A 1\licrosatellite Analysis of the Western Canadian 1\lountain Pine Beetle
(Dendroctonu.\· ponderosae) Epidemic:
Phylogeography and Long Distance Dispersal Patterns

Nicholas V. Bartell

B.Sc .• University of Northern Bnttsh Columbia. 2007

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
June 2008

©Nicholas V. Bartell. 2008

Abstract
The mountain pine beetle (MPB) is an eruptive insect that is currently causing an
outbreak of record size in Western Canada. A lack of long distance MPB dispersal data
has limited our understanding of and ability to manage MPB epidemics. My goal was to
determine the MPBs Western Canadmn population structure. upon wh1ch d1spersal
patterns may be supenmposed. I analyzed MPBs from 35 mfested lodgepole pine stands
at six microsatellite loc1. The MPB exh1bited strong and s1gmticant Western Canadian
population structure. Th1~ population ~tructure wa~ mcongruent w1th the structure of its
primary symbiont. 0. £ hn igerwn. but congruent \\1 ith the structure of 1ts primary host, P.

contorta. Novel fungal selection pres~ure~ ha"Ve probably caused the discrepancy in
beetle/fungus phylogeography. A result of Western Canadmn MPB population structure
alternately contrasts and supports population structure~ previou~ly reported for Scolytids,
including MPBs. The partitioning of MPB populatiOn structure into a Northern and
Southern group is most likely the result of postglacial recolonization and differences in
MPB population dynamics. Primarily using my genetic data, I inferred the historical
movement patterns of the MPB in Western Canada. I found no evidence that the
epidemic spread from an epicenter in Tweedsmuir Provincial Park. My data support
multiple sources for the current epidemic; I suggest that regional population expansions
have caused the rapid escalation in the severity of the current epidemic. MPB movement

patterns and atmospheric wind data were concordant; winds in Western Canada are
predominantly westerly or southwesterly. which was the predominant direction of
inferred movements amo ng MPB populations. In contrast, current MPB population
structure best fits a 30-year climatic suitabi lity distribution for a historical ( 1921-1950) as

II

opposed to the most current ( 1971 -2000) period. The population genetics of long
distance MPB dispersal. an evolutionary theory for MPB population dynamics. and MPB
"range expansion" are extensively d1scussed. Potential biases and research limitations
are noted. Based on my results and inferences. future areas of investigation are noted.
An executive summary. with management recommendations. is prov1ded as a conclusion.

Ill

Table of Contents

Abstract

II

Table of Contents

IV

List of Tables

VI

List of Figures

VII

Acknowledgements

X

I - Introduction
I. I Background
1.2 Research Objectives
1.3 The Mechani~ms Behind Long Distance MPB Dispersal
1.4 Limitations of Mark-Recapture Studies of Long Distance Bark
Beetle Dispersal
1.5 Usmg Population Genetics to Infer MPB D1spersal? Past Genetic
Studies of Long Distance Bark Beetle Dispersal
1.6 Major Influences on the Genetic.., of MPB Populations
- Implications for Study
1.7 Current Methods and Ideas for MPB Management
1.8 The Symbiotic Fungus Grosmannia cia~ igera
1.9 Primary Selective Pressure~ on the MPB and G. ( lavigera
2 - Materials and Methods
2.1 Site Selection and MPB Collection
2.2 DNA Extraction and Evaluation
2.3 PCR of Microsatellites
2.4 Analysis of Microsatellites
2.5 Statistical Analyses
2.5.1 Data Set Validation
2.5.2 AMOVA
2.5.3 Structure
2.5.4 SAMOV A
2.5.5 Isolation by Distance
2.5.6 Other Analyses

I

7
g

10

II
16

20
22
27

30
32
33
34

35
36
37
38
40
41

IV

3- Results
3.1 Hardy-Weinberg Equilibrium and Linkage DiseqUilibrium
3.2 AMOVA
3.3 Structure
3.4 SAMOV A
3.5 Isolation by Distance
3.6 Other Analyses
4- Discussion
4.1 MPB Population Structure in Western Canada
4.1.1. Congruency Among the Phylogeography of the MPB.
its Primary Symbiont. and its Primary Host
4.1.2. Significant MPB Population Structure Context
4.1.3. MPB Genetic Di'.-ersity- Context
4.1.4. Comparison of Beetle/Symbiont Genetic Diversity
-D. ponderosae. D. jeffrevi. and G. da\·igera
4.1.5. We~tern Canadian MPB Population Structure
- A Northern and Southern Group
4.1.6. Western Canadian MPB Population Structure
- Inferred Patterns of Po~t-Ice Age Recolonization
and Population Dynamic~
4.2. The Spread of the Western Canadian MPB Epidemic
4.2.1. Spread of the Epidemic- Southern Group
4.2.2. Spread of the Epidemic- Northern Group
4.2.3. The Population Genetics of Long Distance MPB Dispersal
4.2.4. Concordance with Atmospheric Wind Data
4.2.5. Five Genetically Unique MPB Stands
4.2.6. Ten Comparatively Isolated MPB Population!'.
4.2.7. MPB Population Structure Versus Spread of the Epidemic
4.3. MPB Population Structure and the Geographical Distribution
of Climatic Suitability
4.4. A Model for MPB Population Genetic Dynamics
4.5. MPB "Range Expansion"
4.6. Potential Research Limitations and Biases
4. 7. Recommendations for Future Research
4.8. Executive Summary
4.9. Management Recommendations

S- Literature Cited

43
44
46
49
51
56

67
69
75
81
82
86
90

96
I00
I03
I04
I05
107
I 18
119
121
123
125
127
135
141
142

146

v

List of Tables
Table 1. Sampled stands (35). by region. for the mountain pine beetle with GPS
locations. year sampled. number of beetles fully-genotyped (n). mean observed
heterozygosity. and mean number of alleles. Given locations are mdicative of regions;
sampling did not always occur within noted towns.
31
Table 2. Private alleles. by locus and population. for 35 populations of mountain pine
beetles that were analyzed at six mJcrosatelllte loci.
61

vi

List of Figures
Figure I. Instances of genetic similanty among 35 MPB stands in BC and western AB
analyzed at six microsatellite loci. Connecting lines denote that stands are not
significantly different and thus are genetically similar (Arlequin - AMOV A FsT: a= .05:
9999 permutations). Sampling sites are represented as circles. Similarities 1mply some
level of past and/or recent gene flow. This population structure (Ha =all stands are a
different population) was shallow but highly significant (AMOV A- global FsT = .03828;
P < .00001 ). The inset map denotes the name~ of ~ampling locations.
45
Figure 2. Population structure estnnated u~mg the program Structure. Each individual is
represented by a thin 'r'ert1cal line that is partitioned into K coloured segments that
represent the individual's hkehhood of membersh1p m each of the K clusters. Individual
likelihoods were summed into a mean population likelihood of membership. Black
vertical lines separate populations. ExplanatiOn: at K=2 there are two groups (blue and
orange); at far left Golden has the highest likelihood of membership in the blue group;
membership in the blue group declme~ mo\' ing right; at far nght Houston has the highest
likelihood of membership in the orange group. At K=2 Structure partitioned the 35
stands into a Southern (blue) and Northern (orange) group. Notably, at higher K values,
as represented on this figure by K=3. at far left Golden clearly belongs to the orange
group and at far right Houston clearly belongs to the blue group. However. the
membership of the remaining populations in the middle of the figure is approximately
equally shared among the three groups (yellow. blue. and orange); thus. population
structure dissolved in Structure analyses u~ing K-value~ higher than two.
47
Figure 3. Clines of likelihood of cluster membership (K=2) derived from a Structure
analysis of population structure among 35 sampled MPB stands in BC and western AB.
Structure most strongly ~upported the existence of two groups, Northern and Southern,
which are separated in this figure by a .50/.50 membership isocline; for each cline,
likelihood of stand membership values are for the Northern group on the left and for the
Southern group on the right. Sampling sites are represented as circles. The Northern and
Southern group population structure (all stands above the .50/.50 line in the Northern
group and vice versa) was highly significant (AMOV A- FsT = .05543; FcT = .03665;
both P < .0000 I).
48
Figure 4. Population structure derived from a SAMOV A analysis of 35 stands of MPBs
sampled in BC and western AB. SAMOV A most strongly supported the existence of
two groups (Northern and Southern); the boundary between these groups is the solid
black line. Sampling sites are represented as circles. This Western Canadian MPB
population structure explained the most genetic variation of all of my analyses (AMOV A
50
- FsT = .10 195; FcT = .09162; P < .00001).

vii

Figure 5. Isolation by distance (180) pattern resulting from comparison of FsT-values
and straight-line geographic distances between all pairs of stands. Thirty-five MP8
stands were sampled in BC and western A8 and analyzed at six microsatellite loci . This
relationship was highl y significant (Arlequin- Mantel test with 10.000 permutations; P <
0.000001 ).
52
Figure 6. Isolation by distance (180) pattern resulting from companson of FsT-values
and straight-line geographic distances between all pmrs of stands within only the
SAMOVA-defined Southern MP8 group. Th1s relationship was highly significant
(Arlequin- Mantel test with I 0.000 permutations; P < 0.00000 I).
53

Figure 7. Isolation by distance (180) pattern re~ultmg from comparison of FsT-values
and straight-line geographic distances bet'Aeen all pa1rs of ~tands between only the
SAMOV A-defined Northern and Southern MP8 groups. This relatiOnship was h1ghly
significant (Arlequin Mantel test \\lith 10.000 permutations: P < 0.000001 ).
54
Figure 8. Isolation by d1stance (180) pattern re~ultmg from compari~on of FsT-values
and straight-line geographic distances between all patrs of stands w1thm only the
SAMOVA-defined Northern MP8 group. Thi -.. relationship was not s1gnificant and a
regression line is provided for illustrative purposes (Arlequin Mantel te~t with 10,000
permutation~; P = .6887).
55
Figure 9. An overlay of current (2005/2006) MP8 population structure and the
geographic distribution of regions that are climatically favourable for the MP8 (Left:
1921-1950 climatic data; Right: 1971-2000 climatic data; Carroll et al., 2004 ). The
SAMOV A-derived MP8 population structure is shown, in which the Northern and
Southern groups are demarcated by a solid black line. "Very low" CSCs indicate regions
which are climatically un~uitable for the MP8 while "Extreme" CSCs indicate regions
climatically optimal for MP8s. CSC distribution maps modified and reproduced with
57
permission of A. Carroll.
Figure 10. Inverse relationship between mean number of alleles and straight-line northsouth distance of stands to the 8C!US border (49° latitude). This relationship was highly
significant (P < .00000001 ).
59
Figure 11. Inverse relationship between mean observed heterozygosity and straight-line
north-south distance of stands to the BC!US border (49° latitude). This relationship was

highly significant (P < .00000 I).

60

viii

Figure 12. Map of inferred historical MPB dispersal routes (solid lines) in BC and AB.
These dispersal routes were inferred from: I) generalized patterns of genetic similarities
among 35 MPB stands analyzed at s1x microsatellite loci : 2) atmospheric wind patterns
(Jackson et al.. unpublished data): 3) landfonn influences using a time series of aerial
survey maps of the current epidemic. Note that dashed lines are movements that were
suggested by genetic data but were not apparent from aerial survey maps of the current
epidemic. DISCLAIMER: These "movements" represent an educated best guess using
primarily my genetic data. They may have occurred within the current epidemic or
sometime in the past: statistical methods cannot d1scern between recent and historical
dispersal at this time.
99

ix

Acknowledgement'i
This thesis is the culminatiOn of many successful collaborations on multiple
levels. We acknowledge UNBC and the BC Forest Science Program for funding
research that few expected to succeed. We are grateful for contributions of microsatellite
markers from our Genome BC/Genome AB project affiliation. specifically Drs. Janice
Cooke and Corey Davis at the University of Alberta. and from a USDA Forest Service
Research Team. led by Dr. Karen Mock of Utah State University. We are grateful for
DNA isolates of MPB microbes provided by Alex Plattner of the Colette Breuil lab at the
University of BC. Some beetle ~ampling was conducted by pnvate consultants M.
Duthie-Holt and M. Cleaver as well as by a Canadian Forest Service research team led by
Allan Carroll. The asststance of numerous Canadian Forest Service. BC Mtnistry of
Forests and Range. Alberta Mimstry of Sustamable Resource Development. Provincial
and National Park employees throughout BC and Alberta was mvaluable in helping us
locate infested stands.
Thank you to past research asst~tant s Erin Carbon and Brent Shelest for your
commendable work. Jodi Steinke desene~ recognition for completing an important
initial Undergraduate Thesis on this project. Thank you to Mark Thompson (UNBC
Genetics Facility) for your expertise. To everyone in the Genetics Lab who put up with
my .. awful" music. thanks for the CBC-Radio volume wars. I am also grateful to the
mountain pine beetle for allowing me to sample and genotype you with the goal of
preventing future epidemics. Thank you to UNBC Security for locktng down the
Research Building at haphazard times. especially over the Christmas and Reading
Breaks. To the lab, thanks for letting me frequently work well pa~t midnight without
breaking anything or unexpectedly falling asleep. I regret that I will no longer be racking
any tip boxes on your premises.
A very big thank you to Erin O'Brien for much-needed competition in the final
phases of my lab work. Thank you to my committee: Drs. Brent Murray (supervisor),
Staffan Lindgren. and Allan Carroll, for your expert advice. A special thank you to Dr.
Murray, for helping temper the ideas in this Thesis, and to Dr. Lindgren, for giving me
my start in research. Thank you to UNBC and NSERC for scholarship monies that
indirectly contributed to the Bartell Family Mountaineering Trust, a charity that allows
the Bartell family to spend time together in the mountains. Unfortunately, this
compensation was not adequate, especially given the past six months, and additional
contributions will be required from the author. To my family, I am most grateful.

X

Introduction
1.1. Background
The mountain pine beetle. Dendroctonus ponderosae Hopkins. is the most
destructive pest of pine forests in western North America (Safranyik and Carroll, 2006).
This bark beetle attacks at least I I tree species in the Pinaceae (Kelley and Farrell,
1998). Many of these species, such as lodgepole pine (Pinus contorta var. latifolia and

murravana), western white pine (P. monticola). and ponderosa pine (P. ponderosa), have
wide distributions. The mountain pme beetle (MPB) has killed IOO's of millions of trees
in the United States over the past century (Stock and Guenther. 1979; McGregor, 1985).
However, a current MPB epidemic in British Columbia has caused new tree mortality
over a record I0.1 million hectares based on BC Ministry of Forests and Range surveys
at the end of 2007 (Westfall and Ebata, 2007). This epidemic has extended into Alberta
and is the most severe insect outbreak in Canadian history.
Tree mortality events of this magnitude can have considerable environmental
effects (Uunila et al., 2006). As a positive force, MPB outbreaks are a crucial feedback
mechanism for the regeneration of senescent pine forests (Samman and Logan, 2000).
The sera! consequences of outbreaks, as well as the conditions during outbreaks, are
beneficial for many species ranging from understory vegetation to large mammals
(Schmid and Mata, 1996; Martin eta/., 2006).
Despite positive aspects, the MPB epidemic is overwhelmingly negative,
especially from a human perspective (Samman and Logan, 2000). Losses of mature
forests will negatively impact climax wildlife species and entire aquatic ecosystems due
to habitat loss and changes in water parameters, respectively (Uunila eta/., 2006).

Threatened and endangered species, both terrestrial and aquatic. will face higher
extinction risks. The contiguous expanses of dead trees left by the MPB epidemic will
present an unprecedented risk for large-scale forest fires and muigation costs will be
high. These expanses of dead trees are also making large contributions to global carbon
dioxide emissions (Kurz et al .. 2008). Tourism and property values will also be
impacted, particularly in resort towns. Economic damage from loss of commodity value
is expected to have extremely adverse effects on the Bnt1sh Columbia and Alberta forest
industries (Wagner et al .. 2006).
The mountain pine beetle has been extensively researched because of the severity
of its impacts on forest ecosystems and consequent effects on mdustry (Safranyik and
Carroll. 2006). Particularly well understood is the life history of the MPB. Most MPBs
complete their life cycle within one year (univoltine). Adult MPBs emerge during late
summer by boring to the outside from the subcortical tissues of the host tree. These
MPBs disperse. the majority of beetles flying below the stand canopy until they locate a
suitable host tree. Female beetles initiate construction of galleries in the subcortical
tissues of the host, simultaneously inoculating attacked trees with symbiotic fungi. These
fungi are important for stopping host defenses and killing the tree, as well as for MPB
food sources (Six and Paine. 1998). During gallery construction, volatile chemicals
produced by both the defending host and attacking MPB strongly attract conspecifics,

causing a mass attack of the tree (Borden, 1982).
After mating, females lay eggs along their galleries. Eggs hatch within two
weeks into larvae, which pass through four growth stages, or instars. Larvae excavate
horizontal tunnels in the water- and nutrient-conducting subcortical tissues of the tree.

2

Larval girdling and fungal infection combine to kill the host tree (Safranyik and Carroll.

2006). Depending on the timing of oviposition. as well as site climatic factors. various
instars over-winter under the bark. In spring. larvae resume feeding and as fourth instars.
construct pupal chambers and become pupae. After two to four weeks. pupae develop
into callows. or soft-bodied and pale coloured adults. which in turn develop into adults in
about ten days, completmg the MPB life cycle. Because the MPB life cycle is dictated
by ambient temperatures. in cool locattons such as mountam sites. ltfe cycles may take
two years to complete.
The epidemiology of the MPB ts also well-studted (Safranyik and Carroll, 2006).
In the endemic stage. MPB population densities are so low that only unhealthy
(suppressed) trees are attacked. Three to six trees are tnfested in a mature lodgepole pine
stand per year. at a density of two to three attacks per tree. Thus a maximum population
size for the MPB in the endemic phase is roughly 36 beetles/ha. Through processes that
suppress the health of trees or that increase beetle populations. MPB~ reach the incipientepidemic stage and can successfully mass attack (kill) a large diameter tree in a stand.
Jackson et al. (2008) estimated that -577 beetles are required to successfully attack a
mature lodgepole pine tree. A few years of two- to eight-fold population increases allow
MPBs to reach the epidemic stage, characterized by widespread infestations over large
spatial and temporal scales. Through extreme cold-weather events or resource depletion,
epidemic populations eventually crash and revert to endemic populations.
A lack of long distance MPB dispersal data has limited our understanding of the
development of the current epidemic (Furniss and Furniss, 1972; Safranyik et al., 1992;
Safranyik and Carroll, 2006). The effectiveness of MPB outbreak management has also

3

been affected (Safranyik eta!., 1989; Robertson eta!., 2007). For example, many
complex models assess stand susceptibility to MPB attack (Negron eta/., 1999; Shore et

al .• 2000; Nelson et al.• 2006; Shore eta/., 2006). Such models provide only a
mechanistic understanding of MPB populatiOn dynamics after stands are attacked.
In the assessment of long-range MPB dispersal. the use of mark-recapture
techniques is not economically feasible (Linton eta/., 1987; Salom and McLean. 1990).
The use of genetic techniques. such as allozyme and RAPD (Random Amplification of
Polymorphic DNA) analysis, has also failed to yield significant results (Stock eta/.,

1984; Cal pas eta/., 2002). A relatively new molecular technique. called microsatell ite
analysis, uses genomic markers withm individual beetles and has the greatest potential to
provide critical data on "long distance" (beyond stand) MPB dispersal.
Microsatellites are among the most powerful markers available for genetic
analyses of population structure (Balloux and Lugon-Moulin. 2002). The population
structures derived from microsatellite data allow for high-resolution estimates of
dispersal. As codominant markers, microsatellites produce allele frequency data that
allows for hetero- and homozygotes to be distinguished at a given locus (Goldstein and
SchlOtterer. 1999). Microsatellites are also ideal for population genetic study because
they are neutral and thus free of the confounding effects of selection.
Microsatellites are grouped repeats of short nucleotide sequences, from 2 to 6

nucleotides long, with unique flanking DNA sequences (Beebee and Rowe, 2004).
Primers are developed that anneal to these unique DNA sequences. amplifying (masscopying) the grouped repeats. Different lengths of these repeats at a locus correspond to
different alleles. Microsatellites evolve through genetic drift and mutation, which

4

remove and add repeats. respectively. Because drift and mutation occur randomly. given
sufficient time. isolated populations will show divergence in the number of grouped
repeats at a random locus. allowing for quantification of population differences. Long
distance dispersal patterns can then be elucidated through statistical comparisons of
population allele frequencies.
An undergraduate thesis (Bartell. 2007) on a subset of this Master's dissertation
was among the first to use microsatellite markers to determine the population genetic
structure of a Scolytid beetle (Salle eta!.. 2007). In th1s thesis, the five populations most
likely to be differentiated. primarily because of geographic isolation (distance and
mountain ranges). were selected for analysis. These population~ were Houston. Mount
Robson. Banff. Lac Ia Hache. and Manning Park (see Table I). Using only two
microsatellite loci. 1 demonstrated the efficacy of microsatellite techniques by finding
that Houston was significantly different from the other four populations and that Banff
was significantly different from three other populations (AMOV A at a= 0.05). I
reported that the use of more loci would considerably increase resolution, i.e. the ability
to differentiate populations (Salle eta/., 2007). I concluded that geographic isolation was
a dominating force in MPB population genetics and found no evidence that the current
epidemic originated from dispersal from an epicenter. Instead, my results suggested that
the current MPB epidemic arose from multiple sources. I emphasized that this Master's

dissertation, by analyzing all 35 sampled MPB populations at a greater genetic resolution
(more microsatellite loci), would be able to identify the relative importance of genetic
relationships among populations and the extent of dispersal within the outbreak area.

5

The success of this undergraduate thesis was crucial for determining the feasibility of this
Master's project.
There are numerous plausible explanations for the spread of the current MPB
epidemic. Public perception is that the epidemic originated from an epicenter in
Tweedsmuir Provincial Park. Also, there is a governmental and public perception that
outbreaks in Alberta have onginated via dispersal from numerous. apparently Isolated
BC infestations. Indeed, MPB populations are endem1c to many regions of BC (Wood
and Unger. 1996: Nelson eta/.. 2007c).
Thus. there are two competing theories for the spread of MPB epidemics over
landscapes. Epidem1cs may arise from numerous endemic (native) populations
simultaneously expanding in response to favourable ecological conditions such as
reduced winter mortality and mature. ~usceptible forests. On the other hand, epidemics
may arise from one or a few epicenters from which massive dispersal events take place.
A complex combination of the two theories is also a likely explanation (Namkoong et al.,
1979). Indeed, in an analysis of the current epidemic, Aukema et al. (2006) found
evidence for both a true epicenter in North Tweedsmuir Provincial Park and
simultaneous geographically-isolated outbreaks in southern BC. The main goal of my
research is to use microsatellite analysis to determine the phylogeography and dispersal
patterns of the MPB over the current epidemic area, and in the process, to determine the

most likely mechanisms for the spread of this epidemic.

6

1.2. Research Objectives:
The main goal of my research was to use microsatellite markers to determine the
phylogeography of the MPB over the current epidemic area, upon which dispersal
patterns may be superimposed. In the process, I determined the most likely mechanisms
for the spread of this epidemic. Specific objectives were:
I) To analyze MPB phylogeography and dispersal patterns at regional (fi ne) and
provincial (coarse) scales to determine the relattve influence of historical
isolation versus recent immigration on the current genetic structure of each
MPB population.
2) To compare dispersal patterns derived from MPB population genetic data
with patterns predicted by spatiotemporal and climatic (prevailing
atmospheric winds) analyses of the epidemic's spread, to gain insight into
both the spread of MPB epidemics and the ecology of long distance MPB
dispersal.
3) To compare the population genetic structure found for the MPB to the
population structures reported for the MPBs primary symbiotic fungus, G.

clavigera, and for the MPBs primary host tree, P. contorta.

7

1.3. The M echanism s Behi11d Lo11g Distatlce Mf'll Disp ersal
Long distance bark beetle dispersal is generally viewed as a passive process in
which emerging beetles are caught in updrafts. are moved above the stand canopy. and
are transported many kilometers by atmospheric winds. In Ips tvpographus. the
eightspined spruce bark beetle. Forsse and Solbreck ( 1985) estimated from vertical
trapping and subsequent modeling that -10% of beetles may be above the stand canopy
and disperse many kilometers by wmd during dispersal flights. A similar long distance
dispersal percentage might be expected for MPBs (Safranyik et al .. 1992). which have
similar dispersal ecology and are only slightly larger than these spruce beetles.
Recent research on long distance MPB dispersal (Jackson eta! .. 2008) has found
beetles up to 850 m above the forest canopy. However. most beetles dispersing in the
atmosphere are found within a few hundred meters of the canopy. Jackson eta!. (2008)
suggested that MPBs likely actively regulate their height in the atmosphere in response to
temperature. as hypothesized by Geerts and Miao (2005). Moreover, Jackson eta!.
(2008) conservatively estimated that MPBs may move I I 0 km in a single day. This
estimate assumed an average flight duration of four hours (standard deviation of two
hours) and an average wind speed of 4.8 m/s (standard deviation of 1.3 m/s). If MPB~
undergo passive transport in the atmosphere, then dispersal distances could reach several
hundreds of kilometers, as reported for a 2006 dispersal event into the Peace River region

of BC (Westfall and Ebata, 2007) and as reported for the outbreak which crossed the
Alberta prairies in the 1980's (Cerezke, 1989).
These remarkable dispersal distances are undoubtedly because of the upward
convection currents associated with the warm. fair-weather days needed for Scolytid

8

flight (Chapman, 1962; Safranyik et al., 1989). Furniss and Furniss ( 1972) hypothesized
that these currents could sweep up Scolytids in flight and move them above the stand
canopy; data support this conjecture (Schmid et al., 1992). Though the majority of
dispersing MPB~ fly just above the undergrowth. about 2.4% of dispersing MPBs may be
above the canopy (Safranyik et al.. 1992). Though Botterweg ( 1982) argued that bark
beetles need to disperse longer distances under endemic rather than epidemic conditions.
Safranyik and Carroll (2006) ~uggested an increase in the percentage of above-canopy
dispersers during epidem1cs. when host tree shortages occur. Regardless of how beetles
reach an above-canopy position. once there they may travel I OO's of kilometers before
deposition into new stands (Jackson eta!.. 2008).
Regarding atmospheric transport, it is unknown whether such transport requires
beetles to be actively flying throughout, whether they are "gliding," or are neutrallybuoyant with their wings folded (Jackson et al .. 2008). Thus, at this time. the maximum
potential dispersal distance of bark beetles is a difficult parameter to estimate.
Atmospheric analyses and aerial transects are useful for determining general directions of
long distance bark beetle dispersal but cannot provide data on dispersal between specific
populations or locations (Jackson et al.. 2008). My study will help determine if and how
long-distance dispersal is occurring in epidemic MPB populations. More specifically, I
will determine if there is a correlation between the epidemic phylogeography of the MPB

and the prevailing atmospheric wind patterns of BC and AB. Any correlation will be
examined at both coarse (provincial-level) and fine (regional) scales.

9

1.4. Limitations of Mark-Recapture Studies of Long Distance Bark Beetle Dispersal
The most common method for researching dispersal, particularly in bark beetles
(Scolytinae), is the use of mark-recapture techniques. The scale of these bark beetle
studies has been predominantly within-stand. This scale is directly related to the
limitations of mark-recapture: labour intensive; high cost; prone to considerable error;
prone to study failure. Sal om and McLean ( 1990: 1991) provide an example of the
labour intensive nature of this technique whtle studying dispersal of Trvpodendron

lineatum (Scolytinae) m a small, 3 km 2 valley. They used up to I 00 Lindgren ( 1983)
multiple-funnel traps that had to be pheromone-baited, monitored, and maintained
regularly, while capturing, marking, and releasing beetles over the field season.
Error can be difficult to minimize in mark-recapture. Linton et al. ( 1987) used
florescent powders to mark MPBs and in a test they recaptured significantly fewer
marked compared to unmarked individuals after di spersal flights. Rainfall and high
moisture can remove external markings, as can abrasion and preening (Cook and Hain,
1992), potentially compromising a study. Internal markers, such as rubidium (Rb), are
ingested by beetles but are expensive and are removed by feeding (Thoeny et al., 1992),
placing constraints on the spatial and temporal extent of study. Moreover, the use of
immunomarkers continues to be spatially and economically constrained (Jones et al.,
2006). Safranyik et al. ( 1992) have used mark-recapture to create a solid understanding

of the within-stand mechanisms and extent of MPB dispersal. But because of the
currently insurmountable limitations of mark-recapture, long-distance dispersal is poorly
understood across the Scolytinae (Nilssen, 1984).

10

1.5. Usillg Populatioll Genetics to Infer MP/l Dispersal? Past Genetic Studies of Long
Distance Bark Beetle Dispersal
The primary advantage of a genetic approach to studying di spersal is that markers
are contained within the genome of every sampled individual. With a genetic approach
"resolution" is equivalent to the recapture rate in mark-recapture studies. In the case of
bark beetles. high resolutions are primarily achieved by sampling many stands over a
region (eg. a grid pattern with a I OOkm separation between stands), genotyping -50
beetles/stand. and using a large number of polymorphic neutral markers.
The population genetic approach to determining long distance dispersal. and thus
my research approach. is stepwise. The fundamental requirement for genetically
determining dispersal is significant prior population structure. in other words the
existence of genetically-differentiated populations. Di spersal patterns are superimposed
upon this structure and are detected by instances of genetic similarity between stands.
There are several other advantages associated with a population genetic approach
to dispersal research. Population genetic data allow for the determination of: the source
of invading populations. the presence of selective differences between genotypes, and the
mechanisrn/s that preserve genetic diversity within and between populations (Civetta et

al., 1990; Beebee and Rowe. 2004). Determining patterns of gene flow can provide
critical information for both preventative and reactive forest management (Stauffer eta/.,

1992: Kerdelhue et al., 2003: Mock et al .• 2007). In the long term, population genetic
data may also help us understand how the MPB wil l evolve under a changing climate.

II

Most population genetic studies of Scolytids support the idea that significant
geographic barriers, such as mountain ranges and large distances, can prevent interpopulation gene flow. causing among-population genetic differentiation and generating
population structure. Pioneering population genetic studies that used conservative
allozyme techniques clearly support this idea. In the Douglas-fir beetle (Dendroctonus

pseudotsugae), geographically isolated coastal and inland populations in the Pacific
Northwest may be diverging into separate species (Stock et al., 1979). Moreover. in the
southern pine beetle (D. frontalis), peripatric populations from Arizona and Mexico were
genetically divergent (Anderson eta/., 1979; Namkoong et al., 1979). In a follow-up
study. Roberds eta/. ( 1987) found significant genetic differences among D. frontalis
populations from Texas. Arkansas. North Carolina. and Mississippi. Six et al. ( 1999)
assessed the range-wide population genetic structure of the Jeffrey pine beetle (D.

jeffreyi) and discovered a northern and southern group, consistent with that predicted by
geographic isolation.
Population genetic studies utilizing more modem methods, such as mitochondrial
and nuclear sequencing, consistently support the role of geographic barriers and isolation
in Scolytid population dynamics. The majority of recent Scolytid population genetics
research has occurred in Europe on Ips typographus, Tomicus piniperda, and Tomicus

destruens. In the spruce beetle/. typographus. Stauffer et al. ( 1999) found mitochondrial

evidence for the existence of Central European. Scandinavian, and Central Russian
groups. In the pine shoot beetle T. piniperda, Duan et al. (2004) confirmed the existence
of a new Tomicus species in southern China that was divergent from T. piniperda in
Europe and northern China. Ritzerow et al. (2004) supported this "splitting" ofT.

12

piniperda by providing evidence for barriers to gene flow between the two species. In
Spain, isolation has caused populations ofT. piniperda and its primary host, P. sylvestris.
to evolve in parallel for -0.3 million years (Soranzo et al .• 2000: Ritzerow et al., 2004).
A similar pattern of isolation and parallel evolution has been found between the
pine shoot beetle T. destruens and its host Pinus pinaster in Portugal (Salvador et al.,
2000: Vasconcelos et al .• 2006). T. destruens has well-supported longitudinal population

genetic structures that are embodied by separate groups in Spain, France and Italy
(Faccoli et al .• 2005). Faccoli et al. (2005) also found strong genetic evidence for
separate populations ofT. destruens within Italy. Hornet al. (2006) completed a rangewide study ofT. destruens in the Mediterranean basin and found genetic evidence for
groups in Portugal. Spain/France. Italy. Greece. and the Middle East; populations in
North Africa were hybrids of groups on the Iberian Peninsula (Portugal and
Spain/France).
Considerably fewer population genetic studies of Scolytids have occurred in
North America but these studies have confirmed the importance of geographic barriers in
Scolytid population dynamics. In the western pine beetle (D. brevicomis) there is high
mitochondrial differentiation between western and eastern populations isolated by the
Great Basin (Kelley et al., 1999). Cognato et al. (2003) found three geographicallydefined mitochondrial haplotype lineages in the pinyon pine beetle (Ips confusus). In a
preliminary analysis, Bartell (2007) found significant differences among five isolated

MPB populations in BC and AB at microsatellite loci. Using mitochondrial and
microsatellite techniques. Maroja et al. (2007) delineated three haplotype lineages

13

associated with different geographical regions across the spruce beetle's (D. rufipennis)
North American range.
However. two recent papers have argued that geographical barriers are not a
dominant force in Scolytid population dynamics. Both studies used microsatellite
markers. Salle et al. (2007) found non-significant European population structure in Ips

typographus. the eightspined spruce beetle. They explained this lack of structure by
arguing that/. tvpographus has a consistently high dispersal rate. Kerdelhue et al. (2006)
found a lack of population structure in Tomicus piniperda. the pine shoot beetle, in
France and also argued for a high dispersal rate. However. the potential for sampling and
(fu ngal) contamination bia~ was high in these studies; Stauffer eta!. ( 1992; 1999) and
Kerdelhue et al. (2002) provided opposing results for Salle et al. (2007) and Kerdelhue et

al. (2006), respectively.
Two potentially biased studies on the MPB have also disputed the role of
geographical barriers in Scolytid population dynamics. These studies of the MPB failed
to detect population structure and both doubted that MPB dispersal could be determined
genetically. Calpas eta!. (2002) could not distinguish MPB populations isolated by large
distances. Stock et al. ( 1984) found high genetic similarity between populations from
seven western states in the United States. While Stock et al. ( 1984) used allozyme
analysis, a conservative technique, Calpas et al. (2002) used random amplified

polymorphic DNA (RAPD) analysis, which is highly sensitive to between-population
differences. However, the sample sizes of both studies: 15 per population for Cal pas et

a!. (2002) and not reported for Stock et al. ( 1984). may have caused their negative
results. Calpas eta!. (2002) also used a technique with low reproducibility (Beebee and

14

Rowe, 2004) and did not control for fungal contamination of MPB DNA. Thus Calpas et

al. (2002) may have amplified and analyzed results derived from fungal DNA.
Most studies of the MPB have found some degree of population structure.
relegating the studies of Stock eta!. ( 1984) and Cal pas eta!. (2002) as anomalies. In the
Pacific Northwest, a study of isozyme variation in MPB~ from six sites indicated that the
sites are genetically diverging (Stock and Guenther, 1979). In support of this conclusion,
one population isolated from all others by a large geographic barrier, a desert. was
genetically unique. In Utah and British Columbia, Bentz and Stock ( 1986) found high
levels of allozyme differentiatiOn between D. ponderosae populations. In BC and AB,
high levels of genetic divergence have been found among MPB populations using
allozymes (Langor and Spence. 1991 ). In California, significant differences were found
between northern and central MPB populations using mitochondrial markers (Kelley et

al., 2000). Over its North American range. the MPB exhibited significant population
structure at mitochondrial and AFLP (Amplified Fragment Length Polymorphism) loci
(Mock et al., 2007), with gene flow around rather than across the Great Basin and
Mohave deserts. These studies support the idea that geographic barriers, such as
mountain ranges and large distances, can prevent gene flow and cause genetic
differentiation between populations.

15

1.6. Major Influences 011 the Genetics of MP/l Populations - Implications for Study
In this study I carefully designed a sampling regime that maximized power and
minimized bias. Given that determining MPB population structure was my primary
objective. a major area of concern was controlling for the effects of host species and
epidemic-induced population genetic homogenization on genetic diversity in and
differentiation among MPB populations.
There are genetic ramifications for sampling MPBs from different host tree
species in a study and treating the ~amples as identical. Stock and Amman ( 1980~ 1985)
found that genetic differences in MPBs were highly correlated with host tree species
(ponderosa versus lodgepole pine). Indeed, host species influences the survival and
genetics of attacking MPBs (Stock and Amman. 1985). Langor and Spence ( 1991) also
found significant genetic differences among beetles from different host species (limber
versus lodgepole pine). Notably. all three of the above studies used allozyme analysis.
Allozymes, as protein markers, have low apparent mutation rates (Beebee and Rowe,
2004). Also. allozymes are often not selectively neutral and may have important roles in

cell function and/or maintenance (Bert et al .. 2002). In sum, allozymes are very
conservative markers relative to polymorphic markers such as microsatellites (eg. Ross et

al., 1999).
Considering the conservative nature of allozyme analysis, the results of Stock

and Amman ( 1980~ 1985) and Langor and Spence ( 1991 ), that the MPB exhibits genetic
differences associated with host preference, are more significant than they appear. The
existence of host-associated genetic differences in MPB s was also supported by the
allozyme study of Sturgeon and Mitton ( 1986). However, recent research using a

16

modem, neutral genetic method (AFLP) found no significant differences among MPBs
from different host species (Mock et al., 2007). Kelley et al. (2000), moreover, found
host associated genetic differences among MPBs using allozymes (loci potentially under
selection) but not using neutral mitochondrial markers.
These two recent studies indicate that Stock and Amman ( 1980; 1985), Sturgeon
and Mitton ( 1986), and Langor and Spence ( 1991) detected selective differences among
MPBs from different species of host tree. The results of Kelley et al. (2000) and Mock et

al. (2007) also indicate that the genetic differences they found among MPBs from
different host species were due to host-mediated selection (host effects) and not MPB
host preference. Importantly, modern results suggest that there are no significant
differences among MPBs from different host species at neutral loci. The existence of
sympatric MPB population structure, however, is still an active area of research. I
sampled exclusively from MPB-infested lodgepole pine to prevent any possible genetic
bias in my data.

The fundamental requirement for genetically determining MPB dispersal is
significant prior population structure, in other words the existence of geneticallydifferentiated populations, upon which recent dispersal patterns are superimposed.
However, during epidemics, massive immigration and gene flow may erase the genetic

identities of endemic (native) populations, making the identification of source
populations impossible.
However, based on southern pine beetle epidemics, Scolytid epidemics are not
continuous random-mating populations; some populations that are separated by sufficient

17

distance and/or geographic barriers maintain their genetic identity (Namkoong eta/.,
1979). In addition, Roberds eta/. (1987) found that allele frequencies were similar in
populations that declined from epidemic to endemic status. Among MPB populations in
British Columbia and Alberta, Langor and Spence ( 1991) found high within-site and
higher between-site genetic differences, which suggests that selection and drift are
extremely important in the dynamics and divergence of populations. If epidemics are
relatively infrequent, then the following model of MPB population genetics may be
accurate.
Native MPB populations are randomly distnbuted m habitats. In the endemic
phase, these populations genetically diverge by drift but also by selection at a local scale
(Namkoong eta/., 1979). During epidemics, populations are connected by levels of gene
flow correlated with distance, such that nearby populations become genetically similar
while distant populations remain different. Once the epidemic declines, native
populations in the endemic phase are once again isolated and influenced by drift and
local selection. Bark beetle population genetic structures should be maintained by
barriers, despite the homogenizing effects of infrequent epidemics, allowing for dispersal
inferences.

The population structure derived from neutral genetic markers is a product of the

opposition of two forces. These forces are genetic drift and gene flow. In tandem, these
two forces allow for a population genetic approach to dispersal research. In isolation,
populations will diverge over time due to genetic drift, the random loss of alleles each
generation. Because genetic drift is a random process, isolated populations with an

18

initially identical gene pool will randomly fix or lose alleles at different rates, becoming
differentiated over time. In contrast, gene flow genetically homogenizes populations.
Matings with migrants introduce new alleles and create recombinant offspring. Low
levels of gene flow. such as between isolated populations. simply slow rates of
population divergence but high gene flow can create panmixia (high geneti c similarity)
among populations.

A population genetic approach also allows for the determination of effective
dispersal. Researching effective dispersal entails sampling the surviving brood of
successfully dispersing and reproducing adult MPBs; most stands were sampled in JuneJuly (one exception- May). This approach determines the source/s of immigrating
MPBs that are most important for outbreak propagation. Dispersers may cause tree
mortality at a site but if their brood do not surv1ve, then these dispersers did not
contribute to outbreak propagation. Genotyping effective dispersers provides data
crucial to halting the rapid advance of the current epidemic. Indeed, MPB mortality
during dispersal, especially long-distance dispersal, is extremely high as MPBs must land
in areas with suitable hosts, climate, and in sufficient numbers (Schmid, 1969).
Moreover, after long-di stance dispersal, MPB fat reserves may be exhausted and host
colonization, mating, gallery construction. as well as maternal investment in eggs (Elkin

and Reid, 2005), could be compromised. Identifying source populations producing
effective dispersers capable of overcoming the energetic demands of a long flight and of
surviving and reproducing in a foreign environment, possibly hundreds of kilometers
from their origin, is critical for halting the advance of epidemics.

19

1.7. Current Methods and Ideas for MPB Management
A number of different management options are available to control present MPB
outbreaks and prevent future MPB outbreaks; however. these techniques are highly
limited in scope and effectiveness. Short-term management involves a number of
techniques. such as prescribed burns to infested stands. sanitation harvesting in broodcontaining stands. tree baiting and removal. pesticides. and fall-and-burn treatments. all
of which reduce MPB populations With varying levels of success (Samman and Logan,
2000; Carroll eta! .. 2006). Prescribed burns in unmfested stands have also been used to
create discontinuous host tree distributions and inhibit dispersal, as in some BC
Provincial Parks.
Alternatively the defenses of host trees may be improved. There is a long history
of research supporting the role of stand thinning in reducing tree losses to MPB
outbreaks (Amman and Logan, 1998). As an example, Waring and Pitman ( 1985)
modified lodgepole pine stands in central Oregon by lowering canopy densities or
treating with nitrogen fertilizer. significantly increasing MPB-attack resistance.
However. modern research has disputed the supposed benefits of stand treatments
(Hindmarch and Reid, 200 I; Safranyik eta/., 2004 ). In the Western US, stand thinning
to reduce forest fuel loads increased the incidence of Scolytid attacks on pines by 15
times but caused no significant increase in tree mortality (Fettig eta/., 2006). Safranyik

et al. ( 1999), in the East Kootenays of BC, found that stand thinning had no effect on
Scolytid attacks on lodgepole pine. Nevertheless, many researchers and government
ministries, such as Natural Resources Canada, assert that stand density management is
important for optimizing stand microclimate, tree vigor, and inter-tree spacing, to

20

collectively "beetle proof' stands (Safranyik eta!., 1998; Safranyik eta!., 2004;
Whitehead eta!., 2004; Whitehead and Russo, 2005; Whitehead eta!., 2006).
Regulation of stand composition has also been suggested for minimizing future
MPB outbreaks (Samman and Logan, 2000). Past management practices have promoted
the widespread dominance of climax species, such as lodgepole pine. due in part to fire
suppression and reduced overall disturbance (Taylor and Carroll, 2004), as well as due to
silvicultural practices. Undoubtedly. the abundance of decadent. mature or over-mature
stands of lodgepole pine across BC has been a major factor in the massive scale of the
current MPB epidemic (Taylor eta/., 2006). Samman and Logan (2000) recommend
removing these old-growth stands, which are more susceptible to insect infestations, and
promoting more landscape-level sera] diversity. But effective prevention of future largescale MPB outbreaks may not be as simple as creating more heterogeneous landscapes.
The ideal British Columbia forest landscape of the future must be heterogeneous
in terms of age and genus (Burton, 2006). Since MPBs are generalist mortality agents on

Pinus, which have a very wide distribution in BC, forest managers planning for
landscape heterogeneity cannot only consider the MPBs primary host, lodgepole pine.
Future "landscape heterogeneity" must avoid large regions of contiguous Pinus species,
regardless of age diversity. This distribution of Pinus may not be possible to achieve
given that Pinus species are dominant in many BC ecosystems.

21

1.8. The Symbiotic Fungus Grosmannia clavigera
One of my research goals was to compare the population genetic structure found
for the MPB to the population structures reported for the MPBs primary symbiotic
fungus. G. clavigera (previously Ophiostoma clavigerum), and for the MPBs primary
host tree. P. contorta. A major goal of modem research is to determine the comparative
phylogeography. and thus the comparative evolutionary ecology. of closely associated
species (Stauffer et al .. 1999; Ritzerow et al.. 2004; Hom et al .. 2006; Vasconcelos et al .•
2006; Maroja eta!., 2007). I begin with the MPBs most critical symbiont. Given my
comparative analyses (see Discussion). it is pertinent to thoroughly introduce G.

clavigera and review both its mutualism with the MPB and the factors affecting its
evolution.

Grosmannia clavigera is the main symbiotic fungus of the MPB (Solheim, 1995;
Yamaoka et al .• 1995; Solheim and Krokene, 1998). Both species benefit from their
mutualistic relationship. G. clavigera benefits by being dispersed between suitable host
trees (Harrington, 1993; Paine et al .• 1997). which are scarce during the temporallydominant endemic MPB population phase (Preisler and Mitchell, 1993; Safranyik and
Carroll, 2006). This fungus may be entirely dependent on the MPB for dispersal and
thus long-term persistence (Six and Paine, 1999). Indeed. G. clavigera individuals

produce a wide morphological range of asexual forms, or anamorphs, many of which are
highly adapted for insect dispersal (Harrington, 1993). The fungus may benefit the MPB
through: 1) protecting against antagonistic blue-stain fungi (Klepzig and Wilkens, 1997;
Paine eta!., 1997); 2) helping overcome host defenses (Berryman, 1972; Owen et al ..

22

1987); 3) favourably altering the chemical and/or moisture composition of the phloem
(Reid, 1961; Wagner et al., 1979); 4) providing nutrients required for MPB reproduction
and/or development (Whitney et al., 1987; Goldhammer et al., 1990; Six and Paine.
1998).
The mutualistic relationship between fungi and bark beetles has been widely
supported but some researchers dispute such a relationship (Harrington, 1993 ). The key
traits of mutualistic bark beetle fungi are pathogenicity and a degree of bark beetle
specificity but most bark beetle fungi do not exhibit these traits (Harrington, 1993 ).
However. G. clavigera is highly pathogenic in Pinaceae (Yamaoka eta! .. 1995; Solheim
and Krokene, 1998) and is the main symbiont carried by MPBs (Six, 2003 ). Also, G.

clavigera is only vectored by two species, the MPB and its sister species the Jeffrey pine
beetle (Lee eta! .. 2007). The G. clavigera - MPB relationship is likely mutualistic (Six
and Paine, 1999; Six, 2003 ). 0. montium is also commonly associated with the MPB but
is less pathogenic and species-specific compared to G. clavigera (Harrington, 1993). A
new symbiotic fungus, Leptographium longiclavatum, was recently isolated from the
MPB (Kim et al .. 2005; Lee et al., 2006a) and has pathogenicity similar to G. clavigera
(Lee et al., 2006b), but given the wealth of research on G. clavigera, it is most likely the
MPBs primary symbiont.

G. clavigera, like many fungi associated with bark beetles, is an ascomycete

(Harrington, 1993). Its mycelia are predominantly haploid anamorphs that reproduce
asexually, but during sexual reproduction, short-lived diploid teleomorphs are formed
(Six and Paine, 1999). Asexual spores, or conidia, are produced in slimy masses that line

23

MPB galleries. These conidia are acquired by MPBs by adhering to the exocuticle or by
ingestion into the mycangia of newly eclosed adult MPBs (Six and Paine, 1996).
The population genetic structure of G. clavigera, the MPBs main fungal
symbiont, may yield insight into MPB population genetics. Lee eta/. (2007) found low
genetic diversity in G. clavigera isolated from MPBs or MPB-infested trees at seven sites
in Canada (Houston. Fort St. James. William's Lake, Manning Park. Banff) and the
United States (Hellroaring. Idaho: Hidden Valley, Montana). Lee eta/. (2007) also
found moderate but significant differences among the populations. partially attributable
to distinct Rocky Mountain and BC Interior population genetic structures. They also
found two very strongly supported genetically-distinct groups. Most individuals
belonged to Group I (with representatives from all populations) while Group 2 contained
nine individuals (representing the Rocky Mountain populations). Since both Group I
and 2 were found in the Rocky Mountains but only Group I was found in the BC
Interior. and since the MPB epidemic is spreading from the BC Interior, Lee eta/. (2007)
suggested that Group 2 may represent G. clavigera populations endemic to the Rocky
Mountains while Group I recently came into secondary contact via MPB immigration.
The two groups are so strongly supported (genetically unique) that they may represent
cryptic species (Lee eta/., 2007).
The combination of results from Bartell (2007) and from Lee eta/. (2007) show

significant population structures in the MPB and its main fungal symbiont. These results
support the hypothesis that geographic barriers are a dominant force in MPB population
dynamics. Finally, Lee et al.'s (2007) results also suggest that highly isolated MPB
populations exist in the Rocky Mountains that are only connected to BC Interior

24

populations by immigration during extremely severe MPB epidemics. There is an
inconsistency. however. in Lee et al.'s (2007) results.
Low levels of genetic diversity in populations of MPH-associated G. clavigera
(Lee et al .• 2007) are not consistent with several studies of haploid fungi. which found
levels of genetic variation comparable to outbreeding diploid organisms (Spieth, 1975;
Perkins and Turner. 1988). However. in a close relative to the MPB. Dendroctonus
jeffreyi (Higby and Stock, 1982), Six and Paine ( 1999) also discovered low genetic
variation in G. clavtgera. The low genetic diversity of G. clavigera from both D.
ponderosae and D. jeffreyi may be explained by immediate selection against deleterious
alleles (Perkins and Turner. 1988). dominance of genetic drift over mutation in small
populations (Christiansen and Feldman. 1986; Hartl and Clark, 1997; Beebee and Rowe,
2004), low levels of sexual reproduction (Harrington eta/., 1996), poor dissemination of
sexual spores (Six and Paine. 1999), evolution of clonality to prevent the breaking up of
high fitness genotypes essential for mutualism (Wulff. 1985), or by founder effects (Six
and Paine, 1999).
Given its wide host preference within the genus Pinus (Wood, 1982), the MPB
likely hasn't suffered species-wide founder effects. Moreover, G. clavigera associated
with MPBs commonly reproduces sexually (Robinson-Jeffrey and Davidson, 1968).
However. the ascospores derived from sexually-produced parents may not be produced

fast enough for MPBs to acquire them before dispersal (Six and Paine, 1996); this may
be especially true in most areas, where MPB life cycles are completed within one year.
Notably, MPB contact with G. clavigera ascospores may be possible in mountain
environments such as the Rocky Mountains, where MPB life cycles can last two years or

25

more (Safranyik and Carroll, 2006). A greatly prolonged MPB life cycle may ensure that
a higher proportion of (sexual) ascospores relative to (asexual) conidia are ultimately
acquired by dispersing MPBs. However. in sum. poor dissemination of sexual spores,
the diversity reducing effects of clonality. and the rapid elimination of deleterious alleles
may be the cause of low genetic diversity in MPH-derived G. clavigera. In mountain
environments such as Banff. widespread sexual reproduction, combined with the
predominant disseminatiOn of ascospores, may maintain the higher levels of G. clavigera
genetic diversity from Rocky Mountain populations found by Lee eta/. (2007). Indeed,
the higher levels of genetic variation in G. clavigera from the Rocky Mountains may be
needed for the fungus to maximally respond to selective pressures and evolve (Six and
Paine. 1999).

26

1.9. Primary Selective Pressures 011 the MP/1 and G. clavigera
It is important to understand the major host-mediated selective pressures on the

MPB and its primary symbiont. Such pressures may explain regional differences in
population structure. There are four major additive ways MPBs overcome host tree
defenses and kill trees after attack in late summer/early fall (Franceschi et al., 2005;
Safranyik and Carroll, 2006). First, the beetles use efficient aggregation pheromones to
attack healthy trees with the timing and numbers needed to overcome their defenses.
Second. the beetles need a tolerance to host defense chemicals, most likely a high
tolerance given ever-present constitutive defenses. Third. the beetles physically damage
the bark and inner tissues of the host through adult and larval tunneling. Lastly and
perhaps most critically. upon tree penetration MPB~ introduce symbiotic fungi, present in
their mycangia and/or on their exoskeleton. which infect the tree and prevent water
conduction. This is critical because water and resin conduction must be stopped, or
impeded to very low levels, very rapidly upon MPB attack (Franceschi et al., 2005;
Safranyik and Carroll, 2006). If not, constitutive defenses will pitch out and kill MPBs.
The development of induced defense responses will follow, causing invader-specific
damage.
The first three ways MPBs overcome host defenses most likely show low
variation in MPB populations given their fundamental role in MPB ecology. The low
variation of these three MPB-intrinsic traits is likely persistent across population phases.
During the endemic phase, mass attacks do not occur and both tolerance to host
chemicals and physical damage to the inner bark tissues are probably not important in

27

stressed trees. However. these three traits must be retained to overcome the strong
defenses of mature trees during epidemics.
Of the above four factors, the most important is most likely the presence of
symbiotic virulent fungi. MPB attacks are unsuccessful without these fungi (Yamaoka et

al., 1995; Six and Paine, 1998); there should be heavy selective pressure for fungal
strains that are increasingly virulent and stop sap flow in trees expeditiously. It has been
shown (Solheim. 1995; Solheim and Krokene, 1998) that G. clavigera is the most
virulent of MPB symbiotic fungi. Among MPB symbiotic fungi, only G. clavigera is
aggressive enough to kill lodgepole pine when introduced alone (Yamaoka eta!., 1995).
Within the heavy selective pressure on symbiotic fungi, the Canadian Rocky Mountain
environment, compared to the Interior. may differentially constrain or enhance selection
on individual beetles for novel strains of symbiotic fungi. Since harboring more virulent
strains of G. clavigera. which have regional adaptations, would represent a considerable
fitness advantage for MPBs, this may explain the genetically distinct group of G.

clavigera from Rocky Mountain MPB populations (Lee eta!., 2007).
My research may clarify the relationship between the Rocky Mountain and BC
Interior populations of the MPB and its primary mutualist (G. clavigera). There are
some large genetic differences between Rocky Mountain and BC Interior populations of
G. clavigera. Lee et al. (2007) found that the number of unique AFLP markers (38 out

of 469), heterozygosity, and polymorphism was higher in Rocky Mountain compared to
Interior BC populations of G. clavigera. As previously mentioned, Lee eta!. (2007) also
found that most individuals in their study belonged to Group I (with representatives from
all populations) while Group 2 contained nine individuals (representing the Rocky

28

Mountain populations). They hypothesized that Group 2 was an original endemic
population while Group I was a widespread population that spread from the BC Interior
during the current epidemic. Lee eta/. (2007) suggested that Group I may be more
prevalent because it is more pathogenic or is more beneficial for MPBs while Group 2
may be better adapted to the colder climate of the Rocky Mountains. This latter
suggestion is concordant with my above prediction of differential selection for novel
strains of G. clavigera in the isolated Rocky Mountains compared to the Interior.
In sum. my results will shed light on the association between the MPB, its
primary symbiont, and its primary host, by providmg key data on MPB population
genetic structure. If Rocky Mountam MPB populations are very different from Interior
MPB populations but show recent Immigration from the Interior. matching the structure
of G. clavigera (Lee eta/., 2007). there is evidence that these populations have evolved
allopatrically. But if Rocky Mountain MPB populations are similar to Interior MPB
populations, conflicting with the structure of G. clavigera, there is evidence for
differential fungal selection in the Rocky Mountains.

29

Materials and Methods
2.1. Site Selection and MP/l Collection
In the summers of 2005, 2006, and 2007, MPBs were collected, prior to adult
dispersal, from nine regions throughout BC and Alberta for a total of 35 geographically
distinct sampling sites (Table I). In 2007 only Grande Prairie was sampled. Regions
were selected based on: historical data on past MPB outbreaks; the perceived locations
of native (endemic) populations; and geographic barriers to dispersaL Sampling was
balanced between the edges of the current infestation. to determine the origin of recent
MPB dispersal flights. and a distribution of sites across the epidemic area of BC and
Alberta. Collection permits for Jasper. Banff. and Yoho National Parks, and for several
BC Provincial Parks, were obtained, and landowner permissions were obtained as
needed.
At each sampling site. 13 to 20 infested trees separated by I0 meters or more
were selected. For each tree. bark was removed from each of four sides with a hatchet.
A GPS location was taken at each sampled tree. At least four individual offspring were
sampled from each of four separate breeding galleries per tree. Different breeding
galleries were sampled to ensure genetically independent samples. Only lodgepole pine

(Pinus contorta Douglas var. latifolia) trees were sampled to avoid bias due to potential
host -associated M PB preferences.

30

Table I. Sampled stands (.35), by region, for the mountain pme beetle with GPS locations, year sampled,
number of beetles fully-genotyped (n), mean observed heterozygosity, and mean number of alleles. Given
' t'tve o f I'C!!IOn~: ~a111p1111g
I'
d'd
. h'm note d towns.
Iocaf tons arc .md tea
I not a Iway~ occur wtt
Location by Region
n
Mean He 1
Year
Abbr.
Latitude (N) Longitude (W)
Observed
Sampled
Rocky Mountains
Chct wynd I Pine Pass
Willmore Wilderness
Mount Robson Provincial Park
Banff (Banff National Park)
Lake Louise (Banff National
Parkl
Kootcnav National Park
Golden
Northeast of Rocky Mountains
Tumbler Rid!!c
Grande Prairie
Nechako Plateau
Fort St. James
Francois Lake
Houston
Telkwa
West of Rocky Mountains
Mackenzie
Prince Gcor!!c
McBride
Valemount
Cariboo-Chilcotin

Mean
Number of
Alleles

C/PP
PC
YH
TM
LL

55 .6352
53 5707
52.8949
51 1770
51.4172

122.2522
119.7928
118.7348
115.5593
116.1793

2006
2006
2005
2006
2006

55
48
46
45
48

0.51
0.51
0.54
0.61
0.58

5.3.3
4.8.3
5.17
5.67
6.33

KP
G

50 6436
51 2385

115.9784
116.6530

2005/2006
2005

48
47

0.65
0.66

6.00
5.83

TR
GP

54 ~914
54 7270

121 .2252
118.97.36

2005/2006
2007

48
30

0.5.3
0.47

4.50
4. 17

JP

54.6463
54.0317
51.9938
54 6677

124 4196
124.9.391
126.6522
127.0890

2005
2006
2006
2006

46
48
47
48

0.48
0.51
0.48
0.47

4.3.3
4. 17
4.8.3
4.33

54.6963
53 .9068
53 3120
52.6590

122.8203
122.8066
120.1266
118.9965

2005
2005
2005
2005

47
47
24
48

0.42
0.54
0.49
0.56

4.33
5.17
4. 17
5.67

FL
HO
TE
KL
PG
Rf

v

(,)uc~nc!

(.)U

Bowron Lake Provincial Park
Farwell Canvon
Tatla Lake
Lac La Hache
Wells Gray Provincial Park
Coast Mountains
Whistler
Cascade Mountains
Manning Park
Thompson-Okanagan
Lillooet
Merritt

BL
FC
TA
LA
SP

53.0423
53.2488
51.6590
51.9715
51 .7342
51.7410

122.2519
121.4162
122.9177
124.4130
121.6071
120.0121

2006
2006
2006
2006
2006
2006

46
47
48
44
46
47

0.55
0.54
0.48
0.52
0.57
0.60

5.33
5.50
5.17
4.67
5.50
6.33

GL

50.1683

122.9260

2006

48

0.56

5.50

MP

49.2163

121.0698

2006

46

0.58

5.67

LC
IR
KA
RR
KE

50.4568
50.0343
50.4859
50.5199
49.9978

121.6346
120.6575
120.5321
119.6018
119.6657

2006
2006
2006
2006
2006

44
48
43
48
46

0.56
0.58
0.64
0.58
0.59

6.33
6.83
6.33
6. 17
6.17

NG
VA
WA
AR
ANG

49.2592
49.7500
49.5250
50.1566
49.6416

117.9277
117.4949
117.2321
116.9164
116.2124

2006
2006
2006
2006
2005

48
48
46
48
49

0.64
0.55
0.55
0.64
0.62

6.00
6.00
6.17
5.8.3
6.50

Kamloop~

Falkland
Kelowna
Kootenays
Nancy Greene Provincial Park
Valhalla Provincial Park
West Arm Provincial Park
Arge nta
Ki mhcrlc y

31

The 35 populations were sampled over two years, from mid-June to mid-Ju ly, to
maximize collection of the adult MPB life stage (two exceptions: May- Willmore
Wilderness; from bolts- Grande Prairie). Sampling occurred during this "window"
because most adult MPB offspring disperse in mid-July, requiring an earlier collection
date. Different rates of development and different macro- and microclimatic conditions
between sampling sites, province-wide. resulted in collection of larval, pupal, and adult
life stages. In some high elevation stands, for example, only overwintering adults and
tiny. unsuitable larvae were found because of reduced temperatures, shorter summers,
and reduced phloem thickness as compared to valley bottom sites (Safranyik and Carroll,
2006). Conversely, at sites with a high southern exposure at low elevations, some beetles
had emerged by mid-June. Analyses show no difference in the quality of DNA extracted
from different MPB life stages.
Sampled beetles were placed in labeled vials containing 95% ethanol and stored
at -80°C for DNA preservation. From each gallery, one beetle was randomly selected for
genetic analysis (at least 48 beetles/sampling site).

2.2. DNA Extraction and Evaluation
Using a standard proteinase K and phenol/chloroform procedure (Sambrook and
Russell, 2001 ), DNA was extracted from one beetle per gallery for each site. Adult

offspring were selected when possible. Sterile plastic micropestles were used to
homogenize each beetle sample, followed by two applications of proteinase K and
incubation, culminating in overnight digestion in an incubator at 37°C. DNA was
extracted as above, precipitated in ethanol/sodium acetate solution, and resuspended in

32

Tris-EDTA buffer (pH 8.0) for storage at -80°C. The amount and quality of extracted
DNA was evaluated through electrophoresis on I% agarose gels using ethidium bromide
staining. To normalize DNA to I 00 ng/JlL in Tris-EDTA buffer (pH 8.0). extraction
concentrations were determined using a NanoDrop® ND-1 000 UV-Vis
Spectrophotometer.

2.3. PCR of Microsatellites
Through an initial collaboration with a research team led by Dr. Karen Mock
(Utah State University), I had access to the microsatellite markers they developed. Only
two marker systems, CAL 1-1 and MPB 017, were confirmed at UNBC to amplify beetle
DNA (Steinke, 2006). Even in highly-controlled, sterile laboratory environments with
enrichment protocols, building bark beetle microsatellite libraries is extremely difficult
(Salle et al., 2007). My second collaboration, with Dr. Janice Cooke of the University of
Alberta, is a prime example of such difficulty. Approximately 80% of all primers
developed for the mountain pine beetle (Dendroctonus ponderosae), despite considerable
controls, were developed for fungal mutualists co-occurring in beetle DNA extractions
(C. Davis, unpublished data). Four marker systems were confirmed at UNBC to amplify
beetle DNA (MPB25. MPB30, MPB35, and MPB40) while a fifth marker system
(MPB50) was only sporadically successful and was discontinued. Thus six microsatellite
systems were used in this project.
PCRs were run individually, using one microsatellite per beetle, for each selected
sampling site. The reagent concentrations for PCR were (for CAL 1-1 and MPB 0 17):
2.4mM MgCh, 200J.1M (each) dNTP's, I X PCR buffer, 200nM of dye-labelled forward

33

primer, 200nM of reverse primer, I unit of Taq DNA polymerase, and 20ng of DNA
template, to a final volume of 25!J.L. The reagent concentrations for PCR were (for
MPB25, MP830, MPB35. and MPB40): 2.4mM MgCh, 200!J.M (each) dNTP's, I X
PCR buffer, 80nM of M n-tagged left primer. 320nM of right primer, 640nM of dyelabelled M13 primer. I unit ofTaq DNA polymerase, and 20ng of DNA template, to a
final volume of 15!J.L.
Negative controls substituted I !J.L of PCR water for template beetle DNA. For
CAL 1-1 and MPBOI7. thermal cycling involved temperatures of94°C for4 minutes,
followed by 36 consecutive cycles of: one minute of denaturation at 92°C, annealing
(49°C for Cal 1- 1 and 56°C for MPB017) for I minute, and extension for I minute at
72°C with extension in the final cycle lasting II minutes. For MPB25, MPB35, and
MPB40. thermal cycling was 94°C for 4 minutes, followed by 16 consecutive cycles of
(touchdown cycles): one minute at 94°C, 56°C (decreasing by 0.5°C each cycle) for I
minute, and 72°C for I minute, then 17 consecutive cycles of (main cycles): one minute
at 94°C, one minute at 53°C, and one minute at 72°C with extension in the final cycle
lasting II minutes. Thermal cycling was similar between MPB 25, 35, 40 and MPB 30
except that MPB 30 had a 59°C initial touchdown annealing temperature and a 50°C
main cycle annealing temperature. Completed reactions were stored at -20°C .

2.4. Analysis of Microsatellites
PCR reactions underwent fragment analysis on Beckman-Coulter CEQ8000
automated DNA sequencers. All allele scoring was done manually to prevent both the
erroneous scoring of artefacts and the occurrence of null (unrecognized but real) alleles

34

(Jarne and Lagoda, 1996; Dakin and A vise, 2004). Different alleles were determined for
each microsatellite locus and genotypes were generated for each beetle (-50 beetles per
sampling site).

2.5. Statistical Analyses
2.5.1 Data Set Validation
I analyzed my data set for concordance with Hardy-Weinberg Equilibrium
(HWE) expectations of random mating as well as for linkage disequilibrium, the nonrandom association of alleles between loci. Genetic studies across taxa have confirmed
that the majority of populations conform to HWE expectations of random mating. Under
HWE. population genotype frequencies remain relatively constant over time. Although
HWE is theoretically violated by: assortive mating, inter-population gene flow, natural
selection on markers, mutations, and genetic drift, many populations that are
considerably influenced by one or more of the above factors remain in HWE (Beebee and
Rowe, 2004). Deviation from HWE in a population may imply Wahlund effects
(sampling across populations) or null alleles (alleles that fail to PCR amplify) (Beebee
and Rowe, 2004). I used Arlequin 3.11 (Excoffier et al., 2005) to test stands for HWE
using an expansion of Fisher's exact test (Guo and Thompson, 1992). Six thousand
dememorization steps, to ensure the independence of results from the starting

configuration, and 150,000 MCMC steps were used. Bonferroni corrections for multiple
statistical comparisons (Rice, 1989) were applied at the stand level (.008333 = .05 I 6
loci) and for all pairwise comparisons (.000238 = .05 I 210 tests).

35

I tested for linkage disequilibrium to ensure that loci were independent. I used
some statistical programs that assume linkage equilibrium within populations.
Significant linkage disequilibrium can arise because of inbreeding or hybridization, but
also because loci are physically proximate on chromosomes. I used Arlequin to test
stands for linkage disequilibrium using an extension of Fisher's exact test (Raymond and
Rousset. 1995) with 6.000 dememorizations and 150,000 MCMC steps. Bonferroni
corrections for multiple statistical comparisons were applied at the stand level (.003333 =
.05 I 15 tests per stand) and for all pairwise comparisons (.000095 = .05 I 525 tests).

2.5.2 AI\JOV A

I used Arlequin to assess population structure using AMOV A (Analysis of

Molecular Variance). AMOVA tests a priori population structure in which stands were
defined based on location of collection (35 stands or populations). Creating an xdimensional matrix (x = number of loci). AMOV A calculates the genetic distance
between individuals. Total genetic variation is hierarchically partitioned into: among
groups; among populations within groups; among individuals within populations, and
significance is tested by permutation (Excoffier et al., 1992). Each AMOV A was run
with I 0,000 permutations at a .05 significance level.
Pairwise FsT values were also generated, as FsT is a good correlate of short-term

genetic distance between populations (Reynolds et al., 1983; Slatkin, 1995). The
calculation ofF-statistics has been refined since the pioneering work of Wright ( 1951)
but the theoretical basis remains. For FsT. deviations from Hardy-Weinberg Equilibrium
are quantified as:

36

FsT = (H,- H,) I H,
Where H, is the heterozygosity across populations and H, is the average heterozygosity
of a given population. Arlequin was used to calculate FsT values with Weir and
Cockerham's ( 1984) correction for unequal sample size. Using AMOVA, individual
genotypes are permuted between populations to assess significance. Pairwise FsT values
were calculated with I0,000 permutations (a=.05 ). However, tests other than AMOV A,
such as Nei's ( 1987) multi locus G-test of population differentiation, are also widely used
to determine the significance of pairwise FsT values. I thu~ validated the Arlequin
AMOV A pairwise FsT analyses by conducting the multi locus G-test included in the
program FSTAT 2.9.3.2 (Goudet. 1995) with 11,900 permutations (a=.05).

2.5.3 Structure
To strengthen the population structure inferred from the Arlequin AMOV A
results, I used an innovative population genetics program. Structure 2.2 (Pritchard et al.,
2000; Falush et al., 2003; Falush et al., 2007) does not test for population structure using
populations that are defined a priori, often by collection site. This intellectual advance
allows for the detection of cryptic population structures (eg. Rosenberg et al., 2002;
Harter et al., 2004). Structure is also well-suited to data from a range of genetic markers,
including microsatellites. Using a Bayesian approach, Structure groups individuals with

similar genotypes into clusters based on likelihoods of cluster membership.
I used a model with admixture and correlated allele frequencies. This was the

most biologically-realistic model for the MPB ; each individual MPB has likely inherited
a fraction of its ancestry from each of the K clusters, due to population admixture, while

37

population allele frequencies are likely to be correlated due to inter-population breeding
and/or shared ancestry. I did not use the linkage model as the incidence of linkage
disequilibria in my data was rare and apparently random. Population structure was
determined by examining Structure's output, in terms of stand membership, for a userdefined range of cluster values (K). Typically, population structure will be most
apparent at a lower K. Population structure will decline with increasing K as the
likelihood of individual membership in one cluster is diluted by membership in other
clusters. Runs were conducted forK-values ranging from 2 to 35. For each run, I 00,000
iterations of bum-in and data collection were performed. Multiple runs confirmed the
consistency of results. The program Distruct (Rosenberg, 2004) was used to convert
Structure results to a vi~ual format.

2.5.4 SAI\10VA
SAMOV A (Spatial Analysis of Molecular Variance; Dupanloup et al., 2002) was
used to confirm the population structure supported by the AMOV A and Structure results.
This program uses inputs of both GPS data, with one point for each collected stand, and
genotype data. Similarly to Structure (Falush et al., 2007), SAMOV A determines
population structure using only genotype data. However. while Structure assigns
individuals to groups, SAMOV A assigns stands to groups. SAMOV A additionally

considers but does not require stand membership in a group based on geographic
proximity. SAMOVA uses a simulated annealing procedure in which stands are
randomly assigned to groups until maximum differentiation between groups is achieved;
in AMOV A terms, maximum-differentiation entails optimizing the amount of genetic

38

variance due to differences between groups and is represented by the statistic FcT·
Similar methods, such as the Monmonier algorithm (Simoni et al., 1999), have been
shown to be less powerful than SAMOV A (Dupanloup et al .• 2002). The boundaries
between groups should correspond to genetic barriers. Runs were conducted forK
values from 2 through 15. To minimize the influence of initial configurations on the
results, the simulated annealing process was repeated 100 times for each run.
Molecular distances were calculated as pairwise differences. which are more
typical for DNA sequence data. as well as calculated as sums of squared size differences,
typical of microsatellite data. Both distance calculations were used for all runs to
compare outputs. However. sums of squared size differences were less appropriate for
my data set because 1/2 and 1/3 repeat allele size differences were present.
Microsatellite-based measures calculate distance based on number of mutation events but
my data set made assumptions regarding mutational events unclear. Moreover.
microsatellite-based measures are based on the assumption that allele size differences
and divergence are correlated (Balloux and Lugon-Moulin, 2002). This assumption is
unrealistic for many species (Shriver et al., 1993) and is often violated (Paetkau et al.,
1997; Zhang and Hewitt, 2003). Pairwise measures, in which degree of difference is not
considered, were a more conservative and realistic analysis choice. Population structure
was evaluated by the magnitude of FcT and the significance of the structure (Arlequin-

AMOV A). Multiple runs confirmed the consistency of results.

39

2.5.5 Isolation by Distance
Isolation by distance (IBD) is a correlate for population structure under a stepping
stone model. In IBD. the genetic distance and geographic distance between pairs of
populations is positively correlated. Thus nearby populations are expected to be highly
related and relatedness is expected to decline linearly with increasing distance between
populations. Genetic distances between stands (Arlequin- FsT) were plotted against the
straight-line geographic distances between stands. While some population genetics
studies have used more-derived IBD models (eg. Salle et al .• 2007), these models are
optimized for the estimation of demographic parameters from population genetic data
(Rousset. 1997). In my IBD analyses. I was looking for trends in my data~ I thus did not
seek to indirectly estimate parameters such as the number of migrants between
populations per generation. or population density. as examples.
In my IBD analyses, I did not consider the distribution of suitable host tree
species in the calculation of geographical distances between stands. Unlike Mock et al.
(2007). whose research area included the Great Basin and Mohave Deserts, I studied a
portion of the MPBs range where pine stands are largely contiguous.
Because each IBD data point was not independent, standard regression analyses
were not used. Instead, I used the Mantel test (Mantel, 1967; Smouse et al., 1986) within
Arlequin to test the significance of the relationship between genetic and geographic

distance between stands. The Mantel procedure tests the significance of the correlation
between two matrices, using a permutation procedure to account for auto-correlation.
This test was performed with 10,000 random iterations (a=.05).

40

2.5.6 Other Analyses
I conducted a number of other analyses at either coarse provincial or fine standlevel scales. MPB population genetic structure was overlain with maps of historic
distributions of climatic suitability classes for the MPB (Carroll et al., 2004). Two
related but different measures of the genetic diversity of stands, mean observed
heterozygosity and mean number of alleles. were plotted against the north-south distance
of stands from 49° latitude. the BCIUS border. to investigate possible north-south genetic
gradients. I hypothesized that north-south genetic gradients may exist based on two
ideas: the MPBs genome should reflect Ice Age patterns of refugial isolation and
subsequent recolonization of BC: moreover. the strongest abiotic influence on the MPB
is climate and climate follows a general north-south gradient in BC.
At a stand scale, the incidence of private alleles, alleles which are found in only
one stand. is a simple estimate of the genetic distinctiveness of stands (Kalinowski,
2004). Private alleles were tabulated by locus and stand. The mean pairwise FsT values
of unique stands, those that were significantly different from all others in AMOV A
analyses (Arlequin: a=.05), were compared with mean FsT values across all 35 stands for
evidence of population bottlenecks and/or founder effects.
I did not conduct statistical tests within Bottleneck (Piry et al., 1999) because at

least ten loci are required to obtain reasonable power. I did conduct per-stand tests for
mode-shift distortions of allele frequencies. Stable populations tend to have an L-shaped
allele frequency distribution as most alleles are rare and by definition exist at low
frequencies (< 0.1 0). Bottlenecks cause most rare alleles to be lost such that the

41

dominant peak in the aiJele frequency distribution shifts to an intermediate frequency (eg.
0.2 to 0.3; Luikart eta/., 1998).
Mean number of alleles and number of private alleles were not adjusted for
variation in sample size as statistics (mean n = 46; standard deviation= 5.17) did not
warrant adjustment. Outlier stands, McBride with 24 samples and Grande Prairie with
30 samples. were highlighted instead; these two stands were responsible for 60% of the
above standard deviation.

42

Results
3. I. Hardy-Weinberg Equilibrium and Linkage Disequilibrium
Tests for HWE at each locus for each stand revealed thirteen significant
deviations out of 210 tests using a stand-level Bonferroni correction and only three
deviations using a correction for all tests (data not shown). Of the former. ten deviations
were due to a deficiency of heterozygotes while three deviations were due to an excess of
heterozygotes. Using two levels of Bonferroni correction confirmed that deviations were
distributed across stands and loci. Given the stochastic and rare occurrence of
deviations, I assumed populations were in HWE for subsequent analyses.
Tests for linkage disequilibria between pairs of loci for each stand revealed
twenty significant deviations out of 525 tests using a stand-level Bonferroni correction
and only eight deviations using a correction for all tests (data not shown). Of the former,
most stands had one instance of disequilibria and only five (Banff. Argenta, Tatla Lake,
Kimberley, Golden) had two disequilibria. Using two levels of Bonferroni correction
confirmed that deviations were distributed across stands and loci; deviations remained
distributed after the more stringent correction was applied. The incidence of linkage
disequilibrium was highest between MPB35 and MPB40 (five deviations using the
former correction). Given the stochastic and rare occurrence of deviations, for
subsequent analyses I assumed populations were in linkage equilibrium at all loci.

43

3.2. AMOVA
AMOV A analyses revealed shallow but highly significant MPB population
structure in Western Canada (global FsT = .03828; P < .0000 I). FsT is the variation
among stands relative to the total variance. Nearly 81% of 595 pairwise between-stand
AMOV A tests rejected the null hypothesis and thus found stands were significantly
different. I created a map of genetic similarities where paired stands that were not
significantly different (FsT calculation with AMOV A; a=.05) were joined by lines
(Figure I). Based on my population genetic data, Figure I suggested the existence of a
Northern and Southern group with genetic similarities existing through the Fraser Valley
(defined in this Thesis as the portion of the Fraser River from Prince George to
Vancouver). Besides geographical separation. the groups were also characterized by the
incidence of genetic similarities. Similarities were sparse in the Northern group as
compared to the complicated "web" in the Southern group (Figure I). Notably, five
stands: Whistler. Mackenzie. Tumbler Ridge, McBride, and Willmore Wilderness, were
significantly different from all other sampled stands. A multi locus G-test conducted
using FSTAT was found to be less conservative than the Arlequin AMOV A method for
assessing the significance of pairwise FsT values (data not shown) . The multi locus G-test
found more significant differences between populations. Using the results from FST AT
(a=.05), I created another map of genetic similarities (figure not shown). This map

supported the AMOVA map, confirming the existence of: a Northern and Southern
group with similarities through the Fraser Valley; few similarities in the north and many
similarities in the south; and the same five unique populations.

44

Figure I. Instances of genetic similarity among 35 MPB stands in BC and western AB analyzed
at six microsatellite loci. Connecting lines denote that stands are not significantly different and
thus are genetically similar (Arlequin- AMOV A FsT: a= .05; 9999 permutations). Sampling
sites are represented as circles. Similarities imply some level of past and/or recent gene flow.
This population structure (Ha =all stands are a different population) was shallow but highly
significant (AMOV A- global FsT = .03828; P < .0000 I). The inset map denotes the names of
sampling locations.

45

3.3. Structure
Structure analyses gave the strongest support to the existence of two population
clusters (K=2). Population structure increasingly dissolved at runs with higher K-values.
Figure 2 shows the likelihood of membership in all clusters for each individual beetle,
grouped by stand. Runs with K=2 captured the most structure in my data and subsequent
runs at higher K-values. such as at K=3. quickly dissolved apparent population structure
(Figure 2). Clines of likelihood of cluster membership (K=2) were plotted onto a map of
stand locations (Figure 3 ). This Structure-derived map clearly supported the existence of
a Northern and Southern group. In Figure 3. the Northern and Southern group are
demarcated by a .50/.50 isocline of likelihood of membership in either cluster. In both
the Northern and Southern groups, stands farthest from the .50/.50 isocline had the
highest likelihood of membership in their respective group. Multiple runs yielded
identical likelihood-of-cluster-membership values for each stand. An AMOV A with
Arlequin determined that the Northern and Southern groups defined by Structure were
highly significant (FsT = .05543; FcT = .03665; both P < .0000 I). FcT is the variance
among groups relative to the total variance. The possible existence of groups defined by
the isoclines in Figure 3 was tested with an AMOV A in Arlequin but this five-group
population structure (FsT = .04420: FcT = .02947; both P < .0000 I), while significant,
explained less total genetic variation compared to the two-group Northern and Southern

population structure.

46

K=2

K=3

Figure 2. Population structure estimated using the program Structure. Each individual is
represented by a thin vertical lme that is partitioned into K coloured segments that represent the
individual s ltkelthood of membership m each of the K clusters. Individual likelihoods were
summed into a mean populatiOn ltkelthood of membership. Black vertical lines separate
populations. Explanation: at K=2 there are two groups (blue and orange); at far left Golden has
the highest likelihood of membership m the blue group; membership in the blue group declines
moving right · at far right Houston has the htghest likelihood of membership in the orange group.
At K=2 Structure partitioned the 35 stands mto a Southern (blue) and Northern (orange) group.
Notably, at higher K values, as represented on this figure by K=3, at far left Golden clearly
belongs to the orange group and at far right Houston clearly belongs to the blue group. However,
the membership of the remaining populations in the middle of the figure is approximately equally
shared among the three groups (yellow, blue, and orange)· thus, population structure dissolved in
Structure analyses using K-values higher than two.

47

0

0.35/0.65
0

Figure 3. Clines of likelihood of cluster membership (K=2) derived from a Structure analysis of
population structure among 35 sampled MPB stands in BC and western AB. Structure most
strongly supported the existence of two groups, Northern and Southern, which are separated in
this figure by a .50/.50 membership isocline; for each cline, likelihood of stand membership
values are for the Northern group on the left and for the Southern group on the right. Sampling
sites are represented as circles. The Northern and Southern group population structure (all stands
above the .50/.50 line in the Northern group and vice versa) was highly significant (AMOV AFsT = .05543; Fcr = .03665; both P < .0000 I).

48

3.4. SAMOVA

SAMOV A analyses both confirmed and refined my analyses using Structure.
Similarly to Structure, SAMOV A gave the strongest support to the existence of a
Northern and Southern group (K=2). However, SAM OVA refined the Northern group I
Southern group boundary (Figure 4), which Structure defined very loosely. The
Northern and Southern group population structure defined by SAMOV A explained the
most total genetic variation of all the population structure models (FsT = .05794; FcT =
.03449: both P < .0000 I). Hierarchical partitioning placed 94.2% of total genetic
variance within stands and most of the remainder (3.5%) was among the Northern and
Southern groups; 2.3% of variation was among stands within each group. SAMOV A
analyses at higher K-values did not rearrange stands among groups or split groups.
Instead, a single population was excluded with each successive run at a higher K-value.
Thus, the SAMOV A population structure at K=2 was most strongly supported by both
patterns of stand assignment as well as the maximal amount of total variance explained.
All further references to a Northern and Southern MPB group are specific to the
SAMOVA analyses.
Using pairwise and sums of squared size differences for genetic distance
calculations had no effect on SAMOV A population structures. I thus used results
derived from pairwise differences, which are more conservative. Multiple runs at each

K-value con finned the consistency of results.

49

Figure 4. Population structure derived from a SAMOV A analysis of 35 stands of MPBs sampled
in BC and western AB. SAMOVA most strongly supported the existence of two groups
(Northern and Southern); the boundary between these groups is the solid black line. Sampling
sites are represented as circles. This Western Canadian MPB population structure explained the
most genetic variation of all of my analyses (AMOV A- FsT = .05794; Fer= .03449; P < .0000 I).

50

3.5. Isolation by Distance
I found a highly significant and strong effect of isolation by distance (r =.55; P <
.000001; Figure 5). I then tested whether this IBD pattern was driven by the population
structure indicated by previous analyses. Specifically. I tested whether this strong IBD
pattern was due to the Northern group. the Southern group. or interactions between the
groups. IBD within groups only included within-group stand comparisons while IBD
between-groups only included comparisons between Northern and Southern stands.
Strong and significant IBD was found for the Southern group (r = .60; P < .00000 I;
Figure 6) and for the Northern group versus the Southern group (r =.31; P < .00000 I;
Figure 7). There was no IBD effect for the Northern group (r =.II; P = .6887; Figure 8).

51

02

•
018

016

•
•
• • •
Ffe03064
• •
B
• • •
•
012
•
• • • • •
•
••
•
••• • • •• •• •
• • ••
f 01 •
• •• t
ai
•• ••
•• •
•
,~ 008
•• ••• •
•
•
•
•
•
•
•
•
•
••
•
• ••
•• • •
·~ ~ 006
~··· •
• •
• • '··:
~·:. #
•
•
004
••
•••
••• •• •
002
~:"
••
0
•400
•
100
300
500
0
800
900
~

~

i
c:

014

-#.-.
..
'
..
.....
......
· .. ,.-... ..
.... .....
. ...,... .... .. ..
., . ·+\··.

~

1:5
.!.!

200

600

700

Geog~ 9raigli-Une Di91nl! ~tween PopUatlons (km)

Figure 5. Isolation by distance (180) pattern resulting from comparison of Fs 1-values and
straight-line geographic distances between all pairs of stands. Thirty-five MPB stands were
sampled in BC and western AB and analyzed at six microsatellite loci. This relationship was
highly significant (Arlequin- Mantel test with I 0,000 permutations; P < 0.00000 I).

52

1000

012

"S::>uthern" G'oup~ldentifioo with f!PMOVA

01

Ff =0.3649

~ 008

.~

i

•

.,

002

0
0

100

200

300

400

500

600

~~cal Sraiglt-Une [lS<Wl<l! B!tween PopulalionS(Im)

Figure 6. Isolation by distance (180) pattern re~ultmg from comparison of FsT-value~ and
straight-line geographic distances between all pairs of stands within only the SAMOV A-defined
Southern MPB group. This relationship was highl y significant (Arleq uin - Mantel test with
I 0,000 permutations; P < 0.00000 I ).

53

700

0.2

•
016

"Northern" vs"S>uthern" Qoup~ldentified
withf:PMOVA

016

~

~

I~

012

•

c:

01

t!i

!

•

• • •

014

.

;

A! =0.0975

•
• •
• • •
•
• • • • •
••• • • ••• •
• • ••
• •••
• • ••
•
•
•
•
••
•
••
•
•
•
•
• ••
•
• •

·-

..
.....
• • •
•
..
..
:. ...........
'
....
•
•
•
••
.
-·-.
-• •.....• ...••., ....
••

006

0

u

·~

& 006

• •• •

• ;... •.~+... ' •••
~

004

002

•

~·

• ••
#

0
0

100

200

300

• •

•• ~·

400

500

••

600

700

600

900

G!ogaphical Srcigrt-Une llsax:e ~ween ~ions(km)

Figure 7. Isolation by distance (180) pattern resulting from comparison of F51 -values and
straight-line geographic distances between all pairs of stands between only the SAMOV Adefined Northern and Southern MPB groups. This relationship was highly significant (Arlequin
- Mantel test with I 0,000 permutations; P < 0.00000 I).

54

1000

0.14

"Northa-n" G'oupas ldentifioo with '2/W.OVA
012

•
•

01

~

•
•

•
•

g

J

•

•
•

008

A' -o 0122

•

• • •

•

•
•

• •

002

• •

0
0

100

•

•
•
•• •

•

•

•

•

•

•

•

• •
•
300

200

•

•
•
•

.. •
400

•

•
500

•

Figure 8. Isolation by distance (180) pattern resulting from comparison of Fsr-values and
straight-line geographic distances between all pairs of stands within only the SAMOV A-defined
Northern MPB group. This relationship was not significant and a regression line is provided for
illustrative purposes (Arlequin- Mantel test with I0,000 permutations; P = .6887).

55

600

3. 6. Other Analyses
Overlaying maps of historical distributions of MPB climatic suitability classes
(CSC; Carroll et al., 2004) with my SAMOVA-derived population structure indicated
that climate is a likely driver of MPB population structure (Figure 9). The habitat of the
Southern group is typified by high or extreme CSCs. indicating that the Southern group
generally occupies a climatically-optimal environment. In contrast. the habitat of the
Northern group is typified by roughly equal proportions of extreme CSCs, from very
low/low to high/extreme. This pattern, and thus the concordance between historical
distnbutions of MPB CSCs and current MPB population structure, is weakest for the
most recent 30-year period ( 1971-2000; Figure 9). Current MPB population structure
has the highest visual concordance with the climatic suitability class distribution over the
period of I 92 I- I 950 (Figure 9). As the 201h century progressed, there was a major
expansion of climatic suitability into the regions west of and including Prince George;
however, M PB population structure did not reflect this change.

56

•

Very low

low

n Moderate
D High

Extreme

Figure 9. An overlay of current (2005/2006) MPB population structure and the geographic
distribution of regions that are chmat1cally favourable for the MPB (Left: 1921-1950 climatic
data; Right: 1971 -2000 climatic data; Carroll eta/., 2004). The SAMOV A-derived MPB
population structure is shown, in which the northern and ~outhern groups are demarcated by a
solid black line. "Very low" CSCs mdicate regwns wh1ch are climatically unsmtable for the
MPB while "Extreme·· CSCs mdicate regions climatically optimal for MPB~. CSC distribution
maps modified and reproduced with permissiOn of A. Carroll.

57

I found evidence for two north-south genetic gradients among stands. Mean
number of alleles per stand had a highly significant inverse relationship with the straightline north-south distance of each stand from the BCIUS border (49° latitude; Figure I0; r
= .81; P < .0000000 I). Mean observed stand heterozygosity also had a highly significant
inverse relationship with straight-line north-south distance of each stand from the BCIUS
border (49° latitude; Figure I I; r = .73; P < .00000 I). In these relationships, southern
stands had higher values for each response variable. intermediate populations had
intermediate values. and northern stands had the lowest values. Sample size biases
towards lower genetic diversity values were possible for McBride (n=24) and Grande
Prairie (n=30). Sample size was reduced m McBnde because MPB 30 amplified poorly,
possibly because of one or more stand-specific mutations in the primer annealing site.
However. since McBride was fully genotyped at five of six loci for 48 ~amples, at n=48
at five loci I calculated genetic diversity and found it was identical to diversity at n=24
for six loci. This suggests that a sample size of 24 genetically-independent beetles can be
representative of a MPB population and that the potential for sample size bias in Grande
Prairie was low.

58

7
•

Merritt

A2 : 0.6548
l<iiT'ber16'f

65

•
Ull ~ l<anloops

•

Ulke I.DUI!II
•
• WellsG'Itf

Valhalla

•

Prgenta

•

Val amount

•

Manning Pa-l<
Whllll~

Q)wronlale

+-

Chat wynd Ana Pass

•

P.rwell Olrlyon

•

• R'lflce Ge<Jge

Willmar~

T<ilalale

•

45
Ma;keruie e R>rt S BneS
Telkwa

•

•

Ffa'lOOisl.ake

•
G'me Prarle

4

0

100

200

400

300

~aphka Srai~-Une North-SJuth£l9axE

500

600

700

(km) of Saldsto 49"Laitude

Figure I 0. Inverse relationship between mean number of alleles and straight-line north-south
distance of stands to the BC/US border (49° latitude). This relationship was highly significant (P
< .00000001 ).

59

800

0.7

G:llden
~otenayf>wk

065

•

A-gent ~

Nlrocy
Qeene

•

•

• Klrnloops

F¥ = 0.5365

• l<imberlet

~
...

0"

IS'Iff .
• WellsQay

06

lf.

•

>

·~

Merrill •

~ningf>wk

t 055

•

f

~fmel

;•nceGeo-ge

Yahalla

!

• Tumbler
Rd!J!

Tlilalake

•

~

iQl

we;ount

f Who918f
Ullooet

Was .ltc'm•

05

:E

Oletwynd Are Pam

•

Willmon~

Fa-well Qrlyon

Mc&od~

•

•

Houston

Telkw'

Q<l'lde A'airie

0.45

•

Ma:ken21e

0. 4
0

100

200

400

300

500

600

700

Geogaphica Sni~ -lile Nort~axrt h DiS ll"l<E (km) o1 s Cl1ds to 49" La it ude

Figure II. Inverse relationship between mean observed heterozygosity and straight-line northsouth distance of stands to the BCIUS border (49° latitude). This relationship was highly
significant (P < .00000 I).

60

800

Tabulation of private alleles by locus and stand revealed the existence of 18
private alleles distributed among 13 stands (Table 2). Lake Louise had three private
alleles while three stands. Kimberley. Willmore Wilderness. and Merritt. had two private
alleles. Private alleles occurred singly in the remaining populations. In terms of
bottlenecks. I found no evidence for any distortion of allele frequencies within stands
(data not shown).

Table 2. Private alleles, by locus and population. for ~5 populattons of mountain pme beetles that were
anal \ zed at six microsatellite loci.

Hct

187
203
235
236
256
259
261
265
279
288
305

Willmore
Wilderness
Fort St. James
Ki

Hct
Het
Hct

I

2

Het
Hct
Het
Hct
Het
Hom
Hom
Het

247
252

Het

256
263

Het
Hom

-

61

A comparison of mean pairwise FsT-values between unique stands and the mean
across all 35 stands revealed some between-stand differences. Compared to a mean FsT
of 0.0398 across stands, Mackenzie (mean FsT=.I 051 ), McBride (mean FsT=.0899), and
Willmore Wilderness (mean FsT=.0625) had much higher mean values. Tumbler Ridge
(mean FsT=.0479) and Whistler (mean FsT=.0440) had values close to the across-stand
mean.
Analyzing the incidence of among-stand genetic similarities on a per stand basis
revealed some potentially important patterns. Genetic similarities between populations
(Figure I) imply recurrent inter-population gene flow. In terms of range expansion
within the current epidemic. both Grande Prairie and Chetwynd/Pine Pass were
consistent with coming from sources in North-Central BC. Grande Prairie was not
significantly different from Chetwynd/Pine Pass, Telkwa. Houston, Francois Lake, Fort
St. James, Quesnel, and Bowron Lake. Chetwynd/Pine Pass was not significantly
different from Houston, Fort St. James. Tatla Lake, and Quesnel. In the Northern group,
Telkwa and Houston had many instances of genetic similarities; thus gene flow involving
these populations may have been strong. In contrast, Fort St. James and Francois Lake
had fewer genetic similarities with stands and seem to have been involved in less interpopulation gene flow.
The Prince George/Quesnel region seems to be the hub through which most
similarities, whether ancestral or due to more recent dispersal, occurred between the
Northern and Southern groups. Most gene flow was between stands from the same
group. Thus stands in the Northern group were most likely to be similar to other
Northern stands and vice versa for the Southern group. However. there was an apparent

62

hub, defined by similarities between the two groups, at the boundary between them .
Prince George (PG) was genetically similar to Quesnel (QU) and Bowron Lake (BL), as
well as to three other Southern stands (Farwell Canyon. Lillooet, Lac La Hache) but only
one Northern stand (Telkwa). QU was genetically similar to BLand PG, as well as to
three other Southern stands (Farwell Canyon, Lillooet, Lac La Hache) and five other
Northern stands. BL was genetically similar to QU and PG, as well as to three other
Southern stands (Farwell Canyon. Lillooet, Lac La Hache) and two other Northern
stands. Notably, PG. QU. and BL were all genetically similar to only three other
Southern stands (Farwell Canyon. Lillooet. Lac La Hache). Farwell Canyon was an
intermediate stand by way of its similarity to the PG/QU/BL group in the north and the
Lillooet/Lac La Hache group in the south. By contrast, Tatla Lake was another boundary
stand but was only genetically similar to the Chetwynd/Pine Pass outbreak. The
remaining boundary stands in McBride and the Willmore Wilderness were significantly
different from all other stands.
"Connection" hub stands were common within the current epidemic area. By
labeling stands as hubs, I implied that these stands were focal routes for gene flow
involving recent and/or more ancient dispersal. Only stands that were characterized by
about ten or more instances of genetic similarity were labeled hubs. Numbers of
"connections" are given in parentheses. Lac La Hache (8) and Lillooet (II), Wells Gray
( 12), Lake Louise (I 0), Kootenay Park (II) and Kimberley (l 0), Valhalla (II) and West
Arm ( 15) and Nancy Greene ( 12), and Kamloops (I 0) and Merritt ( 12) and Falkland ( 12)
were all hubs within the current epidemic.

63

Several stands that were not significantly different from all other stands
nonetheless appeared to be comparatively isolated based on the incidence of genetic
similarities, with numbers of "connections" given in parentheses. These ten stands were
Fort St. James (4), Francois Lake (2), Tatla Lake (I), Farwell Canyon (5), Mount Robson
(5). Manning Park (4), Golden (4). Banff (2), Argenta (4), and Kelowna (3). Fort St.

James was connected by gene flow to Houston and Quesnel and may have contributed to
the new infestations in Chetwynd/Pine Pass and Grande Prairie. Francois Lake was
connected by gene flow to Telkwa and also may have contributed to infestations in
Grande Prairie. Tatla Lake was only genetically similar to Chetwynd/Pine Pass;
however. by a ranking of significant Fs1 values. Tatla Lake was next most related to
Lillooet, Farwell Canyon, Bowron Lake. and Quesnel. Farwell Canyon. as mentioned
above, was an intermediary stand connected to boundary hubs in Prince
George/Quesnei/Bowron Lake in the north and Lillooet and Lac La Hache in the south.
Mount Robson had only five connections compared to the neighboring Valemount stand,
which had 14 connections. Mount Robson was genetically similar to Valemount, Well s
Gray. Falkland, West Arm. and Kootenay Park. Given its mountainous location at the
edge of the MPBs Western Canadian range, Mount Robson was likely a sink for these
five regions. Manning Park was genetically similar to nearby stands in Lillooet, Lac La
Hache, Wells Gray, and Merritt. Golden was genetically similar to three stands in the
Kootenays (Nancy Greene. Valhalla. West Arm) as well as to Lake Louise. Banff was
genetically similar to only two stands, Nancy Greene and West Arm, both in the
Kootenays. Argenta was genetically similar to stands from only the Kootenays: Nancy

64

Greene, Valhalla, West Ann. and Kimberley. Kelowna was genetically similar to
Merritt, Valemount. and West Arm.
Relative comparisons of mean pairwise FsT-values (in parentheses) provided
mixed support for the comparative isolation of the ten stands noted above. Fort St. James
and Francois Lake appeared isolated based on their incidences of genetic similarity to
other stands, but the mean FsT-values (-.050) for these stands were consistent with other
northern stands. Farwell Canyon (.034) and Tatla Lake (.036) also appeared isolated
based on "connections" and this isolatiOn was supported by mean FsT-values higher than
nearby stands (Lac La Hache- .020. Lillooet- .0 18. Quesnel - .029, Bowron Lake.022). Mount Robson had a mean FsT-value (.030) slightly higher than neighboring
Valemount (.023) and Wells Gray (.023). Manning Park had a mean Fsrvalue (.030)
slightly higher than nearby Okanagan stands (Merritt- .025; Kamloops- .026). Kelowna
had a mean F51 value (.035) 40lk higher than neighboring stands in the Okanagan
(average for Merritt. Kamloops, and Falkland- .025). Similarly, Argenta had a mean
FsT-value (.044) -30% higher than neighboring stands in the Kootenays (average for
Nancy Greene, Valhalla, West Arm, Kimberley, Kootenay Park - .034 ). Both Golden
(.05 1) and Banff (.047) had mean FsT-values considerably higher than nearby stands:
Lake Louise (.036); Kootenay Park (.033); Kimberley (.036).
Each of the five unique stands that were significantly different from all other
stands exhibited distinctive inter-stand relationships (Fsr data not shown). Since all
genetic distance (FsT) values involving these stands were significant. these non-zero FsT
values could be ranked to determine levels of relatedness. Mackenzie was most closely
related to Tumbler Ridge (FsT=.035). Tumbler Ridge was most closely related to

65

Chetwynd/Pine Pass (FsT=.009), then to Lac La Hache (Fsr-.0 16) and Houston
(FsT=.O 16). McBride was most closely related to Fort St. James (FsT=.030), then
Houston (FsT=.039) and Grande Prairie (FsT=.058). In contrast, McBride's geographical
neighbor in the Willmore Wilderness was most closely related to Quesnel (FsT=.022) and
Bowron Lake (FsT=.026). Whistler was most closely related to Manning Park
(FsT=.013), then Kamloops (FsT=.017) and Lillooet (FsT=.020).

66

Discussion
4. 1. MP/1 Population Structu re in Western Canada
I found shallow but highly significant MPB population structure in Western
Canada (SAMOV A: FsT = .05794: P < .0000 I). This population structure was probably
superficially-shallow because I employed microsatellites, which are highly polymorphic
markers. The high mutation rates of microsatellites deflate FsT statistics (Balloux and
Lugon-Moulin. 2002). As an example. two subpopulations each having ten equifrequent
private alleles are clearly differentiated as much as possible and should have an FsT-value
of 1.0. However. high levels of polymorphism (due to mutations) result in an FsT-value
of only 0.053 (Balloux eta/., 2000). which is indicative of weak population structure. In
sum. seemingly small Fs1values do not imply negligible population structure (Wright,

1978).
Based on these inferences. the MPB exhibits strong and significant Western
Canadian population structure. This population structure was supported and
progressively refined by AMOVA. Structure. and SAMOV A analyses. This study was
the fourth to use microsatellite markers to infer the population structure of a Scolytid
beetle. However. of these studies, this was the first microsatellite analysis of Scolytid
population structure at a high resolution (many sites over a relatively small region).
Kerdelhue et al. (2006) used five microsatellite markers to study populations of the pine
shoot beetle (Tomicus piniperda) sampled from seven sites in France. Also using five
microsatellites, Salle et al. (2007) examined the phylogeography of the European spruce
beetle (Ips typographus) at 28 sampling sites across Europe. Using nine microsatellites,
Maroja et al. (2007) studied the phylogeography of the spruce beetle (D. rufipennis

67

Kirby) at 14 sites across its North American range. My study, in consideration of not
only the three studies above but also the past studies of Scolytid population structure, is
the highest-resolution study of Scolytid population structure to date.
My relatively high-resolution population genetic analysis of the MPB partitioned
stands into a Northern group and a Southern group. I had hypothesized that a larger
number of groups might exist in BC and western AB given that geographic barriers, in
the form of mountain ranges and/or distance, are pervasive and significant in Western
Canada. Geographical barriers are expected to isolate MPB populations. minimizing the
homogenizing effects of inter-population gene flow and allowing for the largely driftmediated divergence of populations into distinct groups. Based on detailed MPB
outbreak maps for BC and western AB covering the past century (Wood and Unger,
1996), I expected distinct MPB groups in the Rocky Mountains (Banff and Lake Louise),
the Kootenays, the Okanagan. the Chilcotin, the Robson Valley, and the NW region of
BC centered on New Hazelton. I also expected several groups in other regions of
Northern BC. However. my results provide no evidence for any of these expected
regionally-based MPB population structures.

68

4.1.1. Congruency Among the Phylogeography of the MPB, its Primary Symbiont, and
its Primary Host
I had hypothesized that the population genetic structure of the MPB may reflect
that of the species with which it has ostensibly associated over countless generations: its
primary symbiont. the pathogenic blue-stain fungus G. clavigera, and its primary host,
the lodgepole pine (Pinus contorta). Comparing the phylogeography of these species
revealed some interesting insights into their respective evolutionary ecology.
Recent research on G. clavigera found weak phylogenetic support for separate
northern and southern groups (Lee eta/., 2007). Though Lee eta/. (2007) only sampled
seven sites, their latitudinally-defined groups match with MPB population structure. In
both species, sites in Houston and Fort St. James were in the northern group and sites in
Manning Park. near Williams Lake, and in Banff were in the southern group. However,
Lee et al.'s (2007) most-supported finding was the significant subdivision of G.
clavigera population structure into distinct Interior BC and Rocky Mountain groups.

Based on the strongest available evidence, MPB population structure is apparently not
concordant with that of its main symbiont. Instead of the longitudinal two-group
population structure found for G. c/avigera (Lee et al., 2007), I found a latitudinal twogroup population structure for the MPB.
Differences in the population structure found by Lee eta/. (2007) for G. c/avigera
and the structure I found for the MPB are not readily explained by a compari~on of
methods. Lee eta/. (2007) employed six AFLP (Amplified Fragment Length
Polymorphism) primer sets, yielding a very large number of markers for analyses (469 in
total). DNA was isolated from 170 different cultures. These cultures were taken from

69

seven sampling sites, five in Canada (Houston, Fort St. James, Williams Lake, Manning
Park, Banff) and two in the United States. Given that their statistical analyses were also
robust, Lee eta/. 's (2007) research was as rigorous as my research. Both studies also
used selectively-neutral markers; neither study's results were confounded by natural
selection.
I hypothesize that the differences in province-wide population structure between
the MPB and G. clavigera are likely because of the different evolutionary pressures
acting on these species. These differential evolutionary pressures are evident when
comparing the genetic diversity of Rocky Mountain and Interior samples of the MPB and
G. clavigera. The genetic diversity reported for G. ( lavigera across the Interior and
Rocky Mountain groups was low (H, = .053). However. G. clavigera from Rocky
Mountain stands had higher numbers/Jevels of unique AFLP markers (38 out of 469;
Interior= 6 out of 469). mean genetic diversity over all loci (Rocky Mountains = .066;
Interior= .0435), and mean polymorphism (Rocky Mountains= 19.3%; Interior=
12.65% ). In contrast, I found that MPBs from the Rocky Mountains (Banff and Lake
Louise) showed levels of genetic diversity comparable to other stands within the southern
group. The incidence of private alleles suggested higher MPB genetic diversity in Rocky
Mountain stands compared to Interior BC stands; Lake Louise and Banff had four of the
18 private alleles found across all 35 stands. If private alleles were randomly distributed
among stands, then an average of -1 per pair of stands would be expected.
Unfortunately, the incidence of private alleles is significantly affected by sampling bias;
though my sample sizes had a mean of 46 and were consistent, I did not explicitly
evaluate this bias and thus my private allele results must be interpreted with caution.

70

In sum, G. clavigera forms a Rocky Mountain group that is strongly supported by
genetic data and has higher levels of genetic diversity than Interior samples. In contrast,
my MPB results support neither a Rocky Mountain group nor higher genetic diversity in
the Rockies. Overall, these results suggest that: I) in the Rocky Mountains. a more
ancient lineage of G. clavigera, compared to the Interior. has been maintained and has
retained considerable genetic diversity while some level of gene flow has occurred
between MPBs from Rockies and Interior stands for some time; and/or 2) as suggested
by Lee eta!. (2007), G. ( lavigera is a species complex and the Rocky Mountain and
Interior BC groups are different species; MPBs may thus carry different species of
symbionts characterized by regional adaptations; and/or 3) the mountain environment of
the Rockies places selective constraints on MPB symbiotic fungi. primarily on G.

clavigera, that are distinct from the constraints of typical Interior environments.
The last hypothesis above is particularly compelling in that it suggests strong
selection for novel strains of G. clavigera in mountain environments. Moreover, all three
of the above hypotheses imply that the Rocky Mountain and Interior BC strains of G.

clavigera are sexually incompatible and/or produce less-fit hybrids. This is important
because the effects of natural selection predicted by hypothesis three could not normally
be detected by the neutral genetic markers of Lee eta!. (2007) unless selection had
considerable effects on G. clavigera population structure.

Any selection for novel strains would be strong in that it would influence
mortality and reproductive success at multiple levels in the MPB/fungus symbiosis. For
the fu ngus alone, there shoul d be an evolutionary arms race for multiple traits which may
include: speed of coloni zation (inter-fungus competition for space and reso urces) and

71

synchrony of sporulation (for dissemination to beetle vectors). From the vantage of the
fungus-vectoring MPB, there should be a fungal pathogenicity arms race for traits that
stop resin flow expeditiously. which would increase beetle survival and fecundity.
Pathogenicity-related selection would act on both beetles and fungi, as their success is
all-or-none; unsuccessful beetle attacks ("pitchouts") usually cause unsuccessful fungal
attacks and vice versa (N. Bartell, pers. obs.). There should also be a fungal arms race in
terms of traits that enhance the provision of nutrients to beetle brood (dispersing beetles
with larger fat reserves should have higher survival and fecundity).
In a Rocky Mountain environment. however. there may be metabolic or life
history constraints on selection for the above traits; specifically. G. clavigera from the
Rocky Mountains may be unique because it is cold-adapted (Lee et al., 2007). Interior
G. clavigera may not face the same level of climatic constraints and thus may be
undergoing selection as above; indeed, the prevalence of G. clavigera strains from the
Interior BC group may be because these strains are more pathogenic and/or beneficial for
MPBs (Lee et al .. 2007). Thus G. clavigera populations endemic to different regions, the
Rocky Mountains versus the Interior, likely evolve very differently over the long-term.
Similarly to adaptations along altitudinal gradients, Rice et al. (2008) hypothesized that
G. clavigera may exhibit growth adaptations that are a response to latitudinal
temperature gradients across the MPBs North American range.

Regional evolutionary pressures may also have immediate, not just long-term.
effects on introduced G. clavigera individuals. Novel selective pressures in the Rocky
Mountains compared to the Interior are strongly supported by Lee et al. 's (2007) result
that although most Rockies G. clavigera samples were genetically tied to Interior

72

sources, Rockies individuals still exhibited higher genetic diversity; this result indicates
sink and not source effects. l suggest sink effects in which incoming G. clavigera from
the Interior underwent selection for higher genetic diversity in the Rocky Mountains.
Higher genetic diversity likely entails higher phenotypic plasticity; thus the Rocky
Mountains environment may favour G. clavigera strains with a greater ability to respond
to conditions atypical of the Interior region.
To complete my argument. l suggest that selective pressures on G. cladgera far
outweigh those on the MPB and that the pressures MPBs face are unlikely to differ
between the Rocky Mountains and Interior. Indeed. without virulent symbiotic fungi
MPB attacks are unsuccessful (Yamaoka et al .. 1995; Six and Paine. 1998). These fungi
play arguably the most important role in MPB ~urvival and reproduction in the host by
quickly stopping water conduction and thus resin flow. By halting water conduction,
these fungi stop constitutive defenses that may kill the MPB (and fungus) and also
prevent the development of induced host defenses. As mentioned previously, there
should be heavy selective pressure for increasingly virulent fungal strains that stop water
conduction more quickly. Such selective pressure likely acts on G. clavigera, the MPBs
most virulent symbiont (Solheim. 1995; Solheim and Krokene, 1998); G. clavigera is the
only MPB symbiont that is aggressive enough to kill lodgepole pine when introduced
alone (Yamaoka et al., 1995).
In sum. symbiotic fungi, primarily G. clavigera, are an important factor allowing
MPBs to successfully colonize their hosts. Selection pressures are likely much greater
for such fungi compared to the MPB because the MPB likely shows little variation in its
three main adaptations for successful host colonization. There are likely regional

73

selection pressures on G. clavigera where in the Rocky Mountains, selection is novel or
constrained compared to the Interior~ such selection on G. clavigera acts immediately
and in the long-term. MPBs are probably more likely to undergo direct selection for
traits that produce more brood, such as locating more suitable hosts or for higher
fecundity (eg. optimization of egg number versus size; Elkin and Reid, 2005), rather than
for fundamental traits involved in overcoming host defenses. If differential selection
does occur for the MPB in lntenor versus Rocky Mountain stands, then effects on overall
population structure (at neutral loci) have been minimal. I believe that the Northern and
Southern groups 1 have found for the MPB have arisen not by selection, but by 2 other
factors/events (see Section 4.1.5.).

The phylogeography of the MPBs primary host, lodgepole pine (Pinus contorta),
appears to reflect both deep and shallow history. Indeed, Marshall et al. (2002) found
post-Ice Age genetic patterns of a rapid recolonization (a "big-bang") starting 12kyr BP
that was overlain by the recent growth and differentiation of local populations ("galaxy
formation") in the past 1000-3000 years. This conclusion was supported by the results of
Fazekas and Yeh (2006) who notably found two population groups: northern BC/Yukon
versus southern BC/ Alberta/Idaho/Montana for P. contorta var. latifolia. The population
genetic structures of the MPB and its primary host, P. contorta var. latifolia, are
concordant in British Columbia. This is consistent with population genetic studies of the
beetles T. piniperda and T. destruens and their primary host trees, which have found
concordant beetle/host population structures in regions of Europe (Salvador et al., 2000;
Soranzo et al., 2000; Korol et al., 2002; Ritzerow eta/., 2004; Horn et a/., 2006;

74

Vasconcelos eta/., 2006). In sum, these results suggest that the MPB and its primary
host were intimately associated during recolonization. Since the MPB and P. contorta
var. latifolia have roughly congruent population structures, these species may be
evolving in response to similar factors, such as climate.

4.1.2. Significant MPH Population Structure- Context
The idea that bark beetles exhibit significant population genetic structure is
opposed by a portion of research on Scolytids, mcluding the MPB. These studies argued
or implied that Scolytids are characterized by consistent long distance dispersal, despite
the presense of significant geographic barriers. In some instances significant
methodological problems were likely. Stock eta/. ( 1984) and Calpas et al. (2002) failed
to genetically differentiate isolated MPB populations using allozyme and RAPD
analyses, respectively. Small sample sizes likely plagued both studies, while Stock et al.
( 1984) used a very conservative genetic technique and Cal pas et al. (2002) may have had

significant fungal contamination of isolated DNA. As an example, both my study and
Cal pas et a!.' s (2002) sampled from four of the same general areas but I was able to
genetically identify stands in Mount Robson, Falkland, Kootenay National Park, and
Banff.
The remaining studies that support an absence of Scolytid population structure
were likely biased. In the first study to use highly polymorphic genetic markers
(microsatellites) to resolve the population structure of a Scolytid, high levels of genetic
differentiation were expected between stands of the European spruce beetle Ips

typographus. Instead, Salle et al. (2007) found a complete lack of Europe-wide

75

population structure. As is often typical when non-significant population structures are
found, Salle et al. (2007) argued that I. tvpographus has a consistently high dispersal rate
but extended this argument to Scolytids in general. However. the potential for sampling
and contamination (fungal DNA) bias was high in this study; indeed, Stauffer et al.
( 1992; 1999) used allozymes and mitochondrial markers. respectively. and found
significant population structure for the same species over the same region.
Kerdelhue et al. (2006) also employed microsatellite markers but in an analysis of
the population structure of the pine shoot beetle, Tonucus piniperda . Though it could be
argued that T. piniperda actually had shallow but sigmficant population structure
subdivided into groups in northern France. central France. and the Pyrenees, Kerdelhue

et al. (2006) argued that T. piniperda exhibited a near-complete lack of population
structure. similarly to Salle et al. (2007). Prototypically. Kerdelhue et al. (2006) argued
that T. piniperda is characterized by frequent long distance migrations that genetically
homogenize populations. Like Salle et al. (2007), sampling and contamination bias were
potentially manifest in this study.
In sum. both Salle et al. (2007) and Kerdelhue et al. (2006) found weak
population structure in two beetle species using microsatellites. Past studies using lowerresolution mitochondrial markers have found more significant European population
structure in these species (Stauffer et al .. 1999; Kerdelhue et al., 2002). Indeed, many
population genetic studies find significant structure using mitochondrial but not
microsatellite markers. This trend is commonly caused by male-biased dispersal;
mitochondrial DNA is maternally-inherited and the development of significant
population structure at these loci is aided by a relative lack of female di spersal.

76

However, there is no evidence for sex-biased dispersal in Scolytids (Safranyik and
Carroll, 2006). The results of Kerdelhue eta/. (2006) and Salle eta/. (2007) are
suggestive of fungal contamination. Mitochondrial sequencing allows for postamplification contamination avoidance through the exclusion of sequences that are
similar to those reported for fungi. However. as suggested by data on the MPB (C.
Davis. unpub. data). microsatellite development for bark beetles most often results in
fungal-specific markers despite considerable efforts to avoid contamination of starting
DNA.
I suggest that the above two potentially contamination-affected studies amplified
fungal and beetle DNA. The microsatellite-derived population structures that Kerdelhue

et al. (2006) and Salle eta/. (2007) obtained probably reflect the population structures of
not only the beetle. but also the structures of various fungi. Indeed, neither study
conducted primer screening on fungal isolates to test for beetle specificity. Both of these
studies' data was in HWE but this would be expected with contamination; indeed, each
individual beetle would have randomly sampled a number of fungal strains of each
species before dispersal. The sum effect of beetle-fungus amplifications would be the
obscuring of any one species' population structure, as seen in both of these studies.
Indeed, as previously mentioned, the MPB and G. clavigera have incongruent population
structures; the population structure of either species would be obscured by DNA
contamination by the other. I suggest that significant population structures, concordant
with mitochondrial results, may be found for the two above species if microsatellites are
screened against fungal DNA isolates to confirm beetle specificity.

77

Patterns of population genetic structure between the MPB and its primary
symbiont, G. clavigera, are very supportive of significant population structuring. Lee et

al. (2007) found that all seven of their MPB-cultured G. clavigera stands were
significantly different from each other; this pattern was similar to my analyses of the
MPB. in which -80% of pairwise tests found significant differences between stands.
Clearly. neither the MPB nor its primary symbiont is panmictic over its range or even
within its genetically-defined groups. As confirmed by patterns of population structure
in an intimately associated fungus. Western Canadian MPB populations exhibit
significant genetic structure.
Most population genetic studies of the MPB have found significant population
structure. Stock and Guenther ( 1979), Bentz and Stock ( 1986), and Langor and Spence
( 1991) all used allozymes and found large differences among MPB populations across

the Pacific Northwest. western United States. and Western Canada, respectively. Kelley

et al. (2000) used allozyme and mitochondrial markers and found large differences
among MPB populations in California. The most recent phylogeographic study of the
MPB used AFLP and mitochondrial markers and found significant North American
population structure (Mock et al., 2007), with gene flow around rather than across the
Great Basin and Mohave deserts. On a per sampling site basis, I cannot compare my
results to past studies of MPB population genetics, which have not sampled BC and AB
to any large extent. However. on a theoretical level, these studies support the idea that
significant geographic barriers, such as mountain ranges and large distances. can prevent
gene flow and cause genetic differentiation between populations.

78

Even early population genetic studies that used conservative allozyme techniques
support the role of geographic barriers and isolation in Scolytid population dynamics.
Speciation may be occurring among specific. geographically isolated populations of the
Douglas-fir beetle (Stock et al., 1979) as well as in the southern pine beetle, D. frontalis
(Anderson et al., 1979; Namkoong et al., 1979). Roberds et al. ( 1987) and Six et al.
( 1999) have found concordance between geographic isolation and significant population
structure in D. frontalis and D. jeffrevi. respectively.
Population genetic studies utilizing more modem methods. such as mtDNA
sequencing, support the role of geographic barriers and isolation in Scolytid population
dynamics. In /. typographus. Stauffer et al. ( 1999) found significant Eurasian population
structure divided into three groups. Duan et al. (2004) confirmed the existence of a new

Tomicus species in southern China that was divergent from T. piniperda in Europe and
northern China. due to geographic barriers (Ritzerow et al., 2004).

T. destruens has well-supported population genetic structure with groups
throughout the Mediterranean basin, which is divided by such barriers as the Pyrenees,
the Alps, and the Mediterranean Sea (Faccoli et al., 2005; Hornet al., 2006). There is
also genetic structure within these groups, as in Italy (Faccoli et al., 2005).
Considerably fewer population genetic studies of Scolytids have occurred in
North America but these studies have confirmed the importance of geographic barriers
and isolation in Scolytid population dynamics. Most of these studies also employed
mitochondrial markers, while Maroja et al. (2007) used both mtDNA and microsatellites.
These studies found significant geographically-influenced population structure in: the
western pine beetle, D. brevicomis. (Kelley et al .. 1999}, the pinyon pine beetle, Ips

79

confusus, (Cognato et al., 2003), and the spruce beetle, D. rufipennis, (Maroja eta!.,
2007).
Many Scolytid population genetic studies that have been mentioned above as
supporting the role of geographic barriers in population divergence have supported such
roles over spatial scales considerab ly larger than the essentially one-province scale of my
study. These above studies (eg. Stauffer eta/ .. 1999: Ritzerow eta!., 2004: Maroja eta! ..
2007; Mock eta!., 2007) have often not found significant differences among Scolytid
populations separated by similar barriers (distance and/or mountains) to those separating
MPB stands that I found differences for. I suggest that these studies' inability to resolve
Scolytid population structure at a finer scale is a direct function of their sampling
intensity and the number and polymorphism of genetic markers they used. Increased
regional sampling and the u~e of a higher number of polymorphic markers may have
found fine scale Scolytid population structure. The lack of fine scale population genetic
studies of Scolytids is currently a major gap left to be filled in this field of research.

80

4.1.3. MPH Ge11etic Diversity- Colltext

Placing the genetic diversity I found among MPB stands into context was made
difficult by the possible fungal contamination of past microsatellite studies of Scolytids.
Observed genetic diversity had a range of 0.42 to 0.66 in the MPB. Kerdelhue eta/.
(2006) found a range of 0.63 to 0.70 among seven infestations ofT. piniperda in France;
however, one marker was suspiciously polymorphic. Salle eta/. (2007) reported a range
of 0.42 to 0.62 among 28 infestations of I. tvpographus in Europe. Unfortunately, as
previously mentioned, Kerdelhue eta/. (2006) and Salle et al. (2007) may have crossamplified beetle and fungal DNA. The probable contamination of these studies by
multiple strains of multiple fungal species may have kept overall .. beetle" genetic
diversity within a normal range. In the North American spruce beetle, Maroja et al.
(2007) found a range of 0.30 to 0.50 among 16 infestations.

81

4. I .4. Comparison of Beetle/Symbiont Genetic Diversity - D. ponderosae, D. jeffreyi,
and G. clavigera
My genetic diversity data on the MPB allow for an evolutionary, not a
phytogeographic (as previously detailed), comparison of bark beetles and their primary
mutualists. As previously mentioned, G. clavigera is the primary symbiont of D.

ponderosae and D. jeffrevi and is vectored only by these beetles. The genetic diversity
values reported in this section are Nei's ( 1973: 1987) gene diversity. which is equal to
observed heterozygosity for randomly mating sexual diploid organisms: these values are
appropriate for comparison between diploid/haploid and sexual/asexual organisms (Six
and Paine. 1999). Thus the genetic diversity values reported in this section can be
compared among beetles and fungi. MPH-associated G. clavigera exhibits low genetic
diversity ( H ~ = .053) over the same general range as my study: however. G. clavigera
from the Rocky Mountains has higher genetic diversity (H ~ = .066). I found high levels
of genetic diversity in the MPB (0.552) but Rocky Mountain stands were not auspicious.
This contrasts with the patterns of genetic diversity found for the G. clavigera mutualism
in the MPB sister species D. jeffreyi (Six and Paine, 1999; Six et al., 1999). Low genetic
diversity was found for both G. clavigera (0.0 14) and D . jeffreyi (0.040). The results for
the D. jeffreyi mutualism may have been an artifact of the use of genetically conservative
allozyme markers. However, using polymorphic mitochondrial loci, Kelley et al. (2000)

confirmed that D. jeffreyi has -10 times less genetic diversity compared to D.
ponderosae. Beyond marker properties, there are more logical explanations for the
discrepancies in genetic diversity between these two symbioses.

82

In both D. ponderosae and D. jeffreyi. the low genetic diversity of G. clavigera
may be because it is a haploid fungus. Although low genetic diversity in G. clavigera
may be explained by factors that influence both haploid fungi and diploid beetles, such as
dominance of genetic drift over mutation in small populations (Christiansen and
Feldman. 1986; Hartl and Clark, 1997; Bee bee and Rowe. 2004) or founder effects (Six
and Paine. 1999), many evolutionary processes, relative to diploid bark beetles, are
unique to haploid fungi. These unique processes include: immediate selection against
deleterious alleles (Perkins and Turner. 1988). low levels of sexual reproduction
(Harrington et al.. 1996). poor dissemination of sexual spores (Six and Paine. 1999},
and/or evolution of clonality to prevent the breaking up of high fitness genotypes
essential for mutualism (Wulff. 1985).
Given the MPBs wide host preference (Wood, 1982) and the high levels of MPB
genetic diversity I have found, the MPB likely did not suffer species-wide bottlenecks
during and/or after speciation. Though G. clavigera associated with MPB~ commonly
sexually reproduces (Robinson-Jeffrey and Davidson, 1968), such reproduction is
delayed relative to asexual reproduction (Six and Paine, 1996). Thus, MPB
dissemination of sexual G. clavigera spores (ascospores) may only be possible in
semivoltine mountain environments, such as the Rocky Mountains (Safranyik and
Carroll, 2006). In sum, poor dissemination of sexual spores, clonality, and the rapid
elimination of deleterious alleles may be the cause of low genetic diversity in MPHderived G. clavigera. In mountain environments such as Banff, the predominant
dissemination of ascospores may maintain higher levels of G. clavigera genetic diversity:
such diversity may maximize responses to selection (Six and Paine, 1999).

83

For the D. jeffreyi I G. clavigera mutualism, when D. jeffreyi diverged from its
common ancestor and became a monophagous beetle with a single-species host range,
the most parsimonious explanation is that both the beetle and its fungus suffered severe
founder effects that have not been rectified (Higby and Stock. 1982; Six and Paine. I 999;
Kelley et al .• 2000). Thus, in the D. jeffreyi I G. clavigera mutualism it is important to
distinguish between speciation-caused founder effects and founder-like population
dynamics (effects of population fluctuations) since speciation . Genetic diversity in D.

jeffreyi-associated G. clavigera has likely been restncted since founding by a rarity of
sexual reproduction (Six and Paine. I 999), as well as factors common to MPH-associated
G. clavtgera: clonality and the rapid elimination of deleterious alleles.
Thus the origin of low genetic diversity in D. jeffrevi is most likely a severe
bottleneck during speciation. Explaining the maintenance of low genetic diversity in D.

jeffreyi relative to the MPB, as they are sister species with similar population dynamics,
is more difficult. It is possible that bark beetle population dynamics, which are
characterized by temporally-dominant small populations and punctuated by widespread
population explosions, may be neutral with respect to the maintenance of population
genetic diversity. In other words, mutation/migration and genetic drift averaged over the
phases of the D. ponderosae I D. jeffreyi population cycle may be in equilibrium. The
diversity-reducing effects of genetic drift during the endemic phase may be offset by the
diversity-increasing effects of massive rates of mutation and migration (Petit eta!., 2005)
during epidemics.

o-

In terms of mutations, a microsatellite mutation rate of 6 X 1 4 per locus per
gamete per generation is standard (Hewitt, 2000). For every locus, there is a probability

84

of -1/14 that a mutation will occur in one of the 60 offspring of a MPB mating pair.
Based on Safranyik and Carro11's (2006) review. I assume that each male and female
effectively contributes 60 gametes ( 120 gametes/pair) and that there is one generation per
year (univoltine). Potentially enormous numbers of mutations are generated because of
MPB epidemics. For example. 1.000.000 matmg pairs yields 72.000 mutations for every
locus in the MPB genome. distributed among their 60.000.000 offspring. This is a rough
estimate of mutation generation. Microsatellite mutation rates vary significantly among
and within both species and loci (Zhang and Hewitt. 2003 ). Within loci. there are upper
limits for allele size (Paetkau eta! .. 1997); moreover. allele size affects both the rate and
direction of mutation at a given locus (Kimmel eta! .• 1998; Ellegren, 2000: Xu et al ..

2000).
When epidemics decline. I hypothesize that most new genetic variation is
maintained in myriad small MPB populations that are distributed over the landscape.
Drift is hypothesized to be heavy in these populations with small effective sizes. as
individuals and populations are extirpated at high rates, while some extirpated regions
are recolonized by migrants. If the population dynamics of bark beetles approximate
(mutation and migration)/drift equilibrium across population phases, then the genetic
diversity of Scolytid species reflects their diversity upon founding. I thus hypothesize
that the MPB, as a generalist mortality agent of Pinus, did not suffer species-wide
founder effects upon speciation and reflects ancient (high) levels of genetic diversity.
concordantly hypothesize that D. jeffreyi, as a host specialist, suffered severe specieswide bottlenecks during and/or since speciation (Six and Paine, 1999; Kelley eta/., 2000)
and reflects consequent levels of genetic diversity.

85

4.1.5. Western Canadian Ml'll Population Structure- A North ern and Southern
Group
I propose two explanations for the existence of a genetically-defined Northern
and Southern group for the MPB in Western Canada. These explanations involve: I)
post-Ice Age range expansion~ or 2) MPB population dynamics. Indeed. population
structure arises from the contribution of past history and current processes (Hewitt, 1999;
Vucetich and Waite, 2003).
There are two extremes of recolonization from refugia: "pioneer" and "phalanx"
(Hewitt, 1996). Both involve stepping stone dispersal with founder effects upon
expansion. These extremes differ in the speed of recolonization (Ibrahim et al., 1996).
Stepping stone dispersal involves spreading from the leading edge, where bottlenecks
reduce diversity (and increase homozygosity) upon the successive founding of pioneer
populations (Nei et al., 1975). "Pioneer" recolonization is very rapid. causing severe
founder events and thus high levels of homozygosity over the colonized range. In
contrast, "phalanx" recolonization is slower and allows more genetic diversity to survive
over the colonized range.
Because I argued that the MPB does not suffer founder effects (see Section 4.2.3),
I alter the above dispersal model in that for the MPB, bottlenecks likely occurred after
founding. This is because during recolonization, newly-founded MPB populations were
located at northern extremes where climate-induced bottlenecks were likely prevalent.
Short and severe climatic osci llations throughout the interglacial warming would have
caused additional bottlenecks (Hewitt, 1999). Thus, the MPB probably fits a non-

86

traditional model of stepping stone recolonization that is likely moderate with respect to
the "pioneer" and "phalanx., extremes.
Under a "moderate" model of post-Ice Age range expansion, I would expect MPB
populations to be the oldest in southern BC as these areas were recolonized first.
Population age (and genetic diversity) should thus decline with increasing north-south
distance to the BC/United States border. A hypothesis of post-Ice Age recolonization
further predicts that older populations in southern BC should be closer to mutation/drift
equilibrium compared to younger populations in northern BC. which should be heavily
influenced by genetic drift. causing reduced diversity.
This predicted genetic pattern is concordant with my results. A similar pattern of
southern richness and northern purity has been found in many taxa (Hewitt, 1999; 2004 ),
such as grasshoppers (Cooper eta/., 1995), newts (Wallis and Arntzen, 1989), whitefish
(Bernatchez and Dodson, 1991 ), and woodrats (Hayes and Harrison. 1992). Such
patterns are also found among taxa endemic to North America's Pacific Northwest (eg.
Green eta/., 1996; Conroy and Cook, 2000).
The resu lts of Mock eta/. (2007) confirm that the population genetic structure of
the MPB reflects post-Ice Age recolonization patterns to some degree. Indeed, Mock et

al. (2007) used other genetic markers (AFLP and mitochondrial sequencing), and thus
studied a different portion of the MPB genome, yet also found that genetic diversity
decreased with increasing distance north from an apparent core MPB range in southern
Idaho and Utah. Combined with my results, this indicates that a molecular signal of
post-Ice Age recolonization exists in the MPB. Indeed. the persistence of such signals

87

has been found for many European taxa (Taberlet and Bouvet, 1994; Lunt eta/., 1998;
Santucci eta/., 1998).
I suggest that the persistence of this signal may have been enhanced by the
probable lag in the MPBs recolonization of BC (Hewitt. 1999). The MPB did not follow
the retreating glacial edge but instead followed the advancing distribution of its host
trees, which likely exhibited time lags in recolonization. For example. the retreat of the
ice sheet did not necessarily create immediately available and suitable habitat for host
trees as there was also a treeless permafrost region in front of the continental ice sheet.
Once the ice sheets and permafrost had retreated from a region, the glacial substrate left
behind would have to drain. stabilize, and be initially colonized by pioneer species such
as lichens and grasses to promote soil "development" at the site; this process may have
taken hundreds of years at a site. Even then. host trees may not have been able to
colonize a recently glacially-abandoned site due to other factors.

MPB population dynamics may also explain the presence of a Northern and a
Southern group of MPBs in Western Canada. Climate is the most important source of
MPB mortality (Safranyik. 1978; Cole, 1981 ). Climate is typically more important to the
(overwintering) mortality of MPB brood in terms of early- or late-season cold snaps or
prolonged periods of -40°C temperatures in mid-winter (Yuill. 1941; Safranyik et al .•
1974 ). However, during the dispersal period, poor weather can cause significant MPB
mortality by causing asynchronous flights (Safranyik and Carroll. 2006).
MPB populations in southern BC are generally located in the most climatically
suitable region of BC (Carroll et al., 2004). Thus southern BC populations are likely

88

larger and more stable in time and space compared to populations from northern BC.
This inference is supported by both general landscape patterns of demography and the
incidence of MPB outbreaks in BC during the past century. Populations nearer the core
of a species' range (southern BC) have higher abundance and lower temporal variability
in abundance compared to peripheral (northern BC) populations (Vucetich and Waite.
2003). Given lower effective population sizes. Vucetich and Waite (2003) estimated that
rates of genetic drift may be two to thirty times higher in peripheral (northern BC)
populations.
Perhaps over 90lk of MPB outbreaks over the past century have occurred in
southern BC (Wood and Unger. 1996). Larger and more stable southern BC populations
will retain and generate considerable genetic diversity as these populations undergo
fewer and less severe bottlenecks. Also, as historical outbreak maps suggest, in southern
BC populations are more numerous and outbreaks are more common compared to
northern BC. The high degree of inter-population breeding among populations in the
south would maintain and increase levels of genetic diversity. Thus, a hypothesis of
population dynamics predicts that: southern BC has larger. more numerous MPB
populations in which outbreaks are more frequent, causing higher genetic diversity
compared to the smaller, fewer MPB populations in northern BC. where outbreaks are
more rare. The predictions of the population dynamics hypothesis are also concordant
with my results.

89

4.1.6. Western Canadian MPJJ Population Structure- Inferred Patterns of Post-Ice
Age Recolonization and Population Dynamics
I suggest that current MPB population structure is the result of a complex synergy
between hypotheses of post-Ice Age recolonization and MPB population dynamics. It is
difficult to explore these questions without further advances in genetic marker
development and in statistics. However. since my results reflect patterns from both
hypotheses, patterns may be explained by either or both of these hypotheses.
Notably, north-south gradients in the genetic diversity or allele frequencies of
MPB populations have been reported by past studies. To this author's knowledge, such
gradients have not been seriously considered until now. Though they did not report this
result, plotting the genetic diversity values found by Mock eta/. (2007) on a map of their
sampling sites revealed that genetic diversity decreased with distance north or south from
an apparent core range in Idaho and Utah. Stock eta/. ( 1984) found a north/south
gradient in allozyme frequencies among MPB populations in Idaho. At two polymorphic
allozyme loci, Kelley eta/. (2000) found that the genetic diversity of MPB populations
decreased with increasing distance south from northern California. The reduced genetic
diversity of populations at both the northern and southern extremes of the MPBs range is
likely due to post-glacial recolonization and population dynamics. The effects of these
two factors have been detailed for the northern extreme in Western Canada. In the
southern extreme, recolonization effects would be similar but climate would have
adverse impacts on population dynamics in terms of multiple, poorly-synchronized
flights per year (Safranyik and Carroll, 2006).

90

Thus the MPB likely survived in a single refugium located in the "core range" of
Idaho and Utah that I inferred from Mock et al. (2007). In this core range, MPBs may
have existed continuously over tens or hundreds of thousands of years, despite repeated
glacial advances and retreats. Post-Ice Age recolonization thus occurred from this core
range and moved north into BC. The decreasing genetic diversity from Oregon into
California and from Utah to Arizona may have been caused by recolonization from the
core range; however. it is possible that MPBs also persisted near the southerly portions of
its current range during the last glacial and have suffered bottlenecks due to the warming
and the increasingly fragmented host distribution of the current interglacial (Hewitt,
2000).
In terms of the last Ice Age. the MPBs primary refugium was either dominantly
or solely populated by Pinus contorta var. latifolia (Fazekas and Yeh, 2006), the
lodgepole pine subspecies from which I exclusively ~ampled MPBs. My results suggest
that the MPB repopulated Western Canada from a single refugium . Following the last
retreat of continental glaciers. the region occupied by the southern MPB group was
recolonized first. Recolonization of Northern BC appears to have occurred through the
Fraser Valley. I found no evidence for recolonization of Northern BC via the Rocky
Mountain Trench (through Valemount and McBride) or via the Chilcotin region of
western BC; such a pattern may have been maintained by the discontinuous Pinus
distributions in these regions.
I found no genetic evidence that the MPB persisted in a refugium north of the ice
sheet as there was no apparent hybrid zone, where colonists from different refugia meet,
within the region occupied by the Northern group. I also found no evidence for other

91

significant MPB refugia south of the ice sheet; no hybrid zone was apparent within the
region occupied by the Southern group. The existence of a single major MPB refugium
during the last glaciation is confirmed by modern population genetic research on the
MPBs primary host, P. contorta var. latifolia (Fazekas and Yeh, 2006).
Modern research appears to confirm a single refugium south of the ice sheet for
P. contorta var. latifolia. This is the most widely distributed subspecies of lodgepole

pine. Fossil pollen data (MacDonald and Cwynar. I 985) indicated that recolonization by
P. contorta var. latifolia progressed northwest from southeast BC to the Yukon . There is

no genetic (Marshall eta/ .. 2002; Fazekas and Yeh. 2006) or fossil pollen (MacDonald
and Cwynar. I 985; Cwynar and MacDonald. I987) evidence that any of the four
lodgepole pine subspecies persisted in Beringia. However. Pleistocene persistence in
refugia along the West Coast of BC. especially in the Queen Charlotte Islands region, has
been supported for P. contorta var. contorta, the shore pine (Heusser. I 960; Peteet, I 99 I;
Fazekas and Yeh, 2006). Since the MPB currently infests all P. contorta subspecies, I
could safely assume that the MPB persisted in West Coast refugia. However, an
evidence-based assumption is possible.
Importantly, Fazekas and Yeh (2006) found no genetic evidence for the
subspecies designations of P. contorta. They suggested that before the current
interglacial and its subsequent recolonization, P. contorta was a panmictic species.
Historical panmixia is suggested by the apparent gene flow between var. latifolia and
var. contorta in northern BC and the Yukon (Wheeler and Guries. I 982; Fazekas and
Yeh, 2006). Thus, P. contorta has probably existed as a single species (not a species
complex) since the last glacial.

92

Since all refugia populated by P. contorta were MPB-suitable. I suggest that the
MPB persisted in var. latifolia in the major refugium south of the ice sheet and that the
MPB persisted in var. contorta in several minor refugia on the West Coast and Queen
Charlotte Islands of BC. The coastal refugia would have been minor for the MPB on
several fronts: shore pine stands have low stem density and uneven age distributions
(long-term effective MPB population sizes would be small); the geographical extent of
these refugia was probably highly constrained; the extirpation of MPB populations was
more probable because West Coast refugial populations were isolated from the main
refugium (the ice sheet covered continental BC and Vancouver Island) and likely from
one another (fiords, mountains, islands. straits). Though the vast majority of MPB
recolonization of BC appears to have originated from the southern refugium. coastal and
inland refugia may have contributed to the MPBs recolonization of northwest BC, such
as via the Skeena or Nass River valleys (Fazekas and Yeh, 2006). My MPB populations
in this region. near Telkwa and Houston, indicate that these have arisen from the major
southern refugium. However, over the past I00 years MPB outbreaks have occurred in
the New Hazelton region, where outbreaks have been absent during the current epidemic.
Efforts should be made to sample MPBs endemic to northwest BC, or to obtain museum
specimens, to make a phylogeographic comparison with beetles that have recolonized
from the southern refugium. MPBs endemic to Vancouver Island should also be sampled
for similar phylogeographic comparison.

In terms of population dynamics, the "mess" of instances of genetic similarities
between stands in the Southern group (Figure I) is likely representative of postglacial

93

recolonization and thousands of years of inter-population breeding since recolonization.
I found a strong and very highly significant pattern of isolation by distance in the south .
Such a pattern is strongly indicative of stepwise (short distance) movements between
populations over thousands of generations. Southern BC is dominated by north-south
geographic barriers in the form of mountain ranges, such as the Monashees, Selkirks, and
Purcell s. Given the pattern of outbreaks in the south over the past century {Wood and
Unger. 1996), it is likely that in any one year or period, one population will erupt and
may breed with an adjacent population/s. Such a pattern of outbreaks is the most
probable contributor to my overall results in the Southern group. However. some of the
genetic similarities in the south may have arisen or been enhanced by widespread
outbreaks, such as the severe Cariboo-Chilcotin outbreak in the 1980's. No other severe
outbreaks occurred in southern BC during the past century {Wood and Unger. 1996).
Unfortunately. MPB outbreak incidence data predating the twentieth century are not
available.
The sparse number of connections in the Northern group seems to imply that long
distance dispersal events, which are known to have occurred within the current epidemic,
are important in this region. I found that isolation by distance was absent in the north; a
dominance of long distance dispersal would disrupt the correlation between gene flow
and geographic distance upon which IBD is predicated. The lack of IBD in the north

also suggests strong genetic drift. As previously detailed, populations in the north are
probably more prone to bottlenecks and probably have fewer opportunities for interpopulation breeding. In sum, in the north, long distance dispersal events are more

94

prl'\aknt than in thl' '-Ollth ;111d arl' ro JJil\\L'd hy j1l' riod-, o r '-trllll~ ~L'nL· tiL· drill that i-,
dJaractl'ri-,tiL· ol al l northnn popu lati(\n'- .

4.2. The Spread of the Western Canadian MP/l Epidemic
MPB outbreaks may arise in two relatively distinct ways. Outbreaks can arise
locally from the expansion of native, endemic-phase MPB populations in response to
favourable ecological conditions. Outbreaks can also arise via migrants, where
expanding local populations supply long distance dispersers, as often occurs during MPB
epidemics. In a spatiotemporal analysis of the current epidemic. Aukema et al. (2006)
found evidence for both. Indeed. their findings supported the occurrence of long distance
dispersal events in Northern BC and also indicated that many parts of the epidemic arose
from the expansion of native MPB populations. especially in Southern BC.
My results do not support the widespread perception that the current MPB
epidemic has spread via mass migrations from an epicenter in Tweedsmuir Provincial
Park. The Houston stand (sampled in 2006) was likely the most representative of the
Tweedsmuir populations. This stand was geographically closest to Tweedsmuir.
Moreover. based on aerial survey maps. Houston was overrun in summer 2004 by a mass
of beetles coming from the south (the Tweedsmuir region) and/or southeast. Contrary to
the Tweedsmuir epicenter hypothesis, most stands were significantly different from
Houston. Indeed, I found no evidence for the homogenization of population allele
frequencies that typically occurs when epidemics spread from epicenters (Namkoong et

al., 1979). Each sampled stand was significantly different from the majority of the other
stands; despite the incidence of genetic similarities among stands, I was able to use these
"connections" to comparatively characterize and distinguish stands.
A time series of outbreak maps of the MPB epidemic indicates that. especially in
southern BC, outbreaks have erupted simu ltaneously in isolated regions (Aukema eta/.,

96

2006). This observation is strongly suggestive that outbreaks in these regions were
initiated by the climate-induced expansion of native MPB populations. I thus suggest
that MPB populations are endemic to the Whistler Valley. Manning Park. various regions
of the Okanagan Plateau. the west Kootenays near stands at Nancy Greene and West
Ann Parks, the east Kootenays near Kimberley. Kootenay National Park. GoldenlY oho
National Park. Valemount, Mount Robson Park. and McBride. In some of these regions
(eg. the Okanagan and the Kootenays). outbreaks seem to have undergone massive
spatial expansions that may be due to significant MPB migrations. contributing to the
population genetic patterns that I found. Inferences regarding the endemism of MPB
populations in Northern BC are made difficult by the apparent dominance of long
distance dispersal events.
In sum, I suggest that most outbreaks in the current epidemic were initiated by the
expansion of MPB populations endemic to various regions of BC and AB . Massive
dispersal events in the years following these initial expansions. especially in NorthCentral BC. have likely caused the large severity and scale of current outbreaks. The
general dispersal patterns I have inferred represent the main factor (massive dispersal
events) in the rapid expansion of MPB infestations. I argue that southern BC is
dominated by significant geographic barriers and that dispersal events are typically
stepwise (not long distance) through weaknesses in these barriers, such as mountain
passes. The long distance dispersal patterns I have inferred should be fully-applicable to
future MPB epidemics. However, I strongly argue that in the absence of long distance
MPB dispersal, the isolated expansion of native MPB populations. especially in southern
BC. would have eventually caused the severity and scale of the current epidemic. The

97

prolonged expansion of isolated native populations may simply make them more
susceptible to climatic events.

Given the above general inferences. I have also inferred main dispersal patterns
throughout the current epidemic (Figure 12). As previously mentioned. genetic
similarities ("connections.. ) between stands most likely reflect: I) ongoing gene tlow, or
2) recent gene flow (dispersal) with insufficient time for divergence.

DISCLAIMER: The movement patterns I have inferred are where beetles are
dispersing and attacking suitable host species. MPBs may be dispersing into sinks
lacking suitable host species; these movements do not contribute to the propagation of
epidemics nor do these movements contribute to the persistence of MPB populations. It
is important to explain my rationale in terms of how I inferred general MPB dispersal
patterns (Figure 12). I used my population genetic patterns primarily, and Jackson et

al. 's (unpublished) wind modeling data secondarily, to establish general directions of
dispersal (eg. Okanagan Plateau to the Kootenays; Figure 12). I then used the incidence
of infestations on outbreak maps to discern finer-scale. landform-influenced dispersal
directions when possible. Although stands are referred to as place names (eg. Prince
George), they represent the general area indicated by the place name.

98

Figure 12. Map of inferred historical MPB dispersal routes (solid lines) in BC and AB. These
dispersal routes were inferred from: I) generalized patterns of genetic similarities among 35
MPB stands analyzed at six microsatellite loci; 2) atmospheric wind patterns (Jackson et al.,
unpublished data); 3) landform influences using a time series of aerial survey maps of the current
epidemic. Note that dashed lines are movements that were suggested by genetic data but were
not apparent from aerial survey maps of the current epidemic. DISCLAIMER: These
"movements" represent an educated best guess using primarily my genetic data. They may have
occurred within the current epidemic or sometime in the past; statistical methods cannot discern

between recent and historical dispersal at this time.

99

4.2.1. Spread of the Epidemic- Southern Group
The Southern MPB group was characterized by isolation by distance; this is a
genetic pattern that takes thousands of generations to develop (Johnson eta/., 2007).
Thus, in the Southern group, inferred movements do not necessari ly represent recent
dispersal but reflect thousands of years of stepwise movements among populations.
History lends credence to my assertion that the movements I have inferred in the
Southern group are the most likely routes for MPB dispersal during any epidemic.
Geographic barriers seem to have played a major role in "directing" MPB
dispersal throughout the epidemic area in Western Canada. Given my genetic data, many
MPBs in the Kootenays and southern Rocky Mountains (the lcefields area south to the
Lake Louise/Banff region) have probably come from regions to the west. Specifically,
MPBs from Lillooet and the Okanagan (Kamloops, Merritt, Falkland) appear to be
penetrating through weaknesses (passes) in the Monashee Mountains, in the vicinity of
Highway 6, into the Columbia River/Arrow Lake vaJiey. From this valley MPB~ are
likely simultaneously moving across part of the Selkirk Mountains into the next valley,
the Slocan Lake valley, which is represented by the site in Valhalla Provincial Park, as
well as moving through the river valley leading to Nelson, as represented by the site in
West Arm Provincial Park. The site in Nancy Greene Provincial Park is only -I 0
kilometers south of the Lower Arrow Lake valley and of the sites, is the most
representative of the MPBs penetrating the Monashees. There is no genetic or outbreak
map evidence for MPB movements south through Kelowna and the Okanagan Lake
valley, then east through weaknesses in the Monashee Mountains along the BCIUS

100

border (though I did not sample south of Kelowna or along the BCIUS border in the
Monashees).
From sites in the Monashee and Selkirk Mountains, MPB movements appear to
have gone around as well as through the bulk of the Purcell Mountain Range . The
former movements likely followed the Kootenay Lake Valley south through Creston,
where there is a major pass through the Purcell Mountains into Cranbrook and the Rocky
Mountain Trench. However. as inferred from the location of MPB infestations within the
current epidemic. many MPBs from the Monashees and Selkirks have penetrated a large
weakness in the Purcells. a west-east mountain valley east of the community of Gray
Creek on Highway 3A. This vaJiey has an acttve logging road on it. the Gray Creek and
Redding Creek forest service roads. and the road/valley apexes at 2000m above sea level
( -6600 feet) before descending eastward to the Trench.
Once in the Rocky Mountain Trench, I suggest that MPBs have moved north
from the site in Kimberley (near Cranbrook) to Kootenay National Park, then through a
major pass in the Rocky Mountains (which Highway 93 follows) to the Icefields
Parkway and north to Lake Louise (but apparently not south to Banff). I also suggest
that MPBs have dispersed long distances northwest up the Rocky Mountain Trench,
apparently bypassing the Golden site, along Kinbasket Lake to the Valemount area.
However. annual MPB outbreak maps do not support recent movements from the Golden
area of the Rocky Mountain Trench to the Valemount area. My genetic data may reflect
connections between Golden and Valemount that occurred during past outbreaks.
MPBs from the Lillooet and the Okanagan (Kamloops, Merritt, Falkland) area
have also apparently dispersed up the Thompson River valley into WeJis Gray Provincial

101

Park and into the Valemount area. MPBs from the Lac La Hache area are suggested to
have dispersed eastward and contributed to movements into Wells Gray and eventually to
Valemount. Again, concerning Valemount, dispersal inferred using outbreaks maps of
the progression of the current epidemic conflict with dispersal inferred from my genetic
data. Annual outbreak maps of the current epidemic do not support recent MPB
dispersal up the North Thompson Valley past Wells Gray Provincial Park. Thus.
historically, MPBs moved up the Thompson River valley north from the Okanagan to the
Valemount area. I also infer long distance MPB movements northward from the lower
Fraser Valley (Lillooet and Lac La Hache) to the Quesnel/Bowron Lake/Prince George
region. where the latter group is apparently the "connection" hub between the Northern
and Southern MPB groups.
The Banff site appeared to be relatively isolated from neighbouring stands
throughout the Rocky Mountain region. Banff was only genetically similar to two sites
in the Kootenays (Nancy Greene and West Arm). Indeed, independent but severe
outbreaks have occurred in the Banff area over the past century (Wood and Unger, 1996).
This result contradicts population genetic results for the MPB s primary symbiont, G.

clavigera. In their phylogeographic analysis of G. clavigera from MPBs, Lee et al. 's
(2007) data was strongly indicative of historical one-way gene flow of fungi (and thus
MPBs) from the BC Interior to the Rocky Mountains. I hypothesize that historically,
native Rocky Mountain MPB populations are isolated from Interior populations;
however, periodic outbreaks in and gene flow from the Interior are sufficient to create
genetic similarities between Rocky Mountain and Interior stands. This hypothesis is
consistent with genetic similarities between Banff and the two stands in the Kootenays.

102

4.2.2. Spread of the Epidemic - Northern Group
In Northern BC, based on historical outbreak records (Wood and Unger, 1996),
Chetwynd/Pine Pass, Tumbler Ridge, and Grande Prairie represent new MPB
populations. For example. Grande Prairie arose from a massive MPB dispersal event in
the summer of 2006, which witnesses described as "beetle rain." My MPB data indicate
that the MPB populations in the Peace River area largely represent an expansion of the
Northern group. It should be noted that Grande Prairie (2007) and Chetwynd/Pine Pass
(2006) were genetically similar but were sampled in different years (given in
parentheses). Thus the beetles represented by Chetwynd/Pine Pass had one year
(summer '06) to potentially contribute to 2007-sampled infestations in the Peace River
area.
However. it should be noted that there were genetic similarities between both new
populations in the Peace River area and the northern populations of the Southern group,
specifically in the Quesnel/Bowron Lake region. Thus, the northern portion of the
Southern group may have contributed dispersers to the new infestations in the Peace
River region. Also, Chetwynd/Pine Pass was the only stand in this study that was
genetically similar to the Tatla Lake stand in the Chilcotin region. If the similarity to
Tatla Lake was not spurious, then some of the MPB colonists of Chetwynd/Pine Pass
came from the Chilcotin region nearly 500 kilometers away (straight-line). This result
has implications in terms of the massive Cariboo-Chilcotin MPB outbreak in the 1980's
in that this outbreak may have created new MPB populations through long distance
dispersal events (up to 500km) to various regions of North-Central BC. In a final note,
as shown in Figure I, there were also genetic similarities from the Telkwa/Houston

103

region through the Bulkley Valley to the Prince George/Quesne]/Bowron Lake region.
However. it is difficult to infer directiona1ity and to date the implied gene flow; these
similarities may have arisen through Ice Age recolonization from the Prince George
region and/or from gene flow between the regions which did not necessarily occur during
the current epidemic.

4.2.3. The Population Genetics of Long Distance Ml'll Dispersal
Though MPB populations in Chetwynd/Pine Pass and Grande Prairie were newly
founded, I did not expect severe founder effects. I expected these populations to retain a
considerable portion of their ancestral gene poo]/s. My data support these expectations.
These stands exhibited levels of genetic diversity comparable to other Northern group
stands that are almost certainly native (eg. Houston, Telkwa). As noted, the founding of
the Grande Prairie population was an exceptional event that involved potentially billions
of MPBs from throughout the north. Chetwynd/Pine Pass was founded less auspiciously.
I hypothesize that the founding of Chetwynd/Pine Pass, in terms of the number of

dispersing beetles, is typical of the long distance MPB dispersal event. For long distance
dispersing MPBs to establish a population, their most basic requirement (besides suitable
hosts) is to land in sufficient numbers. Jackson et al. (2008) conducted airplane mounted
MPB-capture transects near Prince George and estimated: I) an average of 4950 beetles

"are dumped" per ha per year by long distance dispersal in the Centra] Interior during
epidemics of the current severity and scale; 2) -577 beetles are required to successfully
mass-attack a mature lodgepole pine; 3) in leading stands, a conservative average of 8.6
trees will be infested per ha. This prediction, -8.6 infested trees per ha, was surprisingly

104

congruent with the infestation intensities I observed in Chetwynd!Pine Pass and in many
of the other 35 sampled stands. Thus, if new MPB populations are founded, the required
numbers of MPBs preclude significant founder effects. Although long distance dispersal
events may involve one or several source populations, in which the latter involves mixing
of population genotypes, I suggest that the numbers required to found a new population
also preclude sampling effects as long as a sufficient number of beetles (-50) are
sampled.

4.2.4. Concordance with Atmospheric Wind Data
Jackson eta/. (2008) conducted aerial MPB-capture transects at various altitudes
during the dispersal season and determined that MPBs in the atmosphere are found up to

-850 m above ground level. Thus the atmospheric particle modelling that Jackson eta/.
(unpublished data) conducted, where each particle is equivalent to a beetle, used site
climatic data up to 850 m above ground level. The wind patterns reported below are
generalized from site-specific modelling and are for this most-relevant atmospheric layer.
Jackson et al.'s (unpublished data) model simulations help explain both recent long
distance dispersal events as well as trends in historical movements.
Generally in BC, winds are westerly or southwesterly. Thus, MPBs involved in
long distance dispersal events typically move east or northeast. However. there are
exceptions to this pattern. At modeJling sites in the Cariboo-Chilcotin region bounded to
the north at 53°N (approximate latitude of Quesnel), the east by the Fraser River, and the
south at 51 °N (approximate latitude of Cache Creek), predominant winds are NE and
thus transport dispersing MPBs to the southwest. Thus MPB outbreaks in this major

105

portion of the Cariboo-Chilcotin are expected to be contained and spread to the
southwest, into the Coast Mountains.
At a provincial scale, my genetically-inferred dispersal results are consistent with
MPB atmospheric transport patterns modelled by Jackson et al. (unpublished data). Most
movements within both the Northern and Southern groups appear to be to the east or
northeast. I argued (see Section 4.2.6) that the stands in the Chilcotin, Farwell Canyon
and Tatla Lake, were comparatively isolated based on the number of connections they
had with other stands. Indeed, Jackson et al. 's (unpublished data) modelling showed that
Cariboo-Chilcotin stands are unlikely to act as source populations to neighbouring
populations and that neighbouring populations tend to disperse beetles to the east or
northeast, away from the Cariboo-Chilcotin. Thus the MPB~ in the Cariboo-Chilcotin
are unlikely to interbreed with populations outside this region.
In the Fraser Valley. Jackson et al. (unpub. data) found that north-south winds
along this corridor were rarer than W or SW winds. At Jackson et al. 's (unpub. data)
sites in the Fraser Valley, less than one out of twenty (-5%) simulated MPBs travelled
north or south. However. I suggest that the connections between MPB stands through the
Fraser Valley, which I inferred from genetic data, may be maintained by these
periodically favourable winds. It should be noted that MPB transport requires a
combination of favourable atmospheric conditions as well as an outbreak in that region.
The combined probability of these north-south transport events would be much less than
5% in any given year but apparently this is sufficient for the maintenance of genetic

similarities ("connections").

106

4.2.5. Five Genetically Unique M PB Stands
In my analyses I discovered five unique MPB stands; these stands were "unique"
in that they were significantly different from every other stand in this study. They were:
Mackenzie, Tumbler Ridge, McBride, Willmore Wilderness, and Whistler. I will present
different arguments to explain the divergence of these stands. Because these stands were
unique and did not resemble source stands. they likely did not arise from long distance
transport during the current epidemic (Gaggiotti. 1996). An exception to this statement
is the Tumbler Ridge infestation. which as part of the new infestations in the Peace River
region, is known to have recently arisen from long distance dispersal (Wood and Unger.
1996; A. Carroll, unpub. data) . Severe founder events from long distance transport, as

previously mentioned, do not occur and thus did not cause this "uniqueness." Though
bottlenecks after the founding of new populations can quickly increase genetic
divergence (Hey eta/., 2004), I found no evidence for bottlenecks in any stand. The
"uniqueness" of these stands also indicates that they have not significantly interbred
recently with the sampled stands.
The most parsimonious explanation is that these stands (with the exception of
Tumbler Ridge) represent isolated native populations that have diverged by genetic drift
over time. Isolation may be complete, causing rapid divergence, or isolation may involve
some level of inter-population gene flow, which can slow, balance, or erase divergence
due to drift (England et al., 2003; Beebee and Rowe, 2004). The location of three of the
five unique stands in the Northern group (four if the boundary population in the
Willmore Wilderness is included) is strongly suggestive that a scarcity of inter-

107

population connections, which is characteristic of the north. promotes the isolation and
divergence of MPB populations.
A hypothesis of an "old" founding. isolation from other populations. and
bottlenecks can be tested in terms of population outcomes. I looked for evidence of
genetic isolation and geographical isolation in each unique stand. I initially compared
the mean FsT values of each unique population to the mean across all 35 populations (FsT

=0.040). A higher mean FsT value indicates considerable long-term population isolation.

McBride had a mean FsT value (0.090) much higher than the population mean.
This indicates that the McBride stand probably represents a native MPB population that
has been highly isolated by mountains, distance. and/or northern population dynamics for
a long time and even within the current epidemic (to 2005). McBride was significantly
different from Willmore Wilderness , in spite of the -35km straight-line distance between
them that made them the closest stands in this study. Outbreak maps from the past
century suggest the McBride stand in the Rocky Mountain Trench is independent of
infestations and possibly native populations in the adjacent Morkill and Holmes River
valleys (see page 11 0). McBride was genetically closest to Northern stands (Fort St.
James then Houston), suggesting ancestry from these areas. This apparent ancestry may
have been enhanced by migrations from severe outbreaks in the Fort St. James region in
1930-36 and 1946-65 (Wood and Unger. 1996).

Willmore Wilderness, in contrast, had a mean FsT value (0.063) 1.5 times higher
than the population mean, indicating that this is likely a native population that has been

108

historically less-isolated than neighbouring McBride. Thus, the Willmore population
may have undergone recent gene flow with other populations (Willmore Wilderness was
sampled in 2006). Note: for the remainder of this Section, the year given for infestations
has been extrapolated back one year to the date of green attack, or the year when MPB s
flew into an area and attacked trees.

1

MPBs have been continuously recorded in the Willmore Wilderness since at least

1997, when pheromone-baited trees were attacked. In contrast, during the current
epidemic in regions of BC adjacent to the Willmore Wilderness, MPB infestations first
erupted in the Holmes Ri ver valley east of McBride in 1998 (red attacks were recorded in

1999). These infestations were relatively constant to 200 I (20-40 spot infestations per
year). In 2002 and 2003 infestations in the Holmes River valley declined, but slowly
started building again from 2004 to 2006 (red attacks recorded in 2005-07). It is
important to note that most of the spot infestations recorded in the Holmes River valley
were in its upper reaches, near the BC/AB border at Bess Pass and -35 km (straight line
across several mountains) from the Willmore Wilderness stand.
Northwest of McBride in the Morkill River valley, whose headwaters meet the
Willmore Wilderness at the BC/AB border, two spot infestations occurred in the upper
reaches of the watershed in 1999. In 2003, several spot infestations occurred at the
headwaters of this valley on the BC/AB border (red attacks in 2004). 2004 saw a swath

1

MPH -auacked trees take about one year to change colour from green to red. Since aerial surveys detect
red attacks, outbreak data derived from these surveys exhibits a one year lag. Thus an aerial survey
reporting red attacks in 2007, for example, is detecting MPB attacks made from dispersal flights in 2006.
Again, the infestation years noted below are for the year of green attack, or the year of flight In terms of
the spread of outbreaks, the year in which red attacks are noted is the year in which these trees act a~
sources for MPB fli ghts and thus new attacks.

109

of infestations through the Morkill River valley (2005 red attacks), but these infestations
declined annually to 2006. Infestations were located at the mouth of the Morkill River
valley in 1998, 1999, and 2001-2006.
Since MPB infestations were detected in Willmore Wilderness in 1997, at least a
year before infestations were detected in adjacent regions of BC. combined with my
genetic data there is strong evidence for MPB populations endemic to the Willmore
Wilderness (as represented by the sampled stand). Up to and including the 2005 flight
(since I sampled Willmore Wilderness in 2006), I suggest that the Willmore outbreaks
largely arose from native populations supplemented by limited migration from the
Morkill (2005) and Holmes River valleys ( 1999-2002 and 2005-current). Migrations
through the Holmes are consistent with the pre-summer of 2006 advance of infestations
down the Smoky River valley, which is continuous with the Holme~ . The severe 2006
flight likely overwhelmed the Willmore Wilderness and Grande Cache region with
migrants from regions of North-Central BC. Thus a majority of the MPBs currently (as
of the aftermath of the summer of 2007 flight) overrunning the Willmore Wilderness and
the Grande Cache area are likely not descendents of populations endemic to the
Willmore. I predict that if beetles were sampled in these regions in 2007 -current, then
these beetles would be genetically similar to stands in North-Central BC. The Holmes
valley route involves crossing Bess Pass at- I 600m ( -5200 feet) above sea level but as
mentioned previously for the Purcell Mountains, the MPB apparently can disperse to and
through such barriers. Though the Willmore Wilderness stand (as of 2006) probably
reflected some migration from the Morkill and Holmes valleys, Willmore was

110

significantly different from McBride. Thus the McBride, Morkill, and Holmes stands
most likely are historically isolated native MPB populations.
The ancestry of the MPB population endemic to the Willmore Wilderness is
likely more recent compared to its neighbour in McBride. The exact timing of this
founding is very difficult to estimate but two approximate times are most likely.
Notably, the headwaters of the Morkill River are located less than IOkm from the
sampled site for the Willmore Wilderness (in Beaverdam Pass). Historical outbreaks in
the Morkill River valley could have established a native MPB population in the Willmore
Wilderness, specifically in the BC/ AB border region. A founding sometime in the past.
combined with periodic moderate bottlenecks, would be consistent with Willmore
Wilderness's mean FsT (0.063) compared to a population mean FsT of 0.040. Willmore's
mean FsT may also be explained by more severe bottlenecks since a more recent
founding, specifically from the massive Cariboo-Chilcotin outbreak in the 1980's.
Willmore Wilderness was genetically closest to stands near Quesne l and Bowron Lake,
which was the north-eastern extent of the 1980's outbreak. Jackson et al.'s (unpub. data)
atmospheric modelling near these sites suggests that atmospheric winds could
occasionally (<5% of the time) move MPBs from Quesnel/Bowron Lake into the
McBride/Will more Wilderness region of the Rockies.
Mackenzie had a mean FsT value (0.1 05) that was the highest among the 35
populations. This value indicates that Mackenzie represented an endemic (native)
population which has been isolated, even within the current epidemic (to 2005).
Mackenzie was genetically closest to Tumbler Ridge, then more distantly related to a
range of stands in both the Northern and Southern groups. Ancestry was thus difficult to

Ill

discern from the Mackenzie stand, which may be very old and may have genetically
converged. The Mackenzie population may have been old enough to reach a limit on
divergence at these microsatellite loci (Paetkau et al .• 1997), after which convergence
occurs and the population begins to resemble random populations to which it was/is not
connected by gene flow. Because I sampled in 2005. given the large dispersal flights that
have occurred to this region during the summers of 2005 and 2006. I suggest that later
samples would find that Mackenzie beetles are genetically similar to stands in NorthCentral BC. as suggested for the Willmore.

Tumbler Ridge had a mean FsT value (0.048) not much higher than the mean
across populations. This indicates that the Tumbler Ridge stand was not a native
population: indeed, extensive ground surveys have revealed no evidence that the MPB is
endemic to this region (A. Carroll, unpub. data). I suggest that Tumbler Ridge was
founded early in the current epidemic (perhaps -5 years ago) and has suffered climateinduced bottlenecks (eg. overwintering mortality) and limited immigration up to
2004/2005 (samples were collected at this site over 2 years in 2005 and 2006).

The first instance of MPB infestations in the Peace River region during the
current epidemic occurred as a few spot infestations between Moberly Lake and
Chetwynd townsite in 1999 (red attacks in 2000). These infestations disappeared by
2000 (no red attacks in 200 I), when the lone infestation in the Peace River region was a
single spot infestation about 30km NE of Tumbler Ridge townsite. In 200 I, there were a
few spot infestations -40km west of Chetwynd, a spot infestation -60km south of

112

Chetwynd, and a single spot infestation in roughly the same area as recorded NE of
Tumbler Ridge in 2000. By 2002 all of these spot infestations had disappeared (no red
attacks in 2003).
There was a large MPB dispersal event in the summer of 2003. Indeed, in 2003
there were large aggregations of spot infestations 40-80km south of Chetwynd, a large
infestation -35km south of Tumbler Ridge, and isolated spot infestations nearing 20km
from Tumbler Ridge (detected as red attacks in 2004). In 2004 the spot infestations
occurred in the same areas as for 2003 but the number of infestations shrunk
considerably. Thus there were no significant long distance MPB dispersal flights to this
region observed in the summer of 2004. In 2005 spot infestation~ recorded in 2004 grew,
probably due to localized increases; this implies that no significant MPB flights to this
region occurred in 2005. However. in 2006 a massive MPB dispersal event (as
mentioned earlier) involving most of North-Central BC swept over the Rocky Mountains
and inundated the entire Peace River region with continuous, not spot, infestations
(recorded as red attacks in 2007).
In sum, over the nine most recent years of MPB "green attack" infestation data
from the Peace River region ( 1998 to 2006), infestation-causing immigration events only
occurred in five years (1999, 2000,2001,2003, and 2006). Of these events, only the
2003 and 2006 immigration events were considerable. Immigration events in 1999,
2000, and 200 I produced only a few spot infestations. There was also a pattern of
newly-founded spot infestations shrinking, or in most years, disappearing by the next
year in the Peace River region; these contractions or extirpations of new MPB
populations occurred in 2000, 2002, and 2004, could not have occurred in 1998 (no

113

infestations) or 1999 (infestations just founded), did not occur in 2005 (I believe there
were local population increases). and in 2001, 2003, and 2006 population declines could
have been masked by significant immigration. Thus conditions in the Peace River
region. as well as very significant control efforts, caused MPB declines or extirpations in
three of seven years, and in a maximum of six of seven years if not for large immigration
events.
I have noted that the first MPB infestation in Tumbler Ridge. during the current
epidemic. was a spot infestation 30km NE of the townsite in 2000. This infestation was
not part of the MPB stand I sampled in 2005/2006. Indeed. the Tumbler Ridge site was
actually 22km SW of the townsite and thus -50km from this spot infestation. As detailed
above, in 2003 there was a major MPB dispersal event that created, among other
infestations. spot infestations -20km SW of Tumbler Ridge in the same area as the
sampled stand. By 2004 spot infestations at the stand location had disappeared, or been
extirpated. probably because of the winter of 2003/2004. The year 2005 saw the
localized expansion of infestations further SW of Tumbler Ridge; these infestations
apparently survived the winter of 2004/2005 in large numbers and immigrated into the
stand location -20km SW of Tumbler Ridge.
With regard to the hypothesis regarding the origin of the stand in Tumbler Ridge,
the earliest this stand was founded within the current epidemic was 2003. This
population may have survived an abrupt transition to endemic levels in 2004. a year in
which no red-attacks were spotted in the area. If so, this 2003-founded population would
have received immigration in 2005 from outbreaks further south of Tumbler Ridge,
which were also founded in 2003. Based on annual provincial MPB outbreak maps, no

114

significant long distance immigrations of MPBs into the Peace River region occurred in
2004 or 2005; thus immigration from outside of the Tumbler Ridge region did not affect
the beetle offspring I sampled in 2005 and 2006.
I thus speculate that the stand in Tumbler Ridge was founded in 2003. during the
current epidemic. and further speculate that moderate bottlenecks in 2004 and 2005 were
only slightly offset by immigration from less-severely-bottlenecked stands - I Okm to the
SW in 2005. I imply that in the stand I sampled. the '04 and '05 bottlenecks were only
moderate. perhaps reducing the founding population in this stand from I O.OOO's to
I OOO's of MPBs in a single year. This conflicts with aerial survey data. as in the sampled
stand. insufficient numbers of MPB~ apparently survived the winter of 2003/2004 to
mass-attack and kill a single tree in 2004 (no red attacks from this stand reported in
2005). However. as all of my collections were of beetle brood. my 2005 collections of
36 of 51 individual MPB~ represented beetles that were descendants of the 2003founders of this stand. Considerable immigration to the sampled stand likely occurred in
2005. but because I only sampled beetle brood, immigration only influenced the 15
beetles I sampled in 2006. The MPBs in this stand. which were apparently extirpated in
2004 (based on aerial surveys), attacked trees that yielded these 2005 results: five red
attacks had four or more brood-containing galleries while remaining trees were largely
green strip attacks (all sampled galleries had large numbers of brood). Thus, the majority
of the beetles I sampled from Tumbler Ridge were descendants of the original 2003founded population. The 15 individual MPB s sampled from this stand in 2006 probably
had ancestry with the founding population and with dispersers from larger infestations

115

-I Okm SW of the stand. The above explanation is congruent with Tumbler Ridge's
mean FsT value, which was only slightly higher than the mean value.

It is also possible, though unlikely. that Tumbler Ridge was genetically unique
because I failed to sample its source stand. It is also possible that the multi-year
sampling at Tumbler Ridge caused bias. As previously mentioned. 33 beetles were
sampled in 2005 and 15 in 2006. These two beetle samples were largely similar.
Comparing the allele frequencies of the 2005 and 2006 beetles revealed that frequencies
were similar. despite large discrepancies in sample size. An AMOV A also found that
these two stands were not significantly different (Arlequin- FsT = 0.0 14; P = 0.1 04;
I0, I 00 permutations). Although data ~upported my decision to include the 2006 beetles
in my analyses, if they were excluded. then Tumbler Ridge lost its "unique" status.
Tumbler Ridge became connected to Houston, Fort St. James, Lac La Hache, Quesnel,
and Chetwynd/Pine Pass. I suggest that these connections were a direct result of reduced
power caused by a sample of only 33 beetles; if 48 beetles were sampled from Tumbler
Ridge in 2005. then Tumbler Ridge probably would have remained genetically unique.
In terms of ancestry, Tumbler Ridge was most closely related to Chetwynd/Pine
Pass, which suggests their respective foundings in 2003 and 2005 involved similar
regions of North-Central BC.

Finally, based on historical outbreak data, MPBs are endemic to the Whistler
region (Wood and Unger, 1996). However. Whistler's mean FsT value (0.044) was not
much higher than the mean across populations. This suggests that the Whistler stand. the
only unique population from the Southern group, was native but has not been very

116

isolated either historically and/or currently. This lack of isolation opposes dispersal
trends predicted by prevailing wind patterns (Jackson et al., unpub. data). Atmospheric
winds during the MPB dispersal season tend to isolate MPBs in Whistler. Winds are
easterly at Whistler. keeping it from acting as a source to neighbouring populations in the
Fraser Valley and Okanagan. Winds from these neighbouring populations are from the
west, preventing Whistler from acting as a sink. I can, however. hypothesize that since
large outbreaks have occurred in the Whistler region over the past century (Wood and
Unger, 1996), and because Jackson et al. (unpub. data) found that in perhaps one out of
every ten instances. westerly winds occur at Whistler. occasional dispersal of MPBs to
neighbouring populations to the east may be enough to maintain Whistler's relatively low
level of genetic isolation.
In sum, though this is a large extrapolation, if these wind patterns are historically
representative of the Whistler region, then Whistler is most likely to act as a source of
beetles. Occasional gene flow to neighboring MPB populations may maintain Whistler's
low average genetic distance to other populations.

I 17

4.2.6. Ten Comparatively Isolated MP/l Populations
Based on the new outbreaks in the Peace River region. I suggest that populations
connected by significant immigration should show many instances of genetic similarity.
The corollary is that expanding native populations that have been connected by little
immigration before and/or during the current epidemic should have few instances of
similarity. I suggest that these latter populations are isolated given a comparative
regional context where neighbouring stands had many (-I 0) connections. Fort St. James
(4). Francois Lake (2), Tatla Lake (I), Farwell Canyon (5), Mount Robson (5), Golden
(4), Banff (2), Manning Park {4), Kelowna (3), and Argenta (4) were all native
populations based on their comparative isolation (connections m parentheses). Of these,
mean-FsT values suggested that MPB populations in the Chilcotin region (Tatla Lake and
Farwell Canyon), Mount Robson Provincial Park, Manning Park, Kelowna, the northeast
Kootenays (Argenta), Banff. and Golden were the most isolated.
A stark contrast is found in Lake Louise, a "hub" stand where as previously
mentioned, migrants that have apparently originated in the Okanagan have moved
through the Kootenay and Rocky Mountains. A major source for migrants to Lake
Louise has probably also been nearby Golden and Yoho National Park.
In the Northern MPB group, Telkwa and Houston apparently were the major
source stands and Fort St. James and Francois Lake were less important sources.
Connections between the Northern and Southern MPB groups were apparently mediated
via a "hub" involving Prince George, Quesnel, and Bowron Lake. This hub was
connected to Farwell Canyon, Lac La Hache, and Lillooet in the Southern group.

118

4.2.7. MPB Populatioll Structure Versus Spread of the Epidemic
In the Southern MPB group there was a "mess" of connections (AMOV A)
between stands. I thus expected to find little evidence for significant population structure
within the Southern group. The mass movement of beetles among populations during
epidemics is expected to mask or even erase between-stand differences as populations
interbreed and become genetically similar. However. the between-stand connections I
found seemed to follow a pattern of MPB dispersal through weaknesses in barriers,
primarily mountain ranges ; this dispersal may have occurred during the current epidemic,
the recent past, or the distant past. Most between-stand AMOV A comparisons found
significant differences. Moreover, isolation by di stance (180) was strongest within the
Southern group. This highly significant lBO relationship indicates ongoing gene flow
and divergence among populations. Thus in the Southern group, stepwise interpopulation breeding, where nearby populations are most likely to interbreed (and become
genetically similar) compared to distant populations. was dominant. These results, in
sum, suggest that many of the connections between stands arose from stepwise interpopulation breeding. The corollary is that endemism and population structure are being
maintained in the Southern group, despite the expectation that epidemics should mask or
erase population identities. Thus, the spread of the epidemic in the region encompassed
by the Southern group has been largely influenced by the stepwise spread of native
populations.
In the Northern MPB group there were fewer connections (AMOVA) between
stands. I found significant population structure within the Northern group (AMOVA).
However, a finding of non-significant IBD indicated that long distance dispersal events

11 9

were dominant in the north. I suggest that the comparative lack of geographic barriers,
specifically mountain ranges, in the north versus the south has facilitated the dominance
of long distance dispersal in the north. As previously mentioned, a dominance of long
distance dispersal would disrupt the IBD correlation between genetic and geographic
distance between populations. Movements in the north are likely to involve large
"jumps" among populations, with little correlation with distance between populations.
High rates of genetic drift, consistent with historical Northern MPB population
dynamics, have also probably contributed to a lack of IBD in the north. Thus, the spread
of the epidemic in the north has been largely intluenced by long distance MPB dispersal,
facilitated by a lack of geographic barriers.

120

4.3. M PB Population Structure and the Geographical Distribution of Climatic
Suitability
Over the past century. the distribution of regions that are climatically suitable, in
terms of Climatic Suitability Classes (CSC) for the MPB . has expanded across BC and
especially in the North (Figure 9; Carroll et al .. 2004 ). My results suggest that the
Northern group and the northern-most portion of the Southern group are the sources of
recent climatically-induced MPB range expansion into the Peace River region. However.
the MPBs current North-South population genetic structure (2005/2006) has a stronger
concordance with a historical 1921-1950 as opposed to a recent 1971-2000 esc
distribution (Figure 9; Carroll et al .. 2004 ). My results indicate that the genetic structure
of the MPB in Western Canada has not shifted in response to recent expansions of
climatic suitability. I previously argued that the existence of a Northern and Southern
group was a partial consequence of differential climate-driven MPB population
dynamics; Southern BC is typified by high to extreme climatic suitability for the MPB
while Northern BC is markedly less suitable (Carroll et al., 2004). Over the 201h century,
in addition to expansions of climatic suitability in the Peace River region. there was a
major expansion of high to extreme climatic suitability into the regions west of and
including Prince George. Range expansion by the Southern group into this adjoining
region would be expected. However. current MPB population structure did not reflect
this change.
I propose two explanations: I) If the Fraser Valley was an active corridor for
MPB movements between the Northern and Southern groups, then the Southern group
shoul d have shifted into the new ly-suitable region west of Prince George. Since this shift

121

did not occur, the Fraser Valley probably represents a postglacial recolonization route
and/or a little-used corridor. The Northern and Southern groups are probably effectively
separated. The corollary is that MPB responses to climate-induced range expansions will
most likely occur within groups. An expansion of climatic suitability in the Southern
group will probably involve range expansion by the Southern group. and vice versa. 2)
It is also possible that my sampling in the region surrounding the esc shift located west
of Prince George was insufficient to detect a range shift. In the near future. annual
population genetic surveillance of MPB populations will hopefully become a reality.
This surveillance would hopefully occur at a high enough resolution to detect MPB range
shifts in response to climate shifts.

122

4.4. A Model for MPB Population Genetic Dynamics
Based on my literature review and my results regarding the dynamics of MPB
population structure, I propose a model for MPB population genetic dynamics. In terms
of divergence, MPB populations become differentiated by drift and selection during
isolation in the endemic phase: divergence is slowed, balanced, or erased by interpopulation migrations during the epidemic phase. In terms of population genetic
diversity, MPB population dynamics. which are characterized by temporally-dominant
small (endemic-phase) populations and punctuated by widespread population explosions
(epidemics), may be neutral with respect to genetic divers1ty maintenance. In other
words, mutation/migration and genetic drift averaged over the phases of the MPB
population cycle may be in equilibrium . This model for population genetic dynamics,
where divergence and losses of genetic diversity during the endemic phase are balanced
by convergence and gains in genetic diversity during the epidemic phase, may be
characteristic of all eruptive herbivorous insects, such as locusts (Ibrahim, 200 I).
During the endemic phase, local selection is significant. More importantly,
genetic drift should be very heavy as these widely distributed populations have small
effective sizes. Indeed, the MPB endemic phase is characterized by -five to ten beetles
per stand that are outcompeted by other bark beetles in blown-over and suppressed trees
(Safranyik and Carroll, 2006). Thus, during the endemic phase, MPB individuals and
populations are extirpated at high rates due to competition and demographic and
environmental stochasticity, although some extirpated regions are likely recolonized by
migrants.

123

During epidemics, there is enormous generation and retention of mutations.
These mutations are widely distributed by long and short distance migrations (Petit et al.,
2005). Populations are connected by levels of gene flow correlated with geographic
barriers (distance and/or mountain ranges). such that nearby populations become
genetically similar while distant populations remain differentiated. When epidemics
decline. most new genetic variation is likely maintained in myriad small MPB
populations that are distributed over the landscape. Epidemics may not only span
previously occupied regions but may also contribute mtgrants to distant regions. These
migrants may survive and become endemtc-level populations that retain considerable
genetic diversity.

124

4.5. MP/1 "Range Expansion"
The Pleistocene has been characterized by major ice ages (glacials) on a -I OOkyr
cycle that are interrupted by warm interglacials as per current conditions (Hewitt, 1996).
The transitions between these climatic extremes are characterized by gradual warming or
cooling and punctuated by frequent short and severe oscillations. For example, during
the Eemian interglacial ( 135-1 15kyr BP), oscillations involved I0-12°C changes in
average temperature over just 5-I 0 years (Hewitt, 1996); this climate endured for 705000 years depending on the event. Such rapid climatic change would extirpate most or
all of a species over its northern range (sudden cooling) or its southern range (sudden
warming). Under Hewitt's ( 1996) model of range oscillations, only "central"
populations ultimately survive. Thus. if Idaho/Utah approximates the MPBs core
(ancestral Pleistocene) range, then during glacials this would be the northern-most (and
likely smallest) population and during interglacials it would be a central or south-central
(larger) population as per its current location within the MPBs range.

I hypothesize that the current northward range expansion of the MPB , into
"historically climatically-unsuitable areas." is part of a trend of interglacial expansions
and glacial retreats that have occurred over the past several million years (the
Pleistocene). In other words, the MPBs current range expansion. while novel with
reference to the MPBs range over the past century, is a normal event that has recurred
many times over the Pleistocene. When interpreting range shifts, defining an appropriate
frame of reference is critical. The gradual changes and severe oscillations of both the
Pleistocene climate, and of the present Holocene climate. have caused and continue to

125

cause obvious shifts in the geographic distribution of climatic suitability for all
organisms. Retreats during glaciations and expansions during interglacials have occurred
for millions of years (Hewitt, 1996; 1999; 2000), long before humans significantly
influenced the Earth . These cycles will most probably continue long after we are gone.
During this interglacial. like any other. climatic suitability will continue to shift
northwards and the MPB will continue to expand north, reaching an apex somewhere in
the Yukon (Logan and Powell. 200 I). Perhaps in a century or two this prediction will
come to fruition . When the next glacial begins, these northern MPB populations will be
extirpated and the MPB will again. as I hypothesize. persist in Idaho/Utah until the next
interglacial.

126

4.6. Potential Research Limitations and Biases
I identified eight potential areas of limitation or bias in my research: sample sizes;
genetically-i ndependent sampling; adjusting significance levels for the number of tests;
inherent microsatellite limitations; the estimation of recolonization or divergence times;
limitations of the program Structure; potential influences from outbreaks in the United
States; potential anthropogenic influences on dispersal patterns.
Sample sizes were an improbable source of bias for my study. Al l sample sizes
referred to in this paragraph represent final sample sizes. or the number of individual
beetles that were fully genotyped. In my study. I collected more individual beetles for
every site than were represented in the final sample size. All of my sample sizes were in
the range of 43-49 beetles per site (mean= 46) with the exceptions of McBride (n=24),
Grande Prairie (n=30), and Chetwynd/Pine Pass (n=55). The potential n-induced bias
associated with McBride and Grande Prairie has been previously mentioned. The
magnitude and restricted range of my study's sample sizes is robust, especially in
comparison to the three previous microsatellite-based studies of Scolytids. Kerdelhue et

al. (2006) sampled only 30 beetles per site, an average n over 50% smaller than ours.
Salle et al. (2007) sampled an average of only 23.0 beetles/site (28 sites) and their
sample size distribution was bimodal, with most sites having an n=30 or n=20. The most
recent of the three Scolytid microsatell ite studies had the most potential for bias. Maroja

eta/. (2007) sampled an average of 39.7 beetles/site ( 14 sites) but there were notable
outliers in Alaska (n= l 15), Utah (n=73), Montana (n=23), and Washington (n=22). The
magnitude and range of my study's sample sizes were robust.

127

I found low levels of significant linkage disequilibrium distributed evenly across
stands and loci. This was probably a consequence of the fact that the infestations I
sampled likely arose from endemic populations. Endemic populations are characterized
by very few beetles per stand. A direct consequence of small population sizes is strong
bottlenecks and inbreeding. both of which cause significant linkage disequilibrium.
Bottlenecks and inbreeding during the endemic phase most likely play a large role in the
dynamics of all bark beetle populations.
My samples were genetically-independent. I define genetic independence as the
fact that genotyped beetles had different parents. An argument can be made that such
genetic independence caused us to over-represent genetic variation in sampled stands.
However. I feel that my sampling regime is the most robust of any population genetic
study of a Scolytid beetle. As inferred from their methods sections. the past studies
previously mentioned have treated sampling regimes as relatively unimportant.
Protocols for beetle sampling are rarely found. The potential for bias due to the
methodological unknowns of these studies is great. Moreover. the direction and
magnitude of such bias is unknown. I strenuously advocate the use of known, "lesserevils" (methodologies with known small biases) in place of unknown. "greater-evils"
(methodologies with potentially huge biases).
Genetically-independent samples, moreover, are strongly preferable to
pseudorepJicates in population genetic research. Genetic independence maintains sample
sizes while pseudoreplication involves sampling of redundant genotypes. lowering
effective sample size (eg. Yoon Chung et al., 2005). With genetic independence, one can
accurately characterize population genetic diversity and determine population structure

128

using a much lower sample size than studies affected by pseudoreplication. Indeed, the
widespread reporting of no population structure among many Scolytid studies may be
attributable to a failure to correct for pseudoreplication. My sampling regime, however,
was not perfect. MPB populations should be sampled randomly but this poses technical
challenges. Sampling beetles during flight (pheromone trapping) probably does not
represent effective dispersal. Thus brood must be sampled and this can only be done
under the bark of infested trees. How MPBs can be sampled randomly in reference to a
total MPB population. whose geographical boundaries may not exist, is a problem I leave
to future studies.
A fundamental question in my research was whether to be more-conservative or
less-conservative when assessing the genetic similarity between stands. Specifically, the
adjustment (lowering) of significance levels for multiple comparisons has been
haphazardly applied across ecological disciplines (Cabin and Mitchell, 2000). No formal
criteria exist in any scientific fields for when to use such corrections and for which of the
many available corrections to use (Nakagawa, 2004); indeed, the application of multiple
testing procedures is a subjective practice (Pemeger, 1998; Moran, 2003 ). Although I
used a standard Bonferroni correction in my analyses of Hardy-Weinberg Equilibrium
and linkage disequilibrium, I chose not to make any adjustments to the significance level
(a=.05) for the AM OVA analyses. Nearly 81% of 595 pairwise between-stand AMOVA
tests rejected the null hypothesis (H 0 : stands are genetically similar) and thus found
stands were significantly different. Compared to the HWE and linkage disequilibria
results, which showed no consistent patterns, the AMOV A results were strongly
suggestive that I detected something important (Moran, 2003; Verhoeven et al., 2005).

129

Bonferroni corrections would have reduced the rate of Type I error (rejecting a true null
hypothesis) at the expense of a higher rate of Type II error (accepting a false null
hypothesis). In my study. more Type II errors would mean more tests found nonsignificant differences, or genetic similarities. between stands. I considered the
assessment of genetic similarities. or "connections:· within the current epidemic to be of
critical importance and chose to minimize Type II errors by keeping the significance
level at 0.05.
We conducted a sequential Bonferroni correction (Holm. 1979), which reduced
the effective level of significance to 0.000206 (0.05 I -243 ). to determine how correcting
for multiple comparisons would have affected the AMOVA-derived "connection" map
(Figure 1). The sequential Bonferroni correction resulted in 242 non-significant tests
(41%), as compared to only - 19o/c of tests at a 0.05level of significance. Revising
Figure I for Bonferroni-correction (not shown) resulted in more connections within the
Northern and Southern groups. as well as more connections between the groups through
the Fraser Valley. At a stand-level. Bonferroni corrections caused "hub" stands to have
many more connections, comparatively isolated stands to become "hubs," and four of the
five genetically-unique populations to become connected. Whistler became connected to
Manning Park, Lac La Hache, and Kamloops; McBride became connected to Fort St.
James and Houston; Willmore became connected to Quesnel; Tumbler Ridge became
connected to Houston, Fort St. James. Lac La Hache~ Quesnel. Chetwynd/Pine Pass, and
Grande Prairie. In all cases, these four unique populations became genetically similar to
the stands to which they were most related. Mackenzie remained significantly different
from all sampled stands. These additional "connections" confounded relationships

130

among stands. In sum, by keeping a 0.05 significance level and minimizing Type II
errors, I have identified the most important historical and/or recent relationships among
stands.
Microsatellite analysis has inherent sources of error. Null alleles, which are
caused by a point mutation in the primer annealing site (Jame and Lagoda, 1996; Dakin
and A vise, 2004), can result in non-amplification of a sample at a locus. Although null
alleles are typically restricted to just a few individuals m a study, their effects can be
widespread and difficult to detect. Short allele dominance (Wattier et al. , 1998), where
the smaller sized allele of a heterozygote is preferentially amplified, can also be
prevalent. Both types of errors can cause heterozygotes to be scored as homozygotes,
reducing apparent population differentiation (Dakin and A vise. 2004). I minimized these
errors by establishing all alleles at all loci. across all populations, prior to manual scoring
of genotypes; this ensured that valid peaks were recognized as true alleles. Permutation
tests confirmed that most populations did not have excess homozygotes and were well
within Hardy-Weinberg equilibrium.
It would be impossible to estimate recolonization or divergence times from my
microsatellite-derived data set. Such estimates are influenced by levels of population
genetic diversity. fluctuations in population size since divergence/recolonization, the age
of populations, recombination rates at each locus, the number of repeats at each locus.
the location of microsatelJite loci on chromosomes, and many other factors. Calculating
and determining the cumulative effects of these influences on the mutation rates of my
six loci could not be done with any precision. Future studies that aim to calculate the
recolonization and divergence times of MPB populations should employ sequencing

131

analyses that preferably involve both the nuclear and mitochondrial genomes. There is a
considerable body of statistics devoted to estimating various timelines from sequence
data and highly accurate estimates are possible (Marshall eta!., 2002).
The statistical program, Structure, is limited in that it works optimally for data
sets with small numbers of discrete populations. Moreover, data that are characterized
by isolation by distance, where there are allele frequency gradients among populations, is
not ideal. Under isolation by distance, MPB genotypes imply membership in multiple
clusters and population structure is increasingly obscured at higher cluster values. This
pattern of population structure dilution at increasing values of K occurred in my
analyses; thus, the lowest tested value of K=2 yielded the strongest structure.
There is low likelihood that recent MPB outbreaks in the United States have acted
as significant sources of long distance dispersers to the current Western Canadian
epidemic. Maps of outbreaks in the states of Washington, Oregon, and Montana over the
past decade indicate that both the scale and severity of outbreaks pale in comparison to
the Canadian outbreak. While the MPB epidemic newly infested 10.1 million ha at
various mortality levels in 2007 in Western Canada, equivalent infestation areas in the
United States were notably lower. Washington State recorded I 03,000 ha of all pine
beetle mortality in 2006 (Nelson eta!., 2007a). Most of this mortality was MPB-caused
and was focused in the remote northern parts of the state, primarily in the North Cascades
(adjacent to Manning Provincial Park in BC) and west of Franklin D. Roosevelt Lake in
NE Washington State. MPB infestations in Oregon totalled 144,000 ha in 2006 and were
focused on the eastern slope of the Cascade Mountains (Nelson eta!., 2007b). In Idaho,
infestations covered 124,000 ha in 2006 (Idaho Forest Health Protection, 2007) but an

132

outbreak location map was not readily available. Based on host tree distributions, much
of northern Idaho contains suitable host species for the MPB. Lodgepole pine, ponderosa
pine, and western white pine are abundant in northern Idaho and all are ideal MPB hosts.
Of the states mentioned, Montana sustained the largest area of MPB infestations in 2006
at 329,000 ha; most of these outbreaks were in west-central and south-west Montana
(Gannon et a!., 2007).
Overall. only Washington and possibly Idaho have had recent MPB outbreaks
that are proximate to and could possibly act as source populations for Canada. Given the
small relative size of these outbreaks. contributions to the Canadian epidemic, if any,
were and probably continue to be minimal. Given the scale and severity of the current
Western Canadian MPB epidemic. it is more likely that Canadian infestations have been
sources for point infestations across the border in the United States. It is unfortunate that
my sampling did not involve sites in northern Washington State and Idaho State. My
prediction is that infestations at these sites would be genetically contiguous with sites in
southern BC. This prediction awaits a future study that examines the range-wide
population genetic structure of the MPB at a resolution, in terms of distance between
sampling sites, that is as fine as or finer than ours.
My results may have been biased by anthropogenic translocations of MPBs
between normally isolated populations through MPH-infested timber. Such biases are
not uncommon in studies of Scolytids (Nilssen, 1984; Stauffer et al., 1992). Indeed, the
international trade of pine timber may be a contributing cause of apparent gene flow
among isolated populations ofT. piniperda (Faccoli eta/., 2005). Moreover. the
operation of pine plantations may involve movements of possibly infested trees since this

133

beetle does its maturation feeding in shoots (Hornet a/., 2006). Faccoli eta/. (2005)
genetically corroborated that, as previously asserted by forest monitoring, their sampled
stand ofT. destruens in northern Italy represented a population that was introduced with
a plantation from Greece about 50 years ago. Moreover. about 2000 years ago a system
of plantations was created during the Roman era to defend the Adriatic coast and may
have resulted in much translocation of beetles (Faccoli eta! .. 2005).
MPH-infested trees. stacked in lumber yards. can produce dispersing MPBs.
However. no stands were within 80 km of a log yard. Imports of MPH-infested wood for
campfires were only potentially significant in Lake Louise, which sourced MPB-killed
timber from adjacent Yoho National Park (Banff National Park Warden Lee Smith personal communication). However. inspection of the gravel pit where MPH-infested
timber was stored revealed no successful MPB attacks on nearby pines.

134

4. 7. Recommendations for Future Research
Future studies of MPB population structure should use more geneti c markers and
sample at an equivalent or finer spatial resolution than ours but extended over the MPBs
North American range. To maximize potential gains in know ledge, such a study shou ld
also extensively sample stands of D. jeffrevi. A combined approach wou ld not only yield
a high-resolution picture of the population genetic structures of each species, with
information on rates of gene fiow and historical movements, but a combined approach
may also elucidate the mode of speciation between these sister species. Based on their
differential host-use and geographical allopatry (Wood, 1982), I hypothesize that D.

jeffrevi diverged from a MPB-Iike common ancestor. which was a host generalist (Kelley
and Farrell, 1998), via isolation during the Pleistocene. Though there is debate over the
role of Pleistocene glaciations in divergence, many studies support Pleistocene-mediated
speciation (Knowles, 200 I). The ancestor of D. jeffrevi may have been isolated from the
eventual MPB by the Cascade. Klamath, and Sierra Nevada Mountains as well as the
Great Basin and Mohave Deserts in a California refugium, and responded altitudinally
(up and down mountains) rather than latitudinally (Hewitt, 1999), during the frequent
advances and retreats of the past two mil lions years of glaciations (the Pleistocene
epoch). However. due to Pleistocene range dynamics, species distributions may have
changed greatl y since the time of speciation (Knowles, 200 I). A related explanation is
that D. je.ffreyi arose during the Pl eistocene in a refugium adjoining California, from
which it achieved its current distribution.
In a case study of post-Pleistocene recolonization patterns across taxa, Hewitt
( 1999) found that populations in the eastern Mediterranean were not a source of colonists

135

to northern areas. As Hornet al. (2006) indicated for the beetle T. destruens, major
geographical barriers (distance and mountains) and/or lack of suitable hosts played a
large role in this across-taxa pattern of isolation. Thus, if D. jeffrevi was established
during the Pleistocene, geographical barriers and a fragmented host distribution probably
limited its range to California, even during interglacial periods, and prevented northward
expansions and interbreeding with the common ancestor of the MPB, resulting in
allopatric speciation. Indeed, Hornet a!. (2006) estimated that the split between western
and eastern groups ofT. destruens in the Mediterranean occurred during the Pleistocene,
with a mean estimate of 430 000 years bp (range 350 000-1 200 000 years). Although
these groups have not diverged enough to justify the genetic designation of a new
species, T. destruens has two dispersal flights per generation compared to one flight per
generation among D. ponderosae and D. jeffreyi. Thus speciation could have occurred at
a faster rate for D. jeffreyi and could have occurred during Pleistocene events. The
current sympatry between D. ponderosae and D. jeffreyi {Kelley eta!., 2000) would
represent recent secondary contact due to southward recolonization along the West Coast
that I suggest occurred post-Ice Age for the MPB.
Future studies that examine the range-wide population genetics of the MPB at a
high-resolution should simultaneously examine the following hypotheses of post-Ice Age
recolonization: I) post-Pleistocene recolonization was associated with successive founder
events (reduced genetic diversity) away from a single refugium (the "core range" in
southern Idaho and in Utah), as I previously hypothesized; or 2) regional population
structures and hybrid zones exist, suggesting multiple MPB refugia (Knowles, 200 I).

136

Importantly. if future landscape-scale studies of MPB population genetics employ
more genetic markers (Rosenberg eta/., 200 I), then they will be able to explore the
presence of sub-clusters within the Northern and Southern MPB groups. The programs
Structure and SAMOV A shou ld be employed. The hierarchical analyses of population
structure outlined by Rosenberg eta/. (200 I) would entail first testing for clusters using
all individuals, then testing within groups found by the previous analysis. and onwards
until population structure is not apparent at increasingly finer geographic scales.
Future studies should examine the temporal dynamics of MPB population
genetics by sampling stands over consecutive years and detennining how genetic
diversity and allele frequencies change with population status (endemic to incipientepidemic to epidemic to post-epidemic). Ideally. sampled stands would contain native
MPB populations that were highly isolated from influxes of migrants. Information from
this research could Iikely be accurately extrapolated to other Scolytids, including the
Jeffrey pine beetle (D. jeffreyi) and the Douglas-fir beetle (D. pseudotsugae), as
population dynamics are similar across many bark beetle species.
A related recommendation regards employing more marker loci on a similar
number of samples per stand to allow for powerful bottleneck detection. Though I did
not detect bottlenecks among stands, my negative result was consistent with studies
employing low power (Luikart et al., 1998; Spong and Hellborg, 2002; Busch et al.,

2007). Bottleneck detection will be particularly important when explaining the
population structure of recently founded stands in Alberta. especially if these stands are
significantly different from all others. However. caution must be used in bottleneck
detection during MPB epidemics. Even low levels of immigration can erase bottleneck

137

signatures within a few generations (Keller et al .• 200 I). Because of massive amounts of
mutations and the stochastic distribution of these mutations over the landscape via
migrations, bottlenecks may appear to be diffuse. Diffuse bottlenecks allow much allelic
and genetic diversity to be retained, making bottlenecks difficult to detect (England et al .•
2003).
Future studies should also, as a first step. conduct a preliminary investigation of
the comparative pathogenicity of Rocky Mountain and Interior G. clavigera via MPHderived fungal culturing and stand inoculation trials. If a difference in pathogenicity is
detectable. then the study should be conducted comprehensively. The F 1 generation of
MPBs (surviving brood: not colonizing adults) could be sampled from several stands in
the Interior and the Rocky Mountains. G. clavigera could be cultured from an
appropriate number of the beetles collected per stand. Inoculation trials of both groups
of G. clavigera (with various individuals) could be conducted at various inoculation
intensities at multiple sites in both the Interior and the Rocky Mountains. If my
hypothesis is correct and there is stronger. or less-constrained, selection for pathogenicity
in the Interior. then significant differences in G. clavigera pathogenicity between regions
should be apparent.
Future studies should examine the characteristics of weather systems associated
with known times for and known results of MPB dispersal flights during the current
epidemic. The massive summer of 2006 flight in northern BC deserves special attention.
Such research may allow for the prediction of especially favourable conditions for long
distance MPB dispersal.

138

Aukema et al. (2008) have developed a model that incorporates the spatiatemporal and climatic influences on the spread of MPB outbreaks. Given inputs of
outbreak locations in one year. this model has a very high predictive accuracy for
forecasting the location of outbreaks the next year. at a cell resolution of 12km x 12km.
However, the authors found that the accuracy of two and three year forecasts declined
rapidly as severe climatic events (autumnal cold snaps) caused significant MPB
mortality. Thus Aukema et al. 's (2008) model works well under norma] conditions for
MPB outbreaks but stochastic events involving extreme summer or winter climate,
and/or long distance dispersal, can invalidate this model. Hopefully. future MPB models
will incorporate increasingly detailed data on MPB population structure, as well as data
on stepwise and long distance MPB dispersal.
A major gap in MPB research is the lack of detailed understanding of the location
of MPB populations endemic to BC. In the past, outbreaks have occurred in extremely
isolated areas that are decades and/or hundreds of kilometers removed from the last
outbreak. For example, the MPB is endemic to Vancouver Island and outbreaks were
partia11y responsible for the decline of mature white pines on the island from the 1940' s
to the 1960's (WestfaJI, 2006). The next infestation on Vancouver Island occurred in
1984, encompassing 30 trees, and 10 trees were MPB-ki11ed in 2005 about 10km
southeast of Sayward (Westfall, 2006). These dates are synchronous with previous largescale MPB outbreaks that have occurred in BC over the past century.
In contrast, in many regions it is not known if there are extant MPB populations.
MPB outbreaks have been persistent in the New Hazelton region of north-western BC for
the past half-century. Native MPB populations across BC have simultaneously expanded

139

in the current epidemic and one would expect the same in New Hazelton. However.
there is no evidence for localized expansions in this region in the past decade; recent
outbreaks have been caused by long distance dispersal flights. The presence of native
MPBs in a region can be exceptionally variable and difficult to detect. It is possible that
MPBs were extirpated from the New Hazelton region following the outbreak in the mid1990's. that MPB s in this region simply have not responded in a simi lar fashion to other
native MPB populations in BC. and/or that in this region endemic-caused damage has
occurred at low levels and has not been detected.

140

4.8. Executive Summary
I met my objective: to determine the phylogeography of the MPB in Western
Canada and to examine patterns of MPB population structure at varying scales to infer
M PB dispersal. At the largest scale, my data indicated that the current epidemic consists
of a group of populations connected by varying degrees of gene flow. Thus the current
epidemic has arisen by the expansion of a number of native MPB populations. As these
populations have expanded, they have provided migrants at a mostly regional scale
which has caused the rapid escalation of severity ("fill-in" between geographicallyseparated outbreaks) within the epidemic area. Though the expansion of native
populations has been largely synchronous, ~orne populations (eg. Kamloops, Nancy
Greene, North-Central BC) seem to have expanded more rapidly and have become
significant sources of both local and long distance dispersers.

14 1

4.9. Managem ent Recommendations
Outbreaks are a major factor influencing the evolution of the MPB as they
maintain/increase the genetic diversity upon which evolutionary forces act. Thus, there
is a positive-feedback loop in which climate change. by increasing the frequency and
severity of MPB outbreaks. in turn generates higher levels of genetic diversity upon
which MPB adaptation to a changing climate may occur. There are many other. probably
more-important, factors that will influence the MPBs success in the 21 ~t century: the
distribution of host trees of suitable size and species: climate-induced MPB mortality; the
effects of climate change on host tree resistance in various regions of BC and AB;
whether or not the MPB has successfully colonized the jack pine forests of Canada.
From an evolutionary and management perspective, it may be slightly beneficial to limit
the frequency and severity of MPB outbreaks to limit MPB adaptations to a changing
climate. It is also important to realize that the MPB exhibits considerable variation in life
history traits. such as number of generations per year and cold tolerance, over its North
American range; the MPB may exhibit high levels of phenotypic plasticity and strict
adaptations to a changing climate may not be required.
The existence of genetic diversity gradients across the MPBs range is probably
unimportant for MPB management. The reduced genetic diversity of MPB populations
at the northern and southern portions of their range implies that these populations have
reduced fitness (Reed and Frankham. 2003). Low genetic diversity can constrain the
evolution of local adaptations that are important for the long-term persistence of
populations. However. many species' success has not been constrained by reduced
genetic diversity (eg. Frankham. 2005; Pui ll andre et al .• 2008). Indeed, the MPB has

142

high phenotypic plasticity and likely possesses much adaptive variation. Genetic
diversity gradients in the MPB likely do not imply fitness gradients.
Climate change-induced range shifts may be very similar to Pleistocene patterns
of range expansion. I suggest that studying the indicators and patterns of postglacial
range shifts is of maximum predictive value for understanding how species will respond
to a changing climate. For the MPB. by understanding the processes of postglacial
recolonization in terms of dispersal trends and population genetics, we may be able to
predict the range expansion and evolution of the MPB under a changing climate.
Landscape-level sera) diversity, tn combination with tree species diversity. is
probably the best measure for preventing widespread MPB outbreaks. In the near future,
complex landscape level models should aid forest managers in creating such diversity. In
contrast, allowing landscape succession to take its course may allow non-pine understory
species such as spruce. subalpine fir. and Douglas-fir to dominate many stands and create
MPB-resilient landscapes (Burton. 2006). However. given my finding of a Northern and
Southern MPB group, I have management recommendations specific to each group. In
southern BC. by identifying the scale and magnitude of MPB dispersal, my results are
probably best used in planning for the containment of outbreaks to regions.
In southern BC, the majority of MPB dispersal occurs over relatively short
distances and most long distance dispersal seems to be funneled through weaknesses in
geographic barriers. It is thus most pertinent to focus stand management (diversity
management) at regional scales to prevent not only the expansion of outbreaks but also
movements into susceptible stands in adjoining regions ("regional containment"). In a
similar thread, outbreak management in southern BC should also be regionally-focused

143

in terms of strategic harvesting to contain outbreaks to a region. As an example, the
Gray Creek valley that bisects the Purcell Mountains should be a major management
focus in case of outbreaks in the Nelson region. Thus in southern BC. the general
dispersal patterns I have identified are priority areas for containing outbreaks to regions.
In northern BC. creating management recommendations is made difficult by the
apparently widespread occurrence of long distance dispersal events. Such widespread
long distance dispersal is undoubtedly an artifact of the unprecedented scale and severity
of the current epidemic. I suggest that planning for sera! and species diversity across the
northern BC landscape (not regionally) is probably the best option as geographic barriers
and specific dispersal routes are relatively nonexistent in the north. Even in the north,
sera] diversity has the best chance of containing outbreaks to regions and preventing
widespread epidemics.
MPBs are exceptionally difficult to manage (Safranyik et al., 1999).
Unfortunately. there is much evidence for historical MPB epidemics, in the obvious
absence of human interference, whose scale is unknown relative to current epidemics
(Schmid and Mata, 1996; Alfaro et al., 2004). My study, by identifying MPB ..dispersal
highways," may facilitate better management of MPB epidemics; however, it is likely
that MPB outbreaks will occur for the foreseeable future.
My study demonstrates that a population genetic approach to researching long
distance MPB dispersal is highly efficient and effective (Kim eta!., 2006). I recommend
that the British Columbia provincial government fund an annual standardized MPB
genetic monitoring program. Samples could be collected using a standard protocol in the
course of regular Ministry of Forests and Range forest health duties. adding value to

144

current programs (Nelson et al .• 2007c). Sampling entails two to three hours of labour
per site. The genotyping costs of annual population genetic surveillance of MPB
populations would range from $70 to $100 per sampling site. One hundred sites would
provide excellent surveillance across the province. The annual cost for one hundred
sites. including genotyping and labour for typing and data analysis. would be under
$20.000. Such data would not only meet applied (management) objectives in terms of
monitoring MPB population trends and dispersal. but would also provide internationallyunprecedented data for advancing theories of population genetics and evolution.

145

Literature Cited
Alfaro, R.I.. Campbell, R., Vera, P., Hawkes, B., and T.L. Shore. 2004.
Dendroecological reconstruction of mountain pine beetle outbreaks in the
Chilcotin Plateau of British Columbia. Pages 245-256 in T.L. Shore, J.E. Brooks
and J.E. Stone, eds. Challenges and Solutions: Proceedings of the Mountain Pine
Beetle Symposium. Kelowna, British Columbia. October 30-31, 2003. Natural
Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria,
British Columbia, Information Report BC-X-399. 298 p.
Amman, G.D., and J.A. Logan. 1998. Silvicultural control of mountain pine beetle:
prescriptions and the influence of microclimate. Am. Entomol. 44: 166-77.
Anderson, W.W., Berisford, C.W., and R. Kinrich. 1979. Genetic differences among
five populations of the southern pine beetle. Ann. Entomol. Soc. Am. 72: 323-27.
Aukema, B.H., Carroll. A.L.. Zhu. J .. Raffa, K.F., Sickley. T.A., 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.
Ecographv 29: 427-41.
Aukema, B.H., Carroll, A.L., Zheng, Y., Zhu, J., Raffa, K.F., Moore, R.D., Stahl, K., and
S.W. Taylor. 2008. Movement of outbreak populations of mountain pine beetle:
Influence of spatiotemporal patterns and climate. Ecography- Online Early.
Balloux, F., Lugon-Moulin, N., and J. Hausser. 2000. Estimating interracial gene flow
in hybrid zones: how reliable are microsatellites? Acta Theriologica 45: 93-10 I.
Balloux , F., and N. Lugon-Moulin. 2002. The estimation of population differentiation
with microsatellite markers. Mol. Ecol. 1 I: 155-65.
Bartell, N. 2007. Determining the macro-scale dispersal patterns of the mountain pine
beetle (Dendroctonus ponderosae) through a comparison of insect host- fungal
symbiont (Ophiostoma clavigerwn) genetic structure. Undergraduate ThesisUniversity of Northern BC. 49 pgs.
Beebee, T., and G. Rowe. 2004. An Introduction to Molecular Ecology. Oxford
University Press: NY.
Bentz, B.J., and M.W . Stock. 1986. Phenetic and phylogenetic relationships among ten
species of Dendroctonus bark beetles (Coleoptera: Scolytidae). Ann. Entomol.
Soc. Am. 79: 527-34.
Bematchez, L, and J.J. Dodson. 1991. Phytogeographic structure in mitochondrial DNA
of the Lake Whitefish (Coregonus clupeaformis) and its relation to Pleistocene
glaciations. Evolution 45: 1016-35.
146

Berryman, A.A. 1972. Resistance of conifers to invasion by bark beetle- fungal
associations. BioScience 22: 598-602.
Bert, T.M., Seyoum, S., Tringali, M.D., and A. McMillen-Jackson. 2002.
Methodologies for conservation assessments of the genetic biodiversity of aquatic
macro-organisms. Bra~. J. Bioi. 62: 387-408.
Borden. J.H. 1982. Aggregation pheromones. Pages 74-139 in J.B. Mitton and K.B.
Sturgeon, eds. Bark beetles in North American conifers: a system for the study of
evolutionary biology.
Botterweg, P.F. 1982. Dispersal and flight behavior of the spruce bark beetle Ips
tvpographus in relation to sex, size, and fat content. Z. Angew. Entomol. 94: 46689.
Burton, P.J . 2006. Restoration of forests attacked by mountain pine beetle: Misnomer.
misdirected, or must-do? BC J. Ecosvst. Manag. 7: 1- 10.
Busch, J.D., Waser. P.M ., and J.A. DeWoody. 2007. Recent demographic bottlenecks
are not accompanied by a genetic signature in banner-tailed kangaroo rats
(Dipodomys spectabilis). Mol. Ecol. 16: 2450-62.
Cabin, R.J., and R.J. Mitchell. 2000. To Bonferroni or not to Bonferroni: when and how
are the questions. ESA Bulletin 81: 246-8.
Calpas, J., Konschuh, M., and S. Lisowski. 2002. Preliminary investigation into
developing DNA fingerprints for "geographically distinct" populations of the
mountain pine beetle Dendroctonus ponderosae (Coleoptera: Scolytidae).
Alberta Agriculture Research Institute Project# 99£252 Final Report: 19 pgs.
Carroll, A.L., Taylor, S.W., Regniere, J., and L. Safranyik. 2004. Effects of climate
change on range expansion by the mountain pine beetle in British Columbia.
Pages 223-32 in T.L. Shore, J.E. Brooks and J.E. Stone, eds. Challenges and
Solutions: Proceedings of the Mountain Pine Beetle Symposium. Kelowna,
British Columbia. October 30-31, 2003. Natural Resources Canada, Canadian
Forest Service, Pacific Forestry Centre, Victoria, British Columbia, Information
Report BC-X-399. 298 p.
Carroll, A.L., Shore, T.L., and L. Safranyik. 2006. Direct control: theory and practice.
In: Safranyik, L., and B. Wilson, eds. The mountain pine beetle: a synthesis of
biology, management, and impacts on lodgepole pine. Natural Resources
Canada, Canadian Forest Service, Pacific Forestry Centre. Victoria, BC. pgs.
155-72.

147

Cerezke, H.F. 1989. Mountain pine beetle aggregation semiochemical use in Alberta
and Saskatchewan, 1983- 1987. Page 113 in G. D. Amman, Compiler.
Proceedings - Symposium on the management of lodgepole pine to minimize
losses to the mountain pine beetle, July I 2-14, 1988, Kalispell , Montana. USDA
Forest Service, Intermountain Research Station, Ogden, UT. General Technical
Report INT -262.
Chapman, J .A. 1962. Field studies on attack flight and log selection by the ambrosia
beetles Trypodendron lineatum (Oiiv.) (Coleoptera: Scolytidae). Can. Entomol.
94: 74-92.
Christiansen, F.B., and M.W. Feldman. 1986. Population genetics. Blackwell Scientific
Publications, Inc., Oxford.
Civetta, A., Vilardi. J.C., Saidman. 8.0 .. Leanza. C.A .. and J.L. Cladera. 1990.
Estimation of the number of ovipositing females per fruit in the Mediterranean
fruit fly Ceratitis capitata Wied. (Diptera: Tephritidae). Heredity 65: 59-66.
Cognato, A. I., Harlin. A.D .• and M.L. Fisher. 2003. Genetic structure among pinyon
pine beetle populations (Scolytinae: Ips cmzfusus). Environ. Entomol. 32: 126270.
Cole, W .E. 1981. Some risks and causes of mortality in mountain pine beetle
populations: a long-term analysis. Res. Popul. Ecol. 23: 116-44.
Conroy, C.J., and J.A. Cook. 2000. Phylogeography of a post-glacial colonizer:
Microllls /ongicaudus (Rodentia: Muridae). Mol. Ecol. 9: 165-75.
Cook, S.P .. and F.P. Hain. 1992. The influence of self-marking with fluorescent
powders on adult bark beetles (Coleoptera: Scolytidae). J. Entomol. Sci. 27: 26979.
Cooper. S.J .. Ibrahim, K.M., and G.M Hewitt. 1995. Postglacial expansion and genome
subdivision in the European grasshopper Chorthippus parallelus. Mol. Ecol. 4:
49-60.
Cwynar, L.C., and G.M. MacDonald. 1987. Geographical variation of lodgepole pine in
relation to population history. Am. Nat. 129: 463-9.
Dakin, E.E., and J.C. A vise. 2004. Microsatellite null alleles in parentage analysis.
Heredity 93: 504-9.
Duan, Y., Kerdelhue, C., Ye, H., and F. Lieutier. 2004. Genetic study of the forest pest
Tomicus piniperda (Col., Scolytinae) in Yunnan province (China) compared to
Europe: new insights for the systematics and evolution of the genus Tmnicus.
Heredity 93: 416-22.

148

Dupanloup, I., Schneider, S., and L. Excoffier. 2002. A simulated annealing approach to
define the genetic structure of populations. Mol. Ecol. II : 2571-81.
Elkin, C.M., and M.L. Reid. 2005. Low energy reserves and energy allocation decisions
affect reproduction by mountain pine beetles (Dendroctonus ponderosae). Funct.
Ecol. 19: I 02-9.
Ellegren, H. 2000. Heterogeneous mutation processes in human microsatellite DNA
sequences. Nat. Genet. 24: 400-2.
England, P.R., Osler. G.H., Woodworth. L.M., Montgomery. M.E .• Briscoe, D.A., and R.
Frankham. 2003. Effects of intense versus diffuse population bottlenecks on
microsatellite genetic diversity and evolutionary potential. Conservat. Genet. 4:
595-604.
Excoffier. L., Smouse, P.. and J. Quattro. 1992. Analysis of molecular variance inferred
from metric distances among DNA haplotypes: Application to human
mitochondrial DNA restriction data. Genetics 13 1: 479-91.
Excoffier. L.. Laval, G., and S. Schneider. 2005. Arlequin ver. 3.0: An integrated
software package for population genetics data analysis. Evol. Bioinformatics
Online I : 47-50.
Faccoli . M., Pi scedda, A .. Salvato, P., Simonato, M., Masutti. L .• and A. Battisti. 2005.
Genetic structure and phylogeography of pine shoot beetle populations (Tomicus
destruens and T. piniperda, Coleoptera, Scolytidae) in Italy. Ann. Forest. Sci. 62:
361-8.
Falush, D., Stephens, M., and J.K. Pritchard. 2003. Inference of population structure:
Extensions to linked loci and correlated allele frequencies. Genetics 164: 156787.
Falush, D., Stephens, M. , and J.K. Pritchard. 2007. Inference of population structure
using multilocus genotype data: dominant markers and null alleles. Mol. Ecol.
Notes 7: 574-8.
Fazekas, A.J., and F.C. Yeh. 2006. Postglacial colonization and population genetic
relationships in the Pinus contorta complex. Can. J. Bot. 84: 223-34.
Fettig, C.J., McMillin, J.D., Anhold, J.A., Shakeeb, M.H .. Borys, R.R .. Dabney, C.P. and
S.J. Seybold. 2006. The effects of mechanical fuel reduction treatments on the
activity of bark beetles (Coleoptera: Scolytidae) infesting ponderosa pine. Forest
Ecol. Manag. 230: 55-68.
Forsse, E. , and C.H . Solbreck. 1985. Migration in the bark beetle Ips typographus L.:
duration, timing and height of flight. J. Appl. Entomol. 100: 47-57.
149

Franceschi, V.R., Krokene, P., Christiansen, E., and T. Krekling. 2005. Anatomical and
chemical defenses of conifer bark against bark beetles and other pests. New
Phvtologist 167: 353-76.
Frankham, R. 2005. Resolving the genetic paradox in invasive species. Hereditv 94:
385.
Furniss, M.M., and R.L. Furniss. 1972. Scolytids (Coleoptera) on snowfields above
timberline in Oregon and Washington. Can. Entomol. I04: 1471-8.
Gaggiotti, O.E. 1996. Population genetic models of source-sink metapopulations.
Theor. Popul. Bioi. 50: 178-208.
Gannon. A., Sontag, S .. and Skov. K.. eds. 2007. Montana forest insect and disease
conditions and program highlights - 2006. USDA, Forest Service. Northern
Region. 79 pgs.
Geerts, B., and Q. Miao. 2005 . Airborne radar observations of the flight behavior of
small insects in the atmospheric convective boundary layer. Environ. Entomol.
34:361-77.
Goldhammer, D.S., Stephen, F.M .. and T.D. Paine. 1990. The effect of the fungi
Ceratocvstis minor (Hedgecock) Hunt, Ceratocvstis minor (Hedgecock) Hunt var.
Barrasii Taylor. and SJB 122 on reproduction of the southern pine beetle,
Dendroctonusfrontalis Zimmermann (Coleoptera: Scolytidae). Can. Entomol.
122: 407-18.
Goldstein, D.B., and C. Schlotterer (eds). 1999. Microsatellites: Evolution and
applications. Oxford University Press, Oxford.
Goudet, 1. 1995. Fstat version 1.2: a computer program to caJculate F statistics. J.
Heredity 86: 485-6.
Green, D.M., Sharbel, T.F., Kearsley, J., and H. Kaiser. 1996. Postglacial range
fluctuation, genetic subdivision and speciation in the western North American
spotted frog complex. Rana pretiosa. Evolution 50: 374-90.
Guo, S., and E. Thompson. 1992. Performing the exact test of Hardy-Weinberg
proportion for multiple alleles. Biometrics 48: 361-72.
Harrington, T.C. 1993. Diseases of conifers caused by species of Ophiostoma and
Leptographium. In: Ceratocystis and Ophiostoma: taxonomv, ecology, and
pathogenicity. Wingfield, M.J., Seifert, K.A., and J.F. Weber (eds.). American
Phytopathological Society Press, St. Paul, Minn. pgs. 161-71.

150

Harrington, T.C., Steimel, J.P., Wingfield, M.J., and G.A. Kite. 1996. Isozyme variation
and species delimitation in the Ceratocvstis coerulescens complex. Mycologia
88: 104-13.
Harter, A.V., Gardner. K.A .. Falush, D., Lentz, D.L.. Bye, R.A., and L.H. Rieseberg.
2004. Origin of extant domesticated sunflowers in eastern North America.
Nature 430: 201-5.
Hartl, D.L., and A.G. Clark. 1997. Principles of population genetics. 3rd. ed. Sinauer
Associates, Sunderland, Mass.
Hayes, J.P. and R.G. Harrison. 1992. Variation in mitochondrial DNA and the
biogeographic history of woodrats (Neotoma) of the eastern United States. Svst.
Bioi. 41 : 33 1-44.
Heusser. C.J. 1960. Late Pleistocene environments of north Pacific North America. Am.
Geogr. Soc. Spec. Pub/. No. 35.
Hewitt, G.M. 1996. Some genetic consequences of ice ages, and their role in divergence
and speciation. Bioi. J. Linn. Soc. 58: 247-76.
Hewitt, G.M. 1999. Postglacial recolonization of European Biota. Bioi. J. Linn. Soc.
68:87-112.
Hewitt, G.M. 2000. The genetic legacy of the Quaternary ice ages. Nature 405: 907-13.
Hewitt, G.M. 2004. Genetic consequences of climatic oscillations in the Quaternary.
Phil. Trans. R. Soc. Lond. B 359: 183-95.
Hey, J., Won, Y.J., Sivasundar. A., Nielsen. R .. and J.A. Markert. 2004. Using nuclear
haplotypes with microsatellites to study gene flow between recently separated
Cichlid species. Mol. Ecol. 13: 909-19.
Higby. P.K., and M.W. Stock. 1982. Genetic relationships between two sibling species
of bark beetle (Coleoptera: Scolytidae), Jeffrey pine beetle and mountain pine
beetle, in northern California. Ann. Entomol. Soc. Am. 75: 668-74.
Hindmarch, T.D., and M.L. Reid. 200 I. Thinning of mature lodgepole pine stands
increases Scolytid bark beetle abundance and diversity. Can. J. Forest Res. 31:
1502-12.
Holm, S. 1979. A simple sequential rejective multiple test procedure. Scand. J. Stat. 6:
65-70.

151

Hom. A., Roux-Morabito, G .• Lieutier. F., and C. Kerdelhue. 2006. Phytogeographic
structure and past history of the circum-Mediterranean species Tomicus destruens
Woll. (Coleoptera: Scolytinae). Mol. Ecol. 15: 1603-15.
Ibrahim. K.M .• Nichols, R.A .. and G.M. Hewitt. 1996. Spatial patterns of genetic
variation generated by different forms of dispersal during range expansion.
Heredity 77: 282-91.
Ibrahim , K.M. 200 I. Plague dynamics and population genetics of the desert locust: can
turnover during recession maintain population genetic structure? Mol. Ecol. I0:
581-91.
Idaho Forest Health Protection (author unknown). 2007. Idaho forest health highlights 2006. USDA. Forest Service. ID Forest Health Protection. 4 pgs.
Jackson, P.L.. Straussfogel. D .. Lindgren. B.S., Mitchell, S .. and B. Murphy. 2008.
Radar observation and aerial capture of mountain pine beetle, Dendroctonus
ponderosae Hopk. (Coleoptera: Scolytidae) in flight above the forest canopy.
Can. J. Forest Res. -i n press.
Jarne. P .. and P.J. Lagoda. 1996. Microsatellites. from molecules to populations and
back. Trends Ecol. Evol. I I : 424-9.
Johnson. J.A., Dunn, P.O .. and J.L. Bouzat. 2007. Effects of recent population
bottlenecks on reconstructing the demographic history of prairie-chickens. Mol.
Ecol. 16: 2203-22.
Jones, V.P., Hagler. J.R ., Brunner. J.F., Baker. C.C., and T .D. Wilburn . 2006. An
inexpensive immunomarking technique for studying movement patterns of
naturally occurring insect populations. Environ. Entomol. 35: 827-36.
Kalinowski, S.T. 2004. Counting alleles with rarefaction: Private alleles and
hierarchical sampling designs. Conservat. Genet. 5: 539-43.
Keller, L.F., Jeffery, K.J., Arcese, P., Beaumont, M.A., Hochachka, W.M .. Smith, J.N ..
and M.W. Bruford. 2001. Immigration and the ephemerality of a natural
population bottleneck: evidence from molecular markers. Proc. Roy. Soc. Land.
B Bioi. Sci. 268: 1387-94.
Kelley, S.T., and B.D. Farrell. 1998. Is specialization a dead end? The phylogeny of
host use in Dendroctonus bark beetles (Scolytidae). Evolution 52: 1731-43.
Kelley. S.T., Mitton, J.B., and T.D. Paine. 1999. Strong differentiation in mitochondrial
DNA of Dendroctonus brevicomis (Coleoptera: Scolytidae) on different
subspecies of ponderosa pine. Ann. Entomol. Soc. Am. 92: 193-7.

152

Kelley, S.T., Farrell, B.D., and J.B. Mitton. 2000. Effects of specialization on genetic
differentiation in sister species of bark beetles. Heredity 84: 218-27.
Kerdelhue, C., Roux-Morabito, G., Forichon, J., Chambon. J.-M., Robert. A., and F.
Lieutier. 2002. Population genetic structure of Tomicus piniperda L.
(Curculionidae: Scolytinae) on different pine species and validation ofT.
destruens (Woll.). Mol. Ecol. II: 483-94.
Kerdelhue. C .. Mondor-Genson, G .. Rasp Ius, J .- Y.. Robert. A., and F. Lieutier. 2003.
Characterization of five microsatellite loci in the pine shoot beetle Tomicus
piniperda (Coleoptera: Scolytidae). Mol. Ecol. Notes 3: I00-0 I.
Kerdelhue. C., Magnoux. E .. Lieutier. F., Roques. A .. and J. Rousselet. 2006.
Comparative population genetic study of two oligophagous insects associated
with the same hosts. Hereditv 97: 38-45.
Kim, J.J., Allen. E.A .. Humble. L.M., and C. Breuil. 2005. Ophiostomatoid and
basidiomycetous fungi associated with green, red. and grey lodgepole pines after
mountain pine beetle (Dendroctonus ponderosae) infestation. Can. J. Forest Res.

35: 274-84.
Kim, K.S., Cano-Rios, P .. and T.W. Sappington. 2006. Using genetic markers and
population assignment techniques to infer origin of boll weevils (Coleoptera:
Curculionidae) unexpectedly captured near an eradication zone in Mexico.
Environ. Entomol. 35: 813-26.
Kimmel, M., Chakraborty, R., King, J.P. Bamshad, M., Watkins, W.S., and L.B. Jorde.
1998. Signatures of population expansion in microsatellite repeat data. Genetics

148: 1921-30.
Klepzig, K.D., and R.T. Wilkens. 1997. Competitive interactions among symbiotic
fungi of the southern pine beetle. Appl. Environ. Microbial. 63: 621-7.
Knowles, L.L. 2001. Did the Pleistocene glaciations promote divergence? Tests of
explicit refugial models in montane grasshoppers. Mol. Ecol. I0: 691-70 I.
Korol, L., Shklar, G., and G. Schiller. 2002. Diversity among circum-Mediterranean
populations of Ieppo pine and differentiation from Brutia pine in their
isoenzymes: additional results. Silvae Genetica 51: 35-41.
Kurz, W.A., Dymond, C.C., Stinson, G., Rampley, G.J., Neilson, E.T., Carroll, A.L.,
Ebata, T., and L. Safranyik. 2008. Mountain pine beetle and forest carbon
feedback to climate change. Nature 452: 987-90.

153

Langor, D.W., and J.R. Spence. 1991. Host effects on allozyme and morphological
variation of the mountain pine beetle, Dendroctonus ponderosae Hopkins
(Coleoptera: Scolytidae). Can. Entomol. 123: 395-410.
Lee, S., Kim, J.J ., and C. Breuil. 2006a. Fungal diversity associated with the mountain
pine beetle, Dendroctonus ponderosae. and infested lodgepole pines in British
Columbia. Fungal. Divers. 22: 91-105.
Lee, S., Kim, J.J ., and C. Breuil. 2006b. Pathogenicity of Leptographium longiclavatum
associated with Dendroctonus ponderosae to Pinus contorta. Can. J. Forest Res.
36: 2864-72.
Lee, S., Hamelm, R.C ., Six, D.L., and C. Breuil. 2007. Genetic diversity and the
presence of two distinct groups in Ophiostoma { lavigerum associated with
Dendroctonus ponderosae in British Columbia and the Northern Rocky
Mountains. Phvtopathology 97: 1177-85.
Lindgren, B.S. 1983. A multiple funnel trap for Scolytid beetles (Coleoptera). Can.
Entomol. 115: 299-302.
Linton, D.A .. Safranyik. L., McMullen, L.H., and R. Betts. 1987. Field techniques for
rearing and marking mountain pine beetle for use in dispersal studies. Entomol.
Soc. BC 84: 53-58.
Logan, J .A., and J.A. Powell. 2001. Ghost forests, global warming, and the mountain
pine beetle (Coleoptera: Scolytidae). Am. Entomol. 47: 160-73.
Luikart, G., Allendorf. F.W .. Cornuet, J.M., and W.B. Sherwin. 1998. Distortion of
allele frequency distributions provides a test for recent population bottlenecks. J.
Heredity 89: 238-47.
Lunt, D.H ., Ibrahim, K.M ., and G.M. Hewitt. 1998. MtDNA phylogeography and postglacial patterns of subdivision in the meadow grasshopper Chorthippus
parallelus. Heredity 80: 630-641.
MacDonald, G.M ., and L.C. Cwynar. 1985 . A fossil pollen based reconstruction of the
late Quaternary history of lodgepole pine (Pinus contorta ssp. latifolia) in the
western interior of Canada. Can. J. Forest Res. 15 : I 039-44.
Mantel, N. 1967. The detection of disease clustering and a generalized regression
approach. Cane. Res. 27: 209-20.
Maroja, L.S ., Bogdanowicz, S.M. , Wallin, K.F., Raffa, K.F., and R.G. Harrison. 2007.
Phytogeography of spruce beetles (Dendroctonus rufipennis Kirby)
(Curculionidae: Scolytinae) in North America. Mol. Ecol. 16: 2560-73.

154

Marshall, H.D., Newton, C., and K. Ritland. 2002. Chloroplast phylogeography and
evolution of highly polymorphic microsatellites in lodgepole pine (Pinus
contorta). Theor. Appl. Genet. I 04: 367-78.
Martin. K .• Norris, A .. and M. Drever. 2006. Effects of bark beetle outbreaks on avian
biodiversity in the British Columbia interior: Implications for critical habitat
management. BC J. Ecosvst. Manag. 7: 10-24.
McGregor. M.D. 1985. Mountain pine beetle- the conflict between people and the
beetle. In: Loomis, R.C .. Tucker. S .. and T.H. Hofacker. eds. Insect and disease
conditions in the United States. USDA For. Sen•. General Technical Report
W0-46. Asheville. NC. USA: USDA Forest Service. 16-23.
Mock. K.E .. Bentz. B.J .• O'Neill. E.M .. Chong, J.P .. Orwin. J .. and M.E. Pfrender. 2007.
Landscape-scale genetic variation in a forest outbreak species. the mountain pine
beetle (Dendroctonus ponderosae). Mol. Ecol. 16: 553-68.
Moran, M.D. 2003. Arguments for reJecting the sequential Bonferroni in ecological
studies. Oikos I00: 403-5.
Nakagawa, S. 2004. A farewell to Bonferroni: the problems of low statistical power and
publication bias. Behav. Ecol. 15: I044-5.
Namkoong, G .. Roberds. J.H .. Nunnally. L.B .• and H.A. Thomas. 1979. Isozyme
variations in populations of southern pine beetles. Forest Sci. 25: 197-203.
Negron, J.F .. Schaupp Jr., W.C., Gibson, K.E., Anhold, J., Hansen. D .. Thier. R., and P.
Mocettini. 1999. Estimating extent of mortality associated with the Douglas-fir
beetle in the central and northern Rockies. West. J. Appl. Forest. 14: 121-7.
Nei, M. 1973. Analysis of gene diversity in subdivided populations. Proc. Nat!. Acad.
Sci. U.S.A. 70: 3321-3.
Nei, M., Maruyama, T., and R. Chakraborty. 1975. The bottleneck effect and genetic
variability in populations. Evolution 29: 1-10.
Nei, M. 1987. Molecular Evolutionary Genetics. Columbia University Press, New
York, NY, USA.
Nelson, T., Boots, B., Wulder, M.A., Shore, T., Safranyik, L.. and T. Ebata. 2006.
Rating the susceptibility of forests to mountain pine beetle infestations: the
impact of data. Can. J. Forest Res. 36: 2815-25.
Nelson, A., Hubert, J., Ramsey, A .. Ripley, K., and Omdal, D., eds. 2007a. Forest
health highlights in Washington- 2007. Washington State Department of
Natural Resources, Forest Health Program. 24 pgs.
155

Nelson, A., Flowers, R., Kanaskie, A., McWilliams, M., and Sprengel. K., eds. 2007b.
Forest health highlights in Oregon- 2006. Oregon Department of Forestry and
USDA Forest Service (Pacific Northwest Region). 16 pgs.
Nelson, T.A., Boots, B., Wulder, M.A., and A.L. Carroll. 2007c. Environmental
characteristics of mountain pine beetle infestation hot spots . BC 1. Ecosvst.
Manag. 8: 91 - 108.
Nilssen, A.C. 1984. Long-range aerial dispersal of bark beetles and bark weevils
(Coleoptera, Scoytidae and Curculionidae) in northern Finland. Annales
Entomologici Fennici 50: 37-42.
Owen, D.R., Lindahl, K.Q. Jr., Wood, D.L.. and J.R . Parmeter Jr. 1987. Pathogenicity
of fungi isolated from Dendroctonus valens, D. brevicumis, and D. ponderosae to
ponderosa pine seedlings. Phvtopathology 77 : 631-6.
Paetkau, D .. Waits, L.P .. Clarkson, P.L., Craighead, L.. and C. Strobeck. 1997. An
empirical evaluation of genetic distance statistics using microsatellite data from
bear (Ursidae) populations. Genetics 147: 1943-57.
Paine. T.D. , Raffa, K.F., and T.C. Harrington . 1997. Interactions among scolytid bark
beetles, their associated fungi, and live host conifers. Annu. Rev. Entomol. 42:
179-206.
Perkins. D.O., and B.C. Turner. 1988. Neurospora from natural populations: toward the
population biology of a haploid eukaryote. Exp. Mvcol. 12: 91-13 I.
Perneger. T.V. 1998. What's wrong with Bonferroni adjustments. Br. Med. J. 316:
1236-8.
Peteet, D.M. 1991. Postglacial migration history of lodgepole pine near Yakutat,
Alaska. Can. J. Bot. 69: 786-96.
Petit, R.J., Hampe, A .. and R. Cheddadi. 2005. Climate changes and tree
phylogeography in the Mediterranean. Taxon 54: 877-85.
Piry, S., Luikart, G., and J.M. Cornuet. 1999. Bottleneck: a computer program for
detecting recent reductions in the effective population size using allele frequency
data. }. Heredity 90: 502-3.
Preisler, H.K., and R.G. Mitchell. 1993. Colonization patterns of the mountain pine
beetle in thinned and unthinned lodgepole pine stands. Forest Sci. 39: 528-45 .
Pritchard, J.K., Stephens, M., and P. Donnelly. 2000. Inference of population structure
using multi locus genotype data. Genetics 155: 945-59.

156

Puillandre, N., Dupas, S., Dangles, 0 .• Zeddam, J.L., Capdevielle-Dulac. C., Barbin, K.,
Torres-Leguizamon. M .• and J.F. Silvain. 2008. Genetic bottleneck in invasive
species: the potato tuber moth adds to the Iist. Biological Invasions I0: 319-33.
Raymond, M., and F. Rousset. 1995. An exact test for population differentiation.
Evolution 49: 1280-3.
Reed. D.H., and R. Frankham. 2003. Correlation between fitness and genetic diversity.
Conservat. Bioi. 17: 230-7.
Reid, R.W . 1961. Moisture changes in lodgepole pine before and after attack by
mountain pine beetle. The Forestrv Chronicle 37: 368-75.
Reynolds. J., Weir. B.S .. and C.C. Cockerham. 1983. Estimation for the coancestry
coefficient: basis for a short-term genetic distance. Genetics 105: 767-79.
Rice. W .R. 1989. Analyzing tables of statistical tests. Evolution 43: 223-25.
Rice, A.V .• Thormann , M.N .• and D.W. Langor. 2008. Mountain pine beetle-associated
blue-stain fungi are differentially adapted to boreal temperatures. For. Path. 38:
113-23.
Ritzerow, S .• Konrad, H .. and C. Stauffer. 2004. Phylogeography of the Eurasian pine
shoot beetle Tmnicus piniperda (Coleoptera: Scol ytidae ). Eur. J. Entomol. I0 I:
13-9.
Roberds, J.H .• Hain. F.P., and L.B. Nunnally. 1987. Genetic structure of southern pine
beetle populations. Forest Sci. 33: 52-69.
Robertson, C., Nelson. T.A .• and B. Boots. 2007. Mountain pine beetle dispersal: the
spatial-temporal interaction of infestations. Forest Sci. 53: 395-405.
Robinson-Jeffrey. R.C., and R.W. Davidson. 1968. Three new Europhiwn species with
Verticicladiella imperfect states on blue-stained pine. Can. J. Bot. 46: 1523-7.
Rosenberg, N.A., Burke, T., Elo, K .• Feldman, M.W., Freidlin, P.J., Groenen, M.A.,
Hillel, J., Maki-Tanila, A., Tixier-Boichard, M., Vigna). A., Wimmers. K.. and S.
Weigend. 2001. Empirical evaluation of genetic clustering methods using
multilocus genotypes from 20 chicken breeds. Genetics 159: 699-713.
Rosenberg, N.A., Pritchard, J.K .. Weber. J.L., Cann. H.M ., Kidd, K.K., Zhivotovsky.
L.A. , and M.W. Feldman. 2002. The genetic structure of human populations.
Science 298: 2381-85.
Rosenberg, N. 2004. DISTRUCT: a program for the graphical display of population
structure. Mol. Ecol. Notes 4: 137-8.

157

Ross, K.G., Shoemaker. D.O., Kriegen, M.J., Detheer, C.J., and L. Keller. 1999.
Assessing genetic structure with multiple classes of molecular markers: a case
study involving the introduced fire ant Solenopsis invicta. Mol. Bioi. Evol. 16:
525-43.
Rousset, F. 1997. Genetic differentiation and estimation of gene tlow from F-statistics
under isolation by distance. Genetics 145: 1219-28.
Safranyik, L., Shrimpton, D.M ., and H.S. Whitney. 1974. Management of lodgepole
pine to reduce losses from the mountain pine beetle. Environment Canada,
Canadian Forestry Service. Pacific Forest Research Centre. Victoria, BC.
Forestry Technical Report I. 24 pgs.
Safranyik, L. 1978. Effects of climate and weather on mountain pine beetle populations.
Pages 77-84 in Kibbee. D.L., Berryman, A.A .• Amman. G.D., and R.W . Stark,
eds. Theory and practice of mountain pine beetle management in lodgepole pine
forests. Symposium Proceedings, University of Idaho, Moscow, ID.
Safranyik, L., Silversides, R. , McMullen. L.H., and D.A. Linton. 1989. An empirical
approach to modeling the local dispersal of the mountain pine beetle
(Dendroctonus pondernsae Hopk.) (Col., Scolytidae) in relation to sources of
attraction. wind direction and speed. J. Appl. Entomol. I08: 498-51 I.
Safranyik. L.. Linton. D.A .. Silversides, R., and L.H . McMullen. 1992. Dispersal of
released mountain pine beetles under the canopy of a mature lodgepole pine
stand. J. Appl. Entomol. 113: 441-50.
Safranyik, L., Nevill, R., and D. Morrison. 1998. Effects of stand density management
on forest insects and diseases. Pacific Forestry Centre- Technology Transfer
Note No. 12: 4 pgs.
Safranyik, L., Shore, T.L., and D.A. Linton. 1999. Attack by bark beetles (Coleoptera:
Scolytidae) following spacing of mature lodgepole pine (Pinaceae) stands. Can.
Entomol. 131 : 671-85.
Safranyik, L., Shore, T .L., Carroll, A.L., and D.A. Linton. 2004. Bark beetle
(Coleoptera: Scolytidae) diversity in spaced and unmanaged mature lodgepole
pine (Pinaceae) in southeastern British Columbia. Forest Ecol. Manag. 200: 2338.
Safranyik, L., and A.L. Carroll. 2006. The biology and epidemiology of the mountain
pine beetle in lodgepole pine forests. In: Safranyik, L., and B. Wilson, eds. The
mountain pine beetle: a synthesis of biology, management, and impacts on
lodgepole pine. Natural Resources Canada, Canadian Forest Service, Pacific
Forestry Centre. Victoria, BC. pgs. 3-66.

158

Salle, A., Arthofer, W., Lieutier, F., Stauffer, C., and C. Kerdelhue. 2007.
Phylogeography of a host-specific insect: genetic structure of Ips typographus in
Europe does not reflect past fragmentation of its host. Bioi. J. Linn. Soc. 90:
239-46.
Sal om, S.M., and J .A. McLean. 1990. Dispersal of Trvpodendron lineatum (Olivier)
within a valley setting. Can. Entomol. 122: 43-58.
Salom, S.M., and J.A. McLean. 1991. Environmental influences on dispersal of
Trvpodendron lineatum (Coleoptera: Scolytidae). Em•iron. Entomol. 20: 565-76.
Salvador., L., Alia .. R .• Agundez. D .. and L. Gil. 2000. Genetic variation and migration
pathways of maritime pine (Pinus pinaster A it.) m the Iberian peninsula. Theor.
Appl. Genet. I00: 89-95.
Sambrook, J .. and D.W. Russell. 2001. Molecular Cloning: A Laboratorv Manual. 3'd
Edition, Vol. 1-3. Irwin, N .. and K.A . Janssen, eds. Cold Spring Harbor, NY:
Cold Spring Harbour Laboratory Press.
Samman, S .. and Logan. J .. tech. eds. 2000. Assessment and response to bark beetle
outbreaks in the Rocky Mountain area. Report to Congress from Forest Health
Protection, Washington Office, Forest Service. US Department of Agriculture.
Gen. Tech. Rep. RMRS-GTR-62. Ogden, UT: USDA, Forest Service, Rocky
Mountain Research Station. 46 pgs.
Santucci, F .. Emerson, B.C., and G.M. Hewitt. 1998. Mitochondrial DNA
phylogeography of European hedgehogs. Mol. Ecol. 7: I 163-72.
Schmid, J.M. 1969. Laphria gilva (Diptera: Asilidae), a predator of Dendroctonus
ponderosae in the Black Hill s of South Dakota. Ann. Entomol. Soc. Am. 62:
1237-41 .
Schmid, J .M., Mata. S.A., and D.C. Allen. 1992. Potential influences of horizontal and
vertical air movement in ponderosa pine stands on mountain pine beetle dispersal.
USDA Forest Service: Research Note RM-516. 4 pgs.
Schmid, J .M. and S.A. Mata. 1996. Natural variability of specific forest insect
populations and their associated effects in Colorado. General Technical Report
RM-GTR-275. Fort Collins, CO: USDA, Forest Service, Rocky Mountain Forest
and Range Experiment Station. 14 pgs.
Shore, T., Safranyik, L., and J. Lemieux . 2000. Susceptibility of lodgepole pine stands
to the mountain pine beetle: testing of a rating system. Can. J. Forest Res. 30:
44-9.

159

Shore, T.L., Safranyik, L., and R.J. Whitehead. 2006. Principles and concepts of
management. In: Safranyik, L., and B. Wilson, eds. The mountain pine beetle: a
synthesis of biology, management, and impacts on lodgepole pine. Natural
Resources Canada, Canadian Forest Sen,ice, Pacific Forestrv Centre. Victoria,
BC. pgs. 117-22.
Shriver. M.D., Jin. L.. Chakraborty. R., and E. Boerwinkle. 1993. VNTR allele
frequency distributions under the stepwise mutation model: a computer
simulation approach. Genetics 134: 983-93.
Simoni. L., Gueresi, P .. Pettener. D., and G. Barbujani. 1999. Patterns of gene now
inferred from genetic distances in the Mediterranean region. Human Bioi. 71 :
399-415.
Six. D.L., and Paine. T.D. 1996. A technique for the introduction of fungi to bark beetle
mycangia. J. Entomol. S£ i. 3 I : 466-8.
Six , D.L., and T.D. Paine. 1998. Effects of mycangial fungi and host tree species on
progeny survival and emergence of Dendroctonus ponderosae (Coleoptera:
Scolytidae). Environ. Entomol. 27: 1393-140 I .
Six, D.L., and T.D. Paine. 1999. Allozyme diversity and gene flow in Ophiostoma
clavigerum (Ophiostomatales: Ophiostomataceae), the mycangial fungus of the
Jeffrey pine beetle, Dendroctonus jeffreyi (Coleoptera: Scolytidae). Can. J.
Forest Res. 29: 324-31 .
Six. D.L., Paine. T.D., and J.D. Hare. 1999. Allozyme diversity and gene now in the
bark beetle, Dendroctonusjeffreyi (Coleoptera: Scolytidae). Can. J. Forest Res.
29: 315-323.
Six, D.L. 2003. A comparison of mycangial and phoretic fungi of individual mountain
pine beetles. Can. J. Forest Res. 33: 1331-4.
Slatkin, M. 1995. A measure of population subdivision based on microsatellite allele
frequencies. Genetics 139: 457-62.
Smouse, P.E., Long, J.C., and R.R . Sokal. 1986. Multiple regression and correlation
extensions of the Mantel Test of matrix correspondence. Syst. Zoo/. 35: 627-32.
Solheim, H. 1995. Early stages of blue-stain fungus invasion of lodgepole pine
sapwood following mountain pine beetle attack. Can J. Bot. 73: 70-74.
Solheim, H., and P. Krokene. 1998. Growth and virulence of mountain pine beetle
associated blue-stain fungi, Ophiostoma clavigerum and Ophiostoma montium.
Can J. Bot. 76: 561-66.

160

Soranzo, N .• Alia, R., Provan, J., and W. Powell. 2000. Patterns of variation at
mitochondrial sequence-tagged-site provide new insights into the postglacial
history of European Pinus svlvestris populations. Mol. Ecol. 9: 1205- 1 I.
Spieth, P.T. 1975. Population genetics of allozyme variation in Neurospora intermedia.
Genetics 80: 785-805 .
Spong, G .. and L. Hellborg. 2002. A near-extinction event in lynx : do microsatellite
data tell the tale? Conservat. Ecol. 6: 15.
Stauffer. C.H ., Leitinger. R.. Simsek. Z .. Schreiber. J.D .. and E. FUhrer. 1992. Allozyme
variation among nine Austrian Ips typographus L. (Col., Scolytidae) populations.
J. Appl. Entomol. 114: 17-25 .
Stauffer. C ., Lakatos. F .. and G . Hewitt. 1999. Phytogeography and postglacial
colonization routes of Ips f\pographtH L. (Coleoptera. Scolytidae). Mol. Ecol. 8:
763-73.
Steinke, J. 2006. The investigation of dispersal patterns and genetic relatedness in
geographically separated mountain pine beetle (Dendroctonus ponderosae)
populations by microsatellite analysis validation of existing markers.
Undergraduate Thesis - University of Northern BC. 29 pgs.
Stock, M.W .. Pitman. B.. and J.D. Guenther. 1979. Genetic differences between
Douglas-fir beetles (Dendroc tonus pseudotsugae) from Idaho and coastal
Oregon. Ann. Entomol. Soc. Am. 72: 394-97.
Stock, M.W., and J.D. Guenther. 1979. Isozyme variation among mountain pine beetle
(Dendroctonus ponderosae) populations in the Pacific Northwest. Environ.
Entomol. 8: 889-93.
Stock, M.W., and G.D. Amman. 1980. Genetic differentiation among mountain pine
beetle populations from lodgepole pine and ponderosa pine in northeast Utah.
Ann. Entomol. Soc. Am. 73: 472-78.
Stock, M.W., Amman, G.D .. and P.K . Higby. 1984. Genetic variation among mountain
pine beetle (Dendroctonus ponderosae) (Coleoptera: Scolytidae) populations
from seven Western States. Ann. Entomol. Soc. Am. 77: 760-64.
Stock, M.W., and G.D. Amman. 1985. Host effects on the genetic structure of mountain
pine beetle, Dendroctonus ponderosae, populations. Proceedings of JUFRO
Conference on: The Role of the Host in the Population Dynamics of Forest
Insects: 83-95.

161

Sturgeon, K.B., and J.B. Mitton. 1986. Allozyme and morphological differentiation of
mountain pine beetles Dendroctonus ponderosae Hopkins (Coleoptera:
Scolytidae) associated with host tree. Evolution 40: 290-302.
Taberlet, P., and J. Bouvet. 1994. Mitochondrial DNA polymorphism, phylogeography,
and conservation genetics of the brown bear ( Ursus arctos) in Europe. Proc. Rov.
Soc. Lon. B 255: 195-200.
Taylor. S.W., and A.L. Carroll. 2004. Disturbance, forest age, and mountain pine beetle
dynamics in British Columbia: a historical perspective. Pages 41-51 in T .L.
Shore, J.E. Brooks and J.E. Stone, eds. Proceedings of the Mountain Pine Beetle
Symposium: Challenges and Solutions. Kelowna, British Columbia, Canada.
October 30-31, 2003. Natural Resources Canada. Canadian Forest Service,
Pacific Forestry Centre. Victoria. BC, Information Report BC-X-399. 298 p.
Taylor. S.W .. Carroll, A.L .. Alfaro, R.I., and L. Safranyik. 2006. Forest, climate and
mountain pine beetle outbreak dynamics in Western Canada. In : Safranyik. L. ,
and B. Wilson. eds. The mountain pine beetle: a synthesis of biology,
management. and impacts on lodgepole pine. Natural Resources Canada,
Canadian Forest Service. Pacifi( Forestrv Centre. Victoria. BC. pgs. 67-94.
Thoeny, W.T .. Tiarks, A.E., Hayes, J.L., and J.R. Bridges. 1992. Marking the southern
pine beetle (Coleoptera: Scolytidae) with rubidium within loblolly pine for
dispersal studies. Environ. Entomol. 21: 1377-85.
Uunila, L.. Guy. B.. and R. Pike. 2006. Hydrologic effects of mountain pine beetle in
the interior pine forests of British Columbia: key questions and current
knowledge. Streamline Watershed Management Bulletin 9: 1-6.
Vasconcelos, T ., Horn. A., Lieutier, F .. Branco, M., and C. Kerdelhue. 2006.
Distribution and population genetic structure of the Mediterranean pine shoot
beetle Tomicus destruens in the Iberian Peninsula and Southern France. Agr.
Forest Entomol. 8: 103-11.
Verhoeven, K.J ., Simonsen, K.L., and L.M. Mcintyre. 2005. Implementing false
discovery rate control: increasing your power. Oikos I08 : 643-7.
Vucetich, J.A., and T.A. Waite. 2003. Spatial patterns of demography and genetic
processes across the species' range: Null hypotheses for landscape conservation
genetics. Conservat. Genet. 4: 639-45.
Wagner, T .L., Gagne, J.A., Doraiswamy, P.C., Coulson. R.N., and K.W . Brown. 1979.
Development time and mortality of Dendroctonusfrontalis in relation to changes
in tree moisture and xylem water potential. Environ. Entomol. 8: I 129-38.

162

Wagner. W.L., Wilson, B., Peter. B., Wang, S .. and B. Stennes. 2006. Economics in the
management of mountain pine beetle in lodgepole pine in British Columbia: A
synthesis. In : Safranyik, L., and B. Wilson. eds. The mountain pine beetle: a
synthesis of biology. management, and impacts on lodgepole pine. Natural
Resources Canada, Canadian Forest Sen,ice, Pacific Forestn.' Centre. Victoria,
BC. Pgs. 277-300.
Wallis, G.P .. and J.W. Arntzen. 1989. Mitochondrial DNA variation in the crested newt
superspecies: limited cytoplasmic gene flow among species. Evolution 43: 88104.
Waring, R.H., and G.B. Pitman. 1985. Modifying lodgepole pine stands to change
susceptibility to mountain pine beetle attack. Ecologv 66: 889-97.
Wattier. R.. Engel. C.R .. Saumitou-Laprade. P .. and M. Valero. 1998. Short allele
dominance as a source of heterozygote deficiency at microsatellite loci :
Experimental evidence at the dinucleotide locus Gv I CT in Gracilaria gracilis
(Rhodophyta). Mol. Ecol. 7: 1569-73.
Weir. B.S .. and C.C. Cockerham. 1984. Estimating F-statistics for the analysis of
population structure. Evolution 38: 1358-70.
Westfall. J. 2006. 2006 Summary of forest health conditions in British Columbia. BC
Ministry of Forests and Range: Forest Practices Branch. 73 pgs.
Westfall, J., and T. Ebata. 2007. 2007 Summary of forest health conditions in British
Columbia. BC Ministrv of Forests and Range: Forest Practices Branch . 81 pgs.
Wheeler, N.C .. and R.P. Guries. 1982. Biogeography of lodgepole pine. Can. J. Bot.
60: 1805- 14.
Whitehead, R.J., Safranyik, L., Russo, G.L., Shore, T.L., and A.L. Carroll. 2004.
Silviculture to reduce landscape and stand susceptibility to mountain pine beetle.
Pages 233-244 in T.L. Shore, J.E. Brooks and J.E. Stone, eds. Proceedings of the
Mountain Pine Beetle Symposium: Challenges and Solutions. Kelowna, British
Columbia, Canada. Oct. 30 -31, 2003. Natural Resources Canada, Canadian
Forest Service, Pacific Forestry Centre, Victoria, BC. Information Report BC-X399. 298 p.
Whitehead, R.J .. and G.L. Russo. 2005. "Beetle-proofed" lodgepole pine stands in
interior British Columbia have less damage from mountain pine beetle. Natural
Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria.
BC. Information Report BC-X-402. 17 p.

163

Whitehead, R.J., Safranyik, L., and T.L. Shore. 2006. In: Safranyik, L., and B. Wilson,
eds. The mountain pine beetle: a synthesis of biology, management, and impacts
on lodgepole pine. Natural Resources Canada, Canadian Forest Service, Pacific
Forestry Centre. Victoria, BC. pgs. 173-92.
Whitney. H.S .. Bandoni, R.J .. and F. Oberwinkler. 1987. Entomocorticiwn dendroctoni
gen. et. Sp. Nov. (Basidiomycotina). a possible nutritional symbiote of the
mountain pine beetle in lodgepole pine in British Columbia. Can. J. Bot. 65: 95102.
Wood, S.L. 1982. The bark and ambrosia beetles of North and Central America
(Coleoptera: Scolytidae), a taxonomic monograph. Great Basin Nat. Mem. No.6.
Wood, C.S .. and L. Unger. 1996. Mountam pine beetle- a history of outbreaks in pine
forests in British Columbia. 1910 to 1995 . Natural Resources Canada, Canadian
Forest Service. Victoria. BC. 61 pgs.
Wright, S. 1951 . The genetical structure of populations. Ann. Eugen. 15: 323-54.
Wright, S. 1978. Evolution and the Genetu s of Populations: Experimental Results and
Evolutionary Deductions. University of Chicago Press. Chicago.
Wulff. J.L. 1985. Clonal organisms and the evolution of mutualism. In Population
biology and evolution of clonal organisms. Jackson. J .B., Buss, L.W ., and R.E.
Cook (eds). Yale University Press. New Haven, Conn. pp. 437-66.
Xu. X .. Peng, M .. Fang, Z., and X. Xu. 2000. The direction of microsatellite mutations
is dependent upon allele length . Nat. Genet. 24: 396-9.
Yamaoka, Y .. Hiratsuka, Y .. and P.J . Maruyama. 1995. The ability of Ophiostoma
clavigerum to kill mature lodgepole pine trees. Eur. J. Forest Pathol. 25: 401-04.
Yoon Chung, M .. Suh, Y., Oopez-Pujol. J., Nason. J.D., and M.G. Chung. 2005. Clonal
and fine-scale genetic structure in populations of a restricted Korean endemic,
Hostajonesii (Liliaceae) and the implications for conservation. Ann. Bot. 96:
279-88.
Yuill, J.S. 1941. Cold hardiness of two species of bark beetles in California forests. J.
Econ. Entomol. 34: 702-9.
Zhang, D.X., and G.M . Hewitt. 2003. Nuclear DNA analyses in genetic studies of
populations: practice, problems and prospects. Mol. Ecol. 12: 563-84.

164