NatureScot Research Report 1354 - Deer Vehicle Collisions - a review of mitigation measures and their effectiveness
Year of publication: 2024
Authors: Putman, R.J. and Langbein, J. (Langbein Wildlife Associates)
Cite as: Putman, R.J. and Langbein, J. 2024. Deer Vehicle Collisions - a review of mitigation measures and their effectiveness. NatureScot Research Report 1354.
Keywords
deer vehicle collision; wildlife vehicle collision; roadkill prevention; mitigation
Background
Over the past 50 years wild deer in Scotland have increased in range and numbers. During the same period total annual volumes of road traffic have more than doubled, combining to make deer vehicle collisions (DVCs) an increasingly common occurrence. The recent Deer Working Group reporting to Scottish Government (Pepper et al., 2019) concluded that road traffic collisions involving deer are now a major issue. Reducing the risk of deer vehicle collisions helps safeguard the public as well as wild deer welfare.
An earlier report on behalf of the Deer Commission for Scotland: Deer and Road Traffic Accidents: A review of mitigation measures: costs and cost effectiveness (Putman, Langbein and Staines, 2004) provided a review of the various factors known to affect the frequency and risk of DVCs, and of the various mitigation measures available at that time and their relative efficacy. Since then, road traffic collisions involving deer and other ungulates have increased also in many other countries, especially in continental Europe and North America, resulting in a large volume of new publications and ongoing research into this issue.
Building on our previous work referenced above, this report provides an updated review of the latest international research findings into the factors behind DVCs, the variety of mitigation measures identified and trialled for reducing DVCs, the range of mitigation measures widely in use, and the effectiveness of these approaches and their costs.
Main findings
- Collisions between vehicles and deer (DVCs) continue to be a significant and increasing problem on roads throughout Europe and North America. In response there has been an enormous number of recent studies exploring the various factors influencing the frequency of DVCs. These factors include road-type (major/minor road) and traffic volume, habitat characteristics of the roadside and habitat mosaic in the wider landscape, deer species, time of day and season. These different factors interact to affect accident risk. Despite the abundance of these new studies, previous conclusions remain unchanged.
- Certain environmental features consistently emerge as characteristic of sites likely to suffer a high frequency of deer-related road traffic collisions. Thus in addition to road type (number of lanes of traffic/width of road), traffic volume and speed and presence or absence of a central barrier, the following are also typically present: woodland or forest cover close to the carriageway, landscape diversity (variability and patch size), and obvious travel corridors across the roadway, such as rivers, dry gullies or other linear structures leading down at an angle to, or perpendicular to the roadway. Deer density at landscape level may be a relatively minor contributing factor and the relationship between deer density and DVC frequency is clearly non-linear.
- In the same way there has been a proliferation of studies exploring the effectiveness of different measures which may be adopted in an effort to reduce the frequency and severity of DVCs. Once again, despite the number of such studies, in the main these simply reinforce earlier conclusions of our 2004 review about the effectiveness or otherwise of various approaches to mitigation. In essence, measures which seemed to be most effective at that time are reinforced by subsequent studies as being better proven now; whereas deterrents which were considered to be of limited or no efficacy in that earlier review are now confirmed to be of very limited or zero utility.
- In Section 3 of this report, we review the most recent analyses of the effectiveness of the various individual mitigation measures most commonly deployed in Europe and North America.
- We conclude that, despite the attractiveness of their relatively low cost, static (conventional) wildlife warning signs, ‘passive’ wildlife warning reflectors (that reflect vehicle lights to alert deer), ‘active’ warning devices that emit acoustic and/or light signals and chemical (olfactory deterrents) have no proven effect in reducing the frequency of DVCs and nor does a posted vehicle speed restriction along the highway.
- On motorways and major highways, despite high cost, highway fencing remains the most effective option in reducing the frequency of DVCs, although it is crucial that fencing is to appropriate specification and is rigorously maintained. Fencing schemes should also incorporate escape ramps (deer leaps) or one-way gates to allow animals which enter and become trapped within the carriageway to escape. To provide continuity of populations and habitat and to prevent animals forcing an impermeable fence, fencing should be designed to lead animals to safe crossing places such as overpasses or underpasses.
- On unfenced roads, the most effective mitigation would seem to be offered by Dynamic Warning Signs activated only when Roadside Animal Detection Systems (RADS) detect animals approaching or actually on the carriageway, alerting oncoming vehicles and enforcing a temporary speed reduction. A number of vehicle manufacturers are currently developing in-vehicle warning devices which might be linked to such RADS.
- On both major and minor roads, removal of tree cover and shrubby vegetation along verges for some distance away from the carriageway, with regular maintenance to prevent redevelopment of dense understorey has been shown to be effective in reducing the frequency of DVCs, both by increasing visibility to drivers of animals near the roadway (and vice versa) and also by making such areas less attractive to deer for shelter/foraging.
- Recent analysis of the main DVC hotspots on the trunk road network in Scotland has highlighted that a very significant proportion of them are associated with road junctions and major interchanges. The frequency of accidents is likely to be linked to the fact that major road junctions commonly include islands of open ground surrounded by tarmac on all sides, defining an area within which there is very limited public disturbance; these areas are commonly planted for landscape purposes or, through lack of disturbance secondarily develop a cover of trees or woody shrubs by natural colonisation. With the combination of good cover and lack of human disturbance such ‘wooded islands’ thus unintentionally develop into refuges that deer and other wildlife will seek out for shelter. Here too removal of trees and scrubby vegetation, as far as is practicable, would be effective in reducing the frequency of DVCs (paragraph 5.17).
- While, as above, deer density at a landscape level would appear to play only a minor role in the frequency of collisions, it is further clear that any relationship that may exist between local deer density and the frequency of DVCs is non-linear so that any reduction in landscape level deer density is unlikely to be rewarded with an equivalent reduction in the rate of DVCs. That said, a reduction in very local deer numbers (of animals in the immediate vicinity of the target roadway – perhaps sheltering or feeding on highway verges) may indeed be effective in reducing the risk of DVCs. Where this is undertaken in the short-term through targeted culling, the effect is likely to be of comparatively short duration, due to inevitable recolonisation. But effective longer-term reduction in these immediately-local deer populations may be achieved, by manipulation of the roadside vegetation, as above, to make these areas less attractive in harbouring significant numbers of animals.
- Whatever the effectiveness (and cost) of individual measures, it is apparent that to be fully effective any strategy for reduction of the frequency of DVCs in a given context will require the deployment of a combination of measures – with that combination of individual measures differently tailored to specific contexts. In Section 6 we make recommendations separately for the combinations of measures most likely to be effective
- On motorways and major A-dual carriageway trunk roads
- On single carriageway A-roads and more minor roads
- At road junctions, slip roads and major roundabouts
- In urban and peri-urban areas
- We note however that while these recommendations may suggest a shortlist of potential measures which might, in combination, prove effective effective in different contexts, each individual situation is unique and planning for any individual location should always begin with an on-site survey to determine the local feasibility of differing potential measures.
- We end with a plea that where mitigation measures are attempted in the future to reduce collision frequency at established hotspots, funding should be found to support appropriate monitoring of their effectiveness. We would also advocate that specific funding is sought to establish the effectiveness of Roadside Animal Detection Systems (RADS) linked to dynamic signage in a number of known hotspots on more minor roads. We would also urge that future mitigation efforts acknowledge the growing numbers and distribution within Scotland of wild boar/ feral pigs.
Acknowledgements
We would thank Dominic Sargent, the NatureScot officer responsible for this project, for his patience throughout, and would also like to thank various colleagues across continental Europe for sharing their experiences with DVC mitigation in their countries.
Abbreviations
Deer Vehicle Collision (DVC)
Wildlife Vehicle Collision (WVC)
1. Introduction
1.1 As road infrastructures proliferate and traffic volume and speed rise, and ungulate densities also increase throughout Europe (Gill, 1990; Apollonio et al., 2010), so the frequency of road traffic collisions involving wildlife also escalates throughout Europe. The scale and recent escalation of wildlife collisions in Europe are mirrored by figures from North America. In many countries numbers of reported deer-vehicle collisions (DVCs) have increased by more than 50% since 1996, and most recent reviews now indicate that over one million wildlife-vehicle collisions (WVCs) occur per annum in Europe (Langbein et. al. 2011; Seiler et. al., 2016; Rosell et al., 2022); the great majority of which relate to a variety of deer species or wild boar, with collision frequencies over 1.5 million per annum in North America (IIHS, 2008).
1.2 Over the past 50 years annual traffic volumes on the Scottish road network have more than doubled (DfT, 2023). During the same period populations of the four main deer species living wild in Scotland (red deer Cervus elaphus, roe deer Capreolus capreolus, fallow Dama dama and sika Cervus nippon) have expanded their range and increased in numbers, although analyses by Albon et al. (2017) would suggest that, at least for red deer, these would now appear to have largely stabilised. In combination these two main factors have led to deer-vehicle collisions (DVCs) becoming increasingly common, not just in rural but also peri-urban parts of Scotland, and the recent Deer Working Group reporting to Scottish Government (Pepper et al., 2019) concluded that road traffic collisions involving deer are now a major issue.
1.3 To gather information on the distribution and trends in DVC occurrence and to identify DVC hotspots, the Deer Commission for Scotland in 2003 commissioned the first ‘DVCs in Scotland: Monitoring Project’ (Langbein and Putman, 2006), which later continued under the auspices of Scottish Natural Heritage (SNH) (Langbein, 2011a, 2019; Lush and Lush, 2023). Based on a sample of 1,200 to 2,000 DVC reports gathered annually via a small number of sources best able to provide consistent input with coverage from across all of Scotland, trends over the past 15 years indicate a gradual increase in DVCs reported from 2008 to 2017. Thereafter numbers reported have declined slightly and plateaued. As there is no legal obligation for deer road casualties to be reported to, or recorded by, any official authority, there is currently no ready way to determine what proportion of all DVCs across Scotland are captured by the above studies. However, estimates using a range of differing indicators have suggested the national figure for DVCs in Scotland to fall within the broad margins of 8,000-14,000 per year (Langbein, 2018).
1.4 Seen in the context of comparable estimates of 34,000 to 60,000 DVCs for England (Langbein, 2007a, 2011b), the above figures for the number of reported collisions in Scotland may appear comparatively low, not least as the total number of deer is thought to be of a similar order (upwards of 750,000) in both countries. However, while annual traffic volumes in Scotland make up just 9% of all UK road traffic, assessment of insurance data provided by a sample of vehicle insurers showed that over 19% of DVCs occur in Scotland (Langbein and Putman, 2006). Taken together these figures indicate that while the number of DVCs may be lower than that in England, the actual risk to drivers of colliding with a deer per driven mile is close to two-fold higher in Scotland than in the rest of the UK.
1.5 Such collisions may have considerable impact. Recent analyses in the UK suggest that the total mortality imposed through DVCs as a proportion of national spring population sizes is estimated to lie between 3% to 7% for roe deer, 1% to 3% for red deer and from 7% to 13% for fallow deer, making DVCs almost certainly the major cause of annual mortality among wild deer aside from deliberate culls undertaken as part of deer management (Langbein, 2007a). In addition to the impact on deer populations themselves, collisions of vehicles with any of the different deer species represent a significant risk-factor to road safety; and many accidents involving deer result in significant damage to persons and/or property (Putman, 1997; Langbein, 2007a).
1.6 In 2004, Putman, Langbein and Staines provided a review of the various factors known to affect collision frequency or risk of collisions with deer and of the various mitigation measures available at that time and their relative efficacy in reducing collision frequency. The current report offers an update of that earlier review and considers the cost-effectiveness of mitigation measures currently available and their suitability for differing road types and landscapes.
1.7 The large rise in numbers of DVCs as well as collisions with other large wildlife globally, has led to much new research over the past two decades into factors affecting DVCs and the effectiveness of differing measures taken to mitigate them, especially in continental Europe and North America, resulting in an almost overwhelming volume of recent publications. In practice, despite the abundance of this newer literature, the main conclusions of our review in 2004 remain largely unchanged, with newer studies simply reinforcing those earlier conclusions about the environmental and other factors affecting the probability and frequency of DVCs and the effectiveness or otherwise of various approaches to mitigation. In essence, measures which seemed to be most effective at that time (e.g. highway fencing linked to fauna passages) are reinforced by subsequent studies as being better or better proven now; whereas deterrents which were considered to be of limited or no efficacy in that earlier review are now confirmed to be of very limited or zero utility. In the current report, we have not referenced all of this very extensive literature (much of which describes rather local studies) but have selected those more recent articles which are particularly relevant or compelling.
1.8 In addition to our own subsequent review in 2011 (Langbein et al., 2011) two comprehensive recent reviews of WVC mitigation are offered by Seiler et al. (2016) in Safe Roads for Wildlife and People and a more recent North-American Review by Huisjer et al. (2021), while an update to the original European Handbook on Wildlife and Traffic (Iuell et al., 2003) has been presented by Rosell et al. (2022) in the on-line Wildlife and Traffic Handbook. These subsequent reviews, conducted by large collaborative teams of authors from differing countries, provide extremely valuable insights for the current report.
1.9 It is notable, however, that all the above also concluded that there is not, nor is there likely to be in future, any single measure that will effectively reduce DVC risk on its own. Instead, a strategy combining several partially effective approaches tailored to a given local situation is required to reduce and continue to maintain DVCs and other WVCs at acceptable levels.
2. Factors affecting collisions frequency
2.1 Collisions with ungulates are not distributed randomly in space and time, and there are a number of environmental factors which affect the frequency of such incidents. These factors include road-type (major/minor road) and traffic volume, habitat characteristics of the roadside and habitat mosaics in the wider landscape, time of day and season. These different factors interact to affect the level of collision risk.
2.2 Furthermore, the different ungulate species themselves appear to differ in their susceptibility to being involved in traffic collisions (Langbein and Putman, 2006), while, partly due to differences in size, collisions with different species of ungulates also have different implications in terms of severity of damage to vehicles and likelihood and severity of injuries to drivers and other vehicle occupants as well as to the ungulate itself (Ückermann, 1983; Conover et al., 1995; Malo et al., 2004; Langbein, 2007a).
Season and time of day
2.3 Deer of all species regularly cross minor roads (narrow roads of relatively low traffic volume) during routine daily movements within an established home range. Although surprisingly few published data are available for red, sika or fallow deer, it is common experience to encounter deer of all species crossing such roads during regular movements to or from foraging areas within their range (Putman, 1997; Langbein, 2007b). Primary roadways (motorways, dual carriageways), as also railways, appear to constitute more of a recognised barrier to movement. ‘Casual’ crossings of these major routes are less frequent and for most deer species the boundaries of home ranges commonly tend to coincide with major ‘barriers’ of this kind. Major roadways are, however, not totally impermeable: in some cases they may indeed be crossed in the context of daily movement around the home range; in addition there are likely to be peaks in road-crossing coinciding with larger-scale dispersal movements (of juveniles leaving their natal range, or mature males seeking mating opportunities during the rut).
2.4 In response to these patterns of movement we might expect crossings of minor roads of low traffic volume to be frequent, regular and distributed throughout the year, since such roads will be crossed regularly during the course of normal movements around an established home range. For crossings of more major roads by contrast, we might expect the first strong seasonal peak in crossings in late spring and early summer as the consequence of dispersal of juveniles of all species. However, other seasonal peaks in road-crossing activity are rather more species-specific and associated with mating movements of, in particular, mature males at the onset and end of the rut (e.g. in mid-summer in the case of roe deer, and in autumn in the case of red, sika and fallow deer, respectively).
2.5 These expectations are matched by available data on patterns of actual traffic collisions involving deer. For both red and fallow deer, a consistent seasonal peak in reported collisions coincides with the September-November peak in movement associated with the onset of the rut in autumn (Ückermann, 1964; Langbein, 1985; Carsignol, 1989; Desire and Recorbet, 1990; Hartwig, 1991; Groot Bruinderink and Hazebroek, 1996; Pokorny, 2006). In a study of movement patterns of fallow deer and the frequency and distribution of traffic collisions involving deer in an area of north Staffordshire (UK) between 1983 and 1985, Langbein (1985) also noted that minor roads showed a high level of deer crossings throughout the year, but crossings of major roads were notably seasonal and concentrated in autumn (October), coinciding with the peak of the fallow rut in that area, and in the period between February and April (conclusions based on a sample of 986 recorded crossings; Langbein, 1985).
2.6 For roe deer the seasonal peaks in collision frequency appear less well-defined, but the majority of road-related mortality falls within the period April-May. This appears to coincide with the reduced speed of movement of females with very young fawns at heel, with the period of dispersal of yearlings from their natal range, as well as with the period of establishing new territories by adults, particularly bucks (Desire and Recorbet, 1990; Hartwig, 1991; Groot Bruinderink and Hazebroek, 1996; SGS Environment, 1997; Pokorny, 2006). Of 3826 accident reports between 2003-2005 in Great Britain where roe deer were positively identified as the species involved, 26% of incidents occurred in April and May (Langbein and Putman, 2006).
Similarly, in Slovenia 21% of all known collisions with roe deer occurred in these same two months both during 1999–2001 (Pokorny, 2006) and in 2007 (unpublished data of Slovene Hunter’s Association). However, the risk for a collision with roe deer is also high during the summer (particularly in the rut period) and autumn (primarily due to the greater activity of fawns and clearance of maize fields) (Pokorny et al., 2008). Meisingset et al. (2014). Hegland and Hamre (2017) note that the frequency of collisions with red deer in Norway increases over winter.
2.7 As well as reporting such seasonal patterns, a number of studies have found that the majority of road-related collisions occur during the hours of darkness, and particularly at dusk or dawn (e.g., Ückermann, 1964; Langbein 1985, 2007a; Desire and Recorbet, 1990; Lavsund and Sandegren, 1991; Hartwig, 1993; Groot Bruinderink and Hazebroek, 1996; Haikonen and Summala, 2001; Seiler, 2004; Pokorny, 2006; Meisingset et al., 2014; Hothorn et al., 2015). While this coincides with the period of maximum deer activity, the effect may also be partly due to reduced driver visibility at these periods and accentuated when ‘rush hour’ periods coincide with poor light conditions of dawn and dusk in autumn (Sanders, 1985; Langbein, 1985). Langbein (1985) suggested that the coincidence of ‘rush hour’ traffic peaks with twilight in autumn and spring may be important in exacerbating the seasonal peaks in traffic accidents; and that this may contribute to the fact that DVCs overall tend to peak just after rather than at the height of the fallow deer, red deer and sika deer rut. This may also contribute to the commonly noted escalation of collisions in autumn with those species of deer, including roe, which do not rut at this time of year.
2.8 A further factor which in the US has been proposed as contributing to the autumnal peak in DVCs is the start of the hunting season (Etter et al., 2002), though this is generally considered of lesser importance than rutting activity (Puglisi et al., 1974; Gleason and Jenks, 1993). The end of October/ beginning of November also currently mark the start of the open season for females for most deer species in the UK (though more generally across Europe hunting seasons on antler-less deer start in early or mid September in most countries), but no European studies have to date investigated effects of hunting pressure on DVCs.
Road type, traffic volume and traffic speed
2.9 It has been clearly established by a number of authors that the frequency of deer vehicle collisions is related to road density, traffic volume and traffic speed (e.g. Langbein, 2007a; Langbein et al., 2011; Hothorn, Brandl, and Müller, 2012) as well as a number of other environmental factors (e.g. Bashore et al., 1985; Finder et al., 1999; Hubbard et al., 2000; Malo et al., 2004; Seiler, 2004; Putman et al., 2004). In all these studies certain consistent features emerge as characteristic of sites likely to suffer a high frequency of deer-related road traffic collisions (Putman et al., 2004); these are: number of lanes of traffic (width of road); traffic volume and speed; presence or absence of a central barrier; but also woodland or forest cover close to the carriageway; landscape diversity (variability and patch size); and the presence of obvious travel corridors across the roadway, such as rivers, dry gullies or other linear structures leading down at an angle to, or perpendicular to the roadway. [See also: e.g. Meisingset et al., 2014; Hegland and Hamre, 2017 - but there are now many such analyses, most of these confirming these same main environmental features].
2.10 Deer density would appear to play only a comparatively minor role in the frequency of collisions (beyond the obvious that if there are no deer in an area there can be no collisions with deer, whereas where deer densities are extremely high, collision risk is inevitably higher). While Seiler (2004) reported that the overall number of recorded collisions with moose and roe deer in Sweden was closely correlated with changes in annual game bags (harvest) and the increase in traffic volume, in a separate study for the Deer Initiative of collision frequency in different sites within the UK, Uzal (2013) argued that collision frequency was not strongly correlated with deer density, demonstrating that environmental or traffic-related factors accounted for up to 70% of recorded variation in the number of DVCs at a landscape [10km2] scale, leaving comparatively little to be explained by variations in ungulate density.
2.11 If we can pay attention to such features in planning, or alter their disposition in relation to the carriageway even retrospectively, we are likely to achieve a far greater and far longer-lasting reduction in risk of wildlife-related traffic accidents than in any attempts at local reduction of density of deer or other larger wildlife.
2.12 It is self-evident that DVCs will be related to the number of vehicles as well as numbers of deer. However, although the density of roads and the speed of traffic have already been demonstrated as a major risk factors for DVCs (e.g. Romin and Bissonette, 1996; Lode, 2000), some formal studies have shown no relation between average daily traffic flow (as the measure of traffic volume) or posted speed limit (as the measure of traffic speed) and DVC occurrence (Bissonette and Kassar, 2008).
2.13 While the majority of recorded DVCs occur on secondary roads, due to their greater overall length within most national road networks, collision frequency (per unit length of carriageway) has often been found to be higher on primary trunk roads or major throughways where speeds of traffic and most of all total traffic volumes are greatest (Pojar et al., 1975; Hartwig, 1993; Bashore et al., 1985; Desire and Recorbet, 1990; Langbein and Putman, 2006; McShea et al., 2008, Langbein. 2011b). For example, in Germany Hartwig (1993) reported that while motorways accounted for 21.2% of all wildlife related road traffic collisions in the area covered by his analysis, they made up only 7% of the length of major roads in the area. Motorways and primary trunk routes together accounted for 37.5% of all recorded collisions but were just 24% of total road length. Higher incidence of DVCs on more major routes would appear to be a function both of higher traffic volume and higher vehicle speed.
2.14 It should be noted however that there are exceptions to this more general rule in some central European countries (e.g. Austria, Croatia, France, Slovenia) where most major highways are completely fenced, and therefore the possibility for deer to come on the road itself is comparatively low. In France, where major highways and motorways have wildlife fencing installed, Saint-Andrieux et al. (2020) report that in 2008 and 2009, among 86,777 vehicles collisions with red deer, roe deer and wild boar recorded by insurance companies, only 1% occurred on highways, even though highways carried 97% of the daily traffic (4,026,400 versus 1,030 vehicles/day/km for highways vs other roads in 2010).
2.15 In considering these data however, we should note in addition that figures may be somewhat distorted since the likelihood of deer and other wildlife casualties to be reported may be significantly greater on roads of high traffic volume. This is especially pertinent in Scotland, where the trunk road (highways) network encompasses a particularly diverse range of road types and traffic volumes, ranging from multi-lane motorways and dual carriageways mostly in the Central Belt, to some A-single roads of very low traffic volume in the Highlands and elsewhere. To help assess the proportion of deer road casualties reported to the highways authorities, from 2003 to 2010 the Deer Commission for Scotland undertook regular carcase searches for three sections totalling 170 km of the A82, A835 and A87/887 (reported in full in Langbein, 2011). Results suggested that while up to 68% (for 2008 to 2010) of carcasses found on the A835 were also independently reported as collisions to the highways authority, in the case of the A87 and A887 reporting of collisions with deer was less than 24% of the number of carcases recovered in the same period; and even lower in the earlier years. The disparity in reporting of DVCs is likely to be less on major routes (where the trunk road operators are likely to be the primary contact to which deer casualties are reported and who, in any case, regularly patrol their routes), than for example A-single roads in the Highlands, where it is far more likely that a local gamekeeper or other local person will deal directly with removing deer casualties from the roadside.
2.16 In order to address this inherent reporting bias of deer casualties between road types, Langbein (2011b) reported an analysis restricted to DVCs leading to human injury attended by police. Of 1243 personal injury DVCs in England, just 2.9% occurred on motorways, 9.3% on A-class trunk roads, 39.3% on A-principal (non-trunk) roads and 48.5 % on more minor roads). For motorways and other trunk roads, this is far below what would be expected purely in relation to the proportion of national traffic volume they carry (20.0% and 10.9% respectively), whereas in the case of A-principal and more minor roads the proportion of traffic carried by them (32.3% and 36.1%) much more closely reflects the proportion of DVCs resulting in personal injury.
2.17 Table 1 shows that the relative distribution of deer related human injury accidents among non-trunk A-roads (39.3%) and Minor roads (48.5%) is almost directly in line with the distribution of total numbers of personal injury accidents among these road types (41.7% and 49.9% respectively). This suggests that, even though minor roads carry only 36% of all road traffic in England, the risk of an injury DVC is marginally greater on minor roads than major roads. This is in accord also with the findings from a number of US studies (e.g. Allen and McCullough, 1976; Grovenburg et al., 2008) which observed greater mortalities due to DVCs on two-lane paved roads than on divided highways and interstate routes (equivalent to dual carriageways and motorways in the UK).
Table 1. Relative distribution of deer related personal injury accidents among motorways, A-class trunk roads, non-trunk A-roads and minor roads in England (based on Langbein, 2011b).
Traffic and PIA data | Period | Motorway | A-trunk | A-principal | Minor | All road types |
---|---|---|---|---|---|---|
Available number of deer related PIA reports | 2003 – 2010 (percentage) | 36 (2.9%) | 116 (9.3%) | 488 (39.3%) | 603 (48.5%) | 1243 (100%) |
Traffic 2010 as 109 vehicle miles | 2010 (percentage) | 55.0 (20.8%) | 28.7 (10.9%) | 85.3 (32.3%) | 95.3 (36.1%) | 264.3 (100%) |
All reported PIAs on roads in England* | Total (percentage) | 5,940 (4.3%) | 5,578 (4.1%) | 57,194 (41.7%) | 68,551 (49.9%) | 137,263 (100%) |
(*Numbers of personal injury road accidents (PIAs) and road traffic based on DfT, 2011)
2.18 There is clear evidence that both frequency (e.g. Kloeden et al., 1997; Gunter et al, 1998; Bertwistle, 1999; Aarts and van Schagen, 2006; Meisinget et al., 2014) and severity of collisions (Gunter et al., 1998, Ahmed et al., 2021) are, in addition, related to vehicle speed.
2.19 Gunter et al., for example, showed that in Yellowstone National Park, where different sections of the road network are posted at a permitted maximum speed of 15, 25, 35, 40, 45 or 55 mph, 85% of all collisions involving wapiti (Cervus canadensis) and mule deer (Odocoileus hemionus) occurred on those road sections with a posted speed limit of 45 or 55 mph, despite the fact that these represented only a small proportion of the overall road network. Indeed, 41% of collisions occurred in roadway segments with a posted speed limit of 55 mph but these segments represented just 8% of the roadway within Yellowstone National Park (Gunter et al., 1998).
2.20 Meisingset et al. (2014) used the locations of 271 collisions between red deer and vehicles in Norway to model the relationship between collision risk, road type, terrain and posted speed limit. They showed that with a change in speed limit from 50 km per hour (approximately 30 mph) to 60 to 70 km/hr (40 to 45 mph) and from 50 to 80 km/hr (30 to 50 mph), relative risk increased by a factor of 3.9 and 8.6, respectively, although the increase in relative risk from 50 to 80 km/hr was lower for minor rural roads than for major roads. When analysis was restricted to sections of road known to include locations regularly used as crossing places by GPS-tagged deer, an increase in speed limit from 50 km/hr to 60 to 70 km/hr increased the relative risk by a factor of 6.8, whereas from 50 to 80 km/hr gave an increased risk of 14.4 (Meisingset et al., 2014).
2.21 Savolainen and Ghosh (2008) employed a multinomial logit model to investigate the level of severity of driver injuries resulting from deer vehicle collisions using crash data from Michigan. The likelihood of human fatality was found to be greater in roadways with a speed limit over 55 mph, as compared to roadways with lower speed limits. Their analysis also revealed that younger drivers are more likely to sustain injury compared to older drivers, which is attributed to the lack of experience of young drivers in situations when they encounter deer on the roadway. Female drivers were found to be more prone to sustaining injuries, as compared to male drivers, and use of seatbelts was found to reduce the overall likelihood of any injury.
Other environmental and landscape factors
2.22 As noted above, certain environmental features consistently emerge as characteristic of sites likely to suffer a high frequency of deer-related road traffic collisions. Thus, in addition to road type (number of lanes of traffic/width of road), traffic volume and speed and presence or absence of a central barrier, the following features are typical: woodland or forest cover close to the carriageway; landscape diversity (variability and patch size); and the presence of obvious travel corridors across the roadway, such as rivers, dry gullies or other linear structures leading down at an angle to, or perpendicular to the roadway.
2.23 The original analyses of Bashore et al. (1985) considered a number of environmental and ‘traffic-flow’ characteristics associated with the high recorded frequency of DVCs on stretches of two-lane highway in Pennsylvania between July 1979 and October 1980, concluding that the predicted probability of collisions decreases with an increasing number of homes, commercial, and other buildings within the buffer area, and longer sight distance along the roadway. Their model also indicated a decrease in the “high” DVC probability with increases in the proportion of fencing, the distance to woodlands, the ability to see a roadside object (i.e., in- line visibility), non-wooded herbs in the buffer zone, and posted speed limit.
2.24 In a subsequent GIS analysis Finder et al. (1999) measured topographical and habitat-related features within a 0.8 km radius of road segments in Illinois with higher than average collision rates and a series of randomly selected control sites. Once again, high collision rates for white-tailed deer (Odocoileus virginianus) were associated with woodland cover; a logistic regression model developed using only landscape features derived from satellite imaging accurately distinguished between high and low kill sites and related collision frequency to landscape diversity and (shorter) distance from adjacent woodland cover. Hubbard et al. (2002) published similar findings from a multiple regression analysis of land use variables and highway characteristics in Iowa, identifying four landscape features associated with clusters of high collision frequency as the proportional area of woodland and grass adjacent to the roadway, proportion of crop land, and the heterogeneity in size and disposition of land cover patches.
2.25 It is clear from these studies that the risk of colliding with deer is higher in a fragmented landscape, and especially where the forest edge is very long (e.g. Romin and Bissonette, 1996; Finder et al., 1999; Madsen et al., 2003). As reviewed by Staines et al. (2001), one of few consistent findings from analyses of collisions involving deer from both Europe and America is that the majority of DVC hotspots occur within or near wooded areas, particularly where the woodland comes right down to the road edge (e.g. Ückermann, 1964; Bashore et al., 1985; Romin and Bissonette, 1996; Finder et al., 1999). More recent attempts at multivariate analysis of the relative importance of the whole suite of factors that will affect DVC frequency also identify proximity of woodland as a key factor often associated with higher collision rates (Hubbard et al., 2002; Nielsen et al., 2003; Malo et al., 2004; Pokorny, 2006; Hussain et al., 2007; Grovenburg et al., 2008; Hegland and Hamre, 2016).
2.26 While many of the studies discussed above are derived from analyses of deer-vehicle incidents in North America, findings in a more specifically European context are not dissimilar. In an analysis of 115 kills of roe deer at Kalo in Denmark, between 1956 and 1985, Madsen et al. (2002) found no correlations between the pattern of road kills and mean daily traffic flows but noted that collision sites were strongly clumped, and sites associated with higher road-kill tended to have denser vegetation (hedgerows, bushes etc.) present on one or both sides of the road. Malo et al. (2004), based on analysis of the locations of 2067 deer-vehicle collisions occurring between 1988 and 2001 in the province of Soria (central Spain), once more identified the features characteristically associated with locations of high collision frequency as vegetation, fencing or other structures forcing the animals to cross at particular points and natural linear features perpendicular to the roadway associated with natural travel corridors.
2.27 Similar conclusions are reached by other commentators: Seiler (2004); Meisingset et al. (2014) Hegland and Hamre (2016) inter alia.
3. Options for mitigation
3.0.1 With our understanding of the factors associated with collision frequency or risk, what are the options available for mitigation to reduce overall frequency or risk of such collisions?
3.0.2 The ideal goal for measures designed to reduce the risk of DVCs is not that they should seek to prevent deer from crossing a roadway, but that crossing should be effected more safely. Attempts to prevent crossings altogether for long lengths of roadway are likely to prove ineffective and result in animals forcing barriers, potentially injuring themselves in the process and, perhaps more significantly, increasing the risk that they may then become trapped within the carriageway, actually increasing rather than decreasing the risk of a collision. At the very least, where barriers are completely effective at preventing movement, this will result in fragmentation and isolation of previously continuous populations of deer and other larger wildlife (Forman et al., 1997; Forman and Alexander, 1998; Mladenhoff et al., 1999; Huisjer et al., 2021). Thus, the most successful measures will seek not to prevent crossings altogether but to displace these in space or time such that deer cross the road at periods of reduced (or zero) traffic flow, or in places where accident risk is reduced through enhanced visibility and/or driver awareness or provision of traffic free wildlife passages. Alternative measures seek to increase awareness of drivers to general or more specific risks of collisions with deer in the carriageway.
3.0.3 It is important to recognise that some of the different measures we consider below are relevant at different levels of engagement. Thus some may fall more within the remit of road developers/contractors in new construction schemes, some may be of more relevance to road operators (Trunk Road Agencies or Local Authorities) while other measures again (e.g. in-vehicle detection systems, paragraph 3.3.13 – 3.3.17) are more a matter of choice by individual drivers. Further, from the outset we would emphasise, as noted in our earlier review (Putman et al., 2004) that in our opinion few, if any, available mitigation measures are effective when used in isolation, and become truly effective only in combination. This will be a recurring theme throughout the following sections.
3.0.4 By way of preamble, it is also important to note that estimates of effectiveness of the different measures which may be deployed are inevitably controversial. It is in fact extremely difficult to derive any objective measure of effectiveness for many of the deterrents and other mitigation measures in current use to help reduce collisions with deer, especially where this must be based primarily on analyses of changes in collision frequency before and after installation. These are often the only data available on which to base an assessment of the effectiveness of some measure taken to try and reduce collision frequency, but even when more formal statistics are applied to such Before-and-After comparisons (as for example Benten et al., 2018) we should caution:
- There is in any case a great deal of variation in collision frequency between years – simple, stochastic year-on-year variation in accident rates (see for example Voss, 2007), such that any changes recorded before and after installation of some mitigation measure cannot necessarily be attributed unequivocally to the deterrent measure installed; therefore, an adequate number of control (unprotected) road sections should be selected, and ideally those where deterrents are installed should be chosen at random from among all road sections considered.
- Collision frequencies on given (monitored) stretches of road are in any case likely to be relatively few in number (often between zero and five or perhaps at best zero to 10 in most studies), unless trials monitor sections of road over many miles or consider numerous directly comparable replicates. This very restricted range of candidate values compounds the difficulty in determining any statistically valid difference between periods before and after the installation of any attempt at mitigation of collision frequency, simply because such differences will be proportionally extremely low (see also: Lehnert and Bissonette, 1997; Danielson and Hubbard, 1998).
- Numbers of deer killed along the same section of road in consecutive years cannot be assumed to be statistically independent with, for example, years of relatively high DVC mortality potentially reducing overall risk of DVCs in the next if all other mortality factors remain identical.
3.0.5 Records of DVCs are also often obtained retrospectively for past years once a trial commences, which can lead to differing level of accuracy between control sites and sites already identified as ‘problem’ areas requiring mitigation. Furthermore, few studies continue to be monitored for a sufficient number of years after installation due to financial constraints; trials monitored for just one or two years make it difficult to be certain whether any reduction in DVCs found truly relates to the specific type of deterrent or other measure installed, or simply ‘novelty- avoidance’ of any form of change introduced along the trial sections. Even when longer monitoring is possible, there is often an unintentional tendency to discontinue maintenance and monitoring first on those roads where least success has been reported (partly due to difficulty in justifying costs to local authorities in the absence of positive results), thus potentially skewing results towards those parts where more reduction was noted (even if not necessarily resulting from the deterrents).
3.0.6 Thus, where estimates of effectiveness are based on changes in observed frequency of reported DVCs, even when changes Before and After mitigation are related to changes in collision frequency in control (untreated) sections of roadway (BACI analysis: Before and After Control Impact studies), conclusions about the actual effectiveness of mitigation, especially if based on low numbers of incidents and over short duration may still be regarded as ‘insecure’.
3.0.7 In our 2004 review we suggested that the options available to prevent or minimise DVCs might broadly be considered as: attempts to prevent or control crossing, e.g. by the use of highway fencing, roadside wildlife warning devices, reductions in local deer population density, chemical deterrents, or the fitting of warning whistles to vehicles; increase driver awareness, e.g. by use of various driver warning systems – whether through fixed signage, or signage responsive to driver speed, or the presence of deer on the roadside, or removal of roadside vegetation to increase visibility; and provide safer crossing places for deer, e.g. by the installation of dedicated bridges or underpasses, by modification of existing passages to dual use, or by the creation of designated ‘cross-walks’ across the carriageway itself.
3.0.8 Here, we modify this approach slightly to present in the following pages an evaluation of different measures and their effectiveness as:
3.1 Physical measures separating deer from road traffic
3.2 Other measures aimed at affecting or deterring road crossings by deer
3.3 Measures directed at influencing the behaviour of drivers
3.4 Manipulation of deer habitats and deer density adjacent to roads.
3.1 Physical measures separating deer from road traffic
Roadside fencing
3.1.1 Our continuing review produced no literature to change our conclusion (Staines et al., 2001; Putman et al., 2004) that high tensile roadside fencing is likely to remain the primary method used to try and reduce road-crossings and resultant collisions at identified sites of high risk on major roads. Consultation with colleagues in other European countries confirms that this remains the main measure deployed at least on major roadways. However, as noted by Staines et al. and Putman et al. (2004), we reiterate that it is essential that fencing should: “be of adequate specification (height/mesh size) and be designed not with the expectation, or aim, of attempting to prevent road-crossings altogether, but rather to channel animals towards a safer crossing point”.
3.1.2 Suitable specifications for the height and construction of such fencing in relation to differing deer species are offered in the EU handbook Wildlife and Traffic: A European Handbook for Identifying Conflicts and Designing Solutions (European Commission Action 341; Iuell et al., 2003, update 2022 by Rosell et al.), together with a number of helpful illustrations, and are thus not rehearsed in detail here. We merely note that we ourselves do not necessarily agree with the specifications offered in relation to mesh size, suggesting instead that at the base of fences this should not exceed 75 mm x 75 mm in areas where muntjac occur, with mesh no larger than 100 mm x 50 mm in roe deer areas to prevent smaller individuals squeezing through. We would also urge that such fences are sufficiently well anchored to, or below the ground, otherwise badgers and other wildlife may push beneath, opening opportunities for muntjac and roe deer in particular to access the roadway. Securing the bottom of fences adequately is particularly important in areas where wild boar are known to occur or are likely to do so in the future (Rosell et al., 2022) and we would at this point highlight that populations of wild boar or feral pigs are increasing in numbers in many parts of Scotland.
3.1.3 Staines et al. (2001) have emphasised that where deer fencing has not proved effective this has usually been related to inadequate specification of fence construction, to deer getting past the end of fence-lines where insufficient length has been installed (Reed et al., 1979; Ward, 1982; Clevenger et al., 2001, 2002), or at road junctions where fencing is difficult. We would add to this an emphasis on the need for regular monitoring and prompt and proper repair to any such highway fencing which has been breached. By way of illustration, when contracted to inspect a section of a newly constructed dual carriageway in Essex two years after fencing had been installed as mitigation, one of the authors found >10 areas where deer not only could but were getting onto the carriageway side beneath access gates without sills and at breaks in fencing, caused by errant cars, repaired temporarily for many weeks using security (Heras) panels, which most fallow and smaller deer species can bypass either beneath or at upright joints between panels. In all such situations, collision risk may actually be increased where deer become trapped in the road corridor on the wrong side of the fence (Feldhamer et al., 1986; Clevenger et al., 2001).
3.1.4 As well as ensuring regular inspection and prompt and proper maintenance of fences once installed, it is also appropriate in any fencing scheme to incorporate means of exit for deer which may in such ways become trapped within a fenced section of carriageway, such as one-way gates (Reed et al., 1974; Lehnert and Bissonette, 1997) or earthen escape ramps (deer leaps) (see review by Mastro, 2008; Rosell et al, 2022). The Norwegian handbook on traffic safety (Elvik et al., 2009) recommends assuming that fences alone do not increase local traffic safety by more than 55%. However, if fences are combined with jump-outs or exit ramps and with sufficient appropriate wildlife passages, the efficacy of such an inclusive fence system will be increased to over 80%.
3.1.5 Bissonette and Hammer (2000) found white-tailed deer used ramps eight to 11 times more often than one-way gates and we may note that deer-leaps have been installed as exits from a number of sections of railway track in the Scottish Highlands and have proved effective in reducing the frequency of collisions with trains (see Deer leaps to avoid train danger). Once again, detailed specifications are provided in Iuell et al. (2003), updated by Rosell et al, (2022).
3.1.6 Few studies have directly assessed the efficacy of electric fencing to reduce DVCs, although at least in the case of moose (Alces americanus), Leblond et al. (2007) reported significant reductions in numbers of moose tracks leading onto road verges and in numbers of collisions after installation of electric fencing.
3.1.7 Complete barrier fencing attempting to prevent road-crossings altogether is likely to prove ineffective in the longer term, since in the absence of any alternative route animals are eventually likely to force the fence to cross roadways (with the added risk that they may then become trapped within the carriageway, unable to escape). At the very least, where it is effective as a total barrier to movement such fencing causes fragmentation and isolation of previously continuous populations of deer and other larger wildlife (see Iuell et al., 2003; Huisjer et al., 2021). Barrier fencing is thus at its most effective when erected in short lengths and in conjunction with the provision of some alternative and safer means of crossing the carriageway, and designed so as to deflect animal movements towards these safer crossing points.
Safer crossings
Overpasses and underpasses
3.1.8 As noted earlier, highway fencing is at its most effective if it seeks not to prevent animals crossing the road, but rather to direct them to safer crossing points. These safer crossing points may be structures specifically constructed as dedicated overpasses or ‘green bridges’, or underpasses beneath main roads, or else existing crossing structures such as viaducts and agricultural access structures modified for use by wildlife. In assessing wildlife use of existing accommodation structures along sections of the A30 and A38-dual trunk roads in Devon, Langbein (2010) found evidence of use by deer in some cases, even without deer fencing present. More importantly though, many other structures showed potential for movement by deer and other wildlife to be enhanced, either by removing current obstructions (such as gates across entrances) or provision of short stretches of fencing to funnel wildlife to existing crossings. Further evidence of the potential for existing structures of modest dimensions to be enhanced for use by deer has been demonstrated at the M25 London orbital motorway, where minor changes to hedgerows and the substrate of existing bridges has led to frequent movement by fallow and muntjac deer through overpasses, tunnels and culverts with widths from just four metres to seven metres (Langbein, 2015).
3.1.9 Green bridges and underpasses are becoming increasingly widely used in continental Europe, foremost to help reduce the fragmentation of habitats and associated wildlife populations caused by expanding road and rail infrastructure, as well as to improve road safety. Costs of such structures are inevitably relatively high, though they are no longer always as expensive as commonly assumed (Voelk et al., 2001), even when fitted retrospectively to existing roads. Although the majority of major structures to date have been of concrete construction subsequently covered with soil and vegetation, more recently lower cost green bridges made of wood or metal construction or of reduced dimensions have also proved successful (e.g. Voelk et al., 2001; ADAC, 2008). Illustrations of a wide range of such structures and designs are provided by Rosell et al. (2022) in the on-line Wildlife and Traffic Handbook.
3.1.10 Specifications and requirements of crossing structures in order to maximise likelihood of use by a wide range and high proportion of wildlife are reviewed in extensive detail in the COST 341 report Wildlife and Traffic: A European Handbook for Identifying Conflicts and Designing Solutions (Iuell et al., 2003, updated by Rosell et al., 2022) [see also Olbrich, 1984; Oord, 1995; Hlavac and Andel, 2002; Giorgii et al., 2007; Bissonette and Adair, 2008], and only a brief overview is attempted here. The online resource (in 3.1.9 above) provided by Rosell et al. also contains many helpful illustrations of different designs for such crossings.
3.1.11 An early study of the use of a large sample of farm and forestry accommodation bridges not necessarily designed primarily or solely for use by wildlife was undertaken by Olbrich (1984) who assessed the extent to which red, roe and fallow deer used 824 different over- and under-passes of differing construction along 823 km of motorways in Germany. Some evidence of use by roe deer was found for 44.7% of all underpasses available but only 22.4% of available overpasses; fallow used 26.3% of underpasses within their distribution and 16.3% of available overpasses; while red deer used only 8.1% of available underpasses and 4.8% of overpasses. Where underpasses are to be provided, Olbrich (1984) suggested minimum height and breadth should be four metres and stressed that the length of the underpass should be as short as possible. More specifically, Olbrich found, for all species, that the ratio of aperture size to overall length was critical to use (as {height x breadth}/length). He predicted that red and fallow deer would be least likely to use underpasses where this ratio is less than 1.5; and that for roe deer the ratio should be at least 0.75 (ibid.).
3.1.12 Giorgii et al. (2007) investigated usage by animals of a range of structures built in Germany more specifically for wildlife during the last two decades, including 20 green bridges, 10 viaducts, seven underpasses for wild animals, six bridges over waterways and footpaths. Usage was investigated both by mapping of tracks and other signs, and through filming with infrared video cameras during March/April on or below a large part of the structures, providing direct evidence of use by animals ranging from red, fallow, and roe deer to wild boar, hares, foxes, racoons, polecats and otters. Green bridges and viaducts indicated the most intensive use (85% of all records), while the narrowest wildlife underpasses, bridges over waterways and footpaths for small mammals were only frequented by a few animals. Using multiple regression analysis Giorgii et al. (2007) showed that use of green bridges was more intensive the wider and older the structures; while a high proportion of wooded vegetation on the structure, noise level peaks, use by humans, and tilling of the land had a negative impact on numbers and types of animals using the crossings.
3.1.13 The wider European COST 341 review on Wildlife and Traffic (Iuell et al., 2003) recommends a width for overpasses of 40 to 50 metres (between the fences) if to be used by red deer and wild boar as well as smaller mammals, suggesting that this width can be lowered to a minimum of 20 metres where the topography has a channelling effect leading the animals directly onto the crossing (Iuell et al., 2003).
3.1.14 Rosell et al., in their 2022 review of the original COST 341 report suggest, from a range of European studies, that for deer and other larger fauna, overpasses /green bridges should have a width of between 50 and 80 metres with width increasing for longer structures to provide an overall ration of width/length of 0.4 – 0.8. For underpasses, they suggest (for example) that for roe deer and boar, they should have a width of between seven and 10 metres and a height of two to four metres. Like Olbrich, they suggest that what they refer to as the Open-ness Index ({height*breadth}/length) is critical to use and for roe deer and boar should be between 0.75 - 1.0. For larger ungulates such as red deer, Rosell et al. suggest that underpasses should be constructed with a width of between 12 and 20 metres, height of three to four metres and Open-ness Index of 1.5-2.0.
3.1.15 We would note here however that many of the recommendations offered by Iuell et al. (2003) and Rosell et al. (2022) primarily relate to minimising habitat fragmentation and are aimed at maintaining “good ecological connectivity“ rather than merely a reduction of deer/wildlife collisions. Thus, it may be argued that the presence of some smaller passages, which those animals determined to cross can use (such as e.g. seasonal movements of male deer into female areas during the rut or emigration from natal ranges) may still suffice for bringing a significant reduction in the frequency of deer vehicle collisions, even if nevertheless still acting as a barrier to free movement of the wider population as a whole.
3.1.16 We would suggest therefore that the above specifications for crossing structures should be somewhat less stringently interpreted where improvements for road safety rather than isolation of wildlife populations is the major objective. While specifications presented by COST 341 should be the goal where the extra costs can be justified on the basis of addressing multiple environmental objectives, more modest specifications may nevertheless still be of value in enabling at least occasional use or use by a small proportion of deer and other ungulates living to either side of major roads. For example, CTGREF (1978) suggested a minimum width of overpasses for ungulates of six metres; Ballon (1985) suggested a minimum width of eight metres and a minimum ratio of width to length of one to10; SETRA (1993) recommended a minimal width of overpasses for red deer as 12 metres, for roe deer as seven metres; and Langbein (2006) found occasional use by fallow deer of both a non-vegetated concrete overpass less than four metres wide, and another less than six metres wide with limited soil cover across the very heavily trafficked M25 orbital route around London. Following some minor improvements including better hedgerow cover leading to the entrances, much more extensive and frequent use by whole herds of deer was filmed here within five years using CCTV (see Langbein, 2015).
3.1.17 Whichever form of passage is provided it is clear that it takes a period of time for deer to become used to such corridor structures and to use them freely. Reed et al. (1975), Ward (1982), Olbrich (1984), Giorgii et al. (2007) and Ollson et al. (2008) all noted an initial reluctance by deer to use new underpasses until these have ‘mellowed’ or matured, although use of new bridges by roe and fallow deer has in some cases been observed within a few months of completion (Giorgii and Wotschikowski, 2007; authors’ own observations).
Overpasses versus underpasses?
3.1.18 Iuell et al. (2003) noted that where overpasses and (admittedly rather small) underpasses were available close to each other, moose and deer (Odocoileus) preferred to use the overpasses (a conclusion derived from Clevenger et al., 2002; and supported also by Giorgii et al., 2007). Note however that in the survey of Olbrich (1984) use of overpasses (i.e. rather than wide green bridges) by red, roe and fallow deer was lower than that of underpasses. Small data sets hampered detailed analysis of the factors affecting use, but overall breadth again seemed the critical consideration. There are few general rules as to when one is more suitable than the other. The choice is partly determined by the topography. In hilly terrain it is often easy to construct both over- and underpasses, whereas in flat landscapes underpasses may be easier to construct, if the ground water level is not too high.
3.1.19 Overpasses have the advantage that it is easier to provide different microhabitats on them, because vegetation grows more easily than in underpasses. A wider range of species may therefore use them. However, viaducts, which generally retain quite high and wide original pathways beneath the road, can provide equally good results to landscape bridges (e.g. ECONAT, 1992).
How many passages?
3.1.20 Relatively little work has been undertaken to date on how many overpasses and/or underpasses may be needed per km of road length to be effective in reducing collisions with vehicles. However, for red deer and roe deer Hlavac and Andel (2002) recommend that in areas of high importance passages should be provided at two to four kilometre intervals where roe deer are the target species of mitigation, or at five to eight kilometre spacing for red deer.
Highway cross-walks
3.1.21 Feasibility and cost of provision of under- or overpasses for existing roads will depend largely on the local topography, such that, while retro-fitting of a land bridge may perhaps be most readily achievable in undulating landscapes where a road may already run through a cutting, provision of (and landscaping) a similar overpass or tunnel/underpass would be likely to present much greater engineering challenges and higher costs where the existing road runs through a level landscape. Where bridges or accommodation tunnels may not be considered appropriate for existing roadways, an alternative approach is to attempt to provide for safer crossing zones by creating cross-walks across the carriageway surface itself (Lehnert and Bissonette, 1997). In essence this concept builds upon ideas developed earlier, of using fences or other barriers to guide animals to safer, and well-advertised, crossing places. However, rather than provided as underpasses or overpasses, cross-walks aim to provide relatively safe zones through provision of high impact signage, speed limits and modifications to the road surface (rumble strips / speed humps) targeted on short sections of road.
3.1.22 Road surfaces at the ‘cross-walk’ itself may also be modified to encourage use by wildlife, with crossings funnelled (and made more predictable) by fencing the roadside in areas where visibility is poor, and permitting crossing of the carriageway only in a limited number of unfenced stretches of roadway where deer and driver visibility has been improved. Lehnert and Bissonette (1997) tested the efficacy of such cross-walks on two-lane and divided four-lane highways in north-eastern Utah; based on expected kill levels, mortality of mule deer declined by 42.3% and 36.8% along the four-lane and two-lane highways, respectively. Lack of motorist response to warning signs, the tendency for foraging deer to wander from crosswalk boundaries into the carriageway itself, and the ineffectiveness of highway one-way gates in permitting their subsequent escape were considered to contribute most to remaining mortalities within the treatment area, and emphasise that careful design, implementation and monitoring of such crossings is critical.
3.1.23 We would propose that consideration be given to the installation of cattle-grids of appropriate specification across the road at such road crossing areas, and also when possible near fence end-runs more generally, to prevent animals spilling back onto the carriageway itself. In the US, cattle grids (referred to there as wildlife grids or guards) have been developed and tested with promising results, achieving efficiencies of 50 - 89% in ungulates (Allen et al., 2013; Flower, 2016). If used with cross-walks such grids would in effect ‘link’ the fence-lines of opposite sides of the carriageway, providing a close-circuit barrier on each side of the cross-walk. Installation of cattle grids on either side of such crossing zones could have the further incidental advantage of further reducing traffic speed in these targeted crossing areas.
3.1.24 Such cross-walks have not been widely used however and we are unaware of any such structures in current use in Scotland. Despite this we believe there may be some merit in exploring this option, as potentially a relatively inexpensive way of providing safer crossing places. It is essential however that any such cross-walks be adequately signed to raise driver awareness and we should note, after Seiler et al. (2016) that: crosswalks are only feasible on single-lane roads with intermediate to low traffic volumes; but could be considered on entry or exit slip roads on more major roads if adequately signed. Motorways or highways with multiple lanes and central barriers must be mitigated by bridges and tunnels that physically separate animals and vehicles.
3.2 Other measures aimed at affecting or deterring road crossings by deer
3.2.1 Roadside wildlife deterrents of many different types that seek either to prevent deer or other large animals from crossing road sections, or else to control when and where they do so have been deployed for over 60 years. Most roadside deer deterrents emit visual, auditory, or olfactory signals (used separately or sometimes in combination) in an attempt to alert animals to road traffic and in the hope that they may learn to associate the warning signals with motorised vehicles. Despite their use over many decades and in numerous different countries and situations, proof of effectiveness of wildlife deterrents in terms of collision reduction, or even in influencing deer behaviour in the desired manner, remains lacking for the great majority of devices developed to date. Indeed the greater majority of studies tend to confirm that such deterrents have limited effect.
3.2.2 A number of factors contribute to the general lack of good evidence of effectiveness for most roadside wildlife deterrents that are currently available, including:
- Deterrent signals based often on (anthropomorphic) assumptions that they will be perceived by deer (or wild boar) as warning signals and elicit the desired response, rather than on detailed prior study of how deer react to differing signals.
- Study designs with insufficient replication, controls, and duration of trials to ascertain clearly whether any changes in DVC frequency observed are truly attributable to the presence of the deterrent rather than changes in other external factors.
- A tendency for deterrent designs to be manufacturer led with limited input from behavioural ecologists.
- Failure to presume from the outset that deer and other animals will habituate to any signals that they encounter repeatedly when there are no associated negative consequences, and instead accept unfounded presumptions that deer will learn to associate the signals with danger.
- The increased extent to which deer in many countries have become accustomed to living closer to human habitation and infrastructure and its associated noise, light, and olfactory stimuli.
3.2.3 An example video of wild fallow grazing alongside heavy road traffic filmed across differing seasons illustrates that deer can become habituated to the lights and sound from road vehicles even along major dual carriageways. Nevertheless, in view of the very widespread use of roadside deterrents, and continuing development of new devices, a summary of the main types of deterrents available and research on their effectiveness is provided below.
Roadside wildlife warning reflectors
3.2.4 With the same considerations as above, roadside reflectors are designed not to stop animal movement across roads, but to delay these at times when there is traffic in the carriageway until the roadway is clear. Working on the principle that light from approaching headlights is reflected onto the verge to provide a flash warning or continuous wall of light (depending on reflector type and deployment), they are intended to alert deer to approaching traffic at night or startle them to delay crossing until the road is clear.
3.2.5 Even if such optical reflectors are effective at all in delaying deer crossings, by definition they can at best only be effective at night. Research into the daily pattern of accident frequency, suggests in fact that highest periods of accident risk coincide with dawn and dusk (paragraph 2.6 and Figure 1), when approaching vehicles may not in practice have their full headlights on, and when effectiveness of the reflection is to an extent reduced by higher general ambient light conditions.
3.2.6 Furthermore, even during darkness when more likely to be effective, such reflectors can only usefully be installed on roads of relatively low, or sporadic traffic flow: such that there are periods of quiet between vehicles to permit safe crossing (Putman et al., 2004). On roads of high or continuous night-time traffic, where reflectors are continuously activated, deer may be expected to more readily habituate to the reflected light. Even if they do not habituate to the reflected light or traffic in general, if intent on crossing - but not provided with interval periods of no traffic flow - deer are likely simply to force the ‘barrier’ and cross anyway even in the face of oncoming traffic. Despite these basic limitations there have been numerous instances of quite inappropriate deployment of wildlife warning reflectors, including on major motorways in the UK and main roads elsewhere of very high traffic flow.
3.2.7 There has been considerable discussion in the literature over the effectiveness of such reflectors (immediately after installation). From some studies it appears that when properly installed they may have some effectiveness in reducing the incidence of DVCs for a period after installation (e.g. Woodard et al., 1973; Gilbert, 1982; Gladfelter et al., 1982; Schafer and Penland, 1985; Zacks, 1986; Waring et al., 1991; Armstrong, 1992; Ford and Villa, 1993; Reeve and Anderson, 1993; Pafko and Kovach, 1996; Pepper et al., 1998; Jared, 1999; D’Angelo et al., 2006; Voss, 2007). Other studies report limited effectiveness or note that any initial effect wanes over time, either due to deterioration of the reflectors themselves, lack of clearance of vegetation around the installation or due to habituation/learning (e.g. Ujvari et al., 1998).
3.2.8 The majority of these studies are based on analysis of changes in DVC frequency through time. As noted earlier (paragraphs 3.0.4 – 3.0.6) such analyses are inevitably difficult and inevitably inconclusive, given the fact that there is in any case considerable stochastic variation in incident frequency along any given stretch of roadway (Putman et al., 2004; Voss, 2007) and there are, in addition, often a number of confounding variables (Benten et al., 2018a). Recent commentators have noted that simple comparisons of DVCs before and after installation of reflectors are not appropriate for robust analysis of effectiveness and studies should adopt a more formal before-after control-impact (BACI) approach or a cross-over study design with spatial and temporal control sections.
3.2.9 In a detailed meta-analysis of 76 published studies Benten et al (2018a, b) could not find convincing evidence for the effectiveness of wildlife warning reflectors in reducing rates of wildlife collisions with vehicles, concluding that only studies of <12 months effective study duration and <5 km test site length suggested that warning reflectors might have had any effect in reducing DVCs (Benten et al., 2018a, b). Very few studies specifically investigate the effect of reflectors (or other roadside devices) directly on deer behaviour.
3.2.10 One of the few studies which has quantified observed behavioural effects of light reflecting devices on deer (Benten et al. 2019), studied using IR video at 13 different sites with such reflectors, found ungulates were more likely to leave the road side area with reflectors present. However, this effect only lasted 16.5 days and did not influence the risk of a collision with a vehicle. Similarly, Brieger et al. 2017, on studying the behaviour of roe deer where blue semicircle reflectors had been installed found no evidence that these elicited reactions that seem suitable to reduce roe deer deer vehicle collisions. Indeed from one such study, D’Angelo et al. (2006) suggest that installation of such reflectors may even prove counter-productive. They studied the extent to which white-tailed deer altered their behaviour in relation to red, white or amber wildlife reflectors at the roadside, concluding not only that none were effective at reducing DVCs, but based on infrared video recordings suggested that deer may in fact be more likely to be spooked and dart across the road where light reflectors were present.
3.2.11 From all such analyses we must conclude (as also Brieger et al. 2016, 2017; Benten et al., 2018b, 2019; Huisjer et al., 2021; Rosell et al., 2022) that despite the fact that this appears a relatively inexpensive solution - and has thus been comparatively widely adopted by roads authorities in the past - wildlife warning reflectors (of whatever design) have no proven effectiveness in reducing the rate of DVCs.
Acoustic roadside deterrents
3.2.12 In order to attempt to overcome some of the drawbacks of optical-only reflectors outlined above, and to reduce likelihood of habituation to individual deterrent signals, a number of more sophisticated roadside deterrents have been developed over recent years. These generally either combine traditional white or red optical reflectors with acoustic modules to produce whistling sounds (e.g. WEGU-GFT acoustic wildlife warning reflector, EUROCONTOR acoustic reflector), or provide a series of signals at differing frequencies ranging from infrasound through to ultrasound (e.g. EUROCONTOR Ecopillars) based on the assumption that deer may habituate less readily to a mixture of signals. Another type (e.g. DeerDeter, Schalk et. al 2023) produces a series of intermittent whistling sounds readily audible to the human ear and incorporates also a small light emitting alternate blue and yellow flashes. Such devices generally employ solar cells to charge integral batteries during daylight hours, with the ‘deterrent’ signals emitted either at intervals throughout 24hrs a day, or more usually only when triggered by bright car headlights falling on the devices at night. As has often been the case regarding optical reflectors and other wildlife deterrents in the past, various preliminary findings reported in hunting press or other general media claim good results with such devices in terms of DVC reduction, mostly during the first one or two years after installation. However, firm evidence for lasting effects remains lacking in the published scientific literature and results of trials undertaken in differing countries or situations – like those for light reflectors – remain contradictory.
3.2.13 Some more positive initial results were reported in Slovenia, where the effectiveness of three different mitigation measures (ultrasound emitting devices; acoustic reflectors; combination of classic and acoustic reflectors) installed along 23 different road sections (totalling 22 km) was tested between July – December 2006 (Pokorny and Poličnik, 2008). In total, the number of road-killed deer (primarily roe) decreased in the first six months after the installation by 106 individuals (83%) in comparison either with the equivalent period in the year before the trial or with the average value for the equivalent periods in years 2002 – 2005. However, the change in DVCs calculated for an equivalent number of control sections averaged a fall of only 16% (in comparison with the equivalent period in 2005) or 3% (in comparison with average values in the period 2002 - 2005) and was statistically insignificant. Results of continued monitoring over the following 18 months for a reduced set of 16 of the original road sections from January 2007 to June 2008 showed a less pronounced but still statistically significant reduction in the number of road-killed deer (by 47% and 62% in 2007 and first six months of 2008, respectively; Pokorny and Poličnik, 2008).
3.2.14 By contrast, a longer duration study undertaken by the German Insurance Association for Accident Research (Voss, 2007) in the Oberbergische region of Germany to establish the effects of five differing road side measures, including acoustic wildlife reflectors, found no evidence for statistically significant reductions in DVCs from acoustic reflectors or any of the other methods, when compared against a large sample of control sites for which directly comparable before and after data were collected.
3.2.15 In order to investigate more directly the true effect of acoustic deterrents on deer behaviour when crossing roads (and thus circumvent the influence of confounding factors such as variation in cull and population numbers on changes in DVCs between years), Langbein employed periods of remote 24hr day-night roadside video surveillance over an 18-month period (Langbein 2007b; Langbein et al., 2011) along sections of road in two regions of England, one offering a high density of wild fallow and another with mainly red deer, where acoustic deterrents had been installed to establish whether deer delayed crossing in areas with activated deterrents, by comparison to behaviour recorded at control sections with no deterrent present.
3.2.16 For fallow deer crossing the road at night, the median delay after the last vehicle passed before the animals entered the road way lay between 20 and 30 seconds both in the control sections as well as sections fitted with either of the deterrent types. For red deer the median time interval before crossing was 30 to 60 seconds after traffic in sections with activated acoustic reflectors and in control areas. A significant reduction in deer collisions was recorded during the two years post-installation of acoustic reflectors at the red deer site but was of similar magnitude in both control and deterrent sections (Langbein, 2007b), and coincided with a major increase in local deer culls. In further observations of the direct response by wild roe, fallow and red deer to the ultrasonic and low frequency audible signals emitted by EUROCONTOR Ecopillars (set to emit their various acoustic signals every 15 to 30 seconds throughout the day and night) video observations were undertaken for several days before and after switching on the devices at bait stations set up in different areas for each species. In all three species, animals were found to approach and feed calmly close (<1 to 5 m) to the devices within just one day (fallow) to three days (red and roe) after activation, suggesting that habituation to the acoustic signals emitted by this particular device was extremely fast.
3.2.17 A more recent five-year study in Austria, WiConNet, has been assessing both passive deterrents (light reflectors) and active deterrents (with sound and light signals) across a range of trial sites on non-motorway main roads, as well as railways lines in Austria. In some cases, the active deterrents were deployed on their own, and for others combined with networked ‘gateways’ to enable advance activation of series of DeerDeter roadside deterrents (audio + strobe light). However, although the devices were shown to operate well in a technical sense by the end of the project, the final report (Schalk et al. 2023) reported that data collected at the time remained insufficient to show the efficacy of the wildlife deterring devices at reducing DVCs.
3.2.18 There is some suggestion that acoustic deterrents which use sounds with natural meaning, such as distress calls from local species or human voices, and are activated by the approach of a vehicle, may have greater deterrent effect (Biedenweg et al., 2011; Babińska-Werka et al., 2015; Seiler and Olson, 2017; Berndt, 2021; Lodnert, 2021, summarised by Rosell et al., 2022). However, to our understanding such pillars have largely been tested on railway lines (where the volume of traffic is much lower) rather than on roads, and there is no clear indication of whether or not such deterrents would maintain their effectiveness on roads, where the passage of vehicles may be more or less constant.
Chemical deterrents/Olfactory repellents
3.2.19 Proprietary ‘chemical fences’ or ‘scent fences’ (repellent chemicals encapsulated in slow release organic foam and applied to roadside posts or trees) have been trialled extensively in Germany, with claims by the manufacturers as well as the German Automobile Association (ADAC) of some efficacy in reducing the frequency of deervehicle collisions. From trials on six test sections in Bavaria and northern Westphalia, the manufacturers of a proprietary German scent-fence reported that 60% of the animals encountering the treated areas withdrew and crossed the road beyond the ‘scent fence’ at an untreated section.
3.2.20 Twenty percent of the animals crossed despite the treatment but crossed very rapidly without delay; the remaining 20% were unaffected. On one section of treated road, reported collisions with roe deer fell within a year from 22 per year to a total of two (Kerzel and Kirchberger, 1993). More detailed assessment showed that although road-kills were reduced by 30% to 80% within some test sections, collisions outside of the trial areas actually rose (Lebersorger, 1993), and other, independent, studies have suggested that such scent-fences are not in practice as effective as claimed (Lutz, 1994). Scent-fences were also included as one of the preventative measures tested in the longer term study by the German Insurance Association (Voss, 2007); as in the case of acoustic and optical reflectors, the numbers of recorded DVCs along the section where a scent fence was installed and regularly maintained and renewed for several years, actually showed an increase overall and failed to provide any evidence of effectiveness of such fences.
3.2.21 Similar results were obtained in a short-lasting trial (year 2005 only) in Slovenia, where the efficiency of chemical deterrents was tested on 11 problematic road sections (total length of 13.4 km). Although the number of road-killed deer on these sections decreased by 44% in comparison with comparable periods in the immediately preceding year, or 37% across the four years before the trial, the number of DVCs at adjacent road sections dramatically increased in the same time, resulting in a total decrease rate of only 15% (which, however, did not differ from the changes on control road sections). Therefore, the positive influence of chemical deterrents as a countermeasure against DVCs was not confirmed (Pokorny et al., 2008).
3.2.22 More recently, Bil et al. (2020) undertook a study of the effect of olfactory repellents on roe deer in the Czech Republic, analysing for six radio-collared roe deer, the effect of the compound ‘Pacholek’ (produced by the company Ekoplant) both on presence near roads and actual crossing frequency over a period of five months (April–August 2019). The odour repellents were installed along two secondary roads, and along two semi-open habitats (forest–meadow and forest–arable land). As with earlier studies reported above they could detect no clear effect of odour repellents on roe deer behaviour (Bil et al. 2020). Further work was carried out in 2021 and 2022 on approximately 140 road sections in the Czech Republic (Tomas Kusta, pers. comm.) but it seems clear that there is currently no evidence for the effectiveness of any olfactory repellent (or other chemical repellent) on deer behaviour or the frequency of DVCs on roads where such repellents are deployed. Such treatments are in any case ephemeral (requiring frequent re-application) and even where there is an initial deterrent effect (e.g. Kusta et al., 2015) it is likely that, without reinforcement, animals would readily habituate to odours used.
3.2.23 In their 2020 paper, incidentally, Bil et al. suggest that in our 2004 review (Putman et al., 2004) we stated that olfactory repellents may have an effectiveness of between 30 and 80% in reducing the frequency of DVCs. This is incorrect and a misquote of that review; we would reiterate that now, as then, we do not consider olfactory repellents to have any consistent efficacy.
Car-mounted warning whistles
3.2.24 As noted in our review of 2004, various commercial companies offer for sale a device for attachment directly to the front of a motor vehicle, which emits a high frequency whistle claimed to be a deterrent to deer or other roadside wildlife. In a study of the response to such whistles Romin and Dalton (1992) observed no behavioural responses by mule deer to suggest acknowledgement or avoidance of vehicles equipped with such devices, nor could any reduction in the number of deer vehicle collisions be demonstrated. Unpublished work by scientists from the University of Wisconsin, mentioned in a report by the Insurance Institute for Highways in the US in 1993, found that neither deer nor humans could actually detect the sound produced by the air activated whistles in normal operation, and that whistles blown by mouth had no effect on penned deer (see also DVCIC, 2003).
3.2.25 Any effectiveness of deer whistles and horns is dependent on the ability of deer to hear and respond to the emitted sound. Manufacturers’ claims typically suggest that the air activated whistles emit sounds at between 16 and 20 kHz at speeds above 30 mph Overall, there has not been a significant amount of published work on the auditory capabilities of deer; what work there is suggests that the ‘range of greatest hearing sensitivity’ lies between 1 and 8 kHz – which would be well below the sound range claimed for the various whistle designs in commercial production. In independent experimental trials on six different whistle designs, however, Scheifele et al. (1998; 2003) found that the primary operational frequency actually produced by the different whistle designs was 3.3 kHz and 10 kHz. This is closer to the presumed auditory range of (white-tailed) deer; however, it is noted that the 3.3 kHz sound is also within the typical range of normal roadway noise (tyre noise) produced by a vehicle at 45 mph (Scheifele et al., 2003). In an earlier study of the acoustic characteristics of a range of air-activated wildlife warning whistles, Schober and Sommer (1984) found that the sound they produced was at least in principle audible for red and roe deer and hares, but from a moving vehicle no clear positive effects on these animals could be detected for any of the commercial car mounted devices tested.
3.2.26 More recently, Valitzski et al. (2009) conducted a study using sound emitting devices and speakers mounted to the front and sides of a vehicle to test deer reactions. They conclude that auditory deterrents do not appear to be appropriate for prevention of collisions with deer (Valitzski et al. 2009). We are not aware of any subsequent developments in this area.
3.3 Measures directed at influencing driver behaviour
Public Information campaigns
3.3.1 Over the years various campaigns have been established in an attempt to raise general awareness of drivers to the risk of collisions with deer, particularly at times of year when the risk of collision is highest (paragraphs 2.4, 2.5). Thus targeted campaigns have been mounted on the radio, or via other broadcast media including a variety of social media channels, through newspaper campaigns, in preparation of posters or distribution of physical flyers by organisations such as the British Deer Society, and through use of Variable Messaging Signs (both fixed and mobile) by the roadside in particularly sensitive areas. In practice, the effectiveness of such public information campaigns in reducing DVCs is not known and could potentially be zero (Huisjer et al., 2021).
Standard, seasonal and dynamic warning signs
3.3.2 A common approach in attempting to raise driver awareness on roads where collisions are known to occur with some frequency is the erection of fixed, standardised ‘wildlife/ wild animal warning signs’. Such cautionary signs warning drivers of the likelihood of wildlife adjacent to, in or crossing the road are, however, only likely to be of benefit if erected on approaches to known regular crossing points. In practice, warning signs are relatively rarely precisely targeted but rather tend to warn of increased risk for several miles/kilometres.
3.3.3 Further, it is doubtful whether basic signs are in any case very effective in the long-term, since drivers readily habituate to them unless the message is reinforced by actual experience of deer crossings (Putman, 1997; Hedlund et al., 2004; Stanley et al., 2006). Furthermore, in many European countries the overabundance of differing road signs is increasingly leading to signs being overlooked entirely by drivers.
3.3.4 While drivers may reduce vehicle speed in response to standard and enhanced signs (e.g. Pojar et al.,1975; Al-Ghamdi and Al-Gadhi, 2004; Rogers, 2004; Sullivan et al. 2004), the majority of studies of the effectiveness of these sign types in reducing collisions concluded that they were not effective (e.g. Pojar et al., 1975; Rogers, 2004; Meyer, 2006; Bullock et al., 2011). Some have found standard warning signs to be effective (34% reduction in collisions) immediately after installation at recently identified hotspots (Found and Boyce 2011), but in general such warning signs cannot be considered particularly effective.
3.3.5 Although basic warning signs have generally been found to be ineffective for reducing the number of DVCs, official deer warning signs do provide both deer-population managers and road authorities with a degree of judicial safety (e.g. absolving their responsibility in the case of collisions with game/wildlife species).
Enhanced or dynamic signs
3.3.6 Various types of enhanced signage, temporary signs, dynamic message boards and animal activated warning systems have been developed to increase the likelihood that road users will take note of them (Huijser and McGowen, 2003; Huijser et al., 2006; Mastro et al., 2008; Rosell et al., 2022). Enhanced signs should again be used only for warning of known and regular deer-crossing points along a roadway. Driver habituation might also be reduced if signs were only exposed at particular times or seasons where collisions are known to be more frequent (Iuell et al., 2003).
3.3.7 However, Pojar (1972) and Pojar et al. (1975) found no difference in either vehicle speeds or DVC frequency depending on whether signs were visible or not, or even whether illuminated by permanent or intermittent flashing neon lights; although average speed did decrease by 13 km/h when a deer carcass was present on the highway’s emergency lane. Sullivan et al. (2004) did report some success with temporary enhanced signs erected only during autumn and spring migration of mule deer (Odocoilus hemionus), resulting in a fall in the percentage of speeding vehicles from 19% to 8%, and a decrease in DVCs estimated at 50%.
Roadside animal detection systems
3.3.8 In some cases, dynamic sign systems have been developed further by coupling to sensors capable of detecting animals approaching the roadway. Such signs are thus activated only in direct response to animals present or approaching the carriageway, using a variety of sensors based on heat detection, seismic ground vibrations, or breaking of laser or infrared beams along the verge. The sensors trigger fibre-optic-display enhanced wildlife warning signs and can also be combined with speed detection and display of speed limit signs. Numerous differing versions of such systems have now been installed in Europe and North America.
3.3.9 As in the case of other deterrents, firm data on their effectiveness at reducing DVCs remains scarce (Huijser et al., 2006, 2021), although Mosler-Berger and Romer (2003) reported a fall in DVCs by near 80% for a series of infrared activated systems in Switzerland and other studies (Gagnon et al. (2010, 2019); Strein (2010), MnDOT (2011); Sharafsaleh et al. (2012) ) are cited by Huisjer et al. (2021) as suggesting levels of reduction in vehicle collisions with large mammals from 33% - 97%. Several other studies have demonstrated that drivers do slow down in response to activated systems (Kistler, 1998; Muurinen and Ristola, 1999; Gordon et al., 2003; Kinley et al., 2003 ; Hammond and Wade, 2004; Huijser et al., 2017; Grace et al., 2017), which may thus reduce frequency and severity of collisions.
3.3.10 Interactive signage, instead of being triggered by animals, may also be triggered by approaching traffic travelling above a set speed or display messages intermittently (e.g. on digital message boards often mounted permanently above major routes or on temporary trailers at the side of the road). Hardy et al. (2006) found that temporary messages on such boards warning of wildlife crossings led to lower traffic speeds than either a similar board with traffic information or one not displaying any message; and portable signs seemed more effective than permanent signs.
3.3.11 A variant of this same technology (ANIMOT) is currently being trialled in Switzerland (Suter et al. 2021; Animot, 2019). ANIMOT also employs active animal detection devices at the roadside, but these are used to trigger blinking lights on roadside pillars when an animal is detected within 27 metres of the device, with the blinking signal intended to alert drivers to the presence of animals in the vicinity of the verge. Data loggers within the devices can also store and remotely send information on numbers of activations by animals, and be networked with advance gateway signage when animals are active at the roadside.
3.3.12 It does appear that ‘dynamic signage’ overall may have considerable potential as wildlife vehicle collision mitigation – although it remains unclear what may be the actual cost-effectiveness of animal-sensitive devices versus cheaper and often more reliable devices that are triggered by vehicle speed simply to enhance warnings against inappropriate speed at known DVC ‘hotspots’. A fuller review of differing types of dynamic or animal-detection and driver-warning systems deployed in Europe and North America is offered by Huijser et al. (2006, 2021) and Rosell et al. (2022).
In-vehicle warning systems
3.3.13 At the time of our 2004 review, two in-vehicle ‘vision systems’ had been recently developed, designed to enhance driver detection of deer by the roadside, particularly at night. Both used infra-red sensors to offer earlier detection of deer or other wildlife either on, or beside, the carriageway, displaying images continuously on a screen within the dashboard. A number of similar systems have been developed subsequently, but there are no published studies which evaluate the usefulness or the effectiveness of these technologies in reducing the risk of collisions with deer or other larger animals (review in Huisjer et al., 2021).
3.3.14 There are now several such systems available commercially for vehicles in the United States that focus on large animals (Huisjer et al, 2021). A Swedish company has developed a vehicle-mounted night vision detection system used by several car manufacturers, including Audi, BMW, Mercedes and Daimler (Forslund and Bjarkefur 2014). The infrared-based system is reported to have a very low level of false positives - one such event per year (Forslund and Bjarkefur, 2014). Volvo has developed a radar-based system to identify large animals, so that it can operate effectively both day and night (Adams, 2017).
3.3.15 In addition to sensors integrated into vehicles, high risk road sections and high-risk times may also be integrated into satellite navigation systems, to provide alarms when a driver enters road sections with a known high risk of DVCs. Potential high-risk areas may be identified from information on crash data and deer casualty data collected by road maintenance personnel, insurance companies or by the general public through citizen science initiatives collating observations of dead or live animals on or near the road. However, to be effective such systems rely on central collation of DVC data (something not yet undertaken by most European countries; Apollonio et al., 2010) and development of reliable risk maps which are constantly updated (e.g. Nelli et al., 2018).
3.3.16 As noted above, most of the published articles on this topic are restricted to evaluating the technological capabilities and the reliability of the sensors and in-car warning systems and we have been unable to find studies that actually evaluate the effectiveness in reducing DVCs. There is also some concern that over time, with any wide reliance on such systems, drivers encountering serious collisions in areas not flagged up as high-risk, or where on-board sensors have failed to detect animals by the roadside, may see an opportunity to seek claims for damages against development companies or car manufacturers.
3.3.17 There would however seem to be some potential in linking in-vehicle detection systems in real time to Roadside Automatic Detection Systems (RADS; paragraph 3.3.8) detecting animals in or approaching the carriageway and alerting drivers of an impending hazard, and there is currently some considerable commercial interest in development of such systems.
Reducing driver speed
3.3.18 There is clear evidence that both frequency (e.g. Gunter et al., 1998; Bertwistle, 1999; Aarts and van Schagen, 2006; Meisinget et al., 2014) and severity of collisions (Savolainen and Ghosh, 2008, Ahmed et al., 2021) are indeed related to driver speed (paragraph 2.17).
3.3.19 Despite this relationship between vehicle speed and collision risk/severity, permanent reductions in the posted speed limit are rarely effective in actually reducing effective travel speeds (although very short-term restrictions imposed via dynamic signage in response to actual presence of deer by the roadside may have greater impact). As noted by Huisjer et al. (2021): Most drivers drive a speed (operating speed) that is close to or higher than the design speed of a rural road (Fitzpatrick et al., 2003; Jiang et al., 2016; Donnell et al., 2018). If the posted speed limit is substantially reduced below the design speed for a road section through a sensitive area, and if the design speed remains the same for this road section, most drivers will ignore the lower posted speed limit and continue to drive a speed close to or higher than the design speed of the highway. In their study in Yellowstone National Park, for example, Gunter et al. reported that average operating speeds measured along the roadway segments with a 55 mph posted speed limit were about nine to 16 mph higher than that posted (Gunter et al., 1998).
3.3.20 Where speed reductions are imposed, some drivers will respond to the restriction and adhere to the lowered posted speed limit. This will result in a mix of fast and slow-moving vehicles and this is associated with more interaction between vehicles, dangerous driving behaviour (e.g. irresponsible manoeuvres to overtake slow vehicles) and an overall increase in crashes between vehicles (Huang et al., 2013; Elvik, 2014). Huisjer et al. (2017) further calculate that to allow (almost) all drivers to stop their vehicle in time to avoid a collision with a deer or other wild animal, especially at night, operating speed may need to be as low as 25 to 30 mph. This is far lower than the design speed of most roads.
3.3.21 Bertwistle (1999) compared the number of vehicle collisions with bighorn sheep and wapiti in Jasper National Park, Canada for eight years before and eight years after the posted speed limit was reduced along three sections of the Yellowhead Highway from 55 mph (90 kph) to 42 mph (70 kph). Results showed for bighorn sheep a slight increase in the number of vehicle collisions, but this was potentially associated with substantial increases in traffic volume and the sheep habituating to vehicles and not leaving the road when vehicles approached.
By comparison, the number of animal collisions with vehicles decreased by 33% (30% before the change to 20% after) along the 55 mph posted speed limit segments adjacent to marked “Slow Down for Wildlife” zones (Bertwistle, 1999).
3.3.22 Data restrictions allowed the evaluation of wapiti-vehicle collisions within only one of the speed reduction segments selected. Effectiveness of the speed reduction was measured by a statistical comparison of the number of vehicle collisions that did occur to the number of expected collisions. The number of expected wapiti-vehicle collisions was calculated from crash data collected within a 13-mile segment of roadway posted at 55 mph (90 kph), surrounding the reduced speed segment (42 mph). Wapiti vehicle collisions per mile increased by 84% between 1983 and 1998 within the 13-mile roadway segment posted at 55 mph, but by only 24% along the 5.6-mile speed reduction segment posted at 42 mph.
3.3.23 Huisjer et al. (2021) also cite a study conducted by the Colorado Department of Transportation of the effect of reduced night-time speed limits at over 100 miles of marked wildlife crossing zones throughout the state (CDOT 2014). Posted speed limits were reduced to 55 mph (88 km/h) from dusk to dawn. Collisions with large wild animals were compared for two years before and two years after posted speed limit reductions in 14 areas. The study found that the reduced posted night-time speed limits were ineffective in providing the desired reduction in operating speed, with drivers exceeding the night-time posted speed limit by an average of seven miles per hour (11 km/h) (thus continuing at speeds equivalent to day-time limits). In eight of the 14 study areas collisions with wildlife decreased during the study, while in the other six of the 14 study areas these collisions increased (CDOT 2014).
3.3.24 Huisjer et al. conclude that, in general, reducing posted speed limits is not a feasible mitigation measure for through roads that are also meant to provide efficient transportation (i.e. short travel times). However, reducing the posted speed limit on one road can also be used to encourage drivers to use other roads that are safer and better equipped to deal with high traffic volume and high vehicle speed, and that may have more robust mitigation measures in place to reduce collisions with large mammals. Further, as noted previously, temporary restrictions activated by dynamic signage in response to actual presence of deer on the road or close to the carriageway may have greater effect (paragraphs 3.3.8 – 3.3.9).
Improvements to road lighting
3.3.25 While highway lighting is clearly not a measure which will be applied on more minor roads, there is some limited evidence that improvements to lighting on major throughways may have some effect in reducing the frequency of DVCs. Huijser et al. (2021) note that roadway lighting may reduce vehicle collisions with animals by 57 - 68% (McDonald 1991, Riley and Marcoux 2006, Wanvik 2009), but it is unclear if reductions in DVCs along lighted roadways are because of increased visibility of the animals to drivers or because animals avoid the roadway lighting.
3.3.26 Lighting can be considered in high-risk areas to reduce DVCs, but the lights may increase the barrier effect of the road and traffic for some species often acting as a deterrent to road crossing by deer thus damaging population connectivity. In addition, other species may be attracted to the lights and experience higher risk of WVCs. Lighting and improvements to lighting may however be of significant advantage along main roads in urban areas where deer are known to have built up significant populations, where seeing them in advance would no doubt help reduce risks of collision (below: paragraph 5.23).
Management of roadside vegetation
3.3.27 Management of roadside vegetation – and specifically the clearance of woodland or scrub from a margin at the road edge – may have benefits both in increasing driver awareness of deer at the roadside, and increasing visibility of oncoming traffic to the deer themselves (Waring et al., 1991).
3.3.28 Reports of the effectiveness of such measures are however mixed. Numbers of incidents on monitored sections of road (before and after clearance) are usually in single figures and thus inevitably results are subject to the question of whether or not perceived reductions are simply due to inherent stochastic variation. Thus while a number of studies appear to show some effect of vegetation removal on accident frequency (e.g. Lavsund and Sandegren, 1991; Seiler, 2005; Meisingset et al., 2014; Hegland and Hamre 2018) an equal number show no effect (e.g. Voss, 2007; Sivertsen, 2010, Lindstrøm, 2016; Rea et al., 2018). It appears in addition that the timing and frequency of mowing or cutting and potential regrowth contribute to varying effectiveness in collision reduction (Rea et al. 2010, 2014; Canal et al., 2019).
3.3.29 Lavsund and Sandegren (1991) report that clearance of a 20 metre strip either side of the highway decreased moose collisions by near 20%. Sivertsen (2010) reported both a decrease and an increase in vehicle collisions with wildlife after roadside clearance in different parts of Norway while Lindstrøm (2016) found no effect of the removal of trees, shrubs and other vegetation at a distance of six to 25 metres from the road. In an analysis of the relationship between frequency of collisions and landscape characteristics in western Norway, Hegland and Hamre (2018) showed that collisions with red deer were 50% lower when open land on road verges increased from 60% to 100%. In a case study, when they cleared vegetation to a depth of 10 metres from a 600 metre stretch of roadway, the number of DVCs decreased by 44% in the four years following clearance (Hegland and Hamre, 2018), although absolute numbers were small enough that differences might have been due to simple stochastic variation. For a trial in Germany, Voss (2007) did not find that numbers of roe deer collisions changed significantly between three years before / after a more limited strip of around five metres was cleared of woody scrubs and maintained thereafter clear of high vegetation. Similar uncertainty arises from studies of removal of vegetation around railway lines: thus Jaren et al. (1991) found that removal of vegetation from a 20 to 30 metre strip on either side of the railway line caused a 56% reduction in the frequency of collisions between trains and moose; Rolandsen et al. (2015) also found some support for a decrease in moose-train collisions in the years following vegetation clearing but, by contrast, Eriksson (2014) found no effect of tree removal along railways in Sweden.
3.3.30 While, as noted by Staines et al. (2001), one might not advocate a 20 m wide clearance zone more generally alongside all railways or major roads, Hegland and Hamre’s analyses, amongst others, seem to make clear that vegetation immediately adjacent to such thoroughfares does increase the risk of collisions – and a wide strip of vegetation removal in particularly sensitive areas or along road sections with poor forward visibility may well be a viable option.
3.3.31 Rea (2003), however, cautions that where vegetation removal is planned, consideration must be given to timing of such clearance and a possible increase in the subsequent attractiveness of fresh regrowth which, in the longer term, could potentially actually result in an overall increase in the number of deer utilising the roadside verge (see also Rea et al, 2010). Re-sowing verges with special seed mixtures of mainly grasses and herbs of relatively low nutritional value could help negate this issue; although the species and seed provenance should be considered in terms of possible impacts on the wider environment.
3.3.32 In similar context, and by contrast, we should note that lack of active management, particularly of wide motorway/highway verges and, for example, adjacent to entrance and exit slip roads, may lead to them becoming favoured shelter/feeding areas for deer, actively drawing them to close proximity to the roadway (e.g. Keken et al., 2019). In Scotland, clusters of DVCs occur at such slip roads on or off motorways or other major roads (Lush and Lush, 2023) and active management to clear understorey vegetation or make these areas in other ways less attractive to deer may reduce collision frequency (below, paragraph 5.15 – 5.17).
3.4 Manipulation of deer habitats and local deer density adjacent to roads
3.4.1 The potential for reduction in DVCs through changes in management of the vegetational cover of verges is considered above in paragraphs 3.3.27 – 3.3.32. It is clear that appropriate verge management may have advantages not simply in increasing visibility for drivers of deer near the carriageway, but it may also have utility in making roadside areas less attractive to deer for shelter or foraging. Below we consider the likely efficacy in reducing the frequency of DVCs through localised reductions of deer density.
Local reductions in deer density
3.4.2 Deer density at a landscape level would appear to play only a minor role in the frequency of collisions (although it is of course axiomatic that if there are no deer in an area there can be no collisions with deer, whereas where deer densities are extremely high, collision risk is inevitably higher).
3.4.3 A number of published studies have claimed a relationship between the frequency of deer vehicle collisions and local deer densities [e.g. McCaffery, 1973; Wisconsin Department of Natural Resources; Schwabe et al., 2002; Rondeau and Conrad, 2003, Seiler et al., 2004], which suggests that a more general reduction of deer densities, in association with other mitigation techniques, may help to reduce collision frequencies. However, other studies suggest deer density plays a more minor role. In a study for the Deer Initiative of collision frequency in different sites within the UK, Uzal (2013) argued that collision frequency was not strongly correlated with deer density, demonstrating that environmental or traffic-related factors accounted for up to 70% of recorded variation in DVC incidence at a landscape [10km2] scale, leaving comparatively little to be explained by variations in deer density.
3.4.4 It is further clear that any relationship that may exist between local deer density and the frequency of DVCs is non-linear (Putman et al., 2011) so that any reduction in landscape level deer density is unlikely to be rewarded with an equivalent reduction in the rate of DVCs. Perhaps in consequence, published reports on the effectiveness of local reductions in deer population density are somewhat inconsistent and contradictory. A number of published studies do appear to show evidence of a reduction in frequency of deer collisions with (sustained) local reductions in deer density (e.g. Jones et al., 1993; Danielson and Hubbard, 1998; Rondeau and Conrad, 2003; Jenks et al., 2002; Sudharsan et al., 2006); whereas imposed hunting restrictions (Kuser and Wolgast, 1983) or general discontinuation of organised culling (Langbein, 2007a) have been correlated with increases in DVC occurrence. In the latter example, the annual number of DVCs attended by the rangers in Ashdown Forest in Southeast England rose five-fold from 74 in 2000, to 215 in 2005, and over 315 in 2006 and 2007, but had fallen back closer to 200 by 2016 in response to the reintroduction of lethal control [JL pers. comm.]
3.4.5 Huisjer et al. (2021) also cite a number of examples where culling of (white-tailed) deer in urban areas in the US has been reported to reduce DVC frequency by between 30 and 94%. It is notable however that these studies were strongly linked to urban or peri-urban populations, where we know that collision risk is particularly high (Langbein, 2008 in Putman et al., 2014).
3.4.6 Further, while we might cite a number of other instances where deer population reductions have been accompanied by reductions in the frequency of deer vehicle collisions, there are other published cases where no such relationship has been established (e.g. Waring et al., 1991; Doerr et al., 2001). There may be differences in the areas across which “local” population densities were reduced in these different studies, which may account for some of the differences in result, but it seems likely that these differences in outcome rather reflect the fact that the frequency of DVCs is related not to one single or main factor but a multiplicity of causal factors in interaction, amongst which the part played by animal density may be a major factor (e.g. Seiler, 2004; McShea et al., 2008), or may not (Uzal, 2013 and others).
3.4.7 And while it does seem inevitable that there will be some positive association of deer densities, traffic flow and DVCs at a regional or landscape scale, localised control of deer density alone will not necessarily always lead to a predictable or lasting reduction of collisions (indeed it is quite possible that the converse could occur due to fast recolonization of favourable habitats by animals which are less accustomed to live in the vicinity of roads). This once more reinforces our more general view that no single solution to the issue is likely to be effective in isolation, and a sustained reduction in DVCs requires adoption of a suite of appropriate measures tailored to the local conditions.
3.4.8 In concluding this section we would note however that while there is to our view no convincing evidence that reductions of deer density over a comparatively wide “local” area will necessarily result in a reduction of DVCs, reduction in very local deer numbers (of animals in the immediate vicinity of the target roadway – perhaps sheltering or feeding on highway verges) may indeed be effective in reducing the risk of DVCs. Where this is undertaken in the short-term through targeted culling, the effect is likely to be of comparatively short duration, due to inevitable recolonisation. But effective longer-term reduction in these immediately-local deer populations may be achieved, as above (paragraphs 3.3.27 - 3.3.32) by manipulation of the roadside vegetation to make these areas less attractive in harbouring significant numbers of animals.
Immunocontraception
3.4.9 In this context we may note, in passing, that in a number of instances, deer population densities have been reduced in different regions of the US (again typically among urban populations) by reduction of herd fertility through the application of immunocontraceptive vaccines. Inevitably, since these vaccines work through reducing recruitment, there is a time-lag before appreciable reductions may be noted in population density.
3.4.10 Indeed, in a meta-analysis of 479 scientific papers on wildlife fertility control Ransom et al. (2014) concluded that overall, anti-fertility treatments are best suited to maintain population densities rather than reduce them. Huisjer et al. (2021) conclude that anti-fertility treatments are unlikely to prove effective in helping to reduce the frequency of DVCs. We should note in addition that the use of anti-fertility drugs and immunocontraceptive vaccines is not currently permitted in UK except under special licence for specific scientific experimentation and such treatments are extremely expensive to administer.
4. Costs and cost-effectiveness
4.1 It is difficult to offer a definitive summary of costs of the different measures available for reducing frequency and severity of DVCs since much will depend on context. Thus for instance, and as only one example, the installation of an overpass or green bridge will depend on whether it is situated on undulating ground where the road may pass within a natural or artificial cutting, or if it must itself be engineered to achieve sufficient height to cross the roadway; cost will also depend on the specifications (e.g. width) of the structure, whether it is to be installed de novo on a new road development or be retrofitted to an existing road scheme, and whether or not it is developed simply as a modification of an existing structure. Cost will also depend on required frequency per mile of any such structure. As a result, estimated costs reported in the literature for most of the mitigation measures considered in these pages are highly variable.
4.2 In consequence we would emphasise that the detailed costs summarised in Table 2 below are indicative only. Information on costs included in the table is drawn largely from Seiler et al. (2016), Huisjer et al. (2021), Schalk et al. (2023), personal communications from F. Brieger (Germany) and the authors’ own experience. Figures quoted in the original sources are thus variously in Euros, Canadian and US dollars as well as in pounds sterling (GBP) and were assessed at different dates – some as long ago as 2016. In the Table, these figures have all been converted to pounds sterling and uprated by UKCPI to be presented at approximately 2023/’24 equivalents.
4.3 We would further note that costs of almost all measures considered are not only highly variable depending on context, but will quickly become outdated, due both to inflation and the further development of new technologies, so must be viewed as purely indicative. While presenting a range of actual estimated costs for different measures therefore, we have in the main focused on categorising these as High, Medium or Low. Costs presented in the Table are further split into Materials, Installation and Maintenance with an indication offered for expected lifespan of each particular measure.
Table 2. Overview of Approximate Costs of Mitigation Measures
Mitigation measures [selected literature references] | Suitable situations and supporting measures | Relative Costs | Approx cost range per km mitigated on both road sides in £ (GBP) K=1000 M=1,000,000) |
---|---|---|---|
High-tensile wire Wildlife Fencing [1, 2, 3] | Major roads of high traffic flow. Most effective if leads to safer crossing, and with exit ramps / deer-leaps provided within fenced section. | Materials: HIGH Installation: HIGH Maintenance: MEDIUM Lifespan: > 15 years | 40K to 150K (Inclusive of up to two exit ramps per km at c.1K to 5K each) |
Overpasses & Joint-use green bridge [4, 1, 5, 6, 3] | Major high-risk roads. Most effective with lead-in fencing, and natural ground cover. | Materials: HIGH Installation: HIGH Maintenance: MEDIUM Lifespan: > 25 years | 3M to 20M very size dependant (inclusive of fencing for 0.5 km each side) |
Underpasses & Viaducts (New-built) [4, 1, 3] | Major high-risk roads. Most effective with lead-in fencing, and natural ground cover. | Materials: HIGH Installation: HIGH Maintenance: MEDIUM Lifespan: > 25 years | 0.5M to 10M very size dependant (incl. fencing for 0.5 km each side) |
Adapt existing over- or underpasses [7, 5, 3] | Major or medium traffic flow roads. Enhancements may include lead-in fencing, heighten parapets, and/or cover near entrance and part natural substrate. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: > 15 years | 20K to 50K (highly dependent on whether lead-in fence provided and type and length) |
Highway cross-walks [1, 3] | Low to medium speed road sections. Not suitable for high-speed routes where traffic has to be kept flowing. | Materials: HIGH Installation: HIGH Maintenance: MEDIUM Lifespan: > 10 years | 40K to 80K Inclusive of advance flashing signage, speed reduction, power etc. to protect lead-in. |
Optical wildlife warning reflectors [8, 9, 3] | Roads of low traffic volume providing some traffic free periods. | Materials: LOW Installation: LOW Maintenance: MEDIUM Lifespan: c. 10 years | From < 1.5K to 6K (incl. posts for mounting) |
Active wildlife warning devices with audio + light signals [10, 8] | Roads of low traffic volume, where habituation least likely, and providing safe crossing periods. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 5 years | c. 5K to 15K variable by type and whether with gateways and/or networked. Maintenance incl. replacement of any faulty or vandalised. |
Chemical / Olfactory Deterrents [3] | Roads of low to moderate traffic flow | Materials: LOW Installation: LOW Maintenance: HIGH Lifespan: < 0.5 years | c. from <1K but requires renewed application every three to six months, with additional associated labour costs. |
Vehicle mounted whistles and electronic horns | N/A | Materials: LOW Installation: LOW Maintenance: LOW Lifespan: 10+ years | N/A |
Standard wildlife warning signage [1, 3] | Any road type, but should be targeted to forewarn of short, well-defined sections of high risk. | Materials: MEDIUM Installation: MEDIUM Maintenance: LOW Lifespan: 20+ years? | c. 5K to 20K (based on 2 x 1 sign in each direction; cost reflects high cost of secure road furniture installation. |
Interactive speed-activated wildlife plus speed alert signage. [10, 11, 1, 3] | Any road type, but should be targeted to forewarn of short, well-defined sections of high risk. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 5 years | c. 20K to 40K Driver habituation over time, if not reinforced by seeing animals near the crossing point |
Roadside Animal Detection Systems – linked to animal- activated digital signage [1, 2, 3 ] | Low to medium traffic flow and speed road sections. Not suitable for main carriageways of high-speed strategic routes, but potential there for use at entry / exit slip roads. | Materials: MEDIUM to HIGH Installation: MEDIUM to HIGH Maintenance: MEDIUM to HIGH Lifespan: unknown | Highly variable (where driver-warning signal integrate within devices); to c. 25K to 100K (if networked and linked to digital gateway signage); to c. 100K to 500K (with also long lead-in fencing to cross-walks) |
Speed reduction / limits [1, 3] | Low to moderate traffic flow routes. Speed sign at same site as wildlife sign preferable. | Materials: MEDIUM Installation: MEDIUM Maintenance: LOW Livespan: 20+ years? | c. 10K to 20K (based on 2 x 1 sign in each direction; cost reflects high cost of secure road furniture installation. Excludes road safety assessment. |
In-vehicle warning systems
| Likely more suited to roads of low to moderate traffic flow and speed
| Materials: unknown Installation: - Maintenance: - Livespan: - | Vehicle manufacturer / market led |
Road lighting
| High risk un-lit areas in urban DVC areas | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: 20+ years? | c. 20K to 50K Cost reflects high cost of safe road furniture installation and power supply. |
Public awareness raising and driver education [3] | Increasing importance as traffic and collision risk escalates. Animal hazard awareness should be built into national driver syllabuses. | Materials: LOW Installation: LOW Maintenance: LOW Lifespan: annual | from < 1K Dependent on whether local, county, country level campaigns. |
Reduction of local deer density | Site specific assessment needed whether appropriate. Maybe required to prevent increase, if not reduction of deer numbers, for other measures (including fencing) to remain effective. | Materials: LOW Installation: MEDIUM Maintenance: MEDIUM Lifespan: 1 year | from < 1K Highly variable dependent on local context, current deer management, and man-days required. |
Verge clearance and maintenance [3] | Suitable for ALL roads. Ideally verges re-sown with grass mixtures of low digestibility | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 1-3 years | c. 3K to 5K (based on brush clearance along 10m wide verge) |
Wider Habitat / clearance | Mainly Junctions of motorway and other major roads | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 2-3 years? | (see above) (est. 3K per ha, highly variable depending on density and maturity of woody scrub |
Immuno-contraception | Isolated, self-contained populations and / or urban situations only (and only in countries where legal!) | Materials: HIGH Installation: HIGH Maintenance: HIGH Lifespan: 1 year | Cost unclear, but likely high, due to need for close veterinary supervision licencing etc.. |
Numbered Table 2 references: 1 - Seiler et al. (2016), 2 - Brieger (pers. comm), 3 - Huijser et al. (2021), 4 - Rosell et al (2022), 5 - Natural England (2015), 6 - Dodd et al. (2012), 7 - Langbein (2010), 8 - Schalk et al. (2023), 9 - Brieger (2020), 10 - Langbein (2007b), 11 - Sullivan et. al (2004). Where figures from the literature have contributed to ranges of indicative costs given, these have been converted to £ (GBP) and uprated for approximation to 2023 prices using the UK Consumer Price Index (UKCPI).
Cost-effectiveness
4.4 Of all the different potential measures discussed, many appear to have limited or zero effect in reducing the frequency of deer vehicle collisions. In their review of available studies, Huisjer et al. (2021) conclude that Seasonal warning signs (Variable Message Boards) may have an effectiveness of between 9% to 50%; Dynamic Signage, linked to Roadside Animal Detection Systems have a reported effectiveness in reducing DVCs of 33% to 97%. Clearance of roadside verges has the potential to reduce collisions by up to 50% (depending on the density of vegetation in the first instance) while they report that local reductions of deer density through culling might reduce collision rate by 30% to 94% (but see earlier, paragraphs 3.4.2 – 3.4.6). Overpasses or underpasses alone were not necessarily particularly effective in reducing road crossings and collisions, but when combined with appropriate fencing were seen to reduce collision frequency by between 80% and 100% (mean across all reported studies 83%). All other measures considered had unknown or zero effect, although the authors considered there is some potential in the future for in-vehicle warning systems coupled with Roadside Animal Detection systems.
4.5 Seiler et al. in their 2016 review propose a cost-benefit analysis to address the effectiveness of different mitigation measures. Cost–benefit analysis (CBA) is a systematic approach to estimate the strengths and weaknesses of alternative activities for a project. The transport sector applies CBA to evaluate whether an investment in measures to improve traffic flow and safety, for example, is economically worthwhile. They note however that this approach is not without controversy and critics such as Hauer (1994) have opposed the very idea of putting a monetary value on human life as this conflicts with the basic assumptions in CBA. By comparison, cost-effectiveness analyses (CEA) usually compare the relative costs and outcomes of several actions. CEA do not answer the question whether or not a measure should be taken or an investment should be made, but they help select the measure that produces the greatest effect per currency unit invested. The advantage over CBA is that expected benefits do not need to be monetised but instead can be expressed in qualitative or quantitative terms and thus be directly linked to policy objectives and political targets.
4.6 In the event, Seiler et al. use only a stylised calculation with somewhat hypothetical data on accident frequencies and costs to derive the estimates of benefit summarised in part 2 of their Table 2 (printed page 22 - 23 of their 2016 report). It has to be said that the table is difficult to interpret but, while noting the general effectiveness of roadside fencing, the addition of other structures (such as overpasses, underpasses or other crossing structures) in their view adds little to the overall cost-benefit analysis, largely due to the absolute cost of installation of such structures. That said: wildlife crossing structures in combination with wildlife fences reduce collisions, thus effectively reducing the costs to society, e.g., human fatalities, human injuries, property damage, loss of hunting revenue (Conover et al.,1995; Huijser et al., 2009). These estimated annual benefits from reduced wildlife vehicle collisions were estimated in 2012 to have exceeded $200,000/mile. (Dodd et al., 2012).
4.7 The analyses of Seiler et al. (2016) do highlight the enormous benefit of providing escape ramps or other structures to permit animals trapped within a fenced section of carriageway to escape. In relation to other measures, they suggest, as do other commentators, that static warning signs on unfenced roads are ineffective; fixed speed reductions are not cost-effective (we consider they are not particularly effective); and that dynamic signage responsive to vehicle speed or linked to Roadside Animal Detection Systems are not cost-effective (once again in relation to the absolute cost of such measures). In relation to the financial implication implicit in a serious accident involving serious injury or human fatality and the considerations of Dodd et al. above, we might disagree.
4.8 An additional important consideration in assessing the cost-effectiveness of all these various measures is to consider benefits beyond simply those of reducing the frequency or risk of DVCs. Thus for example, green bridges of suitable specification provide essential continuity of populations and habitat for a wide range of species other than simply deer (see for example Giorgii et al, 1997; Huisjer et al., 2021; Rosell et al., 2022).
4.9 We present in Table 3 a summary of the established or perceived effectiveness of different mitigation measures reported in the literature. We conclude that despite the high potential cost, appropriate fencing leading to dedicated crossing places (overpasses or underpasses) remain the most effective way of reducing the frequency of DVCs on major highways. On unfenced roads, it seems clear that the most effective measure is offered by dynamic signage linked to Roadside Animal Detection Systems, although we would advocate that in known trouble spots further consideration also be given to stretches of appropriate fencing leading to well-signed crosswalks. We would also recommend that further consideration be given to the development of in-vehicle warning systems linked to Roadside Animal Detection Systems.
4.10 Finally, while this report is restricted solely to reduction in frequency of deer vehicle collisions, we would note that in the future wild boar/ feral pigs are also likely to be increasingly involved in collisions on our roadways. Wild boar collisions with vehicles are already an increasing issue in parts of England and the Welsh borders (reports from the former Deer Initiative, and see Gatto, 2016) as also elsewhere in Europe (Lagos et al., 2012; Morelle et al., 2013; Putzu et al., 2014; Jägerbrand and Gren (2018); Saint –Andrieux et al., 2020).
Table 3. Potential Effectiveness and Relative Costs of differing Wildlife Vehicle Mitigation Measures. In general, best results are achieved by selecting several complementary measures rather than reliance on any one of the approaches listed below.
Mitigation measures [selected lit. references] | Suitable situations and supporting measures | Relative Costs | Disadvantages | Potential lasting effectiveness
|
---|---|---|---|---|
High-tensile wire [1, 2, 3, 4, 5, 6, 7, 8, 9] | Major high-risk roads of high traffic flow. Most effective when leads to safer crossing point and contains escape ramps / leaps. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: 20+ years? | Barrier effect to non-target wildlife.
| HIGH Well proven, provided appropriate mesh size, height, and bottom dug in or very tight to ground |
Overpasses & joint-use green bridge; new-built [7, 8, 9, 10, 11, 12, 13, 14] | Major high-risk roads. Most effective with lead-in fencing, and natural ground cover. | Materials: LOW Installation: LOW Maintenance: LOW Lifespan: annual | Feasibility dependent on landscape. More readily installed on new-build than for existing roads. | HIGH Well proven; ungulate usage increases with width; but smaller structures can also help alleviate wildlife collision risk. |
Underpasses and viaducts (new-built) [7, 9, 10, 11, 12, 15] | Major high-risk roads. Most effective with lead-in fencing, and natural ground cover. | Materials: LOW Installation: MEDIUM Maintenance: MEDIUM Lifespan: 1 year | Feasibility dependent on landscape. More readily installed on new-build than for existing roads. | HIGH If adequate specification. Mostly lower cost than overpasses of similar size. |
Adapt existing over- or underpasses [11, 12] | Major or medium traffic flow roads; enhancements may include lead-in fencing, heighten parapets, and/or cover near entrance and part natural substrate. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 1-3 years | Variable suitability, requires selection of those with best potential. | MEDIUM Less well studied than above; potential of differing structures likely variable from High to Low, selection should focus on sites where some deer use already apparent. |
Highway cross-walks [8, 9, 13] | Low to medium speed road sections. Not suitable for high speed routes where traffic has to be kept flowing. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 2-3 years | Less likely to be acceptable on major routes where traffic has to be kept flowing. | MEDIUM ? HIGH ? Highly dependant on whether supported by fencing, speed restriction and dynamic signage. |
Optical wildlife warning reflectors [14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24] | Roads of low traffic volume providing some traffic free periods. | Materials: HIGH Installation: HIGH Maintenance: HIGH Lifespan: 1 year | Rapid habituation where lit up by frequent traffic. Can at best only function during night. | LOW Very limited evidence of lasting success. Do not prevent normal range use, but at best work only during night. Vegetation near reflectors needs to be kept clear. |
Active audio plus flashing light signal wildlife warning devices [17, 21, 25, 26] | Roads of low traffic volume, where habituation least likely, and providing safe crossing periods. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 5 years | General effectiveness remains unproven. Limited potential on roads of high traffic volume. | LOW Most unproven to last after novelty response fades; may depend on signal type and variability. |
Chemical / Olfactory Deterrents [19, 27, 28, 29, 30, 31, 32, 33] | Roads of low to moderate traffic flow | Materials: LOW Installation: LOW Maintenance: HIGH Lifespan: < 0.5 years | Requires renewal at regular intervals. Habituation. | LOW Limited convincing evidence of success. Most intend to raise level of alertness, rather than prevent animals crossing. |
Vehicle mounted whistles and electronic horns [34, 35, 36, 37] | N/A | Materials: LOW Installation: LOW Maintenance: LOW Lifespan: 10+ yrs. | Signals mostly drowned out by traffic noise. | NONE Most studies show lack of any response by wildlife. |
Standard wildlife warning signage [8, 9, 38,39,40,41,42,43] | Any road type, but should be targeted to forewarn of short, well-defined sections of high risk. | Materials: MEDIUM Installation: MEDIUM Maintenance: LOW Lifespan: 20+ years? | Over-abundance of wildlife and other signage leading to reduced effect on driver behaviour. | LOW May help absolve legal responsibility of road authorities or deer managers. |
Interactive speed-activated wildlife plus speed signage [41, 42, 43, 45, 46] | Any road type, but should be targeted to forewarn of short, well-defined sections of high risk. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 5 years | Driver habituation if not reinforced by seeing animals and as digital signs becomes common. | MEDIUM Lack widespread assessment of actual effects on DVC reduction. Increased driver perception. |
Roadside Animal Detection Systems – linked to animal activated digital signage [8, 9, 47, 48, 49, 51, 52] | Low to medium traffic flow and speed road sections. Not suitable for high-speed strategic routes, but potential there for use at entry / exit slip roads. | Materials: MEDIUM to HIGH Installation: MEDIUM to HIGH Maintenance: MEDIUM to HIGH Lifespan: unknown | High cost compared to standard or speed- activated signage. | MEDIUM TO HIGH Promising effects on driver awareness and local speed reduction. |
Speed reduction / limits [4, 9, 53, 54, 55, 56, 57, 58, 59] | Low to moderate traffic flow routes. Speed sign at same site as wildlife sign preferable. | Materials: MEDIUM Installation: MEDIUM Maintenance: LOW Lifespan: 20+ years? | Feasibility and acceptability for major roads limited. | MEDIUM Needs to be enforced. Reduces severity of accidents if not necessarily frequency. |
In-vehicle warning systems [9, 60, 61] | Any road type, but likely most effective on routes moderate traffic speed and flow.
| Materials: unknown Installation: - Maintenance: - Lifespan: - | Not yet widely available and tested. | MEDIUM Potential likely to increase as driver-less vehicle technology increases. |
Road lighting [9, 62, 63] | High risk un-lit areas in urban DVC areas. | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: 20+ yrs. | Suitable only in specific situations. | MEDIUM Improving visibility of deer to drivers |
Public awareness raising and driver education [8, 9, 19, 64] | Increasing importance as traffic and collision risk escalates. Animal hazard awareness should be built into national driver syllabuses. | Materials: LOW Installation: LOW Maintenance: LOW Lifespan: annual | Effects unclear even if regularly repeated. | LOW High potential but unproven effects. Integrate with other road safety campaigns, leaflets, press releases to keep cost low. |
Reduction of local deer density [9, 65, 66, 67, 68] | Site specific assessment needed whether appropriate; prevention of increasing deer numbers also helps other measures (including fencing) to remain effective. | Materials: LOW Installation: MEDIUM Maintenance: MEDIUM Lifespan: 1 year | Localised culling may increase problem elsewhere if not done sensitively. | MEDIUM Provided undertaken sensitively and appropriate time, working with neighbours and as part of overall DVC reduction strategy. |
Verge clearance and maintenance [8, 9, 16, 19, 69, 70, 71] | Suitable for ALL roads. Ideally verges re-sown with grass mixtures of low digestibility/ palatability | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 1-3 years | Increase in forage quality on verge needs to be avoided. | HIGH Improved forward visibility for drivers and animals AND to reduce holding cover for deer to settle in verge habitat. |
Wider Habitat / clearance [71, 72] | Mainly Junctions of motorway and other major roads | Materials: MEDIUM Installation: MEDIUM Maintenance: MEDIUM Lifespan: c. 2-3 yrs | Effects on other wildlife to consider and need to maintain. | HIGH To avoid deer settling on roundabouts, and other habitats within highways estate especially near junctions. |
Immuno-contraception [9, 73, 74] | Isolated, self-contained populations and / or urban situations (in countries where legal!) | Materials: HIGH Installation: HIGH Maintenance: HIGH Lifespan: 1 year | Raises animal welfare concern. Not legal in UK. | LOW (in UK context) Non-lethal; higher public acceptability in some countries / situations than culling. Limited short-term effectiveness. |
Numbered Table 3 references: 1 - Reed et al. (1982), 2 - Ward (1982), 3 - Ballon (1985), 4 - Putman et al. (2004), 5 - Mastro et al. (2008), 6 - Feldhammer et al. (1986), 7 - Rosell et al (2022), 8 - Seiler et al. (2016), 9 59 - Huijser et al. (2021) 10 7 - Ohlbrich (1984), 11 8 - Iuell et al. (2003), 12 9 - Georgii et al. (2007), 13 57 - Olsson et al. (2008), 14 58 - Dodd et al. (2012) 15 10 - ECONAT (1992), 11 - Langbein, (2010), 12 Langbein (2015), 13 - Lehnert and Bissonette (1997), 14 - Schaffer and Penland (1985), 15 - Gladfelter (1982), 16 - Waring et al. (1991), 17 - Reeve and Anderson (1993), 18 - Woodward et al. (1973); 19 - Voss (2007); 20 - D’angelo (2006), 21 - Pokorny and Poličnik (2008), 22 - Benten et al. (2018b, 2019), 23 - Brieger et al. (2016), 24 - Brieger and Strein (2020), 25 - Langbein (2007b), 26 - Schalk et al. (2023), 27 - Kerzel and Kirchberger (1993), 28 - Lebensorger (1993), 29 - Lutz (1994), 30 - Pokorny et al. (2008), 31 Bil et al. (2020), 32 Kosta et al. (2015), 33 - Kosta (pers comm), 34 - Deer-Vehicle Crash Information Clearing House (2003), 35 - Romin and Dalton (1992), 36- Schober and Sommer (1984), 37 - Scheifele et al. (2003), 38 - Putman (1997), 39 - Hedlund et al. (2004), 40 - Stanley et al. (2006), 41 - Sullivan et al. (2004), 42 - Hardy et al. (2006), 43 - Pojar et al. (1975); 44 - Huijser et al. (2015) , 45 - Hujser et al. (2006), 46 - ScotlandTranserv (2010), 47 - Gordon et al. (2003), 48 - Hammond and Wade (2004), 49 - Mosler-Berger and Romer (2003), 50 Brieger (pers. Comm), 51- Suter et al. 2021, 52 - AniMot (2019) , 53 - Huijser et al. (2017), 54 - Gunter et al. (1998); 55 - Bertwistle (1999); 56 - Aarts and van Schagen (2006); 57 - Meisinget et al. (2014), 58 - Savolainen and Ghosh (2008), 59 - Ahmed et al. (2021), 60 - Forslund and Bjarkefur (2014), 61 - Adams (2017), 62 - Riley & Marcoux (2006), 63 - Wanvik (2009), 64 - Rea (2003), 65 - McCaffery (1973), 66 - Schwabe et al. (2002), 67 - Rondeau and Conrad (2003), 68 - Sudharsen et al. (2006), 69 - Lavsund and Sandgren (1991), 70 Hegland and Hamre (2018), 71 Keken et al., (2019), 72 Lush and Lush (2023), 73 - Doerr et al. (2001), 74 - Rutberg and Naugle (2008).
5. Combining mitigation measures to develop DVC reduction strategies
5.1 In the previous sections we have considered differing categories of mitigation measures and their effectiveness individually (physical barriers, safe passages, animal deterrents, influencing driver behaviour, manipulation of roadside habitats and local deer density), partly for ease of presentation. In practice, however, there will be very few situations where any one mitigation measure will be effective at leading to a significant and sustained reduction in collisions with deer or other animals when applied in isolation. In an earlier review (Langbein et al., 2011), and here at paragraph 3.0.3 we have stressed that “… in general best results are achieved through use of a range of complementary measures, rather than reliance on any one of the individual approaches listed”. Other reviews (e.g. Seiler et al., 2016, Huijser et al., 2021) also emphasise that “… there is no single magic tool that prevents WVC [wildlife vehicle collisions]”, nor is there ever likely to be; instead, prevention requires a combination of approaches that target differing factors on different scales”.
5.2 Unfortunately, in practice, past attempts at local DVC or WVC reduction in the UK as well as abroad, have all too often focussed on single solutions (not least on standard ‘wildlife warning’ signs or relatively cheap roadside animal deterrents, which in reality are well-established as being ineffective; paragraphs 3.2.4 – 3.2.11 ; 3.3.2 – 3.3.5), rather than treating DVCs in a similar manner to other road safety issues.
5.3 To address hotspots of traffic collisions more generally, it is standard practice for Local Authorities or Highways authorities to design a ‘Road Safety Strategy’. Such strategies will almost always involve a combination of approaches (such as, for example, signage of varying types aimed at reducing traffic speed on approaches to a new pedestrian crossing; or improvement to sightlines and verges along tortuous stretches of road, supported by signage to highlight sharp bends. Similarly, to tackle collision reduction where a DVC hotspot has been identified, the starting point should be to devise a combined DVC strategy; in the first instance this will normally require an on-site assessment to determine the feasibility of differing potential mitigation measures, considering local terrain, habitat, and road layout and affordability. Ideally, the DVC strategy should from the outset be seen and incorporated within the context of any existing local road safety initiatives or strategy, not least as this may shed light on many more local collisions than just those in which deer are implicated and maximise any cross-over benefits.
5.4 It is clear that in terms of both cost and effectiveness different mitigation measures are likely to be more appropriate in some contexts than others (for example fencing appropriate to lengths of motorways or major dual carriageways of high traffic volume and travel speeds would be less suited to more minor roads) or to be more or less effective in rural, peri-urban or urban settings. The one primary DVC reduction measure that we might suggest as desirable in all contexts, is to minimise extensive areas of scrubby vegetation likely to attract and conceal deer close to roadside verges as far as practicable, as well as any woodland encroaching directly on the road verge.
Motorways and major A-dual carriageway trunk roads
5.5 Many published studies have reported that a very high proportion of DVCs overall, and in particular of the most severe incidents, occur on major roadways where speeds of traffic and total traffic volumes are greatest (paragraphs 2.12- 2.16 here, and references therein).
5.6 While it is unlikely in a UK context that such major throughways will be fully fenced against wildlife crossing, attention should be given to the provision of safe crossing places for deer whether in provision of new structures (overpasses or underpasses) or in modification of existing structures (such as maintenance tunnels, pedestrian bridges or farm access routes) to make them more likely to be used by deer (see paragraph 3.1.8 et seq). Some progress has already been made in this regard by Transport Scotland, with one small overpass on the A9 dual carriageway at Luncarty-Birnham recently adapted for wildlife, in addition to two earlier part-green overpasses having been provided as part of the Aberdeen Western Peripheral Route, and we understand that one or more others are likely to follow (Transport Scotland, pers. comm). Equally, modification of existing structures, such as maintenance tunnels, pedestrian bridges or farm access routes to make them more likely to be used by deer - although not excluding deer entirely from crossing the carriageway - may nevertheless reduce collision risk as deer tend to seek out the easiest crossing points (paragraph 3.1.8).
5.7 Such modification may be encouraged further, where feasible, by erecting short lengths (c. 250 metres) of deer fencing on either side of the crossing point (and on both sides of the road), and/or improving hedgerow or other cover near entrances, in order to funnel animals towards the crossing.
Fencing should be to the specifications of COST 341 (see paragraph 3.1.2 and see also Rosell et al., 2022: Fig 7.2.7) and consideration should be given to installing cattle grids/wildlife grids across the verge at fence ends to prevent animals from entering the carriageway at this point. We recommend the provision of earthen ramps (deer-leaps; paragraph 3.1.4, 3.1.5) along extensive fence-lines to facilitate escape of animals which do become trapped within the carriageway (see also Rosell et al., 2022: Figs 7.2.21 and 7.2.22).
5.8 It is critical that any fences erected are to adequate specification and are regularly monitored, with rapid and effective repairs undertaken in case of breakage. New fence-lines should be well-secured at or well below ground, especially in areas where wild boar/feral pigs are known to be present or likely to colonise in the future.
5.9 Work carried out in England (Langbein, 2011b; Enterprise Mouchel, 2013; Nelli et al., 2018) has further established many of the trunk road locations with highest overall DVC incidence to be located at roundabouts and slip roads with large expanses of dense woody scrub cover. In Scotland, recent reports from the national DVC monitoring project (Langbein, 2019; Lush and Lush 2023) have similarly highlighted the very high proportion of DVC hotspots identified on the trunk road network which occur close to motorway and dual carriageway junctions or interchanges. It is indeed notable that in the analysis of Scottish trunk road hotspots by Lush and Lush (2023), 90% relate to road junctions. We pay further specific attention to the issue of mitigation at such junctions as well as entry and exit slip roads below at paragraphs 5.15 – 5.19).
Single carriageway A-roads and more minor roads
5.10 On roads of lower traffic volume and vehicle speeds, once again a major consideration should be removal of scrub or understorey on roadside verges and ensuring that the edges of any woodland blocks are at a safe distance from the carriageway.
5.11 At known crossing places or collision hotspots, attention should be given, as above, to deflecting animals to safer crossing structures (existing bridges or underpasses) with short stretches of deer fencing where this can be done safely; or else by stock fencing and hedges, that may provide a degree of cover for deer or other wildlife approaching the structure.
5.12 Where no accommodating overbridges, underpasses or culverts exist, consideration should be given to the installation of dedicated crossings on the roadway itself at places of higher visibility (cross-walks; paragraph 3.1.21 – 3.1.24), once again with lengths of fencing leading animals towards this safer crossing point and with approaching vehicle drivers alerted by appropriate signage. We would recommend the positioning of Roadside Animal Detection Systems (RADS) at such crossings, linked to dynamic matrix signage activated only when an animal is detected approaching or actually in the carriageway (paragraph 3.3.8). Installation of RADS, linked to dynamic matrix signage, on the approach to known crossing points or known collision hotspots would still be of value even if no formal cross-walks are installed on site.
5.13 Various differing types of such systems are increasingly used in Switzerland, Germany, and Scandinavia (Suter et al., 2021; Strein et al., 2010; Seiler et al., 2016); they are generally a combination of measures, based on enhanced digital signage to alert drivers on approaches to a common deer crossing point, often supported by Roadside Animal Detection Systems. In a Scottish context there may well be opportunities to trial such combined RADS plus connected signage systems not only at DVC hotspots in fairly open countryside or through woodland on A-single carriageway trunk or more minor roads, but also at some of the shortlisted slip-road hotspots at major trunk road junctions, as discussed above.
5.14 We would reiterate that standard wildlife warning signs and light reflectors (whether clear or coloured), while seemingly attractive because of comparatively low cost, have been proven by many, authoritative, independent studies to have no utility in reducing the frequency of DVCs.
Road junctions, slip roads and major roundabouts
5.15 As noted above (paragraph 5.9) a very high proportion of DVCs, especially on more major roads, occur at junctions. We have noted that in their analysis of collision hotspots on the Scottish trunk road network, Lush and Lush (2023) reported that 90% of shortlisted blackspots relate to road junctions. We therefore offer special attention to this specific context here.
5.16 A major factor likely to be responsible for the high frequency of collisions at such junctions and major interchanges is that major road junctions commonly include islands of open ground surrounded by tarmac on all sides, defining an area within which there is very limited public disturbance, not least by dog walkers or other people on foot. These areas are commonly planted for landscape purposes or, through lack of disturbance secondarily develop a cover of trees or woody shrubs by natural colonisation. With the combination of good cover and lack of human disturbance such ‘wooded islands’ thus unintentionally develop into refuges that deer and other wildlife will seek out for shelter, either seasonally or during day-time, especially from late spring onwards when they offer most cover. Where the main highway is fenced an additional contributing factor may arise from the fact that such fences must end at junctions to permit vehicle access and animals otherwise held from the carriageway by extensive fencing may then adopt those points where the fences finish, as major crossing places.
5.17 Clearly, one primary recommendation that would likely feature to some extent in any combination of mitigation measures would be to reduce areas of scrubby vegetation that do provide cover in this way for deer as far as practicable. However, while this may sound obvious and straightforward, design of major highways often deliberately incorporates extensive tree and shrub planting for landscaping and noise mitigation. Such deliberate plantings or incidental development of woodland may also have a further important function in creating corridors for other wildlife and their removal may thus have significant negative impacts in these other aspects. Where such considerations would militate against removal of the tree and shrub cover, serious consideration should be given to deer-fencing the carriageway ‘islands’ created by slip roads or major roundabouts in order to exclude deer from those areas.
5.18 The potential measures in this context might include:
- Enhanced (digital) wildlife signage, activated ideally by animal detection systems (RADS) for slip roads, and/or main carriageway; and/or temporary mobile seasonal VMS.
- Use of alternative traffic calming on slip roads (rumble strips; lighting; flashing signage).
- As above, fencing in the entire junction / interchange; or fencing targeted at the relevant ‘wooded islands’ created between slip roads and the main carriageway, or on roundabouts.
- Enhancing existing trunk underpasses / viaducts or overpasses for deer use.
- Discussions should also be carried out with neighbouring landholders. We noted in earlier sections (paragraphs 3.4.3, 3.4.4) that the relationship between deer culling and DVCs is often non-linear, and maybe affected (positively or negatively) as much by the timing and manner where lethal management is undertaken as by the number of deer culled. Based on local background information about any ongoing deer management on neighbouring land, more informed decisions can then be made as to whether additional – and closely targeted – reduction of deer population numbers may assist within an overall strategy for reduction of DVC frequency at such identified hotspots.
5.19 The actual combination, and local feasibility of the measures above will clearly vary from situation to situation.. Consideration of enhanced signage, backed by RADS and traffic calming for slip roads might well be feasible in many of the areas, and if used could be supported also by reducing scrub cover along verges. Where there is extensive deer-holding cover not only on, but also adjacent to the trunk road estate, close working with neighbours is advised. Working together it may be possible to find solutions, for example adapting the timing and location of any lethal and other deer management to minimise movement of deer onto and settling on the highway estate. Suitability of fencing as part of the solutions will vary greatly between locations.
Mitigation in urban and peri-urban areas
5.20 In Scotland, as in the rest of the UK and many other countries, sightings of deer within peri-urban as well as fully urban environments have become increasingly common over the last five decades. To date, most urban deer sightings in Scotland relate to roe deer that have established small resident populations in many public parks, derelict industrial sites, wildlife corridors along river courses and trunk roads or other bypasses, as well as in other small patches of urban green spaces, such as cemeteries. However, sightings of larger species in urban areas such as red, and more locally fallow, are also increasing. Inevitably DVCs have become an increasingly common occurrence as a result, not only around the periphery of such conurbations but also in the centres of major cities, such as Glasgow, Aberdeen, Dundee and Edinburgh.
5.21 By way of illustration that these are by no means localised occurrences, Figure 2 below show maps of the distribution of ‘reported’ DVCs within the city boundaries of Glasgow and Aberdeen for 2013 to 2017 (from Langbein, 2019).
5.22 Many DVC mitigation measures in urban areas will be akin to those discussed above for A-single and more minor roads, though likely with greater emphasis on traffic calming measures. In view of the difficulties of deer management in general in urban environments (Putman et al., 2014), a primary measure in all urban cases should be the development of a City or County Council wide Deer Management Plan or Strategy. The latter is likely best led by the Local Authority and should seek to involve other stakeholders (e.g. major private or company landholders with deer holding cover, the SSPCA, police and others, aside from rangers’ teams for council owned greenspace).
5.23 Overall DVC mitigation strategies in urban and peri-urban areas are likely to include a combination of several if not all of the following:
- Creation of Local Authority Deer Management Plan involving local stakeholders.
- Raising public awareness of DVC issues and the need for deer management.
- Minimising areas of dense understorey vegetation at roadsides and within roundabouts.
- Enhancing selected existing accommodation structures (footpath under/overpasses, canal sides, culverts) for use by deer and other wildlife to reduce crossing activity over main roads.
- Improving road lighting in locations where deer congregate near roadsides or cross most at night (paragraphs 3.3.25, 3.3.26).
- Traffic calming, such as speed humps, where appropriate, including where advantageous also for wider road safety plan. In areas of high urban fallow deer presence in Plymouth and some London Boroughs, although deer extensively frequent residential housing estates daily and many deer are injured, the numbers of serious human-injury DVCs remains relatively low to date, most likely because of the presence of speed humps and 20 mph speed limits in many of the residential areas that the deer frequent.
5.24 Despite the recommendations included for all scenarios considered above, we should emphasise that while these may suggest a shortlist of potential measures which might, in combination, prove affective in different contexts, each individual situation is unique. Consequently, few general recommendations neither can, nor should be made based on simple investigation of maps and collision statistics for a given DVC hotspot, as many other aspects of the site, including e.g. slope of the terrain, the deer species present, habitat and other wildlife, ownership boundaries, and deer management (if any) on surrounding land need to be taken into consideration. Thus while we offer a broad toolbox of potential measures above, planning for any individual location should always begin with an actual on-site survey to determine the local feasibility of differing potential measures.
Concluding remarks
5.25 One of the disappointments encountered in preparing this review is the limited extent to which mitigation measures which have been introduced in different places in an attempt to address collision hotspots have subsequently been subjected to rigorous monitoring to determine the effectiveness of the measure put in place. Over the years, different approaches to mitigation have been implemented on a number of roads in Scotland, typically in areas identified as hotspots for DVCs (e.g. on the A82 through Glencoe, on the A87 between the Cluanie Inn and Shielbridge and on the A835 between Aultguish and Braemore Junction), yet the effectiveness of these measures in actual reduction of DVCs has not been fully monitored.
5.26 We would conclude this review with a plea that, where mitigation measures are attempted in the future to reduce collision frequency at established hotspots, funding should be found to support appropriate monitoring of their effectiveness. We would also advocate that specific funding is sought to establish the effectiveness of Roadside Animal Detection Systems (RADS) linked to dynamic signage in a number of known hotspots on more minor roads.
References
Aarts, L. and van Schagen, I. 2006. Driving speed and the risk of road crashes: A review. Accident Analysis and Prevention, 38, 215-224.
ADAC, 2008. Grünbrücken – Empfehlungen für die Praxis. [Advice note on green-bridges]; (in German). ADAC, München.
Adams, E. 2017. Volvo's cars now spot moose and hit the brakes for you. Wired Magazine.
Ahmed, A., Cohen, J. and Anastasopoulos, P. 2021. A correlated random parameters with heterogeneity in means approach of deer-vehicle collisions and resulting injury-severities. Analytic Methods in Accident Research, 30, 100160.
Albon, S.D., McLeod, J., Potts, J., Brewer, M., Irvine, J., Towers, M., Elston, D., Fraser, D. and Irvine, J. 2017. Estimating national trends and regional differences in red deer density on open-hill ground in Scotland: identifying the causes of change and consequences for upland habitats. Scottish Natural Heritage Commissioned Report 981.
Al-Ghamdi, A.S. and AlGadhi, S.A. 2004. Warning signs as countermeasures to camel-vehicle collisions in Saudi Arabia. Accident Analysis and Prevention 36, 749-760.
Allen, T.D.H., Huijser, M.P. andWilley, D.W. 2013. Effectiveness of wildlife guards at access roads. Wildlife Society Bulletin, 37, 402-408.
AniMot, 2019. Die intelligente Zukunft in der Wildunfallprävention.
Apollonio, M., Andersen, R., and Putman, R.J. eds. 2008. European Ungulates and their Management in the 21st century. Cambridge University Press.
Allen, R.E. and McCullough, D.R. 1976. Deer-car accidents in Southern Michigan. Journal of Wildlife Management, 40, 317-25.
Armstrong, J.J. 1992. An Evaluation of the Effectiveness of Swareflex deer reflectors. Research and Development Branch, Ministry of Transportation, Ontario.
Babińska-Werka, J., Krauze-Gryz, D., Wasilewski, M. and K. Jasińska, K. 2015. Effectiveness of an acoustic wildlife warning device using natural calls to reduce the risk of train collisions with animals. Transportation Research Part D, 38, 6-14.
Ballon P. 1985. Bilan technique des aménagements réalisés en France pour réduire les impacts des grandes infrastructures linéaires sur les ongulés gibiers. Actes du XVII ème Congrès de l'Union Internationale des Biologistes du Gibier, 679-689.
Bashore, T.L., Tzilkowski, W.M. and Bellis, E.D. 1985. Analysis of deer-vehicle collision sites in Pennsylvania. Journal of Wildlife Management 49, 769-774.
Benten, A., Annighöfer, P. and Vor, T. 2018a. Wildlife warning reflectors’ potential to mitigate wildlife-vehicle collisions–a review on the evaluation methods. Frontiers in Ecology and Evolution, 6, 37.
Benten, A., Hothorn, T., Vor, T. and Ammer, C. 2018b. Wildlife warning reflectors do not mitigate wildlife-vehicle collisions on roads. Accident Analysis and Prevention, 120, 64-73.
Benten, A., Balkenhol, N., Vor, T. and Ammer, A. 2019. Wildlife warning reflectors do not alter the behavior of ungulates to reduce the risk of wildlife-vehicle collisions. European Journal of Wildlife Research, 65, 76.
Berndt, C.V. 2021. Behavioural responses of ungulates to sound systems – Can simulated risk influence behaviour? Master thesis. Swedish University of Agricultural Sciences.
Bertwistle, J. 1999. The effects of reduced speed zones on reducing bighorn sheep and elk collisions with vehicles on the Yellowhead Highway in Jasper National Park. Proceedings of the International Conference on Wildlife Ecology and Transportation., Missoula, Montana, USA, September 13 to 16 1999, pp. 727-735.
Biedenweg, T. A., Parsons, M. H., Fleming, P. A., & Blumstein, D. T. 2011. Sounds scary? Lack of habituation following the presentation of novel sounds. PLoS ONE, 6(1).
Bíl, M., Kušta, T., Andrášik, R., Cícha, V., Brodská, H., Ježek, M. and Keken, Z. 2020. No clear effect of odour repellents on roe deer behaviour in the vicinity of roads. Wildlife Biology, wlb.00744.
Bissonette, J.A. and Adair, W. 2008. Restoring habitat permeability to roaded landscapes with isometrically-scaled wildlife crossings. Biological Conservation 141, 482-488.
Bissonette, J.A. and Hammer, M. 2000. Effectiveness of earthen ramps in reducing big game highway mortality in Utah. Utah Cooperative Fish and Wildlife Research Unit Report Series 1, 1-29.
Bissonette, J. A., Kassar, C. and Cook, L.J. 2008. An assessment of costs associated with deer-vehicle collisions: human death and injury, vehicle damage, and deer loss. Human–Wildlife Conflicts 2, 17-27.
Brieger, F. and Strein, M. 2020. Keine Wirkung von blauen Wildwarnreflektoren auf das Rehverhalten. Straßenverkehrstechnik 64(3), FGSV.FVA-Wildtierinstitut. Bedn-Wuertemmberg, Germany.
Brieger, F., Hagen, R., Vetter, D., Dormann, C.F. and Storch, I. 2016. Effectiveness of light-reflecting devices: A systematic reanalysis of animal-vehicle collision data. Accident Analysis and Prevention, 97, 242-260.
Brieger, F., Hagen, R., Kröschel, M., Hartig, F., Petersen, I., Ortmann, S. and Suchant, R. 2017. Do roe deer react to wildlife warning reflectors? A test combining a controlled experiment with field observations. European Journal of Wildlife Research, 63(5), 72.
Brieger, F., Kämmerle , J-L., Hagen, R. and Suchant, R. 2022. Behavioural reactions to oncoming vehicles as a crucial aspect of wildlife-vehicle collision risk in three common wildlife species. Accident Analysis & Prevention, 168, 106564.
Bullock, K.L., Malan, G. and Pretorius, M.D. 2011. Mammal and bird road mortalities on the Upington to Twee Rivieren main road in the southern Kalahari, South Africa. African Zoology, 46, 60-71.
Canal, D., Camacho, C., Martín, B., de Lucas, M. and Ferrer, M. 2019. Fine‑scale determinants of vertebrate roadkills across a biodiversity hotspot in Southern Spain. Biodiversity and Conservation, 28, 3239-3256.
Carsignol, J. 1989. Dix annees d'observations des collisions vehicules-grand mammiferes gibier sur l'autoroute A4 en Alsace-Lorraine et sur le reseau routier du departement de la Moselle. Office National de la Chasse, Bulletin Mensuelle, 135, 32-37.
CDOT, 2014. Colorado Department of Transportation Final Report – HB 1238 Wildlife Crossing Zones. Denver, Colorado, USA.
Clevenger, A.P., Chruszcz, B. and Gunson, K.E. 2001. Highway mitigation fencing reduces wildlife-vehicle collisions. Wildlife Society Bulletin, 29, 646-653.
Clevenger, AP., Chruszcz, B.,. Gunson, K.E and Wierzchowski, J. 2002. Roads and Wildlife in the Canadian Rocky Mountain Parks - Movement, Mortality and Mitigation. Final report to Parks Canada, Banff, Alberta, Canada, 432 S.
Clevenger, A. P., and Waltho, N. 2000. Factors influencing the effectiveness of wildlife underpasses in Banff National Park, Alberta, Canada. Conservation Biology, 14, 47-56.
Conover, M.R., Pitt, W.C., Kessler, K.K., DuBow, T.J. and Sanborn, W.A. 1995. Review of human injuries, illnesses, and economic losses caused by wildlife in the United States. Wildlife Society Bulletin, 23, 407-414.
CTGREF, 1978. Autoroute et grand gibier. Note technique 42. Groupement technique forestier. Ministère de l'Agriculture. Paris.
D’Angelo, G. and van der Ree, R. 2015. Use of reflectors and auditory deterrents to prevent wildlife-vehicle collisions. In: R. Van der Ree, C. Grilo and D. Smith (eds). Ecology of roads: A practitioner’s guide to impacts and mitigation. John Wiley & Sons Ltd. Chichester, United Kingdom pp. 213-218.
D’Angelo, G. J., D’Angelo, J.G., Gallagher, G.R., Osborn, D.A., Miller, K.V and Warren, R.J. 2006. Evaluation of wildlife warning reflectors for altering white-tailed deer behavior along roadways. Wildlife Society Bulletin, 34, 1175–1183.
Danielson, B.J. and Hubbard, M.W. 1998. A Literature Review for Assessing the Status of Current Methods of Reducing Deer-Vehicle Collisions. Unpublished report to Iowa Department of Transportation and Iowa Department of Natural Resources.
Decker, D.J., Loconti Lee, K.M. and Connelly, N.A. 1990. Incidences and Costs of Deer-related Vehicular Accidents in Tomkins County, New York. HDRU Series 89-7, revised February 1990. Department of Natural Resources, New York State Agricultural College and Cornell University. 22 pp
Deer-Vehicle Crash Information Clearing House; DVCIC, 2003. On-line manual of mitigation measures and perceived effectiveness.
Desire G., and Recorbet G. 1990. Resultats de l'enquete realisee de 1984 a 1986 sur les collisions entre les vehicules at leas grands mammiferes sauvages. Office National de la Chasse, Bulletin Mensuelle, 143, 38-47.
DfT, 2011. Department for Transport National Statistics. Road traffic estimates in Great Britain: 2010. On-line annual statistical update by UK Government National Statistics.
DfT, 2023. Department for Transport National Statistics. Road traffic estimates in Great Britain: 2022. On-line annual statistical update by UK Government National Statistics. [or see DfT figures for Scotland only.
Dodd, N.L., Gagnon, J.W., Boe, S., Ogren, K. and Shweinshurg, R.E. 2012. Wildlife-vehicle collision mitigation for safer wildlife movement across highways: State Route 260. Final Report, FHWA-AZ-12- 603. Research Center, Arizona Department of Transportation, Phoenix, Arizona, USA.
Doerr, M.L., McAninch, J.B and Wiggers, E.P. 2001. Comparison of 4 methods to reduce white-tailed deer abundance in an urban community. Wildlife Society Bulletin, 29, 1105-1113.
Donnell, E., Kersavage, K. and Fontana Tierney, L. 2018. Self-enforcing roadways: A guidance report. Publication No. FHWA-HRT-17-098. Institute of Transportation Engineers, Washington, DC, USA.
ECONAT, 1992. Protection de la faune dans les projets de nouveau traces ferroviaires, 51 S. Yverdon-les-Bains.
Elvik, R. 2014. Speed and road safety - new models. TØI report: 1296/2014. Transportøkonomisk institutt, Oslo, Norway.
Elvik, R., Høye, A., Vaa, T. and Sørensen, M. 2009. Handbook of Road Safety Measures, Emerald Group Publishing, Bingley, UK.
Enterprise Mouchel, 2013. Enterprise Mouchel Area Wide [HA Area 3] Deer Collision Study. Contract Report to the Highways Agency. Enterprise Mouchel, Basingstoke.
Eriksson, C. 2014. Does tree removal along railroads in Sweden influence the risk of train accidents with moose and roe deer? Master thesis, Swedish Agricultural University, 20 p.
Etter, D. R., Van Deelen, T. R., Ludwig, D. R., Kobal, S.N. and Warner, R.E. 2002. Survival and movements of white-tailed deer in suburban Chicago, Illinois. Journal of Wildlife Management 66, 500-510.
Fehlberg, U. 1994. Ökologische Barrierewirkung von Strassen auf wild-lebende Säugetiere. Deutsche Tierärzliche Wochenschrift 101, 81-132.
Feldhamer, G.A., Gates, J.E., Harman, D.M., Loranger, A.J. and Dixon, K.R. 1986. Effects of Interstate highway fencing on white-tailed deer activity. Journal of Wildlife Management, 50, 497-503.
Finder, R.A., Roseberry, J.L. and Woolf, A. 1999. Site and landscape conditions at white-tailed deer collision locations in Illinois. Landscape and Urban Planning 44, 77-85.
Fitzpatrick, K., Carlson, P., Brewer, M.A., Wooldridge, M.D. and Miaou, S.-P. 2003. Design speed, operating speed, and posted speed practices. NCHRP Report 504. National Cooperative Highway Research Program, Transportation Research Board of the National Academies, Washington D.C., USA.
Flower, J.P. 2016. Emerging Technology to Exclude Wildlife from Roads: Electrified pavement and deer guards in Utah, USA. MSc Thesis. Utah State University.
Ford, S.G. and Villa, S.L. 1993. Reflector use and the effect they have on the number of mule deer killed on California highways. Report FHWA/CA/PD94/01. California Department of Transport, Sacramento.
Forman, R.T.T. and Alexander, L.E. 1998. Roads and their major ecological effects. Annual Review of Ecology and Systematics, 29, 207-231.
Forman, R.T.T., Friedman, D.S., Fitzhenry, D., Martin, J.D., Chen, A.S. and Alexander, L.E. 1997. Ecological effects of roads: Toward three summary indices and an overview for North America. In: Canters, K., A. Piepers and D. Hendriks-Heersma (eds.): Habitat fragmentation and Infrastructure. Proceedings of the International Conference on Habitat Fragmentation, Infrastructure and the role of Ecological Engineering, 17-21 September 1995, the Netherlands. pp 40-54.
Forslund, D. and Bjarkefur, J. 2014. Night vision animal detection. Proceedings of the IEEE Intelligent Vehicles Symposium, June 8-11, 2014, Dearborn, Michigan, USA. pp 737-742.
Found, R. and Boyce, M.S. 2011. Warning signs mitigate deer-vehicle collisions in an urban area. Wildlife Society Bulletin, 35, 291-295.
Gagnon, J.W., Dodd, N.L., Sprague, S.C., Ogren, K.S. and Schweinsburg, R.E. 2010. Preacher Canyon wildlife fence and crosswalk enhancement project evaluation. State Route 260. Final report – Project JPA 04-088. Arizona Game and Fish Department, Phoenix, Arizona, USA.
Gagnon, J.W., Dodd, N.L., Sprague, S.C., Ogren, K.S., Loberger, C.D. and Schweinsburg, R.E. 2019. Animal-activated highway crosswalk: long-term impact on elk-vehicle collisions, vehicle speeds, and motorist braking response. Human Dimensions of Wildlife, 24(2), 132-147.
Gatto, G. 2016. Wild boar-vehicle collisions in the Forest of Dean: spatial patterns and available measures for mitigation. Report to the Deer Initiative for England and Wales.
Gilbert, J.R. 1982. Evaluation of Deer Mirrors for Reducing Deer-Vehicle Collisions. United States Federal Highway Administration Report FHWA/RD/82/061; Washington, D.C. 16 pp.
Gill, R.M.A. 1990. Monitoring the Status of European and North American Cervids. GEMS Information Series, 8; Global Environment Monitoring Systems, United Nations Environment Programme, Nairobi. 277 pages.
Georgii, B., Peters-Ostenberg, E., Henneberg, M., Herrmann, M., Müller-Stieß H. and Bach, L. 2007. Nutzung von Grünbrücken und anderen Querungsbauwerken durch Säugetiere. Gesamtbericht zum Forschungs-und Entwicklungsvorhabe. [In German] Forschung Strassenbau und Strassenverkehrstechnik Heft 971. Bonn.
Giorgii, B. and Wotschikowski, U. 2007. Strassen und Wildtiere. [In German] Bundesministerium für Verkehr, Bau und Stadtentwicklung, Bonn.
Gladfelter, J.R. 1982. Effect of Wildlife Highway Warning Reflectors on Deer-Vehicle Accidents. Iowa Highway Research Board Project HR-210, Iowa Department of Transportation, 11 pp.
Gleason, J. S., and Jenks, J. A. 1993. Factors influencing deer–vehicle mortality in east central South Dakota. Prairie Naturalist 25, 281-288.
Gordon, K. M., Anderson, S., Gribble, B. and Johnson, M. 2003. Evaluation of the FLASH (flashing light animal sensing host) system in Nugget Canyon, Wyoming, Report FHWAWYL-01/03F. Wyoming Cooperative Fish and Wildlife Research Unit, Laramie, Wyoming, USA.
Gordon, K.M., McKinstry, M.C. and Anderson, S.H. 2004. Motorist response to a deer-sensing warning system. Wildlife Society Bulletin 32, 565-573.
Grace, M.K., Smith, D.J. and Noss, R.F. 2017. Reducing the threat of wildlife-vehicle collisions during peak tourism periods using a roadside animal detection system. Accident Analysis and Prevention 109, 55-61.
Groot Bruinderink, G.W.T.A. and Hazebroek, E. 1996. Ungulate Traffic Collisions in Europe. Conservation Biology 10, 1059-67.
Grovenburg, T.W., Jenks, J.A., Klaver, R.W., Monteith, K.L., Galster, D.H., Schauer, R.J., Morlock, W.W. and Delger, J.A. 2008. Factors affecting road mortality of white-tailed deer in eastern South Dakota. Human–Wildlife Conflicts, 2, 48-59.
Gunther K., Biel, M. and Robinson, H. 1998. Factors influencing the frequency of road killed wildlife in Yellowstone National Park. In: Evink, G.L., P. Garrett, D. Zeigler and J. Berry (eds.). Proceedings of the International Conference on Wildlife Ecology and Transportation: Fort Myers, Florida, USA. FL-ER-69-98. Florida Department of Transportation, Tallahassee, Florida, USA. pp 32-42.
Guyton, J., Jones, J. and E. Entsminger, E. 2014. Alternative mowing regimes’ influence on native plants and beer: SS228 Final Project Report. Report no. FHWA/MDOT-RD-14-228. Mississippi State University, Starkville, Mississippi, USA.
Haikonen, H. and Summala, H. 2001. Deer-vehicle crashes – extensive peak at 1 hour after sunset. American Journal of Preventive Medicine, 21, 209-213.
Hammond, C. and Wade M.G. 2004. Deer avoidance: the assessment of real-world enhanced deer signage in a virtual environment. Final report. Minnesota Department of Transportation, St. Paul, Minnesota, USA.
Hansen, C.S. 1983. Costs of deer-vehicle accidents in Michigan. Wildlife Society Bulletin, 11, 161-164.
Hardy, A., Lee, S. and Al-Kaisy,A.F. 2006. Effectiveness of animal advisory messages on dynamic message signs as a speed reduction tool. Transportation Research Record, 1973, 64-72.
Hartwig, D. 1991. Erfassung der Verkehrsunfälle mit Wild im Jahre 1989 in Nordrhein-Westfalen im Bereich der Polizeibehörden. Zeitschrift für Jagdwissenschaft 37, 55-62.
Hartwig, D. 1993. Auswertung der durch Wild Verursachten Verehrsunfälle nach der Statistik für Nordrhein-Westfalen. Zeitschrift für Jagdwissenschaft, 39, 22-33.
Hatt, S. 2000. Grünbrücke Lotterbuck A 4.2.9: Eine Erfolgskontrolle nach drei Jahren. Schweitzerische Zeitschrift für Forstwessen, 151, 290-297.
Hedlund, J.H., Curtis, P.D. and Williams, A.F. 2004. Methods to reduce traffic crashes involving deer: what works and what does not. Traffic Injury Prevention, 5, 122-131.
Hegland, S.J. and Hamre, L.N. 2018. Scale-dependent effects of landscape composition and configuration on deer vehicle collisions and their relevance to mitigation and planning options. Landscape and Urban Planning, 169, 178-184.
Hlavac, V. and Andel, P. 2002. On the Permeability of Roads for Wildlife: A Handbook. Agency for Nature Conservation and Landscape Protection of the Czech Republic.
Hothorn, T., Brandl, R., and Müller, J. 2012. Large-scale model-based assessment of deer-vehicle collision risk. PLoS One, 7, e29510.
Hothorn, T., Müller, J., Held, L., Möst, L., and Mysterud, A. 2015. Temporal patterns of deer–vehicle collisions consistent with deer activity pattern and density increase but not general accident risk. Accident Analysis and Prevention, 81, 143-152.
Huang, B., Zhang, Y., Lu, J. and Lu, L. 2013. A simulation study for minimizing operating speed variation of multilane highways by controlling access. Procedia - Social and Behavioral Sciences, 96, 2767-2781.
Hubbard, M.W., Danielson, B.J. and Schmitz, R.A. 2000. Factors influencing the location of deer-vehicle accidents in Iowa. Journal of Wildlife Management, 64, 707-713.
Huijser, M.P., Ament, R.J., Bell, M., Clevenger, A.P., Fairbank, E.R., Gunson, K.E. and McGuire, T. 2021. Animal Vehicle Collision Reduction and Habitat Connectivity – Literature Review. Report 701-18-803 TO 1 to the Nevada Department of Transportation, Carson City, NV 89712.
Huijser, M.P., Fairbank, E.R. and Abra, F.D. 2017. The reliability and effectiveness of a radar-based animal detection system. Report FHWA-ID-17-247. Idaho Department of Transportation (ITD), Boise, Idaho, USA.
Huijser, M. P. and McGowen, P.T. 2003. Overview of animal detection and animal warning systems in North America and Europe. Proceedings of the International Conference on Ecology and Transportation, Lake Placid, New York, USA.
Huijser, M. P., McGowen, P.T., Camel, W., Hardy, A., Wright, P., Clevenger, A., Salsman, L. and Wilson, T. 2006. Animal vehicle crash mitigation using advanced technology phase 1: review, design and implementation. Oregon Department of Transportation Research Unit, Salem, Oregon, and Federal Highway Administration, Washington, D.C.
Huijser, M.P., McGowen, P., Fuller, J., Hardy, A., Kociolek, A., Clevenger, A.P., Smith, D. and Ament, R. 2008. Wildlife-vehicle collision reduction study. Report to Congress. U.S. Department of Transportation, Federal Highway Administration, Washington D.C., USA.
Huijser, M.P., Mosler-Berger, C., Olsson, M. and Strein, M. 2015. Wildlife warning signs and animal detection systems aimed at reducing wildlife-vehicle collisions. In: R. Van der Ree, C. Grilo and D. Smith (eds). Ecology of roads: A practitioner’s guide to impacts and mitigation. John Wiley & Sons Ltd. Chichester, United Kingdom. pp. 198-212.
Hussain, A., Armstrong, J.B., Brown, D.B. and Hogland, J. 2007. Land-use pattern, urbanization, and deer–vehicle collisions in Alabama. Human–Wildlife Conflicts 1, 89-96.
Insurance Institute for Highway Safety, 1993. Deer, Moose Collisions with Motor Vehicles Peak in Spring and Fall. Status Report. 28 (4) April 3, 1993.
Insurance Institute for Highways Safety, 2004. Lots of approaches are under way to reduce deer collisions, but few have proven effective. Status Report, 39 (1), 5-7.
Insurance Institute for Highway Safety, 2008. Collisions with deer and other animals spike in November; Fatal Crashes up 50% since 2000. IIHS Press Release 30 October 2008.
Iuell,B., Bekker, G.J., Cuperus, R. Dufek,J., Fry, G., Hicks,C.,Hlavac, V., Keller, V.B., Rosell, C., Sangwine, T., Torslov, N, Wandall, B. le M. 2003. Wildlife and Traffic: A European Handbook for Identifying Conflicts and Designing Solutions. European Commission Action 341 on "Habitat Fragmentation due to Transportation Infrastructure", Brussels.
Jägerbrand, A.K. and Gren, I-M. 2018. Consequences of increases in wild boar-vehicle accidents 2003–2016 in Sweden on personal injuries and costs. Safety 4, 53.
Jared, D. 1992. Evaluation of Wild Animal Highway Warning Reflectors. Report 98003, Georgia Department of Transportation, Atlanta, Georgia.
Jaren, V., Andersen, R., Ulleberg, M., Pedersen, P.H. and Wiseth, B. 1991. Moose-train collisions: the effects of vegetation removal with a cost-benefit analysis. Alces, 27, 93-99.
Jenks, J.A., Smith, W.P. and DePerno, C.S. 2002. Maximum sustained yield harvest Versus trophy management. Journal of Wildlife Management, 66, 528-535.
Jiang, Z., Jadaan, K. and Ouyang Y. 2016. Speed harmonization—Design speed vs. operating speed. FHWA-ICT-16-019. Illinois Center for Transportation, Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA
Jones, J.M., and Witham, J.H. 1993. Urban Deer “Problem-Solving” in Northeast Illinois: an overview. In the Proceedings of the 55th Midwest Fish and Wildlife Conference – Urban Deer: A Manageable Resource, St. Louis, MO, pp. 58-65.
Kekena, Z., Sedoník, J., Kušta, T., Andrášik, R. and Bíl, M. 2019. Roadside vegetation influences clustering of ungulate vehicle collisions. Transportation Research Part D, 73, 381-390.
Kerzel, H. 2005. Wildunfälle: Zur Notwendigkeit von Verkehrsschutzzäunen und Grünbrücken. In: Grünbrücken für den Biotopverbund. Schriftenreihe des Landesjagdverbandes Bayern e. V., Band 14.
Kerzel, H. and Kirchberger, U. 1993. Erfolge im Kampf gegen Wildunfälle. Die Pirsch, 18, 3-5.
Kinley, T.A., Newhouse, N.J and Page, H.N. 2003. Evaluation of the wildlife protection system deployed on Highway 93 in Kootenay National Park during autumn, 2003. Sylvan Consulting Ltd., Invermere, British Columbia, Canada.
Kistler, R. 1998. Wissenschaftliche Begleitung der Wildwarnanlagen Calstrom WWA-12-S. July 1995-November 1997. Schlussbericht. Infodienst Wildbiologie & Oekologie, Zurich, Switzerland. St. Paul, MN.
Kloeden, C.N., McLean, A.J., Moore, V.M. and Ponte, G. 1997. Traveling speed and the risk of crash involvement. Volume 1 – findings. NHMRC Road Accident Research Unit, University of Adelaide, Adelaide, Australia.
Kuser, J. E., and L. J. Wolgast. 1983. Deer roadkill increases with no-firearms-discharge law. Bulletin of the New Jersey Academy of Science, 28, 71-72.
Kušta, T., Keken, Z. and Kůta, Z. 2015. Effectiveness and costs of odor repellents in wildlife-vehicle collisions: a case study in Central Bohemia, Czech Republic. Transportation Research Part D, 38, 1-5.
Lagos, L., Picos, J. and Valero, E. 2012. Temporal pattern of wild ungulate-related traffic accidents in northwest Spain. European Journal of Wildlife Research, 58, 661-668.
Langbein, J. 1985. North Staffordshire Deer Survey 1983-1985. I. Research and Development. British Deer Society: Fordingbridge, UK.
Langbein, J. 2006. Conservation and management of deer in Epping Forest and its Buffer Land Estate. Corporation of London.
Langbein, J. 2007a. National Deer-Vehicle Collisions Project: England 2003-2005. Final Report to the Highways Agency. The Deer Initiative, Wrexham.
Langbein J. 2007b. Use of remote video surveillance to investigate deer behaviour in relation to wildlife deterrents, roads and vehicles. Presentation at ‘Deer on our Roads Seminar’, Ashridge, UK, October 2007.
Langbein, J. 2008. Deer vehicle collisions in peri-urban areas: A risky life for deer. Presentation to British Deer Society Urban Deer Conference at Linnean Society, London.
Langbein, J. 2010. Pilot study to assess the potential of selected existing structures on the A30 and A38 trunk roads to provide safer crossing places for deer. For Highways Agency, England. Deer Initiative Research report 10/1.
Langbein. J. 2011a. Deer Vehicle Collisions in Scotland Monitoring Project 2008-2011: Final Report to Scottish Natural Heritage. Deer Initiative Research Report 11/2.
Langbein. J. 2011b. Monitoring DVC in England to end 2010: Final Report to Highways Agency. Deer Initiative Research Report 11/3.
Langbein, J. 2015. How do deer cross Britain’s busiest motorway ? On-line Deertails Blog.
Langbein, J. 2019. Deer Vehicle Collisions (DVC) data collection and analysis 2016 - 2018. Final Report to Scottish Natural Heritage. Scottish Natural Heritage Research Report No. 1158.
Langbein, J. and Putman, R.J. 2006. National Deer-Vehicle Collisions Project; Scotland, 2003-2005. Report to the Scottish Executive, June 2006.
Langbein, J., Putman, R.J. and Pokorny, B. 2011. Road traffic accidents involving ungulates and available measures for mitigation. In: Ungulate Management in Europe: Problems and Practices (eds. R.J. Putman, M. Apollonio and R. Andersen), Cambridge University Press, 215-259.
Lavsund, S. and Sandegren, F. 1991. Moose-vehicle relations in Sweden: a review. Alces, 27, 118-126.
Lebersorger, P. 1993. Verkehrspartner Wild. Weidwerk, 11, 47-48.
Leblond, M., Dussault, C., Ouellet, J., Poulin, M., Courtois, R. and Fortin, J. 2007. Electric fencing as a measure to reduce moose-vehicle collisions. Journal of Wildlife Management, 71, 1695-1703.
Lehnert, M.E. and Bissonette, J.A. 1997. Effectiveness of highways crosswalk structures in reducing deer-vehicle collisions. Wildlife Society Bulletin, 25, 809-818.
Lindstrøm, I.M. 2016. No mitigating effects of roadside vegetation clearing on ungulate-vehicle collisions in Nord-Trøndelag. MSc thesis, Norwegian University of Science and Technology, Trondheim, Norway.
Lode, T. 2000. Effect of a motorway on mortality and isolation of wildlife populations. Ambio 29, 163-166.
Lodnert, D. 2021. Evaluating the behavioural response of moose (Alces alces) to acoustic stimuli. Master´s thesis. Swedish University of Agricultural Sciences.
Lush, M.J. and Lush, C.E. 2023. Deer Vehicle Collision analysis 2019-2021. NatureScot Research Report 1329.
Lutz, W. 1994. Ergebnisse der Anwendung eines sogenannten Duftzaunes zur Vermeidung von Wildverlusten durch den Strassenverkehr nach Gehege- und Freilandorientierungen. Zeitschrift für Jagdwissenschaft, 40, 91-108.
Madsen, A.B., Fyhn, H.W. and Prang, A. 1998. Traffic killed animals in landscape ecological planning and research. In: Danish: Trafikdræbte dyr i landskabsøkologisk planlægning og forskning. – DMU Rapport 228, Århus, DK.
Madsen, A.B., Strandgaard, H. and Prang A. 2002. Factors causing traffic killings of roe deer, Capreolus capreolus in Denmark. Wildlife Biology, 8, 55-61.
Malo, J.E, Suarez, F., and Diaz, A. 2004. Can we mitigate animal-vehicle accidents using predictive models? Journal of Applied Ecology, 41, 701-710.
Mastro, L.L., Conover, M.R. and Frey, S.N. 2008. Deer-vehicle collision prevention techniques. Human-Wildlife Conflicts, 2, 80-92.
McCaffery, K.R. 1973. Road-kills show trends in Wisconsin deer populations. Journal of Wildlife Management, 37, 212-216.
McDonald, M. G. 1991. Moose movement and mortality associated with the Glenn Highway expansion, Anchorage, Alaska. Alces, 27(1), 208-219.
McShea, W. J., Stewart, C.M., Kearns, L.J., Liccioli, S. and Kocka, D. 2008. Factors affecting autumn deer–vehicle collisions in a rural Virginia county. Human–Wildlife Conflicts, 2, 110–121.
Meisingset, E.L, Loe, L.E., Brekkum, O. and Mysterud, A. 2014. Targeting mitigation efforts: The role of speed limit and road edge clearance for deer–vehicle collisions. Journal of Wildlife Management, 78, 679-688.
Meyer, E. 2006. Assessing the effectiveness of deer warning signs. Final report. Report no. K-TRAN: KU-03-6. The University of Kansas, Lawrence, Kansas, USA.
Mladenhoff, D.J., Sickely T.A. and Wydeve A.P. 1999. Predicating gray wolf landscape recolonization: Logistic regression models vs field data. Ecological Applications, 9, 37-44.
Morelle, K., Lehaire, F. and Lejeune, P. 2013. Spatio-temporal patterns of wildlife-vehicle collisions in a region with a high-density road network. Nature Conservation, 5, 53-73.
Mosler-Berger, C., and Romer, J. 2003. Wildwarnsystem CALSTROM. Wildbiologie 3, 1–2. [In German.]
Muurinen, I. and Ristola, T. 1999. Elk accidents can be reduced by using transport telematics. Finncontact, 7(1), 7-8.
Mysterud, A. 2004. Temporal variation in the number of car-killed red deer Cervus elaphus in Norway. Wildlife Biology, 10, 203-211.
Natural England, 2015. Green Bridges - Literature Review. Natural England Commissioned Report NECR181.
Nelli, L., Watson, P., Langbein, J. and Putman, R.J. 2018. Mapping Risk: Quantifying and predicting the risk of deer-vehicle collisions on major roads in England. Mammalian Biology, 91, 71-78.
Nielsen, C.K., Anderson, R.G. and Grund, M.D. 2003. Landscape influences on deer-vehicle accident areas in an urban environment. Journal of Wildlife Management, 67, 46-51.
Ng, J.W., Nielsen, C. and St. Clair, C.C. 2008. Landscape factors influencing deer–vehicle collisions in an urban environment. Human–Wildlife Conflicts, 2, 34-47.
Olbrich, P. 1984. Untersuchung der Wirksamkeit von Wildwarnreflektoren und der Eignung von Wilddurchlässen. Zeitschrift für Jagdwissenschaft, 30, 101-116.
Olsson, M.P.O, Widen, P. and Larkin, J.L. 2008. Effectiveness of a highway overpass to promote landscape connectivity and movement of moose and roe deer in Sweden. Landscape and Urban Planning, 85, 133-139.
Oord, J.G. 1995. Handreiking maatregelen voor de fauna langs weg en water, Rijkswaterstaat. Dienst Weg- en Waterbouwkunde & Dienst Landinrichting en Beheer Landbouwgronden,Delft, Utrecht, 278 pp.
Pafko, F. and Kovach, B. 1996. Experience with Deer Reflectors: Trends in Assessing Transportation Related Wildlife Mortality. Minnesota Department of Transportation, St Paul.
Pepper, H.W. 1999. Road Traffic Accidents and Deer Reflectors: A comparative trial of the efficacy of standard red and new blue/green roadside reflectors at preventing motor vehicle and wild deer collisions. Internal report on Project 257, Forestry Commission, Forest Research.
Pepper, H.W., Chadwick, A.H. and Packer, J.J. 1998. Deer Reflectors and Road Traffic Accidents Through Forestry Commission Forests: A review of traffic accident records for roads where deer warning reflectors have been installed. Appendix to contract report VC 0317, Ministry of Agriculture Fisheries and Foods, London.
Pepper, S., Barbour, A. and Glass, J. 2019. The Management of Wild Deer in Scotland. Report of the Deer Working Group to Scottish Government.
Pfister, H.P., Keller, V., Reck H., and Georgii, B. 1997. Bio-ökologische Wirksamkeit von Grünbrücken über Verkehrswege. Bundesministerium für Verkehr, Forschung Straßenbau und Straßenverkehrstechnik, Heft 756. Bonn.
Pfister, H.P., D. Heynen, B. Georgii, V. Keller, and F. von Lerber. 1999. Häufigkeit und Verhalten ausgewählter Wildsäuger auf unterschiedlich breiten Wildtierbrücken (Grünbrücken). Schweizerische Vogelwarte, Sempach, Switzerland. 49 pp.
Pojar, T.M., Prosence, R.A., Reed, D.F. and Woodard, T.N. 1975. Effectiveness of a lighted, animated deer crossing sign. Journal of Wildlife Management, 39, 87-91.
Pojar, T.M., Reseigh, T.C. and Reed, D.F. 1972. Deer crossing signs may prove valuable in reducing accidents and animal deaths. Highway Research News, 46, 20–23.
Pokorny, B. 2006. Roe deer-vehicle collisions in Slovenia: situation, mitigation strategy and countermeasures. Veterinarski Arhiv, 76, 177-S187.
Pokorny, B. and Poličnik, H. 2008. Monitoring učinkovitosti izvedenih ukrepov za preprečevanje trkov vozil z divjadjo [Monitoring of effectiveness of countermeasures implemented for reducing the number of game-vehicle collisions] [In Slovene]. Final report for Slovene Directorate for Roads, Contract no. 2415-07-000721/0. ERICo Velenje, 82 pp.
Pokorny, B., Marolt, J., and Poličnik, H. 2008. Ocena učinkovitosti in vplivov zvočnih odvračalnih naprav kot sredstva za zmanjšanje števila trkov vozil z veliko divjadjo [Assessment of the effectiveness and impacts of acoustic deterrents as a countermeasure for reducing the number of big game-vehicle collisions] [In Slovene]. Final report for Slovene Hunters Association, Contract no. LZS-04/1298. ERICo Velenje, 107 pp.
Pürstl, A. 2006. Tierärztlichles Gutachten zum Farbsehvermögen von Rot und Rehwild. Tierambulanz Türkenschanzplatz, Vienna. (unpublished)
Puglisi, M.J., Lindzey, J.S. and Bellis, E.D. 1974. Factors associated with highway mortality of white-tailed deer. Journal of Wildlife Management, 38, 799-807.
Putman, R.J. 1997. Deer and Road Traffic Accidents: Options for Management. Journal of Environmental Management, 51, 43-57.
Putman, R.J., Langbein, J. and Staines, B.W. 2004. Deer and Road Traffic Accidents; A Review of Mitigation Measures: Costs and Cost-Effectiveness. Report to the Deer Commission for Scotland. Contract RP 23A.
Putman, R.J., Langbein, J., Green, P. and Watson, P. 2011. Identifying threshold densities for wild deer in the UK above which negative impacts may occur. Mammal Review, 41, 175-196.
Putman, R.J., Langbein, J., Watson, P., Green, P. and Cahill, S. 2014. The Management of Urban Populations of Ungulates. Chapter 7 in: Behaviour and Management of European Ungulates (eds. R.J. Putman and M. Apollonio), Whittles Publishing, Caithness, 148-177.
Putzu, N., Bonetto, D., Civallero, V., Fenoglio, S., Meneguz, P.G., Preacco, N. and Tizzani, P. 2014. Temporal patterns of ungulate-vehicle collisions in a subalpine Italian region. Italian Journal of Zoology, 81(3), 463-470.
Ransom, J.I., Powers, J.G., Hobbs, N.T. and Baker, D.L. 2014. Ecological feedbacks can reduce population-level efficacy of wildlife fertility control. Journal of Applied Ecology, 51, 259-269.
Rea,V. 2003. Modifying roadside vegetation management practices to reduce vehicular collisions with moose, Alces alces. Wildlife Biology, 9, 81-91.
Rea, R.V., Child, K.N., Spata, D.P. and MacDonald, D. 2010. Road and rail side vegetation management implications of habitat use by moose relative to brush cutting season. Environmental Management, 46, 101-109.
Rea, R.V., Johnson, C.J. and Emmons, S. 2014. Characterizing moose–vehicle collision hotspots in northern British Columbia. Journal of Fish and Wildlife Management, 5(1), 46-58.
Rea, R.V., Johnson, C.J., Aitken, D.A., Child, K.N and Hesse, G. 2018. Dash cam videos on YouTube™ offer insight into factors related to moose-vehicle collisions. Accident Analysis and Prevention, 118, 207-213.
Reed, D.F., Beck, T.D.I. and Woodard, T.N. 1982. Methods of reducing deer-vehicle accidents: benefit-cost analysis. Wildlife Society Bulletin, 10, 349-54.
Reed, D.F., Pojar, T.M. and Woodard, T.N. 1974. Use of one-way gates by mule deer. Journal of Wildlife Management, 38, 9-15.
Reed, D.F., Woodard, T.N. and Pojar, T.M. 1975. Behavioral response of mule deer to a highway underpass. Journal of Wildlife Management, 39, 361-67.
Reeve, A.F. and Anderson, S.H. 1993. Ineffectiveness of Swareflex Reflectors at reducing deer-vehicle collisions. Wildlife Society Bulletin, 21, 127-132.
Riley, S.J and Marcoux, A. 2006. Deer-vehicle collisions: an understanding of accident characteristics and drivers’ attitudes, awareness, and involvement. Research Report RC-1475. Michigan Department of Transportation, Department of Fisheries and Wildlife, Michigan State University, Lansing, Michigan, USA.
Roedenbeck, I. A., Fahrig, L., Findlay, C.S., Houlahan, J.E., Jaeger, J.A., Klar, N. Kramer-Schadt, S. and E.A. Van der Grift, E.A. 2007. The Rauischholzhausen agenda for road ecology. Ecology and Society, 12(1), 11.
Rogers, E. 2004. An ecological landscape study of deer-vehicle collisions in Kent County, Michigan. Report prepared for Kent County Road Commission, Grand Rapids, Michigan. White Water Associates, Amasa, Michigan, USA.
Rolandsen, C.M., Solberg, E.J., Van Moorter, B. and Strand, O. 2015. Dyrepåkjørsler påjernbanen i Norge 1991–2014. NINA Rapport 1145, 111.
Romin, L.A. and Dalton, L.B. 1992. Lack of response by mule deer to wildlife warning whistles. Wildlife Society Bulletin, 20, 382-384.
Romin, L.A. and Bissonette, J.A. 1996. Deer-vehicle collisions: status of state monitoring activities and mitigation efforts. Wildlife Society Bulletin, 24, 276-83.
Rondeau, D. and Conrad, J.M. 2003. Managing urban deer. American Journal of Agricultural Economics, 85, 266-281.
Rosell, C. et al. 2022. Wildlife and Traffic: Section 7 Solutions to reduce transport infrastructure impacts on wildlife. [This introductory section provides links to other sections]
Rutberg, A.T. and R.E. Naugle 2008. Deer-Vehicle collision trends at a suburban immunocontraception site. Human-Wildlife Conflicts, 2, 60-67.
Saint-Andrieux, C., Calenge, C., and Bonenfant, C. 2020. Comparison of environmental, biological and anthropogenic causes of wildlife–vehicle collisions among three large herbivore species. Population Ecology, 62, 64-79.
Sanders, W. 1985. Fallow Deer: A study of the Possible Factors influencing the Annual Cycle of Road Casualties. HND dissertation, Seal Hayne College, UK.
Savolainen, P. and Ghosh, I. 2008. Examination of Factors Affecting Driver Injury Severity in Michigan's Single-Vehicle—Deer Crashes. Transportation Research Board of the National Academies, Washington,D.C., No. 2078, pp. 17–25.
Schafer, J.A. and Penland, S.T. 1985. Effectiveness of Swareflex reflectors in reducing deer-vehicle accidents. Journal of Wildlife Management, 49, 774-776.
Schalk, A., Aleksa, M. and Forstner, M. 2023. Wildlife Control 4.0 Networks (WiConNET) Ergebnissbericht. Ein Projekt finanziert im Rahmen der Verkehrsinfrastrukturforschung 2016 (VIF 2016). Bundesministerium für Klimaschutz Abteilung Mobilitäts- und Verkehrstechnologien, Vienna.
Sharafsaleh, M., Huijser, M., Nowakowski, C., Greenwood, M.C., Hayden, L., Felder J. and Wang, M. 2012. Evaluation of an animal warning system effectiveness. Phase Two – final report, California PATH research report, UCB-ITS-PRR-2012-12. California PATH Program, University of California at Berkeley, Berkeley, California, USA.
Scheifele, M. P., Browning, D.G. and Scheifele, L.M. 1998. Measurement of several types of ''deer whistles'' for motor vehicles: Frequencies, levels, and animal threshold responses. The Journal of the Acoustical Society of America, 104(3), 1811.
Scheifele, P M., Browning D.G. and Collins-Scheifele, L.M. 2003. Analysis and effectiveness of deer whistles for motor vehicles: frequencies, levels, and animal threshold responses. Acoustics Research Letters Online, 4(3), 71-76.
Schober,F, and Sommer, F. 1984. Untersuchungen akustischer Wildwarngeräten für Kraftfahrzeuge. Zeitschrift für Jagdwissenschaft, 30, 164-176.
Schwabe, K.A., Schuhmann, P.W. and Tonkovich, M. 2002. A dynamic exercise in reducing deer-vehicle collisions: management through vehicle mitigation techniques and hunting. Journal of Agricultural and Resource Economics, 27, 261-80.
Seiler, A. 2004. Trends and spatial pattern in ungulate-vehicle collisions in Sweden. Wildlife Biology, 10, 301-313.
Seiler, A. 2005. Predicting locations of moose-vehicle collisions in Sweden. Journal of Applied Ecology, 42, 371-382.
Seiler, A. and Olson, M. 2017. Wildlife deterrent methods for railways – an experimental study. In: L. Borda-de-Água, R. Barrientos, P. Beja and H.M. Pereira (eds.). Railway Ecology pp.277–291.
Seiler, A., Olsson, M., Rosell, C. and van der Grift, E. 2016. Cost-benefit analyses for wildlife and traffic safety. CEDR Transnational Road Research Programme Call 2013: Roads and Wildlife Technical report No. 4
SETRA, 1998. Collisions véhicules-grands mammifères sauvages - Evolutions des inventaires de 1984-1986 et 1993-1994. Note d'information n° 60. SETRA, Bagneux.
SETRA and MATE, 1993. Passages pour la grande faune, Guide technique. Service d'Etudes Techniques des Routes et Autoroutes, Bagneux. pp. 121.
SGS Environment, 1998. The Prevention of Wildlife Casualties on Roads Through the Use of Deterrents: Prevention of casualties among deer populations. Report to UK Highways Agency SW335/V3/11-98
Sheets, R. and Cason, T. 2005. What deer see and hear. University of Georgia Research Magazine. Georgia.
Sivertsen, T.R., Gundersen, H., Rolandsen, C.M., Andreassen, H.P., Hanssen, F., Hanssen, H.G. and Lykkja, O. 2010. Evaluering av tiltak for å reducere elgpåkjörsler på veg, Høgskolen i Hedmark, Oppdragsrapport nr.1 - 2010.
Stanley, L., Hardy, A. and Lassacher, S. 2006. Responses to enhanced wildlife advisories in a simulated environment. Transportation Research Record, 1980, 126-133.
Staines, B.W., Langbein, J. and Putman, R.J. 2001. Road Traffic Accidents and Deer in Scotland. Report to the Deer Commission, Scotland.
Strein, M. 2010. Restoring permeability of roads for wildlife: wildlife warning systems in practice. Programme and book of abstracts. IENE International Conference on Ecology and Transportation. 27 September–1 October 2010, Velence, Hungary, p 77.
Sudharsan, K., Riley, S.J. and Winerstein, S.R. 2006. Relationships of autumn hunting season to the frequency of deer–vehicle collisions in Michigan. Journal of Wildlife Management, 70, 1161-1164.
Sullivan, T. A., Williams, A.F., Messmer, T.A., Hellinga, L.A. and Kyrychenko, S.Y. 2004. Effectiveness of temporary warning signs in reducing deer–vehicle collisions during mule deer migrations. Wildlife Society Bulletin, 32, 907-915.
Suter, S., Reifler-Bächtiger, M., Koch, T., Stephani, A., Sigrist, B., Graf, R.F., Laube, P., Ratnaweera, N., Kaelin, I., Wróbel, A. 2021. Prävention von Wildtierunfällen auf Strassenverkehrs-infrastrukturen (Prevention of wildlife-vehicle collisions on road infrastructure). Forschungsprojekt VSS 2015/212. Bundesamt für Strassen, Switzerland.
Ückermann, E. 1964. Erhebung uber die Wildverluste durch den Strassenverkehr und die Verkehrsunfälle durch Wild. Zeitschrift für Jagdwissenschaft, 10, 142-168.
Ückermann, E. 1983. Die Auswirkung der Wildverluste durch den Stassen Verkehr auf die Nutzungsfähigkeit der Reviere. Zeitschrift für Jagdwissenschaft, 29, 264-265.
Ujvari, M, Baagoe, H.J. and Madsen, A.B. 1998. Effectiveness of wildlife warning reflectors in reducing deer-vehicle collisions: a behavioral study. Journal of Wildlife Management, 62, 1094-1099.
Uzal, A. 2013. Reported deer road casualties and related accidents in England 2003-2010: their potential to develop an index of deer density. Report to the Deer Initiative, Wrexham, UK.
Valitzski, S.A., D'Angelo, G.J., Gallagher, G.R., Osborn, D.A., Miller, K.V. and Warren, R.J. 2009. Deer responses to sounds from a vehicle-mounted sound-production system. Journal of Wildlife Management, 73(7), 1072-1076.
VerCauteren, K.C. and Pipas, M.J. 2003. A review of the color vision in white tailed deer, Wildlife Society Bulletin, 31(3), 884-691.
VerCauteren, K. C., Gilsdorf, J.M., Hygnstrom, S.E., Fioranelli, P.B., Wilson, J.A. and Barras, S. 2006. Green and blue lasers are ineffective for dispersing deer at night. Wildlife Society Bulletin, 34, 371–374.
Völk, F., Glitzner, I. and Wöss, M. 2001. Kostenreduktion bei Grünbrücken durch deren rationellen Einsatz. Kriterien – Indikatoren – Mindesstandards. Bundesminiterium für Verkehr, Innovation, Technologie u. Stassenforschung, (Austria). Heft 513, Wien, 96 S.
Voss, H. 2007. Unfallhäufungen mit Wildunfällen - Modellversuche im Oberbergischen Kreis. [In German] Gesamtverband der Deutschen Versicherungswirtschaft e. V. Unfallforschung der Versicherer, Berlin.
Wanvik, P.O. 2009. Effects of road lighting: An analysis based on Dutch accident statistics 1987-2006. Accident Analysis and Prevention, 41, 123-128.
Ward, A.L. 1982. Mule deer behavior in relation to fencing and underpasses on Interstate 80 in Wyoming. Transportation Research Record, 859, 8-13.
Waring, G.H., Griffis, J.L. and Vaughn, M.E. 1991. White-tailed deer roadside behavior, wildlife warning reflectors and highway mortality. Applied Animal Behaviour Science, 29, 215-223.
Witmer, G.W. and de Calesta, D.S. 1992. The need and difficulty of bringing the Pennsylvania deer herd under control. Proceedings of the Eastern Wildlife Damage Control Conference, 130-137.
Woodard, T.N., Reed, D.F. and Pojar, T.M. 1973. Effectiveness of Swareflex wildlife warning reflectors in reducing deer-vehicle accidents. Internal Report, Colorado Division of Wildlife, Fort Collins, Colorado.
Wu, E. 1998. Economic analysis of deer-vehicle collisions in Ohio. In the Proceedings of International conference on wildlife ecology and transportation, Fort Myers, Florida, pp. 43-52.
Zacks, J.L. 1986. Do White-tailed deer Avoid Red: An evaluation of the premise underlying the design of Swareflex wildlife reflectors. Transportation Research Record 1025. Transportation Research Board, National Research Council, Washington, DC.