University of Northern British Columbia
September 2009
© Honey-Marie C. de la Giroday

1*1

Library and Archives
Canada

Bibliotheque et
Archives Canada

Published Heritage
Branch

Direction du
Patrimoine de I'edition

395 Wellington Street
OttawaONK1A0N4
Canada

395, rue Wellington
OttawaONK1A0N4
Canada
Your file Votre reference
ISBN: 978-0-494-60827-2
Our file Notre reference
ISBN: 978-0-494-60827-2

NOTICE:

AVIS:

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

L'auteur a accorde une licence non exclusive
permettant a la Bibliotheque et Archives
Canada de reproduce, publier, archiver,
sauvegarder, conserver, transmettre au public
par telecommunication ou par I'lnternet, preter,
distribuer et vendre des theses partout dans le
monde, a des fins commerciales ou autres, sur
support microforme, papier, electronique et/ou
autres formats.

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

L'auteur conserve la propriete du droit d'auteur
et des droits moraux qui protege cette these. Ni
la these ni des extraits substantiels de celle-ci
ne doivent etre imprimes ou autrement
reproduits sans son autorisation.

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

Conformement a la loi canadienne sur la
protection de la vie privee, quelques
formulaires secondaires ont ete enleves de
cette these.

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

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

1*1

ABSTRACT
An immense outbreak of the mountain pine beetle, Dendroctonus ponderosae
Hopkins, currently covers a cumulative area of 14.5 million hectares of mature pine forests
across the provinces of British Columbia and Alberta, Canada. In 2004, the first outbreaking
populations of mountain pine beetle were observed in northeastern British Columbia, an area
not considered part of the insect's native range. My thesis examines how landscape features
and their orientation influence establishment patterns of the insect. Mountain pine beetle
spread between 2004 and 2006 in patterns similar to a propagating wave, likely due to longdistance dispersal into the region. Large glacially-eroded valleys, canyons, deeply incised
streams, local and midslope ridges or small hills in valleys and plains, and open slopes were
often positively associated with infestations, providing evidence that the interaction of mesoscale convective currents and topography can mediate patterns of establishment. The
orientation of landscape features also influenced establishment, as southwest-facing areas
and linear features aligned in northeast-southwest directions were associated with increased
densities of infestations in 2006. Management activities were typically associated with a
decline in the density of mountain pine beetle infestations in the following year, indicating
that such activities were effective in preventing short-distance dispersal of the insect. I found
no evidence that anthropogenic activities such as transport and storage of infested material
increased establishment of mountain pine beetle across the research area. These results may
be used to prioritize preemptive treatments in mountainous regions in the absence of longdistance inputs of mountain pine beetle into expanding ranges.

li

TABLE OF CONTENTS
THESIS ABSTRACT

ii

TABLE OF CONTENTS

iii

ACKNOWLEDGEMENTS

ix

1: INTRODUCTION

1

1.1:

FIGURE CAPTIONS

7

1.2:

LITERATURE CITED

9

2: FIRST STEPS OF A RANGE-EXPANSION EVENT FACILITATED BY CLIMATE CHANGE:
BREACH OF THE ROCKY MOUNTAIN GEOCLIMATIC BARRIER BY MOUNTAIN PINE BEETLE
IN CENTRAL BRITISH COLUMBIA
13
2.1
2.2
2.3

13
14
19
19
20
21
23
27
29
35

ABSTRACT
INTRODUCTION
MATERIAL AND METHODS

2.3.1: Study Area and Spatial Datasets
2.3.2: Statistical Analysis
2.4
2.5
2.6
2.7
2.8

RESULTS
DISCUSSION
ACKNOWLEDGEMENTS
FIGURE CAPTIONS
LITERATURE CITED

3: THE ASSOCIATION BETWEEN ESTABLISHMENT AND SPREAD OF INFESTATIONS OF
MOUNTAIN PINE BEETLE AND LANDSCAPE FEATURES IN THE PEACE RIVER REGION OF
BRITISH COLUMBIA IN NORTHERN CANADA
41
3.1:
3.2:
3.3:

ABSTRACT
INTRODUCTION
MATERIALS AND METHODS

3.3.1:
3.3.2:
3.3.3:
3.4:

41
43
48

Preparation ofInsect Datasets
Preparation ofLandscape Feature Datasets
Statistical Analyses

RESULTS

3.4.1:
3.4.2:
3.4.3:
3.5:
3.6:
3.7:
3.8:

55

Spatial Extent and Landscape ofInvaded Area
Which Landscape Features are Associated with Mountain Pine Beetle Iinfestations?
Does Feature Orientation Influence Invasion Success?

DISCUSSION
ACKNOWLEDGEMENTS
FIGURE CAPTIONS
LITERATURE CITED

4.1.1.
4.1.2.
4.1.3.
4.1.4.
4.1.5.
4.2:
4.3:

55
57
59
60
65
67
86

4: CONCLUSIONS AND AREAS FOR FURTHER RESEARCH
4.1:

48
49
52

CONCLUSIONS

93
93

Influence ofAtmospheric Processes on Aeolian Dispersal
Lack ofEvidence for Anthropogenic Transport
Establishment of Infestations at Higher Elevations
Effect of Management on Spread ofBeetle Infestations
Potential for Continued Spread of Mountain Pine Beetle
AREAS OF FURTHER RESEARCH
LITERATURE CITED

93
94
95
96
96
97
99

APPENDIX A: R CODE FOR FITTING SPATIAL POINT PATTERN REGRESSION MODELS TO
ANNUAL INVENTORY DATASETS
101

ill

APPENDIX B: R CODE FOR CONTINGENCY TESTS (INFESTATIONS RELATIVE TO LANDSCAPE
FEATURES)
104
APPENDIX C: R CODE FOR CONTINGENCY TESTS (INFESTATIONS RELATIVE TO SUSCEPTIBLE
HABITAT)
108
APPENDIX D: EXAMPLE R CODE FOR BOOTSTRAPPING

111

APPENDIX E: EXAMPLE R CODE FOR SELECTION OF SPATIAL POINT PROCESS MODELS 114
APPENDIX F: CONTINGENCY TESTS EXAMINING THE PATTERN OF INFESTATION RELATIVE TO
LANDSCAPE FEATURES AND SUSCEPTIBLE HABITAT
117

IV

LIST OF FIGURES
Figure 1.1 Wind patterns within mountainous terrain that could influence settling patterns
and establishment of mountain pine beetle in the Peace River region of British Columbia. 8
Figure 2.1 Map of the Peace River region of British Columbia including mill locations,
primary rivers, and highways.
30
Figure 2.2 Location of infestations of mountain pine beetle in the Peace River region of
British Columbia, Canada in 2004, 2005, and 2006. Dashed line represents the line of
initiation which is parallel to the Rocky Mountains and perpendicular to the direction of
spread in 2004, the initial year of detection.
31
Figure 2.3 Location of highest intensity of infestations of mountain pine beetle in 2004 and
2005 (line A) and 2006 (line B) within the research area.
32
Figure 3.1 Location of research area in the northeastern British Columbia. The research area
includes a portion of the Northern Rocky Mountains, including the Hart Ranges, and the
Rocky Mountain Foothills, an area of undulate topography that eventually becomes a plateau
at the eastern edge of the research area. The inset map shows the location of the research
area relative to the rest of British Columbia.
69
Figure 3.2 A) Determination of slope position using Topographic Position Index (adapted
from Weiss 2001). B) Determination of general landform using Topographic Position Index
(adapted from Jenness 2006).
70
Figure 3.3 Examples of (A) landscape features of linear canyons or deeply incised
streambeds which are generally linear in nature and whose cardinal direction (i.e., feature
orientation) is measured using azimuth. (B) Landscape features, such as midslope ridges or
small hills in plains which are generally circular or complex in shape and whose orientation
is measured compartmentally using azimuth.
71
Figure 3.4 Area and number of infestations of mountain pine beetle in the Peace River region
of British Columbia in 2004, 2005, and 2006 as determined by aerial surveys of patches of
red trees.
72
Figure 3.5 Location of infestations of mountain pine beetle in the Peace River region of
British Columbia in 2004, 2005, and 2006. Dashed line in the three figures represents the
line of initiation which is parallel to the Rocky Mountains and perpendicular to the direction
ofspreadin2004.
73
Figure 3.6 Map of landscape features within the Peace River region of British Columbia.74
Figure 3.7 Distribution of habitat susceptibility classes in the Peace River region of British
Columbia.
75

v

Figure 3.8 A general wind rose of the Peace River region of British Columbia (produced by
Matthew Lynn of Garrad Hassan Canada Inc. with data from the Canadian Wind Energy
Atlas)
76
Figure Fl Differences between expected and observed area (in hectares) infested by
mountain pine beetle per (A) type of landscape feature and (B) stand susceptibility class
within the Peace River region of British Columbia for initial year of recorded incipientepidemic populations.
119

VI

LIST OF TABLES
Table 2.1 AIC values for the spatial point process models reflecting various hypotheses of
insect dispersal for 2004, 2005, and 2006 in the Peace River region of British
Columbia. Each model uses the spatially explicit intensities of beetle-killed trees for
the respective year as a function of an intercept term and a similarly spatially-explicit
covariate(s), as listed (see Methods). Models with lowest AIC values are judged to fit
the best.
33
Table 2.2 The parameter estimates for spatial point process models judged to best fit
infestation locations for 2004, 2005, and 2006 in the Peace River region of British
Columbia (see Table 2.1). Distance from initiation line is represented by W while
the response variable for the models is log (A), number of infestations of mountain
pine beetle per m2. The location of greatest density of infestations, i.e. distance from
the initiation line, was estimated from the point of inflection for each polynomial
regression equation.
34
Table 3.1 Data fields of annual datasets inventorying patches of dead trees killed by
mountain pine beetle within the Peace River region of British Columbia, 2004-2006.
77
Table 3.2 Classes of landscape features in the Peace River region of British Columbia. 78
Table 3.3 Summary statistics for infestation sizes and their characteristics of their elevations
from annual inventories of mountain pine beetle infestations for 2004, 2005, and 2006
in the Peace River region of British Columbia. Results for /-tests examining the
difference between average elevation for central British Columbia (1000m; Safranyik
and Carroll 2006) conducted by 50 randomly sampled points in each annual dataset
within the research area are reported as the percentage of 100 reiterations that were
significant.
79
Table 3.4 Average size and elevation of landscape features in Peace River region of British
Columbia (see Fig. 3.6 for map).
80
Table 3.5 Numerical summary of "habitat susceptibility classes" across the Peace River
region of British Columbia. See Methods for details of classification scheme. 81
Table 3.6 Percentages of landscape features and total landscape with habitat considered
highly susceptible to outbreaking populations of mountain pine beetle in the Peace
River region of British Columbia (habitat susceptibility class 5; see Table 3.2). 82
Table 3.7 Spatial point process regression estimating the intensity of discrete beetle
infestations as a function of landscape features, topography, distance from a source
line, habitat susceptibility, and distance from control tactics in the previous year,
2004-2006, in the Peace River region of British Columbia. Log(A) is the response
variable for each model, where A is the density of sites successfully attacked by
mountain pine beetle populations per square meter. For example, in 2004, the
estimated density of beetle patches at sites 5,000 kilometres from the source location
in a canyon at 1,500 metres elevation and in highly susceptible habitat would be exp("

vn

7.82+5OOOx-5.O3E-05+-O.O296+,5OOx-3.85E-O3+,.55)

Qf

3 ^ . ^

i n f e s t a t i o n s

p e r

§ q u a r e

m e t e r

(j

g

;

0.034 infestations per hectare or one patch of tree-killing beetles every 30ha). 83
Table 3.8 Spatial point process regression models that identify which orientations of
landscape features best fit the distribution of D. ponderosae infestations in three
initial years of incipient-epidemic outbreaks (2004-2006) in the Peace River region of
British Columbia.
84
Table 3.9 Spatial point process regression models estimating intensity of D. ponderosae
infestations replacing landscape feature with direction of landscape feature as a
covariate. Data is for the years 2004-2006 in the Peace River region of British
Columbia. Other variables are the same as in Table 3.8.
85
Table Fl Test statistics and/»-values for contingency tests examining the effect of landscape
features and distribution of highly susceptible habitat on the occurrence of mountain
pine beetle infestations, assuming an uniform distribution across classes. Data is for
the years 2004-2006 in the Peace River region of British Columbia.
120

Vlll

ACKNOWLEDGEMENTS
There are numerous people I would like to thank for their guidance, patience, and
ability to foster self-driven learning. Dr. Brian Aukema, through your astute guidance you
provided me with an opportunity to learn more about two subject areas I have found
fascinating, forest entomology and spatial statistics. Through you I have come to understand
the joys of academia, life-long learning, and the importance of positivity and team work.
Thank you for selecting me to be part of your team. Dr. Staffan Lindgren, the first
opportunity I had to work with you was an amazing period and fostered a desire to return to
forest entomology for my graduate work. Your quiet and tactful advice about experimental
design and my research questions assisted me with moving my research into an area that
truly interested me. Thank you to Dr. Allan Carroll for taking the time to discuss the many
approaches to address the research questions in my thesis. I would be remiss in not thanking
other researchers at UNBC and beyond that inspired me to complete my graduate degree
including Dr. Josef Ackerman of the University of Guelph, Mr. Bob Drinkwater of the
Ministry of Forests and Range, and Bob Redden and Robert Van Schubert of Environmental
Dynamics Inc. There are also numerous people that provided technical support for this
research as well including Talya Truant, Sandra Allen, Ping Bai, Roger Wheate, Marnie
Duthie-Holt, Aaron McGill, Gurp Thandi, Brian Pate, Colin Johnston, Doug Linton,
Matthew Lynn, and Adrian Baddeley. Fraser McKee, I shall miss our coffee-powered
discussions of our projects, and Matt Klingenberg, I already miss your presence in the office
including your diplomacy and camaraderie. Thank you both for your assistance in this
process.

Finally, I would like to thank my family including my parents Richard and Avis de la
Giroday and siblings (Heath, Jonathan, Nanette and Sherri). Thank you for your guidance,
patience, and support.

x

1: INTRODUCTION
Range expansion of native species creates important challenges to the management of
organisms and ecosystems. For example, the establishment of parks and protected areas is
based on the assumption of temporal continuity of those regions as havens for rare or
sensitive ecosystems and habitats of sensitive species and communities (Lemieux and Scott
2005). However, human alteration of climate has and will influence the distributions of
animal and plant species. For example, the silver-spotted skipper butterfly (Hesperia comma
L. [Lepidoptera: Hesperiidae]), an insect native to southern England, is increasingly
expanding into new habitat types due to climate change (Thomas et al. 2001). Particularly
interesting is the release of species from the northern confines of their distributions, such as
the case of the sachem skipper (Atalopedes campestris Boisduval [Lepidoptera:
Hesperiidae]) in northern California expanding into new habitats previously considered
inhospitable (Crozier and Dwyer 2006).
The recent range expansion of the mountain pine beetle (Dendroctonus ponderosae
Hopkins [Coleoptera: Scolytidae]) into regions thought to be beyond its historical
distribution (Carroll et al. 2004; Robertson et al. 2009) is a striking example of the potential
impact of anthropogenic changes on ecosystem dynamics (Raffa et al. 2008). Although old
mountain pine beetle strip attacks have been observed in northeastern British Columbia (i.e.,
the Peace River region; Allan Carroll pers. comm.1), the region was thought to have a climate
too cold for the successful univoltine reproduction of mountain pine beetle. As a result,
small populations dispersing into the region prior to a change in climatic regime of the region
would not have established successfully due to high levels of mortality during the winter.
However, in 2004, the first incipient-epidemic infestations of mountain pine beetle were
'Allan Carroll Associate Professor, University of British Columbia (allan.caiTollfajubc.ca")
B. Staffan Lindgren, Professor, University of Northern British Columbia (lindgren(q)unbc.ca)

2

1

found in the Peace River region (B. Staffan Lindgren, pers. comm.2). Since then, mountain
pine beetle has spread throughout the region, with large outbreaks coalescing across the
landscape. The forests of the Peace River region contain lodgepole pine {Pinus contorta var.
latifolia), a primary host species of D. ponderosae, and jack pine (P. banksiand), a "novel
host" in which the mountain pine beetle has been demonstrated in laboratory experiments to
successfully reproduce (Safranyik and Linton 1983; Cerezke 1995). Of critical concern is the
potential ability of mountain pine beetle to use jack pine as a conduit for movement further
east (Nealis and Peter 2008).
The life cycle of mountain pine beetle is completed largely in a host tree of the genus
Pinus, such as P. monticola and P. ponderosae (Amman 1982). Upon landing on a suitable
host, a female will bore through the bark into the phloem and begin constructing a gallery by
tunneling upward (Reid 1962). Attacks generally occur at mid-bole, although this can vary
with attack density. The male will enter the gallery and mate with the female, who then lays
eggs in niches along the gallery (Wood 1963). The eggs develop through four larval instars
from the fall through the winter and into spring, although development is arrested during the
coldest periods of the winter. Mountain pine beetle populations can experience large
mortality in the early winter, late fall, or early spring due to unseasonal cold temperatures
(e.g., -25 to 30°C in October or April) or prolonged cold winter temperatures of-40 °C or
colder (Wygant 1940). However, the insect is more cold-tolerant during winter due to
progressive removal of water from their tissues, resulting in high levels of glycerol, between
late fall and the middle of winter (Safranyik and Linton 1998; Bentz and Mullins 1999). In
the spring, development continues through the final instars until mountain pine beetle
pupates and develops into an adult. In June, July or August (depending on the latitude,

2

elevation, and general climate in the region) new adults will emerge from the gallery by
boring through the bark (Safranyik and Carroll 2006).
Upon emergence, mountain pine beetles take flight and seek out a new host tree to
attack. A combination of visual and olfactory cues is used during host-seeking by mountain
pine beetle. Typically, the insect is attracted to pheromones produced by conspecifics and
host volatiles indicating host vigour (Hughes 1973). Attack en masse, with high numbers of
conspecifics in concert with vectored blue-stain fungi, may be effective in overwhelming tree
defenses. Attack densities of 62 attacks/m2 typically characterize maximum per capita
reproduction, with higher mortality due to tree defenses and interspecific competition above
and below these densities, respectively (Raffa and Berryman 1983).
A majority of mountain pine beetles dispersing from a natal host travel short
distances, staying within the stand (Safranyik et al. 1992). While moving short distances,
under the canopy, they can actively fly against the wind at velocities to a maximum of 7.5
km/h if an attractive pheromone stimulus is located upwind (Gray et al. 1972). However, in
the absence of attractants upwind, the insect will often fly downwind. The insect is also
phototactic just after emergence, allowing for an upward trajectory (Reid 1962). Above the
canopy, it may become incorporated into strong wind currents, where its dispersal be can
determined largely by wind velocity and direction (Lewis 1966; Lewis and Stephenson 1966;
Lewis 1970; Ashmole et al. 1983; Jackson et al. 2008). This "passive" insect dispersal will
most likely result in dispersal patterns that differ from those of "active" dispersal due to the
dominance of host seeking behavior in the latter. During "active" dispersal, biological
mechanisms such as pheromone response, host selection, and micro-site conditions influence
patterns of dispersal. Conversely, patterns of establishment resulting from "passive"

3

dispersal will be influenced by the interaction between weather processes, landscape patterns
and particles (i.e., mountain pine beetles) (Robertson et al. 2009), including down-drafts in
zones of convergence, changes in wind direction, and/or deposition or impaction of particles
(Lewis 1966; Lewis and Stephenson 1966; Lewis 1970; Lewis and Dibley 1970; Spalding
1979; Pedgley 1982; Pasek 1988; Jasperson et al. 1990; Jackson et al. 2008). Examining the
role of topography in the spread and establishment of wind-transported insects can provide
information about the impact of mountainous terrain on long-distance dispersal (Fig. 1.1).
The founding populations of mountain pine beetle in the Peace River region are
thought to be from regions southwest to the northern Rocky Mountains (Bartell 2008).
Beginning in the mid-1990s, a large outbreak of D. ponderosae established and continued to
spread at epidemic levels until the present time in the central and southern interior of British
Columbia (Aukema et al. 2006). With a cumulative area of 14.5 million ha of forest affected
(British Columbia Ministry of Forests and Range 2009), the insect has exerted landscapelevel impacts on P. contorta, a tree that comprises -47% of the harvestable timber supply in
British Columbia (Westfall and Ebata 2008). Prior to the present outbreak, mountain pine
beetle in the central interior of British Columbia predominantly existed at endemic levels,
usually taking older, less vigorous trees, and occasionally flaring up into larger outbreaks
such as those in the 1970s and early 1980s (Safranyik and Carroll 2006; Campbell et al.
2007; Aukema et al. 2008). However, fire suppression by humans post-World Wars I and II
permitted an increase in the proportion of large tracts of even-aged stands of suitable hosts on
the landscape (Carroll et al. 2004). Warmer summers and winters have also allowed
mountain pine to successfully reproduce at higher rates with larger populations facilitating
successful mass-attack on vigorous, large diameter trees (Carroll et al. 2004). Numerous

4

factors influencing population dynamics could have contributed to the increased longdistance dispersal as well. For example, large populations of insects attacking trees will
eventually produce pheromones that repel additional insects (Hunt and Borden 1990),
potentially causing increased levels of dispersal through indirect competition for available
resources. Depletion of available hosts adjacent to a natal tree will also facilitate increased
time aloft for mountain pine beetle searching for a host, possibly providing increased
opportunity for the insect to become incorporated into meso-scale convective currents.
These pheromones, coupled with meso-scale atmospheric transport, may have caused
mountain pine beetle to increasingly disperse long-distances over the Rocky Mountains.
Patterns of infestation resulting from long-distance dispersal of mountain pine beetle
have not been studied in the Peace River region of British Columbia. Due to the speed at
which mountain pine beetle spread into this region, it has been speculated that mountain pine
beetle was dispersed into this region primarily by meso-scale atmospheric currents rather
than "active" local dispersal. This thesis examines the spatial patterns of the invasion event
in the Peace River region to gain inference on the primary mechanism(s) of establishment
and the influence of various landscape patterns on occurrences of infestations. The findings
of this research will contribute an understanding of meso-scale atmospheric dispersal of
insects within mountainous terrain, the potential distances that mountain pine beetles can
travel, and may aid forest managers in making decisions for control and prevention tactics as
the beetle potentially continues to spread in the region. In Chapter 2, the patterns of spread
in the Peace River region of British Columbia are examined to gain inference on primary
mechanisms of spread for mountain pine beetles into the Peace River region. These
mechanisms include atmospheric transport, localized stand-level transport, and human-

5

assisted transport along roadways. Annual inventory data for the first three years of
infestation are compared to various dispersal patterns using spatial point process modeling.
Using similar techniques, Chapter 3 examines the influence of topography and orientation of
landforms on occurrence of mountain pine beetle infestations using a landscape feature
classification index and annual inventory data for 2004 to 2006. Chapter 4 provides a
synthesis of findings and recommendations to assist with mitigation of mountain pine beetle
spread.
To facilitate publication of the following work in peer-reviewed journals, this thesis is
written in manuscript format. As a result, there may be redundancy between chapters to
allow each chapter be a stand-alone manuscript. These chapters were produced in
cooperation with a number of authors hence sections will be presented in plural voice. I
anticipate submitting Chapter 2 to Biological Invasions with Allan Carroll and Brian Aukema
as coauthors. I anticipate submitting Chapter 3 to the Journal ofAnimal Ecology with Allan
Carroll, Staffan Lindgren, and Brian Aukema as coauthors.

6

1.1: Figure Captions
Figure 1.1 A schematic diagram of elucidating processes that could influence the longdistance dispersal of mountain pine beetle upon emergence, while dispersing longdistances and during settlement. Wind and topographic interactions are simplified for
mountainous terrain to explore wind patterns that could impact deposition of
mountain pine beetle while aloft.

7

Fig. 1.1
Predominant Wind Direction
Factors Influencing
Dispersal
Wind Speed
Wind Direction
Turbulence

c> r Factors Influencing
Establishment
Landform
Susceptible Hosts
Climate

1.2: Literature Cited

Amman, G. D. 1982. Characteristics of mountain pine beetles reared in four pine hosts.
Environmental Entomology 11(3): 590-593.
Ashmole, N. P., J. M. Nelson, M. R. Shaw and A. Garside. 1983. Insects and spiders on
snowfields in the Cairngorms, Scotland. Journal of Natural History 17(4): 599-613.
Aukema, B. H., A. L. Carroll, Y. Zheng, J. Zhu, K. F. Raffa, R. D. Moore, K. Stahl and S. W.
Taylor. 2008. Movement of outbreak populations of mountain pine beetle: influences
of spatiotemporal patterns and climate. Ecography 31(3): 348-358.
Aukema, B. H., A. L. Carroll, J. Zhu, K. F. Raffa, T. A. Sickley and S. W. Taylor. 2006.
Landscape level analysis of mountain pine beetle in British Columbia, Canada:
spatiotemporal development and spatial synchrony within the present outbreak.
Ecography 29(3): 427-441.
Bartell, N. V. 2008. A microsatellite analysis of the western Canadian mountain pine beetle
{Dendroctonus ponderosae) epidemic: phylogeography and long-distance dispersal
patterns. Natural Resources and Environmental Studies. Prince George, British
Columbia, University of Northern British Columbia. Masters of Science thesis.
Bentz, B. J. and D. E. Mullins. 1999. Ecology of mountain pine beetle (Coleoptera:
Scolytidae) cold hardening in the intermountain west. Environmental Entomology 28:
577-587.
British Columbia Ministry of Forests and Range. 2009. Ministry of Forests and Range: beetle
facts. Accessed on July 27 th , 2009 from
<http://www.for.gov.bc.ca/hfp/mountain_pine_beetle/facts.htm>.
Campbell, E. M., R. I. Alfaro and B. Hawkes. 2007. Spatial distribution of mountain pine
beetle outbreaks in relation to climate and stand characteristics: a dendroecological
analysis. Journal of Integrative Plant Biology 49(2): 168-178.
Carroll, A. L., S. W. Taylor, J. Regniere and L. Safranyik. 2004. Effects of climate change on
range expansion by the mountain pine beetle in British Columbia. Pages 223-232 in
T. L. Shore, J. E. Brooks and J. E. Stone. Challenges and solutions: Proceedings of
the Mountain Pine Beetle Symposium. Kelowna, British Columbia, Canada October
30-31, 2003. Natural Resources Canada, Canadian Forest Service, Pacific Forestry
Centre. Victoria, British Columbia, Canada. Information Report BC-X-399.
Cerezke, H. F. 1995. Egg gallery, brood production, and adult characteristics of mountain
pine beetle, Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae), in three
pine hosts. Canadian Entomologist 127(6): 955-965.

9

Crozier, L. and G. Dwyer. 2006. Combining population-dynamic and ecophysiological
models to predict climate-induced insect range shifts. The American Naturalist
167(6): 853-866.
Gray, B., R. F. Billings, R. I. Gara and R. L. Johnsey. 1972. On the emergence and initial
flight behaviour of the mountain pine beetle, Dendroctonus ponderosae, in eastern
Washington. Journal of Applied Entomology 21: 250-259.
Hughes, P. R. 1973. Dendroctonus: Production of pheromones and related compounds in
response to host monoterpenes. Journal of Applied Entomology 73: 294-312.
Hunt, D. W. A. and J. H. Borden. 1990. Conversion of verbenols to verbenone by yeasts
isolated from Dendroctonus ponderosae (Coleoptera: Scolytidae). Journal of
Chemical Ecology 16(4): 1385-1397.
Jackson, P. L., D. Straussfogel, B. S. Lindgren, S. Mitchell and B. Murphy. 2008. Radar
observation and aerial capture of mountain pine beetle, Dendroctonus ponderosae
Hopk. (Coleoptera: Scolytidae) in flight above the forest canopy. Canadian Journal
of Forest Research 38(8): 2313-2327.
Jasperson, W. H., G. D. Nastrom and D. C. Fritts. 1990. Further study of terrain effects on
the meso-scale spectrum of atmospheric motions. American Meteorological Society
47(8): 979-987.
Lemieux, C. J. and D. J. Scott. 2005. Climate change, biodiversity conservation and
protected area planning in Canada. The Canadian Geographer 49: 384-397.
Lewis, T. 1966. An analysis of components of wind affecting accumulation of flying insects
near artificial windbreaks. Annals of Applied Biology 58(3): 365-370.
Lewis, T. 1970. Patterns of distribution of insects near a windbreak of tall trees. Annals of
Applied Biology 65(2): 213-220.
Lewis, T. and G. Dibley. 1970. Air movement near windbreaks and a hypothesis on the
mechanism of the accumulation of airborne insects. Annals of Applied Biology 66:
477-484.
Lewis, T. and J. W. Stephenson. 1966. The permeability of artificial windbreaks and the
distribution of flying insects in the leeward sheltered zone. Annals of Applied Biology
58(3): 355-363.
Nealis, V. and B. Peter. 2008. Risk assessment of the threat of mountain pine beetle to
Canada's boreal and eastern pine forests. Natural Resources Canada, Canadian Forest
Service, Pacific Forestry Centre. Victoria, British Columbia, Canada, Information
Report BC-X-417.
Pasek, J. 1988. Influence of wind and windbreaks on local dispersal of insects. Agriculture,
Ecosystems & Environment 22-23: 539-554.

10

Pedgley, D. 1982. Windborne Pests and Diseases: Meteorology of Airborne Organisms. Ellis
Horwood Ltd., Chichester, United Kingdom.
Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner and W. H.
Romme. 2008. Cross-scale drivers of natural disturbances prone to anthropogenic
amplification: the dynamics of bark beetle eruptions. Bioscience 58(6): 501-517.
Raffa, K. F. and A. A. Berryman. 1983. The role of host plant resistance in the colonization
behavior and ecology of bark beetles (Coleoptera: Scolytidae). Ecological
Monographs 53(1): 27-49.
Reid, R. W. 1962. Biology of the mountain pine beetle, Dendroctonus monticolae Hopkins,
in the East Kootenay region of British Columbia. I. Life cycle, brood development,
and flight periods. Canadian Entomologist 94: 531-538.
Robertson, C , T. Nelson, D. Jelinski, M. A. Wulder and B. Boots. 2009. Spatial-temporal
analysis of species range expansion: the case of the mountain pine beetle,
Dendroctonusponderosae. Journal of Biogeography 36(8): 1446-1458.
Safranyik, L. and A. L. Carroll. 2006. The biology and epidemiology of the mountain pine
beetle in lodegpole pine forests. Pages 3-66 in L. Safranyik and B. Wilson. The
Mountain Pine Beetle: A Synthesis of Biology, Management, and Impacts on
Lodgepole Pine. Natural Resources Canada, Canadian Forest Service, Pacific
Forestry Centre, Victoria, British Columbia.
Safranyik, L. and D. A. Linton. 1983. Brood production by three spp. of Dendroctonus
(Coleoptera: Scolytidae) in bolts from host and nonhost trees. Journal of the
Entomological Society of British Columbia 80: 10-13.
Safranyik, L. and D. A. Linton. 1998. Mortality of mountain pine beetle larvae,
Dendroctonus ponderosae (Coleoptera: Scolytidae) in logs of lodgepole pine (Pinus
contorta var. latifolia) at constant low temperatures. Journal of the Entomological
Society of British Columbia 95: 81-87.
Safranyik, L., D. A. Linton, R. Silversides and L. H. McMullen. 1992. Dispersal of released
mountain pine beetles under the canopy of a mature lodgepole pine stand. Journal of
Applied Entomology 113: 441-450.
Spalding, J. B. 1979. The aeolian ecology of White Mountain Peak, California: windblown
insect fauna. Arctic and Alpine Research 11(1): 83-94.
Thomas, C. D., E. J. Bodsworth, R. J. Wilson, A. D. Simmons, Z. G. Davies, M. Musche and
L. Conradt. 2001. Ecological and evolutionary processes at expanding range margins.
Nature 411(4837): 577-581.
Westfall, J. and T. Ebata. 2008. 2007 Summary of Forest Health Conditions in British
Columbia. Pest Management Reports. British Columbia Ministry of Forests and
Range, Forest Practices Branch, Victoria, British Columbia, Canada. 72 pp.

11

Wood, S. L. 1963. A revision of the bark beetle genus Dendroctonus Erichson (Coleoptera:
Scolytidae). Great Basin Naturalist 23(1-117).
Wygant, N. D. 1940. Effects of low temperatures on the Black Hills beetle (Dendroctonus
ponderosae Hopk.). Syracuse, NY, New York State College of Forestry. Doctoral
thesis.

12

2: FIRST STEPS OF A RANGE-EXPANSION EVENT FACILITATED BY
CLIMATE CHANGE: BREACH OF THE ROCKY MOUNTAIN
GEOCLIMATIC BARRIER BY MOUNTAIN PINE BEETLE IN CENTRAL
BRITISH COLUMBIA
2.1: Abstract
In 2004, mortality of pine trees due to colonization by mountain pine beetle was
observed northeast of the Northern Rocky Mountains in British Columbia. The insect was
thought to have originally arrived in 2002. This was the first recorded incident of this insect
in the Peace River region. The mountain pine beetle is native to central and southern British
Columbia. Its speed and rate of spread into and within the Peace River region suggests that
long-distance, above-canopy dispersal was the predominant dispersal mechanism rather than
stand-level flight. We explore potential dominant mechanism(s) of dispersal using spatial
point process modeling. Specifically, we examine observed patterns of infestation relative to
covariates reflecting various hypotheses including meso-scale atmospheric dispersal above
the insect boundary layer (an altitude below which insects would determine their own speed
and direction of movement), anthropogenic transport of infested material, and the spread of
insect populations into adjacent stands via corridors of suitable habitat. We select the most
parsimonious models for each of the initial three years of invasion using information criteria
statistics. We found that landscape patterns of trees killed by mountain pine beetle
predominantly suggested a wave front of insects deposited parallel to the Rocky Mountains
following meso-scale atmospheric dispersal above the "insect boundary layer". The area of
highest intensity of infestations advanced up to 25 km to the northeast in a single year.

13

2.2: Introduction
Examining the spatiotemporal patterns of invasive organisms can provide insight into
organisms' rates and typical patterns of spread, which in turn can suggest mechanisms
driving their establishment and persistence (Peterson and Vieglais 2001). Climate plays a
central role in many invasion events, such as by impacting host distribution (Iverson and
Prasad 1998), creating directional gradients in critical temperature thresholds that can restrict
reproduction, or introducing organisms on strong weather fronts or through major
disturbances such as hurricanes (Tackenberg 2003). Humans can play a major role in
introductions of organisms as well, albeit inadvertently (Brockerhoff et al. 2006; Skarpaas
and 0kland 2009). Investigation of resulting patterns of spread allows researchers to
determine which processes contribute to establishment of an organism in a region and can
provide helpful clues for mitigation in the event of an organism's persistence.
The mountain pine beetle {Dendroctonus ponderosae Hopkins) is an insect native to
western North America that intermittently undergoes dramatic population fluctuations
(Safranyik et al. 1974; Amman and Cole 1983; Aukema et al. 2008). A phloeophagus insect
with a short-lived adult phase relative to other insects, mountain pine beetles survive outside
of host trees for only a few days. Numerous species of conifers in the Pinus genus are
susceptible to colonization by mountain pine beetle, including, but not limited to, lodgepole
pine {Pinus contorta Dougl. var. latifolia Engelm.), and ponderosa pine (P. ponderosa
Dougl.). Adult female beetles bore through the bark into the phloem, anywhere on the bole
between the root-collar and crown, depending on the size of the tree. The tree responds with
copious amounts of resin that is both toxic to the beetle and physically impedes access to the
cambium tissues. However, at epidemic population phases, additional conspecifics may be

14

attracted to the host by host volatiles as well as aggregation pheromones released by
pioneering mountain pine beetles (Renwick and Vite 1970; Libbey et al. 1985). Moreover,
mountain pine beetles vector fungi that serve to impede host defenses (Safranyik et al. 1975).
If access is successfully gained, females mate with males under the bark and lay eggs in
small niches along ovipositional galleries. Once hatched, mountain pine beetles progress
through four larval instars from the early fall into the winter. With onset of winter,
development slows, usually by the third or fourth instar (Reid 1962). In the spring, after
amassing enough heat units, the larvae pupate prior to developing into adults in early
summer. In July or early August, new adults take flight en masse after a period of dry and
warm weather (Reid 1962; Bright 1976).
Exposure to cold winter temperatures (Reid 1963; Reid and Gates 1970) and limited
mosaics of suitable hosts in space and time (Shore and Safranyik 1992) generally limits
mountain pine beetle to endemic levels. However, in the late 1990s, a large outbreak erupted
in the central interior of British Columbia, Canada (Aukema et al. 2006), currently
encompassing a total area of 14.5 million hectares of pine forests in that province (British
Columbia Ministry of Forests and Range 2009). The outbreak has been exacerbated by
anthropogenic activities such as fire suppression as well as increasingly warm summer and
winter temperatures reflective of a changing climate (Safranyik and Carroll 2006; Kurz et al.
2008; Raffa et al. 2008).
Outbreaking populations of the insect breached the historic geoclimatic barrier of the
northern Rocky Mountains in approximately 2002 as entomologists discovered that the
mountain pine beetle had become established in the Peace River region of British Columbia,
an area not considered part of the insect's historical distribution (Robertson et al. 2009). This

15

invasion event immediately commanded critical attention due to the proximity of a
potentially suitable new host, jack pine (Pinus banksiana Lamb.) in northeastern Alberta
(Safranyik and Linton 1983; Cerezke 1995) which could provide a potential conduit for
mountain pine beetle to spread toward eastern North America (Wood 2006; Nealis and Peter
2008). In many ways, the mountain pine beetle demonstrates ideal properties of an invasive
scolytid (Holway and Suarez 1999) due to its relatively high fecundity in northern areas
(Cudmore 2009), its ability to attract mates necessary for host procurement via aggregation
pheromone signals (Conn 1981) and its ability to take advantage of seemingly 'novel' hosts
(Safranyik and Linton 1983; Cerezke 1995; Huber et al. 2009).
Studying the positioning of tree-killing infestations of forest insects on the landscape
can elucidate the insects' dominant dispersal mechanism(s). For example, the spread of
organisms via short-distance or within-stand movements (e.g., Jactel 1991) may result in
diffuse patterns (cf. Hengeveld 1988), while long-distance movements can result in wavelike dispersal patterns dependent on the coordination of dispersing individuals (Suarez et al.
2001). Larch budmoth, Zeiraphera diniana Gn., may exhibit wave-propagation patterns,
where large numbers of mated females are blown long-distances (Baltensweiler and Fischlin
1979; Bjornstad et al. 2002; Johnson et al. 2006). Diffuse configurations, occasionally
characterizing mountain pine beetle infestations, are indicative of short-distance dispersal as
insects radiate from natal hosts into adjacent stands (Gamarra and He 2008). Although
mountain pine beetles primarily disperse within a stand (Safranyik et al. 1992) and are
generally not considered strong dispersers at landscape scales (Furniss and Carolin 1977;
Raffa and Berryman 1979; Peltonen et al. 2002), at outbreak levels, the insect may be
transported great distances via meso-scale atmospheric currents (i.e., aeolian dispersal; see

16

Chapter 3; Furniss and Furniss 1972; Safranyik et al. 1992; Jackson et al. 2008). Such longdistance transport can result in a variety of patterns of establishment including leptokurtic
(i.e., stratified) patterns (Tobin and Blackburn 2008; Robertson et al. 2009) and/or wave-like
patterns of distribution across the landscape (Bjornstad et al. 2002).
Despite the importance of wind currents and insect flight to dispersal at either small
or large scales, human activities may also facilitate the spread of invasives, often
unintentionally (Brockerhoff et al. 2006). Notable examples from forest insect systems
include the spread of emerald ash borer, Agrilus planipennis Fairmaire, in the Great Lakes
region via the transport of infested firewood (BenDor et al. 2006), the spread of Asian longhorned beetle (Anoplophora glabripennis Motschulsky) via wood pallets from international
trade (Haack 2006), and the spread of gypsy moth (Lymantria dispar L.) via the transport of
egg masses on automobiles and nursery stock (Johnson et al. 2006). Within the forest
industry of the United Kingdom in the 1970s, harvesting and transport of infested logs
facilitated the spread of a related Dendroctonus species, D. micans (Kugelann) (Gilbert et al.
2003). In the mountain pine beetle system, there are frequent points of contact between
humans and infested host materials throughout the newly invaded areas of British Columbia
and Alberta. For example, control and treatment tactics used throughout the region include
on-site falling and burning as well as harvesting of insect-colonized trees. Although
harvesting companies are careful with infested host material and prioritize its processing at
mill locations, storage of infested host material during periods of insect emergence and flight
could potentially result in increased occurrence of infestations in areas surrounding storage
and processing sites. Moreover, unregulated transportation of infested logs for firewood

17

during periods of emergence could allow mountain pine beetle to spread along transportation
corridors.
The initial movement of mountain pine beetle into the Peace River region was
monitored closely between 2004 and 2006. As a result of this vigilance, the initial years of
invasion are well documented. Using annual inventory maps of insect locations, the present
study aims to examine patterns of establishment within the Peace River region for evidence
suggestive of invasion mechanisms. We focus on two broad hypotheses. First, we examine
evidence of transport within the "insect boundary layer". Although the use of the term
"boundary layer" is typically restricted to the field of fluid dynamics and represents a viscous
layer between two mediums, the term "insect boundary layer" is defined as the altitude below
which an insect will determine its own spread and direction of movement (Taylor 1974).
Above this layer, the insect's movement will be influenced largely by wind speed and
direction. Mechanisms of dispersal occurring primarily within the insect boundary layer
include flights through suitable habitats located primarily along low-elevation corridors such
as river valleys (Robertson et al. 2009), and human-assisted dispersal due to transport of
infested material along roads and storage of infested material at milling sites. Second, we
examine evidence for transport above the insect boundary layer, which we characterize as the
spread of mountain pine beetles at altitudes in which they are blown passively, otherwise
known as aeolian dispersal (Szymkowiak et al. 2007; Jackson et al. 2008; Zhang et al. 2008).
In examining evidence for above-boundary layer transport, we compare establishment
patterns to landscape feature-independent patterns, such as a large gradient, suggesting
progressive movement of mountain pine beetle populations from sources parallel to the
western edge of the Rocky Mountains in BC, or a linear zone containing a high intensity of

18

infestations, suggesting long-distance aeolian dispersal and "rainout" as the primary
mechanisms of invasion respectively.
2.3: Material and Methods
2.3.1: Study Area and Spatial Datasets
The northern and southern extent of the research area, delineated between 54° and
56.5°, spanned from the Great Continental Divide within the North Rockies Mountains to the
British Columbia-Alberta border, covering a total area of approximately 3 million ha (Fig.
2.1). Forestry companies operate over a majority of this Peace River region, as the provincial
government of British Columbia is the predominant landowner and leases tree harvesting
rights to forest management companies. During the initial spread of mountain pine beetle
into the region, forestry companies coordinated with federal and provincial governments to
respond swiftly to the building outbreak. We obtained annual inventories of discrete, treekilling populations of mountain pine beetle (referred to hereafter as infestations), initially
collected to facilitate management planning, from these forest licensees. The infestations
inventoried in a single year were considered to have been attacked in the previous year as
they had dead trees with chlorotic or red foliage and were at least 0.01 ha in size.
Locations of "red-attack" trees had been inventoried from helicopter using GPS with
further ground reconnaissance work in some areas. Because epidemic populations of the
insect must kill their hosts to reproduce, fading foliage can be used to identify trees attacked
one year ago (Safranyik et al. 1974) and serve as a useful proxy for insect abundance
(Wulder et al. 2006; Aukema et al. 2008; Nelson and Boots 2008). Surveys were conducted
in 2004, 2005, and 2006 from approximately May to September. Centre point locations for
each infestation were recorded in Universal Transverse Mercator (UTM) coordinates.

19

Additional data collected for each infestation included its size, the approximate number of
trees affected, ecological land classification scheme, land tenure, and any control strategy
implemented.
In order to consider the potential influence of anthropogenic factors or the effect of
rivers on spread of mountain pine beetle, three datasets representing mill locations, primary
and secondary roads, and rivers were secured from Natural Resources Canada and Global
Forest Watch (2009). From these vector (i.e., point and line) datasets, the Euclidian
distances for all locations in the research area to features of interest (i.e., infestations in the
previous year, mills, roads, and/or rivers) were calculated. This resulted in four raster
datasets with a 73 x 73 metre pixel size (i.e., just over one-half hectare in size).
2.3.2: Statistical Analysis
Spatial point process regression models were fit to each annual inventory dataset with
covariates reflective of different hypothesized mechanisms of dispersal. This class of models
evaluates the intensity of event occurrences, A, (infestations per m2), relative to spatial
location (i.e. x and/or y positions) as well as other covariates (Baddeley and Turner 2008).
Specifically, we tested whether there was evidence for anthropogenic activities contributing
to spread via distances to roads and mills, evidence for natural low-elevation corridors via
associations with rivers, and evidence for aeolian dispersal via diffusion processes (linear
distance gradient from the Rocky Mountains and distances from point-source infestations in
the previous year), and/or wave propagation (polynomial distance terms) (see Appendix A
for R code). In our system, due to the uniform directional spread of infestations in a
northwest direction (see Results), there was a high degree of correlation between x and j>
data. Evaluating the effects of highly collinear variables in regression models can create

20

computational challenges (Graham 2003). Hence, x andy location data for models testing
evidence consistent with gradient or wave-propagation aeolian dispersal were collapsed into
a single variable measuring distance from a line of reference (d) just outside of the research
area. The line of reference, with a 138° bearing, was positioned parallel to the Rocky
Mountains and perpendicular to the dominant direction of spread for 2004 infestations;
reflecting the initial invasion event from the interior of British Columbia (48°; see Chapter
3). A raster dataset was then produced with each pixel representing the distance from the
reference line. Models were compared using Akaike's information criterion (AIC), with the
lowest AIC values identifying the models of best fit (Akaike 1973). All data were handled in
Arc View v.9.2. Analyses were completed using R v.2.8.1 and the spatstat package v. 1.14-9
(Ihaka and Gentleman 1996; R Development Core Team 2009).
2.4: Results
The occurrence of infestations of mountain pine beetle within the research area
increased annually between 2004 and 2006. In 2004 and 2005, 10,536 and 12,275
infestations were mapped respectively. These numbers tripled in 2006 to 35,084 recorded
infestations (Fig. 2.2). The average nearest neighbor distance between infestations was
greatest in 2006 (0.24 km) while in 2005 it was 0.23 km. Infestations in 2004 had the closest
average nearest neighbors at 0.18 km. Mean area of patches of red trees killed by mountain
pine beetle was quite small in 2004, approximately one third of a hectare on average. The
mean size increased to approximately 1 ha by 2006. Sizes of patches of dead trees became
more variable due to expanding areas that were colonized, as the largest areas measured in
2006 were 2,440 ha, versus 6 and 133 ha in 2004 and 2005 respectively. By 2006, the total
area of trees killed by mountain pine beetle had expanded to 35,084 ha (Fig. 2.2).

21

The topography in the region ranges from the sharp relief of the Rocky Mountains at
the western edge, undulate topography of the Foothills in the centre, and plateaus deeply
incised by various rivers in the east (Fig. 2.1). Most rivers in the Peace region flow into the
Arctic Ocean, as the area is east of the Great Divide but south of the Northern Continental
Divide. These rivers include the Kiskatinaw, Moberly, Murray, Narraway, Peace, Pine,
Pouce Coupe, Sukunka, and Wapiti Rivers as well as a large creek in the southern portion of
the area called Red Deer Creek. Primary industries in the region include oil and gas
extraction, mining and forestry such harvesting of soft wood conifers for lumber. There are
two mills in the Peace River region that process lodgepole pine, both in Chetwynd, British
Columbia.
The spread pattern of mountain pine beetles into the region closely resembled that of
a propagating wave in all three years (Table 2.1). Areas closest and furthest from the
initiation source had less infestation than those in the middle of the research area (Fig. 2.2).
These "wave-propogation" models had the lowest AIC values, fitting the data better than
models reflecting alternate modes or mechanisms of dispersal, including anthropogenic
movement of infested material and simple within-boundary layer dispersal between adjacent
stands. Anthropogenic movement of infested materials did not appear to impact the spread
of mountain pine beetle.
By solving the best-fitting spatial regression equations (Table 2.1) for their
maximums, we calculated the locations of greatest infestation densities in the study area
(Table 2.2). Mountain pine beetle exhibited highest density of infestation 46 km from the
edge of the research area in 2004 and 2005. This 'belt' progressed approximately 25 km to

22

the east in 2006, as the maximum density of infestations occurred further from the western
edge of the research area at 71 km (Table 2.2; Fig. 2.3).
2.5: Discussion
Our findings of a northeast-southwest belt of a high density of infestations of
mountain pine beetle within a habitat where they had not previously been detected is
consistent with spread by aeolian dispersal above the insect boundary layer from sources
west of the Rocky Mountains. Such long-distance transport is not unknown in the mountain
pine beetle system (Jackson et al. 2008), occurring in more southern and eastern portions of
its range (e.g., in Washington, Furniss and Furniss 1972; and from Alberta to the Cypress
Hills of Saskatchewan, Cerezke 1981). The present occurrence is the first successful
invasion of new habitat across the historic geoclimatic barrier of the Rocky Mountains in
northern Canada (Kurz et al. 2008; Robertson et al. 2009). Our study offers one of the first
insights of the spatial extent of aeolian dispersal events by studying establishment at a
landscape scale in the absence of populations with landscape signatures of dead trees
confounded by localized endemic-incipient eruptions. The wave front pattern is consistent
with deposition from weather fronts (Drake and Farrow 1988).
Aeolian dispersal is not the norm for mountain pine beetle, as their pheromonemediated host-seeking behaviour predisposes them to diffuse dispersal (Hengeveld 1988)
within the insect boundary layer. Upon emergence, mountain pine beetle, although
phototactic, will generally move within the stand a maximum of 250m from natal host
towards attractive stimuli such as adjacent pine (Safranyik et al. 1992). This movement may
occur against wind speeds to a maximum of approximately 11 km/h if the stimulus is upwind
of its natal host (Schmid et al. 1992). However, mountain pine beetle, like some Ips species,

23

generally require a short period of flight before engaging in host seeking behaviour (Gray et
al. 1972; Jactel 1991; Duelli et al. 1997). Once attracted by host volatiles, the insect uses a
combination of visual and gustatory cues for final host selection (Pureswaran and Borden
2003). Once mountain pine beetles have commenced attack on the new host, at the optimum
attack density, the insects release verbenone produced by microbial symbionts in the gut
(Hunt and Borden 1990). Verbenone, in combination with other semiochemicals, repel other
mountain pine beetles. The repelled beetles tend to attack suitable hosts in the surrounding
area, producing a "spillover" attack. At late outbreak stages, a depletion of suitable hosts
over a large area may predispose insects to move further from natal hosts, increasing abovecanopy flight (Salle and Raffa 2007). Above the forest canopy, insects may have become
entrained in advective currents above the insect boundary layer, facilitating long-distance
transport occasionally observed among other scolytids such as D. autographus Ratz. (Nilssen
1984) and/, typographus L. (Byers 2000).
Although landscape-scale wave-front patterns are not typically noted in mountain
pine beetle outbreaks (Aukema et al. 2006; Gamarra and He 2008), it is a typical pattern
characterizing other bark beetles in the genus Dendroctonus, such as the southern pine beetle
{DendroctonusfrontalisZimm.), as well as defoliators such as larch budmoth {Zeiraphera
diniana Gn.). Approximately 95% of southern pine beetles disperse within 2.3 km of their
natal host (Cronin et al. 2000), causing edges of wave-fronts to be separated by a maximum
of 8 km (Schowalter et al. 1981) as insects progressively move into adjacent stands (Turchin
and Thoeny 1993). Southern pine beetle is initially attracted to host volatiles, including apinene (Payne 1980). Attacking females release frontalin, attracting conspecifics. Once an
optimal attack density is reached, additional attacking individuals are repulsed by endo-

24

brevicomin and verbenone. Southern pine beetles dispersing from their natal hosts are highly
repulsed by already infested trees and attracted to suitable hosts on infestation edges (Payne
1980). As a result, at the edges of infestations of the previous year, new infestations will
establish, causing a wave-like pattern between years (Coulson et al. 1999). The stereotypic
dispersal and attack sequence results in delineated wave-fronts that provide avenues for
successful control in some areas of the southern United States.
Infestations of larch budmoth also progress in wave-like patterns that propagate in
time and space (Bjornstad et al. 2002; Johnson et al. 2004). However, the cause of these
patterns is unclear, with explanations ranging from habitat permeability (Johnson et al. 2004)
to aeolian dispersal to trophic interactions between the moth and its parasitoids (Bjornstad et
al. 2002). Making inference from landscape-level signatures can be challenging because
similar signatures can be caused by different processes. The association of landscape
features with insect establishment is the focus of Chapt. 3, while predators and parasitoids do
not generally play a major role in incipient-epidemic population dynamics of mountain pine
beetle (Cole 1981).
Following 2004, the highest intensity of infestations in the wave front moved a
further 25 km to the northeast from 2005 to 2006, even though the leading edge moved
approximately 80 km (Fig. 2.2). These distances are similar to dispersal exhibited by other
guilds of forest insects with strong dispersal capabilities, such as defoliators like larch
budmoth that can move upwards of 200 km per year (Bjornstad et al. 2002). The slow
progression of the zone of highest intensity of infestations from 2004 to 2005 suggests that
there were fewer weather events facilitating long-distance transport of mountain pine beetle
into the region after the initial invasion. Alternatively, the eastward progression of the zone

25

of highest intensity could reflect harvesting activities changing the densities of patches of
dead trees between years, although this is unlikely as harvesting was focused on the leading,
eastern edge. Eastward progression of the zone of highest intensity may have also reflected
short-distance dispersal after the initial invasion events, or higher reproductive capacity in
novel host trees not previously exposed to historic beetle pressure (Cudmore 2009).
Our study found no evidence that the spread of mountain pine beetle into the Peace
River region was primarily influenced by anthropogenic activities such as storage of
potentially infested logs within the region. Roads, along which infested logs harvested for
firewood could have been transported, and mill locations, at which infested logs were stored,
did not cause an increased occurrence of mountain pine beetle in the surrounding area.
Although invasives can be readily transported along road corridors (Christen and Matlack
2009) and scolytids such as Tomicus piniperda (L.) have been noted to spread from infested
material in mill yards surrounding forests (Poland et al. 2000), our findings are not
unexpected. Schaupp et al. (1993) found that emergence of mountain pine beetles from
infested logs in transport is generally quite small during periods of peak emergence (0.03
insects/m2 July and August). This is considerably lower than the optimal attack density of
mountain pine beetle on standing living trees, 62 insects per m2 (Raffa and Berryman 1983).
In addition, the use of various storage methods, such as increasing sun exposure by avoiding
piling during storage and covering logs in plastic to increase the air temperatures to a level
that causes high mortality in broods infesting the logs, can minimize emergence from stored
logs (Negron et al. 2001) (although we do not know if such methods were utilized). It
appears as though mitigation strategies used by forestry companies in this situation, namely

26

prioritization of processing of infested logs, at the mill locations effectively minimized the
spread of the insect through emergence from host materials.
In conclusion, examination of establishment patterns across years supports the
hypothesis that mountain pine beetles moved into the region via long-distance transportation
by meso-scale atmospheric transport (Robertson et al. 2009). Documenting range expansion
and long-distance dispersal events are particularly important for allowing the estimation of
rate of spread and species' modes of dispersal (Brouwers and Newton 2009). Examining
expansion at the northern margins of ranges, as in this case, also provides an accounting of
impacts due to climate change. With mountain pine beetle first moving over a geoclimatic
barrier and then successfully establishing outbreaking populations in a novel habitat, the
patterns of continued spread can provide valuable information for predicting the potential
extent of further range expansion.
2.6: Acknowledgements
Funding for this study was provided by the Mountain Pine Beetle Initiative of Natural
Resources Canada, Canadian Forest Service, NSERC, and Pacific Forestry Centre Graduate
Student and Association of Professional Biologists Scholarships awarded to HMG. We
thank B. Pate of West Fraser Timber Ltd. and C. Johnson of Canadian Forest Products Ltd.
for providing advice and compiled aerial survey data within the region. G. Thandi, Canadian
Forest Service, Pacific Forestry Centre, A. McGill, Alberta Sustainable Resource
Development, and P. Bai, University of Northern British Columbia provided GIS assistance.
A. Baddeley, School of Mathematics and Statistics, University of Western Australia and J.
Zhu, Department of Statistics, University of Wisconsin - Madison, provided advice on
implementation of spatial point process models. The authors also thank the assistance and

27

support of Marnie Duthie-Holt, Matthew Klingenberg, Fraser McKee, Sandra Allen, Sonia
Wesche, and Doug Linton.

28

2.7: Figure Captions
Figure 2.1 Map of the Peace River region of British Columbia including mill locations,
primary rivers, and highways.
Figure 2.2 Location of infestations of mountain pine beetle in the Peace River region of
British Columbia, Canada in 2004, 2005, and 2006. Dashed line represents the line of
initiation which is parallel to the Rocky Mountains and perpendicular to the direction
of spread in 2004, the initial year of detection.
Figure 2.3 Location of highest intensity of infestations of mountain pine beetle in 2004 and
2005 (line A) and 2006 (line B) within the research area.

29

Fig. 2.1
N

A

-•

L

r

^44ufetnn'ii Hope

- _ • •

'"')

Da\v,son Greek

•

Chetwynd

• i
Pouce Coupe

;

Tumbler Ridge

Location of Mills
Municipalities
Roads
Rivers
Research Area
Lakes
o

-•

\.
t--

lOO "^\

Kilometres

30

I

Fig. 2.2

•*3S^J#>-,-iJft

.„*.

* - ' ' " J t ^•iSi
-%.C*

k

^

k&"

- r ^

>•• -

"1

»'-

•*•.?-.>. , -• ' v V , , ,

Infestations.
2004

'

"3v

20 05
•

2006
0

i()

Initiation Line
Kilometer!

Research Area

31

Fig. 2.3

N

ROCKY
MOUNTAIN
FOOTHILLS

AINS

32

Table 2.1 AIC values for the spatial point process models reflecting various hypotheses of
insect dispersal for 2004, 2005, and 2006 in the Peace River region of British Columbia.
Each model uses the spatially explicit intensities of beetle-killed trees for the respective year
as a function of an intercept term and a similarly spatially-explicit covariate(s), as listed (see
Methods). Models with lowest AIC values are judged to fit the best.
Hypothesis

Covariate(s)

2004

2005

2006

Dispersal primarily occurs

Roads (anthropogenic activities)

333,332

383,898

1,027,245

within insect boundary

Mill locations (anthropogenic activities)

333,532

385,471

1,024,913

layer (human-assisted

River systems (natural corridors)

332,196

384,841

1,027,961

transport/natural corridors/ Roads and mill locations

328,444

381,863

1,024,729

adjacent stands).

Roads and rivers

331,337

383,182

1,027,197

Rivers and mill locations

331,703

384,835

1,024,851

Rivers, roads and mill locations

325,575

380,761

1,024,621

a

370,646

1,023,605

Beetle locations(t.i) (dispersal from previous year's establishment) Dispersal primarily occurs

Distance (linear gradient from source)

324,186

373,781

1,024,656

above insect boundary

Distance + distance2 (wave propagation)

315,602

364,567

1,009,046

-

369,445

1,023,607

-

367,848

1,021,350

layer (aeolian dispersal)
Combination (above and
within insect boundary

Distance + beetle locations^.,)
2

Distance + distance + beetle locations(t.i)

layer)
a

No inventory exists for 2003
Distance from line of initiation (parallel to the Rocky Mountains), see Methods
Models of best fit are indicated in bold.

33

Table 2.2 The parameter estimates for spatial point process models judged to best fit
infestation locations for 2004, 2005, and 2006 in the Peace River region of British Columbia
(see Table 2.1). Distance from initiation line is represented by W while the response
variable for the models is log (A), number of infestations of mountain pine beetle per m2.
The location of greatest density of infestations, i.e. distance from the initiation line, was
estimated from the point of inflection for each polynomial regression equation.
Year

Intercept

Model Parameters
d

Distance to maximum
d2

iinfestation intensity (from initiation

line; see Methods) (km)

2004
2005
2006

-20
-19
-17

2.82E-04
2.56E-04
1.03E-04

34

-3.05E-09
-2.79E-09
-7.23E-10

46.3
45.8
71.1

2.8: Literature Cited
Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle.
Pages 267-281 in B.N. Petrov and F. Csaki, editors. The 2 nd International Symposium
on Information Theory. Akademiai Kiado, Budapest.
Amman, G. D. and W. E. Cole. 1983. Mountain pine beetle dynamics in lodgepole pine
forests. Part II: population dynamics. USDA Forest Service, Intermountain Forest and
Range Experiment Station, Ogden, Utah. General Technical Report.
Aukema, B. H., A. L. Carroll, Y. Zheng, J. Zhu, K. F. Raffa, R. D. Moore, K. Stahl and S. W.
Taylor. 2008. Movement of outbreak populations of mountain pine beetle: influences
of spatiotemporal patterns and climate. Ecography 31 (3): 348-358.
Aukema, B. H., A. L. Carroll, J. Zhu, K. F. Raffa, T. A. Sickley and S. W. Taylor. 2006.
Landscape level analysis of mountain pine beetle in British Columbia, Canada:
spatiotemporal development and spatial synchrony within the present outbreak.
Ecography 29(3): 427-441.
Baddeley, A. and R. Turner. 2008. Spatstat: A package for analyzing spatial point processes
in R. Perth, Australia, <http://www.spatstat.org/>.
Baltensweiler, W. and A. Fischlin. 1979. The role of migration for the population dynamics
of the larch bud moth, Zeiraphera diniana Gn. (Lep. Tortricidae). Bulletin de la
Societe Entomologique Suisse 52: 259-271.
BenDor, T. K., S. S. Metcalf, L. E. Fontenot, B. Sangunett and B. Hannon. 2006. Modeling
the spread of the emerald ash borer. Ecological Modelling 197(1-2): 221-236.
Bjornstad, O. N., M. Peltonen, A. M. Liebhold and W. Baltensweiler. 2002. Waves of larch
budmoth outbreaks in the european Alps. Science 298(5595): 1020-1023.
Bright, D. 1976. The Insects and Arachnids of Canada. Communication Group, Ottawa,
Ontario, Canada.
British Columbia Ministry of Forests and Range. 2009. Ministry of Forests and Range: beetle
facts. Accessed on July 27th, 2009 from
<http://www.for.gov.bc.ca/hfp/mountain_pine_beetle/facts.htm>.
Brockerhoff, E. G., A. M. Liebhold and H. Jactel. 2006. The ecology of forest insect
invasions and advances in their management. Canadian Journal of Forest Research
36(2): 263-268.
Brouwers, N. C. and A. C. Newton. 2009. Movement rates of woodland invertebrates: a
systematic review of empirical evidence. Insect Conservation and Diversity 2(1): 1022.
Byers, J. A. 2000. Wind-aided dispersal of simulated bark beetles flying through forests.
Ecological Modeling 125(231 -243).
Cerezke, H. F. 1981. Mountain pine beetle in southern Alberta and Saskatchewan. Prairie
Landscape Magazine. Environment Canada, Canadian Forestry Service, Northern
Forest Research Centre.
35

Cerezke, H. F. 1995. Egg gallery, brood production, and adult characteristics of mountain
pine beetle, Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae), in three
pine hosts. Canadian Entomologist 127(6): 955-965.
Christen, D. and G. Matlack. 2009. The habitat and conduit functions of roads in the spread
of three invasive plant species. Biological Invasions 11(2): 453-465.
Cole, W. E. 1981. Some risks and causes of mortality in mountain pine beetle populations: A
long-term analysis. Researches on Population Ecology 23(1): 116-144.
Conn, J. E. 1981. Pheromone production and control mechanisms in Dendroctonus
ponderosae Hopkins. Burnaby, British Columbia, Simon Fraser University. Master of
Science thesis.
Coulson, R. N., B. A. McFadden, P. E. Pulley, C. N. Lovelady, J. W. Fitzgerald and S. B.
Jack. 1999. Heterogeneity of forest landscapes and the distribution and abundance of
the southern pine beetle. Forest Ecology and Management 114(2-3): 471-485.
Cronin, J. T., J. D. Reeve, R. Wilkens and P. Turchin. 2000. The pattern and range of
movement of a checkered beetle predator relative to its bark beetle prey. Oikos 90(1):
127-138.
Cudmore, T. 2009. Geographic variation in the effect of tree diameter in mountain pine
beetle productivity in lodgepole pine. Natural Resources and Environmental Sciences.
Prince George, British Columbia, University of Northern British Columbia. Masters
of Science thesis.
Drake, V. A. and R. A. Farrow. 1988. The influence of atmospheric structure and motions on
insect migration. Annual Review of Entomology 33(1): 183-210.
Duelli, P., P. Zahradnik, M. Knizek and B. Kalinova. 1997. Migration in spruce bark beetles
(Ips typographus L.) and the efficiency of pheromone traps. Journal of Applied
Entomology 121(6): 297-303.
Furniss, M. M. and R. L. Furniss. 1972. Scolytids (Coleoptera) on snowfields above
timberline in Oregon and Washington. Canadian Entomologist 104: 1471-1478.
Furniss, R. L. and V. M. Carolin. 1977. Western forest insects., USDA Forest Service.
Miscellaneous Publications. Number 1339.
Gamarra, J. G. P. and F. He. 2008. Spatial scaling of mountain pine beetle infestations.
Journal of Animal Ecology 77(4): 796-801.
Gilbert, M., N. Fielding, H. F. Evans and J. C. Gregoire. 2003. Spatial pattern of invading
Dendroctonus micans (Coleoptera: Scolytidae) populations in the United Kingdom.
Canadian Journal of Forest Research 33(4): 712.
Graham, M. H. 2003. Confronting multicollinearity in ecological multiple regression.
£co/ogy 84: 2809-2815.
Gray, B., R. F. Billings, R. I. Gara and R. L. Johnsey. 1972. On the emergence and initial
flight behaviour of the mountain pine beetle, Dendroctonus ponderosae, in eastern
Washington. Journal of Applied Entomology 21: 250-259.

36

Haack, R. A. 2006. Exotic bark- and wood-boring Coleoptera in the United States: recent
establishments and interceptions. Canadian Journal of Forest Research 36(2): 269288.
Hengeveld, R. 1988. Mechanisms of biological invasions. Journal of Biogeography 15(5/6):
819-828.
Holway, D. A. and A. V. Suarez. 1999. Animal behavior: an essential component of invasion
biology. Trends in Ecology and Evolution 14: 328-330.
Huber, D. P. W., B. H. Aukema, R. S. Hodgkinson and B. S. Lindgren. 2009. Successful
colonization, reproduction, and new generation emergence in live interior hybrid
spruce Picea engelmannii x glauca by mountain pine beetle Dendroctonus
ponderosae. Agricultural and Forest Entomology 11(1): 83-89.
Hunt, D. W. A. and J. H. Borden. 1990. Conversion of verbenols to verbenone by yeasts
isolated from Dendroctonus ponderosae (Coleoptera: Scolytidae). Journal of
Chemical Ecology 16(4): 1385-1397.
Ihaka, R. and R. Gentleman. 1996. R: A language for data analysis and graphics. Journal of
Computational and Graphical Statistics 5(3): 299-314.
Iverson, L. R. and A. M. Prasad. 1998. Predicting abundance of 80 tree species following
climate change in the eastern United States. Ecological Monographs 68(4): 465-485.
Jackson, P. L., D. Straussfogel, B. S. Lindgren, S. Mitchell and B. Murphy. 2008. Radar
observation and aerial capture of mountain pine beetle, Dendroctonus ponderosae
Hopk. (Coleoptera: Scolytidae) in flight above the forest canopy. Canadian Journal
of Forest Research 38(8): 2313-2327.
Jactel, H. 1991. Dispersal and flight behaviour of Ips sexdentatus (Coleoptera: Scolytidae) in
pine forest. Annales des Sciences Forestieres 48: 417-428.
Johnson, D., O. Bj0rnstad and A. Liebhold. 2006. Landscape mosaic induces traveling waves
of insect outbreaks. Oecologia 148(1): 51-60.
Johnson, D. M., O. N. Bjorastad and A. M. Liebhold. 2004. Landscape geometry and
travelling waves in the larch budmoth. Ecology Letters 7: 967-974.
Johnson, D. M., A. M. Liebhold, P. C. Tobin and O. N. Bjornstad. 2006. Allee effects and
pulsed invasion by the gypsy moth. Nature 444(7117): 361-363.
Kurz, W. A., C. C. Dymond, G. Stinson, G. J. Rampley, E. T. Neilson, A. L. Carroll, T.
Ebata and L. Safranyik. 2008. Mountain pine beetle and forest carbon feedback to
climate change. Nature 452: 987-990.
Libbey, L. M., L. C. Ryker and K. L. Yandell. 1985. Laboratory and field studies of volatiles
released by Dendroctonus ponderosae Hopkins (Coleoptera, Scolytidae). Journal of
Applied Entomology 100: 381-392.
Nealis, V. and B. Peter. 2008. Risk assessment of the threat of mountain pine beetle to
Canada's boreal and eastern pine forests. Natural Resources Canada, Canadian Forest
Service, Pacific Forestry Centre, Victoria, British Columbia, Canada. Information
Report BC-X-417.

37

Negron, J. F., W. D. Shepperd, S. A. Mata, J. B. Popp, L. A. Asherin, A. W. Schoettle, J. M.
Schmid and D. A. Leatherman. 2001. Solar treatments for reducing survival of
mountain pine beetle in infested ponderosa and lodgepole pine logs. USDA, Forest
Service, Rocky Mountain Research Station. Fort Collins, Colorado, USA. Research
Paper RMRS-RP-30.
Nelson, T. A. and B. Boots. 2008. Detecting spatial hot spots in landscape ecology.
Ecography 31(5): 556-566.
Nilssen, A. 1984. Long-range aerial dispersal of bark beetles and bark weevils (Coleoptera,
Scolytidae and Curculionidae) in northern Finland. Annales Entomologici Fennici
50(2): 37-42.
Payne, T. L. 1980. Life history and habits. Pages 7-28 in R. C. Thatcher, J. L. Searcy, J. E.
Coster and G. D. Hertel. The Southern Pine Beetle. USDA, Forest Service, Science
and Education Administration.
Peltonen, M., A. M. Liebhold, O. N. Bjornstad and D. W. Williams. 2002. "Spatial
synchrony in forest insect outbreaks: Roles of regional stochasticity and dispersal."
Ecology 83(11): 3120-3129.
Peterson, A. T. and D. A. Vieglais. 2001. Predicting species invasions using ecological niche
modeling: new approaches from bioinformatics attack a pressing problem. Bioscience
51(5): 363-371.
Poland, T. M., R. A. Haack, T. R. Petrice, C. S. Sadof and D. W. Onstad. 2000. "Dispersal of
Tomicus piniperda (Coleoptera: Scolytidae) from operational and simulated mill
yards. Canadian Entomolgist 132: 853-866.
Pureswaran, D. S. and J. H. Borden. 2003. Test of semiochemical mediated host specificity
in four species of tree killing bark beetles (Coleoptera: Scolytidae). Environmental
Entomology 32(5): 963-969.
R Development Core Team. 2009. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, <http://www.Rproject.org>.
Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner and W. H.
Romme. 2008. Cross-scale drivers of natural disturbances prone to anthropogenic
amplification: the dynamics of bark beetle eruptions. Bioscience 58(6): 501-517.
Raffa, K. F. and A. A. Berryman. 1979. Flight responses and host selection by bark beetles.
In A. A. Berryman and L. Safranyik. Dispersal of forest insects: evaluation, theory
and management implications. Pullman, Washington, USA, Washington State
University.
Raffa, K. F. and A. A. Berryman. 1983. The role of host plant resistance in the colonization
behavior and ecology of bark beetles (Coleoptera: Scolytidae). Ecological
Monographs 53(1): 27-49.
Reid, R. W. 1962. Biology of the mountain pine beetle, Dendroctonus monticolae Hopkins,
in the East Kootenay region of British Columbia. I. Life cycle, brood development,
and flight periods. Canadian Entomologist 94: 531-538.

38

Reid, R. W. 1963. Biology of the mountain pine beetle, Dendroctonus monticolae Hopkins,
in the East Kootenay region of British Columbia. III. Interaction between the beetle
and its host, with emphasis on brood mortality and survival. Canadian Entomologist
95(3): 225-238.
Reid, R. W. and H. S. Gates. 1970. Effect of temperature and resin on hatch of eggs of the
mountain pine beetle (Dendroctonus ponderosae). The Canadian Entomologist
102(5): 617-622.
Renwick, J. A. A. and J. P. Vite. 1970. Systems of chemical communication in
Dendroctonus. Contributions of Boyce-Thompson Institute of Plant Research 24: 283292.
Safranyik, L. and A. L. Carroll. 2006. The biology and epidemiology of the mountain pine
beetle in lodegpole pine forests. Pages 3-66 in L. Safranyik and B. Wilson. The
Mountain Pine Beetle: A Synthesis of Biology, Management, and Impacts on
Lodgepole Pine. Natural Resources Canada, Canadian Forest Service, Pacific
Forestry Centre, Victoria, British Columbia.
Safranyik, L. and D. A. Linton. 1983. Brood production by three spp. of Dendroctonus
(Coleoptera: Scolytidae) in bolts from host and nonhost trees. Journal of the
Entomological Society of British Columbia 80: 10-13.
Safranyik, L., D. A. Linton, R. Silversides and L. H. McMullen. 1992. Dispersal of released
mountain pine beetles under the canopy of a mature lodgepole pine stand. Journal of
Applied Entomology 113: 441-450.
Safranyik, L., D. M. Shrimpton and H. S. Whitney. 1974. Management of lodgepole pine to
reduce losses from the mountain pine beetle. Natural Resources Canada, Canadian
Forest Service, Pacific Forestry Centre, Victoria, British Columbia. Technical Report
1.
Safranyik, L., D.M. Shrimpton, and H.S. Whitney. 1975. An interpretation of the interaction
between lodgepole pine, the mountain pine beetle and its associated blue stain fungi in
western Canada. Pages 406-428 in D.M. Baumgartner, editor. Symposium Proceedings:
Management of Lodgepole Pine Ecosystem. Washington State University Cooperative
Extension Service, Pullman, Washington.
Salle, A. and K. Raffa. 2007. Interactions among intraspecific competition, emergence
patterns, and host selection behaviour in Ipspini (Coleoptera: Scolytinae). Ecological
Entomology 32(2): 162-172.
Schaupp, W.C., J.E. Pasek, J.M. Schmid, S.A. Mata, and C.K. Lister. 1993. Mountain pine
beetle emergence from infested logs during hauling. USDA, Forestry Service, Rocky
Mountain Forest & Range Experiment Station, Fort Collins, Co. Research Note RM522.
Schmid, J. M., S. A. Mata and D. C. Allen. 1992. Potential influences of horizontal and
vertical air movement in ponderosa pine stands on mountain pine beetle dispersal.
USDA Rocky Mountain Forest & Range Experiment, Fort Collins, Colorado.
Research Note RM-516.

39

Schowalter, T. D., D. N. Pope, R. N. Coulson and W. S. Fargo. 1981. Patterns of southern
pine-beetle (Dendroctonus frontalis Zimm) infestation enlargement. Forest Science
27(4): 837-849.
Shore, T. L. and L. Safranyik. 1992. Susceptibility and risk rating systems for mountain pine
beetle in lodgepole pine stands. Natural Resources Canada, Canadian Forest Service,
Pacific Forestry Centre, Victoria, British Columbia. Information Report BC-X-336.
Skarpaas, O. and B. 0kland. 2009. Timber import and the risk of forest pest introductions.
Journal of Applied Ecology 46(1): 55-63.
Suarez, A. V., D. A. Holway and T. J. Case. 2001. Patterns of spread in biological invasions
dominated by long-distance jump dispersal: Insights from Argentine ants. Ecology
98(3): 1095-1100.
Szymkowiak, P., G. Gorski and D. Bajerlein. 2007. Passive dispersal in arachnids. Biological
Letters 44(2): 75-101.
Tackenberg, O. 2003. Modeling long-distance dispersal of plant diaspores by wind.
Ecological Monographs 73(2): 173-189.
Taylor, L. R. 1974. Insect migration, flight periodicity and the boundary layer. Journal of
Animal Ecology 43(1): 225-238.
Tobin, P. C. and L. M. Blackburn. 2008. Long-distance dispersal of the gypsy moth
(Lepidoptera : Lymantriidae) facilitated its initial invasion of Wisconsin.
Environmental Entomology 37(1): 87-93.
Turchin, P. and W. T. Theony. 1993. Quantifying dispersal of southern pine beetles with
mark-recapture experiments and a diffusion model. Ecological Applications 3(187198).
Wood, L. 2006. An investigation of natural hybridization between jack pine {Pinus
banksiand) and Lodgepole Pine {Pinus contorta var. latifolid) in Northern British
Columbia. Prince George, British Columiba, University of Northern British
Columbia. Masters of Science thesis.
Wulder, M. A., C. C. Dymond, J. C. White and B. Erikson. 2006. Detection, Mapping, and
Monitoring of the Mountain Pine Beetle. Pages 123-154 in L. Safranyik and B.
Wilson. The Mountain Pine Beetle a Synthesis of Biology, Management, and Impacts
on Lodgepole Pine. Natural Resources Canada, Canadian Forestry Service, Pacific
Forestry Centre, Victoria, British Columbia.
Zhang, Y., L. Wang, K. Wu, K. A. G. Wyckhuys and G. E. Heimpel. 2008. Flight
performance of the soybean aphid, Aphis glycines (Hemiptera: Aphididae) under
different temperature and humidity regimens. Environmental Entomology 37(2): 301306.

40

3: THE ASSOCIATION BETWEEN ESTABLISHMENT AND SPREAD OF
INFESTATIONS OF MOUNTAIN PINE BEETLE AND LANDSCAPE
FEATURES IN THE PEACE RIVER REGION OF BRITISH COLUMBIA
3.1: Abstract
The spread of mountain pine beetle since 2002 across the Rocky Mountains into
northeastern British Columbia, an area not previously exposed to epidemic beetle pressure,
has been facilitated largely by above-canopy dispersal of the insect by meso-scale
atmospheric currents. Terrain-induced thermals likely influence population dispersal and
establishment as insects may behave like inert particles at these scales. Spatial point process
modelling was used to examine the impact of landscape-scale variables, including landscape
features and their orientations, habitat suitability, and elevation, on occurrence of mountain
pine beetle infestations in 2004, 2005, and 2006. The efficacy of treatment and control
efforts in those years is also examined. The terrain within the research area was categorized
according to a landscape feature classification scheme identifying landscape features that
could potentially influence meso-scale convective currents and dispersal patterns of mountain
pine beetle. Although plains, open slopes and U-shaped valleys occupied the largest areas in
the landscape, mountain pine beetle infestations primarily established in U-shaped valleys.
Other landscape features that had a positive association with infestations of mountain pine
beetle differed between years. These features included canyons and deeply incised streams in
2004, local ridges and hills in valleys in 2005, midslope ridges and small hills in plains in
2004 and 2006, and open slopes in 2006. The orientation of landscape features was found to
only influence the occurrence of mountain pine beetle infestations in 2006, with
southwestern slopes of midslope ridges or small hills in plains, southwest facing open slopes,

41

and U-shaped valleys that run in a northeast-southwest cardinal direction positively
influencing intensity. These findings may be used to prioritize treatment areas prior to
infestation by mountain pine beetle based on the landscape features within a target area.

42

3.2: Introduction
Landscape pattern and structure play important roles governing the spread and impact
of insect herbivores in forest ecosystems, both directly by influencing the dispersal
capabilities of insect populations and indirectly by influencing the health and distribution of
host trees (Bjornstad et al. 2002; Ims and Coulson 2004). For example, Adelges tsugae
Annand (Homoptera: Adelgidae) (hemlock woolly adelgid) feeds on native hemlock species
in the eastern United States and disperses via wind (Orwig et al. 2002), severely impacting
stands on xeric ridges and slopes. Similarly, frontal zones of outbreaks of Operophtera
brumata L. (Lepidoptera: Geometridae) (winter moth), tend to vary in size according to
topographic characteristics such as distance from the coast, altitude, and slope (Hagen et al.
2007). Specific landform characteristics can also greatly influence patterns of infestation,
usually due to alteration of host vigour and levels of resistance. For example, infestations of
Dendroctonus frontalis Zimmermann (Coleoptera: Scolytidae) (southern pine beetle) have
been found to be closely associated with landforms including upland plateaus while showing
lower-than-expected infestations on side slopes and steep side slopes (Hicks 1980). In
western North America, recent range expansion events across presumed geoclimatic barriers
by outbreaking populations of D. ponderosae Hopkins (mountain pine beetle) provide a new
opportunity to study the effects of landforms and feature orientation in an invasion process.
The mountain pine beetle is a cryptic herbivore that spends all but a few days of its
life cycle under the bark of mature pine trees. Typically, the insect exists for long periods at
endemic levels, but may intermittently undergo drastic population eruptions (Safranyik et al.
1974; Amman and Cole 1983; Zhang and Alfaro 2003; Gamarra and He 2008). At epidemic
population phases (Raffa and Berryman 1983), the insect must kill its host in order to

43

reproduce, exerting positive feedback in a system with landscape-scale consequences (Raffa
et al. 2008). Outbreaks typically decline when the host supply is exhausted or large-scale,
lethal climatological events result in mortality of a large proportion of the population (Stahl
et al. 2006; Safranyik and Carroll 2006).
Currently, there is a massive outbreak of mountain pine beetle covering
approximately 14.5 million ha of mature forests ofPinus contorta var. latifolia (Engelmann)
Critchfield (lodgepole pine) in the provinces of British Columbia and Alberta in western
Canada (British Columbia Ministry of Forests and Range 2009), exerting carbon-budget
impacts in the order of megatonnes (Kurz et al. 2008). The occurrence of an outbreak is not
unusual per se given the presence of a favourable climate for the insect and a high proportion
of stands with suitable hosts available on the landscape (Thomson and Shrimpton 1984;
Safranyik and Carroll 2006; Campbell et al. 2007; Aukema et al. 2008; Nelson and Boots
2008). However, the magnitude, extent, and severity have been unprecedented as the
outbreak has been exacerbated in large part by anthropogenic impacts such as altered forest
management regimes (i.e., fire suppression) and a changing climate (Carroll et al. 2004;
Taylor et al. 2006; Raffa et al. 2008). The outbreak originated in the central interior of the
province of British Columbia, Canada in the late 1990s (Aukema et al. 2006). However, the
first inventory of incipient-epidemic infestations in northeastern British Columbia on the
eastern slopes of the Rocky Mountains occurred in 2004, from an invasion presumed to have
occurred in 2002 (Allan Carroll, pers. comm.1). Mountain pine beetle was previously
thought not to be endemic to this area, as the Rocky Mountains in the central part of the
province have historically been considered a geoclimatic barrier (Safranyik and Linton
1982). However, over the next few years, outbreaking populations of mountain pine beetle
'Allan Carroll, Associate Professor, University of British Columbia fallan.carroll(a),ubc.ca)
44

quickly spread from the eastern slopes of the Rocky Mountains in British Columbia into
western Alberta. Currently, there is great concern that establishment in hybrid P. contorta x
P. banksiana Lamb, (jack x lodgepole pine), and subsequently P. banksiana further to the
east, may provide a conduit through the boreal forests of Canada to areas and hosts not
previously exposed, such as P. resinosa Aiton (red pine) or P. strobus L. (eastern white pine)
in eastern Canada and United States (Safranyik and Linton 1982; Cerezke 1995; Nealis and
Peter 2008).
Host-seeking behaviour of mountain pine beetle is characterized primarily by shortdistance dispersal in which adult progeny orient toward aggregation pheromones produced
by conspecifics, attacking neighbouring hosts en masse (Safranyik et al. 1989). Dispersal
flight is typically restricted to host-seeking behaviour within the stand (Safranyik et al.
1992), although newly emerged beetles do require a period of flight before becoming
receptive to host volatiles (Shepherd 1966). Newly emerged beetles are attracted to dark
objects on a light background (Shepherd 1966), and as a result may orient themselves
towards spot sources and the canopy resulting in downwind and upward dispersal of
individuals. This may allow beetles to move above the canopy and become incorporated into
convective currents during warm, fair-weather periods (Chapman 1962; Safranyik et al.
1989). Once entrapped by meso-scale atmospheric currents the movement of insects is
similar to the drift of inert particles (Taylor 1974), although insects may maintain a degree of
flight control (Lewis and Dibley 1970). It is likely that population pressure, exacerbated by
high levels of competition and lack of suitable hosts in the central interior of British
Columbia, contributed to increased above-canopy movement, incorporation into advective
currents, and subsequent long-distance dispersal of mountain pine beetle over the Rocky

45

Mountains (Jackson et al. 2005; Jackson et al. 2008; Robertson et al. 2009). Similar longdistance transport events of bark beetles have been noted previously. For example,
infestations of mountain pine beetle in the Cypress Hills on the southern border of Alberta
and Sasketchewan occurred approximately 200 km from outbreaking source populations in
western Alberta in 1980 (Cerezke 1981; Safranyik and Linton 1982) and in the early 1970's
Furniss and Furniss (1972) catalogued scolytids deposited on mountain glaciers in Oregon
and Washington.
In situations where long-distance dispersal dominates, alterations of wind velocity,
turbulence, and direction determine the transport, deposition, and, where there is susceptible
host and suitable climate, the establishment of insect populations. The descent of mountain
pine beetle from convective currents occurs via gravitational settling, active flight, rainout,
and/or impaction (i.e. the contact of biota with objects). In particular, landscape features
provide impactive surfaces for interception of insects (Defant 1951; Bullard et al. 2000), as
demonstrated by arthropods in arctic ecosystems as well as gypsy moth in mountain-valley
systems (Mason and McManus 1981; Ashmole et al. 1983; Antor 1994). The greater the
mass of an airborne object, the more likely it will impact the ground or surface rather than be
swept around the obstacle into lee-ward eddies (Westbrook and Isard 1999). Wind speed
may decrease when the direction of motion is perpendicular to emergent landscape features,
causing increased settlement in areas with reduced wind speeds (Ruel et al. 2001).
Alternatively, wind speed may increase, depending on fetch, within open areas, which may
increase the transportation of insects longer distances within valleys, as the insects become
entrained within air currents in these regions.

46

Despite the importance of topography on settlement patterns of insects exhibiting
aeolian dispersal, the influence of topography on the deposition patterns of mountain pine
beetle in these recent range expansion events has not been studied. The Peace River region
exhibits a diversity of topographical features. At the western edge of the region, the terrain is
dominated by the Northern Rocky Mountains with strongly linear terrain features formed by
erosion of folded and faulted sedimentary rocks (Harcombe 1978). These linear features
include large U-shaped valleys, eroded by glaciers, separated by distinct ridges (British
Columbia Ministry of Environment 1994). The Hart Range, the primary mountain range in
the region containing the Solitude and Murray Ranges, dissects the region at a central belt of
lower elevation in the Rocky Mountains. This range encompasses two low-lying passes,
including the Pine Pass (874m), the lowest of six highway passes through the northern Rocky
Mountains, and the Monkman Pass (1092m). The eastern portion of the research area is
composed of flat-lying or gently dipping sedimentary rocks producing plateau topography
(Harcombe 1978). Due to the relatively consistent directions of meso-scale winds in the
region, it is likely that the interaction between landscape features and convective currents
predictably influences deposition and the establishment patterns of D. ponderosae relative to
feature orientation.
The primary objectives of the present study were fourfold: (1) to examine the
association of landscape features on the establishment and persistence of insect populations
of sufficient size to kill mature trees in a newly-invaded area, (2) to determine whether the
specific orientations of those features provided additional inference on locations of insect
establishment, (3) to examine the relative contributions of other landscape characteristics,
including elevation and susceptible habitat, on the establishment patterns of mountain pine

47

beetle during the initial stages of invasion, and finally, (4) to assess the efficacy of treatment
and control efforts on spread of infestations within the region to better understand the impact
of long-distance dispersal on control efforts. Investigation of factors driving establishment of
mountain pine beetle infestations within this region allows retrospective examination of the
invasion event(s) and may provide critical information to formulate strategies to mitigate
further invasion and range expansion events of this eruptive forest herbivore.
3.3: Materials and Methods
3.3.1: Preparation of Insect Datasets
A study area was delineated that encompassed the invasion process of outbreaking
populations of mountain pine beetle over the northeastern slopes of the Rocky Mountains
from 2004 to 2006. This area encompassed the Peace River region from the Rocky
Mountains to the border of the province of Alberta, an area approximately 3 million hectares
in size (Fig. 3.1).
Detailed annual survey maps of discrete outbreaking populations of mountain pine
beetle were obtained from forest licensees. The provincial government is the predominant
landowner in the province of British Columbia, particularly within the Peace River region,
which in turn leases tree-harvesting rights to forest management companies. As such, we are
confident in the spatial coverage of the data. The surveys were conducted by identifying and
recording locations of "red-attack" trees from helicopter using GPS, in concert with further
ground reconnaissance work in some areas. Because mountain pine beetles in epidemic
population phases must kill their host in order to reproduce, and foliage fades from green to
red within one year after colonization (Safranyik et al. 1974), mapping "red-attack" is a
reliable proxy for estimates of insect abundance (Aukema et al. 2006; Wulder et al. 2006).

48

The surveys had been conducted in 2004, 2005, and 2006 from approximately May to
September. Centre point locations for each infestation were recorded in Universal
Transverse Mercator (UTM) coordinates. Additional data collected for each infestation
included its size, the approximate number of trees affected, ecological land classification
scheme, land tenure, and any control strategy implemented (Table 3.1).
Detailed error checking revealed that only 113 of 58,008 data points were of
questionable quality (duplicates in year 2004; 1% of the data for that year). Duplicate sites
were defined where the distance between sites were recorded as 0 metres. In such instances,
the point with the largest infested area (ha) was retained for that location while the other
point was deleted. Where attribute data were the same, the overlying point was deleted.
A separate dataset of infestation treatment data, including single tree treatments and
harvesting of infested blocks, for 2005 and 2006 was procured from licensees. Polygonal
and point treatment data were combined into a single point dataset using methods similar to
those used for infestation data, such that all points had an affected area associated with them.
Polygons were converted to points based on the centroid of the polygon. A pixel surface of
distance from treatment location within the research area was produced using the Spatial
Analysis distance tool in Arcview 9.2 (ESRI). Each pixel measured 73x73 metres, or
approximately 0.53 ha in size.
3.3.2: Preparation of Landscape Feature Datasets
A stand susceptibility index (SSI) dataset, produced by the Canadian Forest Service,
was used as a surrogate for habitat considered suitable for outbreaking populations of
mountain pine beetle in our analyses. The index is typically calculated using four variables,
including the relative abundance of pine, the ages of dominant and codominant pine, stand

49

density, and location (Shore and Safranyik 1992; Shore et al. 2000). We used an updated SSI
dataset that substituted a climatic factor for the location factor in calculation of the index,
which provided an assessment of habitat suitable for insect colonization and reproduction
(Safranyik et al. 1975; Carroll et al. 2004). This climatic factor predicts the influence of
realized climatic regimes on the susceptibility of forest stands by integrating daily weather
data from meteorological stations in British Columbia with historic temperature regimes into
a mechanistic model of mountain pine beetle population dynamics (Shore and Safranyik
1992; Carroll et al. 2004). Resulting stand susceptibility indices range between 0 and 100,
and are categorized into five stand susceptibility classes with class 0 being the lowest
susceptibility to infestation by mountain pine beetle and class 5 being the highest
respectively. The highest stand susceptibility classes identify areas where there are suitable
hosts (i.e., ages > 60 yr, species composition, diameter, etc.) and the climate is typically
amenable to the establishment and reproduction of mountain pine beetle populations
(Safranyik et al. 1975; Carroll et al. 2004).
The research area was classified into ten classes of landscape features that potentially
influence wind and circulation patterns within the region (Table 3.2).
Each 73x73m pixel within the research area was classified into a feature type based on its
relative elevation using a Topographic Position Index (TPI) tool v. 1.3a (Weiss 2001; Jenness
2006) as an extension within Arcview v.3.2 (ESRI). In brief, pixels are considered relative to
the mean of a predefined neighbourhood of a user-specified size and shape (Fig. 3.2 A). The
combination of TPI at small and large scales allows for predefined landscape features to be
delineated in a digital elevation model (Fig. 3.2 B). The digital elevation model used for the
TPI classification process was composed of 1:250,000 maptiles identical to raster sizes in our

50

other datasets (i.e. 73x73m). The recommended circular neighbourhood was used during
classification (Jenness 2006). Three combinations of small and large neighbourhood sizes
were examined, including 1000 and 2000 metres, 1000 and 3000 metres, and 1000 and 6000
metres, respectively. Resulting TPI rasters were visually inspected and compared to the
source digital elevation model. The first two TPI rasters resulted in a classification with a
pixel resolution too coarse to accurately reflect the topography of the research area. The
third classification scheme satisfactorily represented terrain within the region, as judged by
visual inspection, and was selected for use in statistical analyses.
Using this map of landscape features classified across the study area, the orientation
of certain landscape features was calculated to allow further investigation of the effects of
feature orientation on the occurrence of mountain pine beetle infestations.
The orientation of landscape features was determined according to the type of feature (Table
3.2). Azimuth is defined as the cardinal direction in which a linear feature lies. For example,
within the research area, canyons or deeply incised streambeds and U-shaped valleys may be
characterized by their azimuths (Fig. 3.3 A). Similarly, aspect is defined as the direction in
which non-linear features, such as open slopes or midslope ridges, face (Fig. 3.3 B). The
linear azimuth for linear landscape features with predominantly linear orientations such as
canyons or deeply incised streams, U-shaped valleys, and mountain tops or high ridges was
determined by "breaking" features at changes in direction greater than 45 degrees. The
orientation of the broken feature was then determined using an azimuth measurement tool (v.
1.6) in Arcview v.3.2 (Jenness 2005). The azimuths were then converted from compass
bearing measurements (i.e., angular readings) to one of the following four categories: northsouth, northeast-southwest, west-east, or southeast-northwest orientation. For non-linear

51

features, including open slopes, local ridges or hills in valleys, or midslope ridges or small
hills in plains, aspects of the features were extracted by querying features based on one of
eight directions (i.e., northwest, west, southwest, etc.). All datasets were then converted to
binomial rasters (73 x 73m pixel size) where locations of a particular landscape feature with a
specific orientation were noted by ones (i.e., presence) while the rest of the landscape was
recorded as zeros (i.e., absence).
3.3.3: Statistical Analyses
As an initial step in data exploration, we tested whether the distribution of susceptible
habitat, as well as the types of landscape features, affected the occurrence and distribution of
infestations of mountain pine beetle using contingency tests for the 2004, 2005, and 2006
datasets. For each year, the spatial extent under analysis was restricted to the extent of area
infested in that year plus a 25 kilometre buffer. As a result, the areas considered within each
analysis increased in size from 2004 to 2006 as mountain pine beetle spread across the region
(see Results). For contingency tests, observed values were calculated by summing the area
infested per stand susceptibility class or landscape feature while expected values were
calculated by multiplying proportion of total area occupied by a susceptibility class or
landform type by the total area infested by mountain pine beetle in the research area (see
Appendix B and Appendix C for R code). Boot-strapping procedures with 1000 repetitions
were then used to develop a test statistic distribution, as simple £ tests fail to contend with
spatial autcorrelation within the dataset. This process involved randomizing the spatial
points within the area of analysis, with the sample size equal to that of the dataset under
consideration. The susceptibility class or feature for the location of each point was then
sampled. The area of infestation for each point was also randomly permutated and the

52

contingency test statistic was then calculated for each randomization. The calculated test
statistic for the initial test was then ranked relative to the generated test statistic distribution
to determine if it was significant using a= 0.05 (see Appendix D for R code).
Spatial point process regression models were used to examine the potential influences
of landscape features, their orientation, habitat suitability, elevation, distance from source
infestations, and, for 2005 and 2006, the distances from locations of treatment in the previous
year on the density of outbreaking populations of mountain pine beetle in the study area for
each year (see Appendix E). Candidate landscape feature covariates for inclusion in these
models were restricted to those with an area infested greater than expected (from exploratory
data analysis above). Rather than considering all habitat suitability classes, only those with
the highest susceptibility (i.e. class 5) were included. For our spatial regressions, a spatial
point process (i.e., a spatially-explicit dataset of x and >> locations of infestations of mountain
pine beetle) was examined as a function of other spatially-explicit covariates (such as
presence/absence of management activity, landscape features, orientation, etc.), yielding
regression coefficient estimates and standard errors for all covariates. Substituting covariate
values into the resulting spatial point process regression equation yields an estimated density
X, a spatially-explicit density of infestations per unit area (i.e., pixel of 73><73m in size, just
over one half a hectare resolution). In this class of models, the likelihood is well-defined,
allowing examination of the significance of individual regression coefficients via likelihood
ratio tests among nested models using %2 test statistic distributions (Baddeley and Turner
2008). Moreover, each model yields AIC values useful for model selection, with lower
values judged to be the best (Akaike 1973).

53

Spatial point process regression models share similar challenges in evaluating the
effects of multiple variables as classic linear regression models in that the marginal effects of
highly correlated variables may be difficult to estimate (e.g. inflated standard errors)
(Graham 2003). In our system, due to the settlement patterns of mountain pine beetle in a
general northwest-southeast pattern across the entire research area, there was an extremely
high degree of correlation between x and y variables when examining the effects of distance
from source populations on final establishment densities (see Results). To contend with this
challenge, we reduced spatial location to one dimension by defining a new variable, the
distance from a line of reference along the Rocky Mountains, perpendicular to the direction
of spread. The direction of spread was determined by examining slope of a linear regression
of northing versus easting values (i.e. y versus x) for the first year infestation. The direction
of spread in 2004 was estimated to be northeast or along a 48° bearing. Therefore, the line of
reference had a 138° bearing. Hence, insect locations were measured as a distance from this
line along the western edge of the research area (presumed source of the insects) and their
established locations in the Peace River region of British Columbia.
Our spatial regressions yielded a multiple regression equation for each year (i.e.,
2004, 2005, and 2006) providing inference on the influences of a variety of variables (e.g.,
landscape features, elevation, distance from line of initiation, etc.) on the locations of
mountain pine beetle populations. We then returned to the initial inventory datasets and, for
each year, fit simple linear regressions using one of the eight (for aspect) or four (for
azimuth) landscape feature orientations for each type of landscape feature that was
significant in the previous models. The best-fitting feature orientation was selected for each
feature type in each year by examining AIC values, and then substituted into the previous

54

spatial regression models in place of the overall feature class (see Results). This allowed a
comparison between models with landscape features, and those with features restricted to a
specific best-fitting direction. The model with the lowest AIC value was judged to fit the
best.
To determine if elevations of infestations within the study area each year were similar
to the average elevation of infestations within the central interior (i.e. 1000 m; Safranyik and
Carroll 2006), we used one thousand (1,000) single sample Mests of randomly thinned
subsets of the annual datasets (n = 50). Resampled subsets were used to contend with spatial
autocorrelation in the original dataset. All data was handled in Arcview v.9.2, Arclnfo
Workstation, and Arcview v.3.2 while the spatstat package v. 1.14-9 within R v.2.8.1 was
used for statistical analyses (Ihaka and Gentleman 1996; R Development Core Team 2009).
3.4: Results
3.4.1: Spatial Extent and Landscape of Invaded Area
The number of incipient-epidemic infestations increased annually within the Peace
River region of British Columbia between 2004 and 2006. There were 10,536 infestations
mapped in 2004 and 12,275 in 2005. These numbers approximately tripled to 35,084
infestations in 2006 (Fig. 3.4). In 2004, the mean area of patches of red trees killed by
mountain pine beetle was quite small; approximately one third of a hectare. However, the
mean sizes of individual infestations increased to approximately 1 ha in 2006 (Table 3.3).
Sizes of patches of dead trees became more variable due to expanding areas that were
colonized, as the largest areas measured in 2006 were 2,440 ha, versus 6 and 133 ha in 2004
and 2005 respectively. By 2006, the total area of trees killed by mountain pine beetle had
expanded to 35,084 ha (Fig. 3.4).

55

Patches of mortality were found in lodgepole pine forests between elevations of
approximately 500-1700 m. After the initial invasion event, most successful attacks were at
1,000 m in elevation each year (Table 3.3). In 2005 and 2006, patches of dead trees were
found at significantly higher elevations than those previously noted in outbreaks in the
insect's conventional range. Infestations stretched from the western slopes of the Rocky
Mountains in 2004 to the town of Tumbler Ridge, British Columbia (Fig. 3.5). By 2006, this
extent had expanded two-fold in a northeasterly direction to encompass almost the entire
Peace River region.
Dominant landscape features in the invaded region included plains, open slopes, and
U-shaped valleys (Fig. 3.6). A numerical summary of their areas and elevations are provided
in Table 3.4. Together, plains and open slopes comprised more than 50% of the landscape in
the Peace River region. Other landscape features such as canyons and deeply incised
streams, upper slopes and mesas, and midslope ridges or small hills in plains each comprised
no more than 6% of the area proportionally, but absolute area was always more than 100,00
ha in size. The smallest landscape features in size were local ridges or hills in valley
bottoms, comprising 1,122 ha across the region (0.04% of the total area). Plains exhibited
the lowest mean elevation (877 m) while mountaintops/high ridges were found to have the
highest (1,619 m; Table 3.4).
A map of habitat susceptibility classes is provided in Fig. 3.7, with a numerical
summary of their relative abundances across the landscape in Table 3.5. Most of the Peace
River region across northeastern BC was considered highly unsuitable for outbreaking
populations of mountain pine beetle, as two thirds of the landscape was classified as
suitability class 0. The most susceptible habitat, i.e. class 5, occupied only 3% of the

56

landscape (74,894 ha; Table 3.5). This habitat is stratified across nine different landscape
features (Table 3.6). Of these, plains, open slopes, and U-shaped valleys comprised the
largest proportion of highly susceptible habitat (49, 27, and 14% respectively). Upland
drainages or headwaters, local ridges or hills in valleys, and upper slopes or mesas
collectively comprised less than 1% of habitat classified as highly susceptible to outbreaking
populations of mountain pine beetle (Table 3.6).
3.4.2: Which Landscape Features are Associated with Mountain Pine Beetle
Infestations?
Infestations of mountain pine beetle were not distributed uniformly across all types of
landscape features, as demonstrated by goodness-of-fit tests (Appendix F). Exploratory
analyses identified that certain landscape feature had more (or less) infestation than would be
expected if infestations were distributed uniformly across the landscape. In 2004, canyons or
deeply incised streams, local ridges or hills in valleys, midslope ridges or small hills in plains
and U-shaped valleys had more area infested than expected. This pattern persisted into 2005,
as midslope ridges or small hills in plains, and U-shaped valleys had greater areas infested
than expected, along with canyons or deeply incised streams. In 2006, canyons or deeply
incised streams, local ridges or hills in valleys, midslope ridges or small hills among plains,
open slopes and U-shaped valleys exhibited a greater total area infested by mountain pine
beetle than expected. Although open slopes and plains occupied a large proportion of the
landscape, they had proportionally less area infested than would be expected from uniform
settlement in 2004-2005. Landscape features including mountain tops or high ridges,
midslope ridges or small hills in plains, upland drainages or headwaters, and upper slopes or

57

mesas each contained fewer infestations than would be expected by a uniform distribution
across the landscape in all three years (Appendix F).
Given the differences in relative distribution of infestations of mountain pine beetle
across landscape features, we fit spatial point process regression models to determine which
landscape features might provide the best inference on locations of detected infestations
(Table 3.7). These models, which measure intensity, A, as a spatially-explicit number of
infestation per unit area (m2), also included as covariates elevation, distance from source
location, presence or absence of the most highly suitable habitat (i.e. class 5; see Fig. 3.7),
and, in the latter two years, distance from the nearest sites where insect control tactics, such
as fall and burn or single-tree removal, had been implemented the previous year. Landscape
features associated with increased intensities of infestation included canyons or deeply
incised streams in 2004, local ridges or hills in valleys in 2005, midslope ridges or small hills
in plains in 2004 and 2006, open slopes in 2006, and U-shaped valleys in all three years
(Table 3.7). In all three years, the presence of highly susceptible habitat was also positively
correlated with infestation intensity. In 2004, canyons or deeply incised streams were
associated with lower densities of mountain pine beetle
After accounting for the effects of landscape features, we found, not surprisingly, that
the intensity of infestations decreased with distance from a line of initiation along the Rocky
Mountains (see Methods). Furthermore, there were fewer patches of beetle killed trees at
higher elevations after taking the effects of landscape features and distance from the Rocky
Mountains into account. Beetle control tactics appeared to be efficacious. The density of
infestations decreased with increasing distance from treatment sites in the previous year for
both 2005 and 2006 (Table 3.7).

58

3.4.3: Does Feature Orientation Influence Invasion Success?
To determine which azimuths and aspects of landscape features had the greatest
influence on intensity of patches of trees killed by mountain pine beetle infestations, spatial
point process models were fit to annual inventory data with each of eight cardinal
orientations per landscape feature as single covariates (Table 3.8). In the initial year of
infestation, of the U-shaped valleys, those with a northeast-southwest azimuth best fit the
pattern of infestation. Models of best fit for that year also included west-facing local ridges
or hills in valleys, or midslope ridges or small hills in plains with southwestern aspects.
Canyons or deeply incised streams with a north south azimuth and U-shaped valleys with a
northeast southwest azimuth best fit patterns of infestation in years following initial
establishment, 2005 and 2006. For 2005, a model incorporating southern aspects of local
ridges or hills in valleys best fit patterns of infestation, while midslope ridges or small hills in
plains with southwestern aspects and open slopes with southwestern aspects explained the
most variation in insect establishment on the landscape in 2006.
Once the most-significant azimuth or aspect of each landscape feature was identified
(Table 3.8), we substituted these best terms into the models of Table 3.7; i.e., replacing each
significant global landscape feature with a more targeted subset of that feature class, to see if
the orientation of the landscape feature provided better inference on the establishment
patterns of mountain pine beetle throughout the Peace River region than simply the locations
of features themselves. In 2004 and 2005, incorporating the best orientation of the landscape
feature did not improve models that previously incorporated all orientations, as indicated by
higher AIC values (2004: 311,275 vs. 311,338; 2005: 355,544 vs. 355,774; Table 3.7 and
Table 3.9). However, persistence of tree killing populations of D. ponderosae in the Peace

59

River region in 2006 was readily influenced by orientation of landscape features as indicated
by a lower AIC value (1,011,283 vs. 1,011,290) when feature orientations replaced landscape
features as covariates (Table 3.9). Interestingly, a southwest orientation was common for all
three landscape features included in the 2006 model (Table 3.9). For 2006, southwestern
slopes of midslope ridges or small hills in plains, southwest facing open slopes, and Ushaped valleys that run in a northeast-southwest cardinal direction positively influenced
intensity. As previously established by initial models lacking feature orientation, elevation
and distance from the western edge of research area were inversely correlated to intensity of
D. ponderosae infestations while distance from highly susceptible habitat was positively
correlated. The intensity of infestations decreased with increasing distance from treatment in
the previous year.
3.5: Discussion
Our results demonstrate not only that specific landscape features are associated with
establishment of an invading organism, but orientation of those landscape features is
associated with the organisms potential persistence and spread. Initially, as mountain pine
beetle crossed the Rocky Mountains into new areas of the province of British Columbia in
western Canada, the majority of pockets of invaders were found in valleys in the central
Peace region - which could have been acting as conduits for further dispersal - in the first two
years of detection. This is consistent with the hypothesis that initial invasion of this territory
occurred via aeolian dispersal (Chapter 2; Safranyik and Carroll 2006; Robertson et al.
2009). By 2006, increased densities of infestations of mountain pine beetle were associated
with mid-slope ridges, small hills, and open slopes that were primarily facing in a southwest
direction.

60

There are at least three reasons for the emergence of the importance of southwest
feature orientation in the persistence of these invading herbivores. First, hosts may be more
susceptible on drier, sun-exposed south-west facing slopes (Powers et al. 1999). Exudation
of oleoresin, important in tree defense, is lowered when conifers are water-stressed (Waring
and Pitman 1985). Second, insects established on southwest-facing sites may enjoy higher
reproductive rates due to higher ambient temperatures, than insects on shaded slopes
(Mattson and Haack 1987). Development of mountain pine beetle is strongly temperaturedependent, with progression between life stages critically dependent on accruing a sufficient
number of heat units (Reid 1962; Amman 1973; Safranyik 1978; Bentz et al. 1991; Shore et
al. 2000; Safranyik and Carroll 2006; Powell and Bentz 2009). Moreover, insect populations
on landscape features oriented to receive less sun exposure would be more affected by cold
temperatures during the winter. The cold tolerance of the over-wintering larval stage of
mountain pine beetle increases incrementally over the winter as temperatures get
progressively colder (Bentz and Mullins 1999). However, lethal temperatures, below -40°C,
can cause widespread brood mortality if they occur in October or mid-March, or for extended
periods during the winter (Wygant 1940). Increased mortality and asynchronicity of
development and emergence within populations associated with feature orientations
receiving less sun would exhibit decreased local dispersal and likely lower infestation levels
on those landscape features in successive years.
A third explanation for increased associations of mountain pine beetle with
southwestern slopes, not mutually exclusive to the first two, is that additional episodes of
aeolian dispersal may have deposited higher numbers in the Peace River region on
southwestern slopes, as wind is strongly unidirectional and blows from the west and

61

southwest (Fig. 3.8). Mountain pine beetle tend to accumulate on windward sides of barriers,
as large insects (i.e. those with bodies greater than a 4 or 5 mm2 surface area, like mountain
pine beetle) are less likely to be carried over top of a barrier to enter circulating leeward
eddies because of their great inertia (Lewis and Dibley 1970). Hills that are relatively close
to each other will also increase the amount of deposition on the windward slope of
consecutive hills (Goossens 1996). The topography in the western portion of the Peace River
region is characterized by consecutive ridges and valleys, causing ridgelines downwind to
sustain more insect deposition on windward, southwestern slopes.
Thermal development thresholds linked to differing brood successes on southern
versus northern aspects may be similarly associated with declining population densities at
higher elevations, where temperatures are cooler (Amman 1973; Tishmack et al. 2005;
Brunelle et al. 2008). High elevation sites within British Columbia and the Peace River
region generally have lower mean winter and summer temperatures and a shorter growing
season than low elevation sites. At higher elevations, populations may be particularly
vulnerable to colder temperatures during August and September, forcing them into bivoltine
developmental stages (Amman 1973). For example, Amman (1973) examined the life
history of mountain pine beetle in lodgepole pine between 1923 and 2750 metres elevation in
northwestern Wyoming. Between 1923 and 2130 metres the insects were univoltine, while at
2450 metres, part of the population was bivoltine. At elevations between 2573 and 2750
metres, two years were required to complete one entire life cycle. Asynchronous or mixed
voltinism at high elevations also affects emergence synchronicity critical to the mountain
pine beetles' ability to mass attack and successfully colonize hosts adjacent to their natal
hosts (Amman 1969).

62

Our research area was between approximately 54 and 56° N. Despite the well-known
negative correlation between insect success and elevation, we note that the mean elevation of
outbreaking populations was significantly higher than 1000 metres in 2005 and 2006. In the
insect's historic range within British Columbia, the majority of mountain pine beetle
outbreaks have occurred historically at mean elevations of 1000 metres at 55° N, the same
latitude as our study area (Taylor et al. 2006). The occurrence of more infestations at
elevations higher than those typically exhibited in the insect's native range suggests three
possibilities. First, it could signify that the climate in this newly invaded territory is suitable
for establishment and brood production. Insects are particularly limited by temperatures that
influence survival and reproduction at the northern edges of their habitat (Crozier 2004).
Although dispersal is exceptionally lethal to bark beetles (Schmid 1969; Raffa 2001),
warming of winter temperatures at the northeastern edge of mountain pine beetle's range
may have facilitated increased survival of pioneer dispersers, allowing the insect to thrive in
this new habitat (Carroll et al. 2004). Similar phenomena have been noted in other systems,
such as A talopedes campestris Boisduval (skipper butterfly) (Crozier 2004) and
Thaumetopoeapityocampa Schiff. (pine processionary moth) (Battisti et al. 2005). Second,
increased persistence at higher elevations could signify that host resistance is low in these
areas, since, to our knowledge, populations of lodgepole pine in these areas have not been
previously exposed to outbreaking populations of bark beetles. Of special concern is the
threat of expansion into jack pine, which could provide this eruptive herbivore a conduit to
the east coast of North America (Cerezke 1995). A third explanation, also not mutually
exclusive to the first two, is simply that the persistence of outbreaking populations at higher
elevation sites may reflect prior management activities at easy-to-access areas at lower

63

elevation. However, such activities was accounted for by its own covariate in the models, so
any such signal is in addition to those already noted (Table 3.9).
In our study, as in others (e.g., Preisler and Mitchell 1993; Bentz and Munson 2000;
Fettig et al. 2006; Nelson et al. 2006; Fettig et al. 2007; Trzcinski and Reid 2008)
anthropogenic interventions against tree-killing eruptive herbivores decreased occurrences of
infestations in the year following treatment. Management activities included single tree
treatments using fall and burn, and harvesting of stands infested with mountain pine beetle.
These control tactics decrease the potential for short-distance dispersal to adjacent stands
(Trzcinski and Reid 2008). Treatments may have been efficacious in this region due to
relatively low rates of increase among populations as a whole, either due to temperature or
high levels of competition. Attacked hosts within the research area were noted to have
greater than optimal levels of attack (i.e., 62 attacks/m2; Raffa and Berryman 1983; Allan
Carroll pers. comm.1) resulting in lower per capita reproductive rates due to intraspecific
competition (Safranyik and Carroll 2006). The persistence of a statistical signal for
management activities for 2005 and 2006 also indicates that there were fewer, if any, long
distance inputs of additional beetles into the study region, as the statistical signal for
management activities associated with short-distance dispersal would likely have been
obscured. Aeolian dispersal and rapid range expansion associated with insect outbreaks may
be sporadic, but important events.
In summary, our results indicate that, following the initial invasion event, the
continuing spread of the mountain pine beetle may potentially be facilitated by large, lowelevation valleys orientated along the dominant wind direction, acting as conduits of suitable
habitat for the insect (e.g. Robertson et al. 2009). The orientation of landscape features also
'Allan Carroll Associate Professor, University of British Columbia (allan.carroll@ubc.ca)
64

affects the establishment of mountain pine beetle, either by influencing the susceptibility of
hosts or by allowing populations on warmer and drier sites to have greater reproductive rates
facilitating spread into adjacent stands. These results may be of particular value in
formulating management strategies as the insect progress eastward through the boreal for of
Canada, or in other areas where the insect is at epidemic population phases, such as regions
of the southern Rocky Mountains of Colorado.
A result of this study of particular concern is the successful establishment of
infestations at higher elevations than within their native range in the central interior of British
Columbia. The elevation shift of mountain pine beetle indicates that the host within the
region may be highly susceptible to attacks, or that the climate of such areas may be
conducive to the establishment of infestations at those elevations (Carroll et al. 2004).
Finally, while control measures such as single-tree fall and burn and harvesting of infested
blocks are effective in minimizing dispersal between adjacent stands, it is not likely to be
effective in preventing continuing spread where inputs of insects dispersing long-distances
are present. In these situations, alternate control tactics will need to be used to effectively
deal with long-distance inputs. In particular, predicting the potential spread of new
infestations based on landscape features and their orientation, in concert with new landscapescale spread modeling approaches (e.g., Gamarra and He 2008; Zheng et al. 2009) may
provide newer coarse-scale tools for directing control activities.
3.6: Acknowledgements
Funding for this study was provided by the Mountain Pine Beetle Initiative of Natural
Resources Canada, Canadian Forest Service, NSERC Discovery, and Pacific Forestry Centre
Graduate Student and Association of Professional Biologists Scholarships awarded to HMG.

65

We thank B. Pate of West Fraser Timber Ltd. and C. Johnson of Canadian Forest Products
Ltd. for providing advice and compiled aerial survey data within the region. G. Thandi,
Canadian Forest Service, Pacific Forestry Centre, A. McGill, Alberta Sustainable Resource
Development, and P. Bai, University of Northern British Columbia provided GIS assistance.
T. Shore and B. Riel, Canadian Forest Service, Pacific Forestry Centre provided stand
susceptibility data. A. Baddeley, School of Mathematics and Statistics, University of
Western Australia and J. Zhu, Department of Statistics, University of Wisconsin-Madison,
provided advice on implementation of spatial point process models. J. Jenness, Jenness
Enterprises Ltd. provided the Topographic Position Index tool v. 1.3a for Arcview 3.2. The
authors also thank the assistance and support of Marnie Duthie-Holt, Matthew Klingenberg,
Fraser McKee, Sandra Allen, Sonia Wesche, and Doug Linton.

66

3.7: Figure Captions
Figure 3.1 Location of research area in the northeastern British Columbia. The research area
includes a portion of the Northern Rocky Mountains, including the Hart Ranges, and
the Rocky Mountain Foothills, an area of undulate topography that eventually
becomes a plateau at the eastern edge of the research area. The inset map shows the
location of the research area relative to the rest of British Columbia.
Figure 3.2 A) Determination of slope position using Topographic Position Index (adapted
from Weiss 2001). B) Determination of general landform using Topographic Position
Index (adapted from Jenness 2006).
Figure 3.3 Examples of (A) landscape features of linear canyons or deeply incised
streambeds which are generally linear in nature and whose cardinal direction (i.e.,
feature orientation) is measured using azimuth. (B) Landscape features, such as
midslope ridges or small hills in plains which are generally circular or complex in
shape and whose orientation is measured compartmentally using azimuth.
Figure 3.4 Area and number of infestations of mountain pine beetle in the Peace River region
of British Columbia in 2004, 2005, and 2006 as determined by aerial surveys of
patches of red trees.
Figure 3.5 Location of infestations of mountain pine beetle in the Peace River region of
British Columbia in 2004, 2005, and 2006. Dashed line in the three figures represents
the line of initiation which is parallel to the Rocky Mountains and perpendicular to
the direction of spread in 2004.
Figure 3.6 Map of landscape features within the Peace River region of British Columbia.

67

Figure 3.7 Distribution of habitat susceptibility classes in the Peace River region of British
Columbia.
Figure 3.8 A general wind rose of the Peace River region of British Columbia (produced by
Matthew Lynn of Garrad Hassan Canada Inc. with data from the Canadian Wind
Energy Atlas)

68

Fig. 3.1

A

A

ROCKY
MOUNTAIN

NORTHERN
*

FOOTHILLS

ROCKY
MOUNTAINS
•

'

•

•

'

>

•

.

'

*?
'

s
*

— i

-

.

'

•

^

'

l
rjL'Vcili(j:i (::it'lrt*s)

C^

W

|
!

382

I
Kilinm-i!ri!K

69

j>

3.2

Elevation at point (pt)
Mean elevation of
neighbourhood (u)

pt ~ u and slope ~ 0 = tpi ~ 0 (flat area)
Mean elevation of
neighbourhood (u)
and elevation at
point (pt)

SN: Very low
LN: Very high

SN: Very high
LN: Very low

70

SN: Mid-range
LN: Midrange, steep
slope

SN: Midrange
LN: High

very low
LN: Very low

SN:

Fig. 3.3

S"^!
Cf'

ft

^"

} ' •

"X.-1

!

,—' '

^ '

p'r-

.-

'

>'

• •

(•-••

y - . ;

),y

'

^:^X^y'y-'^}-^7-•••-•

< ^ . ' 1 -L "» ->

" >

V \

>7

•• -

«• .

i

V- ••••'' "

\

f;/"--'.^:
~ . j - v -

Xk*

V'. •••!

•*—).,• A \ i

~

rr

'

'4-vV';X^ ,

71

B

;•

'

1

"• s

%\'*\ik

A

S

• - — - -

. • •

l

•.

• • •• •. -

u^

;..^,,

'

\^

Fig. 3.4

40

•—
—

Number of Infestations (thousands)
Area Infested (thousands; ha)

30

20 h

10 h

2004

2005
Year

72

2006

Fig. 3.5

A;

2004

2005

'--.

x

•
Tumbler Ridge

Infcstatiouh
Initation Line
Research Area
ion

1

1

Tumbler Ridel's... -

* L.- '*

0

•

Kilometers

73

•

2006
*
Tumbler Ridge

no

Kiloim-lrcs

••'"- ":,..'• J._/

Canyons of de^ph imised si renins
MuMopf dnuiia^t,^ ov shallow vallrys
Upland drainages of headwaters
1'-Ah a pod va]3e vs
Plaials
(>peu slopes
i'pper slopes or amesns
Local ridges «r ['tills in valleys
ftlkMope rulyrs or smn33 hjils in plajcis
Moiirttahti lops of hig,3i ridges

74

Fig. 3.7

A

Susceptibility Class
0

3

1

4

o :
"~ i

r

o
L

'^

3
I

Kilometres

75

Fig 3.8

i'

*

0-3

3-6

76

6-9

•10%

'20%

I

I
>9mJs

30%

• 40%

Table 3.1 Data fields of annual datasets inventorying patches of dead trees killed by
mountain pine beetle within the Peace River region of British Columbia, 2004-2006.
Field Name
Description
Mapnum
British Columbia Geographic System Mapsheet Number
Numtrees
Number of trees attacked
Spot_area
Area of attack in hectares
Comments
If there was green, red or grey attack in the site
Source
Source of original data
BECndu
BEC zone, subzone, and variant
Easting
UTM easting
Northing
UTM northing
Park
Is the site within a park? 1 = yes, 0 = no
Ground
Was a ground survey completed? 1 = yes, 0 = no
BMU
Beetle Management Unit
Surveyed
Number of trees surveyed for mountain pine beetle attack within the site
Burned
Number of infested trees removed using fell and burn control treatment
GRratio
Ratio of red attacked trees to unattacked green trees

77

Table 3.2 Classes of landscape features in the Peace River region of British Columbia.
Landscape Features
Determination of Orientation
Canyons or deeply incised streams
Azimuth
Midslope drainages or shallow valleys
Aspect
Upland drainages or headwaters
Not applicable
U-shaped valleys
Azimuth
Plains
Not applicable
Open slopes
Azimuth
Upper slopes or mesas
Not applicable
Local ridges or hills in valleys
Aspect
Midslope ridges or small hills in plains
Aspect
Mountain tops or high ridges
Not applicable

78

Table 3.3 Summary statistics for infestation sizes and their characteristics of their elevations
from annual inventories of mountain pine beetle infestations for 2004, 2005, and 2006 in the
Peace River region of British Columbia. Results for /-tests examining the difference between
average elevation for central British Columbia (1000m; Safranyik and Carroll 2006)
conducted by 50 randomly sampled points in each annual dataset within the research area are
reported as the percentage of 100 reiterations that were significant.
Infestation Sizes (hal
Elevation Summary Stats (m)
Min
Mean
SD. Max
Min Mean SD. Max

Different than historical average
t49
Significant T-tests (% of 100)

2004

0.001

036

034

6

621 1,002 162 1,652

92

100

2005

0.23

0.33

1.43

133

566 1,045 171 1,705

233.9

100

2006

0.001

1.07

17.33 2,440

462 1,049 175 1,686

34.3

100

Mean 0.001

0?78

13.51 860

462 1,040 173 1,705

-

-

79

Table 3.4 Average size and elevation of landscape features in Peace River region of British
Columbia (see Fig. 3.6 for map).
Landform

Area (ha)

1,161,849
Plains
Open slopes
836,946
U-shaped valleys
237,761
Mountain tops/High ridges
168,347
Upper slopes/Mesas
159,454
Midslope drainages/Shallow valleys 158,742
Canyons/Deeply incised streams
122,853
Midslope ridges/Small hills in plains; 116,685
Upland drainages/Headwaters
13,852
Local ridges/Hills in valleys
1,122
2,977,611
Total

Percentage
of Landscape
39.02
28.11
7.98
5.65
5.36
5.33
4.13
3.92
0.47
0.04
100

80

Mean
877
1,091
964
1,619
1,468
1,238
1,007
1,219
1,608
1,119
1,059

Elevation (m)
SD Min.
Max.
192 396
1,988
282 395
2,142
255 382
1,806
277 764
2,808
257 726
2,491
309 396
2,138
256 384
1,817
280 477
2,121
212 860
2,332
202 662
1,759
325 382
2,808

Range
1,592
1,747
1,424
2,044
1,765
1,742
1,433
1,644
1,472
1,097
2,426

Table 3.5 Numerical summary of "habitat susceptibility classes" across the Peace River
region of British Columbia. See Methods for details of classification scheme.
Habitat Susceptibility
Percent
Area (ha)
Class3
of Landscape
0
66
1,958,090
1
15
437,895
2
5
163,763
3
5
163,292
4
6
179,675
5
3
74,894
a
Class 0 is considered least susceptible to outbreaks of D. ponderosae

81

Table 3.6 Percentages of landscape features and total landscape with habitat considered
highly susceptible to outbreaking populations of mountain pine beetle in the Peace River
region of British Columbia (habitat susceptibility class 5).
Landscape Feature

Relative
Distribution (%)

Percent of
Landform Type

Percent of
Landscape

Plains

48.7488

3.142

1.22616

36,510.1

Open slopes

27.3921

2.451

0.68898

20,515.2

U-shaped valleys

14.3966

4.535

0.36211

10,782.3

Midslope ridges, small hills in plains 3.4471

2.213

0.08670

2,581.7

Canyons, deeply incised streams

2.7594

1.682

0.06941

2,066.6

Midslope drainages, shallow valleys

1.9672

0.928

0.04948

1,473.3

Upper slopes, mesas

0.6283

0.295

0.01580

470.6

Mountain tops, high ridges

0.6061

0.270

0.01525

454.0

Local ridges, hills in valleys

0.0537

3.582

0.00135

40.2

Upland drainages, headwaters

0.0007

0.004

0.00002

0.5

100.0000

19.000

3.00000

74,895.0

Total

82

Area (ha)

Table 3.7 Spatial point process regression estimating the intensity of discrete beetle
infestations as a function of landscape features, topography, distance from a source line,
habitat susceptibility, and distance from control tactics in the previous year, 2004-2006, in
the Peace River region of British Columbia. Log(/l) is the response variable for each model,
where X is the density of sites successfully attacked by mountain pine beetle populations per
square meter. For example, in 2004, the estimated density of beetle patches at sites 5,000
kilometres from the source location in a canyon at 1,500 metres elevation and in highly
susceptible habitat would be exp(-™2+5ooo*-5.dW-o.o296+i5oo><-3.85E-o3+l.55) o r 3 4E _ Q6
infestations per square meter (i.e., 0.034 infestations per hectare or one patch of tree-killing
beetles every 30ha).
Model Parameters
Intercept
Distance from source location
Canyons, deeply incised streams
Midslope ridges, small hills in plains
U-shaped valleys
Elevation
Highly susceptible habitat

Estimate
-7.82E+00
-5.03E-05
-2.96E-01
5.91E-01
2.24E-01
-3.85E-03
1.55E+00

Intercept
Distance from source location
Canyons, deeply incised streams
Local ridges, hills in valleys
U-shaped valleys
Elevation
Highly susceptible habitat
Distance from 2004 control tactics

-8.76E+00
-3.96E-05
1.38E-01
8.22E-01
3.87E-01
-2.39E-03
8.87E-01
-9.91E-05

Standard Error Z
7.92E-02
-98.7
4.95E-07 -101.6
3.68E-02
-8.1
4.11E-02
14.4
2.69E-02
8.3
5.36E-05
-71.8
2.86E-02
54.1

P-value
O.0001
<0.0001
O.0001
<0.0001
O.0001
O.0001
O.0001

7.44E-02
5.85E-07
3.23E-02
2.01E-01
2.42E-02
4.58E-05
2.96E-02
1.46E-06

-117.7
-67.7
4.3
4.1
16
-52.1
30
-67.9

<0.0001
<0.0001
O.0001
O.0001
O.0001
<0.0001
O.0001
O.0001

4.22E-02
-1.16E+01
Intercept
Distance from source location
-5.44E-06
2.67E-07
Midslope ridges, small hills in plains
4.49E-01
2.42E-02
Open slopes
2.82E-01
1.23E-02
U-shaped valleys
2.04E-01
1.83E-02
Elevation
-1.18E-03
2.49E-05
Highly Susceptible Habitat
9.77E-01
1.90E-02
Distance from 2005 control tactics
-5.31E-05
7.79E-07
AIC values for final models: 2004, 311,275.3; 2005, 355,543.6; 2006, 1,011,290

-275.4
-20.4
18.6
22.9
11.2
-47.2
51.3
-68.2

O.0001
O.0001
O.0001
O.0001
O.0001
<0.0001
O.0001
O.0001

83

Table 3.8 Spatial point process regression models that identify which orientations of
landscape features best fit the distribution of D. ponderosae infestations in three initial years
of incipient-epidemic outbreaks (2004-2006) in the Peace River region of British Columbia.
2004

P-value
Azimuth/Aspect"
A1C
0.0034 334,079.70
Flat
North south
<0.0001 333,771.80
Northeast southwest <0.0001 333,967.30
West east
<0.0001 333,977.30
Northwest southeast O.0001 333,792.20
U-shaped valleys
Flat
O.0001 334,058.10
North south
O.0001 333,450.30
Northeast southwest 0
331,373.80
West east
O.0001 333,617.20
Northwest southeast 0
331,696.60
Local ridges, hills in valleys
East
0.0087 334,081.40
North
0.0097 334,081.60
Northeast
1.00
334,088.30
Northwest
0.00033 334,075.40
South
0.0047 334,080.30
Southeast
1.00
334,088.20
Southwest
0.41
334,087.60
West
<0.0001 334,069.40
Midslope ridges, small hills in plains
East
0.0066 334,081.00
North
0.064
334,084.90
Northeast
0.19
334,086.60
Northwest
1.00
334,088.20
South
<0.0001 334,025.60
Southeast
<0.0001 334,049.60
Southwest
O.0001 333,973.30
West
<0.0001 334,058.00
Open slopes
East
North
Northeast
Northwest
South
Southeast
Southwest
West
Landscape Feature
Canyons, deeply incised streams

2005
P-value
AIC
O.0001 385,438.70
O.0001 385,036.8
O.0001 385,251.40
<0.0001 385,377.20
O.0001 385,111.40
O.0001 385,427.90
<0.0001 384,065.30
O.0001 382,746.6
O.0001 384,223.50
O.0001 382,934.20
0.012
385,473.20
0.0013 385,469.30
1.00
385,479.50
0.00012 385,464.70
O.0001 385,458.2
1.00
385,479.30
0.47
385,479.00
0.036
385,475.20

2006
P-value
AIC

-

-

0.0050
O.0001
<0.0001
<0.0001
O.0001

1,028,061
1,027,880
1,026,668
1,027,645
1,027,950

-

-

0.059
1,028,065
0.38
1,028,068
1.00
1,028,068
<0.0001
1,028,043
O.0001
1,027,990
<0.0001
1,028,034
O.0001
1,027,893
<0.0001
1,028,039
O.0001
1,028,040
0.00024
1,028,055
0.0011
1,028,058
0.011
1,028,062
<0.0001
1,027,683
<0.0001
1,027,967
O.0001
1,027,523
O.0001
1,027,833
P-value reflects test for homogeneity of the point pattern (significance indicates heterogeneous intensity of
infestations across the region). Best model (i.e., lowest AIC value) for a given landscape feature in a given is
indicated by italics.
a
Canyons/deeply incised streambeds and U-shaped valleys area reported as azimuths, while remainder of
landscape feature directions reflect the aspect (see Table 3.2).

84

Table 3.9 Spatial point process regression models estimating intensity ofD. ponderosae
infestations replacing landscape feature with direction of landscape feature as a covariate.
Data is for the years 2004-2006 in the Peace River region of British Columbia. Other
variables are the same as in Table 3.7.
Year
2004

Model Parameters
Intercept
Distance from source location
Local ridges - West
Midslope ridges - Southwest
Valleys - Northeast-southwest
Elevation
Highly susceptible habitat

2005

Intercept
-8.17E+00
Distance from source location
-4.24E-05
Canyons/Deeply incised streams: North south 1.35E-01
Local ridges/Hills in valleys: South
9.94E-01
U-shaped Valleys: Northeast southwest
1.68E-01
Elevation
-2.70E-03
Highly susceptible habitat
9.09E-01
Distance from 2004 treatment
-1.01E-04

Estimate
Standard Error
Z
-7.97E+00
5.80E-02 -137.4
-4.99E-05
4.23E-07 -117.9
1.06E+00
4.47E-01
2.4
1.11E+00
8.66E-02
12.8
5.02E-01
2.92E-02
17.2
-3.72E-03
4.44E-05
-84.0
1.58E+00
2.84E-02
55.6

P-value
<0.0001
O.0001
0.0087
8.20E-38
1.33E-66
O.0001
O.0001

5.79E-02
5.31E-07
4.97E-02
3.78E-01
2.84E-02
3.97E-05
2.96E-02
1.46E-06

-141.1
-79.8
2.7
2.6
5.9
-67.9
30.7
-69

O.0001
O.0001
6.53E-03
8.60E-03
3.31E-09
<0.0001
2.55E-207
O.0001

Intercept
-1.15E+01
3.67E-02
Distance from source location
-6.04E-06
2.48E-07
5.28E-02
Midslope ridges/Small hills in plains: Southwest7.58E-01
Open slopes: Southwest
5.21E-01
2.33E-02
U-shaped Valleys: Northeast southwest
2.91E-01
2.31E-02
Elevation
-1.19E-03
2.31E-05
Highly Susceptible Habitat
9.67E-01
1.90E-02
Distance from 2005 treatment
-5.40E-05
7.80E-07
AIC values for final models: 2004, 311,337.8; 2005, 355,773.9; 2006, 1,011,283

-312.2
-24.4
14.4
22.4
12.6
-51.6
50.9
-69.3

O.0001
4.47E-131
8.81E-47
5.10E-111
3.47E-36
O.0001
O.0001
O.0001

2006

85

3.8: Literature Cited
Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle.
Pages 267-281 in B.N. Petrov and F. Csaki, editors. The 2 nd International Symposium
on Information Theory. Akademiai Kiado, Budapest.
Amman, G. D. 1969. Mountain pine beetle emergence in relation to depth of lodgepole pine
bark. USDA, Forest Service, Intermountain Forest and Range Experiment Station,
Ogden, Utah. Research Note INT-96.
Amman, G. D. 1973. Population changes of the mountain pine beetle in relation to elevation.
Environmental Entomology 2(4): 541-547.
Amman, G. D. and W. E. Cole. 1983. Mountain pine beetle dynamics in lodgepole pine
forests. Part II: population dynamics. USDA Forest Service, Intermountain Forest and
Range Experiment Station, Ogden, Utah. General Technical Report.
Antor, R. J. 1994. Arthropod Fallout on High Alpine Snow Patches of the Central Pyrenees,
Northeastern Spain. Arctic and Alpine Research 26(1): 72-76.
Ashmole, N. P., J. M. Nelson, M. R. Shaw and A. Garside. 1983. Insects and spiders on
snowfields in the Cairngorms, Scotland. Journal of Natural History 17(4): 599-613.
Aukema, B. H., A. L. Carroll, Y. Zheng, J. Zhu, K. F. Raffa, R. D. Moore, K. Stahl and S. W.
Taylor. 2008. Movement of outbreak populations of mountain pine beetle: influences
of spatiotemporal patterns and climate. Ecography 31(3): 348-358.
Aukema, B. H., A. L. Carroll, J. Zhu, K. F. Raffa, T. A. Sickley and S. W. Taylor. 2006.
Landscape level analysis of mountain pine beetle in British Columbia, Canada:
spatiotemporal development and spatial synchrony within the present outbreak.
Ecography 29(3): 427-441.
Baddeley, A. and R. Turner. 2008. Spatstat: A package for analyzing spatial point processes
in R. Perth, Australia, <http://www.spatstat.org/>.
Battisti, A., M. Stastny, S. Netherer, C. Robinet, A. Schopf, A. Roques and S. Larsson. 2005.
Expansion of geographic range in the pine processionary moth caused by increased
winter temperatures. Ecological Applications 15(6): 2084-2096.
Bentz, B. J., J. A. Logan and G. D. Amman. 1991. Temperature-dependent development of
the mountain pine beetle (Coleoptera: Scolytidae) and simulation of its phenology.
Canadian Entomologist 123(5): 1083-1094.
Bentz, B. J. and D. E. Mullins. 1999. Ecology of mountain pine beetle (Coleoptera:
Scolytidae) cold hardening in the intermountain west. Environmental Entomology 28:
577-587.

86

Bentz, B. J. and A. S. Munson. 2000. Spruce beetle population suppression in northern Utah.
Western Journal of Applied Forestry 15(3): 122-128.
Bjornstad, O. N., M. Peltonen, A. M. Liebhold and W. Baltensweiler. 2002. Waves of larch
budmoth outbreaks in the european Alps. Science 298(5595): 1020-1023.
British Columbia Ministry of Environment. 1994. Monkman Provincial Park: Master Plan.
Prince George, British Columbia.
British Columbia Ministry of Forests and Range. 2009. Ministry of Forests and Range: beetle
facts. Accessed on July 27 th , 2009 from
<http://www.for.gov.bc.ca/hfp/mountain_pine_beetle/facts.htm>.
Brunelle, A., G. E. Rehfeldt, B. Bentz and A. S. Munson. 2008. Holocene records of
Dendroctonus bark beetles in high elevation pine forests of Idaho and Montana, USA.
Forest Ecology and Management 255: 836-846.
Bullard, J. E., G. F. S. Wiggs and D. J. Nash. 2000. Experimental study of wind directional
variability in the vicinity of a model valley. Geomorphology 35(1-2): 127-143.
Campbell, E. M., R. I. Alfaro and B. Hawkes. 2007. Spatial distribution of mountain pine
beetle outbreaks in relation to climate and stand characteristics: a dendroecological
analysis. Journal of Integrative Plant Biology 49(2): 168-178.
Carroll, A. L., S. W. Taylor, J. Regniere and L. Safranyik. 2004. Effects of climate change on
range expansion by the mountain pine beetle in British Columbia. Pages 223-232 in
T. L. Shore, J. E. Brooks and J. E. Stone. Challenges and solutions: Proceedings of
the Mountain Pine Beetle Symposium. Kelowna, British Columbia, Canada October
30-31, 2003. Natural Resources Canada, Canadian Forest Service, Pacific Forestry
Centre. Victoria, British Columbia, Canada. Information Report BC-X-399.
Cerezke, H. F. 1981. Mountain pine beetle in southern Alberta and Saskatchewan. Prairie
Landscape Magazine. Environment Canada, Canadian Forestry Service, Northern
Forest Research Centre.
Cerezke, H. F. 1995. Egg gallery, brood production, and adult characteristics of mountain
pine beetle, Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae), in three
pine hosts. Canadian Entomologist 127(6): 955-965.
Chapman, J. A. 1962. Field studies on attack flight and log selection by ambrosia beetle
Trypodendron lineatum (Oliv.) (Coleoptera: Scolytidae). Canadian Entomologist 94:
74-92.
Crozier, L. 2004. Warmer winters drive butterfly range expansion by increasing survivorship.
Ecology 85(1): 231-241.
Defant, F. 1951. Local winds. Pages 655-672 in T.F. Malone, editor. Compendium of
Meteorology. American Meteorological Society, Boston, Massachusetts, USA.

87

Fettig, C. J., K. D. Klepzig, R. F. Billings, A. S. Munson, T. E. Nebeker, J. F. Negron and J.
T. Nowak. 2007. The effectiveness of vegetation management practices for
prevention and control of bark beetle infestations in coniferous forests of the western
and southern United States. Forest Ecology and Management 238: 24-53.
Fettig, C. J., J. D. McMillin, J. A. Anhold, S. M. Hamud, R. R. Borys, C. P. Dabney and S. J.
Seybold. 2006. The effects of mechanical fuel reduction treatments on the activity of
bark beetles (Coleoptera: Scolytidae) infesting ponderosa pine. Forest Ecology and
Management 230(1-3): 55-68.
Furniss, M. M. and R. L. Furniss. 1972. Scolytids (Coleoptera) on snowfields above
timberline in Oregon and Washington. Canadian Entomologist 104: 1471-1478.
Gamarra, J. G. P. and F. He. 2008. Spatial scaling of mountain pine beetle infestations.
Journal of Animal Ecology 77(4): 796-801.
Goossens, D. 1996. Wind tunnel experiments of aeolian dust deposition along ranges of hills.
Earth Surface Processes andLandforms 21: 205-216.
Graham, M. H. 2003. Confronting multicollinearity in ecological multiple regression.
Ecology 84: 2809-2815.
Hagen, S. B., J. U. Jepsen, R. A. Ims and N. G. Yoccoz. 2007. Shifting altirudinal
distribution of outbreak zones of winter moth Operophtera brumata in sub-arctic
birch forest: a response to recent climate warming? Ecography 30(2): 299-307.
Harcombe, A. 1978. Vegetation resources of the Northeast Coal Study Area 1976-1977.
British Columbia Ministry of Environment. Land Use Commission, Northeast Coal
Development, Victoria, British Columbia.
Hicks, R. R. 1980. Climatic, site, and stand factors. In R. C. Thatcher, J. L. Searcy, J. E.
Coster and G. D. Hertel, editors. The Southern Pine Beetle. USDA, Expanded
Southern Pine Beetle Research and Applications Program, Forest Service, Science
and Education Administration, Technical Bulletin 1631.
Ihaka, R. and R. Gentleman. 1996. R: A language for data analysis and graphics. Journal of
Computational and Graphical Statistics 5(3): 299-314.
Ims, R. A. L., Hans Petter and S. Coulson. 2004. Spatial and temporal variation in patch
occupancy and population density in a model system of an arctic Collembola species
assemblage. Oikos 105(1): 89-100.
Jackson, P. L., B. Murphy and B. J. Burkholder. 2005. Modeling beetle movement by wind.
Bulletin of the American Meteorological Society 86: 28-29.
Jackson, P. L., D. Straussfogel, B. S. Lindgren, S. Mitchell and B. Murphy. 2008. Radar
observation and aerial capture of mountain pine beetle, Dendroctonus ponderosae

88

Hopk. (Coleoptera: Scolytidae) in flight above the forest canopy. Canadian Journal
of Forest Research 38(8): 2313-2327.
Jenness, J. 2006. Topographic Position Index (tpijen.avx) extension for ArcView 3.x, v.
1.3a., Jenness Enterprises, <http://www.jennessent.com/>
Jenness, J. 2005. Distance/Azimuth Tools (distaztools.avx) extension for ArcView 3.x, v.
1.6., Jenness Enterprises, <http://www.jennessent.com/>
Kurz, W. A., C. C. Dymond, G. Stinson, G. J. Rampley, E. T. Neilson, A. L. Carroll, T.
Ebata and L. Safranyik. 2008. Mountain pine beetle and forest carbon feedback to
climate change. Nature 452: 987-990.
Lewis, T. and G. Dibley. 1970. Air movement near windbreaks and a hypothesis on the
mechanism of the accumulation of airborne insects. Annals of Applied Biology 66:
477-484.
Mason, C. J. and M. L. McManus. 1981. Larval dispersal of the gypsy moth. Pages 161-202
in C. C. Doane and M. L. McManus. The Gypsy Moth: Research toward Integrated
Pest Management. USDA, Washington, D.C, USA.
Mattson, W. J. and R. A. Haack. 1987. The role of drought in outbreaks of plant-eating
insects. BioScience 37(2): 110-119.
Nealis, V. and B. Peter. 2008. Risk assessment of the threat of mountain pine beetle to
Canada's boreal and eastern pine forests. Natural Resources Canada, Canadian Forest
Service, Pacific Forestry Centre, Victoria, British Columbia, Canada. Information
Report BC-X-417.
Nelson, T., B. Boots, K. White and A. Smith. 2006. The impact of treatment on mountain
pine beetle infestation rates. British Columbia Journal of Ecosystems and
Management 7(2): 20-36.
Nelson, T. A. and B. Boots. 2008. Detecting spatial hot spots in landscape ecology.
Ecography 31(5): 556-566.
Orwig, D., R. Foster and D. Mausel. 2002. Landscape patterns of hemlock decline in New
England due to the introduced hemlock woolly adelgid. Journal of Biogeography
29(10-11): 1475-1487.
Powell, J. and B. Bentz. 2009. Connecting phenological predictions with population growth
rates for mountain pine beetle, an outbreak insect. Landscape Ecology 24(5): 657672.
Powers, J. S., P. Sollins, M. E. Harmon and J. A. Jone. 1999. Plant-pest interactions in time
and space: A Douglas-fir bark beetle outbreak as a case study. Landscape Ecology 14:
105-120.

89

Preisler, H. and R. Mitchell. 1993. Colonization patterns of the mountain pine beetle in
thinned and unthinned lodgepole pine stands. Forest Science 39(3): 528-545.
R Development Core Team. 2009. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, <http://www.Rproject.org>.
Raffa, K. F. 2001. Mixed messages across multiple trophic levels: the ecology of bark beetle
chemical communication systems. Chemoecology 11: 49-65.
Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner and W. H.
Romme. 2008. Cross-scale drivers of natural disturbances prone to anthropogenic
amplification: The dynamics of bark beetle eruptions. Bioscience 58(6): 501-517.
Raffa, K. F. and A. A. Berryman. 1983. The role of host plant resistance in the colonization
behavior and ecology of bark beetles (Coleoptera: Scolytidae). Ecological
Monographs 53(1): 27-49.
Reid, R. W. 1962. Biology of the mountain pine beetle, Dendroctonus monticolae Hopkins,
in the East Kootenay region of British Columbia. I. Life cycle, brood development,
and flight periods. Canadian Entomologist 94: 531-538.
Ruel, J.-C, D. Pin and K. Cooper. 2001. Windthrow in riparian buffer strips: effect of wind
exposure, thinning and strip width. Forest Ecology and Management 143(1-3): 105113.
Safranyik, L. 1978. Effects of climate and weather on mountain pine beetle populations.
Symposium Proceedings: Theory and practice of mountain pine beetle management
in lodgepole pine forests April 25-27, 1978, Moscow, Idaho. Washington State
University, Pullman Washington, Forest, Wildlife and Range Experiment Station,
University of Idaho, Moscow, USDA Forest Service, Forest Insect and Disease
Research, Washington D.C., USDA Forest Service, Intermountain Forest and Range
Experiment Station.
Safranyik, L. and A. L. Carroll. 2006. The biology and epidemiology of the mountain pine
beetle in lodegpole pine forests. Pages 3-66 in L. Safranyik and B. Wilson. The
Mountain Pine Beetle: A Synthesis of Biology, Management, and Impacts on
Lodgepole Pine. Natural Resources Canada, Canadian Forest Service, Pacific
Forestry Centre, Victoria, British Columbia.
Safranyik, L. and D. A. Linton. 1982. Survival and development of mountain pine beetle
broods in jack pine bolts from Ontario. Natural Resources Canada, Canadian Forestry
Service, Pacific Forestry Centre, Victoria, British Columbia. Research Note 2: 17-18.
Safranyik, L., D. A. Linton, R. Silversides and L. H. McMullen. 1992. Dispersal of released
mountain pine beetles under the canopy of a mature lodgepole pine stand. Journal of
Applied Entomology 113: 441-450.

90

Safranyik, L., D. M. Shrimpton and H. S. Whitney. 1974. Management of lodgepole pine to
reduce losses from the mountain pine beetle. Natural Resources Canada, Canadian
Forest Service, Pacific Forestry Centre, Victoria, British Columbia. Technical Report
1.
Safranyik, L., D. M. Shrimpton and H. S. Whitney. 1975. An interpretation of the interaction
between lodgepole pine, the mountain pine beetle and its associated blue stain fungi
in western Canada. Pullman, WA, Washington State University Cooperative
Extension Service: 406-428.
Safranyik, L., R. Silversides, L. H. McMullen and D. A. Linton. 1989. An empirical
approach to modeling the local dispersal of the mountain pine beetle {Dendroctonus
ponderosae Hopk.) (Col., Scolytidae) in relation to sources of attraction, wind
direction, and speed. Journal of Applied Entomology 108: 498-511.
Schmid, J. M. 1969. Laphria gilva (Diptera: Asilidae), a predator of Dendroctonus
ponderosae in the Black Hills of South Dakota. Annals of the Entomological Society
of America 62: 1237-1241.
Shepherd, R. F. 1966. Factors influencing the orientation and rates of activity of
Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae). The Canadian
Entomologist 91: 207-215.
Shore, T. L. and L. Safranyik. 1992. Susceptibility and risk rating systems for mountain pine
beetle in lodgepole pine stands. Natural Resources Canada, Canadian Forest Service,
Pacific Forestry Centre, Victoria, British Columbia. Information Report BC-X-336.
Shore, T. L., L. Safranyik and J. P. Lemieux. 2000. Susceptibility of lodgepole pine stands to
the mountain pine beetle: testing of a rating system. Canadian Journal of Forest
Research 30(1): 44-49.
Stahl, K., R. D. Moore and I. G. McKendry. 2006. Climatology of winter cold spells in
relation to mountain pine beetle outbreaks in British Columbia, Canada. Climate
Research 32(1): 13-23.
Taylor, L. R. 1974. Insect migration, flight periodicity and the boundary layer. Journal of
Animal Ecology 43(1): 225-238.
Taylor, S. W., A. L. Carroll, R. I. Alfaro and L. Safranyik. 2006. Forest, climate and
mountain pine beetle outbreak dynamics in western Canada. Pages 67-97 in L.
Safranyik and B. Wilson. The Mountain Pine Beetle: A Synthesis of Biology,
Management, and Impacts on Lodgepole Pine. Natural Resources Canada, Canadian
Forest Service, Pacific Forestry Centre, Victoria, British Columbia.
Thomson, A. J. and D. M. Shrimpton. 1984. Weather associated with the start of mountain
pine beetle outbreaks. Canadian Journal of Forest Research 14(2): 255-258.

91

Tishmack, J., S. Mata and J. Schmid. 2005. Mountain pine beetle emergence from lodgepole
pine at different elevations near Fraser, CO. USDA, Forest Service, Rocky Mountain
Research Station, Fort Collins, Colorado. Research Paper RMRS-RN-27.
Trzcinski, M. K. and M. L. Reid. 2008. Effect of management on the spatial spread of
mountain pine beetle {Dendroctonus ponderosae) in Banff National Park. Forest
Ecology and Management 256(6): 1418-1426.
Waring, R. H. and G. B. Pitman. 1985. Modifying lodgepole pine stands to change
susceptibility to mountain pine beetle attack. Ecology 66(3): 889-897.
Weiss, A. 2001. Topographic Position and Landforms Analysis. ESRI User Conference, San
Deigo, California. Poster
Wulder, M. A., C. C. Dymond, J. C. White and B. Erikson. 2006. Detection, mapping, and
monitoring of the mountain pine beetle. Pages 123-154 in L. Safranyik and B.
Wilson. The Mountain Pine Beetle: A Synthesis of Biology, Management, and
Impacts on Lodgepole Pine. Natural Resources Canada, Canadian Forest Service,
Pacific Forestry Centre, Victoria, British Columbia.
Wygant, N. D. 1940. Effects of low temperatures on the Black Hills beetle (Dendroctonus
ponderosae Hopk.). New York State College of Forestry, Syracuse, New York, New
York. Ph.D. thesis.
Zhang, Q.-B. and R. I. Alfaro. 2003. Spatial synchrony of the two-year cycle budworm
outbreaks in central British Columbia, Canada. Oikos 102(1): 146-154.
Zheng, Y., J. Zhu and B. H. Aukema. 2009. Hierarchical dynamic modeling of spatialtemporal binary data. Environmetrics (In review).

92

4: CONCLUSIONS AND AREAS FOR FURTHER RESEARCH
4.1: Conclusions
4.1.1. Influence ofA tmospheric Processes on Aeolian Dispersal
The colonization process of trees by bark beetles is often viewed as an active,
behaviourally-driven process in which the insects take flight and orient towards suitable
hosts, often attracted by a stimulus such as a host kairomone (Borden 1989). However, the
ability of an insect to guide itself toward a stimulus requires the ability to counter forces that
oppose it such as wind. This may occur within a stand, but insect dispersal within advective
currents may be more passive (Jackson et al. 2008). Meso-scale (-100 km) convective
processes within the planetary boundary layer, the zone in which the earth's surface affects
the atmosphere (-1-2 km above the earth's surface) (Drake and Farrow 1988; Westbrook and
Isard 1999), may have largely influenced the dispersal of mountain pine beetle into the Peace
River region. These processes or weather events that may have affected mountain pine
beetle dispersal could have included low-level jets (LLJ), a layer of maximum wind speed in
the planetary boundary layer often at a few hundred meters altitude, and convergent zones,
locations at which two frontal systems cause the descent of wind in a single area (Drake and
Farrow 1988). LLJ tend to occur in areas of deflected large-scale airflows by mountain
ranges, such as the Northern Rocky Mountains. Insects transported by wind are often
uniformly distributed horizontally, lacking convergent circulation cells, but may be vertically
concentrated in a single layer such as a LLJ. Alternatively, convergent zones can cause
horizontal and vertical concentration of insects at high densities within the air column.
Where convergent zones are stationary or form consistently, insects will be deposited (Drake
and Farrow 1988).

93

Rain can also cause the deposition of insects due to "wash out", which is defined as
the removal of particles or insects from the air column due to their collision with water
droplets. However, washout would likely have decreased the amount of successful attack by
the insects after deposition. Rain events are often associated with decreases in temperature
due to cloud cover, and gusty conditions. Mountain pine beetle will generally not emerge
and fly in these conditions, waiting instead for prolonged periods of warm weather (Gray et
al. 1972; Tishmack et al. 2005). Mechanisms of insect convergence in the air column are
critical factors determining the success of insects establishing in a new host after dispersing
long-distances and could result in populations persisting in an epidemic state in locations
where convergence occurs.
The long-distance transport of mountain pine beetle in the Peace River region is most
likely a result of convergent frontal zones (Drake and Farrow 1988). The relatively
stationary location of infestation waves between 2004 and 2005 indicates that some kind of
atmospheric event occurred at that location to allow for the descent of mountain pine beetle
into the area. Alternatively, this pattern could also be indicative of the suitability and
distribution of hosts in the region although such a band (Chapter 2) is more reflective of a
weather front than landscape features per se (Chapter 3).
4.1.2. Lack of Evidence for Anthropogenic Transport
I did not find evidence that anthropogenic spread of mountain pine beetle played a
major role in the establishment and spread of infestations within the Peace River region
between 2004 and 2006. In particular, I found no evidence that direct anthropogenic
manipulation of beetle distribution via movement and/or storage of infested host material
facilitated spread of incipient-epidemic populations. This contrasts, however, with the

94

influence of anthropogenic activities that indirectly influenced the mountain pine beetle
system (Raffa et al. 2008). For example, human alteration of fire regimes in British
Columbia has facilitated large tracts of suitable hosts for mountain pine beetle to establish
(Taylor and Carroll 2004). Taylor and Carroll (2004) found that in the early 1900s
approximately 17% of pine stands were in age classes susceptible to mountain pine beetle,
which rose markedly by 2006 to 55% of pine stands on the landscape, well outside the range
of natural variability. Changes in fire suppression have altered species composition as well
as age class structure, causing a three-way interaction between mountain pine beetle, its host
species, and humans that has allowed mountain pine beetle to take advantage of increased
availability of suitable hosts. Another human-caused process that has indirectly facilitated
spread and establishment of populations of mountain pine beetle in the Peace River region of
British Columbia is climate change (Carroll et al. 2004; Kurz et al. 2008; Raffa et al. 2008).
The ability of mountain pine beetle to successfully establish and spread in the Peace River
region has been due to an increase in summer and winter temperatures that has facilitated
univoltine development as well as high over-wintering survival within populations. The
contrast between direct and indirect impacts of human activities exemplifies the importance
of understanding the complexity of ecosystems when applying management and control
techniques of pests.
4.1.3. Establishment of Infestations at Higher Elevations
A result of this study of particular concern is the successful establishment of
infestations at higher elevations than within their native range. The elevational shift of
mountain pine beetle could indicate that the host within the region is highly susceptible to
attacks, the climate of such areas is conducive to the establishment of infestations (Carroll et

95

al. 2004), and/or that insects dispersing long-distances had a greater relative likelihood of
being dispersed to higher elevation sites by convective currents in this region than in other
areas. It is likely that human mitigated climate change has provided opportunity for this
eruptive herbivore to move into higher elevation sites, allowing them to successfully attack
and kill host trees in this novel habitat.
4.1.4. Effect of Management on Spread of Beetle Infestations
Control tactics, including fall and burn and single tree removal, successfully
diminished the short-distance dispersal potential of mountain pine beetle. However,
management for long-distance dispersal is much more difficult and may require the use of
alternate methods such as use of baited traps outside of host stands (e.g. in large clearings or
stands of non-host species). This would allow for capture of insects dispersing longdistances while limiting "spill-over" attack that could result from luring in mountain pine
beetle populations.
4.1.5. Potential for Continued Spread of Mountain Pine Beetle
The continuing spread of mountain pine beetle is still a large concern for forest
managers across Canada but particularly in northern Alberta and Saskatchewan. Although
mountain pine beetle broods in northern Alberta had an 80% rate of mortality in the winter in
2008 (Alberta Sustainable Resource Development 2008), the beetle has gained a foothold in
northwestern British Columbia and northern Alberta and will probably continue to exist at
endemic levels for a prolonged period (Alberta Sustainable Resource Development 2008).
The continued spread of the insects, either in the Peace River region or in other mountainous
regions such as the southern Rocky Mountains in Colorado, will potentially be facilitated by
large, low-elevation valleys orientated along the dominant wind direction, acting as conduits

96

of suitable habitat for the insect. The orientation of landscape features will also influence the
reproduction of mountain pine beetle and susceptibility of hosts, allowing populations on
warmer and drier sites to have greater rates of increase and spread into adjacent stands. In
order to minimize local spread of outbreaking populations, stake holders including forestry
companies and government organizations should focus control efforts and inventory on such
landscape features. As demonstrated within this study, control measures are effective in
decreasing localized spread of populations; however, where long-distance dispersal is present
alternate control tactics may have to be employed in order to effectively deal with longdistance inputs.
4.2: Areas of Further Research
Examining the influence of pattern of landscape-scale variables on occurrence of
infestations within the Peace River region provides valuable baseline information about the
ability of native insects to spread beyond geoclimatic barriers (Ranta et al. 2002) such as the
Rocky Mountains (Carroll et al. 2004). In particular, movement of mountain pine beetle into
higher elevation sites provides another example of the alteration of fauna to the changing
climate regime. Predicting the spread potential of new infestations based on landscape
features and their orientation, in concert with new spread modeling approaches (e.g.,
Gamarra and He 2008; Zhu et al. 2008; Zheng et al. 2009), may also provide coarse-scale
tools for directing control activities. However, the underlying mechanisms of spread of
mountain pine beetle infestations relative to landscape features and their orientation should
be further examined in the field. For example, windward-facing slopes of various landscape
features should be inventoried for infestations and compared with inventory of leeward sites
to examine differences in establishment rates relative to wind vectors (both within stands and

97

at the meso-scale) and site conditions, such as soil moisture. Deposition patterns of mountain
pine beetle could also be examined relative to multiple hill topography in small-scale
simulated or computer modeling experiments.

98

4.3: Literature Cited
Alberta Sustainable Resource Development. 2008. R-value survey results: Taking action
against mountain pine beetle. Provincial Government of Alberta, Edmonton, Alberta.
Borden, J. H. 1989. Semiochemicals and bark beetle populations: Exploitation of natural
phenomena by pest management strategists. Holarctic Ecology 12(4): 501-510.
Carroll, A. L., S. W. Taylor, J. Regniere and L. Safranyik. 2004. Effects of climate change on
range expansion by the mountain pine beetle in British Columbia. Pages 223-232 in
T. L. Shore, J. E. Brooks and J. E. Stone, editors. Challenges and solutions:
Proceedings of the Mountain Pine Beetle Symposium. Kelowna, British Columbia,
Canada October 30-31, 2003. Natural Resources Canada, Canadian Forest Service,
Pacific Forestry Centre, Victoria, British Columbia. Information Report BC-X-399.
Drake, V. A. and R. A. Farrow. 1988. The influence of atmospheric structure and motions on
insect migration. Annual Review of Entomology 33(1): 183-210.
Gamarra, J. G. P. and F. He. 2008. Spatial scaling of mountain pine beetle infestations.
Journal of Animal Ecology 77(4): 796-801.
Gray, B., R. F. Billings, R. I. Gara and R. L. Johnsey. 1972. On the emergence and initial
flight behaviour of the mountain pine beetle, Dendroctonus ponderosae, in eastern
Washington. Journal of Applied Entomology 21: 250-259.
Jackson, P. L., D. Straussfogel, B. S. Lindgren, S. Mitchell and B. Murphy. 2008. Radar
observation and aerial capture of mountain pine beetle, Dendroctonus ponderosae
Hopk. (Coleoptera: Scolytidae) in flight above the forest canopy. Canadian Journal
of Forest Research 38(8): 2313-2327.
Kurz, W. A., C. C. Dymond, G. Stinson, G. J. Rampley, E. T. Neilson, A. L. Carroll, T.
Ebata and L. Safranyik. 2008. Mountain pine beetle and forest carbon feedback to
climate change. Nature 452: 987-990.
Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner and W. H.
Romme. 2008. Cross-scale drivers of natural disturbances prone to anthropogenic
amplification: the dynamics of bark beetle eruptions. Bioscience 58(6): 501-517.
Ranta, E., P. Lundberg, V. Kaitala and N. C. Stenseth. 2002. On the crest of a population
wave. Science 298(5595): 973-974.
Taylor, S. W. and A. L. Carroll. 2004. Disturbance, forest age dynamics, and mountain pine
beetle outbreaks in BC: a historical perspective. Pages 41-51 in T. L. Shore, J. E.
Brooks and J. E. Stone, editors. Challenges and solutions: Proceedings of the
mountain pine beetle symposium. Kelowna, British Columbia, Canada, October 3031, 2003. Canadian Forest Service, Pacific Forestry Centre, Information Report BCX-399. Ottawa, Ontario, Canada, National Research Council Research Press.
Tishmack, J., S. Mata and J. Schmid. 2005. Mountain pine beetle emergence from lodgepole
pine at different elevations near Fraser, CO. USDA, Forest Service, Rocky Mountain
Research Station, Fort Collins, Colorado, USA. Research Paper RMRS-RN-27.

99

Westbrook, J. K. and S. A. Isard. 1999. Atmospheric scales of biotic dispersal. Agricultural
and Forest Meteorology 97(4): 263-274.
Zheng, Y., J. Zhu and B. H. Aukema. 2009. Hierarchical dynamic modeling of spatialtemporal binary data. Environmetrics In review.
Zhu, J., Y. Zheng, A. L. Carroll and B. H. Aukema. 2008. Autologistic regression analysis of
spatial-temporal binary data via Monte Carlo maximum likelihood. Journal of
Agricultural, Biological, and Environmental Statistics 13: 84-98.

100

Appendix A: R code for fitting spatial point pattern regression models to annual inventory
datasets.
#Pattern of spread by mountain pine beetle infestations in Peace River
#region of British Columbia, Canada between 2004 and 2006.
#Description: Selection of covariates for 2004, 2005, and 2006 that
#provide the most inference about intensity of mountain pine beetle
#infestations.
###########################################################
#Load packages
library(spatstat)
library(shapefiles)
library(maptools)
#Read in shapefile data (2004, 2005, and 2006)
mpb04 <- readShapePoints("2004mpb_licensees")
mpb05 <- readShapePoints("2005mpb_licensees")
mpb06 <- readShapePoints("2006mpb_licensees")
#Extract X and Y coordinates from ShapePoints
X04 <- mpb04$Easting
Y04 <- mpb04$Northing
X05 <- mpb05$Easting
Y05 <- mpb05$Northing
X06 <- mpb06$Easting
Y06 <- mpb06$Northing
#Create object window from research area boundary
landforml <- readAsciiGridCtpi_utm_susl.txt", as.image=TRUE)#Canyons
tpil <- as.im(landforml)
#Create object window for analysis from one of the susceptibility datasets
(these have been clipped by tpi raster in Arcview but are slightly smaller
that tpi boundary in some areas)
tpil.win <- as.owin(tpil)
#Convert ShapePoints to point pattern manually
xxx04 <- ppp(x=mpb04$Easting, y=mpb04$Northing, window=tpil.win)
xxx05 <- ppp(x=mpb05$Easting, y=mpb05$Northing, window=tpil.win)
xxx06 <- ppp(x=mpb06$Easting, y=mpb06$Northing, window=tpil.win)
#Read in distance dataset! Use as covariate to represent location instead
of X and Y coordinates!
d i s t a n c e <- r e a d A s c i i G r i d C d i s t _ s o u r c e . t x t " , as.image=TRUE)
#Distance from roads
r o a d s <- r e a d A s c i i G r i d C r o a d s _ u t m . t x t " ,
•Distance from rivers
r i v e r s <- r e a d A s c i i G r i d C r i v e r s . t x t " ,

a s . image=TRUE)

a s . image=TRUE)

•Distance from mills
mills <- readAsciiGrid("mills_utm.txt", as.image=TRUE)

101

#Create density plot of datasets
dens2004 <- density(xxx04)
dens2005 <- density(xxx05)
•Convert covariates to images
distance <- as.im(distance)
rivers <- as.im(rivers)
mills <- as.im(mills)
roads <- as.im(roads)
###Fit models to see which best fits the dataset
#2004
##Linear gradient
fm.linear <- ppm(xxx04, -distance, Poisson(),
covariates=list(distance=distance))
fm.homo <- ppm(xxx04, ~1, PoissonO)
AIC(fm.linear)
anova(fm.linear, fm.homo, test="Chisq")
##Wave propagation

fm.wave <- ppm(xxx04, ~ I ( d i s t a n c e d ) , P o i s s o n O ,
covariates=list(distance=distance))
AIC(fm.wave)
summary(fm.wave)
sqrt(diag(vcov(fm.wave)))
anova(fm.homo, fm.wave, test="Chisq")
##Distance from roads
fm.roads <- ppm(xxx04, -roads, PoissonO, covariates=list(roads=roads))
AIC(fm.roads)
##Distance from mills
fm.mills <- ppm(xxx04, -mills, PoissonO, covariates=list(mills=mills))
AIC(fm.mills)
##Distance from rivers
fm.rivers <- ppm(xxx04, -rivers, PoissonO,
covariates=list(rivers=rivers) )
AIC(fm.rivers)
#2005
##Linear gradient
fm.linear <- ppm(xxx05, -distance, PoissonO,
covariates=list(distance=distance) )
fm.homo <- ppm(xxx05, -1, PoissonO)
AIC(fm.linear)
##Wave propagation

fm.wave <- ppm(xxx05, - I ( d i s t a n c e d ) , P o i s s o n O ,
covariates=list(distance=distance) )
AIC(fm.wave)
##Stratified dispersal (distance from infestations of the previous year)

102

fm.strat <- ppm(xxx05, ~ dens2004, Poisson(),
covariates=list(dens2004=dens2004) )
AIC (fm.strat)
summary(fm.strat)
sqrt(diag(vcov(fm.strat)))
anova(fm.homo, fm.strat, test="Chisq")
##Distance from roads
fm.roads <- ppm(xxx05, ~roads, PoissonO, covariates=list(roads=roads))
AIC(fm.roads)
##Distance from mills
fm.mills <- ppm(xxx05, -mills, PoissonO, covariates=list(mills=mills))
AIC (fm.mills)
##Distance from rivers
fm.rivers <- ppm(xxx05 , -rivers, PoissonO,
covariates=list(rivers =rivers))
AIC(fm.rivers)
#2006
##Distance from roads
fm.roads <- ppm(xxx06, -roads, PoissonO, covariates=list(roads=roads))
AIC(fm.roads)
##Distance from mills
fm.mills <- ppm(xxx06, -mills, PoissonO, covariates=list(mills=mills))
AIC (fm.mills)
##Distance from rivers
fm.rivers <- ppm(xxx06, -rivers, PoissonO,
covariates=list(rivers=rivers) )
AIC(fm.rivers)
##Linear gradient
fm. linear <- ppm(xxx06, -distance, PoissonO,
covariates=list(distance=distance) )
fm.homo <- ppm(xxx06, - 1, PoissonO)
AIC(fm.linear)
##Wave propagation
fm.wave <- ppm(xxx06, - I (distanceA2) , PoissonO,
covariates=list(distance=distance) )
AIC (fm.wave)
summary(fm.wave)
sqrt(diag(vcov(fm.wave)) )
vcov(fm.wave)
##Stratified dispersal (distance from infestations of the previous year)
fm.strat <- ppm(xxx06, - dens2005, PoissonO,
covariates=list(dens2005=dens2005))
AIC (fm.strat)# 1023605

103

Appendix B: R code for contingency tests (infestations relative to landscape features)
•Contingency test - Distribution of infestations relative to landform
•Description: Compare observed distribution of infestations, by area
•infested (ha), per landform class to expected area infested per flandform
class (i.e. area infested per landform class based on a #uniform
distribution of infestations within those classes)
#######################################################################
###Observed Area Infested per Landform Class
#Load foreign to be able to read dbf files
library(foreign)
#Read in infestation data
data<-read.dbf("mpb_licensees.dbf")
#Examine fields in dataset
summary(data)
#Subset dataset: "Year", "Area", "HAZ_CLASS", "SPOT"
frame <- data[,c("YEAR", "Area", "SPOT")]
#Subset dataset for records by years (2004, 2005, 2006)
yr2004 <- frame[frame$YEAR=='2004',]
yr2005 <- frame[frame$YEAR=='2005', ]
yr200 6 <- frame[frame$YEAR=='2006',]
•Calculate area infested in each landform class (ha)
summary2004 <- aggregate(yr2004[,"Area"], by=list(Landform=yr2004$SPOT),
function(x) sum (x))
summary2005 <- aggregate(yr2005[,"Area"] , by=list(Landform=yr2005$SPOT),
function(x) sum (x))
summary2006 <- aggregate(yr2006[,"Area"], by=list(Landform=yr2006$SPOT),
function(x) sum (x))
summary2004
tRename area columns
colnames(summary2004)[2] <- "Area2004"
colnames(summary2005)[2] <- "Area2005"
colnames(summary2006)[2] <- "Area2006"
#Combine area fields into one matrix
area <- cbind(Landform=summary2004$Landform,
Area2004.Obs=summary2004$Area2004, Area2005.Obs=summary2005$Area2005,
Area2 0 0 6.Obs=summary200 6$Area2 00 6)
area <- as.data.frame(area)
#Change from Number to Name of landform
area[are a$Landform==l, "Landform"] < "Canyons, deeply inci"
area[are a$Landform==2, "Landform"] < "Local ridges/hills i"
area[are a$Landform==3, "Landform"] < "Midslope drainages,"
area[are a$Landform==4, "Landform"] < "Midslope ridges, sma"
area[are a$Landform==5, "Landform"] < "Mountain tops, high"
area[are a$Landform==6, "Landform"] < "Open slopes"
area[are a$Landform==7, "Landform"] < "Plains"

104

area[area$Landform==8, "Landform"] <- "U-shaped valleys"
area[area$Landform==9, "Landform"] <- "Upland drainages, he"
area[area$Landform==10, "Landform"] <- "Upper slopes, mesas"
area
#Total area infested per year (2004, 2005, and 2006 in hectares)
total.04 <- sum(area[,2])
total.05 <- sum(area[,3])
total.06 <- sum(area[,4])
###2004: Expected area infested (hectares per landform class)
#Read in landform dataset that is cut to extent of mountain pine beetle
infestations in 2004
L04<-read.dbf("2004tpihull.dbf")
#Examine fields in dataset
summary(L04)
#Convert GRIDCODE to factor
L04$GRIDCODE <- as.factor(L04$GRIDCODE)
#Sum area (m2) in each landform class
sum.L04 <- aggregate(L04[,c("F_AREA")], by=list(Gridcode=L04$GRIDCODE),
function(x) sum (x))
sum.L04
colnames(sum.L04)[2] <- "Area"
L04.total <- sum(sum.L04$Area)
L04.total*(m2)
•Calculate the proportion of area occupied by each landscape feature
sum.L04$Proportion <- sum.L04$Area/L04.total
sum.L04
•Calculate expected area infested for each landform class in hectares
sum.L04$Area2004.Exp <- (total.04*sum.L04$Proportion)
sum(sum.L04$Area2004.Exp)
total.04
sum.L04
•Convert Gridcode to numeric
sum.L04$Gridcode<-as.numeric(as.character(sum.L04$Gridcode))
class(sum.L04)
summary(sum.L04)
•Replace Gridcode values with equivalent landform class
sum.L04[sum.L04$Gridcode ==1, "Gridcode" <- "Canyons, deeply inci'
sum.L04[sum.L04$Gridcode ==2, "Gridcode" <- "Midslope drainages,"
sum.L04[sum.L04$Gridcode ==3, "Gridcode" <- "Upland drainages, he'
sum.L04[sum.L04$Gridcode ==4, "Gridcode" <- "U-shaped valleys"
sum.L04[sum.L04$Gridcode ==5, "Gridcode" <- "Plains"
sum.L04[sum.L04$Gridcode ==6, "Gridcode" <- "Open slopes"
sum.L04[sum.L04$Gridcode ==7, "Gridcode" <- "Upper slopes, mesas"
sum.L04[sum.L04$Gridcode ==8, "Gridcode" <- "Local ridges/hills i"
sum.L04[sum.L04$Gridcode ==9, "Gridcode" <- "Midslope ridges, sma"
sum.L04[sum.L04$Gridcode ==10, "Gridcode" I <- "Mountain tops, high"
sum.L04

105

###2005: Expected area infested (hectares per landform class)
#Read in landform dataset that is cut to extent of mountain pine beetle
infestations in 2005
L05<-read.dbf("2005tpihull.dbf")
#Sum area (m2) in each landform class
sum.L05 <- aggregate(L05[,c("F_AREA")
function(x) sum (x))
colnames(sum.L05)[2] <- "Area"
L05.total <- sum(sum.L05$Area)
sum.L05

by=list(Gridcode=L05$GRIDCODE),

•Calculate the proportion of area occupied by each landscape feature
sum.L05$Proportion <- sum.L05$Area/L05.total
sum.L05
#Calculate expected area infested for each class in hectares
sum.L05$Area2005.Exp <- total.05*sum.L05$Proportion
sum.L05
•Convert Gridcode to numeric
sum.L05$Gridcode<-as.numeric(as.character(sum.L05$Gridcode))
class(sum.L05)
summary(sum.L05)
•Replace Gridcode values with equivalent landform class
sum.L05[sum.L0 5$Gridcode =1, "Gridcode" <- "Canyons, deeply inci'
sum.L05[sum.L0 5$Gridcode =2, "Gridcode" <- "Midslope drainages,"
sum.L05[sum.L05$Gridcode =3, "Gridcode" <- "Upland drainages, he'
sum.L05[sum.L05$Gridcode =4, "Gridcode" <- "U-shaped valleys"
sum.L05[sum.L05$Gridcode =5, "Gridcode" <- "Plains"
sum.L05[sum.L05$Gridcode = 6, "Gridcode" <- "Open slopes"
sum.L05[sum.L05$Gridcode =7, "Gridcode" <- "Upper slopes, mesas"
sum.L05[sum.L05$Gridcode =8, "Gridcode" <- "Local ridges/hills i"
sum.L05[sum.L05$Gridcode = 9, "Gridcode" <- "Midslope ridges, sma"
sum.L05[sum.L0 5$Gridcode==10, "Gridcode"] <- "Mountain tops, high"
sum.L05
###2006: Expected area infested (hectares per landform class)
#Read in landform dataset that is cut to extent of mountain pine beetle
infestations in 2006
L0 6<-read.dbf("200 6tpihull.dbf")
#Sum area (m2) in each landform class
sum.L06 <- aggregate(L06[,c("F_AREA")
function(x) sum (x))
colnames(sum.L06)[2] <- "Area"
L06.total <- sum(sum.L06$Area)
sum.L06

by=list(Gridcode=L0 6$GRIDCODE) ,

#Calculate the proportion of area occupied by each landscape feature
sum.L06$Proportion <- sum.L06$Area/L06.total
sum.L0 6

106

#Calculate expected area of infestations in each class in hectares
sum.L0 6$Area2 006.Exp <- total.06*sum.L05$Proportion
sum.L06
#Convert Gridcode to numeric
sum.L0 6$Gridcode<-as.numeric(as.character(sum.LO6$Gridcode))
#Repla ce Gr idcode values wi th equivalent landform class
sum.LO 6 [sum.L06$G ridcode ==1, "Gridcode" <- "Canyons, deeply inci'
sum.LO 6 [sum.L06$G ridcode ==2, "Gridcode" <- "Midslope drainages,"
sum.LO 6 [sum.L06$G ridcode ==3, "Gridcode" <- "Upland drainages, he'
sum.LO 6 [sum.L06$G ridcode ==4, "Gridcode" <- "U-shaped valleys"
sum.LO 6 [sum.L06$G ridcode ==5, "Gridcode" <- "Plains"
sum.LO 6 [sum.L06$G ridcode --=6 , "Gridcode" <- "Open slopes"
sum.LO 6 [sum.L06$G ridcode --=1 , "Gridcode" <- "Upper slopes, mesas"
sum.LO 6 [sum.L06$G ridcode ==8, "Gridcode" <- "Local ridges/hills i"
sum.LO 6 [sum.L06$G ridcode: ==9, "Gridcode" <- "Midslope ridges, sma"
sum.LO 6 [sum.L06$G ridcode ==10
1
Gridcode | <- "Mountain tops, high"
sum.LO 6
###Contingency test
#Combine expected and observed values in a single dataframe
chi_area <- cbind(Landform=area$Landform, Area2004.Obs=area$Area2004.Obs,
Area2 005.Obs=area$Area2 005.Obs, Area2006.Obs=area$Area200 6.Obs,
Area2 004.Exp=sum.L04$Area2004.Exp, Area2 005.Exp=sum.L05$Area2 0 05.Exp,
Area2 00 6.Exp=sum.L0 6$Area20 0 6.Exp)
#Convert from matrix to a dataframe
chi_area <- as.data.frame(chi_area)
tConvert factors to numerical data
chi_area$Area2 004.Obs <- as.numeric(as.character(chi_area$Area2 0 04.Obs))
chi_area$Area2 005.Obs <- as.numeric(as.character(chi_area$Area2005.Obs))
chi area$Area2006.Obs <- as.numeric(as.character(chi area$Area2006.Obs))
chi_area$Area2004.Exp <- as.numeric(as.character(chi_area$Area2 004.Exp))
chi_area$Area2005.Exp <- as.numeric(as.character(chi_area$Area2 0 05.Exp))
chi area$Area2006.Exp <- as.numeric(as.character(chi area$Area2006.Exp))
#Double check: Do total area infested in observed equal expected total
area infested per year? Yes!
sum(chi_area$Area2 00 6.Obs)
sum(chi_area$Area2 00 6.Exp)
#Subset expected and observed values for each year
chi2004 <- chi_area[l:10,c("Area2004.Exp", "Area2004.Obs")]
chi2005 <- chi_area[l:10,c("Area2005.Exp", "Area2005.Obs")]
chi2006 <- chi_area[l:10,c("Area2006.Exp","Area2006-Obs")]
#Chi-squared test (Is the distribution of mountain pine beetle
infestations greater in some landforms relative to an expected uniform
distribution?)
chisq.test(chi2004)
chisq.test(chi2005)
chisq.test(chi2006)

107

Appendix C: R code for contingency tests (infestations relative to susceptible habitat)
#Contingency test - Distribution of infestations relative to #susceptible
habitat
•Description: Compares observed distribution of infestations, by area
#infested (ha) per susceptibility class relative to expected (i.e. area
#infested per susceptibility class based on a uniform distribution of
•infestations within those classes).
#######################################################################
###Observed Area Infested per Susceptibility Class
#Load foreign to be able to read dbf files
library(foreign)
#Read in data for infested stands
data<-read.dbf("mpb_licensees.dbf")
#Subset dataset: "YEAR","AREA", "HAZ", "HAZ_CLASS"
frame <- data[,c("YEAR","Area", "HAZ", "HAZ_CLASS")]
#Subset dataset: three years (2004, 2005, 2006)
yr2004 <- frame[frame$YEAR=='2004',]
yr2005 <- frame[frame$YEAR=='2005',]
yr2006 <- frame[frame$YEAR=='2006',]
•Calculate area infested per class
area04 <- aggregate(yr2004[,c("Area")],
by=list(HazardCl=yr2004$HAZ_CLASS), function (x) sum (x))
area05 <- aggregate(yr2005[,c("Area")],
by=list(HazardCl=yr2005$HAZ_CLASS), function(x) sum (x))
area06 <- aggregate(yr2006[,c("Area")],
by=list(HazardCl=yr2006$HAZ_CLASS), function(x) sum (x))
area04
tRename area columns
colnames(area04)[2] <- "Area2004"
colnames(area05)[2] <- "Area2005"
colnames(area06)[2] <- "Area2006"
summary(area04)
•Convert "HazardCl" (i.e susceptibility class) to numeric
area04$HazardCl <- as.numeric(as.character(area04$HazardCl))
summary(area04)
•Combine area fields into one matrix
area <- cbind(HazardCl=area04$Hazard,Area2 004.Obs=area04$Area2 0 04,
Area2 005.Obs=area05$Area2 005, Area200 6.Obs=area06$Area200 6)
area
total.04 <- sum(area[,2])
total.05 <- sum(area[,3])
total.06 <- sum(area[,4])
total.04
total.05
total.06
###2004: Expected area infested (hectares per susceptibility class)

108

Iandscape0 4<-read.dbf("2004hazhull.dbf")
#Subset "HAZ_CLASS" and "area_clip" columns
L04 <- landscape04[,c("HAZ_CLASS","Area_calc")]
#Sum area (meters squared) in each hazard class
sum.L04 <- aggregate(L04[,c("Area_calc")],
by=list(HazardCl=L04$HAZ_CLASS) , function(x) sum (x))
colnames(sum.L04)[2] <- "Area"
sum <- sum(sum.L04$Area)
sum
•Simplify (find proportion)
sum.L04$Proportion <- sum.L04$Area/sum
sum.L04
#Create a column with total.04 as the only number in all rows
sum.L04$Total <- total.04
sum.L04
•Calculate expected area (ha) infested per class based on proportion of
that class on landscape
sum.L04$Area2004.Exp <- (sum.L04$Total*sum.L04$Proportion)
sum.L04
###2005: Expected area infested (hectares per susceptibility class)
landscape05<-read.dbf("2005hazhull.dbf")
landscape05[1:5,]
#Subset "HAZ_CLASS" and "area_clip" columns
L05 <- landscape05[,c("HAZ_CLASS","Area_calc")]
iWere all data imported?
dim(L05)
#Sum area (meters squared) in each hazard class
sum.L05 <- aggregate(L05[,c("Area_calc")],
by=list(HazardCl=L05$HAZ_CLASS), function(x) sum (x))
colnames(sum.L05)[2] <- "Area"
sum05 <- sum(sum.L05$Area)
sum05
•Simplify (find proportion)
sum.L05$Proportion <- sum.L05$Area/sum05
sum.L05
#Calculate expected area of infested in each class based on proportion
area in each class on landscape
sum.L05$Area2005.Exp <- total.05*sum.L05$Proportion
sum.L05
###2006: Expected area infested (hectares per landform class)
landscape0 6<-read.dbf("2 006hazhull.dbf")
landscape0 6[1:5,]

109

#Subset "HAZ_CLASS" and "area_clip" columns
L06 <- landscape06[,c("HAZ_CLASS","Area_calc")]
fWere all data imported?
dim(L06)
#Sum area (meters squared) in each hazard class
sum.L06 <- aggregate(L06[,c("Area_calc")],
by=list(HazardCl=L06$HAZ_CLASS), function(x) sum (x))
colnames(sum.L0 6)[2] <- "Area"
sum06 <- sum(sum.L06$Area)
•Simplify (find proportion)
sum.L06$Proportion <- sum.L06$Area/sum06
sum.L06
#Calc. expected area of infestations in each hazard class based on
proportion area per hazard class on landscape
sum.L06$Area2006.Exp <- total.06*sum.L05$Proportion
sum.L06
###Contingency test
#C"ombine exp and observed area columns fields into a single matrix
overall <- cbind(HazardCl=area0 4$Hazard,Area2 004.Obs=area04$Area20 04,
Area2 0 04.Exp=sum.L04$Area20 04.Exp, Area2 005.Obs=area05$Area200 5,
Area2 0 05.Exp=sum.L05$Area2 00 5.Exp, Area200 6.Obs=area06$Area200 6,
Area2 0 0 6.Exp=sum.L0 6$Area200 6.Exp)
#View result
overall
fConvert to data frame
overall <- as.data.frame(overall)
#Subset expected and observed values for each year
chi2004 <- overall[,c("Area2004.Exp", "Area2004.Obs")]
chi2005 <- overall[,c("Area2005.Exp", "Area2005.Obs")]
chi2006 <- overall[,c("Area2006.Exp","Area2006.Obs")]
chi2004
#Double check: Do total area infested in observed equal expected total
area infested per year? Yes!
sum(overall$Area2 00 6.Obs)
sum(overall$Area2 00 6.Exp)
#Chi-squared test (Is the distribution of mountain pine beetle
infestations greater in some hazard classes relative to an expected
uniform distribution?)
chisq.test(chi2004)
chisq.test(chi2005)
chisq.test(chi2 00 6)

110

Appendix D: Example R code for bootstrapping
•Bootstrapping of contingency tests for stand susceptibility classes
#H0 The distribution of infestations is random relative to •susceptibility
classes
•Description: Compare distribution of infestations to a random
•distribution 1000 times for 2004 dataset
#######################################################################
#Start of program
for(i in 1:1000)
{
cat(c("Test", i))
#Load foreign package (have to read .dbf first otherwise other packages
mask read.dbf function
library(foreign)
#Read in observed infestation data
data<-read.dbf("mpb_licensees.dbf")
data <- as.data.frame(data)
#Change column names: "dbf.YEAR", "dbf.Area", "dbf.HAZ", and
"dbf.HAZ_CLASS"
colnames(data)[6] <- "YEAR"
colnames(data)[3] <- "Area"
colnames(data)[4] <- "HAZ"
colnames(data)[5] <- "HAZ_CLASS"
#Subset dataset: "YEAR","AREA", "HAZ", "HAZ_CLASS"
frame <- data[,c("YEAR","Area", "HAZ", "HAZ_CLASS")]
•Remove "data"
rm(data)
•Subset 2004 dataset
yr2004 <- frame[frame$YEAR=='2004',]
•Remove "frame"
rm(frame)
•Calculate area infested in each susceptibility class (i.e. HAZ_CLASS)
in hectares actual2004 <- aggregate(yr2004[,"Area"] ,
by=list(HazardClass=yr2004$HAZ_CLASS), function(x) sum (x))
actual2004
•Subset 2004 area column
area.yr2004 <- yr2004[,c("YEAR", "Area")]
•Randomize areas
random <- area.yr2004[sample(nrow(area.yr2004)), ]
•Remove "data"
rm(area.yr2004)
•Load spatial data packages
library(shapefiles)

111

library(maptools)
library(maps)
#Read in shapefile with extent of infested area
bndy2004 <- read.shape("2004hull_buf25clip")
•Convert shapefile to polygon
bndy2004.poly <- Map2poly(bndy2004)
#Read in susceptbility polygon data (creates a SpatialPolygonDataFrame)
susceptibility2004 <- readShapePoly("2004hazhull")
•Generate a shapefile of random points in infested area (bndy2004)
random.pnts04 <- dotsInPolys(bndy2004.poly, nrow(yr2004), f="random",
compatible=FALSE)
#Query hazard class for each point
query.haz <- overlay(susceptibility2004, random.pnts04)
summary(query.haz)
•Remove "landforms2004" and "random.pnts04"
rm(susceptibility20 04, random.pnts04)
•Join randomized area column to query.landforms
final <-cbind(HazardClass=query.haz$HAZ_CLASS, Area=random$Area)
final <- as.data.frame(final)
summary(final)
•Calculate area infested in each hazard class (ha; RANDOM).
random2004 <- aggregate(final[,"Area"],
by=list(HazardClass=yr2004$HAZ_CLASS), function(x) sum (x))
•Join RANDOM and ACTUAL areas in a single dataframe
final2004 <- cbind(HazardClass=random2004$HazardClass,
ExpectedArea=random2 0 04$x, ObservedArea=actual2004$x)
final2004 <- as.data.frame(final2004)
•Prepare chi-squared dataframe
chi2004 <- final2004[,c("ExpectedArea", "ObservedArea")]
•Chi-squared test (expected = random infestations, observed = actual
infestations) and write results of consecutive analyses to a single file
without over writing other previous (APPEND)
chi <- chisq.test(chi2004)
chi.clean <- c(chi$statistic, chi$parameter, chi$p.value)
write (chi.clean, "//home/honey/Transfer_linux/Output/2 0 04SR_suscept.txt",
append=TRUE, sep="\t")
•Trigger garbage collection
cat(c("End", i, "Memory",gc()))
•Delete All Objects
rm(list=ls())
•Detach 'shapefiles1 package so this it doesn't mask foreign package
detach(3)

112

#End
}

loop

113

Appendix E: Example R code for selection of spatial point process models
#Influence of covariates on occurrences of mountain pine beetle
#infestations in the Peace River region of British Columbia
•Description: Backwards selection of covariates for 2004 point process
•models
#######################################################################
#Load packages
library(spatstat)
library(shapefiles)
library(maptools)
#Read in shapefile data (2004, 2005, and 2006)
mpb04 <- readShapePoints("2004mpb_licensees")
#Extract X and Y coordinates from ShapePoints
X04 <- mpb04$Easting
Y04 <- mpb04$Northing
#Creat object window from research area boundary
landforml <- readAsciiGridCtpi_utm_susl.txt", as.image=TRUE)#Canyons for
deeply incised streams
tpil <- as.im(landforml)
#Create object window for analysis from one of the susceptibility datasets
(these have been clipped by tpi and susceptibility rasters in Arcview)
tpil.win <- as.owin(tpil)
tConvert ShapePoints to point pattern manually
xxx04 <- ppp(x=mpb04$Easting, y=mpb04$Northing, window=tpil.win)
#Read in elevation data (ASCII)
elevation <- readAsciiGrid("ddcutm_sus.txt", as.image=TRUE)
##Read in landform covariates (ASCII)
#Canyons or deeply incised streams (north-south orientation)
IfIns <- readAsciiGrid("IfIns.txt", as.image=TRUE)
#U-shaped Valleys (northeast-southwest orientation)
lf4nesw <- readAsciiGrid("If4nesw.txt", as.image=TRUE)
#Local ridges or hills in valleys (western aspect)
lf8w <- readAsciiGridClf8w.txt", as . image=TRUE)
#Midslope ridges or small hills in plains (southwest aspect)
lf9sw <- readAsciiGridClf9sw.txt", as . image=TRUE)
#Read in susceptibility covariate (ASCII)
sus5 <- readAsciiGridCsus5.txt", as. image=TRUE)
#Read in distance dataset. Use as covariate to represent location instead
of X and Y coordinates.
distance <- readAsciiGridCdist.txt", as . image=TRUE)
#Convert covariates from list class to image class
elevation.im <- as.im(elevation)
lflns <- as.im(lflns)
lf4nesw <- as.im(If4nesw)
lf8w <- as.im(lf8w)

114

If9sw <- as.im(lf9sw)
sus5 <- as.im(sus5)
distance <- as.im(distance)
#Extract susO's object window
sus5.win <- as.owin(sus5)
#Use cut command to make landform and susceptibility a factor!
sus5 <- cut(sus5, 2)
IfIns <- cut (IfIns, 2)
lf4nesw <- cut(lf4nesw, 2)
lf8w <- cut (lf8w, 2)
lf9sw <- cut (lf9sw, 2)
###Model with all covariates
final2004 <- ppm(xxx04, ~ distance + NS.Canyons + NESW.Ushapedvalleys +
W.Localridges + SW.Midsloperidges + Elevation + High.Susceptibility,
Poisson(), covariates = list(distance=distance, NS.Canyons=lfIns,
NESW.Ushapedvalleys=lf4nesw, W.Localridges=lf8w, SW.Midsloperidges=lf9sw,
Elevation=elevation.im, High.Susceptibility=sus5))
summary(final2 00 4)
AIC(final2004
vcovmatrix <- vcov(final2004, what="vcov")
sqrt(diag(vcovmatrix))
## fm2004.1 <- ppm(xxx04, ~ distance + NS.Canyons + NESW.Ushapedvalleys +
W.Localridges + SW.Midsloperidges + Elevation, Poisson(), covariates =
list(distance=distance, NS.Canyons=lfIns, NESW.Ushapedvalleys=lf4nesw,
W.Localridges=lf8w, SW.Midsloperidges=lf9sw, Elevation=elevation.im))
AIC(fm2004.1)
anova(fm2004.1, final2004, test="Chisq")
## fm2004.2 <- ppm(xxx04, ~ distance + NS.Canyons + NESW.Ushapedvalleys +
W.Localridges + SW.Midsloperidges, PoissonO, covariates =
list(distance=distance, NS.Canyons=lfIns, NESW.Ushapedvalleys=lf4nesw,
W.Localridges=lf8w, SW.Midsloperidges=lf9sw))
AIC(fm2004.2
anova(fm2004.2, fm2004.1, test="Chisq")
fm2004.3 <- ppm(xxx04, ~ distance + NS.Canyons + NESW.Ushapedvalleys +
W.Localridges, PoissonO, covariates = list(distance=distance,
NS.Canyons=lfIns, NESW.Ushapedvalleys=lf4nesw, W.Localridges=lf8w))
AIC(fm2004.3)
anova(fm2004.3, fm2004.2, test="Chisq")
fm2004.4 <- ppm(xxx04, ~ distance + NS.Canyons + NESW.Ushapedvalleys,
PoissonO, covariates = list(distance=distance, NS.Canyons=lfIns,
NESW.Ushapedvalleys=lf4nesw))
AIC(fm2004.4)
anova(fm2004.4, fm2004.3, test="Chisq")

115

fm2004.5 <- ppm(xxx04, ~ distance + NS.Canyons, Poisson(), covariates =
list(distance=distance, NS.Canyons=lfIns))
AIC(fm2004.5)
anova(fm2004.5, fm2004.4, test="Chisq")
fm2004.6 <- ppm(xxx04, ~ distance, PoissonO, covariates =
list(distance=distance))
AIC(fm2004.6)# 324185.7
anova(fm2004.6, fm2004.5, test="Chisq")
fm2004.7 <- ppm(xxx04, ~ 1, Poisson(),covariates=list(distance=distance))
AIC(fm2004.7)#
anova(fm2004.7, fm2004.6, test="Chisq")

116

Appendix F: Contingency tests examining the pattern of infestation relative to landscape
features and susceptible habitat.
Results of contingency tests results indicate a significant difference between observed
area infested per landscape feature and susceptibility classes, assuming uniform distribution
of infested area per class. The expected area per classes was calculated based on the
proportion of that class available within the research area. Results indicate that in 2004 Ushaped valleys, canyons or deeply incised streams, local ridges or hills in valleys, and
midslope ridges or small hills in plains had more area infested than expected (Table Fl and
Fig. Fl). In 2005, U-shaped valleys, canyons or deeply incised streams, and midslope ridges
or small hills in plains exhibited more area infested than expected. This pattern persisted for
U-shaped valleys, canyons or deeply incised streams, local ridges or hills in valleys,
midslope ridges or small hills in plains, as well as open slopes, in 2006.

117

Figure Captions
Figure Fl Differences between expected and observed area (in hectares) infested by
mountain pine beetle per (A) type of landscape feature and (B) stand susceptibility
class within the Peace River region of British Columbia for initial year of recorded
incipient-epidemic populations.

118

Fig. Fl

1500C

-

I
I

I SSJ5
I 2 3 35

1=3

n

1=1

r~i l =•l

51

^\<<<<x.v\x ^./.*<x
'''/y
,:^'

lOOC'C

& 4:

Is

^ H 200-1
1
1

K(ise«ptibi|iiy Class-

119

1 2005
1 7006

Table Fl Test statistics and ^-values for contingency tests examining the effect of landscape
features and distribution of highly susceptible habitat on the occurrence of mountain pine
beetle infestations, assuming an uniform distribution across classes. Data is for the years
2004-2006 in the Peace River region of British Columbia.
Influence of Type of Landscape Feature
X2
P-value

Influence of Habitat Susceptibility
X2
P-value

Year

df

2004

9

1937.81

0.001

1,930.21

0.001

2005

9

2068.00

0.001

1,870.20

0.001

2006

9

9022.51

0.004

13,605.73

0.001

120