Mule Deer Migrations and Highway Underpass Usage in California, USA

Abstract

Roadways may pose barriers to long‐distance migrators such as some ungulates. Highway underpasses mitigate wildlife‐vehicle collisions and can be an important management tool for protecting migration corridors. In northern California, 3 underpasses were built on United States Route 395 (Route 395) in Hallelujah Junction Wildlife Area (HJWA) in the 1970s for a migratory mule deer ( Odocoileus hemionus ) herd that had been negatively affected by highway traffic. To determine whether these underpasses were still reducing mule deer mortalities >40 years after construction, we investigated deer use of the underpasses from 2006–2019 using cameras, global positioning system (GPS) collars, and roadkill records. We used occupancy models, approximations of GPS‐collared mule deer movement paths, and roadkill locations to estimate the highway crossing patterns of deer. From camera data, there was higher use of the underpasses by deer during migration (spring [Mar–Jun], fall [Oct–Dec]) than in summer (Jul–Sep), when only resident deer were present. Higher underpass usage occurred in the spring compared to fall migrations. Eleven of 21 GPS‐collared migrating mule deer crossed Route 395. We estimated 30% of the crossings (by 7 of the 11 deer) occurred south of the underpasses where deer could easily access the highway because of short (1‐m high) and deteriorating highway fencing. Roadkill data confirmed that deer‐vehicle collisions were occurring south of the underpasses and at the underpasses. This was likely due to deteriorating infrastructure at the underpasses that allows wildlife access to the highway. Overall, our study indicated that although underpasses can provide safe passage for migratory deer decades (>40 yr) after their construction, deteriorating infrastructure such as fencing and gates can lead to wildlife mortalities on highways near underpasses. © 2021 The Wildlife Society.

Full text

The Journal of Wildlife Management 1–7; 2021; DOI: 10.1002/jwmg.22043
Research Article
Mule Deer Migrations and Highway Underpass
Usage in California, USA
MOLLY R. CALDWELL ,
1
California Department of Fish and Wildlife, 1701 Nimbus Road, Gold River, CA 95670, USA
J. MARIO K. KLIP, California Department of Fish and Wildlife, 1701 Nimbus Road, Gold River, CA 95670, USA
ABSTRACT Roadways may pose barriers to long‐distance migrators such as some ungulates. Highway
underpasses mitigate wildlife‐vehicle collisions and can be an important management tool for protecting
migration corridors. In northern California, 3 underpasses were built on United States Route 395 (Route
395) in Hallelujah Junction Wildlife Area (HJWA) in the 1970s for a migratory mule deer (Odocoileus
hemionus) herd that had been negatively affected by highway traffic. To determine whether these under-
passes were still reducing mule deer mortalities >40 years after construction, we investigated deer use of the
underpasses from 2006–2019 using cameras, global positioning system (GPS) collars, and roadkill records.
We used occupancy models, approximations of GPS‐collared mule deer movement paths, and roadkill
locations to estimate the highway crossing patterns of deer. From camera data, there was higher use of the
underpasses by deer during migration (spring [Mar–Jun], fall [Oct–Dec]) than in summer (Jul–Sep), when
only resident deer were present. Higher underpass usage occurred in the spring compared to fall migrations.
Eleven of 21 GPS‐collared migrating mule deer crossed Route 395. We estimated 30% of the crossings
(by 7 of the 11 deer) occurred south of the underpasses where deer could easily access the highway because
of short (1‐m high) and deteriorating highway fencing. Roadkill data confirmed that deer‐vehicle collisions
were occurring south of the underpasses and at the underpasses. This was likely due to deteriorating
infrastructure at the underpasses that allows wildlife access to the highway. Overall, our study indicated that
although underpasses can provide safe passage for migratory deer decades (>40 yr) after their construction,
deteriorating infrastructure such as fencing and gates can lead to wildlife mortalities on highways near
underpasses. © 2021 The Wildlife Society.
KEY WORDS highway, mule deer, Odocoileus hemionus, roadkill, underpass, wildlife‐vehicle collisions.
Roadways affect wildlife via mortalities, fragmentation of
movement corridors, and degradation of habitat
(Jackson 2000, Trombulak and Frissell 2000, Huijser
et al. 2008, Fahrig and Rytwinski 2009, Brunton
et al. 2018). For migratory wildlife, roadways may impede
annual migrations, particularly for large terrestrial species,
such as ungulates (Lendrum et al. 2012, Sawyer et al. 2012,
Seidler et al. 2015). Wildlife‐vehicle collisions (WVCs) are
also costly for humans, resulting in human fatalities and
injuries, and an estimated economic cost of 8.4 billion
dollars annually in the United States (Conover et al. 1995,
Huijser et al. 2008).
To reduce WVCs and improve migration corridors across
roadways, crossing structures (i.e., underpasses and over-
passes) are an effective solution in many countries
(Clevenger and Waltho 2000, Olsson et al. 2008, Smith
et al. 2015, Sawyer et al. 2016b, Caldwell and Klip 2020).
Researchers have reported that wildlife crossing structures
are effective at reducing mortalities and improving
permeability along wildlife corridors, particularly for large
migratory species, such as mule deer (Odocoileus hemionus;
Mata et al. 2008, Sawyer et al. 2012, Stewart 2015, Simpson
et al. 2016). Mule deer migrations can range 15–200 km,
and migration paths often cross developed areas and road-
ways (Sawyer and Kauffman 2011, Lendrum et al. 2012,
Sawyer et al. 2016a). Mule deer exhibit fidelity to traditional
migration routes (Monteith et al. 2011, Lendrum
et al. 2013). Roadways that bisect traditional migration
routes often have higher levels of deer‐vehicle collisions
during migration (Coe et al. 2015). Areas with high levels of
WVCs are logical candidates for crossing structures (Sawyer
et al. 2012, Simpson et al. 2016). Crossing structures, such
as overpasses and underpasses, are successful at decreasing
deer‐vehicle collisions at migration corridors, and the
combination of wildlife highway fencing and crossing
structures is considered one of the most effective strategies
to promote connectivity and reduce WVCs (Huijser
et al. 2009, Sawyer et al. 2012). Highway fencing with
insufficient height or length, however, can reduce the
effectiveness of crossing structures (Huijser et al. 2016).
Other factors that can affect crossing structure effectiveness
include traffic volume, human use of crossing structures,
predator‐prey interactions, and crossing type and location
(Clevenger and Waltho 2005, Gagnon et al. 2011,
Received: 1 October 2020; Accepted: 31 January 2021
1
E‐mail: Molly.Caldwell@Wildlife.ca.gov
Caldwell and Klip •Migration and Underpasses 1

Sawyer et al. 2012, Barrueto et al. 2014, Caldwell and
Klip 2020). Most existing research that assessed the effec-
tiveness of crossing structures focused on the first 15 years of
post‐construction, with the majority focusing on the first
5 years (Clevenger and Waltho 2005, Gagnon et al. 2011,
Sawyer et al. 2012, Barrueto et al. 2014). Research is lacking
on the long‐term effectiveness of crossing structures along
migratory corridors, particularly identifying factors that may
reduce effectiveness decades after construction.
In northern California, part of the Loyalton‐Truckee mule
deer herd migrates across United States Route 395 (Route
395) every spring (Mar–Jun) and fall (Oct–Dec). The
Loyalton‐Truckee herd is an interstate population with
summer ranges in California and winter ranges in Nevada
and California. The herd experienced high levels of highway
mortality when crossing Route 395, and in the early 1970s
the California Department of Transportation (CalTrans)
and the California Department of Fish and Wildlife
(CDFW, previously California Department of Fish and
Game) identified the herd's migration route and areas of
high deer mortality along the highway (Kahre 1980). In
1976, CalTrans widened the highway to a 4‐lane divided
highway and completed construction of 3 highway under-
passes and highway fencing in what is now the Hallelujah
Junction Wildlife Area (HJWA). Fencing was 2.4‐m‐tall
metal mesh fencing that was roughly 6 km in length on both
sides of the highway with 1‐way exits for wildlife to leave
the highway near underpasses, human and vehicle access
gates near the underpasses, and 1‐m‐tall barbed wire fencing
approximately 1.5 km south of the underpasses. The 1‐way
exits consisted of angled, horizontal metal bars that allowed
deer and other mid‐sized wildlife to exit, but not enter, the
highway. In the few years following the completion of the
underpasses, CDFW conducted track counts of migrating
deer using the underpasses and assessed highway mortalities
in the area (Kahre 1980). Kahre (1980) reported that about
500–1,500 deer were successfully migrating through the
underpasses and reported an almost total elimination of
highway deer mortalities (with known mortalities attributed
to access gates being left open).
Our study documented mule deer migrations through the
HJWA underpasses on Route 395 during 2006–2019,
>40 years after the crossing structures were completed. Our
objectives were to determine seasonal mule deer use of the
underpasses, whether the underpasses were still effective in
reducing deer‐vehicle collisions, and whether they still
promoted successful migrations of the Loyalton‐Truckee
herd. Based on field observations and previous research, we
predicted that although the underpasses were still effective
corridors for migrating mule deer, factors such as deterio-
rating fencing and 1‐way exits around the underpasses may
contribute to higher deer‐vehicle collisions during migratory
periods (Sawyer et al. 2012, Huijser et al. 2016).
STUDY AREA
The HJWA (39°41′N 120°01′W) was a CDFW‐owned
53.4‐km
2
property in Sierra and Lassen counties, California
(Fig. 1). The property was at 1,585 m in elevation and the
terrain consisted of gentle slopes. We monitored collared
deer in the area in 2006–2015 and monitored underpass use
on Route 395 in HJWA in 2017–2019. The property was
acquired by CDFW in 1989 for the benefit of migratory
Loyalton‐Truckee deer and became the HJWA in 1991.
The CDFW purchased 7 additional parcels of surrounding
areas up to 2019. Much of HJWA is bordered by federal
lands owned by the United States Forest Service and Bureau
of Land Management. Prior to 1991 and during the study
period, portions of the area was grazed by livestock; it also
sustained several wildfires. Extensive wildfires occurred at
HWJA in 2007 and in 2020 that destroyed much of the
area's natural vegetation, including bitterbrush (Purshia
spp.). After the 2007 fire, cheatgrass (Bromus tectorum)in-
vaded the burned areas and reduced suitable habitat for
migratory deer as seen in many other western states
(Clements and Young 1997).
Four creeks, some ephemeral and fed by snowmelt and natural
springs, were the main natural sources of water on the HJWA:
Long Valley, Evans Canyon, Balls Canyon, and Purdy creeks.
Thevegetationoftheareawasmostlysagebrushscrub
(Artemisia spp.), cheatgrass, bitterbrush, juniper woodlands
(western juniper [Juniperus occidentalis], Utah juniper
[J. osteosperma]), wet meadows, and wetlands. The climate of
HJWA was arid with a mean summer (20 Jun–22 Sep) tem-
perature of 22.1°C and a mean winter temperature (21 Dec–
19 Mar) of 3.8°C during 2017–2019. The average monthly
summer precipitation during the study period was 2.5 cm and
the average monthly winter precipitation was 4.4 cm (National
Oceanic and Atmospheric Administration 2020). Spring
was definedas20March–19 June and fall was defined as
23 September–20 December.
Wildlife present in the area included mule deer, prong-
horn (Antilocapra americana), mountain lion (Puma concolor),
bobcat (Lynx rufus), black bear (Ursus americanus), coyote
(Canis latrans), and California quail (Callipepla californica).
The HJWA was seasonally open to the public for recreation
(hunting, hiking, wildlife viewing) during July–February.
Figure 1. The location of 3 underpasses on United States Route 395 in
Hallelujah Junction Wildlife Area, California, USA, during study period
(2006–2019). The location of Hallelujah Junction Wildlife Area in relation
to California counties is shown in the upper left corner.
2 The Journal of Wildlife Management

In May–October, cattle grazed in designated areas. Route
395 bisected the property from north to south; in the 1970s,
3 underpasses (~1.5 km apart) were built underneath the
highway within HJWA. All the underpasses were about
120 m long, 5 m high, and 6 m wide. In the center of the
underpasses, open atria allowed sunlight to pass through,
resulting in vegetation growth. About 6 km of fencing was
built on either side of the highway at the underpasses, with
deer‐proof (2.4‐m high) fencing at the underpasses and
lower (1‐m high) fencing approximately 1.5 km south of the
underpasses. Because of degrading fencing, wildlife could
access the highway in several places. One‐way exits near the
underpasses also degraded, allowing wildlife to access the
highway.
METHODS
Field Methods
Using 6 remote infrared cameras (model HC500;
RECONYX, Holmen, WI, USA), we photographed deer
moving through the 3 underpasses at HJWA. We deployed
4 of the cameras in June 2017 and 2 cameras in August
2017. The data used in this study ended June 2019, but data
collection is ongoing. We set cameras within each entrance
of the underpasses on posts about 50 cm above the ground.
We checked camera batteries and replaced memory cards
monthly. We set cameras to high sensitivity and to take
3 pictures (1 photo/second) every time movement was
detected. The cameras had an infrared flash during
low‐light periods.
We used global positioning system (GPS) data from
26 adult female mule deer from the Loyalton‐Truckee herd
during 2006–2013; 5 of the GPS‐collared deer were
non‐migratory, and we removed them from the analyses. We
captured the mule deer via darting on their summer or winter
ranges and fitted them with GPS collars (Iridium VHF collars
models G2000, G2110B and D; Advanced Telemetry
Systems, MN, USA) set to take fixes once every hour during
spring (May–Jul) and fall (Nov–Dec) migrations and
1–2 times a day at other times for 1–2 years. All procedures
involving wildlife species were approved by the CDFW and
followed guidelines from the California Fish and Game
Wildlife Restraint Handbook (Jessup et al. 2001).
We obtained Route 395 roadkill data from CalTrans for
years 2015–2019, the California Roadkill Observation
System (CROS) for years 2011–2019 (Waetjen and
Shilling 2017), and the California Highway Patrol (CHP)
wildlife‐vehicle collision data for years 2015–2017. The data
from CalTrans and CHP included highway mile or GPS
coordinates during road checks. The CROS data were re-
ported by the general public via the CROS smart phone
application and were opportunistic sightings along the
highway (Waetjen and Shilling 2017). All roadkill data
included the date the roadkill was recorded and the species
killed.
In 2019, we surveyed the highway and underpasses on
foot. We focused on areas within 3 km of the underpasses
and areas with high reported levels of roadkill. We recorded
locations of deer tracks, established wildlife trails, and
wildlife remains to document where wildlife crossed the
highway.
Analysis
We used single‐species occupancy models (MacKenzie
et al. 2002) to assess mule deer use of the underpasses using
the unmarked package (Fiske and Chandler 2011) in
Program R (version 3.5.1; R Core Team 2020). We used a
subset of the camera data that excluded consecutive
detections of the same individuals traveling in the same
underpass within 30 minutes of first detection (Lazenby and
Dickman 2013). We used only the detection probabilities,
not the occupancy estimates, from the models because the
cameras were too close together to meet the assumption of
spatial independence (MacKenzie et al. 2002, Lazenby and
Dickman 2013). We did not include population dynamics
in our modeling because the underpasses were within the
same geographic area (<1.5 km apart) and sampled at
the same time (Royle and Nichols 2003); we assumed the
population of mule deer using each underpass was the same.
To determine whether there were temporal or spatial
patterns of mule deer underpass use, we modeled detections
with the following covariates: ordinal day, season, migration
season, year, underpass location (north, middle, or south),
and camera side (east or west [i.e., side of the underpass the
camera was in]). We defined migration seasons as the pe-
riods in spring or fall when the daily number of mule deer
detections exceeded the maximum daily number of de-
tections during the summer, when only resident mule deer
were present in HJWA. These estimates of migratory
seasons corresponded with peaks of high mule deer activity
within the underpasses. We used Spearman's correlation
coefficients for the detection covariates to confirm there was
no collinearity (Spearman's ρ>0.70, P<0.05). We used
second‐order Akaike's Information Criterion (AIC
c
)to
compare models using the R package AICcmodavg
(Burnham and Anderson 2002, Mazerolle 2019). We se-
lected models with differences in AIC
c
(ΔAIC
c
)<4.0 as the
best approximations for the data and calculated the pre-
dicted detection probabilities, regression coefficients (β),
standard error, and P‐values for the covariates using the
unmarked package in R (Anderson 2008, Fiske and
Chandler 2011).
We determined approximate movement patterns for the
21 collared migratory deer by plotting straight‐line distances
between starting GPS points and subsequent points using
ArcMap (version 10.5; Esri, Redlands, CA, USA). We
determined the approximate migration dates and paths from
straight‐line distances by visually inspecting the GPS data
and plotting when collared deer departed from or arrived at
their winter or summer ranges. We estimated where
collared deer crossed Route 395 by determining where their
straight‐line distances between consecutive GPS points in-
tersected the highway. Although straight‐line distances did
not represent exactly where deer crossed the highway, we
considered it a reliable estimate because most of the
highway crossings occurred during the migration when
Caldwell and Klip •Migration and Underpasses 3

GPS points were taken every hour. We then visually in-
spected the approximate locations of highway crossings and
grouped crossings into 2 categories: crossings that were
likely at the underpasses (<0.75 km from underpass) and
crossings that were likely at the short fencing south of the
underpasses (<0.75 km from shorter, 1‐m highway fence).
These distances corresponded with clusters of crossings
around the underpasses and at the lower fencing south of
the underpasses.
To further investigate where mule deer crossed Route 395,
we also analyzed roadkill data from the area. We removed
roadkill data from different sources at similar locations re-
corded within 7‐day spans to ensure that duplicated reports
were not included in the analysis. We visually inspected the
locations of roadkills and grouped the locations into the
same categories as the GPS‐collar crossings (i.e., roadkill
<0.75 km from the underpasses, and roadkill <0.75 km
from the short fence).
RESULTS
Underpass Cameras
During 2017–2019 we detected 6,112 mule deer within
underpasses at HJWA, with means of 682.83 ±35.16 (SD)
days of camera data per site (Table 1). We determined the
approximate migration dates of mule deer: fall migrations
occurred 27 October 2017–25 December 2017 and
24 October 2018–18 December 2018, and spring migra-
tions occurred 22 March 2018–2 June 2018 and 16 March
2019–21 May 2019.
We compared 16 occupancy models to determine whether
seasonal or location variables affected detection probabilities
at camera sites; migration season, underpass location, and
underpass side of the camera were among the most im-
portant predictive variables (Table 2). Predicted detection
probabilities ranged 0.089–0.577 in the best supported
model. Detection probabilities (p) were positively correlated
with both spring migrations compared to fall 2017 (spring
2018: p=0.409−0.535, β=0.476, P<0.01; spring 2019:
p=0.450−0.577, β=0.645, P<0.001; Fig. 2) and the
southern underpass compared to the middle
(p=0.139−0.577, β=0.508, P<0.001). Detections were
negatively correlated with summer seasons compared to fall
2017 (summer 2017: p=0.170−0.254, β=−0.742,
P<0.001; summer 2018: p=0.167−0.250, β=−0.764,
P<0.001) and both winter seasons compared to fall 2017
(winter 2017: p=0.089−0.139, β=−1.485, P<0.01;
winter 2018: p=0.244−0.349, β=−0.286, P=0.054;
Fig. 2). Detection probabilities had no significant correla-
tion with underpass side of the camera (p=0.088−0.579,
β=0.022, P=0.758).
Table 1. Detection rates (number of individuals detected divided by number of days camera was deployed) and number of detections of mule deer by
migratory season at 6 camera sites in 3 underpasses at Hallelujah Junction Wildlife Area, Sierra County, California, USA, 2017–2019.
Migratory season Dates Detection rate Number of detections
a
Summer 2017 15 Jun 2017–26 Oct 2017 1.602 157
Fall 2017 27 Oct 2017–25 Dec 2017 5.538 327
Winter 2017 26 Dec 2017–21 Mar 2018 3.458 273
Spring 2018 22 Mar 2018–02 Jun 2018 17.311 1,281
Summer 2018 03 Jun 2018–23 Oct 2018 2.831 306
Fall 2018 24 Oct 2018–18 Dec 2018 7.031 387
Winter 2018 19 Dec 2018–15 Mar 2019 17.118 1,403
Spring 2019 16 Mar 2019–21 May 2019 26.853 1,824
a
Total detections excluded consecutive detections of the same individuals. Individuals were not uniquely identifiable; counts are a measure of the overall
level of deer underpass usage, not number of unique individuals.
Table 2. Mule deer top‐ranked (difference in second‐order Akaike's
Information Criterion, ΔAIC
c
<4) single‐species occupancy models at
6 camera sites in 3 underpasses at Hallelujah Junction Wildlife Area, Sierra
County, California, USA, June 2017–June 2019. We tested only the co-
variate effects on the detection probabilities (p) and not the occupancy (Ψ)
of mule deer at camera sites. We also provide the model with no covariates
(.), the number of model parameters (K), and the model weights (w
i
).
Models KAIC
c
ΔAIC
c
w
i
Ψ(.) p(migration season +underpass
location
a
)
11 4,475.11 0.00 0.66
Ψ(.) p(migration season +underpass
location +underpass side
b
)
12 4,476.40 1.33 0.34
Ψ(.) p(.) 2 4,858.80 383.69 0.00
a
Underpass location =which underpass the camera sites were located in
(middle, north, or south).
b
Underpass side =the side of the underpasses where the camera sites
were located (east or west).
Figure 2. The probability of detection of mule deer during migratory and
non‐migratory seasons (fall 2017 =F17, fall 2018 =F18, spring
2018 =SP18, spring 2019 =SP19, summer 2017 =SU2017, summer
2018 =SU18, winter 2017–2018 =W17, winter 2018–2019 =W18)
based on the occupancy model including the effects of migration seasons
and underpass location. We collected data at 6 camera sites in 3
underpasses at Hallelujah Junction Wildlife Area, Sierra County,
California, USA, June 2017–June 2019. The error bars represent 95%
confidence intervals.
4 The Journal of Wildlife Management

GPS Collars and Roadkill Data
We recorded 25 fall and spring migrations during
2006−2010 and 2013−2015 for GPS‐collared migratory
mule deer. For fall migrations, the mean departure from
summer range was 3 November ±22 days (SE) and mean
arrival to winter range was 9 November ±23 days (SE). For
spring migrations, the mean departure from winter range
was 29 April ±14 days (SE) and the average arrival on
summer range was 4 May ±15 days (SE).
Eleven of the collared deer crossed Route 395 a total of
70 times. More highway crossings occurred during fall and
spring migrations (45 spring and 14 fall) compared to other
times (11 winter and 0 summer; exact binomial test,
P<0.001). Twenty‐one estimated crossing points inter-
sected the highway at the shorter fence south of the un-
derpasses, and the remaining crossing points were clustered
around the underpasses (Fig. 3).
Thirty‐two deer roadkills were reported near HJWA on
Route 395 during 2011−2019. More roadkills were
recorded during migrations (13 spring and 9 fall) than other
times (6 winter and 4 summer; exact binomial test,
P=0.05). Roadkills were grouped near the underpasses and
at the shorter fencing (Fig. 3).
From our field observations of the highway and
underpasses during 2019, we observed mule deer tracks and
remains at the highway side of fencing at the underpasses,
where deer were likely getting through holes in the fencing
and were subsequently unable to access the underpasses. We
also observed deer tracks and trails leading to the highway at
the shorter fencing south of the underpasses. Additionally,
our cameras photographed deer on the highway side of the
fence within the underpasses multiple times.
DISCUSSION
The underpasses at HJWA were still important crossing
points for mule deer migrating across Route 395 over
40 years after they were constructed. But roadkill along the
highway and non‐underpass crossing routes were common
near and within HJWA, supporting our hypothesis that
deteriorating highway crossing infrastructure due to age and
deferred maintenance led to an increase of deer‐vehicle
collisions near underpasses. During field visits, we observed
deer tracks going towards the highway at several holes in the
deer‐proof highway fencing and lower (1‐m high) highway
fencing approximately 1 km south of the underpasses.
Additionally, 1‐way exits built near the underpasses were
not functional for all wildlife species in the area, including
mule deer. Future designs planned by CalTrans for Route
395 highway infrastructure at HJWA will focus on wildlife
permeability in general, not just mule deer, and will include
jump‐out ramps, which effectively reduce highway mortal-
ities for a variety of species (Siemers et al. 2015,
Jensen 2018).
Underpass cameras and deer GPS data confirmed that
there was higher use of the underpasses and increased Route
395 crossings during migrations compared to summer when
only resident deer were present. Although 70% of estimated
highway crossing points from GPS data were clustered near
the underpasses, approximately 30% of the crossings were
clustered at the shorter fence south of the underpasses,
indicating this was a highly used highway crossing point.
Additionally, during 2011–2019, 32 deer roadkills were
recorded with most occurring during migrations (69%), at
the underpasses, and at the southern lower fencing; the
actual roadkill volume was likely 10–30 times higher be-
cause factors such as scavengers, injured deer moving offthe
highway, and sampling frequency decrease tallied roadkills
(Slater 2002, Bager and da Rosa 2011, Zimmermann
et al. 2013). Underpass cameras captured deer traveling
through the underpasses on the highway side of the fencing,
where they were unable to access the underpass and were
subsequently trapped on the highway. Clusters of roadkill at
the underpasses illustrated how deteriorating underpass
structures contributed to WVCs. Because the shorter fence
south of the underpasses was a well‐used corridor in our
study but was not reported as an area of WVCs by Kahre's
(1980) study of the same underpasses, deer and other
wildlife may have learned to penetrate the lower fencing and
become accustomed to using this area as a crossing point
over time (Kinsey 1976, Beringer et al. 2003). Habituation
of mule deer to crossing structures and highway fencing can
take several years; patterns of mule deer use of lower
highway fencing may have increased years after construction
(Sawyer et al. 2012). The differences in our roadkill results
from Kahre's (1980) report may have been due to differ-
ences in methods and our ability to use modern sampling
tools such as roadkill reporting smart phone applications.
Figure 3. Locations of mule deer roadkills and estimated highway crossing
points from 21 global positioning system (GPS)‐collared mule deer on
United States Route 395 near Hallelujah Junction Wildlife Area, Sierra
County, California, USA. We collected roadkill data in 2011–2019 and the
GPS‐collar data is from 2006–2015.
Caldwell and Klip •Migration and Underpasses 5

Detection probabilities for deer within the underpasses
during spring migrations were higher than during fall migra-
tions. Our GPS‐collar data confirmed that mule deer followed
roughly the same migration routes in the spring and fall, as
with other mule deer (Monteith et al. 2011, Sawyer and
Kauffman 2011). We had expected that lower use of under-
passes in the fall by deer may have been because nearby routes
did not pass through the underpasses. Both GPS‐collar data
and roadkill data indicated that most of the non‐underpass
highway crossings during migration occurred during spring.
This suggests that other factors may be contributing to less
underpass use during fall migration, but incomplete roadkill
data and few collared deer may have influenced our results.
Lower fall migration underpass use could result if migrating
mule deer used HJWA as a spring stopover where they would
use both sides of Route 395 and thus exploit the underpasses
more, or if fall hunting pressure discouraged deer from using
the underpasses (Garrott et al. 1987, Kufeld et al. 1988,
Kucera 1992, Kamei et al. 2010).
Our results indicated that although underpasses along
migratory routes can effectively serve as highway crossing
points for many decades and reduce WVCs, several factors
can reduce their effectiveness. Deferred maintenance of
crossing structures decades after construction can lead to
higher WVCs around crossing structures because of
deterioration of highway fencing, 1‐way exits, and un-
finished infrastructure. Highway fencing that is <2m in
height and <5 km long around crossing structures can
contribute to WVCs (Huijser et al. 2016). There are plans
by CalTrans to repair and update the highway infrastructure
at HJWA to help reduce highway collisions for all species in
the area, including mule deer. Continued monitoring in the
area will allow us to determine whether these measures are
successful in reducing WVCs and support additional use of
the underpasses. Our study surveyed a small number of
similarly constructed underpasses in a limited geographical
area; therefore, further research is needed to determine
whether these results apply to different types of crossing
structures in a wider variety of settings.
MANAGEMENT IMPLICATIONS
The underpasses in our study were used by mule deer during
migrations immediately after construction and >40 years
later, with similar numbers of spring deer detections
(between 1,000–2,000), suggesting long‐term effectiveness
of the crossing structures. More roadkills and higher use of
non‐underpass crossings were facilitated by deterioration of
underpass and fencing infrastructure and lower (1‐m tall)
highway fencing 1.5 km south of the underpasses.
Therefore, we recommend managers ensure long‐term
maintenance of crossing structures and highway fencing to
promote connectivity and reduce WVCs. Installing taller
(>2 m) deer‐proof highway fencing >2 km from crossing
structures may also make the structures more effective.
ACKNOWLEDGMENTS
For their contributions to the maintenance of cameras and
tagging of photos, we thank C. S. McDonald Ryan, A. J.
Meyer, L. E. Pilatti, and S. A. Thomas. We thank S. M.
Holm for providing the GPS‐collar data used. We also
thank K. Kawsuniak and F. M. Shilling for providing
roadkill data. This study was funded by the CDFW.
LITERATURE CITED
Anderson, D. R. 2008. Model based inference in the life sciences: a primer
on evidence. Springer, New York, New York, USA.
Bager, A., and C. A. da Rosa. 2011. Influence of sampling effort on the
estimated richness of road‐killed vertebrate wildlife. Environmental
Management 47:851–858.
Barrueto, M., A. T. Ford, and A. P. Clevenger. 2014. Anthropogenic
effects on activity patterns of wildlife at crossing structures.
Ecosphere 5:1–19.
Beringer, J., K. C. VerCauteren, and J. J. Millspaugh. 2003. Evaluation of
an animal‐activated scarecrow and a monofilament fence for reducing
deer use of soybean fields. Wildlife Society Bulletin 31:492–498.
Brunton, E. A., S. K. Srivastava, and S. Burnett. 2018. Spatial ecology of an
urban eastern grey kangaroo (Macropus giganteus) population: local decline
driven by kangaroo–vehicle collisions. Wildlife Research 45:685–695.
Burnham, K. P., and D. R. Anderson. 2002. Model selection and multi-
model inference: a practical information‐theoretic approach. Second
edition. Springer‐Verlag, New York, New York, USA.
Caldwell, M. R., and J. M. K. Klip. 2020. Wildlife interactions within
highway underpasses. Journal of Wildlife Management 84:227–236.
Clements, C. D., and J. A. Young. 1997. A viewpoint: rangeland health
and mule deer habitat. Journal of Range Management 50:129–138.
Clevenger, A. P., and N. Waltho. 2000. Factors influencing the
effectiveness of wildlife underpasses in BanffNational Park, Alberta,
Canada. Conservation Biology 14:47–56.
Clevenger, A. P., and N. Waltho. 2005. Performance indices to identify
attributes of highway crossing structures facilitating movement of large
mammals. Biological Conservation 121:453–464.
Coe, P. K., R. M. Nielson, D. H. Jackson, J. B. Cupples, N. E. Seidel, B.
K. Johnson, S. C. Gregory, G. A. Bjornstrom, A. N. Larkins, and D. A.
Speten. 2015. Identifying migration corridors of mule deer threatened by
highway development. Wildlife Society Bulletin 39:256–267.
Conover,M.R.,W.C.Pitt,K.K.Kessler,T.J.Dubow,andW.A.Sanborn.
1995. Review of human injuries, illnesses, and economic losses caused by
wildlife in the United States. Wildlife Society Bulletin 23:407–414.
Fahrig, L., and T. Rytwinski. 2009. Effects of roads on animal abundance:
an empirical review and synthesis. Ecology and Society 14:1–20.
Fiske, I., and R. Chandler. 2011. Unmarked: an R package for fitting
hierarchical models of wildlife occurrence and abundance. Journal of
Statistical Software 43:1–23.
Gagnon, J. W., N. L. Dodd, K. S. Ogren, and R. E. Schweinsburg. 2011.
Factors associated with use of wildlife underpasses and importance of
long‐term monitoring. Journal of Wildlife Management 75:1477–1487.
Garrott, R. A., G. C. White, R. M. Bartmann, L. H. Carpenter, and A.
W. Alldredge. 1987. Movements of female mule deer in northwest
Colorado. Journal of Wildlife Management 51:634–643.
Huijser, M. P., J. W. Duffield, A. P. Clevenger, R. J. Ament, and P. T.
McGowen. 2009. Cost‐benefit analyses of mitigation measures aimed at
reducing collisions with large ungulates in the United States and Canada:
a decision support tool. Ecology and Society 14(2):15.
Huijser, M. P., E. R. Fairbank, W. Camel‐Means, J. Graham, V. Watson,
P. Basting, and D. Becker. 2016. Effectiveness of short sections of
wildlife fencing and crossing structures along highways in reducing
wildlife‐vehicle collisions and providing safe crossing opportunities for
large mammals. Biological Conservation 197:61–68.
Huijser, M. P., P. McGowen, J. Fuller, A. Hardy, A. Kociolek, A. P.
Clevenger, D. Smith, and R. Ament. 2008. Wildlife‐vehicle collision
reduction study: report to congress. U.S. Department of Transportation
Federal Highway Administration, McLean, Virginia, USA.
Jackson, S. D. 2000. Overview of transportation impacts on wildlife move-
ment and populations. Pages 7–20 in T. A. Messmer and B. West, editors.
Wildlife and highways: seeking solutions to an ecological and socio‐
economic dilemma. The Wildlife Society, Nashville, Tennessee, USA.
Jensen, A. J. 2018. Crossing corridors: wildlife use of jumpouts and
undercrossings along a highway with wildlife exclusion fencing. Thesis,
California Polytechnic State University, San Luis Obispo, USA.
6 The Journal of Wildlife Management

Jessup, D. A., W. A. Clark, and M. A. Fowler. 2001. Wildlife restraint
handbook. Eighth edition. California Department of Fish and Game,
Rancho Cordova, California, USA.
Kahre, K. S. 1980. The Lassen‐Washoe interstate deer herd: a status report.
California‐Nevada Wildlife Transactions, Sacramento, California, USA.
Kamei, T., K. Takeda, S. Izumiyama, and K. Ohshima. 2010. The effect of
hunting on the behavior and habitat utilization of sika deer (Cervus
nippon). Mammal Study 35:235–241.
Kinsey, C. 1976. Tests of two electric deer barrier forms. Minnesota
Wildlife Resources Quarterly 36:122–138.
Kucera, T. E. 1992. Influences of sex and weather on migration of mule
deer in California. Great Basin Naturalist 52:122–130.
Kufeld, R. C., D. C. Bowden, and D. L. Schrupp. 1988. Influence of
hunting on movements of female mule deer. Journal of Range
Management 41:70–72.
Lazenby, B. T., and C. R. Dickman. 2013. Patterns of detection and
capture are associated with cohabiting predators and prey. PLoS
ONE 8:1–16.
Lendrum, P. E., C. R. Anderson, R. A. Long, J. G. Kie, and R. T.
Bowyer. 2012. Habitat selection by mule deer during migration: effects
of landscape structure and natural‐gas development. Ecosphere 3:1–19.
Lendrum, P. E., C. R. Anderson, K. L. Monteith, J. A. Jenks, and R. T.
Bowyer. 2013. Migrating mule deer: effects of anthropogenically altered
landscapes. PLoS ONE 8:1–10.
MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle,
and C. A. Langtimm. 2002. Estimating site occupancy rates when de-
tection probabilities are less than one. Ecology 83:2248–2255.
Mata, C., I. Hervás, J. Herranz, F. Suárez, and J. E. Malo. 2008. Are
motorway wildlife passages worth building? Vertebrate use of road‐
crossing structures on a Spanish motorway. Journal of Environmental
Management 88:407–415.
Mazerolle, M. J. 2019. AICcmodavg: model selection and multimodel
inference based on (Q)AIC(c). <https://cran.r-project.org/package=
AICcmodavg>. Accessed 1 Apr 2019.
Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M.
Conner, R. W. Klaver, and T. Bowyer. 2011. Timing of seasonal mi-
gration in mule deer: effects of climate, plant phenology, and life‐history
characteristics. Ecosphere 2:1–34.
National Oceanic and Atmospheric Administration. 2020. National
weather service forcast office. <https://w2.weather.gov/climate/index.
php?wfo=rev>. Accessed 20 Oct 2019.
Olsson, M. P. O., P. Widén, and J. L. Larkin. 2008. Effectiveness of a
highway overpass to promote landscape connectivity and movement of
moose and roe deer in Sweden. Landscape and Urban Planning
85:133–139.
R Core Team. 2020. R: a language and environment for statistical com-
puting. R Foundation for Statistical Computing, Vienna, Austria.
Royle, J. A., and J. D. Nichols. 2003. Estimating abundance from repeated
presence‐absence data or point counts. Ecology 84:777–790.
Sawyer, H., and M. J. Kauffman. 2011. Stopover ecology of a migratory
ungulate. Journal of Animal Ecology 80:1078–1087.
Sawyer, H., C. Lebeau, and T. Hart. 2012. Mitigating roadway impacts to
migratory mule deer—a case study with underpasses and continuous
fencing. Wildlife Society Bulletin 36:492–498.
Sawyer, H., A. D. Middleton, M. M. Hayes, M. J. Kauffman, and K. L.
Monteith. 2016a. The extra mile: ungulate migration distance alters
the use of seasonal range and exposure to anthropogenic risk.
Ecosphere 7:1–11.
Sawyer, H., P. A. Rodgers, and T. Hart. 2016b. Pronghorn and mule deer
use of underpasses and overpasses along U.S. Highway 191. Wildlife
Society Bulletin 40:211–216.
Seidler, R. G., R. A. Long, J. Berger, S. Bergen, and J. P. Beckmann.
2015. Identifying impediments to long‐distance mammal migrations.
Conservation Biology 29:99–109.
Siemers, J. L., K. R. Wilson, and S. Baruch‐Mordo. 2015. Monitoring
wildlife vehicle collisions: analysis and cost‐benefitofescape
ramps for deer and elk on U.S. Highway 550. Colorado
Department of Transportation Report No. CDOT‐2015‐05, Fort
Collins, USA.
Simpson, N. O., K. M. Stewart, C. Schroeder, M. Cox, K. Huebner, and
T. Wasley. 2016. Overpasses and underpasses: effectiveness of crossing
structures for migratory ungulates. Journal of Wildlife Management
80:1370–1378.
Slater, F. M. 2002. An assessment of wildlife road casualties—the potential
discrepancy between numbers counted and numbers killed using con-
ventional census. Web Ecology 3:33–42.
Smith, D. J., R. Van Der Ree, and C. Rosell. 2015. Wildlife crossing
stuctures: an effective strategy to restore or maintain wildlife connectivity
across roads. Pages 172–182 in R. Van Der Ree, D. J. Smith, and C.
Grilo, editors. Handbook of road ecology. John Wiley and Sons,
Chichester, West Sussex, United Kingdom.
Stewart, K. M. 2015. Effectiveness of wildlife crossing structures to min-
imize traffic collisions with mule deer and other wildlife in Nevada.
Nevada Department of Transportation Research Report No. 101‐10‐
803, Reno, USA.
Trombulak, S. C., and C. A. Frissell. 2000. Review of ecological effects of
roads on terrestrial and aquatic communities. Conservation Biology
14:18–30.
Waetjen, D. P., and F. M. Shilling. 2017. Large extent volunteer roadkill
and wildlife observation systems as sources of reliable data. Frontiers in
Ecology and Evolution 5:1–10.
Zimmermann, F., A. Vicente, P. Coelho, I. Beraldi, and A. Kindel. 2013.
Vertebrate road mortality estimates: effects of sampling methods and
carcass removal. Biological Conservation 157:317–323.
Associate Editor: Scott McCorquodale.
Caldwell and Klip •Migration and Underpasses 7