THE EFFECTS OF SEX, TERRAIN, WILDFIRE, WINTER SEVERITY,
AND MATERNAL STATUS ON HABITAT SELECTION BY MOOSE
IN NORTH-CENTRAL ALASKA

Kyle Joly1, Mathew S. Sorum1, Tim Craig2,4, and Erin L. Julianus3

1National Park Service, Gates of the Arctic National Park and Preserve, 4175 Geist Road, Fairbanks,
AK 99709; 2US Fish and Wildlife Service, Kanuti National Wildlife Refuge, 101 12th Avenue, Fairbanks,
AK 99701; 3Bureau of Land Management, Central Yukon Field Office, 1150 University Avenue,
Fairbanks, AK 99709; 4Retired

ABSTRACT: Habitat selection is a central component of the ecology of individual animals as it
affects body condition, survivorship, and reproductive output. We instrumented male and female
moose (Alces alces) in north-central Alaska with GPS radio-collars to assess factors we hypothesized
were important to their habitat selection. Using synoptic modeling techniques, we found that models
with more covariates were better predictors of moose habitat selection than more simplistic models.
As expected, moose selected for habitats with high canopy cover and/or that typically have abundant
forage such as 11-30 year old burned areas. However, we detected differences in habitat selection be-
tween sexes, seasons (i.e., winter versus summer), during winters of varying severity, and females with
differing maternal status. During winter males moved to lower elevations areas, presumably to avoid
greater snow depths, whereas females remained at relatively similar elevations. Females selected
burned habitat and areas that received higher amounts of solar radiation. We found that all moose
selected for lower elevation habitats closer to rivers during moderate and severe winters, but elevation
was not a strong influence during mild winters. We found that females with calves avoided riparian
habitats and selected areas with more forested habitat than females without calves during both summer
and winter. This suggests a trade-off between maximizing forage intake and reducing predation risk for
their offspring. Our and similar data are useful to improve moose management strategies and provide a
benchmark against which the impacts of climate change and industrial development are assessed in
this rapidly-changing region.

ALCES VOL. 52: 101–115 (2016)

Key words: Alces alces, arctic, Brooks Range, maternal status, moose, synoptic modeling, winter
severity

An animal’s use of the landscape affects
its body condition, reproductive output, sur-
vivorship, and fitness (Gaillard et al. 2010).
Thus, studies of habitat selection are also in-
formative to understanding the ecology of
vagile species. Although habitat selection by
moose (Alces alces) has been well documen-
ted in North American populations (see
Peek 1997), there is a paucity of habitat selec-
tion studies in northern Alaska. Patterns of
selection by moose differ among and within
populations, and between sexes and seasons.

Alaskan moose are sexually dimorphic in
body size, and sexual segregation is well
documented (Miquelle et al. 1992, Bowyer
et al. 2001, Oehlers et al. 2011). Barboza
and Bowyer (2000) suggested that sex-
related differences in habitat selection pat-
terns can be explained by differences in
body size and annual changes in the physi-
ology and morphology between sexes. Large
males are able to consume large quantities of
low-quality forage, whereas smaller-bodied
females are better adapted for smaller

101



quantities of high-quality forage. Risk of pre-
dation is also thought to play a major role
in the selection of habitats by moose, particu-
larly females with calves (Dussault et al.
2005, Poole et al. 2007, Oehlers et al. 2011).
Moose reduce risk of predation by avoiding
travel routes used by predators (Kunkel and
Pletscher 1999, Dussault et al. 2005) and
selecting habitats that provide greater
concealment (Oehlers et al. 2011). Terrain
features and snow conditions also influence
patterns of distribution and selection by
moose (Poole and Stuart-Smith 2006). Within
interior mountain areas, moose tend to des-
cend to lower elevation valley bottoms during
winter (Poole and Stuart-Smith 2006). Thus,
differences in habitat selection patterns among
populations of moose are dependent on local
conditions with respect to forage, predators,
and weather.

Climate change is predicted to profound-
ly affect land mammals in the Arctic (Lawler
et al. 2009, Marcot et al. 2015). Wildfire is
already common in the region (Joly et al.
2009) and is predicted to increase under
warming scenarios (Kasischke and Turetsky
2006, Johnstone et al. 2010, Joly et al.
2012). Early seral stage shrub communities,
which follow wildfires, provide abundant
high quality forage for moose (Schwartz
and Franzmann 1989). Moreover, moose
populations have increased where these early
seral habitats have expanded due to wildfire
(Spencer and Hakala 1964, Schwartz and
Franzmann 1989). Increased shrub abun-
dance has been documented around the
Arctic and is thought to be linked to warm-
ing (Tape et al. 2006). Thus, climate change
may produce more moose habitat and more
moose in this region if patterns of selection
for early seral stage and shrubby habitats
by moose are similar to other areas of the
boreal forest (Joly et al. 2012). Understand-
ing current patterns of habitat selection will
aid in assessing the effects of climate change
into the future.

We analyzed habitat selection by moose
on the southern flanks of the Brooks Range
and the adjacent lowlands in north-central
Alaska (Fig. 1), near the northern extent of
moose range in this region. Our goal was to
provide information about habitat selection
patterns in Alaska’s arctic interior to improve
moose management. We focused on selection
within the home range across individuals dur-
ing winter and summer seasons using variables
we believed important to moose. We assessed
whether patterns of habitat selection were
driven primarily by spatial factors related
to abundance of adequate forage, predator
avoidance, or physiography. We hypothe-
sized that habitat selection would be driven
by a complex mix of factors, highlighting
the trade-offs among access to forage, energy
expenditure, and exposure to predation pres-
sure. Further, we hypothesized that maternal
status and winter severity would influence
patterns of habitat selection. We expected
females with calves to select more forested
areas further from rivers than females without
calves, presumably to reduce predation risk,
and that moose would select areas lower in
elevation during more severe winters.

METHODS
Study area

This study took place in the upper
reaches of the Koyukuk River in north-
central Alaska (Fig. 1). The area supported a
low density (~ 0.1 moose/km2) moose popu-
lation (Lawler et al. 2006), as well as the
full complement of naturally occurring spe-
cies including caribou (Rangifer tarandus),
Dall’s sheep (Ovis dalli), wolves (Canis
lupus), grizzly bears (Ursus arctos), and
black bears (U. americanus). The upper
Koyukuk River drainage had a strong contin-
ental climate with short, hot summers and
long, cold winters. Temperatures dropped
below – 45 °C and snow persisted on the
ground from October until May (Western Re-
gional Climate Center, www.wrcc.dri.edu/).

102

HABITAT SELECTION IN ALASKA – JOLY ET AL. ALCES VOL. 52, 2016

www.wrcc.dri.edu/


Snow pack was typically >60 cm most win-
ters and often >90 cm. Summers were brief
but temperatures can exceed 30 °C. Large
wildfires were common during warm dry

summers, particularly south of the Brooks
Range which consisted of boreal forest vege-
tation dominated by fire-prone communities
such as black spruce (Picea mariana) forests.

Fig. 1. Moose habitat selection and use study area (white polygon) in north-central Alaska, 2008–
2013. GPS locations (dots) of individual moose are color-coded.

ALCES VOL. 52, 2016 JOLY ET AL. – HABITAT SELECTION IN ALASKA

103



The northern half of the study area con-
sisted of the central Brooks Range - rugged
mountains that reach up to 2000 m in eleva-
tion that contain narrowly-confined glacial
river valleys, and where wildfire is much
less common. The valleys supported spruce
and birch (Betula papyrifera) forests, tus-
sock tundra, shrub lands (Alnus spp., Salix
spp.), and muskeg. Tall and low shrub commu-
nities occurred on hillsides, but eventually
gave way to alpine vegetation. This area
included the southeastern portion of Gates
of the Arctic National Park and Preserve
(GAAR) and lands managed by the Bureau
of Land Management (BLM) and the state
of Alaska.

The southern portion of the study area
was much less rugged and lower in eleva-
tion; typically about 300 m above sea level
with hills generally lower than 500 m. It
had more wetland habitat, was extensively
forested, and wildfires were prevalent. The
southern portion of the study area primarily
contained lands managed by the Kanuti
National Wildlife Refuge (KNWR), the
state of Alaska, and the BLM. The town of
Bettles, Alaska was in the middle of the
study area.

Moose capture, GPS data, maternal
status, winter severity

We captured adult male and female
moose between March 2008 and April 2011
via aerial darting. Moose were fitted with
GPS radio-collars (Telonics TGW-4780)
that also had a very high frequency (VHF)
radio beacon (Joly et al. 2015a); collars
were removed when the project ended in
April 2013. Collars collected 3 locations/
day except those deployed in March 2008.
For our analyses, all location datasets began
on 15 May, and we excluded all individual-
years that were sampled <330 days.

Maternal status, as indicated by the pres-
ence or absence of a calf, was determined by
tracking collared females in small, fixed-

wing aircraft (e.g., Piper PA-18 Supercub).
We attempted to locate all collared females
just after calving (late May–early June), in
the fall (September–October), and during the
following spring (March–April) to visually
determine if the female was accompanied
by a calf. If we could not make this deter-
mination, the individual was excluded from
analyses related to maternal status.

We classified each winter as mild, mod-
erate, or severe based on the total number
of days with snow and snow depth as
recorded in Bettles, Alaska (Joly et al.
2015a). The 3 classifications were: 1) mild
winters had <135 days with ≥30 cm snow
or <7 days with ≥60 cm snow, 2) moderate
winters had >170 days with ≥30 cm snow,
>50 days with ≥60 cm, or <14 days with
≥90 cm snow, and 3) severe winters had
>170 days with ≥30 cm snow, >100 days
with ≥60 cm, or >30 days with ≥90 cm
snow. We used these non-continuous cate-
gories to highlight that the classifications
were distinctive – all winters fell into a single
category. Two winters (2009–10, 2012–13)
were categorized as mild, 3 (2007–08,
2010–11, and 2011–12) as moderate, and 1
(2008–09) as severe. We defined biological
seasons as summer (1 July–24 August) and
winter (16 December–14 May) based on re-
gional weather patterns.

Study design
Scale is critical to understanding eco-

logical processes (Wiens 1989, Wheatley
and Johnson 2009, DeCesare et al. 2012).
Habitat preferences modify with changes in
the relative amount of available habitat
(Osko et al. 2004, Herfindal et al. 2009).
Due to physiographic differences between
the northern and southern portions of our
study area, we estimated seasonal habitat se-
lection by moose at the home range scale (3rd

order; Johnson 1980) using the synoptic
model of space use (Horne et al. 2008,
Slaght et al. 2013). This model uses a

104

HABITAT SELECTION IN ALASKA – JOLY ET AL. ALCES VOL. 52, 2016



weighted distribution to simultaneously
model an individual’s space use and habitat
selection (Johnson et al. 2008) within its
home range, and is capable of estimating
home range and resource selection simultan-
eously. Thus, the probability of use at loca-
tion x and time t was modeled using:

f u x; tð Þ ¼
f a xð Þ � wðx; tÞR
f a xð Þ � wðx; tÞ

ð1Þ

where f a xð Þ is the null distribution of space
use that models the probability of use in the
absence of habitat selection (i.e., the avail-
ability distribution), and wðx; tÞ is a selection
function that transforms f a xð Þ to f u x; tð Þ by
selectively weighting different areas based
on habitat conditions (Johnson et al. 2008).
We defined f a xð Þ ¼ BVN hð Þ to be a station-
ary (i.e., time invariant) bivariate normal
(BVN) distribution with parameters h de-
scribing the means and variances in the x
and y dimensions and the covariance. By de-
scribing f a xð Þ in this way, the areas consid-
ered available for selection can be thought
of as a BVN distribution characterizing the
entire home range of an individual. The
BVN distribution characterizes the space
use of an animal that biases movement to-
wards a central place (Horne et al. 2008,
Wilson et al. 2014b). We defined the selec-
tion function as:

w x; tð Þ ¼ Exp H xð Þ0bPðtÞ
h i

ð2Þ

where H xð Þ0 is a vector of covariate values
describing the habitat or environmental con-
ditions at location x, b is a vector of para-
meters (i.e., selection coefficients) to be
estimated, and P(t) is an interaction term
representing functions of time (i.e., winter,
summer) to allow for temporal variation in
habitat selection. Others have used similar
approach for modeling habitat selection
through time (see Ferguson et al. 2000,

Forester et al. 2009). We used maximum
likelihood (via numerical optimization) to
estimate the parameters governing the null
model of home range (θ) and the selection
coefficients (β) with a program written in R
(R Development Core Team 2013) with
code developed by J. Horne (see Slaght et al.
2013 for example code). We used odds ratios
to aid interpretation of the estimated coeffi-
cients βi. An odds ratio approximates the
relative change in probability of event x oc-
curring (e.g., a moose being present) given a
1-unit change in a given parameter (Hosmer
and Lemeshow 1989).

Environmental variables
Based on previous research, we formu-

lated 11 models to analyze seasonal habitat
selection by moose (Table 1). Weixelman
et al. (1998) and Maier et al. (2005) sug-
gested that moose select habitats that burned
11–30 years prior to usage because these
areas tend to revegetate with deciduous
shrubs. Riparian zones often have abundant
and high-quality forage that moose use in
Alaska (Collins and Helm 1997, Maier et al.
2005, Stephenson et al. 2006). Areas with
extensive vegetative cover typically have
more deciduous trees (e.g., birch) and tall
shrubs (e.g., willows) that are preferred
moose forage than areas with low cover
(e.g., tussock tundra). We expected moose
to select areas that contained preferred for-
age, such as forested and burned habitat,
and areas closer to rivers. The models that
highlighted the importance of forage con-
tained a mixture of covariates that included
‘Fire’ (if a moose was in habitat that burned
11–30 years prior to use), ‘Forest’ (if a
moose used areas with extensive vegeta-
tive cover based on landcover type), and/or
‘Dist_River’ (distance from a riparian area;
Table 1). We identified areas as ‘fire’ using
the Alaska Fire Service’s geodatabase which
catalogs the extent, number, and location of
large fires mapped from 1950–2014 (Fig. 2;

ALCES VOL. 52, 2016 JOLY ET AL. – HABITAT SELECTION IN ALASKA

105



data at http://fire.ak.blm.gov/predsvcs/maps.
php), and ‘forest’ using the National Land
Cover Database – Alaska 2001 coverage
(http://www.epa.gov/mrlc/nlcd-2001.html).
Major rivers were identified using the
USGS 1:2,000,000 digital line graphs dataset
(https://lta.cr.usgs.gov/DLGs).

We expected moose, particularly females
with calves, to select areas further from ri-
parian areas and that were more forested to
reduce predation pressure. Riparian areas
are often utilized by predators as travel
corridors and forested areas provide more
cover to hide from predators (Peterson
1995, Kunkel and Pletscher 1999, McPhee
et al. 2012). Thus, we interpreted moose
responses to riparian areas as a proxy for

responding to areas of increased predation
risk. The model highlighting the importance
of predation pressure included the covariates
of ‘Dist_River’ and ‘Forest’.

Moose select areas based on physiog-
raphy and 4 models were used to assess the
importance of terrain including a mixture of
the covariates slope, elevation and their
squared terms (to assess non-linear rela-
tions), and ‘SRI’ (a solar radiation index,
Keating et al. 2007; Table 1). These covari-
ates were derived from our digital elevation
model. Higher solar radiation is correlated
with reduced snow depth during winter and
increased net primary productivity (i.e., for-
age) during summer (Crabtree et al. 2009).
We expected moose to select for terrain fea-
tures that reduced snow depth, and subse-
quently increased forage availability, and
that these patterns would be more prominent
during more severe winters. We hypothe-
sized that habitat selection by moose is influ-
enced by a wide array of factors, rather than
just forage abundance, predation pressure, or
terrain acting alone. We used 4 models to as-
sess this hypothesis, and covariates for these
models included the entire suite used in the
previous models.

We used remotely sensed data to quan-
tify the spatial distribution of habitat covari-
ates and included interaction terms between
resource selection coefficients and functions
of time (Ferguson et al. 2000, Forester et al.
2009) to account for temporal variation in
habitat selection. Before modeling resource
selection, we screened predictor variables for
collinearity. We assumed that if │r│< 0.60,
then correlation was not a concern between
predictor covariates (Sawyer et al. 2006,
Ciarniello et al. 2007). Slope and elevation
were considered positively correlated and
were not included together in any model.

Model selection
We used an information-theoretic ap-

proach for evaluating synoptic models of

Table 1. Models and their structure used to analyze
different hypotheses related to moose habitat
selection in north-central Alaska, USA, 2008–
2013.

Model Covariates

Fire Firea

Forage Fire+Dist_Riverb+Forestc

Predator Dist_River+Forest

Terrain1 SRId+Eleve

Terrain2 SRI+Slope

Terrain3 SRI+Elev+Elev2

Terrain4 SRI+Slope+Slope2

Complexity1 Fire+Forest+Dist_River+SRI+Elev

Complexity2 Fire+Forest+Dist_River+SRI
+Slope

Complexity3 Fire+Forest+Dist_River+SRI+Elev
+Elev2

Complexity4 Fire+Forest+Dist_River+SRI
+Slope+Slope2

a ‘Fire’ denoted if a moose location was in habitat that
burned 11-30 years prior to use
b ‘Dist_River’ is distance to a riparian area a moose
was located
c ‘Forest’ denoted if a moose location was in habitat
that was extensively vegetated (i.e. forest or tall
shrubs)
d ‘SRI’ is a solar radiation index
e ‘Elev’ is elevation

106

HABITAT SELECTION IN ALASKA – JOLY ET AL. ALCES VOL. 52, 2016

http://fire.ak.blm.gov/predsvcs/maps.php
http://fire.ak.blm.gov/predsvcs/maps.php
http://www.epa.gov/mrlc/nlcd-2001.html
https://lta.cr.usgs.gov/DLGs


habitat selection and determined a set of a
priori candidate models that we deemed bio-
logically relevant (Burnham and Anderson
2002). We fit models to location data for
each individual and year. We ranked the
models for each moose and year using the
difference in Akaike Information Criterion
adjusted for small sample size (AICc) from
the model with the smallest value (ΔAICc),
and determined the relative likelihood of
each model using Akaike weights (Burnham
and Anderson 2002). Models, including the
top model, which had an AICc score of <2
from the top model were designated as being
in the top model set.

We evaluated habitat selection by sex,
winter severity, and maternal status. We
averaged estimates of selection coefficients
across models based on Akaike weights for
each individual and year. We scaled the
weights to total 1 across models containing
each variable (Burnham and Anderson
2002). For individuals that we observed dur-
ing multiple years, we averaged the value of
estimated-coefficients across years. To make
class-level (i.e., sex, maternal status, and se-
verity of the winter) inferences, we calcu-
lated the means and standard errors of
univariate parameter estimates across all
individuals for each parameter. If a param-
eter (e.g., fire) was not used by an individual,
then no estimate was included for that in-
dividual for class-level inferences. For a
conservative measure of precision at the
class-level, we considered a coefficient to
be significant if 2 times the standard error
of the mean did not contain zero (Boyce
2006, Fieberg et al. 2010).

Complete separation of the data occurred
where habitats were available but not used
by a moose. For these individuals, we did
not estimate a coefficient for the variable
but simply noted avoidance (e.g., Nielsen
et al. 2002). We entered elevation and slope
as quadratic terms to allow for selection, or

avoidance, at intermediate values of eleva-
tion and slope.

RESULTS
We retrieved 71,675 GPS locations

from 37 moose between March 2008 and
April 2013 via remote download and
collar retrieval; 6 moose did not provide
enough data to be included in our analyses.
The remaining 31 moose (20 females and
11 males) produced 70 moose-years of data
(range: 1–4 years per individual).

For male moose, the Complexity3 model
(which included the covariates Fire, Forest,
Dist_River, SRI, Elev, and Elev2) best
described habitat selection within home
ranges during both winter and summer
(Tables 1 and 2). Complexity3 was in the
top model set for 48% and 46% of the indi-
vidual moose-years during the winter and
summer, respectively. Complexity1 was in
the top model set for 29% and 38% of the in-
dividual moose-years during the winter and
summer, respectively. Both Complexity4
and Terrain3 were in the top model set for
10% of the individual moose-years during
the winter, and Complexity4 in the top
model set for 8% of the individual moose-
years during the summer. The remaining
models were in the top model set for ≤10%
of the individual moose-years during either
season (Table 2).

For female moose, the Complexity3
model best described habitat selection within
home ranges during both winter and summer
(Tables 1 and 3). Complexity3 was in the top
model set for 49% and 41% of the individual
moose-years during the winter and summer,
respectively. Complexity2 was in the top
model set for 14% and 24% of the indivi-
dual moose-years during the winter and sum-
mer, respectively. Terrain3 was in the top
model set for 16% of the individual moose-
years during the winter, and Complexity4
in the top model set for 16% of the indi-
vidual moose-years during the summer.

ALCES VOL. 52, 2016 JOLY ET AL. – HABITAT SELECTION IN ALASKA

107



The remaining models were in the top model
set for ≤10% of the individual moose-years
during either season (Table 3).

Nearly half (11 of 23) of the moose in
the northern portion of the study area, where
wildfire is less common than in the southern
portion, did not use burned habitat during
either winter or summer. All 8 moose in the
southern portion of the study area used
burned habitat, with 1 animal located only
within burned habitat.

Seasonal selection patterns by moose
Patterns of selection by moose varied

between season and sex (Table 4). Male
moose consistently selected areas that were
forested, lower in elevation, and with gentler
slopes in winter; during summer they selected
areas that were forested. During both seasons
males were more variable in their selection
of areas that received higher amounts of
solar radiation, that were closer to riparian

habitat, or that had been burned. Across
seasons, female moose consistently selected
areas that were forested, burned, and lower
in elevation. Further, during winter females
selected areas that received higher amounts
of solar radiation, and during summer they
avoided steeper slopes. Distance to riparian
habitat was not consistently selected or
avoided by females during either season.

Winter severity
The severity of the winter influenced

habitat selection. As expected, moose selected
areas lower in elevation with gentler slopes
during more severe winters suggesting that
snow depth influenced habitat selection.
Based on average probability ratios, moose
were 56% less likely to select a location for
every 100 m higher during severe winters,
but only 6% less likely during mild winters.
In addition, moose selected areas closer to riv-
ers during more severe winters. During mild

Table 2. Top models of habitat selection by male moose in north-central, Alaska, USA, 2008–2013. The
number of individual-years of data (n) for which each of the top 3 models of habitat selection received
the most support, average and range of Akaike weights, and percent of times (%) each model occurred in
the top model set (<2 AICc of the top model) are presented by season.

Winter Summer

Model n Akaike weight % n Akaike weight %

Complexity3 10 0.95 (0.62–1.00) 48 11 0.85 (0.22–1.00) 46
Complexity1 6 0.92 (0.55–1.00) 29 9 0.77 (0.30–1.00) 38
Complexity4 2 0.69 (0.38–1.00) 10 2 0.47 (0.24–0.70) 8
Terrain3 2 1.00 (1.00–1.00) 10

Table 3. Top models of habitat selection by female moose in north-central, Alaska, USA, 2008-2013. The
number of individual-years of data (n) for which each of the top 3 models of habitat selection received the
most support, average and range of Akaike weights, and percent of times (%) each model occurred in the
top model set (<2 AICc of the top model) are presented by season.

Winter Summer

Model n Akaike weight % n Akaike weight %

Complexity3 31 0.88 (0.32–1.00) 49 26 0.79 (0.25–1.00) 41
Complexity1 9 0.73 (0.27–1.00) 14 15 0.77 (0.27–1.00) 24
Complexity4 10 0.66 (0.17–0.98) 16
Terrain3 10 0.79 (0.23–1.00) 16

108

HABITAT SELECTION IN ALASKA – JOLY ET AL. ALCES VOL. 52, 2016



winters moose were more variable in their se-
lection of most land-cover classes and land-
scape features (Table 4).

Maternal status
Six females successfully raised at least 1

calf through to the following spring, 11 lost
calves by fall, and 11 either did not give birth
or lost their calves during the first month
post-birth. We were unable to determine the
maternal status of 9 females. During both
seasons, females with calves selected areas
further from rivers, more forested, and with
less burned habitat than females without
calves (Table 5). For example, based on
average probability ratios, females with
calves were 20% more likely to select a site
1000 m further from a river, whereas females
without calves were 13% less likely to be

found there. Females with calves were 70%
more likely to be in forested habitat, whereas
females without calves were only 40% more
likely to be there.

DISCUSSION
Similar to studies in other northern

regions, we found that moose in north-central
Alaska selected for habitats with extensive
canopy cover. Where habitat that burned
11–30 years previous was available, moose,
particularly females, selectively used it (pre-
sumably) because habitats at this seral stage
tend to have abundant forage (MacCracken
and Viereck 1990, Weixelman et al. 1998,
Maier et al. 2005). This appears to support
the hypothesis that moose habitat selection is
primarily driven by availability of forage abun-
dance and quality (Peek 1997). However,

Table 4. Average parameter estimates (β) used to characterize selection by moose in north-central, Alaska,
USA, 2008–2013. Bold values were significant at the class level (i.e., sex and population). Values in
parentheses represent n for each class. ‘F’ denotes female and ‘M’ male.

Winter Summer Winter

Male
(11)

Female
(20)

Male
(11)

Female
(20)

Mild
(19 F, 8 M)

Mod/Severe
(18 F,11M)

Firea,x �1.12 1.37 0.02 3.57 1.15 0.10

SRIb �1.98 0.70 2.26 0.60 0.19 �0.64

Non-linear Elevc

Elev 5.08 35.89 65.27 52.71 50.68 3.23

Elev2 �30.66 �86.05 �103.20 �114.05 �107.74 �33.61

Non-linear Slope

Slope 2.87 2.58 12.53 2.52 5.12 0.62

Slope2 �11.24 �5.74 �18.94 �13.66 �8.72 �7.09

Dist_Riverd �3.08 �1.18 �2.17 0.50 �0.25 �2.29

Foreste 0.34 0.39 0.76 0.36 0.46 0.33

Elev �22.38 �4.63 �2.47 �6.94 �1.21 �17.19

Slope �3.67 �0.89 0.41 �2.85 �0.32 �2.73

a ‘Fire’ denoted if a moose location was in habitat that burned 11–30 years prior to use
b ‘SRI’ is a solar radiation index
c ‘Elev’ is elevation
d ‘Dist_River’ is distance to a river a moose was located
e ‘Forest’ denoted if a moose location was in habitat that was extensively vegetated (i.e. forest or tall shrubs)
x Not all moose utilized recently burned habitat so sample sizes were: winter male, n=4; winter female, n=12;
summer male, n=7; summer female, n=10

ALCES VOL. 52, 2016 JOLY ET AL. – HABITAT SELECTION IN ALASKA

109



this hypothesis was not supported by our top
models that included the greatest number of
variables. The majority (>66%) for both
males and females included indices of forage
abundance (time since last fire), extensive
vegetative cover (forest), distance to river, in
addition to elevation and solar radiation. While
many of these covariates can be associated
with forage abundance, our results suggest
that a wide array of factors likely influence
habitat selection by moose – supporting the
hypothesis that habitat selection by moose is
driven by a complex interaction of diverse
factors.

We found that patterns of habitat selec-
tion differed between sex and season. Male
and female moose exhibited similar patterns
of selection for terrain features, particularly
elevation during summer. During winter,
however, sex-related differences were evident.

As expected, males moved to lower elevations,
but unexpectedly, females remained at similar
elevations throughout the winter. This behav-
ioral difference might provide smaller-bodied
females some benefit of higher quality forage
(i.e., in burned areas) and terrain features
that reduced snow pack (i.e., higher SRI).
However, predation on moose is often greater
in lowland areas (Fuller and Keith 1980).

Within the region, wolves often travel
along riparian corridors (Lake et al. 2013).
We suspect that predators focus their hunting
along riparian corridors, owing to the con-
centration of prey in areas of lower snow
depth and that travel is probably easier for
predators due to smooth, hard surfaces
afforded by rivers (Peterson 1995, Kunkel
and Pletscher 1999, McPhee et al. 2012).
Moose reduce their vulnerability to wolf pre-
dation by avoiding areas used by wolves for

Table 5. Average parameter estimates (β) used to characterize selection by maternal status of female moose
in north-central, Alaska, USA, 2008–2013. Bold values were significant. Values in parentheses represent
n for each class.

Winter Summer

Calf (6) No Calf (22) Calf (17) No Calf (11)

Fireax �0.30 0.73 4.16 7.23

SRIb 0.14 0.39 0.41 0.58

Non-linear Elevc

Elev 43.02 39.93 31.59 47.95

Elev2 �84.31 �79.63 �95.45 �129.86

Non-linear Slope

Slope 2.47 2.31 0.63 4.93

Slope2 �1.79 �5.42 �10.49 �17.71

Dist_Riverd 1.62 �1.26 1.90 �0.59

Foreste 0.51 0.34 0.38 0.24

Elev �3.26 �0.60 �6.80 �5.79

Slope 0.27 �0.52 �3.24 �1.13

a ‘Fire’ denoted if a moose location was in habitat that burned 11–30 years prior to use
b ‘SRI’ is a solar radiation index
c ‘Elev’ is elevation
d ‘Dist_River’ is distance to a river a moose was located
e ‘Forest’ denoted if a moose location was in habitat that was extensively vegetated (i.e. forest or tall shrubs)
x Not all moose utilized recently burned habitat so sample sizes were: winter calf, n=1; winter no calf, n=10;
summer calf, n=2; summer no calf, n=7

110

HABITAT SELECTION IN ALASKA – JOLY ET AL. ALCES VOL. 52, 2016



travel (Kunkel and Pletscher 1999). Our
findings are consistent with the hypothesis
that females try to minimize predation risk,
whereas males adopt a strategy to maximize
forage intake (Fuller and Keith 1980, Oehlers
et al. 2011).

Moose, with their large size and for-
midable strength, are well-adapted to snow
(Telfer and Kelsall 1979, Peek 1997).
Nevertheless, deep (65–70 cm) snow can
affect moose movement, distribution, and
home range size (van Ballenberghe 1977,
Miquelle et al. 1992, Peek 1997, Ball et al.
2001, Joly et al. 2015b). As expected, during
severe winters moose selected habitats that
were at lower elevations, with gentler slopes,
and closer to rivers than during mild winters.
Deep snow at mid- to high elevations, or in
early successional stages of burns, can cover
preferred browse inducing moose to move to
lower elevations and use areas where forage
is more concentrated within riparian areas
(Weixelman et al. 1998). Relatively higher
moose densities in valley bottoms during se-
vere winters may attract predators such as
wolves, and thus increase localized predation
risk (McPhee et al. 2012, Lake et al. 2013).
Thus, harsh winters may have indirect, as
well as direct, negative impacts on moose
in our study area that may reduce their prod-
uctivity and survivorship. We found that
moose were more variable in their selection
of land-cover classes and landscape features
during mild winters, suggesting that tenden-
cies for moose to select lower elevational
areas closer to rivers during winter may be
more related to snow depth, and subsequently,
forage availability not forage quality.

Interestingly, we did not find that moose
utilized habitats with higher canopy cover
during severe winters which may reflect
that many trees are diminutive in our high-
latitude study area. Even though forest
stands can have relatively high canopy
cover, there may be insufficient overhead

foliage to intercept snow and reduce under-
lying depths, or provide thermal cover.

We found that maternal status influenced
patterns of habitat selection by females. As
expected, females with calves avoided ripar-
ian habitats and selected areas with more
forested habitat than females without calves
during both summer and winter. In addition,
females with calves selected areas with less
burned habitat than females without calves.
Both riparian habitat and burned areas tend
to provide more high quality moose forage
than other habitat types (Collins and Helm
1997, Maier et al. 2005, Stephenson et al.
2006). These results suggest that maternal
status-related differences in habitat selection
patterns were likely more related to the spe-
cific needs of females with regard to protec-
tion of calves. However, due to the small
sample size of females with calves (n = 6),
our results should be considered preliminary.

Habitat selection is fundamental to the
ecology of wildlife species. Understanding
patterns of habitat selection by moose can
improve their management. An obvious ex-
ample to use this information is to help guide
where and when development occurs to min-
imize loss of critical moose habitat. Our
work is timely, given that a proposed in-
dustrial road would bisect the study area
(Wilson et al. 2014a, Guettabi et al. 2016).
Further, enhanced knowledge of moose
movements, distribution, and habitat selec-
tion should be useful to abate conflicts be-
tween subsistence and non-subsistence
hunters by spatially or temporally separating
users near high quality moose habitat.

The Arctic is undergoing rapid warming
which will result in measurable ecological
changes (Hinzman et al. 2005, IPCC 2007).
Further, wildfire is predicted to increase in
the region, potentially creating more pro-
ductive foraging habitats for moose (Joly
et al. 2012). By collecting baseline data on
habitat selection and use, future researchers
will be better able to assess the impacts of

ALCES VOL. 52, 2016 JOLY ET AL. – HABITAT SELECTION IN ALASKA

111



climate change on moose at their northern
extent of range in North America.

ACKNOWLEDGEMENTS
Funding for this project was provided

by the National Park Service, US Fish
and Wildlife Service, BLM, and the Alaska
Department of Fish and Game. We thank
pilots T. Cambier, M. Spindler, M. Webb,
P. Zaczkowski, C. Cebulski, P. Christian,
N. Guldager, H. Bartlett, A. Greenblatt, L.
Dillard, D. Sowards, and S. Hamilton for
efficient and safe flying. J. Burch, T. Hollis,
J. Lawler, N. Pamperin, C. Roberts, and G.
Stout provided expert assistance with cap-
tures. C. Harwood, S. Miller, and R. Sarwas
provided database and GIS expertise. T.
Hollis and other biologists from the Alaska
Department of Fish and Game conducted
the radiotracking flights during calving.
G. Stout provided critical assistance with
project management. All moose captures
adhered to State of Alaska Animal Care
and Use Committee (ACUC) guidelines
(#07-11). We thank anonymous reviewers
and N. DeCesare for constructive reviews
that improved our manuscript.

LITERATURE CITED

BALL, J. P., C. NORDENGREN, and K.
WALLIN. 2001. Partial migration by large
ungulates: characteristics of seasonal
moose Alces alces ranges in northern
Sweden. Wildlife Biology 7: 39–47.

BARBOZA, P. S., and R. T. BOWYER. 2000.
Sexual segregation in dimorphic deer: a
new gastrocentric hypothesis. Journal of
Mammalogy 81: 473–489.

BOWYER, R. T., B. M. PIERCE, L. K. DUFFY,
and D. A. HAGGSTROM. 2001. Sexual
segregation in moose: effects of habitat
manipulation. Alces 37: 109–122.

BOYCE, M. S. 2006. Scale for resource selec-
tion functions. Diversity and Distribu-
tions 12: 269–276.

BURNHAM, K. P., and D. R. ANDERSON.
2002. Model Selection and Multimodel
Inference: A Practical Information-
Theoretic Approach. Springer-Verlag,
New York, New York, USA.

CIARNIELLO, L. M., M. S. BOYCE, D. R. SEIP,
and D. C. HEARD. 2007. Grizzly bear
habitat selection is scale dependent. Eco-
logical Applications 17: 1424–1440.

COLLINS, W. B., and D. J. HELM. 1997.
Moose, Alces alces, habitat relative to
riparian succession in the boreal forest,
Susitna River, Alaska. Canadian Field-
Naturalist 111: 567–74.

CRABTREE, R., C. POTTER, R. MULLEN, J.
SHELDON, S. HUANG, J. HARMSEN, A.
RODMAN, and C. JEAN. 2009. A model-
ing and spatio-temporal analysis frame-
work for monitoring environmental
change using NPP as an ecosystem indi-
cator. Remote Sensing of Environment
113: 1486–1496.

DECESARE, N. J., M. HEBBLEWHITE, F. K. A.
SCHMIEGELOW, D. HERVIEUX, G. MCDER-
MID, L. NEUFELD, M. BRADLEY, J. WHIT-
TINGTON, K. SMITH, L. E. MORGANTINI,
M. WHEATLEY, and M. MUSIANI. 2012.
Transcending scale-dependence in iden-
tifying habitat with resource selection
functions. Ecological Applications 22:
1068–1083.

DUSSAULT, C., J.-P. OUELLET, R. COURTOIS,
J. HUOT, L. BRETON, H. JOLICOEUR, and
D. KELT. 2005. Linking moose habitat
selection to limiting factors. Ecography
28: 619–628.

FERGUSON, S. H., M. K. TAYLOR, and F. MES-
SIER. 2000. Influence of sea ice dynamics
on habitat selection by polar bears. Ecol-
ogy 81: 761–772.

FIEBERG, J., J. MATTHIOPOULOS, M. HEBBLE-
WHITE, M. S. BOYCE, and J. L. FRAIR.
2010. Correlation and studies of habitat
selection: problem, red herring or oppor-
tunity? Philosophical Transactions of
the Royal Society London B Biological
Sciences 365: 2233–2244.

112

HABITAT SELECTION IN ALASKA – JOLY ET AL. ALCES VOL. 52, 2016



FORESTER, J. D., H. K. IM, and P. J.
RATHOUZ. 2009. Accounting for animal
movement in estimation of resource
selection functions: sampling and data
analysis. Ecology 90: 3554–3565.

FULLER, T. K., and L. B. KEITH. 1980. Wolf
population dynamics and prey relation-
ships in northeastern Alberta. Journal of
Wildlife Management 44: 583–602.

GAILLARD, J.-M., M. HEBBLEWHITE, A.
LOISON, M. FULLER, R. POWELL, M.
BASILLE, and B. VAN MOORTER. 2010.
Habitat–performance relationships: find-
ing the right metric at a given spatial
scale. Philosophical Transactions of the
Royal Society of London B: Biological
Sciences 365: 2255–2265.

GUETTABI, M., J. GREENBERG, J. LITTLE,
and K. JOLY. 2016. Evaluating potential
economic effects of an industrial road
on subsistence in north-central Alaska.
Arctic 69: 305–317.

HERFINDAL, I., J.-P. TREMBLAY, B. B.
HANSEN, E. J. SOLBERG, M. HEIM, and
B.-E. SÆTHER. 2009. Scale dependency
and functional response in moose habitat
selection. Ecography 32: 849–859.

HINZMAN, L. D., ET AL. 2005. Evidence and
implications of recent climate change in
northern Alaska and other arctic regions.
Climatic Change 72: 251–298.

HORNE, J. S., E. O. GARTON, and J. L.
RACHLOW. 2008. A synoptic model of
animal space use: Simultaneous estima-
tion of home range, habitat selection,
and inter/intra-specific relationships.
Ecological Modeling 214: 338–348.

HOSMER, D. W., and S. LEMESHOW. 1989.
Applied Logistic Regression. Wiley Series
in Probability and Statistics. John Wiley
& Sons, Inc., New York, New York, USA.

IPCC (Intergovernmental Panel on Climate
Change). 2007. Climate Change 2007:
Synthesis Report. Contribution of Work-
ing Groups I, II and III to the Fourth As-
sessment Report of the Intergovernmental
Panel on Climate Change. Core Writing

Team, R. K. Pachauri, and A. Reisinger,
editors. IPCC, Geneva, Switzerland.

JOHNSON, D. H. 1980. The comparison of
usage and availability measurements for
evaluating resource preference. Ecology
61: 65–71.

JOHNSON, D. S., D. L. THOMAS, J. M. vER
HOEF, and A. CHRIST. 2008. A general
framework for the analysis of animal re-
source selection from telemetry data.
Biometrics 64: 968–976.

JOHNSTONE, J. F., F. S. CHAPIN, III, T. N.
HOLLINGSWORTH, M. C. MACK, V.
ROMANOVSKY, and M. TURETSKY.
2010. Fire, climate change, and forest
resilience in interior Alaska. Canadian
Journal of Forest Research 40:
1302–1312.

JOLY, K., T. CRAIG, M. S. SORUM, J. S.
MCMILLAN, and M. A. SPINDLER.
2015a. Moose (Alces alces) movement
patterns in the upper Koyukuk river
drainage, northcentral Alaska. Alces 51:
87–96.

———, ———, ———, ———, ———.
2015b. Variation in fine-scale move-
ments of moose (Alces alces) in the upper
Koyukuk river drainage, northcentral
Alaska. Alces 51: 97–105.

———, P. A. DUFFY, and T. S. RUPP. 2012.
Simulating the effects of climate change
on fire regimes in Arctic biomes: impli-
cations for caribou and moose habitat.
Ecosphere 3: 1–18.

———, T. S. RUPP, R. R. JANDT, and F. S.
CHAPIN. 2009. Fire in the range of the
Western Arctic Caribou Herd. Alaska
Park Science 8: 68–73.

KASISCHKE, E. S., and M. R. TURETSKY.
2006. Recent changes in the fire regime
across the North American boreal region
– spatial and temporal patterns of burning
across Canada and Alaska. Geophysical
Research Letters 33: 1–5.

KEATING, K. A., P. J. P. GOGAN, J. M. VORE,
and L. R. IRBY. 2007. A simple solar radi-
ation index for wildlife habitat studies.

ALCES VOL. 52, 2016 JOLY ET AL. – HABITAT SELECTION IN ALASKA

113



Journal of Wildlife Management 71:
1344–1348.

KUNKEL, K., and D. H. PLETSCHER. 1999.
Species-specific population dynamics of
cervids in a multipredator ecosystem.
Journal of Wildlife Management 63:
1082–1093.

LAKE, B. C., M. R. BERTRAM, N. GULDAGER,
J. R. CAIKOSKI, and R. O. STEPHENSON.
2013. Wolf kill rates across winter in a
low‐density moose system in Alaska.
Journal of Wildlife Management 77:
1512–1522.

LAWLER, J. J., S. L. SHAFER, D. WHITE, P.
KAREIVA, E. MAURER, A. R. BLAUSTEIN,
and P. J. BARTLEIN. 2009. Projected cli-
mate-induced faunal change in the west-
ern hemisphere. Ecology 90: 588–597.

———, L. SAPERSTEIN, T. CRAIG, and G.
STOUT. 2006. Aerial Moose Survey in
Upper Game Management Unit 24, Al-
aska, Fall 2004, Including State Land,
and Lands Administered by the Bureau
of Land Management, Gates of the Arctic
National Park and Preserve, and Kanuti
National Wildlife Refuge. National Park
Service Technical Report NPS/AR/NR/
TR-2006-55. Fort Collins, Color-
ado, USA.

MACCRACKEN, J. G., and L. A. VIERECK.
1990. Browse regrowth and use by
moose after fire in interior Alaska. North-
west Science 64: 11–18.

MAIER, J. A. K., J. M. vER HOEF, A. D.
MCGUIRE, R. T. BOWYER, L. SAPER-
STEIN, and H. A. MAIER. 2005. Distribu-
tion and density of moose in relation to
landscape characteristics: effects of scale.
Canadian Journal of Forest Research 35:
2233–2243.

MARCOT, B. G., M. T. JORGENSON, J. P. LAW-
LER, C. M. HANDEL, and A. R.
DEGANGE. 2015. Projected changes in
wildlife habitats in Arctic natural areas
of northwest Alaska. Climatic Change
130: 145–154.

MCPHEE, H. M., N. F. WEBB, and E. H. MER-
RILL. 2012. Hierarchical predation: wolf

(Canis lupus) selection along hunt paths
and at kill sites. Canadian Journal of Zo-
ology 90: 555–563.

MIQUELLE, D. G., J. M. PEEK, and V. vAN
BALLENBERGHE. 1992. Sexual segrega-
tion in Alaskan moose. Wildlife Mono-
graph 122: 1–57.

NIELSEN, S. E., M. S. BOYCE, G. B. STEN-
HOUSE, and R. H. M. MUNRO. 2002.
Modeling grizzly bear habitats in the
Yellowhead ecosystem of Alberta: taking
autocorrelation seriously. Ursus 13:
45–56.

OEHLERS, S. A., R. T. BOWYER, F. HUETT-
MANN, D. K. PERSON, and W. B. KESS-
LER. 2011. Sex and scale: implications
for habitat selection by Alaskan moose
Alces alces gigas. Wildlife Biology 17:
67–84.

OSKO, T. J., M. N. HILTZ, R. J. HUDSON, and
S. M. WASEL. 2004. Moose habitat pre-
ferences in response to changing avail-
ability. Journal of Wildlife Management
68: 576–584.

PEEK, J. M. 1997. Habitat relationships.
Pages 351-375 in A. W. Franzmann and
C. C. Schwartz, editors. Ecology and
Management of the North American
Moose. Wildlife Management Institute,
Washington, D. C., USA.

PETERSON, R. O. 1995. The Wolves of Isle
Royale: A Broken Balance. Willow
Creek Press, Minocqua, Wisconsin,
USA.

POOLE, K. G., R. SERROUYA, and K. STUART-
SMITH. 2007. Moose calving strategies in
interior montane ecosystems. Journal of
Mammalogy 88: 139–150.

———, and K. STUART-SMITH. 2006. Winter
habitat selection by female moose in
western interior montane forests. Canadian
Journal of Zoology 84: 1823–1832.

R CORE TEAM. 2013. R: A Language and En-
vironment for Statistical Computing. R
Foundation for Statistical Computing,
Vienna, Austria.

SAWYER, H., R. M. NIELSON, F. LINDZEY, and
L. L. MCDONALD. 2006. Winter habitat

114

HABITAT SELECTION IN ALASKA – JOLY ET AL. ALCES VOL. 52, 2016



selection of mule deer before and during
development of a natural gas field.
Journal of Wildlife Management 70:
396–403.

SCHWARTZ, C. C., and A. W. FRANZMANN.
1989. Bears, wolves, moose, and forest
succession: some management consid-
erations on the Kenai Peninsula, Alaska.
Alces 25: 1–10.

SLAGHT, J. C., J. S. HORNE, S. G. SURMACH,
and R. J. GUTIÉRREZ. 2013. Home
range and resource selection by animals
constrained by linear habitat features: an
example of Blakiston's fish owl. Journal
of Applied Ecology 50: 1350–1357.

SPENCER, D. L., and J. B. HAKALA. 1964.
Moose and fire on the Kenai. Proceeding
of the Tall Timbers Fire Ecology Confer-
ence 3: 11–33.

STEPHENSON, T. R., V. VAN BALLENBERGHE,
J. M. PEEK, and J. G. MACCRACKEN.
2006. Spatio-temporal constraints on
moose habitat and carrying capacity in
coastal Alaska: vegetation succession
and climate. Rangeland Ecology and
Management 59: 359–372.

TAPE, K. D., M. STURM, and C. RACINE.
2006. The evidence for shrub expan-
sion in northern Alaska and the pan-
Arctic. Global Change Biology 12:
686–702.

TELFER, E. S., and J. P. KELSALL. 1979. Studies
of morphological parameters affecting
ungulate locomotion in snow. Canadian
Journal of Zoology 57: 2153–2159.

vAN BALLENBERGHE, V. 1977. Migratory be-
havior of moose in southcentral Alaska.
Transactions of the 13th International
Congress of Game Biologists 13:
103–109.

WEIXELMAN, D. A., R. T. BOWYER, and V.
vAN BALLENBERGHE. 1998. Diet selec-
tion by Alaskan moose during winter:
effects of fire and forest succession.
Alces 34: 213–238.

WHEATLEY, M., and C. JOHNSON. 2009.
Factors limiting our understanding of
ecological scale. Ecological Complexity
6: 150–159.

WIENS, J. A. 1989. Spatial scaling in ecology.
Functional Ecology 3: 385–397.

WILSON, R. R., D. D. GUSTINE, and K. JOLY.
2014a. Evaluating potential effects of an
industrial road on winter habitat of cari-
bou in north-central Alaska. Arctic 67:
472–482.

———, J. S. HORNE, K. D. RODE, E. V.
REGEHR, and G. M. DURNER. 2014b.
Identifying polar bear resource selection
patterns to inform offshore development
in a dynamic and changing Arctic. Eco-
sphere 5 (10): 136.

ALCES VOL. 52, 2016 JOLY ET AL. – HABITAT SELECTION IN ALASKA

115


	THE EFFECTS OF SEX, TERRAIN, WILDFIRE, WINTER SEVERITY, AND MATERNAL STATUS ON HABITAT SELECTION BY MOOSE IN NORTH-ENTRAL ALASKA
	Methods
	Study area
	Moose capture, GPS data, maternal status, winter severity
	Study design
	Environmental variables
	Model selection

	Results
	Seasonal selection patterns by moose
	Winter severity
	Maternal status

	Discussion
	Acknowledgements
	Literature CITED