# Comparison of Methods of Monitoring Wildlife Crossing-Structures on Highways

> Are tracks in the sand or cameras the better way to tell which animals use highway underpasses and overpasses? In Canada’s Banff National Park, track-pads picked up more coyotes and grizzly bears, while cameras caught more deer and elk and became cheaper for studies over a year. The study also shows how busy, trampled track-pads lead to undercounting.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2009-ford-clevenger-comparison-methods-monitoring-wildlife/

## Citation and access

- **Authors:** Adam T. Ford, Anthony P. Clevenger and Andrew Bennett
- **Year published:** 2009
- **Journal or source:** Journal of Wildlife Management
- **Full citation:** Ford, A. T., Clevenger, A. P. and Bennett, A. 2009. Comparison of methods of monitoring wildlife crossing-structures on highways. Journal of Wildlife Management 73(7): 1213-1222.
- **DOI:** 10.2193/2008-387
- **Access status:** Publisher link only
- **Where to find it:** https://doi.org/10.2193/2008-387

## Tags

- **Species:** Mammals > American black bear; Mammals > Bighorn sheep; Mammals > Brown bear; Mammals > Coyote; Mammals > Domestic dog; Mammals > Elk; Mammals > Grey wolf; Mammals > Horse; Mammals > Moose; Mammals > Mountain lion; Mammals > Mule deer; Mammals > White-tailed deer
- **Region:** North America > Canada
- **Tracking:** Track counts; Tracking; Tracks; Tracks and signs identification
- **Behaviour and ecology:** Behaviour; Connectivity; Movement; Road crossings; Wildlife corridors
- **Methods and technology:** Akaike Information Criterion (AIC); Camera traps; Detectability; Logistic regression; Mann–Whitney U test; Survey costs; Wilcoxon matched pairs test
- **Data type:** Modelling or theoretical; New field data
- **Observer qualification:** Not reported
- **Discipline:** Behaviour and ecology; Wildlife tracking
- **Year published:** 2009

## Detailed keywords

- **Species:** Studied (large mammals coyote-sized and larger): grizzly bear (Ursus arctos), black bear (Ursus americanus), wolf (Canis lupus), coyote (Canis latrans), cougar (Puma concolor), deer (Odocoileus sp.; mule deer and white-tailed deer, whose tracks were almost impossible to tell apart on the track-pads), elk (Cervus elaphus), sheep (Ovis canadensis) and moose (Alces alces). Also recorded: humans (hikers, bikers and horseback riders), horses, a domestic dog, and tracks recorded only as “bear spp.”, “small canid” or “large canid”. Mentioned as rare species whose tracks could be mistaken for common ones: wolverine (Gulo gulo, for wolf), lynx (Lynx canadensis, for cougar) and mountain goat (Oreamnos americanus, for deer).
- **Location and habitat:** Bow River Valley, Trans-Canada Highway in and near Banff National Park, about 120 km west of Calgary, Alberta, Canada. About 45 km of fenced, four-lane highway (about 17,630 vehicles a day in 2007). 15 of the 24 crossing-structures in the area, of six designs: creek-bridge underpasses, an elliptical metal culvert, a round metal culvert, concrete box underpasses, open-span bridge underpasses and two 50-m-wide overpasses.
- **Topics:** road ecology; highway mitigation; wildlife crossing-structures (underpasses and overpasses); wildlife-proof fencing; landscape connectivity; comparing track-pads with motion-triggered cameras; species-specific detection bias; agreement between methods; track-pad condition (Good, Fair, Poor, Too many); track misidentification (sheep and moose tracks possibly made by deer and elk); undercounting when many animals cross; weather and structure design; cost-effectiveness of monitoring; human disturbance at crossings; hair-snagging barbed wire for genetic sampling (another study).
- **Statistics or models:** detection bias index, (track detections minus camera detections) divided by (track plus camera detections), from -1 (camera-biased) to 1 (track-biased); Wilcoxon signed-rank tests for each species (the Table 2 footnote calls this a Wilcoxon rank sums test); logistic regression of agreement between methods, with model selection by Akaike’s Information Criterion corrected for small samples (AICc); odds ratios; Mann–Whitney U test and Wilcoxon signed-rank test for counts by track-pad condition; two economic models (cost per animal detection and cost per unit time) over 3 months, 1, 4 and 10 years. Possible inconsistencies in the paper: (1) the text says human detections were 2.65 times more likely by camera, but Table 1 gives 26 human track detections and 165 by camera (a ratio of 0.16, or about 6 times); (2) the results give an example for a “4-month study period”, while the modelled durations were 3 months, 1, 4 and 10 years; (3) the track-pad threshold is given as under 100 crossing events per 3 months and the camera threshold as 1,000 or more crossing events per month; (4) Table 4 has no “Too many” column although the text reports 5.0% of checks as Too many; (5) the Figure 2 caption lists dollar values (US$43.36 herbivores, US$6.48 carnivores) that read as monthly detections and would make carnivores cheaper, while the text says carnivores were the most costly to monitor.
- **Data type:** New field data: track-pad and camera checks every other day, 29 June to 24 October 2007 (917 checks; 51 sampling days excluded). The track-pads had been checked continuously since 1996. Economic modelling from study costs.

## Synopsis

### 1. What they did

Highways kill wildlife and cut landscapes in two. Fences keep animals off the road, and crossing-structures (underpasses and overpasses) let them cross safely. To know whether these crossings work, managers need to know which species use them and how often. Two common ways to find out are track-pads (strips of soft soil that record footprints) and motion-triggered cameras, but no published study had compared them at crossings.

The authors compared the two methods at 15 crossing-structures on the Trans-Canada Highway in and near Banff National Park, Alberta, Canada:

- Track-pads. Each crossing had track-pads of dry, loamy sand, silt and clay, 1 to 4 cm deep, spanning its width. They were about 2 m long at underpasses and 4 m long at overpasses. Every other day, technicians recorded the species, numbers and direction of travel of large mammals (coyote-sized and larger) and people, rated the track-pad condition (Good, Fair, Poor or Too many), and raked the pads smooth.

- Cameras. In June 2007 they installed 17 infrared motion-triggered cameras, one at each underpass and two on each overpass, and downloaded the photos at each visit.

- Comparing the records. From 29 June to 24 October 2007, they measured which species each method was more likely to detect, and what made the two methods agree for a given visit.

- Costs. They built two economic models (cost per animal detected and cost per unit of time) for track-pads only, cameras only, and both together, over 3 months, 1 year, 4 years and 10 years.

### 2. What they found

Detections

- At least one animal used a crossing on 78% of 917 checks.

- Track-pads recorded 3,405 crossing events and cameras 4,430.

- Deer made up more than 65% of the crossing events.

- People were recorded far more often by camera than on the track-pads, probably because signs asked people to stay off the track-pads.

Which method detected which species

- Coyotes and grizzly bears were significantly more likely to be detected on track-pads.

- Elk and deer were significantly more likely to be detected by cameras.

- Wolves, sheep and moose leaned slightly towards track-pads, and black bears and cougars slightly towards cameras, but not significantly.

- The authors suggest coyotes are small and fast enough to slip past the camera sensor. They suggest the sheep and moose bias may come from technicians misidentifying deer tracks as sheep and elk tracks as moose. They had no explanation for the grizzly bear result.

When the two methods agreed

- The methods agreed on species, number and direction for only 32% of crossing events.

- Agreement fell as more animals crossed between checks, and rose with longer track-pads. Camera height and angle and the crossing’s design made no difference.

- For deer on a Good track-pad, lengthening the track-pad from 2 m to 10 m raised the chance of agreement from 0.53 to 0.62, while going from 1 to 8 crossing events between checks dropped it from 0.53 to 0.13.

Track-pad condition

- 83.9% of checks were rated Good, 4.6% Fair, 6.5% Poor and 5.0% Too many.

- Overpasses, open to the weather, had the most Poor conditions (21% of visits).

- Track-pads became “Too many” after about 18 camera-detected passages. Technicians tended to undercount: cameras recorded 19% more animals than track-pads on Good pads, and 68% more on “Too many” pads.

Costs

- In the short term, cameras cost US$3 to US$45 more per crossing event than track-pads, depending on the species.

- For studies longer than one year, cameras alone were always cheaper at the levels of animal activity seen.

- Carnivores were the most expensive to monitor because they crossed less often than ungulates (hoofed animals).

- Across a hypothetical range of activity, track-pads were cheapest for studies of 4 years or less with fewer than 100 crossing events per 3 months.

### 3. What they concluded

- Cameras and track-pads have complementary strengths. Track-pads are cheaper to start, more reliable for coyotes and grizzly bears, and at low risk of theft. Cameras give more confident species identification, cheaper long-term running costs, longer intervals between visits, and less sensitivity to weather and to heavy animal use.

- Track-pads have two main limits: similar tracks (mule deer and white-tailed deer, and rare species such as wolverine, lynx and mountain goat that could be recorded as wolves, cougars or deer), and data quality that drops as conditions worsen and animals trample the pad.

- Animal activity matters more for monitoring design than camera placement or track-pad design. More frequent visits reduce crossings per interval, but human presence can disturb wildlife.

- Neither method can tell individuals apart or show whether the crossings help populations persist. Methods such as capture-mark-recapture, genetic sampling, transponder tags and video can help with that.

### 4. How tracking was or was not used

Tracking was one of the two methods being tested. Technicians identified footprints of large mammals and people on raked track-pads every other day, counted the individuals, and recorded their direction of travel. Tracks that could not be identified to species were recorded as “bear spp.”, “small canid” or “large canid”. No trailing was reported.

### 5. Tracker or observer certification

Not reported. The track-pads were read by technicians, with no description of their experience, training or any test of their skill. This matters to the findings: the authors suggest some sheep and moose records may be technician misidentifications of deer and elk tracks. The cost model allowed 16 hours of technician training for the track-pad method, but this is a cost assumption, not a description of the study’s technicians.

### 6. Other technologies, skills or innovations

- Infrared motion-triggered cameras (Reconyx PM35M15) with an infrared light-emitting-diode (LED) flash for day and night use, taking five photos per trigger.

- Photos downloaded to a handheld computer (Hewlett-Packard iPAQ) and classified in a customized database form.

- A project-cost worksheet for planning monitoring, available from the corresponding author.

### 7. Implications for conservation (from the paper)

- Camera-based monitoring is more cost-effective in the long term and detects crossing events of most large mammals more efficiently, especially where animal activity is high.

- Tracking remains a viable way to monitor crossings, and the best choice depends on the study’s logistics and goals.

- For long-term track-pad monitoring, it is worth combining several detection methods, since in this study the long-term cost of using both was similar to track-pads alone.

- Where rare or threatened species use crossings alongside common species with similar tracks, methods that keep identification errors low are worthwhile.

### 8. Implications for tracking (Original Wisdom's observations)

_These are Original Wisdom's observations, not the authors'._

- Track-pads were outperformed partly on identification: similar tracks, possible misidentifications and undercounting on busy pads. Nothing is said about the technicians’ tracking skill, so the study cannot separate the limits of the method from the limits of the observers. Evaluating track-pad readers would make comparisons like this one easier to interpret.

## Cite this summary

Lawrence, K. 2026. Summary of: Comparison of Methods of Monitoring Wildlife Crossing-Structures on Highways (Ford and Clevenger et al., 2009). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2009-ford-clevenger-comparison-methods-monitoring-wildlife/