Best Practices Manual: Wildlife Vehicle Collision Reduction Study

CHAPTER 6: GUIDELINES FOR MONITORING

This chapter provides guidelines for major elements of a monitoring/evaluation plan as well as some common pitfalls to be avoided. Of the dozens of mitigations listed in table 2, section 4.1, most are considered experimental in nature. Additional monitoring and evaluation is needed to continue to improve best practices and adequately assess the performance of the mitigation measures that are more experimental in nature, both with regard to WVC reduction as well as habitat connectivity for wildlife. A monitoring and evaluation program is the capstone of any WVC reduction program (figure 153).

Figure 153. Monitoring and evaluation as part of a strategy for reducing WVCs.

A plan for evaluating the effectiveness of mitigation measures should be developed well before the mitigation is implemented. One of the greatest challenges in executing an evaluation plan is lack of pre-installation data.

The mitigation plan should identify the goals of the mitigation. The goals of the mitigation measures typically have a safety and a habitat connectivity component:

• Safety-the desired level of reduction in WVCs.
• Habitat connectivity-the acceptable number of animal movements of the target species across or under the road.

Based on these goals the evaluation plan should state specific questions to be answered, such as:

• Were WVCs reduced, and if so, by how much?
• Did the potential safe crossing opportunities provide "sufficient" connectivity across or under the road for the target species?
• Were there negative side effects of the mitigation measure?
• What was the cost of construction and installation?
• What were the maintenance requirements?
• Were there specific modifications in the design of the mitigation measure to address the local situation or project goals and, if so, which modifications?
• What changes should be made when designing similar measures in the future?
• What are the perceptions of staff involved (e. g. , State DOT maintenance and resource agency staff) regarding the effectiveness, costs and benefits of the mitigation?

These types of questions can be used to define the specific parameters that must be measured to evaluate the performance of the mitigation, and decide on the specific methods that are to be used for the evaluation.

For the monitoring plan, it is important to consider the appropriate data study design, data collection protocols, and the required sample sizes. When designing the WVC or habitat connectivity research, be as specific as possible with regard to the research question, and make sure that the dependent variable (e. g. , number of WVCs or numbers of animal crossings of the target species) and the associated data collection and data analysis procedures allow for answering the research questions. If some data are available already from the same or a different location, consider conducting power analyses to make sure the potential sample sizes will be large enough to detect the differences or effects that might be expected or desired. Consider the empirical Bayes approach and BACI-analyses (Before-After-Control-Impact) as mitigation location selection will impact the needed sample size.

A critically important aspect of monitoring effectiveness is to publish the results, regardless of the outcome or thoroughness of the evaluation effort. Providing information that other agencies can use is extremely helpful in the continuing effort to reduce WVCs nationwide.

The remainder of this chapter discusses common data collection methods and some of the related mistakes and pitfalls commonly encountered therein.

Back to top

6.1. MEASURING REDUCTION IN WVCS

The most common method for evaluating effectiveness of mitigation measures on WVCs, is to compare the number of WVCs that occurred before and after implementation of the mitigation. Typically periods of three or five years of crash data before and after implementation are compared. Mistakes commonly made in these comparisons are discussed below.

• Simply comparing mean crash frequencies before and after the mitigation measure is installed to assess effectiveness: Typically evaluations of WVC mitigations are implemented at locations with the highest frequency of WVCs. This incorporates a selection bias, known as "regression to mean error," that could skew results. By contrast, this error would not occur if locations for mitigation are chosen randomly. Refer to Hauer for a methodology, known as the empirical Bayes method, to account for this error. ⁴⁷
• Assuming a change in crash frequency is solely the result of the mitigation that was implemented: The frequency of wildlife crossing the road obviously has a strong impact on the number of crashes. Animals may avoid the area because of construction, maintenance and/or monitoring activities associated with the implementation, a wider road and increased traffic volume, or changes in the landscape adjacent to the road, all of which can affect the number of attempts by wildlife to cross the road and, thus, the number of WVCs. Another important factor is fluctuating herd sizes; fewer or more WVCs may simply be related to lower or higher population density. There are several approaches to normalizing results to account for changes in animal movement:
  • Use control sites (comparison in space) as well before/after data from the same sites (comparison in time).
  • Use population size estimates as a measure of exposure.
  • Monitor animal movement in the area before and after the installation.
• Changing crash data collection methods after the mitigation is installed: It is important to compare apples to apples. For instance, crash reports should not be compared to carcass counts. Carcass search and reporting efforts are sometimes increased after the mitigation is installed when carcass counts are used in the before/after comparison. It is important that the same methods for collecting WVC data be used both before and after the installation. It is not essential to report every crash or carcass, but it is essential to have consistency in the search and reporting effort.

In the case of mitigation measures that are aimed at changing driver behavior (e. g. , alerting drivers to the potential hazard so they are more attentive), vehicle speeds can be collected as an indicator to numbers of WVCs. This approach may be desirable as more data can be collected (e. g. , thousands of vehicle speeds compared to a few WVCs per year). A statistically significant difference is easier to find if the sample size is larger. This method can be used to verify that the mitigation does, in fact, modify driver behavior. However, there is no agreed upon relationship between reduced driver speed and reduced WVCs, and reduced speed may not necessarily translate into reduced WVCs.

Back to top

6.2. MONITORING WILDLIFE MOVEMENT

The most common method for evaluating effectiveness measures on habitat connectivity is to simply measure the animal movements across the roadway. Monitoring wildlife use of crossing structures with fencing may reveal substantial wildlife movement across the roadway. However, without a reference or pre-stated goal, it is difficult to conclude whether the observed wildlife use of the crossing structures is sufficient to deem them effective. Examples of references that can be used for comparison are:

• The number of animal movements across the road before the mitigation measures were put in place.
• The number of animal movements in a comparable but undisturbed landscape, away from the road and beyond the zone that may be affected by roads and traffic.

Examples of references that can be used to determine habitat connectivity are:

• A certain minimum number of animal movements by target species over a certain time period. This may be a management goal only, not necessarily with a biological basis.
• Enough movement to maintain or restore viable populations of the target species on both sides of the road. Population viability analysis can provide insight into the minimum number of animal movements across or under the road over a certain time period in order to maintain or restore the population viability of selected species in an area.

Methods to monitor animal presence or movements include tracking beds, cameras, DNA sampling, GPS collars and trapping/tagging.

Tracking beds of sand, or ink beds, or other tracking media can be installed to measure the number of animal movements across a road. Tracking beds may be used before construction of the mitigation measures to provide a reference for the wildlife use of safe crossing opportunities. ⁴⁸ Tracking beds may also be used to measure wildlife use of the safe crossing opportunities. Depending on the medium, environmental conditions, and the number of animal movements, tracking beds may have to be checked daily, twice a week, or weekly. In northern locations, snow can be an inexpensive medium used for tracking beds.

Another method to document wildlife movements along roadways is the use of wildlife cameras. Camera sampling stations are usually placed in the study area along the road corridor, or at safe crossing opportunities. The range of the sensors, the delay between a detection and taking a picture, the potential disturbance of wildlife (e. g. , normal visible flash vs. infrared), and potential theft or vandalism are important factors to consider.

Counters, such as those that count animals when they break an infra-red beam, can be a surrogate for cameras. Adjusting the height of the counter may help focus the count on species of interest, but all species that trigger the counter will be pooled together, making it hard to relate the data to one or more specific species.

Non-invasive DNA sampling can be a practical method to distinguish between individual animals. For example, ten passages of a species through an underpass could represent ten passages of different individuals, or ten passages by the same individual. Minimum population size estimates can be acquired by also sampling DNA in a grid in the areas adjacent to the road. Strands of barbed wire are often used to collect hair samples from animals. A track bed, barbed-wire hair snag, and camera are shown in figure 154.

Figure 154. Tracking bed, photo camera and DNA hair snag (copyright: Marcel Huijser).

GPS collars and radio telemetry can be used to track individuals to investigate when and where they cross the roadway. This approach allows for detailed movement tracking, but only for the individuals that have been equipped with the tracking device.

Trap, tag and recapture methods (typically for smaller animals) can be used to identify individuals that cross the roadway. Individuals are captured at grid locations on each side of the roadway, tagged to identify the initial location, and recaptured sometime later. If an individual is recaptured at a location on the opposite side of the roadway, it is certain the individual crossed the roadway. Typically, life traps are used that are customized for the target species.

When collecting and analyzing wildlife movement data consider the following:

• With wildlife fencing in combination with wildlife crossing structures, it may take three to four years (ungulates) or five to seven years (grizzly bears) before the increase in wildlife use of crossing structures levels off, indicating that the animals have learned the location of the crossing structures and that they are safe to use. This does not necessarily mean that all species and all individuals use the crossing structures after seven years, but the use of the crossing structures is likely to have stabilized by then. Measuring wildlife use in the first years after completion of the crossing structures is likely to show an increase in use that represents a "learning curve," whereas measuring wildlife use five or more years after completion of the crossing structures is more likely to indicate the eventual normal level of wildlife use. Note that it may take longer for the increase in wildlife use to level off if relatively naive individuals are targeted that may not have had exposure to roads, traffic and mitigation measures, and may be unfamiliar with the location of the crossing opportunities (e. g. , those in seasonal migration or dispersal movements).
• Consider measuring variables that are likely to influence animal movements across or under the road over time. For example, changes in traffic volume, changes in human disturbance or changes in the landscape adjacent to the road, and changes in wildlife population sizes can influence the number of animal movements across or under a road. Thus potential changes in the number of animal movements across or under a road are not necessarily related to just the presence and performance of the mitigation measures.

For more detail on monitoring methods refer to Hardy and others. ⁴⁸