Vehicular collisions with moose Alces alces are currently a serious problem throughout much of the range of moose (Oosenbrug, Mercer & Ferguson 1991, Rattey & Turner 1991, Gundersen & Andreassen 1998). Collisions with moose and other ungulates appear to be on the rise worldwide (Groot Bruinderink & Hazebroek 1996) and have increased by more than 200% in some regions in less than a decade (Cook & Daggett 1995).

It is estimated that 29,000 humans are injured and 211 die annually in the US due to vehicular collisions with deer (the term deer in this work refers to members of the genus Odocoileus) alone (Conover, Pitt, Kessler, DuBow & Sanborn 1995). In France, approximately 50 people die and 2,500 are injured in ungulate-related vehicular collisions each year (Groot Bruinderink & Hazebroek 1996). In Sweden, 5–20 deaths and 500 injuries are reported each year as a direct result of moose-related vehicular collisions (MRVCs; Lavsund & Sandegren 1991). In northern New England, one in every 50 MRVCs results in a human fatality (Forman & Deblinger 1998).

Material damage claims following ungulate collisions cost billions of dollars each year; more than USD 50 million were spent on deer collision repairs in a single year in the state of New York alone (Decker, Loconti-Lee & Connelly 1990). The average cost for repairing vehicles can run from USD 4,000 per vehicle following a collision with a deer (Del Frate & Spraker 1991) to USD 15,150 per vehicle following a collision with a moose (Thomas 1995).

Wildlife-related vehicular collisions negatively impact animal numbers (Harrison, Hooper & Jacobson 1980, Cook & Daggett 1995, Thomas 1995) and are considered a long-term threat to populations of ungulates in certain areas (Jackson & Griffin 1998). In Newfoundland, Canada, approximately 4,800 moose roadkills were reported between 1988 and 1994 (Joyce & Mahoney 2001). These numbers are generally considered conservative because up to half of the ungulates killed by vehicles are never reported (Allen & McCullough 1976, Lavsund & Sandegren 1991); animals involved in collisions may wander from the corridor before dying (Moen 1979, Del Frate & Spraker 1991), are salvaged or scavenged (Child, Barry & Aitken 1991) or simply go undetected (Sielecki 2000). In some areas, collisions kill more ungulates than do hunters (Cook & Daggett 1995). In some parts of North America, roadkills are often reported as the chief cause of moose mortality second only to legal hunting (Del Frate & Spraker 1991) and may exceed 10% of the total annual harvest (Belant 1995). On a yearly basis, collisions with moose (automobiles and trains combined) claim approximately 6% of the annual allowable harvest nationwide in Canada (Child 1998).

Animal losses to road traffic can in part be attributed to the placement of human transportation corridors. These corridors tend to be routed through lowlands that follow the natural contours of the land (Thomas 1995) and often bisect or parallel prime habitat and natural routes traditionally used by ungulates and other wildlife for travel and migration (Andersen, Wiseth, Pedersen & Jaren 1991). Because of this overlap, road corridors are an integral part of many species' home range (Case 1978).

Roadsides often comprise remnants of natural vegetation in areas that tend to otherwise be heavily developed. Corridors provide islands and conduits of habitat for a variety of species and are used for feeding, breeding, nesting, dispersal and recolonisation (Bennett 1991). Some species rely exclusively on roadside habitat (Oetting & Cassell 1970, Way 1977). Roadside areas can also harbour feral animals and noxious weeds (Saunders & Hobbs 1991), creating a paradox for managers faced with the task of managing corridors with multiple objectives in mind (Bennett 1991).

Although reindeer Rangifer tarandus fennicus and caribou R. t. tarandus tend to avoid transportation corridors (Curatolo & Murphy 1986, Klein 1971), many ungulates, including moose (Kelsall & Simpson 1987, Thomas 1995), are known to use corridors for a variety of purposes (Table 1). For example, corridors may be used by ungulates for travel during periods of deep snow, but appear to be used predominantly for feeding (Peek & Beilis 1969, Puglisi, Lindzey & Beilis 1974, Groot Bruinderink & Hazebroek 1996).

Table 1.

Various corridor activities engaged in by ungulates.

Roadside forage

Ungulate activity in utility and transportation corridors increases in spring and autumn and appears to be linked to the utilisation of early greening and late senescing forages that are found in these areas (Harrison et al. 1980, Bashore, Tzilkowski & Beilis 1985, Kelsall & Simpson 1987, Lavsund & Sandegren 1991). These peaks in foraging activity correspond with those times of year when most collisions with moose and other ungulates occur (McDonald 1991, Gleason & Jenks 1993, Sutton 1996, Sielecki 2000). In general, clearings and corridors provide an abundant source of preferred foods for ungulates (Bédard, Crête & Audy 1978, Thompson & Stewart 1998, Finder, Roseberry & Woolf 1999) that are superior in nutritional quality (Hughes & Fahey 1991, Ricard & Doucet 1999) and more spatially concentrated than those found in adjacent woodlands (Carbaugh, Vaughan, Beilis & Graves 1975, Groot Bruinderink & Hazebroek 1996).

The quality and availability of browse along managed roadsides tend to remain relatively constant. This is largely due to roadside brush-cutting that is aimed at increasing sight lines and driver visibility by suppressing plant maturation and forest succession. Although this is done to increase road safety, this practice perpetuates the growth of early successional vegetation that is attractive to herbivores like moose. For this reason, highway transportation corridors have been described as long pastures bisected by highspeed lanes (Beilis & Graves 1971) and serve as foraging grounds for elk Cervus elaphus (H. Flygare, unpubl. data), mountain goats Oreamnos americanus (Leedy & Adams 1982), bighorn sheep Ovis canadensis (Harrison et al. 1980, Leedy & Adams 1982) , wild boar Sus scrofa (Groot Bruinderink & Hazebroek 1996), bison Bison bison (Damas & Smith 1983), deer (Puglisi et al. 1974, Carbaugh et al. 1975, Waring, Griffis & Vaughn 1991), moose (Kelsall & Simpson 1987, Child et al. 1991, Thomas 1995) and other herbivores (Arnold, Weeldenburg & Steven 1991, Bennett 1991).

Ungulates increase their foraging activities between dusk and dawn when they can move about under the protective cover of darkness (Peek & Beilis 1969, Carbaugh et al. 1975). Given that dark coloured animals such as moose are more difficult for motorists to see at night (Moen 1979, Thomas 1995, Sutton 1996), increased foraging activity and ungulate mobility between dusk and dawn are, not surprisingly, intimately tied to peaks in ungulate-related collisions (Carbaugh et al. 1975, Jaren, Andersen, Ulleberg, Pedersen & Wiseth 1991). Ungulate collisions appear to occur consistently between dusk and dawn regardless of the time of year or the ungulate population in question (Grenier 1973, Oosenbrug, McNeily, Mercer & Folinsbee 1986, Rattey & Turner 1991, Waring et al. 1991, Garrett & Conway 1999).

I reviewed the literature on patterns of ungulate- related collisions, plant response to tissue removal and vegetation management in transportation corridors as well as ungulate foraging behaviour. My objective was to elucidate new ways to manage roadside vegetation to reduce corridor attractiveness and moose utilisation of roadsides with an aim to reduce collisions with moose.

Countermeasures

A variety of countermeasures have been used in an attempt to reduce collisions with ungulates (Damas & Smith 1983). Many of these countermeasures, however, have proven ineffective. Deer reflectors, for example, are commonly installed on roadsides in an attempt to scare ungulates but have proven to be ineffective (see Groot Bruinderink & Hazebroek 1996) and cost USD 7,500 per km to install (Sielecki 2000). Exclusionary fencing is extremely effective at keeping ungulates out of transportation corridors but costs USD 45,000 per km to install. Furthermore, fencing is unsightly, requires frequent repair, and often prevents animals that make it into the corridor from escaping (Kent 1994, Sielecki 2000). In addition, the widespread use of fencing can greatly increase the fragmentation effect of transportation corridors on the movements of various species. On the other hand, managing corridor vegetation in a way that makes the corridor less attractive to species such as moose appears to be a more practical and promising tool for mitigation (Jaren et al. 1991, Lavsund & Sandegren 1991, Gundersen, Andreassen & Storaas 1998).

Planting unpalatable species within the corridor and luring animals away to strategically located feeding areas far from the road is an effective means of reducing wildlife collisions (Harrison et al. 1980, Cook & Daggett 1995, Romin & Bissonette 1996), as is completely eliminating palatable corridor brush such as birch Benda spp., poplar Populus spp. and willow Salix spp. (Jaren et al. 1991, Lavsund & Sandegren 1991). Unfortunately, these strategies are generally cost-prohibitive (Jaren et al. 1991, Sielecki 2000) and, in some cases, destroy habitat for other wildlife on a long-term basis (Oetting & Cassell 1970).

Manipulating the existing forage base within the corridor to produce low-quality browse may be a more cost- effective alternative for deterring feeding within the corridor (Sielecki 2000). Reducing the quality of roadside vegetation can be accomplished through applying noxious chemicals such as lithium chloride directly to the browse (Harrison et al. 1980). However, such strategies tend to be expensive and environmentally unsound. Although previously unreported, stimulating the growth of less palatable roadside browse through more carefully designed brush-cutting may prove less costly and equally, or more, effective.

Plant response to damage

It has long been established that mechanical damage to plants alters plant morphology, chemistry, the overall growth patterns and subsequently, the palatability of plant tissues for herbivores (Bryant, Danell, Provenza, Reichardt, Clausen & Werner 1991, Singer, Mark & Cates 1994). This type of response appears to have evolved as part of a generalised adaptive response against tissue removal by herbivores (Rhoades 1985, Bryant et al. 1991, Whitham, Maschinski, Larson & Paige 1991) but also occurs following other forms of stem breakage or tissue removal, including pruning, wind-breakage, snow press, ice scouring (Danell, Elmqvist, Ericson & Salomonson 1987), and brush-cutting (Oldemeyer & Regelin 1987, Nellemann 1990, Rea 1999). The morphology of current annual shoots (hereafter referred to as shoots) of broadleaf trees and shrubs often changes in response to damage. Plants generally respond to damage by producing large shoots (Willard & McKell 1978, Hjeljord & Grønvold 1988, Rea 1999) or by producing shoots that are more heavily armed (Gowda 1997). Depending on the intensity of damage, the overall architecture of the plant (tree-like vs hedge or shrub-like) may also be altered (Rea 1999).

Plants regenerating from mechanical damage also tend to produce shoots that are chemically different from the shoots of undamaged plants. Some woody browse plants, for example, produce shoots that contain higher concentrations of plant defensive compounds such as tannins, and are less digestible and contain lower concentrations of mineral elements following damage (Scotter 1980, Rhoades 1985), albeit plant chemical responses to damage vary significantly (Bryant, Wieland, Clausen & Kuropat 1985, Rhoades 1985, Singer et al. 1994).

Changes in the leafing phenology of plants also occur in response to mechanical damage. Plants can delay leaf senescence in the autumn and flush leaves earlier in the spring following damage relative to undamaged plants (Danell & Bergström 1985, Rea & Gillingham 2001). These changes alter the availability of leafy vegetation for herbivores at times of the year when nutritious plants are generally scarce (Renecker & Schwartz 1998).

Extensive research on plant response to damage (see Rhoades 1985, Bryant et al. 1991, Whitham et al. 1991) has shown that plant response varies with, among other things, the intensity, timing and frequency of damage (Danell & Bergström 1985, Whitham et al. 1991). For example, the timing of cutting (DeBell & Alford 1972, Harrington 1984, Kays & Canham 1991, Lepage, Pollack & Coates 1991) and clipping (Willard & McKell 1978, Bergström & Danell 1987a) stimulates plants to alter the morphology of browse shoots produced following damage. It has recently been concluded that the timing of browsing affects the chemistry of regenerating shoots and thus their palatability to ungulates (Alpe, Kingery & Mosley 1999), as does the timing of brush-cutting (Rea & Gillingham 2001).

Ungulate forage preferences and the corridor

Ungulates such as moose select browse based predominantly on quality (Thompson & Stewart 1998). Ungulates prefer browse plants that delay leaf senescence in the autumn and possess large shoots high in digestible energy and protein but low in plant defensive compounds (Bergström & Danell 1987b, Singer et al. 1994). Because late autumn and winter are times of nutritional deprivation for ungulates (Hobbs, Baker, Ellis & Swift 1981), roadside brush-cutting operations that inadvertently stimulate nutritious regrowth may act to increase the attractiveness of roadsides to moose. If corridors become more attractive to moose, roadside utilisation would tend to increase, as would the likelihood of collision.

Inarguably, other landscape features and animal behaviours influence ungulate use of areas such as roadsides (Treweek, Watt & Hambler 1997, Finder et al. 1999) and subsequently the frequency of collision. For example, collisions with moose often occur at distinct locations such as drainages (Thomas 1995) and the outlets of side valleys (Gundersen et al. 1998). The risk of ungulate collisions may also be greater near wooded, rather than open areas such as fields (Damas & Smith 1983). However, some authors report that deer collisions are randomly scattered within transportation corridors, with little concentration according to landscape features (Allen & McCullough 1976, Gleason & Jenks 1993). This suggests that other small-scale attributes such as browse diversity (R.V. Rea, unpubl. data) or other forage-based features of the corridor might influence animal activity.

Design features such as ditch depth and cut slope as well as corridor width may also influence how animals use the corridor (Kelsall & Simpson 1987, McGuire & Morrall 2000). Moose are particularly influenced by corridor width, for example, given that they predominantly use forest edges (Child 1998), and narrower corridors contain relatively more edge per cleared area (Bashore et al. 1985, Finder et al. 1999).

Driver visibility as well as the proximity of animals using the forest edge to the roadbed also varies with corridor width. Edge location in the corridor is generally considered fixed following corridor construction. Because it is not practical to relocate corridor edges, reducing browse attractiveness at the forest edge-corridor interface through post-construction vegetation management practices may be the only practical way to reduce the use of corridor edge by herbivores (Harrison et al. 1980, Damas & Smith 1983, Kelsall & Simpson 1987). Reducing the quality of forages growing near the corridor edge has been recommended by several authors studying the problem of ungulate-related vehicular collisions (Jaren et al. 1991, Cook & Daggett 1995, Ricard & Doucet 1999).

To date, studies on reducing the appeal of roadside for age for reducing ungulate collisions have primarily focused on the removal of browse from corridors. Cutting (Jaren et al. 1991, Lavsund & Sandegren 1991) and steam killing (Schwartz & Bartley 1991) vegetation within transportation corridors, for example, have proven effective (as much as a 56% reduction in train collisions; Jaren et al. 1991), but costly when practised repeatedly (Jaren et al. 1991, Sielecki 2000).

Cutting time as a countermeasure

Although several studies report the effects of the timing of cutting on shrub and tree regeneration, most have focused on how the physical and not the chemical characteristics of shoots and sprouts change following coppicing or silvicultural treatments (Belanger 1979, Kays & Canham 1991, Lepage et al. 1991, Babeux & Mauffette 1994). And while the nutritional quality of browse shoots is generally correlated with shoot morphology (Danell & Bergström 1985), this is not invariably true, particularly in the first two years after cutting when the effects of cutting time are considered (Rea & Gillingham 2001).

It is known that the quality of regenerating shoots of willow Salix scouleriana increases in the first two years after cutting when willows are cut during the middle of the growing season. Willows cut in mid-July produce shoots that, when collected in winter, are low in plant defensive compounds (tannin/lignin) and high in digestible energy and protein and delay leaf senescence into late autumn relative to plants cut at other times of the year and uncut controls (Rea & Gillingham 2001). These findings suggest that summer roadside brush-cutting operations could, inadvertently, be stimulating plants to produce nutritious regrowth that is attractive to moose.

Delays in leaf senescence due to roadside brush-cutting could alone be problematic where concerns for collisions with ungulates exist. Moose prefer greener vegetation (Bergerud & Manuel 1968, Hobbs et al. 1981) and, like other ungulates, will concentrate foraging efforts on leaves rather than shoots in autumn as long as leaves are available (Hobbs et al. 1981, Renecker & Schwartz 1998). Delayed leaf senescence in corridor plants could potentially extend the period of increased foraging activity and mobility that moose demonstrate when switching from decomposing summer forages to nutrient-rich browse shoots (Kelsall & Simpson 1987), thereby increasing their exposure to vehicular traffic. Similar problems are likely to occur in the spring given that ungulates are attracted to early-greening road side forages (Kelsall & Simpson 1987, Anderson 1991, Renecker & Schwartz 1998) and the timing of brush-cutting alters the timing of leaf flush in spring (Rea 1999).

Altering the timing of brush-cutting can stimulate the production of less nutritious browse by willow (Rea & Gillingham 2001). Cutting plants at a time that reduces plant quality could potentially discourage moose from foraging in the corridor and decrease the probability of collision. Brush-cutting in early June for example, results in the production of browse that is significantly less nutritious for the first two years after brush-cutting than browse produced by plants cut later in the growing season or by uncut controls (Rea & Gillingham 2001). Although it has yet to be tested, cutting immediately following leaf flush could result in the production of even lower quality regrowth. Plant resources flushed into newly expanding leaves would be lost to early cutting before photosynthesis could restore root reserves (Bryant et al. 1991, Kays & Canham 1991). Reduced nutrient stores weaken the plant's capacity for vegetative regrowth and the building of nutrient-rich shoots (Kays & Canham 1991). Plants cut earlier in the year are also less likely to delay leaf senescence when compared to later cutting dates that tend to promote delayed senescence for at least two years after brush-cutting (Rea & Gillingham 2001).

Recommendations

I recommend cutting brush in early spring shortly after woody plants have flushed their leaves. For reasons previously discussed, regrowth from this treatment regime should be lower in nutritional value and palatability for moose relative to plants cut in the middle of the growing season, when most roadside brush-cutting operations are currently carried out. The later in the season that plants are cut, the more likely it is that they will produce nutritious regrowth in the years following brush-cutting. Although regrowth from plants cut later (e.g. autumn) will not be available to moose in the first winter after brush-cutting and is not as nutritious as regrowth from plants cut in July in the second winter after cutting, such regrowth, when available, is more nutritious than regrowth from plants cut early in the year (Rea & Gillingham 2001). Based on my review of the literature, cutting from July to March is not recommended in areas where concerns for collisions with ungulates exist.

Cutting roadside brush in the early spring means that conventional, tractor brush-cutting practices may not be feasible to use. If the corridor is too wet and the ground too soft for tractors to be used, other techniques such as manual brush-cutting may be required. Using manual brush-cutting would not only allow brush management regardless of season but would also allow further experimentation with the height and angle of the stump cut, which is also known to alter plant response (Belanger 1979, Harrington 1984, Babeux & Mauffette 1994). Techniques such as girdling and torching permanently kill woody browse species (Olson, Macrigeanis & Davis 1981, Danell et al. 1987) and could also prove effective means, either alone or in combination with specific cutting times, for reducing the appeal of the roadsides to ungulates. Although ‘ecological side-effects’ should be considered prior to use, silvicultural herbicides may also prove useful in some situations where other techniques fail to reduce collisions with moose. The use of any or all of these alternatives as countermeasures should be applied across the entire width of the corridor section being treated (including highway medians) and should be closely monitored. This strategy will ensure that the efficacy of the treatment and its implications for road safety can be tested in isolation.

Practices such as cutting only tall-growing plants under corridor utility lines (pers. obs.) should be discouraged. Such practices may promote the growth of low- growing, palatable species in the corridor that must no longer compete with taller plants and can utilise nutrients from the decomposing slash (plant cuttings) of taller cut plants (Payne & Bryant 1998). Furthermore, because slash is attractive to ungulates (Alkon 1961, Renecker & Schwartz 1998), all slash should be mulched or removed from the corridor. Incidentally, similar measures should be considered when more mature vegetation is felled during corridor construction and widening given that the crowns of many tree species are attractive forage for moose (pers. obs.).

Although cutting brush in corridors more than once per season can be expensive, inhibiting regrowth through repeated brush-cutting may also prove feasible (Jaren et al. 1991) if limited to areas where ungulate collisions are recurrent, assuming such management does not simply displace moose to the next section of the corridor. It should be kept in mind, however, that the consequence of multiple cuttings can lead to carbon exhaustion of the plants being cut (DeBell & Alford 1972), killing shrubs and altering roadside plant composition and serai trajectories (Parr & Way 1988, Anderson & Katz 1993). Understanding the effects of repeated cuttings on corridor vegetation is relevant considering that browse diversity appears to influence the number of collisions per site (R.V. Rea, unpubl. data).

Currently, no information exists on changes in plant quality or moose foraging behaviour relative to the length of the vegetation control cycle (Ricard & Doucet 1999). Although the effects of brush-cutting on plant quality can last for at least five years (Rea 1999), precisely how long the effects of the timing of cutting on quality persist are unknown. Preliminarily, control cycles should be scheduled on a three-year rotation to test the effect of treatments because plants can reassume some characteristics of their pre-treatment growth form in as little as two to three growing seasons following brush-cutting (Rea 1999). Assessing plant response on a yearly basis could help to determine the long-term effects of brush-cutting on plant quality and help to determine how often roadside plants should be cut.

Regardless of the brush management strategy employed, corridor vegetation must be managed in a way that considers both the forage and non-forage values of the corridor for other organisms as well as moose. Even closely-related species of ungulates may respond to similar management strategies in different ways (Kent 1994), emphasizing the need to understand and manage for multiple values (Anderson 1991, Lautenschlager, Bell, Wagner & Reynolds 1998). This may mean concentrating brush management activities in certain sections of the corridor or within a specified distance from the road surface while employing current or alternative practices aimed at conserving other habitat values elsewhere in the corridor.

It must be remembered that these recommendations are based largely on mechanical brush-cutting operations that were tested in a conifer plantation setting. Plantation brush-cutting differs from roadside brush-cutting in two important ways. Firstly, during roadside cutting all plants are removed. In the plantation setting, however, conifers (and deciduous plants that are not in direct competition with conifers; Härkönen 1998) are left uncut and continue to grow, consuming surrounding resources. This makes nutrient acquisition easier for plants cut in plantations versus transportation corridors (Blair 1971) and may, therefore, in part determine the plants ability to compensate for damage. Secondly, although brush in plantations may be cut more than once before the conifers reach a free-to-grow stage, it is rarely cut more than two or three times. Roadside plants, alternatively, tend to be cut back on a regular basis for the life of the corridor. For these reasons, spring cuttings can be implemented but their effects should be tested using long-term monitoring programs to assess the quality of various browse species regenerating from cutting. Because ungulate food preferences and plant responses vary by both species and geographic area (Kelsall & Simpson 1987), indiscriminate implementation of these and future research findings to all possible management areas is not recommended and should be approached with caution.

Conclusions

Current vegetation management practices in transportation corridors are often based on operational and logistical constraints; roadsides are cut when the ground is dry and brush-cutting tractors can be used. Although these maintenance practices are aimed at increasing road safety, they may also inadvertently, create ideal foraging habitat for animals such as moose (Damas & Smith 1983) depending on the time of the year that vegetation management is performed (Fig. 1). Understanding the effects of these management activities in relation to plant response and ungulate behaviour should therefore be considered by agencies responsible for managing vegetation in and near transportation corridors (Cook & Daggett 1995, Romin & Bissonette 1996, Jackson & Griffin 1998). Several authors have suggested that highway authorities, state/provincial and federal agencies, insurance companies, conservation groups and industry must collaborate more closely on research that aims to reduce such collisions (Scotter 1980, Kent 1994, Cook & Daggett 1995, Child 1998) before impacts to animal populations, the danger to motorists and publie costs escalate further (Child et al. 1991, Groot Bruinderink & Hazebroek 1996, Thompson & Stewart 1998).

Figure 1.

Theoretical relationships developed for application to roadside vegetation management as a result of a review and synthesis of currently published works on ungulate-related vehicular collisions and plant response to the timing of cutting.

Finally, there will always be a risk of collision where moose and vehicles co-exist (Jaren et al. 1991) and no countermeasure, forage-based or otherwise, will ever completely eliminate MRVCs. However, even a small reduction in collision frequency substantially reduces societal costs and the deleterious effects on animal populations (Gleason & Jenks 1993). In this respect, management strategies aimed at reducing MRVCs can only provide positive returns and should, therefore, be viewed in terms of an investment for current and future generations of both humans and moose.