SMALL MAMMAL COMMUNITIES IN TWO GRASSLAND ECOSYSTEMS
IN BRITISH COLUMBIA, CANADA

by

Gerad Shane Hales
B. Sc. The University of Victoria, 2000

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
Master of Science
in
Environmental Science
Thompson Rivers University

Spring Convocation 2011

© Gerad Shane Hales, 2011

DEDICATION

To all the mice, voles and shrews who begrudgingly
allowed themselves to captured, handled, tagged and
weighed during the course of this study.

To the ones who led us on a merry chase:
Big Hungry Joe
Brandine
Buckshot
Cilantro
Cousin Merl
Daisy Duke
Dirty Rotten Scoundrel
Dozer
Muffy
Penelope
Ruprecht the Monkey Boy
Sir Sniffs-A-Lot
Veruca Salt

To Barbie Lightning II, who carried my gear and
never once let me down.

And finally to deer mouse # 2101,
whose contribution shall not be forgotten.

ii

ACKNOWLEDGEMENTS
I would like to thank all of those who provided support for this project. Foremost, I
would like to thank Dr. Karl Larsen and Dr. Lauchlan Fraser for their guidance and
humour during all stages of this thesis. As well, I would like to acknowledge Cindy
Haddow of the Ministry of Environment for all her work in securing funds for this project
and for her enthusiastic contributions to the visual obscurity aspects of the study. I wish
to acknowledge the cast of thousands who provided field support, namely Helen Kerr and
Alysa McCall for their tireless efforts and support, and for putting up with all my
ignorance, mindless chatter and perpetual fear of the weather. This project would not
have been possible without the support of Michael Burwash and the rest of the Kamloops
Ministry of Environment crew who provided traps, trucks, gas, personnel, time and
expertise to this project. You all rock! My thanks also goes out to Eleanor Bassett, Mark
Boyce, Chuck Bulmer, Montana Burgess, Cameron Carlyle, Bruno Delesalle, Lisa
DeSandoli, Cynthia Fell, Alex Francis, Wendy Gardner, Louis Gosselin, Purnima
Govindarajulu, Felice Griffiths, Bill Harrower, Brian Heise, Chris Johnson, Joyce
Kabloona, Richard Klafki, Walt Klenner, Jo-Anne Mosterd, David Nagorsen, Dustin
Oaten, Morgan Rankin, Beckie Rozander, Carl Schwarz, Kate Sidlar, Eric Stromgren,
Dana and Brandon Taylor, Don Thompson, Rick Tucker, Michele Waring, Rich Weir
and Gary White for all of your assistance. I apologize for any omissions. And of course
my family and friends, all of whom I know took sick pleasure in constantly berating me
to get this thesis done! Your badgering was appreciated. This work was generously
supported by Thompson Rivers University, the Natural Sciences and Engineering
Research Council of Canada, the Grasslands Conservation Council of British Columbia,
the Ministry of Environment and the Habitat Conservation Trust Fund, the Ministry of
Forests and Range and the Forests and Range Evaluation Program, and the Endangered
Species Recovery Fund (World Wildlife Fund-Canada and Environment Canada). This
work was conducted under permit number KA05-15057 and KA08-4891, granted by the
British Columbia Ministry of Environment, and was approved by the Thompson Rivers
University Animal Care Committee, protocol no. 2008-01 and 2008-02.

iii

ABSTRACT
This study compared small mammal communities between upper and lower elevation
grassland systems using mark-recapture, and examined third and fourth order resource
selection of daytime refuge sites by deer mice (Peromyscus maniculatus) in lower
elevation grasslands using radio telemetry, near Kamloops, British Columbia, Canada.
Small mammal densities showed high levels of variability. Deer mice were found in both
habitat types, and survival rates between the two grassland types were not significantly
different. Voles (Microtus spp.) were confined to the upper grasslands. Radio-collared
deer mice selected daytime refuge sites in areas with increased slope and decreased litter
(third order), at sites with large-diameter shrubs, decreased levels of bare ground and
increased levels of coarse woody debris (fourth order). Land managers can use this
information to begin filling knowledge gaps in species-specific recovery plans, and to
help inform anthropogenic-related activities in grasslands so as to maintain rodent
populations on the landscape.
Key words: British Columbia, grassland, mark-recapture, Microtus, Peromyscus
maniculatus, radio telemetry, resource selection, small mammal

iv

TABLE OF CONTENTS
DEDICATION .................................................................................................................... ii
ACKNOWLEDGEMENTS ............................................................................................... iii
ABSTRACT ........................................................................................................................iv
LIST OF FIGURES ............................................................................................................vi
LIST OF TABLES ............................................................................................................ vii
CHAPTER 1 – INTRODUCTION: OF MICE AND MEN AND
GRASSLANDS...............................................................................................................1
LITERATURE CITED ....................................................................................................................... 9

CHAPTER 2 – SMALL MAMMAL COMMUNITIES IN TWO
TEMPERATE GRASSLAND ECOSYSTEMS OF BRITISH
COLUMBIA, CANADA .............................................................................................. 15
INTRODUCTION .......................................................................................................................... 15
METHODS ................................................................................................................................... 17
RESULTS...................................................................................................................................... 25
DISCUSSION................................................................................................................................ 42
LITERATURE CITED ..................................................................................................................... 47

CHAPTER 3 – SELECTION OF NESTS AND DAYTIME REFUGE SITES
BY DEER MICE IN A SEMI-ARID GRASSLAND IN BRITISH
COLUMBIA, CANADA .............................................................................................. 55
INTRODUCTION .......................................................................................................................... 55
METHODS ................................................................................................................................... 57
RESULTS...................................................................................................................................... 63
DISCUSSION................................................................................................................................ 67
LITERATURE CITED ..................................................................................................................... 72

CHAPTER 4 – CONCLUSIONS ...................................................................................... 79
LITERATURE CITED ..................................................................................................................... 85

v

LIST OF FIGURES
Figure 1.1. Photographs of typical upper grassland (A) and lower grassland
(B) landscapes. ................................................................................................................4
Figure 1.2. Study site locations in and around Lac du Bois Grasslands
Provincial Park, near Kamloops, British Columbia, Canada. Map
developed by M. Wolowicz. ........................................................................................... 7
Figure 1.3. Mean monthly total precipitation (mm) and mean monthly
average temperature (°C) for the 1976-2005 (●) and the 2006-2008 (○)
time periods. Data compiled from Environment Canada‟s National Climate
Data and Information Archive available at:
http://climate.weatheroffice.gc.ca/. November and December 2008 data for
mean total precipitation was not available. .....................................................................8
Figure 2.1. Snow shelter used to facilitate winter trapping (A), and the snow
shelter in situ with snow removed from the shelter‟s top (B). ......................................20
Figure 2.2. Mean population densities (animals/ha) for P. maniculatus in
upper (○) and lower (●) grasslands across all 15 trap sessions. ....................................28
Figure 2.3. Mean population densities (animals/ha) for Microtus spp. in upper
grasslands across all 15 trap sessions. ...........................................................................30
Figure 2.4. Mean population densities (animals/ha) for small mammals in
upper (○) and lower (●) grasslands across all 15 trap sessions. ....................................31
Figure 2.5. Mean percent female P. maniculatus in upper (○) and lower (●)
grasslands across all 15 trap sessions, and including a 50% reference line. .................33
Figure 2.6. Mean percent adult P. maniculatus in upper (○) and lower (●)
grasslands across all 15 trap sessions. ...........................................................................35
Figure 2.7. Small mammal population densities (animals/ha) the lower
grassland (●) and lower grassland winter (○) sites, and the upper grassland
( ) and upper grassland winter ( ) sites across the final three trapping
sessions of 2007, and the three lower grassland and two upper grassland
winter trapping sessions of 2008. ..................................................................................39
Figure 3.1. Layout of methodology for collecting habitat metrics at the third
order and fourth order levels for each telemetry location. ............................................60

vi

LIST OF TABLES
Table 2.1. Comparison of habitat characteristics (mean + s.e.) in the upper
and lower grassland sites using the Kruskal-Wallis test (d.f. = 1). ............................... 26
Table 2.2. Comparisons of the densities of P. maniculatus captured by year,
and between upper and lower elevation grasslands and trapping session
using two-way repeated-measures ANOVA tests. ........................................................ 29
Table 2.3. Comparisons of the densities of small mammals captured by year,
and between upper and lower elevation grasslands and trapping session
using two-way repeated-measures ANOVA tests. ........................................................ 32
Table 2.4. Comparisons of the percentages of female P. maniculatus captured
by year, and between upper and lower elevation grasslands and trapping
session using two-way repeated-measures ANOVA tests. ...........................................34
Table 2.5. Comparisons of the percentages of adult P. maniculatus captured
by year, and between upper and lower elevation grasslands and trapping
session using two-way repeated-measures ANOVA tests. ...........................................35
Table 2.6. Top Cormack-Jolly-Seber apparent survival models for P.
maniculatus by year for sex and elevation treatments. Survival, Φ;
recapture, p; time effects, (t); no effects, (.). .................................................................37
Table 2.7. Apparent survival rates of P. maniculatus per trap session interval,
by year and for lower and upper elevations. When > 1 model is listed,
model averaging was used to infer survival rates. Survival, Φ; recapture, p;
treatment effects, (g); time effects, (t); no effects, (.); additive effects (no
interaction terms), (g+t). ............................................................................................... 38
Table 2.8. Numbers of over-winter surviving individuals of P. maniculatus
and Microtus spp. for the 2006/07 and 2007/08 over-winter periods in the
lower and upper elevation grasslands. ..........................................................................40
Table 2.9. Spearman‟s rank correlation matrix between site characteristics
and mean densities of P. maniculatus and Microtus spp. Within each cell,
the uppermost number equals Spearman‟s correlation ρ and the lower
number equals the P-value. Dark grey cells indicate a strong positive
correlation (ρ > 0.50), and light grey cells indicate a strong negative
correlation (ρ < -0.50). ..................................................................................................41
Table 3.1. Habitat metrics measured at telemetry locations, third order and
fourth order habitat plots using a 1 × 1 m frame. .......................................................... 59
Table 3.2. Top 10 models used to examine refuge site selection by deer mice
at the fourth order scale, and detailing the number of model variables (K),
Akaike‟s Information Criterion adjusted for small sample sizes (AICc), the
change in AICc from most parsimonious model (ΔAICc), and the model
weight (w)...................................................................................................................... 65

vii

Table 3.3. β-coefficients and average measurements of the habitat metrics
included in the top model of the fourth order habitat analysis. Reported
average measurements are taken from the telemetry locations and fourth
order plots, with SageD recorded in centimetres, and Ground and CWD in
percent cover of the 1 × 1 m frame. ..............................................................................65
Table 3.4. Top three models used to examine refuge site selection by deer
mice at the third order scale, and detailing the number of model variables
(K), Akaike‟s Information Criterion adjusted for small sample sizes (AICc),
the change in AICc from most parsimonious model (ΔAICc), and the model
weight (w)...................................................................................................................... 66
Table 3.5. β-coefficients and average measurements of the habitat metrics
included in the top model of the third order habitat analysis. Reported
average measurements are taken from the third order and fourth order plots,
with Slope recorded in degrees, and Litter in percent cover of the 1 × 1 m
frame. ............................................................................................................................ 66

viii

CHAPTER 1 – INTRODUCTION: OF MICE AND MEN AND GRASSLANDS
Why are some species abundant in certain habitats, and yet absent from others? This
seemingly simple question has formed the basis of a large body of scientific work, and is
a matter ecologists continue to investigate. In general, the distribution and abundance of
a particular species can be influenced by dispersal habits, biotic (e.g. competition) and
abiotic (e.g. weather) factors, and by the selection and exploitation of resources (e.g.
food, shelter) available on the landscape (Krebs 1994). As adequate quantities of
resources are needed to support populations, understanding which resources are
important to animals provides valuable insights into how they meet their requirements for
survival (Manly et al. 2002).
The manner in which certain organisms choose to occupy a particular habitat or
exploit a resource can be viewed as a hierarchical process in which individuals make
selection choices at varying spatial scales. Johnson (1980) described this hierarchy of
choices as ordered selections. For example, members of a species may occupy a number
of distinct areas (e.g. forest vs. grassland) within the extent of its range: a first order
selection. Within each area, individuals may occupy a home range (second order
selection), and within each home range select areas for specific activities such as feeding
or nesting (third order selection). The location where a specific activity occurs (e.g. nest
site location) may be considered a fourth order selection. The selection criteria may vary
at each level (Johnson 1980, Orians and Wittenberger 1991), and so to gain better insight
into why a species occupies a certain area or displays changes in abundance, it is essential
to understand which resources may be important to it at various spatial scales.
Research on species within the Order Rodentia can offer excellent insights into
resource and habitat selection processes, and these animals have been extensively studied
with respect to this field (Krebs 1994). Their ubiquitous nature (i.e. found on all
continents except Antarctica and in a vast range of habitats), relatively short breeding
cycle and varied social structures have allowed scientists to use rodents to study a host of
ecological processes, including population cycles (e.g. Krebs et al. 1973, Korpimäki and
Krebs 1996, Krebs 1996, Boonstra et al. 1998, Korpimäki et al. 2004, Getz et al. 2006)

1

and competition and community assembly rules (e.g. Valone and Brown 1995, Brown et
al. 2000, 2002, Eccard and Ylönen 2003). Rodents have been extensively studied in
forest systems (e.g. Carey and Johnson 1995, Carey and Harrington 2001, Moses and
Boutin 2001, Sullivan and Sullivan 2001, 2004, Sullivan et al. 2001, Klenner and
Sullivan 2003, 2009, Larsen et al. 2007, Oaten and Larsen 2008), in desert and
desert/grassland systems (e.g. Brown and Heske 1990, Valone and Brown 1995, Brown
et al. 2000, Brown and Ernest 2002, Hernandez et al. 2005) and in the temperate
grasslands of North America (e.g. French et al. 1976, Grant and Birney 1979, Grant et al.
1982, Brady and Slade 2001, 2004, Howe and Brown 2001, Howe et al. 2002, 2006,
Howe and Lane 2004, Reed et al. 2007).

Grasslands, not surprisingly, are those areas where grasses (Poaceae) and grass-like
plants and/or forbs dominate the vegetative cover. As with rodents, grasslands are found
on every continent except Antarctica, and generally fall into two broad categories:
temperate and savannah. The former is characterized by the complete lack of trees, and
the latter containing scattered individual trees (Daubenmire 1978, Wikeem and Wikeem
2004). Temperate grasslands are known by many names, from the “puszta” of Hungary
and the “pampas” of Argentina and Uruguay, to the “veldts” of South Africa and the
“steppes” of Russia (Campbell et al. 1999). In North America, grasslands occupy
approximately 28% of the continent‟s land area (Wikeem and Wikeem 2004), with the
most prominent grassland being the central great plains of the United States of America.
Small mammals are important in the structuring and functioning of grasslands. In
these systems, rodents play vital roles as consumers of invertebrate species (Churchfield
et al. 1991), as seed dispersers (La Tourrette et al. 1971, McAdoo et al. 1983) and
consumers (Hulme 1994, Edwards and Crawley 1999), and as consumers of herbaceous
and woody material (Lindroth and Batzli 1984, Howe et al. 2002, 2006). They provide
ecological functions as prey for a variety of grassland predators (Reich 1981, Sera and
Early 2003), and as dispersers of spores of mycorrhizal fungi (Maser et al. 1988). Further
to their important trophic position in grassland systems, some rodents act as ecological

2

engineers who increase landscape heterogeneity by establishing unique patches different
from the surrounding landscape, and create habitat for other plant and wildlife species
(Davidson and Lightfoot 2008, Davidson et al. 2008). Their exceptional abilities to
drastically alter grassland vegetation and processes (Keesing 2000, Howe and Lane 2004,
Howe et al. 2006) highlight the need to understand how rodent populations change
temporally and spatially across different grassland ecosystems.
In British Columbia, Canada, grasslands cover only about 1% of the land base (0.74
million hectares) scattered over 11° of latitude and 25° of longitude, and can be divided
into two broad groups: the cooler grasslands found north of 52° N latitude, and the hotter
semi-arid grasslands which occur south of 52° N latitude where almost 90% of the
province‟s grasslands occur (Wikeem and Wikeem 2004). Previous studies into the
organization of these semi-arid grasslands revealed they have a definite sequence of
vegetative zones that occur on an elevation/precipitation gradient and divide the
grasslands into three fairly distinct zones: the lower, middle and upper grassland zones
(Tisdale 1947, van Ryswyk et al. 1966). The lower grasslands are confined to valley
bottoms, and are the hottest and driest of the grassland systems (Figure 1.1). Vegetation
in this zone consists of widely spaced bunchgrasses such as bluebunch wheatgrass
(Pseudoroegneria spicata) associated with big sagebrush (Artemisia tridentata). The
inter-shrub areas often are bare or partially lichen-covered (Tisdale 1947, van Ryswyk et
al. 1966). The cooler, more mesic upper grassland zone is characterized by nearly
continuous grass coverage, the lack of shrubs, greater plant richness, higher amounts of
yield and substantive levels of vegetative litter (Figure 1.1). In this zone, rough fescue
(Festuca campestris) can be the dominant plant species on northern and eastern slopes,
and often co-dominates with P. spicata on more xeric sites (Tisdale 1947, van Ryswyk et
al. 1966). The middle grasslands are transitional areas between the lower and upper
grassland zones, with moderate levels of grass coverage, plant richness, and vegetative
litter. Shrub species prevalent in the lower grasslands mostly are absent from this zone,
as are most specimens of F. campestris. Instead, species such as needle-and-thread grass
(Hesperostipa comata), Sandberg bluegrass (Poa secunda) and P. spicata can dominate

3

A

B

Figure 1.1. Photographs of typical upper grassland (A) and lower grassland (B)
landscapes.

4

the landscape of middle grassland climax communities (Tisdale 1947, van Ryswyk et al.
1966).
Although British Columbia‟s grasslands constitute only 1% of the province‟s land
base, they are utilized to some degree by 30% of the at-risk species in the province
(Wikeem and Wikeem 2004). A number of these species, including the burrowing owl
(Athene cunicularia), badger (Taxidea taxus), Great Basin gopher snake (Pituophis
catenifer deserticola), and the western rattlesnake (Crotalus oreganus) depend on small
mammals as sources of prey, further highlighting the need to understand small mammal
community dynamics in grassland settings.

The ultimate goal of my research was to increase the knowledge of small mammal
communities in temperate grasslands of British Columbia, Canada. To gain a more
complete understanding of these communities, my investigation had two fundamental
objectives:
1.

To collect and compare small mammal population and demographic information,
as well as habitat data, across two grassland communities; and

2.

To examine the third and fourth order resource selection of a specific habitat
feature by a dominant semi-arid grassland rodent.

To accomplish the first objective, an intensive small mammal live-trapping and markrecapture program took place from 2006 through 2008 using study sites in the upper and
lower grasslands. Small mammal densities, demographics and rates of apparent survival
were compared between the two grassland types, as were a number of habitat
characteristics, in order to confirm differences between the two grassland types. This
work is presented in Chapter 2 of this thesis. My second objective was accomplished by
examining the use of daytime refuge sites by deer mice (Peromyscus maniculatus) in the
lower grasslands using radio telemetry techniques. Chapter 3 details this work. In
Chapter 4, I discuss the potential management implications of my study.

5

My research was conducted in grassland systems near Kamloops, British Columbia,
Canada (50°43‟ N; 120°25‟ W). In total 10 sites were used for the study, with 5
occurring in each grassland type. Eight sites were utilized for the majority of the small
mammal trapping activities, while an additional two were established in the fall of 2007
to investigate the over-winter survival of small mammals (Figure 1.2). Following
provincial terminology, all of the lower grassland sites occurred within the Thompson
Very Dry Hot Bunchgrass Variant biogeoclimatic zone, and the upper grassland sites
were situated within the grassland phase of the Thompson Very Dry Hot Interior
Douglas-Fir Variant (termed „BGxh2‟ and „IDFxh2a‟ respectively, by Meidinger and
Pojar 1991). The eight main sites were established in 2006 by researchers at Thompson
Rivers University (M. Rankin working under the supervision of K. Larsen and L. Fraser),
and some trapping data also were collected that year. I incorporated these data into parts
of this thesis, using it to provide a more thorough picture of the small mammal
communities through time.
Climate in the area of my study generally is driven by weather systems moving east
from the Pacific Ocean, as well as by local topography (Tisdale 1947). As the study area
is situated within the rain shadow of the Coast and Cascade mountains, the local climate
is typical of the dry interior portion of the province. Precipitation levels follow the
previously-described elevation gradient, varying from about 240 mm in lower grasslands
at the valley bottom to 380 mm in grasslands above 850 m elevation (Wikeem and
Wikeem 2004), with the 30-year average (1976-2005) annual precipitation, as recorded at
the Kamloops International Airport within the lower grasslands, being 287 mm (data
compiled from Environment Canada‟s National Climate Data and Information Archive
available at: http://climate.weatheroffice.gc.ca/). Much of the precipitation falls during
the summer months in the form of thunder showers (Tisdale 1947), and relatively less
during winter months. An examination of the mean monthly total precipitation for the 30
years previous to the study and the mean monthly total precipitation that occurred during
the study indicates that the study period may have been drier than the years preceding it
(Figure 1.3).

6

Figure 1.2. Study site locations in and around Lac du Bois Grasslands Provincial Park,
near Kamloops, British Columbia, Canada. Map developed by M. Wolowicz.

7

Mean total precipitation (+ s.e.)

40

30

20

10

Mean average temperature (+ s.e.)

0

20

10

0

-10
Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep

Oct

Nov Dec

Month

Figure 1.3. Mean monthly total precipitation (mm) and mean monthly average
temperature (°C) for the 1976-2005 (●) and the 2006-2008 (○) time periods. Data
compiled from Environment Canada‟s National Climate Data and Information Archive
available at: http://climate.weatheroffice.gc.ca/. November and December 2008 data for
mean total precipitation was not available.

8

Average annual snowfall in the lower grasslands is 75 cm, and 126 cm in the upper
grasslands (Wikeem and Wikeem 2004). Rarely is a prolonged continuous snow cover
established in the lower grassland, as ground snow is intermittently removed by warmer
Chinook winds (Tisdale 1947). Snow in the upper grasslands can, however, form a
continuous and often deep snow layer that may last through April (Tisdale 1947). Snow
depth in the upper grasslands is highly variable, with winter winds creating snowdrifts
across the local topography.
Low elevation grasslands are considered to have some of the most extreme
temperature conditions in the province. Wikeem and Wikeem (2004) reported average
July temperatures in the lower grasslands to be 22.7°C, but daily highs can exceed 40°C.
Average January temperatures are relatively mild at -10°C, but extreme cold events can
decrease the temperature to below -20°C. Upper grasslands have milder summer
conditions, with average July temperatures being 18.9°C, but colder average January
temperatures (-13.4°C; Wikeem and Wikeem 2004). Mean monthly average
temperatures during the study period appear to closely mirror the mean monthly average
temperatures of the preceding 30 years (Figure 1.3).
The majority of the sites used in this study occurred either within, or very near to, Lac
du Bois Grasslands Provincial Park (Figure 1.2). Established in 1996, the park is one of
only three in the province that provides a significant amount of protection for grassland
ecosystems (Ministry of Environment, Lands and Parks 2000). Its proximity to the city
of Kamloops means it receives extensive amounts of anthropogenic use, including allterrain vehicle and motorcycling, horseback riding, cycling and hiking, and the grazing of
livestock. In general, these uses were not felt to exert a direct effect on the results of this
study, but their presence on the landscape must be acknowledged as part of the general
ecological stressors operating in the area of study.
LITERATURE CITED
BOONSTRA, R., C. J. KREBS, AND N. C. STENSETH. 1998. Population cycles in small
mammals: the problem of explaining the low phase. Ecology 79: 1479-1488.

9

BRADY, M. J., AND N. A. SLADE. 2001. Diversity of a grassland rodent community at
varying temporal scales: the role of ecologically dominant species. Journal of
Mammalogy 82: 974-983.
BRADY, M. J., AND N. A. SLADE. 2004. Long-term dynamics of a grassland rodent
community. Journal of Mammalogy 85: 552-561.
BROWN, J. H., AND S. K. M. ERNEST. 2002. Rain and rodents: complex dynamics of desert
consumers. BioScience 52: 979-987.
BROWN, J. H., B. J. FOX, AND J. A. KELT. 2000. Assembly rules: desert rodent
communities are structured at scales from local to continental. American Naturalist 156:
314-321.
BROWN, J. H., AND E. J. HESKE. 1990. Temporal changes in a Chihuahuan Desert rodent
community. Oikos 59: 290-302.
BROWN, J. H., J. A. KELT, AND B. J. FOX. 2002. Assembly rules and competition in desert
rodents. American Naturalist 160: 815-818.
CAMPBELL, N. E., J. B. REECE, AND L. G. MITCHELL. 1999. Biology. 5th ed.
Benjamin/Cummings, Menlo Park, California.
CAREY, A. B., AND C. A. HARRINGTON. 2001. Small mammals in young forests:
implications for management for sustainability. Forest Ecology and Management 154:
289-309.
CAREY, A. B., AND M. L. JOHNSON. 1995. Small mammals in managed, naturally young,
and old-growth forests. Ecological Applications 5: 336-352.
CHURCHFIELD, S., J. HOLLIER, AND V. K. BROWN. 1991. The effects of small mammal
predators on grassland invertebrates, investigated by field exclosure experiment. Oikos
60: 283-290.
DAUBENMIRE, R. F. 1978. Plant geography with special reference to North America.
Academic Press, New York.
DAVIDSON, A. D., AND D. C. LIGHTFOOT. 2008. Burrowing rodents increase landscape
heterogeneity in a desert grassland. Journal of Arid Environments 72: 1133-1145.

10

DAVIDSON, A. D., D. C. LIGHTFOOT, AND J. L. MCINTYRE. 2008. Engineering rodents
create key habitat for lizards. Journal of Arid Environments 72: 2142-2149.
ECCARD, J. A., AND H. YLÖNEN. 2003. Interspecific competition in small rodents: from
populations to individuals. Evolutionary Ecology 17: 423-440.
EDWARDS, G. R. AND M. J. CRAWLEY. 1999. Rodent seed predation and seedling
recruitment in mesic grasslands. Oecologica 118: 288-296.
FRENCH, N. R., W. E. GRANT, W. GRODZINSKI, AND D. M. SWIFT. 1976. Small mammal
energetics in grassland ecosystems. Ecological Monographs 46: 201-220.
GETZ, L. L., M. K. OLI, J. E. HOFMANN, AND B. MCGUIRE. 2006. Vole population
fluctuations: factors that initiate and determine intervals between them in Microtus
pennsylvanicus. Journal of Mammalogy 87: 841-847.
GRANT, W. E., AND E. C. BIRNEY. 1979. Small mammal community structure in North
American grasslands. Journal of Mammalogy 60: 23-36.
GRANT, W. E., E. C. BIRNEY, N. R. FRENCH, AND D. M. SWIFT. 1982. Structure and
productivity of grassland small mammal communities related to grazing-induced changes
in vegetative cover. Journal of Mammalogy 63: 248-260.
HERNANDEZ, L., A. G. ROMERO, J. W. LAUNDRE, D. LIGHTFOOT, E. ARAGON, AND J.
LOPEZ PORTILLO. 2005. Changes in rodent community structure in the Chihuahuan Desert
Mexico: comparisons between two habitats. Journal of Arid Environments 60: 239-257.
HOWE, H. F., AND J. S. BROWN. 2001. The ghost of granivory past. Ecology Letters 4:
371-378.
HOWE, H. F., J. S. BROWN, AND B. ZORN-ARNOLD. 2002. A rodent plague on prairie
diversity. Ecology Letters 5: 30-36.
HOWE, H. F., AND D. LANE. 2004. Vole-driven succession in experimental wet-prairie
restorations. Ecological Applications 14: 1295-1305.
HOWE, H. F., B. ZORN-ARNOLD, A. SULLIVAN, AND J. S. BROWN. 2006. Massive and
distinctive effects of meadow voles on grassland vegetation. Ecology 87: 3007-3013.

11

HULME, P. E. 1994. Post-dispersal seed predation in grasslands: its magnitude and
sources of variation. Journal of Ecology 82: 645-652.
JOHNSON, D. H. 1980. The comparison of usage and availability measurements for
evaluating resource preference. Ecology 61: 65-71.
KEESING, F. 2000. Cryptic consumers and the ecology of an African savanna. BioScience
50: 205-215.
KLENNER, W., AND T. P. SULLIVAN. 2003. Partial and clear-cut harvesting of highelevation spruce–fir forests: implications for small mammal communities. Canadian
Journal of Forest Research 33: 2283-2296.
KLENNER, W., AND T. P. SULLIVAN. 2009. Partial and clearcut harvesting of dry Douglasfir forests: implications for small mammal communities. Forest Ecology and
Management 257: 1078-1086.
KORPIMÄKI, E., P. R. BROWN, J. JACOB, AND R. P. PECH. 2004. The puzzles of population
cycles and outbreaks of small mammals solved. BioScience 54: 1071-1079.
KORPIMÄKI, E., AND C. J. KREBS. 1996. Predation and population cycles of small
mammals. BioScience 46: 754-764.
KREBS, C. J. 1994. Ecology: the experimental analysis of distribution and abundance. 4th
ed. HarperCollins College Publishers, New York.
KREBS, C. J. 1996. Population cycles revisited. Journal of Mammalogy 77: 8-24.
KREBS, C. J., M. S. GAINES, B. L. KELLER, J. H. MEYERS, AND R. H. TAMARIN. 1973.
Population cycles in small rodents. Science 179: 35-41.
LA TOURRETTE, J. E., J. A. YOUNG, AND R. A. EVANS. 1971. Seed dispersal in relation to
rodent activities in seral big sagebrush communities. Journal of Range Management 24:
118-120.
LARSEN, K. W., ADAMS, I. T., AND D. L. HAUGHLAND. 2007. Small mammal communities
in a pyrogenic habitat mosaic. International Journal of Wildland Fire 16: 728-740.
LINDROTH, R. L., AND G. O. BATZLI. 1984. Food habits of the meadow vole (Microtus
pennsylvanicus) in bluegrass and prairie habitats. Journal of Mammalogy 65: 600-606.

12

MANLY, B. F. J., L. L. MCDONALD, D. L. THOMAS, T. L. MCDONALD, AND W. P.
ERICKSON. 2002. Resource selection by animals – statistical design and analysis for field
studies. 2nd ed. Kluwer Academic Publishers, Dordrecht, Netherlands.
MASER, C., Z. MASER, AND R. MOLINA. 1988. Small-mammal mycophagy in rangelands
in central and southeastern Oregon. Journal of Range Management 41: 309-312.
MCADOO, J. K., C. C. EVANS, B. A. ROUNDY, J.A. YOUNG, AND R. A. EVANS. 1983.
Influence of heteromyid rodents on Oryzopsis hymenoides germination. Journal of Range
Management 36: 61-64.
MEIDINGER, D., AND J. POJAR. 1991. Ecosystems of British Columbia. British Columbia
Ministry of Forests, Research Branch, Victoria, Special Report Series Number 6.
MINISTRY OF ENVIRONMENT, LANDS AND PARKS. 2000. Lac du Bois Grasslands Park
management plan background document. Ministry of Environment, Lands and Parks,
Kamloops, British Columbia, Canada.
MOSES R. A., AND S. BOUTIN. 2001. The influence of clear-cut logging and residual leave
material on small mammal populations in aspen-dominated boreal mixedwoods.
Canadian Journal of Forest Research 31: 483-495.
OATEN, D. K., AND K. W. LARSEN. 2008. Aspen stands as small mammal “hotspots”
within dry forest ecosystems of British Columbia. Northwest Science 82: 276-285.
ORIANS, G., AND J. WITTENBERGER. 1991. Spatial and temporal scales in habitat selection.
American Naturalist 137: 29-49.
REED, A. W., G. A. KAUFMAN, AND B. K. SANDERCOCK. 2007. Demographic response of a
grassland rodent to environmental variability. Journal of Mammalogy 88: 982-988.
REICH, L. M. 1981. Microtus pennsylvanicus. Mammalian Species 159: 1-8.
SERA, W. E., AND C. N. EARLY. 2003. Microtus montanus. Mammalian Species 716: 1-10.
SULLIVAN, T. P., AND D. S. SULLIVAN. 2001. Influence of variable retention harvests on
forest ecosystems. II. diversity and population dynamics of small mammals. Journal of
Applied Ecology 38: 1234-1252.

13

SULLIVAN, T. P., AND D. S. SULLIVAN. 2004. Influence of granivorous diversionary food
on population dynamics of montane vole (Microtus montanus), deer mice (Peromyscus
maniculatus), and western harvest mice (Reithrodontomys megalotis). Crop Protection
23: 191-200.
SULLIVAN, T. P., D. S. SULLIVAN, AND E. J. HOGUE. 2001. Influence of diversionary foods
on vole (Microtus montanus and Microtus longicaudus) populations and feeding damage
to coniferous tree seedlings. Crop Protection 20: 103-112.
TISDALE, E. W. 1947. The grasslands of the southern interior of British Columbia.
Ecology 28: 346-382.
VALONE, T. J., AND J. H. BROWN. 1995. Effects of competition, colonization, and
extinction on rodent species diversity. Science 267: 880–883.
VAN RYSWYK, A. L., A. MCLEAN, AND L. S. MARCHAND. 1966. The climate, native

vegetation, and soils of some grasslands at different elevations in British Columbia.
Canadian Journal of Plant Science 46: 35-50.
WIKEEM, B., AND S. WIKEEM. 2004. The grasslands of British Columbia. Grasslands
Conservation Council of British Columbia, Kamloops, British Columbia, Canada.

14

CHAPTER 2 – SMALL MAMMAL COMMUNITIES IN TWO TEMPERATE
GRASSLAND ECOSYSTEMS OF BRITISH COLUMBIA, CANADA
INTRODUCTION
The processes that drive small mammal community structures are complex and have a
lengthy history of debate. One long-standing theory involves the idea of resource
availability, wherein increased primary productivity is positively associated with rodent
densities (Hernandez et al. 2005, Krebs et al. 2010). Simply put, increased precipitation
(or sunlight, heat, nutrients, etc.) leads to increased plant growth and reproduction, which
can support higher populations of consumers such as rodents (Brown and Ernest 2002).
This form of bottom-up trophic cascade has been shown to be particularly prevalent in
arid, pulse-driven systems (Beatley 1969, Ernest et al. 2000, Baez et al. 2006), and in
other terrestrial systems (Ostfeld and Keesing 2000, Falls et al. 2007). Moisture
gradients also have been found to be dominant drivers of rodent community structure in a
number of North American grassland systems (Grant and Birney 1979, Reed et al. 2007).
These changes often are observed at the functional group level, with increased plant
productivity and the accumulation of litter (dead plant matter) resulting in increased
abundances of herbivore species such as voles, and a decrease in granivorous small
mammals through processes such as a reduction in seed foraging efficiency (Kaufman
and Kaufman 1990, Reed et al. 2006a, 2006b). Yet primary productivity alone does not
dictate the structure of small mammal communities. Top-down forces (i.e. predation)
also have been shown to affect small mammal community structures (Hairston et al.
1960, Meserve et al. 1999), as have other biotic and abiotic factors such as competition
(Redfield et al. 1977, Valone and Brown 1995, Brady and Slade 2001) and microhabitat
availability (Morris 1984, 1987).
Understanding the mechanisms that affect small mammal communities is important, as
rodent populations often play vital roles in their ecosystems. In grassland systems,
rodents act as consumers of invertebrate species (Churchfield et al. 1991), as seed
dispersers (La Tourrette et al. 1971, McAdoo et al. 1983) and consumers (Hulme 1994,
Edwards and Crawley 1999), and as consumers of herbaceous and woody material

15

(Lindroth and Batzli 1984, Howe et al. 2002, 2006). They provide ecological functions
as prey for a variety of grassland predators (Reich 1981, Sera and Early 2003), and as
dispersers of spores of mycorrhizal fungi (Maser et al. 1988). Further to their important
trophic position in grassland systems, some rodents act as ecological engineers and
habitat modifiers who increase landscape heterogeneity by establishing unique patches
different from the surrounding landscape, and create habitat for other plant and wildlife
species (Davidson and Lightfoot 2008, Davidson et al. 2008).
The abilities of small mammals to drastically alter grassland vegetation and processes
(Keesing 2000, Howe and Lane 2004, Howe et al. 2006) highlight the need to understand
how their populations change temporally and spatially across different grassland
ecosystems. This need is particularly acute with respect to the grasslands of British
Columbia, Canada, which make up approximately 1% of the provincial land base (0.74
million hectares) yet are home to nearly 30% of the threatened and endangered species in
the province (Wikeem and Wikeem 2004). And while small mammal communities have
been studied extensively in other semi-arid and grassland systems throughout the world
(see previous references), they have received little attention in temperate British
Columbia grassland systems where they often are subject to extreme summer and winter
conditions.
Plant productivity and precipitation have been positively correlated in North American
grasslands (Sala et al. 1988). In some British Columbia grasslands, an elevation gradient
exists with precipitation increasing with increased elevation. This results in the
establishment of three distinct zones of vegetation: the lower, middle and upper elevation
grassland zones (Tisdale 1947, van Ryswyk et al. 1966). Low elevation sites (valley
bottoms to about 700 m elevation) are the hottest and driest of the grassland systems, and
are characterized by low plant diversity and the dominance of big sagebrush (Artemisia
tridentata), and large, bare inter-shrub areas with widely spaced bunchgrasses. Upper
elevation sites (850 m to 1130 m) are relatively more mesic, and are characterized by
increased plant species diversity, the lack of big sagebrush, nearly complete grass
coverage and a layer of dead vegetation, and by species such as rough fescue (Festuca

16

campestris) and bluebunch wheatgrass (Pseudoroegneria spicata) which often codominate in the mid- to late-seral stages of these communities (van Ryswyk et al. 1966,
Wikeem and Wikeem 2004). Given this elevation increase in plant productivity, one
might expect upper grasslands to support higher densities of small mammal species,
particularly litter-dwelling species that require higher amounts of vegetative litter and
cover to persist.
The primary objective of this study was to examine and compare the abundance,
composition and apparent survival of small mammal communities in two distinct
grassland ecosystems (lower and upper elevation grasslands). It is predicted that 1) total
small mammal densities and densities of herbivorous small mammal species (e.g.
Microtus spp.) will be higher in upper elevation grassland sites; 2) densities of more
omnivorous small mammal species will generally be higher in the lower grasslands; and
3) small mammal apparent survival rates will be higher in upper grassland sites.
METHODS
Study Area and Site Selection
This study took place in grassland ecosystems near Kamloops, British Columbia,
Canada (50°43‟ N; 120°25‟ W). Eight sites were utilized for the majority of the study,
and two supplementary sites were established in 2007 for winter surveys of small
mammal communities (see Figure 1.2). Five of the sites occurred within the lower
grasslands, and the other five sites within the upper grasslands. The four main sites in the
lower grasslands ranged in elevation from 434 m to 584 m, and the four main upper
grassland sites from 846 m to 981 m.
Small Mammal Trapping – Spring, Summer and Fall
Small mammals were live-trapped for three sessions between June and September
2006, and for six sessions between May and October in both 2007 and 2008. Most trap
sessions were separated by approximately 28 days (Getz et al. 2006). At each site, small

17

mammals were sampled with a 10 × 5 trapping grid (50 traps total) with 14.3 metres
between trap stations. A single Longworth-style live trap (Little Critter Traps, Rogers
Manufacturing, Kelowna, BC) was placed within a 2 m radius of each trap station, and
covered with a 15 × 30 cm board to provide protection from exposure to sun and rain.
Synthetic cotton bedding was added to the nesting chamber of each trap, and traps were
locked open and placed at each study site a week before the first trap session in order for
small mammals to become accustomed to their presence. Two days before the start of
each trapping session, traps were pre-baited with a small amount of a mixture of whole
oats and sunflower seeds and left open for two consecutive nights (Edalgo and Anderson
2007). After each pre-bait session, small mammals were live-trapped for three
consecutive nights. Traps were baited with approximately 5 g of the same mixture of
oats and sunflower seeds with a piece of apple or carrot included as a moisture source.
Traps were set within two hours of sunset and checked within two hours after sunrise the
following morning. Captured animals were identified to species and weighed using 60 g
spring scales (PESOLA AG, Baar, Switzerland). The sexual condition of each individual
was assessed using scoring techniques similar to McCravy and Rose (1992) and Moses
and Boutin (2001). For males, testes position (abdominal or scrotal) was noted, and for
females, teats were scored as either inactive (small and difficult to see), enlarged (teats
large; pregnant), lactating (teats large and fur worn), or returning to normal (teats healing
with fur re-growth). Each animal was tagged with a uniquely-numbered ear tag (Monel
#1005-1, National Band and Tag, Newport, KA) and released at the point of capture.
Whenever the number of animals caught at a particular site equalled or exceeded 35 on a
single morning (> 70% trap saturation), the number of traps at the site was increased by
50% for the remainder of the trapping session by adding an additional trap to every
second trap station (Parmenter et al. 2003). At the end of each trap session, residual bait
was removed from the traps to avoid supplemental feeding of small mammal populations
between trap sessions. Traps then were locked open and left in situ until the following
session.

18

Small Mammal Trapping – Winter
Small mammals were live-trapped at the two dedicated winter sites for three sessions
from August through October 2007, using methods identical to those described in section
above. These trapping sessions were completed in order to obtain estimates of small
mammal populations prior to the winter months. Three winter trapping sessions were
completed between January through March 2008 at the dedicated winter site and one
regular site (i.e. a site used for the spring, summer and fall trapping sessions) in the lower
grasslands, and two winter trapping sessions occurred at the dedicated winter site and one
regular site in the upper grasslands between January and February 2008. One winter site
and one regular site, as opposed to two regular sites, were trapped in each grassland type
so as to minimize any potential negative effects such winter trapping may have on the
small mammal populations at sites used for the majority of this study.
In November 2007, prior to the first snowfall of the season, each Longworth-style live
trap was placed inside a plastic trap shelter, based on the designs of Pruitt (1959) and
Iverson and Turner (1969), in order to prevent snow from hindering trap function. Each
shelter measured 33 × 28 × 13 cm (l × w × h), and on each side a 5.7 cm diameter hole
was cut to allow small mammal entrance. The shelters were covered with a 25.5 × 30.5
cm wooden removable top, to allow access to the trap while minimizing snow
disturbance around the trap site. To prevent the shelters from being overturned by wind
prior to snow cover, and to facilitate finding the traps under the snow, each shelter was
pinned to the ground using a 53 cm metal surveyor stake flag (Figure 2.1).
Trapping methods and animal processing otherwise followed procedures outlined for
the summer work, however, peanut butter was added to the bait and traps were set for a
maximum of 10 hours in order to minimize potential trap mortality (Pruitt 1959). To
access traps, snow was gently removed from the top of the shelters, and replaced once the
trap set/check was completed. Snowshoes were worn at all times while on the trapping
sites, and a single track was used to access all traps in an effort to minimize potential
disturbance to the subnivean space.

19

A

A

B

Figure 2.1. Snow shelter used to facilitate winter trapping (A), and the snow shelter in
situ with snow removed from the shelter‟s top (B).

20

Habitat Sampling
On each of the eight main live-trapping grids, eighteen 3 × 3 m semi-permanent plots
were systematically placed between trap stations, and were used to ascertain shrub
species and percent cover. Grass and forb species identification and percent cover, as
well as bare ground, crust (lichens and mosses), rock, coarse woody debris (CWD:
woody material > 2.5 cm diameter) and litter (above ground dead vegetation) cover
estimates were gauged by systematically placing three 20 × 50 cm frames in each of the
semi-permanent plots, and assessing the percent cover of each metric within the frame.
To assess herbaceous litter and biomass (above ground live vegetation) levels, as well
as sagebrush litter levels (un-rooted woody material < 2.5 cm diameter) at the lower
grassland sites only, three randomly-placed 70 m transects were established at each site.
From each transect, four litter and biomass samples were collected using 1 × 1 m frames
randomly placed along each transect. All litter and biomass samples were oven-dried for
48 hours and weighed.
Vertical and horizontal visibility levels were assessed in accordance to Carlyle et al.
(2010). To assess vertical visibility, four vole-sized pieces of square dowel (10 × 2.5 ×
2.5 cm) were placed systematically beneath the litter within a 0.5 × 0.5 m frame. A
digital photograph of the dowels was taken from 1 m above the ground directly above the
dowels with the assistance of a camera tripod. To assess horizontal visibility, a digital
photograph of a 50 × 50 cm board was taken at a distance of 4 m with the camera 1 m off
the ground. The dowels and board were painted fluorescent orange so as to contrast with
the surrounding vegetation. Using the 70 m random transects described above, photos of
both the board and dowels were taken at 7 m intervals along each transect. For
comparative purposes, reference photographs were taken of the board and dowels in
similar manners, but with no foliar obscurity. Digital photos were assessed using an
open-source photo software package called the GNU Image Manipulation Package
(Kimball and Mattis 2007). For each photograph, the number of orange pixels exposed
through the foliage was counted twice using the photo software and averaged. This

21

number was divided by the number of orange pixels counted on the reference
photographs to obtain a percent of the board or dowels exposed through the foliage.
Soil compaction was estimated by using a HFG-110 hand-held force gauge
(Transducer Techniques, Temecula, CA). The instrument measures the force needed to
push a gauge with a 12.57 mm2 tip through the soil, and expresses that measure in units
of kilogram-force (kgf). One kilogram-force is equivalent to 9.81 newtons. At each of
the eight live trapping grids, 20 square holes approximately 20 cm deep were
systematically dug between trap stations. Compaction was estimated by pushing the
force gauge into each of the four sides of the hole at depths of 1 and 10 cm, and
averaging the resulting measurements at each hole for each soil depth.
Small Mammal Abundance and Apparent Survival
Program CAPTURE (Otis et al. 1978) was used to generate closed population
estimates by trap session and site for small mammal species when 11 or more individuals
were captured (Moses and Boutin 2001). When 10 or fewer individuals were captured
during a trap session, the total number of individuals caught during the trap session was
used as a measure of abundance. When CAPTURE selected a model for which
abundance was not estimable, the next best model as picked by CAPTURE was used to
estimate abundance. All abundance estimates were then converted to densities (animals
per hectare).
To determine if small mammal survival differed between upper and lower elevation
sites, Cormack-Jolly-Seber (CJS) recapture models were developed using program
MARK (White and Burnham 1999). MARK allows for the estimation of apparent
survival (Φ) and recapture rate (p) parameters using the method of maximum likelihood
estimation (White 2008), and can incorporate treatment (g) and time (t) effects and their
interactions (Larsen et al. 2007). Within each year of study, apparent survival rates
between males and females by species were first compared and then comparisons
between upper and lower grasslands were made. An a priori set of 25 candidate models
were used to compare treatments. Models ranged from the fully parameterized global

22

model (Φ(g*t) p(g*t)), to the “no effect” model for both treatment and time (Φ(.) p(.)).
Akaike‟s Information Criterion adjusted for small sample sizes (AICc) was used to
compare models (Burnham et al. 1995).
Prior to examining the models, the goodness-of fit of the most global model (Φ(g*t)
p(g*t)) was tested using bootstrapping techniques (200 simulations) available within
MARK. The over-dispersion factor of the data (ĉ) was determined by dividing the
observed deviance of the global model by the mean deviance of the simulated
(bootstrapped) data, and was used to account for data over-dispersion by adjusting the
AICc to the quasi-AICc (QAICc; Cooch and White 2008). The number of parameters for
each model was hand-calculated and adjusted as necessary in order to ensure the correct
ranking of models by MARK. Model selection was based on the model with the lowest
QAICc value (Lebreton et al. 1992). Overall survival estimates were derived from the
model with the lowest QAICc value. When the best model‟s QAICc differed from the
next best model(s) by < 2.00, model averaging of these models was performed to estimate
survival parameters (Cooch and White 2008).
Small Mammal Demographics
A number of demographic characteristics were used to compare small mammal
communities between treatments. These characteristics included sex ratios, proportion of
populations composed of juveniles and the proportion of populations composed of
reproductive adult females.
Weight at sexual maturity has been used to distinguish adults from juveniles in small
mammal populations (Keller and Krebs 1970, Fairbairn 1977b). However, the age and
weight at which some small mammal species become sexually mature can vary between
years (Negus et al. 1977), can be influenced by social condition and diet (Negus and
Pinter 1966), and can vary between the sexes (Hoffman 1958). In accordance with
Sullivan and Sullivan (2004), mass at sexual maturity was ascertained by determining the
weight at which 50% of the animals displayed signs of sexual maturity. This was
accomplished by plotting the percent of males and females showing signs of sexual

23

maturity (males = testes scrotal; females = nipples enlarged, lactating or returning to
normal) in each one-gram weight category. A least squares regression was performed on
the data, and the predicted weight at which 50% of the animals showed signs of sexual
maturity was used as the point at which sexual maturity was reached. Female deer mice
were classified as adults if they weighed > 21 g, male deer mice if they weighed > 19 g,
and male voles if the weighed > 26 g. Due to the highly variable nature of the female
vole data, a least squares regression did not produce reliable results. For female voles,
weight at sexual maturity was determined by ascertaining the lowest weight class in
which > 50% of the animals showed signs of sexual maturity; which was > 26 g. These
mass limits are consistent with those described by Sullivan and Sullivan (2004).
Statistical Considerations
Small mammal density and demographic data were analyzed separately by year, using
two-way repeated-measures ANOVA tests with elevation as the independent variable and
trapping session as the repeated-measure factor. For each repeated-measures ANOVA, a
variety of covariance models were developed in a manner consistent with Littell et al.
(2002). The model with the lowest AICc value was used to draw conclusions from the
repeated-measures data. When a significant interaction was observed, the simple effects
were further analyzed using „slicing‟ methods described by Littell et al. (2002). In
addition, mean population densities of the June, July and September trapping sessions
were compared between years for both the upper and lower grasslands separately using a
one-way ANOVA. Paired t-tests were used to examine the differences between lower
and upper elevation mean small mammal densities and demographic information per trap
session over the course of the entire study.
Habitat metrics were compared between upper and lower elevation sites using a
nonparametric Kruskal-Wallis test. Further, plant cover data were used to assess plant
species diversity using diversity indices calculated with the Microsoft Office Excel (vers.
2007) add-in module Diversity.xla (available at:
http://www.reading.ac.uk/ssc/software/diversity/diversity.html). Diversity indices

24

generally fall into two types: those that place more emphasis on rare species in the
sample (i.e. Type I) and those that put most weight on the common species (i.e. Type II;
Peet 1974). As such, both the Shannon-Weiner diversity index (Pielou 1966), a Type I
index, and the Simpson‟s diversity index (Simpson 1949), a Type II index, were
calculated for the analysis.
Spearman‟s rank correlation analysis (Zar 1999) was used to examine the association
between small mammal species densities at each site and the habitat metrics collected at
those sites, including plant species richness and diversity. The mean abundances of each
small mammal species across the 2007 and 2008 trapping sessions were used for the
analysis, and correlations were considered strongly positive when ρ > 0.50 and strongly
negative when ρ < -0.50 (Cohen 1988).
All data were analyzed using SAS 9.2 software (SAS Institute Inc. 2008). For the
one-way ANOVA tests and t-tests, assumptions of normality were tested for all data and
when these assumptions were not met, data were either transformed or the corresponding
nonparametric analysis was performed using methods described in Schlotzhauer and
Littell (1997). A significance level of α = 0.05 was used for all tests, except that
Bonferroni tests with the appropriately-adjusted α- or P-values were used for all post-hoc
multiple comparisons and when the simple effects for significant interactions were
examined.
RESULTS
Site Characteristics
Elevational differences between the eight main upper and lower sites were significant
(lower elevation mean = 532.3 m, upper elevation mean = 906.5 m; H = 5.33, d.f. = 1, P
= 0.02), and Table 2.1 summarizes the differences in habitat variables between the upper
and lower grassland sites. As expected, upper elevation sites had higher amounts (kg/ha)
of herbaceous litter and biomass, and higher levels of litter and grass coverage (m2/ha).

25

Table 2.1. Comparison of habitat characteristics (mean + s.e.) in the upper and lower
grassland sites using the Kruskal-Wallis test (d.f. = 1).
Habitat Variable

Lower Elevation
Mean

Upper Elevation
Mean

H

P

Herbaceous litter (kg/ha)
Herbaceous biomass (kg/ha)
Shrub litter (kg/ha)

457.8 (133.9)
490.3 (138.5)
779.3 (62.1)

2088.6 (604.2)
1520.7 (188.7)
0.0

5.33
5.33
-

0.02
0.02
-

Bare ground (m2/ha)
Coarse woody debris (m2/ha)
Crust (m2/ha)
Grass (m2/ha)
Litter (m2/ha)
Rock (m2/ha)
Shrub (m2/ha)

1999.3 (362.5)
631.0 (72.7)
3019.7 (90.9)
3322.5 (537.4)
5116.7 (361.9)
343.8 (183.1)
2765.3 (400.0)

169.4 (75.8)
0.0
756.9 (372.5)
8144.9 (393.4)
9251.9 (295.0)
80.3 (45.4)
0.7 (0.7)

5.33
5.33
5.33
5.33
3.00
5.60

0.02
0.02
0.02
0.02
0.08
0.02

2.0

2.5

2.08

0.15

4.8

4.7

0.08

0.77

0.5

0.6

2.08

0.15

0.7

0.4

4.08

0.04

17.8 (3.2)

24.3 (2.3)

2.08

0.15

1.7 (0.3)

2.0 (0.1)

1.33

0.25

0.72 (0.1)

0.79 (0.0)

0.33

0.56

Soil compaction:
1 cm depth (kgf)A
Soil compaction:
10 cm depth (kgf)A
Horizontal visibility
(% visibility)B, C
Vertical visibility
(% visibility)B
Plant Richness
Plant Shannon-Weiner
diversity index
Plant Simpson‟s diversity
index
A

The means presented here are the logarithmic means back-transformed to the original
measurement scale. Standard errors are not provided as per Krebs (1999).
B
The means presented here are the arcsine means back-transformed to the original measurement
scale. Standard errors are not provided as per Krebs (1999).
C
Normality could not be obtained for this metric.

26

On the other hand, lower grasslands had greater levels of bare ground, crust and shrub
coverage. The amount of rock coverage was not significantly different between the two
habitat types, nor was soil compaction at either the 1 cm or the 10 cm depths. The lower
grasslands showed lower amounts of horizontal visibility than the upper grasslands,
probably due to the higher amounts of shrub coverage, whereas the upper grasslands had
significantly lower amounts of vertical visibility, most likely a result of the upper
grassland‟s higher amounts of herbaceous litter and biomass. Upper grassland sites
tended to have higher plant species richness than lower grassland sites, but the
differences were not significant. Sites did not differ with respect to plant species richness
or diversity. Abundant species, based on percent cover estimates, at upper grassland sites
included rough fescue (Festuca campestris), Kentucky bluegrass (Poa pratensis), and
needle-and-thread grass (Hesperostipa comata). In the lower grasslands, bluebunch
wheatgrass (Pseudoroegneria spicata), Sandberg bluegrass (Poa secunda) and needleand-thread-grass commonly were the abundant species. In addition, each lower grassland
site had varying amounts of big sagebrush (Artemisia tridentata) cover.
Small Mammal Totals
Over the course of the study 1,420 individual small mammals of three main species
were captured over 21,086 trap nights. Deer mice (Peromyscus maniculatus) were the
most numerous, with 906 animals caught (63.8% of total). Montane voles (Microtus
montanus) were the second most abundant with an estimated 483 individuals captured
(34.0% of total), followed by an estimated 31 meadow voles (Microtus pennsylvanicus;
2.2% of total). In addition, 15 shrews (Sorex spp.) were captured in the upper grasslands.
Shrews were considered incidental captures, and were released at the point of capture
without being identified to species. Due to the apparently low numbers of M.
pennsylvanicus captured in this study, and the difficulty of differentiating this species
from M. montanus in the field, data on these two species were combined for all
subsequent analyses. Deer mice were captured on all sites, whereas voles were captured
almost exclusively in the upper grasslands. Only 14 of the 514 voles captured in this

27

study were found in the lower grasslands, and all were caught during the 2006 trapping
season.
Small Mammal Density Estimates
The mean densities of deer mice in the upper and lower grasslands for all trap sessions
are illustrated in Figure 2.2. Densities were relatively high in 2006, then followed similar
patterns in 2007 and 2008 in both the upper and lower grasslands by remaining relatively
low or decreasing in May through July and then steadily increasing to peak in October.
A paired t-test examining the differences between lower and upper elevation mean deer
mouse densities per trap session over the course of the study showed no significant
differences between the two grassland types (t = 1.68, d.f. = 14, P = 0.12).

Mean deer mouse density (+ s.e.)

40
35
30
25
20
15
10
5
0
06-06 06-07 06-09 07-05 07-06 07-07 07-08 07-09 07-10 08-05 08-06 08-07 08-08 08-09 08-10

Trap session (yy-mm)

Figure 2.2. Mean population densities (animals/ha) for P. maniculatus in upper (○) and
lower (●) grasslands across all 15 trap sessions.

Table 2.2 details the results of the repeated-measures ANOVA tests for deer mouse
densities. In 2006, lower grasslands had significantly higher densities of deer mice than
28

Table 2.2. Comparisons of the densities of P. maniculatus captured by year, and between
upper and lower elevation grasslands and trapping session using two-way repeatedmeasures ANOVA tests.
Elevation
2006
2007
2008

Session

Elevation × Session

F

d.f.

P

F

d.f.

P

F

d.f.

P

22.71
1.82
0.57

1, 6
1, 6
1, 6

< 0.01
0.23
0.48

2.77
9.00
5.11

2, 11
5, 30
5, 30

0.11
< 0.01
< 0.01

0.65
4.56
1.99

2, 11
5, 30
5, 30

0.54
< 0.01
0.18

upper grasslands. In 2007, a significant difference between trap sessions was observed;
however, a significant interaction was also observed. An examination of the simple
effects showed that, when the data were sliced by trap session (α = 0.008) only trap
session 07-05 had a significant difference in densities between the two grassland types (P
< 0.01), whereas the other trap sessions did not (07-06: P = 0.02; 07-07: P = 0.58; 07-08:
P = 0.53; 07-09: P = 0.41; 07-10: P = 0.45). Slicing the data by elevation (α = 0.025)
revealed that both the lower and upper elevations had significant differences in deer
mouse densities between trap sessions (P < 0.01 and P < 0.01, respectively). For 2008,
densities differed significantly by trap session, but not by elevation. Post-hoc analyses of
the trap session data showed that session 08-05 was significantly different than session
08-10, and that sessions 08-06 and 08-07 were both significantly different than sessions
08-09 and 08-10.
A one-way ANOVA comparing densities between all three years at each elevation
separately showed a significant difference between years for the lower grasslands (F =
29.36, d.f. = 2,33, P < 0.01), with 2006 densities significantly higher than 2007 and 2008
densities, and 2007 densities significantly higher than those of 2008. Overall, no
substantive differences were seen between years in the upper grasslands (F = 0.23, d.f. =
2,32, P = 0.80).
As voles (Microtus spp.) were all but absent from the lower grasslands, no
comparisons between upper and lower elevations were made. Further, one-way repeatedmeasures ANOVA tests could not be performed due to insufficient „error‟ degrees of

29

freedom. Alternatively, one-way ANOVA tests or the equivalent non-parametric tests
were used to examine the changes, per year, in vole densities over trap sessions.
In the upper grasslands, vole populations expressed high levels of variability in 2006,
with some sites having densities as high as 118 animals per hectare, and some sites
having no animals in each of the three trap sessions that year. Vole densities decreased
dramatically from September 2006 to May 2007, and remained low throughout the 2007
season. Densities further declined in 2008, and voles were essentially absent from all
trapping sites from May through August 2008, and then came back to an average high of
11 animals per hectare in October 2008 (Figure 2.3). In all years, there were no
significant differences in densities between trapping sessions (2006: H = 2.14, d.f. = 2, P
= 0.37; 2007: F = 0.88, d.f. = 5,18, P = 0.51; 2008: H = 6.85, d.f. = 5, P = 0.23). A nonparametric comparison of average vole densities between years indicated significant
differences in densities between years (H = 9.54, d.f. = 2, P = 0.01), with 2006 densities
probably significantly higher than either 2007 or 2008 levels.

120

Mean vole density (+ s.e.)

100

80

60

40

20

0
06-06 06-07 06-09 07-05 07-06 07-07 07-08 07-09 07-10 08-05 08-06 08-07 08-08 08-09 08-10

Trap session (yy-mm)

Figure 2.3. Mean population densities (animals/ha) for Microtus spp. in upper grasslands
across all 15 trap sessions.

30

The densities of deer mice and voles were combined by trap session in order to
compare total small mammal densities between upper and lower elevations, and between
trapping sessions. Generally, small mammal densities were greater in upper grasslands
than in lower grasslands (Figure 2.4). The exceptions were the first two trap sessions of
2007 and the first session of 2008, where lower grassland mean densities were marginally
higher than those at the upper sites. Pooled across years, mean densities between upper
and lower grasslands were just significantly different (t = -2.16, d.f. = 14, P = 0.05), with

Mean small mammal density (+ s.e.)

the upper grasslands having higher mean densities than the lower grasslands.

120

100

80

60

40

20

0
06-06 06-07 06-09 07-05 07-06 07-07 07-08 07-09 07-10 08-05 08-06 08-07 08-08 08-09 08-10

Trap session (yy-mm)

Figure 2.4. Mean population densities (animals/ha) for small mammals in upper (○) and
lower (●) grasslands across all 15 trap sessions.

Table 2.3 details the results of the two-way repeated-measures ANOVA tests for total
small mammal densities. In 2006, the interaction between elevation and trap session was
significant. An examination of the simple effects showed that, when the data were sliced
by trap session (α = 0.02), none of the trap sessions showed a significant difference in
densities between the two grassland types (06-06: P = 0.59; 06-07: P = 0.04; 06-09: P =

31

Table 2.3. Comparisons of the densities of small mammals captured by year, and
between upper and lower elevation grasslands and trapping session using two-way
repeated-measures ANOVA tests.
Elevation
2006
2007
2008

Session

Elevation × Session

F

d.f.

P

F

d.f.

P

F

d.f.

P

3.09
1.46
3.80

1, 6
1, 6
1, 6

0.13
0.27
0.10

7.11
1.66
6.61

2, 11
5, 30
5, 30

0.01
0.17
< 0.01

6.46
2.50
2.04

2, 11
5, 30
5, 30

0.01
0.05
0.10

0.10). Slicing the data by elevation (α = 0.025) revealed that the upper elevation had
significant differences in deer mouse densities between trap sessions (P = 0.01) whereas
the lower grasslands did not (P = 0.95). A significant interaction was again observed in
2007, and when sliced by trap session (α = 0.008), once more none of the trap sessions
showed a significant difference in densities between the two elevations (07-05: P = 0.40;
07-06: P = 0.82: 07-07: P = 0.16; 07-08: P = 0.01; 07-09: P = 0.05; 07-10: P = 0.64).
Slicing the data by elevation (α = 0.025) showed that neither the lower nor upper
grasslands had significant differences in densities between trap sessions (P = 0.05 and P
= 0.06, respectively). For 2008, densities between trap sessions were significantly
different, and the post-hoc analysis showed that sessions 08-05 through 08-08 were all
significantly different than sessions 08-09 and 08-10.
A one-way ANOVA comparing small mammal densities by year in the upper and
lower grasslands separately showed significant differences between years in the lower
grasslands (F = 29.79, d.f. = 2,33, P < 0.01), with 2006, 2007 and 2008 densities all
significantly different from each other. Densities also were significantly different by year
in the upper grasslands (F = 3.48, d.f. = 2,32, P = 0.04) with 2006 densities significantly
higher than 2008 levels.
Deer Mouse Demographics
As P. maniculatus was the only species caught in both the upper and lower grasslands,
only their demographics are considered in depth here.
32

The percentages of female deer mice captured per trap session were highly variable,
ranging from 0.00 to 1.00 (Figure 2.5). Overall, upper grassland sites had marginally
significantly higher mean percentages of females in the population than lower sites, when
analyzed using a paired t-test (t = -2.11, d.f. = 14, P = 0.05), yet this was not always the

Mean percent female deer mice (+ s.e.)

case for each year of study.

1.0

0.8

0.6

0.4

0.2

0.0
06-06 06-07 06-09 07-05 07-06 07-07 07-08 07-09 07-10 08-05 08-06 08-07 08-08 08-09 08-10

Trap session (yy-mm)

Figure 2.5. Mean percent female P. maniculatus in upper (○) and lower (●) grasslands
across all 15 trap sessions, and including a 50% reference line.

Table 2.4 details the results of the repeated-measures ANOVA tests comparing the
mean percentages of female P. maniculatus between upper and lower elevation
grasslands and trapping session, and separated by year. In 2006, there was a significant
interaction between the independent variables elevation and session, as well as a
significant difference in trap session values. An evaluation of the simple effects revealed
that, when sliced by trap session (α = 0.017), none of the sessions showed significant
density differences between the upper and lower grasslands (06-06: P = 0.11; 06-07: P =
0.93; 06-09: P = 0.33). When sliced by elevation (α = 0.025), the upper grasslands

33

Table 2.4. Comparisons of the percentages of female P. maniculatus captured by year,
and between upper and lower elevation grasslands and trapping session using two-way
repeated-measures ANOVA tests.
Elevation
2006
2007
2008

Session

Elevation × Session

F

d.f.

P

F

d.f.

P

F

d.f.

P

0.08
6.94
0.91

1, 6
1, 6
1, 6

0.78
0.04
0.38

76.36
1.26
0.32

2, 10
5, 30
5, 26

< 0.01
0.31
0.89

46.80
1.48
0.06

2, 10
5, 30
5, 26

< 0.01
0.23
> 0.99

showed significant differences between trap sessions (P < 0.01), but the lower grasslands
did not (P = 0.07). 2007 was the only year to show a significant difference between
elevation, and for 2008, the percentages of female deer mice caught did not significantly
differ between either elevation or trap session, nor was the interaction significant.
The one-way ANOVA comparing small mammal densities by year in the upper and
lower grasslands separately showed there were no significant differences in percentages
of females caught between years in both the lower grasslands (F = 0.48, d.f. = 2,32, P =
0.62) and the upper grasslands (F = 1.25, d.f. = 2,31, P = 0.30).
The proportion of adult deer mice in the population was usually higher at lower
grassland sites than at upper elevations, and comparing the mean proportions between the
two grassland types using a paired t-test showed the differences to be highly significant (t
= 28.42, d.f. = 14, P < 0.01). Generally, the proportion of adults increased through the
summer months, and then decreased dramatically with the onset of fall (Figure 2.6).
In 2006, both a significant interaction and a significant difference between trap
sessions were noted (Table 2.5). An analysis of the simple effects showed that, when
sliced by trap session (α = 0.017), session 06-06 showed significant differences between
elevations (P < 0.01), session 06-07 a nearly significant difference (P = 0.04), and
session 06-09 a non-significant difference (P = 0.10). When sliced by elevation (α =
0.025), the upper grasslands had significant differences between trap sessions (P < 0.01),
whereas the lower grasslands did not (P = 0.34). In 2007, significant differences in
elevation and by trapping session were noted, with the lower grasslands having higher

34

Mean percent adult deer mice (+ s.e.)

1.0

0.8

0.6

0.4

0.2

0.0
06-06 06-07 06-09 07-05 07-06 07-07 07-08 07-09 07-10 08-05 08-06 08-07 08-08 08-09 08-10

Trap session (yy-mm)

Figure 2.6. Mean percent adult P. maniculatus in upper (○) and lower (●) grasslands
across all 15 trap sessions.

Table 2.5. Comparisons of the percentages of adult P. maniculatus captured by year, and
between upper and lower elevation grasslands and trapping session using two-way
repeated-measures ANOVA tests.
Elevation
2006
2007
2008

Session

Elevation × Session

F

d.f.

P

F

d.f.

P

F

d.f.

P

0.02
18.42
27.4

1, 6
1, 6
1, 6

0.90
< 0.01
< 0.01

4.06
3.90
4.50

2, 10
5, 30
5, 26

0.05
< 0.01
< 0.01

12.10
1.39
5.87

2, 10
5, 30
5.26

< 0.01
0.26
< 0.01

levels of adults in the population. Post-hoc analysis of the trapping sessions showed that
the 07-07 and 07-10 trapping sessions were significantly different. In 2008, a significant
interaction and significant differences between elevations and trap sessions were
observed. An analysis of the simple effects showed that, when sliced by trap session (α =

35

0.008), sessions 08-06, 08-07 and 08-09 showed significant differences between
elevations (P < 0.01, P < 0.01 and P < 0.01, respectively), whereas sessions 08-05, 08-08
and 08-10 did not (P = 0.83, P = 0.02 and P = 0.02, respectively). When sliced by
elevation (α = 0.025), the upper grasslands had significant differences between trap
sessions (P = 0.01), as did the lower grasslands (P < 0.01).
Compared across years, there were significant yearly differences of proportions of
adults caught in the lower grasslands (F = 4.32, d.f. = 2,32, P = 0.02) with 2006 and
2007 having significantly different proportions of adults in their populations. In the
upper grasslands, the proportion of adults in the populations also were significantly
different between years (H = 8.71, d.f. = 2, P = 0.01).
Due to the small numbers of adult female deer mice trapped over the course of the
study, a comparison of the proportions of reproducing adult females by year and trapping
session was not performed. Instead, data for the June, July and September trapping
sessions for all years of study were pooled by elevation, and the proportions of
reproductively active adult female deer mice compared between the upper and lower
grasslands using a Kruskal-Wallis test. As well, data for all trap session in 2007 and
2008 were pooled by elevation, and the proportions of reproductively-active adult female
deer mice compared between the upper and lower grasslands. Comparing the June, July
and September values across all years of study showed that there was no significant
difference between the two elevations (H = 2.08, d.f. = 1, P = 0.15). The comparison of
the pooled 2007 and 2008 data showed similar results (H = 3.00, d.f. = 1, P = 0.08).
Deer Mouse Apparent Survival and Elevation
As P. maniculatus was the only the species caught in both upper and lower grasslands,
only their apparent survival across the elevation gradient was compared.
The most appropriate models describing monthly P. maniculatus apparent survival
rates in 2006, 2007 and 2008 were ones that did not include differences between males
and females (Table 2.6), and as such, the sexes were combined for all subsequent
analyses. The analysis showed that P. maniculatus survived equally well in upper and

36

Table 2.6. Top Cormack-Jolly-Seber apparent survival models for P. maniculatus by
year for sex and elevation treatments. Survival, Φ; recapture, p; time effects, (t); no
effects, (.).
Year

Treatment

Top Model

2006
2007
2008
2006
2007
2008

Sex
Sex
Sex
Elevation
Elevation
Elevation

Φ(.) p(.)
Φ(.) p(t)
Φ(t) p(.)
Φ(.) p(.)
Φ(.) p(t)
Φ(t) p(.)

lower grasslands in each year of study, with the top models for each year of study lacking
any treatment (i.e. elevation) effect on the survival parameter (Table 2.6). There were
indications of slight differences in apparent survival rates between the two elevation
classes, however, as the next best models describing the data (i.e. models that had
QAICcs which differed from the top model by < 2.00) did include an elevation treatment
effect (Table 2.7).
Overall apparent survival rates for P. maniculatus in the lower grasslands ranged from
0.28 + 0.11 to 0.81 + 0.10, and in the upper grasslands from 0.29 + 0.11 to 0.82 + 0.10
(Table 2.7). In 2006 and 2007, apparent survival rates remained relatively constant for
all trap session intervals, respectively, and only slight differences in rates were observed
between upper and lower elevations. In 2008, a time component was included in the top
model, and for that year, apparent survival rates in both elevations began the trapping
season quite low, steadily increased throughout the trapping season, and peaked in the
final trap session interval.

37

Table 2.7. Apparent survival rates of P. maniculatus per trap session interval, by year
and for lower and upper elevations. When > 1 model is listed, model averaging was used
to infer survival rates. Survival, Φ; recapture, p; treatment effects, (g); time effects, (t);
no effects, (.); additive effects (no interaction terms), (g+t).

Year

Model(s) Used

Elevation

2006

Φ(.) p(.), Φ(.) p(g+t), Φ(.) p(t),
Φ(t) p(.), Φ(g) p(.), Φ(.) p(g)

Lower
Upper

2007

Φ(.) p(t), Φ(g) p(t),

2008

Φ(t) p(.), Φ(t) p(g), Φ(g+t) p(.)

Lower
Upper
Lower

Upper

Trap Session
Interval
(yy-mm to yy-mm)

Apparent
Survival Rate
(+ s.e.)

06-06 to 06-07
06-07 to 06-09
06-06 to 06-07
06-07 to 06-09
All sessions
All sessions
08-05 to 08-06
08-06 to 08-07
08-07 to 08-08
08-08 to 08-09
08-09 to 08-10
08-05 to 08-06
08-06 to 08-07
08-07 to 08-08
08-08 to 08-09
08-09 to 08-10

0.46 (0.15)
0.44 (0.14)
0.44(0.17)
0.43 (0.16)
0.52 (0.04)
0.53 (0.05)
0.28 (0.11)
0.45 (0.13)
0.61 (0.14)
0.60 (0.09)
0.81 (0.10)
0.29 (0.11)
0.47 (0.13)
0.63 (0.13)
0.62 (0.09)
0.82 (0.10)

Over-Winter Abundance and Apparent Survival
Winter appeared to be a particularly difficult time for small mammals in both the
upper and lower grasslands, with all sites showing considerable declines in small
mammal densities between the final 2007 trapping season and the first winter session of
2008 (Figure 2.7). The two upper grassland sites had the highest decreases, dropping
from densities of 56 and 38 animals per hectare in the 07-10 trapping session, to zero
animals in both the 08-01 and 08-02 sessions.

38

70

60

Small mammal density

50

40

30

20

10

0

07-08

07-09

07-10

08-01

08-02

08-03

Trap session (yy-mm)

Figure 2.7. Small mammal population densities (animals/ha) the lower grassland (●) and
lower grassland winter (○) sites, and the upper grassland ( ) and upper grassland winter
( ) sites across the final three trapping sessions of 2007, and the three lower grassland
and two upper grassland winter trapping sessions of 2008.

Over-winter apparent survival (October 2006 to May 2007, and November 2007 to
May 2008) for both P. maniculatus and Microtus spp. was exceedingly low, making
estimates of such unattainable; hence, only the actual numbers of recaptured individuals
are presented here. In total, 11 individuals out of 434 apparently survived through either
the winter of 2006/07 or the winter of 2007/08. Of those 11, seven apparently survived
through the 2006/07 and four through the 2007/08 winter season (Table 2.8). No
individuals were recaptured after both the 2006/07 and 2007/08 over-winter periods (i.e.
no one animal survived through both over-winter periods), and all individuals that were
recaptured following an over-winter period were P. maniculatus.

39

Table 2.8. Numbers of over-winter surviving individuals of P. maniculatus and Microtus
spp. for the 2006/07 and 2007/08 over-winter periods in the lower and upper elevation
grasslands.

Species

Overwinter
period

P. maniculatus 2006/07
2007/08
Microtus spp.

2006/07
2007/08

Elevation

Total number
of individuals
caught in the
last 2006 or
2007 trap
session

Total number
of individuals
that survived
the overwinter period

Percent of
individuals
trapped in the last
session that
survived the overwinter period

Lower
Upper
Lower
Upper
Lower
Upper
Lower
Upper

91 (2006)
39 (2006)
75 (2007)
57 (2007)
11 (2006)
137 (2006)
0 (2007)
24 (2007)

6
1
1
3
0
0
0

6.59
2.56
1.33
5.26
0.00
0.00
0.00

Correlation Analysis
The mean abundance of P. maniculatus and Microtus spp. across the 2007 and 2008
trapping sessions were correlated with site characteristics at each of the eight trapping
sites (Table 2.9). Peromyscus maniculatus densities were not significantly correlated
with any of the site characteristics measured, whereas Microtus spp. densities were
correlated with most site attributes. Correlations for Microtus spp. generally followed
habitat differences at the elevational level, showing strong correlations with herbaceous
litter and biomass levels, grass cover, litter cover and plant richness, and strong negative
correlations to sage litter levels, bare ground, coarse woody debris, crust, rock and shrub
cover, and vertical visibility. Mean P. maniculatus densities and Microtus spp. densities
were not significantly correlated (ρ= 0.30, P = 0.46). Also, P. maniculatus and Microtus
spp. densities were not correlated when analyzed on a per trap session basis across all
three years of study in the upper grasslands (ρ = 0.19, P = 0.15), nor were they for any
individual trapping grid on a per trap session basis, across all three years of study.

40

Table 2.9. Spearman‟s rank correlation matrix between site characteristics and mean
densities of P. maniculatus and Microtus spp. Within each cell, the uppermost number
equals Spearman‟s correlation ρ and the lower number equals the P-value. Dark grey
cells indicate a strong positive correlation (ρ > 0.50), and light grey cells indicate a strong
negative correlation (ρ < -0.50).
Site Characteristics
Herbaceous litter levels
Herbaceous biomass levels
Sage litter levels
Bare ground cover
Coarse woody debris cover
Crust cover
Grass cover
Litter cover
Rock cover
Shrub cover
Soil compaction: 1 cm depth
Soil compaction: 10 cm depth
Horizontal visibility
Vertical visibility
Plant richness
Plant Shannon-Weiner diversity index
Plant Simpson‟s diversity index

P. maniculatus
0.05
0.91
0.05
0.91
-0.10
0.81
-0.14
0.74
-0.13
0.77
0.38
0.35
-0.36
0.39
-0.24
0.57
-0.07
0.87
-0.22
0.60
-0.21
0.61
-0.05
0.91
-0.18
0.69
0.02
0.96
-0.29
0.49
-0.29
0.49
-0.48
0.23

Microtus spp.
0.84
0.01
0.86
0.01
-0.87
0.01
-0.86
0.01
-0.87
0.01
-0.71
0.05
0.71
0.05
0.76
0.03
-0.66
0.08
-0.86
0.01
0.34
0.401
0.01
0.98
0.44
0.27
-0.77
0.02
0.52
0.19
0.43
0.29
0.15
0.72

41

DISCUSSION
Total small mammal densities tended to be higher in upper grassland habitats as
opposed to lower grasslands, yet this pattern did not hold for all species trapped in this
study. While voles and shrews were caught almost exclusively in the upper grasslands,
deer mice were caught in both habitat types, and appear to be the dominant species of the
lower grasslands. This shift in species composition is likely due to the significant
differences in plant communities and the associated habitat characteristics found in each
grassland type.
Previous studies into small mammal community assemblages have documented
similar shifts in functional groups (e.g. herbivore, omnivore, granivore) based on habitat
characteristics, similar to that seen in this study. French et al. (1976) and Grant and
Birney (1979) were one of the firsts to describe rodent communities across a number of
North American grasslands, and demonstrated a shift in rodent assemblages based on
vegetation density and composition, with microtine herbivores occurring in areas of high
vegetative cover, omnivores at intermediate cover sites, and granivorous heteromyids at
low cover sites. Similar findings have been reported in later studies (e.g. Pearson et al.
2001, Reed et al. 2006b), and it appears rodent communities in this study followed
similar patterns.
Vegetative cover, and particularly litter, is an important habitat component for voles,
and may explain why these animals were all but absent from the lower grassland sites.
For rodent herbivores with relatively low digestion efficiency (French et al. 1976), who
can quickly cut and consume plant matter (Howe et al. 2002, 2006), increased vegetative
cover can equate to increased food availability. As semi-fossorial animals who often
tunnel through plant litter, increased litter levels may reduce the risk of predation,
particularly from diurnal predators, and may decrease the chances of aggressive
interspecific and intraspecific interactions (Warnock 1965). Decreased levels of
vegetative cover may negatively affect vole population dynamics, including recruitment
and survival (Peles and Barrett 1996), and it has been hypothesized that a threshold level
of cover is required for some vole populations to persist, increase, and in some instances,

42

undergo cyclic fluctuations (Birney et al. 1976). For their study, Birney et al. (1976)
suggested that cover levels between 400 and 600 g/m2 were needed for voles at their sites
to increase to the point of cycling, and on the lower grassland sites for this study,
vegetative cover levels were well below those values, averaging 94.81 g/m2. This lack of
herbaceous plant cover and the biotic and abiotic benefits it provides most likely exclude
voles from persisting in the lower grasslands.
Unlike voles, deer mice were found in both grassland habitats, often at higher densities
in the lower grasslands, but at times higher densities in the upper grasslands. Generally,
however, there were few significant differences in deer mouse densities between the two
grasslands types. This observation supports previous works that document the ubiquitous
nature of the species (e.g. French et al. 1976, Grant and Birney 1979, Pearson et al.
2001), given that their generalist lifestyle allows them to exploit a wide variety of
habitats. Using abundance as an indicator of habitat quality has been questioned however
(Van Horne 1983, Wheatley et al. 2002, Battin 2004), and looking at other population
metrics such as survival and reproduction may provide deeper insights into the quality of
the habitat in which wildlife persist (Schorr et al. 2007).
In this study, apparent survival rates of deer mice differed only slightly between the
upper and lower grasslands, with the treatment effect (i.e. elevation) excluded from the
top survival models in all years of study. The percent of reproductively active females
also did not differ significantly between elevations, and although the upper grasslands did
tend to have higher portions of females in their populations, on a year-by year basis these
differences were seldom significant. Such results indicate that both the upper and lower
grasslands may provide similar levels of habitat quality to deer mice, and yet conflict
somewhat with the density results which showed that at times the lower grasslands could
have significantly higher densities of deer mice. The answer to this disparity may lie in
differences in over-winter survival at the two habitats.
The over-winter period appears to be a difficult time for deer mice in grassland
ecosystems, as it can be for other rodent species in different habitats (e.g. Merritt and
Merritt 1978, Boonstra and Krebs 2006, Larsen et al. 2007). In general, rodents living in

43

areas with complete snow cover will persist in the subnivean space. This space protects
them from external winter conditions and predators, and allows for the exploration and
utilization of under-snow food resources. At the same time, the subnivean space may not
be contiguous, and it has been suggested that depending on the conditions under which it
is formed, snow cover can reduce the over-winter survival of rodents by encasing
vegetation in ice, and significantly reducing the levels of available food resources
(Korslund and Steen 2006). It has also been hypothesized that late winter/early spring
thaw can produce extreme sub-snow conditions by filling much of the subnivean space
with water, or by creating a layer of ice over vegetation, leading to the potential drowning
and/or freezing of its rodent inhabitants, and further restricted access to food resources
(Merritt and Merritt 1978, Aars and Ims 2002).
This study saw significant differences in snow depths between the upper and lower
grasslands during the two over-winter trapping sessions, with the upper grasslands having
continuous snow cover, and the lower grasslands remaining relatively snow-free during
the course of the winter. Although the lack of snow in the lower grasslands would offer
deer mice little in the way of thermal protection, it would allow them better winter
foraging opportunities and protection from any adverse conditions during spring thaw.
Further, although the two grassland ecosystems showed similar May through October
population demographics and apparent survival rates, winter conditions in the upper
grasslands may prove to be a tighter bottleneck for deer mice than in the lower
grasslands. Further study into the winter dynamics of rodents in these two ecosystems is
suggested.
Both 2007 and 2008 saw similar patterns of within-year deer mouse densities in both
the lower and upper grasslands, with densities dropping or remaining low for the first
three trapping sessions, and then significantly increasing into the fall season. Similar
patterns have been reported elsewhere (Fairbairn 1977a, Gilbert and Krebs 1991, Falls et
al. 2007), with the initial spring time reduction thought to be the result of increased
aggressiveness and spacing behaviour of breeding males, and an increased mortality of
early-breeding females (Fairbairn 1977a). The increase in densities later in the year may

44

be the result of increased breeding success, and the presence of more juvenile individuals
in the populations. This study‟s analysis of the percent of adult deer mice trapped per
trap session lend support to this supposition, as these percentages generally decreased
during fall trapping sessions while densities continued to increase.
Beyond the differences in densities at the habitat scale, vole populations in the upper
grasslands showed a significant decrease in densities across the three years of study.
Deer mice showed similar trends in the lower grasslands, although not to the same
severity.
Studies into the nature of vole population changes are extensive (e.g. Krebs et al.
1973, Korpimäki and Krebs 1996, Krebs 1996, Boonstra et al. 1998, Oli and Dobson
2001, Korpimäki et al. 2004) and microtine species have often been used as models to
examine the nature of animal population cycles. It is generally accepted that many vole
populations, particularly those at northern latitudes, can undergo multi-annual cycles,
generally 3-5 years, but such cycles have also been noted in grassland rodent
communities (Brady and Slade 2004), and multiannual fluctuations have been observed
in some populations of deer mice (Drost and Fellers 1991, Brady and Slade 2004, Bartell
et al. 2008). Hypotheses to explain these cycles are extensive, with Stenseth and Ims
(1993) dividing them into three broad categories: abiotic (e.g. weather), biotic extrinsic
(e.g. predation, competition, food availability), and biotic intrinsic (e.g. genetic and
behavioural), and the interactions between these categories. Unfortunately the current
study was not long enough to determine if the observed declines from 2006 vole densities
represented the onset of a low-phase of a population cycle (Boonstra et al. 1998), the
short-term decline of a generally higher population, or the return of the population to a
lower density steady-state after a previous population increase.
A cursory review of the historic annual rainfall levels in the area of the study showed
that 2004 and 2005 precipitation levels were above the previous 30 year (1974-2003)
average (281.4 mm) at 378.9 mm and 306.5 mm, respectively, while the 2006 and 2007
levels were below this historic average (268.5 mm and 230.0 mm, respectively). It is
well known that precipitation levels are highly correlated to net primary productivity in

45

arid and semi-arid systems (Sala et al. 1988, Lauenroth and Sala 1992), with small
mammal densities also increasing as precipitation increases, but lagging up to a season
behind (Brady and Slade 2004, Shenbrot et al. 2010, Thibault et al. 2010), and decreasing
over a relatively small reduction in mean precipitation (Reed et al. 2007). The higher
densities of voles in the upper grasslands and deer mice in the lower grasslands in 2006
may have been the result, in part, of higher than average precipitation levels in 2004 and
2005, with their declines in 2007 and 2008 a result of the less than average amounts of
precipitation observed in 2006 and 2007. Longer-term monitoring of small mammal
densities, precipitation and primary production would be needed to fully understand these
interactions.
Small mammals are important components of grassland ecosystems, and this study is
one of the first to provide insights into small mammal communities in the grasslands of
British Columbia, Canada. It has been shown that although upper grasslands tend to have
higher total small mammal densities, this relationship is not constant. Further, a shift of
small mammal functional groups appears to occur, with vole species confined to the
upper elevations, and deer mice found in both habitats and almost exclusively dominating
the lower grasslands: a separation believed to be based on the decreased levels of plant
litter at the lower grassland sites. Inspection of the apparent survival rates between deer
mice in the upper and lower grasslands, however, showed no tangible differences, and so
both grassland types may be of equivalent quality for this species, at least during nonwinter months. And finally, this study documented the general decrease of both vole and
deer mouse densities over the course of three years, a trend that may be the result, in part,
of previous years‟ levels of precipitation. This study provides information on small
mammal communities in British Columbia grasslands that was previously lacking, and
offers a foundation for further research into the structure and workings of these grassland
ecosystems.

46

LITERATURE CITED
AARS, J., AND R. A. IMS. 2002. Intrinsic and climate determinants of population
demography: the winter dynamics of tundra voles. Ecology 83: 3449-3456.
BAEZ, S., S. L. COLLINS, D. LIGHTFOOT, AND T. L. KOONTZ. 2006. Bottom-up regulation
of plant community structure in an aridland ecosystem. Ecology 87: 2746-2754.
BARTEL, R. A., F. F. KNOWLTON, AND C. STODDART. 2008. Long-term patterns in
mammalian abundance in northern portions of the Great Basin. Journal of Mammalogy
89: 1170-1183.
BATTIN, J. 2004. When good animals love bad habitats: ecological traps and the
conservation of animal populations. Conservation Biology 18: 1482-1491.
BEATLEY, J. C. 1969. Dependence of desert rodents on winter annuals and precipitation.
Ecology 50: 721-724.
BIRNEY, E. C., W. E. GRANT, AND D. D. BAIRD. 1976. Importance of vegetative cover to
cycles of Microtus populations. Ecology 57: 1043-1051.
BOONSTRA, R., AND C. J. KREBS. 2006. Population limitation of the northern red-backed
vole in the boreal forests of northern Canada. Journal of Animal Ecology 75: 1269-1284.
BOONSTRA, R., C. J. KREBS, AND N. C. STENSETH. 1998. Population cycles in small
mammals: the problem of explaining the low phase. Ecology 79: 1479-1488.
BRADY, M. J., AND N. A. SLADE. 2001. Diversity of a grassland rodent community at
varying temporal scales: the role of ecologically dominant species. Journal of
Mammalogy 82: 974-983.
BRADY, M. J., AND N. A. SLADE. 2004. Long-term dynamics of a grassland rodent
community. Journal of Mammalogy 85: 552-561.
BROWN, J. H., AND S. K. M. ERNEST. 2002. Rain and rodents: complex dynamics of desert
consumers. BioScience 52: 979-987.
BURNHAM, K. P., G. C. WHITE, AND D. R. ANDERSON. 1995. Model selection strategy in
the analysis of capture-recapture data. Biometrics 51: 888-898.

47

CARLYLE, C. N., L. H. FRASER, C. M. HADDOW, B. A. BINGS, AND W. HARROWER. 2010.
The use of digital photos to assess visual cover for wildlife in rangelands. Journal of
Environmental Management 91: 1366-1370.
CHURCHFIELD, S., J. HOLLIER, AND V. K. BROWN. 1991. The effects of small mammal
predators on grassland invertebrates, investigated by field exclosure experiment. Oikos
60: 283-290.
COHEN, J. 1988. Statistical power analyses for the behavioral sciences. 2nd ed. LawrenceErlbaum, Hillside, New Jersey.
COOCH, E., AND G. WHITE. 2008. Program MARK “A gentle introduction”. 7th ed.
http://www.phidot.org/software/mark/docs/book/.
DAVIDSON, A. D., AND D. C. LIGHTFOOT. 2008. Burrowing rodents increase landscape
heterogeneity in a desert grassland. Journal of Arid Environments 72: 1133-1145.
DAVIDSON, A. D., D. C. LIGHTFOOT, AND J. L. MCINTYRE. 2008. Engineering rodents
create key habitat for lizards. Journal of Arid Environments 72: 2142-2149.
DROST, C. A., AND G. M. FELLERS. 1991. Density cycles in an island population of deer
mice, Peromyscus maniculatus. Oikos 60: 351-364.
EDALGO, J. A., AND J. T. ANDERSON. 2007. Effects of prebaiting on small mammal
trapping success in a Morrow‟s honeysuckle-dominated area. Journal of Wildlife
Management 71: 246-250.
EDWARDS, G. R. AND M. J. CRAWLEY. 1999. Rodent seed predation and seedling
recruitment in mesic grasslands. Oecologica 118: 288-296.
ERNEST, S. K. M., J. H. BROWN, AND R. R. PARMENTER. 2000. Rodents, plants, and
precipitation: spatial and temporal dynamics of consumers and resources. Oikos 88: 470482.
FAIRBAIRN, D. J. 1977a. The spring decline in deer mice: death or dispersal? Canadian
Journal of Zoology 55: 84-92.
FAIRBAIRN, D. J. 1977b. Why breed early? A study of reproductive tactics in Peromyscus.
Canadian Journal of Zoology 55: 862-871.

48

FALLS, J. B., E. A. FALLS, AND J. M. FRYXELL. 2007. Fluctuation of deer mice in Ontario
in relation to seed crops. Ecological Monographs 77: 19-32.
FRENCH, N. R., W. E. GRANT, W. GRODZINSKI, AND D. M. SWIFT. 1976. Small mammal
energetics in grassland ecosystems. Ecological Monographs 46: 201-220.
GETZ, L. L., J. E. HOFMANN, L. SIMMS, AND B. MCGUIRE. 2006. Efficacy of monthly and
semi-monthly trapping of small mammal populations. American Midland Naturalist 155:
422-427.
GILBERT, S., AND C. J. KREBS. 1991. Population dynamics of Clethrionomys and
Peromyscus in southwestern Yukon 1973-1989. Holarctic Ecology 14: 250-259.
GRANT, W. E., AND E. C. BIRNEY. 1979. Small mammal community structure in North
American grasslands. Journal of Mammalogy 60: 23-36.
HAIRSTON, N. G., F. E. SMITH, AND L. B. SLOBODKIN. 1960. Community structure,
population control, and competition. American Naturalist 94: 421-425.
HERNANDEZ, L., A. G. ROMERO, J. W. LAUNDRE, D. LIGHTFOOT, E. ARAGON, AND J.
LOPEZ PORTILLO. 2005. Changes in rodent community structure in the Chihuahuan Desert
Mexico: comparisons between two habitats. Journal of Arid Environments 60: 239-257.
HOFFMAN, R. S. 1958. The role of reproduction and mortality in population fluctuations
of voles (Microtus). Ecological Monographs 28: 79-109.
HOWE, H. F., J. S. BROWN, AND B. ZORN-ARNOLD. 2002. A rodent plague on prairie
diversity. Ecology Letters 5: 30-36.
HOWE, H. F., AND D. LANE. 2004. Vole-driven succession in experimental wet-prairie
restorations. Ecological Applications 14: 1295-1305.
HOWE, H. F., B. ZORN-ARNOLD, A. SULLIVAN, AND J. S. BROWN. 2006. Massive and
distinctive effects of meadow voles on grassland vegetation. Ecology 87: 3007-3013.
HULME, P. E. 1994. Post-dispersal seed predation in grasslands: its magnitude and
sources of variation. Journal of Ecology 82: 645-652.
IVERSON, S. L., AND B. N. TURNER. 1969. Under-snow shelter for small mammal trapping.
Journal of Wildlife Management 33: 722-723.

49

KAUFMAN, D. W., AND G. A. KAUFMAN. 1990. Influence of plant litter on patch use by
foraging Peromyscus maniculatus and Reithrodontomys megalotis. American Midland
Naturalist 124: 195-198.
KEESING, F. 2000. Cryptic consumers and the ecology of an African savanna. BioScience
50: 205-215.
KELLER, B. L., AND C. J. KREBS. 1970. Microtus population biology; III. Reproductive
changes in fluctuating populations of M. ochrogaster and M. pennsylvanicus in Southern
Indiana, 1965-67. Ecological Monographs 40: 264-294.
KIMBALL, S. AND P. MATTIS. 2007. The GNU image manipulation program. Version
2.4.3. The GIMP Development Team, Berkeley, California.
KORPIMÄKI, E., P. R. BROWN, J. JACOB, AND R. P. PECH. 2004. The puzzles of population
cycles and outbreaks of small mammals solved. BioScience 54: 1071-1079.
KORPIMÄKI, E., AND C. J. KREBS. 1996. Predation and population cycles of small
mammals. BioScience 46: 754-764.
KORSLUND, L., AND H. STEEN. 2006. Small rodent winter survival: snow conditions limit
access to food resources. Journal of Animal Ecology 75: 156-166.
KREBS, C. J. 1996. Population cycles revisited. Journal of Mammalogy 77: 8-24.
KREBS, C. J. 1999. Ecological methodology. 2nd ed. Welsey Educational Publishers, Inc.,
Menlo Park, California.
KREBS, C. J., K. COWCILL, R. BOONSTRA, AND A. J. KENNEY. 2010. Do changes in berry
crops drive population fluctuations in small rodents in the southwestern Yukon? Journal
of Mammalogy 91: 500-509.
KREBS, C. J., M. S. GAINES, B. L. KELLER, J. H. MEYERS, AND R. H. TAMARIN. 1973.
Population cycles in small rodents. Science 179: 35-41.
LA TOURRETTE, J. E., J. A. YOUNG, AND R. A. EVANS. 1971. Seed dispersal in relation to
rodent activities in seral big sagebrush communities. Journal of Range Management 24:
118-120.

50

LARSEN, K. W., ADAMS, I. T., AND D. L. HAUGHLAND. 2007. Small mammal communities
in a pyrogenic habitat mosaic. International Journal of Wildland Fire 16: 728-740.
LAUENROTH, W. K., AND O. E. SALA. 1992. Long-term forage production of North
American shortgrass steppe. Ecological Applications 2: 397-403.
LEBRETON, J-D., K. P. BURNHAM, J. CLOBERT, AND D. R. ANDERSON. 1992. Modeling
survival and testing biological hypotheses using marked animals: a unified approach with
case studies. Ecological Monographs 62: 67-118.
LINDROTH, R. L., AND G. O. BATZLI. 1984. Food habits of the meadow vole (Microtus
pennsylvanicus) in bluegrass and prairie habitats. Journal of Mammalogy 65: 600-606.
LITTELL, R. C., W. W. STROUP, AND R. J. FREUND. 2002. SAS for linear models. 4th ed.
SAS Institute, Inc., Cary, North Carolina.
MASER, C., Z. MASER, AND R. MOLINA. 1988. Small-mammal mycophagy in rangelands
in central and southeastern Oregon. Journal of Range Management 41: 309-312.
MCADOO, J. K., C. C. EVANS, B. A. ROUNDY, J.A. YOUNG, AND R. A. EVANS. 1983.
Influence of heteromyid rodents on Oryzopsis hymenoides germination. Journal of Range
Management 36: 61-64.
MCCRAVY, K. W., AND R. K. ROSE. 1992. An analysis of external features as predictors
of reproductive status in small mammals. Journal of Mammalogy 73: 151-159.
MERRITT, J. F., AND J. M. MERRITT. 1978. Population ecology and energy relationships of
Clethrionomys gapperi in a Colorado subalpine forest. Journal of Mammalogy 59: 576598.
MESERVE, P. L., W. B. MILSTEAD, J. R. GUTIERREZ, AND F. M. JAKSIC. 1999. The
interplay of biotic and abiotic factors in a semiarid Chilean mammal assemblage: results
of a long-term experiment. Oikos 85: 364-372.
MORRIS, D. W. 1984. Patterns and scale of habitat use in two temperate-zone, small
mammal faunas. Canadian Journal of Zoology 62:1540-1547.
MORRIS, D. W. 1987. Ecological scale and habitat use. Ecology 68:362-369.

51

MOSES R. A., AND S. BOUTIN. 2001. The influence of clear-cut logging and residual leave
material on small mammal populations in aspen-dominated boreal mixedwoods.
Canadian Journal of Forest Research 31: 483-495.
NEGUS, N. C., BERGER, P. J., AND L. G. FORSLUND. 1977. Reproductive strategy of
Microtus montanus. Journal of Mammalogy 58: 347-353.
NEGUS, N. C., AND A. J. PINTER. 1966. Reproductive responses of Microtus montanus to
plants and plant extracts in the diet. Journal of Mammalogy 47: 596-601.
OLI, M. K., AND F. S. DOBSON. 2001. Population cycles in small mammals: the alphahypothesis. Journal of Mammalogy 82: 573-581.
OSTFELD, R. S., AND F. KEESING. 2000. Pulsed resources and community dynamics of
consumers in terrestrial ecosystems. Trends in Ecology & Evolution 15: 232-237.
OTIS D., K. P. BURNHAM, G. C. WHITE, AND D. R. ANDERSON. 1978. Statistical inference
from capture data on closed animal populations. Wildlife Monographs 62: 3-135.
PARMENTER, R. R., T. L. YATES, D. R. ANDERSON, K. P. BURNHAM, J. L. DUNNUM, A. B.
FRANKLIN, M. T. FRIGGENS, B. C. LUBOW, M. MILLER, G. S. OLSON, C. A. PARMENTER, J.
POLLARD, E. REXSTAD, T. M. SHENK, T. R. STANLEY, AND G. C. WHITE. 2003. Smallmammal density estimation: A field comparison of grid-based vs. web-based density
estimators. Ecological Monographs 73:1-26.
PEARSON, D. E., Y. K. ORTEGA, K. S. MCKELVEY, AND L. F. RUGGIERO. 2001. Small
mammal communities and habitat selection in northern Rocky Mountain bunchgrass:
implications for exotic plant invasions. Northwest Science 75: 107-117.
PEET, R. K. 1974. The measurement of species diversity. Annual Review of Ecology and
Systematics 5: 285-307.
PELES, J. D., AND G. W. BARRETT. 1996. Effects of vegetative cover on the population
dynamics of meadow voles. Journal of Mammalogy 77: 857-869.
PIELOU, E. C. 1966. The measurement of diversity in different types of biological
collections. Journal of Theoretical Biology 13: 131-144.
PRUITT, W. O., JR. 1959. A method for live-trapping small taiga mammals in winter.
Journal of Mammalogy 40: 139-143.

52

REDFIELD, J. A., C. J. KREBS, AND M. J. TAITT. 1977. Competition between Peromyscus
maniculatus and Microtus townsendii in grasslands of coastal British Columbia. Journal
of Animal Ecology 46: 607-616.
REED, A. W., G. A. KAUFMAN, AND D. W. KAUFMAN. 2006a. Effect of plant litter on seed
predation in three prairie types. American Midland Naturalist 155: 278-285.
REED, A. W., G. A. KAUFMAN, AND D. W. KAUFMAN. 2006b. Species richnessproductivity relationship for small mammals along a desert –grassland continuum:
differential responses of functional groups. Journal of Mammalogy 87: 777-783.
REED, A. W., G. A. KAUFMAN, AND B. K. SANDERCOCK. 2007. Demographic response of a
grassland rodent to environmental variability. Journal of Mammalogy 88: 982-988.
REICH, L. M. 1981. Microtus pennsylvanicus. Mammalian Species 159: 1-8.
SALA, O. E., W. J. PARTON, L. A. JOYCE, AND W. K. LAUENROTH. 1988. Primary
production of the central grasslands region of the United States. Ecology 69: 40-45.
SAS INSTITUTE INC. 2008. SAS/STAT software: statistical analysis software version 9.2.
SAS Institute Inc., Cary, North Carolina.
SCHLOTZHAUER, S., AND R. C. LITTELL. 1997. SAS system for elementary statistical
analysis. 2nd ed. SAS Institute, Inc., Cary, North Carolina.
SCHORR, R. A., J. L. SIEMERS, P. M. LUKACS, J. P. GIONFRIDDO, J. R. SOVELL, R. J.
RONDEAU, AND M. B. WUNDER. 2007. Using survival of rodents to assess quality of
prairie habitats. Southwestern Naturalist 52: 552-563.
SERA, W. E., AND C. N. EARLY. 2003. Microtus montanus. Mammalian Species 716: 1-10.
SHENBROT, G., B. KRASNOV, AND S. BURDELOV. 2010. Long-term study of population
dynamics and habitat selection of rodents in the Negev Desert. Journal of Mammalogy
91: 776-786.
SIMPSON, E. H. 1949. Measurement of diversity. Nature 163: 688-688.
STENSETH, N. C., AND R. A. IMS. 1993. Population dynamics of lemmings: temporal and
spatial variation – an introduction. Pp. 61-96 in The biology of lemmings (N. C. Stenseth
and R. A. Ims, eds.). The Linnean Society of London, London, United Kingdom.

53

SULLIVAN, T. P., AND D. S. SULLIVAN. 2004. Influence of granivorous diversionary food
on population dynamics of montane vole (Microtus montanus), deer mice (Peromyscus
maniculatus), and western harvest mice (Reithrodontomys megalotis). Crop Protection
23: 191-200.
THIBAULT, K. M., S. K. M. ERNEST, E. P. WHITE, J. H. BROWN, AND J. R.GOHEEN. 2010.
Long-term insights into the influence of precipitation on community dynamics in desert
rodents. Journal ofMammalogy 91: 787-797.
TISDALE, E. W. 1947. The grasslands of the southern interior of British Columbia.
Ecology 28: 346-382.
VALONE, T. J., AND J. H. BROWN. 1995. Effects of competition, colonization, and
extinction on rodent species diversity. Science 267:880-883.
VAN HORNE, B. 1983. Density as a misleading indicator of habitat quality. Journal of
Wildlife Management 47: 893-901.
VAN RYSWYK, A. L., A. MCLEAN, AND L. S. MARCHAND. 1966. The climate, native

vegetation, and soils of some grasslands at different elevations in British Columbia.
Canadian Journal of Plant Science 46: 35-50.
WARNOCK, J. E. 1965. The effects of crowding on the survival of meadow voles
(Microtus pennsylvanicus) deprived of cover and water. Ecology 46: 649-664.
WHEATLEY, M., K. W. LARSEN, AND S. BOUTIN. 2002. Does density reflect habitat quality
for North American red squirrels during a spruce-cone failure? Journal of Mammalogy
83: 716-727.
WHITE, G. C. 2008. Closed population estimation models and their extension in Program
MARK. Environmental and Ecological Statistics 15: 89-99.
WHITE, G. C., AND K. P. BURNHAM. 1999. Program MARK: survival estimation from
populations of marked animals. Bird Study 46(Suppl.): 120-138.
WIKEEM, B., AND S. WIKEEM. 2004. The grasslands of British Columbia. Grasslands
Conservation Council of British Columbia, Kamloops, British Columbia, Canada.
ZAR, J. H. 1999. Biostatistical analysis. 4th ed. Prentice Hall, Upper Saddle River, New
Jersey.

54

CHAPTER 3 – SELECTION OF NESTS AND DAYTIME REFUGE SITES BY
DEER MICE IN A SEMI-ARID GRASSLAND IN BRITISH COLUMBIA,
CANADA
INTRODUCTION
The manner in which certain organisms choose to occupy a particular habitat can be
viewed as a hierarchical process in which individuals make habitat selection choices at
varying spatial scales. Johnson (1980) described this hierarchy of choices as ordered
selections. For example, members of a species may occupy a range of different
macrohabitats (e.g. forest vs. grassland): a first order selection. Within each
macrohabitat, individuals may occupy a home range (second order selection), and within
each home range select areas for specific activities such as feeding or nesting (third order
selection). The location where the specific activity occurs (e.g. nest site location) may be
considered a fourth order selection (Johnson 1980). Small mammals have been shown to
respond to habitat characteristics at both the macrohabitat and microhabitat levels (Morris
1984, 1987, Stapp 1997), and these responses can be influenced by a number of
interacting factors, including food and shelter availability, levels of predation, and
conspecific and inter-specific interactions (Redfield et al. 1977, Tait 1981, Dooley and
Dueser 1996). Two such habitat attributes that may affect small mammal habitat
selection and that have been considered potential limiting resources for small mammals
are nests and daytime refuge sites (Dooley and Dueser 1990, Bright and Morris 1991).
Characteristics of nests and daytime refuge site locations (hereafter referred to
collectively as refuge sites) have been described for a number of small mammal species.
For example, cotton mice (Peromyscus gossypinus) and golden mice (Ochrotomys
nuttali) in south-central Florida often are associated with burrows of the gopher tortoise
(Gopherus polyphemus; Frank and Layne 1992), and for cotton mice, coarse woody
debris (CWD) in some forest systems (Hinkelman and Loeb 2007). Wolff and Hurlbutt
(1982) and Wolff and Durr (1986) detailed a strong affinity of deer mice (P. maniculatus)
for arboreal nest sites in Virginia mixed-deciduous forests. Refuge sites may provide a
decreased risk of predation (Klein and Layne 1978), can facilitate the coexistence of

55

similar species (Wolff and Hurlbutt 1982, Barry et al. 1984, Harney and Dueser 1987),
and may provide protection from the elements through a moderated living environment
(Frank and Layne 1992); such resources may be particularly important for rodents living
in xeric or semi-arid habitats.
In British Columbia, Canada, the deer mouse is a ubiquitous rodent found throughout
all of the province‟s biogeoclimatic zones (Nagorsen 2005). In some of the province‟s
low elevation grasslands, the deer mouse plays a significant role in the functioning of the
ecosystem, including prey for a number of grassland predators like western rattlesnakes
(Crotalus oreganus), gopher snakes (Pituophis catenifer), coyotes (Canis latrans) and
raptor species. These low-elevation grassland sites are characterized by the lack of trees,
but are dominated by big sagebrush (Artemisia tridentata) and large, bare inter-shrub
areas with widely spaced bunchgrasses (van Ryswyk et al. 1966, Wikeem and Wikeem
2004). Summer daytime air temperatures can exceed 40°C, and effective refuge sites
may be critical to deer mice for managing thermal stress, maximizing reproductive
success, and ultimately, increasing fitness. Thus, given the relatively short season of
productivity (due to latitude) and the relatively hot summer temperatures, this habitat
provides an interesting backdrop for investigating potential limiting factors on deer mice,
such as the selection and availability of refuge sites.
Deer mice have been shown to orient their movements towards shrubs, and to display
a preference towards shrub microhabitats in grassland environments (Stapp and Van
Horne 1997). Although no formal studies have examined refuge sites for deer mice in
British Columbia‟s semi-arid grasslands, which lack trees and large quantities of CWD,
they have been observed burrowing under Artemisia (Nagorsen 2005). Understanding
where deer mice build refuge sites and how they choose refuge locations will help us
understand the relationship of this dominant grassland rodent to its dry semi-arid
environment, and will help assess the impacts of land management decisions,
environmental change and habitat alterations.
The objective of this study was to model deer mouse refuge site selection within a
low-elevation grassland macrohabitat at two scales corresponding to Johnson‟s (1980)

56

third and fourth order selection scheme. To do this I used radio telemetry and examined
a suite of a priori above-ground habitat variables. Due to the previously reported affinity
of deer mice to shrubs, it was hypothesized that shrub cover would be a significant
predictor of refuge sites at both scales of habitat selection.
METHODS
Study Area
Data for this study were collected from a shrub-steppe grassland near Kamloops,
British Columbia, Canada (50°43‟ N; 120°25‟ W). The area is consistent with the “lower
grasslands” described by Tisdale (1947) and van Ryswyk et al. (1966) and consists of a
large expanse of Artemisia-dominated grassland, interspersed with bare inter-shrub areas,
and grass species such as bluebunch wheatgrass (Pseudoroegneria spicata), needle-andthread grass (Hesperostipa comata) and crested wheatgrass (Agropyron cristatum). The
area receives infrequent grazing by foraging livestock.
Capture and Radio Telemetry
Deer mice were live-trapped at two sites between June 9, 2008 and September 3, 2008.
At each site, deer mice were sampled with a 10 × 5 trapping grid (50 traps total) with
14.3 m between trap stations. A single Longworth-style live trap (Little Critter Traps,
Rogers Manufacturing, Kelowna, BC) was placed within a 2 m radius of each trap
station, and covered with a 15 × 30 cm board to protect it from exposure to sun and rain.
Synthetic cotton bedding was added to the nesting chamber of each trap, and traps were
baited with approximately 5 g of a mixture of oats and sunflower seeds with a piece of
apple or carrot included as a moisture source. Traps were set within 2 hours of sunset
and checked within 2 hours after sunrise the following morning. Captured deer mice
were sexed and weighed using 60 g spring scales (PESOLA AG, Baar, Switzerland). As
this study was part of a larger project examining small mammal communities in grassland
ecosystems (see Chapter 2), most captured animals had been previously marked with a

57

uniquely numbered ear tag (Monel #1005-1, National Band and Tag, Newport, KA), and
newly captured animals were tagged thusly.
Deer mice weighing > 20 g were considered candidates for radio telemetry, in order to
keep the weight of the radio transmitter at < 5 % of the body weight of collared
individuals, and to avoid collaring transient animals attempts were made to only collar
mice that had been previously tagged as part of the community study referenced above.
The reproductive status of each individual was assessed using scoring techniques similar
to McCravy and Rose (1992) and Moses and Boutin (2001). For males, testes position
(abdominal or scrotal) was noted, and for females, teats were scored as either inactive
(small and difficult to see), enlarged (teats large; pregnant), lactating (teats large and fur
worn), or returning to normal (teats healing with fur re-growth). Mice were each fitted
with a BD-2NC transmitter (Holohil Systems Ltd., Carp, ONT), observed for a minimum
of 15 minutes to ensure the animal‟s welfare and to confirm the transmitter‟s operability,
and released at the point of capture.
For the purposes of this study, “daytime” refers to the time between 2 hours after
sunrise to 2 hours prior to sunset. At the onset of the study, deer mice were located twice
daily at irregular intervals until it was determined they were not changing locations, after
which they were located once a day. Telemetry locations were marked using a 53 cm
metal surveyor stake flag and their Universal Transverse Mercator (UTM) coordinates
determined using a handheld global positioning system unit. Individual deer mice were
tracked for 7 to 9 consecutive days, at the end of which they were recaptured and their
collars removed, and then released.
Habitat Measurements
Habitat measurements were taken after the tracking session for each particular mouse
had been completed, so as not to interfere with the animal‟s daily movement patterns. At
each telemetry location, the habitat metrics listed in Table 3.1 were estimated by
centering a 1 × 1 m frame over the location and visually estimating the percent cover of
each metric. To compare the habitat elements at each telemetry location with the

58

Table 3.1. Habitat metrics measured at telemetry locations, third order and fourth order
habitat plots using a 1 × 1 m frame.
Habitat Metric Description
Herb
Litter

Rock
Ground
Shrub
CWD
StemTotal
SageD
SageH
SageL
RbD
RbH
RbL
Slope

Percent cover of live grasses, forbs, and cacti.
Percent cover of dead grasses, forbs and cacti, as well as downed
shrub material, including leaves and woody material < 2.5 cm in
diameter.
Percent cover of rocks and gravel, but excluding sand and silt.
Percent cover of sand, silt and lichen.
Percent cover of live shrubs.
Percent cover of rooted and unrooted coarse woody debris > 2.5 cm in
diameter.
Number of shrubs rooted within the 1 × 1 m frame.
A. tridentata trunk diameter measured at 10 cm above the ground and
recorded to the nearest 0.1 cm.
A. tridentata height measured vertically from the ground to the
uppermost crown and recorded to the nearest 0.5 cm.
A. tridentata length measured from the ground along dominant
branch to the uppermost crown and recorded to the nearest 0.5 cm.
Chrysothamnus nauseosus trunk diameter measured at 10 cm above
the ground and recorded to the nearest 0.1 cm.
C. nauseosus height measured vertically from the ground to the
uppermost crown and recorded to the nearest 0.5 cm.
C. nauseosus length measured from the ground along dominant
branch to the uppermost crown and recorded to the nearest 0.5 cm.
Measured at telemetry location and at centre-point of the third order
plots. Recorded in percent slope.

immediate surrounding area (i.e. fourth order selection), the habitat metrics in Table 3.1
were again assessed by placing the 1 × 1 m frame at a distance of 5 m at bearings of 0°,
120° and 240° from the telemetry location. To collect third order selection data, a
random distance between 20 and 50 m from the telemetry location, and at a random
compass bearing, was selected and marked. From that position, the 1 × 1 m frame was
placed at a distance of 5 m at bearings of 0°, 120° and 240°, and the habitat metrics in
Table 3.1 were again collected. This process was then repeated by going back to the

59

telemetry location and measuring out a position the same distance from the telemetry
location as the first set of third order habitat metrics, but offset by 180°. Hence, for each
telemetry location, three sets of fourth order measurements and six sets of third order
measurements were taken (Figure 3.1).

5 m 0°
20-50 m

20-50 m
240°

120°

Telemetry location
Fourth order plot
Third order plot
Centre-point location of slope measurements for third order analysis
Figure 3.1. Layout of methodology for collecting habitat metrics at the third order and
fourth order levels for each telemetry location.

To investigate the levels of thermal protection afforded by refuge sites, a temperature
data logger (DS1921G Thermochron iButton, Maxim Integrated Products, Sunnyvale,
CA) was placed within each identified refuge location. A second data logger was placed
within 5 m of the telemetry location, and attached to the north side of an Artemisia trunk
approximately 30 cm above the ground in order to measure external ambient daytime
temperatures. The loggers recorded temperatures every 15 minutes, and collected data
for 5 days at each location.
Data Analysis
For descriptive purposes, refuge sites were classified as: 1) underground; 2) above
ground (sites where the mouse was visible on the ground‟s surface); 3) arboreal (sites in
shrub canopy); 4) inside shrub trunk; and 5) anthropogenic (sites in or on man-made

60

structures). Sites were further classified according to the dominant landscape feature at
the telemetry location, and included shrub, grass, rock, bare ground and anthropogenic.
To determine the minimum daily movements (i.e. the straight-line distance between
consecutive telemetry locations) of deer mice, telemetry UTM coordinates were mapped
and the distances determined. Telemetry locations separated by more than 36 hours were
excluded from this portion of the analysis.
Conditional (i.e. case-control) logistic regression was used to analyze the selection of
habitats at both the third and fourth order scales. Conditional logistic regression allows
the pairing of used sites (i.e. cases) with multiple available sites (i.e. controls), which
reduces autocorrelation issues often associated with spatial and temporal data
(Whittington et al. 2005), and allows for a better sampling of the habitat available to the
animal being tracked. Fourth order selection was analyzed using a 1:3 case:control
design (i.e. predictor variables at each telemetry location compared to variable
measurements taken from each of the three associated fourth order plots), and third order
selection was analyzed using a 1:2 design, with the average of each predictor variable
from each set of fourth order plots compared to predictor variable averages of each of the
two sets of associated third order habitat plots, for every telemetry location.
Predictor variables were selected for inclusion in the model after first examining the
collinearity between variables using Spearman‟s rank correlation analysis (Zar 1999).
When variable pairs had correlations > 0.7, the variable with the lowest Akaike‟s
Information Criterion adjusted for small sample sizes (AICc) using a single variable
model was selected for model inclusion (Burnham and Anderson 2002). Further, the
Wald statistic was calculated for each predictor variable using a single variable model,
and variables with a P-value > 0.10 were considered uninformative, and excluded from
further model development (Harrower 2007).
The goodness-of-fit of the global model (i.e. the model with all variables included)
were assessed by plotting the Δχ2 statistic against the model‟s fitted values, and
examining data for matched case-control sets that disproportionately affected the model‟s
fit (Hosmer and Lemeshow 2000). Sets deemed to be outliers were removed from the

61

dataset. Models were then developed by examining all combinations of the predictor
variables. Model development was constrained by limiting the number of variables in the
model to approximately 10% of the number of observations (Peduzzi et al. 1996), in this
case four variables, and model selection was based on the model with the lowest AICc
value.
As the hallmark of any good habitat selection model lies in its ability to accurately
assess a species‟ habitat choices, the predictive success of the top models were
determined using a form of k-fold cross validation, adapted from Boyce et al. (2002). Kfold cross validation generally involves splitting a dataset into three or greater subsets,
and using a portion of the subsets to train the model, and the remaining subsets to test the
model (Fielding and Bell 1997). For this study, a k-fold partition of three groups was
used. Cases and their matched controls were systematically assigned to one of the three
groups, and two of the three groups were used to train the model. This was repeated
three times so that each partition was excluded from the training set once. For each
iteration, data from the training set were used to determine the β-coefficients for each of
the predictor variables in the top model, using conditional logistic regression. The βcoefficients for each variable were then averaged across all three training-set runs, and
for each control (i.e. available) location, the resource selection function [RSF; w(x)] was
determined using the averaged β-coefficients for each variable with the equation:
w(x) = exp(β1x1 +…+ βpxp)
These results were then divided into eight categories, or bins, with each bin containing
approximately the same number of RSF scores (i.e. equal area bins; Boyce et al. 2002,
Wiens et al. 2008). Resource selection function scores for the case (i.e. used) locations
were calculated in the same manner, using the averaged β-coefficients derived from the
training datasets and the equation above, and assigned to the appropriate bin. A
Spearman‟s rank correlation analysis was then performed on the number of cases and the
mid-point RSF score of each bin. A model with good predictive ability would have a
strong positive correlation, as more case sites should occur in bins with higher RSF
scores (Boyce et al. 2002).

62

Temperature data were analyzed in two ways: 1) the mean daily high external ambient
temperature was compared with the corresponding mean daily temperature within the
refuge site; and 2) the mean daily high external ambient temperature was compared with
the mean daily high temperature within the refuge site. All comparisons were made
using paired t-tests.
All data were analyzed using SAS 9.2 software (SAS Institute Inc. 2008). The
goodness-of-fit for each model was examined using the SAS macro MCSTRAT, and all
logistic regression models were built and analyzed using the PROC LOGISTIC
procedure. The normality of the temperature data was confirmed using PROC
UNIVARIATE, and the paired t-tests were performed using PROC TTEST.
RESULTS
Over the course of the study, 13 individual deer mice were collared (♂ = 8, ♀ = 5),
with one male deer mouse collared and tracked twice, approximately 3 months apart. All
collared males were considered reproductively-active based on testes position, and all
females were considered pregnant or lactating, based on nipple morphology. In total, 42
unique daytime refuge sites were located, with 71.4% located underground, 19.1% above
ground, 4.8% inside the bole of an Artemisia, 4.7% in an anthropogenic structure, and 0%
in arboreal sites. The majority of the telemetry locations had shrubs as the dominant
above-ground characteristic (76.2%), followed by grasses (16.7%), rock and bare ground
(9.5% each), and finally anthropogenic structures (4.8%). These numbers exceed 100%
as several sites were included in two or more categories (e.g. shrub and grass). The
anthropogenic sites included one within the metal tube of a cattle guard crossing, and
another under a rock and asphalt pile in a gravel quarry. These sites were removed from
any further analyses. Mean male minimum daily movement was 122 m (n = 36), with a
maximum daily movement of 359 m, and mean female daily movement was 69 m (n =
18), with a maximum daily movement of 262 m.

63

Fourth Order Selection Modelling
Of the 13 variables originally measured for inclusion in the candidate models, the
variables Herb and Litter, SageD and SageH and SageL, as well as RbD and RbH and
RbL were highly correlated with each other. Of these, Herb, SageD and RbD explained
relatively more of the variation in the data (i.e. had the lowest AICc scores), and so were
retained for model development, while the others were omitted. As well, the variables
StemTotal and RbD both had Wald statistic P-values > 0.10 and so were removed from
further model development.
Using the six remaining habitat variables, 56 models were developed to examine the
fourth order selection of deer mice, with the top ten models based on AICc value
provided in Table 3.2. An examination of the β-coefficient estimates of the parameters in
the top model (SageD + Ground + CWD) indicate that the deer mice in this study were
selecting refuge sites at areas with relatively larger-diameter Artemisia, decreased levels
of bare ground and increased amounts of CWD than was typically available in the
immediately surrounding microhabitats (Table 3.3). Subsequent evaluation indicated that
the top model did a good job of predicting deer mouse fourth order selection. The
correlation coefficient between the midpoint RSF score for each bin and the observed
frequency was 0.82, with P = 0.01.

64

Table 3.2. Top 10 models used to examine refuge site selection by deer mice at the
fourth order scale, and detailing the number of model variables (K), Akaike‟s Information
Criterion adjusted for small sample sizes (AICc), the change in AICc from most
parsimonious model (ΔAICc), and the model weight (w).
Model

K

AICc

ΔAICc

w

SageD + Ground + CWD

4

68.69

-

0.36

SageD + Ground + CWD + Rock

5

70.30

1.62

0.16

SageD + Ground + CWD + Herb

5

71.18

2.49

0.10

SageD + Ground + CWD + Shrub

5

71.36

2.67

0.10

SageD + Ground

3

71.62

2.94

0.08

SageD + Ground + Rock

4

72.89

4.21

0.04

SageD + Ground + Rock + Shrub

5

73.53

4.84

0.03

SageD + Ground + Herb

4

73.54

4.86

0.03

SageD + Ground + Shrub

4

73.59

4.90

0.03

SageD + Ground + Rock + Herb

5

75.27

6.59

0.01

Table 3.3. β-coefficients and average measurements of the habitat metrics included in the
top model of the fourth order habitat analysis. Reported average measurements are taken
from the telemetry locations and fourth order plots, with SageD recorded in centimetres,
and Ground and CWD in percent cover of the 1 × 1 m frame.
Habitat Metric

β

Telemetry
Location

Fourth Order
Plots

SageD

0.31

6.97

4.25

Ground

-0.04

37.62

54.74

CWD

0.05

8.62

4.02

65

Third Order Selection Modelling
Fourteen predictor variables were originally measured for third order model building.
Of these, all but two variables (Litter and Slope) had Wald statistic P-values > 0.10, and
so were excluded from further model development. As a result, only three models were
built to examine third order habitat selection of deer mice (Table 3.4). An examination of
the β-coefficient estimates of the parameters in the top model (Slope + Litter) indicate
that the deer mice in this study were selecting areas with increased slope and decreased
litter (Table 3.5). Model evaluation indicated that the top model did a reasonable job of
predicting deer mouse third order habitat selection. The correlation coefficient between
the midpoint RSF score for each bin and the observed frequency was 0.73, with P = 0.04.

Table 3.4. Top three models used to examine refuge site selection by deer mice at the
third order scale, and detailing the number of model variables (K), Akaike‟s Information
Criterion adjusted for small sample sizes (AICc), the change in AICc from most
parsimonious model (ΔAICc), and the model weight (w).
Model

K

AICc

ΔAICc

w

Slope + Litter

3

57.69

-

0.95

Slope

2

63.38

5.69

0.06

Litter

2

83.24

25.55

0.00

Table 3.5. β-coefficients and average measurements of the habitat metrics included in the
top model of the third order habitat analysis. Reported average measurements are taken
from the third order and fourth order plots, with Slope recorded in degrees, and Litter in
percent cover of the 1 × 1 m frame.
Habitat Metric

β

Third Order
Plots

Fourth Order
Plots

Slope

0.11

24.29

37.10

Litter

-0.05

47.78

40.89

66

Temperature Analysis
In total, 35 refuge sites were included in the temperature analysis. The mean 5-day
average daily high ambient temperature was 32.7°C (s.e. + 0.9°C) with the corresponding
mean 5-day average daily high temperature in refuge sites being 28.0° C (s.e. + 0.9°C).
The mean 5-day average temperature of refuge sites when the ambient temperatures were
at their hottest was 26.8°C (s.e. + 0.9°C). The ambient daily high temperatures were
significantly greater than the daily high temperatures in the refuge sites (t = 5.91, d.f. =
34, P = < 0.01) and the temperatures in the refuge sites when the ambient temperature
was at its peak (t = 7.32, d.f. = 34, P = < 0.01).
DISCUSSION
Nest and daytime refuge site selection by deer mice in the current study appeared to be
strongly influenced at the fourth order level by the presence of Artemisia, and in
particular, relatively large-sized Artemisia, as originally hypothesized. That a majority of
refuge sites were found underground at or near the base of an Artemisia is consistent with
observations by Nagorsen (2005).
Previous studies have shown the importance of shrubs to deer mice. For example,
Stapp and Van Horne (1997) revealed that deer mice in shortgrass prairies oriented their
movements towards shrubs, and selected shrub microhabitats in areas where shrubs were
rare. A standing theory is that shrubs provide refuge sites from predators (Kotler 1984,
Kotler and Brown 1988), and while deer mice may be choosing subterranean daytime rest
sites beneath larger shrubs to help evade detection by avian and mammalian predators,
such sites may not provide increased security from snake predation (Pierce et al. 1992).
Indeed, we often were able to see the resting mouse in the refuge site, suggesting the
animals would be vulnerable to snake and other ground-based predators. A more
probable reason why the deer mice in this study were selecting microhabitats with
relatively large Artemisia could be the favourable environmental conditions these shrubs
provide.

67

Rodents living in desert and semi-arid climates are challenged with thermoregulation
under exceptionally hot and dry conditions, where minimizing the loss of body water is
of paramount importance (Walsberg 2000). Sagebrush are known to establish resource
islands in their sub-canopies (the area below the canopy) by creating moderate
microclimates when compared to inter-shrub areas (Parmenter and MacMahon 1983,
Davies et al. 2009). Locating refuge sites below the canopy of large Artemisia may
facilitate deer mice in thermoregulation as these sites have moderated maximum daily
soil temperatures when compared to inter-shrub spaces (Pierson and Wight 1991, Davies
et al. 2007). The current study appears to lend support to these previous findings, with
average daily high temperatures at refuge sites being significantly lower than average
daily high ambient temperatures. As sagebrush sub-canopies also tend to have higher soil
moisture levels (Chambers 2001, Davies et al. 2007), selecting them for refuge sites may
promote decreased rates of body water loss. Finally, as sub-canopy microhabitats can
promote herbaceous plant growth (Chambers 2001, Davies et al. 2007) and offer retreat
sites for desert insects (Parmenter et al. 1989), deer mice may orient their movements to
sub-canopy areas, selecting daytime refuge locations at those areas due to the proximity
to potential food sources (Parmenter and MacMahon 1983). These final benefits may be
tempered somewhat by the allelopathic nature of Artemisia plants (Weaver and Klarich
1977, Groves and Anderson 1981).
Deer mice in this study selected sites with decreased amounts of bare ground than was
typically available on the landscape: a result that was not unexpected. Bare ground may
be viewed as an antonym to the microhabitat conditions provided by shrubs and their subcanopies, with increased bare ground offering reduced amounts of protection from
predators, increased exposure to extreme environmental conditions, and increased
distances to potential food resources (see previous citations). An increased amount of
bare ground is analogous to reduced levels of microhabitat heterogeneity, an
environmental component that may be important in rodent habitat selection. Increased
heterogeneity can provide more crevices and spaces to establish refuge locations, as well

68

as multiple routes to hide and evade predator detection (Bertolino and Cordero Di
Montezemolo 2007).
Increased levels of CWD was the final predictor variable for deer mouse refuge site
selection at the fourth order level, and CWD has been shown to be an important habitat
component for small mammals, particularly in forested habitats (e.g. Loeb 1999,
Bowman et al. 2000, Butts and McComb 2000, Johnston and Anthony 2008). Studies
examining the nesting and refuge site selection of rodents in forests have found small
mammals using CWD as primary nest and refuge sites (e.g. McCay 2000, Hinkelman and
Loeb 2007), and in this study, 2 of the 40 refuge sites were found inside the boles of dead
sagebrush.
Further to CWD providing primary refuge site locations, it may provide secondary
benefits to the establishment of refuge sites. In hot sagebrush habitats, CWD may act in
ways similar to live shrub cover, with standing woody debris providing sites of decreased
risks of predation and of moderated environmental conditions conducive to refuge site
selection. Downed (i.e. horizontal) woody debris are used as silent, efficient travel routes
by species of Peromyscus in forest settings, presumably to avoid detection by predators
(Barnum et al. 1992, Roche et al. 1999, McCay 2000). Downed Artemisia trunks,
although generally much smaller in diameter than forest CWD, may provide similar
services in shrub-steppe settings by providing travel corridors that reduce the risk of
predation for deer mice entering and exiting refuge locations.
At the third order level, neither the presence of large Artemisia nor the levels of shrub
or herbaceous plant cover were included in the final models predicting deer mouse use;
this was an unexpected result. At that scale, deer mice were selecting areas with
increased slope and decreased levels of plant litter, a result that supports Johnson‟s
(1980) supposition that animals can make selection choices at different spatial scales, and
supports previous works that examined habitat selection of small mammals at different
habitat levels (Morris 1984, 1987, Stapp 1997).
The use of steep terrain by deer mice might have its basis in the physical
characteristics such slopes can provide. The fine-grained top soil at the study site

69

appeared quite prone to erosion, with summer thunderstorms creating numerous rills and
small cavities on the steeper areas of the site. The apparent increased heterogeneity of
the ground surface on steeper slopes may provide greater opportunities for the
establishment of refuge sites. Further, Artemisia boles and roots on steep slopes often
were observed to create overhangs above the downhill slope, and were targeted by deer
mice for burrow placement on several occasions. Deer mice may select refuge sites at the
bases of large shrubs on steeper slopes, due to the increased protection such shrubs may
provide.
There may be additional benefits to deer mice selecting refuge locations on steep
slopes. The study site is subject to periodic grazing by cattle and it has been shown that
cattle reduce their grazing on steeper slopes (Holechek 1988, Bailey 2005, Bailey et al.
2006), preferring to congregate and forage on flatter areas. Cattle observed at the study
site appeared to follow these patterns, restricting their foraging and movement patterns to
flat or gentle terrain in the area. Signs of recent or historic cattle use (e.g. hoof prints,
feces, disturbed/flattened shrubs) were virtually absent from all areas with steep slopes.
Deer mice selecting refuge sites on steep slopes may benefit from a reduced risk of
livestock disturbance. The study site also receives quite a bit of human disturbance,
namely from vehicle traffic (e.g. four-wheel-drive trucks, all-terrain vehicles and dirt
bikes), as well as from hikers and the occasional cyclist. Human activity was generally
concentrated on flatter terrain or ridgelines, and so as with livestock disturbance,
selecting steep slopes for refuge sites could reduce the potential of human disturbance.
Litter was the second variable included in the top mode for third order habitat
selection, with deer mice choosing areas with decreased levels of litter at this scale. This
result is a bit surprising, yet previous studies have documented the preferential selection
of areas with reduced litter by deer mice (Kaufman et al. 1988). A long-standing theory
is that such areas allow deer mice to more easily predate on grass seeds that would
otherwise be more difficult to detect in areas with increased litter (Kaufman et al. 1988,
Kaufman and Kaufman 1990, Reed et al. 2004). As such, deer mouse fourth order
habitat selection may be influenced by increased seed-forage opportunities at a higher

70

spatial scale. At this stage, however, further study is required to confirm the importance
of decreased litter levels of deer mouse habitat selection at the third order level.
It needs to be recognized that the top third order model did not do an exceptional job
of predicting deer mouse habitat use. This indicates that deer mice may be basing refuge
site area selection choices on habitat characteristics other than those measured in this
study (e.g. soil compaction/friability, arthropod densities), that the third order habitats
were not adequately surveyed, or that deer mice were not selecting habitats at the scale
measured.
An important assumption of both paired logistic regression and the calculation of
RSFs with a used-available design is that the points designated as “available” represent
habitats that are actually available to the animal or individual in question (Compton et al.
2002): an assumption that is difficult not to violate with cryptic wildlife such as deer
mice. The current study failed to determine the home ranges of collared individuals, and
as such, „available‟ habitat plot measurements may have fallen outside of an individual
animal‟s home range, and thus would not represent truly available habitat. It was
assumed that fourth order plots were available to the deer mouse being tracked, due to
their proximity to their paired telemetry location, but the third order habitat plots may
have violated this assumption. However, because of the low deer mouse densities
observed over the course of the study, averaging six individuals per hectare, it is believed
that conspecific interference was minimal and that collared animals had access to optimal
daytime refuge sites on the landscape (Rosenzweig 1989, Thompson 2004).
In summary, deer mice can play a significant role in semi-arid grassland ecosystems,
and the selection of nests and daytime refuge sites may be one factor that affects the local
distribution of the species. Shrub cover has been shown to be important to small
mammals, and in the shrub-steppe grasslands of British Columbia, deer mice appear to
select refuge sites at locations that not only have larger-diameter Artemisia, but also
decreased levels of bare ground and increased amounts of CWD. These locations seem
to offer mice a form of thermal protection from the extreme daytime conditions of the
low-elevation grasslands, and managers wishing to keep deer mice present on the

71

landscape should consider preserving some of these habitat attributes when making land
management decisions, in a manner akin to maintaining large snags for cavity nesting
animals within forest ecosystems (Carey 2000, Payer and Harrison 2003, Walter and
Maguire 2005, Oaten 2007). At the third order level, deer mice may be selecting refuge
sites in areas with increased slope and decreased litter, although other habitat traits may
be influencing deer mouse resource selections at this scale, and further research is
warranted. Despite the limitations of our study, the information presented here is the first
to provide insights into the refuge site selection of deer mice in northern semi-arid
grasslands. Information presented in this study can be used to direct future research into
the selection habits of deer mice and other small mammal species.
LITERATURE CITED
BAILEY, D. W. 2005. Identification and creation of optimal habitat conditions for
livestock. Rangeland Ecology & Management 58: 109-118.
BAILEY, D. W., H. C. VANWAGONER, AND R. WEINMEISTER. 2006. Individual animal
selection has the potential to improve uniformity of grazing on foothill rangeland.
Rangeland Ecology & Management 59: 351-358.
BARNUM, S. A., C. J. MANVILLE, J. R. TESTER, AND W. J. CARMEN. 1992. Path selection
by Peromyscus leucopus in the presence and absence of vegetative cover. Journal of
Mammalogy 73: 797-801.
BARRY, R. E., JR., M. A. BOTJE, AND L. B. GRANTHAM. 1984. Vertical stratification of
Peromyscus leucopus and P. maniculatus in southwestern Virginia. Journal of
Mammalogy 65: 145-148.
BERTOLINO, S., AND N. CORDERO DI MONTEZEMOLO. 2007. Garden dormouse (Eliomys
quercinus) nest site selection in an alpine habitat. Ethology Ecology & Evolution 19: 5160.
BOWMAN, J. C., D. SLEEP, G. J. FORBES, AND M. EDWARDS. 2000. The association of
small mammals with coarse woody debris at log and stand scales. Forest Ecology and
Management 129: 119-124.

72

BOYCE, M. S., P. R. VERNIER, S. E. NIELSEN, AND F. K. A. SCHMIEGELOW. 2002.
Evaluating resource selection functions. Ecological Modelling 157: 281-300.
BRIGHT, P. W., AND P. A. MORRIS. 1991. Ranging and nesting behaviour of the dormouse,
Muscardinus avellanarius, in diverse low-growing woodland. Journal of Zoology 224:
177-190.
BURNHAM, K. P., AND D. R. ANDERSON. 2002. Model selection and multimodel inference:
a practical information-theoretic approach. 2nd ed. Springer-Verlag, New York.
BUTTS, S. R., AND W. C. MCCOMB. 2000. Associations of forest-floor vertebrates with
coarse woody debris in managed forests of western Oregon. Journal of Wildlife
Management 64: 95-104.
CAREY, A. B. 2000. Effects of new forest management strategies on squirrel populations.
Ecological Applications 10: 248-257.
CHAMBERS, J. C. 2001. Pinus monophylla establishment in an expanding Pinus-Juniperus
woodland: environmental conditions, facilitation and interacting factors. Journal of
Vegetation Science 12: 27-40.
COMPTON , B. W., J. M. RHYMER, AND M. MCCOLLOUGH. 2002. Habitat selection by
wood turtles (Clemmys insculpta): an application of paired logistic regression. Ecology
83: 833-843.
DAVIES, K. W., J. D. BATES, AND J. J. JAMES. 2009. Microsite and herbaceous vegetation
heterogeneity after burning Artemisia tridentata steppe. Oecologia 159: 597-606.
DAVIES, K. W., J. D. BATES, AND R. F. MILLER. 2007. The influence of Artemisia
tridentata ssp. wyomingensis on microsite and herbaceous vegetation heterogeneity.
Journal of Arid Environments 69: 441-457.
DOOLEY, J. L. JR., AND R. D. DUESER. 1990. An experimental examination of nest-site
segregation by two Peromyscus species. Ecology 71: 788-796.
DOOLEY, J. L. JR., AND R. D. DUESER. 1996. Experimental tests of nest site competition in
two Peromyscus species. Oecologia 105: 81-86.

73

FIELDING, A. H., AND J. F. BELL. 1997. A review of methods for the assessment of
prediction errors in conservation presence/absence models. Environmental Conservation
24: 38-49.
FRANK, P. A., AND J. N. LAYNE. 1992. Nests and daytime refugia of cotton mice
(Peromyscus gossypinus) and golden mice (Ochrotomys nuttalli) in south-central Florida.
American Midland Naturalist 127: 21-30.
GROVES, C. R., AND J. E. ANDERSON. 1981. Allelopathic effects of Artemisia tridentata
leaves on germination and growth of two grass species. American Midland Naturalist
106: 73-79.
HARNEY, B. A., AND R. D. DUESER. 1987. Vertical stratification of two Peromyscus
species: an experimental analysis. Ecology 68: 1084-1091.
HARROWER, W. L. 2007. Nesting requirements of the northern goshawk (Accipiter
gentilis atricapillus) in southeastern British Columbia. M.Sc. thesis, University of
Victoria, Victoria, Canada.
HINKELMAN, T. M., AND S. C. LOEB. 2007. Effect of woody debris abundance on daytime
refuge use by cotton mice. Southeastern Naturalist 6: 393-406.
HOLECHEK, J. L. 1988. An approach for setting the stocking rate. Rangelands 10: 10-14.
HOSMER, D. W., AND S. LEMESHOW. 2000. Applied logistic regression. 2nd ed. John
Wiley & Sons, Inc., New York.
JOHNSON, D. H. 1980. The comparison of usage and availability measurements for
evaluating resource preference. Ecology 61: 65-71.
JOHNSTSON, A. N., AND R. G. ANTHONY. 2008. Small-mammal microhabitat associations
and response to grazing in Oregon. Journal of Wildlife Management 72: 1736-1746.
KAUFMAN, D. W., AND G. A. KAUFMAN. 1990. Influence of plant litter on patch use by
foraging Peromyscus maniculatus and Reithrodontomys megalotis. American Midland
Naturalist 124: 195-198.
KAUFMAN, G. A., D. W. KAUFMAN, AND E. J. FINCK. 1988. Influence of fire and
topography on habitat selection by Peromyscus maniculatus and Reithrodontomys
megalotis in ungrazed tallgrass prairie. Journal of Mammalogy 69: 342-352.

74

KLEIN, H. G., AND J. N. LAYNE. 1978. Nesting behavior in four species of mice. Journal of
Mammalogy 59: 103-108.
KOTLER B. P. 1984. Risk of predation and the structure of desert rodent communities.
Ecology 65: 689-701.
KOTLER, B. P., AND J. S. BROWN. 1988. Environmental heterogeneity and the coexistence
of desert rodents. Annual Review of Ecology and Systematics 19: 281-307.
LOEB, S. C. 1999. Responses of small mammals to coarse woody debris in a southeastern
pine forest. Journal of Mammalogy 80: 460-471.
MCCAY, T. S. 2000. Use of woody debris by cotton mice (Peromyscus gossypinus) in a
southeastern pine forest. Journal of Mammalogy 81: 527-535.
MCCRAVY, K. W., AND R. K. ROSE. 1992. An analysis of external features as predictors
of reproductive status in small mammals. Journal of Mammalogy 73: 151-159.
MORRIS, D. W. 1984. Patterns and scale of habitat use in two temperate-zone, small
mammal faunas. Canadian Journal of Zoology 62: 1540-1547.
MORRIS, D. W. 1987. Ecological scale and habitat use. Ecology 68: 362-369.
MOSES R. A., AND S. BOUTIN. 2001. The influence of clear-cut logging and residual leave
material on small mammal populations in aspen-dominated boreal mixedwoods.
Canadian Journal of Forest Research 31: 483-495.
NAGORSEN, D. W. 2005. Rodents & lagomorphs of British Columbia. Royal British
Columbia Museum, Victoria, British Columbia, Canada 4: 1-410.
OATEN, D. K. 2007. Biodiversity within dry forests of the interior of British Columbia:
the role of aspen and stand structure. M.Sc. thesis, University of British Columbia,
Vancouver, Canada.
PARMENTER, R. R., AND J. A. MACMAHON. 1983. Factors determining the abundance and
distribution of rodents in shrub-steppe ecosystems: the role of shrubs. Oecologia 59: 145156.

75

PARMENTER, R. R., C. A. PARMENTER, AND C. D. CHENEY. 1989. Factors influencing
microhabitat partitioning among coexisting species of arid-land darkling beetles
(Tenebrionidae): behavioral responses to vegetation architecture. Southwestern Naturalist
34: 319-329.
PAYER, D. C., AND D. J. HARRISON. 2003. Influence of forest structure on habitat use by
American marten in an industrial forest. Forest Ecology and Management 179: 145-156.
PEDUZZI, P., J. CONCATO, E. KEMPER, T. R. HOLFORD, AND A. R. FEINSTEIN. 1996. A
simulation study of the number of events per variable in logistic regression analysis.
Journal of Clinical Epidemiology 49: 1373-1379.
PIERCE, B. M., W. S. LONGLAND, AND S. K. JENKINS. 1992. Rattlesnake predation on
desert rodents: microhabitat and species-specific effects on risk. Journal of Mammalogy
73: 859-865.
PIERSON, F. B., AND J. R. WIGHT. 1991. Variability of near-surface soil temperature on
sagebrush rangeland. Journal of Range Management 44: 491-497.
REDFIELD, J. A., C. J. KREBS, AND M. J. TAITT. 1977. Competition between Peromyscus
maniculatus and Microtus townsendii in grasslands of coastal British Columbia. Journal
of Animal Ecology 46: 607-616.
REED, A. W., G. A. KAUFMAN, AND D. W. KAUFMAN. 2004. Influence of fire, topography,
and consumer abundance on seed predation in tallgrass prairie. Canadian Journal of
Zoology 82: 1459-1467.
ROCHE, B. E., A. I. SCHULTE-HOSTEDDE, AND R. J. BROOKS. 1999. Route choice by deer
mice (Peromyscus maniculatus): reducing the risk of auditory detection by predators.
American Midland Naturalist 142: 194-197.
ROSENZWEIG, M. L. 1989. Habitat selection, community organization, and small mammal
studies. Patterns in the structure of mammalian communities. Pp. 5-21 in Patterns in the
structure of mammalian communities (D. W. Morris, Z. Abramsky, B. J. Fox, and M. R.
Willig, eds.). Texas Technical University Press, Lubbock, Texas.
SAS INSTITUTE INC. 2008. SAS/STAT software: statistical analysis software version 9.2.
SAS Institute Inc., Cary, North Carolina.
STAPP, P. 1997. Habitat selection by an insectivorous rodent: patterns and mechanisms
across multiple scales. Journal of Mammalogy 78: 1128-1143.

76

STAPP, P., AND B. VAN HORNE. 1997. Response of deer mice (Peromyscus maniculatus)
to shrubs in shortgrass prairie: linking small-scale movements and the spatial distribution
of individuals. Functional Ecology 11: 644-651.
TAIT, M. J. 1981. The effect of extra food on small rodent populations: I. deermice
(Peromyscus maniculatus). Journal of Animal Ecology 50: 111-124.
THOMPSON, I. D. 2004. The importance of superior-quality wildlife habitats. Forestry
Chronicle. 80: 75-81.
TISDALE, E. W. 1947. The grasslands of the southern interior of British Columbia.
Ecology 28: 346-382.
VAN RYSWYK, A. L., A. MCLEAN, AND L. S. MARCHAND. 1966. The climate, native

vegetation, and soils of some grasslands at different elevations in British Columbia.
Canadian Journal of Plant Science 46: 35-50.
WALSBERG, G. E. 2000. Small mammals in hot deserts: some generalizations revisited.
BioScience 50: 109-120.
WALTER, S. T., AND C. G. MAGUIRE. 2005. Snags, cavity-nesting birds, and silvicultural
treatments in western Oregon. Journal of Wildlife Management 69: 1578-1591.
WEAVER, T. W., AND D. KLARICH. 1977. Allelopathic effects of volatile substances from
Artemisia tridentata Nutt. American Midland Naturalist 97: 508-512.
WHITTINGTON, J., C. C. ST. CLAIR, AND G. MERCER. 2005. Spatial responses of wolves to
roads and trails in mountain valleys. Ecological Applications 15: 543-553.
WIENS, T. S., B. C. DALE, M. S. BOYCE, AND G. P. KERSHAW. 2008. Three way k-fold
cross-validation of resource selection functions. Ecological Modelling 212: 244-255.
WIKEEM, B., AND S. WIKEEM. 2004. The grasslands of British Columbia. Grasslands
Conservation Council of British Columbia, Kamloops, British Columbia, Canada.
WOLFF, J. O., AND D. S. DURR. 1986. Winter nesting behavior of Peromyscus leucopus
and Peromyscus maniculatus. Journal of Mammalogy 67: 409-412.
WOLFF, J. O., AND B. HURLBUTT. 1982. Day refuges of Peromyscus leucopus and
Peromyscus maniculatus. Journal of Mammalogy 63: 666-668.

77

ZAR, J. H. 1999. Biostatistical analysis. 4th ed. Prentice Hall, Upper Saddle River, New
Jersey.

78

CHAPTER 4 – CONCLUSIONS
Review and Research Limitations
The overall goal of this thesis was to increase knowledge about small mammal
communities in British Columbia temperate grasslands. To achieve this goal, two
objectives were pursued: 1) to compare small mammal population and demographic
information across markedly different upper and lower elevation grassland communities;
and 2) to examine the third and fourth order resource selection of daytime refuge sites by
deer mice (Peromyscus maniculatus) in a semi-arid grassland setting. Major results
related to the first objective included: (a) total small mammal densities tended to be
higher in the upper grasslands, and were significantly higher when data were pooled
across years but not when analyzed on a per-year basis; (b) vole species (Microtus spp.)
were found almost exclusively in the upper grasslands; (c) deer mice were found in both
grassland habitats and densities tended to be higher in the lower grassland. These
differences were not significantly different when data were pooled across years, and
rarely significantly different when analyzed on a per-year basis; and (d) deer mouse
apparent survival estimates did not differ between the upper and lower grasslands.
Further, small mammal populations showed, at times, high levels of variability between
both trapping sessions and years. For the second objective, deer mice appeared to be
making third order selections based on the slope and percent cover of the landscape, and
selecting daytime refuge sites (fourth order selections) at locations with large diameter
big sagebrush (Artemisia tridentata), decreased amounts of bare ground, and increased
amounts of coarse woody debris (CWD).
Small mammal populations are notorious for being highly dynamic, both spatially and
temporally. Population irruptions and declines can occur over small or large scales
(Korpimäki and Krebs 1996, Korpimäki et al. 2004), and although there is evidence that
climatic variables may play a role in population fluctuations (Brady and Slade 2004,
Korpimäki et al. 2004, Reed et al. 2007, Shenbrot et al. 2010, Thibault et al. 2010), a
complete understanding of the underlying mechanisms has yet to be ascertained. The

79

cyclic nature of some small mammal populations (Korpimäki and Krebs 1996, Brady and
Slade 2004, Korpimäki et al. 2004, Bartel et al. 2008) can further complicate studies into
the characteristics of their communities. Such studies require long-term data sets
collected from adequately replicated research sites in order to avoid making erroneous
conclusions based on interpretations of spurious results (Brady and Slade 2004). The
present study collected data from eight main sites over three seasons of study, and so
there is a risk that the patterns observed may not reflect the long-term structure and
dynamics of small mammal communities in the areas studied. Future research efforts
should endeavour to increase the number of replicate study sites in each grassland type,
and attempt to collect data for 8 to 10 years, minimum. However, as always, funding
avenues for this type of work will be a constraining factor.
Site selection may limit the ability to extrapolate these results to other areas of the BC
grasslands, particularly those subject to invasion by alien plant species. Study sites in
both the upper and lower grasslands were selected based on size (minimum one hectare)
and plant community, with areas having appreciative amounts of non-native or invasive
plant species avoided (Rankin unpubl.). Invasive plants are prevalent across most
grasslands in British Columbia (Wikeem and Wikeem 2004), with the areas in and
around Lac du Bois Grasslands Provincial Park inundated with species such as diffuse
knapweed (Centaurea diffusa), spotted knapweed (Centaurea stoebe), Dalmatian toadflax
(Linaria genistifolia) and cheatgrass (Bromus tectorum). The presence of these invasive
species may alter the habitat characteristics (e.g. cover) of the plant community upon
which small mammals rely, and so may have effects on the small mammal communities
themselves (Pearson et al. 2000, 2001, Ostoja and Schupp 2009). Because the sites
chosen for the current study purposely avoided areas with invasive plants, my results may
not translate well to the majority of the grasslands in the area, where invasive plant
species can be amply present.
With respect to the radio telemetry portion of the study, I was limited in the number of
animals I could track due to logistics and resources, and a larger focused study would be
able to provide a more thorough understanding of not only refuge site selection, but

80

foraging behaviour, home range attributes, and other important life-history attributes of
the animals. And although there is a belief that habitat selection at low population levels
allows individuals to use the „best‟ habitat resources (Rosenzweig 1989, Thompson
2004), the relatively small number of mice tracked for this study warrant caution when
drawing conclusions from the results. A further limitation to this portion of the study was
the inability to demarcate home ranges for the deer mice that were tracked. This is an
important factor when attempting to determine what habitats are available to an
individual when making resource selections (Compton et al. 2002), and may have been a
reason for the marginal ability of the third order model to predict deer mouse habitat use.
Application of Research Findings
Land or wildlife managers rarely state that they are managing for small mammal
populations, and indeed unless the small mammal in question is threatened with
extinction or extirpation, rodent management normally falls very far down the list of
priority actions or consideration. However, given their important roles in grassland
systems (see examples in Chapter 2), land managers tasked with overseeing grasslands
(particularly the endangered British Columbia grasslands) would be wise to consider
maintaining rodent populations when planning management activities.
Anthropogenic activities such as grazing and prescribed burning to promote forage
growth can have long-term effects on Artemisia and on levels of CWD (Harniss and
Murray 1973, Wikeem and Strang 1983). Artemisia, generally the primary source of
CWD in the lower grasslands, is a relatively slow-growing plant: Perryman and Olson
(2000), for example, showed that in Wyoming USA, Artemisia could take up to 30 years
to reach the average stem diameter of those used by deer mice in this study. Land
managers interested in maintaining deer mouse populations should consider preserving
some large-diameter Artemisia on the landscape to benefit this rodent species, in a
manner akin to maintaining or creating large snags for cavity nesting animals within
managed forest ecosystems (Carey 2000, Payer and Harrison 2003, Walter and Maguire
2005, Oaten 2007).

81

One of the important ecological functions of rodents is their role as sources of prey for
a number of threatened and endangered grassland predators, including the western
rattlesnake (Crotalus oreganus), gopher snake (Pituophis catenifer) and badger (Taxidea
taxus). The provincial recovery strategies for both the badger and the deserticola
subspecies of gopher snake (P.c. deserticola) cite the need for a better understanding of
how small mammals respond to land management activities, such as grazing. The
recovery strategy for badger further states that more detailed information on badger prey
ecology is also necessary for the species‟ recovery, and Hoodicoff (2006) recommended
collecting more localized information on historic, current and future prey population
trends. Overall, this is a key knowledge gap that needs to be addressed to help aid in the
recovery of these and other predator species in British Columbia (jeffersonii Badger
Recovery Team 2008, Southern Interior Reptile and Amphibian Recovery Team 2008).
Data presented in this study begins to fill this gap by providing local information on
current small mammal populations in two grassland ecosystems.
Future Research
The constraints of this study notwithstanding, a number of recommendations can be
made for future work, the first being a recommendation to establish long-term monitoring
sites in the upper, middle and lower grasslands.
As discussed, long-term monitoring projects are essential to understanding small
mammal populations, due to their highly dynamic nature. A long-term study should help
further elucidate and confirm typical small mammal densities in the three grassland types.
Any such study should endeavour to concurrently collect vegetative and climate data, as
well as information on key grassland predators, particularly those threatened and
endangered species listed above, that would be using rodents as prey sources. Such
community-level projects are few and far between (e.g. Krebs et al. 1995, Bartel et al.
2008), but would be fundamental to understanding the factors that may be driving any
observed small mammal population fluctuations.

82

As shown in this study and others (e.g. Merritt and Merritt 1978, Boonstra and Krebs
2006, Larsen et al. 2007), winter appears to be a difficult time for small mammals, and
any long-term monitoring program should attempt to collect population and habitat data
during the winter months. The snow shelters used in this study to protect the live-traps
from the elements (see Figure 2.1) proved to be somewhat ineffectual. Although they
were inexpensive and easy to build, their low profile made access to the traps difficult
and time consuming. Future winter studies should consider using taller trap “chimneys”
that allow researchers to access traps below the snow cover with minimal disturbance to
the subnivean space (see for example Merritt and Merritt 1978, Korslund and Steen
2006).
Any long-term monitoring program would have the added benefit of providing data on
long-term changes to the grassland community as a whole, which may be particularly
relevant given the ongoing speculation with respect to the effects of climate change in
British Columbia. How future changes in climate will affect provincial ecosystems
remains a hotly debated topic. Hamman and Wang (2006) predicted large latitudinal and
elevational expansions of the Bunchgrass and the Interior Douglas Fir biogeoclimatic
ecological zones. Should this occur, the upper and middle grasslands in and around Lac
du Bois may experience an ecological shift towards systems more like those found in the
lower grasslands today, and accompanied by a shift in rodent functional groups, with the
Microtus species in the upper grasslands replaced by the more xeric-tolerant deer mouse.
Other studies have shown, however, that deer mouse populations may decrease
significantly in times of reduced precipitation (Reed et al. 2007), so if future climate
change results in an overall decrease in precipitation levels, deer mouse populations in
both the lower or upper grasslands may suffer. A long-term monitoring program could
catalogue these trends and provide useful information for predicting future impacts of
climate change on grassland rodent communities.
Habitat and resource selection studies are gaining popularity with wildlife researchers,
and the tools to analyze selection data are becoming more elegant and advanced (Boyce
et al. 2002, Compton et al. 2002, Whittington et al. 2005, Wiens et al. 2008). The current

83

study focussed on one narrow aspect of deer mouse resource selection (i.e. the selection
of daytime refuge sites in the lower grasslands), and there is still much to learn about the
selection habits of both deer mice and species of Microtus. With lessons learned from
this study, future researchers may want to undertake more fulsome habitat selection
projects to determine other potential resources that may be influencing the densities and
distributions of small mammals in grassland ecosystems. Such studies will provide
valuable information on rodent ecology and social structures, such as home range sizes,
that can be lacking for several grassland small mammal species in the interior of British
Columbia (Nagorsen 2005).
Concluding Remarks
Rodent species occupy pivotal positions in grassland ecosystems, both as primary
consumers and modifiers of grassland resources, and as prey for a variety of grassland
predators. And although their roles may be particularly acute in rare British Columbia
grasslands, where they serve as sources of prey for a number of imperilled grassland
species, little work has been done to asses and monitor rodent populations in these
ecosystems. The current study is one of the first to offer information to begin filling this
knowledge gap, and its results provide valuable insights into the densities and
composition, survival rates and demographics of local rodent communities in two
grassland types. Further, the telemetry portion of the study provided new information on
habitat resources that may be important to a particular rodent species within a semi-arid
grassland system. Wildlife and land managers can use this information to help inform
species-specific recovery plans as well as anthropogenic-related activities in grasslands
so as to maintain rodent populations on the landscape. Finally, this study has laid some
of the groundwork needed to guide future studies in the intricacies of rodent communities
and resource selection in grassland ecosystems, including those in British Columbia.

84

LITERATURE CITED
BARTEL, R. A., F. F. KNOWLTON, AND C. STODDART. 2008. Long-term patterns in
mammalian abundance in northern portions of the Great Basin. Journal of Mammalogy
89: 1170-1183.
BOONSTRA, R., AND C. J. KREBS. 2006. Population limitation of the northern red-backed
vole in the boreal forests of northern Canada. Journal of Animal Ecology 75: 1269-1284.
BOYCE, M. S., P. R. VERNIER, S. E. NIELSEN, AND F. K. A. SCHMIEGELOW. 2002.
Evaluating resource selection functions. Ecological Modelling 157: 281-300.
BRADY, M. J., AND N. A. SLADE. 2004. Long-term dynamics of a grassland rodent
community. Journal of Mammalogy 85: 552-561.
CAREY, A. B. 2000. Effects of new forest management strategies on squirrel populations.
Ecological Applications 10: 248-257.
COMPTON , B. W., J. M. RHYMER, AND M. MCCOLLOUGH. 2002. Habitat selection by
wood turtles (Clemmys insculpta): an application of paired logistic regression. Ecology
83: 833-843.
HAMANN, A., AND T. WANG. 2006. Potential effects of climate change on ecosystem and
tree species distribution in British Columbia. Ecology 87: 2773-2786.
HARNISS, R. O., AND R. B. MURRAY. 1973. 30 years of vegetal change following burning
of sagebrush-grass range. Journal of Range Management 26: 322-325.
HOODICOFFF, C. 2006. Badger prey ecology: the ecology of six small mammals found in
British Columbia. British Columbia Ministry of Environment, Wildlife Working Report
No. WR-109.
JEFFERSONII BADGER RECOVERY TEAM. 2008. Recovery strategy for the badger (Taxidea

taxus) in British Columbia. British Columbia Ministry of Environment, British Columbia
Recovery Strategy Series.
KORPIMÄKI, E., P. R. BROWN, J. JACOB, AND R. P. PECH. 2004. The puzzles of population
cycles and outbreaks of small mammals solved. BioScience 54: 1071-1079.

85

KORPIMÄKI, E., AND C. J. KREBS. 1996. Predation and population cycles of small
mammals. BioScience 46: 754-764.
KORSLUND, L., AND H. STEEN. 2006. Small rodent winter survival: snow conditions limit
access to food resources. Journal of Animal Ecology 75: 156-166.
KREBS, C. J., S. BOUTIN, R. BOONSTRA, A. R. E. SINCLAIR, AND J. N. M. SMITH. 1995.
Impact of food and predation on the snowshoe hare cycle. Science 269: 1112-1115.
LARSEN, K. W., ADAMS, I. T., AND D. L. HAUGHLAND. 2007. Small mammal communities
in a pyrogenic habitat mosaic. International Journal of Wildland Fire 16: 728-740.
MERRITT, J. F., AND J. M. MERRITT. 1978. Population ecology and energy relationships of
Clethrionomys gapperi in a Colorado subalpine forest. Journal of Mammalogy 59: 576598.
NAGORSEN, D. W. 2005. Rodents & lagomorphs of British Columbia. Royal British
Columbia Museum, Victoria, British Columbia, Canada 4: 1-410.
OATEN, D. K. 2007. Biodiversity within dry forests of the interior of British Columbia:
the role of aspen and stand structure. M.Sc. thesis, University of British Columbia,
Vancouver, Canada.
OSTOJA, S. M., AND E. W. SCHUPP. 2009. Conversion of sagebrush shrublands to exotic
annual grasslands negatively impacts small mammal communities. Diversity and
Distributions 15: 863-870.
PAYER, D. C., AND D. J. HARRISON. 2003. Influence of forest structure on habitat use by
American marten in an industrial forest. Forest Ecology and Management 179: 145-156.
PEARSON, D. E., K. S. MCKELVEY, AND L. F. RUGGIERO. 2000. Non-target effects of an
introduced biological control agent on deer mouse ecology. Oecologia 122: 121-128.
PEARSON, D. E., Y. K. ORTEGA, K. S. MCKELVEY, AND L. F. RUGGIERO. 2001. Small
mammal communities and habitat selection in northern Rocky Mountain bunchgrass:
implications for exotic plant invasions. Northwest Science 75: 107-117.
PERRYMAN, B. L. AND R. A. OLSON. 2000. Age-stem diameter relationships of Big
Sagebrush and their management implications. Journal of Range Management 53: 342346.

86

REED, A. W., G. A. KAUFMAN, AND B. K. SANDERCOCK. 2007. Demographic response of a
grassland rodent to environmental variability. Journal of Mammalogy 88: 982-988.
ROSENZWEIG, M. L. 1989. Habitat selection, community organization, and small mammal
studies. Patterns in the structure of mammalian communities. Pp. 5-21 in Patterns in the
structure of mammalian communities (D. W. Morris, Z. Abramsky, B. J. Fox, and M. R.
Willig, eds.). Texas Technical University Press, Lubbock, Texas.
SHENBROT, G., B. KRASNOV, AND S. BURDELOV. 2010. Long-term study of population
dynamics and habitat selection of rodents in the Negev Desert. Journal of Mammalogy
91: 776-786.
SOUTHERN INTERIOR REPTILE AND AMPHIBIAN RECOVERY TEAM. 2008. Recovery strategy
for the gopher snake, deserticola subspecies (Pituophis catenifer deserticola) in British
Columbia. British Columbia Ministry of Environment, British Columbia Recovery
Strategy Series.
THIBAULT, K. M., S. K. M. ERNEST, E. P. WHITE, J. H. BROWN, AND J. R.GOHEEN. 2010.
Long-term insights into the influence of precipitation on community dynamics in desert
rodents. Journal ofMammalogy 91: 787-797.
THOMPSON, I. D. 2004. The importance of superior-quality wildlife habitats. Forestry
Chronicle. 80: 75-81.
WALTER, S. T., AND C. G. MAGUIRE. 2005. Snags, cavity-nesting birds, and silvicultural
treatments in western Oregon. Journal of Wildlife Management 69: 1578-1591.
WHITTINGTON, J., C. C. ST. CLAIR, AND G. MERCER. 2005. Spatial responses of wolves to
roads and trails in mountain valleys. Ecological Applications 15: 543-553.
WIENS, T. S., B. C. DALE, M. S. BOYCE, AND G. P. KERSHAW. 2008. Three way k-fold
cross-validation of resource selection functions. Ecological Modelling 212: 244-255.
WIKEEM , B. M., AND R. M. STRANG. 1983. Prescribed burning on B.C. rangelands: the
state of the art. Journal of Range Management 36: 3-8.
WIKEEM, B., AND S. WIKEEM. 2004. The grasslands of British Columbia. Grasslands
Conservation Council of British Columbia, Kamloops, British Columbia, Canada.

87