Background information and definitions
Schemes designed to reduce collisions between vehicles and wild mammals may use multiple interventions. Two of the most common ones, installing barrier fencing and providing routes for mammals to travel underneath roads, are often employed within the same scheme. This may entail regular roadside fencing with entrances to underpasses set further back away from the road or fencing may be designed to adjoin the sides of underpass entrances. Sometimes, fencing may be installed to form a funnel leading towards underpass entrances.
This intervention includes studies where these two actions are in place at the same site. In most studies, all underpasses (where there are multiple crossings) are beneath stretches or roads that have barrier fencing. In a minority, just some of the underpasses monitored are along stretches with barrier fencing. Studies included use of either conventional fencing, electric fences or other barriers, such as walls. Most studies report solely on the use of crossings or trends in numbers of mammals killed on roads. There is an absence of studies reporting on wider population-level effects of the presence of these structures.
See Install tunnels/culverts/underpass under roads for studies where underpasses are either installed without use of barrier fencing or where it is not clear from the study that barrier fencing was installed. See also Install barrier fencing along roads for studies which, in some cases, included underpasses but where the specific effect of fencing was evaluated.
Supporting evidence from individual studies
A before-and-after study in 1970–1973 along a highway in Colorado, USA (Reed et al. 1975; same experimental set-up as Reed 1981) found that an underpass, in areas with roadside fencing and one-way gates, reduced road mortalities and allowed most local mule deer Odocoileus hemionus to migrate safely under a highway. There were 14 deer-vehicle accidents/year within the fenced section compared to 36/year before installation of the underpass and fencing. On average, 345 mule deer (61% of the local population) used the culvert each season, with up to 17 crossings/day. Underpass use was not affected by artificial lighting. On average, 17% of deer used one-way gates to escape the highway and 17% went round the ends of fences or did not cross. In 1970, a concrete box underpass (3 × 3 × 30 m, with two skylights) was installed under a 3.2-km section of highway. The 2.4-m-high barrier fencing either side had eight one-way gates. Underpass-use was monitored by track counts and mechanical counters daily and a video camera at night during spring–summer and autumn migrations in 1970–1973. Artificial lighting was alternately turned on and off over 28 nights, in June and October 1973. Tracks at gates and deer movements along the fence were monitored each morning.Study and other actions tested
A study in 1974–1979 along a highway in Colorado, USA (Reed 1981; same experimental set-up as Reed et al. 1975) found that an underpass, in an area with roadside fencing, continued to be used by mule deer Odocoileus hemionus 4–9 years after installation Between 1.3 and 5.8 deer/morning (average 2.3) were observed exiting the underpass each year (total 298 deer). Deer behavior suggested that 75% of animals exiting the underpass were reluctant, wary, or frightened. Eleven hesitated just inside the exit and 23 showed wariness or excitability after exiting the underpass. Behavioural responses of deer to the underpass were reported not to have changed substantially over 10 years (1970–1979) of spring-summer use. In 1970, a concrete box underpass (3 m high, 3 m wide, 30 m long) was installed under a 3.2-km section of highway. Entrances were separated from the road by 2.4-m-high barrier fencing. Deer were observed from 130 m away, at 05:00–07:00 h, on 9–30 days (average 16), during each spring/summer migration in 1974–1979. Behavioral responses were likened (but not compared numerically) with those from earlier monitoring that commenced in 1970.Study and other actions tested
A study in 1977–1979 along a highway through shrubland in Wyoming, USA (Ward 1982) found that underpasses, in areas with roadside fencing, were used by mule deer Odocoileus hemionus to cross under the road. During four migration periods (two spring, two autumn–winter) immediately after underpasses were connected to a fence, >4,000 crossings through underpasses were made by deer (precise figure not stated). The study was conducted along a 7.8-mile stretch of highway constructed in late 1970. The highway was located on a migration route of 1,600–2,000 mule deer. Over four migratory periods, seven underpasses (length: 110–393 feet; width: 10–50 feet; height: 10–17 feet) were monitored for deer use. Underpasses were connected to 8-foot-high roadside fencing that guided animals towards entrances. From 1978, an attempt was made to attract deer to six of the seven underpasses by baiting with alfalfa hay, supplemented with apple pulp or by vegetable trimmings. Deer movements were monitored by track counts and surveillance cameras.Study and other actions tested
A before-and-after study in 1975–1981 in Montana, USA (Singer & Doherty 1985) found that two underpasses and roadside fencing increased highway crossing success by mountain goats Oreamnos americanus. After construction, 90% of highway crossing attempts were successful compared to 86% during and 74% before construction (unsuccessful attempts were when the crossing was temporarily thwarted). Crossing hesitations and run-backs decreased by 80% after underpass construction, delay time before crossing declined by about 30% and signs of fear (measured by an index) decreased. All crossings were successful when there was no disturbance, but success decreased to 85% when humans or traffic were present. A large underpass (3–8 m high, 23 m wide, 11 m long) was constructed where goats were observed crossing. In addition, a new road bridge included a ledge underneath for goats to cross (3 m high, 3 m wide, 11 m long). A sheer wall downhill and barrier fencing prevented goats crossing between underpasses. Old goat trails were removed and new trails to underpasses dug. Goat crossings were monitored before (1975), during (May–October 1980) and after underpass construction (October 1980–September 1981).Study and other actions tested
A before-and-after study in 1977–1990 along a highway in Alaska, USA (McDonald 1991) found that barrier fencing with one-way gates, along with an underpass and road lighting, reduced vehicle collisions with moose Alces alces. Effects of fencing and the underpass could not be separated from those of gates and lighting. There were fewer moose-vehicle collisions after installation of fencing with one-way gates, an underpass and lighting (0.7/year) than before (17/year). There was no significant difference in the distribution of moose in relation to the highway after and before fence installation. A total of 17 moose were observed using one-way gates and tracks suggested gates were used frequently. However, this meant that moose were regularly getting onto the highway. The first gates installed stayed open if swung all the way open and gates got stuck open below 0°C, because of the lubricant used. In October 1987, road lighting was installed along 11.5 km of the highway. Fencing and 30 one-way gates were installed along 5.5 km of this section and an underpass was created. Moose-vehicle collisions were monitored before (1977–1987) and after (1987–1990) installation. One-way gates were monitored using track counts in snow.Study and other actions tested
A study in 1994–1995 in Florida, USA (Foster & Humphrey 1995) found that four underpasses beneath a highway, in areas with roadside fencing, were used by Florida panthers Felis concolor coryi and a range of other mammal species. Ten crossings were recorded through underpasses by panthers, as were 361 by white-tailed deer Odocoileus virginianus, 133 by bobcats Lynx rufus, 167 by raccoons Procyon lotor and two by black bears Ursus americanus. Panther records were thought to relate to two individuals. Four concrete bridge underpasses (21–26 m wide, 49 m long) were monitored along a 64-km stretch of a four-lane, divided highway. Barrier fencing, 3 m high, ran along the highway. Infrared game counters and cameras were used to monitor underpasses for 2, 10, 14 and 16 months in 1994–1995.Study and other actions tested
A replicated study in 1995 along two highways in Florida, USA (Land & Lotz 1996) found that large underpasses and box culverts, in areas with roadside fencing, were used by a range of mammal species. Mammals recorded using large underpasses were white-tailed deer Odocoileus virginianus (5.1 crossings/month), panther Felis concolor (2.2), bobcat Lynx rufus (1.3) and raccoon Procyon lotor (1.4). Box culverts were additionally used by red foxes Vulpes vulpes and otters Lontra canadensis. Two box culverts (2.4 m high, 7 m wide, 15 m long) were monitored along a 6.4-km section of a highway. Two of nine large underpasses (21–25 m wide, 49 m long) with vegetation were monitored along a 15-km section of a different highway. Highways had barrier fencing 3.4 m high with a 1-m overhang. Underpasses were monitored from March or April 1995 (end date not stated) using an infra-red digital counter and camera and by counting tracks.Study and other actions tested
Referenced paperLand D. & Lotz M. (1996) Wildlife crossing designs and use by Florida panthers and other wildlife in southwest Florida. Proceedings of the Trends in Addressing Wildlife Mortality: Transportation Related Wildlife Mortality Seminar, FL-ER-58-96, Florida Department of Transportation,Tallahassee, Florida USA, 379-386.
A before-and-after study in 1993–1995 of a highway in Florida, USA (Roof & Wooding 1996) found that an underpass beneath a highway, in an area with roadside fencing, was used by mammals but the road-kill rate was not reduced. Nine mammal species used the crossing. Most crossings were by rabbits Sylvilagus palustris (69 crossings), racoons Procyon lotor (61), armadillos Dasypus novemcinctus (36), opossums Didelphis virginiana (36), foxes Vulpes vulpes (29) and bobcats Lynx rufus (27). The number of mammals of squirrel size or larger killed on the fenced road section was not significantly different in the 11 months after fence installation (13 animals) relative to the 11 month before (10 animals). A wildlife crossing (14.3 m long, 7.3 m wide and 2.4 m tall) was constructed under the two-lane highway between summer and December 1994. A 3-m-high fence extended along both sides of the highway, 0.6 km in one direction and 1.1 km in the other. Underpass use was determined in December 1994 to December 1995 by footprint surveys and by using a motion-triggered camera. Road-kills were surveyed three times/week from November 1993 to December 1995.Study and other actions tested
A study in 1996–1997 along a highway through forest and grassland in Alberta, Canada (Clevenger 1998; same experimental set-up as Clevenger & Waltho 2000) found that underpasses, in areas with roadside fencing, were used by at least 10 species of medium- and large-sized mammals. Over 12 months at 11 underpasses, there were 1,338 detections of elk Cervus canadensis, 538 of deer Odocoileus spp., 373 of coyotes Canis latrans, 97 of black bears Ursus americanus, 77 of wolves Canis lupus, 29 of cougars Puma concolor and six of grizzly bears Ursus arctos. Most visits resulted in completed passages (96–100%, depending on species). Bighorn sheep Ovis canadensis, mountain goats Oreamnos americanus and moose Alces alces were also detected (frequency not reported). Elk, deer and coyotes used all 11 underpasses, black bears used nine, wolves used six, cougars used five and grizzly bears used three underpasses. The study was conducted along 27 km of a four-lane highway. Wildlife movements were monitored through seven cement open-span underpasses, under two bridges over creeks and through two metal culverts. Barrier fencing, 2.4 m high, ran alongside the highway. Underpasses, constructed in 1986–1991, were located in twinned highway sections. Animal tracks were monitored at each end of each crossing within a sand, silt and clay mix (2 × 4 m), every 3–4 days from November 1996 to October 1997.Study and other actions tested
Referenced paperClevenger A.P. (1998) Permeability of the Trans-Canada highway to wildlife in Banff National Park: importance of crossing structures and factors influencing their effectiveness. Proceedings of the 1998 International Conference on Wildlife Ecology and Transportation, Florida Department of Transportation, Tallahasee, Florida USA, 109-119.
A study in 1999 along a highway in Alberta, Canada (Clevenger & Waltho 1999) found that drainage culverts, in areas with roadside wildlife exclusion fencing, were used by small- and medium-sized mammals. Crossings at 24 culverts included snowshoe hare Lepus americanus (13 crossings at 8 culverts), red squirrel Tamiasciurus hudsonicus (6 crossings at 4 culverts), deer mouse Peromyscus maniculatus (161 crossings at 14 culverts), voles Arvicolinae spp. (5 crossings at 3 culverts) and shrews Sorex spp. (43 crossings at 16 culverts). Weasels Mustela sp., and martens Martes americana also used culverts. Culvert use positively correlated with traffic volume and road width (hare, squirrel, vole), road clearance (squirrel) and culvert length (hare, vole) and negatively correlated with distance to cover (vole), age (hare, squirrel) and openness (squirrel, vole). Shrews preferred larger, more open culverts. Vegetation cover effected use by hares, squirrels and voles. The Trans-Canada highway was expanded to four lanes, with 2.4-m-high wildlife exclusion fencing, in three sections, completed in 1986, 1988 and 1997. Twenty-four drainage culverts were monitored along a 55-km highway section, using multiple sooted track-plates (75 × 30 cm) in each culvert. Plates were checked weekly in January–March 1999. Structural and landscape variables were recorded at culverts.Study and other actions tested
Referenced paperClevenger A.P. & Waltho N. (1999) Dry drainage culvert use and design considerations for small- and medium–sized mammal movement across a major transportation corridor. Proceedings of the 1999 International Conference on Wildlife Ecology and Transportation, Florida Department of Transportation, Tallahassee, Florida USA, 264-270.
A study in 1995–1998 along a highway in Alberta, Canada (Clevenger & Waltho 2000; same experimental set-up as Clevenger 1998) found that underpasses, in areas with roadside barrier fencing, were used by large herbivores and carnivores. A total of 8,959 elk Cervus canadensis appearances, 2,411 deer Odocoileus sp. appearances and two moose Alces alces appearances were recorded at 11 underpasses. There were also 193 appearances of black bears Ursus americanus, seven of grizzly bears Ursus arctos, 117 of cougars Puma concolor and 311 of wolves Canis lupus. On 98% of visits, the animal passed through. Features that positively influenced use of underpasses included increased length, noise level and distance to drainage. Increased width, openness, distance to forest and human activities negatively influenced their use. Nine cement open-span underpasses and two metal culverts (length: 26–96 m, width: 4–15 m, height: 2.5–4.0 m) were monitored along a 27-km stretch of the four-lane Trans-Canada Highway. Barrier fencing, 2.4 m high, ran alongside the highway. Tracks were monitored in sand or clay at each end of each crossing, every 3–4 days, from January 1995 to March 1996 and November 1996 to June 1998. Information about structure, landscape and human activity were recorded for each underpass.Study and other actions tested
A study in 1997 along a highway in Jutland, Denmark (Mathiasen & Madsen 2000) found that an underpass, in an area with roadside barrier fencing, was used by four mammal species. These were red fox Vulpes vulpes (122 observations, 161 tracks), badger Meles meles (16 observations, 22 tracks), stone marten Martes foina (18 observations, 41 tracks) and roe deer Capreolus capreolus (20 observations, 41 tracks). The roe deer records were all accounted for by a single male, with other animals present in the area not using the underpass. Three brown hares Lepus europeaus were observed entering the underpass, but all turned around and did not pass through. The entrance of a tunnel underpass (13 m wide, 7.5 m high, 155 m long) was monitored using a video camera and two infra-red lamps for 30 days in April–May and in August–September 1997 (total 495 hours). Tracks in sand at either end of the stream through the underpass were recorded daily. There was 1.8-m-high fencing both sides of the highway, for 1 km in each direction from the underpass.Study and other actions tested
A before-and-after study in 1981–1999 in temperate mixed woodland forest and grassland in Alberta, Canada (Clevenger et al. 2001) found that underpasses and overpasses, along with roadside fencing, reduced road deaths of large mammals. Wildlife-vehicle collisions were significantly lower during the two years after fencing (5–28/year) compared to the two years before (18–93/year) for all three road sections, despite an increase in traffic flow. Ungulate casualties declined by 80%. Species included coyote Canis latrans, black bear Ursus americanus, wolf Canis lupus, bighorn sheep Ovis Canadensis, moose Alces alces, deer Odocoileus spp. and elk Cervus canadensis. Most road deaths were within 1 km of the end of the fences. Deaths also occurred close to drainage structures. The Trans-Canada highway was expanded to four lanes and had 2.4-m-high wildlife exclusion fence installed in three phased sections, completed in 1984 (10 km), 1987 (16 km) and 1997 (18 km). Twenty-two wildlife underpasses and two overpasses were constructed along these sections. Wildlife-vehicle collisions were monitored from May 1981 to December 1999.Study and other actions tested
A study in 2002–2003 of a highway bisecting forest blocks in Victoria, Australia (Abson & Lawrence 2003) found that an underpass, along with roadside fencing, was used by 13 native mammal species. These comprised 76% of mammal species recorded in the adjacent forest (bats not included). The underpass was used by koalas Phascolarctos cinereus, wombats Lasiorhinus latifrons, echidnas, macropods (e.g. kangaroos, wallabies), rodents and carnivorous marsupials (four of five species), and gliders and possums (four of seven species). In 1997, a 70-m wide underpass was built under a split dual-carriageway bridge. Some vegetation was retained and some planted within the underpass. Barrier fencing, 2 m high, ran the length of the highway (with koala escape poles). Intensive sampling was carried out for one week/month in July 2002–June 2003, within the underpass and at two forest sites, 100 m and 320 m from the underpass. Small mammal traps, hair tubes, nest boxes for arboreal mammals, spotlight counts, track surveys and scat surveys were used to monitor wildlife.Study and other actions tested
Referenced paperAbson R.N. & Lawrence R.E. (2003) Monitoring the use of the Slaty Creek wildlife underpass, Calder Freeway, Blackforest, Macedon, Victoria, Australia. Proceedings of the 2003 International Conference on Ecology and Transportation, Center for Transportation and the Environment, North Carolina State University, Raleigh NC, USA, 303-308.
A replicated study in 2000–2003 along a highway in Pennsylvania, USA (Brudin 2003) found that a range of mammals used box culverts and bridge underpasses, some of which were in areas with roadside fencing. In the first phase, eight of nine culverts were used by mammals, with white-tailed deer Odocoileus virginianus (one culvert), raccoon Procyon lotor (seven), opossum Didelphis marsupialis (two), feral cat Felis catus (one), long-tailed weasel Mustela frenata (one), red fox Vulpes fulva (one), striped skunk Mephitis mephitis (one) and black bear Ursus americanus (one) recorded. In the second phase, white-tailed deer used nine of 20 larger culverts (with higher cross-section:length ratios). Black bears, opossums, raccoons and muskrats Ondatra zibethicus also used these culverts. Deer did not use culverts >90 m long, but use was not affected by substrate (concrete, natural or water). In September–November 2000, nine culverts were monitored using infrared-triggered cameras. Approximately half of the culverts had sediment on their floors. Twenty larger culverts that were considered suitable for deer (out of 70) were monitored using cameras, 10 in September–November 2002 and 10 in May–July 2003. Entrances to 13 of these were separated from roads by right-of-way fencing.Study and other actions tested
A before-and-after study in 2002–2003 along a highway in Arizona, USA (Dodd et al. 2003; same experimental set-up as Dodd et al. 2006) found that two open-span bridge underpasses, in areas with roadside elk-proof fencing, were used by elk Cervus canadensis but not by white-tailed deer Odocoileus virginianus and vehicle-deer collisions did not decrease after installation. A total of 181 collisions were reported, with no difference in rates along the section before and after the two underpasses were constructed. GPS collars recorded 675 highway crossings by elk, only 6% of which were through underpasses. Overall, 62% of 1,435 elk, but only 0.4% of 257 white-tailed deer recorded on cameras at underpasses crossed through them. Two open-span bridge underpasses (<250 m apart) along the State Route 260 highway were monitored using video cameras and track counts (inside and 60 m from entrances). Cameras were also installed at the ends of the short sections of elk-proof fencing. Thirty elk were tracked using GPS collars (May 2002 to July 2003). Vehicle-deer collisions were recorded before and after underpass installation.Study and other actions tested
Referenced paperDodd N.L., Gagon J.W. & Schweinsburg R.E. (2003) Evaluation of measures to minimize wildlife-vehicle collisions and maintain wildlife permeability across highways in Arizona, USA. Proceedings of the 2003 International Conference on Ecology and Transportation, Center for Transportation and the Environment, North Carolina State University, Raleigh NC, USA, 353-354.
A study in 2001–2003 along two highways in Wyoming, USA (Gordon & Anderson 2003) found that use of underpasses, in areas with roadside fencing, by mule deer Odocoileus hemionus decreased with a decrease in underpass width. Only one of the six underpasses was consistently used by mule deer, accounting for 91% of the 1,028 recorded crossings made through all underpasses. It had a high cross section:length ratio and was near a historic migration route. At an experimental underpass, the percentage of deer turning away from the underpass increased significantly as the cross section:length ratio decreased. Six (of 12) underpasses along a section of Interstate 80 were monitored. Four were box type and two were small gravel road underpasses. Use was assessed using infrared-triggered cameras and track surveys. One experimental underpass was installed in 2001. It was 18 m long. The width was experimentally manipulated from 3–6 m and height from 2–3 m. Video cameras recorded deer behaviour. Underpasses were monitored from autumn 2001 to spring 2003. Fences, 2.4 m high, ran alongside the highway.Study and other actions tested
Referenced paperGordon K.M. & Anderson S.H. (2003) Mule deer use of underpasses in western and southeastern Wyoming. Proceedings of the 2003 International Conference on Ecology and Transportation, Center for Transportation and the Environment, North Carolina State University, Raleigh NC, USA, 309-318.
A study in 2001–2002 along a highway in Wyoming, USA (Plumb et al. 2003) found that an underpass, in an area with roadside deer-proof fencing, was used by pronghorn Antilocapra americana. A total of 70 pronghorns passed through the underpass over 11 occasions between December and April (group size 1–57). These animals did not hesitate before crossing. An additional 19 pronghorns approached the structure but did not cross. All but two crossings took place at dusk or pre-dawn and most were in the presence of mule deer Odocoileus hemionus. A 2.4-m-high deer-proof fence was constructed in 1989 alongside 11 km of United States Highway 30. In 2001, a wildlife underpass was constructed. Underpass use was monitored using motion sensors with infrared-triggered cameras at either end from October 2001 to May 2002.Study and other actions tested
A study in 2002–2003 along a highway in Montana, USA (Servheen et al. 2003) found that seven bridge underpasses, in areas with roadside fencing, were used by white-tailed deer Odocoileus virginianus, mule deer Odocoileus hemionus and elk Cervus canadensis. White-tailed deer were photographed 791 times, mule deer 379 times and elk 100 times. Between 38 and 430 deer were recorded at each underpass, but none in culverts. Smaller numbers were recorded of striped skunk Mephitis mephitis (nine photographs), raccoon Procyon lotor (three), red fox Vulpes vulpes (one), coyote Canis latrans (three) and black bear Ursus americanus (one). There was no significant relationships between wildlife use and underpass structural features. Distribution of mammal road deaths was independent of underpass locations. Seven bridge underpasses and three culverts were monitored along an 80-km highway section from October 2002 to July 2003. Crossings connected with roadside fencing, though this was inadequately maintained and was permeable to deer. Heat- and motion- sensitive cameras were used at underpasses (for 101–700 camera days/underpass). Details about location, structure, vegetation cover and human activities were recorded for each underpass. Road deaths were opportunistically recorded and combined with data collected by road maintenance crews (spanning 1998–2002).Study and other actions tested
Referenced paperServheen C., Shoemaker R. & Lawrence L. (2003) A sampling of wildlife use in relation to structure variables for bridges and culverts under I-90 between Alberton and St. Regis, Montana. Proceedings of the 2003 International Conference on Ecology and Transportation, Center for Transportation and the Environment, North Carolina State University, Raleigh NC, USA, 331-341.
A study in 2000–2001 in coastal lowlands in New South Wales, Australia (Taylor & Goldingay 2003) found that concrete wildlife culverts, in areas with roadside fencing, were used by small and medium-sized mammals. Mammal tracks made up 82% of all vertebrate tracks recorded. These were made by bandicoots Perameloidea (25% of all tracks), rats (25%), wallabies (13%), mice Muridae (10%), feral cat Felis catus (<2%) and red foxes Vulpes vulpes (<2%). Koala Phascolarctos cinereus tracks were recorded twice. In cage traps, house mouse Mus musculus (29 individuals) and swamp rat Rattus lutreolus (16 individuals) were the most common among six species (67 individuals) caught. Nine concrete culverts along a 2.5-km section of highway were monitored. They were 2.4 m wide, 1.2 m high and 18 m long. A 1.8-m-high fence ran along either side of the road. Tracks were recorded on sand in culverts from 22–30 September 2000 and 1–9 December 2000. Between 15 and 17 cage traps were set in and next to each culvert on four nights in September 2000 (560 trap-nights).Study and other actions tested
A study in 2001–2002 along a highway in Florida, USA (Dodd et al. 2004) found that culverts, in areas with roadside barrier walls, were used by mammals but road casualties still occurred. Ten mammal species (and one species pair) were recorded using culverts. These included rice rat/hispid cotton rat Oryzomys palustris/Sigmodon hispidus (in five culverts), cotton mouse Peromyscus gossypinus (three culverts), round-tailed muskrat Neofiber alleni (three culverts) and southeastern short-tailed shrew Blarina carolinensis (two culverts). Other species used one culvert each. During the same period, ≥13 mammal species were recorded dead on the road. The most frequent casualties were rice rat (25), Virginia opossum Didelphis virginianus (15) and nine-banded armadillo Dasypus novemcinctus (10). Culverts reduced overall vertebrate road mortality, but separate mammal figures were not reported for before culverts were installed. Eight culverts (from 0.9 m diameter to 2.4 × 2.4 m cross-section, all 44 m long) were connected using prefabricated concrete barrier walls. Culverts were monitored from 14 March 2001 to 5 March 2002 using funnel traps, camera traps and sand track stations. Roadkills were monitored by walking the 3.2-km road over three consecutive days each week.Study and other actions tested
A study in 1999–2000 in Alberta, Canada (McDonald & St Clair 2004) found that small culverts, in areas with roadside barrier fencing, were used by mice and voles more than were larger underpasses. More translocated animals returned to their capture location through 0.3-m-diameter culverts (deer mice Peromyscus maniculatus: 100% returned; red-backed voles Clethrionomys gapperi: 86%; meadow voles Microtus pennsylvanicus: 58%) than through 3-m-wide underpasses (69, 49, 10% respectively). More animals successfully returned through underpasses (and overpasses) with 100% vegetation cover at entrances (55–100% of animals returned) compared to those with 50% (20–76%) or no cover (0–66%). Animals crossed within 1–4 days. Nine vegetated soft-bottomed, unvegetated arch-shaped underpasses (64–73 m long) and nine metal drainage culverts with grass cover (63–72 m long) were studied. Crossings were linked to roadside fencing that limited movements of large animals. Territorial mice and voles were captured using Longworth live traps (166 caught), ear-tagged, coated with fluorescent powder, taken across the road, released at standardized distances from crossings (20, 40, 60 m) and followed as they returned. Vegetation cover 2 m inside and outside entrances was varied using spruce branches to 100%, 50% and no cover. Traps at original capture sites were monitored for four days after translocation. Monitoring was undertaken in July–October 1999 and 2000.Study and other actions tested
A study in 1997–2000 of a highway in Alberta, Canada (Clevenger & Waltho 2005) found that underpasses, in areas with roadside fencing, were used by large mammals. The 11 underpasses were visited by elk Cervus canadensis (1302 records), deer Odocoileus sp. (543), cougars Puma concolor (105), black bears Ursus americanus (103), wolves Canis lupus (43) and grizzly bears Ursus arctos (six). The majority of animals that visited underpasses crossed through the structures. Underpass height and width were both positively correlated with the number of animals using them. Two bridge underpasses (3 m high, 11 m wide), four concrete box underpasses (2.5 × 3.0 m) and five metal culverts (4 m high, 7 m wide) were monitored along an 18-km stretch of the four-lane Trans-Canada Highway. Barrier fencing, 2.4 m high, ran along the highway. Tracks were monitored at each end of each crossing, in a 2 × 4-m sand, silt and clay tracking station, every 3–4 days from November 1997 to August 2000. Information about each structure, the surrounding landscape, and human activity were recorded for each underpass.Study and other actions tested
A study in 2002 of a highway in Zamora, Spain (Mata et al. 2005; same experimental set-up as Mata et al. 2008) found that underpasses and culverts, in areas with roadside barrier fencing, were used by mammals. Circular culverts were used by hedgehog Erinaceus europaeus, garden dormouse Eliomys quercinus, badger Meles meles, common genet Genetta genetta and red fox Vulpes vulpes. Adapted (enlarged) culverts were used by red squirrel Sciurus vulgaris, badger and red fox. Open-span underpasses were used by hedgehog, badger, red fox and red deer Cervus elaphus. Wildlife underpasses were used by hedgehog, badger, common genet and red fox. Crossings were also used by rodents and shrews, rabbit Oryctolagus cuniculus, Iberian hare Lepus granatensis, weasel Mustela nivalis, European wildcat Felis silvestris and wolf Canis lupus (see paper for details). Sixty-four underpasses/culverts (30–150 m long) under a 72-km section of motorway were monitored. These included 33 circular drainage culverts (2 m diameter), 10 wildlife-adapted box culverts (2–3 m wide, 2 m high), 14 open-span underpasses (rural tracks/paths, 4–9 m wide, 4–6 m high) and seven wildlife underpasses (20 m wide, 5–7 m high). The motorway was barrier-fenced. Animal tracks were monitored over 10 days in June–September 2002 using marble dust (1-m-wide cross). Camera traps verified species identifications in some underpasses.Study and other actions tested
A study in 2002–2005 along a highway through riparian meadows in Arizona, USA (Dodd et al. 2006; same experimental set-up as Dodd et al. 2003) found that two open-span bridge underpasses, in areas with roadside ungulate-proof fencing, were used by Rocky Mountain elk Cervus canadensis nelsoni, with a more open, shorter underpass with natural sides being used most frequently. In total, 3,708 elk, in 1,266 groups, were recorded at the two underpasses (91% of all mammals recorded) with 2,612 elk in 905 groups passing through the underpasses. More elk groups passed through the shorter underpass (663 groups) than through the longer underpass (242 groups). Seven additional mammal species were recorded at the two underpasses (species not stated in paper). Two open-span bridge underpasses (<250 m apart), along the State Route 260 highway, were studied. Fencing, 2.4 m high, along 0.6 km of highway, funneled animals towards underpasses. Underpasses were monitored using four video cameras, in September 2002 to September 2005. The shorter underpass was 7 m high, 10 m wide and 53 m long, with open, natural sides. The longer underpass was 12 m high, 16 m wide and 111 m long, with concrete walls.Study and other actions tested
A replicated study in 2004 along two roads through agricultural land in Alentejo, Portugal (Ascensão & Mira 2007) found that all 34 monitored culverts, some in areas with roadside fencing, were used by mammals. Crossings were made by small mammals (289 crossings, 34 culverts), hedgehogs Erinaceus europaeus (55 crossings, 15 culverts), hares and rabbits (71 crossings, 15 culverts), weasels Mustela nivalis (16 crossings, 9 culverts), stone martens Martes foina (93 crossings, 28 culverts), Eurasian badgers Meles meles (55 crossings, 10 culverts), otters Lutra lutra (2 crossings, 2 culverts), common genets Genetta genetta (65 crossings, 20 culverts), Egyptian mongooses Herpestes ichneumon (82 crossings, 21 culverts) and red foxes Vulpes vulpes (27 crossings, 12 culverts). A total of 34 culverts (<1.0 m wide, 8–25 m long) were monitored along two roads (17 culverts along each). Road sections studied were 16 and 30 km long. There was 1.5-m-high roadside fencing along the 30-km section. Tracks were monitored using marble dust (60–100 cm wide) which was placed inside each end of each culvert. Tracks were recorded on four days in each of spring, summer and autumn 2004 (total 408 culvert monitoring days).Study and other actions tested
A study in 2007 along a road, in Northumberland, UK (Baker et al. 2007) found that three underpasses, with entrances fenced off from the road, were used by several species of small and medium-sized mammals to make crossings. Tracks were identified of western hedgehog Erinaceus europaeus, brown rat Rattus norvegicus, badger Meles meles and American mink Mustela vison. The number of underpasses used and frequency of use was not detailed in the paper. Underpasses, 0.6–0.9 m wide, were constructed in 2003–2006 along a 46-km stretch of road and were fenced off from the road. Mammal use was monitored in August–October 2007. Clay-based drain seals (45 × 45-cm surface and 0.5 cm thick), used as footprint pads, were placed at entrances to three dry culverts and checked weekly for footprints.Study and other actions tested
A before-and-after study in 1990–2005 along a highway in Québec, Canada (Leblond et al. 2007) found that an underpass was used by moose Alces alces and, along with electric fences, it reduced moose-vehicle collisions. Twenty-three sets of moose tracks were recorded in the underpass over three years. There were fewer moose-vehicle collisions after fence construction (zero) than before (1.4/year). An underpass (23 m long, 16 m wide, 7 m high) was established along both side of a river, under a bridge along the highway. Electric fences (1.5 m high, wires 0.3 m apart) were installed along both sides of a 5-km highway section, encompassing the underpass, in 2002. Data on moose-vehicle collisions before fence installation were collated by the Ministère des Transports du Québec, between 1990 and 2002. Details of monitoring collisions after installation are not given.Study and other actions tested
A before-and-after study in 2004–2007 along a highway through eucalypt woodland in Queensland, Australia (Bond & Jones 2008) found that two underpasses, in areas with roadside barrier fencing, were used by mammals and the mammal road casualty rate fell after construction. There were three wild mammal road casualties over 29 months post-construction and six during four months pre-construction. This comparison was not tested for statistical significance. Tracks detected in underpasses were from rodents (370 tracks), house mice Mus musculus (115), Dasyurid sp. (most likely Common dunnart Sminthopsis murina) (17), northern brown bandicoots Isoodon macrourus (179), possums (16), red-necked wallabies Macropus rufogriseus (3), short-beaked echidnas Tachyglossus aculeatus (2) and from feral cats Felis catus, dogs Canis lupus familiaris and brown hares Lepus europaeus. Proportions of tracks representing full crossings varied by species with the highest figure for wild mammals being for possums (18–40% of records). In 2004, a 1.3-km section of highway was upgraded to four lanes and a variety of wildlife crossings constructed, linked by barrier fencing (2.5 m high). Use of two underpasses (2.4 m high, 2.5 m wide, 48 m long) with water flowing through and ledges attached to side walls, was monitored, starting six months after construction. Tracks were counted on sand within each entrance, twice weekly from August 2005–February 2006 and monthly from June 2006–June 2007. Road-kill was monitored twice weekly before (April–July 2004) and weekly after construction until June 2007.Study and other actions tested
A before-and-after study in 1996–2004 in Florida, USA (Braden et al. 2008, same experimental set-up as Parker et al. 2008 and Parker et al. 2011) found that two underpasses, along with roadside barrier fencing, reduced Florida Key deer Odocoileus virginianus clavium collisions with vehicles by 94%. There were 2 collisions/year over two years after fence construction compared to 12–20 collisions/year over five years before construction (total 79 collisions). Underpass use increased over time, with 22 photographs of deer/month over the first six months and 59/month over the following six months. Average annual deer ranges and core areas did not change after underpass construction. Only 45% (5/11) of radio-collared deer were located on both sides of the highway after construction compared to 100% (9/9) before. In 2002, two box underpasses (14 × 8 × 3 m) were constructed with 2.6-km-long barrier fencing (2.4 m high) and four deer guards (modified cattle guards) installed between them, along a two-lane highway. Deer mortalities on roads were recorded from 1996, by direct sightings, law enforcement reports and observations of vultures. Underpass use was monitored using infrared-triggered cameras from February 2003–January 2004. Deer were radio-tracked between January 1998 and December 2000 (44 deer) and between February 2003 and January 2004 (32 deer) and were located 6–7 times/week.Study and other actions tested
A replicated study in 2004 along two roads in southern Portugal (Grilo et al. 2008) found that underpasses and culverts along roads bounded by livestock fencing were used by carnivore species to cross highways. Crossing rates of underpasses were similar to those of culverts for red fox Vulpes vulpes (underpasses: 0.25 crossings/day; culverts: 0.11), badger Meles meles (underpasses: 0.30; culverts: 0.15), genet Genetta genetta (underpasses: 0.15; culverts: 0.9) and Egyptian mongoose Herpestes ichneumon (underpasses: 0.29; culverts: 0.22). Stone marten Martes foina used underpasses more (0.22 crossings/day) than they used culverts (0.05 crossings/day). Fifty-seven passages under 252 km of two major roads were monitored. They comprised 1.2 circular culverts/km (1 and 1.5 m diameters), 0.3 box culverts/km (2 × 2 m to 5 × 5 m), and 0.5 underpasses/km (5 m high and 8 m wide). Crossing structures were 5–1,566 m apart. Livestock fencing, 1.5 m high, ran along both sides of both roads. A 1-m2 plot of marble dust was placed at each end and in the middle of each passage. This was checked for tracks every five days, over 20 consecutive days of monitoring, in both spring and summer 2004.Study and other actions tested
A study in 2001 along a highway in Zamora province, Spain (Mata et al. 2008; same experimental set-up as Mata et al. 2005) found that road underpasses and culverts, in areas with roadside barrier fencing, were used by mammals. Wildlife underpasses were the most used out of four structure types, by polecats Mustela putorius (detected on average on 0.2/10 days/underpass), roe deer Capreolus capreolus (0.4/10), red deer Cervus elaphus (0.4/10), wild boar Sus scrofa (0.6/10) and rabbits and hares (1.2/10). Open-span underpasses was the most used structure by small-spotted genets Genetta genetta (0.3/10) and red foxes Vulpes vulpes (4.7/10). European badgers Meles meles (3.1/10) and rats (0.4/10) used wildlife-adapted box culverts more than other structure. Small mammals (1.6/10) were most frequently recorded in circular culverts. Thirty-three crossings were monitored. These comprised five wildlife underpasses (14–20 m wide, 5–8 m high, 30–96 m long), seven open-span underpasses (rural tracks/paths, 4–9 m wide, 4–6 m high, 32–72 m long), seven wildlife-adapted box culverts (2–4 m wide, 2–3 m high, 36–45 m long) and 14 circular drainage culverts (2 m diameter, 35–62 m long). The motorway had barrier fencing along its length. Animal tracks were recorded using marble dust (1-m-wide cross) over 10 days in March–June 2001.Study and other actions tested
A before-and-after study in 2002–2005 along a highway through mixed forest and farmland in southwestern Sweden (Olsson & Widen 2008; same experimental set-up as Olsson et al. 2008) found that following installation of an underpass, overpasses and barrier fencing, moose Alces alces road casualties declined but moose did not use the underpass. There were fewer moose-vehicle collisions after fence construction (zero/year) than before (2.7/year). During construction, 1.8 collisions/year were recorded. Moose were recorded crossing the highway 47 times before construction of crossing features, 76 during and 12 times after features were installed. All crossings after fencing prevented direct road access were via the two wildlife overpasses. Two 6-km sections of a highway were converted to a fenced four-lane highway in 2000–2004. The sections contained one wildlife underpass (35 m long, 4.7 m high, 13 m wide), two wildlife overpasses, three conventional road tunnels and two conventional bridges that could be crossed. Twenty-four moose were radio-collared. Locations were recorded every two hours before construction (February–September 2002), during construction (October 2002–May 2004) and after construction (June 2004–December 2005; 8,830 moose days).Study and other actions tested
A before-and-after study in 2000–2005 in forest and farmland in southwestern Sweden (Olsson et al. 2008; same experimental set-up as Olsson & Widen 2008) found that barrier fencing and three road crossings reduced moose Alces alces and roe deer Capreolus capreolus road-kills. Deaths were reduced 70% from averages of 2.7 moose killed/year and 5.3 roe deer killed/year over the 12 years pre-construction. In 2000–2004, a 12-km section of the European Highway 6 was converted from two to four lanes and 2.2-m-high exclusion fencing was installed along its length. Two overpasses and one underpass were also constructed. Moose and deer casualty rates were collated from casualties reported to police pre-construction (1990–2001) and post-construction (up to 2005).Study and other actions tested
A before-and-after study in 1996–2005 along a highway in Florida, USA (Parker et al. 2008; same experimental set-up as Braden et al. 2008 and Parker et al. 2011) found two underpasses with associated barrier fencing reduced vehicle collisions with Florida Key deer Odocoileus virginianus clavium. Fewer deer were killed on the fenced road section after underpass and fence installation (0–3/year) than before (11–20/year). There were more collisions on unfenced road sections after installation (40/year) than before (24/year), so collisions were not reduced overall. However, deer densities increased and the ratio of collisions to deer numbers suggested that risks of collisions decreased after construction. Deer use of two underpasses increased from the first year after construction (871 detections) to the second and third years (1,857 and 1,629 deer detections respectively). A 2.6-km-long system with two underpasses (dimensions not stated), 2.4-m-high fencing and four deer guards were constructed on US Highway 1. An infrared trail monitor and camera monitored deer passages at the centre of each underpass for three years post-construction (2003–2005). Deer-vehicle collisions were recorded (from 1996) from direct sightings, citizen and law enforcement reports and observations of vultures before (1996–2000) and after (2003–2005) fence and underpass construction.Study and other actions tested
A site comparison study in 2006 along a Highway in New South Wales, Australia (Hayes & Goldingay 2009) found that two underpasses were used by mammals and that presence of crossing-structures along with barrier fencing reduced road-kills. There were fewer road-kills over seven weeks along the section with crossing-structures (0.02/km of survey) than along a section without crossings (0.09/km of survey). The most frequently recorded road casualties were bandicoots (16 casualties) and kangaroos and wallabies (nine casualties). Bandicoots used the two underpasses more than they used the two overpasses (87 vs 28 tracks) as did rodents (82 vs 15). Kangaroos and wallabies used underpasses less than they used overpasses (36 vs 104 tracks). Use was similar between structure types for possums (14 vs 9). There were two concrete box culverts (3 × 3 m, 42–63 m long) and two wildlife bridges (9–37 m wide, with vegetation) with 5 km of exclusion fencing, along a 12-km section of dual-carriageway highway. Tracks were monitored on sand plots across each crossing. Road-kill surveys were conducted along the 12-km section and along a 51-km two-lane section without crossings or fencing. Track and road-kill surveys were conducted up to three times/week over seven weeks in August–September 2006.Study and other actions tested
A study in 2001–2005 along a motorway through forest and agricultural land in Germany (Klar et al. 2009) found that most underpasses and overpasses, in areas with roadside fences, were used by wildcats Felis silvestris to cross roads. Wildcats used crossing structures on 18 of the 21 occasions on which they were recorded <50 m from the motorway. The three underpasses were each used by one cat from a total of eight wildcats that had underpasses located within their home ranges. One 40-m-wide underpass and two road underpasses (9–14 m wide), along with two open-span viaducts and two forest road overpasses, were monitored in 2002–2005. All underpasses were 29 m long. Underpasses were connected to fencing that was designed specifically to exclude wildcats from the road. Twelve wildcats were radio-collared between January 2001 and February 2005. Animals were tracked at night for 3–30 months each.Study and other actions tested
A study in 1999–2001 along a road through beech and fir forest in Gorski kotar, Croatia (Kusak et al. 2009) found that an underpass below a section of road on a viaduct, and separated from the road by barrier fencing, was used by medium to large-sized mammals. Tracks were recorded of roe deer Capreolus capreolus (total 20 tracks), red deer Cervus elaphus (12) wild boar Sus scrofa (1), brown bear Ursus arctos (4), grey wolf Canis lupus (1) and Eurasian lynx Lynx lynx (1). However, the underpass had five times fewer mammal crossings/day than did three overpasses (100–835 m wide). A new highway was constructed in 1998–2004 with 44 wildlife crossings and 2.1-m barrier fencing along a 9-km section. An underpass (569 m wide, below a 25-m-high road viaduct) was monitored. Tracks (in snow, mud or sand) and other animal signs were counted 23 times in January 1999–January 2001.Study and other actions tested
A site comparison study in 2000–2007 along a highway in North Carolina, USA (McCollister & van Manen 2010) found that underpasses and barrier fencing facilitated road crossings by a range of mammals but did not reduce road casualties. Camera traps showed crossings through the three underpasses by white-tailed deer Odocoileus virginianus (2,258 times), raccoon Procyon lotor (125), American black bear Ursus americanus (15), bobcat Lynx rufus (11), gray fox Urocyon cinereoargenteus (eight), Virginia opossum Didelphis virginiana (six), rabbits Sylvilagus spp. (two) and Canis spp. (two). Track counts indicated an additional 3,552 mammal crossings by 15 species, with 90% by white tailed deer. A similar number of mammals was killed over one year on road sections with underpasses and fencing (5.0/km) as on sections without (5.1/km). A four-lane highway was constructed with three underpasses. Barrier fencing, 3 m high, was installed ≥800 m along the highway from each underpass. Gates allowed trapped animals to escape the highway. Underpass use was monitored by 2–3 camera traps /underpass. Twice-weekly track surveys were conducted (on 2.5-m-wide plates across underpasses). Road deaths were recorded along 6 km of road with fencing and underpasses and 11 km without, twice/week, from July 2006–July 2007.Study and other actions tested
A review of 30 papers reporting on monitoring of 329 crossing structures in Australia, Europe and North America (Taylor & Goldingay 2010) found that mammals used most culverts and underpasses, among which some were in areas with roadside barrier fencing. Small mammals used pipes (demonstrated by 6/7 relevant studies), drainage culverts (5/5 studies), adapted culverts (5/5 studies), wildlife underpasses (3/4 studies) and bridge underpasses (2/3 studies). Arboreal mammals used pipes (1/1 studies), drainage culverts (4/4 studies), adapted culverts (4/4 studies) and bridge underpasses (1/1 studies). Medium-sized mammals used pipes (8/11 studies), drainage culverts (12/13 studies), adapted culverts (8/8 studies), wildlife underpasses (6/8 studies) and bridge underpasses (6/7 studies). Large mammals used pipes (6/9 studies), drainage culverts (11/12 studies), adapted culverts (11/11 studies), wildlife underpasses (24/24 studies) and bridge underpasses (14/15 studies). Larger mammals tended to use more open underpasses. Small and medium-sized mammals used underpasses with funnel-fencing or adjoining walls and those with vegetation cover close to entrances. Those with vegetation cover tended to be avoided by some ungulates. Thirty papers reporting monitoring of 329 crossing structures were reviewed. Fourteen papers investigated multiple structure types, resulting in a total of 52 studies of different structure types. Underpasses, from small drainage pipes to dry passage bridges, comprised 82% of crossings.Study and other actions tested
A study in 2003–2007 at six sites along a highway through forest and shrubland in Arizona, USA (Taylor & Goldingay 2010) found that underpasses, in areas with ungulate-proof fencing, were used by white-tailed deer Odocoileus virginianus and that underpass use was not affected by traffic levels. Crossing rates of white-tailed deer that approached underpasses did not differ significantly between traffic volume levels of 0 vehicles/minute (0.28 crossings/approach), 1–2 vehicles/minute (0.34 crossings/approach), 2–4 vehicles/minute (0.40 crossings/approach), 4–6 vehicles/minute (0.27 crossings/approach) and >6 vehicles/minute (0.28 crossings/approach). Deer passage rates and traffic flows were monitored at six wildlife underpasses beneath 27 km of an upgraded four-lane highway. Underpasses were 53–128 m long and 5–15 m high. Five underpasses had a fenced above-ground section (11–48 m long) between the two carriageways. Roadside fencing, 2.4 m high, was gradually installed with the full road section fenced by 2006. Four video cameras with infrared beams monitored traffic and deer at each underpass in 2003–2007. The number of deer approaching within 50 m of underpasses and the number crossing the highway through underpasses was counted.Study and other actions tested
A replicated study in 2010 at 38 sites along nine roads in England, UK (Eldridge & Wynn 2011) found that underpasses, in areas with roadside fencing, were used by badgers Meles meles, Eurasian otters Lutra lutra, red foxes Vulpes vulpes, European hedgehogs Erinaceus europaeus and brown rats Rattus rattus to cross roads. Of 38 underpasses monitored, 34 were used by badgers. Eurasian otters, red foxes, European hedgehogs and brown rats used underpasses, but the number of underpasses used or crossing frequencies are not reported. Badger footprints were recorded 7–8 times in 14 underpasses, 4–6 times in 11 underpasses and 1–3 times in 9 underpasses. Mammals were monitored in 38 underpasses, installed in 2003–2007, under single carriageway roads (16 underpasses), dual carriageways (20 underpasses), a motorway (one underpass) and a junction (one underpass). Underpasses were 20–120 m long, 0.3–1 m in diameter (most were 0.6 m diameter) and were made of concrete and corrugated iron. Roadside fence characteristics are not specified. Mammals were surveyed weekly, between August and October 2010, by monitoring footprints in a clay mat (45 × 45 cm) at the entrance of each underpass.Study and other actions tested
A replicated study in 2002–2008 along a highway in Arizona, USA (Gagnon et al. 2011) found that wildlife underpasses, in areas with roadside ungulate-proof fencing, were used by mammals. Six underpasses were approached 14,683 times by wild mammals, of 15 species. Of all animals recorded (which included also 450 records of domestic animals and one of a bird) 72% crossed through underpasses. Elk Cervus canadensis accounted for 70% of visits by wild mammals to underpasses, white-tailed deer Odocoileus virginianus for 13% and mule deer Odocoileus hemionus for 7%. Other crossings comprised coyote Canis latrans (1%), gray fox Urocyon cinereoargenteus (2%), raccoon Procyon lotor (2%) and other mammals (4%). Reconstruction of a 27-km stretch of State Route 260 was undertaken in 2000–2006 and included creation of 11 large wildlife underpasses, connected to ungulate-proof fencing. Six underpasses (34–41 m wide, 5–12 m high and 53–128 m long) were monitored for an average 4.7 (2.5–5.5) years using animal-triggered multi-camera video surveillance.Study and other actions tested
A before-and-after, site comparison study in 1996–2009 along a highway through woodland and developed areas in Florida, USA (Parker et al. 2011; same experimental set-up as Braden et al. 2008 and Parker et al. 2008) found that underpasses beneath the highway, along with roadside fencing, reduced vehicle collisions with Florida Key deer Odocoileus virginianus clavium. Fewer deer were killed on the road over seven years after underpass and fence installation (1.6/year) than in the five years before installation (15.6/year). Concurrently, along an unfenced section without underpasses, 43 deer/year were killed in the latter period and 24/year were killed in the earlier period. Underpass use increased from 185 passages during the first year after construction to 1,337 passages in the seventh year after construction. A highway was upgraded to increase vehicle capacity, with construction completed in 2002. Two box culvert underpasses (14 m long, 8 m wide, 3 m high) were installed under a 2.6-km-long fenced road section through undeveloped land. Deer-vehicle collisions were monitored along this section and along an adjacent 3.0-km-long unfenced section through a developed area, before culvert installation (1996–2000) and after (2003–2009). Culvert use was monitored using camera traps.Study and other actions tested
A before-and-after study in 1990–2011 of scrubland in Wyoming, USA (Sawyer et al. 2012) found that underpasses beneath a highway, in areas with roadside game-proof fencing, were extensively used by mule deer Odocoileus hemionus and collisions between deer and vehicles reduced. Over three years, 49,146 mule deer were recorded moving through seven underpasses. Passage rates through underpasses of deer approaching to ≤50 m increased over three years, from 54% to 92%. After underpass construction, there were 1.8 collisions/month between deer and vehicles compared to 9.8 collisions/month before. Underpasses were also used by elk Cervus canadensis (1,953 crossings), pronghorns Antilocapra americana (201), coyotes Canus latrans (13), bobcats Lynx rufus (77), badgers Taxidea taxus (9), moose Alces alces (13), raccoons Procyon lotor (3) and cougars Puma concolor (1). Seven concrete underpasses (approximately 6 m wide, 3 m high and 18 m long) and 21 km of fencing were installed in 2001–2008. Three camera traps/underpass were operated from 1 October (16 December in first year) to 31 May between 2008–2009 and 2010–2011. Vehicle-deer collision data were collated before (1 January 1990–1 October 2001) and after underpass construction (1 October 2008–1 May 2011).Study and other actions tested
A study in 2006–2008 of 18 wildlife crossings under a highway, along with roadside fencing, in a national park in Alberta, Canada (Sawaya et al. 2013) found that American black bears Ursus americanus and grizzly bears Ursus arctos used underpasses. Over three years, 218 crossings of American black bears and 153 of grizzly bears were detected. These were through 13 culverts (black bear: 44 crossings; grizzly bear: 36) and five open-span underpasses (black bear: 174 crossings; grizzly bear: 117). Bear crossings were monitored at 20 of 25 wildlife crossing structures in Bow Valley, Banff National Park, including 18 culverts and underpasses. Fencing (2.4 m high) was installed alongside the road. Bear tracks were counted in May–October 2006, April–October 2007 and April–October 2008 on track pads, comprising 1.5–2 m of sandy loam, spanning the width of the wildlife crossing. Track pads were checked every two days and the species, direction of travel, and number of animals was recorded.Study and other actions tested
A study in 1997–2009 along a major road in California, USA (Alonso et al. 2014) found that all 19 culverts under the road (most of which were in areas with roadside fencing) were used as road crossing points by coyotes Canis latrens, bobcats Lynx rufus, and mule deer Odocoileus hemionus. Coyotes used 18–19 of the 19 culverts studied, and bobcats used 13–19 culverts. Mule deer used 1–4 of the five underpasses considered suitable for them. Ranges represent the numbers of culverts used in each of two survey periods. Sixteen culverts were part of a road upgrade programmme, conducted in 2005, that included installation of 3-m-high roadside fencing. From November 1997 to January 2000, remotely triggered cameras were placed in each culvert. Cameras were again placed in each culvert from August 2008 to September 2009. Between the two surveys, the road network was expanded and adjacent habitat was restored.Study and other actions tested
A review published in 2014 of eleven studies in Australia (Bond & Jones 2014) found that underpasses, separated from roads by fencing, were used by red-necked wallabies Macropus rufogriseus, swamp wallabies Wallabia bicolor, red-legged pademelons Thylogale stigmatica, long-nosed potoroos Potorous tridactylus and Lumholtz’s tree-kangaroos Dendrolagus lumholtzi. At all road underpasses, fencing was used to deter animals crossing roads rather than using underpasses. Underpasses in the study were 1.2–3.4 m high, 2.4–3.7 m wide, and 20–52 m long.Study and other actions tested
A before-and-after study in 2000–2008 along a highway through swamp and woodland in New South Wales, Australia (Taylor & Goldingay 2014) found that after being extended, underpasses beneath a newly constructed carriageway (in areas with roadside fencing), were used less by northern brown bandicoots Isoodon macrourus and long-nosed bandicoots Perameles nasuta. Bandicoot crossings through underpasses averaged 0.03/day after underpass extension, compared to 0.5/day during road widening and 1.1/day before widening. Construction of a single-carriageway by-pass finished in 1998. Six underpasses, 90–240 m apart, along 750 m of bypass, were studied. Underpasses were 2.4 m wide, 1.2 m high and 17–19 m long. In 2005–2006, an additional highway carriageway was constructed, with a 20–30-m-wide vegetated central strip. Four underpasses were extended, with an above-ground, enclosed section across the central strip, one underpass ran continuously under both carriageways and one linked with a creek bridge under the new carriageway. Crossings were 49–58 m long. Crossing entrances were separated from the road by 1.8-m-high fencing. Footprint sand pads were checked daily over 4–8 days to document tunnel passages. Underpasses were surveyed five times before widening (spring 2000 to autumn 2005), four times during widening (spring 2005 to spring 2006) and four time after widening (summer 2007 to autumn 2008). Not all underpasses were surveyed each time.Study and other actions tested
A study in 2012–2013 in six urban sites in Western Australia, Australia (Chambers & Bencini 2015) found that underpasses, separated from roads by fencing, were used by mammals to cross the road. Southern brown bandicoots Isoodon obesulus fusciventer crossed 540 times, western grey kangaroos Macropus fuliginosus crossed 186 times and brushtail possums Trichosurus vulpecula crossed twice. Underpasses were also used by several invasive mammal species. Road crossings were monitored through 10 underpasses from May 2012 to May 2013, using camera traps. Underpasses were round (0.6–0.9 m diameter) or square culverts (0.6–1.2 m wide, 0.5–1.2 m high). They were 23–88 m long and separated from roads by 0.6–1.8-m-high fences. The time since construction ranged from two to 19 years.Study and other actions tested
A study in 2010–2012 of a desert region of California, USA (Murphy-Mariscal et al. 2015) found that underpasses in areas with roadside fencing were used by a range of native mammals. There were 3,778 wildlife occurrences (mammals and birds) recorded over 4,279 monitoring days (where a monitoring day is one underpass monitored for one day). Rodents made up 32% of occurrences. Rabbits and hares, mainly desert cottontails Sylvilagus audubonii, made up 29%. Birds made up 27% of wildlife occurrences. Other mammals recorded included mule deer Odocoileus hemionus, mountain lion Puma concolor, bobcat Lynx rufus, coyote Canis latrans and ground squirrels (frequencies not reported). Seven underpasses, measuring 18–150 m wide, 3–9 m high and 12–112 m long, were studied. Roads were fenced, but gaps allowed animal passage and fences did not funnel animals towards underpasses. Wildlife movements were monitored from July 2010 to November 2012, using camera traps and track pads.Study and other actions tested
A replicated, site comparison study in 2013 along a highway in Montana, USA (Huijser et al. 2016) found that underpasses connected with long roadside fences were used by similar numbers of large mammals compared to those with no fences or very short fences. The rate of large mammal crossings through underpasses connected to 6.1–6.2-km-long roadside fences (0.44 mammals/underpass/day) and 1.4–2.7-km-long fences (0.77 mammals/underpass/day) was not significantly different to the rate crossing through underpasses with no fencing or with fences up to 0.4 km long (0.22 mammals/underpass/day). Mammals identified using underpasses were white-tailed deer Odocoileus virginianus, mule deer Odocoileus hemionus, American black bear Ursus americanus, mountain lion Puma concolor, grizzly bear Ursus arctos and elk Cervus canadensis. Twenty-three underpasses were monitored along US Hwy 93 North. Roads were fenced alongside underpasses for 0.0–6.2 km length with 2.4-m high fencing. Wildlife crossings were monitored using ≥1 camera trap/underpass in January–December 2013.Study and other actions tested
Referenced paperHuijser M.P., Fairbank E.R., Camel-Means W., Graham J., Watson V., Basting P. & Becker D. (2016) Effectiveness of short sections of wildlife fencing and crossing structures along highways in reducing wildlife-vehicle collisions and providing safe crossing opportunities for large mammals. Biological Conservation, 197, 61-68
A study in 2012–2013 along a highway in Montana, USA (Huijser et al. 2016) found that underpasses, in areas with roadside fencing, were used by white-tailed deer Odocoileus virginianus for crossing the road more often than was the road surface. This result was not tested for statistical significance. There were 727 road crossings with 721 by white-tailed deer, three by American black bear Ursus americanus and three by either this species or grizzly bear Ursus arctos. Eighty-two percent of all crossings were through underpasses and 18% were above the road. Ten fenced underpasses were monitored along US Hwy 93 North. Underpasses were 2–5 m high and 4–40 m wide. Fences were 2.4 m high and 3–256 m long. The proportion of wildlife crossings did not change with fence length (data presented as regression results). Between June 2012 and October 2013, road crossings were monitored for two weeks/underpass using one camera trap at each fence end and at least one at an underpass entrance. Only highway crossings in which animals entered or exited underpasses or accessed or left the highway at a fence end (not returning within ≤3 minutes) were considered.Study and other actions tested
Referenced paperHuijser M.P., Fairbank E.R., Camel-Means W., Graham J., Watson V., Basting P. & Becker D. (2016) Effectiveness of short sections of wildlife fencing and crossing structures along highways in reducing wildlife-vehicle collisions and providing safe crossing opportunities for large mammals. Biological Conservation, 197, 61-68
A study in 2010–2014 of two sites along a highway in Nevada, USA (Simpson et al. 2016) found that underpasses, in areas with roadside fencing, were used by migratory mule deer Odocoileus hemionus to cross a road, but less so than were overpasses. Fewer mule deer crossed the road through three underpasses (44–629 deer crossings/underpass/season) than across two overpasses (234–4,007 deer crossings/overpass/season). Crossing structures, 1.5–2.0 km apart, at important crossings for migratory deer, were completed by August 2010 (August 2011 for one overpass). One site had two underpasses and one overpass. The other had one of each structure. Underpasses, 8 m wide, 28 m long and 6 m tall, were oval in cross-section. Concrete arch overpasses, were 31–49 m wide and 8–20 m long. All structures had soil bases. Fencing, 2.4 m high, deterred deer access to the highway between crossings and extended 0.8–1.6 km beyond crossings at each site. Crossings were monitored during eight mule deer migratory periods (autumn 2010 to spring 2014), using camera traps, over 10 weeks in each migration (15 September to 1 December and 1 March to 15 May). Cameras were positioned 12 m apart along crossing structures.Study and other actions tested
A study in 1996–2014 of a major highway in Alberta, Canada (Ford et al. 2017) found that culverts, in areas with roadside fencing, were used as crossing points by grizzly bears Ursus arctos, but less often than were overpasses, especially by family groups. Over 18 years, grizzly bears used culverts less often (122 crossings/structure) than they used overpasses (241 crossings/structure). Over eight years, bear family groups used culverts less often (0.0–0.3 family groups/year/structure) than they used overpasses (1.4 family groups/year/structure). In 1996–2006, 2-m-wide pads, were covered in sandy-loam soil to survey bear movements at 23 crossing structures. From 2008 to 2014, remote cameras were installed at all crossing structures. As more crossing structures were built in the area, they were added to the survey, up to a maximum of 19 culverts and 18 overpasses. Crossing structure entrances were separated from the road by fencing.Study and other actions tested