SNOWSHOE HARE (LEPUS AMERICANUS) HABITAT USE IN THE SUB-BOREAL
FORESTS OF NORTH-CENTRAL BRITISH COLUMBIA

by
JOSEPH DUNCAN CHISHOLM
BSc Honours, University of Northern British Columbia

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF SCIENCE IN ENVIRONMENTAL SCIENCE

in the Department of Natural Resource Sciences

Thesis Examining Committee:
Karl Larsen (PhD), Thesis Supervisor and Professor,
Department of Natural Resource Sciences, Thompson Rivers University
Dexter Hodder (MSc), Thesis Co-Supervisor,
Adjunct Professor, Department of Natural Resource Sciences, Thompson Rivers University
and Director of Research and Education, John Prince Research Forest
David Hill (PhD), Committee Member and Professor,
Department of Environment, Culture and Society, Thompson Rivers University
Douglas Ransome (PhD), External Examiner, Instructor, School of Construction and the
Environment, British Columbia Institute of Technology

April 2023
Thompson Rivers University

© Joseph Chisholm 2023

ii
Thesis Supervisor: Professor Karl Larsen (PhD)

ABSTRACT
Understanding a species’ habitat use allows researchers and managers to identify areas
and/or features essential to management and conservation. For populations that cycle in
abundance, habitat associations may change through time complicating our understanding of
how different habitat metrics may be related to preferences shown by the species. Snowshoe
Hares (Lepus americanus) are a keystone species as they are a major food source for a diverse
array of predators. Understanding their specific habitat associations provides for broader
management planning (e.g., the conservation of associated predators). Although hare habitat use
has been well studied, gaps in our knowledge still exist including how habitat use differs with
ecological scale or throughout their population cycles. In this thesis I examined hare habitat use
in the sub-boreal forests of the John Prince Research Forest and surrounding area in north-central
British Columbia, Canada. I assessed this relationship at multiple scales, with multiple methods
and at varying hare occupancy and densities in order to address these knowledge gaps. I used
wildlife camera trapping (2015, 2016 and 2020) and hare pellet counts (2018 – 2020) to
passively collect hare detection data and used LIDAR data to quantify habitat variables. My
study population also showed evidence of being cyclical and cameras proved useful at detecting
population oscillations. I detected a large decline in camera-detection rates during winter from
0.09 (SD = 0.13) in 2015 to 0.03 (SD = 0.05) in 2020. Naive occupancy declined during the
period from 0.42 to 0.21. Pellet detections also declined between 2018/2019 and 2020. Distance
to riparian area appeared to be the best metric for explaining hare camera detections in 2015,
versus canopy closure in 2016 and distance to edge in 2020. Model selection using pellet counts
was similar between the two years but canopy closure proved the most useful at explaining hare
use at the landscape-scale, versus distance-to-edge for the within-stand models. Habitat use
appeared to be density-dependent and also varied at different ecological scales and this should be
considered in future habitat studies of Snowshoe Hare, as it has often gone unaddressed in the
past.
Keywords: Snowshoe Hare, Lepus americanus, camera trapping, pellet count, occupancy,
density estimation, habitat use, habitat biology, population cycle, sub-boreal forests.

iii

ACKNOWLEDGEMENTS
I would like to thank my supervisor Karl Larsen for enthusiastically taking me and this
project on. This has been a great learning opportunity for me and I am sure it would have not
been the same without his dedication, guidance, patience and support. His wealth of knowledge,
experience and connection with the broader wildlife research community provided an
opportunity to deepen my own.
This project would not have been possible without the support and guidance of Dexter
Hodder and Shannon Crowley from the John Prince Research Forest (JPRF). They provided me
with the data, project set-up and opportunities to conduct my own fieldwork at the JPRF. During
my time at JPRF they supported my fieldwork and shared their local knowledge. They further
supported me through every aspect of crafting this thesis. I would also like to thank the Mitacs
program for financial support in partnership with the JPRF.
I acknowledge and thank Tl’azt’en Nation and Binche Whut’en First Nation for the use
of their traditional territory for this research. Thanks to JPRF research technician Steven
Murdock who worked hard to help me with most of my fieldwork, including a grizzly encounter
or two.
Lastly thank you to my partner, Hannah Sungaila, who has supported me through this
long process and my mother-in-law Brenda Sawatzky, for frequently caring for our son, Miles;
without them I would not have finished this thesis.

iv

TABLE OF CONTENTS
Abstract ........................................................................................................................................... ii
Acknowledgements ........................................................................................................................ iii
Table of Contents ........................................................................................................................... iv
List of Figures ................................................................................................................................ vi
List of Tables ................................................................................................................................ vii
Chapter 1: Introduction ................................................................................................................... 1
Snowshoe Hare Ecology ............................................................................................................. 2
Behaviour and Reproduction .................................................................................................. 3
The Role of Snowshoe Hares as a Keystone Species ............................................................. 3
The Snowshoe Hare Cycle ...................................................................................................... 5
Snowshoe Hare Habitat Use ....................................................................................................... 6
The Importance of Cover for Snowshoe Hares....................................................................... 6
Factors Influencing Snowshoe Hare Habitat Use ................................................................... 7
Wildlife Camera Trapping Studies ............................................................................................. 9
Hare Pellet Counts .................................................................................................................... 11
Thesis Objectives ...................................................................................................................... 12
Study Site .................................................................................................................................. 14
Literature Cited ......................................................................................................................... 19
Chapter 2: Density-Dependent Habitat Use in a Cyclical Snowshoe Hare (Lepus americanus)
Population in the Sub-Boreal Forests of North-Central British Columbia. .................................. 35
Introduction ............................................................................................................................... 35
Methods..................................................................................................................................... 38
Study site............................................................................................................................... 38
Snowshoe Hare Detection Data ............................................................................................ 41
Camera Data...................................................................................................................... 41
Pellet Data ......................................................................................................................... 42

v
Habitat Metrics...................................................................................................................... 43
Habitat Modelling ................................................................................................................. 45
Camera Habitat Models .................................................................................................... 45
Pellet Habitat Models........................................................................................................ 45
Model Set .......................................................................................................................... 46
Results ....................................................................................................................................... 49
Camera Detection Variance Across Years ............................................................................ 49
Camera Habitat Models .................................................................................................... 51
Pellet Detections Across Multiple Scales ............................................................................. 55
Pellet Habitat Models........................................................................................................ 57
Discussion ................................................................................................................................. 61
Habitat Use............................................................................................................................ 61
Habitat Use Across Ecological Scales .................................................................................. 64
Density-Dependent Habitat Use and the Snowshoe Hare Cycle .......................................... 65
Conclusions and Management Implications ......................................................................... 66
Literature Cited ......................................................................................................................... 68
Chapter 3: Conclusion................................................................................................................... 75
Highlights of the Study ............................................................................................................. 75
Management Implications ......................................................................................................... 75
Implications of Study to Snowshoe Hare Ecology ............................................................... 75
Implications of Study to Passive Snowshoe Hare Surveys................................................... 78
Forest Management of a Keystone Species .......................................................................... 79
Limitations and Future Research .............................................................................................. 80
Conclusion ................................................................................................................................ 83
Literature Cited ......................................................................................................................... 84
Appendix 1 – Full Set of Logistic Regression Camera Models (n = 8 per camera trapping
session).............................................................................................................................. 88
Appendix 2 – Full Set of Logistic Regression Pellet Models (n = 7 per camera trapping session).
........................................................................................................................................... 91

vi

LIST OF FIGURES
Figure 1.1. Location of the John Prince Research Forest (JPRF) and surrounding Snowshoe
Hare (Lepus americanus) study area in north-central British Columbia, Canada.......….15
Figure 1.2. Example of common forest types found in and around the John Prince Research
Forest in north-central, British Columbia, Canada. Outer (clockwise from the top left):
Mature warm and dry Douglas-fir site, moist aspen stand, saturated bog-like black spruce
stand (note the presence of the bog species Labrador tea (Rhododendron groenlandicum))
and mature open park-like lodgepole pine stand. Centre: Typical dominant hybrid sprucefir stand..............................................................................................................................17
Figure 2.1. Study area with the locations of the 66 camera trapping stations and 10 pellet sites
used for the five-year (2015 – 2020) Snowshoe Hare (Lepus americanus) habitat study in
the John Prince Research Forest (JPRF) and surrounding area in north-central, British
Columbia, Canada..............................................................................................................40

vii

LIST OF TABLES
Table 2.1. Habitat and site covariates used in Snowshoe Hare (Lepus americanus) logistic
regression models for camera and pellet detection data (2015 – 2020) from the John
Prince Research Forest and surrounding area in north-central, British Columbia (BC),
Canada. Data collected using remote sensing LIDAR with additional data from the
Vegetation Resource Index (VRI; Ministry of Forests, BC). Canopy closure is in meters
(m) above the ground. *Used in camera models only....………...................……………44
Table 2.2. Models used in logistic regression analysis of camera (2015, 2016 and 2020) and
pellet (2019 and 2020) detection data from the John Prince Research Forest and
surrounding area in north-central, British Columbia, Canada. BA = basal area, CC =
canopy closure between 0 – 1 m, 1 – 3 m, 3 – 10 m and greater than (>) 10 m, DistEdge =
distance to edge, DistRip = distance to riparian area and Elev = elevation................…...48
Table 2.3. Total camera-days (maximum of 4620 per year for winter (W) and 8918 for springsummer (SS)), mean camera-days per camera station (maximum of 70 per station for W,
182 for SS) (SD), total camera-days with detections, mean detection rate (total cameradays with detections/total camera-days) (SD) and occupancy (stations with hares
present/total stations) for Snowshoe Hare (Lepus americanus) camera trapping study
during W (February and March) 2015, 2016 and 2020 and SS (April – September) 2016
in the John Prince Research Forest and surrounding area, north-central British Columbia,
Canada................................................................................................................................50
Table 2.4. Logistic regression models for Snowshoe Hare (Lepus americanus) camera trapping
study in the John Prince Research Forest and surrounding area, north-central British
Columbia, Canada. Only models deemed competitive with the top model (i.e., within two
Delta AIC units) are displayed. K = number of parameters, AICci = corrected Akaike’s
Information Criterion, AICcΔi = Delta AIC = difference in AIC score between top model
and each model, AICcWi = AIC weight = proportion of the total amount of predictive
power of each model when assessing the full set of models, AUC = area under the curve
for the receiver operating characteristic. See Table 2.2 for model abbreviations. * Denotes
significant coefficient (P < 0.05).......…........................................................................…52

viii
Table 2.5. Coefficients (β) for significant variables in logistic regression models for Snowshoe
Hare (Lepus americanus) camera detections in the John Prince Research Forest and
surrounding area, north-central British Columbia, Canada. See Table 2.2 for model
abbreviations......................................................................................................................53
Table 2.6. Estimated hare density (hares/ha) using the Yukon Equation (Krebs et al. 2001) for
Snowshoe Hare (Lepus americanus) pellet sites in the John Prince Research Forest
(JPRF) and surrounding area, north-central British Columbia (BC), Canada. Each site
contained 100 plots for a total of 1000 plots (600 plots in 2018) counted twice annually
and combined to obtain a single annual estimate. *These sites were eliminated from the
within-stand model analysis. **Forest stand age estimates obtained from the BC
Vegetation Resource Inventory and confirmed by JPRF harvest records.........................56
Table 2.7. Logistic regression habitat models for Snowshoe Hare (Lepus americanus) derived
from pellet counts in 2019 and 2020 within the John Prince Research Forest and
surrounding area, north-central British Columbia, Canada. All four study periods used the
same set of seven models (See Table 2.2 for model abbreviations) but only those deemed
competitive with the top model (i.e., within two Delta AIC units) are displayed. K =
number of parameters in model including the intercept, Akaike’s Information Criterion
scores (AICci), Delta AIC (AICcΔi) = difference in AIC score between top model and each
model, AIC weight (AICcWi) = proportion of the total amount of predictive power
provided by each model when assessing the full set of models, and area under the curve
(AUC) for the receiver operating characteristic. *Denotes significant coefficient (P <
0.05)...................................................................................................................................58
Table 2.8. Coefficients for significant variables in logistic regression models for Snowshoe Hare
(Lepus americanus) pellet detections in 2019 and 2020 in the John Prince Research
Forest and surrounding area, north-central British Columbia, Canada. WS = Withinstand-scale and LS = Landscape-scale models, see Table 2.2 for model abbreviations....60

1

CHAPTER 1: INTRODUCTION
Wildlife habitat can most broadly be defined as the natural environment an animal
occurs in (Grebner et al. 2013) and may include any space that provides food, shelter, or
breeding grounds (Sharpe and Undersander 2019). Densities and patterns of landscape
occupancy vary across heterogenous landscapes for many wildlife species based on habitat
requirements that increase each individual’s chance of survival and fitness (Stirnemann et al.
2015, Zhang et al. 2018). For example, many small rodents (e.g., mice, voles, shrews) are
more active and abundant in areas with taller vegetation as it may provide protection from
predators (Jacob 2008). Female White-Tailed Deer (Odocoileus virginianus) may seek areas
of their home ranges with the greatest cover during and following parturition (Hasapes and
Comer 2017). All species, sub-species and local populations have these types of habitat
associations that may change during different life-stages, seasons or with population cycles
and understanding them is key to studying and managing most wildlife.
Studies that examine how habitat quality affects the distribution and abundance or
occupancy of species across a landscape have been a foundation of ecological research
(Royle and Nichols 2003, Weir et al. 2003, Fuller et al. 2016). By combining species
detection data and local habitat data we can assess how different biotic and abiotic factors
influence habitat use (MacKenzie et al. 2002, 2006). These types of studies are commonly
used by researchers and managers to assess habitat quality while identifying important
features and/or habitats essential to long-term conservation and management (MacKenzie et
al. 2002, Di Marco et al. 2017).
Assessing habitat use is difficult for species or local populations that cycle in
abundance over time (Mordecai et al. 2011). Many species are known to exhibit densitydependent habitat use, in which they exploit different habitats during population highs
versus lows (Wolff 1980, van Beest 2014). As populations increase high-quality habitat can
become oversaturated forcing some individuals into lower-quality habitat (Fretwell and
Lucas 1969). For example, high-quality habitats are known to sustain healthy and stable
cyclical vole (subfamily Arvicolinae) populations, whereas marginal habitats are only
occupied during population peaks (Sundell et al. 2012). Habitat studies undertaken at

2
population highs and lows may lead to different and even conflicting results and
interpretations so there is a need to standardize surveys at known points within the cycle.
The well-known 8 – 11-year population cycle of Snowshoe Hare (Lepus americanus)
has been recognized by scientists for over a century and is one of the most studied ecological
phenomena in the world (Meslow and Keith 1968, Keith and Windberg 1978, Krebs et al.
2018). I undertook a hare habitat use study, using passive detection methods, to answer
some key questions about the relative importance of different habitats and features to hares
and the local hare cycle in north-central British Columbia (BC), Canada.
In this chapter I give an overview of Snowshoe Hare ecology and habitat use,
highlight much of what is already known about Snowshoe Hares and identify knowledge
gaps. I then lay out the objectives of my study and how it aims to address some of these
gaps. Finally, I introduce my dynamic study site area in north-central BC.
SNOWSHOE HARE ECOLOGY
Snowshoe Hares are medium-sized (~1.1 – 1.5 kg; Black 1965, Sullivan and Sullivan
1988, Murray 2002, Ellsworth et al. 2016) lagomorphs, native to North America. They
occupy a range of habitat types and are distributed from the boreal forests of Alaska to
Newfoundland and Labrador (introduced to the island of Newfoundland), south through the
sub-boreal forests of southern Canada and into the montane forests of the Rocky, Sierra
Nevada and Appalachian Mountains. They are absent from the far north of the Arctic, prairie
grasslands and coastal islands of BC.
Snowshoe Hares draw their name from their recognizably large hindfeet that help
them move across snow. The soles of their feet have dense fur, another anatomical feature
that hints to their evolution in and for the cold snowy winter conditions of their northern and
montane range. Across most of their range they are known to turn from a rusty- or grayishbrown colour in summer to white in winter; this provides camouflage from predators (Mills
et al. 2018, Zimova et al. 2020) and may also increase the insulating properties of hair to
assist thermoregulation in winter (Russell and Tumlison 1996, Gigliotti et al. 2017).

3
Behaviour and Reproduction
Snowshoe Hares are nocturnal and/or crepuscular and active year-round (Keith 1964,
Feierabend and Kielland 2014, Studd et al. 2019). During the day they hide under dense
brush, in hollow trees and logs, or anywhere else that provides structural protection. They
eat a large variety of woody and non-woody plant species across their range (Brooks 1955,
Telfer 1972, Smith et al. 1988) and also are known to scavenge carrion (Peers et al. 2018).
Most Snowshoe Hare populations have multiple litters per breeding season; the
average is typically around three (Dodds 1965, Meslow and Keith 1968, O’Donoghue and
Krebs 1992). Breeding begins in early-spring followed by a 35 – 40-day gestation period,
with the first litter occurring in May followed by short intervals between subsequent litters
(Severaid 1945, Dodds 1965, Kuvlesky and Keith 1983). Litter sizes can vary from one to
14 leverets but the typical average is three to five per litter (Meslow and Keith 1968,
O’Donoghue and Krebs 1992, Stefan and Krebs 2001). Leverets usually are weaned in under
a month with females often giving birth to a new litter shortly afterwards (Severaid 1942,
Rongstad and Tester 1971). Litters generally are synchronized among females so that
distinct litter groups occur within a population (Meslow and Keith 1968, O’Donoghue and
Boutin 1995, Peers et al. 2021).
Pregnancy rates vary with each litter group but generally exceed 90% (Kuvlesky and
Keith 1983, O’Donoghue and Krebs 1992, Krebs et al. 2001a). Females normally first breed
as yearlings, this group making up approximately 70% of the breeding population, given that
Snowshoe Hares have a short average life span of one year (Krebs et al. 2018). Surprisingly
the main cause of leveret mortality is predation by other small mammals such as American
Red Squirrels (Tamiasciurus hudsonicus) and Arctic Ground Squirrels (Urocitellus parryii;
O’Donoghue 1994, Krebs et al. 2018).
The Role of Snowshoe Hares as a Keystone Species
The International Union for the Conservation of Nature (IUCN) presently lists
Snowshoe Hares as a species of Least Concern (LC - G5), which corresponds to their
appearance on the Yellow List in BC (secure - S5). While there is currently not broad
concern for their conservation, they are widely recognized as playing a critical role in food

4
webs. Snowshoe Hares, particularly during periods of high abundance, represent a
substantial amount of the vertebrate biomass available to many mesocarnivores, including
Canada Lynx (Lynx canadensis), American Marten (Martes americana), Fisher (Pekania
pennanti), Wolverine (Gulo gulo), Coyote (Canis latrans), Red Fox (Vulpes vulpes) and
others. This is especially true in winter when other small mammals are hibernating or have
reduced activity (Krebs et al. 2018). Overall, the main cause of Snowshoe Hare mortality is
predation (Boutin et al. 1986, Feierabend and Kielland 2015, Majchrzak et al. 2022).
Larger apex predators, such as Gray Wolves (Canis lupus) and Cougars (Puma
concolor), also are known to occasionally consume hares (Spalding and Lesowski 1971,
Floyd et al. 1978, Gable et al. 2018). Avian predators, such as the Great Horned Owl (Bubo
virginianus; Rohner and Krebs 1996) and Northern Goshawk (Accipiter gentilis; Boutin et
al. 1995), as well as other members of Orders Accipitriformes and Strigiformes also predate
hares (Luttich et al. 1970, Krebs et al. 2018, Schmidt et al. 2018). In fact, avian predation
can account for up to 40% of predation events in some hare populations (Krebs et al. 1995,
Feierabend and Kielland 2015). Other species prey on leverets during spring and summer,
including squirrels. This critical and significant role played by Snowshoe Hares in food
webs is why they are considered a keystone species in many parts of their range (Bois et al.
2012, Morris and Vijayan 2018, Jensen et al. 2022).
Canada Lynx are known to be a specialist predator of Snowshoe Hares across much
of their range and account for a significant proportion of hare mortality; hares in fact often
constitute the primary prey species of lynx (Breitenmoser et al. 1993, O’Donoghue et al.
1997, Boutin 1998). Lynx are a medium-sized felid species that have evolved large
snowshoe-like feet and other features for the snowy winter climates they inhabit. Their range
overlaps almost completely with that of Snowshoe Hares but they are absent from the
southernmost portions of the hare’s range in the contiguous United States of America
(USA). The connection between lynx and hares is so strong that their populations are known
to cycle together, with lynx typically demonstrating a time-lag behind hare numbers, across
much of their joint range (Breitenmoser et al. 1993, Vik et al. 2008, Tyson et al. 2010).

5
The Snowshoe Hare Cycle
The hare-lynx association is a classic ecological tale found in many textbooks. It
often is used to demonstrate the Lotka-Volterra equation that describes the dynamics of
interacting predator-prey species (Krebs et al. 2001a). Reports of regular oscillations in
Snowshoe Hare populations were made by fur trappers in the 18th and 19th centuries (Krebs
et al. 2001a). The hare-lynx cycle was first detected by scientists in the early 20th century
when The Hudson Bay Company’s detailed fur trading records were analyzed by biologists
(Elton and Nicholson 1942). These records demonstrated approximately 10-year intervals
between population peaks in many discrete locations across Canada (Elton and Nicholson
1942). The amplitude of the population peaks also fluctuates between cycles (Ginzburg and
Krebs 2015, Myers 2018). The Snowshoe Hare cycle was thereafter confirmed with field
studies (Green and Larson 1938, Green and Evans 1940, Dodds 1965).
The documentation of the hare cycle has led to considerable work seeking to explain
the underlying cause for such an extensive, synchronous cycle (Keith and Windberg 1978,
Breitenmoser et al. 1993, O’Donoghue et al. 1997 and reviewed in Krebs et al. 2018).
Hypotheses have included predation (Krebs et al. 1995, 2018, Sheriff et al. 2011), food
availability (Pease et al. 1979, Vaughan and Keith 1981, Keith 1983), secondary plant
compounds (Fox and Bryant 1984, Sinclair and Smith 1984, Sinclair et al. 1988), solar
activity (Sinclair et al. 1993, Ranta et al. 1997, Sinclair and Gosline 1997) and parasites,
diseases and other stressors (Green and Larson 1938, Erickson 1944, Keith et al. 1985).
Studies across their range, but especially those in Alberta (Keith 1966, Meslow and Keith
1968, Cary and Keith 1979) and Southern Yukon (reviewed in Krebs et al. 2001a and 2018),
have helped explain this phenomenon, with predation now appearing to be the primary cause
of the cycle (Krebs et al. 2018). Predators not only cause direct mortality but interactions
with them cause stress in reproductive females that is passed down via the maternal effect
and sustains the low-phase of the hare cycle by decreasing reproductive output (Sheriff et al.
2009, 2010, 2011). The constant threat of predation influences Snowshoe Hare habitat use as
they select habitats that offer the greatest levels of protection.

6

SNOWSHOE HARE HABITAT USE
Snowshoe Hare habitat use has received considerable research attention across much
of their range (Sullivan and Sullivan 1988, Ferron and Ouellet 1992, Kawaguchi and
Desrochers 2018). Snowshoe Hares generally are restricted to continuous forested habitats,
avoiding open grasslands, coastal beaches, alpine habitats and tundra. They do, however,
occupy a diverse array of forest types: middle- to mature-aged conifer-predominated forests
in western North America (Ausband and Baty 2005, Griffin and Mills 2007, Sullivan et al.
2012), black spruce (Picea mariana) forests of northern and eastern Canada (Ferron et al.
1998, St-Laurent et al. 2008), mixed hardwood-softwood forests of their southeastern range
(Litvaitis et al. 1985, Homyack et al. 2006, Gigliotti et al. 2018) and subalpine fir (Abies
lasiocarpa) stands found in the southern extent of the species’ range in southwestern USA
(Frey and Malaney 2006, Malaney and Frey 2006). Hares also use stands and patches of
deciduous species such as trembling aspen (Populus tremuloides; Wolfe et al. 1982, Strong
and Jung 2012), red maple (Acer rubrum; Conroy et al. 1979) and scrub oak (Quercus
ilicifolia; Gigliotti et al. 2018).

The Importance of Cover for Snowshoe Hares
Despite the wide range of forest types used by hares, there appear to be some
commonalities. Cover, primarily provided by vegetation, often is cited as the most important
determinant and predictor of hare habitat use, as it provides visual obstruction from
predators (Litvaitis et al. 1985, Bois et al. 2012, Simard et al. 2018). Given the intensity of
predation on hare populations, it is not surprising that individuals select habitats that provide
protection and increase their fitness. This relationship supports the so-called “Landscape of
Fear” theory that postulates the perceived threat or fear of potential predators cause animals
to choose habitats that offer protection, even if doing so requires the sacrifice of optimal
foraging (Brown et al. 1999, Laundré et al. 2001). Cover also could be correlated with other
important life-history requirements of hares, such as thermoregulation and food quantity
and/or quality.

7
Horizontal cover, also called lateral, understory or shrub cover, frequently has been
found to be the most important type of cover for Snowshoe Hares (Litvaitis et al. 1985, Bois
et al. 2012, Gigliotti et al. 2018). Dense understories provide places where hares can feed
and rest while reducing their susceptibility to terrestrial predators (Bois et al. 2012, Suffice
et al. 2015). This type of cover also may visually obstruct hares from aerial predators,
although canopy closure is likely more important in this role (Thomas et al. 2019). Increased
canopy closure has been found to be correlated with increased hare use (Roy et al. 2010,
Thornton et al. 2013, Cheng et al. 2015). However, Bois et al. (2012) did not find a strong
correlation between canopy closure and hare use, suggesting canopy closure in fact may
reduce horizontal cover, possibly by reducing light availability (Simard et al. 2018). Another
habitat feature that may impart protection to hares is tree density, with hares being more
abundant in areas with greater numbers of trees and/or volume (i.e., basal area) of trees
(Hodson et al. 2010a, Ewacha et al. 2014, Thomas et al. 2019).
The relationship between hares and distance to habitat edges also has been explored.
Some of this work has suggested hares use edge habitats (Conroy et al. 1979), while others
lean towards avoidance (Roy et al. 2010, Sultaire et al. 2016). In winter, when hares feed
more on woody browse, they are known to forage in densely vegetated interior sites. In
summer, however, it is the interspersion of interior habitat (providing protection) and open
habitat and gaps (providing herbaceous foliage) that may drive habitat use (Wolff 1980).
Although edge habitat may provide an important interspersion of foraging and cover habitat
(Conroy et al. 1979), predation rates of some prey species are known to be greater closer to
edge habitat (Andrén and Angelstam 1988, Andrén 1995).
Factors Influencing Snowshoe Hare Habitat Use
In addition to cover, other site factors known to influence hare use include elevation
and slope (Berg et al. 2012, Thornton et al. 2013). Coarse woody debris also has been
positively associated with hare use (Berg et al. 2012). The local soil type may influence
vegetation and subsequently hare use (Berg et al. 2012). Stand-age and silviculture practices
have also been linked to hare habitat use patterns (Sullivan 1994, Sullivan et al. 2007, Bois
et al. 2012). All told, there appears to be a multitude of factors and interactions that

8
influence hare habitat use and patterns of landscape occupancy. Snowshoe Hare habitat use
also varies regionally and more studies have occurred in the northern boreal and southern
extents of their range.
Despite the relatively large number of hare habitat studies, we still lack a clear
picture of how habitat use changes throughout and between population cycles. Snowshoe
Hares are known to exhibit density-dependent habitat use (Wolff 1980, Hodson et al.
2010b). When the population is low, during the crash and recovery phase of the cycle, the
remaining individuals only may occupy the best quality habitat, usually areas with dense
vegetation (Wolff 1980). During population highs, however, the higher density of hares may
result in occupancy across a wider array of available habitats. This would conform to the
“Ideal Free Distribution” theory that predicts the occupancy of suboptimal habitat at
population peaks (Fretwell and Lucas 1969). This in turn raises the question of when is the
right time to survey cyclic populations to quantify and assess habitat use. Studies undertaken
at different points during a population cycle may yield different and even conflicting results
and interpretations. Short-lived occupancy patterns (i.e., patch colonization and extinction)
also may render it harder to classify what is or is not hare habitat in variegated landscapes
(Thornton et al. 2013). If Snowshoe Hares do retreat to high-quality refugia during cyclic
lows, then those habitats may be relatively important when managing for hares and their
many predators.
Hare habitat use also can vary at different ecological scales (Lewis et al. 2011, Fuller
and Harrison 2013, Gigliotti et al. 2018). Most hare studies have been at landscape- or standlevel and only a few have examined within-stand or plot-level habitat use (Ausband and
Baty 2005, Fuller and Harrison 2013, Holbrook et al. 2017). Horizontal cover has been
found to drive within-stand habitat selection with structure/visual obstruction being more
important than species composition (Gigliotti et al. 2018). Holbrook et al. (2017) found that
horizontal cover drove habitat use at the within-stand and landscape-scale but that snow
depth was significant only at the landscape-level. Similarly, Sultaire et al. (2016) found that
horizontal cover was the only significant predictor of habitat use at the plot-scale but not at
the landscape-level. Selection for forested and lower edge density areas at the landscapescale showed how the effects of habitat fragmentation outweighed the influence of
horizontal cover at a broader scale. This illustrates how studies undertaken at multiple scales

9
can be important in determining the full scope of hare habitat use (Thornton et al. 2013,
Holbrook et al. 2017).
Holbrook et al. (2017) pointed out that not only does hare habitat use differ at
varying scales but with different response variables (e.g., occupancy, density, detection
rates, etc.). Most ecological studies focus only on a single response but it has been shown
that habitat factors can influence both occupancy and density in different ways (Holbrook et
al. 2016 and 2017). They suggest studies that evaluate multiple responses may be able to
improve our understanding of habitat use by Snowshoe Hares. They further suggest that
study areas should be in continuous habitat with a gradient of natural conditions over typebased or stratified approaches (e.g., sampling only pre-defined high-quality habitat based on
existing cover maps; Holbrook et al. 2017).
Snowshoe Hare home range size varies across different habitat types but is typically
7 – 14 ha and averages around 9 ha, of which hares spend 80% or more of their time in a
core-area of less than 3 ha (Wolff 1980). Male hares generally have larger home ranges than
females (Wolff 1980). Snowshoe Hares exhibit seasonal patterns of habitat use (Wolff 1980,
Ferron and Ouellet 1992, Ivan et al. 2014). In winter they may spend their time in densely
vegetated interior forests that provide not only the cover they require but also winter diet
items such as spruce, willow, birch and alder species (Wolff 1980). During summer they
may move to areas with a mix of open and densely vegetated areas (e.g., near forest edges or
gaps) to access herbaceous foliage that grows in areas where increased light penetration
encourages growth (Wolff 1980, Ferron and Ouellet 1992).
WILDLIFE CAMERA TRAPPING STUDIES
Over the years various methods have been used to collect detection, movement,
density and other data in to assess Snowshoe Hare habitat use. Mark-recapture studies
historically were a common way to gather this type of information, but they remain labour
intensive and invasive (Meslow and Keith 1968, Mills et al. 2005). The attachment of VHFand GPS-collars provides detailed movement data that can give insights into fine-scale
habitat use (Feierabend and Kielland 2014, Majchrzak et al. 2022) but similarly requires
invasive capture, restraint and handling. Pellet (scat) counts provide a relative measure of

10
use that has most commonly been used at the stand-level. Wildlife camera studies have
become common over the last two decades and have also been used to assess hare habitat
use (Thomas et al. 2019).
The use of wildlife camera trapping (CT) began with the development of new
technology in the 1990s and quickly became a common way to passively monitor wildlife
species (Shannon et al. 2014, Newey et al. 2015, Neilson et al. 2018). Cameras can be
deployed remotely to passively gather detection data on one or more focal species. They also
may be used in place of labour intensive and stress-inducing traditional trapping studies,
depending on the objectives of the study. Historically these studies were limited by camera
battery life and storage capacity but both of these features have continued to improve over
time. They have been used to assess wildlife abundance and distribution (Linkie et al. 2013,
Rich et al. 2017, Evans et al. 2019), habitat use (Fuller et al. 2016, Thomas et al. 2019,
Wilson et al. 2020), behaviour (Peers et al. 2018), to detect rare species (Linkie et al. 2013)
and even to estimate density (Rowcliffe et al. 2008, Jensen et al. 2022).
Many different sampling designs and types of analyses have been used for CT
studies and several papers have stressed the importance of standardizing and validating the
different approaches (Meek et al. 2014, Burton et al. 2015, Newey et al. 2015). For example,
using multiple cameras, with a minimum of two per station, may greatly improve detection
rates and reduce false negatives (O’Connor et al. 2017, Evans et al. 2019). However, it is
challenging to standardize approaches when habitats or species ecology require different
methods (Burton et al. 2015). It is critical to assess the relationship between camera
detections and abundance because if this relationship does not exist, or if it is heavily
affected by animal movements or other factors, then research conclusions may not be valid
(Burton et al. 2015).
Once CT data have been collected, they must be categorized into discrete
presence/absence detection histories that can vary from a day to weeks (Burton et al. 2015,
Rich et al. 2017, Neilson et al. 2018). The detection history data can then be used in models
to assess habitat use. Occupancy modelling commonly has been used with CT data as it
incorporates detection probabilities that account for false negatives (Shannon et al. 2014,
Fuller et al. 2016, Neilson et al. 2018). This may not, however, be necessary if you can
assume high levels of detection, especially with abundant species and other modelling

11
approaches, such as logistic regression, also may be appropriate (Ngoprasert et al. 2012,
Chaiyarat et al. 2019)
Researchers can add covariates to models to assess the use of different habitat types
and features or factors influencing detection (Burton et al. 2015, Neilson et al. 2018).
Several Snowshoe Hare studies have used CT recently for a variety of objectives including
multi-species monitoring (Foresman and Pearson 1999, Evans et al. 2019, Jensen et al.
2021). Thomas et al. (2019) investigated the effects of salvage logging induced structural
changes on hare habitat use. Wilson et al. (2020) used cameras to study a population of
translocated hares and their colonization and extinction of different habitats. Other aspects
of Snowshoe Hare ecology have been studied using cameras including risk management
strategies (Morris 2019), molting and camouflage (Zimova et al. 2020) and carrion
scavenging (Peers et al. 2018). Recently Jensen et al. (2022) used CT to estimate Snowshoe
Hare density, suggesting that with further refinement, CT studies may replace traditional
pellet counts.

HARE PELLET COUNTS
The technique of fecal pellet counts has been used by wildlife biologists to estimate
abundance for many years (Neff 1968, White and Eberhardt 1980, Litvaitis et al. 1985).
Originally the use of pellet data was considered inappropriate for estimating density (Wolfe
et al. 1982); however, the work of Krebs et al. (1987) set out a method of estimating
Snowshoe Hare density through pellet counts, creating a standard for hare population
assessments. In their seminal paper the researchers used Jolly-Seber mark-recapture
estimates from a hare trapping study along with pellet counts to create an equation (the
“Yukon Equation”) that could be used to estimate hare density (i.e., hares/ha). They
determined the relationship between pellet and mark-recapture estimates to be so close as to
recommend the use of pellet counts as a precise and accurate way to estimate hare density.
Some studies simply have used pellet counts to determine the abundance and
distribution of hares (Homyack et al. 2006, Hodges et al. 2009). Other studies have used
pellet counts to assess habitat use generally, or more specifically the impacts of different
forestry treatments (e.g., precommercial thinning, fertilization) on hares (Sullivan et al.

12
2006, Griffin and Mills 2007, Homyack et al. 2007). Pellet counts have been used as a
stand-alone approach (Newbury and Simon 2005, Thornton et al. 2013, Sultaire et al. 2016),
or combined with complimentary methods such as mark-recapture and snow-tracking
(Ferron et al. 1998, Homyack et al. 2006, Hodges et al. 2009).
Since the publication of the Yukon Equation other studies have developed their own
regression models linking pellet count data to hare habitat use (Mills et al. 2005, McCann et
al. 2008, Berg and Gese 2010). Mills et al. (2005) found that the Yukon Equation provided
comparable density estimates to their locally derived equation despite the fact that their
study took place in very different habitat in Montana. They suggested that the Yukon
Equation may be applicable across a broader spectrum of habitats, especially if studies do
not require highly accurate measure of absolute abundance. The Yukon Equation has since
been revised by some of the original researchers and other collaborators (Krebs et al.
2001b).
The precise method of collecting Snowshoe Hare pellet counts has involved a variety
of study designs and plot sizes. Hodges and Mills (2008) suggested using long thin
2

2

rectangles (5.08 cm x 305 cm = 0.155 m ) or circles (1 m ). The original Yukon Equation
2

papers used 0.155 m rectangular plots (Krebs et al. 1987, Krebs et al. 2001b) and
subsequently it has been suggested that using different plot sizes with this equation may not
be appropriate (McKelvey et al. 2002, Mills et al. 2005). It has been suggested the optimal
number of plots per site is between 50 – 100 and shown that increasing the number past that
does not improve estimates (Krebs et al. 2001b, Hodges and Mills 2008). Increasing the
number of sites, however, can improve accuracy and a minimum of three sites per study
(150 – 300 plots) was suggested by Krebs et al. (2001b).

THESIS OBJECTIVES
After an in-depth review of over a century worth of Snowshoe Hare research, with a
particular focus of hare habitat use, it was clear several knowledge gaps still exist. While it
is known that hares exhibit density-dependent habitat use, few studies have directly
compared habitat use at different population densities or occupancy. Snowshoe Hare studies
also have been mainly concerned with habitat use at a stand-level scale and the nuances of

13
within-stand habitat use have often gone ignored. In this thesis I address both these issues
and also expanded on traditional hare habitat use studies by using multiple ecological
responses (i.e., occupancy and density) in a range of habitats and methods newer to hare
research (i.e., wildlife CT).
For this thesis I partnered with the John Prince Research Forest (JPRF) in northcentral BC. A primary focus of the JPRF is to monitor multiple species, particularly
mesocarnivores, over time. This has made understanding the relationship between hares,
their population cycles and habitat particularly relevant. Staff at JPRF and I collected
passive hare detection data between 2015 and 2020, which I then analyzed for this thesis.
In consultation with JPRF researchers and following Holbrook et al. (2017) I chose
to conduct a multi-scale, multi-response study of hare occupancy patterns over a
heterogenous landscape that represented a gradient of potential habitat. In a broad sense my
goal was to establish baseline population and habitat analysis that could be used to aid
further Snowshoe Hare research at the JPRF. Additionally, this analysis may be useful to
others studying hare predators or larger ecosystem dynamics at the JPRF and more broadly.
My specific objectives were to determine what habitat components were most
important to hares in my study area and to develop models that could predict hare use.
Additionally, I wanted to compare habitat use at the stand- and within-stand-scale.
Furthermore, I wanted to determine if the local hare population oscillated in abundance over
time and whether cameras were useful in detecting that change. If a population oscillation
did occur, I was curious to determine whether the local hare population exhibited densitydependent habitat use.
In the following chapter I use overlapping CT and pellet counts to assess hare habitat
use using logistic regression models. I further used the pellet data to assess whether habitat
use differed at the landscape compared to the within-stand-scale. With the camera data I also
evaluated whether cameras can detect population oscillations to assess their use for
monitoring the Snowshoe Hare cycle. In the final chapter of this thesis, I summarize the
results of my work, discuss management implications, highlight limitations encountered
during my study and suggest areas for future research.

14
STUDY SITE
In addition to field work, I also utilized historical data collected at the JPRF. The
JPRF is located in north-central BC approximately 45 km north of Fort St. James and served
as the main study area for this thesis (Figure 1.1).
The JPRF is cooperatively managed by the Tl’azt’en Nation, Binche Whut’en First
Nation and the University of Northern British Columbia but works with other academic
institutions across BC and beyond. Their primary objective is to “develop innovative
approaches to natural resource conservation and management that combines First Nations’
traditional and western scientific approaches to understanding the natural world”. For over
twenty years researchers with the JPRF and their external partners have carried out a diverse
array of wildlife and environmental studies in the region. Some of the species previously
studied at the JPRF include River Otter (Lontra canadensis; Crowley et al. 2013b, 2018),
Canada Lynx (Crowley et al. 2013a, Crowley and Hodder 2017), Moose (Alces alces;
Chisholm et al. 2021), American Mink (Neovison vison; Hodder et al. 2018), American
Marten (Seip et al. 2018) and others.
The JPRF and the surrounding region is within the Sub-Boreal Spruce
biogeoclimatic zone, typical of north-central BC, that experiences cold snowy winters and
relatively short cool summers (Meidinger and Pojar 1991). Mean daily average temperatures
(1981 – 2010) were 3.5 ºC; the lowest monthly mean daily average occurred in January (9.5º C) while the warmest occurred in July (15.4º C) (Environment and Climate Change
Canada 2022). Mean annual precipitation (1981 – 2010) was 487 mm, with more than a
third of it falling as snow (173 cm) (Environment and Climate Change Canada 2022).
The JPRF is bordered to the north by Tezzeron Lake and Pinchi Lake to the south
(see Figure 1.1). Both year-round and ephemeral streams of varying size criss-cross much of
the landscape. Wetlands, bogs and small lakes are numerous. The valley bottoms sit at
approximately 700 m above sea level with rolling terrain and low mountains to 1500 m
(Crowley and Hodder 2017). Land-use operations in the area include forestry, ranching and
historically, mining. There are no major highways or urban areas within the study area but
there are several active logging roads and backroads for access to remote recreation sites and
rural properties.

15

Figure 1.1. Location of the John Prince Research Forest (JPRF) and surrounding Snowshoe
Hare (Lepus americanus) study area in north-central British Columbia, Canada (UTM Zone
10 – NAD83).

16
Much of this region is predominated by hybrid white spruce (Picea x albertiana) and
subalpine fir (Abies lasiocarpa) (Meidinger and Pojar 1991; See Figure 1.2 for examples).
Lodgepole pine (Pinus contorta) is an early-seral species but also occurs frequently on drier
mature sites (Bedford and Sutton 2000). Black spruce (Picea mariana) is the dominant
species in bogs and saturated areas (Wali and Krajina 1973).
Interior Douglas-fir (Pseudotsuga menziesii var. glauca) is less predominant but
occurs on dry warm sites throughout the region (Meidinger and Pojar 1991) where the
species is at its northern limit (Crowley and Hodder 2017). Smaller patches and stands of
deciduous trees occur throughout the area. Trembling aspen stands are fairly common in
moist historically harvested areas (Wali and Krajina 1973). Other deciduous species such as
paper birch (Betula papyrifera) and poplar (Populus balsamifera) also are common but less
abundant (Wali and Krajina 1973).
Close to a century of forestry has created a patchwork of seral stages in the JPRF and
it is now predominantly composed of regenerating-to-mature conifer stands with patches of
deciduous and clear-cut areas (Crowley and Hodder 2017). In the summer of 2015, JPRF
staff conducted a variety of timber cruising and vegetation field surveys at all camera
stations confirming the description of sub-boreal spruce forest composition was accurate for
my study area. Hybrid spruce was by far the predominant tree and accounted for nearly half
(41%) of all trees counted in cruise plots. Several other species occurred in varying amounts:
lodgepole pine (16%), subalpine fir (13%), trembling aspen (8%), paper birch (6%) and
Douglas-fir (5%) with minor amounts of black spruce and poplar.
A dense understory is found throughout much of the study area (Crowley and
Hodder 2017). Visual obstruction cover pole measurements at camera stations ranged from 0
to 90% with an average of 35%. Common shrub species counted during brush cover surveys
included willows (Salix spp.), red osier dogwood (Cornus sericea), black huckleberry
(Vaccinium membranaceum), saskatoon berry (Amelanchier alnifolia), twinberry (Lonicera
involucrata), highbush-cranberry (Viburnum edule), soopolallie (Shepherdia canadensis),
devil’s club (Oplopanax horridus) and alders (Alnus spp.).

17

Figure 1.2. Example of common forest types found in and around the John Prince Research
Forest in north-central, British Columbia, Canada. Outer (clockwise from the top left):
Mature warm and dry Douglas-fir site, moist aspen stand, saturated bog-like black spruce
stand (note the presence of the bog species Labrador tea (Rhododendron groenlandicum))
and mature open park-like lodgepole pine stand. Centre: Typical predominant hybrid sprucefir stand.

18

A diverse terrestrial mammalian mesocarnivore community is found in the JPRF and
includes Canada Lynx, Coyote, Fisher, American Marten, American Mink, Red Fox, ShortTailed Weasel (Mustela erminea), Striped Skunk (Mephitis mephitis) and Wolverine.
Several species of raptors also are known to inhabit the area including known hare predators
such as Great Horned Owls (Rohner and Krebs 1996) and Northern Goshawks (Boutin et al.
1995), as well as other members of the Orders Accipitriformes and Strigiformes. Wolves,
Cougars, Black (Ursus americanus) and Grizzly (Ursus arctos horribilis) Bears also
occasionally prey on hares (Spalding and Lesowski 1971, Floyd et al. 1978, Gable et al.
2018) and have all been observed locally. The American Red Squirrel (Tamiasciurus
hudsonicus) also occurs locally.

19
LITERATURE CITED
Andrén, H. 1995. Effects of landscape composition on predation rates at habitat edges.
Pages 225–255 in L. Hansson, L. Fahrig, and G. Merriam, editors. Mosaic
landscapes and ecological processes. Chapman and Hall, United Kingdom.
Andrén, H., and P. Angelstam. 1988. Elevated predation rates as an edge effect in habitat
islands: experimental evidence. Ecology 69:544–547.
Ausband, D. E., and G. R. Baty. 2005. Effects of precommercial thinning on Snowshoe Hare
habitat use during winter in low-elevation montane forests. Canadian Journal of
Forest Research 35:206–210.
Bedford, L., and R. F. Sutton. 2000. Site preparation for establishing lodgepole pine in the
sub-boreal spruce zone of interior British Columbia: the Bednesti trial, 10-year
results. Forest Ecology and Management 126:227–238.
Berg, N. D., and E. M. Gese. 2010. Relationship between fecal pellet counts and Snowshoe
Hare density in Western Wyoming. The Journal of Wildlife Management 74:1745–
1751.
Berg, N. D., E. M. Gese, J. R. Squires, and L. M. Aubry. 2012. Influence of forest structure
on the abundance of Snowshoe Hares in Western Wyoming. The Journal of Wildlife
Management 76:1480–1488.
Black, H. C. 1965. An analysis of a population of Snowshoe Hares, Lepus americanus
washingtonii Baird, in western Oregon. Master’s thesis, Oregon State University.
Bois, G., L. Imbeau, and M. J. Mazerolle. 2012. Recovery time of Snowshoe Hare habitat
after commercial thinning in boreal Quebec. Canadian Journal of Forest Research
42:123–133.
Boutin, S. 1998. Functional responses of Coyotes and Lynx to the Snowshoe Hare cycle.
Ecology 79:1193–1208.
Boutin, S., and C. J. Krebs. 1986. Estimating survival rates of Snowshoe Hares. The Journal
of Wildlife Management 50:592–594.
Boutin, S., C. J. Krebs, R. Boonstra, M. R. T. Dale, S. J. Hannon, K. Martin, A. R. E.
Sinclair, J. N. M. Smith, R. Turkington, M. Blower, A. Byrom, F. I. Doyle, C. Doyle,
D. Hik, L. Hofer, A. Hubbs, T. Karels, D. L. Murray, V. Nams, M. O’Donoghue, C.

20
Rohner, and S. Schweiger. 1995. Population changes of the vertebrate community
during a Snowshoe Hare cycle in Canada’s boreal forest. Oikos 74:69–80.
Boutin, S., C. J. Krebs, A. R. E. Sinclair, and J. N. M. Smith. 1986. Proximate causes of
losses in a Snowshoe Hare population. Canadian Journal of Zoology 64:606–610.
Breitenmoser, U., B. G. Slough, and C. Breitenmoser-Würsten. 1993. Predators of cyclic
prey: is the Canada Lynx victim or profiteer of the Snowshoe Hare cycle? Oikos
66:551–554.
Brooks, M. 1955. An isolated population of the Virginia varying hare. The Journal of
Wildlife Management 19:54–61.
Brown, J. S., J. W. Laundré, and M. Gurung. 1999. The ecology of fear: optimal foraging,
game theory, and trophic interactions. Journal of Mammalogy 80:385–399.
Burton, A. C., E. Neilson, D. Moreira, A. Ladle, R. Steenweg, J. T. Fisher, E. Bayne, and S.
Boutin. 2015. REVIEW: Wildlife camera trapping: a review and recommendations
for linking surveys to ecological processes. Journal of Applied Ecology 52:675–685.
Cary, J. R., and L. B. Keith. 1979. Reproductive change in the 10-year cycle of Snowshoe
Hares. Canadian Journal of Zoology 57:375–390.
Chaiyarat, R., N. Youngpoy, P. Kongsurakan, S. Nakbun. 2019. Habitat preferences of
reintroduced Banteng (Bos javanicus) into the Salakphra Wildlife Sanctuary,
Thailand. Wildlife Research 46:573–586.
Cheng, E., K. E. Hodges, and L. S. Mills. 2015. Impacts of fire on Snowshoe Hares in
Glacier National Park, Montana, USA. Fire Ecology 11:119–136.
Chisholm, J. D., D. P. Hodder, S. M. Crowley, R. V. Rea, and S. Marshall. 2021. Seasonal
movements of migratory and resident female moose (Alces alces) in north-central
British Columbia, Canada. Mammal Research 66:419–427.
Conroy, M. J., L. W. Gysel, and G. R. Dudderar. 1979. Habitat components of clear-cut
areas for Snowshoe Hares in Michigan. The Journal of Wildlife Management
43:680–690.
Crowley, S. M., D.P. Hodder, and K. Larsen. 2013a. Canada Lynx (Lynx canadensis)
detection and behaviour using remote cameras during the breeding season. Canadian
Field Naturalist 127:310–318.

21
Crowley, S. M., C. J. Johnson, and D. P. Hodder. 2013b. Spatio-temporal variation in river
otter (Lontra canadensis) diet and latrine site activity. Écoscience 20:28–39.
Crowley, S. M., and D. P. Hodder. 2017. An assessment of the efficacy of rub stations for
detection and abundance surveys of Canada Lynx (Lynx canadensis). Canadian
Journal of Zoology 95:885–890.
Crowley, S. M., D. P. Hodder, C. J. Johnson, and D. Yates. 2018. Wildlife health indicators
and mercury exposure: A case study of river otters (Lontra canadensis) in central
British Columbia, Canada. Ecological Indicators 89:63–73.
Di Marco, M., J. E. M. Watson, H. P. Possingham, and O. Venter. 2017. Limitations and
trade-offs in the use of species distribution maps for protected area planning. Journal
of Applied Ecology 54:402–411.
Dodds, D. G. 1965. Reproduction and productivity of Snowshoe Hares in Newfoundland.
The Journal of Wildlife Management 29:303–315.
Ellsworth, E., M. R. Boudreau, K. Nagy, J. L. Rachlow, and D. L. Murray. 2016.
Differential sex-related winter energetics in free-ranging Snowshoe Hare (Lepus
americanus). Canadian Journal of Zoology 94:115–121.
Elton, C., and M. Nicholson. 1942. The ten-year cycle in numbers of the lynx in Canada.
Journal of Animal Ecology 11:215–244.
Environment and Climate Change Canada. 2022. Canadian climate normals 1981-2010
station data, Fort St James, BC. <http://climate.weather.gc.ca/climate_normals>.
Accessed 31 Mar 2020.
Erickson, A. B. 1944. Helminth infections in relation to population fluctuations in Snowshoe
Hares. The Journal of Wildlife Management 8:134–153.
Evans, B. E., C. E. Mosby, and A. Mortelliti. 2019. Assessing arrays of multiple trail
cameras to detect North American mammals. PLOS ONE 14:1–18.
Ewacha, M. V. A., J. D. Roth, and R. K. Brook. 2014. Vegetation structure and composition
determine Snowshoe Hare (Lepus americanus) activity at arctic tree line. Canadian
Journal of Zoology 92:789–794.

22
Feierabend, D., and K. Kielland. 2014. Movements, activity patterns, and habitat use of
Snowshoe Hares (Lepus americanus) in interior Alaska. Journal of Mammalogy
95:525–533.
Feierabend, D., and K. Kielland. 2015. Seasonal effects of habitat on sources and rates of
Snowshoe Hare predation in Alaskan boreal forests. PLOS ONE 10:e0143543.
Ferron, J., C. Dussault, and F. Potvin. 1998. Short-term effects of logging on Snowshoe
Hares in the boreal forest. Canadian Journal of Forest Research 28:1335–1343.
Ferron, J., and J. P. Ouellet. 1992. Daily partitioning of summer habitat and use of space by
the Snowshoe Hare in southern boreal forest. Canadian Journal of Zoology 70:2178–
2183.
Floyd, T. J., L. D. Mech, and P. A. Jordan. 1978. Relating wolf scat content to prey
consumed. The Journal of Wildlife Management 42:528–532.
Foresman, K. R., and D. Pearson. 1999. Activity patterns of American Martens, Fishers,
Snowshoe Hares, and Red Squirrels in westcentral Montana. Canadian Field
Naturalist. 113:386–389.
Fox, J. F., and J. P. Bryant. 1984. Instability of the Snowshoe Hare and woody plant
interaction. Oecologia 63:128–135.
Fretwell, S. D., and H. L. Lucas. 1969. On territorial behavior and other factors influencing
habitat distribution in birds. Acta Biotheoretica 19:16–36.
Frey, J. K., and J. L. Malaney. 2006. Snowshoe Hare (Lepus americanus) and Mountain
Cottontail (Sylvilagus nuttallii) biogeography at their southern range limit. Journal of
Mammalogy 87:1175–1182.
Fuller, A. K., and D. J. Harrison. 2013. Modeling the influence of forest structure on
microsite habitat use by Snowshoe Hares. International Journal of Forestry Research
4:1–7.
Fuller, A. K., D. W. Linden, and J. A. Royle. 2016. Management decision making for Fisher
populations informed by occupancy modeling. The Journal of Wildlife Management
80:794–802.

23
Gable, T. D., S. K. Windels, J. G. Bruggink, and S. M. Barber-Meyer. 2018. Weekly
summer diet of gray Wolves (Canis lupus) in Northeastern Minnesota. American
Midland Naturalist 179:15–27.
Gigliotti, L. C., D. R. Diefenbach, and M. J. Sheriff. 2017. Geographic variation in winter
adaptations of Snowshoe Hares (Lepus americanus). Canadian Journal of Zoology
95:539–545.
Gigliotti, L. C., B. C. Jones, M. J. Lovallo, and D. R. Diefenbach. 2018. Snowshoe Hare
multi-level habitat use in a fire-adapted ecosystem. Journal of Wildlife Management
82:435–444.
Ginzburg, L., and C. Krebs. 2015. Mammalian cycles: internally defined periods and
interaction-driven amplitudes. PeerJ 3:e1180.
Grebner, D. L., P. Bettinger, and J. P. Siry. 2013. Introduction to forestry and natural
resources. Elsevier Incorporated, United Kingdom.
Green, R. G., and C. A. Evans. 1940. Studies on a population cycle of Snowshoe Hares on
the Lake Alexander Area I. gross annual censuses, 1932-1939. The Journal of
Wildlife Management 4:220–238.
Green, R. G., and C. L. Larson. 1938. Shock disease and the Snowshoe Hare cycle. Science
87:298–299.
Griffin, P. C., and L. S. Mills. 2007. Precommercial thinning reduces Snowshoe Hare
abundance in the short term. The Journal of Wildlife Management 71:559–564.
Hasapes, S. K., and C. E. Comer. 2017. White-Tailed Deer fawn survival, home range, and
habitat composition in northwest Louisiana. Wildlife Society Bulletin 41:499–507.
Hodder, D. P., K. W. Larsen, and S. M. Crowley. 2018. Effects of scale and gender on
winter habitat selection by American Mink. The American Midland Naturalist
179:261–274.
Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for Snowshoe Hares.
Forest Ecology and Management 256:1918–1926.
Hodges, K. E., L. S. Mills, and K. M. Murphy. 2009. Distribution and abundance of
Snowshoe Hares in Yellowstone National Park. Journal of Mammalogy 90:870–878.

24
Hodson, J., D. Fortin, and L. Bélanger. 2010a. Fine-scale disturbances shape space-use
patterns of a boreal forest herbivore. Journal of Mammalogy 91:607–619.
Hodson, J., D. Fortin, M.-L. LeBlanc, and L. Bélanger. 2010b. An appraisal of the fitness
consequences of forest disturbance for wildlife using habitat selection theory.
Oecologia 164:73–86.
Holbrook, J. D., R. S. Arkle, J. L. Rachlow, K. T. Vierling, D. S. Pilliod, and M. M. Wiest.
2016. Occupancy and abundance of predator and prey: implications of the firecheatgrass cycle in sagebrush ecosystems. Ecosphere 7:e01307.
Holbrook, J. D., J. R. Squires, L. E. Olson, R. L. Lawrence, and S. L. Savage. 2017.
Multiscale habitat relationships of Snowshoe Hares (Lepus americanus) in the mixed
conifer landscape of the Northern Rockies, USA: cross-scale effects of horizontal
cover with implications for forest management. Ecology And Evolution 7:125–144.
Homyack, J. A., D. J. Harrison, and W. B. Krohn. 2007. Effects of precommercial thinning
on Snowshoe Hares in Maine. The Journal of Wildlife Management 71:4–13.
Homyack, J. A., D. J. Harrison, J. A. Litvaitis, and W. B. Krohn. 2006. Quantifying
densities of Snowshoe Hares in Maine using pellet plots. Wildlife Society Bulletin
34:74–80.
Ivan, J. S., G. C. White, and T. M. Shenk. 2014. Density and demography of Snowshoe
Hares in Central Colorado. The Journal of Wildlife Management 78:580–594.
Jacob, J. 2008. Response of small rodents to manipulations of vegetation height in agroecosystems. Integrative Zoology 3:3–10.
Jensen, P. O., A. J. H. Meddens, S. Fisher, A. J. Wirsing, D. L. Murray, and D. H. Thornton.
2021. Broaden your horizon: the use of remotely sensed data for modeling
populations of forest species at landscape scales. Forest Ecology and Management
500:119640.
Jensen, P. O., A. J. Wirsing, and D. H. Thornton. 2022. Using camera traps to estimate
density of Snowshoe Hare (Lepus americanus): a keystone boreal forest herbivore.
Journal of Mammalogy 103:693–710.
Kawaguchi, T., and A. Desrochers. 2018. A time-lagged effect of conspecific density on
habitat selection by Snowshoe Hare. PLOS ONE 13:e0190643.

25
Keith, L. B. 1964. Daily activity pattern of Snowshoe Hares. Journal of Mammalogy
45:626–627.
Keith, L. B. 1966. Habitat vacancy during a Snowshoe Hare decline. The Journal of Wildlife
Management 30:828–832.
Keith, L. B. 1983. Role of food in hare population cycles. Oikos 40:385–395.
Keith, L. B., J. R. Cary, T. M. Yuill, and I. M. Keith. 1985. Prevalence of heleminths in a
cyclic Snowshoe Hare population. Journal of Wildlife Diseases 21:233–253.
Keith, L. B., and L. A. Windberg. 1978. A demographic analysis of the Snowshoe Hare
cycle. Wildlife Monographs 58:3–70.
Krebs, C. J., R. Boonstra, and S. Boutin. 2018. Using experimentation to understand the 10year Snowshoe Hare cycle in the boreal forest of North America. Journal of Animal
Ecology 87:87–100.
Krebs, C. J., R. Boonstra, S. Boutin, and A. R. E. Sinclair. 2001a. What drives the 10-year
cycle of Snowshoe Hares? The ten-year cycle of Snowshoe Hares, one of the most
striking features of the boreal forest, is a product of the interaction between predation
and food supplies, as large-scale experiments in the Yukon have demonstrated.
BioScience 51:25–35.
Krebs, C. J., R. Boonstra, V. Nams, M. O’Donoghue, K. E. Hodges, and S. Boutin. 2001b.
Estimating Snowshoe Hare population density from pellet plots: a further evaluation.
Canadian Journal of Zoology 79:1–4.
Krebs, C. J., S. Boutin, R. Boonstra, A. R. E. Sinclair, J. N. M. Smith, M. R. T. Dale, K.
Martin, and R. Turkington. 1995. Impact of food and predation on the Snowshoe
Hare cycle. Science 269:1112–1115.
Krebs, C. J., B. S. Gilbert, S. Boutin, and R. Boonstra. 1987. Estimation of Snowshoe Hare
population density from turd transects. Canadian Journal of Zoology 65:565–567.
Krebs, C. J., B. S. Gilbert, S. Boutin, A. R. E. Sinclair, and J. N. M. Smith. 1986. Population
biology of Snowshoe Hares. I. Demography of food-supplemented populations in the
Southern Yukon, 1976-84. Journal of Animal Ecology 55:963–982.
Kuvlesky, W. P., and L. B. Keith. 1983. Demography of Snowshoe Hare populations in
Wisconsin. Journal of Mammalogy 64:233–244.

26

Laundré, J. W., L. Hernández, and K. B. Altendorf. 2001. Wolves, Elk, and Bison:
reestablishing the “landscape of fear” in Yellowstone National Park, U.S.A.
Canadian Journal of Zoology 79:1401–1409.
Lewis, C. W., K. E. Hodges, G. M. Koehler, and L. S. Mills. 2011. Influence of stand and
landscape features on Snowshoe Hare abundance in fragmented forests. Journal of
Mammalogy 92:561–567.
Linkie, M., G. Guillera-Arroita, J. Smith, A. Ario, G. Bertagnolio, F. Cheong, G. R.
Clements, Y. Dinata, S. Duangchantrasiri, G. Fredriksson, M. T. Gumal, L. S.
Horng, K. Kawanishi, F. R. Khakim, M. F. Kinnaird, D. Kiswayadi, A. H. Lubis, A.
J. Lynam, Maryati, M. Maung, D. Ngoprasert, W. Novarino, T. G. O’Brien, K.
Parakkasi, H. Peters, D. Priatna, D. M. Rayan, N. Seuaturien, N. M. Shwe, R.
Steinmetz, A. M. Sugesti, Sunarto, M. E. Sunquist, M. Umponjan, H. T. Wibisono,
C. C. T. Wong, and Zulfahmi. 2013. Cryptic mammals caught on camera: Assessing
the utility of range wide camera trap data for conserving the endangered Asian Tapir.
Biological Conservation 162:107–115.
Litvaitis, J. A., J. A. Sherburne, and J. A. Bissonette. 1985. Influence of understory
characteristics on Snowshoe Hare habitat use and density. The Journal of Wildlife
Management 49:866–873.
Luttich, S., D. H. Rusch, E. C. Meslow, and L. B. Keith. 1970. Ecology of Red-Tailed Hawk
predation in Alberta. Ecology 51:190–203.
MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. Andrew Royle, and C. A.
Langtimm. 2002. Estimating site occupancy rates when detection probabilities are
less than one. Ecology 83:2248–2255.
MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines.
2006. Occupancy estimation and modeling: inferring patterns and dynamics of
species occurrence. Academic Press, Cambridge, USA.
Majchrzak, Y. N., M. J. L. Peers, E. K. Studd, A. K. Menzies, P. D. Walker, S. Shiratsuru,
L. K. McCaw, R. Boonstra, M. Humphries, T. S. Jung, A. J. Kenney, C. J. Krebs, D.
L. Murray, S. Boutin, and A. Sih. 2022. Balancing food acquisition and predation
risk drives demographic changes in Snowshoe Hare population cycles. Ecology
Letters 25:981–991.

27
Malaney, J. L., and J. K. Frey. 2006. Summer habitat use by Snowshoe Hare and Mountain
Cottontail at their southern zone of sympatry. The Journal of Wildlife Management
70:877–883.
McCann, N. P., R. Moen, and G. Niemi. 2008. Using pellet counts to estimate Snowshoe
Hare numbers in Minnesota. The Journal of Wildlife Management 72:955.
McKelvey, K. S., G. W. McDaniel, L. S. Mills, and P. C. Griffin. 2002. Effects of plot size
and shape on pellet density estimates for Snowshoe Hares. Wildlife Society Bulletin
30:751–755.
Meek, P. D., G. Ballard, A. Claridge, R. Kays, K. Moseby, T. O’Brien, A. O’Connell, J.
Sanderson, D. E. Swann, M. Tobler, and S. Townsend. 2014. Recommended guiding
principles for reporting on camera trapping research. Biodiversity and Conservation
23:2321–2343.
Meidinger, D., and J. Pojar. 1991. Special report series 6: ecosystems of British Columbia.
Ministry of Forests, Victoria, Canada.
Meslow, E. C., and L. B. Keith. 1968. Demographic parameters of a Snowshoe Hare
population. The Journal of Wildlife Management 32:812–834.
Mills, L. S., E. V. Bragina, A. V. Kumar, M. Zimova, D. J. R. Lafferty, J. Feltner, B. M.
Davis, K. Hackländer, P. C. Alves, J. M. Good, J. Melo-Ferreira, A. Dietz, A. V.
Abramov, N. Lopatina, and K. Fay. 2018. Winter color polymorphisms identify
global hot spots for evolutionary rescue from climate change. Science 359:1033–
1036.
Mills, L. S., P. C. Griffin, K. E. Hodges, K. McKelvey, L. Ruggiero, and T. Ulizio. 2005.
Pellet count indices compared to mark-recapture estimates for evaluating Snowshoe
Hare density. Journal of Wildlife Management 69:1053–1062.
Mordecai, R. S., B. J. Mattsson, C. J. Tzilkowski, and R. J. Cooper. 2011. Addressing
challenges when studying mobile or episodic species: hierarchical Bayes estimation
of occupancy and use. Journal of Applied Ecology 48:56–66.
Morris, D. W. 2019. Contingent strategies of risk management by Snowshoe Hares.
FACETS 4:407–422.

28
Morris, D. W., and S. Vijayan. 2018. Trade-offs between sight lines and escape habitat
determine spatial strategies of risk management by a keystone herbivore. FACETS
338–357.
Murray, D. L. 2002. Differential body condition and vulnerability to predation in Snowshoe
Hares. Journal of Animal Ecology 71:614–625.
Myers, J. H. 2018. Population cycles: generalities, exceptions and remaining mysteries.
Proceedings of the Royal Society B: Biological Sciences 285:20172841.
Neff, D. J. 1968. The pellet-group count technique for big game trend, census, and
distribution: a review. The Journal of Wildlife Management 32:597–614.
Neilson, E. W., T. Avgar, A. C. Burton, K. Broadley, and S. Boutin. 2018. Animal
movement affects interpretation of occupancy models from camera-trap surveys of
unmarked animals. Ecosphere 9:e02092.
Newbury, T. L., and N. P. P. Simon. 2005. The effects of clearcutting on Snowshoe Hare
(Lepus americanus) relative abundance in central Labrador. Forest Ecology and
Management 210:131–142.
Newey, S., P. Davidson, S. Nazir, G. Fairhurst, F. Verdicchio, R. J. Irvine, and R. van der
Wal. 2015. Limitations of recreational camera traps for wildlife management and
conservation research: a practitioner’s perspective. Ambio 44:624–635.
Ngoprasert, D., A.J. Lynam, R. Sukmasuang, B. Tantipisanuh, W. Chutipong, R. Steinmetz,
K.E. Jenks, G.A. Gale, L.I. Grassman Jr., S. Kitamura, J. Howard, Pa. Cutter, Pe.
Cutter, P. Leimguber, N. Songsasen, D.H. Reed. 2012. Occurrence of three felids
across a network of protected areas in Thailand: prey, intraguild, and habitat
associations. Biotropica 44:810–817.
O’Connor, K. M., L. R. Nathan, M. R. Liberati, M. W. Tingley, J. C. Vokoun, and T. A. G.
Rittenhouse. 2017. Camera trap arrays improve detection probability of wildlife:
Investigating study design considerations using an empirical dataset. PLOS ONE
12:e0175684.
O’Donoghue, M. 1994. Early survival of juvenile Snowshoe Hares. Ecology 75:1582–1592.
O’Donoghue, M., and S. Boutin. 1995. Does reproductive synchrony affect juvenile survival
rates of northern mammals. Oikos 74:115–121.

29
O’Donoghue, M., S. Boutin, C. J. Krebs, and E. J. Hofer. 1997. Numerical responses of
Coyotes and Lynx to the Snowshoe Hare cycle. Oikos 80:150–162.
O’Donoghue, M., and C. J. Krebs. 1992. Effects of supplemental food on Snowshoe Hare
reproduction and juvenile growth at a cyclic population peak. Journal of Animal
Ecology 61:631–641.
Pease, J. L., R. H. Vowles, and L. B. Keith. 1979. Interaction of Snowshoe Hares and woody
vegetation. The Journal of Wildlife Management 43:43–60.
Peers, M. J. L., J. R. Reimer, Y. N. Majchrzak, A. K. Menzies, E. K. Studd, R. Boonstra, A.
J. Kenney, C. J. Krebs, M. O’Donoghue, and S. Boutin. 2021. Contribution of latelitter juveniles to the population dynamics of Snowshoe Hares. Oecologia 195:949–
957.
Peers, M. J., Y. N. Majchrzak, S. M. Konkolics, R. Boonstra, and S. Boutin. 2018.
Scavenging by Snowshoe Hares (Lepus americanus) in Yukon, Canada.
Northwestern Naturalist 99:232–235.
Ranta, E., J. Lindstrom, V. Kaitala, H. Kokko, H. Linden, and E. Helle. 1997. Solar activity
and hare dynamics: a cross-continental comparison. The American Naturalist
149:765–775.
Rich, L. N., C. L. Davis, Z. J. Farris, D. A. W. Miller, J. M. Tucker, S. Hamel, M. S.
Farhadinia, R. Steenweg, M. S. D. Bitetti, K. Thapa, M. D. Kane, S. Sunarto, N. P.
Robinson, A. Paviolo, P. Cruz, Q. Martins, N. Gholikhani, A. Taktehrani, J.
Whittington, F. A. Widodo, N. G. Yoccoz, C. Wultsch, B. J. Harmsen, and M. J.
Kelly. 2017. Assessing global patterns in mammalian carnivore occupancy and
richness by integrating local camera trap surveys. Global Ecology and Biogeography
26:918–929.
Rohner, C., and C. J. Krebs. 1996. Owl predation on Snowshoe Hares: consequences of
antipredator behaviour. Oecologia 108:303–310.
Rongstad, O. J., and J. R. Tester. 1971. Behavior and maternal relations of young Snowshoe
Hares. The Journal of Wildlife Management 35:338–346.
Rowcliffe, J. M., J. Field, S. T. Turvey, and C. Carbone. 2008. Estimating animal density
using camera traps without the need for individual recognition. Journal of Applied
Ecology 45:1228–1236.

30
Roy, C., L. Imbeau, and M. J. Mazerolle. 2010. Transformation of abandoned farm fields
into coniferous plantations: is there enough vegetation structure left to maintain
winter habitat of Snowshoe Hares? Canadian Journal of Zoology 88:579–588.
Royle, J. A., and J. D. Nichols. 2003. Estimating abundance from repeated presence–
absence data or point counts. Ecology 84:777–790.
Russell, J. E., and R. Tumlison. 1996. Comparison of microstructure of white winter fur and
brown summer fur of some arctic mammals. Acta Zoologica 77:279–282.
Schmidt, J. H., C. L. McIntyre, C. A. Roland, M. C. MacCluskie, and M. J. Flamme. 2018.
Bottom‐up processes drive reproductive success in an apex predator. Ecology &
Evolution 8:1833–1841.
Seip, C. R., D. P. Hodder, S. M. Crowley, and C. J. Johnson. 2018. Use of constructed
coarse woody debris corridors in a clearcut by American Martens (Martes
americana) and their prey. Forestry: An International Journal of Forest Research
91:506–513.
Severaid, J. H. 1942. The Snowshoe Hare: its life history and artificial propagation. Maine
Department of Inland Fisheries and Game, Augusta, USA.
Severaid, J. H. 1945. Breeding potential and artificial propagation of the Snowshoe Hare.
The Journal of Wildlife Management 9:290–295.
Sharpe, P., and D. J. Undersander. 2019. Co-existing with wildlife. Pages 329–354 in P.
Sharpe, editor. Horse pasture management. Elsevier Incorporated, United Kingdom.
Shannon, G., J. S. Lewis, and B. D. Gerber. 2014. Recommended survey designs for
occupancy modelling using motion-activated cameras: insights from empirical
wildlife data. PeerJ 2:e532.
Sheriff, M. J., C. J. Krebs, and R. Boonstra. 2009. The sensitive hare: sublethal effects of
predator stress on reproduction in Snowshoe Hares. Journal of Animal Ecology
78:1249–1258.
Sheriff, M. J., C. J. Krebs, and R. Boonstra. 2010. The ghosts of predators past: population
cycles and the role of maternal programming under fluctuating predation risk.
Ecology 91:2983–2994.

31
Sheriff, M. J., C. J. Krebs, and R. Boonstra. 2011. From process to pattern: how fluctuating
predation risk impacts the stress axis of Snowshoe Hares during the 10-year cycle.
Oecologia 166:593–605.
Simard, V., L. Imbeau, and H. Asselin. 2018. Effects of selection cuts on winter habitat use
of Snowshoe Hare (Lepus americanus) in northern temperate forests. Canadian
Journal of Forest Research 48:1049–1057.
Sinclair, A. R. E., and J. M. Gosline. 1997. Solar activity and mammal cycles in the
Northern Hemisphere. The American Naturalist 149:776–784.
Sinclair, A. R. E., J. M. Gosline, G. Holdsworth, C. J. Krebs, S. Boutin, J. N. M. Smith, R.
Boonstra, and M. Dale. 1993. Can the solar cycle and climate synchronize the
Snowshoe Hare cycle in Canada? Evidence from tree rings and ice cores. The
American Naturalist 141:173–198.
Sinclair, A. R. E., C. J. Krebs, J. N. M. Smith, and S. Boutin. 1988. Population biology of
Snowshoe Hares. III. Nutrition, plant secondary compounds and food limitation.
Journal of Animal Ecology 57:787–806.
Sinclair, A. R. E., and N. M. Smith. 1984. Do plant secondary compounds determine feeding
preferences of Snowshoe Hares? Oecologia 61:403–410.
Smith, J. N. M., C. J. Krebs, A. R. E. Sinclair, and R. Boonstra. 1988. Population biology of
Snowshoe Hares. II. Interactions with winter food plants. Journal of Animal Ecology
57:269–286.
Spalding, D. J., and J. Lesowski. 1971. Winter food of the cougar in south-central British
Columbia. The Journal of Wildlife Management 35:378–381.
Stefan, C. I., and C. J. Krebs. 2001. Reproductive changes in a cyclic population of
Snowshoe Hares. Canadian Journal of Zoology 79:2101–2108.
Stirnemann, I., A. Mortelliti, P. Gibbons, and D. B. Lindenmayer. 2015. Fine-scale habitat
heterogeneity influences occupancy in terrestrial mammals in a temperate region of
Australia. PLOS ONE 10:e0138681.
St-Laurent, M-H., M. Cusson, J. Ferron, and A. Caron. 2008. Use of residual forest by
Snowshoe Hare in a clear-cut boreal landscape. Northeastern Naturalist 15:497–514.

32
Strong, W. L., and T. S. Jung. 2012. Stand-level attributes of Snowshoe Hare (Lepus
americanus) habitat in a post-fire trembling aspen (Populus tremuloides)
chronosequence in Central Yukon. Canadian Field Naturalist 126:295–305.
Studd, E. K., M. R. Boudreau, Y. N. Majchrzak, A. K. Menzies, M. J. L. Peers, J. L. Seguin,
S. G. Lavergne, R. Boonstra, D. L. Murray, S. Boutin, and M. M. Humphries. 2019.
Use of acceleration and acoustics to classify behavior, generate time budgets, and
evaluate responses to moonlight in free-ranging Snowshoe Hares. Frontiers in
Ecology and Evolution 7.
Suffice, P., G. Joanisse, L. Imbeau, M. J. Mazerolle, and G. Lessard. 2015. Short-term
effects of irregular shelterwood cutting on yellow birch regeneration and habitat use
by Snowshoe Hare. Forest Ecology and Management 354:160–169.
Sullivan, T. P. 1994. Influence of herbicide-induced habitat alteration on vegetation and
Snowshoe Hare populations in sub-boreal spruce forest. Journal of Applied Ecology
31:717–730.
Sullivan, T. P., and D. S. Sullivan. 1988. Influence of stand thinning on Snowshoe Hare
population dynamics and feeding damage in lodgepole pine forest. Journal of
Applied Ecology 25:791–805.
Sullivan, T. P., D. S. Sullivan, P. M. F. Lindgren, and D. B. Ransome. 2006. Influence of
repeated fertilization on forest ecosystems: relative habitat use by Snowshoe Hares
(Lepus americanus). Canadian Journal of Forest Research 36:2080–2089.
Sullivan, T. P., D. S. Sullivan, P. M. F. Lindgren, and D. B. Ransome. 2007. Long-term
responses of ecosystem components to stand thinning in young lodgepole pine forest:
IV. Relative habitat use by mammalian herbivores. Forest Ecology and Management
240:32–41.
Sullivan, T. P., D. S. Sullivan, P. M. F. Lindgren, and D. B. Ransome. 2012. Silviculture and
wildlife: Snowshoe Hare abundance across a successional sequence of natural and
intensively managed forests. International Scholarly Research Notices 2012:593103.
Sultaire, S. M., J. N. Pauli, K. J. Martin, M. W. Meyer, and B. Zuckerbeg. 2016. Extensive
forests and persistent snow cover promote Snowshoe Hare occupancy in Wisconsin.
The Journal of Wildlife Management 80:894–905.
Sundell, J., C. Church, O. Ovaskainen. 2012. Spatio-temporal patterns of habitat use in voles
and shrews modified by density, season and predators. 81:747–755.

33
Telfer, E. S. 1972. Browse selection by deer and hares. The Journal of Wildlife Management
36:1344–1349.
Thomas, J. P., M. L. Reid, R. M. R. Barclay, and T. S. Jung. 2019. Salvage logging after an
insect outbreak reduces occupancy by Snowshoe Hares (Lepus americanus) and their
primary predators. Global Ecology and Conservation 17:e00562.
Thornton, D. H., A. J. Wirsing, J. D. Roth, and D. L. Murray. 2013. Habitat quality and
population density drive occupancy dynamics of Snowshoe Hare in variegated
landscapes. Ecography 36:610–621.
Tyson, R., S. Haines, and K. E. Hodges. 2010. Modelling the Canada Lynx and Snowshoe
Hare population cycle: the role of specialist predators. Theoretical Ecology 3:97–
111.
van Beest, F. M., P. D. McLoughlin, E. Vander Wahl, and R. K. Brook. 2014. Densitydependent habitat selection and partitioning between two sympatric ungulates.
Oecologia 175:1155–1165.
Vaughan, M. R., and L. B. Keith. 1981. Demographic response of experimental Snowshoe
Hare populations to overwinter food shortage. The Journal of Wildlife Management
45:354–380.
Vik, J. O., C. N. Brinch, S. Boutin, and N. C. Stenseth. 2008. Interlinking hare and lynx
dynamics using a century’s worth of annual data. Population Ecology 50:267–274.
Wali, M. K., and V. J. Krajina. 1973. Vegetation-environment relationships of some subboreal spruce zone ecosystems in British Columbia. Vegetatio 26:237–381.
Weir, R. D., H. Davis, and C. Hoodicoff. 2003. Conservation strategies for North American
Badgers in the Thompson and Okanagan regions. Habitat Conservation Trust Fund
and Artemis Wildlife Consultants, Armstrong, Canada.
White, G. C., and L. E. Eberhardt. 1980. Statistical analysis of Deer and Elk pellet-group
data. The Journal of Wildlife Management 44:121–131.
Wilson, E. C., B. Zuckerberg, M. Z. Peery, and J. N. Pauli. 2020. The past, present and
future impacts of climate and land use change on Snowshoe Hares along their
southern range boundary. Biological Conservation 249:108731.

34
Wolfe, M. L., N. V. Debyle, C. S. Winchell, and T. R. McCabe. 1982. Snowshoe Hare cover
relationships in Northern Utah. The Journal of Wildlife Management 46:662–670.
Wolff, J. O. 1980. The role of habitat patchiness in the population dynamics of Snowshoe
Hares. Ecological Monographs 50:111–130.
Zhang, G., A.-X. Zhu, S. K. Windels, and C.-Z. Qin. 2018. Modelling species habitat
suitability from presence-only data using kernel density estimation. Ecological
Indicators 93:387–396.
Zimova, M., A. P. K. Sirén, J. J. Nowak, A. M. Bryan, J. S. Ivan, T. L. Morelli, S. L. Suhrer,
J. Whittington, L. S. Mills, and P. Morellato. 2020. Local climate determines
vulnerability to camouflage mismatch in Snowshoe Hares. Global Ecology &
Biogeography 29:503–515.

35

CHAPTER 2: DENSITY-DEPENDENT HABITAT USE IN A CYCLICAL SNOWSHOE
HARE (LEPUS AMERICANUS) POPULATION IN THE SUB-BOREAL FORESTS OF
NORTH-CENTRAL BRITISH COLUMBIA.
INTRODUCTION
Species and populations have different habitat requirements for survival and
reproduction that cause their densities to vary across different habitat types (Stirnemann et
al. 2015, Zhang et al. 2018). Consequently, many ecological studies have assessed habitat
quality in an attempt to explain wildlife distribution, abundance or occupancy across a
landscape (Royle and Nichols 2003, Weir et al. 2003, Fuller et al. 2016). Modeling species
detection and habitat data allows for the assessment of relative use and identification of
important habitat components (MacKenzie et al. 2002, 2006). This common type of study
can inform wildlife managers by identifying high-quality or important habitats for protection
or management action (MacKenzie et al. 2002, Di Marco et al. 2017).
For species or local populations that cycle in abundance over time, assessing habitat
use becomes more challenging. Studies completed during population peaks may yield
different results and interpretations compared to those undertaken during population lows
and long-term studies across many predictable population cycles may offer the greatest
insights (e.g., the Southern Yukon hare studies – reviewed in Krebs et al. 2018)
Additionally, the amplitude of cycles may vary over time (Ginzburg and Krebs 2015, Myers
2018) furthering complicating the issue of when to study these populations.
The population cycle of Snowshoe Hares (Lepus americanus) is one of the most well
documented and studied ecological phenomena in the world (Meslow and Keith 1968, Keith
and Windberg 1978, Krebs et al. 1986). Across much of their range, populations of this
species are known to fluctuate on a predictable 8 – 11-year interval with varying amplitudes
(Krebs et al. 2014). Snowshoe Hares account for a large proportion of the vertebrate prey
biomass in the boreal and sub-boreal forests of Canada and thus are critical links in the
complex food webs of these ecosystems (Krebs et al. 2018). Canada Lynx (Lynx canadensis)
specialize on Snowshoe Hares, and populations of this small felid cycle in accordance with
hare populations over much of their shared range (Breitenmoser et al. 1993, O’Donoghue et

36
al. 1997, 1998). Predation has been recently argued to be the main driver of the Snowshoe
Hare cycle (Krebs et al. 2018).
Many studies have evaluated Snowshoe Hare habitat use across their range and
cover, particularly horizontal/understory cover, is often considered the most important
predictor of use (Litvaitis et al. 1985, Thornton et al. 2013, Simard et al. 2018). Horizontal
cover provides visual obstruction so that hares can feed and rest without being easily seen
and preyed upon by terrestrial predators. Canopy closure also may be important as it
provides protection from avian predators (Gigliotti et al. 2018, Thomas et al. 2019); this is
significant given raptors can be responsible for up to 40% of predatory losses (Southern
Yukon – see Krebs et al. 1995). Other habitat features that may be correlated with Snowshoe
Hare habitat use include tree density (Thomas et al. 2019) or volume, distance to edge
habitat (Roy et al. 2010) and other regionally significant factors.
Although it is known that Snowshoe Hares exhibit density-dependent habitat use
(Wolff 1980, Hodson et al. 2010) few studies have investigated how hare habitat use
changes throughout a population cycle. Wolff (1980) found that when hare populations were
low, the remaining individuals only occupied the best quality habitat, usually areas with
dense vegetation, but during a population high they occupied a wider array of available
habitats. This type of density-dependent habitat use and occupancy of suboptimal habitat at
population peaks is predicted by the “Ideal Free Distribution” theory (Fretwell and Lucas
1969). If during population lows Snowshoe Hares retreat to high quality refugia those
habitats may be important when managing for hares and their predators. If habitat use does
change during the cycle, then the timing of population surveys is important. Studies
undertaken at different points during a population cycle may yield different and even
conflicting results and interpretations. Short-lived occupancy patterns (i.e., patch
colonization and extinction) also make it harder to determine what is or is not hare habitat
(Thornton et al. 2013) and this is why there is a need to better understand how habitat use
changes with population cycles.
Most Snowshoe Hare studies have been at landscape- or stand-level and only a few
have examined within-stand or plot-level habitat use (Ausband and Baty 2005, Fuller and
Harrison 2013, Holbrook et al. 2017); therefore, there is still a need to better understand hare
habitat use at a fine-scale as it is known to vary across ecological scales (Lewis et al. 2011,

37
Fuller and Harrison 2013, Gigliotti et al. 2018). For example, horizontal cover has been
found to drive hare within-stand habitat selection while other factors may influence standlevel selection (Holbrook et al. 2017, Gigliotti et al. 2018). This is why studies undertaken at
multiple scales can be important in determining the full scope of hare habitat use (Thornton
et al. 2013, Holbrook et al. 2017).
I assessed Snowshoe Hare habitat use in the sub-boreal forests of north-central
British Columbia (BC), Canada, using habitat metrics known to be of importance to hares
(e.g., canopy closure, tree volume, distance to edge). I conducted this work using camera
trapping (CT) and pellet counts, in a region where Snowshoe Hares exist alongside a diverse
community of mesocarnivores including mustelid, felid and canid mammals, as well as
raptors. I used cameras partly to assess their efficacy in detecting hare population changes
and also carried out pellet counts to confirm population oscillations detected by the cameras
and to compare and contrast the results of the two methods. An additional motivation for
adding in pellet surveys was to be able to compare this study with existing hare habitat use
studies that have primarily used pellet surveys as their detection and/or enumeration method.
My main objective was to determine what habitat components were most important
to hares in the region and to develop models using fine-scale habitat covariates that could
predict habitat use. I also wanted to assess and compare habitat use at the landscape- and
within-stand-scale. An additional objective was to determine if the local hare population
oscillated in abundance over time. Given that the landscape was highly fragmented by
forestry, a condition suggested to lead to non-cyclical hare populations (Wirsing et al. 2002),
it was unknown whether the local hare population followed a typical cyclical pattern. The
hare cycle can have far reaching impacts on other species and the ecosystem as a whole and
it was important to know to what extent the local population cycled over time. If a
population oscillation was detected, I wanted to assess whether there was a concomitant
density-dependent change in habitat use.
I developed several hypotheses based on commonalities that ran through existing
hare habitat literature. I predicted hare use would increase with increasing horizontal cover
and canopy closure. I also expected hare use to increase with tree volume (i.e., basal area)
and I additionally assumed that hare use would increase closer to habitat edges. Since
vegetation is usually denser in riparian areas, I predicted hare use would increase closer to

38
them. Finally, I assumed that, if hare occupancy were to increase or decrease appreciably, I
would observe a density-dependent change in habitat use.
Snowshoe Hare habitat use has been well studied but some gaps in our knowledge
still exist including how habitat use inferences change with scale or at varying hare densities
throughout their population cycles. These data were collected over a period of time when
Snowshoe Hare occupancy ranged considerably across the landscape allowing me to assess
density-dependent habitat use. I also evaluated hare habitat use at the within-stand-scale,
which has gone understudied, and was able to compare this to landscape-scale models to
assess hare habitat use at multiple scales. Thus, this work adds to our understanding of
Snowshoe Hare habitat use across landscapes and fluctuations of population occupancy or
density.

METHODS

Study Site
In this study I utilized Snowshoe Hare detection data from 2015 to 2020 in the John
Prince Research Forest (JPRF) and neighbouring provincial lands, approximately 45 km
north-east of Fort St. James, BC (54° 40’ 14” N, 124° 25’ 13” W; Figure 2.1, also see Figure
1.1). This region occurs within the Sub-Boreal Spruce biogeoclimatic zone of the province
and experiences cool summers and long cold winters (Meidinger and Pojar 1991). Over 70
years of forestry has altered the landscape, creating a patchwork of seral stages
predominantly composed of regenerating-to-mature conifer stands of hybrid white spruce
(Picea x albertiana), subalpine fir (Abies lasiocarpa), lodgepole pine (Pinus contorta) and
Interior Douglas-fir (Pseudotsuga menziesii var. glauca; Crowley and Hodder 2017).
Smaller deciduous-predominant patches of trembling aspen (Populus tremuloides), paper
birch (Betula papyrifera) and poplar (Populus balsamifera) also occur, as well as newly
harvested areas. A dense understory of willows, alders and other shrubs is found throughout
most of the study area (Crowley and Hodder 2017; See Chapter 1 for species description).
There also is a diverse mesocarnivore community that includes Canada Lynx, Marten
(Martes americana), Fisher (Pekania pennanti), Wolverine (Gulo gulo), Coyote (Canis

39
latrans), Red Fox (Vulpes vulpes), American Mink (Neogale vison) as well as several
species of avian predators (Orders Accipitriformes and Strigiformes). Known hare leveret
predators, American Red Squirrels (Tamiasciurus hudsonicus) are common in the area.

40

Figure 2.1. Study area with the locations of the 66 camera trapping stations and 10 pellet
sites used for the five-year (2015 – 2020) Snowshoe Hare (Lepus americanus) habitat study
in the John Prince Research Forest (JPRF) and surrounding area in north-central, British
Columbia, Canada (UTM Zone 10 – NAD83).

41
Snowshoe Hare Detection Data

Camera Data
Researchers at the JPRF divided the study area into a grid of 66 hexagons with 2.5
km between the center of each 5.41 km2 hexagon, creating a 357 km2 study area. They
positioned one digital infrared trail camera at the center of each hexagon; if this was not
possible (e.g., centre was in the middle of a stream) they chose the nearest suitable location
to establish a camera station. In 2015 and 2016 they deployed Bushnell Trophy Cameras
(models 119467 and 11947; Bushnell Outdoor Products, Missouri, USA). They programmed
the cameras to record a 30-second video when motion was detected, with a one-second
interval between videos. In 2020 they switched out all cameras for Browning Dark Ops HD
Pro model BTC-6HDP (Browning, Utah, USA) with video lengths of 10 seconds, with a
one-second interval between videos or continuous recording if motion was detected.
Snowshoe Hares are nocturnal and/or crepuscular (Feierabend and Kielland 2014,
Studd et al. 2019); therefore, I considered my independent sampling event to be one cameraday: the 24-hour period of active camera recording starting at 12:00:00 and ending at
11:59:59 the following calendar day. Doing this fully encompassed one active period rather
than splitting a single active period over two sampling events. I used camera data from
January 20 in 2015 and from January 21 in 2016 and 2020 until March 31st to create a 70
camera-day study period for each year, while accommodating for leap years. I reviewed the
camera footage and recorded whether a hare was detected or not during each camera-day. I
also gathered detection data over a single 26-week spring-summer session from 12:00:00 on
April 1, 2016 to 11:59:59 on September 30, 2016 at 49 camera stations. Staff at JPRF
accessed camera stations by snowmobile in winter. During spring and summer, they could
not access all 66 stations due to changing conditions (e.g., ephemeral streams and wetlands)
so for my analyses I only included data from cameras that were functioning for a minimum
of 75% of camera-days.
Since all camera stations were not equally surveyed due to camera malfunctions,
battery issues, etc., I used the detection rate (camera-days with hare detections/camera-days
per camera station) to determine whether hares were present or absent at each station. I used

42
detection rates greater than the 50th percentile of all detection rates for the first year (2015 =
0.04) to set a threshold for classifying hares as present (1) at a station, whereas hares at
stations with detection rates equal to or less than 0.04 were classified as absent (0). This
allowed me to distinguish between stations with moderate to high detection rates, likely
because the station fell within an individual’s home range (i.e., suggesting some level of
habitat use) and stations where hares were completely absent or had low detection rates,
suggesting less and possibly only temporary use (e.g., a foray out of its home range for
predator avoidance). In addition to using the above system to create a binary database (1 =
present, 0 = absent for each camera-day), I also calculated occupancy (stations with hare
present/total number of stations), sometimes referred to as naïve occupancy as it does not
incorporate detection probabilities (Nichols et al. 2007), to compare trends between study
years.
Pellet Data
I established a total of 10 hare pellet sites in a variety of forest types (e.g., mature
forests (>50 years old), regenerating forests (15 – 50 years old), clearcuts (<15 years old))
across the study area from 2017 – 2019. Each pellet site consisted of four transects located
100 m apart; each transect was composed of 25 pellet plots, located 15 m apart, for a total of
100 pellet plots per site. Each plot was positioned directionally (first plot = north-south, the
second = northeast-southwest, third = east-west, fourth = northwest-southeast). I cleared
each 0.155 m2 (305 cm x 5 cm) rectangular plot (Krebs et al. 2001, Hodges and Mills 2008)
of all pellets when I first established six sites in fall 2017 and added an additional four sites
in spring 2018. Following establishment of each site, I counted pellets in spring and fall of
2019 and 2020. I also counted pellets at six of the 10 sites in 2018, which I used to estimate
density for each site but did not include in my models. I established these pellet sites to
provide a way to corroborate population trends or habitat use conclusions drawn from the
camera data.
I removed pellets from each plot after counting to avoid recounting. I combined
spring and fall pellet count totals to estimate density over a one-year time span that ranged
from the clearing of pellets in fall of one year to the counting of pellets in fall of the

43
subsequent year (e.g., the 2019 study year went from the clearing of pellets in September
2018 to the second counting of pellets in September 2019, representing the total of the
spring and fall 2019 counts). This ensured the results were comparable with other pellet
studies where a one-year sampling period is most common (McCann et al. 2008). I obtained
forest stand age estimates for each pellet site from the publicly available Vegetation
Resource Inventory (VRI; Ministry of Forests, BC) and confirmed them with timber harvest
records from the JPRF.
In order to estimate hare density from the pellet counts I used the revised “Yukon
Equation” (hares/ha = -1.203 + 0.889 loge (mean no. of pellets) x 1.567; Krebs et al. 2001).
Other local equations have been created from paired pellet and mark-recapture studies across
Snowshoe Hare range (Mills et al. 2005, Homyack et al. 2006, Berg and Gese 2010). Mills
et al. (2005) found that the Yukon Equation provided comparable density estimates to their
locally derived equation even though it was in different habitat in Montana. They also
suggested that it may be appropriate to use at moderate hare densities if absolute density is
not needed. That was less important in this study than obtaining a measure of relative use to
assess hare habitat. I used 0.155 m2 rectangular plots to align with the methods of Krebs et
al. (2001) as this may be important for producing reliable density estimates from the Yukon
Equation (McKelvey et al. 2002, Mills et al. 2005).

Habitat Metrics
The JPRF had remote-sensing LIDAR data collected from a fixed-winged aircraft for
the entire study area in August 2015. This provided high-resolution habitat data (10 m x 10
m pixel) that allowed for the quantification of dozens of candidate site and habitat covariates
to use in models. LIDAR data were initially processed by Forsite Consultants Ltd., Salmon
Arm, BC. Out of the dozens of candidate covariates I selected nine that had been previously
identified as important to Snowshoe Hares, allowing me to explore my hypotheses (Table
2.1). I obtained basal area for the camera models and distance to edge for the camera and
pellet models from the publicly-available VRI. I treated canopy closure from 0 to 1 m as a
surrogate for horizontal cover, as it was highly correlated with cover pole measurements
taken at camera stations.

44
Table 2.1. Habitat and site covariates used in Snowshoe Hare (Lepus americanus) logistic
regression models for camera and pellet detection data (2015 – 2020) from the John Prince
Research Forest and surrounding area in north-central, British Columbia (BC), Canada. Data
collected using remote sensing LIDAR with additional data from the Vegetation Resource
Index (VRI; Ministry of Forests, BC). Canopy closure is in meters (m) above the ground.
*Used in camera models only.
Covariate
Abbreviation Definition
Data Source
2
Basal Area*
BA
Estimated BA (m /ha) from VRI
VRI
polygon for each location.
Canopy Closure
0–1m
1–3m
3 – 10 m
> 10 m

CC0–1
CC1–3
CC3–10
CC>10

•

•

LIDAR
Camera Models: percent canopy
closure using raster zonal statistics
for each height category within 50
m-radius buffer around each
camera station
Pellet Models: percent canopy
closure for each height category for
each pellet plot within 10 x 10 m
LIDAR tile it occurs in.

Distance to
Edge

DistEdge

Distance (m) to VRI polygon stand
edge.

VRI

Distance to
Riparian Area*

DistRip

Distance (m) to the closest wetland,
stream or lake.

LIDAR

Elevation (m)*

Elev

Mean elevation using raster zonal
statistics within 50 m-radius buffer.

LIDAR

•

LIDAR

Slope

•

Camera Models: slope (%) using
raster zonal statistics within 50 mradius buffer surrounding the
camera.
Pellet Models: slope (%) for each
pellet plot within 10 x 10 m
LIDAR tile it occurs in

45
Habitat Modelling
Camera Habitat Models
To explore habitat relationships between binary detection data (present vs. absent) I
chose logistic regression over occupancy models, although both are commonly used in
analyses of detection data (Fuller et al. 2016, Pease et al. 2016, Thomas et al. 2019). The
cameras provided near-continuous monitoring and had good sightlines that made false
negatives unlikely. I performed preliminary occupancy modelling that confirmed the
addition of detection probabilities was not necessary as they did not change occupancy
estimates for any of the years. Given this, I used more parsimonious logistic regression
models (i.e., less parameters). To explore for multicollinearity between the habitat metrics, I
generated and visibly inspected a correlation matrix of all the variables. Any variables that
appeared to be correlated were not included in the same model.
I fitted all logistic regression models using R version 4.2.1 (R Foundation Statistical
Computing, Vienna, Austria). I used corrected Akaike’s Information Criterion (AICc) and
AIC model averaging to assess the relative rank of each model (Hurvich and Tsai 1989,
Burnham and Anderson 2002) using the R package “AICcmodavg.” I considered models
that were within two AIC units of the top model to be competitive (Burnham and Anderson
2002). To assess goodness-of-fit for the models I calculated the area under the curve (AUC)
for the Receiver Operating Characteristic (ROC) curve (Boyce et al. 2002) using the R
package “pROC.” I used the commonly referenced 0.7 AUC threshold (Boyce et al. 2002)
for defining models that have acceptable predictive power for assessing habitat use. From
these models, I compared the explanatory covariates between stations where hares were
present with those where they were absent using Mann Whitney U tests (non-parametric ttests).

Pellet Habitat Models
I pooled the data from the plots at each site together to assess habitat use across the
entire landscape, where hares were both present and absent. Hereafter I refer to these models

46
as “landscape models.” To investigate use at a finer-scale I looked at 7/10 sites where hares
were present above 0.3 hares/ha, a threshold necessary to prevent an overestimation of
density through the Yukon Equation (Mills et al. 2005). This set of models (based on 700
plots) I refer to as “within-stand models.” I decided to only use covariates that varied at
similar levels among plots as they did among sites to eliminate potential bias. I therefore
used fine-scale covariates that were measured at the plot-level, resulting in the exclusion of
basal area and elevation from the pellet analysis. Additionally, I omitted the distance to
riparian area metric from the fine-scale analysis based on the fact it was not considered
biologically meaningful at this scale.
I used logistic regression models for pellet analysis, by assigning all pellet plots to a
binary category (plots with pellets (i.e., hares present) = 1; plots without pellets (i.e., hares
absent) = 0). This was appropriate because, although it was possible to estimate density at a
site by averaging many plots, it was impossible to equate the number of pellets at each
individual plot with intensity of use due to variation in pellet deposition rates and
topography. Logistic models also fit better, had less parameters and allowed for some
comparison with the camera models. I performed all pellet logistic regression models for
2019 and 2020 using the same software and model averaging methods as the camera data
(See Camera Habitat Models).

Model Set
I generated a set of eight models for the camera detection data and seven for the
pellet data to evaluate the relationship among hare use and cover, tree volume, edge and
riparian habitat (Table 2.2). The “Site” models contained only geographic features while the
“Cover” models contained only canopy closure. The “Hyp1” model, short for hypothesis,
combined near-ground canopy closure and site covariates, with the idea that near-ground
cover provides protection from terrestrial predators while overstorey cover covariates
contained in “Hyp2” may protect from avian predators. The “Hyp3” and “Hyp4” combine
all site and canopy closure covariates to investigate whether all levels of cover may interact
to influence use. Finally, models “Esc1” and “Esc2,” short for escape, included features that
could provide quick escape cover from predators such as horizontal cover (i.e., canopy

47
closure from 0 to 1 m), cover provided by larger tree stems (i.e., basal area) and the distance
to edge and riparian area. I did not include basal area, distance to riparian area or elevation
in the pellet models; this made Hyp3 and Hpy4 equivalent so I eliminated Hyp4.

48
Table 2.2. Models used in logistic regression analysis of camera (2015, 2016 and 2020) and
pellet (2019 and 2020) detection data from the John Prince Research Forest and surrounding
area in north-central, British Columbia, Canada. BA = basal area, CC = canopy closure
between 0 – 1 m, 1 – 3 m, 3 – 10 m and greater than (>) 10 m, DistEdge = distance to edge,
DistRip = distance to riparian area and Elev = elevation.
Name
Camera Models
Pellet Models
Site

Elev + Slope + DistEdge + DistRip

Slope + DistEdge

Cover

CC0–1 + CC1–3 + CC3–10 + CC>10

CC0–1 + CC1–3 + CC3–10 + CC>10

Hyp1

Elev + Slope + CC0–1 + CC1–3 + BA

Slope + CC0–1 + CC1–3

Hyp2

Elev + Slope + CC3–10 + CC>10 + BA

Slope + CC3–10 + CC>10

Hyp3

Slope + CC0–1 + CC1–3 + CC3–10 +

Slope + CC0–1 + CC1–3 + CC3–10 +

CC>10 + BA

CC>10

Hyp4

Elev + Slope + CC0–1 + CC1–3 +
CC3–10 + CC>10 + BA

Esc1

CC0–1 + DistEdge + DistRip + BA

CC0–1 + DistEdge

Esc2

Elev + Slope + CC0–1 + DistEdge +

Slope + CC0–1 + DistEdge

DistRip + BA

49
RESULTS

Camera Detection Variance Across Years
I detected hares during 885 out of a total of 12,840 winter camera-days (7%) across
all three study years (Table 2.3). The mean detection rate was three times greater in 2015 (𝑋"
= 0.09 (SD = 0.13)) than in 2020 (𝑋" = 0.03 (SD= 0.05)). Detection rates decreased at 41/66
camera stations between 2015 and 2020. I detected hares at least once at 59 stations across
all three study years in winter and at 25 out of 49 in spring-summer 2016.
Detection rates in 2015 ranged from zero at 19/66 camera stations to a maximum of
0.66, while the majority (71%) of stations had detections rates of less than 0.10. I did not
detect hares at 19 stations in winter 2016, whereas one camera station had the study
maximum detection rate of 0.77. Detection rates were the lowest overall in 2020 with hares
only detected at half (52%) of the stations and a maximum detection rate of 0.23 at one.
Although detection rates generally declined over the three years of study, the
majority of camera stations retained the same relative position when compared to other
stations within the same year (i.e., most stations that had the lowest/highest detection rates in
2015 also had the lowest/highest in 2016 and 2020). The 50th percentile of detection rates for
2015 was 0.04 and I used it as the threshold for classifying hares present (detection rate >
0.04) or absent (detection rate ≤ 0.04) at each station. Using this threshold, I classified hares
as present at 28/66 camera stations in 2015, 27/66 in winter 2016, 25/49 in spring-summer
2016 and 14/66 in 2020. This meant occupancy also decreased by half, from a winter high of
0.42 in 2015 to a low of 0.21 in 2020.

50
Table 2.3. Total camera-days (maximum of 4620 per year for winter (W) and 8918 for
spring-summer (SS)), mean camera-days per camera station (maximum of 70 per station for
W, 182 for SS) (SD), total camera-days with detections, mean detection rate (total cameradays with detections/total camera-days) (SD) and occupancy (stations with hares
present/total stations) for Snowshoe Hare (Lepus americanus) camera trapping study during
W (February and March) 2015, 2016 and 2020 and SS (April – September) 2016 in the John
Prince Research Forest and surrounding area, north-central British Columbia, Canada.
2015 - W

2016 – W

2016 - SS

2020 - W

Total camera-days

4254

4453

8621

4133

Mean camera-days per station

64 (8)

68 (7)

176 (12)

63 (8)

Total camera-days with detections

389

361

628

135

Mean detection rate
Occupancy

0.09 (0.13)
0.42

0.08 (0.10) 0.08 (0.11)
0.41

0.51

0.03 (0.05)
0.21

51
Camera Habitat Models
Most of the 2015 models had a level of predictive power that made them useful for
assessing habitat use (AUC > 0.7; Table 2.4). The majority of the model averaged weight
was from models containing the distance to riparian area covariate, which was significant in
all models it was included in. These models also had the greatest predictive power (i.e.,
AUC scores) and suggested hare use was positively associated with increasing distance from
riparian areas (Table 2.5). The median distance to riparian areas for stations where hares
were present was 41 m (Interquartile range (IQR) = 11 – 78 m) and was over four times
greater than the 9 m median at the absent stations (IQR = 3 – 21 m; W28,38 = 328, P = 0.008).

52
Table 2.4. Logistic regression models for Snowshoe Hare (Lepus americanus) camera trapping study in the John Prince Research
Forest and surrounding area, north-central British Columbia, Canada. Only models deemed competitive with the top model (i.e.,
within two Delta AIC units) are displayed. K = number of parameters, AICci = corrected Akaike’s Information Criterion, AICcΔi =
Delta AIC = difference in AIC score between top model and each model, AICcWi = AIC weight = proportion of the total amount of
predictive power of each model when assessing the full set of models, AUC = area under the curve for the receiver operating
characteristic. See Table 2.2 for model abbreviations. * Denotes significant coefficient (P < 0.05).
Model

K

AICci

AICcΔi

AICcWi

AUC

Esc2

Elev + Slope* + CC0–1 + DistEdge + DistRip* + BA

7

82.96

0.00

0.30

0.84

Hyp2

Elev + Slope* + CC3–10* + CC>10 + BA

6

83.58

0.62

0.22

0.80

Esc1

CC0–1 + DistEdge + DistRip* + BA*

5

84.03

1.06

0.18

0.79

Site

Elev + Slope* + DistEdge + DistRip*

5

84.36

1.40

0.15

0.81

2015 – Winter (n = 66)

2016 – Winter (n = 66)
Hyp2

Elev* + Slope + CC3–10* + CC>10 + BA

6

85.80

0.00

0.28

0.77

Hyp1

Elev*+ Slope + CC0–1 + CC1–3* + BA

6

85.82

0.01

0.28

0.70

Cover

CC0–1 + CC1–3 + CC3–10 + CC>10

5

86.28

0.48

0.22

0.65

Hyp4

Elev* + Slope + CC0–1 + CC1–3 + CC3–10 + CC>10 + BA

8

87.28

1.48

0.13

0.80

2016 – Spring-Summer (n = 49)
Cover

CC0–1 + CC1–3* + CC3–10* + CC>10*

5

64.56

0.00

0.37

0.80

Hyp3

Slope + CC0–1* + CC1–3 + CC3–10* + CC>10* + BA

7

64.68

0.12

0.35

0.83

2020 – Winter (n = 66)
Esc1

CC0–1 + DistEdge* + DistRip + BA

5

66.55

0.00

0.44

0.76

Site

Elev + Slope + DistEdge* + DistRip

5

67.19

0.64

0.32

0.76

53
Table 2.5. Coefficients (β) for significant variables in logistic regression models for
Snowshoe Hare (Lepus americanus) camera detections in the John Prince Research Forest
and surrounding area, north-central British Columbia, Canada. See Table 2.2 for model
abbreviations.
Study Period

Model Covariate

β

SE

P

2015

Esc2

Slope

-0.12

0.06

0.04

2015

Esc2

DistRip

0.03

0.01

0.01

2015

Hyp2

Slope

-0.14

0.05

0.01

2015

Hyp2

CC3–10

0.05

0.02

0.01

2015

Esc1

DistRip

0.03

0.01

0.004

2015

Esc1

BA

-0.06

0.03

0.013

2015

Site

Slope

-0.11

0.05

0.025

2015

Site

DistRip

0.02

0.01

0.039

2016 – Winter

Hyp2

Elev

0.01

0.01

0.037

2016 – Winter

Hyp2

CC3–10

0.05

0.02

0.021

2016 – Winter

Hyp1

Elev

0.01

0.01

0.039

2016 – Winter

Hyp1

CC1–3

0.06

0.02

0.024

2016 – Winter

Hyp4

Elev

0.01

0.01

0.050

2016 – SS

Cover

CC1–3

-0.11

0.05

0.027

2016 – SS

Cover

CC3–10

0.09

0.04

0.014

2016 – SS

Cover

CC>10

-0.08

0.03

0.004

2016 – SS

Hyp3

CC0–1

0.08

0.04

0.033

2016 – SS

Hyp3

CC3–10

0.09

0.04

0.024

2016 – SS

Hyp3

CC>10

-0.09

0.03

0.010

2020

Esc1

DistEdge

0.03

0.01

0.009

2020

Site

DistEdge

0.02

0.01

0.045

54
In 2015 slope was significant in all models where it was included, being negatively
associated with hare use. In fact, the median slope was almost twice as steep for camera
stations where hares were absent (14% (IQR = 8 – 18%)) than camera stations where they
were present (8% (IQR = 6 – 12%); W28,38 = 744, P = 0.005) in 2015. I also observed a
similar difference in 2020 with flatter median slopes at stations where hares were present
(8% (IQR = 5 – 11%)) than at stations where they were absent (12% (IQR = 7 – 17%); W14,
52 = 495, P = 0.041). In winter 2016 there was no difference in slope (W27,39 = 645, P =

0.124); however, median elevation was greater at stations where hares were present (809 m
(IQR = 781 – 851 m)) than where they were not (786 m (IQR = 767 – 810 m); W27,39 = 321,
P = 0.007). In Winter 2016, elevation was positively and significantly associated with hare
use in all models it was included.
Canopy closure from 3 – 10 m was significant in the Hyp2 model in 2015 and in the
top ranked model of winter 2016, was in several of the models with the highest AUC values
and was positively associated with hare use in all models. It was also significant in the top
two ranked models in spring-summer 2016. Median canopy closure from 3 – 10 m was
greater at camera stations where hares were present than where they were absent in both
2015 (58% (IQR = 44 – 70 %) vs. 48% (IQR = 38 – 57%); W28,38 = 364, P = 0.029) and
winter 2016 (57% (IQR = 48 – 70%) vs. 49% (IQR = 38 – 58%); W27,39 = 351, P = 0.022).
Model selection differed in winter 2016 from winter 2015 as many of the models
with the highest AUC and lowest AIC contained canopy closure covariates, with distance to
riparian area seeming less influential. Canopy closure from 1 – 3 m was significant in the
Hyp3 model in winter 2016 and camera stations with hares present also had greater median
canopy closure from 1 – 3 m (49% (IQR = 40 – 59%)) than where they were absent (42%
(IQR = 32 – 53%); W27,39 = 374, P = 0.047). It was also significant in the top model in
spring-summer 2016 but was negatively associated with hare use.
Distance to edge appeared to drive model selection in 2020. It was significant and
positively associated with hare use and included in the top three ranked models. Median
distance to edge where hares were present was almost double (29 m (IQR = 17 – 59 m)) that
of camera stations where they were absent (16 m (IQR = 5 – 35 m); W14, 52 = 219, P =
0.023). An almost two-fold increase in this metric also was seen at stations where hares were
present (15 m (IQR = 5 – 25 m) vs 29 m (IQR = 12 – 50 m); W24,25 = 415, P = 0.021) in

55
spring-summer 2016 but did not seem as important to model selection as it was only
included in models with relatively low model average weight.
Canopy closure over 10 m was significant in the Hyp3 model and negatively
associated with hare use in 2020. It was also significant in all models in spring-summer
2016 and the median was 23% greater at camera stations where hares were absent (54%
(IQR = 32 – 62%)) than where they were present (31% (IQR = 21 – 50%); W24,25 = 180, P =
0.015). Canopy closure from 0 – 1 m was not significant in any models in winter in any of
the three years but it was in two models in spring-summer 2016 and positively associated
with hare use. However, there was no difference in the median between stations where they
were present or absent (W24,25 = 242, P = 0.254).

Pellet Detections Across Multiple Scales
I detected hare pellets at 408/1000 plots in 2019 and 322/1000 plots in 2020. When I
eliminated recently cut and mature sites pellet detection rates increased to 403/700 plots in
2019 and 320/700 plots in 2020. I also used pellet counts from each site to estimate annual
hare density for 2019 and 2020 and for the six sites that were also sampled in 2018 (Table
2.6). These numbers suggested density dropped at 7/10 sites from 2019 to 2020. Density
also decreased or was unchanged at five out of six sites from 2018 to 2019. Density was the
lowest at the mature sites (>80 years) and the recently clear-cut site (~ six years old).

56
Table 2.6. Estimated hare density (hares/ha) using the Yukon Equation (Krebs et al. 2001)
for Snowshoe Hare (Lepus americanus) pellet sites in the John Prince Research Forest
(JPRF) and surrounding area, north-central British Columbia (BC), Canada. Each site
contained 100 plots for a total of 1000 plots (600 plots in 2018) counted twice annually and
combined to obtain a single annual estimate. *These sites were eliminated from the withinstand model analysis. **Forest stand age estimates obtained from the BC Vegetation
Resource Inventory and confirmed by JPRF harvest records.
Site 2018 Density 2019 Density 2020 Density Age**
1

0.18

0.16

0.07

43

2

1.74

2.19

0.71

16

3

0.47

0.26

0.16

48

4*

0.01

0.00

0.00

>80

5*

0.16

0.05

0.06

>80

6*

0.00

0.01

0.00

6

7

0.47

0.66

23

8

2.04

1.52

34

9

1.20

1.43

35

10

1.54

0.92

40

57
Pellet Habitat Models
Model selection outcomes were different when comparing high-ranking habitat
covariates at the within-stand-scale to those at the landscape-scale (Table 2.7). The
landscape models had the same AIC ranking in 2019 and 2020 with the majority of the AIC
weight held by the same top ranked model (Hyp3 AICcWi: 2019 = 0.83, 2020 = 0.96) that
contained all four canopy closure categories and slope. The cover covariates in combination
with slope seems to have driven model selection at the landscape-scale, whereas the distance
to edge seemed more important for within-stand models.

58
Table 2.7. Logistic regression habitat models for Snowshoe Hare (Lepus americanus)
derived from pellet counts in 2019 and 2020 within the John Prince Research Forest and
surrounding area, north-central British Columbia, Canada. All four study periods used the
same set of seven models (See Table 2.2 for model abbreviations) but only those deemed
competitive with the top model (i.e., within two Delta AIC units) are displayed. K = number
of parameters in model including the intercept, Akaike’s Information Criterion scores
(AICci), Delta AIC (AICcΔi) = difference in AIC score between top model and each model,
AIC weight (AICcWi) = proportion of the total amount of predictive power provided by each
model when assessing the full set of models, and area under the curve (AUC) for the
receiver operating characteristic. *Denotes significant coefficient (P < 0.05).
Model
K AICci AICcΔi AICcWi AUC
2019 – Within-stand Models (n = 700)
Site

Slope + DistEdge*

3

810.61

0.00

0.54

0.73

Esc2

Slope + CC0–1 + DistEdge*

4

811.98

1.37

0.27

0.74

Esc1

CC0–1 + DistEdge*

3

812.65

2.04

0.19

0.73

2020 - Within-stand Models (n = 700)
Site

Slope* + DistEdge*

3

858.10

0.00

0.68

0.69

Esc2

Slope* + CC0–1 + DistEdge*

4

860.11

2.01

0.25

0.69

6

983.09

0.00

0.83

0.84

6

981.17

0.00

0.96

0.81

2019 - Landscape Models (n = 1000)
Hyp3 Slope* + CC0–1 + CC1–3* + CC3–10* +
CC>10*
2020 - Landscape Models (n = 1000)
Hyp3 Slope* + CC0–1 + CC1–3* + CC3–10* +
CC>10*

59
Canopy closure from 1 – 3 m was significant in all landscape models and also was
positively associated with hare use. Median canopy closure from 1 – 3 m was one-third
greater for plots with hares present for the landscape data in 2019 (58% (IQR = 19 – 52%)
vs. 44% (IQR = 43 – 78; W408,592 = 59870, P < 2.2e-16) and more than two-thirds greater in
2020 (61% (IQR = 46 – 80%) vs. 35% (IQR = 20 – 57%); W322,678 = 54136, P < 2.2e-16;
Table 2.8). Canopy closure from 3 – 10 m was significant in all models and was positively
associated with hare use. Canopy closure over 10 m was significant in all models as well but
was negatively associated with hare use. Median canopy closure over 10 m was two-thirds
greater for plots with hares absent in 2019 (37% (IQR = 0 – 87%) vs. 22% (IQR = 0 – 48%);
W408,592 = 157197, P = 2.3e-16) and nearly three times greater in 2020 (53% (IQR = 0 –
85%) vs. 20% (IQR = 0 – 46%); W322,678 = 140946, P = 5.1e-14) for the landscape data. It
was also over two times greater for plots with hares absent in 2019 (53% (IQR = 5 – 83%)
vs. 22% (IQR = 0 – 47%); W403,297 = 77822, P = 8.1e-12) and 2020 (43% (IQR = 2 – 79%)
vs. 19% (IQR = 0 – 46%); W320,380 = 77135, P = 6.3e-10) for the within-stand data.
It was harder to determine the relationship between hare use and canopy closure
from 0 – 1 m, as even in those models where the covariate was significant, there was both
positive and negative values. For the within-stand data in both years, canopy closure from 0
– 1 m was not significantly different for hare present sites from absent sites (2019 - W403,297
= 55705, P = 0.113, 2020 - W320,380 = 55777, P = 0.059).
The within-stand models were similarly ranked in 2019 and 2020. The distance to
edge covariate was significant for both years and was also included in the top three AIC
ranked models in both years as well, accounting for all of the model averaged AIC weight.
The AUC values for 2020 were low, however, suggesting not all the models may have had
adequate predictive power to use for assessing habitat use (6/7 below 0.7 AUC value).
Median distance to edge was almost three times less for plots with hares present (50 m (IQR
= 22 – 101 m)) than for those with hares absent in 2019 (149 m (IQR = 44 – 282 m); W403,297
= 122792, P = 0.001) and over two times less in 2020 (50 m (IQR = 33 – 103 m) vs. 116 m
(35 – 257 m); W320,380 = 82141, P = 1.25e-15) for hare habitat plots.

60
Table 2.8. Coefficients for significant variables in logistic regression models for Snowshoe
Hare (Lepus americanus) pellet detections in 2019 and 2020 in the John Prince Research
Forest and surrounding area, north-central British Columbia, Canada. WS = Within-standscale and LS = Landscape-scale models, see Table 2.2 for model abbreviations.
Study Period Model Covariate
β
SE
P
2019 – WS

Site

DistEdge

-0.010

0.001

<2e-16

2019 – WS

Esc2

DistEdge

-0.010

0.001

< 2e-16

2019 – WS

Esc1

DistEdge

-0.010

0.001

< 2e-16

2020 – WS

Site

Slope

-0.014

0.001

0.0283

2020 – WS

Site

DistEdge

-0.009

0.001

< 2e-16

2020 – WS

Esc2

Slope

-0.015

0.007

0.0307

2020 – WS

Esc2

DistEdge

-0.009

0.001

< 2e-16

2019 – LS

Hyp3

Slope

-0.015

0.006

0.022

2019 – LS

Hyp3

CC1–3

0.018

0.005

0.006

2019 – LS

Hyp3

CC3–10

0.043

0.004

< 2e-16

2019 – LS

Hyp3

CC>10

-0.031

0.004

< 2e-16

2020 – LS

Hyp3

Slope

-0.018

0.006

0.005

2020 – LS

Hyp3

CC1–3

0.026

0.005

6.23E-07

2020 – LS

Hyp3

CC3–10

0.031

0.004

4.98E-16

2020 – LS

Hyp3

CC>10

-0.023

0.004

2.89E-10

61
DISCUSSION
My results showed that habitat use changed at different scales and population
densities and occupancy. For example, habitat model selection and fit differed for all three
years of CT as Snowshoe Hare occupancy declined by almost half; this suggested densitydependent habitat use. It was also different at the within-stand- compared to the landscapescale for the pellet counts. This study furthers the understanding of Snowshoe Hare habitat
use by demonstrating some of the shortcomings of previous hare habitat surveys that have
primarily focused on the stand-level and either ignored or did not discuss the impact densitydependent habitat use on their results and conclusions (Holbrook 2017). Furthermore, I was
able to produce habitat models that were biologically realistic and had good predictive
accuracy that help deepen our understanding of hare habitat use in the sub-boreal forests of
north-central BC.
Habitat Use
In 2015, when occupancy was highest, increased distances to riparian areas seemed
to drive camera model selection. This is counter to what I predicted as I believed the dense
vegetation and abundance of winter food species (e.g., willows) provided by riparian areas
would be associated with increased hare use. Tape et al. (2016) discovered that changes to
the Arctic climate have allowed riparian ecosystems to expand north and these habitats may
have further supported the expansion of Snowshoe Hare range north as well. I was not,
however, able to identify other studies that specifically looked at hare habitat use in relation
to riparian areas. It is possible that hares avoid riparian areas and this was most apparent in
2015 when they occupied more available habitat. This needs further investigation to confirm
as many species, including mesocarnivores, use riparian areas as travel corridors (Hilty and
Merenlender 2016, Santos et al. 2016) and it is possible predation is greater for hares in
these habitats.
In 2016 camera model selection and fit appeared to be influenced by canopy closure.
I predicted hares in my study area would use areas with greater horizontal cover (i.e., canopy
closure from 0 – 1 m) but it was not significant in any winter camera or pellet models. This

62
LIDAR variable was highly correlated with cover pole visual obstruction measurements
taken at each camera station; therefore, I believed it was a robust measure of
understory/horizontal cover. This type of cover has been found to be the most important to
Snowshoe Hares, especially in winter, and is often the biggest predictor of use (Sullivan et
al. 2007, Berg et al. 2012, Sultaire et al. 2016).
Snow may explain the fact canopy cover from 0 – 1 m was significant and positively
associated with hare use in spring-summer 2016 but was not in many of the winter models,
when 1 – 2 m of snow on the ground is common. Snow also could have influenced this layer
in the pellet models, which span a year of pellet counts, as there is generally snow on the
ground in the region for about half the year (Environment and Climate Change Canada
2022). Canopy closure from 1 – 3 m was significant and positively associated with hare use
in one 2016 camera model and in many pellet models. It is possible that this height category
acted as horizontal cover in these models as it represented the crucial 0 – 2 m above ground
horizontal cover layer (Conroy et al. 1979, Litvaitis et al. 1985) when there was snow on the
ground. Potvin et al. (2005) did not find a significant effect of horizontal cover on hare use
and suggested the uniformity of cover across their study area may be the cause. That could
also be true for my study but it is more likely snow caused the 1 – 3 m layer to be more
representative of horizontal cover in the winter than the 0 – 1 m layer, indicating horizontal
cover is important to hares in the region.
Canopy closure above the understory provides a different type of cover (e.g., visual
obstruction from aerial predators) and is often correlated with increased hare use (Roy et al.
2010, Cheng et al. 2015, Holbrook et al. 2016), which my data also supported. Canopy
closure from 3 – 10 m was a significant predictor of hare use for several camera models in
2015 and 2016 and many of the pellet models. Thomas et al. (2019) found canopy closure
more important to hares than horizontal cover and suggested this could be due to hares
avoiding the risk of avian predation. A diverse and abundant raptor community is found in
the area with several known hare predators, including Great Horned Owl (Bubo virginianus)
(Rohner and Krebs 1996) and Northern Goshawk (Accipiter gentilis) (Boutin et al. 1995),
and it is possible hares use areas with increased canopy closure to reduce the risk of aerial
predation. Canopy closure also could be correlated with thermoregulation, food quality
and/or quantity or other benefits.

63
Mature forests do not always provide suitable habitat for hares even though canopy
closure is often greatest in these habitats (Sullivan et al. 2012). Canopy closure over 10 m
was significant and negatively associated with hare use in camera models in spring-summer
2016 and winter 2020 as well as several pellet models. In 2020 population pressure to
occupy all available, including suboptimal, habitat was likely reduced and so it is possible
the remaining hares avoided sites with the greatest canopy closure over 10 m, indicating
they avoided mature sites. This idea was further supported by very low pellet detections at
mature sites in all years.
For the camera models in 2020 and the pellet within-stand models the distance to
edge seemed to be an important predictor of use and was significant in the top three models.
Interestingly, hares were positively associated with distance to edge for the camera models
in 2020 but negatively associated for the pellet models. The difference between the pellet
and camera models could be due to differences in scale but could also be attributed to
seasonality. Hares often forage in more densely vegetated interior sites in winter when they
feed on more woody browse; whereas in summer it is the interspersion of edge and open
habitat that provides seasonally available herbaceous foliage (Wolff 1980). The 2016 springsummer data contradicts this, however, as distance to edge also was significant and positive
but was not highly influential on model selection. Although edge habitat may provide the
interspersion of foraging and cover habitat required by hares in some populations (Conroy et
al. 1979) it may also be correlated with greater predation for some species (Andrén 1995).
Several studies have associated sites with lower edge density or greater distances
from edge habitat with higher Snowshoe Hare occupancy and use (Roy et al. 2010, Sultaire
et al. 2016). Distance to edge was not significant in winter 2015 or 2016 but as occupancy
substantially decreased it became highly influential in the 2020 cameras models. It is
possible in 2020 that the remaining hares stayed in or moved to areas further from habitat
edges, potentially to avoid predation (Wolff 1980). This highlights the differences in model
selection in winter 2015 and 2016 compared to 2020 that suggest density-dependent habitat
use. Furthermore, the different model selection and importance of distance to edge in the
within-stand compared to the landscape pellet models suggests a potential difference in use
at varying ecological scales.

64
Habitat Use Across Ecological Scales
As evidenced in many species, but largely unexplored with Snowshoe Hares, model
selection appears to vary across different ecological scales. In the current study, pellet model
selection was similar for 2019 and 2020 and was more influenced by scale (within-stand vs.
landscape). At the landscape-scale canopy closure appeared to have had the greatest
influence on model selection. This contrasts with the finer-scale within-stand models where
distance to edge drove model selection and fit but aligns with the few other studies that have
investigated and found a change in hare habitat use at different scales (Lewis et al. 2011,
Fuller and Harrison 2013, Gigliotti et al. 2018).
The concept of ecological scale is fundamental to wildlife research yet most habitat
studies focus on a single scale without exploring or addressing the impact of scale on their
results and conclusions (Wheatley and Larsen 2018). One approach is to test all the same
independent and dependent variables at each scale measured (Wheatley and Larsen 2018); a
continuum of scales is ideal for this as it may allow for cross-scale predictability (Wheatley
and Johnson 2009). Wheatley and Larsen (2018) demonstrated how model selection varied
for Northern Flying Squirrels (Glaucomys sabrinus) across 14 observational scales and
showed how the effect differed between sex and age classes. Unfortunately, I was not able to
do this type of in-depth scale analysis but it may be possible in the future with a long-term
hare dataset and LIDAR, which allows for the quantification of the same habitat variables at
different scales.
I only examined hare habitat use at two scales but I was able to demonstrate
differences in model selection between the two. This illustrates how studies undertaken at
multiple scales can be important in determining the full scope of hare habitat use (Thornton
et al. 2013, Holbrook et al. 2017) and further supports the recommendations of others to
expand beyond typical stand-level hare habitat studies (Ausband and Baty 2005, Fuller and
Harrison 2013, Holbrook et al. 2017).

65
Density-Dependent Habitat Use and the Snowshoe Hare Cycle
I observed a large decline in camera detections between the first two years of study
and the third; this population oscillation suggested to me that my study population may
follow a cyclical pattern observed in many other Snowshoe Hare populations (Keith and
Windberg 1978, Krebs et al. 2018, Sullivan and Sullivan 1988). Pellet counts also declined
between 2018 and 2020 and subsequent pellet counts in 2022 have shown the population is
beginning to rebound (unpublished JPRF data); this further corroborated a local hare cycle.
It is also worth noting that lynx detections decreased during this same period to their lowest
levels in 2020 and 2021 (unpublished JPRF data). This also suggested the occurrence of the
classic hare-lynx cycle in my study area.
Without observations spanning the entirety of a typical cycle length, or multiple
cycles, I was unable to determine if 2015 represented the peak and 2020 the low of the
cycle. I did, however, believe it was useful to compare the camera models in winter 2015
and 2016, which had similar occupancy, to 2020 when occupancy was significantly reduced.
Occupancy was highest in spring-summer 2016. This was not surprising as hare populations
generally are higher in summer when many offspring are produced before many young-ofthe-year and adults do not survive winter (Kielland et al. 2010). The winter models
supported my hypothesis that if occupancy and/or density changed so would habitat use and
supported the theory that hares exhibit density-dependent habitat use.
Determining the most effective time to quantify and assess habitat and life history
characteristics of species that cycle in abundance has long been a question of debate in
ecology. The Ideal Free Distribution theory suggests that as a population increases, some
individuals will have to move to lower quality habitat (Fretwell and Lucas 1969). Snowshoe
Hares are known to exhibit density-dependent habitat preferences and occupy most available
habitat during population peaks and retreat to refugium during lows (Wolff 1980, Hodson et
al. 2010). These refugia appear to be high-quality areas with dense vegetation (Wolff 1980).
This change in landscape occupancy could potentially lead to conflicting results and
interpretations when assessing habitat use for hare or other cyclic populations that exists in a
heterogenous environment. This could explain the differences in model selection for the
three years of camera data. More broadly it could explain the varying and sometimes

66
contrasting hare habitat use inferences from previous studies, in addition to their broad
distribution across many habitat types and varying methods used.

Conclusions and Management Implications
My comparisons are limited in that the camera and pellet studies did not have
complete temporal overlap. Also, the spring-summer data only represented a subset of all
camera stations. A useful continuation of this study would be to monitor hares in my study
area over the entire cycle. Cameras proved to be a useful tool for collecting hare detection
data that I used to look at habitat use but they could be used to examine other aspects of their
ecology. The detection data we passively gathered with cameras allowed me to develop
models that seemed to fit the data (i.e., AUC > 0.7). With my study design and cameras, I
was also able to detect a large decline in the population confirmed by more traditional and
widely-tested hare pellet plots. This supports the argument that CT is a useful way to
monitor the Snowshoe Hare cycle, especially with new methods for estimating hare densities
from cameras (Jensen et al. 2022).
My results suggested that both horizontal cover and canopy closure influence habitat
use. It should be a priority of anyone managing forests for Snowshoe Hares, or one or more
of their many predators, to ensure contiguous areas with abundant horizontal cover and
canopy closure (Holbrook et al. 2017). In practice this might include dividing forestry
operations over a larger area to avoid large clearcuts, retaining more trees, or longer harvest
intervals (Thomas et al. 2019). Managing for a mosaic of different interconnected habitat
types (e.g., stand ages, composition) may allow both forestry and conservation priorities to
be met as long as some contiguous forested patches with dense cover are maintained. This
will also provide the diversity of habitats that allow hares to avoid or use areas near habitat
edges and riparian areas.
My results showed that habitat modelling can be heavily influenced by the method
and scale used in each study and hence the results of this and other studies need to be
interpreted within their own confines. This in turn supports the recommendations of
Holbrook et al. (2017) to assess habitat use at multiple ecological scales (e.g., assess withinstand use in addition to the traditional stand-level approach), with multiple response

67
variables (e.g., density and occupancy) and across a gradient of potential habitats not just
those assumed to be high quality.
Density-dependent habitat use can occur in cyclic populations and further complicate
the interpretations of these studies (Kawaguchi and Desrochers 2018). Habitats that provide
refugia during population lows may be the most important to protect and manage for
Snowshoe Hares (Wolff 1980, Hodson et al. 2010) and their many obligate and
opportunistic predators, especially lynx. Similar to Wolff (1980) my results suggested
interior sites (i.e., further from habitat edges) with abundant horizontal cover and canopy
closure may provide this type of refugia. Multi-year studies that span the entirety, with
regular intermittent sampling, of one or more population cycles may provide better insights
into Snowshoe Hare habitat use and may also be important for other species or populations
that cycle in abundance over time as well.
Camera trapping proved useful in detecting changes in the local hare population and
for evaluating habitat use. Deploying cameras and maintaining storage devices and batteries
across a fairly large study area was labour intensive. As storage and battery technology
continues to improve and easier and cheaper remote access wildlife cameras become
available, they may provide a more effective way to monitor cyclic populations like
Snowshoe Hares. These populations may also require more frequent monitoring to fully
understand habitat use, especially if it is density-dependent, to ensure proper management.
This is important for a keystone species like Snowshoe Hares that sustain so many other
species and are an integral part of their local ecosystems.
This study has expanded our understanding of hare habitat use in the sub-boreal
forests of north-central BC. It also has confirmed the need for more hare habitat studies to
go beyond the stand-level and explore hare habitat use at a fine-scale. I have also been able
to provide more support for the idea that Snowshoe Hares exhibit density-dependent habitat
use; this is useful as it needs to be addressed more frequently in hare habitat studies. Finally,
I was able to provide some evidence that the local hare population cycles in abundance and
that wildlife CT studies were useful in detecting population changes. This project
contributes to the long-term conservation and understanding of this keystone species that is
integral to the healthy functioning of northern forested ecosystems.

68
LITERATURE CITED
Andrén, H. 1995. Effects of landscape composition on predation rates at habitat edges.
Pages 225–255 in L. Hansson, L. Fahrig, and G. Merriam, editors. Mosaic
Landscapes and ecological processes. Chapman and Hall, United Kingdom.
Ausband, D. E., and G. R. Baty. 2005. Effects of precommercial thinning on Snowshoe Hare
habitat use during winter in low-elevation montane forests. Canadian Journal of
Forest Research 35:206–210.
Berg, N. D., and E. M. Gese. 2010. Relationship between fecal pellet counts and Snowshoe
Hare density in Western Wyoming. The Journal of Wildlife Management 74:1745–
1751.
Berg, N. D., E. M. Gese, J. R. Squires, and L. M. Aubry. 2012. Influence of forest structure
on the abundance of Snowshoe Hares in Western Wyoming. The Journal of Wildlife
Management 76:1480–1488.
Boutin, S., C. J. Krebs, R. Boonstra, M. R. T. Dale, S. J. Hannon, K. Martin, A. R. E.
Sinclair, J. N. M. Smith, R. Turkington, M. Blower, A. Byrom, F. I. Doyle, C. Doyle,
D. Hik, L. Hofer, A. Hubbs, T. Karels, D. L. Murray, V. Nams, M. O’Donoghue, C.
Rohner, and S. Schweiger. 1995. Population changes of the vertebrate community
during a Snowshoe Hare cycle in Canada’s boreal forest. Oikos 74:69–80.
Boyce, M. S., P. R. Vernier, S. E. Nielsen, and F. K. A. Schmiegelow. 2002. Evaluating
resource selection functions. Ecological Modelling 157:281–300.
Breitenmoser, U., B. G. Slough, and C. Breitenmoser-Würsten. 1993. Predators of cyclic
prey: is the Canada Lynx victim or profiteer of the Snowshoe Hare cycle? Oikos
66:551–554.
Burnham, K. P. and D. R. Anderson. 2002. Model selection and multimodal inference: a
practical information-theoretic approach. Springer, USA.
Cheng, E., K. E. Hodges, and L. S. Mills. 2015. Impacts of fire on Snowshoe Hares in
Glacier National Park, Montana, USA. Fire Ecology 11:119–136.
Conroy, M. J., L. W. Gysel, and G. R. Dudderar. 1979. Habitat components of clear-cut
areas for Snowshoe Hares in Michigan. The Journal of Wildlife Management
43:680–690.

69
Crowley, S. M., and D. P. Hodder. 2017. An assessment of the efficacy of rub stations for
detection and abundance surveys of Canada Lynx (Lynx canadensis). Canadian
Journal of Zoology 95:885–890.
Di Marco, M., J. E. M. Watson, H. P. Possingham, and O. Venter. 2017. Limitations and
trade-offs in the use of species distribution maps for protected area planning. Journal
of Applied Ecology 54:402–411.
Environment and Climate Change Canada. 2022. Canadian climate normals 1981-2010
station data, Fort St James, BC. <http://climate.weather.gc.ca/climate_normals>.
Accessed 31 Mar 2020.
Feierabend, D., and K. Kielland. 2014. Movements, activity patterns, and habitat use of
Snowshoe Hares (Lepus americanus) in interior Alaska. Journal of Mammalogy
95:525–533.
Fretwell, S. D., and H. L. Lucas. 1969. On territorial behavior and other factors influencing
habitat distribution in birds. Acta Biotheoretica 19:16–36.
Fuller, A. K., and D. J. Harrison. 2013. Modeling the influence of forest structure on
microsite habitat use by Snowshoe Hares. International Journal of Forestry Research
4:1–7.
Fuller, A. K., D. W. Linden, and J. A. Royle. 2016. Management decision making for Fisher
populations informed by occupancy modeling. The Journal of Wildlife Management
80:794–802.
Gigliotti, L. C., B. C. Jones, M. J. Lovallo, and D. R. Diefenbach. 2018. Snowshoe Hare
multi-level habitat use in a fire-adapted ecosystem. Journal of Wildlife Management.
82:435–444.
Ginzburg, L., and C. Krebs. 2015. Mammalian cycles: internally defined periods and
interaction-driven amplitudes. PeerJ 3:e1180.
Hilty, J. A., A. M. Merenlender. 2004. Use of riparian corridors and vineyards by
mammalian predators in Northern California. Conservation Biology 18:126-135.
Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for Snowshoe Hares.
Forest Ecology and Management 256:1918–1926.

70
Hodson, J., D. Fortin, M.-L. LeBlanc, and L. Bélanger. 2010. An appraisal of the fitness
consequences of forest disturbance for wildlife using habitat selection theory.
Oecologia 164:73–86.
Holbrook, J. D., J. R. Squires, L. E. Olson, R. L. Lawrence, and S. L. Savage. 2016.
Multiscale habitat relationships of Snowshoe Hares (Lepus americanus) in the mixed
conifer landscape of the Northern Rockies, USA: cross-scale effects of horizontal
cover with implications for forest management. Ecology And Evolution 7:125–144.
Holbrook, J. D., J. R. Squires, L. E. Olson, R. L. Lawrence, and S. L. Savage. 2017.
Multiscale habitat relationships of Snowshoe Hares (Lepus americanus) in the mixed
conifer landscape of the Northern Rockies, USA: cross-scale effects of horizontal
cover with implications for forest management. Ecology And Evolution 7:125–144.
Homyack, J. A., D. J. Harrison, J. A. Litvaitis, and W. B. Krohn. 2006. Quantifying
densities of Snowshoe Hares in Maine using pellet plots. Wildlife Society Bulletin
34:74–80.
Hurvich, C. M., C. L. Tsai. 1989. Regression and time series model selection in small
samples. Biometrika 76:297–307.
Jensen, P. O., A. J. Wirsing, and D. H. Thornton. 2022. Using camera traps to estimate
density of Snowshoe Hare (Lepus americanus): a keystone boreal forest herbivore.
Journal of Mammalogy 103:693–710.
Kawaguchi, T., and A. Desrochers. 2018. A time-lagged effect of conspecific density on
habitat selection by Snowshoe Hare. PLOS ONE 13:e0190643.
Keith, L. B., and L. A. Windberg. 1978. A demographic analysis of the Snowshoe Hare
cycle. Wildlife Monographs 3–70.
Kielland, K. O., K. Olsen and E. Euskirchen. 2010. Demography of Snowshoe Hares in
relation to regional climate variability during a 10-year population cycle in Interior
Alaska. Canadian Journal of Forest Research. 40: 1265–1272.
Krebs, C. J., R. Boonstra, and S. Boutin. 2018. Using experimentation to understand the 10year Snowshoe Hare cycle in the boreal forest of North America. Journal of Animal
Ecology 87:87–100.

71
Krebs, C. J., R. Boonstra, V. Nams, M. O’Donoghue, K. E. Hodges, and S. Boutin. 2001.
Estimating Snowshoe Hare population density from pellet plots: a further evaluation.
Canadian Journal of Zoology 79:1–4.
Krebs, C. J., S. Boutin, R. Boonstra, A. R. E. Sinclair, J. N. M. Smith, M. R. T. Dale, K.
Martin, and R. Turkington. 1995. Impact of food and predation on the Snowshoe
Hare cycle. Science 269:1112–1115.
Krebs, C. J., J. Bryant, K. Kielland, M. O’Donoghue, F. Doyle, S. Carriere, D. DiFolco, N.
Berg, R. Boonstra, S. Boutin, A. J. Kenney, D. G. Reid, K. Bodony, J. Putera, H. K.
Timm, T. Burke, J. A. K. Maier, and H. Golden. 2014. What factors determine cyclic
amplitude in the Snowshoe Hare (Lepus americanus) cycle? Canadian Journal of
Zoology 92:1039–1048.
Krebs, C. J., B. S. Gilbert, S. Boutin, A. R. E. Sinclair, and J. N. M. Smith. 1986. Population
biology of Snowshoe Hares. I. Demography of food-supplemented populations in the
Southern Yukon, 1976-84. Journal of Animal Ecology 55:963–982.
Lewis, C. W., K. E. Hodges, G. M. Koehler, and L. S. Mills. 2011. Influence of stand and
landscape features on Snowshoe Hare abundance in fragmented forests. Journal of
Mammalogy 92:561–567.
Litvaitis, J. A., J. A. Sherburne, and J. A. Bissonette. 1985. Influence of understory
characteristics on Snowshoe Hare habitat use and density. The Journal of Wildlife
Management 49:866–873.
MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A.
Langtimm. 2002. Estimating site occupancy rates when detection probabilities are
less than one. Ecology 83:2248–2255.
MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines.
2006. Occupancy estimation and modeling: inferring patterns and dynamics of
species occurrence. Academic Press, USA.
McCann, N. P., R. Moen, and G. Niemi. 2008. Using pellet counts to estimate Snowshoe
Hare numbers in Minnesota. The Journal of Wildlife Management 72:955–958.
McKelvey, K. S., G. W. McDaniel, L. S. Mills, and P. C. Griffin. 2002. Effects of plot size
and shape on pellet density estimates for Snowshoe Hares. Wildlife Society Bulletin
30:751–755.

72
Meidinger, D., and J. Pojar. 1991. Special report series 6: ecosystems of British Columbia.
Ministry of Forests, Victoria, Canada.
Meslow, E. C., and L. B. Keith. 1968. Demographic parameters of a Snowshoe Hare
population. The Journal of Wildlife Management 32:812–834.
Mills, L. S., P. C. Griffin, K. E. Hodges, K. McKelvey, L. Ruggiero, and T. Ulizio. 2005.
Pellet count indices compared to mark-recapture estimates for evaluating Snowshoe
Hare density. Journal of Wildlife Management 69:1053–1062.
Myers, J. H. 2018. Population cycles: generalities, exceptions and remaining mysteries.
Proceedings of the Royal Society B: Biological Sciences 285:20172841.
Nichols, J. D., J. E. Hines, D. I. MacKenzie, M. E. Seamans, and R. J. Gutierrez. 2007.
Occupancy estimation and modeling with multiple states and state uncertainty.
Ecology 88:1395–1400.
O’Donoghue, M., S. Boutin, C. J. Krebs, and E. J. Hofer. 1997. Numerical responses of
Coyotes and Lynx to the Snowshoe Hare cycle. Oikos 80:150–162.
O’Donoghue, M., S. Boutin, C. J. Krebs, D. L. Murray, and E. J. Hofer. 1998. Behavioural
responses of Coyotes and Lynx to the Snowshoe Hare cycle. Oikos 82:169–183.
Pease, B. S., C. K. Nielsen, and E. J. Holzmueller. 2016. Single-camera trap survey designs
miss detections: impacts on estimates of occupancy and community metrics. PLOS
ONE 11:e0166689.
Potvin, F., N. Bertrand, and J. Ferron. 2005. Attributes of forest strips used by Snowshoe
Hare in winter within clear-cut boreal landscapes. Canadian Journal of Forest
Research 35:2521–2527.
Rohner, C., and C. J. Krebs. 1996. Owl predation on Snowshoe Hares: consequences of
antipredator behaviour. Oecologia 108:303–310.
Roy, C., L. Imbeau, and M. J. Mazerolle. 2010. Transformation of abandoned farm fields
into coniferous plantations: is there enough vegetation structure left to maintain
winter habitat of Snowshoe Hares? Canadian Journal of Zoology 88:579–588.
Royle, J. A., and J. D. Nichols. 2003. Estimating abundance from repeated presence–
absence data or point counts. Ecology 84:777–790.

73
Santos, M.J., L. M. Rosalino, G. M. Matos and M. Santos-Reis. 2016. Riparian ecosystem
configuration influences mesocarnivores presence in Mediterranean
landscapes. European Journal Wildlife Research 62:251–261.
Simard, V., L. Imbeau, and H. Asselin. 2018. Effects of selection cuts on winter habitat use
of Snowshoe Hare (Lepus americanus) in northern temperate forests. Canadian
Journal of Forest Research 48:1049–1057.
Stirnemann, I., A. Mortelliti, P. Gibbons, and D. B. Lindenmayer. 2015. Fine-scale habitat
heterogeneity influences occupancy in terrestrial mammals in a temperate region of
Australia. PLOS ONE 10:e0138681.
Studd, E. K., M. R. Boudreau, Y. N. Majchrzak, A. K. Menzies, M. J. L. Peers, J. L. Seguin,
S. G. Lavergne, R. Boonstra, D. L. Murray, S. Boutin, and M. M. Humphries. 2019.
Use of acceleration and acoustics to classify behavior, generate time budgets, and
evaluate responses to moonlight in free-ranging Snowshoe Hares. Frontiers in
Ecology and Evolution 7.
Sullivan, T. P., and D. S. Sullivan. 1988. Influence of stand thinning on Snowshoe Hare
population dynamics and feeding damage in lodgepole pine forest. Journal of
Applied Ecology 25:791–805.
Sullivan, T. P., D. S. Sullivan, P. M. F. Lindgren, and D. B. Ransome. 2007. Long-term
responses of ecosystem components to stand thinning in young lodgepole pine forest:
IV. Relative habitat use by mammalian herbivores. Forest Ecology and Management
240:32–41.
Sullivan, T. P., D. S. Sullivan, P. M. F. Lindgren, and D. B. Ransome. 2012. Silviculture and
wildlife: Snowshoe Hare abundance across a successional sequence of natural and
intensively managed forests. International Scholarly Research Notices 2012:593103.
Sultaire, S. M., J. N. Pauli, K. J. Martin, M. W. Meyer, and B. Zuckerbeg. 2016. Extensive
forests and persistent snow cover promote Snowshoe Hare occupancy in Wisconsin.
The Journal of Wildlife Management 80:894–905.
Tape, K. D., K. Christie, G. Carroll, and J. A. O'Donnell. 2016. Novel wildlife in the Arctic:
the influence of changing riparian ecosystems and shrub habitat expansion on
Snowshoe Hares. Global Change Biology 22: 208-219.
Thomas, J. P., M. L. Reid, R. M. R. Barclay, and T. S. Jung. 2019. Salvage logging after an
insect outbreak reduces occupancy by Snowshoe Hares (Lepus americanus) and their
primary predators. Global Ecology and Conservation 17:e00562.

74

Thornton, D. H., A. J. Wirsing, J. D. Roth, and D. L. Murray. 2013. Habitat quality and
population density drive occupancy dynamics of Snowshoe Hare in variegated
landscapes. Ecography 36:610–621.
Weir, R. D., H. Davis, and C. Hoodicoff. 2003. Conservation strategies for North American
Badgers in the Thompson and Okanagan regions. Habitat Conservation Trust Fund
and Artemis Wildlife Consultants, Armstrong, Canada.
Wheatley, M., and C. Johnson. 2009. Factors limiting our understanding of ecological scale.
Ecological Complexity 6:150–159.
Wheatley, M., and K. Larsen. 2018. Scale relativity of species-habitat models. Ecological
Complexity 36:7–15.
Wirsing, A. J., T. D. Steury, and D. L. Murray. 2002. A demographic analysis of a southern
Snowshoe Hare population in a fragmented habitat: evaluating the refugium model.
Canadian Journal of Zoology 80:169–177.
Wolff, J. O. 1980. The role of habitat patchiness in the population dynamics of Snowshoe
Hares. Ecological Monographs 50:111–130.
Zhang, G., A.-X. Zhu, S. K. Windels, and C.-Z. Qin. 2018. Modelling species habitat
suitability from presence-only data using kernel density estimation. Ecological
Indicators 93:387–396.

75

CHAPTER 3: CONCLUSION
HIGHLIGHTS OF THE STUDY
I produced Snowshoe Hare (Lepus americanus) habitat models that were biologically
realistic and had good predictive accuracy. Model selection and fit aligned with many other
hare habitat studies in that cover and canopy closure were important predictors of hare use.
My results, however, showed that habitat modelling can be influenced by scale and also by
changes in population density or occupancy. For example, distance to edge was positively
associated with hare use for the within-stand pellet models but it appeared they avoided
edges for the camera models in the same year (2020). In the landscape pellet models
distance to edge was not significant highlighting differences in use at varying ecological
scales. It was also not significant in the 2015 or 2016 camera models potentially showing
density-dependent habitat use in Snowshoe Hares.
Other significant site and habitat variables included elevation, slope and distance to
riparian areas in some years/models. This study has helped to deepen our understanding of
hare habitat use in the sub-boreal forests of north-central British Columbia (BC). It also
furthers the understanding of Snowshoe Hare habitat use in general by addressing some of
the previous knowledge gaps by assessing hare habitat use at different scales with two
methods over a gradient of potential habitats and while hare occupancy and density was
declining. I was also able to document density-dependent habitat use.
MANAGEMENT IMPLICATIONS

Implications of Study to Snowshoe Hare Ecology
My results showed that a variety of site and habitat components influenced hare
habitat use. Cover often is cited as the single most important determinant and predictor of
hare habitat use (Litvaitis et al. 1985, Bois et al. 2012, Gigliotti et al. 2018). My results
supported this conclusion although other variables such as distance to edge also influenced
model selection. Horizontal/understory often is cited as the most important type of cover to

76
hares (Sullivan et al. 2007, Berg et al. 2012, Simard et al. 2018), while my models suggested
it did influence habitat use they also showed the importance of canopy closure higher in the
canopy (i.e. >3 m).
Canopy closure is known to influence hare habitat use (Roy et al. 2010, Thornton et
al. 2013, Cheng et al. 2015). Thomas et al. (2019) found canopy closure to be more
important to hares than horizontal cover and suggested hares used areas with greater canopy
closure to avoid aerial predation. Although Canada Lynx (Lynx canadensis) are known to
predate hares in the area (as evidenced by the camera footage), the exact proportion each
predator species contributes to hare mortality in the area is unknown. In the Yukon, Krebs et
al. (1995) found that avian predators accounted for 40% of hare predation. It is possible
hares select for different types of cover based on the composition of the local predator
community. It is, therefore, important to include several measures of cover when modelling
hare habitat use including both horizontal and canopy closure. My models indicated hares
were positively associated with canopy closure under 10 m but negatively associated with it
over 10 m so it may also be useful for other studies to stratify canopy closure to assess its
influence at multiple levels.
Distance to edge was an important predictor of use and was significant in the top
three camera models in 2020 and the within-stand pellet models in both years. Hare habitat
use was positively associated with distance to edge for the camera models in 2020 but
negatively associated for the pellet models. This may have indicated that the relationship
with distance to edge changes with scale and/or hare abundance. Other significant site and
habitat variables included elevation, slope and distance to riparian areas in some
years/models.
Snowshoe Hare densities vary greatly across their range and throughout the hare
cycle so it can be difficult to compare them. In the southern part of their range in Colorado,
Ivan et al. (2014) estimated densities in different forest stands that ranged from 0.01 to 0.66
hares/ha. Whereas Krebs et al. (1995) documented densities in the southern Yukon as high
as 7 hares/ha near a population peak that later crashed to 1 hare/ha. Estimated densities at
my pellet sites ranged from 0 to 2.2 hares/ha and generally declined from 2018 to 2020 but
appeared to be rebounding in 2021 and 2022 (unpublished JPRF data). Originally it was
believed that the hare cycle was synchronized across their range but fur trapper data from

77
the 1930s and -40s suggested population peaks travelled in a wave across Canada with
delays between peaks up to four years (Krebs et al. 2018). Hare densities most recently hit
their peak at the famous Kluane Lake study area in 2017 (Oli et al. 2020). Krebs et al.
(2018) suggested about a two-year delay in peak number between north-central BC, where
this study was located, and the Kluane Lake study area in Southern Yukon. If 2015 was, or
was close to, the peak for my study population then it would align with that suggestion,
adding support to my conclusion that it cycles in abundance over time.
As emphasized throughout this thesis, Snowshoe Hares have received considerable
research attention. Studies across their range and especially long-term studies, such as those
in the Southern Yukon (Krebs et al. 1986, 2001a, Smith et al. 1988) are not just important
for understanding hare ecology but are also landmark ecological studies. Some knowledge
gaps still exist though; for example, most hare habitat studies have been conducted at the
stand- or patch-level and it is important that future research broadens the focus to multiple
scales (e.g., within-stand; Thornton et al. 2013, Holbrook et al. 2017). With my pellet
models I assessed hare habitat use at the landscape-scale (i.e., a representative selection of
all habitat types) vs. areas with moderate to high hare use and demonstrated a difference in
model selection between the two scales. Although camera and pellet models were not
directly comparable, model selection differed between the 7854 m2 camera trapping (CT)
scale and the 100 m2 pellet plot scale. I believe this shows it is important that those
managing forests for hares and their predators or any other species to assess habitat use at
multiple scales or conservation planning may leave important habitats unprotected
(Wheatley and Larsen 2018, Wheatley and Johnson 2009).
Another common practice in hare habitat studies has been to use type-based
approaches where habitats are predefined and sampled accordingly (e.g., using forest cover
maps to stratify habitats and only sampling areas with high hare use potential) but it has
been argued surveying a gradient of potential habitats may be more appropriate (Holbrook et
al. 2017). The CT grid was randomly generated in order to capture a representative sample
of the broader landscape. This included areas where hares were more likely to occur (e.g.,
regenerating hybrid spruce-fir forests or lodgepole pine stands) and where they likely were
not (e.g., open habitats, mature stands). With the pellet counts I also included a variety of
sites with varying hare use potential, such as clear-cuts and old-growth stands that are not

78
commonly included. By dropping these sites for part of the pellet analysis I was able to
show that model selection, and therefore likely habitat use, differed when assessing all sites
or just those that supported moderate to high hare population densities. This leads me to
agree with the conclusions of Holbrook et al. (2017) and recommend biologists survey a
broader range of habitat types not just those assumed to support high hare densities.
With this study I was able to demonstrate density-dependent habitat use that also has
been observed in other Snowshoe Hare populations (Wolff 1980, Hodson et al. 2010).
Although it has been acknowledged by other hare researchers, the impacts of this shift in
landscape occupancy throughout the hare cycle largely goes undiscussed. Studies that are
able to investigate hare habitat use over the course of a cycle, or several cycles, may provide
better insights than short-term studies. It would therefore be useful for researchers to
explicitly state which phase of the cycle their study took place in, as habitat conclusions
made at population peaks can differ from those made during lows, as I demonstrated. I state
this with the understanding that long-term research funding is rare and planning studies that
may span decades is very challenging (e.g., a graduate study typically lasts two to four
years).
Implications of Study to Passive Snowshoe Hare Surveys
The cameras used in this study also detected many other species and these data are
currently being analyzed by JPRF researchers and other graduate students. Their data
showed population and/or detection declines in hare predators and changes in the
composition of the mesocarnivore community (unpublished JPRF data). I believe my study
can help inform this ongoing work. First, I would suggest including cover (horizontal and
canopy closure) and distance to edge covariates in any habitat models, as it is reasonable to
assume overlap in habitat use between hares and their predators. I would also recommend
assessing whether local predator species also experience density-dependent habitat use as
exhibited by hares in this study.
The use of CT for Snowshoe Hare density estimation (Jensen et al. 2022) may
provide a logistically and economically feasible way to assess hare habitat cycles alone or in
conjunction with other types of surveys. I was able to detect a large decline in occupancy

79
(50%) with cameras alone and confirmed this decline with traditional pellet counts. This
could eliminate the need for labour intensive and invasive mark-recapture trapping, except
in certain circumstances; these might include studies where more accurate measures of
abundance are required (e.g., addressing conservation concerns) or investigations into diet,
movement or other factors.
We also might see CT density estimation replace labour intensive pellet surveys,
especially with continuing technological improvements and new methods discussed in
Jensen et al. (2022). I also was able to demonstrate that traditional pellet surveys can be used
to explore hare habitat use at different scales; this also could be done using CT methods,
especially when combined with LIDAR data, which can allow for the quantification of the
same variable at a variety of scales. I believe the combination of camera and pellet methods
allowed me to address my hypothesis without the need for challenging mark-recapture
surveys. I also demonstrated the CT can be used to monitor the Snowshoe Hare cycle, which
could be very useful to researchers looking for a non-invasive way to do that.

Forest Management of a Keystone Species
Many aspects of Snowshoe Hare ecology have been studied over the years. This is
due to both their critical importance as a food source for many other species and also their
dramatic population cycles that allows for the exploration of many key ecological concepts
(e.g., Lotka-Volterra predator-prey dynamics). I chose to focus on hares due to their
importance to the local ecosystem in the sub-boreal forests of north-central BC and because
preliminary camera analysis suggested a change in the composition of the local predator
community (e.g., decreases in Canada Lynx and American Marten (Martes americana)
detections and increases in detections of rarer species such as Fisher (Pekania pennanti) and
Wolverine (Gulo gulo)). This study will help JPRF researchers further explore these
dynamics and determine whether they are related to the Snowshoe Hare cycle.
These results support the idea that a mosaic of seral stages and habitat types may
provide the interspersion of interior and open sites that hares require at different scales and
at varying points along the hare cycle (Conroy et al. 1979, Wolff 1980). Snowshoe Hares
occupy a variety of forested habitats across their range and so it is up to land managers to

80
determine what harvest regimes and combination of different age patches best support local
hare populations.
It was clear that recent clear-cut sites, with no or low cover, and mature sites, with
lower shrub and tree densities, did not support as many hares as sites of intermediate ages.
This has also been observed in other studies and hares usually do not recolonize harvested
areas for 15 – 30 years (Paragi et al. 1997, de Bellefeuille et al. 2001). My models indicated
both cover and canopy closure influenced habitat use so patches that provide both (e.g., midsuccessional conifer- or mixed-stands) should be maintained, preferably they also should be
connected within a mosaic of other seral stages and land uses. Therefore, for my study
region, and most likely others, I would recommend staggering harvest timing geographically
so that the appropriate composition of seral stages exist for hares and avoiding large clear
cuts, increasing tree retention in and adjacent to harvested areas and allowing for longer
harvest intervals (Thomas et al 2019).
LIMITATIONS AND FUTURE RESEARCH
Most of the biggest limitations of this study revolved around timing. With the data
series I was unable to fully determine which years represented the absolute hare population
peak and low. I had camera data for 2015 and 2016, then a three-year data gap and then
pellet data in 2019 and both pellet and camera data in 2020. This five-year period did not
cover the span of a typical hare cycle, which is around 10 years and sometimes presented as
a range from 8 – 11 years (Krebs et al. 2018). Camera trapping and pellet counts completed
subsequent to my study, in 2021 and 2022, seem to indicate that the local hare population is
beginning to recover (unpublished JPRF data). This does suggest that 2020 was the
population low and further implies that the population was near its peak in 2015. Another
limitation was that the camera and pellet data did not fully align temporally. A longer-term
dataset would have enabled me to plot out the entire cycle but I do believe I was still able to
draw conclusions about density-dependent hare habitat use during or near a population peak
and low. Linden et al. (2017) concluded that models using detection data can be used in
place of other methods (e.g., measure of abundance such as density) as they often offer
similar habitat inferences, while remaining more cost effective.

81
Going forward the JPRF will have annual hare density estimates and CT every three
to five years. This dataset will allow them to determine and predict hare population highs
and lows. Using the analysis framework in this thesis, it should be relatively simple to use
the same models to assess hare habitat use for subsequent years. As I established with this
study, the hares in this region exhibit density-dependent habitat use. It would be very
interesting to not only assess habitat use near the population peak and crash but also during
the recovery and decline phase. Furthermore, it has been established that the amplitude of
each population cycle can vary (Ginzburg and Krebs 2015, Myers 2018) and it would be
interesting to determine whether habitat use also varies with cyclic amplitude.
The use of CT wildlife studies has grown exponentially over the last 20 years and a
variety of field methods and analytical approaches have been used. Although some have
advocated for the standardization of CT methods it is challenging when habitats and species
ecology vary so much; also, rare species may require specialized approaches (Burton et al.
2015). One commonly cited issue with CT studies is that researchers must organize
continuous detection data into discrete presence/absence “surveys” that are analogous to
independent sampling events and can vary from a day to weeks (Burton et al. 2015, Rich et
al. 2017, Neilson et al. 2018). This subjectivity can bias results and interpretations and
different survey lengths could led to different estimates of abundance or conflicting model
selection. Cameras also do not allow the identification of individuals, with the exception of
some species (e.g., distinct markings), and this makes estimating density difficult. Camera
data is also time consuming to process and so traditional mark-recapture studies are still
useful for many research questions.
In order to define what constituted an independent sampling event I considered the
ecology of my model species. Snowshoe Hares are mostly active at night, dawn and dusk
and are almost completely sedentary during the day (Keith 1964, Feierabend and Kielland
2014, Studd et al. 2019). This meant that their activity already was organized into discrete
time periods and hence I chose a 24-hour camera-day that ran from noon on one calendar
day until noon the next. Of course, this did not eliminate all biases and the results of this
study must be interpreted within the confines of the methods used. This includes the fact that
I further defined presence/absence using the 50th percentile of detection rates for the first
year of study (2015), during the studies’ population high. This may seem subjective but the

82
detection rate data was clearly bimodal, where the stations with detections either had very
few (one or two camera-days out of 70) or moderate to high detections (10 – 40 camera-days
out of 70). The approach I chose helped to delineate between the two clear categories.
Another reason I chose to do this was because a common weakness of CT studies is that
they fail to established a clear relationship between camera detections and abundance
(Burton et al. 2015). I believed by delineating very low use stations I could be more
confident that stations where hares were classified as present represented stations they
actually used.
I chose not to incorporate detection probabilities into my models, which has been
cited as a flaw of some camera studies (Burton et al. 2015). I did, however, perform
exploratory occupancy modeling with my data and the occupancy estimates were not
different for any of the years. I chose to use logistic regression models because they have
less parameters and I sought to compare, in whatever capacity, the pellet and camera
models. Another weakness of some camera studies is the arbitrary definition of scale as
detections at a single camera are often broadened and interpreted in the context of a much
larger geographic unit (e.g., a forest stand). To limit this type of bias I paired the camera
detection data with fine-scale habitat data (50-m radius), which was only possible due to the
LIDAR data.
The LIDAR data were collected in 2015, given it is expensive to collect and analyze
these data it was not feasible to update it for the later models. I visited each camera station in
summer of 2019, as did JPRF researchers in winter 2020, and did not observe any major
disturbances or habitat changes within the buffer of any camera. My study area experiences
long cold winters that limit annual plant growth and rapid changes in vegetative biomass so I
believe that it was still appropriate to use the 2015 LIDAR for all my study years.
One final limitation of the pellet models was that I used the Yukon-derived
regression equation (Krebs et al. 1987, Krebs et al. 2001b) to estimate density and not a
locally derived equation because that would have required an entire hare trapping and
tagging mark-recapture study. As shown by Mills et al. (2005) the Yukon Equation may
provide comparable density estimates across a wider geographic range, especially if studies
do not require highly accurate measures of absolute abundance, which I did not. I also
followed similar pellet count methods to limit bias (McKelvey et al. 2002, Mills et al. 2005).

83
A hare mark-recapture study in the JPRF would answer this question. More broadly other
future research that would complement my thesis includes verifying new methods for
estimating hare density from cameras (Jensen et al. 2022) and directly comparing them to
pellet studies. Eventually this could eliminate the need for labour intensive pellet and markrecapture studies and standardize Snowshoe Hare density estimating methods across their
range. As suggested by Wheatley (2009) I also believe a more in-depth study of hare habitat
use over a larger scale-domain continuum would be beneficial instead of just the two I
investigated in this study. The JPRF LIDAR dataset may provide the opportunity to do this
in the future.
CONCLUSION
Although Snowshoe Hares have been heavily researched this study is one of a few
that has assessed hare habitat use at multi-scales, with multiple responses and field methods
and sampled a wide-range of habitats, not just those suspected to be high-quality (Holbrook
et al. 2017). With this study I was able to show that hares exhibit density-dependent habitat
use that varied at different scales. I was also able to confirm that cameras can be useful in
detecting oscillations in hare populations such as those occurring over the full population
cycle. I also showed that cover, especially horizontal cover and canopy closure under 10 m,
distance to edge and several other site and habitat features were important and useful in
predicting hare use.
I believe this study expands our understanding of Snowshoe Hare ecology potentially
leading to better management of not only hares but the many predators who rely on them.
My ability to demonstrate the effects of different detection methods, scale and
density/occupancy on habitat model selection influences not only Snowshoe Hare studies
but also those of their predators and other species who cycle in abundance overtime. I
showed that cameras were useful in detecting changes in occupancy/density associated with
population cycles and this has implications for species other than hares as CT studies could
be used to monitor their cycles as well. This study increases the scope and depth of our
knowledge of this important keystone species, and in turn provides for better management
and conservation of forest processes.

84
LITERATURE CITED
de Bellefeuille, S., L. Bélanger, J. Huot, and A. Cimon. 2001. Clear-cutting and
regeneration practices in Quebec boreal balsam fir forest: effects on Snowshoe Hare.
Canadian Journal of Forest Research 31:41–51.
Berg, N. D., E. M. Gese, J. R. Squires, and L. M. Aubry. 2012. Influence of forest structure
on the abundance of Snowshoe Hares in Western Wyoming. The Journal of Wildlife
Management 76:1480–1488.
Bois, G., L. Imbeau, and M. J. Mazerolle. 2012. Recovery time of Snowshoe Hare habitat
after commercial thinning in boreal Quebec. Canadian Journal of Forest Research
42:123–133.
Burton, A. C., E. Neilson, D. Moreira, A. Ladle, R. Steenweg, J. T. Fisher, E. Bayne, and S.
Boutin. 2015. REVIEW: Wildlife camera trapping: a review and recommendations
for linking surveys to ecological processes. Journal of Applied Ecology 52:675–685.
Cheng, E., K. E. Hodges, and L. S. Mills. 2015. Impacts of fire on Snowshoe Hares in
Glacier National Park, Montana, USA. Fire Ecology 11:119–136.
Conroy, M. J., L. W. Gysel, and G. R. Dudderar. 1979. Habitat components of clear-cut
areas for Snowshoe Hares in Michigan. The Journal of Wildlife Management
43:680–690.
Feierabend, D., and K. Kielland. 2014. Movements, activity patterns, and habitat use of
Snowshoe Hares (Lepus americanus) in interior Alaska. Journal of Mammalogy
95:525–533.
Gigliotti, L. C., B. C. Jones, M. J. Lovallo, and D. R. Diefenbach. 2018. Snowshoe Hare
multi-level habitat use in a fire-adapted ecosystem. Journal of Wildlife Management
82:435–444.
Ginzburg, L., and C. Krebs. 2015. Mammalian cycles: internally defined periods and
interaction-driven amplitudes. PeerJ 3:e1180.
Hodson, J., D. Fortin, M.-L. LeBlanc, and L. Bélanger. 2010. An appraisal of the fitness
consequences of forest disturbance for wildlife using habitat selection theory.
Oecologia 164:73–86.

85
Holbrook, J. D., J. R. Squires, L. E. Olson, R. L. Lawrence, and S. L. Savage. 2017.
Multiscale habitat relationships of Snowshoe Hares (Lepus americanus) in the mixed
conifer landscape of the Northern Rockies, USA: cross-scale effects of horizontal
cover with implications for forest management. Ecology And Evolution 7:125–144.
Ivan, J. S., G. C., White, and T. M., Shenk. 2014. Density and demography of Snowshoe
Hares in Central Colorado. The Journal of Wildlife Management 78:580–594.
Jensen, P. O., A. J. Wirsing, and D. H. Thornton. 2022. Using camera traps to estimate
density of Snowshoe Hare (Lepus americanus): a keystone boreal forest herbivore.
Journal of Mammalogy 103:693–710.
Keith, L. B. 1964. Daily activity pattern of Snowshoe Hares. Journal of Mammalogy
45:626–627.
Krebs, C. J., R. Boonstra, and S. Boutin. 2018. Using experimentation to understand the 10year Snowshoe Hare cycle in the boreal forest of North America. Journal of Animal
Ecology 87:87–100.
Krebs, C.J., R. Boonstra, S. Boutin, and A. R. E. Sinclair. 2001a. What drives the 10-year
cycle of Snowshoe Hares? The ten-year cycle of Snowshoe Hares, one of the most
striking features of the boreal forest, is a product of the interaction between predation
and food supplies, as large-scale experiments in the Yukon have demonstrated.
BioScience 51:25–35.
Krebs, C.J., R. Boonstra, V. Nams, M. O’Donoghue, K. E. Hodges, and S. Boutin. 2001b.
Estimating Snowshoe Hare population density from pellet plots: a further evaluation.
Canadian Journal of Zoology 79:1–4.
Krebs, C. J., S. Boutin, R. Boonstra, A. R. E. Sinclair, J. N. M. Smith, M. R. T. Dale, K.
Martin, and R. Turkington. 1995. Impact of food and predation on the Snowshoe
Hare cycle. Science 269:1112–1115.
Krebs, C. J., B. S. Gilbert, S. Boutin, and R. Boonstra. 1987. Estimation of Snowshoe Hare
population density from turd transects. Canadian Journal of Zoology 65:565–567.
Krebs, C. J., B. S. Gilbert, S. Boutin, A. R. E. Sinclair, and J. N. M. Smith. 1986. Population
biology of Snowshoe Hares. I. Demography of food-supplemented populations in the
Southern Yukon, 1976-84. Journal of Animal Ecology 55:963–982.

86
Linden, D. W., A. K. Fuller, J. A. Royle, and M. P. Hare. 2017. Examining the occupancydensity relationship for a low-density carnivore. Journal of Applied Ecology
54:2034–2052.
Litvaitis, J. A., J. A. Sherburne, and J. A. Bissonette. 1985. Influence of understory
characteristics on Snowshoe Hare habitat use and density. The Journal of Wildlife
Management 49:866–873.
McKelvey, K. S., G. W. McDaniel, L. S. Mills, and P. C. Griffin. 2002. Effects of plot size
and shape on pellet density estimates for Snowshoe Hares. Wildlife Society Bulletin
30:751–755.
Mills, L. S., P. C. Griffin, K. E. Hodges, K. McKelvey, L. Ruggiero, and T. Ulizio. 2005.
Pellet count indices compared to mark-recapture estimates for evaluating Snowshoe
Hare density. Journal of Wildlife Management 69:1053–1062.
Myers, J. H. 2018. Population cycles: generalities, exceptions and remaining mysteries.
Proceedings of the Royal Society B: Biological Sciences 285:20172841.
Neilson, E. W., T. Avgar, A. C. Burton, K. Broadley, and S. Boutin. 2018. Animal
movement affects interpretation of occupancy models from camera-trap surveys of
unmarked animals. Ecosphere 9:e02092.
Oli, M. K., C. J. Krebs, A. J. Kenney, R. Boonstra, S. Boutin, and J. E. Hines. 2020.
Demography of Snowshoe Hare population cycles. Ecology 101:e02969.
Paragi, T. F., W. N. Johnson, D. D. Katnik, and A. J. Magoun. 1997. Selection of post-fire
seres by lynx and Snowshoe Hares in the Alaskan Taiga. Northwestern Naturalist
78:77–86.
Rich, L. N., C. L. Davis, Z. J. Farris, D. A. W. Miller, J. M. Tucker, S. Hamel, M. S.
Farhadinia, R. Steenweg, M. S. D. Bitetti, K. Thapa, M. D. Kane, S. Sunarto, N. P.
Robinson, A. Paviolo, P. Cruz, Q. Martins, N. Gholikhani, A. Taktehrani, J.
Whittington, F. A. Widodo, N. G. Yoccoz, C. Wultsch, B. J. Harmsen, and M. J.
Kelly. 2017. Assessing global patterns in mammalian carnivore occupancy and
richness by integrating local camera trap surveys. Global Ecology and Biogeography
26:918–929.
Roy, C., L. Imbeau, and M. J. Mazerolle. 2010. Transformation of abandoned farm fields
into coniferous plantations: is there enough vegetation structure left to maintain
winter habitat of Snowshoe Hares? Canadian Journal of Zoology 88:579–588.

87
Simard, V., L. Imbeau, and H. Asselin. 2018. Effects of selection cuts on winter habitat use
of Snowshoe Hare (Lepus americanus) in northern temperate forests. Canadian
Journal of Forest Research 48:1049–1057.
Smith, J. N. M., C. J. Krebs, A. R. E. Sinclair, and R. Boonstra. 1988. Population biology of
Snowshoe Hares. II. Interactions with winter food plants. Journal of Animal Ecology
57:269–286.
Studd, E. K., M. R. Boudreau, Y. N. Majchrzak, A. K. Menzies, M. J. L. Peers, J. L. Seguin,
S. G. Lavergne, R. Boonstra, D. L. Murray, S. Boutin, and M. M. Humphries. 2019.
Use of acceleration and acoustics to classify behavior, generate time budgets, and
evaluate responses to moonlight in free-ranging Snowshoe Hares. Frontiers in
Ecology and Evolution 7.
Sullivan, T. P., D. S. Sullivan, P. M. F. Lindgren, and D. B. Ransome. 2007. Long-term
responses of ecosystem components to stand thinning in young lodgepole pine forest:
IV. Relative habitat use by mammalian herbivores. Forest Ecology and Management
240:32–41.
Thomas, J. P., M. L. Reid, R. M. R. Barclay, and T. S. Jung. 2019. Salvage logging after an
insect outbreak reduces occupancy by Snowshoe Hares (Lepus americanus) and their
primary predators. Global Ecology and Conservation 17:e00562.
Thornton, D. H., A. J. Wirsing, J. D. Roth, and D. L. Murray. 2013. Habitat quality and
population density drive occupancy dynamics of Snowshoe Hare in variegated
landscapes. Ecography 36:610–621.
Wheatley, M. 2009. Domains of scale in forest-landscape metrics: implications for specieshabitat modeling. Acta Oecoloigca 36:259–267.
Wheatley, M., and C. Johnson. 2009. Factors limiting our understanding of ecological scale.
Ecological Complexity 6:150–159
Wheatley, M., and K. Larsen. 2018. Scale relativity of species-habitat models. Ecological
Complexity 36:7–15.
Wolff, J. O. 1980. The role of habitat patchiness in the population dynamics of Snowshoe
Hares. Ecological Monographs 50:111–130.

88

APPENDIX 1 – FULL SET OF LOGISTIC REGRESSION CAMERA MODELS (N = 8 PER CAMERA TRAPPING SESSION).
Table 1. Full set of logistic regression models with number of parameters (K), corrected Akaike’s Information Criterion (AICci), Delta
AIC (AICcΔi) = difference in AIC score between top model and each model, AIC weight (AICcWi) = proportion of the total amount of
predictive power of each model when assessing the full set of models, and area under the curve (AUC) for the receiver operating
characteristic = a measure of predictive accuracy, for Snowshoe Hare (Lepus americanus) camera trapping study during winter 2015,
2016 and 2020 and during spring-summer 2016 in the John Prince Research Forest and surrounding area, north-central British
Columbia, Canada. BA = basal area, CC = canopy cover between 0 – 1 m, 1 – 3 m, 3 – 10 m and over 10 m, DistEdge = distance to
edge, DistRip = distance to riparian and Elev = elevation. * Denotes significant coefficient (P < 0.05).
Model

K

AICci

AICcΔi

AICcWi

AUC

Esc2

Elev + Slope* + CC0–1 + DistEdge + DistRip* + BA

7

82.96

0.00

0.30

0.84

Hyp2

Elev + Slope* + CC3–10* + cc>10 + BA

6

83.58

0.62

0.22

0.80

Esc1

cc0–1 + DistEdge + DistRip* + BA*

5

84.03

1.06

0.18

0.79

Site

Elev + Slope* + DistEdge + DistRip*

5

84.36

1.40

0.15

0.81

Hyp3

Slope* + CC0–1 + CC1–3 + CC3–10 + cc>10 + BA

7

85.72

2.76

0.08

0.79

Hyp4

Elev + Slope* + CC0–1 + CC1–3 + CC3–10 + CC>10 + BA

8

86.06

3.09

0.06

0.81

Hyp1

Elev + Slope* + CC0–1 + CC1–3 + BA

6

89.13

6.16

0.01

0.76

Cover

CC0–1 + CC1–3 + CC3–10 + CC>10

5

93.40

10.44

0.00

0.67

2015 – Winter (n = 66)

2016 – Winter (n = 66)
Hyp2

Elev* + Slope + CC3–10* + CC>10 + BA

6

85.80

0.00

0.28

0.77

Hyp1

Elev*+ Slope + CC0–1 + CC1–3* + BA

6

85.82

0.01

0.28

0.70

Cover

CC0–1 + CC1–3 + CC3–10 + CC>10

5

86.28

0.48

0.22

0.65

89
Hyp4

Elev* + Slope + CC0–1 + CC1–3 + CC3–10 + CC>10 + BA

8

87.28

1.48

0.13

0.80

Hyp3

Slope + CC0–1 + CC1–3 + CC3–10 + CC>10 + BA

7

89.16

3.36

0.05

0.76

Site

Elev* + Slope + DistEdge + DistRip

5

91.20

5.40

0.02

0.70

Esc2

Elev* + Slope + CC0–1 + DistEdge + DistRip + BA

7

92.23

6.43

0.01

0.74

Esc1

CC0–1 + DistEdge + DistRip + BA*

5

93.30

7.49

0.01

0.69

2020 – Winter (n = 66)
Esc1

CC0–1 + DistEdge* + DistRip + BA

5

66.55

0.00

0.44

0.76

Site

Elev + Slope + DistEdge* + DistRip

5

67.19

0.64

0.32

0.76

Esc2

Elev + Slope + cc0–1 + DistEdge* + DistRip + BA

7

69.66

3.11

0.09

0.78

Cover

CC0–1 + CC1–3 + CC3–10 + CC>10*

5

70.01

3.46

0.08

0.72

Hyp2

Elev + Slope + CC3–10 + CC>10 + BA

6

72.31

5.76

0.02

0.74

Hyp3

Slope + CC0–1 + CC1–3 + CC3–10 + CC>10 + BA

7

72.57

6.02

0.02

0.76

Hyp1

Elev + Slope + CC0–1 + CC1–3 + BA

6

73.28

6.73

0.02

0.72

Hyp4

Elev + Slope + CC0–1 + CC1–3 + CC3–10 + CC>10 + BA

8

74.60

8.05

0.01

0.76

2016 – Spring-Summer (n = 49)
Cover

CC0–1 + CC1–3* + CC3–10* + CC>10*

5

64.56

0.00

0.37

0.80

Hyp3

Slope + CC0–1* + CC1–3 + CC3–10* + CC>10* + BA

7

64.68

0.12

0.35

0.83

Hyp4

Elev + Slope + CC0–1* + CC1–3 + CC3–10* + CC>10* + BA

8

67.53

2.97

0.08

0.83

Site

Elev + Slope + DistEdge* + DistRip

5

67.80

3.24

0.07

0.75

Hyp2

Elev + Slope + CC3–10 + CC>10* + BA

6

67.93

3.38

0.07

0.79

Esc1

CC0–1 + DistEdge* + DistRip + BA

5

69.29

4.74

0.03

0.73

90
Esc2

Elev + Slope + CC0–1 + DistEdge* + DistRip + BA

7

70.22

5.66

0.02

0.78

Hyp1

Elev + Slope* + CC0–1 + CC1–3 + BA

6

72.99

8.43

0.01

0.72

91
APPENDIX 2 – FULL SET OF LOGISTIC REGRESSION PELLET MODELS (N = 7 PER CAMERA TRAPPING SESSION).
Table 1. Full set of logistic regression models with the number of parameters (K), Akaike’s Information Criterion scores (AICci), Delta
AIC (AICcΔi) = difference in AIC score between top model and each model, AIC weight (AICcWi) = proportion of the total amount of
predictive power provided by each model when assessing the full set of models, and area under the curve (AUC) for the receiver
operating characteristic = a measure of predictive accuracy for each model, for Snowshoe Hare (Lepus americanus) pellet study for
2019 and 2020 in the John Prince Research Forest and surrounding area, north-central British Columbia, Canada. CC = canopy cover
between 0 – 1 m, 1 – 3 m, 3 – 10 m and over 10 m, DistEdge = distance to edge. * Denotes significant coefficient (P < 0.05).
Model

K

AICci

AICcΔi

AICcWi

AUC

2019 –Within-stand Models (n = 700)
Site

Slope + DistEdge*

3

810.61

0.00

0.54

0.73

Esc2

Slope + CC0–1 + DistEdge*

4

811.98

1.37

0.27

0.74

Esc1

CC0_1 + DistEdge*

3

812.65

2.04

0.19

0.73

Hyp2

Slope + CC3–10* + CC>10*

4

835.47

24.86

0.00

0.74

Hyp3

Slope + CC0–1 + CC1–3 + CC3–10* + CC>10*

6

839.17

28.56

0.00

0.74

Cover

CC0–1 + CC1–3 + CC3–10* + CC>10*

5

840.41

29.81

0.00

0.73

Hyp1

Slope + CC0–1* + CC1–3

4

910.93

100.32

0.00

0.66

2020 – Within-stand Models (n = 700)
Site

Slope* + DistEdge*

3

858.10

0.00

0.68

0.69

Esc2

Slope* + CC0–1 + DistEdge*

4

860.11

2.01

0.25

0.69

Esc1

CC0_1 + DistEdge*

3

862.81

4.72

0.06

0.68

Hyp3

Slope* + CC0–1 + CC1–3* + CC3–10* + CC>10*

6

883.22

25.12

0.00

0.70

Hyp2

Slope* + CC3–10* + CC>10*

4

885.20

27.10

0.00

0.69

92
Cover

CC0–1 + CC1–3* + CC3–10* + CC>10*

5

888.32

30.22

0.00

0.69

Hyp1

Slope* + CC0–1* + CC1–3*

4

913.13

55.03

0.00

0.67

2019 - Landscape Models (n = 1000)
Hyp3

Slope* + CC0–1 + CC1–3* + CC3–10* + CC>10*

6

983.09

0.00

0.83

0.84

Cover

CC0–1 + CC1–3* + CC3–10* + CC>10*

5

986.38

3.29

0.16

0.84

Hyp2

Slope + CC3–10* + CC>10*

4

992.21

9.12

0.01

0.84

Hyp1

Slope* + CC0–1* + CC1–3*

4

1134.64

151.55

0.00

0.77

Esc1

CC0–1* + DistEdge*

3

1316.11

333.02

0.00

0.58

Esc2

Slope + CC0–1* + DistEdge*

4

1318.06

334.97

0.00

0.58

Site

Slope + DistEdge*

3

1320.84

337.75

0.00

0.56

2020 - Landscape Models (n = 1000)
Hyp3

Slope* + CC0–1 + CC1–3* + CC3–10* + CC>10*

6

981.17

0.00

0.96

0.81

Cover

CC0–1* + CC1–3* + CC3–10* + CC>10*

5

987.43

6.26

0.04

0.81

Hyp2

Slope* + CC3–10* + CC>10*

4

1004.93

23.76

0.00

0.80

Hyp1

Slope* + CC0–1* + CC1–3*

4

1058.77

77.59

0.00

0.77

Esc1

CC0–1* + DistEdge*

3

1224.64

243.47

0.00

0.58

Esc2

Slope + CC0–1*+ DistEdge*

4

1225.81

244.64

0.00

0.58

Site

Slope + DistEdge*

3

1229.61

248.44

0.00

0.56