EVALUATING THE RESILIENCE OF NORTHERN INTERIOR CEDARHEMLOCK FORESTS TO WESTERN HEMLOCK LOOPER DEFOLIATION
EVENTS
by
Michelle Janette Connolly
B.Sc., University of British Columbia, 2006

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTERS OF SCIENCE
IN
NATURAL RESOURCES AND ENVIRONMENTAL STUDIES

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
February, 2014

© Michelle Connolly, 2014

UMI Number: 1525793

All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

Di!ss0?t&iori Publishing
UMI 1525793
Published by ProQuest LLC 2014. Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code.

ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346

ABSTRACT

Forest response to western hemlock looper outbreaks in the inland temperate rainforest
(ITR) can be quite variable. Sixty-three plots sampled 17 years after collapse o f the 199194 outbreak east o f Prince George, British Columbia, revealed multiple aspects o f
ecosystem resilience. Overstory recovery appears to be greatest in stands with lower site
productivity and in stands with a more severe outbreak history. Forest regeneration is
strongly constrained by shrub cover, which is promoted by canopy openness and soil
nutrients, both o f which can be accentuated by defoliators. Site index and soil moisture
regime are primarily responsible for niche differentiation among regenerating tree
species. Management for tree production should consider differences in site index, site
series, post-outbreak canopy density and annual heat:moisture index among stands. These
patterns can be mapped and extrapolated throughout the ITR, but forest management
options after a defoliation event must be evaluated on a site by site basis.

ACKNOWLEDGEMENTS

I would like to express my gratitude to my supervisor Phil Burton for enabling this research
opportunity. He encouraged me to think independently and was a patient mentor. He was
also an encyclopaedic resource on all matters forest and ecology related.
Darwyn Coxson shared his lab’s resources with me, which allowed me to carry out the
statistical analyses necessary for this study. Susan Stevenson provided constructive feedback
throughout the research process. I want to thank both these researchers for being on my
committee. I would also like to thank Ellen Macdonald for prompting improvements in the
final document.
I would like to thank the Pacific Institute for Climate Solutions for providing the financial
backing for this study, and the Natural Resource and Environmental Studies program at the
University o f Northern B.C. for creating the academic environment in which it could take
place. I was able to thrive in the NRES program in a way that wouldn’t have been possible at
a different university.
Dan Crawford with the B.C. Forest Service provided me with the field maps necessary to do
this study. Work would have been far more difficult without his easy-to-use and understand
maps. Tammy Baerg was instrumental in facilitating this relationship and introducing me to
Stacy Perkins. These women are foresters I respect very much.
Sybille Haeussler generously answered my questions about shrubs, Craig DeLong shared his
knowledge of ITR site series whenever I asked for it and Bruce Rogers enthusiastically
imparted his experience with the ecology o f the ITR in the field. Paul Sanbom helped me
acquire equipment I need to dig soil pits and obligingly shared his familiarity with ITR soils
with me. I am grateful to have had these experts to rely on for help with this study.
Chris Konchalski kindly let me traipse around his research plots before my own study began,
and showed me parts o f the inland rainforest I probably wouldn’t have stumbled upon on my
own.
Alana Clason and Benoit Gendreau-Bethiaume pointed me in the right direction (several
times) about statistics. I appreciate their collegiality and frankness.
Thank you to my trusty and diligent field technician Matt Colley. My friend and office-mate
Benita Kaytor tolerated me asking her lots o f questions and made graduate school that much
more fun. I also appreciate the support o f Maria Walsh.
I want to thank Nowell Senior and the Caledonia Ramblers for being so devoted to the
conservation of the ITR, and for their willingness to share their experience o f this ecosystem
with strangers.
Lastly, thank you to my sister Allison and my parents.

TABLE OF CONTENTS

ABSTRACT.......................................................................................................................................i
ACKNOWLEDGEMENTS............................................................................................................ ii
TABLE OF CONTENTS............................................................................................................... iii
LIST OF TABLES...........................................................................................................................v
LIST OF FIGURES.........................................................................................................................vi
1.0 Introduction................................................................................................................................ 1
1.1 Herbivory as a disturbance....................................................................................................1
1.2 Study objectives..................................................................................................................... 4
1.3 Biology and ecology o f the hemlock looper........................................................................4
1.4 Hemlock looper as an agent o f disturbance.........................................................................6
1.5 Hemlock looper defoliation history, geography and control............................................. 8
1.5.1 Western hemlock looper in British Columbia............................................................. 9
1.6 Tree mortality, survival and stand development................................................................11
1.6.1 Overstory.......................................................................................................................11
1.6.2 Stand development and succession............................................................................. 14
1.6.3 Understory.....................................................................................................................16
1.6.3.1 Coarse woody debris in a defoliated stand......................................................... 19
1.7 Nutrients and defoliation.................................................................................................... 20
1.8 Research questions..............................................................................................................22
2.0 M ethods.................................................................................................................................... 24
2.1 Study a rea.............................................................................................................................24
2.3 Data preparation.................................................................................................................. 30
2.4 Statistical analyses............................................................................................................... 34
3.0 R esults...................................................................................................................................... 41
3.1 Summary o f site variables.................................................................................................. 41
3.2 Summary o f overstory.........................................................................................................43
3.3 Summary o f understory...................................................................................................... 47
3.3.1 Plant community...........................................................................................................47
3.3.2 Regeneration................................................................................................................. 51
3.4 Overstory recovery............................................................................................................53
3.4.1 Effect of outbreak
severity and site variables................................................. 53
3.5 Understory recovery............................................................................................................55
3.5.1 Effect of outbreak
severity...............................................................................55
3.5.2 Effect of site variables................................................................................................ 56
3.5.3 Plant community...........................................................................................................63
3.5.3.1 Cluster analysis..................................................................................................... 67
3.5.3.2 Plant community and outbreak severity............................................................. 71
3.5.4 Regeneration................................................................................................................. 71
3.5.4.1 Regeneration niche............................................................................................... 85
3.5.4.2 Regeneration substrate..........................................................................................90
4.0 Discussion................................................................................................................................. 95

4.1 Overstory recovery.............................................................................................................. 95
4.1.1 Effect o f outbreak severity and site variables............................................................ 95
4.2 Understory recovery............................................................................................................97
4.2.1 Effect o f outbreak severity..........................................................................................97
4.2.2 Effect o f site variables................................................................................................. 99
4.2.3 Plant community......................................................................................................... 101
4.2.3.1 Cluster analysis....................................................................................................102
4.2.3.2 Plant community and outbreak severity........................................................... 106
4.2.4 Regeneration............................................................................................................... 106
4.2.4.1 Regeneration niche............................................................................................. 107
4.2.4.2 Regeneration substrate........................................................................................ 109
5.0 Conclusions and synthesis.....................................................................................................110
6.0 Management of defoliated stands......................................................................................... 112
References.....................................................................................................................................116

LIST OF TABLES

Table 1. Study-wide summary o f site variables measured at each plot (n-63)........................ 41
Table 2. Overstory density by species. Mean density values are from tree counts in 0.04 ha
plots (n=63).............................................................................................................................44
Table 3. Overstoiy basal area by species. All dead trees are trees that likely died directly or
indirectly from the WHL defoliation in the 1990s..............................................................44
Table 4. All plant species, listed by functional group, found in stands sampled for this study
(n=63), with species codes for plant community RDA output interpretation and cluster
analysis interpretation............................................................................................................ 48
Table 5. Total and well-spaced regeneration by species and size class (n=63)........................51
Table 6. Total regeneration summaries by species, size class and substrate.............................52
Table 7. Well-spaced regeneration summaries by species and size class..................................53
Table 8. Spearman’s correlations between Overstory Recovery Index (ORIave) and site
variables.................................................................................................................................. 54
Table 9. Spearman’s correlations between three understory recovery responses and two
indices o f defoliation severity, the Defoliation Severity Index (DSI) and the proportion
o f dead stems.......................................................................................................................... 56
Table 10. Spearman’s correlations between understory recovery variables and site variables.
.................................................................................................................................................57
Table 11 a) Summary of the independent effects o f all individual explanatory variables on
overall understory plant community composition; b) Summary of the conditional effect
o f each predictor.....................................................................................................................64
Table 12. Summary o f plant community ordination forward selection. Only significant
predictors (p<0.05) were retained.........................................................................................65
Table 13. Summary o f plant community ordination results from the final constrained RDA.
.................................................................................................................................................65
Table 14. Spearman’s correlations for the total number o f total understory trees by size class
and two classes o f shrub cover..............................................................................................75
Table 15. Spearman’s correlations between the number of well-spaced understory trees by
size class and two classes o f shrub cover.............................................................................80
Table 16. a) Summary of the independent effects o f all explanatory variables on overall
regeneration density by species and size class; b) Summary o f the conditional effect of
each predictor........................................................................................................................ 87
Table 17. Summary o f tree regeneration niche ordination results from forward selection.
Only significant predictors (p<0.05) were retained............................................................ 88
Table 18. Summary o f results from the final constrained RDA for tree regeneration niche.. 88
Table 19. Spearman’s correlations for regeneration density (stems/ha) on coarse woody
debris by size class and two classes o f shrub cover............................................................ 93
Table 20. Spearman’s correlations for the proportion of total stems on coarse woody debris
by size class and two classes o f shrub cover....................................................................... 94

v

LIST OF FIGURES

Figure 1. Examples of “light”, “moderate” and “severe” categories of defoliation, from the
Forest Health Aerial Overview Survey Standards for British Columbia..........................26
Figure 2. Representation of plot and subplot layout....................................................................27
Figure 3. Mean live stems/ha and dead stems/ha tallied according to their health and decay
class......................................................................................................................................... 45
Figure 4. Mean diameter at breast height o f live and dead stems by species (mean, +/- SE). 45
Figure 5. Tree species by diameter at breast height (mean, +/- SE) by subzone: a) live; and b)
dead..........................................................................................................................................46
Figure 9. Scatterplots with linear least-squares regression lines for total regeneration density
as a function of: a) canopy density; and b) site index.........................................................60
Figure 10. Scatterplots with linear least-squares regression lines for well-spaced regeneration
density as a function of: a) steepness; b) annual heat: moisture index; c) site index; and
d) canopy density....................................................................................................................62
Figure 11. a) Redundancy analysis (RDA) correlation biplot for plant community response
variables and explanatory variables identified during forward selection; and b) only
plant community response.................................................................................................... 66
Figure 12. Two-way similarity analysis o f defoliated stands in the ITR, using a flexible beta
linkage method and Sorenson dissimilarity coefficient with PC-ORD.............................69
Figure 13. Plant community RDA showing the arrangement o f understory shrubs and trees
and plots in ordination space.................................................................................................70
Figure 15. Scatterplots with linear least-squares regression lines for well-spaced regeneration
density and percent competitive shrub cover.......................................................................74
Figure 16. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) total Tsuga
seedlings; b) total Tsuga saplings; and c) total Tsuga (all size classes)............................76
Figure 17. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) total Abies
seedlings; b) total Abies saplings; and c) total Abies (all size classes)..............................78
Figure 18. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) total seedlings
(all species); and b)total saplings (allspecies).....................................................................79
Figure 19. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) well-spaced
Tsuga seedlings; b) well-spaced Tsuga saplings; and c) well-spaced Tsuga (all size
classes).....................................................................................................................................81
Figure 20. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) well-spaced
Abies seedlings; b) well-spaced Abies saplings; and c) well-spaced Abies (all size
classes).....................................................................................................................................83
Figure 21. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) well-spaced
seedlings (all species); and b) well-spacedsaplings (all species).......................................84

Figure 22. a) Redundancy analysis (RDA) correlation biplot for tree regeneration density by
species and size class response variables and explanatory variables identified during
forward selection.................................................................................................................... 88
Figure 23. Total regeneration density RDA showing the arrangement of understory tree
species by size class and plots in ordination space............................................................. 89
Figure 24. Mean tree regeneration density (+ S.E.) by species, size class and substrate
92
Figure 25. Forest stand types representative of cluster analysis groupings: groups A) to D).
............................................................................................................................................... 105

1.0 Introduction

Projections of future climatic conditions in the forests o f western North America consistently
anticipate greater suitability for outbreaks of both bark beetles and defoliators (Ayres and
Lombardero 2000, Haughian et al. 2012, Weed at al. 2013). The western hemlock looper
(WHL; Lambdina fiscellaria spp. lugubrosa Hulst) is a Lepidopteran defoliator that acts as a
landscape-scale disturbance in the Interior Cedar-Hemlock (ICH) biogeoclimatic zone
(Ketcheson et al. 1991) of British Columbia, Canada, an area also known as the Inland
Temperate Rainforest (ITR; Stevenson et al. 2011). The WHL defoliation events o f the
1950s and 1990s continue to influence current characteristics o f the ITR, and this study aims
to ecologically evaluate the consequencess o f this natural disturbance, and to forecast the
recovery potential of these forests should insect outbreaks become more frequent and more
severe, as predicted.

In particular, sampling and analysis was conducted to determine what outbreak, site and
stand conditions are associated with stands in which the overstory tree cover has recovered,
stands in which tree regeneration or regeneration release has occurred, and stands which have
become ingrown with dense shrub cover.

1.1 Herbivory as a disturbance

Herbivory is the consumption o f living plant parts by animals (Schowalter 2000). It is a key
ecosystem process that reduces density o f plants or plant materials, transfers biomass and

1

nutrients to other trophic levels and eventually the soil, and affects habitat and resource
conditions for other organisms (Mattson and Addy 1975, Bardgett et al. 1998, Schowalter
2000). Defoliation by insects is a type o f herbivory (also called folivory or phytophagy), and
is the direct consumption o f photosynthetically active material (Schowalter 2000). It has
been argued that phytophagous insects occupy a special “regulator” position in some
ecosystems because they affect leaves, the primary site o f energy and biochemical synthesis
in plants (Mattson and Addy 1975).

Whether herbivory is seen as a normal trophic process (Schowalter 1985, Schowalter and
Lowman 1999, Willig and Walker 1999) or a disturbance (Pickett and White 1985), a
defoliation event brought about by especially high population densities o f an herbivorous
insect can dramatically alter ecosystem structure and function over large areas. As a
disturbance, the effect o f herbivorous insects is not uniform, and varies in severity across
forest stands (Stone and Wolfe 1996). Low or moderate levels o f herbivory sometimes
stimulate productivity through an over-compensation response by plants (McNaughton 1979,
Belsky 1986, Dyer et al. 1993, Trumble et al. 1993) or by cascading trophic interactions
(Carpenter et al. 1985). Severe herbivory usually results in decreased fitness or mortality
(Marquis 1984, Williamson et al. 1989) by reducing a plant’s ability to carry out necessary
metabolic processes (Kessler and Baldwin 2002).

Entomologists who study the population dynamics o f insects have defined some species as
“outbreaking” (also known as epidemic) and others as “non-outbreaking” (also known as
endemic) (Wallner 1987). This difference has also been described as a matter o f population

2

fluctuation versus population balance (Richards 1961). Outbreak behavior is what
distinguishes a disturbance event from the background herbivory that constrains primary
productivity.

Herbivory by insects can influence the structure and composition o f forest stands by causing
mortality among some species or sizes o f trees and not others (Evenden 1940, Mattson and
Addy 1975, Alfaro et al. 1982, Romme et al. 1986, Knight 1987, Alfaro et al. 1999), thereby
reducing the abundance of the host species and allowing other species or cohorts to flourish
(Wohlgemuth et al. 2001). In addition to causing the direct mortality of canopy and sub­
canopy trees and sometimes shrubs (Jardine 1969, Harris et al., 1982, Alfaro et al. 1999,
McCloskey 2007), a defoliating insect outbreak will affect the way a forest develops by
reducing plant density, opening up the canopy, stimulating or limiting the growth o f plants,
transferring nutrients and altering resource conditions for soil organisms (Bardgett et al.
1998, Schowalter 2000). Along with other variables such as light, moisture and temperature,
the presence and accessibility of soil nutrients over different scales will have both immediate
and gradual consequences for the makeup o f plant communities present at any point in time
following a defoliation event (Uriarte 2000).

In this study, “defoliation event” and “outbreak” will refer to the years o f a sudden increase
in population density o f an herbivorous insect, including the western hemlock looper. The
term “hemlock looper” refers to all subspecies o f hemlock looper, Lambdina fiscellaria,
whereas references to the “WHL” refer specifically to the lugubrosa subspecies.

3

1.2 Study objectives

This study uses previous research and new data to characterize how stands are recovering
from the defoliation event that took place between 1991 and 1994 and was motivated, in part,
to determine if the short-term trends observed in defoliated stands (Hoggett 2000, Hoggett
and Negrave 2001) persist after 20 years of succession. This study aimed to sample
documented and mapped WHL outbreaks in the ITR and compile the resulting stand
structure, tree regeneration and understory plant composition patterns that have arisen since
the event. Moreover, the ambition of this study was to characterize these sites and extrapolate
the results to predict the trajectory o f forest dynamics on similar sites in the ITR following a
WHL outbreak.

1.3 Biology and ecology of the hemlock looper

The hemlock looper (Lambdina fiscellaria Guenee) is a native herbivorous insect that goes
through cyclic outbreaks in temperate and boreal forests across North America (Otvos et al.
1979, Jobin and Desaulniers 1981, Harris et al. 1982, Hebert et al. 2001). The subspecies
fiscellaria and lugubrosa are the eastern and western forms of this insect, respectively, but
studies have been unable to morphologically or genetically distinguish the two (McGuffm
1987, Sperling et al. 1999). A third subspecies, somniaria, is found in southwestern B.C. in
the Garry oak (Quercus garryana Dougl. ex Hook.) ecosystem.

4

In eastern Canada, the main host trees o f the hemlock looper, in order of the observed level
of relative feeding “preference,” are balsam fir (Abies balsamea (L.) Mill.) and eastern
hemlock (Tsuga canadensis (L.) Carr.) (Carroll 1956, Davidson and Prentice 1967,
Martineau 1984, MacLean and Ebert 1999). In western Canada, the hemlock looper feeds on
many species including western hemlock (Tsuga heterophylla (Raf.) Sarg.), Douglas-fir
(.Pseudotsuga menziesii (Mirb.) Franco), western redcedar (Thujaplicata Donn ex D. Don),
Sitka spruce (Picea sitchensis (Bong.) Carr.), Engelmann spruce (Picea engelmannii Parry ex
Engelm.), white spruce (Picea glauca (Moench) Voss), true firs (Abies spp.), western white
pine (Pinus monticola Dougl. Ex D. Don), western larch (Larix occidentalis Nutt.),
lodgepole pine (Pinus contorta ex Loudon.), and some broad-leaved species and understory
shrubs (Jardine 1969, Koot 1994, Parfett et al. 1995, Alfaro et al.1999, McCloskey 2007),
although the food species within these latter groups are not named in the literature.

The hemlock looper has one generation per year, with eggs laid on tree branches, lichens,
leaves and in bark crevices in September and October (Hopping 1934, Shore 1989, Koot
1994, Alfaro et al.1999, Hebert et al. 2001). Oviposition may also occur in litter on the forest
floor, but only at the height o f an infestation (Hopping 1934). Hemlock looper eggs
overwinter and hatch in May and June, when early instars (stages of larval development)
feed on current-year needles and buds on upper tree crowns (Erickson 1984, Koot 1994,
Raske et al. 1995, Alfaro et al. 1999).

In July and August, later-instar larvae feed on foliage in all age classes and often do not
consume leaves completely (Carroll 1956, Carolin and Lejeune 1967, Otvos et al. 1971,

5

Furaiss and Carolin 1977, Dobesberger 1989, Raske et al. 1995, Carroll 1999). Feeding
larvae move down to lower sections o f trees during later stages, where they feed voraciously
(Koot 1994, Alfaro et al. 1999). In late August, larvae find pupation sites on lower parts of
the tree on leaves, tree trunks, on the ground around trees, and on other surfaces, and emerge
as adults in 10 to 15 days (Watson 1934, Carroll 1956, Jardine 1969, Ostaff et al. 1974,
Raske et al. 1995). Mating occurs soon after (Jardine 1969, Ostaff et al. 1974) and females
lay eggs, likely near their emergence site (Thomson 1958).

1.4 Hemlock looper as an agent o f disturbance

Little is known about the factors that facilitate WHL population growth and bring a
population to outbreak levels (McCloskey et al. 2009). Climatic factors have been linked to
vacillations in population densities o f Lepidoptera, with above-average temperature and
below-average precipitation being associated with WHL outbreaks (McCloskey 2007). A
favourable climate may allow high survival and fecundity (McCloskey et al. 2009) and may
lead to synchronous outbreaks over wider areas (Williams and Liebhold 1995, Myers 1998).
Spatially independent outbreaks of WHL have developed in B.C. in the same years in the
past, suggesting that regional climate is a factor influencing the start of an outbreak (Parfett
et al. 1995, McCloskey et al. 2009).

During a defoliation event, trees may be stripped of leaves in one season, or they appear
orange-to-brown with the surrounding ground littered with needles that were chewed at their
base and then fell to the forest floor, while other needles were partially chewed, dried out and

6

then fell (Koot 1994, Alfaro et al. 1999). Outbreaks often last three years, and extensive tree
mortality may occur after the first and second years o f a defoliation event (Watson 1934,
Otvos et al. 1979, Jobin and Desaulniers 1981, Alfaro et al. 1999, Bordeleau 2000, Hebert et
al. 2001). A WHL outbreak leaves the soil and forest floor undisturbed (Hoggett 2000).

Populations may become limited by wasp parasitism or by one or more o f the 47 viruses that
affect all life stages on this insect (Torgersen 1971, Tumquist 1991, Koot 1994, Otvos 2004,
Whittome-Waygood et al. 2009). The fungal pathogen Entomophthora sp. has been a
significant natural control agent o f the hemlock looper (Koot 1994), as have the common
parasitic wasps Aoplus velox occidentalis (Harrington), Telenomus dalmani (Ratzeburg) and
Trichogramma minutum (Riley), and the nucleopolyhedrovirus (Torgersen 1971, Otvos
2004, Whittome-Waygood et al. 2009). In situations when foliage becomes a limiting factor,
larval starvation may end an outbreak, and there is speculation that heavy precipitation and
cool weather during the adult flight and egg-hatching periods can also lead to population
collapse (Koot 1994). Some bird species prey on the larval and pupal stages o f the hemlock
looper, but their effect on population densities during outbreak periods is unknown and
unlikely to be limiting (Otvos et al. 1979).

In B.C., mature Tsuga heterophylla dominated forests were considered to be most
susceptible to WHL defoliation (Kinghom 1954, Koot 1994). Forests previously defoliated
by hemlock looper are at high risk for future defoliation (Borecky and Otvos 2001).

7

Western hemlock looper outbreaks are said to be less damaging than eastern hemlock looper
outbreaks (Harris et al., 1982), but to date no study specifically comparing the severity and
consequences of outbreaks o f each subspecies has been carried out.

1.5 Hemlock looper defoliation history, geography and control

Macrofossil and dendroecological data suggest that there was a major decline in eastern
hemlock abundance across its entire range about 5000 years before present (Bhiry and Filion
1996, Foster 2000). Because this decline was only experienced by one tree species over a
wide area at the same time, it has been suspected to be a result of an outbreak of a defoliating
insect like the hemlock looper, as opposed to climate change or aboriginal impact (Davis
1981, Bhiry and Filion 1996, Foster 2000). Lending support to this hypothesis is evidence in
the paleoecological record of large numbers o f looper fossils in sediments from the same
period in which pollen and macrofossils indicate a hemlock decline (Bhiry and Filion 1996).

Hemlock looper outbreaks have been recorded in all the Atlantic provinces (Hudak et
al.1978, Otvos et al., 1979, Clarke and Carew 1987, Magasi 1990, Hartling et al. 1991,
Carter and Hartling, 1992, Iqbal and MacLean 2010, Iqbal et al. 2011), Quebec (Jobin 1980,
MacLean and Ebert 1999, Martel 1999, Chouinard and Filion 2005), Ontario (Carroll 1956,
Torgersen 1971) and B.C. (Kinghom 1954, Alfaro et al. 1999, McCloskey 2007, McCloskey
et al. 2009). Alaska, Oregon and Washington, Idaho and Montana, Michigan and Wisconsin
have also been subject to hemlock looper outbreaks in the past (Graham 1956, Torgersen and

8

Baker 1967, Johnson et al. 1970, Fumiss and Carolin 1977, Western North American
Defoliator Working Group 2002).

Efforts to control hemlock looper outbreaks with such agents as Bacillus thuringiensis var.
kurstaki (commonly known as B.t.), dichlorodiphenyltrichloroethane (DDT), lead and
calcium arsenates, carbaryl and phosphamidon have been documented (Buffam 1965,
Carolin and Thompson 1967, Tumquist 1991, Koot 1994, Fuxa et al. 1998, Charles 2000),
and the introduction of naturally occurring pathogens has also been studied (Wood and Van
Sickle 1993, Levin et al. 1997, Whittome-Waygood 2009). Early tests o f the general
effectiveness o f B.t. found that its effectiveness is in proportion to the concentration applied
to infested trees (Buffam 1965), but limited success was found in achieving high levels of
mortality of hemlock looper when it was applied to control an infestation (Fuxa et al. 1998).
Aerial sprayings o f DDT and lead arsenate have been found to curtail the adverse effects of
feeding forest insects, including hemlock looper (Paananen et al. 1987), but have other
unacceptable environmental effects.

1.5.1 Western hemlock looper in British Columbia

In B.C., WHL outbreaks have occurred primarily south o f 56 degrees latitude, in valleybottom stands with a high proportion o f mature open-growing Tsuga heterophylla (Parfett et
al. 1995). The biogeoclimatic units most frequently attacked by the WHL in B.C. are wet
cool (wk), very wet cool (vk) and moist warm (mw) subzones of the Interior Cedar-Hemlock
zone (ICHvk, ICHwk, ICHmw), and the wk and very wet cold (vc) subzones o f the

9

Engelmann Spruce - Subalpine Fir (ESSF) zone, known as the ESSFwk and ESSFvc
(Ketcheson et al. 1991, Coupe et al. 1991, Parfett et al. 1995), all o f which are in the interior
o f the province. Most infestations have only lasted for one to two years in any one area;
however, two small coastal areas (east of Indian Arm and south o f Lake Cowichan) have
defoliation records showing four-year long outbreaks (Parfett et al. 1995). The first recorded
WHL outbreak in B.C. was in 1911 in Stanley Park, Vancouver (Harris et al. 1982, Parfett et
al. 1995).

WHL infestations with the highest proportion o f trees killed have occurred in coastal and
interior wet belt regions, principally in mature hemlock and hemlock-cedar stands (Alfaro et
al. 1999). In coastal B.C., major outbreaks have been documented on Vancouver Island and
the South Coast (1945-1946, 2000-2003). In interior B.C., major outbreaks have taken place
in the Upper Fraser River valley (1954-1955, 1991-1994), the north Thompson (1976 and
1991-1992), Arrow Lakes (1972-1973), Shuswap Lake (1983-1984), Horsefly Lake-Quesnel
Lake (1946, 1984, 1991-1992), and Revelstoke (1945-1947, 1972-1973, 1982-1983, 19911993) areas (Thomson 1958, Tumquist 1991, Duncan 2006, McCloskey et al. 2009). The
fifteen distinct outbreaks in 100 years have increased in size, distribution and intensity over
that period, particularly those in the Interior (Borecky and Otvos 2001, McCloskey 2007).
Some research suggests that outbreak patterns differ between interior and coastal forests o f
B.C. due to differences in regional climate and tree species composition (McCloskey 2007).

In the ITR o f the Upper Fraser River valley, a WHL outbreak in 1954 and 1955 affected
45,500 ha of forest at Eaglet Mountain, Lunate Creek, Penny and McBride (Harris et al.

10

1982, Parfett et al. 1995). Another defoliation event over 39,000 ha took place forty years
later from 1991 to 1994 (Taylor 1996).

1.6 Tree mortality, survival and stand development

1.6.1 Overstory

Tree death is often the consequence o f multiple factors (Franklin et al. 1987). Defoliation
affects the growth and vigour o f individual trees (McCloskey 2007), but secondary attack by
bark beetles, root rot such as Armillaria mellea (Vahl. ex. Fr.) and dwarf mistletoe
(Arceuthobium spp.) has been thought to exacerbate tree stress further and eventually kill
defoliated trees (Kinghom 1954, Kulman 1971). A WHL defoliation will kill some trees and
not others, increasing heterogeneity at the stand scale, with variation in tree mortality
producing distinct patches of defoliated trees at the landscape scale (McCloskey 2007).
Sometimes defoliation can help tree survival; grand fir (Abies grandis Douglas ex. D. Don
Lindl.) seedlings defoliated by the western spruce budworm (Choristoneura occidentalis
Freeman) showed higher survival during a period o f drought stress than non-defoliated
seedlings (Parks 1993), possibly due to the reduced demands of water and nutrient uptake of
defoliated plants (Webb 1978).

In the context o f forest disturbance research, disturbance severity typically refers to the
proportion of trees killed during a more or less discrete disturbance event (Frelich and Reich
1999). In general, WHL-induced tree mortality is related to defoliation level, tree species and

11

tree size (Alfaro et al. 1999, Hoggett and Negrave 2001), and usually happens after a tree has
been defoliated for two consecutive seasons (McCloskey 2007). It has been found that low
levels o f defoliation only slowed the growth of Tsuga heterophylla in the canopy, and severe
defoliation resulted in the death o f hemlock in the canopy (Alfaro et al. 1999). Generally, all
species of trees that are 100% defoliated by WHL are killed outright, while trees 50% or
more defoliated have a high degree of mortality due to subsequent stresses to the weakened
trees (Kinghom 1954, Tumquist 1991). Mortality can continue to occur up to three years
following the collapse o f an infestation (Kinghom 1954, Tumquist 1991).

In a study in the submontane variant o f the very wet maritime subzone o f the Coastal
Western Hemlock biogeoclimatic zone (CWHvml; Pojar et al. 1991), the most severely
defoliated stands had been dominated by Tsuga heterophylla, and the most lightly defoliated
sites had the lowest proportion o f living T. heterophylla pre-defoliation (McCloskey 2007).
Four years after the outbreak had ended, T. heterophylla had the highest cumulative
mortality, with amabilis fir (Abies amabilis (Dougl. ex Loud.) Dougl. ex J. Forbes) and Thuja
plicata suffering the least mortality, thus increasing their proportion in the surviving canopy
(McCloskey 2007). A study examining mortality ten years post-disturbance in stands with
similar overstory composition found comparable results (Kinghom 1954). The consequence
of this differential mortality on overstory composition was reduced absolute and relative
abundance of Tsuga heterophylla and increased relative abundance of other species
(McCloskey 2007).

12

A study that contained elements analogous to the coastal study began approximately six
years following the end of the 1991-1994 WHL outbreak in the ITR. Within the Goat variant
of the wet cool subzone of the Interior Cedar Hemlock biogeoclimatic zone (ICHwk3), Abies
lasiocarpa and Tsuga heterophylla had the highest mortality following the outbreak, while
Thujaplicata and Picea engelmanii x glauca suffered the least mortality (Alfaro et al. 1999,
Hoggett and Negrave 2001). However, stands with a high basal area composition of Tsuga
and Abies did not always experience commensurately high mortality; high variability in
mortality between stands masked these species-related mortality differences within stands
(Hoggett and Negrave 2002).

In both coastal and interior rainforests, the effects o f WHL are more severe for sub-canopy
trees than for canopy trees, with smaller-diameter trees showing significantly higher levels o f
mortality than larger trees (McCloskey 2007, Hoggett and Negrave 2001). Although Hoggett
and Negrave (2001) reported high variability in mortality among diameter classes in
defoliated stands in the Interior, in general the largest diameter classes suffered the lowest
rates o f mortality. Trees with a diameter at breast height (DBH) of greater than 50 cm had
significantly less mortality than those in smaller size classes (Hoggett and Negrave 2001).
Forest floor depth was one measured site factor that was found to be significantly associated
with the severity o f the WHL defoliation; there was a strong negative relationship between
forest floor depth and Thuja plicata mortality (Hoggett and Negrave 2001).

In the Interior, a strong positive correlation between total live basal area, proportion o f trees
to survive and mean live cover with the basal area o f live Thuja plicata suggested that

13

overstory tree survival in stands was closely linked with the presence and survival o f Thuja
(Hoggett 2000). Furthermore, the mean DBH and height o f live canopy trees was positively
correlated with the basal area o f Thuja, meaning that stands with greater proportions o f Thuja
tended to have surviving canopies that consisted o f taller and larger trees (Hoggett 2000).
Changes in plant community composition following a defoliation event happen partially as a
result o f differences in survival rates among different tree species (Kinghom 1954, Alfaro et
al. 1999, McCloskey 2007).

1.6.2 Stand development and succession

Defoliation events are an important instrument of succession (Bhiry and Filion 1996). A
study that examined two mid-Holocene hemlock looper outbreaks in eastern North America
using paleoecological methods showed a long-term decline in eastern hemlock in areas
where an outbreak was severe (Bhiry and Filion 1996, Fuller 1998). This defoliation was
described as a “major biotic and biostratigraphic event” (Oswald and Foster 2011) that
initiated a successional process that led to stands with entirely different species compositions
than pre-disturbance ones (Fuller 1998, Foster 2000).

Succession is a directional change in the composition of a plant community over time
(Pickett et al. 1987, Barbour et al. 1999). Mechanisms of change in plant species dominance
following a disturbance can reflect interactions such as competition or herbivory, a limitation
of a particular resource, the availability o f propagules, life history differences, changes in
environmental factors or canopy architecture (Pickett et al. 1987, Veblen 1992). Since

14

neither the soil nor the propagules within it are disrupted by a WHL defoliation event, the
series of plant community changes following an outbreak o f this kind can be described as
secondary succession, which occurs following a perturbation that is not severe enough to kill
all plants or the existing seed bank (Connell and Slayter 1977, Barbour et al. 1999).
Some ecological models describing succession in forested ecosystems have included the
influence of herbivores (Dyer and Shugart 1992, Fuhlendorf and Smeins 1997, Tester et al.
1997, Blatt et al. 2001, Seidl et al. 2011). Insect herbivores influence competitive interactions
within a plant community and therefore affect plant species composition (Dyer and Shugart
1992, Blatt et al. 2001, Hartley and Jones 2004). Changes in the proportion and vigour of
overstory trees following a WHL defoliation have been found to lead to increased dominance
of understory shrub species in some stands, while in other stands understory densities of
climax tree species are greater. Different stands follow different successional pathways
following a WHL disturbance (McCloskey 2007), and one o f the primary objectives o f this
study is to determine why this is the case.

The WHL defoliation event in coastal B.C. increased structural and compositional diversity
within stands by increasing the density o f snags, thinning the forest canopy, shifting the
relative composition of canopy trees and modifying biomass and competition in the
understory plant community (McCloskey 2007). These effects likely depend on the severity
o f the defoliation event, including the role o f biological legacies such as standing dead trees
(Turner et al. 1998, Franklin et al. 2000). Canopy openness increases significantly with
increasing severity of defoliation, and a severe defoliation event may slow the redevelopment
of a closed-canopy forest because young trees are killed and a dense understory plant

15

community is created, often with little or no tree regeneration (McCloskey 2007). This
pattern was also found in the ITR, where increased canopy mortality was associated with
increasing cover of understory shrub species, which in turn was negatively correlated with
tree regeneration density (Hoggett and Negrave 2001).

1.6.3 Understory

In a four-year post-outbreak study of a 2000 to 2003 WHL defoliation in coastal (CW Hvml;
Pojar et al. 1991) forests, it was found that more severely defoliated stands (>65% visible
defoliation) were significantly richer in shrub, fern and herbaceous species, and had higher
shrub cover than lightly or moderately defoliated stands (McCloskey 2007). Conversely,
there was lower tree seedling and sapling cover with increasing defoliation severity
(McCloskey 2007). Red elderberry (Sambucus racemosa L.) and salmonberry (Rubus
spectabilis Pursh) are known to dominate sites following WHL disturbances in coastal
forests, with salmonberry being particularly responsive and persistent (Kinghom 1954). The
dense canopies of these shrub species limit the regeneration of trees, and one study predicted
that it would take decades for coniferous trees to become re-established unless management
actions were taken to control shrubs in these stands (Kinghom 1954). It is not known
whether the dominating shrub and fern species originated from growth release o f preestablished plants, from propagule reservoirs in the soil, or an input of propagules from
elsewhere.

16

Five years after the 1991-1994 ITR WHL outbreak, sampled stands in the ICHwk3 generally
had two distinct, co-existing understory layers: a lower layer made up of climax community
species, and an upper layer made up of species more common on disturbed sites, such as
Sambucus racemosa, fireweed (Epilobium angustifolium L.) and red raspberry (Rubus idaeus
L.) (Haeussler et al. 1990, Hoggett 2000, Wilhelm 2004). Devil’s club (Oplopanax horridus
(Sm.) Miq.), a species usually associated with later serai stages (Mason 1990) also occurred
in the upper understory layer. In areas o f higher canopy mortality, the development o f this
upper understory layer was pronounced (Hoggett 2000, Wilhelm 2004). In a study o f sites
with 100% canopy mortality, Rubus idaeus had a higher cover and frequency than other
species (Wilhelm 2004).

The development o f the tall shrub understory layer was associated with a decrease in
regeneration density (Hoggett 2000). Regeneration density was highly variable among plots
(between 1,000 stems/ha to 12,000 stems/ha) but was found to be associated with higher
levels o f living canopy cover (Hoggett 2000). In stands where there was tree regeneration,
Tsuga heterophylla seedlings and saplings occurred at the highest densities, with somewhat
less Thuja plicata and minor amount of Abies lasiocarpa, while paper birch (Betula
papyrifera Marsh.) and spruce were absent (Hoggett 2000).

In light of these results, Hoggett and Negrave (2001) identified two possible trajectories for
stands following a WHL disturbance in the ITR: stands with moderate mortality and a sparse
shrub understory should recover to a full tree canopy in the medium term (not defined), while
stands with high levels of mortality (>60%) and strong shrub understory development (>90%

17

cover) would enter a period o f shrub dominance. Hoggett and Negrave (2001) and Wilhelm
(2004) —both working in the ICHwk3 —and McCloskey (2007) —working in the CWHvm - noted that a period o f understory dominance by shrubs and ferns appears to be part o f the
natural dynamic of heavily defoliated stands affected by WHL in B.C. A white spruce underplanting trial set up immediately after the 1991-1994 WHL outbreak in the ICHvk2 in a
severely defoliated stand found that seedlings had low growth rates and height-to-diameter
ratios in the first three years following the plantings (Coopersmith et al. 1998). These results
were likely due to low light levels due to competition from understory plants and shade from
Tsuga heterophylla and Thujaplicata snags (Coopersmith et al. 1998).

Severe defoliation in CWHvm and ICHwk3 stands altered canopy structure and understory
composition in ways that are expected to delay the re-development of a closed canopy
(Hoggett and Negrave 2002, McCloskey 2007). As observed in moderately and severely
defoliated stands, the death of subcanopy trees means a loss of opportunity for trees to recruit
from subcanopy to overstory strata and replace the canopy trees killed by defoliation
(McCloskey 2007). Thus, there will most likely be a prolonged period o f open canopy and a
lag in canopy closure as new seedlings establish in the understory and recruit to the canopy
(Hoggett and Negrave 2002, McCloskey 2007). This effect was exacerbated in severely
defoliated stands where the understory plant community was already well developed prior to
defoliation (McCloskey 2007).

18

1.6.3.1 Coarse woody debris in a defoliated stand

Coarse woody debris (CWD) in the ITR provides habitat for wildlife (Stevenson et al. 2006).
Some o f the existing CWD in this ecosystem is likely the result of defoliated trees that have
died and fallen to the forest floor, creating a potential substrate for regenerating trees. Coarse
woody debris may help regenerating trees avoid competition with species on the forest floor
(Harmon and Franklin 1989, Daniels and Gray 2006) and could also function to hide or block
access to regenerating trees from browsing ungulates (Ripple and Larsen 2001). Both of
these possibilities facilitate tree regeneration, although the rate o f falling and decay in looperdefoliated trees may limit the availability of this substrate.

Defoliated trees have reduced decay resistance; trees that become stressed due to prolonged
defoliation lose both their capacity to block rot fungi from entering, and their ability to
prevent their spread to other parts o f the tree by compartmentalizing (Engelhardt 1957, Neely
1970, Shortle et al. 1977, Wargo 1977, Shortle 1979, Shortle and Ostrovsky 1983). In a wet
coastal forest, standing looper-defoliated trees deteriorated much faster than dead trees on the
ground not killed by WHL over a similar period (Johnson et al. 1970). Additionally, 20% o f
the wood volume of trees killed by WHL can be expected to show advanced decay after only
two years, whereas dead trees on the ground for the same period will suffer only slight losses
after the same amount of time (Johnson et al. 1970). Decay rates o f post-WHL trees are
useful to an understanding of recovery and regeneration in defoliated stands and are an
important consideration in determining their carbon balance.

19

1.7 Nutrients and defoliation

The delivery of frass and partially-eaten leaves to the forest floor is one o f the ways forest
canopies and soils are linked by the process of herbivory (Chew 1974, Mattson and Addy
1975, Swank et al. 1981, Seastedt and Crossley 1984, Grace 1986, Hollinger 1986,
Schowalter et al. 1991, Risley and Crossley 1993, Lovett and Ruesink 1995, Reynolds et al.
2000, Reynolds and Hunter 2001, Hunter 2001, Reynolds et al. 2003). A defoliation event
can modify the functional attributes of a forest stand by changing conditions o f soil nutrients,
light and moisture in a way that may shape the direction o f succession and recovery
(Assmann 1970, Franklin et al. 1987, Hicke et al. 2012). Defoliating insects affect the total
amount of nutrients available to plants and also cause temporal and spatial variation in that
availability (Hartley and Jones 2004). An outbreak also increases canopy openness, thereby
increasing forest floor temperatures and decomposition rates (Schlesinger 1991, Reynolds
and Hunter 2001).

Defoliators such as the hemlock looper will consume the carbohydrates and proteins in
leaves and release large fluxes o f nitrogen, carbon and other nutrients into forest ecosystems
through frass (feces), partially eaten “greenfall,” dead insects, and exuviae (Mattson and
Addy 1975, Swank et al. 1981, Grace 1986, Hollinger 1986, Schowalter 2000, Frost and
Hunter 2004, Hicke et al. 2012). Also, defoliation can increase the nutrient content of
precipitation as it falls through the forest canopy (“throughfaH”) (Schowalter et al. 1991,
Hunter 2001). Increased canopy openness results in higher forest floor temperatures, which

20

in combination with more moisture availability from throughfall, facilitates decomposition
and nutrient cycling (Gabriel and Kellman 2011).

There appears to be no consistent trend in soil nutrient availability to plants in insect
defoliated stands (Stadler et al. 2004). Some authors have observed increased nitrogen
mobilization brought about by enhanced frass and litter inputs (Swank et al. 1981, Hollinger
1986, Christensen et al. 2002), and others report nitrogen immobilization in reponse to
greater levels o f carbon compounds (Lovett and Ruesink 1995, Michalzik and Stadler 2000,
Lovett et al. 2002). This might be due to site-specific factors such as the microbial soil
community, nutrient status (e.g., N saturation,) litter quality (Stadler et al. 2004), or other
abiotic variables.

At non-outbreak densities, defoliating insects don’t have an especially important effect on
the decomposer system, with significant effects occurring mainly during short, intense
outbreak periods (Wardle and Bardgett 2004). At non-outbreak densities, frass has not been
found to affect soil respiration or litter decomposition rates, even though frass inputs
stimulate micro-arthropod densities (Reynolds and Hunter 2001, Reynolds et al. 2003).
Thoughfall and greenfall associated with a defoliation event have been found to reduce soil
respiration, likely because soil microbes were outcompeted by mycorrhizal fungi for uptake
of the added mineral nitrogen (Reynolds and Hunter 2001).

Below ground, plants can increase root exudation in response to herbivory, and this may
have effects on soil fauna (Wardle 2002), which in turn may affect nutrient availability to

21

plants (Bardgett and Chan 1999). The physiological responses o f plants to above-ground
herbivory may be important in influencing both soil organisms and key soil processes such as
decomposition and nutrient mineralization in terrestrial ecosystems (Bardgett et al. 1998),
although there is limited research on this effect in forest ecosystems.

Importantly, differences among studies o f what happens to nitrogen or other nutrients
following a defoliation event can be the result of differences in soil parent material, local
topography, edaphic conditions, soil fauna, variation in forest age, quality o f frass, and other
factors (Hunter 2001, Christenson et al. 2002). In addition to nutrient inputs and nutrient
availability, light availability, soil moisture and temperature changes may influence the
quality of litter by affecting leaf chemistry and its decomposition dynamics (Strand et al.
1999). These effects can create stand-specific defoliation outcomes.

1.8 Research questions

With this study spanning a wide range of site types approximately two decades after the
WHL outbreak, I can meet the objectives identified in Section 1.2 by answering the
following four research questions about the resilience of defoliated stands in the ICHwk3 and
ICHvk2:

1.

Is greater canopy recovery after defoliation significantly related to site productivity or

climate factors?
2.

Are different levels o f shrub cover more closely associated with differences in site

productivity factors (e.g., soil moisture availability), climate or canopy openness?

22

3.

Are tree regeneration densities more closely associated with canopy openness, shrub

cover, site productivity or climatic factors?
4.

To what degree does understory plant community composition reflect site productivity,

climatic factors, WHL outbreak history or current canopy conditions?

23

2.0 Methods

2.1 Study area

Plots were located within the vk2 and wk3 variants o f the Interior Cedar-Hemlock
biogeoclimatic zone (ICH; Ketcheson et al. 1991). The ICHvk2 occurs between elevations o f
680 and 1,180 m in the valley of the Fraser River between Dome Creek and Sinclair Mills,
and includes areas around the Torpy and MacGregor Rivers, Slim Creek and Purden Lake
(DeLong 2003). Mean annual precipitation in this variant is about 840 mm and mean annual
temperature is approximately 3.3°C. The ICHwk3 occurs at elevations between 670 and
1,225 m between Dome Creek and McBride, has a mean annual temperature of
approximately 3.1°C (Meidinger et al. 1988), and is slightly drier than the ICHvk2 (DeLong
2007).

Soils in the ITR are mostly of the Podzolic order, with Brunisols and Regosols occurring on
unstable slopes and in areas o f recent fluvial deposition, and Luvisols forming where there
are morainal deposits (Stevenson et al. 2011). Forest floors in the ITR can be two to four
times thicker than in drier forest types to the west in central B.C. (Arocena and Sanborn
1999). Mormoders, Humimors and Hemimors tend to be the dominant humus forms in the
vk2 and wk3 variants (Green et al. 1993).

Sixty-three plots were placed in stands that had been defoliated by the WHL between 1992
and 1994. Only stands that had been classified and mapped as ‘moderately’ or ‘severely’

24

defoliated by Forest Health Aerial Overview Surveys (CFS and MoF 2000) and had
remained unharvested were sampled. Stands in the ‘moderate’ defoliation class are defined
as having trees with pronounced discoloration, thin foliage, the top third o f trees defoliated,
and some trees completely stripped (CFS and MoF 2000; Figure 1). Stands in the ‘severe’
defoliation class are defined as having trees with bare branch tips, completely defoliated tops
and where most trees sustain more than 50% total defoliation (CFS and MoF 2000). Maps
provided by the B.C. Forest Service were used to locate plots in stands that met these criteria.

McCloskey (2007) found that lightly defoliated stands (<25% defoliated) do not differ
substantively from the surrounding non-disturbed landscape, and so only stands that
experienced higher defoliation levels were sampled. Only stands mapped as having been
defoliated in the 1990s, but not in the 1950s, were sampled in order to isolate the effects of
the more recent defoliation event and to avoid obscuring the circumstances under which
regeneration and succession are occurring.

Plots were located in stands that had been sampled previously by other researchers, as well as
in some stands with no history o f having been sampled in the past. Plots were placed in
stands sampled by Deschamps et al. (Walker Creek, unpublished 1994), Alfaro et al. (1999),
Hoggett and Negrave (2002), Conder (unpublished 2006), and Konchalski (unpublished) that
met the sampling criteria for this study. Maps, latitude/longitude coordinates and UTM
coordinates provided by these researchers were used to find the plots they had sampled, and
new plots were placed in the same stands when possible. When an attempt to find a

25

previously sampled plot location in the field was unsuccessful, a new plot was placed within
the same forest cover polygon, based on the B.C. Forest Service maps o f the study area.

Additional plots accessible by road were located to capture the ecological variation o f stands
within the study area. After walking into a stand (away from the nearest road) 170-250 m, a
random number between zero and 360 generated in advance using MS Excel was used as a
compass bearing then followed for 70 paces. All plots were at least 100 m from a road except
for three plots located at Walker Creek, which were approximately 30 m away from the
Highway 16 right-of-way. Plots were at least 30 m away from any cut tree sighted.

Figure 1. Examples o f “light”, “moderate” and “severe” categories o f defoliation, from the
Forest Health Aerial Overview Survey Standards for British Columbia (CFS and MoF 2000).
Copyright (c) Province of British Columbia. All rights reserved. Reproduced with permission
of the Province of British Columbia, www.ipp.gov.bc.ca

26

2.2 Plot layout and sampling design

The centre of each 400 m2circular, fixed-area plot was marked with a flagged wire ‘pigtail’
pin. Three 100 m2 circular subplots, each with a nested 50 m2 circular subplot, were located
along the edge of the main plot at 20° (“North”), 140° (“Southeast”) and 260°(“Southwest”)
azimuths from plot centre; see Figure 2. Within each 400 m2 plot, the following site data
were collected: slope, aspect, elevation and UTM coordinates. Stand data collected within
each plot were: percent canopy cover (using a concave spherical densiometer in each subplot
for a total of three readings), the species and DBH o f all live trees >7.5 cm DBH, and a
visual estimate of their current percent defoliation were recorded. This included trees with
tops that had snapped off. Percent defoliation values were not assigned to Betula papyri/era.

Figure 2. Representation of plot and subplot layout.

27

Dead trees were also tallied by species, measured for DBH, and recorded as being
“standing,” “leaning,” “snapped” or “fallen.” Coarse woody debris (CWD) classes 4 and 5
and snag classes 6 and up (Maser et al. 1979) were excluded from this count, so that the dead
trees counted are assumed to have died directly or indirectly from the 1990s WHL
defoliation. This method is consistent with studies determining the relationship between
decay and time o f tree death (Daniels et al. 1997). We did not, however, look for or record
evidence o f butt rots, root rots, bark beetles or other causes of tree damage or death, none o f
which are recorded as being widespread forest health issues in the Robson Valley during the
1990s. We took four digital photographs, one in each cardinal direction, from the centre of
every 400 m2 plot.

A 50 cm deep soil pit was dug in every plot and the following data were collected: soil
texture class, per cent coarse fragments, effective rooting depth, humus form, mesoslope
position, whether there was gleying or mottling, and a determination of soil nutrient regime
and soil moisture regime (DeLong 2003, 2007). Soil nutrient regime (SNR), soil moisture
regime (SMR) and site series (Pojar et al. 1987) were validated using a combination of soils
data and plant cover data (DeLong 2003, 2007).

Although no charcoal particles were found in soil pits to indicate fire activity, these particles
are difficult to observe without a trained eye (P. Sanborn pers. comm.). There is datable
evidence of fires as recently as 182 years ago in both forest floors and buried soil profiles in
climax stands sampled in the ICHwk3 (Sanborn et al. 2006) so it is likely that some o f the 63

28

stands sampled for this study have fire in their history. This study considers the effects o f the
WHL defoliation in isolation from their possible fire history.

All saplings (>1.3 metres in height to 7.5 cm DBH) were tallied by species in each 100 m 2
subplot. In each 50 m2 subplot, all seedlings (10 cm to <1.3 metres in height) and germinants
(<10 cm in height) were searched for and tallied by species as well. Germinants were likely
under-counted because o f their small size. Two tallies were done for each subplot: one for
total regeneration and one for only well-spaced regeneration. Well-spaced germinants,
seedlings and saplings were the tallest and most vigorous o f each size class and at least 0.5 m
from other individuals. For total regeneration tallies, the substrate on which each germinant,
seedling, and sapling was growing was recorded. It is not necessarily possible to correctly
identify the establishment substrate o f a sapling because o f the time elapsed since the
establishment of trees in that size class. All regeneration was assigned to one o f three
substrate categories: forest floor, coarse woody debris and mineral soil.

A prism sweep using a glass wedge prism having a basal area factor (BAF) o f 4 m2/ha, o f all
overstory trees was taken from the centre o f each subplot to determine local basal area for
overstory trees by species. Canopy density was also measured from the centre o f each
subplot using a spherical densiometer. Subplot data were aggregated so that there was one
value per plot for basal area and canopy density.

Vegetation cover for all vascular plant and bryophyte species that could be seen on the forest
floor was visually estimated in 50 m2 subplots. Some lichen species in some stands were

29

recorded in this study, but lichens were not sampled reliably, and therefore not included in
any analyses. It was observed that terricolous lichens were uncommon and most lichens were
epiphytic. For more information on this functional group in the ITR see Goward and
Arsenault (2000), Benson and Coxson (2002), Radies (2008), Radies et al. (2009), Spribille
et al. (2009) and MacDonald (2013).

2.3 Data preparation

Canopy density, regeneration densities, vegetation cover and basal area data were aggregated
by calculating the arithmetic mean for each variable at the plot level.

In order to analyze plots in both subzones together, I created a synthetic index for relative
soil moisture using SMR categories and their correspondence to absolute soil moisture
regime (ASMR) values (DeLong et al. 201 la). Absolute Soil Moisture Regime is an estimate
of water availability on a site, considering climatic as well as terrain influences (DeLong et
al. 201 la) and these values were used to compare moisture levels in both subzones. Plots in
this study fell into SMR categories 3 to 6 (‘submesic’ to ‘hygric’) in the wk3 biogeoclimatic
variant and categories 3 to 7 (‘submesic’ to ‘subhydric’) in the vk2. The synthetic index,
standardized to ICHwk3 relative moisture, captures the range of soil moisture possibilities
within the plots in this study; the driest category (3) includes submesic plots in the wk3,
while the wettest category (8) includes subhydric plots in the vk2. The intermediate
categories (4-7) include plots in both subzones because moisture levels in this range likely

30

overlap. In general, relative moisture regimes in the ICHvk2 were offset by one relative SMR
unit to correspond to approximately the same ASMR score.

A frost hazard rating was assigned to each plot using the Stand Level Frost Hazard
Assessment and Management Tool (DeLong et al. 201 lb). Using latitude, slope and aspect,
potential direct incident radiation (referred to hereafter as simply “radiation”) was estimated
for each plot (McCune and Keon 2002). Estimates o f mean annual precipitation (MAP),
mean annual temperature (MAT) and annual heatrmoisture index (AHM) were estimated for
each plot using ClimateWNA ver. 4.71 (Wang et al. 2012). ClimateWNA calculates sitespecific climate data based on 30-year climate norms using geo-referenced interpolations,
correlations, and elevation adjustment (Wang et al. 2012). Those variables were selected for
inclusion because their coefficients o f variation were the highest among the 21 climate
variables produced by ClimateWNA for the 63 plot locations, and hence should have the
greatest potential for distinguishing plot differences.

Site index, a measure of site productivity, was included in this analysis in order to compare
the productivity potential among stands (MoFR 2011). Site index values were assigned to
each sampled stand on the basis of biogeoclimatic variant and site series, using values
published by MoFR (2011). Site index, as used here, is the mean reported height (m) for
dominant trees across five species (Picea engelmanii x glauca, Pinus contorta, Abies
lasiocarpa, Tsuga heterophylla and Thuja plicata) in stands at 50 years of age.

31

Data collected in Forest Health Aerial Overview Surveys in the years 1992, 1993 and 1994
were used to create an independent index o f defoliation severity (Defoliation Severity Index,
DSI). Based on estimates of living canopy cover, Forest Health Aerial Overviews Surveys
assigned an annual damage classification rating, such as “moderate” or “severe” to defoliated
parts o f the landscape during the 1991-1994 outbreak (CFS and MoF 2000). These damage
ratings and their duration were integrated to create an ordinal index with four classes o f
cumulative defoliation severity. MacLean and Ebert (1999), in studying eastern hemlock
looper outbreaks, suggest that two years o f moderate defoliation have a similar effect as one
year o f severe defoliation, so I make comparable assumptions o f additive cumulative impacts
here. Consequently, one year only of moderate defoliation was assigned a DSI score o f 1,
stands with two years of moderate defoliation or one year only o f severe defoliation were
assigned a DSI score o f 2, stands with one year o f moderate defoliation and one year of
severe defoliation were assigned a score of 3, while DSI = 4 for stands with two years o f
severe defoliation. Nine sample plots which clearly had experienced a history o f WHL did
not have multi-year damage classification ratings from the Forest Health Overview Surveys,
so these plots were assigned an index score that reflected the mapped severity according to
the Pest Infestation polygons provided by B.C.’s Land and Resource Data Warehouse.
Where both sources of information were available, those Pest Infestation polygons generally
corresponded to the 1994 defoliation survey results, but could only be scored as DSI=1 or
DSI=2, so may underestimate the disturbance severity experienced by those nine plots.

A second indicator of defoliation severity —the proportion of dead stems per hectare —was
calculated using the number o f dead stems (including fallen, leaning and snapped trees in

32

snag classes 3 to 5; Maser et al. 1979) and the number of live stems in sampled stands. Using
the proportion o f dead stems as an index o f severity assumes that this is a measure o f net
canopy impact or “final” severity. This plays down the situation where some stems may have
been defoliated for one or more years, but have since recovered fully or partially.

The damage classes reported in Forest Health Aerial Overviews Surveys were also used to
infer their converse, namely the proportion o f unaffected trees as documented during the
1992-1994 outbreak. With a damage class o f “moderate” denoting 11-29% defoliation, its
midpoint is 20% defoliation and its converse is implied to portray a forest stand that was
80% intact at that time; “severe” damage portrays 30-100% defoliation, with a midpoint of
65%, and a converse o f 35% intact trees.

An index was developed to denote the relative recovery or deterioration o f the forest canopy
from the peak o f the outbreak, as estimated at the time. An average Overstory Recovery
Index (ORIave) divides percent live stems in 2012 by the mean of intact stem percentages
inferred for 1992, 1993 and 1994, assuming 100% vigour o f stands not mapped as
moderately or severely defoliated in a particular year. High values o f these ORI scores more green trees now compared to the past —indicate a higher level of tree recovery, while
low ORI values denote poorer tree recovery at the plot or stand level.

“Competitive shrub cover” refers to the percent cover o f shrubs considered competitive with
tree regeneration and associated with soils that are higher in nitrates (as per Haeussler et al.
1990). These shrubs include Rubus idaeus, thimbleberry (Rubus parviflorus Nutt.),

33

Sambucus racemosa and Oplopanax horridus. “Tall shrub cover” refers to cover o f all non­
dwarf shrub species found in sampled stands.

2.4 Statistical analyses

To explore how overstory recovery is related to defoliation severity, ORIavc scores were
tested against cumulative damage class ratings (i.e., “moderate” or DSI=1, “moderatemoderate” or “severe” for which DSI=2; “moderate-severe” or DSI=3, and “severe-severe”
or DSI=4) in a Spearman Rank correlation. This non-parametric test does not assume that
variables are drawn from a normal distribution. All correlations were performed using Stata
version 12.1 (Stata Corporation 2011). For significant correlations, linear least-squares
regression was used to characterize the relationship. Significant results are shown in figures.

An exploratory Spearman Rank correlation analysis was done to test associations between
the ORIave and 13 site variables (excluding canopy density) to explore how overstory
recovery is associated with site variables.

To explore how basic understory attributes are associated with measured and inferred
defoliation severity, Spearman Rank correlations were performed to test three understory
recovery variables - shrub cover, total regeneration density and well-spaced regeneration
density —against two indices of 1990s WHL outbreak severity, the ordinal DSI and percent
dead stems.

34

A Spearman Rank correlation analysis was done to test associations between understory
recovery variables (shrub cover, total regeneration density and well-spaced regeneration
density) against all 14 site variables. Where visual inspection revealed a unimodal or
parabolic distribution, a quadratic term (x2) was added to the regression model. Where logtransformed values o f the dependent variable or the independent variable resulted in a higher
R2 value, only the regression based on the log-transformed variable is reported.

To examine understory plant community patterns and the relationships between species
composition and site characteristics in defoliated stands in the ITR, redundancy analysis
(RDA; Rao 1964, van den Wollenburg 1977) was performed. Redundancy analysis is a
canonical ordination method, and is the direct extension of multiple regression to the
modelling of multivariate response data (Legendre and Legendre 2012). All RDA analyses
were conducted using CANOCO 5.0 (ter Braak and Smilauer 2012).

Redundancy analysis represents the variation of a data matrix in a reduced number of
dimensions. A plant community data matrix composed of percent cover for all individual
vascular plant and bryophyte species in defoliated stands was tested against an explanatory
data matrix made up of site variables. Prior to RDA analysis, I removed from the plant
community data matrix any species that occurred in fewer than 5% of plots in order to reduce
noise and enhance the detection o f relationships between community composition and site
variables (McCune and Grace 2002).

35

Species gradient lengths are a measure o f the heterogeneity in community composition along
an ordination axis (McCune and Grace 2002, Leps and Smilauer 2003). To determine
whether the plant community dataset was appropriate for a linear ordination method such as
RDA, rather than a unimodal method, I used detrended correspondence analysis (DCA) in
PC-ORD 6.07 (McCune and Mefford 2011) to determine the longest species gradient length.
Plant community data had a gradient <3.0 standard deviation units long and thus was
appropriate for a linear method (ter Braak and Smilauer 2002). A linear technique was also
recommended by CANOCO 5.0 (ter Braak and Smilauer 2012).

For all multivariate plant community analysis, plant cover data were first converted using the
Hellinger transformation:

Y’HOfy/YH)

where Y ’jj= Hellinger transformed abundance value, YtJ =% cover o f a species at a site,
i=sites (rows), j =species (columns), and Y,+= row sums (Legendre and Gallagher 2001). The
Hellinger transformation allows the use of Euclidean-based ordination methods such as RDA
without having to deal with the problems associated with Euclidean distance (Legendre and
Gallagher 2001).

To reduce the asymmetry of distributions of site variables and dampen the influence of
outliers (Legendre and Legendre 2012), some o f these variables were transformed where
appropriate. Percent live stems and canopy density (both proportions) were transformed

36

using arcsine square root, and elevation, steepness, live total basal area, mean annual
temperature, mean annual precipitation and annual heat:moisture index were square root
transformed. The variables slope position, radiation, soil nutrient regime, soil moisture index,
site index and frost hazard were not transformed.

I used a modified forward selection procedure that corrects for the overestimation of the
proportion of explained variance, which often occurs in forward selection (Blanchet et al.
2008). A ‘global’ test o f the relation between the plant community and the whole set o f 14
site variables was run using a constrained ordination using RDA (ter Braak and Smilauer
2012). The overall statistical significance o f the whole model was evaluated using Monte
Carlo permutation tests (n=999). Two stopping criteria used for the forward selection of
variables were the model-adjusted R2 and a variable retention permutation p-value (stopping
significance value) of 0.05 (Blanchet et al. 2008).

Stepwise selected variables had to have variance inflation factors (VIF) < 1 0 (Borcard et al.
2011). Site variables were tested for multi-collinearity in Stata v. 12.1. Only mean annual
temperature (MAT), mean annual precipitation (MAP) and annual heat:moisture index
(AHM) were multicollinear (VIF>10). Mean annual precipitation was pulled out o f the
forward selection process in RDA because it was not significant and had a low percent
contribution (<5%) to the regression model. Axis scores derived from the global RDA for
plant community were graphed in order to examine relationships among all explanatory
variables. A final parsimonious RDA model was run after stepwise selection o f the combined
data set. This process identified the smallest number of significant predictors necessary to

37

model plant community and regeneration trends. Final biplots (two RDA axis scores plotted
orthogonally to each other) were created to display the results o f the most parsimonious
model.

Spearman Rank correlations were performed to explore how the proportions of competitive
shrub cover (Rubus idaeus, Oplopanax horridus, Rubus parviflorus and Sambucus racemosa)
and tall shrub cover (all shrub species except Rubuspedatus Sm.) associated with overall
regeneration density (total and well-spaced), and regeneration densities of each tree species
by size class (total and well-spaced). Where log-transformed values of the dependent
variable or the independent variable resulted in a higher R2 value, only the regression based
on the log-transformed variable is reported.

Redundancy analysis was used also to examine relationships between environmental
variables and the composition o f the regenerating tree assemblage among defoliated stands in
the ITR. This regeneration RDA followed the same methodology as the plant community
RDA (i.e., treating size classes as well as species o f tree regeneration in the manner
analogous to individual plant species), including gradient length determination,
transformation of variables and forward selection procedure.

In order to investigate possible vegetation trends in the plant community RDA analysis as
related to defoliation severity, a Kruskal-Wallis analysis of variance was used to test the first
three RDA axis scores against cumulative damage classification ratings.

38

Spearman Rank correlations were used to evaluate how the proportion of total regeneration
on coarse woody debris relates to both competitive and tall shrub cover, and how the
individual proportions of tree species by size class on coarse woody debris related to both
shrub groups. Availability o f each substrate (coarse woody debris, forest floor, mineral soil)
was not accounted for in this study and so it is not possible to quantitatively determine the
relative importance o f each substrate to regenerating trees in defoliated stands.

Cluster analysis is a tool used to objectively identify groups of sample units that are similar
in composition (Peck 2010). A hierarchical cluster analysis was accomplished using PCORD 6.07 (McCune and Mefford 2011) to group defoliated stands in terms o f their similarity
based on the relative abundance of species in the understory. This analysis was produced
using Sorensen/Bray-Curtis distance coefficients and a flexible beta linkage method (beta
value o f =-0.25), and was applied to cluster samples (stands), not species. Only understoiy
species that occurred in more than 5% o f the sampled stands were included. A dendrogram
portraying multivariate compositional relatedness of stands is used to portray the output of
this analysis.

A second examination o f possible groupings among stands according to their recovery
response was also undertaken. Using the plant community RDA biplot, stands were simply
colour-coded according to their site series and the resulting output was scrutinized for any
clustering among stands. Using the same technique, the regeneration RDA biplot was colourcoded in order to better visualize regeneration patterns by site series. Collectively, these

39

clustering techniques facilitated distinction o f the two main understory development
trajectories, namely dominance by shrubs or by regenerating trees.

40

3.0 Results

3.1 Summary o f site variables

Site variables measured at each plot are listed, described and summarized in Table 1. It is
worth noting that those with the greatest variability (coefficient of variation, CV, > 30%) are
steepness, live basal area, percent live stems, and frost hazard.

Table 1. Study-wide summary of site variables measured at each plot (n=63).
Full variable name and
description

Measurement
method

Units

elev

Elevation. Distance above
m ean sea level

steepness

Steepness. Horizontal slope
gradient

slopepos

Slope position. Relative
m esoslope position (Banner
et al. 1993)

radiation

Radiation. Potential annual
direct incident radiation

snr

Soil nutrient regime. A
relative scale o f available
nutrients for plant growth
Indexed Soil Moisture
Regime. A synthetic index
o f soil m oisture regim e, a
relative scale o f available
w ater for plant growth

GPS unit
(M agellan
M eridian Basic)
Clinom eter
(Suunto
PM 5/360PC)
Visual
estim ation in the
field and G oogle
Earth (2012)
Equation # 1 in
M cCune and
Keon (2002)
using folded
aspect, slope and
latitude,
collected in the
field
From key in
D eLong (2003,
2007)
A ssigned index
based on
Drought
A ssessm ent Tool
(DeLong et al.
2011a)
Estim ates by site
series, from
M FLNRO
(2011)

Variable

ismr

Site index

Site index. A measure o f site
productivity

41

Mean

CV

Min.

Max.

M etres

837.6

(% )
12.6

653

1,145

%

11.5

90.3

0

51

Index

1.8

25.3

1

3

M J c m '2 y r '1

0.57

13.0

0.32

0.78

Index

4.4

23.4

2

6

Index

5.3

16.8

3

8

m height at
50 yrs.
B ased on a
five-species
m ean (Tsuga
heterophylla,

18.6

8.7

12.6

19.8

V ariab le

lb a t o t

% live
stems

canopdens

Full v aria b le nam e an d
d escrip tio n

Live basal area. Live total
basal area o f all overstory
tree species taken from
centre o f regeneration
subplots
Percent live stems.
Proportion o f living stems o f
all live and dead overstory
trees
Canopy density. The
proportion o f area that is
covered by a crown o f trees.

frsthaz

Frost hazard rating

MAT

Mean annual temperature

MAP

Mean annual precipitation

AHM

Annual heat:m oisture index,
(M A T + 10)/(M A P /1000))

M e a su re m e n t
m eth o d

W edge prism,
basal area factor
4

Counted in the
field in fixed
area (400 m 2)
plots
Spherical
densiom eter
(Lem m on Model
C)
A ssigned index
based on Stand
Level Frost
Hazard
A ssessm ent and
M anagem ent
Tool (D eLong et
al. 2011b)
For 1960-1999,
interpolated
using
Clim ate WNA
ver. 4.71 (W ang
et al. 2012)
For 1960-1999,
interpolated
using
Clim ate WNA
ver. 4.71 (W ang
et al. 2012)
For 1960-1999,
interpolated
using
Clim ate WNA
ver. 4.71 (W ang
et al. 2012)

42

U nits

M ean

CV

M in.

M ax.

(% )
Thuja
plicata,
Abies
lasiocarpa,
Picea glauca
x engelm anii
and Pinus
contorta)
n ^ h a '1

34.0

55.0

2

81.3

%

49.3

48.4

5.2

92.1

%

80.3

19.8

40.0

95.1

Index

2.14

40.0

1

4

°C

3.0

11.2

1.7

3.6

mm

825.2

5.9

751

969

Index

15.8

6.1

13.7

17.3

3.2 Summary of overstory

A summary o f overstory tree density is presented in Table 2. Tsuga heterophylla had the
highest mean density among live and dead trees, while Betula papyrifera had the lowest
mean density in sampled stands. Overstory basal area by species is displayed in Table 3 and
shows that among live trees, Thuja plicata had the highest mean basal area and Tsuga
heterophylla had the highest mean basal area among dead trees.

Figure 3 presents the current health class distribution of stems in sampled stands. Most living
trees of all species are in the highest health class (have the highest percentage o f live
canopy), and Tsuga heterophylla has the highest mean number o f stems in the highest decay
class (fallen trees). Most dead trees o f all species are in the lowest decay class (standing).
Figure 4 shows mean DBH o f live and dead stems by species. The mean DBH o f dead Tsuga
heterophylla was greater than that of live individuals, whereas the abundance o f largerstemmed Thuja plicata implies that they had greater WHL survivorship than stems with a
lower DBH. This information is further divided to display mean DBH of live and dead stems
by subzone (ICHvk2 and ICHwk3: Figures 5a and 5b, respectively).

43

Table 2. Overstory density by species. All dead trees are trees that likely died directly or
indirectly from the WHL defoliation in the 1990s. Older dead trees (snag classes 6 and
greater; Maser et al. 1979) were excluded from tree counts in this study. Mean density values
are from tree counts in 0.04 ha plots (n=63).______________________________________
Overstory density (stems ha'1)
1st quartile
3rd quartile
Live
mean median
SD min. max
(25%)
(75%)
Tsuga
159.1
50.0
192.7 0.0 775.0
25.0
225.0
Thuja
121.8
75.0
25.0
137.0 0.0 775.0
200.0
Abies
70.2
0.0
151.3 0.0 800.0
0.0
50.0
Picea
49.2
25.0
0.0
74.7
0.0 450.0
75.0
Betula
18.7
0.0 300.0
0.0
0.0
55.4
0.0
Subtotal
419.0
Dead
Tsuga
150.0
182.9
188.4 0.0 975.0
50.0
250.0
Thuja
89.7
75.0
0.0 375.0
25.0
93.8
125.0
Abla
0.0 450.0
83.3
50.0
93.8
25.0
100.0
Picea
9.5
0.0
75.0
0.0
0.0
15.8
25.0
Betula
0.0
4.0
0.0
15.0
75.0
0.0
0.0
Subtotal
369.4
% of total
46.9

Table 3. Overstory basal area by species. All dead trees are trees that likely died directly or
indirectly from the WHL defoliation in the 1990s. Old trees (snag classes 6 and greater;
Maser et al. 1979) were excluded from tree counts in this study. Basal area values are from
wedge prism (BAF 4) sweeps, three per each 0.04 ha plot, (n^63) averaged for each plot.
Overstory basal area (m2/ha)
..
1st quartile
3rd quartile
uve
mean median SD min. max
(25%)
(75%)
Tsuga
8.7
4.7
10.3 0.0 41.3
1.3
14.3
Thuja
17.4
10.0
2.7
17.3 0.0 72.0
26.3
Abies
4.5
18.7
0.0
3.1
1.3
0.0
4.0
Picea
4.0
5.4
0.0
2.0
0.0 26.0
6.0
Betula
0.4
1.0
0.0
4.0
0.0
0.0
0.0
Subtotal
33.6
Dead
Tsuga
7.6
5.3
8.7
0.0 38.7
1.3
11.3
Thuja
7.4
4.7
8.1
1.0
11.0
0.0 36.0
Abla
2.7
2.0
2.9
11.3
0.3
0.0
4.0
Picea
0.0
1.0
3.3
0.0
0.7
0.5
0.0
Betula
0.0
0.0
0.1
0.7
0.0
0.0
0.0
Subtotal
18.2
% of total
35.1
44

200

-

■ Health class 1
■ Health class 2
Health class 3
Health class 4

X!

es

JS

Tsuga

Thuja

Abies

Picea

E -50
4>

■ Decay class 4
■ Decay class 3
* Decay class 2

I -100

Decay class 1

■150

Figure 3. Mean live stems/ha (above ‘O’ line) and dead stems/ha (below ‘O’ line) tallied
according to their health (if live) and decay (if dead) class. Health classes go from highest to
lowest percentage o f live canopy (1=76-100%, 2=51-75%, 3=26-50%, 4=0.1-25%). Decay
classes go from least decayed to most decayed (l=standing, 2=leaning, 3=snapped, 4=fallen).
Dead trees are trees that likely died directly or indirectly from the WHL defoliation in the
1990s. Coarse woody debris classes 4 nd 5 and snag classes 6 and up (Maser et al. 1979)
were excluded from this count.

60
50

g

40

-

1 30

-

s

10

0
Tsuga

Thuja

A bies

Picea

B etula

Species

Figure 4. Mean diameter at breast height of live and dead stems by species (mean, +/- SE).
45

a)

C? 50

■ ICHvk2
* ICHwk3

Tsuga

Picea

Thuja

B etula

VL

50

■ ICHvk2
■ ICHwk3

Picea

Betula

Dead

Figure 5. Tree species by diameter at breast height (mean, +/- SE) by subzone: a) live; and b)
dead.

46

3.3 Summary o f understory

3.3.1 Plant community

Eighty-seven species of vascular and non-vascular plants were found in plots sampled in this
study. Table 4 presents a list o f the plant species encountered, with mean abundance and
frequency by species. Five trees, 19 shrubs, 41 forbs, five ferns, two grasses, one sedge, 10
mosses and one clubmoss species were identified across the 63 plots sampled. In terms of
frequency, the most common tree species was Tsuga heterophylla, with germinants and/or
seedlings and/or saplings growing in 97% of plots. The next most common tree species was
Thuja plicata, which was present in 94% of plots.

The most common shrubs were five-leaved bramble (Rubus pedatus Sm.), which was found
in 97% of plots, and Oplopanax horridus, which occurred in 95% of plots. Rosy twistedstalk
(Streptopus lanceolatus (Aiton) Reveal var. curvipes (Vail) Reveal) was found in all plots,
and other common herbs were foamflower (Tiarella spp.) and bunchberry (Cornus
canadensis L.), found in 98% and 97% of plots, respectively. Oak fern (Gymnocarpium
dryopteris (L.) Newman) and lady fern (Athyrium filix-femina (L.) Roth ssp. cyclosorum
(Rupr.) C. Chr.) were the most common ferns, appearing in 100% and 86% o f plots,
respectively. Knight’s plume moss (Ptilium crista-castrensis (Hedw.) De Not.) occurred in
90% o f plots and red-stemmed feather moss (Pleurozium schreberi (Brid.) Mitt.) occurred in
87% o f plots.

47

All plots sampled had some regenerating trees in the understory. Tsuga heterophylla was
regenerating in 97% of plots and Thuja plicata regeneration was found in 94% o f plots. Abies
lasiocarpa and Picea engelmannii xglauca were less common, occurring in 71% and 43% o f
plots, respectively. Forest floor was the most frequent substrate used by regenerating trees in
all plots. Coarse woody debris was used as a substrate by regenerating trees in 96% o f plots.
Exposed mineral soil, typically resulting from tree tip-ups, was a substrate used by
regenerating trees in fewer than 5% o f plots. As a size class, saplings (>130 cm tall) were the
most frequent, occurring in all plots, whereas germinants (<10 cm tall) were the least
frequent, occurring in 43% o f plots. Among all size classes and species on any substrate, T.
heterophylla saplings on forest floor were found in the most plots (81%).

Table 4. All plant species, listed by functional group, found in stands sampled for this study
(n=63), with species codes for plant community RDA output interpretation and cluster
analysis interpretation. Presented in this table are the mean, coefficient o f variation, median,
minimum and maximum values for percent cover (abundance values) of each species.
Species highlighted in blue occurred in fewer than 5% o f plots sampled and were thus not
included in the plant community RDA analysis or the cluster analysis. Three lichen genera
found in plots are underlined in this list but this functional group was not reliably sampled in
this study.
Species
Trees
Abies lasiocarpa
Betula papyrifera
Picea engelmannii x glauca
Thuja plicata
Tsuga heterophylla
Shrubs
Acer glabrum
Alnus tenuifolia
A melanchier alnifolia
Cornus sericea
Lonicera involucrata

code

Mean

C.V.

Med.

Min.

Max.

tabla
tbepa
tpigl
tthpl
ttshe

1.21
0.34
0.45
2.81
6.90

2.22
3.24
2.31
1.46
1.26

0.17
0.00
0.00
1.33
3.08

0.00
0.00
0.00
0.00
0.00

16.83
6.67
5.00
23.00
34.00

sacgl

1.09
0.03
0.01
0.01
0.80

3.52
5.77
6.52
7.94
2.65

0.00
0.00
0.00
0.00
0.00

0.00
0.00
0.00
0.00
0.00

27.00
1.33
0.67
0.50
11.00

sloin

48

Species
Menziesia ferruginea
Oplopanax horridus
Ribes glandulosum
Ribes lacustre
Rubus idaeus
Rubus parvijlorus
Rubus pedatus
Rubus pubescens
Sambucus racemosa
Sorbus scopulina
Vaccinium membranaceum
Vaccinium ovalifolium
Vibernum edule
Herbs
Ac tea rubra
Aquilegiaformosa
Aralia nudicaulis
Arnica spp.
Aruncus dioicus
Asarum caudatum
Aster spp.
Chimaphila umbellata
Circaea alpina
Cirsium palustre
Clintonia uniflora
Cornus canadensis
Epilobium angustifolium
Equisetum arvense
Equisetum pratense
Equisetum sylvaticum
Galium triflorum
Geum macrophyllum *
Goodyeraa oblongifolia
fmpatiens noli-iangere
Linnaea borealis
Listera cordata
Lysichiton americanus
Maianthemum racemosa
Mitella spp.
Moneses uniflora
Orthilia secunda

code
smefe
sopho
srigl
srila
sruid
srupa
srupe
ssara
ssosc
svame
svaov
svied
hacru
harau
hardi

hchum
hcial
hcipa
hclun
hcoca
hepan
heqar
heqsi
hgatr
hgoob
hlibo
hlico
hmara
hmitell
hmoun
horse

Mean
0.28
17.31
0.25
2.08
2.31
3.70
2.25
0.12
1.35
0.11
0.48
2.55
0.11

C.V.
2.54
1.08
2.33
1.23
1.51
2.02
1.28
6.39
1.99
2.93
2.00
1.53
3.87

Med.
0.00
9.67
0.00
1.08
0.67
0.33
1.50
0.00
0.13
0.00
0.17
0.92
0.00

Min.
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00

Max.
3.33
66.67
3.50
13.00
15.50
38.00
15.50
5.83
15.17
2.00
5.92
20.33
2.50

0.40
0.01
1.19
0.00
0.23
0.01
0.02
0.02
1.04
0.20
0.99
5.36
1.52
0.82
0.03
1.07
0.76
0.02
0.09
0.01
0.45
0.08
0.00
0.46
0.29
0.02
0.23

2.39
7.94
1.88
7.94
3.24
5.23
4.33
4.46
2.36
3.29
1.67
1.05
2.70
2.53
7.94
1.81
1.41
4.98
3.89
7.94
2.58
4.10
7.54
1.69
3.24
3.06
2.02

0.00
0.00
0.07
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.08
4.00
0.00
0.00
0.00
0.33
0.18
0.00
0.00
0.00
0.00
0.00
0.00
0.07
0.00
0.00
0.00

0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00

4.50
0.50
9.17
0.03
4.00
0.23
0.50
0.75
11.83
4.33
7.17
31.00
22.83
10.17
1.67
9.83
3.83
0.75
2.50
0.67
6.33
2.42
0.25
4.75
4.67
0.37
2.33

Species
code
Mean C.V.
Petasites pahmitus
0.03 5.57
Platanthera dilatata
0.00 7.94
Platanthera orbicu/ata
0.00 5.60
hpyas
Pyrola asarifolia
0.28 3.55
Pyrola grandiflora
0.01
5.68
Senecio triangularis
0.02 7.94
Spiraea spp.
0.24 6.10
Streptopus amplexifolius
hstam
0.33
1.53
Streptopus roseus
hstro
3.68 0.95
Thalictrum occidentale
0.00 7.94
Tiarella spp.
htiarel
4.86 1.01
Valeriana sitchensis
0.03 7.94
hvevi
Veratrum viride
0.64 2.27
Viola spp.
hviola
0.17 2.95
Ferns, fern allies, mosses, grasses and
lichens
Athyrium filix-femina
fatfi
6.41
1.24
Botrychiitm virginiamun
0.00 7.94
Calamagrostis canadensis
gcaca
0.13 2.31
Cladina sod.
Cladonia son.
Dicranum polysetum
mdipo
0.11 2.36
fdrex
Dryopteris expansa
2.02 1.32
Glyceria elata
gglel
0.08 3.67
fgydr
Gymnocarpium dryopteris
7.41
1.13
Hylocomnium splendens
mhysp
9.15 2.25
Lycopodium annotinum
mlyan
0.81 2.43
mmnium 0.72
Mnium spp.
3.21
Peltigera spp.
mplsc
Pleurozium schreberi
1.78 1.33
Polytrichum juniperum
mpoju
0.04 3.57
mptcr
Ptilium crista-castrensis
2.34 1.23
Racomitrium spp.
0.01
7.03
Rhytidiadelphus triquetrus
mrhtr
0.94 1.92
Sphagnum spp.
msphag
0.22 4.09
Thelypteris phegopteris
fthph
0.60 2.78
*Species mistakenly left out o f plant community RDA

50

Med.
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.13
2.67
0.00
2.83
0.00
0.00
0.00

Min.
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.05
0.00
0.00
0.00
0.00
0.00

Max.
1.08
0.08
0.08
6.67
0.20
1.00
10.75
2.17
16.67
0.07
19.00
1.83
6.42
3.17

2.83
0.00
0.00

0.00
0.00
0.00

31.67
0.17
1.58

0.00
0.67
0.00
3.83
1.00
0.07
0.03

0.00
0.00
0.00
0.13
0.00
0.00
0.00

1.67
9.50
2.00
46.67
86.67
14.00
17.00

0.83
0.00
1.33
0.00
0.17
0.00
0.00

0.00
0.00
0.00
0.00
0.00
0.00
0.00

11.00
0.67
12.00
0.67
9.33
6.00
10.17

3.3.2 Regeneration

Trees regenerating in defoliated stands o f the ITR were mostly Tsuga heterophylla, followed
by Thuja plicata (Table 5) in both total and well-spaced tallies. The mean number o f
germinants of Tsuga heterophylla was greater than all other species. Germinants, as a size
class, constituted the majority o f regeneration.

Table 6 summarizes total regeneration by species, size class and substrate. There was a
greater mean number of germinants on coarse woody debris (CWD) than other size classes,
and Tsuga heterophylla as a species was found in higher densities on this substrate than other
species. Mineral soil was the least-used substrate for all size classes and tree species. These
general findings were similar for well-spaced regeneration (Table 7).

Table 5. Total and well-spaced regeneration by species and size class (n=63).
Regeneration density (stems ha'1)
Germinant
Seedling
Sapling
Mean
S.E.
Mean
S.E.
Mean
S.E.
627.5
649.2
Tsuga
Total
992.6
514.3
158.5
92.8
171.4
322.2
Well-spaced
255.0
123.3
46.8
43.6
122.2
346.0
62.4
267.7
43.6
Thuja
Total
159.8
157.7
23.3
120.6
22.4
27.7
Well-spaced
15.2
221.2
216.4
5.3
47.3
56.4
Abies
Total
3.5
117.5
2.1
1.5
78.3
16.1
30.9
Well-spaced
4.2
41.3
13.6
36.5
11.1
2.6
Picea
Total
18.0
23.8
7.7
2.1
6.1
Well-spaced
1.5
All
1,236.0
1,169.8
187.8
123.6
conifers Total
1,161.9
534.5
388.4
621.2
59.2
61.3
Well-spaced
282.5
125.3

51

Table 6. Total regeneration summaries by species, size class and substrate. All numbers are
in stems/ha, and were obtained by tally.___________________________________
Mean
C.V.
Min.
Max.
Tsuga germinants
4.1
992.6
0
28,867
627.5
Tsuga seedlings
2.0
0
8,267
Tsuga saplings
649.2
1.1
0
2,600
Thuja germinants
159.8
6.1
0
7,600
Thuja seedlings
346.0
1.4
0
2,600
Thuja saplings
267.7
1.3
0
1,733
Abies germinants
5.3
5.2
0
200
Abies seedlings
221.2
1.7
0
2,333
216.4
2.1
0
Abies saplings
2,800
Picea germinants
4.2
4.8
0
133
Picea seedlings
41.3
0
2.6
733
Picea saplings
36.5
2.4
0
533
1,130.2
3.7
Germinants on CWD
0
29,000
2.9
31.7
Germinants on FF
0
467
Germinants on MS
0.0
n/a
0
0
1.7
Seedlings on CWD
644.4
0
6,333
Seedlings in FF
1.1
590.5
0
2,800
7.9
Seedlings on MS
0
67
1.1
Saplings on CWD
0.9
448.1
0
1,433
720.6
1.0
Saplings on FF
0
2,833
Saplings on MS
1.1
7.9
0
67
1,680.4
2.5
Tsuga on CWD
0
28,900
Tsuga on FF
586.8
1.1
0
2,900
Tsuga on MS
5.6
0
67
2.1
2.6
Thuja on CWD
366.1
0
7,333
1.3
Thuja on FF
407.4
0
2,667
0.0
n/a
Thuja on MS
0
0
Abies on CWD
1.8
0
1,267
123.3
1.9
Abies on FF
319.6
0
3,867
Abies on MS
0.0
n/a
0
0
Picea CWD
52.9
3.0
0
1,133
Picea on FF
29.1
2.2
0
300
Picea on MS
0.0
0
n/a
0
All conifers
29,000
1,161.9
3.6
0
Germinants
Seedlings
1,235.9
1.2
0
8,867
Saplings
1,169.8
0.8
33
4,267
On CWD
2,222.7
1.9
0
29,200
On FF
1,342.8
0.8
33
5,300
On MS
0
67
2.1
5.5
52

Table 7. Well-spaced regeneration summaries by species and size class. All numbers are in
stems/ha, and were obtained by tally._____________________________________
Max.
Mean
C.V.
Min.
6,867
0
255.0
3.8
Tsuga germinants
2,200
171.4
2.2
0
Tsuga seedlings
1,267
322.2
1.1
0
Tsuga saplings
0
933
5.2
Thuja germinants
23.3
933
1.5
0
120.6
Thuja seedlings
1.4
0
1,167
Thuja saplings
157.7
5.6
0
67
2.1
Abies germinants
1.6
0
667
Abies seedlings
78.3
1,500
2.1
0
Abies saplings
117.5
5.6
0
67
2.1
Picea germinants
2.7
0
267
Picea seedlings
18.0
2.6
0
400
Picea saplings
23.8
All conifers
0
6,933
3.5
282.5
Germinants
1.2
0
2,600
388.4
Seedlings
0.7
0
2,266
621.2
Saplings

3.4 Overstory recovery

3.4.1 Effect of outbreak severity and site variables

There was a significant relationship between overstory recovery (ORIave) and defoliation
severity (DSI) in defoliated stands (rs =0.35, P=0.004). Figure 6 shows that overstory
recovery was greatest in stands with a more severe outbreak history. There was a significant
negative correlation between the ORIave and site index (rs =-0.39, P=0.001; Table 8). Figure 7
shows that overstory recovery was greatest in stands with lower site productivity.

53

1

2

3

4

Def ol i at i on Severity Index (DSI)

Figure 6. Scatterplot with linear least-squares regression line relating Overstory Recovery
Index ( O R I av e ) to Defoliation Severity Index (DSI) and the Overstory Recovery Index
( O R I a v e ) . y=0.12x + 0.36, R2=0.14, p=0.002.

Table 8. Spearman’s correlations between Overstory Recovery Index
( O R I a v e ) and site variables.

ORIave
elevation
site index
steepness
slope position
radiation
soil nutrient regime
indexed soil moisture regime
frost hazard
mean annual temperature
mean annual precipitation
annual heat:moisture index
*Correlation is significant at a = 0.05.

rs
-0.1699
-0.3913
-0.1940
0.0374
-0.0626
-0.0766
0.0532
0.0579
-0.0941
0.2361
-0.2164

54

P
0.1831
0.0015*
0.1277
0.7713
0.6257
0.5508
0.6790
0.6520
0.4632
0.0625
0.0884

12

13

14

15

16

17

18

19

20

Site index

Figure 7. Scatterplot with linear least-squares regression line relating Overstory Recovery
Index ( O R I av e ) to site index. y=-0.09x + 2.47, R =0.18, p<0.001.

3.5 Understory recovery

3.5.1 Effect o f outbreak severity

There was no relationship between shrub cover and damage class as defined by the
Defoliation Severity Index (DSI), nor between either of the two groups o f regeneration
density (total and well-spaced) and the DSI. Shrub cover increased with the proportion of
dead stems (rs =0.46, p<0.001), while in contrast, both total and well-spaced regeneration
densities were negatively associated with the proportion o f dead stems (rs =-0.34, p=0.005,

Table 9. Spearman’s correlations between three understory recovery responses and two
indices o f defoliation severity, the Defoliation Severity Index (DSI) and the proportion of
dead stems.
DSI
% dead stems
rs
P
rs
P
0.4645
% Shrub cover
-0.0582
<0.001*
0.6504
-0.3445
0.005*
Total regeneration (stems ha"')
0.0093
0.9426
-0.3415
Well-spaced regeneration (stems ha ')
0.8979
0.006*
-0.0165
*Correlation is significant at a = 0.05.

3.5.2 Effect of site variables

There was a significant positive relationship between shrub cover and site index (rs =0.70,
p<0.001) and slope steepness (rs =0.42, p<0.001), and a significant negative relationship with
canopy density (rs =-0.56, p<0.001; Table 10). On the other hand, there are indications that
shrub cover peaks under conditions o f 50% to 90% canopy density (Figure 8a) and less steep
sites (Figure 8b). Perhaps the most practical revelation of this analysis is that shrub cover
appears to be at a minimum at site index 16 (Figure 8c), although the lack of shrub cover
data points for lower site indices suggests that this result should be interpreted cautiously.

Surprisingly, total regeneration density was positively associated with canopy density (rs
=0.46, p<0.001; Figure 9a) and negatively associated with site index (rs =-0.34, p=0.006;
Figure 9b). A regression analysis confirms that total regeneration decreases with site index,
however, this variable explains little o f the variation in total regeneration density (R2=0.06).

Well-spaced regeneration was similarly correlated with canopy density in a positive manner
(rs =0.44, p<0.001), and also with annual heat:moisture index (rs =0.29, p=0.018), and

56

negatively associated with site index (rs =-0.33, p=0.006) and slope steepness (rs =-0.27,
p=0.03). There was more well-spaced regeneration on sites of lower steepness (Figure 10a)
and on wanner sites (Figure 10b). Again, well-spaced regeneration appears to decrease with
site index (Figure 10c) and to increase with canopy density (Figure lOd).

Relationships between tree regeneration densities and shrub cover are reported later in this
chapter.

Table 10. Spearman’s correlations between understory recovery variables and site variables.
Total regeneration
Well-spaced
% Shrub cover
regeneration (stems ha"1)
(stems ha"1)
rs
rs
rs
P
P
P
-0.0592
0.6449
-0.0875
0.4953
-0.1227
0.3382
elevation
0.7070
0.0062*
-0.3369
0.0069*
site index
<0.001*
-0.3413
-0.2184
0.0854
-0.2726
0.0307*
0.4252
<0.001*
steepness
-0.1617
0.2054
0.2691
-0.1414
-0.0891
0.4873
slope position
-0.0852
-0.0150
0.9069
radiation
-0.1335
0.2969
0.5067
soil nutrient
0.4835
0.1428
0.2641
0.0899
0.1170
0.3610
regime
indexed soil
0.3616
0.2512
0.0281
0.8269
0.1169
0.1467
moisture regime
<0.001*
-0.5645
0.4646
0.4427
canopy density
<0.001*
<0.001*
-0.0725
0.5723
0.7158
0.0873
0.4963
frost hazard
0.0468
mean annual
0.2099
-0.0492
0.7016
0.1602
0.1469
0.2506
temperature
mean annual
0.6922
0.3515
-0.2396
0.0585
0.0509
-0.1193
precipitation
annual
-0.0940
0.0184*
0.4638
0.1892
0.1376
0.2963
heat:moisture
index
♦Correlation is significant at a = 0.05.

57

o“
% C om petitive shrub cover

% Competitlve shrub cover

H No Wo ^ o
y i oO 'oM oC O
v O
o o

o

100

o

o

% Canopy den sity
00

o

§n-

m

c)
100

U. 90
o

ofj> 80

pfi

70

5
■8 60
>

50

V

40

a

30
E
o
u 20
10

0
12

13

14

15

17

16

18

19

20

Site index

Figure 8. Scatterplots with linear least-squares regression lines for competitive shrub cover
as a function of: a) canopy density; b) steepness; and c) site index,
in
a) y = -0.05x2 + 7.3 86x •
202.02, R2 = 0.32, p=0.003;; b) log y= 0.027x + 0.66, R =0.11, p=0.007; c) log y=0.37x 6.04, R2=0.51, p<0.001.

59

a)
30000

1 1 15000
Ai rtt

10000

SSSSSSBfc
60

70

80

% Canopy d en sity

b)
30000
25000

V
■o

„

20000
e
o to
zs
to J3
t_
V | 15000
s v

«es
v

10000
■•mm

15

16

17

Site index

Figure 9. Scatterplots with linear least-squares regression lines for total regeneration density
as a function of: a) canopy density; and b) site index, a) log y= 0.009x + 2.55, R2=0.13,
p=0.002; b) log y=-0.06x + 4.56, R2=0.06, p=0.04.

60

00
In

W ell-spaced regeneration density
stem s h a 1

W ell-spaced regeneration density
stem s h a 1

ui ooi owoui owoi noui ocn
ooooooooooooooo
oooooooooooooooo

MMNNJWW^4kU1t nO' C\ Nl ' >l
cnocnouioortouiOLnOcnoui
O
O
OOOOOOOOOOOOO
oooooooooooooooo

o

1 !
*sl* M l : :

o

m
s

9
C

01

M

M
o
% Steepness

3
Oi
vi
s■1
(P
9

afP H-l
X

ON

In

4*

o

cn
o

c)
7500
7000
wi 6500
B
■V
o 6000
s
5500
o
■a „ 5000
2 2 4500
S v) 4000
* g 3500
XJ 3000
0w) 2500
CB 2000
a(A 1500
a> 1000
500
£

0
15

16

17

19

18

20

Site in d ex

d)
*

7500 r -------- - --- -----------------------------------------7000 f ---------*--------------------g
6500 | ------------------------------------------------ -----------------------------------£
6000 f~ -------------------- -- ------- ----------------------------------------------c
5500 f - ------------------------------------------------------------------------------o
5000 | ----------------------------------------------------------- -------------------------2 * 4500 f
........... — ..............................-.........................
-g— .............
g ■** 4000 1....................................................................................................—....
g, I 3500 I
--------------------------------------------' ------- g - -------2 2 3000 I
-............... — ..............—.........— ...............
♦ - ... -....
■g " 2500 I....................................................................
* ........9 t * .........

£

§

2000 f

-

~

— ................ ...... ......

S'

t

*

JL

1000 *r-r

|

500 fV

^

*

40

50

60

-

-

V
» -* " # * “ I f

»

-

tip

♦ .-* - -............................... .....
# A

f 4 ...............................................
70

80

90

100

% Canopy den sity

Figure 10. Scatterplots with linear least-squares regression lines for well-spaced regeneration
density as a function of: a) steepness; b) annual heat: moisture index; c) site index; and d)
canopy density, a) log y=-0.01x + 3.09, R2=0.09, p=0.01; b) log y=0.14x + 0.64, R2=0.15,
p=0.001; c) log y=-0.06x + 4.20, R2=0.08, p=0.02; d) log y= O.Olx + 2.29, R2=0.13, p=0.003.

62

3.5.3 Plant community

An overall test of significance in the RDA showed that the canonical relationship between
species composition and the explanatory variables was highly significant (p=0.001 after 999
permutations; pseudo-F=3.3 on all axes), meaning that there was a relationship between the
explanatory variables and plant community composition. Table 11 shows the explanatory
influence o f each variable and its contribution to the model.

The forward selection process identified nine o f the 14 predictors as a sufficient subset: site
index, total live basal area, index o f soil moisture, slope position, MAT, AHM, steepness,
percent live stems and soil nutrient regime (Table 12). I included these nine variables in the
final constrained RDA analysis (Table 13). The first two axes summarize about 26% of the
variation in the plant community data from defoliated stands. Only these two axes are
represented in correlation biplots (Figures 11a and 1lb).

The nine site variables accounted for 43.1% o f the total variation in the plant community of
defoliated stands. The explained variation (adj. R2) o f this final model (which included all
the site variables) was 33.5% (pseudo-F=4.5, p=0.001).

63

Table 11 a) Summary o f the independent effects o f all individual explanatory variables on
overall understory plant community composition; b) Summary o f the conditional effect of
each predictor, representing the variation and significance explained by a predictor after
accounting for the effect of the predictors placed above it in the list. The predictors were
chosen in the order of their decreasing explained variation.____________
a) Independent Term Effects
Variable
Explains %
pseudo-F
P
16.6
site index
12.2
0.001
8.2
5.5
0.001
canopy density
% live stems (all)
7.2
4.7
0.001
total live basal area
6.3
4.1
0.001
steepness
5.9
3.8
0.002
annual heat:moisture index
4.5
2.9
0.008
4.1
2.6
slope position
0.016
3.7
2.3
frost hazard
0.027
elevation
3.6
2.3
0.024
mean annual temperature
3.5
2.2
0.025
3.4
mean annual precipitation
2.1
0.027
3.0
soil nutrient regime
1.9
0.057
2.8
1.7
indexed soil moisture regime
0.073
radiation
1.8
1.1
0.284
b) Conditional Term Effects:
Variable
site index
total live basal area
indexed soil moisture regime
slope position
mean annual temperature
annual heat:moisture index
steepness
% live stems (all)
soil nutrient regime
mean annual precipitation
elevation
radiation
frost hazard
canopy density

Explains %
16.6
5.5
4.3
3.9
3.4
2.7
2.7
2.1
1.9
1.2
1.1
1.1
1.0
1.1

pseudo-F
12.2
4.2
3.5
3.2
2.9
2.4
2.5
1.9
1.8
1.2
1.1
1.0
0.9
1.0

64

P
0.001
0.001
0.001
0.001
0.002
0.003
0.010
0.021
0.041
0.260
0.347
0.416
0.498
0.435

Table 12. Summary of plant community ordination forward selection. Variables are in
selection order and the model adjusted R2 after inclusion o f each successive variable is
shown. Only significant predictors (p<0.05) were retained._______________________
Explains %
Contribution % pseudo-F
Variable
P
16.6
34.2
12.2
site index
0.001
total live basal area
5.5
11.3
4.2
0.001
indexed soil moisture regime
4.3
8.9
3.5
0.002
3.9
3.2
0.001
slope position
7.9
3.4
mean annual temperature
7
2.9
0.002
2.7
5.5
2.4
0.001
annual heat:moisture index
2.7
5.6
2.5
steepness
0.005
2.1
4.3
% live stems (all)
1.9
0.028
1.9
3.9
soil nutrient regime
1.8
0.031

Table 13. Summary of plant community ordination results from the final constrained RDA.
Axis 1
Axis 2
Statistic
Axis 3
Axis 4
0.042
Eigenvalues
0.2059
0.0643
0.0567
Explained variation (cumulative)
20.59
27.01
36.88
32.68
0.8355
Pseudo-canonical correlation
0.8619
0.8309
0.8586
47.72
62.62
Explained fitted variation (cumulative)
75.75
85.49
a)

hepan

t

..

heqsi

sruia

I

V hcipa gcaca

t>k sloin
hlibo
ssara V . Snr*K. k .
srupa .w

<N
22

S ite ln d x s r ila ^ i

'S

<
<

ttlw l %

hac
fa,fih®
g a fhviojt
^ X .im*-

B°5nglsc hchume.

Q

CsL

sopho

S te e g n e s ^

svarnt

MATmsghag mptcr
iorse hmoun

h

mhysp

hvevi -'"v**
■fey mmnium
% liv e s te

o

Ib a to t
I
-

1.0

1.0

RDA Axis 1

65

b)
oo
o'

h ep a n

(3

h vq si

t.

s ru id
V

h cip u
ssar

v

yectca
sloins ■igl

heqar

h lib u

.. sv a o v J ,cOCa

harimihgpa
.srupiisiru svied b sm efe

<N
t/3
*
<

srila -

<

,

ttshe

4 sri/p e

,.•*»..................................hcrcrtr .....

Q
oc

4

f atf i hyair <*fdrex
/ /'■ I

f
hnntg!! tyti

.

/jfdvdn

tnjJttSZ

p-mptcr
nispj/iig

^ & I j c ^ f a a i r c / u i i f p i g i l ^ ^ c o / , ^r ^e hniotimdrysp
so p h o

o

hnicira

&

I/veri &
mm ft! an

I
-

0.8

hp ya s

sacgl

hgoob

RDA Axis 1

0.8

b)
Figure 11. a) Redundancy analysis (RDA) correlation biplot for plant community response
variables and explanatory variables identified during forward selection; and b) only plant
community response, to show more clearly the positioning o f understory species in
ordination space. Shown in a) are the two first axes of a canonical ordination of 63 sites,
Hellinger-transformed percent cover data for 59 understory plant species and nine significant
explanatory variables. Explanatory variables are shown as vectors and species names are
coded. The longer the vector arrow, the stronger the inferred correlation between the
explanatory variable and the species (ter Braak and Smilauer 2002). Species codes are given
in Table 4. The four competitive shrub species are in dark red and the remaining tall shrubs
are purple to show their position in ordination space. The five tree species are in dark green.

66

3.5.3.1 Cluster analysis

Four somewhat dissimilar stand types emerged from the two-way cluster analysis of stands
by their floristic composition (Figure 12; see groups A, B, C and D). Stands dominated by
Oplopanax horridus cover in combination with some Thuja plicata cover comprise one
group (A), which includes site series vk2-01, wk3-05a and wk3-05b. Stands containing an
intermediate level o f Oplopanax in combination with higher proportions o f other shrub
species (Rubus idaeus, Sambucus racemosa and Rubus parviflorus), as well as some
understory Thuja plicata, Tsuga heterophylla and Abies lasiocarpa, comprise another group
(B), and almost all belong to the wk3-05a and wk3-05b site series. A third group (C) contains
little shrub cover, little Thuja regeneration, but ample Tsuga regeneration and some Abies
and Picea cover. This grouping is made up primarily of wk3-01a, wk3-01b, and vk2-06 site
series. Finally, wk3-07 stands dominated by Hylocomium splendens form a distinct cluster
(D).

Figure 13 shows site series groupings visible in biplots of RDA plant community ordination.
Stands were coloured according to their site series to see if they formed clusters based on
their post-defoliation composition. This output revealed definite grouping among some site
series in ordination space. Site series vk2-01 and wk3-07 occupy dissimilar parts o f the RDA
biplot; vk2-01 stands have high Oplopanax horridus cover and wk3-07 sites have minimal
shrub cover and some regenerating Abies lasiocarpa and Tsuga heterophylla. Site series vk206 and wk3-05a and 05b stands overlap in ordination space and may be indistinguishable in
terms o f their recovery response, which appears to include higher shrub cover of all the four

67

competitive species and some Thuja plicata, Picea engelmanii x glauca and Betula
papyrifera regeneration.

If site vectors from the correlation biplot in Figure 1 la are overlaid on Figure 13, the
environmental variables contributing to the way stands are grouped can be visualized. In
particular, wk3-07 stands appear to be warmer (have a higher MAT), vk2-01 stands are
wetter, and the remaining stands in site series vk2-6, wk3-01a and b and wk3-05a and b do
not appear to cluster based on an ecological gradient.

68

■■■■■«■■■•■■■■■ ■■■■■■■££«Hi ■■■■■■■■
mtmmmmmmm
■■■■■■■■■m b b b b b b

ib b b b b b b b b ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ •■ ■ ■ ■ ■ ■ ■

-■■■•■■■>■■■■■■■■■■■■■■■■■■■■■■■■■■■■ ■■■■■■■■

■■■■■

•

mgmm.. mmu.

I C I I U I I I

I I I I I

BBBBBBBB ■■■■

a a a a ^ i :ia a ic : ■ b h h b b b b b b b b b b b b b b ■ ■ bbbbbb
B
BBBBBBB^g^SeBB^eB
CiBBg£ s mm:
iiiiiiK iiin n
. sz
ibssbsbbs^bb mummmmwmmmmmmmmmmmmmmmm
B a B B B B B B ^ S S S D B i lB B e i iB B

mrnmum

m

B

b b b b b sh b b b b b

D

Figure 12. Two-way similarity analysis o f defoliated stands in the ITR, using a
flexible beta linkage method and Sorenson dissimilarity coefficient with PC-ORD. Samples
(stands) are arranged on the basis of species dendrogram sequences. Darker squares represent
higher percent cover of a species. This diagram enables the relationship between samples
(stands) and species groups to be more easily recognized (Kent 2012). The four groups of
stands (A, B, C, D) are indicated by red boxes.

69

CO

o
sr u id

svaov

IA

sru pa

svam e

(N

'x

<
<

'SSOSC

Q
ai

CD

Oi
-

0.8

RDA Axis 1

1.0

Plots by site series
ICHvk2 ICHwk3

Figure 13. Plant community RDA showing the arrangement of understory shrubs and trees
and plots in ordination space. Score scaling is focused on species scores. Stands, represented
by circles, are colour-coded according to site series. The distance between the symbols
approximates the dissimilarity of their species composition as measured by their Euclidean
distance.

70

3.5.3.2 Plant community and outbreak severity

There was a significant association with RDA Axis-2 scores by cumulative damage class
rating (Kruskal-Wallis X2=15.351, df=3, P=0.0015), but no statistical difference was evident
between RDA Axis-1 or Axis-3 and cumulative damage class rating (Axis-1: Kruskal-Wallis
X2=2.967, df=3, p=0.3967; Axis-2: X2=7.499, df=3, p=0.0576). Multiple comparisons
between cumulative damage class ratings revealed that against RDA Axis-2 scores, ratings
“moderate” and “severe” (p<0.001), and “severe” and “severe-severe” (p<0.001) were
significantly different from each other.

3.5.4 Regeneration

There was a significant negative correlation between the total density of germinants,
seedlings and saplings and both percent competitive shrub cover (rs=-0.3883, p=0.0017) and
percent tall shrub cover (rs=-0.3979, p=0.0012). This relationship is shown in Figure 14,
although the slope o f the least squares regression line appears subtle because o f the scale of
the Y-axis. Similarly, there is a significant negative correlation between well-spaced
stems/ha and both percent competitive shrub cover (rs=-0.3685, p=0.003; Figure 15) and
percent tall shrub cover (rs=-0.3818, p=0.002). Only relationships with competitive shrub
cover are displayed in figures.

O f total regeneration by size class, Tsuga heterophylla and Abies lasiocarpa had significant
relationships with shrub cover (Table 14). Tsuga seedling density is negatively correlated

71

with competitive shrub cover (rs=-0.52, p<0.001; Figure 16a) and tall shrub cover (rs=-0.52,
p<0.001). Tsuga sapling density is negatively correlated with competitive shrub cover (rs=0.51, p<0.001; Figure 16b) and tall shrub cover (rs=-0.55, p<0.001). Analyzed as one group,
Tsuga regeneration is negatively associated with competitive shrub cover (rs—0.38,
p<0.0017; Figure 16c) and tall shrub cover (rs—0.41, p<0.001).

Total Abies lasiocarpa seedling density is negatively correlated with competitive shrub cover
(rs=-0.39, p<0.001; Figure 17a) and tall shrub cover (rs=-0.36, p=0.002). Abies sapling
density is negatively correlated with competitive shrub cover (rs—0.54, p<0.001; Figure 17b)
and tall shrub cover (rs—0.49, p<0.001). Analyzed as one group, Abies regeneration is
negatively correlated with both competitive shrub cover (rs—0.48, p<0.001; Figure 17c) and
tall shrub cover (rs—0.44, p<0.001).

Analyzed as a cohort with all species included, total seedlings and saplings are both
significantly negatively correlated with competitive and tall shrubs. Seedlings are negatively
associated with competitive shrub cover (rs—0.53, p<0.001; Figure 18a) and tall shrub cover
(rs=-0.51, p<0.001). Saplings are also negatively associated with competitive shrub cover (rs
—0.53, p<0.001, Figure 18b) and tall shrub cover (rs—0.54, p<0,001).

O f well-spaced regeneration by size class, only Tsuga heterophylla and Abies lasiocarpa had
significant relationships with shrub cover (Table 15). Tsuga seedling density is negatively
correlated with competitive shrub cover (rs—0.49, p<0.001; Figure 19a) and tall shrub cover
(rs—0.46, p<0.001). Tsuga sapling density is negatively correlated with competitive shrub

72

cover (rs=-0.55, p<0.001; Figure 19b) and tall shrub cover (rs =-0.59, p<0.001). Analyzed as
one group, Tsuga regeneration is negatively associated with competitive shrub cover (rs=0.44, p<0.001; Figure 19c) and tall shrub cover (rs=-0.46, p<0.001).

Well-spaced Abies lasiocarpa seedling density is negatively correlated with competitive
shrub cover (rs=-0.43, p<0.001; Figure 20a) and tall shrub cover (rs=-0.40, p=0.001). Abies
sapling density is negatively correlated with competitive shrub cover (rs=-0.48, p<0.001;
Figure 20b) and tall shrub cover (rs=-0.45, p<0.001). Analyzed as one group, Abies
regeneration is negatively associated with competitive shrub cover (rs=-0.55, p<0.001; Figure
20c) and tall shrub cover (rs=-0.51, p<0.001).

Analyzed as a cohort with all species included, well-spaced seedlings and saplings are both
significantly negatively correlated with competitive and tall shrubs. Seedlings are negatively
associated with competitive shrub cover (rs=-0.38, p<0.001; Figure 21a) and tall shrub cover
(rs=-0.35, p<0.0044). Saplings are also negatively associated with competitive shrub cover
(rs=-0.53, p<0.001; Figure 21b) and tall shrub cover (rs=-0.53, p<0.001).

73

30000

1/1

s

25 0 0 0

V

■o
e - 20000
.2 «
2 g 15000
jS S
0S>b ^0)
0tm> 10000
IS
w
o
H

.JL.

-•

^ « I I • k
Lt—
• —#%
§#-— ——
20

40

•
-

—
60

80

100

% C om petitive shrub cover

Figure 14. Scatterplots with linear least-squares regression lines for total regeneration density
and percent competitive shrub cover, log y=-0.16(log x) + 3.51, R2=0.11, p=0.006. All zeros
in this data were changed to ‘0.1’ to accommodate a log transformation.

7500
7 000
6 500
0>
■o 6 0 0 0
eo
5500
5000
« •?
!r 18 4 5 0 0
c ■“ 4 0 0 0
§M E
2 3500
t a 3000
-a » 2 5 0 0

0)
m
a)
V

2000
1500

1000 M

*

500

0 * _ !? § * •
20

40

60

80

100

% C om petitive shrub cover

Figure 15. Scatterplots with linear least-squares regression lines for well-spaced regeneration
density and percent competitive shrub cover, log y=-0.15(log x) + 3.12, R =0.13, p=0.002.
All zeros in this data were changed to ‘0.1 ’ to accommodate a log transformation.

74

Table 14. Spearman’s correlations for the total number o f total understory trees by size class
and two classes o f shrub cover.
%Tall shrub cover
% Competitive shrub cover
Total regeneration
rs
-0.1096
Tsuga
Germ.
-0.5272
Seed.
-0.5174
Sap.
-0.3881
All Tsuga
0.0787
Thuja
Germ.
-0.0727
Seed.
0.1564
Sap.
0.1093
All Thuja
0.0730
Abies
Germ.
-0.3928
Seed.
-0.5487
Sap.
-0.4800
All Abies
Picea
Germ.
0.1433
-0.1986
Seed.
0.0786
Sap.
-0.0470
All Picea
-0.0695
All species Germ.
-0.5364
Seed.
-0.5332
Sap.
C orrelation is significant at a = 0.05.

rs
0.5079
-0.5292
-0.5562
-0.4148
0.0655
-0.0827
0.1305
0.0716
0.0643
-0.3692
-0.4974
-0.4463
0.1602
-0.1184
0.1238
0.0304
-0.0485
-0.5144
-0.5484

P

0.3925
<0.001*
<0.001*
0.0017*
0.5397
0.5712
0.2209
0.3937
0.5697
0.0015*
<0.001*
<0.001*
0.2625
0.1188
0.5402
0.7148
0.5885
<0.001*
<0.001*

a)
.
.■S'

8500
8000
2 7500
5
7000
■?
6500
o
6000
■S
5500
2 - 5000
g 2 4500
4000
S? £ 3500
m S 3000
s “ 2500
=
2000
S
1500
8
iooo
a
500

a
H

•
•1

-1
3
4
4
i

I .
f •
•
w
<

,*

,

♦
*« * V *

0
0

20

40

60

% C om petitive shrub cover

75

*
80

» .....................
-............
100

P

0.5079
<0.001*
<0.001*
<0.001*
0.6100
0.5192
0.3082
0.5770
0.6164
0.0029*
<0.001*
<0.001*
0.2098
0.3555
0.3338
0.8130
0.7056
<0.001*
<0.001*

0

20

40

60

80

100

% C om petitive shrub cover

c)
30000

u
•o
c
o

« '1
V -C

GJQfi

25000
20000
15000

g a
nBJD “ 10000
3</)
H
5000 i"W
3o
H
o
0

(ft®
20

40

*
60

% *
80

100

% C om petitive shrub cover

Figure 16. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) total Tsuga
seedlings; b) total Tsuga saplings; and c) total Tsuga (all size classes), a) y=-696.26(log x) +
1309.06, R2=0.22, p=<0.001; b) y=-409.48(log x) + 1049.96, R2=0.22, p=<0.001; c) log y=0.28(log x) + 3.14, R2=0.06, p=0.03. All zeros in shrub cover data were changed to ‘0.1 ’ to
accommodate a log transformation, as were zeros in regeneration density data in c).

Abies seed lin g regeneration density
stem s h a 1

Abies sapling regeneration density
stem s h a 1
o

O
©

o

U
1
o
o

2000

o

U
1
o

o

«
#

<

"J
-vl

2
or
oo
<
fD

% Competitive shrub cover

on
*re
T3

o

C
/1
©

o

►
-*
©

o

o

1—
‘
C/1

©

o

N3
©
©

o

NJ
C/1

©

o

c)
5500
5000 iJL.
W
!
B
01 4500
■o
e
4000
o
'•O•? 3500
B«
s ■« 3000
C g
V
oo>n.3&
s 2500
u in 2000
V
1500
S
<
1000
3
o
500

H

0
40

60

80

100

i C om petitive shrub cover

Figure 17. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) total Abies
seedlings; b) total Abies saplings; and c) total Abies (all size classes), a) log y=-0.70(log x) +
1.64, R =0.12, p=0.004; b) log y=-0.981(log x) + 1.81, R2=0.25, p=<0.001; c) log y=0.81(log x) + 2.29, R2=0.17, p=<0.001. All zeros in this data were changed to ‘0.1’ to
accommodate a log transformation.

78

a)
9000

t/i 8000
c
4>
•O 7000
c
.2
*8 6000
2V5000
S J3
bfi £2 4000
u0) 0c)
gp ts 3000
c
54) 2000
cV
u)
1000

3o

•

0

H

f

®----- 1—■
—% —♦♦
20

40

100

60

% C om petitive shrub cover

b)
4500

£
3) 4000
S
0)
■o 3500
e
© 3000
*•8
ffi *
u©
R)
JS 2500 f -*g

"W~

V)

ac a 2000

2L> §
1500
s
a(8 1000
130 *

(A

3o

H

500

0
20

40

60

80

100

% Com petitive shrub cover

Figure 18. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) total seedlings (all
species); and b) total saplings (all species), a) y=-933.52(log x) + 2149.77, R2=0.28,
p=<0.001; b) y=-19.55 x + 1652.27, R2=0.22, p=<0.001. Shrub cover values in a) that were
zero were changed to ‘0.1 ’ to accommodate a log transformation.

79

Table 15. Spearman’s correlations between the number o f well-spaced understory trees by
size class and two classes of shrub cover.
% Competitive shrub cover
%Tall shrub cover
Well-spaced regeneration
rs
Germ.
Tsuga
-0.0481
Seed.
-0.4960
Sap.
-0.5596
All Tsuga
-0.4404
Germ.
Thuja
0.1533
Seed.
0.0063
Sap.
0.2197
All Thuja
0.2040
Abies
Germ.
0.0299
Seed.
-0.4396
Sap.
-0.4839
All Abies
-0.5511
Germ.
Picea
0.1792
Seed.
-0.0319
Sap.
0.0025
All Picea
0.0543
All species Germ.
0.0399
Seed.
-0.3877
Sap.
-0.5320
*Correlation is significant at a = 0.05.

P
0.7083
<0.001*
<0.001*
<0.001*
0.2304
0.9611
0.0836
0.1088
0.8162
<0.001*
<0.001*
<0.001*
0.1598
0.8041
0.9844
0.6725
0.7562
0.0017*
<0.001*

rs
-0.0601
-0.4642
-0.5915
-0.4689
0.1306
-0.0197
0.2098
0.1714
0.0398
-0.4037
-0.4550
-0.5118
0.1792
0.0345
0.0375
0.1119
0.0281
-0.3545
-0.5309

a)
2500
60

*3
6* 2000
4) .—
v> a

5)TJ '* 1500
s e G
£ -c
° V)
H
E
■oja w
u
«
1000
u «
n e
a, v

in 60
341■ uV

5 00

♦

t
♦

20

40

60

% C om petitive shrub cover

80

80

-

100

P
0.6401
<0.001*
<0.001*
<0.001*
0.3076
0.8780
0.0988
0.1792
0.7566
0.001*
<0.001*
<0.001*
0.1599
0.7885
0.7705
0.3826
0.8267
0.0044*
<0.001*

0

20

40

60

80

100

% C om petitive shrub cover

c)
7500
7000
6 500
6 000
SP'5
5 500
5 s
5 000
£ ^ « 4500
« .Sg -2
u
i2 4 000
3500
2.
g
I
w r *3 3 000
■ 2c t«
V V 2 500

>
60
9 u
v

2000
1500

1000
500

0
100
% C om petitive shrub cover

Figure 19. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) well-spaced Tsuga
seedlings; b) well-spaced Tsuga saplings; and c) well-spaced Tsuga (all size classes), a) y=221.81 (log x) + 388.55, R2=0.25, p=<0.001; b) y=-197.48(log x) + 515.53, R2=0.23,
p=<0.001; c) log y = -0.015x + 2.76, R2=0.13, p=0.003. Shrub cover values in a) and b) that
were zero were changed to ‘0.1 ’ to accommodate a log transformation, as were regeneration
density values in c).

fc3

W ell-spaced Abie seed lin g
regeneration density
stem s h a 1

W ell-spaced Abie sapling
regeneration density
stem s h a 1
L

N

^

^

CO

O

i

N

|— I

O
o

O
n
O
O

r-o
o
% Competitive shrub cover

o

00

o

oo

o

O
o

u?
o
o

o
o

cn

©
o

o

o
o

Vi

o
o

c)
30 0 0 0
25 0 0 0

£ B a 15000

10000
5000

0

20

60

40

80

100

% C om petitive shrub cover

Figure 20. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) well-spaced Abies
seedlings; b) well-spaced Abies saplings; and c) well-spaced Abies (all size classes),
a) y=-89.1 l(log x) + 165.54, R2=0.55, p=<0.001; b) log y=-0.75(log x) + 1.41, Rz=0.16,
p=<0.001; c) log y = -0.27(log x) = 3.46, R2=0.17, p=<0.001. All shrub cover values that
were zero were changed to ‘0.1’ to accommodate a log transformation, as were regeneration
density values in b) and c).

83

a)
3000

cf £?

2500

5 I

2000
0) "
o
re
V) ■
e js
■41O .2
u a “£ 1500
41

re 2 I
a
IA B
4>

&
C
4)
U

1000
500

0
0

20

60

40

80

100

pC om petitive shrub cover

b)
2500

M £> 2000
£ '5
a 4) *4
*9

" e- js2 1500
■a
9) O up
U*3
C
rarer
g- i « iooo

j. a
5S S
o
V
>

u

500
ft

_ f
20

r”

40

—

60

100

% C om petitive shrub cover

Figure 21. Scatterplots with linear least-squares regression lines for associations between
understory regeneration density and percent competitive shrub cover: a) well-spaced
seedlings (all species); and b) well-spaced saplings (all species), a) y=-291.07(log x) +
673.28, R2=0.26, p=<0.001; b) y=-8.85x + 839.59, R2=0.19, p=<0.001. All shrub cover
values in a) that were zero were changed to ‘0.1’ to accommodate a log transformation.

84

3.5.4.1 Regeneration niche

I treated regeneration densities by tree species and size class as multivariate response
variables in an RDA analysis. An overall test o f significance showed that the canonical
relationship between regeneration density by species and size class and the explanatory
variables was highly significant (p = 0.001 after 999 permutations; pseudo-F = 2.4 on all
axes), meaning that there was a meaningful association between the explanatory variables
and regeneration densities. Total variation o f the RDA model was 26%. The resulting
explained variation (adj. R2) o f the global model (which included all the site variables) was
24%.

Table 16 shows the explanatory influence o f each variable and its contribution to the model.
The forward selection process identified four o f the 14 predictors as a sufficient subset: site
index, index of soil moisture, percent live stems and total live basal area (Table 17). I
included these four variables in the final constrained RDA analysis (Table 18). The first two
axes summarize approximately 18% o f the variation in the regeneration density by species
and size class data from defoliated stands. Only these two axes are represented in the
scatterplot (Figure 22).

The four site variables accounted for 23.7% o f the total variation in the regeneration
attributes o f defoliated stands. The explained variation (adj. R2) o f this model was 18.4%
(pseudo-F=4.5, p=0.001).

85

A greater abundance o f Thuja plicata o f all size classes (germinants, seedlings and saplings)
in high site index stands was also seen in the plant community RDA analysis. Tsuga
heterophylla seedlings, Picea engelmanii x glauca and Abies lasiocarpa appear to tolerate
lower-productivity sites. Picea and Abies seedlings and saplings occur in higher densities on
sites with higher relative moisture. If a higher proportion o f live stems is interpreted as an
indicator of a lower-severity defoliation, then Abies seedling and sapling densities are higher
in stands that experienced lower-severity outbreaks.

The regeneration niche RDA (Figure 23) shows that regenerating trees group themselves
across site series by species, and that the site series wk3-07, vk2-01 and vk2-06 form distinct
clusters, while the remaining site series are not as distinguishable from each other in their
regeneration response to defoliation history. Tsuga heterophylla seedlings are concentrated
in wk3-07 site series stands, Abies lasiocarpa seedlings and saplings occur mostly in vk2-06
site series stands, and Thuja plicata regeneration o f all size classes is found in the vk2-01 and
wk3-05. Site series wk3-01 stands contain regenerating trees of all species. Picea engelmanii
x glauca regeneration is sparse but found primarily in the vk2 biogeoclimatic variant. The
stands that do not cluster based on site series (wk3-01 (a and b) and wk3-05 (a and b)) appear
to fall along a moisture gradient.

86

Table 16. a) Summary of the independent effects of all explanatory variables on overall
regeneration density by species and size class; b) Summary o f the conditional effect o f each
predictor, representing the variation and significance explained by a predictor after
accounting for the effect of the predictors placed above it in the list. The predict
chosen in the order of their decreasing explained variation.
a) Independent Term Effects
Explains %
pseudo-F
Variable
P
0.001
9.8
6.6
site index
6.2
4
0.001
indexed soil moisture regime
3.9
0.001
6
canopy density
5.7
3.7
0.002
% live stems (all)
4.7
3
0.006
steepness
4.7
3
0.006
total live basal area
2.7
4.3
0.012
slope position
3.9
2.5
0.022
elevation
3.6
2.3
0.021
frost hazard
mean annual temperature
2.9
1.8
0.091
2.2
1.4
0.195
radiation
0.9
0.45
1.5
soil nutrient regime
1.1
0.7
0.67
annual heat:moisture index
0.8
0.5
0.868
mean annual precipitation
b) Conditional Term Effects:
Variable
site index
indexed soil moisture regime
% live stems (all)
total live basal area
mean annual temperature
elevation
slope position
mean annual precipitation
annual heat.moisture index
steepness
soil nutrient regime
frost hazard
radiation
canopy density

Explains %
9.8
5.4
5.2
3.4
2.1
1.5
2
1.9
5.5
1.2
0.9
1
0.9
0.5

87

pseudo-F
6.6
3.8
3.8
2.6
1.6
1.2
1.5
1.5
4.6
1
0.8
0.8
0.7
0.4

..

P
0.001
0.002
0.004
0.012
0.13
0.313
0.143
0.18
0.001
0.38
0.659
0.587
0.665
0.903

Table 17. Summary o f tree regeneration niche ordination results from forward selection.
Variables are in selection order and the model adj. R2 after inclusion of each successive
variable is shown. Only significant predictors (p<0.05) were retained._______________
Variable
site index
indexed soil moisture regime
% live stems (all)
total live basal area

Explains %
9.8
5.4
5.2
3.4

Contribution %
23.8
13
12.5
8.3

pseudo-F
6.6
3.8
3.8
2.6

P
0.001
0.002
0.001
0.019

Table 18. Summary of results from the final constrained RDA for tree regeneration niche.
Axis 2
Axis 1
Axis 3
Axis 4
Statistic
0.1064
0.0824
0.0339
Eigenvalues
0.0143
18.88
Explained variation (cumulative)
10.64
22.27
23.7
Pseudo-canonical correlation
0.7296
0.6243
0.4409
0.3217
44.89
79.64
93.96
Explained fitted variation (cumulative)
100

oo
o
tssap

tsseed

<N
v.

'x

<
<

thseed
thsap

D
a:

S ite ln d x
CD

ihgerm

abseed
■£
»
.
absap

tsgerm

pis up
lb a _ t o t is m r

OI
-

1.0

RDA Axis 1

0.8

Figure 22. a) Redundancy analysis (RDA) correlation biplot for tree regeneration density by
species and size class response variables and explanatory variables identified during forward
selection. This figure shows the two first axes of a canonical ordination o f 63 sites, 10
Hellinger-transformed regeneration densities by species and size class and four significant
explanatory variables. Explanatory variables are shown as vectors and tree species names are
coded. The longer the vector arrow, the stronger the inferred correlation between the
explanatory variable and the species (ter Braak and Smilauer 2002). Tree species codes are
based on the first two letters of the genus followed by the first letters of the size class.
“thsap” represents the Thuja plicata saplings.

88

o

tssap

tsse e d

<Dl
tsg erm
p is e e d

p isa p

CD

O
-

RDA Axis 1

0.6

0.6

Plots by site series
ICHvk2 ICHwk3

■

E
m

m

Figure 23. Total regeneration density RDA showing the arrangement of understory tree
species by size class and plots in ordination space. Score scaling is focused on tree species
scores. Stands, represented by circles, are colour-coded according to site series. The distance
between the symbols approximates the dissimilarity of their regeneration composition as
measured by their Euclidean distance.

89

3.5.4.2 Regeneration substrate

The availability of each substrate (forest floor, coarse woody debris and mineral soil) was not
measured in this study, and so it is not possible to determine the relative importance o f each
substrate to regenerating trees in defoliated stands. In general, however, it was observed that
forest floor was always more abundant than coarse woody debris (CWD), which was more
abundant than exposed mineral soil. Across stands, CWD was the substrate with the highest
number of regenerating trees (Figure 24a), with forest floor being the substrate type with the
second highest number of regenerating trees, and mineral soil having the fewest germinants,
seedlings or saplings using it. Tsuga heterophylla, as the most abundant regenerating species,
was the most prolific user of CWD, and germinants were the size class most numerous on
this substrate type (Figure 24b). Again, it is not always possible to correctly identify the
establishment substrate for a sapling, and this should be considered in when interpreting the
regeneration substrate results for this size class.

Analyzed as one cohort, Tsuga heterophylla regeneration on CWD is negatively correlated
with competitive and tall shrub cover (rs=-0.27, p<0.001; rs=-0.27, p=0.030; Table 19).
Tsuga seedlings and saplings analyzed separately are negatively correlated with competitive
shrub cover (rs=-0.38, p=0.001; rs=-0.48, p<0.001) and tall shmb cover (rs=-0.38, p=0.001;
rs=-0.51, p<0.001). All Abies lasiocarpa regeneration on CWD, as well as Abies saplings on
CWD analyzed on their own, are also negatively correlated with competitive and tall shrub
cover. Analyzed as a cohort with all species included, seedlings and saplings are both
significantly correlated with competitive shrubs (rs=-0.32, p=0.008; rs=-0.43, p<0.001) and

90

tall shrubs (rs=-0.28, p=0.022; rs=-0.45, p=0.001), showing negative associations in both
instances. The exception to this pattern is Thuja plicata regeneration, which has a positive
association with both competitive and tall shrubs (rs=0.31, p=0.0129; rs=0.2940, p=0.0193).

There was no significant correlation between the proportion of total regeneration of all
understory tree species on CWD and competitive shrub cover (rs = 0.1970, p = 0.1218) or tall
shrub cover (rs =0.2105, p = 0.0977). However, there was a positive association between the
proportion o f Thuja plicata saplings on CWD and both competitive shrub cover (rs =0.30, p
= 0.014) and tall shrub cover (rs =0.29, p = 0.017; Table 20). Analyzed as one group (all size
classes), Thuja regeneration on CWD is positively associated with competitive shrub cover
(rs =-0.39, p = 0.0017) and tall shrub cover (rs =-0.38, p = 0.0019).

91

1600

n 1400
W
)

JS

E

A 1200
--

1000

■ Tsuga
■ T huja
□ Abies

&

400

n

200

HPicea

G erm jSeed.j Sap. |GermjSeed.[ Sap. perm JS eed.i Sap.
CWD

|

FF

|

MS

b)
1600
it 1400

£ 1000
■ Germ.
USeed.
□ Sap.

O Cju
£ ! u.
U !

00
S

o

u.
a.

00
2

u
P icea

Abies

Figure 24. Mean tree regeneration density (+ S.E.) by species, size class and substrate,
grouped by substrate a) and by species b). CWD: coarse woody debris, FF: forest floor, MS:
mineral soil.

92

Table 19. Spearman’s correlations for regeneration density (stems/ha) on coarse woody
debris by size class and two classes of shrub cover._______________________________
% Competitive shrub cover
%Tall shrub cover
stems/ha on CWD

rs
-0.0984
Tsuga
Germ.
-0.3878
Seed.
Sap.
-0.4846
-0.2707
All Tsuga
Germ.
Thuja
0.1707
Seed.
0.0569
Sap.
0.2350
0.3118
All Thuja
Abies
Germ.
0.0838
Seed.
-0.1657
Sap.
-0.3418
-0.2947
All Abies
Germ.
Picea
0.1433
Seed.
-0.0445
Sap.
0.1108
All Picea
0.0703
All species Germ.
-0.0603
Seed.
-0.3287
Sap.
-0.4380
^Correlation is significant at a = 0.05.

P
0.4429
0.0017*
<0.001*
0.0319*
0.1811
0.6576
0.0637
0.0129*
0.5137
0.1943
0.0061*
0.019*
0.2625
0.7290
0.3874
0.5840
0.6388
0.0085*
<0.001*

93

rs
-0.0669
-0.3873
-0.5113
-0.2730
0.1825
0.0750
0.2130
0.2940
0.0559
-0.1467
-0.2981
-0.2654
0.1602
0.0431
0.1509
0.1458
-0.0280
-0.2877
-0.4595

P
0.6022
0.0017*
<0.001*
0.0304*
0.1522
0.5592
0.0937
0.0193*
0.6636
0.2512
0.0177*
0.0355*
0.2097
0.7372
0.2377
0.2542
0.8276
0.0222*
<0.001*

Table 20. Spearman’s correlations for the proportion of total stems on coarse woody debris
by size class and two classes o f shrub cover.______________________________________
% Competitive shrub cover
%Tall shrub cover
% on CWD

rs
Germ.
-0.1531
Tsuga
-0.1074
Seed.
Sap.
-0.0725
0.1128
All Tsuga
Germ.
0.1701
Thuja
Seed.
0.1613
Sap.
0.3057
0.3901
All Thuja
Abies
Germ.
0.0838
Seed.
-0.0727
Sap.
-0.2125
-0.1066
All Abies
Picea
Germ.
0.1435
Seed.
-0.0438
Sap.
0.0957
All Picea
0.0578
All species Germ.
-0.1102
Seed.
0.0793
Sap.
0.0364
’■‘Correlation is significant at a = 0.05.

P
0.2311
0.4023
0.5722
0.3786
0.1827
0.2065
0.0148*
0.0017*
0.5137
0.5714
0.0945
0.4057
0.2620
0.7329
0.4557
0.6525
0.3901
0.5367
0.7773

94

rs
-0.1162
-0.0862
-0.0763
0.1371
0.1825
0.1827
0.2996
0.3868
0.0559
-0.0707
-0.1865
-0.0927
0.1599
0.0434
0.1331
0.1285
-0.0671
0.1268
0.0079

P
0.3643
0.5020
0.5524
0.2840
0.1522
0.1518
0.0170*
0.0019*
0.6636
0.5821
0.1432
0.4699
0.2108
0.7354
0.2984
0.3156
0.6013
0.3219
0.9512

4.0 Discussion

4.1 Overstory recovery

4.1.1 Effect of outbreak severity and site variables

The effect o f a defoliation event on overstory trees is a function of the interacting factors o f
severity, timing and duration o f defoliation, site characteristics, the initial condition of trees
and presence of secondary insects and disease organisms (Kulman 1971, Laursen 1977). In
comparing the growth response effects o f different severities o f hemlock looper defoliation
to those of other defoliators, such as the western spruce budworm (Iqbal 2011), researchers
have found that it may be the stimulation o f suppressed buds that greatly increases foliage
regrowth and tree recovery from defoliation (Piene 1989, Piene and MacLean 1999, Iqbal
2011). Similarly to the spruce budworm, the hemlock looper consumes all age classes of
foliage, including buds, which may contribute to the quicker observed recovery o f hemlock
looper defoliated trees when compared to trees defoliated by other species, such as the
balsam fir sawfly (Neodiprion abietis Harris) (Iqbal 2011).

Aerial survey data o f insect defoliation can be more economical than large-scale field
surveys and provide greater spatial coverage o f the symptoms caused by forest pests (Harris
and Dawson 1979). They are known, however, for lacking the accuracy necessary for local
or operational purposes (MacLean and MacKinnon 1996, Magnussen and Alfaro 2012). In
this study, Forest Health Aerial Overview Survey data were used to create a Defoliation

95

Severity Index (DSI), and were also used in combination with on-the-ground data of
percentages of live stems at the stand level to derive the Overstory Recovery Index (ORIave)In other words, aerial survey data were relied upon in this study for information on
defoliation history and hence the degree o f contrast (recovery) o f stand overstories.

To the extent that the Forest Health Aerial Overview Surveys from 1991 to 1994 can be
relied upon, overstory recovery in the ITR has been greatest in stands with a more severe
outbreak history, and in those with lower site productivity. One possible explanation is a
greater probability of individual tree recovery and growth in stands that are sparsely stocked
with mature, living trees (which therefore experience less competition). Another possibility
is that there is higher nutrient availability (from frass) is more severely defoliated stands. Yet
the poor resolution and reliability o f the aerial surveys still makes these very tenuous
relationships. It is also unclear why stands wirh a lower site index have greater recovery than
stands with a higher site index, as one would expect higher-productivity sites to have more
soil resources available for tree recovery.

The demographic response o f trees - their survival and reproduction - only portrays one
dimension of recovery after a WHL defoliation. There is likely a continuum o f growth
responses of trees, both in the surviving overstory and suppressed “pole-sized” trees. That
growth response can be described with the analysis o f tree ring increments. Follow-up
dendrochronological research is recommended for a more accurate description o f overstory
recovery at the stand level.

96

One study examining Tsuga heterophylla recovery in managed stands following an outbreak
of western blackheaded budworm (Acleris gloverana Walsingham), found that recovery was
greatest in stands that had been spaced, even where the defoliation had been most severe
(Nealis and Tumquist 2010). This may be due to greater resource availability for individual
trees in stands that have been spaced. Other researchers suspect that greater resource
availability in the form o f nutrients, water and sunlight in higher productivity stands provide
trees with the ability to produce foliage with greater resistance to defoliators (Day 2013).
These researchers suggest that tree recovery following defoliation may be tied to stores of
non-structural carbohydrates in sapwood that can be used to grow new foliage and produce
compounds such as lignins and tannins (Day 2013).

4.2 Understory recovery

4.2.1 Effect of outbreak severity

Any influence of mapped outbreak severity on shrub cover and tree regeneration was not
apparent in defoliated stands when evaluated in a correlation analysis using the Defoliation
Severity Index as the independent variable. These results could also indicate that the
cumulative damage class ratings and the DSI are too coarse to capture the variation in
severity across landscapes aerially surveyed, or that a broader range of defoliation severities
should have been included in this study.

The proportion o f dead stems is a weaker representation o f defoliation severity than the DSI,
since this metric may be representing mortality due to other factors, and here we assume that
97

it is an adequate surrogate for the proportion of trees killed directly or indirectly by the
WHL. Shrub cover has a distinctly positive association with the proportion o f dead stems,
and both regeneration types (total and well-spaced stem counts) have a slightly weaker
negative association with this measure o f defoliation severity. Tree regeneration appears to
be suppressed by the growth o f shrubs observed in high-severity outbreak areas.

Shrubs are presumably responding to an increase in light levels and/or a pulse in nutrients
brought about by the return o f frass to the forest floor (Reichle 1973). Although there is
research showing that atmospheric nitrogen deposition o f inorganic nitrogen leads to
fertilization effects in terrestrial ecosystems (Throop and Lerdau 2004), and that frass
deposited by defoliators contributes to forest floor nitrogen levels (Grace 1986, Hunter 2001,
Lovett et al. 2002), there seems to be little published evidence demonstrating an increase in
understory growth due to the influence o f frass from other defoliating insects such as the
gypsy moth (Lymantria dispar L.), western spruce budworm or forest tent caterpillar
(Malascosoma disstria Hubner). In general, the long-term consequences of the transfer of
foliage to insect biomass and then to the forest floor are not well understood (Russell et al.
2004).

Twenty years following a WHL outbreak, the influence o f nutrient inputs to the forest floor
may be less important than the effect o f higher light levels in more severely defoliated
stands. High precipitation could decrease the contact of nitrogen from frass with soil during
or after a defoliation event, flushing this nutrient before it can be retained (Lovett et al.
2002), and this scenario seems plausible in a high-precipitation system such as the ITR.

98

Defoliation increases light penetration to the forest floor (Collins 1961, Campbell and Sloan
1977) and even though tree growth rates can recover to pre-defoliation levels within five
years of an outbreak (Iqbal 2011) it likely takes longer than that for a severely defoliated
canopy to return to creating pre-disturbance light levels on the forest floor.

4.2.2 Effect of site variables

Shrub cover responded strongly to site index in defoliated stands, being greater in higherproductivity stands. Although site index is commonly used for silvicultural purposes and
forest planning based on specified managed tree species (MoFR 2011), it was designed to
reflect the net effects o f climate, terrain and soil on tree growth, and so it is a suitable
indicator of general plant productivity. The four competitive shrub species (analyzed together
due to their similar functional effects on tree regeneration) are also all nitrophilous to varying
degrees, and might also be reacting to higher soil nitrogen resulting from the WHL
defoliation event while already exhibiting good productivity on high site index stands. In
particular, Rubus idaeus tends to be profuse immediately after a nitrate-releasing disturbance
event, before abating quickly after nutrients have subsided, unless the site is exceptionally
rich (S. Haeussler, pers. comm.). Sambucus racemosa similarly establishes and grows
immediately after a disturbance but persists longer than Rubus idaeus (Haeussler et al. 1990).
Rubus parviflorus and Oplopanax horridus are longer lived and more tolerant o f later serai
conditions (S. Haeussler, pers. comm.). It is not known what the abundance o f these species
was in sampled stands before the defoliation event, but it is likely that the increased light

99

and/or nitrate inputs from defoliation encouraged their establishment or growth following the
disturbance.

Shrub cover was higher on steeper slopes and lower in stands with a higher canopy density.
A comparison o f means between generally North-facing slopes and generally South-facing
slopes found no significant difference between shrub cover based on aspect, indicating that
higher shrub cover on steeper slopes may be a function o f better water drainage in those
stands rather than a function o f radiation. Oplopanax horridus is said to require waterreceiving but well-drained sites (Klinka et al. 1989, Beaudry et al. 1999) and higher shrub
cover on steeper sites may be reflecting this species primarily.

In general, shrub cover is largely a function o f canopy openness in defoliated stands. The
four competitive shrub species included in this analysis have different shade tolerances,
however. Rubus idaeus is considered shade-intolerant (Beaudry et al. 1999) and requires
canopy openings in order to establish and survive (Haeussler et al. 1990). Sambucus
racemosa, Rubus parviflorus and Oplopanax horridus are all considered somewhat shadetolerant (Beaudry et al. 1999). Sambucus, although tolerant of deep shade, grows best in
forest openings (Harcombe et al. 1983). Rubus parviflorus grows under closed canopies and
in open conditions, but has higher cover and greater vigour as understory light increases
(Mueggler 1965, Irwin and Peak 1979). Oplopanax horridus will grow towards canopy gaps
and sunflecks in low light conditions (Lantz and Antos 2002) and thereby maximizes light
capture in forest understories (Roorbach 1999). Oplopanax dominates the shrub layer across
most stands, and was absent from only three o f the 63 plots included in this study.

100

The correlation analysis of relationships between regeneration densities and various site and
stand factors suggests that perhaps regeneration might be responding to the abundance of
competitive shrubs, and that the results are not necessarily a reflection of how regeneration
directly responds to the site variables measured. That total regeneration increased with
canopy cover and decreased with site index, while shrub cover did the opposite, probably
indicates that shrubs were able to react sooner or more strongly to these site variables and
flourish before regeneration was able to establish post-disturbance. Well-spaced regeneration
appears to respond to sites that are warmer and drier, indicating that the most vigourous
germinants, seedlings and saplings have somewhat more o f a competitive advantage in
stands on such sites.

4.2.3 Plant community

The multivariate plant community analysis revealed that competitive shrubs occupy a
different portion o f multivariate space than do most regenerating trees. With the exception o f
Thuja plicata, which appears to tolerate stands with more competitive shrub cover,
regenerating tree cover of Tsuga heterophylla, Abies lasiocarpa and Picea engelmannii x
glauca is higher in stands where there is less shrub cover. As in the exploration of
associations through bivariate correlation analysis, site index is the strongest explanatory
driver of competitive shrub cover. Regenerating tree species generally fell in different
quadrats o f the RDA output from each other, reflecting their relative abundance across plant
community types in defoliated stands, the site variables that underpin that variation, and their
inherent niche differentiation that facilitates their coexistence in the ITR.

101

When considered together, the site variables with the most explanatory influence on
understory vegetation pattern are (in decreasing order): site index, total live basal area, soil
moisture index, slope position, mean annual temperature, annual heat:moisture index,
steepness, percent live stems and relative soil nutrients. Site index exerts a powerful positive
influence on shrub cover, with steepness and relative soil nutrient regime influencing shrub
cover to a lesser degree. Soil moisture index, annual heatmoisture index, mean annual
temperature and the proportion of live stems in the overstory have a positive association with
Tsuga heterophylla, Abies lasiocarpa, Picea engelmanii x glauca and Betulapapyrifera.
Thuja plicata appears to endure a shrub canopy and thrives under the same site
characteristics exploited by competitive shrub species in this ecosystem.

4.2.3.1 Cluster analysis

The motivation for the cluster analysis was primarily to see whether plant communities could
be objectively grouped twenty years following the WHL outbreak. Using arbitrary “stopping
points,” four reasonably dissimilar stand types emerged from the hierarchical cluster
analysis. Figure 25 shows photographs of the site series that typify each o f the four clusters
(groups A to D) resulting from this analysis. The RDA cluster analysis was useful is
visualizing how stands clustered by site series based on their plant community response to
the defoliation event. There appears to be overlap among some site series, and clear
clustering among others. The RDA cluster analysis also allowed us to see which site
characteristics underpin those patterns, namely mean annual temperature and soil moisture
availability.

102

I expected that some group structure (Kent 2012) was present in the data, and anticipated that
similar stand types might group together according to site series in a hierarchical cluster
analysis. The overlap in site series between clusters observed using both methods
(hierarchical cluster analysis and the RDA cluster analysis) indicates that it might not be
possible to create distinct categories that can be used to predict trajectories of understory
development following a WHL outbreak. It also confirms that stands recovering from a
WHL defoliation in the ITR are part o f a continuous plant community, with stands
distinguished by subtle site differences. This conclusion is in agreement with the
interpretation of many scientists who believe that the classification of vegetation samples can
produce artificial and misleading results, since plant communities exist as a continuum
(McIntosh 1967, Kent 2012) and do not necessarily fall naturally into distinct groups. More
study is needed to identify the other factors responsible for discriminating recovery patterns
in defoliated stands.

103

104

c)

Figure 25. Forest stand types representative o f cluster analysis groupings: group A) an ICH
vk2-01 stand with a prominent Oplopanax horridus understory; group B) an ICHwk3-05b
stand with mixed shrub cover and mixed conifer regeneration; group C) an ICHwk3-01a
stand with little shrub cover and mixed conifer regeneration; and group D) an ICH wk3-07
stand with few understory shrubs or large herbs.

105

4.2.3.2 Plant community and outbreak severity

Results of the Kruskal-Wallis test are taken to mean that the cumulative damage
classification rating (i.e., local WHL outbreak severity) increased with RDA Axis-2 scores.
In other words, the constrained community was related to cumulative damage classification.
When this result is used to interpret the RDA biplot, the four competitive shrub species
appear to be organized along a severity gradient, with Rubus idaeus having the highest
relative abundance in stands that were most severely defoliated and Oplopanax horridus
having the highest relative abundance in less severely defoliated stands. When this
suggestion is subjected to an attendant correlation analysis, however, only non-significant
relationships are found that mostly contradict the RDA axis scores. Rubus idaeus and R.
parviflorus and Oplopanax horridus all show higher abundances in stands that were less
severely defoliated (rs=-0.04, p=0.715; rs=-0.171, p=0.180; rs=-0.01, p=0.902). Only
Sambucus racemosa correlation results are more consistent with their arrangement in the
RDA biplot; that is, Sambucus appears to increase with defoliation severity (rs=0.162,
p=0.203). There are two possible explanations for these results: the RDA Axis-2 is an
imperfect proxy for defoliation severity or the cumulative damage classification ratings are
too crude to detect subtle differences between plant community responses to defoliation
severity.

4.2.4 Regeneration

Tree regeneration (both total and well-spaced) in defoliated stands is a negative function of

106

shrub cover. Tsuga heterophylla and Abies lasiocarpa have a particularly negative response
to higher shrub cover, indicating lower shade tolerance or poorer competition tolerance than
Thuja plicata. There also may be other elements o f competition that these species are
deficient in when compared to Thuja. Picea engelmanii x glauca is said to have lower shade
tolerance than Tsuga heterophylla and Abies lasiocarpa (Klinka et al. 2000), so the reason
for its placement with shade tolerant Thuja plicata is unclear. That larger size classes o f all
species (seedlings and saplings) respond negatively to higher shrub cover may indicate that
there is enough light available for germination but not enough for newly established
seedlings to grow well.

In general these findings confirm that twenty years following a WHL event, the patterns
found by Hoggett and Negrave (2001) seven years after the outbreak continue to persist.
Increased canopy mortality (lower canopy density) is associated with increasing cover of
understory shrub species, which in turn was negatively correlated with tree regeneration
density (Hoggett and Negrave 2001). The degree o f canopy mortality is somewhat a function
o f tree layer composition, but also may reflect proximity to outbreak centers. The trajectory
of forest recovery subsequently depends on terrain position and site variables in addition to
the degree of canopy opening.

4.2.4.1 Regeneration niche

Autotrophic plants all need light, carbon dioxide, water and the same mineral nutrients
(Grubb 1977) but differences in regenerating tree species composition and abundance among

107

defoliated stands suggest additional requirements for germination and establishment by
different species. Species in a community often differ from one another during the
regeneration phase o f their life cycle (Silvertown 2004). These differences are described by
the ‘regeneration niche’ o f each species, which is the component o f the niche that is
concerned with processes such as seed production and germination and by which one mature
individual is replaced by another (Gmbb 1977).

The results o f the regeneration niche RDA analysis suggest that site index and soil moisture
regime are responsible for variation in the composition o f the regenerating tree assemblage
among defoliated stands of the ITR. There are likely other nuanced differences among stands
that were not captured by this study but that matter to the germination, successful
establishment, growth and survival o f trees. Small-scale heterogeneity within stands affects
patterns o f regeneration (Hannon and Franklin 1989, Nakashizuka 1989, Hulme 1996). In
particular, litter (forest floor) depth has been found to create differential effects on the
regeneration success of tree species with different seed mass (Christie and Armesto 2003).
Soil organisms are of special interest in stands that have been subject to foliar herbivory
(Bardgett et al. 1998, Hunter 2001) and the presence o f root mycorrhizae may influence the
growth of seedlings (Wright 1963, Clark 1963, Wallander 1995). Herbivory from browsing
ungulates may also create differential success of regenerating species in recovering stands
(Hulme 1996).

108

4.2.4.2 Regeneration substrate

Conifer regeneration is rarely randomly distributed on the forest floor (Maguire and Forman
1983); young conifer seedlings are associated with particular microsites (Simard et al. 1998).
In stands o f the ITR that have not experienced a recent WHL outbreak, coarse woody debris
(CWD) is common (Stevenson and Keisker 2002, Stevenson et al. 2006). In defoliated stands
of the ITR, CWD is the most favourable substrate type. Coarse woody debris may be helping
regenerating trees avoid competition with species on the forest floor (Harmon and Franklin
1989, Daniels and Gray 2006) and could also function to hide or block access to regenerating
trees from browsing ungulates (Ripple and Larsen 2001).

Since the total availability of CWD was not measured, this study cannot conclude that this
substrate is comparatively more important to regenerating trees than either forest floor or
mineral soil. Although all regenerating tree species in the ITR use CWD as a substrate, Thuja
plicata seems to increase its use of CWD as shrub cover increases, appearing to take
advantage of this substrate in defoliated stands. This effect is consistent with the results of
the plant community analysis showing Thuja plicata in stands with higher competitive shrub
cover; the presence of CWD in these stands may have facilitated the germination and growth
o f this tree species. This substrate preference o f Thuja plicata in shrubby stands has been
seen in Thuja occidentalis L. as well (Simard et al. 1998).

109

5.0 Conclusions and synthesis

Forest response to western hemlock looper (WHL) outbreaks in the inland temperate
rainforest (ITR) can be quite variable. Overstory recovery appears to be most pronounced on
lower productivity sites and in stands with a more severe outbreak history. Forest
regeneration is strongly constrained by shrub cover, which is promoted by canopy openness
and soil nutrients, both o f which can be accentuated by defoliators, as well as being more
pronounced on steep slopes. Immediately post-disturbance, shrubs are likely able to react
sooner or more strongly than trees to nutrient inputs to the forest floor, particularly in stands
with a higher site index, and flourish before regeneration is able to establish or release from a
seed bank. Twenty years following a WHL outbreak, the influence o f nutrient inputs to the
forest floor from the defoliation event may be less important than the effect o f canopy
openness and stand site index.

Regenerating Tsuga heterophylla and Abies lasiocarpa have a particularly negative response
to higher shrub cover, indicating lower shade tolerance or poorer competition tolerance than
Thuja plicata, which appears to tolerate stands with higher shrub cover. The reason for this
species’ apparent tolerance for stands with higher shrub cover might also be its proficient use
o f CWD, the substrate most commonly used in defoliated stands by all regenerating tree
species. Thuja plicata appears to take advantage o f CWD in order to escape competition with
shrubs, but the relative importance of this substrate to Thuja plicata or to other species
cannot be determined without considering the availability o f CWD in all stands, which was
not measured in this study. In terms of site characteristics, site index and soil moisture

110

availability are primarily responsible for the niche differentiation among regenerating tree
species in defoliated stands o f the ITR.

Ill

6.0 Management of defoliated stands

The WHL is a natural part of the ITR and plays a critical role in shaping the spatial patterns
and functioning of this ecosystem (Rhemtulla et al. 2001, Stevenson et al. 2011). Any
management that takes place in the ITR needs to consider maintaining ecological patterns
and processes, including those created by this defoliating insect. In anticipation of more
suitable conditions for outbreaking insects as climate change proceeds (Haughian et al.
2012), more old forest needs to be protected from harvest in order to mitigate the effects of
the WHL. Sound landscape-level management decisions can enhance the resilience of the
ITR to climate change (Stevenson et al. 2011), species loss and other human-created impacts.

In addition to landscape-level management, current industrial timber harvesting practices
need to be reconsidered, as some studies suggest that they are contributing to an increase in
the severity and extent of insect outbreaks (Rhemtulla et al. 2001). In eastern Canada and
Oregon, fire suppression, clearcutting and the selective harvesting of non-susceptible species
have been found to be primary contributors to an increase in the severity, frequency and area
affected by the spruce budworm (Lewis and Lindgren 2000). In the ITR, it has been
postulated that large clearcuts in the vicinity of a WHL infestation near Crescent Spur in
2000 may have helped to trigger an outbreak by creating drying winds that increased water
stress in adjacent stands (Rhemtulla et al. 2001).

Stand-scale management o f defoliated stands for tree production should consider differences
in site index, site series, canopy density and annual heat:moisture index among stands.

112

Stands with a higher site index are likely to require much more shrub control than stands
with a lower site index; it appears that a multi-species average site index o f 16 m @ 50 yrs is
optimal for tree regeneration, although this is a weak relationship. Stands with site indices
greater than 16 would require silvicultural intervention to accelerate reforestation, although
Thuja plicata would probably become established without treatment. Stands suitable for
regeneration by natural ingress generally have canopy densities over 75%, while stands with
an annual heat:moisture index 16 or greater have optimal warmth for the establishment and
growth of young trees.

In this study, site series with the lowest site indices were ICHvk2-03, ICHvk2-06 and
ICHwk3-07. Stands falling into these site series tended to have higher Tsuga heterophylla
regeneration with some Abies lasiocarpa and Picea engelmanii x glauca cover and would
likely require minimal shrub management. For Thuja plicata promotion, stands within the
ICHwk3-05 site series are best, and should require little shrub management. In contrast,
ICHvk2-01 site series stands also have high cover o f Thuja plicata regeneration, but also
have the highest site index (19.8), thereby making shrubs more abundant and their control
more important where tree establishment is required.

The following management guidelines for stand types identified in the hierarchical cluster
analysis (Figure 12) will encourage forest regeneration in the understory and recovery in the
overstory. Stands found in Group A (Figure 12) will have the poorest overstory recovery
potential, as they tend to have the highest site indices (19.8 for vk2-01 stands and 19.6 for
wk3-05 stands). These stands have heavy Oplopanax cover but also tend to have robust

113

densities of Thuja in the understory. These stands should be left for natural recovery. Manual
mechanical release around individual seedlings may accelerate redcedar vigour and growth.

Group B stands (Figure 12) are made up of stands that fall in to site series wk3-05, and so do
not have the strongest overstory potential. They do, however, have mixed conifer
regeneration, making them a more resilient stand type than Group A stands. B-type stands
are likely able to recover naturally, and inherently have more flexibility (diversity) for coping
with a changing climate than stands with fewer regenerating tree species.

Group C stands (Figure 12), made up mostly o f wk3-01 and vk2-06 site series, have an
intermediate level of overstory recovery potential, due to intermediate site indices (19.1 and
17.4 respectively) and mixed conifer regeneration (in particular Tsuga). One stand o f the site
series vk2-03, which has the lowest site index (12.6) and thus the highest overstory recovery
potential, fell in the C-type stand grouping because of the similarity of its’ understory to
wk3-01 and vk2-06 stands. Little shrub cover in C-type stands makes them ideal for
underplanting, with light unlikely to be a limiting factor in most stands of this type. C-type
stands, like B-type stands appear to be resilient due to the presence of several regenerating
tree species in the understory.

In Group D stands (Figure 12), overstory recovery potential is the greatest (save for vk2-03
stands), with wk3-07 stands having a relatively low site index (16). The understory o f D-type
stands is moss-dominated, with little tree regeneration or shrub cover, most likely because
the thick moss carpet inhibits natural regeneration from seed. If the overstory is mostly alive,

114

it should be left to recover naturally by an infilling crown. If the overstory is mostly dead,
Group D stands might benefit from underplanting and standard silvicultural stand renewal
practices, which may include salvage logging, mechanical site preparation, and planting.
Information about the environmental impacts, implications and limitations of salvage logging
can be found in Lindenmayer et al. 2008.

In summary, management of WHL-affected stands in the ITR should be undertaken
cautiously. A full range of responses to outbreak severity and site factors were found across
stands, and subtle differences between stands classified as similar mean that it is difficult to
predict with high precision the outcome o f management activities. The ITR is much more
robust and resilient to disturbances such as the WHL than expected, so management
interventions should rarely be considered “necessary” and require careful planning and
restraint where they are carried out.

115

References

Alfaro, R. I., G.A. Van Sickle, A.J. Thomson and E. Wegwitz. 1982. Tree mortality and
radial growth losses caused by the western spruce budworm in a Douglas-fir stand in British
Columbia. Canadian Journal of Forest Research 12: 780-787.
Alfaro, R. I., S. Taylor, G. Brown and E. Wegwitz. 1999. Tree mortality caused by the
western hemlock looper in landscapes o f central British Columbia. Forest Ecology and
Management 124: 285-291.
Arocena, L. and P. Sanbom. 1999. Mineralogy and genesis o f selected soils and their
implications for forest management in central and northeastern British Columbia. Canadian
Journal o f Soil Science 79: 571-592.
Assmann, E. 1970. The principles of forest yield study. Pergamon, New York.
Ayres, M.P. and M. J. Lombardero. 2000. Assessing the consequences o f global change for
forest disturbance from herbivores and pathogens. The Science o f the Total Environment
262: 263-286.
Banner, A., W. MacKenzie, S., Haeussler, S. Thomson, J. Pojar, and R. Trowbridge. 1993. A
field guide to site identification and interpretation for the Prince Rupert Forest Region. Land
Management Handbook 26. B.C. Ministry o f Forests and Range. Victoria, B.C., Canada.
Barbour, M.G., J.H. Burk, W.D. Pitts, F.S. Gilliam, M.W. Schwartz. 1999. Terrestrial plant
ecology, 3rd Edition. Benjamin-Cummings, Menlo Park, California.
Bardgett, R., D.A. Wardle and G.W. Yeates. 1998. Linking above-ground and below-ground
interactions: how plant responses to foliar herbivory influence soil organisms. Soil Biology
and Biochemistry 30: 1867-1878.
Bardgett, R. and K. Chan. 1999. Experimental evidence that soil fauna enhance nutrient
mineralization and plant nutrient uptake in montane grassland ecosystems. Soil Biology and
Biochemistry 31: 1007-1014.
Bardgett, R. D. and D.A. Wardle. 2010. Aboveground-belowground linkages: biotic
interactions, ecosystem processes and global change. Oxford University Press, Oxford, U.K.
Beaudry, L.J., R.A. Coupe, C. DeLong and J. Pojar. 1999. Plant indicator guide for Northern
British Columbia: Boreal, Sub-boreal and Subalpine Biogeoclimatic Zones; SBS, BWBS,
SBPS and Northern ESSF. Land Management Handbook 46. Research Branch, B.C. Ministry
of Forests and Range. Victoria, B.C.
Belsky, A. J. 1986. Does herbivory benefit plants? A review o f the evidence. American
Naturalist 127: 870-92.

Benson, S. and D.S. Coxson. 2002. Lichen colonization and gap structure in wet-temperate
rainforests o f northern interior British Columbia. The Bryologist 105: 673-692.
Berryman, A. A. and R.W. Stark. 1985. Assessing the risk of forest insect outbreaks. Journal
o f Applies Entomology 99: 199-208.
Bhiry, N. and L. Filion. 1996. Mid-Holocene hemlock decline in eastern North America
linked with phytophagous insect activity. Quaternary Research 45: 312-320.
Blais, J.R., R.M. Prentice, W.L. Sippell and D.R. Wallace. 1955. Effects o f weather on the
forest tent caterpillar Malacosoma disstria Hbn., in central Canada in the spring o f 1953.
Canadian Entomologist 87: 1-8.
Blanchet, F.G., P. Legendre and D. Borcard. 2008. Forward selection of explanatory
variables. Ecology 89: 2623-2632.
Blatt, S.E., J.A. Janmaat and R. Harmsen. 2001. Modelling succession to include a herbivore
effect. Ecological Modelling 139 : 123-136.
Borcard, D., F. Gillet and P. Legendre. 2011. Numerical ecology with R. Springer, New
York.
Bordeleau, C. 2000. Insectes et maladies des arbres au Quebec. Ministere des Ressources
naturelles du Quebec, Direction de la conservation des forets, Division des releves, Quebec,
Quebec.
Borecky, N. and I.S. Otvos. 2001. Coarse-scale hazard rating of western hemlock looper in
British Columbia. In Proceedings of the integrated management and dynamics o f forest
defoliating insects, August 15-19, 1999, Victoria B.C., Canada. Gen. Tech. Rep. NE-277
Edited by A.M. Liebhold, M.L. McManus, I.S. Otvos and S.L.C. Fosbroke. USDA Forest
Service, Northeastern Research Station, Newtown Square, Pennsylvania, pp. 6-15.
Buffam, P.E. 1965. Carbaryl, phosphamidon, and DDT tests on the western hemlock looper
in Washington. Journal of Economic Entomology 55: 1006-1008.
Campbell, R.W. and R.J. Sloan. 1977. Forest stand responses to defoliation by the gypsy
moth. Forest Science Monographs 19: 1-19.
Canadian Forest Service (CFS) and B.C. Ministry of Forests (MoF). 2000. Forest health
aerial overview survey standards for British Columbia. Pacific Forestry Centre, Canadian
Forest Service, and B.C. Ministry of Forests, Forest Practices Branch. Victoria, B.C.
Carolin, V.M. and C.G. Thompson. 1967. Field testing of Bacillus thuringiensis for control
o f western hemlock looper, Research Paper 38. U.S. Department o f Agriculture, Forest
Service, Pacific Northwest Forest and Range Experiment Station, Portland, Oregon. 24 p.

Carolin, V.M. and R.R. Lejeune. 1967. The western hemlock looper, Lambdina fiscellaria
lugubrosa (Hulst). In Important forest insect and diseases o f mutual concern to Canada, the
United States and Mexico. Edited by A.G. Davidson and R.M. Prentice. Canadian
Department of Rural Development Ottawa, Ontario, pp. 123-124.
Carpenter, S.R., Kitchell, J.F. and Hodgson, J.R. 1985. Cascading trophic interactions and
lake productivity. BioScience 35: 634-639.
Carroll, W.J. 1956. History o f the hemlock looper, Lambdina fiscellaria fiscellaria
(Guen.), (Lepidoptera : Geometridae) in Newfoundland, and notes on its biology. The
Canadian Entomologist 88: 587-599.
Carroll, A.L. 1999. Physiological adaptation to temporal variation in conifer foliage by a
caterpillar. The Canadian Entomologist 131: 659-669.
Carter, N. and L. Hartling. 1992. Hemlock looper in New Brunswick. In Proceedings o f the
Annual Forest Pest Control Forum, November 17-19, Canadian Forest Service, Ottawa,
Ontario, pp. 161-164.
Charles, J.F., A. Delecluse and C. Nielsen-LaRoux. 2000. Entomopathogenic bacteria: from
laboratory to field application. Kluwer Academic Publishers, Netherlands.
Chew, R.M. 1974. Consumers as regulators of ecosystems: an alternative to energetics. Ohio
Journal of Science 74: 359-370.
Chouinard, A. and L. Filion. 2005. Impact o f introduced white-tailed deer and native insect
defoliators on the density and growth of conifer saplings on Anticosti Island, Quebec.
Ecoscience 12: 506-518.
Christie, D. and J. Armesto. 2003. Regeneration microsites and tree species coexistence
in temperate rain forests of Chiloe Island, Chile. Journal of Ecology 91: 776-784.
Christenson L.M., G.M. Lovett, M.J. Mitchell and P.M. Groffman. 2002. The fate o f
nitrogen in gypsy moth frass deposited to an oak forest floor. Oecologia 131: 444-452.
Clark, B. 1963. Endotrophic mycorrhizae influence yellow poplar seedling growth. Science
140: 1220-1221.
Clarke, L.J. and G.C. Carew. 1987. Forest insect and disease conditions in Newfoundland
and Labrador in 1986. Information Report N-X-259. Newfoundland Forestry Centre,
Canadian Forest Service, St. John’s, Newfoundland.
Conder, N. 2006. Western hemlock looper trapping study. Unpublished report. Pacific
Forestry Centre, Canadian Forest Service, Victoria, B.C.

Connell, J.H. and R.O. Slatyer. 1977. Mechanisms o f succession in natural communities and
their role in community stability and organization. The American Naturalist 111: 1119-1144.
Coopersmith, D., S. Taylor, M. Jull, B. Rogers, K. Deschamps andN , Humphreys. 1998. A
comparison of the growth o f planted white spruce seedlings in small openings and beneath
hemlock looper attacked cedar/hemlock stands in the Robson Valley, Prince George Forest
Region. Draft manuscript, Prince George Forest Region, B.C. Ministry o f Forests, Prince
George, B.C. 16 p.
Coupe, R. A.C. Stewart and B.M. Wikeem. 1991. Engelmann Spruce - Subalpine Fir Zone.
In Ecosystems of British Columbia. Edited by D. Meidinger and J. Pojar. Special Report No.
6. B.C. Ministry of Forests, Victoria, pp. 223-236.
Daniels, L.D., J. Dobry, K. Klinka and M.C. Feller. 1997. Determining year o f death o f logs
and snags of Thuja plicata in southwestern coastal British Columbia. Canadian Journal of
Forest Research 27: 1132-1141.
Daniels, L.D. and R.W. Gray. 2006. Disturbance regimes in coastal British Columbia. B.C.
Journal o f Ecosystem Management 7: 44-56.
Davidson, A.G. and R.M. Prentice. 1967. Important forest insects and diseases o f mutual
concern to Canada, the United States and Mexico. Publ. 1180. Canadian Department for
Rural Development, Ottawa, Ontario.
Davis, M. B. 1981. Outbreaks of forest pathogens in Quaternary history. In Proceedings of
the IV International Palynological Conference (1976-77). Birbal Sahni Institute of
Paleobotany, Lucknow, India, pp. 216-227.
Day, M. 2013. Influence of commercial thinning on recovery from defoliation in spruce-fir
forests (unpublished). Available from: http://nsrcforest.org/project/influence-commercialthinning-recovery-defoliation-spruce-fir-forests [viewed September 30, 2013]
DeLong, C. 2003. A field guide to site identification and interpretation for the southeast
portion of the Prince George Forest Region. Land Management Handbook 51. Research
Branch, B.C. Ministry o f Forests and Range. Victoria, B.C., Canada.
DeLong, C. A. 2007. Draft field guide to site identification and interpretation for the Rocky
Mountain Trench portion o f the Prince George Forest Region. Land Management Handbook
15, unpublished. Research Branch, B.C. Ministry of Forests and Range. Victoria, B.C.,
Canada.
Delong, S. C., H. Griesbauer and C. R. Nitschke. 201 la. FFESC Project B5: risk analysis
and decision support tool final report Appendix 1. Assessing the risk of drought in British
Columbia forests using a stand-level water balance approach. Available from:
http://www.for.gov.bc.ca/hfp/silviculture/TSS/Drought_Risk.html [viewed April 2, 2013]

DeLong, S. C., H. Griesbauer and C. R. Nitschke. 201 lb. FFESC Project B5: risk analysis
and decision support tool final report (Stand level frost hazard assessment and management
tool). Available from:
http://www.for.gov.bc.ca/hfp/silviculture/TSS/frost_hazard.html [viewed April 26, 2013]
Deschamps, K., N. Humphreys and T. LaPointe, 1994. ICH hemlock looper Sw
underplanting trial: working plan. Pacific Forestry Centre, Canadian Forest Servicce, Prince
George, B.C. 7 p.
Dobesberger, E J. 1989. A sequential decision plan for the management o f the eastern
hemlock looper, Lambdina fiscellaria fiscellaria (Lepidoptera: Geometridae) in
Newfoundland. Canadian Journal o f Forest Research 19: 916-991.
Duncan, R.W. 2006. Conifer defoliators of British Columbia. Pacific Forestry Centre,
Canadian Forest Service, Victoria, B.C. 359 p.
Dyer, M.I. and H.H. Shugart. 1992. Multi-level interactions arising from herbivory: a
simulation analysis of deciduous forests utilizing FORET. Ecological Applications 2: 376386.
Dyer, M.I., C.L. Turner and T.R. Seastedt. 1993. Herbivory and its consequences. Ecological
Applications 3: 10-16.
Engelhardt, N.T. 1957. Pathological deterioration o f looper-killed western hemlock on
southern Vancouver Island. Forest Science 3: 125-136.
Erickson, R.D. 1984. Western hemlock looper. FPL 21. Pacific Forestry Centre, Canadian
Forest Service, Victoria, B.C.. 4. p.
Evenden, J. C. 1940. Effects o f defoliation by the pine butterfly upon ponderosa pine.
Journal o f Forestry 38: 949-955.
Foster, D.R. 2000. Hemlock’s future in the context of its history: an ecological perspective.
In Proceedings of the Symposium on sustainable management o f hemlock ecosystems in
eastern North America. Edited by K.A. McManus, K.S. Shields and D.R. Souto. General
Technical Report NE-267. United States Department of Agriculture, Forest Service,
Northeastern Forest Experiment Station, Newtown Square, Pennsylvania, pp. 1-4.
Franklin, J.F., H.H. Shugart and M.E. Harmon. 1987. Tree death as an ecological process.
BioScience 37: 550-556.
Franklin, J.F., D. Lindenmayer, J.A. MacMahon, A. McKee, J. Magnuson, D.A. Perry, R.
Waide and D. Foster. 2000. Threads o f continuity: Ecosystem disturbances, biological
legacies and ecosystem recovery. Conservation Biology in Practice 1:8-16.

Frelich, L.E. and P.B. Reich. 1999. Neighborhood effects, disturbance severity, and
community stability in forests. Ecosystems 2: 151-166.
Frost, C.J. and M.D. Hunter. 2004. Insect canopy herbivory and frass deposition affect soil
nutrient dynamics and export in oak mesocosms. Ecology 85: 3335-3347.
Fuhlendorf, S.D. and F.E. Smeins. 1997. Long-term vegetation dynamics mediated by
herbivores, weather and fir in a Juniperus-Quercus savanna. Journal of Vegetation Science 8:
819-828.
Fuller, J. L. 1998. Ecological impact o f the mid-Holocene hemlock decline in southern
Ontario, Canada. Ecology 79: 2337-2351.
Fumiss, R.L. and V.M. Carolin. 1977. Western forest insects. Pub. No. 1339. United States
Department of Agriculture Forest Service, Washington, D.C.
Fuxa, J. R., R. Ayyappath and R.A. Goyer. 1998. Pathogens and microbial control of North
America forest insect pests. Forest Health Enterprise Team FHTET-97-27. Louisiana State
University Agricultural Center, Baton Rouge, Louisiana.
Gabriel, C. E. and L. Kellman 2011. Examining moisture and temperature sensitivity o f soil
organic matter decomposition in a temperate coniferous forest soil. Biogeosciences
Discussions 8:1369-1409.
Google Earth, 2012. V 6.2.2.6613. British Columbia, Canada 53° 38’ 22.00”S, 120° 50’
32.84”W, CNES/SPOT image. Available from: http://www.earth.google.com [accessed
November 15, 2012],
Goward, T. and A. Arsenault 2000. Inland old growth rainforests: safe havens for rare
lichens? In Proceedings of a Conference on the Biology and Management o f Species and
Habitats at Risk, Kamloops, B.C., 15-19 Feb., 1999. Edited by L. Darling, B.C. Ministry of
Environment, Lands and Parks, Victoria, B.C. and University College of the Cariboo,
Kamloops, B.C. pp. 759-766.
Grace, J.R. 1986. The influence o f gypsy moth on the composition and nutrient content o f
litter fall in a Pennsylvania oak forest. Forest Science 32: 855-870.
Graham, S.A. 1956. Ecology o f forest insects. Annual Review o f Entomology 1: 261-280.
Green, R. N., R.L. Trowbridge and K. Klinka. 1993. Towards a taxonomic classification o f
humus forms. Forest Science Monographs 29: 1-49.
Grubb, P.J. 1977. The maintenance o f species richness in plant communities: the importance
o f the regeneration niche. Biological Reviews 52: 107-145.

Haughian, S.R., P.J. Burton, S.W. Taylor and C.L. Curry. 2012. Expected effects o f climate
change on forest disturbance regimes in British Columbia. British Columbia Journal of
Ecosystems and Management 13: 1-24.
Haeussler, S., D. Coates and J. Mather. 1990. Autecology o f common plants in British
Columbia: a literature review. Forest Resource Development Agreement Report 158.
Canadian Forest Service and British Columbia Ministry of Forests, Victoria, B.C.
Haeussler, S. Personal communication. July 12, 2013.
Harcombe, A., B. Pendergast, B. Petch and D. Janz. 1983. Elk habitat management: Salmon
River Valley. Working Report. No. 1. British Columbia Ministry o f Environment, Victoria,
B.C.
Harmon, M. E. and J.F. Franklin. 1989. Tree seedlings on logs in Picea-Tsuga forests of
Oregon and Washington. Ecology 70: 48-59.
Harris, J.W.E. and A.F. Dawson. 1979. Evaluation o f forest pest damage survey techniques
in British Columbia. Pacific Forestry Centre, Canadian Forest Service, Victoria, B.C. 22 p.
Harris, J.W.E., A.F. Dawson and R.G. Brown. 1982. The western hemlock looper in British
Columbia 1911-1980. Report no. BC-X-234, Pacific Forestry Centre, Canadian Forest
Service, Victoria, B.C. 18 p.
Hartley, S. E. and T.H. Jones. 2004. Insect herbivores, nutrient cycling and plant
productivity. In Insects and ecosystem function. Edited by W.W. Weisser and E. Siemann.
Springer-Verlag, New York. pp. 27-52.
Hartling L, P.M. MacNutt and N. Carter. 1991. Hemlock looper in New Brunswick: notes on
biology and survey methods. New Brunswick Department o f Natural Resources and Energy,
Fredericton, New Brunswick.
Hebert, C., L. Jobin, R. Berthiaume, C. Coulombe and A. Dupont. 2001. Changes in pupal
distribution o f the hemlock looper (Lepidoptera: Geometridae) through an outbreak.
Phytoprotection 82: 57-63.
Hicke, J.A., C.D. Allen, A.R. Desai, M.C. Dietze, R.J. Hall, E.H. Hogg, D.M. Kashian, D.
Moore, K.F. Raffa, R.N. Sturrock and J. Vogelmann. 2012. Effects o f biotic disturbances on
forest carbon cycling in the United States and Canada. Global Change Biology 18: 7-34.
Hoggett, A. 2000. Western hemlock looper and forest disturbance in the ICH wk3 o f the
Robson Valley - Stage 2: The effects o f western hemlock looper - report and silviculture
recommendations. Research Note. Faculty of Forestry, University of British Columbia,
Vancouver, B.C. 93 p.

Hoggett, A. and R. Negrave. 2001. Western hemlock looper forest disturbance in the
ICHwk3 of the Robson Valley: Stage 3: Effect o f western hemlock looper and disturbance
classification - progress report and ecosystem management recommendations, Unpublished
report for the Robson Valley EFMPP, University o f British Columbia Department o f Forest
Sciences, Vancouver, B.C.
Hoggett, A. and R. Negrave. 2002. Western hemlock looper and disturbance in the ICHwk3
forests of the Robson Valley. Draft Research Note. B.C. Ministry o f Forests, Prince George,
B.C.
Hollinger, D.Y. 1986. Herbivory and the cycling o f nitrogen and phosphorous in isolated
California oak trees. Oecologia 70: 291-297.
Hopping, G.R. 1934. An account of the western hemlock looper Ellopia somniaria Hulst, on
conifers in British Columbia. Scientific Agriculture 15: 12-28.
Hudak, J., G. LaFlamme and J.P. Meades. 1978. Deterioration o f balsam fir damaged by the
eastern hemlock looper in Newfoundland. Information Report N-X-157. Newfoundland
Forest Research Centre, Canadian Forest Service, St. John’s, Newfoundland.
Hulme P.E. 1996. Herbivory, plant regeneration and species coexistence. Journal o f Ecology
84: 609-615.
Hunter, M.D. 2001. Insect population dynamics meets ecosystem ecology: effects of
herbivory on soil nutrient dynamics. Agricultural Forest Entomology 3: 77-84.
Iqbal, J. and D. MacLean. 2010. Estimating cumulative defoliation of balsam fir from
hemlock looper and balsam fir sawfly using aerial defoliation survey in western
Newfoundland, Canada. Forest Ecology and Management 259: 591-597.
Iqbal, J., D.A. MacLean and J.A. Kershaw. 2011. Impacts o f hemlock looper defoliation on
growth and survival of balsam fir, black spruce and white birch in Newfoundland, Canada.
Forest Ecology and Management 261: 1106-1114.
Irwin, L.L. and J.M. Peek 1979. Shrub production and biomass trends following five logging
treatments within the cedar-hemlock zone o f northern Idaho. Forest Science 25: 415-429.
Lantz, T.C. and J.A. Antos. 2002. Clonal expansion in the deciduous understory shrub,
devil's club (Oplopanax horridus; Araliaceae). Canadian Journal o f Botany 80: 1052-1062.
Jardine, A.K. 1969. Western hemlock looper (Lambdina Jiscillaria lugubrosa) in British
Columbia. Forest Pest Leaflet no. 21. Pacific Forestry Centre, Canadian Forest Service,
Victoria, B.C. 5 p.
Jobin, L. 1980. L’arpenteuse de la pruche, une insecte insulaire. Racine, 2: 5-7.

Jobin, L. and R. Desaulniers. 1981. Results o f aerial spraying in 1972 and 1973 to control the
eastern hemlock looper {Lambdina fiscellaria fiscellaria (Guen.)) on Anticosti Island.
Information Report LAU-X-49E. Laurentian Forest Research Centre, Canadian Forest
Service, Environment Canada, Sainte-Foy, Quebec.
Johnson, N.E., K.R. Shea and R.L. Johnsey. 1970. Mortality and deterioration o f looperkilled hemlock in western Washington. Journal of Forestry 68: 162-163.
Kessler, A. and I.T. Baldwin. 2002. Plant response to insect herbivory: the emerging
molecular analysis. Annual Review of Plant Biology 53: 299-328.
Kent, M. 2012. Vegetation description and data analysis: a practical approach. John Wiley,
New York.
Ketcheson, M.V., T.F. Braumandl, D. Meidinger, G. Utzig, D.A. Demarchi and B.M.
Wikeem. 1991. Interior Cedar - Hemlock Zone. In Ecosystems o f British Columbia. Special
Report No. 6. Edited by D. Meidinger and J. Pojar. B.C. Ministry o f Forests, Victoria, pp.
167-181.
Kinghom, J.M. 1954. The influence o f stand composition on the mortality o f various
conifers, caused by defoliation by the western hemlock looper {Ellopia fiscellaria
lugubrosa (Hulst)) on Vancouver Island, British Columbia. Forestry Chronicle 30: 380-400.
Klinka, K., V.J. Krajina, A. Ceska and A.M. Scagel. 1989. Indicator plants o f coastal British
Columbia. University o f British Columbia Press, Vancouver, B.C.
Klinka, K., J. Worrall, L. Skoda and P. Varga. 2000. The distribution and synopsis o f
ecological and silvical characteristics of tree species o f British Columbia’s forests. Canadian
Cartographies Ltd., Coquitlam, B.C.
Knight, D. H. 1987. Parasites, lightning, and the vegetation mosaic in wilderness landscapes.
In Landscape heterogeneity and disturbance. Edited by M. G. Turner. Springer-Verlag, New
York. pp. 59-83.
Koot, H.P. 1994. Western hemlock looper, Forest Pest Leaflet 21. Forest Insect and Disease
Survey, Pacific Forestry Centre, Canadian Forest Service, Victoria, B.C.
Konchalski, C. Unpublished. Sensitivity o f Western Redcedar to Climate and Western
Hemlock Looper in British Columbia's Inland Temperate Rainforest M.Sc. thesis. University
o f Northern B.C., Prince George, B.C.
Kulman, H.M. 1971. Effects of insect defoliation on growth and mortality o f trees. Annual
Review of Entomology 16: 289-324.
Laursen, S. 1977. The effect of defoliation on the growth of Douglas-fir, Pseudotsuga
mensiezii. Directed study 502-01. B.C. Ministry of Forests Research Branch, Victoria, B.C.

Legendre, P. and E. D. Gallagher. 2001. Ecologically meaningful transformations o f
ordinations of species data. Oecologia 129: 271-280.
Legendre, P. and L. Legendre. 2012. Numerical ecology, 3rd English edition. Elsevier
Science BV, Amsterdam. 990 p.
Leps, J. and P. Smilauer. 2003. Multivariate analysis of ecological data using CANOCO.
Cambridge University Press, Cambridge, U.K.
Levin, D.B., A.M. Laitinen, T. Clarke, C.J. Lucarotti, B. Morin and I.S. Otvos. 1997.
Characterization of nuclear polyhedrosis viruses from three subspecies of Lambdina
fiscellaria. Journal of Invertebrate Pathology 69: 125-134.
Lindenmayer, D.B., P.J. Burton, and J.F. Franklin. 2008. Salvage logging and its ecological
consequences. Island Press, Washington, D.C. 227 p.
Lovett, G.M. and A.E. Ruesink. 1995. Carbon and nitrogen mineralization from
decomposing gypsy moth frass. Oecologia 104: 133-138.
Lovett, G.M., L.M. Christenson, P.M. Groffman, C.G. Jones, J.E. Hart and M.J. Mitchell.
2002. Insect defoliation and nitrogen cycling in forests. Bioscience 52: 335-341.
Lewis, K.J. and B.S. Lindgren. 2000. A conceptual model of biotic disturbance ecology in
the central interior of B.C.: How forest management can turn Dr. Jekyll into Mr. Hyde. The
Forestry Chronicle 76: 433-443.
MacDonald, A.M. 2013. Lichens in the inland rainforest: climate biomonitoring and
population structure of Lobaria pulmonaria (L.) Hoffm. M.Sc. thesis, University of Northern
B.C., Prince George, B.C.
MacLean, D.A. and P. Ebert. 1999. The impact of hemlock looper (Lambdina fiscellaria
fiscellaria (Guen.)) on balsam fir and spruce in New Brunswick, Canada. Forest Ecology and
Management 120: 77-87.
MacLean, D.A. and W.E. MacKinnon. 1996. Accuracy of aerial sketch-mapping estimates of
spruce budworm defoliation in New Brunswick. Canadian Journal of Forest Research 35:
2099-2108.
Magasi, L.P. 1990. Forest pest conditions in the Maritimes in the 1990s. Information Report.
M-X-178, Atlantic Forestry Centre, Canadian Forest Service, Fredericton, N.B. 68 p.
Magnussen, S. and R.I. Alfaro. 2012. Linking aerial survey data o f forest insect defoliation
and tree ring data to estimate forest level growth losses. Dendrochronologia 30: 287-294.
Maguire, D.A. and R.T.T. Forman. 1983. Herb cover effects on tree seedling patterns in a
mature hemlock-hardwood forest. Ecology 64: 1367-1380.

Martel, M. 1999. Analyse dendroecologique des effects de la defoliation par l’arpenteuse de
la pruche Lambdina fiscellaria (Guen) sur la population forestieres de file d’Anticosti,
Quebec. M.Sc. thesis, Universite Laval, Quebec, Quebec.
Maser, C., R.G. Anderson, K. Cromack Jr., J.T. Williams and R.E. Martin. 1979. Dead and
down woody material. In Wildlife habitats in managed forests: the Blue Mountains of
Oregon and Washington. Agricultural Handbook No. 553. Edited by J.W. Thomas. United
States Department o f Agriculture Forest Service, Washington, D.C. pp.78-79.
Mason, R. 1990. Analytical study of plant/environment interactions in thimbleberry and
devil’s club. M.Sc. thesis, University o f B.C.,Vancouver, B.C.
Mattson, W.J. and N.D. Addy. 1975. Phytophagous insects as regulators o f forest primary
production. Science 190: 515-522.
McCloskey, S.P.J. 2007. Western hemlock looper: a biological agent of disturbance in
coastal forests o f British Columbia. PhD dissertation. The University of British Columbia,
Vancouver, British Columbia.
McCloskey, S.P.J., L.D. Daniels and J.A. McLean. 2009. Potential impacts of climate change
on western hemlock looper outbreaks. Northwest Science 83: 225-238.
McCune, B. and J. B. Grace. 2002. Analysis o f ecological communities. MjM Software,
Gleneden Beach, Oregon. 304 p.
McCune, B. and D. Keon. 2002. Equations for potential annual direct incident radiation and
heat load. Journal o f Vegetation Science 13: 603-606.
McCune, B. and M.J. Mefford. 2011. Multivariate analysis of ecological data. PC-ORD.
Version 6.07. MjM Software, Gleneden Beach, Oregon.
McGuffm, W.C. 1987. Guide to the Geometridae of Canada (Lepidoptera). II. Subfamily
Ennominae. 4. Memoirs of the Entomological Society o f Canada 138: 64-65.
McIntosh, R. P. 1967. Continuum concept o f vegetation. Botanical Review 33: 130-187.
McNaughton, S.J. 1979. Grazing as an optimization process: grass-ungulate relationships in
the Serengeti. American Naturalist 113: 691-703.
Marquis, R.J. 1984. Leaf herbivores decrease fitness o f a tropical plant. Science 226: 537539.
Martineau, R. 1984. Insects harmful to forest trees. Publication No. 64-32/1984E. Canadian
Forestry Service, Ottawa.

Meidinger, D.V., A. McLeod, A. McKinnon, C. DeLong and G. Hope. 1988. A field guide
for identification and interpretation of ecosystems o f the Rocky Mountain Trench, Prince
George Forest Region. Land Management Handbook 15. Research Branch, B.C. Ministry of
Forests and Range. Victoria, B.C.
Ministry of Forests, Lands and Natural Resource Operations (MFLNRO). 2011. Site index
estimates by site series: report by region (2011 Approximation). Available from:
http://www.for.gov.bc.ca/hre/sibec/reports/sisuByRegion.pdf [viewed June 7, 2013]
Ministry of Forests and Range (MoFR). 2011. Site index estimates by site series (SIBEC).
Available from:
http://www.for.gov.bc.ca/hre/sibec/SIBEC_RDM_Section_2.htm [from June 8, 2013]
Mueggler, W.F. 1965. Ecology of serai shrub communities in the cedar-hemlock zone of
northern Idaho. Ecological Monographs 35: 165-185.
Myers, J.H. 1998. Synchrony in outbreaks o f forest Lepidoptera: a possible example o f the
Moran effect. Ecology 79: 1111-1117.
Nakashizuka, T. 1989. Role of uprooting in composition and dynamics o f an old-growth
forest in Japan. Ecology 70: 1273-1278.
Nealis, V.G. and R. Tumquist. 2010. Impact and recovery o f western hemlock following
disturbances by forestry and insect defoliation. Forest Ecology and Management 260: 699706.
Neely, D. 1970. Healing of wounds on trees. Journal o f the American Society of
Horticultural Science 95: 536-540.
Ostaff, D.P., J.H. Borden and R.F. Shepherd. 1974. Reproductive biology o f Lambdina
fiscellaria lugubrosa (Lepidoptera: Geometridae). Canadian Entomologist 106: 659-665.
Otvos, I.S., R.C. Clark and L.J. Clarke. 1971. The hemlock looper in Newfoundland: the
outbreak, 1966 to 1971; and aerial spraying, 1968 and 1969. Information Report N-X-68
Newfoundland Forest Research Centre, Canadian Forest Service, St. John’s, Newfoundland.
Otvos, I.S., L.J. Clarke and D.S. Durling. 1979. A history o f recorded eastern hemlock
looper outbreaks in Newfoundland. Information Report N-X-179. Newfoundland Forest
Research Centre, Canadian Forest Service, St. John’s, Newfoundland.
Otvos, I.S. 2004. Monitoring western hemlock looper populations in British Columbia. Final
Technical Report - Part C. Report to the Forest Innovation Investment (#R04-007CFS).
Pacific Forest Centre, Canadian Forest Service, Victoria, B.C.
Oswald, W.W. and D.R. Foster. 2011. A record of late-Holocene environmental change from
southern New England, USA. Quaternary Research 76: 314-318.

Paananen, D.M, R.F. Fowler and L.F. Wilson 1987. The aerial war against eastern region
forest insects, 1921-1986. Journal of Forest History 34: 173-186.
Parfett, N., I.S. Otvos and G.A. van Sickle. 1995. Historical western hemlock looper
outbreaks in British Columbia: input and analysis using a geographical information system.
Forest Resource Development Agency Report 235, Pacific Forest Centre, Canadian Forest
Service, Victoria, B.C., Copublished by the B.C. Ministry o f Forests.
Parks, C.G. 1993. The influence of induced host moisture stress on the growth and
development o f western spruce budworm and Armillaria ostoyea on grand fir seedlings.
Ph.D. Dissertation, Oregon State University, Corvallis, Oregon.
Peck, J.E. 2010. Multivariate analysis for community ecologists: step-by-step using PCORD. MjM Software Design, Gleneden Beach, Oregon. 162 p.
Pickett, S. T. A. and P.S. White, editors. 1985. The ecology of natural disturbance and patch
dynamics. Academic Press, Orlando, Florida.
Pickett, S.T.A., S.C. Collins and J.J. Armesto. 1987. Models, mechanisms, and pathways o f
succession. Botanical Reviews 53: 335-371.
Piene, H. 1989. Spruce budworm defoliation and growth loss in young balsam fir: recovery
of growth in spaced stands. Canadian Journal o f Forest Research 19:1616-1624.
Piene, H. and D.A. MacLean. 1999. Spruce budworm defoliation and growth loss in young
balsam fir: pattern of shoot, needle and foliage weight production over a nine-year outbreak
cycle. Forest Ecology and Management 123: 115-133.
Pojar, J., K. Klinka, and D.V. Meidinger. 1987. Biogeoclimatic ecosystem classification in
British Columbia. Forest Ecology and Management 22: 119-154.
Pojar, J., K. Klinka and D.A. Demarchi 1991. Coastal Western Hemlock Zone. In
Ecosystems of British Columbia. Special Report No. 6. Edited by D. Meidinger and J. Pojar.
B.C. Ministry o f Forests, Victoria, pp. 95-111.
Radies, D.N. 2008. Canopy macrolichen distribution in a very wet old growth forest
landscape of the upper Fraser River watershed. M.Sc. thesis. University o f Northern B.C.,
Prince George, B.C.
Radies, D.N., D.S. Coxson, C.J. Johnson and K. Konwicki. 2009. Predicting canopy
macrolichen diversity and abundance within old-growth inland temperate rainforests. Forest
Ecology and Management 259: 86-97.
Rafes, R.M. 1978. Biogeocoenological theory of the population dynamics o f phytophagous
forest insects. In Mathematical modeling in ecology. Edited by A.M. Molchanov. Translated
from Russian by P. Rubtzoff. Moscow, USSR. pp. 34-51.

Rao, R. 1964. The use and interpretation of principal component analysis in applied research.
Sankhya: The Indian Journal o f Statistics, Series A 26: 329-358.
Raske, A.G., R.J. West and A. Retnakaran. 1995. Hemlock looper, Lambdina fiscellaria. In
Forest insect pests in Canada. Edited by J.A. Armstrong and W.G.H. Ives. Canadian Forest
Service, Ottawa, pp 141-147.
Reichle, D. E., R. A. Goldstein, R. I. Van Hook, Jr. and G. J. Dodson. 1973. Analysis of
insect consumption in a forest canopy. Ecology: 54:1076-1084.
Reynolds, B.C., D.A. Crossley and M.D. Hunter. 2003. Response o f soil invertebrates to
forest canopy inputs along a productivity gradient. Pedobiologia 47: 127-139.
Reynolds, B.C. and M.D. Hunter. 2001. Responses of soil respiration, soil nutrients and litter
decomposition to inputs from canopy herbivores. Soil Biology and Biochemistry 33: 16411652.
Reynolds, B.C., M.D. Hunter and D.A. Crossley. 2000. Effects o f canopy herbivory on
nutrient cycling in a northern hardwood forest in western North Carolina. Selbyana 21: 7478.
Rhemtulla, J., S.H. Marshall and H. Hammond. 2001. Fraser headwaters proposed
conservation plan. Silva Forest Foundation, Slocan Park, B.C.
Ripple, W. and E.J. Larsen. 2001. The role o f postfire coarse woody debris in aspen
regeneration. Western Journal o f Applied Forestry 16: 61-64.
Risley, L.S. and D.A. Crossley Jr. 1988. Herbivore-caused greenfall in the southern
Appalachians. Ecology 69: 1118-1127.
Risley, L.S. and D.A. Crossley. 1993. Contribution of herbivore-caused greenfall to litterfall
nitrogen flux in several southern Appalachian forested watersheds. American Midland
Naturalist 129: 67-74.
Romme, W.H., D.H. Knight and J.H. Yavitt. 1986. Mountain pine beetle outbreaks in the
Rocky Mountains: regulators of primary productivity? American Naturalist 127: 484-494.
Roorbach, A.H. 1999. The ecology of devil’s club (Oplopanax horridum (J.E. Smith) Miq.)
in western Oregon. M.Sc. thesis, Department o f Forest Science, Oregon State University,
Corvallis, Oregon.
Russell, C.A., K.R. Kosola, E.A. Paul and G.P. Robertson. 2004. Nitrogen cycling in poplar
stands defoliated by insects. Biogeochemistry 68: 365-381.
Sanborn, P. Personal communication. May 8, 2013.

Sanborn, P., M. Geertsema, T. Jull and B. Hawkes. 2006. Soil and sedimentary charcoal
evidence for Holocene forest fires in an inland temperate rainforest, east-central British
Columbia, Canada. The Holocene 16: 415-427.
Schlesinger, W.H. 1991. Biogeochemistry: An analysis o f global change. Academic Press,
San Diego 443 p.
Schowalter, T.D. 1985. Adaptations of insects to disturbances. In The ecology of natural
disturbance and patch dynamics. Edited by S.T.A. Pickett and P.S. White. Academic Press,
New York. pp. 235-252.
Schowalter, T. D. 2000. Insect ecology: an ecosystem approach. Academic Press, San Diego,
California, pp. 322-341.
Schowalter, T.D. and M.D. Lowman. 1999. Insect herbivory in forests. In Ecosystems of
Disturbed Ground. Edited by L. R. Walker. Elsevier, Amsterdam, pp. 253-269.
Schowalter, T.D. and T.E. Sabin. 1991. Litter microarthropod responses to canopy herbivory,
season and decomposition in litterbags in a regenerating conifer ecosystem in western
Oregon. Biology and Fertility o f Soils 11: 93-96.
Seastedt. T.R. and O.A. Crossley Jr. 1984. The influence o f arthropods on ecosystems.
BioScience 34: 157-161.
Seidl, R., P.M. Fernandes, T.F. Fonseca, F. Gillet, A.M. Johnsson, K. Merganicova, S.
Netherer, A. Arpaci, J.D. Bontemps, H. Bugmann, J.R. Gonzalez-Olabarria, P. Lasch, C.
Meredieu, F. Moreira, M. Schelhaas and G .M J. Mohren. 2011, Modelling natural
disturbances in forest ecosystems: a review. Ecological Modelling 222:903-924.
Shore, T.L. 1989. Sampling western hemlock looper pupae (Lepidoptera: Geometridae)
using burlap traps. Journal o f the Entomological Society o f B.C. 24: 348-354.
Shortle, W. C. 1979: Mechanisms of compartmentalization of decay in living trees.
Phytopathology 69: 1147-1151.
Shortle, W.C. and A. Ostrovsky. 1983. Decay susceptibility o f wood in defoliated fir trees
related to changing physical, chemical, and biological properties. European Journal o f Forest
Pathology 13: 1-11.
Shortle, W.C., A.L. Shigo, A.L., P. Berry and J. Ahusamra. 1977: Electrical resistance in tree
cambium zone: relationship to rates o f growth and wound closure. Forest Science 23: 326329.
Simard, M.J., Y. Bergeron and L. Sirois. 1998. Conifer seedling recruitment in a
southeastern Canadian boreal forest: the importance o f substrate. Journal o f Vegetation
Science 9: 575-582.

Silvertown, J. 2004. Plant coexistence and the niche. Trends in Ecology and Evolution 19:
605-611.
Sperling, F.A.H., A.G. Raske and I.S. Otvos. 1999. Mitochondrial DNA sequence variation
among populations and host races of Lambdina fiscellaria (Guen.) (Lepidoptera:
Geometridae). Insect Molecular Biology 8: 97-106.
Spribille, T., C.R. Bjork, S. Ekman, J.A. Elix, T. Goward, C. Printzen, T. Tonsberg and T.
Wheeler. 2009. Contributions to an epiphytic lichen flora o f northwest North America: I.
Eight new species from British Columbia's inland rainforests. The Bryologist 112: 109-137.
Stadler, B. and B. Michalzik. 2000. Effects o f phytophagous insects on microorganisms and
throughfall chemistry in forested ecosystems: herbivores as switches for the nutrient
dynamics in the canopy. Basic Applied Ecologist 1: 109-116.
Stadler, B., E. Muhlenberg and B. Michalzik, 2004. The ecology driving nutrient fluxes in
forests. In Insects and ecosystem function. Edited by W.W. Weisser and E. Siemann.
Springer-Verlag, New York. pp. 213-239
Stata Corporation 2011. Stata Statistical Software: Release 12.1. College Station, Texas.
Stevenson, S., H. Armleder, A. Arsenault, D. Coxson and C. DeLong. 2011. British
Columbia’s Inland Rainforest. UBC Press, Vancouver, B.C..
Stevenson, S.K. and D.G. Keisker. 2002. Evaluating the effects o f partial cutting on wildlife
trees and coarse woody debris. In Proceedings of the ecology and management o f dead wood
in western forests, Reno, Nevada, 2-4 November 1999. General technical report PSW-GTR181. W.F. Laudenslayer Jr., P.J. Shea, B.E. Valentine, C.P. Weatherspoon, and T.E. Lisle
(technical coordinators), pp. 753-763.
Stevenson, S.K., Jull, M.J. and B.J. Rogers 2006. Abundance and attributes o f wildlife trees
and coarse woody debris at three silvicultural systems study areas in the Interior CedarHemlock Zone, British Columbia. Forest Ecology and Management 233: 176-191.
Stone, W.E. and M.L. Wolfe. 1996. Response of understorey vegetation to variable tree
mortality following a mountain pine beetle epidemic in lodgepole pine stands in northern
Utah. Vegetatio 122: 1-12.
Strand, M., D.A. Herms, M.P. Ayres, M.E. Kubiske, M.G. Kaufman, E.D. Walker, K.S.
Pregitzer and R.W. Merritt. 1999. Effects o f atmospheric C 0 2, light availability and tree
species on the quality of leaf detritus as a resource for treehole mosquitoes. Oikos 84: 277283.
Swank, W.T., J.B. Waide, D.A. Crossley Jr. and R.L. Todd. 1981. Insect defoliation
enhances nitrate export from forest ecosystems. Oecologia 51: 297-299.

Taylor, S.P. 1996. The bark beetle and hemlock looper programs 1995/96. Forest Health
Note #1. B.C. Ministry o f Forests, Victoria, B.C. 8 p.
ter Braak, C.J.F. and P. Smilauer. 2002. CANOCO Reference manual and CanoDraw for
Windowsuser’s guide: Software for canonical community ordination (version 4.5).
Microcomputer Power, Ithaca, New York.
ter Braak, C.J.F. and P. Smilauer. 2012. CANOCO for Windows 5.0 Microcomputer Power,
Ithaca, New York..
Tester, J.R., A.M. Starfield and L.E. Frelich 1997. Modelling for ecosystem management in
Minnesota pine forests. Biological Conservation 80: 313-324.
Thomson, G.A. 1958. Egg sampling for the western hemlock looper. Forestry Chronicle 34:
248-256.
Throop, H.L. and M.T. Lerdau. 2004. Effects of nitrogen deposition on insect herbivory:
implications for community and ecosystem processes. Ecosystems 7: 109-133.
Torgersen, T.R. 1971. Parasites of the western hemlock looper, Lambdina fiscellaria
lugubrosa (Hulst), in southeast Alaska. Pan-Pacific Entomology 47: 215-219.
Torgersen, T.R. and B.H. Baker. 1967. The occurrence of the hemlock looper (Lambdina
fiscellaria [Guenee]) (Lepidoptera: Geometridae) in southeast Alaska, with notes on its
biology. Research Note PNW-61. Pacific Northwest Forest and Range Experiment Station.
Unites States Department of Agriculture Forest Service, Portland, Oregon. 6 p.
Turner, M.G., W.L. Baker, C.J. Peterson and R.K. Peet. 1998. Factors influencing
succession: lessons from large, infrequent natural disturbances. Ecosystems 1: 511-523.
Tumquist, R. 1991. Western hemlock looper in British Columbia. Forest Insect and Disease
Survey Report 91-8. Pacific Forestry Centre, Canadian Forest Service, Victoria, B.C.
Trumble, J.T., D.M. Kolodny-Hirsh and I.P. Ting. 1993. Plant compensation for arthropod
herbivory. Annual Review o f Entomology 38: 93-119.
Uriarte, M. 2000. Interactions between goldenrod (Solidago altissima L.) and its insect
herbivore (Trirhabda virgata) over the course of succession. Oecologia 122: 521-528.
van den Wollenburg, A.L. 1977. Redundancy analysis, an alternative for canonical
correlation analysis. Psychometrika 42: 207-219.
Veblen, T.T. 1992. Regeneration dynamics. In Plant succession: theory and prediction.
Edited by D.C. Glenn-Lewin, R.K. Peet and T.T. Veblen. Chapman and Hall, London, pp.
152-187.

Wallander, H. 1995. A new hypothesis to explain allocation of dry matter between
mycorrhizal fungi and seedlings in relation to nutrient supply. Plant and Soil 62: 243-248.
Wallner, W.E. 1987. Factors affecting insect population dynamics: differences between
outbreak and non-outbreak species. Annual Review of Entomology 32: 317-340.
Wang, T., A. Hamann, D.L. Spittlehouse and T.Q. Murdock. 2012. ClimateWNA - High resolution spatial climate data for western North America. Journal o f Applied Meteorology
and Climatology 51: 16-29.
Wardle, D. A. 2002. Communities and ecosystems: linking the aboveground and
belowground. Princeton University Press, Princeton, New Jersey.
Wargo, P. M. 1977. Wound elosure in sugar maple: adverse effects of defoliation. Canadian
Journal of Forest Research 7: 410-414.
Watson, E. B. 1934. An account o f the eastern hemlock looper, Ellopia fiscellaria Gn., on
balsam fir. Scientific Agriculture 14: 669-678.
Webb, W.L. 1978. Effects o f defoliation and tree energetics. In The Douglas-fir tussock
moth: a synthesis. Technical Bulletin 1585. Edited by M.H. Brookes, R.W. Stark and R.W.
Campbell. United States Department o f Agriculture Forest Service, Washington, D.C. pp.
77-81.
Weed, A.S., M.S. Ayres and J. Hicke. 2013. Consequences o f climate change for biotic
disturbances in North American forests. Ecological Monographs 83: 1-15.
Western North American defoliator working group and the hemlock looper committee. 2002.
Meeting minutes. Available at
http://prdp2fs.ess.usda.gov/Intemet/FSE_DOCUMENTS/fsbdev2_026022.pdf [viewed May
2, 2012]
Whittome-Way good, B.H., J.C. Fraser, C.J. Lucarotti, I.S. Otvos, N. Conder and D.B. Levin.
2009. In vitro culture of Lambdina fiscellaria lugubrosa nucleopolyhedrovirus in
heterologous cell lines. In Vitro Cellullar and Development Biology - Animals 45: 300-309.
Wilhelm, L. 2004. Understorey vegetation response in western hemlock looper-kill sites of
the Interior Cedar-Hemlock zone (ICKwk3). Undergraduate thesis in Natural Resource
Management. University of Northern British Columbia, Prince George, B.C. 65 p.
Williams, D.W. and A.M. Liebhold. 1995. Influence o f weather on the synchrony o f gypsy
moth (Lepidoptera: Lymantriidae) outbreaks in New England. Environmental Entomology
24: 987-995.
Williamson, S.C., J.K. Detling, J.L Dodd and M.I. Dyer. 1989. Experimental evaluation of
the grazing optimization hypothesis. Journal of Range Management 42: 149-152.

Willig, M. R. and L.R. Walker. 1999: Disturbance in terrestrial ecosystems: salient themes,
synthesis, and future directions. In Ecology of disturbed ground. Edited by L.R. Walker.
Elsevier, Amsterdam, pp. 747-767.
Wohlgemuth, P.M., K.R. Hubbert and P.R. Robichaud. 2001. The effects o f log erosion
barriers on post-fire hydrologic response and sediment yield in small forested watersheds,
southern California. Hydrological Processes 15: 3053-3066.
Wood, C.S. and G.A. Van Sickle. 1993. Forest insect and disease conditions: British
Columbia and Yukon - 1992. Information Report BC-X-340. Pacific Forestry Centre,
Canadian Forest Service, Victoria, B.C. pp. 22-24.
Wright, E. 1963. Ectotrophic mycorrhizae on pine seedlings in Oregon. Ecology 44: 173175.