Starr_H2013.pdf

Media

Part of Analyzing the Influence of Interstate 90 on Elk Home Range Establishment and Resource Selection

extracted text
Analyzing the Influence of Interstate 90 on
Elk Home Range Establishment and Resource Selection

by
Hailey Starr

A Thesis
Submitted in partial fulfillment
of the requirements for the degree
Master of Environmental Studies
The Evergreen State College
June 2013

I

©2013 by Hailey Starr. All rights reserved.

This Thesis for the Master of Environmental Studies Degree
by
Hailey Starr

has been approved for
The Evergreen State College
by
________________________
Dina Roberts, Ph.D.
Member of the Faculty

________________________
Kelly McAllister, B.S.
Wildlife Biologist,
Washington State Department of Transportation

________________________
Date

ABSTRACT
Analyzing the Influence of Interstate 90 on
Elk Home Range Establishment and Resource Selection

Hailey Starr

Negative effects from roads are evident throughout many natural systems.
Habitat fragmentation is among the most severe of these effects, with some wildlife
species experiencing consequences on population viability. High volume interstates are
among the most detrimental to wildlife. Interstate-90 (I-90) transects the North Bend
area, as the primary East-West traffic corridor in Washington State, resulting in
significant habitat fragmentation effects and a high number of elk-vehicle collisions. A
partnership between The Upper Snoqualmie Valley Elk Management Group (USVEMG)
and the Washington State Department of Transportation (WSDOT) was formed to study
elk movement to inform management decisions of how to reduce elk-vehicle collisions
while ensuring connectivity across I-90 for elk. I used locations from 10 Global
Positioning System (GPS) collared female elk during the years 2010-2012. Home range
and resource selection analyses were executed using the Brownian Bridges Movement
Model and second order resource selection analysis to understand how elk are influenced
by I-90. A majority of elk home ranges were located bordering I-90 with slight overlap,
only two home ranges largely overlapped I-90 and only one individual had core use areas
located on both sides. This suggests that some individuals approached I-90, but that few
crossed and spent abundant time on the opposite side of the interstate. I addition, elk were
found to avoid medium (35-45 mph) and high intensity (>55mph) roads. When space use
was evaluated at different distances from I-90, elk were found to avoid areas at distances
less than 50 meters. Therefore, it would appear that for many elk in this population, I-90
is a partial barrier to their movement. Identifying areas of potential connectivity across
this partial barrier where bridges exist in accordance with riparian habitat can inform
areas where connectivity exists and should be managed for if barrier fencing is
implemented in order to prevent collisions.

Table of Contents
Chapter 1: Literature Review
Introduction: Roads and Wildlife…………………..……..……….………1
Landscape Level Road Planning for Increased Connectivity……………..7
Crossing Structures………………………………………………..9
Fencing and Signs………………………………………………..10
Evaluation of Effectiveness…………………………………...…10
Wildlife Movement, Advancement in Technology and Methodology –
A novel approach to understanding landscape level effects……………..11
Conclusions…...………………………………………………………….14
Benefits of Multiagency Collaboration and Future Research……………15
Chapter 2: Analysis of Elk Home Ranges and Resource Selection……………...18
Introduction – Road Impacts on Wildlife………………………………..18
Materials and Methods …………………………………………………..20
Study Area……………………………………………………….20
Elk Capture and Telemetry………………………………………22
Home Ranges and Utilization Distributions – BBMM………..…23
Estimation Resource Selection………………………………......24
Statistical Analysis/Resource Selection………………………….26
Results……………………………………………………………………27
Home Ranges…………………………………………………….27
Core Use Areas…………………………………………………..28
Resource Selection……………………………………………….28
Discussion………………………………………………………………..30
Home Ranges…………………………………………………….30
iv

Core Use Areas…………………………………………………..33
Resource Selection……………………………………………….33
Conclusions………………………………………………………………37
Chapter 3: Conclusions and Management Implications…………………………41
Continued Research and Recommendations……………………………..42
Interdisciplinary Effort…………………………………………...………44
References………………………………………………………………………..64

iv

Appendices
A.

Example of the Brownian Bridge Movement Model R Script for elk
337…………………………………………………………………..………......75

B.

Schedules for collared elk in North Bend, WA……………………...………….79

C.

Land use layer created in ArcGIS 10...……………………………………….....80

D.

Road intensity layer created in ArcGIS 10..………………………………..…..81

E.

I-90 distance band layer created in ArcGIS 10..………………………………..82

F.

Riparian habitat at different distance bands from I-90 layer created in ArcGIS
10..…………………………………………………………...……….…………83

G.

Topographic Position Index layer created in ArcGIS 10..…………………...…84

iv

List of Figures
1

Map of Washington State and defined project area around North Bend, WA
(area=363km2)………………………………………………………………..…47

2

Elk home ranges (95% contour, in km2) of ten elk in North Bend, WA
determined from a Brownian Bridge Movement Model……………………..…51

3

Overlapping elk home ranges (95% contour, in km2) of ten elk identified as four
groups in North Bend, WA determined from a Brownian Bridge Movement
Model……………………………………………………………………………52

4

Composition of Land Use within the project area, North Bend, WA…………...54

5

Elk home ranges (50% contour, in km2) of ten elk in North Bend, WA
determined from a Brownian Bridge Movement Model……………………..…55

6

Bonferroni 95% confidence intervals for selection of land use variables by elk in
the project area, North Bend, WA…………………………………………...….58

7

Bonferroni 95% confidence intervals for selection of road intensity variables by
elk in the project area, North Bend, WA………………………………………..59

8

Bonferroni 95% confidence intervals for selection of distances from I-90 by elk
in the project area, North Bend, WA…………………………………...……….60

9

Bonferroni 95% confidence intervals for selection of riparian habitat at different
distances from I-90 by elk in the project area, North Bend, WA…………….....61

10

Bonferroni 95% confidence intervals for selection of topographic positions by elk
in the project area, North Bend, WA……………………………….…………...62

11

Elk captured by Reconyx game cameras utilizing riparian habitat and bridge
structures to cross safely under I-90 near North Bend, WA………………….…63

v

List of Tables
1

Number of elk carcasses removed and number of elk-vehicle collisions along I90 within the project area per 1 mile road segment of I-90 and Hwy 202. Data
Source: the Washington State Traffic Data base………………………………..48

2

Home range (95% contour, in km2) and core use area (50% contour, in km2) for
elk in North Bend, WA……………………...…………………………………. 49

3

Definitions of resource variables find in the project area, North Bend, WA…...50

4

Land use composition for the study area and elk home ranges. Core use area 50%
contour and home range 95% contour………………………...………………...53

5

Estimated resource selection log-likelihood chi-square test statistics for elk in
North Bend, WA. ……………………………………………………………….56

6

Estimated resource selection indices for elk in North Bend, WA.
estimated
habitat selection ratio,
= standard error of selection ratio,
and
are Bonferroni -adjusted 95% lower and upper confidence intervals…..57

vi

Acknowledgements
I am very fortunate to have the people in my life that I do and for the
supportive network they provide. I would like to give special thanks to
everyone who supported me in this great endeavor. This project was made
possible through the partnership between the Washington State
Department of Transportation and the Upper Snoqualmie Valley Elk
Management Group. At WSDOT I would like to thank Kelly McAllister,
wildlife biologist, for providing fantastic guidance, support and an ever
available ear, Mark Bakeman, biologist, for statistical guidance, Stacey
Plumley, GIS wiz, for GIS troubleshooting and Marion Carey, fish and
wildlife program manager, for supporting my internship in the WSDOT
Environmental Services office. Harold Erland at The Upper Snoqualmie
Valley Elk Management Group for his assistance with elk collaring and
data management. Martha Henderson, the director of the Graduate
Program on the Environment at The Evergreen State College for
sponsoring my internship and Dina Roberts, faculty, for the copious
amount of time spent providing me with vital support in all things thesis
related. Andy Duff at WDFW for his generous offering of time, invaluable
guidance and support on the Brownian Bridge Movement Model and
methods. The Muckleshoot Indian Tribe for sharing coordinate data. And
last but most certainly not least my family, friends and partner in crime
Andy Leveque since they were the ones that received the brunt of my
crankiness. Without their unconditional love and support I would not be
who I am today.

vii

Chapter 1: Literature Review
Introduction
Roads and Wildlife
The extent to which North America’s landscape has been modified to meet
human demands has wide ranging impacts. Roads have been a major form of
development, transecting the continent in linear patterns, linking people and commerce.
Road development and construction increased dramatically in the 20th century in the
United States to meet growing demands of automobile drivers (Forman et al. 2002). The
human population has now become dependent on roads and vehicles for daily activities,
resulting in a particularly expansive road system in America, with 4 million miles of
public roads (Forman et al. 2002). The Federal Interstate Highway System carries 22.8%
of all traffic in the US, despite only representing 1.2% of the country’s total public road
length (Fed. Hwy Adm. 2002, Forman et al. 2002). Between 15-20% of the US land area
is ecologically affected by this highway system (Fed. Hwy Adm. 2002). Washington
State has accumulated 7,046 miles of state and federal highways receiving 31.6 billion
miles of vehicle travel annually, which has doubled since 1960 (Washington State
Department of Transportation 2005).
Losses of wildlife due to wildlife vehicle collisions are among the most
noticeable ecological effects of roads. Wildlife-vehicle collisions across the United States
are estimated to be 300,000 each year, estimated to have grown from 200,000 to 300,000
during the 1990-2004 time frame (Huijser et al. 2008). Reasons for this increase could
include growing deer populations in many regions of the U.S., but could also be due to
increased traffic (Huijser et al. 2008). On Washington State highways vehicle collisions
involved at minimum, 14,969 deer and 415 elk over the five year period 2000-2004
(Myers et al. 2008). These minimum values were based on carcass removals. Actual

1

numbers of collisions with deer and elk in Washington State are unknown since not all
collisions with deer and elk result in instant mortality, with some animals not being
accounted for when death occurs at some distance from the roadway. Additionally, data
are only available for state maintained roads, local road departments rarely record carcass
removals.
Wildlife-vehicle collisions have safety consequences. Large ungulates, such as
elk can cause serious injury to drivers and substantial property damage (Huijser et al.
2008). Large ungulates are highly mobile and are more likely to enter roadways than less
mobile species, increasing the chance of collision (Gibbs and Shriver 2002, Forman et al.
2002). Driver safety is a primary goal for many transportation agencies and reducing or
eliminating collisions with large ungulates is a common problem for DOT management.
Therefore, transportation departments have invested in studies of wildlife-vehicle
collisions with the goal of reducing impacts on both humans and wildlife. Studies
completed to date underscore the complexity of factors that contribute to these accidents.
Wildlife-vehicle collisions are influenced by many factors including road characteristics
and human behavior (Bashore et al. 1985, Jaeger et al. 2005, Parris and Schneider 2008).
Commonly studied road characteristics include road geometry, dimensions,
spatial distribution, density, traffic volume, speed limit, and placement on the landscape.
Jaeger and colleagues (2005) found that road width and speed limit negatively impact
wildlife, but not as significantly as traffic volume. This suggests that wider roads and a
higher traffic speed result in greater wildlife-vehicle impacts (Forman and Alexander
1998, Jaeger et al. 2005). In addition, Gagnon and colleagues (2007) found that as traffic
volume increases, wildlife mortality and collisions increase. Highways with high traffic
volume have higher wildlife-vehicle collision rates and wildlife mortality, which
negatively affect wildlife populations (Gunther et al. 1998; Gunson et al. 2005).

2

The placement of roads on the landscape in relation to topography also
influences wildlife-vehicle collisions. The majority of roads in the US were constructed
in locations where transportation agencies could minimize difficulty of construction. For
example, several roads built in mountainous landscapes were placed in valley bottoms
where terrain was least resistant (Kaszynski 2000). Unfortunately, for montane wildlife,
preferred road locations often coincide with chosen travel corridors and wintering
grounds in these milder valley areas (Moen 1976). Consequently, many highways built in
valley floors of montane regions are faced with high wildlife-vehicle collisions and
wildlife mortality rates because of this conflict (Gagnon et al. 2007).
In addition to road characteristics, behavior of humans and wildlife also influence
wildlife-vehicle collisions and road mortality. Studies that have temporally quantified
wildlife-vehicle collision data have discovered that collision rates are higher at night than
during the day (Bashore et al. 1985, Gunson et al. 2005). At night, drivers have reduced
visibility and decreased reaction time, subsequently increasing driver and wildlife
vulnerability to vehicle collisions (Rost and Bailey 1979, Mastro et al. 2008). Some
species of wildlife are most active at dawn, dusk and night, contributing to increased
collisions at night (Jaarsma et al. 2006). In addition, most motorists don’t actively pay
attention to wildlife; instead, they usually look for other vehicles, which are usually the
most dangerous objects encountered on roadways (Rumar 1990). Many different collision
trends can be attributed to these variables but it is suggested that driver behavior is a
major influence (Rumar 1990). Unfortunately, few studies have quantified human
behavior as a factor influencing wildlife-vehicle collisions; therefore, further research is
necessary.
Traffic disturbance can have negative impacts to wildlife that aren’t as noticeable
as mortality. Traffic disturbances include road noise and traffic volume that can interfere

3

with normal wildlife behaviors, communication and reproduction, such as the functions
of bird songs (Gagnon et al. 2007, van der Ree et al. 2011). Some species of birds have
been reported to change their pitch and frequency to compete with road noise (Parris and
Schneider 2008). Subsequently, hampered detection of songs by other birds can lead to
difficulty in establishing and maintaining territories, attracting mates, and maintaining
pair bonds (Parris and Schneider 2008). Such interferences could lead to reduced
breeding success in noisy roadside habitats as found by Halfwerk and colleagues (2011)
where traffic noise caused females to lay smaller clutches in noisier areas. The variation
in the traffic noise frequency band overlapped most of the lower frequency part of the
great tit (Parus major) song (Halfwerk et al. 2011). Unfortunately, this study is one of
only a few that has researched road noise effects on breeding success of avian species. In
general, little research has been done on analyzing the effects of road noise on the
breeding success of wildlife.
Less obvious than wildlife-vehicle collisions, but likely more important impacts
of roads occur at a landscape level, where habitat fragmentation is a result. Effects from
habitat fragmentation influence habitat loss and reduced connectivity at both fine and
broad scales. Such changes at a landscape level can indirectly influence behavior,
survival, growth and reproductive success of individual animals, resulting in cumulative
effects at the population level (Harrison and Bruna 1999, Crooks and Sanjayan 2006).
Habitat loss and fragmentation occur when new roads are built by destroying
habitat, reducing patch size, and increasing the distance between patches (Andrén 1994).
Roads fragment the environment by transecting the landscape with dense impervious
surfaces and high volume traffic which may reduce wildlife access to essential resources
(Van der Ree et al. 2011). When patch size is reduced and distance between patches
increases, the result is often isolation effects; this can negatively impact population

4

viability (Andrén 1994, Fahrig 1997). Population viability can become compromised
when species cannot access resources, such as food, mates, and breeding sites (Jackson
and Fahrig 2011). Inaccessibility to these essential resources can lead to lower
reproductive and survival rates, which may reduce overall population persistence
(Thomas and Hanski 2004).
Behavioral changes in wildlife as a response to roads and traffic are also known
effects of fragmentation (Jaeger et al. 2005). Examples of behavioral modification
include home range shifts, altered movement patterns, and reproductive success
(Trombulack and Frissell 2000). When behavioral modification occurs, such as in road
avoidance, populations can become isolated when individuals are unable to move
between populations (Trombulak and Frissell 2000). Roads can alter an animal’s home
range selection, often as a road avoidance response, which has consequences when
important resources are located near roads (Trombulak and Frissell 2000). For example,
elk (Cervus elaphus) in Montana prefer feeding sites far from their visibility of roads
reducing availability of resource located near roads (Grover and Thompson 1986).
Several species, including frogs and snakes, have been found to avoid crossing roads
(Row et al. 2007, Bouchard et al. 2001). In these cases, roads are considered barriers to
movement, which can have negative effects at a population level (Beckmann et al. 2010).
When movement of animals between populations is inhibited, gene flow is reduced and
may cause significant genetic differentiation among populations (Crooks and Sanjayan
2006). In Germany, Reh and Seitz (1990) observed genetic drift caused by roads in the
common frog (Rana temporaria). When connectivity between populations is reduced and
populations are subdivided, they inevitably become smaller and more vulnerable. With
reduced connectivity a population becomes less likely to receive immigrants from other
habitats and, as a result may suffer from lack of genetic input and subsequently exhibit

5

inbreeding effects (Jaeger et al. 2005, Crooks and Sanjayan 2006). Lack of genetic input
and the resultant inbreeding contribute to genetic defects which may lower the probability
of population persistence (Fahrig 1997, Jackson and Fahrig 2011). Stochastic events can
further exacerbate isolation effects by increasing risk of extinction through random
demographic, genetic or environmental events (Crooks and Sanjayan 2006). Therefore,
chances of recolonization after local extinction are reduced in a fragmented landscape
(Hanski 1999).
Species life history traits may predispose populations to effects of roads caused
by habitat fragmentation and reduced connectivity. Species that occur in low densities,
have low reproductive rates and long generation times have increased risk of population
level effects caused by road mortality and barriers (Beckmann et al. 2010). For example,
many carnivore species have low reproductive rates, suggesting low turn over time to
compensate for high mortality rates caused by roads, thereby leading to population
declines. Many reptile species are inherently attracted to roads for thermoregulation
benefits, most notably snakes, are attracted to road surfaces. In addition, some reptiles lay
their eggs in gravel roads or on road shoulders (Sullivan 1981, Aresco 2005, Steen et al.
2006). Some research has shown that various animals do not behaviorally avoid roads,
including some frogs and snakes, which increase their risk of road mortality (Row et al.
2007). Therefore, these species are more likely to enter road surfaces and experience
higher mortality rates, ultimately affecting population persistence. Highly mobile species
are also vulnerable to road movement. Species with large movement ranges encounter all
types of landscape features more frequently than species with small movement ranges,
which increases their likelihood of crossing a major roadway (Gibbs and Shriver 2002).
For example, deer and elk are highly mobile species; males increase their mobility during
certain seasons of the year to find mates and seek release from hunting pressure (Moeller

6

2010, Cleveland et al. 2012) Road mortality rates of these species are highest during
individual dispersal in the fall (Rost and Bailey 1979). Unfortunately, few studies
elucidate the effects roads have on population viability. Most of the knowledge about
such effects has been acquired through monitoring animal abundance in relation to
roadways. Conversely, more research is needed to determine how life history traits lend
themselves to species specific vulnerabilities to roads. Testing theoretical knowledge of
what life history traits influences species vulnerability will inform road management and
mitigation to reduce impacts to wildlife.
Landscape Level Road Planning for Increased Connectivity
It was not until the mid-1990s that road ecologists increased their efforts to
examine the effects of habitat fragmentation and reduced connectivity caused by roads.
Scientists increasingly explored the dynamics at work over larger landscape scales. With
the development of tools, such as remote sensing, Global Information Systems (GIS), and
genetic techniques, scientists and road ecologists can now address land use change and
ecological impacts simultaneously across multiple spatial and temporal scales. The field
of landscape ecology has advanced by incorporating studies from road ecology. Road
ecology studies broadened to combine ideas from the fields of wildlife management and
conservation biology. This transdisciplinary approach to understanding how roads
influence the environment fostered the growth of a discipline known as road ecology.
Road ecology derives its theories from many different disciplines and its growth and
maturation have increased research and planning at a landscape level.
By addressing wildlife interactions at a landscape scale, mitigation targeting
wildlife habitat connectivity began to gain importance in North America, with Parks
Canada leading the way (Forman et al. 2002). The Trans-Canada highway twinning

7

process was planned for the late ‘90s; Parks Canada decided to take advantage of this
highway improvement opportunity, recognizing reduced costs for mitigation during a
highway upgrade in comparison to retrofitting efforts (Forman et al. 2002). While
planning for the twinning of the Trans-Canada highway in Banff National Park in
Alberta, Canada, the Department of Transportation understood the affects that the new,
larger divided highway would have on the wildlife. Therefore, research on large mammal
movement and highway crossings was implemented to inform the DOT of where and
what type of mitigation was necessary to reduce negative impacts of a larger highway.
Mitigation included installation of wildlife crossing bridges and culverts. After
completing the first stage of construction in 1996, Anthony Clevenger and colleagues
began monitoring 11 large mammal species, including bears, elk and cougars. They have
documented these species using crossing structures more than 200,000 times (Parks
Canada Agency 2012). For some individuals the use of crossing structures has been
incorporated into daily movements. For example, a grizzly bear traveled 1,600 kilometers
during the summer of 2012, using crossing structures 66 times (Highway Wilding 2012).
This demonstrates the success of crossing structures to facilitate the movement of wildlife
across the landscape. This project has resulted in being a seminal mitigation project,
demonstrating the importance of reducing road effects and the benefits of increasing
connectivity. Several projects in the United States have ensued since, recognizing the
importance of this type of mitigation.
Since Banff’s seminal mitigation project, management actions intended to
minimize negative ecological effects of roads have increased in the US. In Washington
State several actions have been taken to minimize the effects of roads on wildlife. The
Washington Habitat Connectivity Working Group was formed in 2007 to “ create tools
and analyses that identify opportunities and priorities to provide habitat connectivity in

8

Washington and surrounding habitats” (WHCWG 2010). An aspect of this group’s intent
is to minimize the effects of roads by mitigating for areas of most concern. Current work
with WSDOT has led to ranking sections of state highways based on specific habitat
connectivity concerns to prioritize mitigation efforts (Kelly McAllister, personal
communication, February 2013). In 2011 highway construction to reduce road effects and
increase connectivity across I-90, east of Snoqualmie Pass was launched, culminating
over a decade of negotiations and environmental permitting. Mitigation plans include fish
barrier corrections, installment of bridges, box and round culverts for terrestrial wildlife,
barrier fencing and two wildlife overpasses (Long et al. 2012). These are typical methods
used to mitigate for road effects.
Crossing Structures
Crossing structures are commonly used to increase connectivity and safe
crossings (Hardy et al. 2003). Overpass structures, sometimes referred to as wildlife
bridges or wildlife overpasses, are an effective mitigation measure implemented for
aiding most wildlife to safely cross roads (Clevenger and Waltho 2003). Ungulate species
like deer, elk and antelope have been found to prefer utilizing these wide open structures
because of their prey behavior (Kintsch and Cramer 2011). Carnivores prefer a more
intimate structure, such as a culvert, where they have more cover which they require for
stalking and hiding (Kintsch and Cramer 2011).Many recent structure designs have not
been tested for their attractiveness to multiple species, thus increasing the necessity for
on-going collaboration between engineers and biologists. Overpasses are costly, therefore
not commonly implemented as a mitigation measure. Perhaps the most challenging aspect
of crossing structure design is finding a structure that addresses the entire community of
wildlife that require rescue from the barrier effects caused by roads. Often, a particular
design is deficient in aiding all wildlife because it only functions well for select

9

taxonomic groups. Despite the challenges for meeting the needs a diverse wildlife
community, crossing structures are an effective measure for lessening barrier effects of
roads for some taxonomic groups (Kinstch and Cramer 2011).
Fencing and Signs
Fencing and warning systems are additional methods used to mitigate wildlifevehicle collisions. Fencing prevents wildlife from becoming casualties and guides them
to safe crossing opportunities, preventing wildlife-vehicle collisions and ensuring safe
crossings (Forman et al. 2002). By directing wildlife toward safe crossing opportunities,
ecological effects caused by roads are reduced (McCollister and Van Manen 2010).
Warning systems, such as wildlife crossing signs are another commonly used and cost
effective way to mitigate for wildlife-vehicle collisions (Beckmann et al. 2010). By
warning drivers that they are entering a wildlife-vehicle collision prone area, these
systems attempt to alter driver behavior, increasing awareness and potentially mitigating
for mortality caused by driver inattentiveness. Collision records are usually an indicator
of where to deploy signs.
Evaluation of Effectiveness
Much of the literature in road ecology is composed of evaluations on the
effectiveness of various mitigation techniques. Trail cameras are among the most
frequently used method for evaluation. Camera traps are a noninvasive way to observe
wildlife utilizing crossing structures (Hardy et al. 2003). Scientists analyze crossing rates
as a way to quantify use of crossing structures. By counting the number of times an
animal approaches and uses a structure, scientists can evaluate the structure’s
effectiveness. If wildlife are crossing and crossing at high volume, more than
approaching and retreating, structures are deemed a success. Unfortunately, much of the

10

surrounding species community composition is not accounted for when analyzing
wildlife use of structures. Therefore, camera traps are not an accurate method to use when
analyzing responses of the total community to crossing structures or potential gene flow.
They can only observe those animals choosing to approach the crossing structure.
Therefore, this is an insufficient way to understand which animals choose to avoid roads.
There are more effective methods used to quantify wildlife movement in relation to
highways and crossing structures.
Wildlife Movement, Advancement in Technology and Methodology – A novel
approach to understanding landscape level effects

Commonly used technologies like Global Positioning Systems (GPS) and Very
High Frequency (VHF) collars are put on wildlife as a way to monitor their movements
when constant observation is impossible. GPS collars are technologically more advanced
than VHF collars, as they store more location data derived from satellite communications,
and, as a result, they are used more frequently despite being expensive (Coulombe et al.
2006). In conjunction with Geographic Information Systems (GIS), the movements of
collared wildlife can now be observed and analyzed at larger scales, mapping movements
in relation to roads to indicate behavioral responses like road avoidance behavior (Dodd
et al. 2007). These methods are a tremendous advancement over older methods for
quantifying road effects which entailed the collection of wildlife-vehicle collision and
carcass removal data. GPS telemetry and GIS can ultimately be used to pinpoint where
wildlife are avoiding roads and where they are crossing roads identifying potential
mitigation sites (Dodd et al. 2007, Lewis et al. 2011, Vaughan et al. 2012). Identifying
key habitat variables, land use characteristics, and road characteristics in GIS is an
effective way to understand what variables influence wildlife movement in relation to
roadways.

11

Home range analyses identify areas used by tracked animals, including both size
of area used and intensity of use (‘utilization distribution’ or UD) (Burt 1943, Worton
1989). Home range analysis using GIS allows scientists to quantify home ranges
spatially. These are commonly used methods for understanding wildlife movements in
the field of wildlife management. When applied to studies of road effects of wildlife,
transportation agencies can better understand wildlife behaviors and barriers to their
movements. Commonly used wildlife-vehicle collision and carcass removal records
provide limited information describing where wildlife are being hit, while analyzing
wildlife movements can show where they avoid crossing and suggest reasons for
choosing to cross where they are crossing (Lewis et al. 2011). Knowing how wildlife
move in relation to roads provides critical information to resource managers and
transportation planners, as they can identify areas where wildlife are likely to cross roads
and cause wildlife-vehicle collisions (Vaughan et al. 2012). Subsequently, they can also
identify where wildlife aren’t crossing roads and why. Further research on the
applicability of such methods is needed. With advancement in technology and methods
for analyzing wildlife movement in conjunction with multiagency collaboration, new
insights can be gained and improved mitigation of road effects on wildlife can be
developed. There are multiple ways to estimate home ranges or UDs of wildlife
including the convex polygon, kernel density estimation and Brownian bridge movement
model.
The convex polygon method is one of the oldest methods used to estimate home
ranges; it is a relatively simple approach to defining an animal’s home range. In this
method, a polygon circumscribes all of the known locations of the animal. This indicates
to the researcher that all of the animal’s activities are confined within this area.
Unfortunately, this method does not describe where animals spend a lot of time or chosen

12

movement paths. Therefore, this method does not lend itself well to analyzing where
animals choose to cross a roadway. The commonly used fixed kernel density method
describes an animal’s activities with nested polygons that encompass increasing
proportions of the locations gathered for the animal (Fieberg 2007). This approach
summarizes home ranges and highlights where animals spend most of their time;
additionally, it describes the outer bounds of area the animal inhabits. Many authors have
chosen to use this approach over the convex polygon method because it does not assume
that sample points lie on the home range boundary. Instead, it generates contours of
relative density (Worton 1989, Fieberg et al. 2010).
The Brownian bridge movement model (BBMM) is a recent advance in
methodology for analyzing wildlife movement that is thought to more accurately portray
paths and responses to roads. BBMM uses coordinate data that contains information on
the date and time of each point, calculates the average movement rates of individuals and
incorporates time intervals between GPS points to establish contour probabilities of an
animal being in an area (Bullard 1991, Horne et al. 2007). This method incorporates
location errors commonly associated with coordinate data into the equation, reducing bias
of results. Minimum convex polygon and fixed kernel density do not account for location
errors. BBMM can accommodate more detailed animal tracks provided by modern GPS,
providing a more realistic depiction of animal paths (Kie et al. 2010). By providing a
more realistic depiction of paths, scientists can then estimate attributes of an animal’s
preferred path. Therefore, the BBMM provides a more robust view of an animal’s use of
its surroundings, which transportation agencies can then use to better understand how
wildlife move in response to roads. This method can be used by DOTs to identify areas
where wildlife are likely to cross roads, predicting locations based on the identification of
preferred attributes through resource selection analyses. In addition, this method can be

13

used to identify areas where habitat connectivity objectives call for improvements,
identifying barriers to wildlife movement.
Conclusions

Roads are a prevalent linear features found throughout developed landscapes in
North America. Their negative effects are evident among many natural systems, with
habitat fragmentation effects among the most severe, affecting population viability of
some wildlife species. Transportation agencies are inherently invested in securing the
safety of drivers including prevention of wildlife-vehicle collisions. Mitigation for
fragmentation effects by roads is growing among transportation agencies as the
importance of connected landscapes becomes widely recognized by road planners and
biologists alike. Wildlife-vehicle collision and carcass removal data alone cannot be used
to understand the large landscape level effects of roads. Therefore, other methods, such as
monitoring wildlife movements in relation to roadways are good alternatives to choose
when trying to understand how roads fragment wildlife populations. With the
advancement in technologies and methods, DOTs and road ecologist are more equipped
then ever to fully understand the ecological costs associated with roads. Unfortunately,
technologies used to analyze wildlife movements remotely are expensive, and when
coupled with budget cutbacks, agencies are limited in the analyses they can choose from.
Therefore, multiagency collaboration could be an alternative to counteract costs involved
with these technologies, increasing the attainability of wildlife movement data. We know
that landscape level effects are the most costly of road effects and, therefore, it is of great
importance for DOTs to account for such affects when planning for mitigation. A way to
better understand those landscape level effects is through monitoring wildlife movements
in relation to roadways. By analyzing wildlife movements road planners and road

14

ecologists can more fully understand the effects of roads including many effects not
apparent when using wildlife-vehicle collision and carcass removal data alone.
Benefits of Multiagency Collaboration and Future Research
With advanced technology and techniques come costs. GPS collar tracking of
individual animals is an expensive approach and, therefore, it is not commonly used by
DOTs when planning for mitigation. In addition, many agencies are in a time of budget
cut backs, reducing the likelihood of using such methods. DOTs can benefit from
collaborating with wildlife agencies and other entities through sharing wildlife movement
data to better understand how roads they manage are affecting wildlife species of interest,
thus cutting costs of performing their own field research. Consequently, wildlife agencies
are then better able to achieve their wildlife protection goals when transportation agencies
increase their consideration of wildlife during planning and design of highway
improvements. The BBMM is available to DOTs and can lend itself to more accurately
understanding where wildlife are moving and what resources they are using in relation to
roadways. By understanding how wildlife are moving and what resources they are
selecting for in relation to roads, appropriate mitigation at a landscape level can occur. In
the long term, DOT agencies will be saving money and time, while increasing the
efficiency of mitigation for fragmentation effects.
The Upper Snoqualmie Valley Elk Management Group (USVEMG), the
Washington State Department of Transportation (WSDOT) and myself, a graduate
student at The Evergreen State College (TESC) have teamed up to better manage for elk
in the Snoqualmie Valley around North Bend Washington. The community of North
Bend has a relatively new problem with elk that have become habituated to humans in a
heterogeneous landscape near a large, high volume interstate. Interstate-90 (I-90)

15

transects the North Bend area, as the primary East-West traffic corridor in Washington
State, resulting in significant habitat fragmentation effects and high wildlife-vehicle
collisions. The fragmentation effects, include elk behavior changes caused by Interstate
90 are poorly understood. As mentioned earlier, Forman and Alexander (1998) suggest
that behavioral avoidance of roads may have the most pervasive effects. Gagnon and
colleagues (2007) found that elk may avoid roads during periods of high traffic volume.
If elk behavioral changes are positively correlated with traffic volumes, such affects
could be magnified in light of future population growth and traffic volume increases on I90. Therefore, I will be studying GPS collared individuals from this herd to understand
their current home range distribution and resource use in relation to I-90. This will give
insight into how large human habituated ungulates respond to high trafficked interstates.
Most research on how roads affect elk and large ungulates have taken place in remote
areas (Rost and Bailey 1979, Grover and Thompson 1986, Cole et al. 1997)
Few studies have looked at human habituated elk (Lee and Miller 2003,
Cleveland et al. 2012) with none analyzing how habituated elk respond to large
interstates. In light of future human population growth and ever increasing wild-urban
interfaces, understanding how human habituated elk respond to heavy traffic interstates
necessitates further investigation. In addition, the gray wolf (Canis lupus) an apex
predator of elk has been absent from this ecological system since the 1930s (Becker et al.
2013). Its recent return to Washington state, continual population growth and potential
return to forests near North Bend, has the potential to change the dynamics of the herd.
To manage for the resiliency of this herd for the future, all large ecologically influential
factors need to be understood. We currently lack a complete understanding of the
behavioral response of these human-habituated elk to high-traffic interstates and the
potential damaging fragmentation effects of roads on their persistence. Few studies have

16

quantified fragmentation effects through the approach of analyzing movement patterns
remotely (Dodd et al. 2007, Gagnon et al. 2007, St. Clair and Forrest 2009) and none
using the BBMM method. As presented in Chapter 2, I address these unknown aspects of
road effects of Interstate 90, by analyzing the North Bend herd’s movement and resource
selection using BBMM and ArcGIS 10.

17

Chapter 2 – Analysis of Elk Home Ranges and Resource Selection
Introduction – Road Impacts on Wildlife
Roads are a main form of development transecting vast areas of the Earth’s
surface, often negatively affecting ecosystems and associated wildlife (Forman et al.
2002, Coffin 2007). Effects of roads include increased wildlife mortality rates; with
vehicle collisions among the most noticeable and in some cases primary causes of
mortality for large vertebrates (Huijser et al. 2007, Coffin 2007). The less obvious but
also influential impact of roads on ecosystems is habitat fragmentation. Road networks
fragment landscapes and populations by impeding wildlife movement through physical
barriers and behavioral avoidance, impacting population viability and resilience to
changing environmental conditions (Beckmann et al. 2010). Restriction of movements
can reduce migration, dispersal and opportunities for mating, leading to population
subdivision and genetic differentiation (Andrén 1994, Fahrig 1997). Maintaining
connectivity between subdivided populations of large ungulates in landscapes fragmented
by road networks can be challenging, however efforts to mitigate road effects are
necessary (Gibbs and Shriver 2002).
Human conflicts with large ungulates can be serious when the animals attempt to
cross roads, resulting in collisions. Large ungulates can cause substantial property
damage and human injury when wildlife-vehicle collisions occur (Nielsen et al. 2003). It
is therefore of interest to transportation agencies to manage for driver safety in high
wildlife-vehicle collision prone areas. Mitigating for large ungulate connectivity and
driver safety by transportation departments has historically been a product of analyzing
wildlife-vehicle collisions (Huijser et al. 2007). Unfortunately, this approach undermines
barrier and behavioral effects of roads, which are better observed by analyzing wildlife
movement. By analyzing wildlife movements and resource selection in relation to roads,

18

a more insightful representation of behavioral response to roads can be obtained and
inform mitigation decisions made by transportation planners and wildlife managers.
Near the community of North Bend in western Washington State, a high
incidence of elk-vehicle collisions has become a relatively new problem. Elk have
become habituated to humans in a heterogeneous landscape near a large, high volume
interstate. Habituation to humans can be a result of high elk density, maximization of
reproductive fitness, and reduced lethal interactions with humans, and when human
activities are consistent and predictable (Thompson and Henderson 1998, Walter et al.
2010, Cleveland et al. 2012). Interstate-90 (I-90) transects the North Bend area, as the
primary East-West traffic corridor in Washington State, resulting in significant habitat
fragmentation effects and a high number of wildlife-vehicle collisions. A partnership
between The Upper Snoqualmie Valley Elk Management Group (USVEMG) and the
Washington State Department of Transportation (WSDOT) was formed to study elk
movement and minimize elk-vehicle collisions. USVEMG was initially created as an
effort to gain information to help minimize property damage and public safety risks
associated with human habituated elk in North Bend. One outgrowth of this partnership
involved equipping local elk with GPS collars to monitor their movements. USVEMG
has accumulated several years’ worth of coordinate data on members of this habituated
herd that has not been analyzed. Therefore, it is currently unknown how these elk
behaviorally respond to roads and utilize resources in the highly human modified area of
North Bend. To manage this herd and mitigate for collisions, understanding elk
movement and space use in relation to I-90 is important.
This research addressed movement patterns and resource selection of members of
this elk herd in North Bend, WA. Spatial locations of elk home ranges in relation to I-90
were examined to understand whether I-90 influenced elk movement. I asked whether elk

19

would establish home ranges and core use areas away from I-90. A comprehensive
review of ungulate-highway interactions found that high volume interstates like I-90
interrupt elk behavior or had a “road effect zone” up to 425 meters (Gagnon et al. 2007).
Therefore, it was expected that elk would avoid areas at distances less than 425 meters
from I-90. A resource selection approach was used to gain insight into elk space use at a
fine scale. I also hypothesized that elk used resources disproportionately to what is
available in the study area, with elk displaying preference or avoidance for specific
resources. By understanding the location of home ranges and which resources were
selected for, transportation planners can improve mitigation efforts for ungulate species
by ensuring safe crossing areas and prevention of crossing at unsafe sites where selected
resources are located; thus both ensuring connectivity and reducing wildlife-vehicle
collisions.
Materials and Methods

Study Area

The study area was centered around North Bend, WA (UTM 10T 591432,
5260765) within a small southern section of game management unit (GMU) 460, located
50 km east of Seattle in the foothills of the Cascade mountain range. The study area
encompassed 363 km2 in the upper Snoqualmie Valley (Figure 1). Elevation ranged from
130 m to 4,167 m, from the valley bottom to the top ridge line with Mount Si, the tallest
topographic feature within the study area. The project area was a matrix of different land
use and habitat types. Within the valley, land uses included housing, subdivisions, private
agriculture lands, commercial buildings, and main county and state arterial roadways.
Roadways consisted of residential, main city arterials and major state highways and

20

interstates (Hwy 202 and I-90). I-90, the largest interstate located in the project area is
located in the middle of the valley as the main West/East interstate. Traffic volumes
average 28,000 vehicles per day and are increasing by ~ 2.1% per year (WSDOT 2008).
The North Fork Snoqualmie River flows through the middle of the valley floor and the
South Fork Snoqualmie River follows the I-90 corridor, providing abundant riparian
habitat. Upland from the valley, forests are dominated by Douglas Fir (Pseudotsuga
menziesii), Western Hemlock (Tsuga heterophylla), and Pacific Silver Fir (Abies
amabilis). Weather is characterized by maritime conditions with average annual
precipitation approximately 1,500 mm (NOAA 2012). The average summer temperature
was 16 deg. C. and the average winter temperature was 4 deg. C.
During the mid to late 1800s, human encroachment and over hunting led to local
extinction of the Snoqualmie Valley herd. Rocky Mountain elk were then shipped by
railcar from Montana to reestablish a herd within the valley during the early 19th century
(Couch 1935). The non-migratory behavior of this herd, in conjunction with human
development and human habituation has led to considerable human-wildlife conflicts.
Common human-wildlife conflicts found in wildland-urban interfaces like North Bend
include damage to agriculture and private property (Walter et al. 2010). Columbian
Black-tailed deer (Odocoileus hemionus columbianus), another native ungulate, was
present in the study area. Predators included the occasional presence of cougar (Puma
concolor), black bear (Ursus americanus) and humans. Some of the study area allowed
limited hunting with special damage tags. Additionally, elk encountered fatal interactions
with motorists on roadways, with I-90 being the main contributor. Between the years
2009 and 2011 a total of 62 elk carcasses were removed from I-90 and Hwy 202 within
the project area (WSDOT 2011)(Table 1).

21

Elk Capture and Telemetry
The Upper Snoqualmie Valley Elk Management Group captured 9 adult female
elk from 2010-2012. An additional female elk that entered the study area during this time
period was monitored and included in this study. This 10th elk was originally captured
and collared by the Muckleshoot Indian Tribe and fitted with a Vectronics GPS Plus
collar (Vectronics, Starkville, Mississippi). The other 9 elk were fitted with global
positioning system (GPS) telemetry collars, LOTEK 4400S and 4400M (Lotek Wireless,
Newmarket, Ontario, Canada). Seven LOTEK 4400S collars were refurbished collars
supplied by WSDOT and two LOTEK 4400M collars were purchased new. Elk were
captured using clover traps (Thompson et al. 1989). One collared elk died during the
years 2010-2012, this collar was then reused, totaling ten females collared. If necessary,
immobilization was accomplished using telazol/xylazine HCL with Yohimbine as the
reversal drug. Handling procedures were under the direct control of state or Muckleshoot
Tribal biologists or a veterinarian experienced at handling elk (USVEMG 2010). Due to
the random nature of elk capture, collars were deployed at varying dates with a variety of
collar schedules (Table 2 and Appendix B). Downloaded GPS locations were converted
to North America Datum (NAD) of 1983, Universal Transverse Mercator (UTM) Zone
10 using ArcGIS 10 Convert Coordinate Notation (ESRI 2013). Global positioning
system-collar fix-rates varied greatly between collars (19% - 96.15%) and location error
was marginal (error = 24 m). Location error was obtained by testing one LOTEK 4400S
collar for position accuracy with a handheld Trimble GPS GEOXT explorer 6000 series
unit. Due to the variability in collar fix-rates and location error, habitat could bias the
location data (Frair et al. 2010). Dense canopy cover and steep terrain found in the
project area could have decreased fix-rates and increased location error.

22

Home Ranges and Utilization Distributions- BBMM
To delineate home ranges and utilization distributions (UDs) for each animal, the
Brownian bridge movement model was used (Horne et al. 2007). Calculated home ranges
were used to explore the spatial relationship of elk space use with I-90, to determine if I90 influences elk movement. UDs were used to define use contours, described in detail
later. The Brownian bridge movement model is a continuous-time stochastic movement
model in which the probability of an animal being in an area is calculated. BBMM
requires (1) time-specific location data, (2) the estimated error associated with location
data, (3) the distance and time between successive locations (4) the animals average
movement rate and (5) the grid-cell size for the output (Horne et al. 2007). The BBMM is
based on the properties of a conditional random walk between successive pairs of
locations, dependent on the time between locations, the distance between locations, and
the Brownian motion variance that is related to the animal's mobility (Horne et al. 2007).
The BBMM estimates the probability of various animal paths between sequential
locations irrespective of the density of locations where the width of the Brownian bridge
is conditioned on time duration between beginning and ending GPS locations and GPS
location error. Unlike other kernel density methods, BBMM is able to predict animal
movement paths. A program developed in the R language for statistical computing (R
Development Core Team 2007); (Appendix A) was used to create home ranges. Since
collar types and schedules differed among different individual’s BBMM max lag and
location error, inputs were unique to individual collar schedules. A grid-cell size of 30 X
30 meter was used to provide high-resolution mapping, while maintaining a reasonable
processing time.
Cell values for each elk’s UD were summed and then re-scaled with their
cumulative cell values summing to 1, such that the home ranges of each elk was

23

represented by one UD. As the accumulated cell values reach 1, use is considered to be
low. Core use areas were defined by 50% contour lines and home ranges boundary by
95% contour lines. Normal activity of an individual is commonly accepted at 95% of the
locations of an animal within the entire home range area (White and Garrott 1990). Core
use areas are areas within the home range that are used more frequently than any other
area (Samuel et al. 1985). They usually contain home sites and areas of most dependable
resources (Kaufmann 1962). Both home ranges and core use areas were used to explore
space use in relation to I-90. It is important to examine both since activity patterns differ
between the two.
In order to explore space use by elk in relation to I-90, digitized polylines
between core use areas and I-90 buffer (pixel size 9m) were created in ArcGIS 10 as a
measure of the average distance between core use areas and I-90. Visual observations of
core use areas and home range locations were explored in ArcGIS 10 to understand
compass location of home range and core use areas in relation to I-90 (north, east, south,
west), if home ranges overlapped I-90 and to what extent. Visual observation of core use
areas and home ranges in relation to I-90 can give insight into behavioral avoidance, with
elk spending a majority of their time far from I-90 or vice versa. Previous research shows
that the higher the traffic volume of a road is, the less wildlife cross and the greater the
distance is that wildlife spend from the road (Gagnon et al. 2007b). Average land use
composition within each home range was also calculated
Estimating Resource Selection
GPS locations of 10 female elk were analyzed to describe second order resource
selection (Manly et al. 2002). Resource selection for each elk was determined by
overlaying coordinate points contained by the calculated 95% home range contour on

24

resource category layers and then summed in ArcGIS 10. Proportions of use for each
variable in each category were calculated to define use. For estimating proportions of
available resources, each resource category was censused for the extent of the project
area in ArcGIS 10. The project area was defined by mapping all the coordinate points
found within the 95% contour line for each home range and using the minimum bounding
geometry tool in ArcGIS 10 to draw a minimum convex polygon around all points,
defining the project area boundary.
The term resource will be used here when referring to 27 variables in five broad
“resource” categories. The five categories included 1) distance to I-90, 2) distance to I-90
and use of riparian habitat, 3) road intensity, 4) land use and 5) topographic position
index. Distance bands were created using ArcGIS 10 buffer tool, to understand space use
in relation to I-90 at different distances (<50 m, 50-250 m, 250-450 m, 450-1,000 m and
>1,000 m) (Appendix E). Distances were based on distances used by Dodd and
colleagues (2007) with adjustments to accommodate for the smaller size of our study area
(Gagnon et al. 2007b). Previous research found that large mammals like elk, exhibit a
behavioral response to highway disturbance up to 425 m (Gagnon et al. 2007a). Use of
riparian habitat was also evaluated at different distances from I-90, using the same
distance bands as noted above (Appendix F). Riparian habitat was chosen because of its
use by elk as a corridor for movement (Arizona Game and Fish Department 2011). A
road layer obtained from ESRI was classified into 3 different variables defining road
intensities based on speed limit, with roads greater than 55mph classified as high
intensity, 35-45mph classified as medium intensity, and less than 25mph were classified
as low intensity (Table 3 and Appendix D). Speed limit was used to define intensity level
because research has shown that roads with higher speed limits are found to impact
wildlife more than roads with lower designated speed limits (Jaeger et al. 2005). Land use

25

variables were digitized spatially using ArcGIS editor tool and a 2011 NAIP image at a
1:24,000 scale (USDA 2012). Variables included development intensity (high and low),
developed open space, open/forage, wetland, riparian, forest and open water (Table 3 and
Appendix C). Variables were chosen based off of similar land classifications used in the
REGAP analysis done by the Washington Wildlife Habitat Connectivity Working Group
(WHCWG 2010). The Topographic Position Index (TPI) was used to calculate the
influence of slope and topographic features on elk movement. This information was used
as opposed to other topographic information due to the accuracy at which TPI defined
slope position (Weiss 2006). TPI can distinguish between valley floors and ridge lines
that resemble the same percent slope. Variables included ridge, upper slope, middle
slope, flat slope, lower slope and valley. The TPI layer was developed from a 30 x 30
meter pixel USGS Digital Elevation Model (DEM) data using TPI v. 2.3a (Jenness
2006)(Appendix G).
Statistical Analysis/Resource Selection
Once the proportions of used and available resources were known, selection was
assessed by estimating log-likelihood chi-square test statistics and selection ratios (ratio
of the proportion of resource used and available) for different resource variables (Manly
et al. 2002). This is a widely used method to test for selection of resources by wildlife
(Neu et al. 1974). The log-likelihood chi-square test was calculated as
, where

is the expected value of

, to test the null hypothesis

that resource selection is proportional to availability or that resource selection is random
(4.27 Manly et al. 2002). When the

statistic was significantly larger than the chi-

square distribution, with n(I-1) df, there was evidence of non-random selection by at least
some of the elk, suggesting resource selection occurred.

26

Selection ratios were used to test the null hypothesis that elk do not display
preference or avoidance of resources. Selection ratios (
resource variable
all elk, and

, where

i) were calculated for each

is the proportion of used units in variable i by

is the proportion of available resource units in variable i (4.31 Manly et al.

2002). Standard errors for selection ratios were calculated as s.e.(
where

is the used resource units in variable i ,

used units sampled and

)
is the total number of

is the proportion of available resource units in variable i (4.14

Manly et al. 2002). Since multiple tests were computed across variables within a category
simultaneous Bonferroni adjusted confidence intervals were calculated as
for each variable in order to locate significant selections. The Bonferroni correction is
considered the simplest and most conservative method to control for type 1 errors.
Adjusted confidence intervals were calculated as

, where

I is the

number of resource variables in the category (4.33 Manly et al. 2002). Significant
selection was considered to occur when 1 < the confidence interval, significant avoidance
occurred when the 1 > confidence interval, neither selection nor avoidance occurred
when 1 was found inside the confidence interval (Manly et al. 2002).
Results
Home Ranges
Ten female elk were fitted with GPS collars between the years of 2010 and 2012.
Location data for all ten individuals were used to estimate home ranges. The number of
locations from the 10 individuals ranged from 149 to 14,119 (Table 2). For this
population of elk, the average home range (95% contour) was 9 km2 (range = 4 to 23
km2) (Table 2). Eight individuals excluding elk 1326 and 601_2 had home ranges
bordering I-90 (Figure 2). Six out of these 8 overlapped slightly with I-90 (Figure 2).
Two elk, 351 and 324 had home ranges that overlapped I-90 substantially, displaying

27

abundant space use on the North and South side of I-90 (Figure 2). Every home range
was found to overlap with at least one other forming 4 distinct groups, group 1 (601_2
and 1326), group 2 (351 and 324), group 3 (3870 and 1550) and group 4 (341, 601_1,
337, and 339) (Figure 3). Average composition of land use within home ranges was
primarily forest (53.39%), open/forage (18.37%) and developed-low (9.54%) (Table 4,
Figure 4).
Core Use Areas
For this population of elk, the average core use area (50% contour) was 2 km2
(range = 0.41 km2 to 4 km2) (Table 2). Eight individuals (all elk 1326 and 601_2) had
their core use areas bordering I-90, with elk 1550’s core use area slightly overlapping
(Figure 5). Elk 324 was the only elk that displayed core use areas on the North and South
side of I-90 (Figure 5). Average distance between elk core use areas and I-90 was 1,647
meters (range = 384 m to 2,759 m). When elk 3870 was excluded distance decreased to
1,400 meters. Elk 324 had the smallest average distance between core use area and I-90
with 384 meters (range = 145 m to 894 m). Average composition of land use within core
use areas differed from composition within home ranges, with forest at 48.73%,
open/forage 23.22% and developed-low 8.08% (Table 4, Figure 4).
Resource Selection
Only coordinate data that resided within 95% contours from all ten female elk
were used to analyze resource selection. Since elk 3870’s coordinate data comprised 60%
of the total dataset, a subsample of 1,176 random points were sampled using excel to
avoid resource selection bias by the individual. A total of 10,297 points were used to
assess second order resource selection. For all five resource categories the null hypothesis
that selection is proportional to availability was rejected at significant p value of 0.001

28

(Table 5). This result suggested that selection of resources by elk was not random.
Further investigation of resource selection using selection ratios and confidence intervals,
revealed selection of variables within categories differed significantly, displaying
preference for some variables and avoidance of other variables (Table 6). Therefore, the
null hypothesis that all resource variables are selected equally was rejected.
Within the land use category elk selection differed between different land use
variables, with significant differentiation of selection between developed-open and
open/forage from the rest (Figure 6 and Appendix C). Developed-low and riparian were
similarly selected for while wetland and open water were similarly selected against, with
all other variables showing unique selection or avoidance (Figure 6). Elk showed
preference for developed high, developed low, developed open, open/forage and riparian
habitat, avoiding forest, wetland and open water variables (Table 6). However, the
selection for developed high and selection against forest was slight.
Within the road intensity category elk avoided medium and high intensity roads,
while slightly selecting for low intensity roads within the road intensity category (Table
6). Selection against medium and high intensity roads was similar (Figure 7).
Elk displayed differential selection for different distances from I-90, selecting for
distance bands of 50-250 m, 250-450 m and 450-1,000 m (50-1,000 m range), while
avoiding distance bands close to I-90 at <50 m and distance bands far from I-90 at >1,000
m (Table 6). Selection for distances of 50-250m and 250-450 were similarly selected
while distances 250-450 m and 450-1,000 m were similarly selected against, showing that
selection of 50-250 m and 450-1,000 m differed (figure 8).
Riparian habitat selection differed at varying distance from I-90. Elk selected for
riparian habitat far from I-90 at distances of >450 meters, while avoiding distances of

29

<50 meters, 50-250 m and 250-450 m (Table 6). Selection against distances of 50-250
and 250-450 m were similar, while all other selections displayed unique selection (Figure
9).
Lastly, elk displayed a strong selection for flat slopes, avoiding all other
topographic positions (Table 6). Topographic positions ridge, lower slope and valley
were similarly selected against, while elk significantly selected for flat slopes (Figure 10).
Discussion
Home Ranges
The results show that most elk home ranges are located bordering I-90, with
some slightly overlapping I-90; only two home ranges largely overlapped I-90. These
results suggest some individuals approach I-90, but few crossed and spent abundant time
on opposite sides of the interstate. Only two individuals, 351 and 324, displayed abundant
time spent on both the north and south sides of I-90, suggesting that they crossed I-90
multiple times to access resources while others did not. These results contradict
expectations that elk would avoid I-90 at great distances. Previous research suggested
that large mammals like elk are negatively influenced by large interstates similar to I-90,
displaying strong behavioral avoidance (Rost and Bailey 1979, Dodd et al. 2007, Gagnon
et al. 2007a). However these studies evaluated space use of non-human habituated elk
which behave differently than habituated elk. Human habituated elk are less disturbed
and display a more mild behavioral response to constant human presence in contrast to
their more wild counterpart (Stankowich 2008, Walter et al. 2010). Reasons for
habituation include the need for elk to maximize reproductive fitness, and due to learned
behavioral responses to non-lethal interactions with humans (Thompson and Henderson

30

1998). Therefore, human habituated elk may respond to high volume interstates and roads
in general differently than remote, non-habituated elk populations.
It is important to note that home ranges were calculated using the highly accurate
Brownian bridge movement model, however poor fix rates and high lag time between
fixes could influence the precision of the contours, increasing contour width, possibly
accounting for the slight overlap of home ranges on I-90. Therefore, the results could
show an overly conservative home range size, creating larger home ranges than what
actually occurred. Future research with standardized collar schedules and reduced lag
time between fixes could provide a more refined home range and better depiction of
behavioral response to I-90 by the elk. Nonetheless, these results suggest that the human
habituated elk in the North Bend area are spatially influenced by I-90, with some
displaying behavioral avoidance. Therefore, I-90 could be considered a partial barrier to
elk in the North Bend area but not a completely impassible structure, due to the riparian
underpasses or bridges present, as will be discussed below.
Average annual home range size (95% contour) of 9 km2 (range = 4 km2 to 23
km2) falls within the lower range for what is commonly found in the literature (Anderson
et al. 2005). Annual home range size of elk can be as small as 3 km2 and as large as 245
km2, depending upon many different factors (Peek 2003, Anderson et al. 2005). A study
on two non-migratory female groups located in a mesic California redwood forest,
reported annual home ranges of 3 km2 (Franklin et al. 1979). Some individuals within
North Bend displayed similar home range sizes as to what was found by Franklin and
colleagues, with half displaying annual home range size of less than 6 km2. It is important
to note that these individuals were found in spatially different groups (Table 2, Figure 3).
However, non-migratory home ranges of elk found by Moeller, south of this study area,
located south of Mount Rainier had average annual home ranges of 62 km2, ranging

31

between 5.40 km2 and 102 km2 (Moeller 2010). This area is fairly undeveloped in
comparison to North Bend, potentially influencing the difference found in home range
size between the two herds. In addition, they had a longer telemetry monitoring period.
Many factors can influence small home range size. For example, elk may reduce
travel distance in order to balance the needs of minimizing predation risk and energy
demands, while meeting forage uptake, minimizing thermal stress and maintaining social
contacts (Anderson et al. 2005). Home range size must meet the energy and nutritional
demands of wildlife, when such demands are not met, wildlife increase distances traveled
to access additional resources. Therefore, when forage is scare or patchily distributed,
wildlife range over large areas (Ford 1983, Relyea et al. 2000). Consequently, the small
home range size of the elk in this population suggests that energy and nutritional
demands are being met. High-quality forage such as lawns, gardens, golf courses,
pastures, and hay meadows found in this urban setting could be part of the reason for
their small home range sizes, in addition to human habituation (Thompson and
Henderson 1998). If North Bend increases development concurrent with elk population
growth, resources for elk could become scarcer, forcing elk to increase home range size
to access and compete for resources, thus increasing interactions with humans and I-90.
Therefore, it is important to identify potential safe crossing opportunities so that elk can
access additional resources that may be located on the other side of I-90.
Substantial space use overlap was found among several individuals. Groups
appeared to utilize similar spaces in relation to I-90, with some groups staying away from
I-90, as seen with group 1 and some staying close, as seen with groups 2, 3 and 4. Group
2 displayed space use on both North and South sides of I-90, however only 324 had core
use areas on both sides. This could indicate that these individuals belong to similar family
groups. Therefore, these individuals’ home ranges may be spatially auto correlated,

32

displaying similar resource selection, reducing the effective sample size. Further testing
for autocorrelation should be executed.
Core Use Areas
Most core use areas bordered I-90, but at a great distance from I-90 on average >
1,500 m, suggesting elk spend considerable time away from I-90. Only one elk displayed
core use areas located on both sides of I-90. These results suggest that this elk crossed I90 multiple times to access resources, while most individuals did not. Therefore, these
findings imply that I-90 has a spatial influence on core use location. In addition, evidence
that few individuals displayed core use areas on opposite sides of I-90 despite bordering
I-90, also supports the conclusions that most elk could be behaviorally avoiding I-90,
implying that I-90 is a partial barrier to their movement.
Resource Selection
The results confirm that elk select for resources in their home range (95%
contour) disproportionately to what is available in the study area, displaying preference
and avoidance of specific resources. As such, the null hypothesis that resources are
selected proportionately to availability was not supported. It is important to note that
these results were treated with conservative log-likelihood chi-square test and Bonferroni
adjusted confidence intervals and were found to be significant. However, habitat bias and
collar schedules could potentially affect the results of selection.
Within the land use layer elk were found to select for developed-low, developedopen, and open/forage. Selection of these variables was as expected since North Bend
offers high-quality forage and security. Human habituated elk have been found to prefer
forage offered by lawns, ornamental plants, golf courses and pasture due to the
accessibility and quality in an urban setting (Thompson and Henderson 1998). In

33

addition, some land owners enjoy viewing elk from the comfort of their homes, providing
artificial feed. Non-lethal interactions with humans teach elk that security corresponds
with urban settings. Subsequently, elk seek the refuge of urban areas to increase their
reproductive fitness (Thompson and Henderson 1998). Therefore, the selection for
developed-low, developed-open and open/forage is evidence of habituation. In addition,
the avoidance of the forest resource variable could also be an indication of security in this
study area, displaying reduced need to seek refuge from natural and human predators in a
forest environment (Lee and Miller 2003, Anderson et al. 2005, Cleveland et al. 2012).
I-90 and highway 202 structures were expected to be avoided in all categories.
Out of all roads, highways and interstates are found to have the most influential effects on
wildlife species (Forman et al. 2002, Gagnon et al. 2007a, Fahrig and Rytwinski 2009).
Road effect zones increase as traffic volume and size of road increases (Forman et al.
2002, Dodd et al. 2007, Gagnon et al. 2007b, 2007a). Results found in the road intensity
category supported previous research results of elk-highway interactions, showing
avoidance of medium (Hwy 202) and high (I-90) intensity roads. Contrarily, within the
land use category elk selected for developed-high, which included I-90 and Highway 202
structures. Reasons for such selection could be because this classification included many
different land use types which could have influenced the selection of the variable as a
whole.
Few previous studies have addressed road interactions with human habituated
elk, therefore it is not fully understood how habituated elk respond to roads (Rost and
Bailey 1979, Gagnon et al. 2007a). Low intensity roads have reduced effects on elk than
medium and high intensity roads. Most research is done on elk-road interactions in areas
with low intensity roads, and these studies have shown smaller road effect zones of 200
meters than what is found with high intensity roads like highways (Gagnon et al. 2007a).

34

In contrast, this study found that elk in North Bend select for low intensity roads. The
difference in this result is perhaps explained by the fact that most studies on elk-road
interactions are with non-human habituated elk. The more habituated elk in this study
area may have become familiar to low intensity roads due to the overall abundance of
such roads in the project area and likelihood of interaction (Thompson and Henderson
1998, Walter et al. 2010). Low intensity roads were found to intersect every home range
in this study. The selection for low intensity could be an indication of human habituation
and reduced wariness near such road types or an artifact of the correspondence between
elk home ranges and the valley bottom where roads are more numerous that they are
throughout the peripheral higher elevations of the project area.
The evaluation of space use in relation to I-90 at different distance bands found
that elk prefer distances between 50 meters and 1,000 meters, avoiding distances less
than 50 meters and greater than 1,000 meters. Selection against distances less than 50
meters was as expected however; selection for distances less than 450 meters was not as
expected. A literature review of ungulate interactions with roads by Gagnon and
colleagues found that elk were affected by highways at distances up to 425 meters,
therefore it was expected that elk would select against distances less than 425 meters
(Gagnon et al. 2007a). This affected distance is termed the “road effect zone”. Therefore,
the road effect zone of 50 m for these human habituated elk is far less than what is found
in the literature. Therefore, these habituated elk may be less affected by highways than
their non-human habituated counterparts. Moreover, available open areas and other
preferred habitats provide important resources at distances relatively close to I-90 in the
valley floor. Thus the juxtaposition of where transportation planners constructed I-90, the
development of North Bend near the Interstate, and the local topography, likely
influenced selection of elk use between 50 meters and 1,000 meters, avoiding distances

35

greater than 1,000 meters, where topographic relief increases. Therefore, elk movement
could be confined by these steep slopes due to their preference for flat slopes. In addition,
elk were found to select for flat slopes within the topographic position index category,
avoiding all other positions. Exclusive selection of only one topographic position was not
as expected, since elk are capable of utilizing a variety of slope positions. However, this
could be an indication of habitat quality and security located on flat slopes that
correspond with much of the urban and residential valley bottom. Therefore, selection for
flat slopes could also be an artifact of their non-migratory and human habituated
behavior. Migratory elk usually cross a variety of topographic positions during their
seasonal movements (Anderson et al. 2005). Unfortunately, many high intensity roads are
built on flat slopes, as such for I-90 and Hwy 202 in the project area, which could be an
explanation for the high collision rates in this area. Therefore, if these elk choose to cross
I-90 it may be in topographically flat areas.
Riparian habitat at distances greater than 450 meters from I-90 was selected by
elk, avoiding this cover type at distances less than 450 meters. This was not as expected
since riparian habitat is known to be utilized as corridors for movement (Arizona Game
and Fish Department 2011). It was expected that riparian habitat would be selected for at
all distances. Two forks of the Snoqualmie River flow through the study area, one
following right along I-90 and another residing at greater distances, elk may be choosing
to utilize the riparian corridor farther away from I-90. Their preferential use of riparian
habitat at distances greater than 450 meters could indicate that they like to utilize such
corridors and cover away from I-90. Therefore, elk avoidance of both available and
known preferred habitat type adjacent to I-90, demonstrates I-90’s influence on elk
resource use.

36

However, two individuals, elk 351 and 324 were found to move along the South
Fork Snoqualmie River corridor near I-90, crossing under I-90 to utilize resources located
on the southern side. These two elk were the only individuals that both crossed I-90 and
spent abundant time on both sides of I-90. Evidence obtained from game cameras
deployed by WSDOT for habitat connectivity research at bridges along I-90 that cross
riparian habitat, have captured elk utilizing these structures to pass safely under I-90
(Figure 11). One of the I-90 bridges along the South Fork Snoqualmie River is also
located along elk 324’s and 351’s movement paths. Therefore, these two individuals
could be using this structure to pass safely under I-90 (Figure 11 A. and B). Elk 324 was
the only elk that had core use areas on both sides of I-90. Within the project area, I-90
crossed the South Fork Snoqualmie River several times, creating several opportunities for
safe passages at bridge structures. Previous literature shows that bridges are preferred
crossing structures used by ungulates like elk; however in this study, only two individuals
utilized these structures (Forman et al. 2002, Beckmann et al. 2010). This analysis of elk
351 and 324 home ranges, as well as, game camera images obtained at bridge locations
along the South Fork Snoqualmie River, provide proof that elk utilize riparian habitat and
bridge structures to cross I-90 safely if they chose to move close enough to I-90, however
most elk did not despite available structures and habitat.
Conclusions
Roads can significantly impact wildlife through a variety of mechanisms. Prior to
this study, home ranges and road interactions of elk located around North Bend was
generally undocumented. This research informed where elk establish home ranges in
relation to I-90 and how elk utilize resources in the North Bend area. Since this herd is
human habituated, results from this study offer insight into previously under-researched
aspects of the behavior of human habituated elk and their road interactions. Most

37

previous research on elk-highway interactions were conducted with non-human
habituated elk which behave differently than their habituated counterparts. Results from
this study supports previous research showing habituated ungulate tendencies for why
certain resources were chosen or avoided. Interactions with I-90 were different than what
is commonly found in the literature, with elk utilizing space fairly close to I-90, but at
distances greater than 50 meters. However, few elk chose to cross the interstate to utilize
resources located on the opposite side, riparian habitat was generally avoided at distances
close to I-90 and high intensity roads were avoided, suggesting I-90 may be a partial
barrier to elk movement. Those elk that crossed followed a riparian corridor, most likely
utilizing a bridge structure to pass safely under I-90.
Despite the conservative measures taken to analyze home ranges and resource
selection, it must be noted that there were some limitations with the data. The variety in
collar schedules between elk and the limited number of fixes per day for some individuals
gave a less than complete view of daily movements. Low fix rates of some collars could
have been a product of local satellite blocked by topographic features or vegetation,
potentially biasing the analysis of selection. These data were accrued during the early
stages of collaboration, when funds and staffing were limited. Collars were scheduled to
maximize battery life and deployment time by limiting the number of transmissions each
day. Currently, more elk are being collared, new collars have been purchased and collar
schedules are improving. Therefore, the quality of data gathered for future analysis will
likely improve the accuracy to detect fine scale movements of these elk. Regardless of
limitations with the coordinate data, this study is one of only a few that has researched
human habituated elk interactions with high traffic interstates. In addition, it is a great
example of interagency collaboration.

38

Suggestions for future research would include temporally analyzing home ranges
and space use for trends, which could be used to predict if seasonality influences elk
movement in relation to I-90. Secondly, increasing the sample size (collaring more elk)
and standardizing collar schedules would improve the accuracy of analyses on elk
crossings of I-90. Lastly, research should be implemented on how selections of resources
are correlated. A more in-depth resource selection function using utilization distributions
could give a more detailed depiction of elk spaces use.
In conclusion, this research brings insight into how human habituated elk
respond to I-90. By understanding how elk respond to high traffic interstates like I-90 and
utilize resources and space adjacent to this high-volume interstate, transportation planners
and wildlife managers gain invaluable information to better manage for connectivity and
to reduce wildlife-vehicle collisions. It appears that for many elk in the population, I-90 is
a barrier to elk movement; however two individuals followed a riparian corridor and
crossed I-90 safely, mostly at a bridge structure. Understanding habitat selection
combined with existing knowledge of riparian habitat corridors used by wildlife can
pinpoint linkages and opportunities for safe crossings where bridges are located (Arizona
Game and Fish Department 2011). Several authors stress the need to identify linkages
across barriers and maintaining connectivity between preferred resources when placing
crossing structures (White and Ernst 2004, Singleton et al. 2004, Kindall and van Manen
2007). In addition, studies have found that road mortality sites and road crossings by
wildlife occur near preferred resources (Cain et al. 2003). By ensuring connectivity
between existing bridges where I-90 crosses riparian areas, costs associated with
implementation of crossing structures can be avoided while ensuring connectivity.
Construction of barrier fencing is a measure that can be taken to prevent elk from
crossing over the surface of I-90 and function to direct them to safe crossing

39

opportunities, preventing wildlife-vehicle collisions and ensuring connectivity. As North
Bend becomes more developed, resources for elk may become scarce and fragmented. If
such effects occur, elk may be forced to increase the size of their home ranges, and thus
an increase in elk/human interactions is expected. Therefore, human wildlife conflicts
may increase in the area of North Bend. Further research is recommended to ascertain
where permeability for elk in landscapes adjacent to high-traffic interstates exist, in order
to provide safe movement of animals between resources and to mitigate for associated
negative road effects.

40

Chapter 3 – Conclusions and Management Implications

Negative effects from roads are evident throughout many natural systems.
Habitat fragmentation is among the most severe of these effects, with some wildlife
species experiencing consequences on population viability. Transportation agencies must
confront complex issues of how roads affect natural systems, while simultaneously
creating safe transportation corridors for humans. Mitigation for fragmentation effects is
a growing priority among transportation agencies as the importance of maintaining
connected landscapes becomes recognized by road planners. Traditionally, road planners
used data from wildlife-vehicle collisions and carcass removals as the basis for mitigation
decisions. However, this data cannot be used alone to understand the larger landscapelevel effects of roads. With the advancement in wildlife tracking technologies and
methods, transportation agencies and road ecologists are more equipped than ever to fully
understand ecological impacts of roads. By analyzing wildlife movements, road planners
and road ecologists can comprehensively understand the effects of roads which are not
apparent when using wildlife-vehicle collision and carcass removal data alone.
Knowledge of how human habituated elk respond to I-90 in the North Bend, WA
area was largely undocumented prior to this study. Overall, knowledge of human
habituated elk is lacking in the greater body of literature, let alone interactions with and
response to high volume interstates. Elk herds habituated to human-dominated
environments respond to human infrastructure, especially developed structures like roads
and developed spaces differently than their non-habituated counterparts. In light of
continued human population growth and development in many regions, it is important to
understand how habituated elk respond to high volume interstates and developed areas, to
ensure appropriate management.

41

This research studied how human habituated elk responded to I-90 by analyzing
home range establishment and resource selection in a developed area. A majority of the
home ranges were established away from I-90 with few individuals crossing I-90 to
access resources on the other side, suggesting that I-90 may be a partial barrier to their
movement. Based on the spatial location of home ranges in relation to I-90 one can infer
that I-90 does in fact influence elk movement and behavior up to at least 50 meters.
However, to fully understand the relationship between I-90’s influence and resource
allocation, further multivariate tests are recommended. Camera evidence showed that
utilization of riparian corridors under bridge structures by elk, provided safe passages
under I-90 and informs efforts to reduce elk-vehicle collisions while ensuring
connectivity. Several authors stress the need to identify linkages across barriers and
between preferred resources when applying mitigation techniques (White and Ernst 2004,
Singleton et al. 2004, Kindall and van Manen 2007). By using existing bridges where I90 crosses riparian areas, transportation planners can reduce costs associated with the
implementation of creating crossing structures while ensuring connectivity. Additionally,
constructing barrier fencing in strategic locations to prevent elk from crossing I-90 can
direct them to these safe crossing opportunities. Although road ecologists and planners
have gained substantial knowledge about mitigation actions, further research to
understand what constitutes connectivity between resources is necessary, as well as site
specific information on how to provide safe crossing opportunities across high-volume
interstates.
Recommendations for further research include the following:
1. Continued monitoring of elk movement with these suggested changes:


Increase Sample Sizes by Collaring Additional Elk

42



Improve Collar Schedules to Obtain More Frequent Locations



Analyze Elk Crossings Spatially and Temporally

As we had access to data on only 10 elk for this study, a larger sample size would
improve future analyses. With a larger sample size a more accurate depiction of how I-90
affects elk at a population level can be done. With more elk collared and collar schedules
improved, road ecologists can improve their understanding of crossing behavior.
Currently, this dataset does not provide enough accounts of elk crossing to conduct a
detailed analysis. With additional data on elk crossings, road ecologists can understand
both temporal and spatial patterns of when and where elk cross I-90. Collar schedules
will need to be improved if such analyses are to be conducted. Currently, collar schedules
receive fixes too infrequently to get precise data of when crossings actually occurred.
However, schedules are being improved so that they receive fixes at higher frequency and
at standardized schedules.
2. Analysis of bridge structures:


Improve Accessibility To and Connectivity Between Bridges.



Implement Passage Assessment System (PAS)

Elk are selective when utilizing structures to pass safely under roadways. For
passage, open span bridges are preferred, but there are things that can prevent the elk
from utilizing them. The surrounding environment could prevent the utilization of bridges
if conditions conducive to connectivity do not exist. Further spatial analysis could
evaluate the surrounding environment for potential barriers that might prevent the elk
from utilizing otherwise available structures. In addition, connectivity between structures
should be evaluated if mitigation measures such as fencing are to be implemented. If
structures are inaccessible or connectivity between structures is highly fragmented,
fencing could potentially increase the barrier effects associated with I-90 by inhibiting

43

crossings at grade. (McCollister and Manen 2010). Lastly, implementing PAS, a
“Passage Assessment System” developed to assess the permeability of existing structures
for terrestrial wildlife by Julia Kintsch and Patricia Cramer (2011), can be used to
determine if existing bridges in the North Bend area are attractive to and accessible to
elk. Rating each structure can inform improvement actions necessary to make the
structure more suitable for elk.
3. Improved Resource Selection Analysis


Conduct Additional Multivariate Analysis Using Utilization Distributions (UDs)

The resource selection analysis implemented in this study was fairly straightforward,
analyzing resources for selection at an individual level. Performing a full resource
selection function can bring insight into how resources influence the selection of certain
resources. A full resource selection function can pin point what combination of resources
are most preferred by elk.
4. Ensure good management of riparian corridors near and adjacent to bridges:
Riparian habitats are known corridors for wildlife movement, however within this
study elk disproportionally selected for habitat away from I-90. However, game cameras
have caught images of elk utilizing bridge structures along riparian habitat to cross safely
under I-90. Therefore, these areas should be managed appropriately to ensure that
excellent habitat quality exists around bridges now and into the future, providing
connectivity for elk and other wildlife.
Interdisciplinary Effort
Society is increasingly faced with complex environmental issues that require
dynamic and thoughtful solutions. Many environmental problems today are inter-tangled

44

between balancing our need to protect the environment with the growing human demands
on natural systems. Humans are no longer able to perceive ourselves and our activities
separate from the environment; we are now experiencing feedback from our past actions.
Many of our roads were built long before we knew their environmental impact; therefore
a lot of our current management decisions are to rectify that damage. As scientists and
transportation managers seek to understand ways to mitigate current or past damage and
reduce future impacts of roads, it has become critical that the field of road ecology
“quantifies the ecological effects of roads, with the ultimate goal of avoiding, minimizing
and compensating for their negative impacts on individuals, populations, communities,
and ecosystems” (van der Ree et al. 2011). To meet these objectives it takes an
interdisciplinary effort with many professionals from a variety of backgrounds. The
collaboration between biologists, road planners and structural engineers is essential in
planning for mitigation measures to minimize the negative effects of roads. Without this
collaborative approach between multiple disciplines mitigation would fail to meet all
objectives. Within an interdisciplinary framework, mitigation can ensure that wildlife, the
environment and human structures are resilient to future change.
In Washington State, the problem with elk-vehicle collisions in the North Bend
area requires a dynamic solution where driver safety is increased while ensuring
connectivity of elk populations living near I-90. To address this problem, we took a
dynamic approach to this applied research question, combining the disciplines of road
ecology, road planning, and landscape ecology with methods commonly found in wildlife
management. This research involved a collaborative effort between WSDOT and the
USVEMG to tackle a complex, multidimensional problem. In an era of budget cutbacks
and ever-growing natural resource management issues, collaboration between agencies,
non-governmental groups, academia and local citizen science groups is necessary. It took

45

the collaborative effort between WSDOT and USVEMG to research elk movement since
neither group alone had the resources to conduct this research. It is these collaborative
efforts that should continue and be enriched in other regions facing similar budget
shortfalls and decreased funding. Wildlife movement research used to analyze wildlifehighway interactions is costly, but fortunately collaboration is a solution.

46

Figure 1. Map of Washington State and defined project area around North Bend, WA
(area=363km2).

47

Table 1. Number of elk carcasses removed and number of elk-vehicle collisions along I90 and Hwy 202 within the project area. Data Source: the Washington State Traffic Data
base.
Year
Elk Carcasses Removed
Elk-Vehicle Collision
7
NA
2007
10
NA
2008
36*
18*
2009
10*
9*
2010
16*
9*
2011
*Note the difference between elk carcasses removed and elk-vehicle collisions. Collision
records are recorded when an officer is present at the accident and require a minimum of
$750 in property damage or a human injury. Collision records are fewer because not all
collisions with elk are reported. Elk-vehicle collisions were not tracked separate from
other wildlife-vehicle collisions until 2009.

48

Table 2. Home range (95% contour, in km2) and core use area (50% contour, in km2) for
elk in the vicinity of North Bend, WA.
Elk
Number
50% 95%
Fixes Per Day+
1.28
7.23
5
601_2
0.74
5.02
6
324
1.39
5.47
6
341
0.41
4.60
6
351
3.13 10.14
6
339
0.79
4.13
7
1326
0.98
4.73
7
337
1.19
7.23
7
601_1
3.35 14.03
12
1550
3.81 22.53
29
3870
*Subsample for resource selection analysis.
+
Reference Appendix B for collar schedule.

Total Number of
Fixes
267
742
1933
1050
1300
1176
149
1445
1261
14119, 1176*

Start Date
12/10/2011
8/15/2010
2/14/2011
2/18/2012
3/11/2011
4/12/2012
4/9/2010
4/7/10
3/11/2011
3/27/2011

End Date
4/25/2012
3/30/2011
4/11/2012
9/18/2012
4/12/2012
8/22/2012
6/28/2010
2/4/2011
4/18/2012
8/3/2012

49

Table 3. Definitions of resource variables found in the project area, North Bend, WA.
Category
Land Use

Variable
Developed-High

Open Water
Low

Definition
High traffic roads, I-90, highway 202, North Bend
Way, commercial development, quarries, mines, and
gravel pits.
Residential roads, subdivisions, rural houses.
City parks and recreational fields.
North pacific Douglas-fir, western hemlock, spruce,
and silver fir forest.
Pasture, lawns, and successional fields.
North pacific lowland riparian forest and shrubland.
North pacific bog, shrub swamp, and hardwoodconifer swamp.
Lakes, ponds, and open bodies of water.
< 25 mph

Medium
High

35- 45 mph
> 55 mph

Developed-Low
Developed-Open
Forest
Open/Forage
Riparian
Wetland

Road
Intensity

50

Figure 2. Elk home ranges (95% contour, in km2) of ten elk in North Bend, WA
determined from a Brownian Bridge Movement Model.

51

Figure 3. Overlapping elk home ranges (95% contour, in km2) of ten elk
identified as four groups in North Bend, WA determined from a Brownian
Bridge Movement Model.

52

Table 4. Land use composition for the study area and elk home ranges. Core use area
50% contour and home range 95% contour.

Class
Developed-High
Developed-Low
Developed-Open
Forest
Open/Forage
Riparian
Wetland
Open Water
Total

Study Area
2.49%
3.14%
0.82%
87.22%
1.22%
2.03%
2.72%
0.38%

Home Ranges
50%
3.51%
8.08%
10.93%
48.73%
23.22%
5.04%
0.46%
0.03%

95%
3.68%
9.54%
8.92%
53.39%
18.37%
5.70%
0.36%
0.04%

100.00%

100.00%

100.00%

53

Land Use Composition
100.00%
90.00%
80.00%
70.00%
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%

Study Area
50%
95%

Figure 4. Composition of Land Use within the project area, North Bend, WA.

54

Figure 5. Elk home ranges (50% contour, in km2) of ten elk in North Bend, WA
determined from a Brownian Bridge Movement Model.

55

Table 5. Estimated resource selection log-likelihood chi-square test statistics for elk in
North Bend, WA.
Category
Land Use
Road Intensity
Distance to I-90
Distance to I-90:
Riparian
Topographic Position
Index

11226.90
153.18
28164.64

Df
70
40
40

p-value
0.001
0.001
0.001

642.41

40

0.001

14778.61

50

0.001

56

Table 6. Estimated resource selection indices for elk in North Bend, WA.
= estimated
habitat selection ratio,
) = standard error of selection ratio,
and
are
Bonferroni -adjusted 95% lower and upper confidence intervals.

Category
Land Use

Road
Intensity

I-90
Distances

I-90
Distances:
Riparian

Topographic
Position
Index

Bonferroni
Confidence

Interval

)

Variable
DevelopedHigh
DevelopedLow
DevelopedOpen
Forest
Open/Forage
Riparian
Wetland
Open Water

Selection
+
+

1.48
3.04

0.07
0.09

1.28
2.79

1.68
3.29

+
+
+
-

10.91
0.61
15.11
2.81
0.13
0.10

0.34
0.01
0.31
0.11
0.02
0.05

9.99
0.60
14.26
2.51
0.07
0.00*

11.84
0.63
15.95
3.12
0.19
0.24

Low
Medium
High

+
-

1.09
0.44
0.10

0.01
0.12
0.04

1.08
0.15
0.01

1.11
0.73
0.19

<50 m
50-250 m
250-450 m
450-1000 m
>1000 m

+
+
+
-

0.22
2.10
2.60
2.91
0.67

0.05
0.10
0.11
0.06
0.01

0.08
1.86
2.34
2.76
0.65

0.35
2.34
2.87
3.06
0.69

<50 m
50-250 m
250-450 m
450-1000 m
>1000 m

+
+

0.04
0.45
0.62
2.06
1.30

0.04
0.06
0.11
0.17
0.03

0.00*
0.29
0.35
1.62
1.22

0.13
0.61
0.88
2.49
1.37

Ridge

-

0.31

0.01

0.28

0.35

Upper Slope
0.74
0.03
0.66
Middle Slope 0.53
0.01
0.49
Flat Slope
+
3.25
0.02
3.19
Lower Slope 0.35
0.02
0.31
Valley
0.28
0.01
0.25
+ Significant selection above what would be expected by chance.
- Significant selection against what would be expected by chance.
*A zero replaced a negative value, as a proportion cannot take a negative value.

0.82
0.56
3.31
0.39
0.32

57

Land Use
18
16

Selection Ratio

14

Developed-High
Developed-Low

12

Developed-Open

10

Forest

8

Open/Forage

6

Riparian

4

Wetland

2

Open Water

0

Figure 6. Bonferroni 95% confidence intervals (CI) for selection ratios ( ) of
land use variables by elk in the project area, North Bend, WA. When 1<CI
selection occurred, 1>CI avoidance occurred and when 1 is found within CI
neither selection or avoidance occurred.

58

Road Intensity
1.2

Selection Ratio

1
0.8
Low
0.6

Medium
High

0.4
0.2
0

Figure 7. Bonferroni 95% confidence intervals (CI) for selection ratios ( ) of
road intensity variables by elk in the project area, North Bend, WA. When 1<CI
selection occurred, 1>CI avoidance occurred and when 1 is found within CI
neither selection or avoidance occurred.

59

Distance from I-90
3.5

Selection Ratio

3
2.5
2

<50 m
50-250 m
250-450 m

1.5
1

450-1,000 m
>1,000 m

0.5
0

Figure 8. Bonferroni 95% confidence intervals (CI) for selection ratios ( ) of
distances from I-90 by elk in the project area, North Bend, WA. When 1<CI
selection occurred, 1>CI avoidance occurred and when 1 is found within CI
neither selection or avoidance occurred.

60

I-90 Distance: Riparian Habitat
3

Selection Ratio

2.5
2

<50 m
50-250 m

1.5

450-250 m
450-1,000 m

1

>1,000 m

0.5
0

Figure 9. Bonferroni 95% confidence intervals (CI) for selection ratios ( ) of
riparian habitat at different distances from I-90 by elk in the project area, North
Bend, WA. When 1<CI selection occurred, 1>CI avoidance occurred and when 1
is found within CI neither selection or avoidance occurred.

61

Topographic Position Index Selection
3.5

Selection Ratio

3
2.5

Ridge
Upper Slope

2
1.5

Middle Slope
Flat Slope
Lower Slope

1

Valley

0.5
0

Figure 10. Bonferroni 95% confidence intervals CI for selection ratios ( ) of
topographic positions by elk in the project area, North Bend, WA. When 1<CI
selection occurred, 1>CI avoidance occurred and when 1 is found within CI
neither selection or avoidance occurred.

62

A. MP 31.6

B. MP 31.6 West Bank

C. MP 39 West Bank

D. MP 38 East Bank

Figure 11. Elk captured by Reconyx game cameras utilizing riparian habitat and bridge
structures to cross safely under I-90 near North Bend, WA.

63

References
Abramsky, Z. 1978. Small mammal community ecology. Oecologia 34:113-123.
Anderson, D. P., J. D. Forester, M. G. Turner, J. L. Frair, E. H. Merrill, D. Fortin, J. S.
Mao, and M. S. Boyce. 2005. Factors influencing female home range sizes in elk
(Cervus elaphus) in North American landscapes. Landscape Ecology 20:257–
271.
Andrén, H. 1994. Effects of Habitat Fragmentation on Birds and Mammals in Landscapes
with Different Proportions of Suitable Habitat: A Review. Oikos 71:355.
Aresco, R.A. and S. J. Richards. 1999. Global amphibian declines: A problem in applied
ecology. Annual Review. Ecological Systems 30:133-165.
Arizona Game and Fish Department. 2011. The Coconino County Wildlife Connectivity
Assessment: report on Stakeholder Input.
Bashore, T. L., W. M. Tzilkowski, and E. D. Bellis. 1985. Analysis of Deer-Vehicle
Collision Sites in Pennsylvania. The Journal of Wildlife Management 49:769–
774.
Becker, S. A. P. F. Frame, D. Martorello, and E. Krausz. 2013. Washington Gray Wolf
Conservation and Management 2012 Annual Report. Pages WA-1 to WA-16 in
U.S. Fish and Wildlife Service Rocky Mountain Wolf Program 2012 Annual
Report. USFWS, Ecological Services, 585 Shepard Way, Helena Montana,
59601.
Beckmann, J. P., A. P. Clevenger, M. Huijser, and J. Hilty. 2010. Safe Passages:
Highways, Wildlife, and Habitat Connectivity. Island Press.
Bouchard, J., A. T. Ford, F. E. Eigenbrod, and L. Fahrig. 2009. Behavioral response of
northern leopard frogs (Rana pipiens) to roads and traffic: implication for
population persistence. Ecology and Society 14:23.
Bullard, F. 1991. Estimating the home range of an animal: a Brownian Bridge approach.

64

Thesis, University of North Carolina at Chapel Hill, Chapel Hill
Burt, W.H. 1943. Territoriality and home range concepts as applied to mammals. Journal
of Mammalogy 24:346-352.

Cain, A., V. Tuovila, D. Hewitt, and M. Tewes. 2003. Effects of a highway and
mitigation projects on bobcats in Southern Texas. Biological Conservation
114:189–197.
Cleveland, S. M., M. Hebblewhite, M. Thompson, and R. Henderson. 2012. Linking Elk
movement and resource selection to hunting pressure in a heterogeneous
landscape. Wildlife Society Bulletin 36:658–668.
Clevenger, A. P., and N. Waltho. 2003. Long-term, year-round monitoring of wildlife
crossing structures and the importance of temporal and spatial variability in
performance studies. Pages 293–302.
Coffin, A. W. 2007. From roadkill to road ecology: A review of the ecological effects of
roads. Journal of Transport Geography 15:396–406.
Cole, E. K., M. D. Pope, and R. G. Anthony. 1997. Effects of Road Management on
Movement and Survival of Roosevelt Elk. Journal of Wildlife Management
61:1115–1126.
Couch, L. K. 1935. Chronological Data on Elk Introduction into Oregon and Washington.
The Murrelet 16:2.
Coulombe, M. L., A. Masse, and S. D. Cote. 2006. Quantification and Accuracy of
Activity Data Measured with VHF and GPS Telemetry. Wildlife Society Bulletin
34:81–92.
Crooks, K. R., and M. Sanjayan. 2006. Connectivity Conservation. Cambridge University
Press.

65

Dodd, N. L., J. W. Gagnon, S. Boe, and R. E. Schweinsburg. 2007. Assessment of Elk
Highway Permeability by Using Global Positioning System Telemetry. The
Journal of Wildlife Management 71:1107–1117.
Fahrig, L. 1997. Relative Effects of Habitat Loss and Fragmentation on Population
Extinction. The Journal of Wildlife Management 61:603.
Fahrig, L., and T. Rytwinski. 2009. Effects of Roads on Animal Abundance: an
Empirical Review and Synthesis. Ecology and Society 14:21.
Federal Highway Administration. 2002a. Our Nation’s Highways. Publication FHWAPL-00-014. Washington, D.C.: U.S. Department of Transportation.
Fieberg, J. 2007. Kernel density estimators of home range: smoothing and the
autocorrelation red herring. Ecology 88:1059-1066.
Fieberg, J. Matthipoulos, J., Hebblewhite, M., Boyce, M.S. and Frair, J. L. 2010.
Correlation and studies of habitat selection: problem, red herring or
opportunity? Philosophical Transactions of the Royal Society B:
Biological Sciences 365:2233-2244.
Ford, R. G. 1983. Home range in a patchy environment: optimal foraging predictions.
American Zoologist 23:315-326.
Forman, R. T. T., and L. E. Alexander. 1998. Roads and their major ecological effects.
Annual review of ecology and systematics.
Forman, R. T. T., D. Sperling, J. A. Bissonette, A. P. Clevenger, C. D. Cutshall, V. H.
Dale, L. Fahrig, R. L. France, C. R. Goldman, K. Heanue, J. Jones, F. Swanson,
T. Turrentine, and T. C. Winter. 2002. Road Ecology: Science And Solutions.
Island Press.
Franklin, William L.; Lieb, James W. 1979. The social organization of a sedentary
population of North American elk: a model for understanding other populations.
In: Boyce, M. S.; Hayden-Wing, L. D., eds. North American elk: ecology,

66

behavior and management: Proceedings of a symposium; 1978 April 3-5;
Laramie, WY. Laramie, WY: University of Wyoming: 185-198
Frair, J. L., J. Fieberg, M. Hebblewhite, F. Cagnacci, N. J. DeCesare, and L. Pedrotti.
2010. Resolving issues of imprecise and habitat-biased locations in ecological
analyses using GPS telemetry data. Philosophical Transactions of the Royal
Society B: Biological Sciences 365:2187–2200.
Gagnon, J. W., R. E. Schweinsburg, and N. L. Dodd. 2007a. Effects of Roadway Traffic
on Wild Ungulates: A Review of the Literature and Case Study of Elk in
Arizona.
Gagnon, J. W., T. C. Theimer, N. L. Dodd, S. Boe, and R. E. Schweinsburg. 2007b.
Traffic Volume Alters Elk Distribution and Highway Crossings in Arizona.
Journal of Wildlife Management 71:2318.
Gibbs, J. P., and W. G. Shriver. 2002. Estimating the effects of road mortality on turtle
populations. Conservation Biology 16:1647–1652.
Grover, K. E., and M. J. Thompson. 1986. Factors Influencing Spring Feeding Site
Selection by Elk in the Elkhorn Mountains, Montana. The Journal of Wildlife
Management 50:466.
Gunson, K. E., B. Chruszcz, and A. P. Clevenger. 2005. What features of the landscape
and highway influence ungulate vehicle collisions in the watersheds of the
Central Canadian Rocky mountains: a fine-scale perspective? Pages 543–555
International Conference on Ecology and Transportation.
Gunther, K. A., M. J. Biel, and H. L. Robison. 1998. Factors influencing the frequency of
road-killed wildlife in Yellowstone National Park. International Conference on
Wildlife Ecology and Transportation.

67

Halfwerk, W., L. J. M. Holleman, C. M. Lessells, and H. Slabbekoorn. 2011. Negative
impact of traffic noise on avian reproductive success. Journal of Applied Ecology
48:210–219.
Hanski, I. 1999. Habitat Connectivity, Habitat Continuity, and Metapopulations in
Dynamic Landscapes. Oikos 87:209.
Hardy, A., A. P. Clevenger, M. Huijser, and G. Neale. 2003. An overview of methods
and approaches for evaluating the effectiveness of wildlife crossing structures:
emphasizing the science in applied science. Pages 319–330 International
Conference on Ecology and Transportation.
Harrison, S., and E. Bruna. 1999. Habitat fragmentation and large-scale conservation:
what do we know for sure? Ecography 22:225-232.
Hemson, G. Johnson, P., South, A., Kenward, R., Ripley, R. and Mcdonald, D. 2005. Are
kernels the mustard? Data from global positioning
system (GPS) problems for kernel home-range analyses
with least squares cross-validation. Journal of Animal
ecology 74:455-463.
Highway Wildling. 2012. Accessed online at http://highwaywilding.org/media.php
Horne, J. S., E. O. Garton, S. M. Krone, and J. S. Lewis. 2007. Analyzing Animal
Movements Using Brownian Bridges. Ecology 88:2354–2363.
Hujser, M. P., K. J. S. Paul, L. Oechsli, R. Ament, A. P. Clevenger, and A. Ford. 2008.
Wildlife-vehicle collision and crossing mitigation plan for Hwy 93S in
Kootenay and Banff National Park and the roads in and around
Radium Hot Springs. Report 4W1929 B, Western Transportation
Institute – Montana State University, Bozeman, Montana, USA.
Huijser, M. P., P. T. McGowen, J. Fuller, A. Hardy, and A. Kociolek. 2007. WildlifeVehicle Collision Reduction Study: Report to Congress.

68

Jaarsma, C. F., F. van Langevelde, and H. Botma. 2006. Flattened fauna and mitigation:
Traffic victims related to road, traffic, vehicle, and species characteristics.
Transportation Research Part D: Transport and Environment 11:264–276.
Jackson, N. D., and L. Fahrig. 2011. Relative effects of road mortality and decreased
connectivity on population genetic diversity. Biological Conservation 144:3143–
3148.
Jaeger, J. A. G., J. Bowman, J. Brennan, L. Fahrig, D. Bert, J. Bouchard, N.
Charbonneau, K. Frank, B. Gruber, and K. T. von Toschanowitz. 2005.
Predicting when animal populations are at risk from roads: an interactive model
of road avoidance behavior. Ecological Modeling 185:329–348.
Jenness, J. 2006. Topographic Position Index (tpi_jen.avx) extension for ArcView 3.x, v.
1.3a. Jenness Enterprises. Available at:
http://www.jennessent.com/arcview/tpi.htm.
Kaufmann, J. H. 1962. Ecology and social behavior of the coati, Nasua nirica on Barro
Colorado Island Panama. University of California Publications in Zoology,
60:95-222
Kaszynski, W. 2000. The American Highway: The History and Culture of Roads in the
United States. McFarland.
Kindall, J. L., and F. T. van Manen. 2007. Identifying habitat linkages for black bears in
eastern North Carolina. Journal of Wildlife Management 71:487-495.
Kintsch, J., and P. Cramer. 2011. Permeability of existing structures for terrestrial
wildlife: A passage assessment system. Research Report No. WA-RD 777.1.
Washington State Department of Transportation, Olympia, WA.
Kie, J.G., J. Matthiopoulos, J. Fieberg, R. A. Powell, F. Cagnacci, M. S. Mitchell, J.
Gaillard, and P. R. Moorcroft. 2010. The home-range concept: are traditional
estimators still relevant with modern telemetry technology? Philosophical

69

transactions of the Royal Society B: Biological Sciences 365:2221-2231.
Lee, M. E., and R. Miller. 2003. Managing elk in the wildland-urban interface: attitudes
of Flagstaff, Arizona residents. Wildlife Society Bulletin:185–191.
Lewis, J. S., J. L. Rachlow, J. S. Horne, E. O. Garton, W. L. Wakkinen, J. Hayden, and P.
Zager. 2011. Identifying habitat characteristics to predict highway crossing areas
for black bears within a human-modified landscape. Landscape and Urban
Planning 101:99–107.
Long, R. A., J. S. Begley, and P. MacKay. 2012. Baseline wildlife monitoring at I-90
Snoqualmie Pass East, prior to the installation of wildlife crossing structures.
Final report to the Washington State Department of Transportation, Yakima,
Washington. Western Transportation Institute, Montana State University,
Bozeman, Montana. 270pp.
Manly, B. F., L. McDonald, D. L. Thomas, T. L. McDonald, and W. P. Erickson. 2002.
Resource Selection by Animals: Statistical Design and Analysis for Field
Studies. Springer.
Mastro, L. L., M. R. Conover, and S. N. Frey. 2008. Deer-vehicle collision prevention
techniques. Human Wildlife Interactions Journal 2:80-92.
McCollister, M. F., and F. T. van Manen. 2010. Effectiveness of Wildlife Underpasses
and Fencing to Reduce Wildlife–Vehicle Collisions. Journal of Wildlife
Management 74:1722–1731.
Meyer, W. L., W. Y. Chang, S. S. Germaine, W. M. Vander Haegen, and T. E. Owens.
2008. Ana analysis of deer and elk-vehicle collision sites along state highways in
Washington State. Completion Report, Washington Department of Fish and
Wildlife. Olympia, WA 40p.

70

Moeller, B. J. 2010. Migratory and non-migratory movements and habitat use by female
elk in the Cascade range of Washington. . Masters of Environmental Studies, The
Evergreen State College.
Moen, A. N. 1976. Energy Conservation by White-Tailed Deer in the Winter. Ecology
57:192.
Nielsen, C. K., R. G. Anderson, and M. D. Grund. 2003. Landscape Influences on DeerVehicle Accident Areas in an Urban Environment. The Journal of Wildlife
Management 67:46.
National Oceanic Atmospheric Administration (NOAA) 2012. Accessed online at
http://www.nws.noaa.gov/climate/xmacis.php?wfo=sew
Parks Canada Agency, G. of C. 2012, March 12. Parks Canada - Trans-Canada Highway
Twinning - Banff National Park of Canada - Trans-Canada Highway Twinning.
Parris, K. M., and A. Schneider. 2008. Impacts of traffic noise and traffic volume on
birds of roadside habitats. Ecology and Society 14:29.
Peek, J. M. 2003. Wapiti (Cervus elaphus). In: Feldhamer, George A.; Thompson,
Bruce C.; Chapman, Joseph A., eds. Wild mammals of North America: Biology,
management, and conservation. 2nd ed. Baltimore, MD: The Johns Hopkins
University Press: 877-888.
Reh, W. and A. Seitz. 1990. The influence of land use on the genetic structure of
populations of the common frog Rana temporaria. Biological
Conservation 54:239-249.
Relyea, R.A., Lawrence R.K. and S. Demarais. 2000. Home range of desert mule deer:
testing the body size and habitat productivity hypothesis. Journal of Wildlife
Management 63:739-747.

71

Rost, G. R., and J. A. Bailey. 1979. Distribution of Mule Deer and Elk in Relation to
Roads. The Journal of Wildlife Management 43:634–641.
Row, J.R., G. Blouin-Demers, and P. J. Weatherhead. 2007. Demographic effects of road
mortality in black ratsnakes (Elaphe obsolete). Biology Conservation 137:117124.
Rumar, K. 1990. The basic driver error: late detection. Ergonomics 33:1281–1290.
Samuel, M. D., D. J. Pierce, and E. O. Garton. 1985. Identifying Areas of Concentrated
Use within the Home Range. Journal of Animal Ecology 54:711–719.

Singleton, P. H., W. L. Gaines, and J. F. Lehmkuhl. 2004. Landscape
permeability for grizzly bear movements in Washington and southwestern
British Columbia. Ursus 15:90–103.
Sullivan, B.K. 1981. Observed differences in body temperature and associated behavior
of four snake species. Herpetol 15:245-246.
St. Clair, C. C., and A. Forrest. 2009. Impacts of vehicle traffic on the distribution and
behavior of rutting elk, Cervus elaphus. Behaviour 146:393–413.
Stankowich, T. 2008. Ungulate flight responses to human disturbance: A review and
meta-analysis. Biological Conservation 141:2159–2173.
Steen, D. A. and L. L. Smith. 2006. Road surveys for turtles: consideration of possible
sampling biases. Conservation Biology 1:9-15.
Thompson, M. J., and R. E. Henderson. 1998. Elk Habituation as a Credibility Challenge
for Wildlife Professionals. Wildlife Society Bulletin 26:477–483.
Trombulak, S., and C. Frissell. 2000. Review of Ecological Effects of Roads on
Terrestrial and Aquatic Communities. Conservation Biology 14:18–30.
United States Department of Agriculture (USDA) 2012. Accessed online at
http://datagateway.nrcs.usda.gov/

72

Upper Snoqualmie Valley Elk Management Group (USVEMG). 2010. Upper
Snoqualmie Valley Elk Telemetry Study Cooperative Work Plan 2010. North
Bend, WA.
Van der Ree, R., J. A. G. Jaeger, and A. P. Clevenger. 2011. Effects of roads and traffic
on wildlife populations and landscape function: Road ecology is moving toward
larger scales. Ecology and Society 16:48.
Vaughan, Michael R., M. Kelly, J. A. Trent, and C. Proctor. 2012. Evaluating potential
effects of widening US highway 64 on red wolves, Washington, Tyrrell, and
Dare Counties, North Carolina. National Conference of The Wildlife Society.
Walter, W. D., M. J. Lavelle, J. W. Fischer, T. L. Johnson, S. E. Hygnstrom, and K. C.
VerCauteren. 2010. Management of damage by elk (Cervus elaphus) in North
America: a review. Wildlife Research 37:630–646.
Washington State Department of Transportation (WSDOT). 2008. I-90 Snoqualmie Pass
East: Final environmental impact statement and section 4(f) evaluation. Yakima,
WA.
Washington State Department of Transportation (WSDOT). 2011. Unpublished data from
the Statewide Travel and Collision Data Office. WSDOT headquarters, Olympia,
WA.
Washington Wildlife Habitat Connectivity Working Group (WHCWG). 2010.
Washington Connected Landscapes Project: Statewide Analysis. Washington
Departments of Fish and Wildlife, And Transportation, Olympia, WA.
Weiss, A. D. 2006. Topographic position and landforms analysis. The Nature
Conservancy. Access online. http://www.jennessent.com/downloads/tpi-poster
tnc_18x22.pdf
White, P. A., and M. Ernst. 2004. Second Nature: improving transportation without
putting nature second. Surface Transportation Policy Project, Defenders of

73

Wildlife, www.defenders.org.
White G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic
Press, New York, NY.
Wilkins, K. T. 1982. Highways as barriers to rodent dispersal. Southwest Nat. 27:459460.
Wilkins, K. T., and D. J. Schmidly. 1980. Highway mortality of vertebrates in
southeastern Texas. Texas Journal of Science 4:343-350.
Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in HomerRange studies. Ecology 70:164-168.

74

Appendices
Appendix A. Example of the Brownian Bridge Movement Model R Script for elk 337.
##Set the working directory
##My working directory
directory <- "h:/Collars"
setwd(directory)
##Read a csv file into data frame. This is an example for elk collar 337
tele <- read.csv("Collar337.csv", header = TRUE)
## variable for range id (elk) ##this one is best for trajectory example
range <- "337"
## Get the current range from the data frame
tele.range <- subset(tele, tele$AnimalID == toString(range)) ##I modified this,
tele$RangeID references a field in the dataset for subsetting, yours should be
tele@AnimalID
##Get only the coords
tele.range.xy <data.frame("x"=tele.range$EastingUTM83,"y"=tele.range$NorthingUTM83)
##Need sp to make spatial objects
library(sp)
##Define projection of coords
proj4string <- CRS("+proj=utm +zone=10 +datum=NAD83 +units=m +no_defs
+ellps=GRS80 +towgs84=0,0,0")
##Make SpatialPointsDataFrame
tele.range.spdf <- SpatialPointsDataFrame(tele.range.xy, tele.range, proj4string =
proj4string , match.ID = TRUE)
plot(tele.range.spdf) ##Run this to see your data
##DF is used for a number of things including attaching additional attributes to the
trajectory (activity needed by BRB)
tele.range.df <data.frame("x"=tele.range$EastingUTM83,"y"=tele.range$NorthingUTM83,
"ObsStepMin"=tele.range$ObsStepMin, "ObsDaText"=tele.range$ObsDaText)
##REmoved variables not needed

75

##Set up for Elk data
##Home ranges
##First create a bounding box for a mask grid we will project home ranges to.
##Get the bounding box from subset data from exercise above(tele.range) which
##will be modified to make a region of interest grid for calculating homeranges.
##For home range calculations, some packages require evaluation points (KS) while
others require
##a grid as spatial pixels (adehabitat). In preperation I made several different versions.
##Set the expansion value for the grid and get the bbox
expandValue <- 2500 #This value in meters is used in the calculation
boundingVals <- tele.range.spdf@bbox
##Get the change in x and y and adjust using expansion value
deltaLong <- as.integer(((boundingVals[1,2]) - (boundingVals[1,1])) + (2*
expandValue))
deltaLat <- as.integer(((boundingVals[2,2]) - (boundingVals[2,1])) + (2* expandValue))
##200 meter grid for testing, watch part in BBMM where cell size is set too
gridRes <- 30
gridSizeX <- deltaLong / gridRes
gridSizeY <- deltaLat / gridRes
##Offset the bounding coordinates
boundingVals[2,1] <- boundingVals[2,1] - expandValue
boundingVals[2,2] <- boundingVals[2,2] + expandValue
boundingVals[1,1] <- boundingVals[1,1] - expandValue
boundingVals[1,2] <- boundingVals[1,2] + expandValue
##load raster
library(raster)
##Grid Topology object is basis for sampling grid (offset, cellsize, dim)
gridTopo <- GridTopology((boundingVals[,1]), c(gridRes,gridRes),
c(gridSizeX,gridSizeY))
##Define the projection of the coords
proj4string <- CRS("+proj=utm +zone=10 +datum=NAD83 +units=m +no_defs
+ellps=GRS80 +towgs84=0,0,0")
##Using the Grid Topology create a SpatialGridClass

76

sampGrid <- SpatialGrid(gridTopo, proj4string = proj4string)
##Cast over to SP
sampSP <- as(sampGrid, "SpatialPixels")
##convert the spatialgrid class to a raster
sampRaster <- raster(sampGrid)
##set all the raster values to 1
sampRaster[] <- 1
##Get the center points of the mask raster with values set to 1
evalPoints <- xyFromCell(sampRaster, 1:ncell(sampRaster))
##Here we can see how grid has a buffer around the locations
plot.new()
plot(sampRaster)
points(tele.range.spdf, pch=1, cex=0.5)
##BBMM home range
library(BBMM)
#Run the BBMM using the data frame
BBMM <- brownian.bridge(x=tele.range.df$x, y=tele.range.df$y,
time.lag=tele.range.df$ObsStepMin, area.grid=evalPoints, time.step=10,
location.error=24, max.lag=300)

# Create a data from of x,y,z
BBMM.df <- data.frame("x"=BBMM$x,"y"=BBMM$y,"z"=BBMM$probability)
# Rescale the Probabilities to PDF
#BBMM.df$z <- BBMM.df$z/sum(BBMM.df$z)
##Make a raster from the x, y, z values, watch cell size parameter
tele.range.df.bbmm.raster <- rasterFromXYZ(BBMM.df, res=c(30,30), digits=5)
plot(tele.range.df.bbmm.raster)
library(adehabitatHR)
tele.range.bbmm.px <- as(tele.range.df.bbmm.raster, "SpatialPixelsDataFrame")
tele.range.bbmm.ud <- new("estUD", tele.range.bbmm.px)
tele.range.bbmm.ud@vol = FALSE

77

tele.range.bbmm.ud@h$meth = "BBMM"
##Convert the UD values to volume
tele.range.bbmm.ud.vol <- getvolumeUD(tele.range.bbmm.ud, standardize=TRUE)
##Create a raster object
tele.range.bbmm.ud.vol.raster <- raster(tele.range.bbmm.ud.vol)
tele.range.bbmm.99vol <- getverticeshr(tele.range.bbmm.ud, percent = 99, ida = NULL,
unin = "m", unout = "ha", standardize=TRUE)
tele.range.bbmm.95vol <- getverticeshr(tele.range.bbmm.ud, percent = 95, ida = NULL,
unin = "m", unout = "ha", standardize=TRUE)
tele.range.bbmm.50vol <- getverticeshr(tele.range.bbmm.ud, percent = 50, ida = NULL,
unin = "m", unout = "ha", standardize=TRUE)

##Put the HR, volume, volume contours, trajectory, and points on a plot
plot.new()
breaks <- c(0, 50, 95, 99)
plot(tele.range.bbmm.ud.vol.raster, col=heat.colors(3), breaks=breaks,
interpolate=TRUE, main="Brownian Bridge Movement Model", xlab="Coord X",
ylab="Coord Y", legend.shrink=0.80, legend.args=list(text="UD by Volume (%)",side=4,
font=2, line=2.5, cex=0.8))
plot(tele.range.bbmm.50vol, add=TRUE)
plot(tele.range.bbmm.95vol, add=TRUE)
plot(tele.range.bbmm.99vol, add=TRUE)
points(tele.range.spdf, pch=1, cex=0.5)

##Write out the BBMM raster for external GIS
writeRaster(tele.range.df.bbmm.raster , paste(directory, "/bbmm_", range, ".tif", sep=""),
overwrite=TRUE)
writeRaster(tele.range.bbmm.ud.vol.raster , paste(directory, "/bbmm_", range, "_vol.tif",
sep=""), overwrite=TRUE)
##Write out the
writeOGR(tele.range.bbmm.99vol, ".", paste("bbmm_vol99_", range, sep=""),
driver="ESRI Shapefile",overwrite_layer=TRUE)
writeOGR(tele.range.bbmm.95vol, ".", paste("bbmm_vol95_", range, sep=""),
driver="ESRI Shapefile",overwrite_layer=TRUE)
writeOGR(tele.range.bbmm.50vol, ".", paste("bbmm_vol50_", range, sep=""),
driver="ESRI Shapefile",overwrite_layer=TRUE)

78

Appendix B. Schedules for collared elk in North Bend, WA.
Elk
Number
601_2
324

Fixes Per
Day
5
6

341

6

351

6

339

6

1326
337
601_1
1550
3870

7
7
7
12
29

Collar Schedule
Every 5 hrs
Every 1.5 hrs, 3 times after 12:00am
and 12:00pm
Every 1.5 hrs, 3 times after 12:00am
and 12:00pm
Every 1.5 hrs, 3 times after 12:00am
and 12:00pm
Every 2hrs from 4pm to 12:00am
then every hour from 12:01am to
2:00am
Every 2.5 hrs
Every 2.5 hrs
Every 2.5 hrs
Every 2 hrs
Every 50 mins

79

Appendix C. Land use layer created in ArcGIS 10.

80

Appendix D. Road intensity layer created in ArcGIS 10.

81

Appendix E. I-90 distance band layer created in ArcGIS 10.

82

Appendix F. Riparian habitat at different distance bands from I-90 layer
created in ArcGIS 10.

83

Appendix G. Topographic Position Index layer created in ArcGIS 10.

84