Wildlife-vehicle collision mitigation for safer wildlife movement across highways : State Route 260

              Wildlife-Vehicle Collision
Mitigation for Safer Wildlife
Movement across Highways:
State Route 260
Final Report 603
December 2012
Arizona Department of Transporta􀆟􉽯on
Research Center
Wildlife‐Vehicle Collision Mitigation
for Safer Wildlife Movement across
Highways: State Route 260
Final Report 603
December 2012
Prepared by:
Arizona Game and Fish Department
Research Branch
5000 West Carefree Highway
Phoenix, Arizona 85068
Prepared for:
Arizona Department of Transportation
in cooperation with
U.S. Department of Transportation
Federal Highway Administration
This report was funded in part through grants from the Federal Highway Administration,
U.S. Department of Transportation. The contents of this report reflect the views of the
authors, who are responsible for the facts and the accuracy of the data, and for the use
or adaptation of previously published material, presented herein. The contents do not
necessarily reflect the official views or policies of the Arizona Department of
Transportation or the Federal Highway Administration, U.S. Department of
Transportation. This report does not constitute a standard, specification, or regulation.
Trade or manufacturers’ names that may appear herein are cited only because they are
considered essential to the objectives of the report. The U.S. government and the State
of Arizona do not endorse products or manufacturers.
Technical Report Documentation Page
1. Report No.
FHWA-AZ-12-603
2. Government Accession No. 3. Recipient's Catalog No.
4. Title and Subtitle
Wildlife-Vehicle Collision Mitigation for Safer Wildlife Movement across
Highways: State Route 260
5. Report Date
December 2012
6. Performing Organization Code
7. Authors
Norris L. Dodd, Jeffrey W. Gagnon, Susan Boe, Kari Ogren,
and Raymond E. Schweinsburg
8. Performing Organization Report No.
9. Performing Organization Name and Address
Arizona Game and Fish Department
Research Branch
5000 W. Carefree Highway
Phoenix, AZ 85068
10. Work Unit No.
11. Contract or Grant No.
SPR 000 1(069) 603
12. Sponsoring Agency Name and Address
Research Center
Arizona Department of Transportation
206 S. 17th Avenue MD075R
Phoenix, AZ 85007
13.Type of Report & Period Covered
FINAL REPORT
14. Sponsoring Agency Code
15. Supplementary Notes
Prepared in cooperation with the U.S. Department of Transportation, Federal Highway Administration
16. Abstract
Researchers investigated wildlife-highway relationships in central Arizona from 2002 to 2008 along a 17-mile stretch
of State Route (SR) 260, which is being reconstructed in five phases and will have 11 wildlife underpasses and
6 bridges. Phased reconstruction allowed researchers to use a before-after-control experimental approach to their
research. The objectives of the project were to:
 Assess and compare wildlife use of underpasses (UPs)
 Evaluate highway permeability and wildlife movements among reconstruction classes
 Characterize wildlife-vehicle collision (WVC) patterns and changes with reconstruction
 Assess relationships among traffic volume and WVCs, wildlife crossing patterns, and UP use
 Assess the role of ungulate-proof fencing with WVCs, wildlife UP use, and wildlife permeability
Researchers used video surveillance to assess and compare wildlife use of six UPs, at which 15,134 animals and 11
species were recorded; 67.5 percent crossed through UPs. Modeling found that UP structure type and placement was
the most important factor influencing the probability of successful crossings by elk (Cervus elaphus) and Coues white-tailed
deer (Odocoileus virginianus). Researchers used Global Positioning System (GPS) telemetry tracking of 100 elk
and 13 white-tailed deer to assess and compare permeability. Elk permeability on reconstructed sections was 39
percent lower than controls, while deer permeability was 433 percent higher on reconstructed sections. The elk-vehicle
collision (EVC) rate on fenced reconstructed sections was the same as before-reconstruction levels, but on
unfenced sections the EVC rate was nearly four times higher. In addition to a safer and more environmentally friendly
highway, the economic benefit from reduced EVCs on SR 260 averaged $2 million/year since the completion of
three reconstructed highway sections.
17. Key Words
Cervus elaphus, deer, elk, GPS telemetry, fencing,
highway impact, Odocoileus spp., permeability,
traffic volume, video surveillance, white-tailed deer,
wildlife underpasses, wildlife-vehicle collisions
18. Distribution Statement
This document is available to the
U.S. public through the National
Technical Information Service,
Springfield, Virginia 22161
23. Registrant’s Seal
19. Security Classification
Unclassified
20. Security Classification
Unclassified
21. No. of Pages
128
22. Price
CONTENTS
Executive Summary ...........................................................................................................1
1.0 Introduction ...............................................................................................................5
1.1 Background ........................................................................................................5
1.2 Highway Reconstruction and Study Approach ..................................................7
1.2.1 Phased Construction and Adaptive Management ................................. 7
1.2.2 Experimental Approach ........................................................................ 7
1.2.3 Research Phases.................................................................................... 8
1.3 Research Objectives ...........................................................................................9
1.4 Phases I and II Final Report Summary ............................................................13
1.4.1 Assessment of Wildlife Underpass Use ............................................. 13
1.4.2 Traffic Effects on Elk Underpass Crossings ...................................... 13
1.4.3 Elk Permeability from GPS Telemetry ............................................... 14
1.4.4 Traffic Effects on Elk Highway Crossings ......................................... 14
1.4.5 Role of Ungulate-Proof Fencing ........................................................ 14
1.4.6 Wildlife-Vehicle Collision Relationships ........................................... 15
1.4.7 Economic Benefit of Wildlife Measures ............................................ 16
1.4.8 Conclusion .......................................................................................... 16
1.5 Report Organization .........................................................................................16
2.0 Study Area ...............................................................................................................17
2.1 Reconstructed Sections and Chronology .........................................................18
2.1.1 Preacher Canyon Section .................................................................... 19
2.1.2 Christopher Creek Section .................................................................. 19
2.1.3 Kohl’s Ranch Section ......................................................................... 19
2.2 Natural Setting .................................................................................................20
2.3 Traffic Volume .................................................................................................22
3.0 Evaluation of Factors Influencing Wildlife Use of Highway Underpasses ........25
3.1 Introduction ......................................................................................................25
3.2 Methods............................................................................................................27
3.2.1 Video Surveillance Systems ............................................................... 27
3.2.2 Assessment of Wildlife Use of Underpasses ...................................... 27
3.3 Results ..............................................................................................................33
3.3.1 Wildlife Underpass Use ...................................................................... 33
3.3.2 Factors Influencing Successful Elk Underpass Crossings.................. 35
3.3.3 Factors Predicting Successful White-Tailed Deer Underpass
Crossings ............................................................................................ 36
3.3.4 Influence of Duration of Video Surveillance Monitoring .................. 38
3.3.5 Influence of Season ............................................................................ 39
3.3.6 Influence of Time of Day ................................................................... 41
3.3.7 Wildlife Use of the Indian Gardens Underpass .................................. 41
3.4 Discussion ........................................................................................................42
3.4.1 General Efficacy of Underpasses ....................................................... 43
3.4.2 Influence of Underpass Structural Characteristics ............................. 43
3.4.3 Influence of Duration of Video Surveillance Monitoring .................. 47
3.4.4 Influence of Fencing on Wildlife Underpass Use .............................. 48
4.0 Effectiveness of Passage Structures and Fencing in Minimizing
Wildlife-Vehicle Collisions .....................................................................................51
4.1 Introduction ......................................................................................................51
4.2 Methods............................................................................................................53
4.2.1 Wildlife-Vehicle Collision Tracking .................................................. 53
4.2.2 Comparison of Elk-Vehicle Collision Rates by Highway
Reconstruction Classes ...................................................................... 53
4.2.3 Economic Benefit of Reduced Elk-Vehicle Collisions ...................... 54
4.3 Results ..............................................................................................................55
4.3.1 Wildlife-Vehicle Collision Tracking .................................................. 55
4.3.2 Comparison of Elk-Vehicle Collision Rates by Highway
Reconstruction Classes ...................................................................... 57
4.3.3 Economic Benefit of Reduced Elk-Vehicle Collisions ...................... 59
4.4 Discussion ........................................................................................................61
5.0 Influence of Passage Structures and Their Spacing on Elk
Highway Permeability ............................................................................................65
5.1 Introduction ......................................................................................................65
5.2 Methods............................................................................................................67
5.2.1 Elk Capture and GPS Collars ............................................................. 67
5.2.2 GPS Data Analysis of Elk Movements and Permeability .................. 68
5.2.3 Elk Permeability by Highway Reconstruction Classes ...................... 70
5.2.4 Elk Permeability by Passage Structure Spacing ................................. 70
5.3 Results ..............................................................................................................70
5.3.1 GPS Data Analysis of Elk Movements and Permeability .................. 70
5.3.2 Elk Permeability by Highway Reconstruction Classes ...................... 71
5.3.3 Elk Comparison of Permeability by Passage Structure Spacing ........ 72
5.4 Discussion ........................................................................................................73
5.4.1 Impact of Highway Reconstruction on Elk Permeability ................... 73
5.4.2 Comparison of Elk Permeability by Passage Structure Spacing ........ 74
6.0 Influence of Underpasses and Traffic Volume on Coues White-Tailed
Deer Highway Permeability ...................................................................................77
6.1 Introduction ......................................................................................................77
6.2 Methods............................................................................................................78
6.2.1 Deer Capture and GPS Telemetry ...................................................... 78
6.2.2 GPS Data Analysis of Deer Movements and Permeability ................ 78
6.2.3 Traffic Volume and Deer At-Grade Highway Crossings ................... 79
6.2.4 Traffic Volume and Deer Below-Grade Underpass Crossings .......... 81
6.3 Results ..............................................................................................................82
6.3.1 Deer Capture and GPS Data Analysis of Movements and
Permeability ....................................................................................... 82
6.3.2 Traffic Volume and Deer At-Grade Highway Crossings ................... 84
6.3.3 Traffic Volume and Deer Below-Grade Underpass Crossings .......... 86
6.4 Discussion ........................................................................................................90
6.4.1 Deer Movements and Highway Permeability ..................................... 90
6.4.2 Deer Response to Highway Reconstruction with Passage
Structures ........................................................................................... 90
6.4.3 Influence of Traffic Volume on Deer Movements and
Permeability ....................................................................................... 91
7.0 Conclusions and Recommendations ......................................................................95
7.1 Highway Planning and Monitoring ..................................................................95
7.2 Wildlife Underpasses .......................................................................................96
7.3 Influence of Traffic Volume on Wildlife .........................................................98
7.4 Wildlife Permeability Relationships ................................................................98
7.5 Highway Safety and Wildlife-Vehicle Collisions ............................................99
7.6 Role of Ungulate-Proof Fencing ....................................................................100
7.7 Future SR 260 Reconstruction Sections ........................................................101
References .......................................................................................................................103
LIST OF FIGURES
Figure 1. Location of the 17-mi SR 260 Study Area and the Five Highway
Reconstruction Sections with Wildlife Underpasses, Bridges, and
Riparian-Meadow Habitats. .........................................................................................17
Figure 2. Existing Two-Lane Roadway, Doubtful Canyon Section (Left),
Being Reconstructed into a Four-Lane Divided Highway, Preacher
Canyon Section (Right). ...............................................................................................18
Figure 3. SR 260 Study Area at the Pedestrian-Wildlife Underpass on the
Christopher Creek Section, with Mogollon Rim Escarpment (Background)
and Solar Panels for Powering Video Camera Surveillance System
(Foreground). ...............................................................................................................20
Figure 4. Aerial View of Little Green Valley Riparian-Meadow Complex
Adjacent to Preacher Canyon Section of the SR 260 Study Area. ..............................21
Figure 5. Average Annual Daily Traffic Volume Levels for SR 260 (at the
ADOT Control Road Monitoring Station), 1994–2008. ..............................................22
Figure 6. SR 260 Vehicular Traffic Patterns by Time of Day. Top Graph:
Traffic Volume for All Vehicles. Bottom Graph: Proportion of
Commercial Vehicles. ..................................................................................................23
Figure 7. Aerial and Ground Photographs of SR 260 Underpasses, Including
Little Green Valley West (Top) and East (Middle) Underpasses and Indian
Gardens Underpass (Bottom; Aerial Photograph on Bottom Left Depicts
Underpass Construction). .............................................................................................28
Figure 8. Aerial and Ground Photographs of SR 260 Underpasses, Including
Pedestrian-Wildlife Underpass (Top), Wildlife 2 Underpass (Middle), and
Wildlife 3 Underpass (Bottom). ...................................................................................29
Figure 9. Layout of Video Surveillance System Components at Six SR 260
Wildlife Underpasses. ..................................................................................................30
Figure 10. Probability of a Successful Crossing by Elk at Five SR 260
Wildlife Underpasses (LGV = Little Green Valley). ...................................................38
Figure 11. Probability of a Successful Crossing by White-Tailed Deer at
Five SR 260 Wildlife Underpasses (LGV = Little Green Valley). ..............................39
Figure 12. Number of Elk Underpass Crossings (Left) and Mean Passage
Rates (Right) by Season at Five Underpasses along SR 260, 2002–2008. ..................40
Figure 13. Probability of a Successful Elk Crossing by Season at
Five Underpasses along SR 260. .................................................................................40
Figure 14. Mean Elk Passage Rates (Crossings/Approach) at
Five Underpasses along SR 260, 2002–2008. .............................................................41
Figure 15. Number of Elk Underpass Crossings (Left) and Passage Rates
(Right) by Time of Day along SR 260, 2002–2008. ....................................................42
Figure 16. Mean Elk Passage Rates (Crossings/Approach) for Elk
Approaching and Crossing Indian Gardens Underpass, 2006–2008. ..........................42
Figure 17. Photographs Showing Open Nature and Preserved Native
Vegetation of Indian Gardens Underpass on the SR 260 Kohl’s Ranch
Section.........................................................................................................................44
Figure 18. Annual Frequency of Documented Elk-Vehicle Collisions along
SR 260 with Completion of the First Three Phases of Highway
Reconstruction. ............................................................................................................56
Figure 19. Mean Elk-Vehicle Collisions/mi by SR 260 Highway
Reconstruction Class (without Fencing) Data, 2001−2008. ........................................58
Figure 20. Mean Elk-Vehicle Collisions/mi by SR 260 Highway
Reconstruction Class. ...................................................................................................59
Figure 21. Annual Number of Actual and Expected Elk-Vehicle Collisions
along SR 260, 2001−2008. ..........................................................................................60
Figure 22. Annual Number of Documented Elk-Vehicle Collisions along
SR 260 Control Sections and Corresponding Annual Elk Population
Estimates for Game Management Units 22 and 23. ....................................................61
Figure 23. Cow Elk Caught in a Clover Trap and Blindfolded (Left) and Fitted
with a GPS Receiver Collar and Ear Tag (Right). .......................................................68
Figure 24. GPS Locations and Lines between Successive Fixes to Determine
Highway Approaches and Crossings in 0.1-mi Segments. ..........................................69
Figure 25. Relationships between Mean Elk Passage Rates and Mean Passage
Structure Spacing on Three Reconstructed Sections along SR 260. ...........................76
Figure 26. Female White-Tailed Deer Caught in a Clover Trap (Left) and
Male Deer Fitted with a GPS Receiver Collar (Right) along SR 260. ........................79
Figure 27. GPS Relocations for Male White-Tailed Deer 104 (Red) and 111
(Yellow), Determined from GPS Telemetry Conducted along SR 260,
2004–2007....................................................................................................................83
Figure 28. Mean Probability of GPS-Collared Elk (Top, A–F) and White-
Tailed Deer (Bottom, A–F) Occurring within Each 330-ft Distance Band
along SR 260 at Varying Traffic Volumes, between 2004 and 2007. .........................85
Figure 29. Comparison of At-Grade Highway and Below-Grade Underpass
Passage Rates for Elk (Top) and White-Tailed Deer (Bottom) at Varying
Traffic Volume Levels along SR 260, 2003–2007. .....................................................87
Figure 30. Linear Regression Analyses for Association between White-Tailed
Deer Highway Crossing Passage Rates and Traffic Volume along SR 260,
2004–2007....................................................................................................................88
LIST OF TABLES
Table 1. SR 260 Reconstruction Dates and Duration of Research by Highway
Section...........................................................................................................................8
Table 2. Summary of SR 260 Reconstruction Sections. ....................................................18
Table 3. Physical Characteristics Associated with SR 260 Wildlife
Underpasses Monitored by Video Camera Surveillance, 2002–2008. ........................30
Table 4. Number and Types of Animals Recorded by Video Surveillance at
Six SR 260 Underpasses, 2002–2008. .........................................................................34
Table 5. Likelihood-Ratio Test Results for Factors Influencing Elk Crossings
at SR 260 Underpasses. ................................................................................................35
Table 6. Significant Factors in Predicting the Probability of Elk Crossings at
SR 260 Underpasses. ...................................................................................................35
Table 7. Probability of Successful Elk and White-Tailed Deer Crossings at
SR 260 Underpasses. ...................................................................................................36
Table 8. Comparison of Odds of a Successful Elk Crossing at SR 260 Wildlife
Underpasses. ................................................................................................................36
Table 9. Likelihood-Ratio Test Results for Factors Influencing White-Tailed
Deer Crossings at SR 260 Underpasses. ......................................................................37
Table 10. Significant Factors in Predicting Probability of White-Tailed Deer
Crossings at SR 260 Underpasses. ...............................................................................37
Table 11. Comparison of Odds of a Successful White-Tailed Deer Crossing at
SR 260 Wildlife Underpasses. .....................................................................................37
Table 12. Frequency of SR 260 Wildlife-Vehicle Collisions by Species and
Year, 2001−2008. ........................................................................................................56
Table 13. Proportion of Single-Vehicle Accidents Involving Wildlife,
Documented by Highway Section by the Department of Public Safety. .....................57
Table 14. Mean Annual Number of Elk-Vehicle Collisions/mi ( SE) by
SR 260 Highway Reconstruction Class and Section. ..................................................58
Table 15. Annual Number of Actual and Expected Elk-Vehicle Collisions
(from Modeling of AADT and Elk Population Estimates) and Economic
Benefit from Reduced Elk-Vehicle Collisions. ...........................................................60
Table 16. Mean Elk Passage Rates (Crossings/Approach) by Highway Section
and Coefficients of Variation of Means by SR 260 Reconstruction Class,
Determined from GPS Telemetry, 2002−2008. ...........................................................71
Table 17. Mean Elk Passage Rates for After-Reconstruction–After-Fencing
Class, Determined from GPS Telemetry, 2002−2008. ................................................72
Table 18. SR 260 Reconstruction Status and Number of White-Tailed Deer
Caught and Relocated by Highway Section, 2004–2007. ............................................79
Table 19. Mean Highway Crossing and Passage Rates for White-Tailed Deer
on Control and Reconstructed Sections of SR 260, 2004–2007. .................................84
Table 20. Parameters for the Best Four Models Supported by Akaike’s
Information Criterion of the Probability of 13 White-Tailed Deer Crossing
SR 260. .........................................................................................................................88
Table 21. Number of Successful and Unsuccessful White-Tailed Deer
Crossings and Passage Rates at Five Wildlife Underpasses along SR 260,
2003–2007....................................................................................................................89
Table 22. Number of Individual White-Tailed Deer Exhibiting Flight
Responses at Six Wildlife Underpasses along SR 260 with Varying
Overhead Traffic Levels, 2003–2007. .........................................................................89
ACRONYMS AND ABBREVIATIONS
AADT average annual daily traffic
AC alternating current
ADOT Arizona Department of Transportation
AGFD Arizona Game and Fish Department
AIC Akaike’s Information Criterion
ANCOVA analysis of covariance
ANOVA analysis of variance
BACI before-after–control-impact (design)
CV coefficient of variation
df degrees of freedom
DPS Department of Public Safety
EVC elk-vehicle collision
ft foot
FHWA Federal Highway Administration
GIS Geographic Information System
GMU Game Management Unit
GPS Global Positioning System
hr hour
HR0.5 home-range distance (linear metric)
IGA intergovernmental agreement
MCP minimum convex polygon
mi mile
min minute
ROW right-of-way
SE standard error
SR State Route
TNF Tonto National Forest
UP underpass
V volt
VHF very high frequency
WT white-tailed (deer)
WVC wildlife-vehicle collision
yr year
LIST OF SPECIES
Animals
black bear Ursus americanus
caribou Rangifer tarandus
coyote Canis latrans
elk Cervus elaphus
gray fox Urocyon cinereoargenteus
grizzly bear Ursus arctos
javelina Tayassu tajacu
moose Alces alces
mountain lion Puma concolor
mule deer Odocoileus hemionus
pronghorn Antilocapra americana
raccoon Procyon lotor
roe deer Capreolus capreolus
Coues white-tailed deer Odocoileus virginianus couesi
wolf Canis lupus
Plants
Douglas-fir Pseudotsuga menziesii
Gambel oak Quercus gambelii
juniper Juniperus spp.
manzanita Arctostaphylos spp.
pinyon pine Pinus edulis
ponderosa pine Pinus ponderosa
scrub live oak Quercus turbinella
white fir Abies concolor
ACKNOWLEDGMENTS
The Arizona Department of Transportation’s (ADOT) Research Center and the Arizona
Game and Fish Department (AGFD) funded this research project. The research team
thanks and commends ADOT for its long-term commitment to this project, which has
added substantially to the body of road ecology knowledge. The Tonto National Forest
(TNF) and the Federal Highway Administration (FHWA) provided additional funding
that made the application of Global Positioning System (GPS) telemetry possible. The
research team also thanks Terry Brennan, Robert Ingram, and Duke Klein of the TNF and
Paul Garrett and Steve Thomas of FHWA for their early commitment which made this
project possible.
Many individuals at ADOT provided endless support and guidance in this project and
were instrumental to its success, especially Estomih Kombe, Bruce Eilerts, Siobhan
Nordhaugen, and Doug Brown (now at the Arizona Department of Administration).
Mark Catchpole, Doug Eberline, and Jami Rae Garrison of the Multimodal Planning
Division provided invaluable traffic data and support. The research team is grateful to
Tom Goodman, James Laird, Tom Foster, Myron Robison, David Gerlach, William
Pearson, Jack Tagler, and Dallas Hammit, formerly with the Prescott District, for their
commitment to adaptive management and willingness to consider and respond to our
ideas and recommendations. Their commitment maximized the effectiveness of
underpasses, fencing, and highway safety.
The cooperation of John Anderson, Walt Cline, Bob Ochoa (Boy Scouts of America),
Mikey Marazza, and Tom Dunney (Arizona State University) allowed researchers to trap
elk on private lands and contributed greatly to the success of the GPS telemetry portion
of the study.
The Game and Fish Department’s Mesa Region played a crucial role in the project,
especially Tim Holt, Henry Apfel, John Dickson, Craig McMullen, and Jon Hanna.
Research Branch personnel Scott Sprague, Rob Nelson, and Tim Rogers assisted with the
labor intensive task of keeping the video camera surveillance systems fully operational.
The logistical support provided by Tonto Creek Fish Hatchery personnel was invaluable
to our project, particularly the hospitality and assistance provided by Larry Peterson
(deceased), John Diehl, Larry Duhamell, Mike Weisser, and Trevor Nelson.
The research team offers special thanks to the Arizona Department of Public Safety
(DPS) highway patrol officers in the Payson District. Their efforts to document wildlife-vehicle
collisions were not only instrumental to the success of the project but invaluable
in helping resolve wildlife-vehicle conflicts across Arizona, thus making Arizona’s
highways safer.
The ADOT Research Center’s Technical Advisory Committee for this project provided
many suggestions for improving the project’s effectiveness and applicability. The
research team greatly appreciates the committee’s tremendous support, oversight, and
commitment throughout the duration of the project.
1
EXECUTIVE SUMMARY
This report concludes eight years of continuous wildlife-highway relationships research
conducted along a 17 mile section of State Route (SR) 260 in central Arizona, from 2001
to 2008. This stretch of highway was being reconstructed in five phases from a two-lane
roadway to a four-lane divided highway to incorporate 11 wildlife underpasses (UPs) and
6 bridges. Phased reconstruction made it possible for researchers to use a before-after-control
experimental approach to assess the impact of the construction and success of
measures to reduce wildlife-vehicle collisions (WVCs) and promote wildlife
permeability. The objectives of this research project were:
 Assess and compare wildlife use of wildlife UPs and examine factors that
influence wildlife UP use.
 Evaluate wildlife movements across SR 260 among highway reconstruction
classes (before, during, and after reconstruction) using GPS telemetry.
 Characterize WVC changes associated with SR 260 highway reconstruction and
assess the potential economic benefit of wildlife UPs and other measures.
 Evaluate the relationships among highway traffic volume, wildlife highway
crossing patterns, and wildlife use of UPs.
 Assess the role that ungulate-proof fencing plays in the incidence of WVCs,
wildlife use of UPs, and overall wildlife highway crossings.
To evaluate factors influencing wildlife use of underpasses, researchers monitored
wildlife crossings at six UPs constructed on three sections of SR 260. The focus of this
monitoring was (1) to document wildlife use of UPs by video camera surveillance and to
compute passage rates (number of animals crossing/number of animals approaching) and
compare the probabilities of successful UP crossing by different species and among
different UPs, (2) to evaluate the influence of UP structural characteristics, duration of
monitoring, and other factors on successful UP crossings by elk and white-tailed deer,
and (3) to develop recommendations to maximize UP effectiveness.
To assess the effectiveness of passage structures and fencing in minimizing highway
collisions, the research team documented WVCs along SR 260 to determine the success
and benefits of wildlife UPs and ungulate-proof fencing. The aim was to assess (1) the
incidence of WVCs and the relationship of elk-vehicle collision (EVC) rates to highway
reconstruction classes, (2) the role of ungulate-proof fencing in conjunction with UP
structures in minimizing EVC, and (3) highway safety and economic benefits associated
with reduced EVC. Researchers compared mean EVC rates (EVCs per mile) for highway
sections by analysis of covariance, controlling for annual traffic effects. Two separate
analyses were done using different highway reconstruction classes. The first analysis
compared EVC rates among three treatment classes (before, during, and after
reconstruction). The second analysis assessed the influence of ungulate-proof fencing on
EVC rates by comparing the before-fencing and after-fencing treatment sections of each
2
reconstructed highway section; the after-fencing treatments reflected fencing added to the
limited amount originally planned by ADOT.
This overall research effort underscores the ability to integrate transportation and
ecological objectives into highway reconstruction, yielding tangible benefits to both
highway safety and wildlife permeability. The combination of phased construction,
adaptive management, and effective monitoring of UPs and ungulate-proof fencing were
instrumental in achieving these objectives. It is recommended that such an approach to
highway construction be pursued whenever possible at the time of initial highway design
or reconstruction. The paragraphs that follow highlight the key conclusions and
recommendations of arising from the study.
Wildlife UPs were highly effective in promoting below-grade wildlife crossings, with
two-thirds of more than 15,000 animals recorded on videotape having crossed through an
UP. These UPs were important to improving highway safety through the reduction of
WVCs and promoting wildlife permeability. Structural design characteristics and
placement of UPs are important considerations in maximizing their success in promoting
wildlife passage, and structural characteristics were the most important factor in
determining the probability of achieving successful crossings by wildlife. UP openness is
crucial to achieving high probability of successful use.
The distance that animals must travel through a UP is an especially important factor in
maximizing crossing success and should be minimized. Elk avoided a UP where concrete
retaining walls were erected, compared to a neighboring UP with 2:1 earthen slopes. The
use of concrete walls for wildlife UPs should be avoided. UPs with clear through
visibility should be maximized, and adjacent bridges should be placed in line whenever
possible to maximize visibility by animals through the structures. Wildlife UP placement
should avoid concentrated areas of human disturbance or congregation that occur outside
daytime hours. Elk and deer exhibited dramatically different passage rates for the same
UPs, pointing to the need to address multispecies passage and permeability requirements.
Wildlife UPs in conjunction with adequate ungulate-proof fencing substantially reduced
the incidence of EVCs compared to before-fencing levels. The limited-fencing approach
with highway reconstruction resulted in a nearly fourfold increase in EVCs over before-reconstruction
EVC levels; once fenced under an adaptive management approach
informed by GPS telemetry, EVCs declined 76 percent to before-reconstruction levels.
Such fencing is necessary to funnel elk toward UPs to cross SR 260 below grade, thus
contributing to substantially improved highway safety.
Just as previous SR 260 research found that traffic volume differentially affected elk
depending on whether they approached and crossed at grade or at UPs below grade,
similar results were obtained for white-tailed deer. Traffic volume had minimal impact on
deer crossings at UPs, especially compared to animals attempting to cross at grade; this
finding was of paramount importance to understanding the success of UPs in promoting
both elk and deer permeability.
3
GPS telemetry afforded an unprecedented opportunity to assess and compare wildlife
permeability among reconstruction classes for two ungulate species with different levels
of mobility. For white-tailed deer, a species with relatively limited mobility, mean
passage and crossing rates on reconstructed highway sections were considerably higher
than for control sections; UPs and bridges on the widened upgraded sections improved
deer permeability over the narrow control sections that were a significant barrier to deer
passage. By contrast, highway control sections had the highest mean passage rate for elk,
a more far-ranging species. The mean control section elk passage rate, at 39 percent, was
lower than the reconstructed section mean of 44 percent of approaches. However, this
level of reduced permeability between two-lane undivided and four-lane divided
highways was considerably lower than that documented elsewhere, reflecting the benefit
of combining UPs with ungulate-proof fencing.
The spacing of passage structures on reconstructed highway sections had a significant
influence on elk passage rates, with a strong inverse relationship between permeability
and passage structure spacing. A minimum spacing distance of 1.0 mile between
structures is recommended to balance cost of structures and provide adequate opportunity
for elk to cross highways. Placement of passage structures in areas of high concentrations
of EVCs or preferred habitats (e.g., meadows) is important; however, this may not be
feasible owing to factors including, but not limited to, right-of-way (ROW) clearances,
terrain, the presence of structures at the time of reconstruction, and roadway design
considerations.
With the completion of the three reconstruction sections of SR 260 that exhibited the
worst historical incidence of WVCs, the integration of wildlife UPs and fencing yielded
not only substantial benefits to improved highway safety and wildlife permeability but
also a significant economic benefit. In the three years following the completion of the
reconstructed sections, the economic benefit tied to reduced incidence of EVCs averaged
$2 million per year. The collective benefit to wildlife, highway safety, and financial
savings underscores the degree to which wildlife UPs and fencing along SR 260 can be
considered a great success.
4
5
1.0 INTRODUCTION
1.1 BACKGROUND
Direct and indirect highway impacts have been characterized as some of the most
prevalent and widespread forces altering ecosystems in the United States (Noss and
Cooperrider 1994; Trombulak and Frissell 2000; Farrell et al. 2002). Estimates of annual
collisions involving deer in the United States have ranged from 700,000 (Schwabe and
Schuhmann 2002) to as high as 1.5 million (Conover 1997). Wildlife-vehicle collisions
(WVCs) cause human injuries, deaths, and tremendous property loss (Reed et al. 1982;
Schwabe and Schuhmann 2002). Over 38,000 human deaths attributable to WVCs
occurred in the United States between 2001 and 2005, and the economic impact exceeds
$8 billion/year (Huijser et al. 2007).
WVCs disproportionately affect threatened or endangered species populations and
recovery efforts (Foster and Humphrey 1995; Parker et al. 2008). Forman and Alexander
(1998) estimated that highways have affected more than 20 percent of the nation’s land
area through habitat loss and degradation. Perhaps the most pervasive impact of
highways on wildlife is the barrier and fragmentation effects resulting in diminished
habitat connectivity (Noss and Cooperrider 1994; Forman and Alexander 1998; Forman
2000).
Highways block animal movements between seasonal ranges or other vital habitats. This
barrier effect fragments habitats and populations, reduces genetic interchange (Gerlach
and Musolf 2000; Epps et al. 2005), and limits dispersal of young (Beier 1995); all
disrupt viable wildlife population processes. Long-term fragmentation and isolation
renders populations more vulnerable to the influences of catastrophic events and may
lead to extinctions (Hanski and Gilpin 1997). Fencing constructed to block wildlife and
livestock access across highways without provisions for adequate passage may
exacerbate barrier effects.
Though numerous studies have alluded to highway barrier effects on wildlife (e.g.,
Forman et al. 2003), few have yielded quantitative data relative to animal passage rates,
particularly in an experimental (e.g., before- and after-highway reconstruction) context.
Many studies have focused on the efficacy of passage structures in maintaining wildlife
permeability (Clevenger and Waltho 2003; Ng et al. 2004) or have relied on modeling to
assess permeability (Singleton et al. 2002). Assessments of highway fragmentation
effects on relatively low-mobility small mammals (Swihart and Slade 1984; Conrey and
Mills 2001; McGregor et al. 2003) have proven easier to accomplish than assessments for
far-ranging species that are limited by cost-effective capture and tracking techniques.
Paquet and Callaghan (1996) used winter track counts adjacent to highways and other
barriers to determine passage rates by wolves, something few other studies have reported.
Some studies have used very high frequency (VHF) radio telemetry to assess wildlife
movements and responses to highways. Such studies have often pointed to avoidance of
highways and roads (Brody and Pelton 1989; Rowland et al. 2000) but have seldom
directly addressed permeability, as Gibeau et al. (2001) did for grizzly bears.
6
Numerous assessments of WVC patterns have been conducted, most focusing on deer
(Reed and Woodard 1981; Bashore et al. 1985; Romin and Bissonette 1996a; Hubbard et
al. 2000). Only recently have WVC assessments specifically addressed elk-vehicle
collision (EVC) patterns (Gunson and Clevenger 2003; Biggs et al. 2004). Insights gained
from such assessments have been instrumental in developing strategies to reduce WVC
incidents (Romin and Bissonette 1996a; Farrell et al. 2002), including planning passage
structures to reduce at-grade crossings and to maintain permeability (Clevenger et al.
2002). Consistent tracking of WVCs constitutes a valuable tool to assess the impact of
highway construction (Romin and Bissonette 1996b) and efficacy of passage structures
and other measures (e.g., fencing) in reducing WVCs (Reed and Woodard 1981; Ward
1982; Clevenger et al. 2001a; Dodd et al. 2007b).
Though WVC data are valuable, no study investigated or validated the relationships
between WVCs and spatial and temporal crossing patterns exhibited by wildlife involved
in collisions until recently (Dodd et al. 2006). Barnum (2003) reported that WVC data
were not useful in identifying crossing zones, largely due to inaccurate reporting of
locations. Efforts to increase the accuracy of WVC reporting will provide valuable
information to transportation agencies for planning purposes (Gunson and Clevenger
2003). However, for those species that avoid roadways and seldom cross them (e.g.,
pronghorn), tracking WVCs may be ineffective, and wildlife movement data may be
needed to make sound management decisions on the placement of passage structures.
Increasingly, structures designed to promote wildlife passage across highways are being
implemented throughout North America, especially large bridges (e.g., underpasses or
overpasses) designed specifically for large-animal passage (Clevenger and Waltho 2000;
Bissonette and Cramer 2008). Whereas early passage structures were typically
approached as single-species mitigation measures to address direct impact (Reed et al.
1975), the focus today is more on preserving ecosystem integrity and landscape
connectivity benefiting multiple species (Clevenger and Waltho 2000). Transportation
agencies are increasingly receptive to integrating passage structures into highways to
address both safety and ecological needs (Farrell et al. 2002). However, there is
increasing expectation that such structures will indeed benefit multiple species and
enhance connectivity (Clevenger and Waltho 2000) and that the effectiveness of such
structures will improve with continued scientifically sound monitoring and evaluation
of wildlife responses to them (Clevenger and Waltho 2003; Hardy et al. 2003).
Corlatti et al. (2009) argued for long-term monitoring of wildlife passages to evaluate
their effectiveness in maintaining connectivity and promoting population and genetic
viability, thus justifying their high cost.
Wildlife use of crossings has been measured differently by researchers. Most studies have
reported underpass (UP) use based on track counts (Clevenger and Waltho 2000; Gloyne
and Clevenger 2001), event recorders (Foster and Humphrey 1995), or single-frame
camera images (Ng et al. 2004). Using information about frequency of animal occurrence
to compare passage structure use is potentially biased due to heterogeneous animal
distribution or differential funneling of varying amounts of wildlife-proof fencing; this
fails to account for animals not using passage structures or those exhibiting behaviors
7
such as resistance to crossing. To address such biases, Clevenger et al. (2001b) estimated
expected passage frequencies derived from track assessments of relative abundance, and
Clevenger and Waltho (2003) calculated species performance ratios from radio telemetry,
pellet transects, and habitat suitability indices. Reed et al. (1975) compared animal
evidences at the entrance and exits of UPs to calculate activity indices, while Gordon and
Anderson (2003) used behavioral quantification as a measure of wildlife response. Dodd,
Gagnon, Manzo, et al. (2007) demonstrated video surveillance of UPs as a useful
measurement to assess wildlife passage rates.
1.2 HIGHWAY RECONSTRUCTION AND STUDY APPROACH
The reconstruction of State Route (SR) 260, incorporating 11 large-wildlife passage
structures and 6 bridges (1 passage structure/mi) to address wildlife permeability and
highway safety considerations, constitutes one of the most comprehensive wildlife
connectivity projects in North America. This project compares with landmark efforts to
address wildlife permeability and WVCs in Banff National Park, Alberta, Canada, with
24 passage structures in 28 miles (mi) (0.86/mi; Clevenger and Waltho 2003), as well as
those planned for U.S. Highway 93 reconstruction in Montana, with 42 passage structures
over 56 miles (0.75/mi; Huijser et al. 2010).
1.2.1 Phased Construction and Adaptive Management
In addition to addressing WVCs, two other aspects of the SR 260 reconstruction project
are noteworthy: (1) its phased construction approach and (2) its application of adaptive
management. The phasing of the highway reconstruction in five separate sections has
facilitated effective construction oversight by ADOT and allowed reconstruction to occur
on priority sections with limited funding sources. The incidence of WVCs was a key
factor in the planning and prioritization of the order in which highway sections have been
upgraded.
The phased reconstruction of SR 260 has also facilitated the feedback of preliminary
research findings and insights to ADOT project managers to address wildlife-related
issues. Such insights have been applied to SR 260 sections already under construction or
planned for construction to improve wildlife passage structure design and to identify
appropriate stretches needing ungulate-proof fencing to maximize UP effectiveness and
minimize WVCs. Such an adaptive management approach, where research data are used
to make continuous improvements during subsequent construction activities, can benefit
the quality of highway construction, especially relating to highway safety. However,
adaptive management carries the potential risk of increased costs should construction
delays and increased project budget expenditures occur.
1.2.2 Experimental Approach
The phased reconstruction of SR 260 allowed the researchers to assess the impact of
highway reconstruction on wildlife at various stages. Hardy et al. (2003), Roedenbeck et
8
al. (2007), and Underwood (1994) stressed the value of conducting before-after–control-impact
(BACI) assessments to determine the effects of highway construction and the
efficacy of measures to reduce WVCs and promote permeability. The phasing of SR 260
reconstruction into five highway sections and the presence of experimental controls
provided the opportunity to conduct such an assessment. During the project, the research
team has been able to assess wildlife relationships and response to various stages of
highway reconstruction (Table 1).
Table 1. SR 260 Reconstruction Dates and Duration
of Research by Highway Section.
Highway Section
Reconstruction Upgrade Research Duration (in years)
Begun Completed Beforea Duringa Aftera
Preacher Canyon 1999 2001 0 0 8
Christopher Creek 2002 2004 1 2 5
Kohl’s Ranch 2003 2006 2 3 3
Little Green Valley Control 8 0 0
Doubtful Canyon Control 8 0 0
a Before, during, or after highway reconstruction.
This project focused on evaluating the effectiveness of design measures along SR 260 to
minimize the incidence of WVCs, especially those involving elk, and maintaining
wildlife permeability across the highway. This research was initiated in 2001and
consisted of eight continuous years of field evaluation and monitoring, making it one of
the longest-running monitoring projects in the United States, especially since the average
length of such projects has been only 1.4 years (Clevenger and Waltho 2003). This
commitment to environmental protection, research, and adaptive management has made it
one of the most comprehensive projects of its type in the United States (Cramer and
Bissonette 2007), and one that has garnered well-deserved recognition for ADOT and its
partners—including a 2004 FHWA Exemplary Ecosystem Initiative Award and the 2008
National Association of Environmental Professionals’ National Environmental
Excellence Award.
1.2.3 Research Phases
The SR 260 research project, which was funded by ADOT’s Arizona Transportation
Research Center, occurred in three phases from 2001 to 2008.
Phase I
The first phase, initiated under an intergovernmental agreement (IGA) executed with
ADOT in January 2002 (JPA 01-152), focused on the Preacher Canyon section, the first
reconstructed section (Table 1). Research under this phase served as a “pilot study” for
9
the development and evaluation of several research techniques (e.g., video camera
surveillance and GPS telemetry) to assess the effectiveness of various measures to
minimize WVCs and facilitate wildlife passage across the highway corridor. Field
activities under this phase were initiated in early 2001, before the execution of the first
IGA.
Phase II
Research continued during Phase II under a second IGA executed with ADOT in
December 2003 (JPA 04-024T). This phase focused on the Christopher Creek section,
where reconstruction was completed in late 2004 (Table 1), with continued monitoring of
the first Preacher Canyon section. This IGA extended research through July 2006. After
the second phase, the research team completed the first comprehensive final project
report on research findings (Dodd, Gagnon, Boe, et al. 2007); Section 1.4 below
summarizes the findings of that first research report.
Phase III
Phase III of the research project, which was conducted under a third IGA executed with
ADOT in November 2005 (JPA 06-004T), focused on the Kohl’s Ranch reconstruction
completed in early 2006 (Table 1) and continued monitoring of previously completed
sections. Research under this phase continued through December 2008. This final report
addresses Phase III research findings, as well as those of the previous phases.
1.3 RESEARCH OBJECTIVES
Research conducted during all three phases of the SR 260 study addressed the following
six primary objectives.
Objective 1 Assess and compare wildlife use of SR 260 wildlife UPs and examine
factors that influence wildlife UP use
The study of wildlife response to passage structures has employed various approaches
(Hardy et al. 2003), including use of track counts (Rodríguez et al. 1997; Clevenger et al.
2001b; Clevenger and Waltho 2000, 2003), event recorders (Reed et al. 1975; Foster and
Humphrey 1995), and infrared-motion or heat-sensor single-frame cameras (Servheen et
al. 2003; Brudin 2003; Ng et al. 2004). Only a few studies have used video cameras to
assess passage structure use (Reed et al. 1975; Gordon and Anderson 2003; Plumb et al.
2003). Video surveillance has an advantage over other techniques because animal
behavior can be assessed, especially when crossing resistance or failed crossings occur
(Hardy et al. 2003). Video surveillance also allows for identification and classification
(e.g., sex, age) of individual animals when compared to track counts (Hardy et al. 2003).
Though video camera surveillance has been used minimally, such monitoring has
10
nonetheless provided insights not obtained from other methodologies (Reed et al. 1975;
Gordon and Anderson 2003).
Under this objective, Phase I research evaluated the application of video surveillance to
assess and compare wildlife response to UPs constructed during the reconstruction of
SR 260 (Dodd, Gagnon, Manzo, et al. 2007) by focusing on the two Preacher Canyon
section UPs. The research team developed an unbiased, comparable metric to evaluate
wildlife use of UPs (Dodd, Gagnon, Manzo, et al. 2007). In Phase II of the research
project, the researchers expanded video surveillance to a total of five UPs, and in Phase
III to a total of six UPs. The researchers compared wildlife use at these UPs to relate
differences in response to UP structural characteristics and placement, traffic volume
(Gagnon, Theimer, Dodd, Manzo, et al. 2007), and duration of monitoring.
The comprehensive findings for video surveillance conducted in 2002−2008 under this
objective are presented in Chapter 3.
Objective 2 Evaluate highway permeability and wildlife movements across SR 260
among highway reconstruction classes (before, during, and after
highway reconstruction) using GPS telemetry
The application of GPS telemetry in wildlife movement studies has become increasingly
popular, cost effective, and reliable (Rodgers et al. 1996). With continuous automated
tracking at set time intervals, reduced observer bias (compared to VHF telemetry), and
potential to collect large datasets, GPS telemetry has revolutionized wildlife movement
assessment, and it holds tremendous potential to facilitate highway permeability
assessment and determine spatial and temporal highway crossing patterns by wildlife.
Under this objective, the research team used GPS telemetry to investigate wildlife
permeability across SR 260, comparing before- and after-highway-reconstruction passage
rates of sections under various stages of reconstruction. Under Phase I of the project, the
research team developed and evaluated quantitative measures of elk highway
permeability using GPS telemetry, assessed spatial and temporal influences on elk
movements, and compared permeability as a function of highway reconstruction classes,
within a single reconstructed section (Dodd et al. 2007a). Under Phase II, the research
team assessed the role of ungulate-proof fencing on elk permeability (Dodd et al. 2007b)
and assessed permeability by highway reconstruction classes with two reconstructed
sections. Under Phase III, the research team continued its assessment of elk permeability
across highway reconstruction classes, now with three reconstructed sections exhibiting a
variation in the distance between passage structures that allowed the team to evaluate the
influence of UP spacing on elk permeability. Also under Phase III, the team expanded
telemetry data gathering to include white-tailed deer to assess permeability relationships
for another ungulate species.
The findings for this objective are presented in Chapter 5 (elk permeability) and Chapter
6 (white-tailed deer permeability).
11
Objective 3 Characterize WVCs and changes associated with SR 260 highway
reconstruction (before, during, and after reconstruction) and assess
economic benefit of wildlife UPs and other measures
WVCs present a serious and growing problem for wildlife safety, motorist safety, and
property loss (Reed et al. 1982; Farrell et al. 2002; Schwabe and Schuhmann 2002). The
incidence of WVCs along SR 260 was a major impetus for incorporating wildlife UPs
and ungulate-proof fencing into the highway reconstruction project. Most assessments of
WVCs in North America have focused on deer (Reed and Woodward 1981; Bashore et
al. 1985; Romin and Bissonette 1996b; Hubbard et al. 2000). Only recently have
assessments specifically addressed EVC patterns (Gunson and Clevenger 2003; Biggs et
al. 2004; Dodd et al. 2006; Dodd, Gagnon, Boe, et al. 2007).
The reconstruction of SR 260 in phases allowed the research team to assess the impact of
highway reconstruction on WVCs, including EVCs, across reconstruction classes and
with and without fencing (Dodd et al. 2006; Dodd et al. 2007b). During Phase I, the
research team characterized the nature of EVC patterns along SR 260 and compared
collision incidence associated with the highway under various stages of reconstruction.
The team compared spatial and temporal patterns of EVCs to elk highway crossings
determined by GPS telemetry to validate the usefulness of collision data in developing
strategies to reduce collisions and promote permeability. Under Phases II and III,
continued monitoring of WVCs was accomplished, with a minimum of three years of
after-reconstruction assessment accrued on three reconstructed sections. Research under
these phases also addressed the economic benefit of wildlife mitigations.
The comprehensive 2001−2008 findings for minimizing WVCs under this objective are
presented in Chapter 4.
Objective 4 Evaluate the relationships among highway traffic volume, wildlife
highway crossing patterns, and wildlife use of UPs
Traffic may serve as a “moving fence” that can render highways impermeable to wildlife
(Bellis and Graves 1978). One theoretical model (Iuell et al. 2003) predicted that
highways become impermeable barriers to most wildlife at 10,000 vehicles/day,
potentially leading to fragmentation and rapid genetic isolation of wildlife populations
like that documented for bighorn sheep (Epps et al. 2005). Alternatively, because traffic
volume varies by season, day, and time, some animals may be able to cross even high-traffic-
volume highways during periods when the volume is relatively low.
During the latter part of Phase I and during Phase II, the research team investigated the
relationship of average annual daily traffic (AADT) levels with elk GPS telemetry
relocations to determine the influence on at-grade crossings and elk distribution (Gagnon,
Theimer, Dodd, and Schweinsburg 2007). This was made possible by a permanent traffic
counter installed by ADOT along the study stretch of SR 260. The team also used video
surveillance to assess the influence of traffic volume on elk crossings below grade at five
UPs (Gagnon, Theimer, Dodd, Manzo, et al. 2007). Under Phase III, the team assessed
12
the influence of AADT on at-grade crossings and traffic volume on below-grade UP
crossings by deer, complementing telemetry research previously conducted on elk.
The findings for this objective are presented in Chapter 5 (elk permeability) and Chapter
6 (white-tailed deer permeability).
Objective 5 Assess the role that ungulate-proof fencing plays in the incidence of
WVCs, wildlife use of UPs, and wildlife permeability across the
highway
Though fencing is effective in reducing WVCs (Romin and Bissonette 1996a; Forman et
al. 2003), some studies have reported mixed results (Falk et al. 1978; Feldhamer et al.
1986). Since fences constitute effective barriers to ungulate passage across highways
(Falk et al. 1978), fencing itself may exacerbate the reduction in wildlife permeability
associated with highways alone, particularly where effective measures to accommodate
animal passage are lacking. In addition, fencing is costly and can require substantial
maintenance (Forman et al. 2003). Therefore, transportation agencies have been reluctant
to fence extensive stretches of highways, including SR 260, especially without
information or guidelines for the application of fencing in conjunction with wildlife
passages.
During the reconstruction of SR 260, ADOT applied a general model for integrating 8-ft
ungulate-proof fencing with UPs and bridges. Limited (<300 ft) wing fences were erected
outward from bridge abutments to funnel animals toward the structures. Research was
needed to evaluate both this limited-fencing approach and the strategic placement of
fencing to intercept crossing wildlife as determined from GPS telemetry under adaptive
management (Dodd et al. 2007a).
Under Phases I and II, the researchers looked at the role of both the limited-fencing
approach on WVCs (Dodd et al. 2006) and the strategic fencing approach based on GPS
telemetry-based elk crossing patterns (Dodd et al. 2007b). The researchers further
evaluated this during Phase III.
The findings for this objective are presented in Chapter 3 (UP use by elk and deer),
Chapter 4 (WVC minimization), and Chapter 5 (elk permeability).
Objective 6 Provide ongoing, on-site highway construction implementation
guidance and instruction throughout all reconstruction phases
As the research project was integrated with an ongoing adaptive management approach to
SR 260 reconstruction, the research team provided recommendations and guidelines for
maintaining wildlife permeability, minimizing WVCs, and improving wildlife UP design.
Under Phases I and II, adaptive management activities were focused on applying research
findings to improving UP design (Dodd, Gagnon, Manzo, et al. 2007) and determining
the extent of ungulate-proof fencing needs based on GPS telemetry (Dodd et al. 2007a,
13
2007b). Under Phase III, the research team continued to assess the effectiveness of the
various adaptive management modifications made to UP design and fencing applications.
Research findings related to adaptive management are presented in Chapter 3 (UP use),
Chapter 4 (wildlife-vehicle collisions), and Chapter 5 (elk permeability).
1.4 PHASES I AND II FINAL REPORT SUMMARY
With completion of Phases I and II, the research team prepared a final report (Dodd,
Gagnon, Boe, et al. 2007) detailing the findings of its research activities conducted
through 2006. The following is a summary of the extensive findings from the first two
phases of SR 260 research.
1.4.1 Assessment of Wildlife Underpass Use
Researchers recorded 11 different wildlife species and 8,455 animals, of which elk
accounted for 74 percent. UP passage rates ranged from 0.10 to 0.68 crossings/approach.
UPs were highly effective in promoting below-grade wildlife crossings, with two-thirds
of recorded animals having crossed through one. UPs were instrumental in improving
highway safety through the reduction of WVCs, and in promoting wildlife permeability.
Structural design characteristics and placement of UPs were important considerations to
maximizing their efficacy in promoting wildlife passage. Structural characteristics were
the most important factor in determining the probability of achieving successful crossings
by wildlife.
UP openness is crucial to achieving high probability of successful UP use. The distance
the animals must travel through a UP was an especially important factor in maximizing
efficacy. Elk more often avoided UPs with concrete retaining walls that were erected for
soil stabilization than neighboring UPs with more natural 2:1 sloped earthen sides.
Researchers documented a recurring seasonal pattern where elk UP passage rates dropped
from summer levels >0.90 crossings/approach to below 0.40 during the fall when
migratory elk moved through the SR 260 corridor. Migratory elk did not appear to exhibit
the same propensity for habituation to UPs as resident elk. Ungulate-proof fencing in
conjunction with UPs should expedite the wildlife learning process and help address this
seasonal drop in passage rates.
1.4.2 Traffic Effects on Elk Underpass Crossings
Traffic levels on SR 260 fluctuated greatly on an hourly, daily, and seasonal basis and
nearly doubled from an AADT volume of 4,500 in 2001 to 8,700 in 2003. At the five UPs
where video surveillance occurred, the researchers documented whether traffic levels
affected elk passage rates when elk approached and crossed by simultaneously counting
traffic passing above the UPs. Passage rates at low, intermittent traffic volume
(0.59–0.75 passage rate) and at higher traffic levels (0.71–0.73) did not differ from the
mean passage rate determined when no vehicles were present (0.65). Passage rates varied
14
seasonally due to the presence of migratory elk, but even during migratory periods,
traffic volume levels had minimal effect on passage rate. Thus, the researchers found that
traffic volume had no effect on elk passage rates when they crossed the highway below
grade at UPs. This finding was crucial to understanding the efficacy of UPs in promoting
wildlife permeability.
1.4.3 Elk Permeability from GPS Telemetry
GPS telemetry afforded the researchers an unprecedented opportunity to assess and
compare wildlife permeability among highway reconstruction classes. In the first phase
of GPS telemetry (2002–2004), the researchers fitted 33 elk with GPS receiver collars.
These collars accrued 101,506 GPS location fixes, with 45 percent occurring within
0.6 mi of the highway. Nearly two times the proportion of locations occurred within
0.6 mi of the highway compared to randomly generated locations. Elk were attracted to
the highway corridor by riparian-meadow foraging habitats that were seven times more
concentrated within the 0.6-mi zone around the highway, compared to the mean
proportion within elk home ranges. Elk crossed the highway 3,057 times; crossing
frequency and distribution were strongly aggregated rather than randomly distributed.
The mean passage rate for elk crossing the highway section where reconstruction was
completed (0.43 crossings/approach) was half that of the sections under reconstruction
and control sections combined (0.86). Permeability was jointly influenced by the size of
the widened highway and associated vehicular traffic on all lanes. The researchers used
crossing frequency to delineate where ungulate-proof fencing yielded maximum benefit
in intercepting and funneling crossing elk toward UPs and in reducing EVCs.
1.4.4 Traffic Effects on Elk Highway Crossings
A permanent traffic counter was installed within the study area to provide continuous
traffic data to compare to elk highway crossing data. From 44 elk collared in both
telemetry phases, researchers linked 38,709 GPS locations to hourly traffic volume data
(6,470,000 vehicles) to determine how elk distribution varied with traffic and how elk
highway crossings were affected by traffic volume. The probability of elk occurring near
the highway decreased with increasing traffic volume; elk primarily used the habitat
near the highway when traffic volumes were relatively low (<100 vehicles/hr). The
researchers found that increasing traffic volume reduced the overall probability of
at-grade elk highway crossings, but this effect depended on both seasonality and the
proximity of riparian-meadow habitats. Elk crossings occurred later in the evening
when traffic levels abated, and unsuccessful attempts, or “repels,” by elk to cross
SR 260 at grade typically coincided with high traffic volume.
1.4.5 Role of Ungulate-Proof Fencing
In the second phase of GPS telemetry (2004–2005), the research team compared
permeability on one reconstructed section nearly one year before and one year after
ungulate-proof fencing was erected. The research team fitted 22 elk with GPS receiver
15
collars and accrued 87,745 GPS locations. The elk highway passage rate after SR 260
was opened to traffic, but before fencing was erected (0.54 crossings/approach), was
32 percent lower than the level during reconstruction work (0.79). However, once
ungulate-proof fencing was erected, the passage rate increased 52 percent to
0.82 crossings/approach. Thus, fencing with UPs promoted wildlife permeability as
animals were funneled toward UPs by fencing where they crossed below grade with
minimal impact from traffic (compared to crossings at grade where traffic did have an
influence).
In addition to playing an instrumental role in promoting permeability, ungulate-proof
fencing was crucial to achieving effective use of UPs, especially those not located near
meadow habitats. Without fencing, elk and deer continued to cross SR 260 at grade
immediately adjacent to UPs. With just 49 percent of one section strategically fenced to
intercept peak elk highway crossings determined from GPS telemetry, an 87 percent
reduction in ECVs was realized in the year after fencing. Fencing constitutes an integral
component of wildlife mitigations in promoting permeability.
1.4.6 Wildlife-Vehicle Collision Relationships
The research team assessed spatial and temporal patterns of EVCs from 1994 to 2006
(n = 571). The team used data from the first phase of GPS telemetry to assess spatial and
temporal patterns of elk highway crossings and compare those patters with EVC patterns.
Annual EVCs were related to traffic volume and elk population levels. EVCs occurred in
a nonrandom pattern. The EVC mean for sections under reconstruction (up until
ungulate-proof fencing was erected) was higher (11.6 EVCs/yr) than the before-reconstruction
EVC mean (4.4 EVCs/yr) and the after-reconstruction EVC mean (6.5
EVCs/yr). On the first section completed in 2001 with limited fencing (13 percent),
EVCs did not differ among before, during, and after reconstruction classes, even though
mean traffic volume increased 67 percent from before- to after-reconstruction levels,
pointing to the benefit of three passage structures and fencing. On another section, EVCs
increased more than 2.5 times when opened to traffic but before strategically located
ungulate-proof fencing was erected. Once fencing was erected along half the section
linking passage structures, EVCs dropped 87 percent.
The researchers compared EVCs and crossings at five spatial scales; the strongest
relationship was at the highway section scale. Strength of the relationship and
management utility were optimized at the 0.6-mi (ca. 1 km) scale. The strong association
between EVCs and highway crossings underscored the utility and value of WVC data in
planning wildlife mitigation measures ranging from passage structures to ungulate-proof
fencing. EVCs were associated with proximity to riparian-meadow habitats adjacent to
the highway. Although EVCs and crossings during the fall season exceeded expected
levels, the proportion of EVCs in September-November (49 percent) exceeded the
proportion of crossings and coincided with the breeding season, elk migration, and high
use of riparian-meadow habitats adjacent to the highway. A higher proportion of EVCs
(59 percent) occurred relative to crossings (33 percent) in the evening (1700–2300 hr); 34
percent of EVCs occurred within one hour after sunset, and 55 percent within two hours
16
after sunset. EVC data are valuable in developing strategies, including locating passage
structures, to maintain permeability and increase highway safety.
1.4.7 Economic Benefit of Wildlife Measures
With reconstruction of just two SR 260 sections completed with UPs and ungulate-proof
fencing, 2006 was the first year that the incidence of actual EVCs dropped below the
level predicted from modeling based on traffic volume and elk population levels.
Modeling predicted even greater benefit as traffic volume is anticipated to increase. The
complement of measures implemented to date has achieved its objective in mitigating the
impact of highway reconstruction and increasing traffic volume. The researchers expect
the benefit to grow now that the third section is complete and the entire first
reconstructed section has been fenced under an enhancement grant project. In 2006, the
researchers estimated the annual economic benefit from reduced EVCs to be $850,000.
With only a modest increase in traffic volume, the researchers estimated that the annual
benefit will exceed $1 million/year.
1.4.8 Conclusion
This study of Phases I and II underscored the ability to integrate transportation and
ecological objectives into highway construction activities, yielding tangible benefits to
highway safety and wildlife permeability, as well as economic benefits from reduced
WVCs. The combination of phased construction, adaptive management during
reconstruction, and effective monitoring was instrumental to jointly achieving
transportation and ecological objectives.
1.5 REPORT ORGANIZATION
This final report is organized as follows: Chapter 2 sets the highway reconstruction and
biological context for the research project; Chapters 3–6 detail the SR 260 research
objectives and associated research findings; and Chapter 7 synthesizes those findings
across objectives and summarizes key recommendations that reflect the increased
understanding of the complex interactions between wildlife and highways. Literature
cited throughout the report is listed in a single reference section at the end of the report.
Scientific names for plant and animals species used throughout are listed in the report’s
front matter; scientific names are not used elsewhere in the report.
17
2.0 STUDY AREA
This study was conducted along a 17-mi stretch of SR 260 (mileposts 260−277),
beginning 9 mi east of Payson and extending to the base of the Mogollon Rim in central
Arizona (latitude 34o15’–34o18’N, longitude 110o15’–111o13’W; Figure 1). SR 260 links
metropolitan Phoenix to several tourism-dependent White Mountain communities (e.g.,
Show Low, Pinetop-Lakeside, Springerville-Eagar) and popular summer (e.g., camping,
fishing) and winter (e.g., skiing) recreation areas on the White Mountain Apache
Reservation and the Apache-Sitgreaves National Forest. SR 260 also serves as the
primary connector to Interstate 40.
Figure 1. Location of the 17-mi SR 260 Study Area
and the Five Highway Reconstruction Sections with
Wildlife Underpasses, Bridges, and Riparian-Meadow Habitats.
18
Starting in 2000, sections of the two-lane highway have been upgraded to a four-lane
divided highway (Figure 2). In places, the footprint of the upgraded highway exceeds
0.3 mi wide (Figure 2). The reconstructed highway will incorporate 11 wildlife UPs
specifically intended to reduce at-grade elk crossings and WVCs, as well as 6 bridges
over large canyons and streams that will accommodate wildlife use (Figure 1; Table 2).
Reconstruction of three sections with 7 UPs and all 6 bridges is now completed
(Figure 1; Table 2). Reconstruction of the last two sections, Little Green Valley and
Doubtful Canyon, with 4 UPs started in 2010.
Figure 2. Existing Two-Lane Roadway, Doubtful Canyon Section (Left),
Being Reconstructed into a Four-Lane Divided Highway,
Preacher Canyon Section (Right).
Table 2. Summary of SR 260 Reconstruction Sections.
Highway Section
Reconstruction
Status
Highway
Mileposts
Length
(mi)
Wildlife Passages
UP Bridge
Preacher Canyon Completed 2001 260.0–263.0 3.0 2 1
Little Green Valley Control 263.1–265.5 2.5 1 0.5
Kohl’s Ranch Completed 2006 265.6–269.5 4.0 1 1.5
Doubtful Canyon Control 269.6–272.5 3.0 3 0
Christopher Creek Completed 2004 272.6–277.0 4.5 4 3
All 260.0–277.0 17.0 11 6
2.1 RECONSTRUCTED SECTIONS AND CHRONOLOGY
Understanding the status of the three reconstructed highway sections and the time frames
associated with the reconstruction process is important for contextualizing the BACI
experimental design (Underwood 1994; Hardy et al. 2003; Roedenbeck et al. 2007) that
was integral to this research project. Also important is an understanding of the
19
modifications made to the original reconstruction plans under the adaptive management
process, primarily those involving the application of 8-ft ungulate-proof fencing along
each section. Design characteristics and photos of the completed wildlife UPs are
included in Chapter 4.
2.1.1 Preacher Canyon Section
The first highway section, Preacher Canyon, was completed in November 2001. This
section included two UPs and a large bridge over Preacher Canyon (Figure 1; Table 2).
Upon completion, only 0.4 mi (13 percent) of the section’s length was fenced with
ungulate-proof fencing, associated with the two UPs near Little Green Valley. As a result
of continuing WVCs (see Chapter 5), an enhancement project was implemented to raise
the existing 3.5-ft right-of-way (ROW) fence to 7.5 ft (with barbed-wire and electric
fence) along the remaining unfenced portion of the section. The 2.5 mi of fence
modification were completed in December 2006 (Gagnon et al. 2010). The research team
conducted six years of after-reconstruction–before-fencing treatment monitoring and two
years of after-reconstruction–after-fencing treatment evaluation.
2.1.2 Christopher Creek Section
The majority of heavy reconstruction on the Christopher Creek section, including
construction of 3 bridges and 4 UPs, was completed by May 2003. Upon completion,
wildlife could pass through the unfenced passage structures (Figure 1; Table 2); however,
vehicular traffic was confined to two lanes until early July 2004, when all four lanes were
opened to traffic. Erection of ungulate-proof fencing was not completed until
mid-December 2004. Original construction designs incorporated ungulate-proof fencing
along 0.7 mi of the section (22 percent). This extent of fencing was increased to 2.4 mi
(49 percent) by raising the ROW fence through the adaptive management process to
address peak elk highway crossing zones determined by GPS telemetry (Dodd, Gagnon,
Manzo, et al. 2007). Research during Phases I and II projected that added fencing would
intercept 45 percent of elk crossings, for a total of 58 percent crossing interception
(Dodd, Gagnon, Manzo, et al. 2007). Initially, a 0.2-mi gap was left in the fence midway
along a 2.0-mi stretch of fenced highway due to complexities associated with integrating
fencing at a lateral access road into the community of Christopher Creek; this gap was
fenced in November 2007. Overall, the research team’s data collection and evaluation
covered 1 year of before-reconstruction, 3.5 years of during-reconstruction, 1 year of
after-reconstruction–before-fencing, and 3.5 years of after-reconstruction–after-fencing.
(Figure 3).
2.1.3 Kohl’s Ranch Section
The Kohl’s Ranch section, completed in March 2006, included 1 wildlife UP and 1.5
bridges (only one bridge span was built over Thompson Draw, and the other will be built
under the Little Green Valley section). The original wildlife UP designs were modified
substantially under adaptive management (Dodd, Gagnon, Manzo, et al. 2007; see
20
Chapter 4 of this report), as was the planned length of ungulate-proof fencing (<0.5 mi;
12 percent). The fencing was increased to include the eastern third of the section (1.3 mi),
projected to intercept 60 percent of the GPS-determined elk crossings. Here, however,
only limited fencing was extended westward from the peak crossing area associated with
the Indian Gardens UP. For this section, the research team conducted two years of before-reconstruction,
three years of during-reconstruction, and nearly three years of after-reconstruction–
after-fencing treatment evaluation.
Figure 3. SR 260 Study Area at the Pedestrian-Wildlife Underpass on the
Christopher Creek Section, with Mogollon Rim Escarpment (Background) and
Solar Panels for Powering Video Camera Surveillance System (Foreground).
2.2 NATURAL SETTING
The study area lies within the ponderosa pine association of the montane coniferous
forest community (D. Brown 1994). Elevations along SR 260 range from 5,220 to
6,560 ft. The Mogollon Rim escarpment to the north is the dominant landform, rising
precipitously to 7,860 ft (Figures 1 and 3). Vegetation adjacent to the highway grades
ranges from mixed ponderosa pine, pinyon pine, juniper, and live oak forest on the lower-elevation
Preacher Canyon and Little Green Valley sections to forests dominated by
ponderosa pine interspersed with Gambel oak at higher elevations to the east on the
21
Christopher Creek section. Chaparral (e.g., manzanita) with sparse pinyon pine, live oak,
and ponderosa pine is prevalent on the drier south-facing slopes. Mixed-conifer forests of
ponderosa pine, Douglas fir, white fir, and Gambel oak occur in canyons emanating from
the Mogollon Rim. Numerous riparian and wet meadow habitats occur at several
locations along the highway corridor (Figure 1), with some meadows more than 60 acres
in size (Figure 4). Several perennial streams flow adjacent to portions of the highway,
including Little Green Valley, Tonto, Christopher, Hunter, and Sharp creeks (Figure 1).
Figure 4. Aerial View of Little Green Valley Riparian-Meadow Complex
Adjacent to Preacher Canyon Section of the SR 260 Study Area.
Climatic conditions within the study area are mild, with a mean maximum monthly
temperature of 90.3° F (July) and a mean minimum monthly temperature of 19.6° F
(January). Annual precipitation averages 20.7 inches, with a mean of 21.3 inches of
snowfall in winter; precipitation has averaged two-thirds of normal since 2002.
The research team focused its study on Rocky Mountain elk for several reasons. First, elk
accounted for more than 80 percent of all collisions between vehicles and wildlife (Dodd
et al. 2006; see Chapter 5 of this report) and for the vast majority of property loss and
human injuries associated with collisions with vehicles. Elk are large animals that are
relatively easy to trap and that can readily support GPS telemetry collars, which can yield
substantial long-term data on wildlife highway movements.
22
Both resident and migratory elk herds occurred within the study area. Resident elk were
common, especially in proximity to wet meadows. Nonresident elk migrate off the
Mogollon Rim with the first snowfall greater than 12 inches, typically in late October
(R. Brown 1990, 1994). Brown (1990) reported that 85 percent of the elk residing within
this Mogollon Rim herd unit migrate to an area below but within 6 mi of the base of the
Mogollon Rim, which encompasses the SR 260 study area. Migratory elk return to
summer range with forage green-up at higher elevations (Brown 1990). The 2008
estimated resident elk population in Game Management Units (GMUs) 22 and 23
encompassing the study area was approximately 2,500 (Arizona Game and Fish
Department, Game Management Branch, Phoenix), though not all elk resided in
proximity to SR 260. White-tailed deer were frequently present near SR 260, while mule
deer were less common and more localized on the eastern portion of the study area.
2.3 TRAFFIC VOLUME
AADT volume on this portion of SR 260 (at the ADOT Control Road traffic monitoring
station) nearly tripled in 10 years from 3,100 in 1994 to 8,700 in 2003 but has been static
since (Figure 5; ADOT Data Management Section). Since 2002, AADT has been
determined by a permanent traffic counter installed at the center of the study area along
the Little Green Valley section. Traffic volumes were highest during daytime hours
(Figure 6) when passenger cars accounted for 81 percent of all vehicles traveling along
SR 260 (2004–2007); commercial vehicles account for 19 percent of the traffic volume
but often exceeded 40 percent during nighttime hours (Figure 6).
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
1994 1996 1998 2000 2002 2004 2006 2008
Average annual daily traffic Year
Figure 5. Average Annual Daily Traffic Volume Levels for SR 260
(at the ADOT Control Road Monitoring Station), 1994–2008.
23
Figure 6. SR 260 Vehicular Traffic Patterns by Time of Day. Top Graph: Traffic
Volume for All Vehicles. Bottom Graph: Proportion of Commercial Vehicles.
Note: Data obtained from traffic recorded by a
permanent traffic counter on SR 260 between 2004 and 2008.
24
25
3.0 EVALUATION OF FACTORS INFLUENCING
WILDLIFE USE OF HIGHWAY UNDERPASSES
3.1 INTRODUCTION
As road and highway networks throughout the world are upgraded to accommodate
increasing traffic, opportunities for wildlife to cross at grade diminish as animals suffer
increased mortality from vehicle collisions or exhibit road avoidance (Jaeger et al. 2005).
Highways act as barriers to free movement of wildlife, fragmenting and isolating habitats
and resources, reducing genetic interchange (Epps et al. 2005) and the probability of
population persistence (Jaeger et al. 2005), and increasing population susceptibility to
stochastic events (Forman and Alexander 1998; Trombulak and Frissell 2000). Mortality
from vehicle collisions is a serious and growing problem for wildlife populations,
motorist safety, and property loss (Reed et al. 1982; Farrell et al. 2002).
Highway reconstruction projects are increasingly incorporating wildlife passage
structures to promote wildlife passage across highways and to preserve landscape
connectivity; these structures have proven successful for many species (Clevenger and
Waltho 2000, 2005; Foster and Humphrey 1995; Dodd, Gagnon, Manzo, et al. 2007;
Bissonette and Cramer 2008). Passage structure use can minimize or eliminate the
effects of vehicular traffic (Mueller and Berthoud 1997), allowing for unimpeded
movement across roadways (Gagnon, Theimer, Dodd, Manzo, et al. 2007).
Transportation agencies have been receptive to integrating passage structures in projects
to address safety and ecological needs (Farrell et al. 2002), and there is increasing
expectation that these structures will yield tangible biological and economic benefits
(Clevenger and Waltho 2000). As such, scientifically sound monitoring of wildlife use of
passage structures is vital to improving future effectiveness and justifying their continued
application (Clevenger and Waltho 2003; Hardy et al. 2003).
Structural characteristics and placement of wildlife crossing structures are important to
maximizing wildlife use (Reed et al. 1975; Reed et al. 1979; Foster and Humphrey 1995;
Clevenger and Waltho 2000, 2003; Dodd, Gagnon, Manzo, et al. 2007). Prior studies
modeled structural factors accounting for differences in wildlife use (Clevenger and
Waltho 2000, 2005; Ng et al. 2004). Design and placement is important to passage
structure success, particularly if flawed design or inadequate funnel-fencing results in
animals avoiding a passage structure altogether and crossing the highway at grade,
presenting a risk to motorists and animals (Dodd, Gagnon, Manzo, et al. 2007).
Various techniques have been used to assess passage structure usage by wildlife,
including track counts (Rodríguez et al. 1997; Clevenger et al. 2001a; Clevenger and
Waltho 2000, 2003), triggered event recorders or counters (Reed et al. 1975; Foster and
Humphrey 1995), and infrared-motion or heat-sensor single-frame cameras (Brudin 2003;
Servheen et al. 2003; Ng et al. 2004), including digital infrared cameras (Olsson et al.
2008). Only limited use of video cameras has occurred (Reed et al. 1975; Sips et al. 2002;
Gordon and Anderson 2003; Plumb et al. 2003; Dodd, Gagnon, Manzo, et al. 2007).
26
Video surveillance has advantages over other techniques because it allows for evaluation
of animal behavior, especially when avoidance or failed crossings occur (Hardy et al.
2003; Gordon and Anderson 2003; Dodd, Gagnon, Manzo, et al. 2007), and for
simultaneous observation of passing traffic (Gagnon, Theimer, Dodd, Manzo,
et al. 2007).
Several measures have been used to quantify wildlife use of passage structures. Most
studies have enumerated frequency of use (Clevenger and Waltho 2000; Gloyne and
Clevenger 2001; Sips et al. 2002; Ng et al. 2004; Olsson et al. 2008). However, frequency
of use can be a biased index because it may be subject to differential funneling of animals
by topography, varying amounts of fencing, and heterogeneous animal distribution.
Frequency of use does not account for nonuse attributable to structural characteristics or
alternative crossing locations (Reed et al. 1975; Clevenger et al. 2001a; Clevenger and
Waltho 2003, 2005). Dodd, Gagnon, Manzo, et al. (2007) used passage rate (number of
crossing animals/number of animals approaching) as a comparative measure of passage
structure use to address this bias. Passage rates determined by video surveillance are
relatively unbiased by differential wildlife densities associated with various passage
structures, and such rates provide a calculation of the proportion of animals that refuse to
cross through structures (Dodd, Gagnon, Manzo, et al. 2007). Dodd, Gagnon, Boe,
et al. (2007) also modeled probability of use by logistic regression, which was useful in
comparing underpass use by wildlife and complemented passage rate as a metric of
UP use.
Hardy et al. (2003) and Clevenger and Waltho (2003) stressed the importance of long-term
monitoring of wildlife passage structure use. The latter reported that for 18 studies,
the average monitoring duration was 1.4 years. They documented dramatic changes in
wildlife use patterns over the course of their five-year evaluation of newly constructed
passage structures. Numerous studies have reported that ungulates and other wildlife
require time to adapt to crossing structures (Reed et al. 1975; Clevenger and Waltho
2000, 2003; Dodd, Gagnon, Manzo, et al. 2007; Olsson et al. 2008). While Clevenger and
Waltho (2003) and Dodd, Gagnon, Manzo, et al. (2007) found relatively rapid acceptance
of new UPs by elk, achieving peak use within two years, other species took longer to
habituate to passage structures. Use of passages reflected both structural characteristics
and species-specific adaptation to them over time (Clevenger and Waltho 2003).
Olsson et al. (2008) believed that differential learning rates by species were related to
differences in home-range sizes and exposure to passages. Dodd, Gagnon, Manzo, et al.
(2007) reported dramatically different seasonal elk passage rates along SR 260 (<0.40 in
winter and >0.80 in summer) attributable to the influx of migratory elk in winter that,
unlike resident animals, lacked regular exposure to UPs; they believed that this might
pose a long-term impediment to achieving consistent yearlong use by elk. Since those
reported results, nearly four additional years of monitoring have occurred along the
highway, including monitoring at an additional four UPs (six total). As with Clevenger
and Waltho’s (2003) assessments based on long-term monitoring, the researchers now
have the ability to assess wildlife use patterns and passage rates over time.
The objectives were to:
27
 Assess wildlife use of UPs by video camera surveillance and compute passage
rates as a comparative measure of UP use by different species and among UPs.
 Evaluate the influence of UP structural characteristics and other factors important
in predicting successful UP crossings by elk and white-tailed deer, species for
which sufficient data were collected across all UPs.
 Consider the influence of the duration of UP monitoring and how it might
influence interpretations of the efficacy of wildlife UPs.
 Develop recommendations to maximize the effectiveness of UPs in promoting
wildlife permeability, thus providing transportation agencies additional options
for resolving wildlife-highway conflicts.
3.2 METHODS
3.2.1 Video Surveillance Systems
The research team monitored wildlife use at six UPs constructed on three sections of
SR 260 (Figures 7 and 8), with monitoring ongoing at individual UPs anywhere from 2.5
to 5.5 years (Table 3). Video surveillance of the Preacher Canyon section began in late
2002, yielding 5.5 years of monitoring; surveillance of the Christopher Creek section
began in early 2004, yielding 4 years of monitoring; and surveillance of the Kohl’s Ranch
section began in spring 2006, yielding 2.5 years of monitoring.
The team used integrated animal-triggered four-camera video surveillance systems to
examine the number and types of wildlife species that used the six UPs. Each
surveillance system included two cameras that recorded animals approaching the UP
from one side of each UP; the other two cameras recorded animals as they passed through
the UP (Figure 9). The Indian Gardens and Pedestrian-Wildlife UP surveillance systems
were powered by arrays of solar panels, while the other four systems were powered by
120 V AC. Dodd, Gagnon, Manzo, et al. (2007) used time-lapse validation to show that
the use of photo-beam triggers to detect approaching and crossing animals was an
accurate and reliable mode of video recording, with benefits of efficient videotape
analysis time and costs.
3.2.2 Assessment of Wildlife Use of Underpasses
The research team limited the overall analysis of results for the six UPs to a comparison
of passage rates and did not include behavioral response as reported by Dodd, Gagnon,
Manzo, et al. (2007). Passage rates were determined by the proportion of animals
crossing through each UP to those that approached each UP. The research team
considered a UP approach to occur when animals crossed over the 4-ft ROW fence
approximately 130���150 ft from the mouths of the UPs and showed movement toward the
mouths. Passage rates were calculated from animals approaching from only one side of
the UPs.
28
Figure 7. Aerial and Ground Photographs of SR 260 Underpasses,
Including Little Green Valley West (Top) and East (Middle) Underpasses
and Indian Gardens Underpass (Bottom; Aerial Photograph on Bottom Left
Depicts Underpass Construction).
29
Figure 8. Aerial and Ground Photographs of SR 260 Underpasses,
Including Pedestrian-Wildlife Underpass (Top), Wildlife 2 Underpass (Middle),
and Wildlife 3 Underpass (Bottom).
30
Table 3. Physical Characteristics Associated with SR 260 Wildlife Underpasses
Monitored by Video Camera Surveillance, 2002–2008.
Wildlife Underpass Highway Section
Span
(ft)
Height
(ft)
Length
(ft)a
Atrium
(ft)b
Monitoring
Duration
(in years)
East Little Green Valley Preacher Canyon 135 22 175 36 5.5
West Little Green Valley Preacher Canyon 135 38 365 36 5.5
Pedestrian-Wildlife Christopher Creek 110 22 420 155 5
Wildlife 2 Christopher Creek 130 32 390 105 5
Wildlife 3 Christopher Creek 125 17 210 None 4.5
Indian Gardens Kohl’s Ranch 135 41 215 120 2.5
a Length = distance for animals to fully negotiate passage structure, from mouth to mouth, including fill material.
b Atrium = width of opening between eastbound and westbound bridge spans.
Figure 9. Layout of Video Surveillance System Components at
Six SR 260 Wildlife Underpasses.
Note: Video cameras were oriented to record wildlife approaching the underpass (two cameras),
animals crossing through the underpass from both directions (one camera), and simultaneous
traffic on the highway while animals approached and crossed the underpass (one camera).
31
The research team used multiple logistic regression analysis to select factors important in
predicting a successful crossing through the UPs (Agresti 1996). These calculations were
limited to data for elk and white-tailed deer, since they were the only species adequately
represented across all UPs. The binomial response variable was based on a successful
crossing or noncrossing once a group (≥1) of elk or deer approached a UP. The research
team deemed factors important by using likelihood-ratio tests to test the significance of
each selected factor given the other factors incorporated in the model (Agresti 1996). The
team selected factors for analysis based on what previous studies reported were important
in affecting elk and deer movements associated with highways. The researchers also
believed that temporal availability of structure use was a potentially important factor
influencing UP use. Although other wildlife species used the UPs, the sample sizes for
those species were inadequate across all UPs to predict their probability of crossing. The
research team instead provided overall passage rates and use by all species associated
with each UP monitored.
The research team limited its analytical modeling to five factors that generally influenced
ungulate movements to determine whether the temporal movements of elk outweighed
the importance of UP structure:
 UP structure and placement (Clevenger and Waltho 2000, 2005; Gagnon et al.
2006; Dodd, Gagnon, Manzo, et al. 2007)—This factor served as a categorical
variable to evaluate the importance of UP structure among the other variables and
to compare differences in wildlife use among UPs.
 Months monitored (Clevenger and Waltho 2003; Dodd, Gagnon, Manzo, et al.
2007; Olsson et al. 2008)—This factor served as a continuous variable to
determine changes in wildlife use since completion of construction.
 Season (Bruinderink and Hazebroek 1996; Gunson and Clevenger 2003; Dodd,
Gagnon, Manzo, et al. 2007)—This factor evaluated changes in seasonal weather
conditions and elk migration patterns:
 Winter December–February
 Spring March–May
 Summer June–August
 Fall September–November
 Time of day (Bruinderink and Hazebroek 1996; Haikonen and Summala 2001;
Dodd, Gagnon, Boe, et al. 2007; Dodd, Gagnon, Manzo, et al. 2007)—This factor
evaluated four 6-hr periods:
 Morning 0400–0959 hr
 Daytime 1000–1559 hr
 Evening 1600–2159 hr
 Nighttime 2200–0359 hr
32
 Day of week (Rost and Bailey 1979; Witmer and deCalesta 1985; Gunson and
Clevenger 2003; Gagnon 2006)—This factor served as a surrogate variable for
traffic level, since SR 260 traffic levels were typically 30 percent higher on
weekends than weekdays (Gagnon 2006). Based on local traffic levels, weekday
(Monday through Thursday  6,000 AADT) and weekend (Friday through
Sunday  8,000 AADT) days served as categorical variables.
The research team limited its logistic regression analysis to the five UPs that were
monitored for at least four years to give an adequate representation of seasonal
differences and use over time. The team did not analyze data for the Indian Gardens UP
since only 2.5 years of data existed and this UP reflected structural changes made
adaptively from prior monitoring (Dodd, Gagnon, Manzo, et al. 2007). Once the research
team determined the factors that were important to predicting elk and white-tailed deer
UP crossing probability, they further analyzed these factors graphically to assess
associated patterns. The team examined the significance of the influence that each factor
(except months monitored) had in the model in each of the four years. The research team
tested model fit using a Hosmer and Lemeshow goodness-of-fit test (Hosmer and
Lemeshow 1989). The team used a general linear model with a logistic regression link to
determine probabilities of a successful crossing for each of the selected factors and to
further provide the odds ratios of a successful crossing for each of the scenarios selected
as important by the analysis. The research team used the following equation to calculate
probabilities of successful UP crossing:
exp (α + x)
Probability =
1 + exp (α + x)
This calculation can be interpreted as the probability of a successful crossing under a
given scenario versus that of a failure (1 – probability) once an elk approaches a UP. The
 and  terms represent the intercept and log odds, respectively. The research team used
months monitored, the only continuous variable, in combination with all other significant
factors separately for graphical representation. To calculate the comparative odds ratios
for successful elk and deer crossings at any two UPs, the research team divided the odds
of a successful crossing at one UP by the odds for the other one being compared.
To evaluate whether elk crossing probabilities at the five UPs changed over the first four
monitoring years (potentially affecting conclusions regarding UP efficacy), the research
team used analysis of variance (ANOVA) to compare differences among mean UP
crossing probabilities for each year (Hays 1981). The team tested the null hypothesis that
no differences in elk crossing probabilities and passage rates existed as a function of year.
A Tukey test for unequal sample sizes assessed the statistical significance of post hoc
pairwise comparisons among years (Hays 1981). The team transformed all proportion
data for the ANOVA using an arcsine transformation before analysis (Neter et al. 1996).
33
3.3 RESULTS
3.3.1 Wildlife Underpass Use
From 2002 to 2008, the research team logged 9,305 days of video surveillance
monitoring and recorded 1,428 hours of videotape footage of approaching and crossing
wildlife at the six UPs. The video surveillance systems recorded 15,134 animals and
11 different species (Table 4); 10,216 animals, or 67.5 percent, crossed through the UPs.
Elk accounted for 68 percent of all animals documented at the UPs, while white-tailed
deer and mule deer accounted for 13 percent and 6 percent, respectively (Table 4). The
average passage rate for all species at the six UPs was 0.58 crossings/approach; the
Indian Gardens UP had the highest overall passage rates (0.78 crossings/approach) for all
species combined (Table 4).
In general, the research team noted an increasing degree of species diversity and evenness
in distribution recorded at the UPs along a gradient from west to east, corresponding to an
increase in elevation. At the west end of the study area, elk accounted for more than 90
percent of all animals recorded on videotape approaching and crossing the two Preacher
Canyon section UPs (East and West Little Green Valley); at these same UPs, white-tailed
deer accounted for 6 percent, and mule deer <1 percent. At the Indian Gardens UP near
the midpoint of the study area, elk accounted for 64 percent of the total animals recorded,
white-tailed deer 13 percent, and mule deer still <1 percent (Table 4). At the three UPs on
the Christopher Creek section at the eastern end of the study area, elk accounted for 47
percent of all recorded animals, while white-tailed deer accounted for 19 percent and
mule deer 15 percent.
In addition to the species listed in Table 4, surveillance systems at five UPs recorded 14
black bears, 7 (50 percent) of which passed through, and 22 mountain lions, 9 of which
(41 percent) passed through. The research team did not document any predator-prey
interactions at any of the UPs, as described by Little et al. (2002). The surveillance
systems recorded javelina at four UPs; the vast majority of javelina (>450) were recorded
at the Indian Gardens UP.
The research team recorded an overall mean UP passage rate for elk of 0.61 crossings/
approach, ranging from 0.20 at the Wildlife 3 UP to 0.83 at the Indian Gardens UP. For
white-tailed deer, the team documented an overall mean passage rate of 0.39 crossings/
approach, ranging from 0.06 crossings/approach at the East Little Green Valley UP to
0.96 at the Wildlife 3 UP; the Pedestrian-Wildlife UP had the highest white-tailed deer
use (936 on videotape) and a mean passage rate of 0.51 crossings/approach (Table 4).
34
Table 4. Number and Types of Animals Recorded by Video Surveillance at Six SR 260 Underpasses, 2002–2008.
Wildlife Species Recorded on Videotape
All
2,445
4,482
2,455
2,778
487
2,189
15,134
1,819
3,299
1,672
1,989
341
1,844
10,216
0.67
0.58
0.46
0.48
0.62
0.78
0.58
Other
146
248
71
136
24
461
1,084
143
256
16
83
18
341
857
0.83
0.90
0.10
0.53
0.45
0.57
0.56
Raccoon
5
24
168
20
99
17
333
5
24
68
9
98
14
218
1.00
1.00
0.57
0.35
0.90
0.67
0.75
Gray Fox
15
49
127
55
2
4
252
10
27
78
35
2
4
156
0.67
0.58
0.64
0.59
—
1.00
0.70
Coyote
14
91
45
27
2
21
200
5
23
24
9
2
14
77
0.67
0.22
0.28
0.55
—
0.67
0.46
Mule Deer
9
2
66
798
90
2
967
0
0
37
647
72
2
758
—
—
0.42
0.61
0.61
—
0.55
WT Deer
198
279
965
96
145
286
1,996
30
42
660
46
116
174
1,068
0.11
0.06
0.51
0.27
0.96
0.44
0.39
Elk
2,179
3,789
1,098
1,743
125
1,398
10,332
1,626
2,927
789
1,160
33
1,295
7,830
0.71
0.71
0.68
0.48
0.20
0.83
0.61
Wildlife UP
West Little Green Valley
East Little Green Valley
Pedestrian-Wildlife
Wildlife 2
Wildlife 3
Indian Gardens
Total
West Little Green Valley
East Little Green Valley
Pedestrian-Wildlife
Wildlife 2
Wildlife 3
Indian Gardens
Total
West Little Green Valley
East Little Green Valley
Pedestrian-Wildlife
Wildlife 2
Wildlife 3
Indian Gardens
Total
No. of Animals
on Video
No. of Animals
Crossing in
Underpass
Passage Rate
35
3.3.2 Factors Influencing Successful Elk Underpass Crossings
Of the five factors included in the logistic regression model, four were important in
predicting the probability of a successful elk crossing during the first four years of
monitoring (Table 5). These factors included UP structure and placement, months
monitored, season, and time of day. Day of the week, the surrogate factor for traffic
volume, did not have a significant influence on crossing probabilities when elk crossed
the highway below grade at UPs, similar to that found by Gagnon, Theimer, Dodd,
Manzo, et al. (2007); model fit was adequate for continued analysis (Hosmer and
Lemeshow test; χ2 = 7.58, df = 8, P = 0.480).
Table 5. Likelihood-Ratio Test Results for Factors Influencing
Elk Crossings at SR 260 Underpasses.
Model Factora dfb Likelihood-Ratio χ 2 χ 2 Probability
UP structure and placement 4 170.6 <0.001c
Months monitored 1 52.1 <0.001c
Season 3 27.5 <0.001c
Time of day 3 4.7 0.019c
Day of week 1 <0.1 0.990
a Factors modeled by logistic regression for determining the probability of a successful elk crossing at
five underpasses during the first four monitoring years by video camera surveillance, 2002–2008.
b df = degrees of freedom
c Corresponds to those factors that had a significant influence on elk underpass-crossing probabilities.
Modeling identified UP structure and placement as the most important factor, therefore
suggesting that this factor likely was of primary importance in predicting the probability
of successful elk passage (Table 5). The duration of UP monitoring was the second most
important factor, followed closely by season. Time of day had the least influence on
probability of elk successfully crossing at a UP. UP structure and placement was a
significant influence in all four years, season in the first three years, and time of day only
in the first year; day of week did not have a significant influence in any individual year
(Table 6).
Table 6. Significant Factors in Predicting the Probability of
Elk Crossings at SR 260 Underpasses.
Model Factor
Year Monitoreda
1 2 3 4
UP structure and placement X X X X
Season X X X NS
Time of day X NS NS NS
Day of week NS NS NS NS
a X = significant factor; NS = not significant factor.
36
The probability of a successful elk crossing among UPs ranged from 0.76 at the East
Little Green Valley UP to only 0.08 at the Wildlife 3 UP (Table 7). Statistical analysis
using pairwise comparisons showed that the odds of elk crossing at the East Little Green
Valley UP were higher than all others—ranging from 37.7:1 odds, compared to a
successful crossing at the Wildlife 3 UP, to 1.3:1 odds, compared to a successful crossing
at the West Little Green Valley (Table 8). The odds of a successful elk crossing at the
Wildlife 3 UP were lower than all other UPs.
Table 7. Probability of Successful Elk and White-Tailed Deer
Crossings at SR 260 Underpasses.
Wildlife Underpass Elk White-Tailed Deer
East Little Green Valley 0.76 0.08
West Little Green Valley 0.73 0.09
Pedestrian-Wildlife 0.65 0.52
Wildlife 2 0.52 0.20
Wildlife 3 0.08 0.67
Table 8. Comparison of Odds of a Successful Elk Crossing
at SR 260 Wildlife Underpasses.
Wildlife Underpass
East Little
Green Valley
West Little
Green Valley
Pedestrian-
Wildlife Wildlife 2 Wildlife 3
East Little Green Valley 1.3:1 1.8:1 3:1 37.7:1
West Little Green Valley 1:1.3 1.4:1 2.3:1 29.8:1
Pedestrian-Wildlife 1:1.8 1:1.4 1.7:1 21.6:1
Wildlife 2 1:3 1:2.3 1:1.7 12.9:1
Wildlife 3 1:37.7 1:29.8 1:21.6 1:12.9
Note: Number on the left side of each ratio is associated with the structures listed in each column.
3.3.3 Factors Predicting Successful White-Tailed Deer Underpass Crossings
Of the five factors included in the logistic regression model, UP structure and placement
was the only factor important in predicting the probability of a successful white-tailed
deer crossing during four years of monitoring (Table 9). None of the other factors had a
significant influence on deer crossing probability in any individual year (Table 10).
The probability of a successful deer crossing among UPs contrasts to those for elk,
ranging from 0.08 at the East Little Green Valley UP to 0.67 at the Wildlife 3 UP
(Table 7). A statistical analysis with pairwise comparisons between the five UPs
indicated that the odds of successful deer crossing at the Wildlife 3 UP were higher
than for others, ranging from 21.4:1 odds compared to the East Little Green Valley UP to
37
2:1 odds compared to the Pedestrian-Wildlife UP (Table 11). While the odds of a
successful crossing were lowest for elk at the Wildlife 3 UP, the odds were higher for
deer at that UP than all other UPs (Table 8).
Table 9. Likelihood-Ratio Test Results for Factors Influencing
White-Tailed Deer Crossings at SR 260 Underpasses.
Model Factor dfa Likelihood-Ratio χ 2 χ 2 Probability
Underpass structure and placement 4 85.3 <0.001b
Months monitored 1 <0.1 0.982
Season 3 2.6 0.457
Time of day 3 3.8 0.294
Day of week 1 <0.1 0.845
a df = degrees of freedom.
b Corresponds to factors that had a significant influence on elk underpass crossing probabilities.
Table 10. Significant Factors in Predicting Probability of
White-Tailed Deer Crossings at SR 260 Underpasses.
Model Factor
Year Monitoreda
1 2 3 4
Underpass structure and placement X X X X
Season NS NS NS NS
Time of day NS NS NS NS
Day of week NS NS NS NS
a X = significant factor; NS = not significant factor.
Table 11. Comparison of Odds of a Successful White-Tailed Deer
Crossing at SR 260 Wildlife Underpasses.
Wildlife Underpass
East Little
Green Valley
West Little
Green Valley
Pedestrian-
Wildlife Wildlife 2 Wildlife 3
East Little Green Valley 1:1.3 1:10.6 1:2.6 1:21.4
West Little Green Valley 1:1.3 1:7.9 1:1.9 1:16
Pedestrian-Wildlife 10.6:1 7.9:1 1.4:1 1:2
Wildlife 2 2.6:1 2.3:1 1:4.1 1:8.3
Wildlife 3 21.4:1 16:1 2:1 8.3:1
Note: Number on the left side of each ratio is associated with the structures listed in each column.
38
3.3.4 Influence of Duration of Video Surveillance Monitoring
The second most important factor influencing the probability of successful elk crossings,
the time elapsed since UP installation, measured by UP monitoring, likely relates to the
learning curve associated with elk habituation to the structures since construction. The
overall probability of a successful elk crossing increased steadily over four years: 0.55 in
the first, 0.62 in the second, 0.68 in the third, and 0.73 in the fourth year. The
probabilities of a successful elk crossing at individual UPs also steadily increased over
time for all UPs except the Wildlife 3 UP (Figure 10). By the fourth year, the
probabilities of elk crossing converged for four UPs, all >0.70 (Figure 10); the initially
low (<0.10) probability of crossing at the Wildlife 3 UP actually decreased over the four
years. Mean elk passage rates followed the same trend as crossing probabilities: 0.50
crossings/approach in the first year, 0.66 in the second, 0.64 in the third, and 0.82
crossings/approach by the fourth.
Figure 10. Probability of a Successful Crossing by Elk at
Five SR 260 Wildlife Underpasses (LGV = Little Green Valley).
Though the elapsed time (duration of monitoring) was not a significant factor in the
logistic regression model predicting white-tailed deer crossings, since probabilities were
static over time, a large increase in crossing probability occurred at the Wildlife 3 UP
over the four years (Figure 11).
The ANOVA of elk crossing probabilities among the first four monitoring years at the
five UPs found that there were differences among the UP means by year (F3, 4 = 4.06,
P = 0.033). Post hoc comparisons among years indicated that the differences among years
in the ANOVA was limited to that of the first year (mean = 0.47) versus fourth year
(mean = 0.62) (P = 0.045), over which the mean probability increased 32 percent
(Figure 10).
39
Figure 11. Probability of a Successful Crossing by White-Tailed Deer
at Five SR 260 Wildlife Underpasses (LGV = Little Green Valley).
3.3.5 Influence of Season
Season had nearly as great an influence on the probability of successful elk crossing as
the elapsed time since installation. The highest number of elk UP crossings occurred in
spring during the period of forage green-up in meadows adjacent to SR 260, coupled with
elk migration back to the summer range atop the Mogollon Rim (Figure 12). The mean
elk passage rate for the five UPs was at its highest in spring and summer (>0.65) but
dropped to its lowest (0.55) in fall and winter when nonhabituated migratory elk were
present along SR 260 (Figure 12). For the first four monitoring years combined, seasonal
crossing probabilities ranged widely from 0.48 in winter to 0.53 in spring to 0.72 in
summer, then dropping to 0.31 in fall.
However, like elk crossing probabilities by UP over the four years, crossing probabilities
by season converged to >0.65 by the fourth year (Figure 13). Likewise, the recurring
pattern of seasonal fluctuations that the research team noted in mean elk UP passage rates
evident in the first three years (e.g., <0.40 crossings/approach in fall-winter and >0.80 in
spring-summer) did not occur in the fourth year; there was a general upward trend in
passage rates over the first four years at the five UPs (Figure 14).
40
Figure 12. Number of Elk Underpass Crossings (Left) and Mean Passage Rates
(Right) by Season at Five Underpasses along SR 260, 2002–2008.
Figure 13. Probability of a Successful Elk Crossing by Season at
Five Underpasses along SR 260.
0.5
0.6
0.7
Winter Spring Summer Fall
Season
Passage rates
0
500
1000
1500
2000
2500
Winter Spring Summer Fall
Season
Number of crossings
41
Figure 14. Mean Elk Passage Rates (Crossings/Approach)
at Five Underpasses along SR 260, 2002–2008.
3.3.6 Influence of Time of Day
Time of day had an influence on the probability of elk successfully crossing the five UPs
(Table 5), though when considered by individual year its contribution was only
significant in the first year (Table 6). This is not to say that time of day was not an
important factor in elk UP crossings, which showed a strong bimodal pattern of crossings
in the evening and morning (Figure 15). Both the elk UP passage rate and the probability
of a successful UP crossing (0.55) were highest during the nighttime hours; crossing
probabilities were somewhat lower in the evening (0.47) and morning (0.39) and
considerably lower during daytime hours (0.22). However, like the convergence of
probabilities of crossing by UP and season, the researchers noted a similar convergence
in probability of crossing by time of day over the first four years, given that the
probability of crossing increased for all four time periods across each of the four
monitoring years.
3.3.7 Wildlife Use of the Indian Gardens Underpass
Though not included in the logistic regression analysis, monitoring of the Indian Gardens
UP nonetheless provided valuable insights to understanding wildlife use of these
structures. Unlike all other UPs that exhibited relatively low elk passage rates in their
first year (mean = 0.50 crossings/approach), the passage rate at the Indian Gardens UP
after six months was >0.75 crossings/approach and exceeded 0.80 by the end of the first
year (Figure 16). The mean elk passage rate for the other five UPs did not attain this
passage rate level until the fourth year. The Indian Gardens UP exhibited an above-average
passage rate among all six UPs for white-tailed deer (0.44 crossings/approach
versus the mean of 0.39), and the highest overall passage rate across all species (0.78
versus the mean of 0.58).
42
Figure 15. Number of Elk Underpass Crossings (Left) and
Passage Rates (Right) by Time of Day along SR 260, 2002–2008.
Figure 16. Mean Elk Passage Rates (Crossings/Approach) for Elk
Approaching and Crossing Indian Gardens Underpass, 2006–2008.
3.4 DISCUSSION
Video camera surveillance constituted a valuable means to assess and compare wildlife
use of the six UPs, particularly with passage rate and probability of UP crossing as
metrics for comparison and evaluation of UP efficacy (Dodd, Gagnon, Manzo, et al.
2007). Compared to the extensive replications of similar types and placements of UPs
available to Clevenger and Waltho (2000, 2005) and Ng et al. (2004) in their modeling of
0
100
200
300
400
500
600
700
1500
1700
1900
2100
2300
0100
0300
0500
0700
Time of day
Number of crossings
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1500
1700
1900
2100
2300
0100
0300
0500
0700
Time of day
Passage rate
43
structural factors, the replications available for the SR 260 experimental design modeling
were limited. Nonetheless, the results still provide compelling insights relative to the
influence of UP design, placement, and other factors on wildlife use, including different
species’ responses to the same UPs. The research team’s long-term monitoring illustrates
the influence that UP monitoring duration has on formulating conclusions about the
efficacy of the UPs, as stressed by Clevenger and Waltho (2003).
3.4.1 General Efficacy of Underpasses
Regardless of the metric, the fact that over two-thirds of the 15,134 animals recorded on
videotape at the six UPs successfully crossed SR 260 below grade via the UPs
underscores the overall efficacy of these structures in promoting wildlife passage and
motorist safety. The research team believes that an equal or greater number of animals
likely crossed below grade at the seven other passage structures that the research team did
not monitor; those passage structures have larger span widths that make them difficult to
monitor but that make them highly suitable for animal passage. Where UPs occur with
fencing, the incidence of EVCs has declined dramatically and highway safety has
increased (Dodd, Gagnon, Boe, et al. 2007; Dodd et al. 2007b; Gagnon et al. 2010). By
the fourth year of monitoring, the mean elk passage rate among UPs exceeded 80 percent
and no longer exhibited the dramatic seasonal fluctuations tied to migratory animals that
previous research documented and considered a limitation to year-round UP efficacy
(Dodd, Gagnon, Manzo, et al. 2007). Given these recent monitoring results, and that elk
have historically accounted for the vast majority of WVCs, property damage, and human
injuries, the application of wildlife UPs along SR 260 can be considered a success.
3.4.2 Influence of Underpass Structural Characteristics
Because modeling determined UP structure and placement as the most important factor
influencing the probability of a successful elk crossing, the research team believes that
this factor reflects variation among UPs relative to structural design, placement, or both.
However, the team’s interpretation is based on limited replications of UP design and
placement. Little Green Valley, with its two adjacent UPs (East and West), was the only
section where placement could be controlled to allow for a comparison of design alone
(Dodd, Gagnon, Manzo, et al. 2007).
Given the differences in UP structural design and placement characteristics among the
five UPs, the research team nonetheless believes that the results provide valuable insights
on the influence of structural design and placement. Other studies have reported such
attributes as crucial to achieving successful wildlife use of UPs (Reed et al. 1975; Beier
and Loe 1992; Foster and Humphrey 1995; Clevenger and Waltho 2000, 2005; Forman
et al. 2003). Even though elk crossing probabilities converged for four of the five UPs
analyzed by the fourth monitoring year, passage rates and comparative odds of successful
UP crossing still point to differences in use. Furthermore, while the convergence of the
probabilities of crossing at these UPs reflects the habituation and learning potential of elk
over time, even among nonresident migratory animals, the differences in use and learning
44
curves associated with each UP structure still reflect important UP structural and
placement characteristics.
Five of the six monitored UPs were of a similar large, twin open-span bridge design
(Figures 7 and 8) varying in span length (110−135 ft) and height (17−41 ft). Excluding
the Wildlife 3 UP (the only single bridge structure without an atrium) and given enough
time to allow habituation to UPs, all SR 260 structures became effective for elk passage;
in some instances (e.g., Wildlife 2 UP) they became effective in spite of structural or
placement limitations. That elk overcame these limitations should not be construed to
imply that transportation agencies should ignore the design characteristics that created the
limitations. Rather, the goal should remain to construct the best passage structures
possible to maximize both wildlife use and habituation given constraints such as funding,
topography, and other factors. The Indian Gardens UP (Figure 17) exemplifies the
importance and benefit of adaptive management improvements (Dodd, Gagnon, Boe, et
al. 2007; Dodd, Gagnon, Manzo, et al. 2007). Modifications to the Indian Gardens UP
design eliminated concrete walls for soil stabilization below the bridge spans, thus
opening up the floor of the UP and preserving natural vegetation. These improvements
resulted in a range of species rapidly accepting the Indian Gardens UP as a passageway.
Considerations to maximize use and learning that reflect insights gained from monitoring
structures along SR 260 and elsewhere need not add cost to the construction of passage
structures.
Figure 17. Photographs Showing Open Nature and Preserved Native Vegetation of
Indian Gardens Underpass on the SR 260 Kohl’s Ranch Section.
45
Dodd, Gagnon, Manzo, et al. (2007) previously reported significantly different
probabilities of elk crossing at the East and West Little Green Valley UPs after just 2.5
years of video surveillance. Now, after another three years of monitoring, both UP elk
passage rates and probabilities of successful elk crossing are identical. Over time, elk
have habituated to the West UP as reflected in steadily increasing passage rate and
probability of elk crossing. However, due to the proximity of the two UPs, the research
team believes that elk have also learned to avoid approaching the West UP altogether,
instead approaching and crossing at the East UP. With 1.8 times more elk having crossed
through the East UP (Table 4), some animals have likely habituated to using the West
UP, while some simply have avoided approaching it and instead use the East UP,
reducing the proportion of failed crossings and thus resulting in higher passage rates and
probabilities.
While the success of both Little Green Valley UPs is attributable to their proximity to the
preferred meadow foraging area and placement in established drainage travel corridors
(Dodd, Gagnon, Manzo, et al. 2007), the higher elk use of the East UP reflects
differences in structural attributes. The East UP has a twofold higher openness ratio
(Reed et al. 1979), half the distance for animals to traverse through the UP, and 2:1
earthen sloped sides. In addition to these structural attributes, the research team still
believes that the concrete retaining walls at the West UP (Figure 7) have continued to
influence the lower incidence of elk use compared to the East UP, as described by Dodd,
Gagnon, Manzo, et al. (2007). The research team frequently observed animals standing at
the mouth or just inside the West UP and looking upward from side to side.
Although the researchers did not specifically address predator-prey interactions, they did
not document any such interactions either, as was also the case with Little et al. (2002).
Elk nonetheless appeared hypervigilant of predators potentially lurking atop the concrete
walls of the West UP. Little et al. (2002) recommended designing UPs for prey species
(e.g., elk, deer) to minimize predation risk with short, wide, and high passages. Though
several factors contributed to the difference in elk use of the UPs, the research team
believes that differential use is largely attributable to ledge effect, unnatural feel, and
possible noise properties associated with its concrete walls. As such, even with the now-identical
elk passage rate and probability of UP crossing, the research team still
recommends that use of mechanically stabilized earth retaining walls be avoided in UP
design and construction where possible, as Dodd, Gagnon, Boe, et al. (2007) and Dodd,
Gagnon, Manzo, et al. (2007) previously recommended.
Of the five UPs monitored, the Wildlife 2 UP exhibited the most dramatic increase in the
probability of successful elk crossing over the first four years of monitoring (Figure 10).
This UP was unique in its bridge placement and alignment; the bridges at the other UPs
were constructed in line, allowing approaching animals to see completely through the
structures. The Wildlife 2 UP bridges were offset along the existing drainage alignment
(Figure 8). Fill slopes due to the offset bridge placement obstructed elk views through the
UP at floor level. During the first year of monitoring (2004), the elk passage rate for the
Wildlife 2 UP was only 0.12 crossings/approach. Since then, the elk passage rate has
improved steadily to >0.80 crossings/approach (and the probability of crossing >0.70),
46
pointing to both the ability of elk to habituate to UPs (