Deer-vehicle collisions pose a sizable hazard in the United States. Numerous methods have been tried to reduce such crashes, often without scientific foundation or evaluation. A new Institute-sponsored review summarizes the various approaches, finding that some are effective at least in some situations. Some methods show promise, but more research is needed. And some methods simply do not work.
Big problem is getting bigger
"The problem is definitely growing," says the study's lead author, James H. Hedlund of Highway Safety North. "Populations of deer are increasing. It's not a problem that's going to go away." The best estimate is that more than 1.5 million deer motor vehicle crashes occur each year on U.S. roads. These collisions result in about 150 occupant deaths and more than $1 billion in vehicle damage. More precise data are hard to come by because collisions with deer often aren't reported to police. When they are, they're usually categorized along with collisions with horses, cattle, moose, elk, and other animals.
Insurers pay the costs of many deer collisions, but most companies don't code deer strikes separately under comprehensive losses. Erie Insurance Group is one company that does track deer claims separately and publishes results. The data, which come primarily from eastern states with large deer populations, estimate an average of 12 deer claims per 1,000 insured vehicles in 2002. This represents a 12 percent increase since 1998. Claims vary widely from state to state, but overall the rate of claims has increased every year except for a small decrease from 2001 to 2002.
The average cost of a deer claim in 2002 was $1,960. Such costs represented 38 percent of all comprehensive losses.
There are three general strategies to reduce deer-vehicle collisions. These include modifying driver behavior, modifying deer behavior, and reducing the number of deer. Each method reviewed in the study falls into one of these general strategies.
Fencing is effective
The only broadly accepted method of reducing deer collisions that's theoretically sound and proven effective is to install fencing, combined with underpasses and overpasses where appropriate. Fencing that's sufficiently high, long, strong, and anchored with no gaps or tunnels will prevent deer from crossing roads. "This approach certainly works," Hedlund says. But he adds that "it's expensive and can be intrusive."
Other measures show promise
Reducing the size of deer herds also will reduce collisions with vehicles. But this approach is controversial. There's public resistance to deer kills. There also are technical questions including how much herd reduction is necessary and over how wide an area a reduction must occur to reduce collisions with motor vehicles.
Establishing broad areas of cleared ground alongside roadways reduces the foliage that might attract deer toward the road. This approach also increases the likelihood that drivers will see deer approaching the road. More studies are needed to confirm these effects.
Displaying temporary signs during deer migration periods has been shown to reduce collisions by about half.
Another potentially promising approach involves signs that activate when deer are detected near a road. Detection methods include infrared light, radar, laser, radio frequency beams parallel to roads, and heat detection cameras. Little research is available on effectiveness.
Approaches with limited effects
For more than 30 years reflectors and mirrors have been used along roads in Europe and some U.S. states. The most common system, made by Swareflex, involves reflectors on posts installed at regular intervals along the roadside. Reflected light from vehicle headlights is thought to form a visual "fence" that deer aren't expected to cross.
More than 10 studies of this and similar systems yield conflicting results. The basic behavioral questions about reflectors are whether deer can see light in the wavelengths, whether deer are reluctant to cross such light beams, and whether deer become habituated to the light beams over time. The balance of research evidence indicates little in the way of long-term effects for this approach.
Research indicates that feeding deer at stations removed from roadways can be somewhat effective in keeping deer from crossing the roads (see "Deer, moose collisions with motor vehicles peak in spring and fall," April 3, 1993). But there are downsides including the continuing costs of this approach, the possibility of attracting more deer to the roadsides, and the likelihood that deer will become dependent on the food offered at the stations.
Limited studies have been conducted of the effects of repellents with unpleasant tastes applied to food and/or area repellents that smell unpleasant to deer. The research findings are mixed. Repellents can be effective in changing the patterns of deer feeding and movement, but they aren't likely to keep deer away from roads.
Methods that don't work
The most common approach to deer-vehicle collisions is to post permanent signs at fixed locations warning drivers of deer crossings. The locations of the signs don't vary throughout the year, regardless of whether collisions with deer are likely or not. The effects haven't been evaluated, but these passive signs are thought to be widely ignored.
The ineffectiveness of another approach has been proven. Deer whistles that attach to vehicles have been available for more than 20 years. The whistles produce ultrasonic noise (16 to 20 kHz) when a vehicle exceeds about 30 mph. The presumption is that deer will hear the noise and be warned away. It's unclear whether deer do hear the noise, but in any event studies show the whistles have no effect on deer behavior.
"People approach this hoping to find quick and easy solutions, but there aren't any. Whistles don't work," Hedlund says.
What would help
Better reporting of motor vehicle collisions with deer would help. It also would be useful to record the precise locations of the deer collisions to identify the problem areas.
"Now that we know some measures are effective and other measures show promise, we need better data to help decide where and how to apply these measures," Hedlund concludes.