!
"#$%&$'()*!+,'-)'($%!-..-/'0!,.!1(2-)()*!34!56!,)!7-2!1,%#-0!()!
8$09()*',):!;<77-%%:!$)2!=$7-!>,&)'(-0:!?,7'9!>$7,%()$!!
!!
!!
!
!!!!!!!!!!!!!!
@()$%!A-+,7'!
!
B7()/(+$%!C)#-0'(*$',70D!
E(/9$-%!A!F$&*9$):!B9=!$)2!E$7/-%%$!G!H-%%<:!B9=!
!
F(7*()($!;-/9!
=-+$7'I-)'!,.!@(09!$)2!8(%2%(.-!>,)0-7#$'(,)!
JKK!>9-$'9$I!L$%%!!
M%$/N0O&7*:!FP!Q6K5J!
!
=-/-IO-7!QKJJ!
! ((!
P/N),1%-2*-I-)'0!
!
Thank you to the North Carolina Department of Transportation for providing the necessary
funding to complete this project. We would like to offer a special thank you to Anne Burroughs
and John Kirby, North Carolina Department of Transportation, for their guidance, input, and
support throughout this project. Our gratitude is also extended to Mike Bryant, Scott Lanier,
Dennis Stewart, Colleen Olfenbuttel, and the entire Red Wolf Recovery Team (David Rabon, Art
Beyer, Chris Lucash, Ford Mauney, Michael Morse, and Ryan Nordsven). Their input, support,
and field assistance was invaluable to the success of this project. Our appreciation is also
extended to all of the field and lab technicians for their tireless efforts: Thomas Esson, J.
Bernardo Mesa, Jacob Humm, David Drewett, Casey Carbaugh, Rebecca Landis, John Vanek,
Jaya Kannan, Chris Haggard, Bryan Will, and Rebecca Fraenkel. This research would not have
been possible without their assistance. We would like to acknowledge Justin Dellinger at
Auburn University for his statistical expertise and assistance in evaluating habitat data.
4+,)0,70!
!
NC Department of Transportation, Red Wolf Recovery Program Team,
Office of Research & Development US Fish and Wildlife Service,
Raleigh, NC Manteo, NC
NC Department of Transportation, NC Wildlife Resources Commission,
Division 1 Raleigh, NC
Edenton, NC
Alligator River National Wildlife Refuge, Weyerhaeuser Company,
US Fish and Wildlife Service, North Carolina Timberlands
Manteo, NC Vanceboro, NC
Pocosin Lakes National Wildlife Refuge,
US Fish and Wildlife Service,
Columbia, NC
!
Cite this report as: Vaughan, M.R., Kelly, M.J., Proctor, C.M., Trent, J.A. 2011. Evaluating
potential effects of widening US 64 on red wolves in Washington, Tyrrell, and Dare Counties,
North Carolina. Final Report. VT-NCDOT Contract No. 09-0776-10. 55pp.!
! (((!
;$O%-!,.!>,)'-)'0!
Page No.
ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .iv
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .vi
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
BACKGROUND INFORMATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Problem Need/Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Research Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
STUDY AREA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Red Wolf Collar Deployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Objective 1: Red Wolf Home Range and Habitat Selection . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Objective 2: Significance of Habitat Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Objective 3: Identification of Significant Red Wolf Crossing Locations . . . . . . . . . . . . . . . .12
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
!
Red Wolf Collar Deployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
Objective 1: Red Wolf Home Range and Habitat Selection . . . . . . . . . . . . . . . . . . . . . . . . . .15
Objective 2: Significance of Habitat Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Objective 3: Identification of Significant Red Wolf Crossing Locations . . . . . . . . . . . . . . . .28
!
DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
Objective 1: Red Wolf Home Range and Habitat Selection . . . . . . . . . . . . . . . . . . . . . . . . . .37
Objective 2: Significance of Habitat Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Objective 3: Identification of Significant Red Wolf Crossing Locations . . . . . . . . . . . . . . . 40
LITERATURE REVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43
LITERATURE CITED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51
!
! (#!
R(0'!,.!;$O%-0!
Table No. Page No.
1. Summary of collar statistics for red wolves, 2009 – 2011 . . . . . . . . . . . . . . . .14
2. Home range size (average and range) for juvenile, disperser, and adult red
wolves in Dare, Tyrrell, Washington, Beaufort, and Hyde Counties, NC . . . .16
3. Home range size (average and range) for male and female red wolves in
Dare, Tyrrell, Washington, Beaufort, and Hyde Counties, NC . . . . . . . . . . . .16
4. Average habitat composition of home ranges for red wolves . . . . . . . . . . . . . 17
5. Average monthly home range size for red wolves . . . . . . . . . . . . . . . . . . . . . .17
6. Average habitat composition of monthly home ranges for red wolves . . . . . .18
7. Most parsimonious 2nd order resource selection function model . . . . . . . . . . .21
8. Most parsimonious 3rd order resource selection function model . . . . . . . . . . .22
9. Permeability index for US 64 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
10. The proportion of a red wolf pack home range that will be directly affected
by highway construction in Tyrrell County. . . . . . . . . . . . . . . . . . . . . . . . . . . 24
11. The proportion of a red wolf pack home range that will be directly affected by
highway construction in Dare County. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
12. Most parsimonious habitat model for crossing sites identified using GPS
-collar data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
13. Most parsimonious habitat model for crossing sites identified using remote
camera trap data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
! #!
R(0'!,.!@(*&7-0!
Figure No. Page No.
1. Project study area, Dare, Tyrrell, Washington, Beaufort, and Hyde
Counties, NC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8
2. Illustration of red wolf road crossings and approaches . . . . . . . . . . . . . . . . . .11
3. Rarefaction curves for cumulative weekly red wolf home ranges . . . . . . . . . 16
4. Relative probability of occurrence of red wolves in northeastern NC based
on 2nd order resource selection functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5. Relative probability of occurrence of red wolves in northeastern NC based
on 3rd order resource selection functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
6. Regression analysis of the relationship between traffic flow and monthly
permeability indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7. Number of red wolf packs per 82 ft. increment to be displaced by a
highway widening in Tyrrell County, NC. . . . . . . . . . . . . . . . . . . . . . . . . . . .25
8. Number of red wolf packs per 82 ft. increment to be displaced by a
highway widening in Dare County, NC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
9. Area of red wolf habitat per 82 ft. increment to be affected by a
highway widening in Tyrrell County, NC. . . . . . . . . . . . . . . . . . . . . . . . . . . 27
10. Area of red wolf habitat per 82 ft. increment to be affected by a
highway widening in Dare County, NC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
11. Photo of an adult female collared red wolf photographed at a remote
camera station along US 64 in Tyrrell County, NC . . . . . . . . . . . . . . . . . . . .28
12. GPS-collar locations for a red wolf whose home range paralleled US 64 . . .29
13. Comparison of observed red wolf crossings and randomly generated
crossings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
14. Number of red wolf highway crossings per 0.10 mi. segments as
Determined using GPS-collar locations collected on a 5-hour schedule . . . .31
15. Number of red wolf highway crossings per 0.10 mi. segments as
Determined using GPS-collar locations collected on a 30-minute schedule .31
16. Map illustration that red wolf crossing locations coincide with where
home ranges approach the highway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
17. Number of red wolf crossing events per 0.10 mi segments along US 64
calculated from combined GPS-collar and remote camera data . . . . . . . . . . 33
18. Map of important red wolf crossing locations along US 64 in
Dare County, NC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
19. Map of important red wolf crossing locations along US 64 in
Tyrrell County, NC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
! !
! #(!
"S-/&'(#-!4&II$7<!!
!!
Information reported here is the result of a 2-year study evaluating red wolf (Canis rufus) habitat
use and crossing patterns along US 64 from Columbia, NC east to the 64/264 intersection in
Manns Harbor, NC. This report includes a problem statement and background information,
evaluation of red wolf home range size and habitat selection, an analysis of probable effects to
red wolf home range and habitat availability in the event of a highway widening, identification of
current red wolf crossing locations, and suggestions for the location and type of crossing
structures to mitigate potential adverse effects of a highway widening of US 64.
We used data from 16 red wolves fitted with GPS-collars between January 2009 and April 2011
to evaluate home range size and habitat selection. Home range size for red wolves averaged 13.7
mi2 with no significant difference between males and females. Although we found no significant
difference in home range size among age classes, dispersers tended to have larger home ranges
than adults and juveniles. Red wolf home ranges were larger during winter than during other
seasons. Red wolves avoided wetter habitats such as pocosins, wetlands, and lowland forests,
leaving agriculture the best predictor of red wolf presence. Red wolves also selected for the
presence of agriculture/forest road systems for travel.
Road permeability, calculated using GPS-collar data, was 100%, thus the current 2-lane highway
does not impose a barrier effect on the red wolf population. This increases the risk of road
mortality events. A decrease in the red wolf population to the west of Columbia, NC, prevented
collaring of red wolves where widening to a 4-lane highway was completed. Therefore, we were
not able to compare highway permeability between 2- and 4-lane highways. Using a 3281 ft. (1
km) buffer, construction north of the current US 64 in Tyrrell County has the potential to remove
up to 0.16 mi2 of red wolf habitat and 6% of the home range area used by a current red wolf pack
while construction to the south will impact only 0.09 mi2 of red wolf habitat and will not displace
any current red wolf packs. East of Alligator River in Dare County, a widening of the current
highway to the south has the potential to remove up to of 0.07 mi2 of red wolf habitat and 20% of
the home range used by the only existing red wolf pack in Alligator River National Wildlife
Refuge if construction disturbs out to 3281 ft. (1 km) from the current road. Construction to the
north of US 64 in Dare County has the potential to remove up to 0.04 mi2 of red wolf habitat and
will not overlap with any current packs, based on 95% home ranges.
Through the use of GPS-collars and remote camera traps, we identified 5 important red wolf
crossing locations, 4 in Tyrrell County west of Alligator River and 1 in Dare County east of
Alligator River. The presence of agricultural fields, successional fields, and/or upland forests 328
to 492 ft. from the road provided the most parsimonious explanation for the location of crossing
sites identified using GPS-collar locations; trail/road width provided the best explanation for the
location of crossing sites identified by remote camera traps. The presence of agricultural fields,
successional fields, and upland forests as well as proximity to maintained agricultural/forest
roads at crossing sites corresponds to habitat selection results.
! #((!
Four of the 5 red wolf crossing locations we identified are suitable for crossing structures. The
most western crossing site is located within the town of Colombia, NC where retro-fitting a
wildlife underpass is not practical. Well maintained trails at least 26.24 ft. (8 m) in width leading
to and from underpasses, which connect habitats selected for by red wolves (e.g. agriculture,
successional fields, and upland forests), is suggested to optimize efficacy.
Detailed results and discussion are provided in the report below.
! !
! J!
"#$%&$'()*!+,'-)'($%!-..-/'0!,.!1(2-)()*!34!56!,)!7-2!1,%#-0!()!
8$09()*',):!;<77-%%:!$)2!=$7-!>,&)'(-0:!?,7'9!>$7,%()$!
!
@()$%!A-+,7'!
!
B7()/(+$%!C)#-0'(*$',70D!
E(/9$-%!A!F$&*9$):!B9=!$)2!E$7/-%%$!G!H-%%<:!B9=!
!
C)'7,2&/'(,)!
!
Roads have profound effects, both direct and indirect, on natural ecosystems (see reviews by
Forman et al. 2003 and Coffin, 2007). Forman and Alexander (1998) estimated that the
approximately 10.9 million hectares of public roads in the US and their related habitat loss and
degradation has affected >20% of land area in the United States. In a 43-year period (1960 –
2003), the number of registered cars increased from 74 to 231 million nation-wide and the annual
distance traveled by car in the US grew from approximately 720,000 to 2.8 million miles (Ouren
and Watts, 2005). As transportation needs increase, a rise in the amount of habitat lost and
degraded due to road construction can be expected. Indeed, the link between economic
development and transportation expansion was well documented by the mid 1960’s (Kansky,
1963; Taaffe et al., 1963; Haggett 1965). Such large-scale and multifaceted changes to
ecosystems have many detrimental impacts on wildlife (Jackson, 1999), including direct
mortality (Lalo, 1987; Harris and Scheck, 1991; Schwabe and Schuhmann, 2002), habitat
destruction (Theobald et al., 1997; Angelsen and Kaimowitz, 1999), barrier effects (Forman et al,
2003), and increased human land use activities (Bjurlin and Cypher, 2003; Coffin, 2007).
There is perhaps no other human impact as transportation infrastructure whose far-reaching,
cumulative effects on wildlife are so devastating and demanding of attention, yet so commonly
underestimated. Direct wildlife mortality results from construction injury and vehicular
collisions (Trombulak and Frissell, 2000) and has now surpassed hunting as the leading direct
human cause of vetebrate mortality on land (Forman and Alexander, 1998), with an estimated
720,000 to 1.5-million deer-vehicle crashes reported annually (Conover et al., 1995; Forman et
al., 2003). While the number of common species killed along roads is staggering, in most cases
road mortality of wildlife does not translate into population level effect. However, road related
impacts to threatened and endangered species are of particular concern. Road kill is the primary
cause of mortality in Florida for Florida panthers, black bears, key deer, and crocodiles (Harris
and Scheck, 1991) and accounts for a high percentage of deaths in Iberian lynx (Ferreras et al.,
1992). In addition, road impacts have been found in gray wolves (Thiel, 1985; Paquet and
Callaghan, 1996), desert tortoises (Boarman, 1996), and some populations of San Joaquin kit
foxes (Bjurlin and Cypher, 2003). For the declining copperbelly water snake in Indiana, road
mortality accounts for approximately 17% of all deaths (Roe et. al, 2006). Road mortality was
the second highest cause of death for red wolves in North Carolina within the 5 county recovery
zone between 1999 and 2006, accounting for 14% of mortality overall (USFWS, 2007). When
broken down by age class, vehicle strikes were the leading cause of death in dispersing red
! Q!
wolves, accounting for 19% of mortality (USFWS, 2007). The increasing incidence of wildlife-vehicle
collisions also presents a real issue for humans, as they claim hundreds of lives, cause
tens of thousands of injuries, and inflict an enormous monetary cost for medical treatment and
vehicle repair each year nationwide (Forman et al., 2003). For example, in 1993 1.5 million deer-vehicle
crashes were reported in the US leading to $1.1 billion in vehicle damages (Durbin,
2004). Deer-vehicle collisions have been reported to cause 29,000 human injuries and claim 211
lives (Conover et al., 1995).
Beyond direct mortality, roads can negatively affect wildlife populations by degrading habitat
quality (Theobald et al., 1997; Carr et al., 2002), fragmenting habitat and populations (Oxley et
al., 1974; Trombulak and Frissell, 2000; Nellermann et al., 2001), hindering gene flow (Gerlach
and Musolf, 2000; Epps et al., 2005; Riley et al., 2006), skewing sex ratios (Gibbs and Steen,
2005), and limiting dispersal (Beier, 1995). Most roads exhibit a distinct trade-off between
permeability and road kill (Forman and Alexander, 1998). A highly permeable road might result
in a high level of wildlife/vehicular collisions, whereas an impermeable road might have few
road kill events. Yet this decrease in road kill comes at the expense of habitat connectivity. As
individuals lose their mobility and gene flow is reduced, portions of the population become
isolated. In the event of local extinction due to some stochastic event, fragmentation can make
recolonization of previously-occupied habitat impossible (Theobald et al., 1997). These affects
are of particular concern in small populations. Small populations have an increased risk of
extinction due to demographic stochasticity, decreased heterozygosity, genetic drift, inbreeding,
and low effective population size (Caughley, 1994), all of which can be exacerbated though road
construction and expansion-related barrier effects. Social organization may also be affected by
spatial change leading to population instability (Krausman et al., 2004).
Low population densities and large home ranges make carnivores particularly vulnerable to the
effects of habitat fragmentation by roads (Whittington et al., 2005). A highway was found to
restrict gene flow in a Cleveland, Ohio coyote population and direct the movements of migrants
towards urbanizing centers (Rashleigh et al., 2008). Riley et al. (2006) found that coyote and
bobcat populations in southern California separated by a major freeway exhibited genetic
differentiation, suggesting that the freeway is a barrier to dispersal. Even when they do not
constitute an absolute physical barrier, high-use roads can lead to avoidance behavior in canids
affecting their ability to move across a landscape (Kaartinen et al., 2005, Paquet and Callaghan,
1996, Whittington et al., 2004). For those that do cross, heightened territorial behavior along
roadways can discourage reproductive success, again limiting gene flow (Riley et al., 2006). The
degree to which a road affects canid survival is dependent on the specific situation, and
sometimes no detrimental effects are observed, as is the case with a California population of San
Joaquin kit foxes (Cypher et al., 2009). Some documentation exists regarding wolves in the
vicinity of highways. A study tracking gray wolf dispersal in Minnesota found that wolves were
willing to cross major highways to colonize areas in Wisconsin and Michigan (Mech et al.,
1995). For gray wolves (Canis lupus), a 4-lane unfenced highway in Wisconsin seemed to not
influence wolf movements (Kohn et al., 1999). In contrast, a 4-lane fenced highway in Banff
National Park in Alberta, Canada, appears to hinder wolf movements (Paquet and Callaghan,
1996), although crossing structures mitigated its barrier effect to some degree (Clevenger and
Waltho, 2000, 2005). In Spain, wolves whose home ranges were greater than 5 km from the
highway crossed a 4-lane, fenced highway via vehicle bridges (Blanco et al., 2005), however
! T!
those living in close proximity to the highway (<5 km) crossed the highway only after severe
habitat disturbance. Gray wolves in Canada and Spain seem to prefer large, open wildlife
overpasses (Forman et al., 2003). In general, a large void exists in addressing factors that affect
how large carnivores use passages. Major transportation corridors bisect and potentially fragment
most of the major ecosystems that still support wide-ranging carnivores. Increased concerns
expressed by transportation and natural resources agencies regarding mitigation planning for
large carnivores highlight the need for more information and research in this area (Forman et al.,
2003).
The use of wildlife crossing structures can mitigate some of the negative effects associated with
highways (Forman and Alexander, 1998). The appropriate type of crossing structure to mitigate
road effects varies with species (Mata et al., 2008). A study monitoring the success of multi-species
highway underpasses following the highway-widening just west of this current study area
found that bobcats, black bears, and foxes utilized underpasses (McCollister and van Manen,
2010), however there were no confirmed detections of coyotes or red wolves using the
underpasses. Because the red wolf population of both this study and the one cited above is the
only wild red wolf population, there does not exist literature on the preference of red wolves for
particular crossing structures. However several studies have found that while gray wolves will
use wide tunnels and underpasses (Clevenger and Waltho, 2000, Kusak et al., 2009), they prefer
open overpasses (Forman et al., 2003, Kusak et al., 2009). In Banff, wolves select for taller
underpasses close to town (Clevenger and Waltho, 2000). Coyotes on the other hand have been
found to use underpasses of a wide variety of sizes, from pipe culverts to wide underpasses, as
long as they did not connect developed areas (Ng et al., 2004). Likewise, a study in Virginia
found that coyotes readily used a variety of underpasses (Donaldson, 2007). The Wildlife
Crossing Structure Handbook published in 2011 by the Federal Highway Administration
recommends that underpasses geared towards large mammals (deer, bears, and wolves) and high
mobility medium sized mammals (coyote, fox, and likely the category red wolves would be
placed) should be greater than 32 ft. in width and greater than 13 ft. in height and that overpasses
be at least 50 ft. wide (Clevenger and Huijer, 2011). If designing mitigating structures solely for
high mobility medium sized mammals, underpasses and culverts with a diameter of 4ft. have
been effective (Clevenger and Huijer, 2011). However, it appears that the structural components
of crossing structures play a larger role in determining success for ungulates (Clevenger and
Waltho, 2005; Gagnon et al., 2011) while habitat connectivity plays a larger role in the
successful use of crossing structures for carnivores (White and Ernst, 2004; Singleton et al.,
2005; Riley et al, 2006; Kindall and van Manen, 2007).
Underpasses can function as effective crossing structures for wolves, but high variability in use
indicates that consideration of social interactions, placement, construction specifications and
distance between crossings is essential for success (Paquet and Callaghan, 1996). Animals do not
treat all sections of a roadway indiscriminately, so crossing funnel areas and natural habitat
linkages at the landscape level must be identified. White and Ernst (2004), Singleton et al.
(2005), and Kindall and van Manen (2007) all stress the need to identify habitat linkages across
barriers to properly place crossings. In addition, Roger and Ramp (2009) discuss the importance
of species-specific habitat use data in determining roadway impacts. Thus, it is imperative not
only that wildlife underpasses be constructed in areas identified as high use for crossings
(Scheick and Jones, 1999), but habitat variables at crossing locations be collected as well to
! 6!
model what landscape factors predict crossings. Often times it is also necessary to install
exclusionary fencing in addition to a crossing structure to guide animals to the crossing point and
discourage crossing at road-level (Baker, 2005).
Indeed, one study in Portugal found that fencing had a funneling effect, directing larger animals
towards culverts (Ascensao and Mira, 2006). Similarly, fencing along a highway reduced elk-vehicle
collisions by 97% and other wildlife-vehicle collisions by 64% in Arizona (Gagnon et al.,
2010) and likely lead to a decease in white-tailed deer-vehicle collisions in North Carolina (Jones
et al., 2010). A study in Germany found that fencing successfully reduced wildcat road mortality
by 83% (Klar et al., 20009) and fencing along with culverts lowered wildlife-vehicle collisions
by 93.5% in Paynes Prairie State Preserve (Dodd et al., 2004). Ungulate-vehicle collisions were
reduced by 80% following the installation of roadside fencing in Banff National Park, Canada
(Clevenger et al., 2001). Jaeger and Fahrig (2004) developed a model to look at the trade-off
between reductions of road kill and increased barrier effect due to fencing installation. They
found that below a certain traffic volume, the barrier effect of fencing is harmful to a population
and therefore they only recommend the use of roadside fencing when traffic volume is high (e.g.
high risk of road mortality) and the target species does not show behavioral avoidance of roads
(Jaeger and Fahrig, 2004). Both Clevenger et al. (2001) and McCollister and van Manen (2010)
found that while fencing reduced wildlife-vehicle collisions close to underpasses, wildlife-vehicle
collisions increase approaching fence ends. McCollister and van Manen (2010) found
that road mortality was higher in fenced highway segments as compared to unfenced segments
due to the increased mortality where roadside fencing ends. Therefore, if non-continuous fencing
is used, it may be necessary to modify fence ends to direct wildlife away from the highway
(Clevenger et al., 2001). Ungulates are the focal species for most studies that successfully
demonstrate the effectiveness of roadside fencing. Fencing may be less effective for carnivores
as they often go over (e.g. black bears) or under (e.g. coyotes) fencing (Clevenger et al., 2001).
Indeed, the use of fencing did not increase culvert use by bobcats in Texas (Cain et al., 2003) and
actually lead to an increase in wolf road mortality in Spain (Colino-Rabinal et al., 2011).
Burying roadside fencing can help to discourage some species from digging under the fence
(Clevenger et al., 2001).
Clearly, an understanding of red wolf activity patterns, movements, and habitat use are all
needed in the vicinity of US 64 and across the Albemarle Peninsula. This study assessed red wolf
home range, habitat selection, and highway crossing patterns along the US 64 corridor with the
use of GPS collars and remote cameras to determine important red wolf habitat and to identify
significant red wolf highway crossing locations. In addition, this research examined which
landscape attributes promote red wolf use of crossing locations to increase the success of
mitigating structures.
M$/N*7,&)2!C).,7I$'(,)!
!
The North Carolina Department of Transportation (NCDOT) is planning a highway
improvement project for US 64 in Tyrrell and Dare Counties North Carolina, which will extend
across the full length of the Albemarle Peninsula when completed, separating the northern
section of the 5 county (Washington, Tyrrell, Dare, Hyde, and Beaufort counties) red wolf
recovery zone. The effects of the highway widening on red wolf recovery and conservation could
! U!
be substantial, potentially creating a barrier to movement and gene flow of red wolves and other
wildlife from one side of the highway to the other. In addition, the habitat loss associated with a
highway widening likely will disrupt red wolves living adjacent to the existing highway causing
a shift in current home ranges. Any shifts in home ranges have the potential to affect social
order. Even in the absence of a barrier effect, the project may lead to an increase in vehicle
related deaths as wolves attempt to cross a wider highway with increased speed limits. In
addition, there is a potential to concentrate prey and herbaceous food sources at highway edges,
attracting wolves, coyotes, black bears, white-tailed deer and other wildlife, increasing the risk of
vehicle collisions. Highway barrier effects, habitat loss, social disruptions, and road mortality
resulting from the highway widening may culminate in reduced red wolf population viability.
Problem Need/Definition
Viable populations of wildlife depend, in part, on dispersal to maintain genetic diversity.
Whether natural or man-made, barriers to dispersal are of concern to wildlife managers. For
restored or recovering populations, potential barriers such as highways or large fenced areas
magnify in importance because of their potential to restrict or retard growth and genetic diversity
in small wildlife populations. Roads, in particular, recently received attention with respect to
large carnivore population dynamics related to increased direct (vehicle collisions) and indirect
(changes in behavior that affect food acquisition) mortality (Trombulak and Frissell, 2000).
Forced spatial change also may affect area-wide social organization and thus population stability,
and increased noise or activity levels may initially affect wildlife behavior (Krausman et al.,
2004).
For the past several years the North Carolina Department of Transportation (NCDOT) has been
planning a proposed project to widen US 64 from 2 to 4 lanes from Raleigh to Manteo, North
Carolina. With respect to the segment of US 64 already widened and elevated between Plymouth
and Columbia by 2005, preliminary data collected by the U.S. Fish and Wildlife Service
indicates red wolf (Canis rufus) movements and gene flow, including dispersal and home range
size, may already be restricted by that highway segment. Remaining sections of US 64 planned
for widening are the approximate 15.5-mile section from Columbia to Alligator River, and 11.8-
mile section that runs through Alligator River National Wildlife Refuge.
The nature of the US 64 widening project calls into question important ecological and regulatory
considerations that, together, mean data collection is needed to assist with science-based
decisions and project design. Red wolves will be involved in two federal regulatory processes
pertinent to widening of US 64, namely, project consultation under Section 7 of the U.S.
Endangered Species Act of 1973, and assessment of “refuge compatibility” under the National
Wildlife Refuge System Improvement Act of 1997 (Public Law 105-57), along with the National
Wildlife Refuge System Administration Act of 1966 (16 U.S.C. 668dd-668ee), as amended.
These processes allow cooperation toward achieving a project that takes into account human
safety, traffic management, and wildlife concerns that include passage, mortality, large-sized
animals, and multiple wildlife refuge values. Refuge considerations include endangered species
conservation, waterfowl management, wildlife habitat with associated species, hydrology,
wetlands, reptiles and amphibians, public use, fire management, exotic species management, etc.
! 5!
The effects of US 64 widening on red wolf recovery and conservation could be substantial.
Widening US 64 may be accompanied by increased speed limits, and likely will create a barrier
to movement of red wolves and other wildlife from one side of the highway to the other. Thus, it
is imperative that wildlife crossing structures be constructed in areas identified as high use
crossings by red wolves, bears, deer, and other species (Scheick and Jones, 1999). Completed
and planned phases of the US 64 widening project extend across the full length of the Albemarle
Peninsula, separating and otherwise affecting the entire northern quarter of the 5-county red wolf
experimental population area.
Construction of the highway itself most likely will directly disrupt the red wolf population, along
with other wildlife populations (e.g., black bear, white-tailed deer) living adjacent to the existing
highway during the 1-2 year construction period. These disruptions may cause red wolves to
shift out of their current home ranges or territories during the construction phase and move into
areas already occupied by other red wolves, causing social disruptions and ripple effects across
the Albemarle Peninsula. While the disruption due directly to construction will be short-term,
effects on the red wolf population may be long lasting and even permanent. Habitat loss, social
disruptions, and ripple effects, as a result of direct, indirect, and cumulative effects of a highway
widening, may result in a reduction of the red wolf population, its gene flow, and gene diversity,
by some unknown quantity.
Vehicle strike mortality significantly impacts the wild red wolf population on the Albemarle
Peninsula in North Carolina (USFWS, 2007). Of 166 known adult red wolf loses since 1999, 23
were killed in vehicle strikes. Vehicle strikes are three times higher in non-breeder (19%) vs.
breeder (6%) red wolves in the designated experimental population area. This is partly explained
by single red wolves dispersing or roaming over large distances.
Studies are needed to assess how the red wolf population has utilized the area since restoration
began. An examination of which landscape attributes promote red wolf use would be helpful,
along with a thorough assessment of site-specific habitat availability.
More specifically, the potential problems or benefits examined for red wolves in association with
US 64 widening should include the following concerns.
1. Vehicle mortality of red wolves and associated human safety.
2. Reproduction and survival.
3. Considerations of placement of underpasses or overpasses.
4. Changes in red wolf habitat, prey, home range size, dynamics, and associated landscape
fragmentation.
5. Effects upon red wolf activity, movements, gene flow, dispersal, territory dynamics,
social organization, pack integrity, habitat use, and land occupancy.
6. Ripple effects throughout the red wolf population, across the Albemarle Peninsula.
7. Changes in red wolf numbers pre-project, during project, and post-project.
8. Influences on eastern coyotes, a competitor and threat to red wolves.
9. Effects upon monitoring of red wolves and eastern coyotes.
! V!
It is important to understand the effect of canid activity and movements out of or into the
experimental area along the expanded highway in western portions of the Albemarle Peninsula.
Possible study topics include coyote/red wolf interactions and retrospective examination of
adaptability of coyote/hybrids vs. red wolves in the face of significant habitat change and/or
significant project construction.
Research Objectives
The objectives of this research project are to:
1. Evaluate wolf habitat use along the entire US 64 corridor from Plymouth to the US
64/264 intersection
2. Evaluate the significance of red wolf habitat changes anticipated from the proposed
highway project from Columbia to the US 64/264 intersection in terms of movements,
survival, reproduction, home range shifts, and social organization.
3. Identify significant red wolf crossing areas to determine where wildlife crossing
structures or other design features could be placed to minimize adverse project effects on
red wolves.
!
4'&2<!P7-$!
The only wild population of red wolves occurs on more than 2,567 mi2 of federal, state, and
private lands in 5 counties (Beaufort, Dare, Hyde, Tyrrell, and Washington) in northeastern
North Carolina (Figure 1), known as the Red Wolf Recovery Zone (RWRZ). Two of the northern
counties within the RWRZ, Tyrrell and Dare Counties, were the focal point of this study because
they contain the remaining 27.34 mi of US 64 to be widened. Federal lands within the study area
include Pocosin Lakes National Wildlife Refuge, Alligator River National Wildlife Refuge, and
a bombing range shared by the Navy and Air Force. State land consists of numerous game
management properties, while private lands are primarily made-up of timber plantations,
agricultural fields, and a few developed residential and commercial properties.
The most prevalent land cover types within the study area, as identified by the North Carolina
Gap database (2009), are agricultural fields (~30%) planted primarily with wheat, corn, soybean,
cotton, and potatoes; commercial pine plantations (~15%); pocosin (~15%); non-riverine swamp
forests (~10%); and saltwater marsh or open water (~10%). Climate within the study area is
characterized by 4 full seasons of nearly equal length with annual precipitation averaging 50 in.
Temperatures range from a mean of 41°F in winter to 80.6°F in summer. Elevation ranges from
sea level to 164 ft. (Beck et al., 2009). Carnivores that co-occur with red wolves within the study
area include gray foxes (Urocyon cineroargenteus), red foxes (Vulpes vulpes), coyotes (Canis
latrans), feral dogs (Canis lupus familiaris), bobcats (Lynx rufus), black bears (Ursus
americanus), and various mustelids.
! W!
Figure 1. The study area (highlighted in gray) is located within 2 of the northern counties,
Tyrrell and Dare, of the 5 county red wolf recovery zone in northeastern North Carolina. The
study area focuses on the remaining 27.3 mi section of US 64, between Columbia, NC and the
US 64/264 intersection, to be expanded from a 2- to 4-lane highway.
! X!
E-'9,20
Capturing and collaring of animals: From January 2009 to April 2011, adult and juvenile red
wolves were captured by USFWS biologists and fitted with mortality-sensitive Lotek GPS 4400S
collars (Lotek Wireless, Inc., Ontario, Canada). Red wolves > 2 years old were classified as
adults, < 2 years old as juveniles, and < 9 months old as pups. Pups were not fitted with GPS
collars because typically they were too small to safely wear collars. Prior to deployment, GPS
collars were remotely programmed to record locations every 5 hours with a nested program to
collect a position every 30-minutes for a 5 hour period daily. The nested 30-minute program was
scheduled to rotate around the 24-hour clock to capture detailed movements. Each collar emitted
a VHF locator beacon each day from 0900 – 1200, allowing us to locate collared animals every
12 weeks on the ground and remotely download stored data.
Objective 1: !"#$%#&'()*$+(,#-.&#&(%/'(#$*01(&,'('0&.2'(34(56(7*22.8*2(+2*9(:$;9*%&,(
<=(&*(&,'(34(56>?56(.0&'2/'7&.*0(
Independence of animal movements: To address the issue of correlation of GPS location data
between pack mates, we calculated home ranges (Getz et al., 2007) for all collared wolves in a
single pack then associated locations for each animal with the corresponding isopleth. Next,
Spearman correlation matrices were used to determine the similarity of home ranges and habitat
use among all collared animals within the pack. This determined if animals within a pack should
be treated separately or if habitat use, selection, and home range of one collared animal was
representative of the entire pack.
Home range analyses: Following the conclusion of field work, rarefaction curves of cumulative
weekly home ranges were calculated on all complete data sets for each collared animal to
determine the relationship between length of time collar was deployed and when size of home
range stabilized (Bekoff and Mech, 1984). Starting by calculating size of home range of an
animal during the first week of collar deployment, we calculated size of home range of the
animal during the second week of collar deployment and so on until the complete data set for
that animal was included in calculating size of home range. Ninety-five percent home range
isopleths were constructed using adaptive nearest neighbor convex hull methods (Getz et al.,
2007). Animals whose home ranges did not stabilize in size were excluded from subsequent
analyses. Given the varying age, dominance, and sex of the animals that were collared, and that
home range composition between packs with collared animals may vary; we assumed that all
factors influencing stabilization of size of home range were captured sufficiently. For individuals
whose home range stabilized, monthly home ranges were constructed according to Getz et al.
(2007) to examine short-term and seasonal variations in home range composition and size.
Overall and monthly home ranges were overlaid onto habitat maps developed by NC GAP to
determine percent composition of home ranges. Habitat types included agricultural fields,
wetlands, upland forests, lowland forests, successional fields, and pocosin (areas covered with
evergreen vegetation and inundated with water). We used one-way ANOVA to test for
differences in overall home range size among age classes and Student’s t-tests to test for a
difference in home range size between sexes. Student’s t-tests were also used to determine if
seasonal variation in monthly home range size and composition for each habitat type were
! JK!
significantly different. Following previous studies (Phillips et al., 2003; Chadwick et al., 2010;
Hinton et al., 2010), we recognized a summer (April – September) and winter (October – March)
season. Significance was set at ! "0.05.
Habitat use and selection analyses: Resource selection functions (RSFs; Manly et al., 2002)
were used to examine 2nd order (home range) and 3rd order (within home range) habitat use by
red wolves (Johnson, 1980). Resource selection functions were developed using use/availability
data with a binomial distribution (Manly et al., 2002).
For 2nd order habitat use, we considered the entire 5 county red wolf recovery area as available
habitat and all locations of each GPS-collared animal occurring within its respective 95% home
range (Getz et al., 2007) as used habitat. For 3rd order habitat use, the entire 95% home range
(Getz et al., 2007) was considered to be available. All locations of each animal contained within
its respective 95% home range were combined to examine 2nd and 3rd order habitat use for the
entire population. An equal number of random points, compared to locations, were generated
within the available areas for 2nd and 3rd order habitat use, respectively. Distance to road and
water, human density (people per square mile), and habitat type were determined for all used and
random locations. Habitat types were the same as those for determining home range
composition. After combining used and random locations for each order of habitat use, RSFs
were developed for each order of habitat use which contained habitat type, distance to roads and
water, human density, and all biologically meaningful interactions (habitat type by distance to
roads, habitat type by human density, and distance to roads by human density). Animals were
monitored for varying lengths of time, had different numbers of locations, were of different age
classes, and different sexes. Therefore each animal could have potentially influenced the RSFs
more or less than another animal. Thus to make sure that no animal biased the RSFs, preliminary
2nd and 3rd order RSFs were developed using a sampling with replacement method in which each
animal was excluded once from calculation of a RSF while all other animals were included. For
the 3rd order RSF, a random effect for animal was included in the RSFs to account for differences
in habitats available to each animal. Akaike’s information criterion corrected for small sample
sizes (AICc) was used to choose the most parsimonious RSF from the global (all possible
variables included) RSF and all possible subsets for each order of habitat use (Burnham and
Anderson, 2002). Twenty-five percent of used and random locations for each order of habitat use
were not used in developing all RSFs to evaluate fit of most parsimonious RSFs using cross-validation
(Johnson et al., 2006). The most parsimonious RSFs that were shown to have a good
fit to the data were projected in a GIS to create habitat suitability maps depicting areas of high,
medium, and low quality habitat and probability of occurrence of red wolves.
AICc weights of most parsimonious 2nd and 3rd order RSFs were compared to determine whether
habitat type, distance to roads and water, and density of humans were scale dependent for red
wolves. The RSF with the greatest AICc weight demonstrated the scale at which the variables of
interest and associated interactions influenced habitat use the most. Statistical analyses were
conducted in R 2.11.1 (R Development Core Team 2010) and spatial analyses using ArcGIS 10
(ESRI® ArcMap™ 10, Copyright © 1999-2010 ESRI Inc.) and Geospatial Modeling
Environment 0.5.3 (Beyer, H. L., Copyright © 2001-2010 Spatial Ecology LLC).
! JJ!
Objective 2: !"#$%#&'(&,'(/.10.+.7#07'(*+(2'8()*$+(,#-.&#&(7,#01'/(#0&.7.@#&'8(+2*9(
&,'(@2*@*/'8(,.1,)#;(@2*A'7&(+2*9(=*$%9-.#(<=(&*(&,'(34(56>?56(.0&'2/'7&.*0B(
(
Quantifying barrier effects using passage rates: Following Dodd et al. (2007), we quantified
the barrier effects of US 64 by calculating a permeability index. A permeability index is a
passage rate measuring an individuals willingness to attempt a road crossing and is calculated by
using the following equation: #crossings/(#crossings + #approaches), where an approach is
defined as a red wolf entering into a 164 ft. buffer zone around the highway without crossing
(see Figure 2). The 164 ft. (50 m) buffer zone was determined by measuring the distance
between US 64 and the boundary of the closest red wolf home range (95% MCP) to the highway.
This was done to exclude movements within a home range from being counted as an approach.
The permeability index ranges from 0 to 1, with 0 indicating an impermeable road and 1
indicating 100% permeability. Permeability indices were calculated for the 30-minute and 5-hour
data sets separately. An overall permeability index (using total number of crosses and approaches
from all study animals) for the duration of the study was calculated as well as monthly
permeability indices. We used a paired t-test to compare monthly permeability indices calculated
using the 30-minute and 5-hour data sets.
Using 30-minute monthly permeability indices, regression analysis was then used to determine if
a relationship existed between monthly permeability indices and monthly traffic flow along the
existing US 64.
Figure 2. A crossing was defined as a line connecting two points on opposite sides of a road that
intersects the roadway. An approach was defined as any excursion from a point further than 164
ft. from the road to a point within 164 ft., and then back, without crossing.
Assessing the effect of the highway widening on current red wolf territories: To determine the
potential for the highway widening to displace current red wolf packs, buffers at ~ 164 ft. (50 m)
intervals were constructed around the current US 64. The buffers were then overlaid on current
red wolf home range locations and the number of home ranges intersected by each buffer was
counted. Where the buffers intersected home ranges, the percent of total home range intersected
was calculated. As with the home range analysis above, only one home range per pack was used
when the movements among individuals of a pack were correlated.
! JQ!
Assessing the effect of the highway widening on important red wolf habitat: To determine the
potential for the highway widening to affect important red wolf habitat, buffers around the
current US 64 at ~164 ft. (50 m) intervals were overlaid on a habitat map. The area of available
red wolf habitat was then calculated within each buffer zone.
Objective 3: C8'0&.+;(/.10.+.7#0&(2'8()*$+(72*//.01(#2'#/(&*(8'&'29.0'(),'2'().$8$.+'(
72*//.01(/&2%7&%2'/(*2(*&,'2(8'/.10(+'#&%2'/(7*%$8(-'(@$#7'8(&*(9.0.9.D'(#8"'2/'(
@2*A'7&('++'7&/(*0()*$"'/B!
Determining crossing locations and rates using GPS collars: Because of the occurrence of
different collar schedules, wolf locations were sub-sampled into both 5-hour and 30-minute
intervals. The following methods were used to analyze data collected for each frequency, 5-hour
and 30-minute, respectively. Using ArcGIS v9.3 (ESRI® ArcMap™ 9.3, Copyright © 1999-
2010 ESRI Inc.), we divided the 27.3 mi-section of US 64 into 273 segments, each 0.10 miles
long. To determine road crossings, we used the Home Range Tools v9 extension for ArcGIS to
calculate the travel path of each individual by connecting consecutive GPS fixes. We then
overlaid the travel paths on the segmented highway layer and counted the number of crossings
per highway segment for each individual. A crossing was defined as two consecutive fixes on
opposite sides of the highway (see Figure 2). Crossing rates for each individual were determined
by dividing the number of crossings by the number of days the collar was actively collecting data
for each collection frequency sub-sample. Total crossing frequencies per segment were plotted in
a histogram to identify the location of key red wolf crossing areas.
Statistical Analysis: To test the hypothesis that the crossing distribution calculated using GPS
collar locations was different from a random crossing distribution, an equivalent number of
random line segments were drawn between the GPS locations for each red wolf. To approximate
actual red wolf movement, random segment lengths were constrained to less than or equal to the
maximum distance moved by a red wolf for the 30-minute and 5-hour data sets, respectively.
Crossing frequencies for the random segments were calculated for each highway segment
following the methods above. The distributions for the GPS crossing frequencies and the random
crossing frequencies were compared using a Kolmogorov-Smirnov test (Clevenger et al., 2001).
We used a t-test to test for differences in crossing rates between male and female wolves and a
one –way ANOVA to test for differences in crossing rates among age classes.
Determining crossing locations using Camera Traps: Even with 30-minute locations, the GPS
collars likely did not catch all red wolf crossing events. To capture additional crossing events,
remote cameras were placed at canal crossings along the 27.3 mi stretch of US 64 within 328 ft.
(100 m) of the roadside. Because drainage canals exist along the entire length of US 64, canal
crossings serve as an access point for animals to reach the highway. We used both film and
digital remote cameras triggered by laser or heat disturbance. All cameras were active 24-hours
per day to maximize the number of crossings captured. Cameras were active from July 2009 to
March 2011. However, the number of trap nights varied for each camera station so captures were
reported per 100 trap nights. To avoid pseudoreplication, consecutive photos of an individual
animal were considered a single event.
! JT!
Using GPS locations, camera stations were associated with one of the 273 segments along US
64. Not all segments had camera stations. Total crossing frequencies per segment were plotted in
a histogram to identify the location of key red wolf crossing areas as identified by cameras.
Evaluating habitat characteristics at crossing sites identified by GPS collar locations: Using
the NC GAP habitat map, we extracted the habitat type for each of the 273 highway segments at
164 ft. (50 m) intervals starting at the road to a distance of 656 ft. (200 m) perpendicular to the
segment (ArcGIS v9.3). Segments that had at least one crossing were coded with a 1 and
segments without crossings were coded with a 0. Logistic regression was used to evaluate 5 a
priori models developed using site-specific habitat type at different distance intervals and the
occurrence of a red wolf crossing. The most parsimonious model was chosen using AIC
corrected for small sample size (AICc) (Burnham and Anderson, 2002), with models ranked
using !AICc.
Evaluating habitat characteristics at crossing sites identified by camera traps:
Using the NC GAP habitat map, we extracted the habitat type for each of the camera stations at
164 ft. (50 m) intervals starting at the camera sites to a distance of 656 ft. (200 m) perpendicular
to US 64 (ArcGIS v9.3). In addition, the width of the access road/trail was measured at each
camera station. Camera sites that captured red wolves were coded with a 1 and camera sites that
did not capture red wolves were coded with a 0. Logistic regression was used to evaluate 5 a
priori models developed using habitat variables and trail width at camera placement. The most
parsimonious model was chosen using AIC corrected for small sample size (AICc) (Burnham
and Anderson, 2002), with models ranked using !AICc.
! J6!
A-0&%'0!
Capturing and collaring of animals: Between January 2009 and April 2011, the USFWS Red
Wolf Team deployed 32 of 40 collars. Due to a decrease in red wolf population in Washington
County, North Carolina, the 8 collars reserved for red wolves living in the vicinity of the
previously expanded portion of US 64 could not be deployed. Thirteen of the 32 collars deployed
were placed on females (8 adults, 5 juveniles) and 19 on males (8 adults, 11 juveniles). The
average collar deployment was 14.8 months (range: 4 to 30 months) and average collar success
in obtaining GPS locations was 86.0% (range: 63.6% to 97.5%). In total, 39, 573 successful red
wolf locations were collected. We used 6 different collar schedules: 30-minute locations for 5
hours per day, 5-hour locations, 5-hour locations with the nested 30-minute schedule for 5-hours
per day, 11-hour locations, 12-hour locations, and 23-hour locations (Table 1).
Table 1. Summary of collar statistics for collared red wolves in Dare, Tyrrell, Washington,
Beaufort, and Hyde Counties, NC.
! JU!
Objective 1: !"#$%#&'()*$+(,#-.&#&(%/'(#$*01(&,'('0&.2'(34(56(7*22.8*2(+2*9(:$;9*%&,(
<=(&*(&,'(34(56>?56(.0&'2/'7&.*0(
(
Home range analyses: Movements of individuals within the same pack were highly correlated
(rs = 0.87–0.91); therefore only 1 animal per pack (chosen randomly) was used in the following
analyses. After removing those individuals where multiple animals in a pack were collared, we
calculated cumulative weekly home ranges for 21 of 32 animals (Figure 3). Following analysis,
we removed 5 (1 juvenile, 2 dispersers, 2 adults) additional individuals from our sample due to
an inadequate number of locations to capture a complete home range. Overall home range varied
between 2.61 mi2 and 38.19 mi2 with a mean of 12.93 ± 9.50 mi2!
(
Figure 3. To determine if an adequate number of locations were obtained from each wolf to
capture home range area, rarefaction curves of cumulative weekly home ranges were calculated
for 21 red wolves of different age groups and sexes collared from January 2009 to April 2011. A
home range is considered to be at equilibrium at the point that the home range area no longer
increases and reaches a plateau (Bekoff and Mech, 1984). Home range did not reach equilibrium
for 5 of the 21 wolves in our sample (6, 7, 11, 13, 14), thus they were excluded from further
home range analyses. Collared red wolves were located in Tyrell, Dare, Washington, and Hyde
Counties, NC.
! J5!
Although home range size among age classes (F =2.71, P =0.14) and between sexes (t10 = 2.10,
P = 0.57) did not differ significantly (Tables 2 and 3), the home range size of dispersers tended
to be larger than those of juveniles or adults. Five of 8 animals that died while collared were
dispersing and 2 additional dispersers were removed from further home range analyses due to
inadequate data.
(
Table 2. Average and range of 95% home range areas for three age classes of red wolves. A
local convex hull method was used to calculate home range from GPS collar locations. The home
range analysis was generated from 16 red wolves collared in Washington, Tyrell, Dare, Hyde,
and Beaufort Counties, NC from January 2009 to April 2011.(
!
!!!
!!
Table 3. Average and range of 95% home range areas for male and female red wolves. A local
convex hull method was used to calculate home range from GPS collar locations. The home
range analysis was generated from 16 red wolves collared in Washington, Tyrell, Dare, Hyde,
and Beaufort Counties, NC from January 2009 to April 2011.
!!!
Home ranges were composed primarily of agricultural fields with 95% home range isopleths on
average containing 55% agricultural fields (Table 4). Summer home ranges were between 0.77
and 3.09 mi2 smaller (t10 = -4.84, P < 0.01) than winter home ranges (Table 5). Average monthly
home range percent composition was different between summer and winter. Red wolves
increased their use of pocosin (t10 = -2.65, P =0.03), wetlands (t10 = -4.29, P < 0.01), and upland
forests (t10 = -4.17, P < 0.01) in late winter and increased use of agricultural fields (t10 = 3.44, P
< 0.01) in summer months (Table 6). Agricultural fields and successional fields account for over
65% of habitat composition regardless of season.
! JV!
Table 4. Average composition of 95% home ranges for 16 red wolves collared in Washington,
Tyrell, Dare, Hyde, and Beaufort Counties, NC from January 2009 to April 2011.
!
!!(((((((((((((
Table 5.!Average monthly 95% home range areas for 16 red wolves collared in Washington,
Tyrell, Dare, Hyde, and Beaufort Counties, NC from January 2009 to April 2011.(
!(((((((((((((((((((((((
! JW!
Table 6Y!Average monthly percent habitat composition calculated using a 95% home range for
16 red wolves collared in Washington, Tyrell, Dare, Hyde, and Beaufort Counties, NC from
January 2009 to April 2011.
Habitat use and selection analyses: We used 29,680 locations of red wolves to construct the 2nd
and 3rd order resource selection functions (RSFs), respectively (Johnson, 1980). Second order
RSF predicted a patchy distribution of red wolves across the 5-county red wolf recovery area
(Figure 4). Third order RSF predicted a relatively equal probability of habitat use by red wolves
across a given home range (Figure 5). RSFs calculated for each individual wolf included the
same variables as the most parsimonious 2nd and 3rd order RSFs. Thus, despite the fact that
collars were deployed on wolves of all ages and both sexes over a range of collar deployment
periods (2009, 2010, and 2011) and deployment lengths (4 to 30 months), no one animal was
considered to bias the RSFs in a unique way different from other animals. The most
parsimonious 2nd order RSF contained: habitat type, distance to roads and water, human density,
an interaction between distance to road and habitat type, and an interaction between human
density and habitat type (Table 7). The AICc weight of the most parsimonious 2nd order RSF was
0.98. The next most parsimonious RSF included an interaction between human density and
distance to road, and had a #AICc of 8 and an AICc weight of 0.02. Agricultural fields were
more likely to be used than all other habitat types. Likelihood of habitat use by red wolves
decreased as human density increased, distance to road increased, and distance to water sources
(e.g. steams and ponds) decreased. As distance to road increased, lowland forest, pocosin, and
wetland habitats were disproportionately less likely to be used by red wolves than other habitat
! JX!
types. As human density increased, upland forests and wetlands were more likely to be used by
red wolves than other habitat types.
The most parsimonious 3rd order RSF contained: 5 habitat types, distance to roads, and distance
to water sources (e.g. steams and ponds) (Table 8). The AICc weight of the most parsimonious
3rd order RSF was 0.75. The next most parsimonious RSF included habitat type, distance to
water, and human density, and had a #AICc of 2 and an AICc weight of 0.25. Again, agricultural
fields were more likely to be used than all other habitat types. Likelihood of habitat use by red
wolves decreased as distance to roads and water increased.
To test the validity of our selected 2nd and 3rd order RSF models, we overlaid 9,893 red wolf
locations withheld from the initial analysis on the resulting probability maps (Figures 4 and 5).
The GPS-collar locations (observed) overlapped areas identified as high probability of red wolf
occurrence (expected) for both 2nd (t1 = 0.79, P > 0.05) and 3rd order (t1 = 1.06, P > 0.05) RSFs.
!!!!!
Figure 4. Relative probability of occurrence of red wolves (Canis rufus) across Washington,
Tyrrell, Dare, Hyde, and Beaufort Counties, North Carolina with respect to 2nd order habitat use,
2009-2011. a) Relative location of packs no longer in existence but identified as habitat with
high relative probability of occurrence of red wolves; b-e) Relative location of packs not
represented in our dataset but in existence at the time of this study.
! QK!
Figure 5. Relative probability of occurrence of red wolves (Canis rufus) across Washington,
Tyrrell, Dare, Hyde, and Beaufort Counties, North Carolina with respect to 3rd order habitat use,
2009-2011.
! QJ!
Table 7. Most parsimonious 2nd order RSF, according to AICc, for habitat use of red wolves in
Washington, Tyrrell, Dare, Hyde, and Beaufort Counties, North Carolina from 2009-2011.
! QQ!
Table 8. Most parsimonious 3rd order RSF, according to AICc, for habitat use of red wolves in
Washington, Tyrrell, Dare, Hyde, and Beaufort Counties, North Carolina from 2009-2011.
Objective 2: !"#$%#&'(&,'(/.10.+.7#07'(*+(2'8()*$+(,#-.&#&(7,#01'/(#0&.7.@#&'8(+2*9(
&,'(@2*@*/'8(,.1,)#;(@2*A'7&(+2*9(=*$%9-.#(&*(&,'(34(56>?56(.0&'2/'7&.*0B(
Quantifying barrier effects using passage rates: Though 3 times as many crossings were
recorded using the 30-minute collar schedule compared to the 5-hour schedule, the overall
permeability for US 64 calculated from both collar schedules was approximately 100% (Table
9). No difference in monthly permeability index was found between the 30-minute and 5-hour
collar scheduled (t10 = 0.045, P = 0.48). No relationship (F=0.021, P=0.89, r2=1.0) was found
between monthly permeability and monthly traffic flow (Figure 6).
Table 9. Permeability index for US 64 between Columbia, NC and the US 64/264 intersection in
Manns Harbor, NC. The permeability index was calculated by dividing the number of highway
crossings by (the number of highway crossings + the number of approaches). Road crossings
were determined using red wolf GPS-collar locations collected between January 2009 and April
2011. A permeability index of 1 represents a highly permeable road while an index of zero
indicates impermeability.
! QT!
Figure 6. Regression analyzing the relationship between monthly permeability index and
average monthly traffic flow rates (vehicles/hour). The monthly permeability index was
calculated using GPS-collar data on the 5-hour schedule from 6 red wolves between March 2009
and May 2010 in Tyrrell and Dare Counties, NC.
Assessing the effect of the highway widening on current red wolf territories: Buffers around
US 64 to a distance of 3281 ft. (1000 m) in 164 ft. (50 m) increments overlaid on a map
displaying current red wolf home ranges showed that 2 red wolf packs would be directly affected
by a highway widening. One pack is located north of the current 2-lane highway in Tyrrell
County and the second is south of the highway in Dare County, the only existing pack in
Alligator River National Wildlife Refuge. The proportion of home range to be affected in Tyrell
County ranges between 0.11% and 6.63% (Table 10) and is located between 820 ft. (250 m) and
3218 ft. (1000 m) from the existing highway just east of Columbia, NC where the previously
widened portion of US 64 narrows to a 2-lane highway (Figure 7). In Dare County, the
proportion of home range that would be affected ranges between 0.01% and 20.31% (Table 11)
and is located south of US 64 between River Rd. and Bear Rd. on Alligator River National
Wildlife Refuge (Figure 8).
Assessing the effect of the highway widening on important red wolf habitat: Important red wolf
habitat is defined as, following the results of the resource selection function analysis in objective
1, agricultural fields, successional fields, and upland forests. Buffers at 164 ft. (50 m) increments
extending out to a distance of 3281 ft. (1 km) from US 64 overlaid on the NC GAP habitat map
revealed that construction to the north of the current US 64 in Tyrrell County would remove
more red wolf habitat than construction to the south. The opposite was found in Dare County,
with more red wolf habitat at risk south of the current US 64 than north (Figures 9 and 10). If
highway construction were to disturb the entire area between the existing US 64 and the 3281 ft.
(1 km) buffer, a total of 0.16 mi2 of red wolf habitat will be removed north of the highway vs.
0.09 mi2 south of the highway in Tyrrell County. For Dare County, 0.04 mi2 would be removed
north of the highway vs. 0.07 mi2 south of the highway if construction were disturb the whole
area between the existing US 64 and the 3281 ft. (1 km) buffer.
! Q6!
Table 10. The proportion of a red wolf pack home range that will be directly affected by
highway construction in Tyrell County between 820 ft. and 3218 ft. from the existing highway.
The red wolf pack is located north of US 64 east of Columbia, NC where the previously widened
portion of US 64 narrows to a 2-lane highway.
Table 11. The proportion of a red wolf pack home range that will be directly affected by
highway construction in Dare County between 264 ft. and 3218 ft. from the existing highway.
The home range is located south of US 64 in Alligator River National Wildlife Refuge between
River Road and Bear Road.
! QU!
Figure 7. Location of the Tyrrell County red wolf pack home range with potential to be directly
affected by highway construction. The home range extends to within 820 ft. from the existing
highway where 4-lanes merge into 2-lanes just east of Columbia, NC. The highway buffer lines
above (black lines) start at 820 ft. from US 64 and end at 3281 ft. in 164 ft. increments. If
highway construction were to disturb the area between 820 ft. to 3281 ft. from the existing
highway, 6.63% of this pack’s home range would be removed. The numbers along US 64
indicate the number of red wolf highway crossings per 0.10 mi. segment captured using GPS-collar
data collected between January 2009 and April 2011.
! Q5!
Figure 8. Location of the Dare County red wolf pack home range with potential to be directly
affected by highway construction. The home range extends to within 264 ft. just south of the
existing highway between River Rd. and Bear Rd. (east of Milltail Rd – not pictured) on
Alligator River National Wildlife Refuge. This is the only red wolf pack on the refuge. The
highway buffer lines above (black lines) start at 264 ft. from US 64 and end at 3281 ft. in 164 ft.
increments. If highway construction were to disturb the area between 264 ft. to 3281 ft. from the
existing highway, 20.31% of this pack’s home range would be removed. The numbers along US
64 indicate the number of red wolf highway crossings per 0.10 mi. segment captured using GPS-collar
data collected between January 2009 and April 2011.
! QV!
Figure 9. The area of important red wolf habitat per 164 ft. buffer for Tyrrell County, NC.
Important red wolf habitat for eastern North Carolina includes agricultural land, upland forests,
and early successional fields.
Figure 10. The area of important red wolf habitat per 164 ft. buffer for Dare County, NC.
Important red wolf habitat for eastern North Carolina includes agricultural land, upland forests,
and early successional fields.
! QW!
Objective 3: C8'0&.+;(/.10.+.7#0&(2'8()*$+(72*//.01(#2'#/(&*(8'&'29.0'(),'2'().$8$.+'(
72*//.01(/&2%7&%2'/(*2(*&,'2(8'/.10(+'#&%2'/(7*%$8(-'(@$#7'8(&*(9.0.9.D'(#8"'2/'(
@2*A'7&('++'7&/(*0()*$"'/B(
(
Determining crossing locations and rates using GPS collars: Six wolves on the 5-hour
schedule (3 M: 3 F) and 8 on the 30-minute schedule (4 M: 4 F), with 3 wolves that started on
the 5-hour and were switched to the 30-minute schedule, displayed crossing activity around US
64 out of a total of 32 collars. Only wolves with home ranges along US 64 crossed the highway.
Wolves on the 5-hour schedule crossed between 2 and 9 times while wolves on the 30-minute
crossed between 2 and 20 times. Five wolves (1 M: 4 F) on the 30-minute schedule, subsampled
every 5-hours for a paired t-test, crossed 53 (30-minute) and 19 times (5-hour), respectively
(P=0.030), showing that the 30-minute schedule captured nearly 3 times the road crossings as
compared to the 5-hour rollover.
An additional 5-hour wolf, (8-year-old female #1880; Figure 11), crossed the highway 266 times.
On reviewing the distribution of her points, which extended about 11 miles along the highway, it
was determined that US 64 bisected the core of her home range (Figure 12). Additionally, her
movements in a narrow band surrounding the road increased the likelihood of “false crossings,”
where the line connecting consecutive points on either side of the highway did not necessarily
represent the true crossing location. For these reasons, and because 1880 represented an unusual
circumstance that heavily skewed the rest of the data, this wolf was considered an outlier and
removed from all further analysis of US 64 GPS-collar data.
Figure 11. Photo of wolf #1880 (adult female) obtained by camera trap along US 64 in Tyrrell
County, North Carolina in July 2009.
! QX!
Figure 12. GPS-collar locations collected between April 2009 and November 2009 of wolf
#1880 along US 64, Tyrrell County, North Carolina within its 95% MCP home range.
Although observed red wolf crossings and randomly generated crossings were both normally
distributed, observed red wolf crossings occurred at a significantly lower frequency (t=1.196,
P=0.03) and were bimodal as compared to the random crossings (Figure 13). Data from both the
5-hour and 30-minute schedules pointed to 2-crossing locations (Figures 14 and 15,
respectively), 1 east of Alligator River in Dare County between miles 8 and 10 and 1 to the west
in Tyrrell County centered on mile 28. However, the 30-minute data were more tightly
concentrated and obvious. The two clusters of crossings identified by the GPS-collar data (Figure
16) coincided with where home ranges approached US 64.
Using the 30-minute collar data, red wolf highway crossing rates did not differ by wolf age
(F=5.14, P = 0.13, n = 13; 3 juveniles, 3 dispersers, 7 adults) or sex (t=0.32, P = 0.76; n = 13; 7
males, 6 females).
Determining crossing locations using Camera Traps: Crossing data were collected at 39
camera stations along US 64 accumulated over 8,154 trap nights. The average and median
number of trap nights per station was 204 and 160, respectively. The number of trap nights per
station ranged from 35 to 617 nights. Four red wolf crossing sites were identified from camera
data, 3 west of Alligator River in Tyrrell County at miles 19, 20.5 and 23 - 24 and 1 east of
Alligator River in Dare County between miles 9 and 10 (Figure 17). The crossing site in Tyrrell
County between miles 23 – 24 and the crossing site in Dare County between miles 9 – 10 were
considered one location each due to proximity and habitat continuity.
The combined GPS and camera crossing data indicated 5 important crossing sections along US
64 between Columbia, NC and the US 64/264 intersection, 4 west of Alligator River in Tyrrell
County and 1 east of Alligator River in Dare County (see Figures 17 - 19). The crossing site in
Dare County identified by the cameras overlaps with the crossing site identified using GPS-collar
locations, however that was not the case in Tyrrell County. The 3 crossing sites identified
in Tyrrell County using cameras are from crossings made by wolf #1880, the wolf excluded from
collar analyses. The crossing site in Tyrrell County identified with the GPS-collar data occurred
in an area where no cameras were placed, within the town limits of Columbia, North Carolina.
! TK!
Figure 13. Observed red wolf crossings (red bars) occurred at a significantly lower frequency
(t=1.196, P=0.03) and were bimodal as compared to the random crossings (black bars).
Observed crossings are based on GPS locations taken at 30- minute intervals from red wolves in
Washington, Tyrrell, Dare, Hyde, and Beaufort Counties, North Carolina collared between
October 2009 and March 2011. An equivalent number of random line segments were drawn
between the GPS locations for each red wolf. To approximate actual red wolf movement, random
segment lengths were constrained to less than or equal to the maximum distance moved by a red
wolf for the 30-minute data sets.
! TJ!
Determining crossing locations using historical road kill data: From May 1988 to February
2009, 58 wolves (31 M, 27 F) died as a result of vehicle collisions in the recovery zone, with an
average of nearly 3 wolves per year. Twelve of these occurred on US 64. While not significant,
the locations of current known road-kills appear to be generally clustered around crossing sites
identified in this analysis, particularly on US 64 (Figures 18 – 19). However many of the historic
road kill events highlight the location of packs no longer present.
!
(
(
Figure 14. The number of red wolf crossings identified by GPS locations per 0.10 mile segments
along US64 between Columbia, NC and the US64/US264 intersection. Crossings are based on
GPS locations taken at 5- hour intervals between January 2009 and March 2011.
!!!!
Figure 15. The number of red wolf crossings identified by GPS locations per 1 mile segments
along US64 between Columbia, NC and the US64/US264 intersection. Crossings are based on
GPS locations taken at 30- minute intervals between October 2009 and March 2011.
!!!!!!!!
! TQ!
!
$Z!
!!
b)!
Figure 16. The two clusters of crossings identified by the GPS-collar data in both a) Dare and b)
Tyrrell Counties, North Carolina coincided with the location where home ranges approached US
64. Crossings are based on GPS locations taken at 30- minute intervals between October 2009
and March 2011.
! ""!
!!
Figure 17. Combined GPS- collar (red bars) and camera (black bars) data, collected from 2009 – 2011, revealed 5 areas along US 64
currently used by red wolves as crossing sites, 4 in Tyrrell County west of Alligator River and 1 in Dare County east of Alligator
River, North Carolina. Camera data are presented as number of captures per 100 trap nights. The red wolf crossing site identified by
camera data in Dare County (miles 9 – 10) overlaps with the crossing identified by GPS-collar data (miles 8 – 10). However, crossing
sites identified by camera data in Tyrrell County (miles 19, 20.5 and 23 – 24) do not overlap with the crossing site identified using
GPS-collar data (mile 28). The crossing site in Dare County between miles 9 – 10 and the crossing site in Tyrrell County between
miles 23 – 24 were considered one location each due to proximity and habitat continuity.
!
! "#!
Legend
Number of Road Crossings per .10 mi segment
Num_Cross
0
1 - 3
4 - 6
7 - 19
!!!!!!!!!!!!!!!!!!!!!!
Figure 18. Map of red wolf highway crossing locations along US 64 in Dare County, NC. Crossing locations were calculated from GPS
collar and remote camera data. The numbers indicate how many crossings per 0.10 mi. segment were recorded. The yellow marker shows
the suggested location for a red wolf crossing structure in Dare County. The blue points indicate the location of a red wolf road mortality
event.
!
! "$!
Legend
Number of Road Crossings per .10 mi segment
Num_Cross
0
1 - 3
4 - 6
7 - 19
Legend
Number of Road Crossings per .10 mi segment
Num_Cross
0
1 - 3
4 - 6
7 - 19
%&'!()!*#!
%&'!()!*#!
+,&-./012!3+!
Legend
Number of Road Crossings per .10 mi segment
Num_Cross
0
1 - 3
4 - 6
7 - 19
Figure 19. Map of red wolf highway crossing locations along US 64 in Tyrrell County, NC. Crossing locations were calculated from GPS collar
and remote camera data. The numbers indicate how many crossings per 0.10 mi. segment were recorded. The yellow markers show the location
of the 3 suggested locations for red wolf crossing structures in Tyrrell County. The blue point indicates the location of a red wolf road mortality
event.
!
! "#!
Evaluating habitat characteristics at crossing sites identified by GPS collar locations:
The most parsimonious habitat model at crossing locations determined using GPS-collar
data included habitat type at distances of 328 ft. and 492 ft. from the crossing site (Table
12). The AICc weight of the most parsimonious habitat model was 0.69. The second most
parsimonious model included habitat type at 656 ft. from the crossing site and had a
!AICc of 3.19 and an AICc weight of 0.07. The habitat types at distances of 328 ft. and
492 ft. from crossing locations correspond to those identified by resource selection
functions for red wolves: agriculture, upland forests, and early successional fields.
Evaluating habitat characteristics at crossing sites identified by camera traps: The most
parsimonious habitat model at crossing locations determined using camera trap data
included width of the road/trail at the camera location (Table 13). The AICc weight of the
most parsimonious habitat model was 0.89. The second most parsimonious model
included road/trail width and habitat type at 164 ft. from the camera site and had a !AICc
of 6 and an AICc weight of 0.03. The trail widths (which ranged from 1.64 ft. to 65.6 ft.)
at camera trap locations with recorded red wolf crossings were 26.24 ft. or wider.
Table 12. Most parsimonious habitat model for red wolf crossing sites in Tyrrell and
Dare Counties, NC identified using GPS-collar data collected between January 2009 and
April 2011.
!!
Table 13. Most parsimonious habitat model for red wolf crossing sites in Tyrrell and
Dare Counties, NC identified using camera trap data collected between March 2009 and
April 2011.
!!!!!
!
! "$!
"#$%&$$#'(!
Objective 1: )*+,&+-.!/',0!1+2#-+-!&$.!+,'(3!-1.!.(-#4.!56!78!%'44#9'4!04':!
;,<:'&-1!=>!-'!-1.!56!78?@78!#(-.4$.%-#'(!
Understanding of basic species survival needs is required before completing any wildlife
management plan. This study used data from 16 wolves from 16 different packs to
estimate home range size and habitat selection of red wolves (Canis rufus) in eastern
North Carolina. The home range sizes we calculated (2.61 mi2 – 38.19 mi2) were smaller
than those reported in 2 earlier studies that followed 3-red wolf packs each (Phillips et al.,
2003: 13.40 mi2 - 78.10 mi2; Chadwick et al., 2010: 31.51 mi2 - 57.72 mi2). However, if
the red wolf pack with the largest home range size in the Phillips et al. (2003) study is
excluded, the home ranges for the remaining 2 packs fall within the range of our findings
(13.40 mi2 - 30.00 mi2). In addition, Phillips et al. (2003) used minimum convex
polygons to determine home range size where as we used "-NNCH, a more conservative
method of home range estimation (Getz et al., 2007), which could account for the
discrepancy in home range sizes between the two studies. Chadwick et al. (2010) tracked
males 2 – 3-years in age, two of which were brothers, and therefore may have been
dispersing individuals. Though not significantly different, our study showed that
dispersing animals tended to have larger home ranges than adults or juveniles, which
could account for the differences in home range size between our study and the one
completed by Chadwick et al. (2010). Small sample size likely accounts for no significant
difference in home range size among age classes. Seven dispersers were eliminated from
this study, 5 due to death and 2 because of inadequate data. Summer (June – September)
home range size (4.84 mi2 – 5.73 mi2) averaged for all 16 packs over 2-years (2009 and
2010) corresponded to summer home ranges reported (1.34 mi2 – 4.72 mi2) for one red
wolf pack monitored during the summer of 2005 (Hinton et al., 2010). We did not look at
the influence of pack size on home range size, as previous research suggests that a
relationship does not exist between pack size and home range size in gray wolves
(Jedrzejewski et al., 2007).
Similar to Phillips et al. (2003) and Chadwick et al. (2010), our study revealed that home
range size varied with season, being smaller during summer months and larger in winter
with monthly home range size peaking in January. Smaller home ranges in summer are
likely due to the presence of pups (Phillips et al. 2003; Chadwick et al., 2010). Mating,
den preparation and whelping for red wolves typically occurs between February and
April (C. Lucash, per. comm.), which coincides with the reduction of monthly home
range sizes. This study found that monthly home range continually reduced in size
starting in February and continued until reaching the smallest size in April. Monthly
home ranges remained small until September when they started a steady increase that
peaked in January. Jedrzejewski et al. (2001) showed home range size and movement
patterns of gray wolf (Canis lupus) packs were also influenced by reproductive cycles.
Habitat and prey availability also may influence seasonal fluctuations in home range size.
We found that habitat in the home ranges was primarily composed of agricultural fields
! "%!
year round. However, the percentage of agriculture within home ranges was highest in
summer and lowest in winter. Increased use of agricultural fields in summer could be due
to increased food resources available to prey species of red wolves such as white-tailed
deer. A recent study found that red wolves readily prey on adult white-tailed deer
(Odocoileus virginianus) and fawns during summer months (Dellinger et al., In Press).
Growth of crops in agricultural fields in summer could help concentrate prey
(Vercauteren and Hygnstrom, 1998). Additionally, the birth of fawns in early summer
could provide a source of prey that is easier to catch, thus allowing red wolf packs to
gather adequate food in a smaller area. Variation in home range size due to prey
availability also has been shown in gray wolves (Ballard et al., 1987).
A decrease in percentage of agricultural fields making up home ranges in winter may be
related to the harvesting of crops. Harvesting eliminates food resources available to prey
and eliminates potential cover for red wolves. This study found an increase in non-agricultural
habitats, such as upland forests, pocosins, and wetlands, during the fall and
winter months (Table 6). Chadwick et al. (2010) noted that increased use of non-agricultural
habitats corresponded to the harvesting of row crops between September and
November and with the onset of the hunting season. Although this study showed that red
wolves typically selected against non-agricultural habitats, cover types such as early
successional fields, upland forests, and pocosins could be providing essential cover for
red wolves after crop harvesting. Also, red wolves tend to prefer cover types with denser
ground vegetation for den sites (Phillips et al., 2003), thus leading to a switch in habitat
use during late winter and spring months.
Another important habitat finding is the selection for areas closer to roads. Most roads in
the red wolf recovery zone are unpaved gravel or dirt roads used for agricultural purposes
(C. Lucash pers. comm.). Red wolves likely used the road network as travel corridors,
which could allow for packs to persist in areas where habitats are highly interspersed and
large parcels of quality habitats are few.
Conclusion: White and Ernst (2004), Singleton et al. (2005), and Kindall and van Manen
(2007) all stress the need to identify habitat linkages across barriers to properly place
crossings. Thus, it is imperative not only that wildlife underpasses are constructed in
areas identified as high use for crossings (Scheick and Jones, 1999), but also that
crossings are placed in a manner that connects habitat being selected by the species of
concern. This study suggests that red wolf crossing structures should connect agricultural
landscapes that are interspersed with upland forests, successional fields and pocosins. In
addition, avoiding the aforementioned cover types during construction will minimize
direct impacts to the red wolf population.
Objective 2: )*+,&+-.!-1.!$#3(#0#%+(%.!'0!4.9!/',0!1+2#-+-!%1+(3.$!+(-#%#A+-.9!
04':!-1.!A4'A'$.9!1#31/+<!A4'B.%-!04':!>',&:2#+!=>!-'!-1.!56!78?@78!
#(-.4$.%-#'(C!
!
This was the first study to employ the use of a permeability index to a non-seasonal
migrating species. This provided a challenge in determining what could be considered a
! "&!
road “approach”, as we had to be careful not to include normal movements within a home
range as an approach. The resulting buffer width of 164 ft., which is similar to the buffer
width suggested for gray wolves (Paquet and Callaghan, 1996), illustrates the willingness
of red wolves to establish home ranges in close proximity to the current 2-lane highway.
The resulting permeability indices calculated for the 2-lane section of US 64 using both
5-hour and 30-minute data were 1.0 and 0.99, respectively. This suggests that the current
2-lane highway is not discouraging the red wolf population from attempting to cross US
64. However, it is important to note that only 14 of the 32 collared red wolves crossed a
highway within the 5-county recovery zone, and 8 of those only crossed either once or
during dispersal. Just 6 wolves from 3 packs crossed a highway regularly, and all 3 packs
had home ranges that were adjacent to or straddled US 64.
The original goal was to compare permeability indices between the previously widened
4-lane section of US 64 in Washington County to the permeability index for the 2–lane
section. However, a decrease in and near disappearance of the red wolf population to the
west of Columbia, NC, prevented the collaring of red wolves where the widening to a 4-
lane highway already was completed.
Although the current 2-lane highway is not discouraging wolves from attempting to
cross, it is important to note that most roads exhibit a distinct trade-off between
permeability and road kill (Forman and Alexander, 1998). A highly permeable road
might result in a high level of wildlife/vehicular collisions, whereas an impermeable road
might have few road kill events. Yet this decrease in road kill comes at the expense of
habitat connectivity. This trade-off indeed holds true for US 64. Though the 2-lane
portion of US 64 may not be hindering attempts to cross, road mortality is the second
leading cause of death among red wolves accounting for 14% of mortalities (USFWS,
2007). !
Permeability was expected to behave inversely to traffic flow, decreasing during the busy
summer months and increasing during the winter. However, due to the high permeability
of the highway, no such relationship existed. In addition, time of day may play an
instrumental role in the event that crossing times (typically at night) do not coincide with
peak traffic hours (midday). Such a pattern could be determined by separating traffic flow
and permeability data by time. It should also be noted that while traffic fluctuates heavily
on US 64 between summer and winter, the highway experiences relatively low traffic
volume (maximum 250 vehicles per hour during the peak season) in comparison to other
highways in the vicinity of the Outer Banks outside of the study area (~791 vehicles per
hour; Currituck Development Group, 2011). The Federal Highway Administration
reports that relatively few animals avoid crossing the road at traffic volumes below 2,500
cars per day and, that while road avoidance increases at moderate volumes (2,500 –
10,000 cars per day), it is not until traffic volume surpasses 10,000 cars per day that a
large portion of animals will avoid highway crossing attempts (Clevenger and Huijer et
al., 2011). The average daily traffic volume for the study site is 1,995 cars per day with a
peak of 6,500 cars per day in July, placing the focal section of US 64 in the low to
moderate traffic flow category as defined by the Federal Highway Administration.
! '(!
Construction north of the current US 64 in Tyrrell County has the potential to remove a
maximum of 0.16 mi2 of red wolf habitat and 6% of the home range area used by a
current red wolf pack while construction to the south will directly impact only 0.09 mi2 of
red wolf habitat and will not displace any current red wolf packs. East of Alligator River
in Dare County, a widening of the current highway to the south has the potential to lead
to a loss of 0.07 mi2 of red wolf habitat and 20% the home range used by the only
existing red wolf pack in Alligator River National Wildlife Refuge. Construction to the
north of US 64 in Dare County will only remove up to 0.04 mi2 of red wolf habitat and
will not overlap with any current packs. Therefore, limiting construction to the south of
the existing US 64 in Tyrrell County and north of the highway in Dare County will avoid
direct effects to the current red wolf population.
We highlight that these results quantify only direct effects on current wolf home ranges.
Road construction can have many indirect effects through changing hydrology; air, water,
noise, and light pollution levels; wind flow; humidity; temperature; vulnerability to
invasive species; and habitat continuity (Forman et al., 2003; Coffin, 2007). These
indirect effects of the construction can disrupt red wolves living adjacent to the existing
highway causing a shift in current home ranges. Any shift in home ranges has the
potential to affect social order, mating, and ability to locate prey. At this time we are not
able to quantify these effects, but these potential indirect effects may be measured in the
“during-“ and “post-” construction phases of the project.
Conclusion: Road permeability, calculated using GPS-collar data, was 100%, thus the
current 2-lane highway does not discourage the red wolf population from attempting to
cross US 64. This does, however, increase the risk of road mortality events. A decrease in
the red wolf population to the west of Columbia, NC, prevented collaring of red wolves
where widening to a 4-lane highway was completed. Therefore, we were not able to
compare highway permeability between 2- and 4-lane highways. To avoid any direct
effects to the current red wolf population, highway construction should be limited to the
south of the existing US 64 in Tyrrell County and north of the highway in Dare County.
Potential indirect effects of highway widening activities were not quantified, as they
could not be quantified using GPS or camera data. It is important to note that indirect
effects can negatively effect the red wolf population.
Objective 3: D9.(-#0<!$#3(#0#%+(-!4.9!/',0!%4'$$#(3!+4.+$!-'!9.-.4:#(.!/1.4.!
/#,9,#0.!%4'$$#(3!$-4&%-&4.$!'4!'-1.4!9.$#3(!0.+-&4.$!%'&,9!2.!A,+%.9!-'!
:#(#:#E.!+9*.4$.!A4'B.%-!.00.%-$!'(!/',*.$C!
Movement patterns obtained through tracking of GPS locations and remote camera traps
clearly demonstrate that red wolves crossed US 64 and two other highways in the red
wolf recovery zone, with some frequency. This implies there is potential for one of two
outcomes of widening the road from 2 to 4 lanes: (1) either increased traffic or the
increased width itself may increase road mortality, or (2) these factors may decrease road
permeability. The degree to which these threats are deemed relevant and serious will have
a significant bearing on NC DOT’s planning and execution of the construction project.
! ')!
The distribution of GPS-collar locations and camera trap photos along US 64 revealed 5
distinct red wolf crossing sites, 4 west of Alligator River in Tyrrell County and 1 east of
the river in Dare County. Though they do not overlap completely, GPS-collar data and
camera trap data are in accordance for the location of the crossing site in Dare County.
However, that is not the case in Tyrrell County. GPS-collar data revealed 1 crossing site
in Tyrrell County, while camera trap data identified the 3 additional crossing sites. Two
factors contributed to the 2-methods not overlapping in Tyrrell County. First, due to the
number of false crossings, GPS-collar data from wolf #1880 was eliminated from our
analysis, thus eliminating GPS-collar data along the section of US 64 coinciding with the
cameras. False crossings were obtained from this wolf because US 64 bisected her home
range (Figure 12). Secondly, cameras were not set up where GPS-collar data identified a
crossing location in Tyrrell County due to increased risk of camera theft within the town
limits of Columbia, NC.
The potential for “false crossings,” which may suggest that a crossing took place in a
different location from where it actually occurred, exists in the remainder of the GPS-collar
data as well. However, the presence of such distinct activity clusters, particularly
under the 30-minute schedule, suggests that our results captured real movement trends.
Although we captured nearly 3 times as many crossing events and the data displayed
tighter clustering with better defined locations using the 30-minute collar schedules, the
location of crossing sites identified using GPS-collar data was generally consistent
between the 2 schedules (5-hour and 30-minute).
The red wolf crossing site identified in Dare County is within the Alligator River Wildlife
National Wildlife Refuge and is centered on Hickory Road. This matches the location of
an important black bear crossing site (Vaughan et al., 2011), and therefore is a candidate
site for the placement of a multi-species crossing structure for large wildlife. Likewise,
the 3 red wolf crossing sites in Tyrrell County located via camera trap data overlap with
candidate areas for large wildlife crossing structures identified in an earlier study by
University of Central Florida (UCF) (Smith, 2011). The red wolf crossing site between
miles 23 and 24 (cameras W12 and W14) overlaps with “Area 1” of the UCF study
(Smith, 2011), which is centered on the western intersection of Old US 64 and US 64.
The red wolf crossing sites at mile 20.5 (camera W24) and mile 19 (camera W30S)
overlap with “Area 3”and “Area 5” of the UCF study, respectively (Smith, 2011) and are
located near the eastern intersection of Old US 64 and US 64. Using the eastern
intersection of Old US 64 and US 64 as a reference point, “Area 3” is 0.31 miles west of
the intersection and “Area 5” is 1.16 miles east of the intersection.
The red wolf crossing site in Tyrrell County identified using GPS-collar locations is
located within the town limits of Columbia, NC where US 64 narrows from 4 to 2 lanes.
Placement of a crossing structure here may not be practical because of proximity to
residential areas.
Crossing rates suggest that there is no difference in highway crossing behavior between
sexes or among ages. As with the home range analysis, the lack of any significant
difference in either sex or age class might simply be a function of low sample size, and it
! '*!
is possible that a real relationship could be hidden by an interaction between the two
variables, which we were unable to test for the same reason. This possibility is supported
by data on road mortality, which has impacted dispersers hardest among the age classes.
Although road mortality accounts for 14% of deaths for the red wolf population over all,
when broken down by age class, road mortality accounts for 19% of dispersers but only
6% of breeding adults (USFWS, 2007). Low sample size and high expense per individual
is a common hurdle in research involving large carnivores, as was true for this study. In
addition, the status of the red wolf as critically endangered puts a major constraint on
population size from which to draw a sample.
In addition to identifying the location of important red wolf crossing sites, we also
investigated which habitat variables were correlated with those locations. The presence of
agricultural fields, successional fields, and/or upland forests 328 ft. to 492 ft. (100 to 150
m) from the road best predicted where a red wolf chose to cross when using GPS-collar
data while trail/road greater than 26.24 ft. (8 m) provided the best explanation for the
location of crossing sites identified by remote camera traps. The presence of agricultural
fields, successional fields, and upland forests as well as proximity to maintained
agricultural/forest roads at crossing sites corresponds to habitat selection results.
Conclusion: The distribution of GPS-collar locations and camera trap photos along US
64 revealed 5 distinct red wolf crossing sites, 4 west of Alligator River in Tyrrell County
and 1 east of the river in Dare County. Four of the 5 red wolf crossing locations we
identified are suitable for crossing structures. The most western crossing site is located
within the town of Columbia, NC where retro fitting a wildlife underpass may be
impractical. All 4 crossing sites suitable for placement of a crossing structure overlap
with large wildlife crossing locations identified in previous studies. The 1 red wolf
crossing site located in Dare County is centered on Hickory Road and the 3 crossing sites
in Tyrrell County are approximately where US Old 64 intersects with US 64. Although
no significant difference in crossing behavior was found during this study, high road
mortality among dispersers suggests they may cross the highway more frequently than
adults or juveniles. The most parsimonious models looking at the relationship between
habitat variables at 164 ft. increments from US 64 and road/trail widths measured at road
access points where cameras were placed (e.g. dikes, logging roads, public property
access roads) indicates that well maintained trails at least 26.24 ft. (8 m) in width leading
to and from underpasses and connect habitats selected for by red wolves (e.g. agriculture,
successional fields, and upland forests), will optimize efficacy.
The data presented here are reflective of the current population’s behavior. In the event
that wildlife crossing structures are deemed necessary, our results identify locations
where crossing structures would have the greatest effect on the red wolf population.
This project is only one of several examining the use of US 64 by numerous wildlife
species. The results of those studies, in addition to this one, should be taken into account
in determining the need for mitigation, the type of mitigation to use, and the layout that
would be most compatible with all target species. The direct and indirect effects of the
road widening project remain difficult to predict, yet the potential for a negative effect on
the red wolf must be considered. Careful monitoring of the red wolf population
! '"!
throughout and following the construction process will be crucial to ensuring red wolf
survival and will aid management decisions in future road issues.
F#-.4+-&4.!G.*#./!
G.9!H',*.$I!;+$-J!;4.$.(-J!+(9!K&-&4.!
Red wolves (Canis rufus) were originally described in 1851 by Audubon and Bachman
and considered a subspecies of the gray wolf. However, red wolf heritage came under
debate in the mid-1900’s when Goldman suggested that all of the southeastern wolf
subspecies should be combined into the distinct species of Canis rufus, separate from
gray wolves. Many supported this decision until the advent of genetic methodologies in
the 1990’s. Genetic studies in the 1990’s provided support for the hypothesis that red
wolves evolved from a natural hybridization between gray wolves and coyotes (Wayne
and Jenks 1991; Wayne 1992; Roy et al. 1994, 1996; Wayne and Gittleman 1995; Wayne
et al. 1998; Reich et al. 1999). However, Wilson et al. (2000) suggested that red wolves
and Algonquin wolves (Canis lupus lycaon) diverged from gray wolves 1.2 million years
ago and then diverged from coyotes 150,000 to 300,000 years ago. Work by Hendrick et
al. (2000) investigating major histocompatibility complex genetics data indicates that red
wolves are more closely related to coyotes than to gray wolves, adding support to the
claims made by Wilson et al. (2000).
The current stance that the red wolf is a species in its own right, separate from gray
wolves, coyotes, and domestic dogs is based on mtDNA sequencing of 340 base pairs
showing a unique sequence for red wolves (Adams, 2002; Adams et al., 2003). However,
the debate over red wolf taxonomy is far from over. Both Wilson et al. (2000, 2003) and
Kyle et al. (2006, 2007) now suggest that red wolves and Algonquin wolves are
genetically similar enough to be combined into one species, the eastern wolf (Canis
lycaon). In 2007 Murray and Waits, while acknowledging the genetic similarity between
red wolves and Algonquin wolves and the plausibility that they are conspecifics, argue
that combining the two species would hinder red wolf conservation efforts and the ability
to secure conservation funds because red wolf extinction would become an issue of
population extinction rather than species extinction. In 2008, Kyle et al. rebutted the
article by Murray and Waits stating that taxonomy embracing conservation agendas
rather than scientific scrutiny should be avoided. Kyle et al. (2008) go on to say that
while they agree with Murray and Waits (2007) that there are instances in which
genetically unique populations warrant protection, that the genetic uniqueness of the red
wolf population is not supported scientifically. Red wolves and Algonquin wolves are
only separated genetically by one mtDNA haplotype differing by one base pair (Wilson
et al. (2000, 2003). Kyle et al. (2008) suggest that any difference between red wolves and
Algonquin wolves may be an artifact of a low effective population size, a founder effect,
a by-product of artificial selection, and/or because of current management strategies that
remove individuals that are <80% red wolf from the breeding population (potentially
removing important red wolf genes from the population). For now, the taxonomy of red
wolves remains under debate.
! ''!
The historical range of red wolves was originally described as occurring from south
central Texas east to Florida and then north to the Ohio River (Nowak, 1979). The
historical range was then extended north to Pennsylvania in 1995 (Nowak) and then north
again to south central Maine in 2002 (Nowak) in support of the theory that there is one
eastern wolf species. Red wolves declined initially with European colonization (USFWS,
2007). Predator control programs and habitat fragmentation in the 1960’s dramatically
reduced red wolf populations. By the 1970’s, red wolves were reduced to remnant
populations along the Texas and Louisiana coast. In 1973, the red wolf achieved
endangered status with the passing of the Endangered Species Act of 1973. The United
States Fish and Wildlife (USFWS) Service worked to capture the remaining wild red
wolves between 1974 and 1980 to establish a captive breeding population as a last ditch
effort to save the red wolf (USFWS, 2007). The USFWS successfully captured 17
individuals, 14 of which were used as founders for the captive breeding program
(USFWS, 2007). As a result of capturing the remaining wild animals, red wolves were
declared extinct in the wild in 1980.
Through the establishment of a captive red wolf breeding program with the Association
of Zoos and Aquariums (AZA), enough red wolves were bred in captivity to attempt a
reintroduction in 1987. The reintroduction began with the release of 4 breeding pairs on
the Alligator River National Wildlife Refuge. By 1988, the first pups (2 litters) were born
post reintroduction (USFWS, 2007). The USFWS started two additional red wolf
reintroduction programs; in 1991 at the Great Smoky Mountains National Park at the
Tennessee/North Carolina border and in 1993 at the Pocosin Lakes National Wildlife
Refuge in North Carolina just 27 miles west of the original reintroduction site. The
reintroduction in the Great Smoky Mountains did not succeed, but the reintroduced
populations at Alligator River National Wildlife Refuge and the Pocosin Lakes National
Wildlife Refuge continued to expand and merged to form the current, and only, red wolf
population in the wild. The current red wolf recovery zone has expanded to 5 counties in
North Carolina’s Albemarle Peninsula (Dare, Tyrrell, Washington, Beaufort, and Hyde
Counties – see current range in Figure 1) and contains between 100 and 130 red wolves
forming 20 packs (USFWS, 2007). Red wolves remain listed under the Endangered
Species Act of 1973 (USFWS, 2007) and are recognized by IUCN as one of the most
endangered canid species in the world (IUCN, 2006). The re-introduced population +,!
-.,+/012.-!1,!0304.,,.02+15!.67.8+9.0215:
The USFWS has a population goal of 220 individuals, yet the population has fluctuated
between 100 and 130 individuals over the past 12 years (USFWS, 2007). USFWS
biologists with the Red Wolf Recovery team feel that the population can still expand
further west allowing population growth to continue. However, non-USFWS researchers
on the Red Wolf Implementation Team believe that the red wolf population may have
reached carrying capacity within the recovery zone (USFWS, 2007). Models suggest that
carrying capacity for red wolves within the current 5 county recovery zone is
approximately 138 individuals (Murray, unpublished data). Starting in 2002, to help
bolster the wild population, captive-born pups have been fostered to wild parents with
similarly aged pups (USFWS). However, a better understanding of habitat requirements
is needed to determine the ability of the peninsula to hold more animals.
! ';!
Management of the red wolf gene pool and genetic fitness are the primary focus of the
red wolf recovery and species survival plan due to a low effective population and
potential founder effects. Genetic drift and inbreeding depression are of concern with
small populations (Caughley, 1994). A study by Kalinowski et al. (1999) reported to find
no evidence of inbreeding depression within the captive red wolf population. Long and
Waddell (2006) reported that the captive population retained 89.65% of the genetic
diversity of the founding captive population. Despite these results, there have been
reports of physical anomalies in the captive red wolf population such as progressive
retinal atrophy, malocclusion and undescended testicles (USFWS, 2007). Although a
study by Miller et al. (2003) showed that only a few individuals per generation were
needed to maintain sufficient genetic diversity in a grizzly bear population, further
studies are needed to determine if genetic drift and inbreeding depression are impacting
the wild red wolf populations.
For now, management of the reintroduced red wolf population focuses on a different
genetic problem, the introgression of coyote genetics. Kelly et al. (1999) reported
interbreeding between coyotes and red wolves resulting in coyote gene introgression into
the wild red wolf population. As a result, an adaptive management plan was developed
(Fazio et al., 2005). The plan calls for either the complete removal of coyotes and hybrids
or the sterilization of hormonally intact coyotes and hybrids via vasectomy and tubal
ligation, depending on the location within the recovery zone. In Zone 1 of the plan, all
coyotes and hybrids are removed. In zones 2 and 3, coyotes and hybrids are sterilized and
then used as territorial “place-holders” until replaced by wild red wolves. The sterilized
coyotes and hybrids cannot interbreed with wild red wolves and they exclude intact
coyotes or hybrids from the territory they hold. The idea is that these sterilized animals
act as “place-holders” until red wolves replace them either naturally via displacement or
through management actions to make room for translocation of a red wolf pair. The
effectiveness of the management plan is evaluated via non-invasive genetic monitoring of
canid scats (Waits 2004; Waits and Paetkau, 2005; Adams, 2006; Adams and Waits
2007). Through continued genetic monitoring, Adams noted strong evidence that a single
hybridization event in 1993 resulted in most introgression of coyote genes into the red
wolf population observed to date. From this evidence, Adams (2006) infers that
hybridization with coyotes has had less genetic impact on the restored red wolf
population than originally thought by Kelly et al. (1999), largely because backcrossing
has been rare in the population.
Due to the immediate attention required to address the hybridization of red wolves and
coyotes, less is known about red wolf home range, habitat, and diet requirements. Two
recent studies examined red wolf home range and habitat use. The first study (Hinton and
Chamberlain, 2010) used VHF collars to follow two red wolf packs during summer 2005
(July to September), one with pups and one without pups. This study found that the pack
with pups had a smaller average home range size than the pack without pups, 5.74 km2
vs. 9.55 km2 for diurnal home range and 8.24 km2 vs. 9.40 km2 for nocturnal home
ranges, respectively (Hinton and Chamberlain, 2010). Although it is important to note
that the larger averaged home range calculated for the non-breeding pack is likely driven
! '#!
by one male whose home range was 2-3 times larger than any other wolf in the study.
Adults in both packs increased home range size nocturnally (1800-0559 hours) and both
packs spent approximately 98% of their time in agricultural fields, defined as corn,
soybean, and cotton (Hinton and Chamberlain, 2010).
The second study investigating red wolf home range and habitat use employed GPS
collars to monitor 4 male wolves from 3 packs over a period ranging from 11 to 18
months (Chadwick et al., 2010). Chadwick et al. (2010) corroborated the finding that red
wolf packs primarily utilize agricultural fields during summer and early fall months, with
highest use of agricultural fields occurring July through September. However, they noted
a seasonal switch to grass/brush and forested habitats during winter and early spring
months, November to May (Chadwick et al., 2010). Though results of both studies
showed similar summer habitat preferences, the home range size estimates calculated by
Chadwick et al. (2010) were several magnitudes larger than those calculated by Hinton
and Chamberlain (2010). Home ranges reported in Hinton et al. (2010) ranged from 3.48
km2 to 12.24 km2 while those calculated by Chadwick et al. (2010) ranged from 81.6 km2
to 148.1 km2. Both studies employed kernel density estimators to estimate home range
size. Chadwick et al. (2010) did mention that they found a 40 to 63% reduction in home
range size during summer months, but that places their summer home range estimates
between 51.4 km2 and 59.24 km2, still considerably larger than those estimated by Hinton
and Chamberlain (2010).
Though Hinton and Chamberlain (2010) did calculate home range size for both sexes and
all age classes, they only collected point locations for a period of 3 months and the
number of daily locations varied. Chadwick et al. (2010), while focusing only on
nocturnal movements of males, collected point locations over a period of 11 to 18 months
and were able to consistently collect 4 locations per day with the use of GPS collars.
This suggests that the discrepancy in home range estimates between the two publications
may be the result of Hinton and Chamberlain (2010) not collecting enough locations to
accurately capture the entire home range size. Until data on all sexes and age classes
collected covering the entire 24-hour period and across all seasons is made available,
conclusions concerning red wolf home range and habitat requirements cannot be made.
Phillips et al. (2003) reports that the primary prey species of red wolves include: white-tailed
deer, raccoon, rabbits, nutria, and other small rodents. A more recent diet
assessment via scat analysis lists white-tailed deer as the primary prey item of red wolves
(Dellinger et al. in review). However, packs will increase the amount of small rodents
and human-sourced foods (e.g. hog pits) in their diet during periods of increased energy
demands such as pup rearing (Dellinger et al., in review).
For red wolf management to move forward, the current gaps in knowledge of red wolf
natural history need to be filled. Furthermore, before model building to predict the effect
of a highway widening through the red wolf recovery zone starts, base knowledge of
home range and habitat selection is required.
! !
! '$!
G'+9!)%','3<!!
Accompanying the rapid expansion of our transportation network was a growing concern
over the environmental effects of roadways. The emergence of road ecology, coined by
Richard T.T. Forman (1998), as a distinct discipline has brought together scientists from
many disciplines (e.g. landscape ecology, wildlife biology, toxicology, hydrology,
limnology, etc) and engineers to tackle the ecological challenges posed by transportation
systems. For several decades now, researchers have studied the effects of roads on both
the abiotic and biotic components of ecosystems. As a result, we now know that roads
affect hydrology, air, water, noise, light pollution levels, wind flow, humidity,
temperature, vulnerability to invasive species, and habitat continuity (Forman et al.,
2003; Coffin, 2007). Such large-scale and multifaceted changes to ecosystems have many
detrimental effects on wildlife (Jackson, 1999), including direct mortality (Lalo, 1987;
Harris and Scheck, 1991; Schwabe and Schuhmann, 2002), habitat destruction (Theobald
et al., 1997; Angelsen and Kaimowitz, 1999), barrier effects (Forman et al, 2003), and
increased human land use activities (Bjurlin and Cypher, 2003; Coffin, 2007). !
Before road mortality can be effectively mitigated, it is important to understand the
factors that influence wildlife-vehicle collisions to occur in the first place. Jaarsma et al.
(2006) modeled several road, traffic, vehicle, and species characteristics to find which
had the greatest influence on the occurrence of a wildlife vehicle collision event. They
found that traffic volume and the animal’s traversing speed were the greatest predictors in
determining a road mortality event, with higher traffic volumes and slower crossing
speeds more likely to lead to a collision (Jaarsma et al., 2006). Two separate studies
investigating the relationship among road kill events, body size, and diet found that
carnivores were less likely to be hit along a road as compared to herbivores and
omnivores (Ford and Fahrig, 2007, Barthelmess and Brooks, 2010). Those same two
studies found a peaked relationship between road mortality and body size, with small (<1
kg) and large (>10 kg) body animals less like to be killed by vehicles as compared to
medium (1 – 10 kg) sized animals (Ford and Fahrig, 2007, Barthelmess and Brooks,
2010). All three of the above cited articles suggest that direct mortality resulting from
roads may not have a significant negative impact on carnivore populations as many
carnivores are faster moving and larger bodied.
However, a study in southern Texas that looked at the influence of habitat variables on
the location of bobcat road mortality events found that suitable habitat adjacent to the
highway best explained the location of mortality events (Cain et al., 2003). These results
were corroborated by another bobcat study in southern Illinois (Kolowski and Nielsen,
2008). Likewise, red wolves in northeastern North Carolina cross highways at locations
adjacent to preferred habitat and established home ranges (Proctor, unpublished data).
These results are similar to studies evaluating the use and success of highway crossing
structures. The most successful wildlife crossing structures are the ones that connect
preferred habitats of the targeted species (Ng et al. 2003, White and Ernst 2004,
Singleton et al., 2005, Kindall and van Manen, 2007).
When vehicle strikes do occur, they account for a low percentage of mortality in
! '%!
carnivores and do not translate into population level effects. Even for endangered San
Joaquin kit foxes, road mortality rarely accounted for over 10% of mortality, with
predators accounting for most mortality events (Bjurlin and Cypher, 2003). In a 3-year
study that followed 60 radio collared kit foxes that lived in close proximity to a 2-lane
paved highway, only one was lost to a vehicle strike (Cypher et al. 2009). However, prior
to mitigation efforts, road mortalities did account for 49% of mortality in Florida panthers
(Maehr et al., 1991). The carnivore populations with the highest reports of road kill
events in the United States are black bears. In Virginia, black bears and white-tailed deer
account for the most frequently recorded road kill events (Donaldson, 2007). Two studies
in Florida found increased road mortality of black bears in areas of higher road density
(Hostetler et al., 2009, McCown et al., 2009). These results differ from studies focusing
on other carnivore populations where an increase in road density lead to increased road
avoidance rather than increased mortality events (Dickson et al. 2005, Chetkiewicz and
Boyce 2009). However, the studies by Ford and Fahrig (2007) and Barthelmess and
Brooks (2010) did find that omnivores are more likely to be stuck by vehicles as
compared to carnivores. Although black bears are classified as carnivores, their diet is
omnivorous.
A barrier effect blocking access to resources, dispersal, and gene flow is the greatest
impact of highways and roads on carnivore population in the United State. A study in
southern California found that while cougars often made use of dirt roads, they actively
avoided paved roads (Dickson et al., 2005). Similar results were found in another study
with cougars negatively associated with roads, particularly during winter months
(Chetkiewicz and Boyce, 2009). Riley et al. (2006) found that coyote and bobcat
populations in southern California separated by a major freeway exhibited genetic
differentiation, suggesting that the freeway is a barrier to dispersal. For those that do
cross, heightened territorial behavior along roadways can discourage reproductive
success, again limiting gene flow (Riley et al., 2006). Likewise, a study found that a
highway in southern Canada is acting as a dispersal barrier for grizzly bears at the US-Canada
border, as evidenced through genetic differentiation between the two populations
(Proctor et al., 2005). The result is the creation of two vulnerably small populations
(Proctor et al., 2005). A highway was found to restrict gene flow in a Cleveland, Ohio
coyote population and direct the movements of migrants towards urbanizing centers
(Rashleigh et al., 2008). Even when they do not constitute an absolute physical barrier,
high-use roads can lead to avoidance behavior in canids affecting their ability to move
across a landscape (Kaartinen et al., 2005, Whittington et al., 2004). The degree to which
a road impacts canid survival is dependent on the specific situation, and sometimes no
detrimental effects are observed, as in the case with San Joaquin kit foxes (Cypher et al.,
2009).
The amount to which a road constitutes a movement barrier for black bears is dependent
of the level of traffic volume (McCown et al., 2009). A study documenting the
movements of two black bear populations along the same highway in Florida found that
the population living in the area with lower traffic volume crossed the highway more
often (McCown et al., 2009). In Maryland, black bears avoided the larger primary
highways, but readily crossed all other road classes (Fecske et al., 2002). In the northern
! '&!
Rockies, just under 50% of collared black bears were willing to cross a highway at least
once were (Lewis et al. 2011). A study in North Carolina found that site occupancy of
black bears deceased from 0.81 to 0.35 a highway in the study area was widened from 2-
lanes to 4-lanes (Nicholson and van Manen 2009). For black bears, roads appear to
exhibit the distinct trade-off between permeability and road kill discussed by Forman and
Alexander (1998).
Though the last 10 years has documented many adverse effects of highways and roads on
carnivore populations, there have been positive developments as well. Highway crossing
structures have been successful at mitigating some negative impacts of highways on
carnivore populations. In Texas, bobcats did make use of culverts to cross a highway
when the culverts were placed adjacent to suitable habitat (Cain et al., 2003). A study in
California found that a large variety of species, including reptiles, small mammals,
carnivores, and mule deer use highway underpasses, even underpasses not designed
specifically for wildlife (Ng et al., 2004). A study investigating wide variety of structures,
including culverts, modified box-culverts, underpasses, and overpasses, found that
culverts were the least used preferences between underpasses and overpasses varied with
species (Mata et al., 2008). A study in Portugal found that red foxes, badgers, genet, and
Egyptian mongooses used underpasses and culverts without preference (Grilo et al.,
2008). However, a study of wildlife underpasses in Virginia revealed that while they
were effective for foxes and coyotes, they did not find evidence of black bears utilizing
highway underpasses (Donaldson, 2007). Likewise, a study monitoring the success of
multi-species highway underpasses following a highway-widening project found that
bobcats, black bears, and foxes utilized underpasses, but not coyotes or red wolves
present in the area (McCollister and van Manen, 2010).
In all documented success of highway crossing structures, the authors noted that the
successful structures connected areas of suitable habitat for the target species. The non-detection
of all area carnivore species in the multi-species crossing structures may be the
result of not being located in an area that contains suitable habitat for all species. Though
multi-species structures may be may appear to be more cost effective initially, a lower
success rate will decrease the cost effectiveness in the long run.
While the subjects above have gotten considerable coverage in the peer-reviewed
literature, relatively little research has been directed at determining the placement of
highway underpasses. In may be beneficial to focus future research efforts on
determining the effective placement of highway crossing
Of the studies that focused on placing mitigating structures, methodologies have varied
widely and range from non-invasive to the capture and handling of target species. Non-invasive
techniques include the use of track/trail counts (Van Dyke et al., 1986;
Rodriguez et al., 1996; Alexander and Waters, 1999; Scheick and Jones, 1999, 2000;
Barnum, 2001, 2003), remote cameras (Scheick and Jones, 1999, 2000), barbed wire hair
traps (Wills and Vaughan, 2005), road kill surveys (Clevenger et al., 2003b; Mazerolle,
2004; Smith et al., 2009), and GIS based modeling (Smith et al. 1998; Klein, 1999;
Kobler and Adamic, 1999; Sheick and Jones, 1999, 2000; Clevenger et al., 2003a; Lloyd
! ;(!
et al., 2005). Non-invasive techniques are relatively inexpensive and can be effective,
however all are time intensive. A constraint of track counts is the requirement of an
appropriate substrate, the use of sand, or, in some areas, the presence of fresh snow.
Remote camera traps provide crossing location, date, time, and work for a wide variety of
species. Yet, cameras cannot cover the entire length of the highway simultaneously.
Running barbed wire the length of the study area provides crossing location and, with the
addition of genetics, crossing frequency on the level of the individual. Drawbacks of
using barbed wire include the added cost of genetics and the limited number of mammals
this technique is appropriate for. Road-kill surveys to detect crossing hotspots are an
excellent method for collecting data on a wide range of species simultaneously. However
rate of decay, scavenger activity, and method of survey (driving vs. walking) can affect
results and must be considered in planning survey interval times. Road kill surveys may
also miss animals that wander away from the collision site before dying. It has also been
suggested that while road kill events do represent failed attempts to cross, they do not
necessarily indicate important linkage areas. The primary weaknesses of the GIS-based
techniques are data availability and data quality. GIS models are most effective when
data on habitat use patterns of the subject species are well known and where the habitat is
diverse and heterogeneous. A limitation of non-invasive techniques as a whole, with the
exception of GIS based models, is inherent bias unless the entire length of the proposed
highway construction project is