# Habitat, Highway Features, and Animal-Vehicle Collision Locations as Indicators of Wildlife Crossing Hotspots

> Where do wild animals really cross the road? Tracking along 41 km of New Hampshire highways in one winter and spring turned up over 7000 track records of 22 species, from moose and deer to fisher, bobcat and a Canada lynx. Moose, deer and predators each chose different crossing places, and moose crossings did not match where moose had been hit by cars.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2007-barnum-rinehart-habitat-highway-features-animal/

## Citation and access

- **Authors:** Sarah Barnum, Kurt Rinehart and Mark Elbroch
- **Year published:** 2007
- **Journal or source:** Proceedings of the 2007 International Conference on Ecology and Transportation (ICOET 2007)
- **Full citation:** Barnum, S., Rinehart, K. and Elbroch, M. 2007. Habitat, highway features, and animal-vehicle collision locations as indicators of wildlife crossing hotspots. In: Irwin, C. L., Nelson, D. and McDermott, K. P. (eds.), Proceedings of the 2007 International Conference on Ecology and Transportation, pp. 511-518. Center for Transportation and the Environment, North Carolina State University, Raleigh, NC.
- **Access status:** Hosted (freely available elsewhere)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/2007_Barnum_and_Rinehart_etal_habitat-highway-features-and-collision-locations-as-indicators-of-wildlife-crossing-hotspots.pdf

## Tags

- **Species:** Mammals > American black bear; Mammals > American marten; Mammals > Bobcat; Mammals > Canada lynx; Mammals > Coyote; Mammals > Domestic cat; Mammals > Domestic dog; Mammals > Fisher; Mammals > Grey fox; Mammals > Long-tailed weasel; Mammals > Mink; Mammals > Moose; Mammals > North American porcupine; Mammals > North American river otter; Mammals > Raccoon; Mammals > Red fox; Mammals > Snowshoe hare; Mammals > Stoat; Mammals > Striped skunk; Mammals > White-tailed deer; Birds > Wild turkey
- **Region:** North America > United States
- **Tracking:** Snow tracks; Spoor age estimation; Track counts; Tracking; Tracks; Tracks and signs identification
- **Behaviour and ecology:** Behaviour; Connectivity; Habitat selection; Movement; Road crossings; Wildlife corridors
- **Methods and technology:** Chi-square test; Descriptive statistics; Geographic Information System (GIS); Global Positioning System (GPS); Linear regression; Pearson correlation
- **Data type:** Existing data; New field data
- **Observer qualification:** Reported but not validated
- **Discipline:** Behaviour and ecology; Wildlife tracking
- **Year published:** 2007

## Detailed keywords

- **Species:** Studied (three analysis groups): moose (Alces alces), white-tailed deer (Odocoileus virginianus, spelled “Odocoilius” in the paper), and “wide-ranging predators”: coyote (Canis latrans), fisher (Martes pennanti) and bobcat (Lynx rufus). Red fox (Vulpes vulpes) was recorded often but left out of the analysis because they seemed to live along the roadside. Also recorded at the roadside (22 species in all): gray fox (Urocyon cinereoargenteus), snowshoe hare (Lepus americanus), mink, long-tailed weasel (Mustela frenata), ermine or short-tailed weasel (M. erminea), raccoon (Procyon lotor), black bear, turkey, striped skunk, squirrels, American marten (Martes americana), porcupine (Erethizon dorsatum), otter (Lutra candensis, as spelled in the paper), Canada lynx (a single track record), domestic cats and dogs, and one unidentified weasel.
- **Location and habitat:** 31 km (19 miles) of US Route 2 and 10 km (6 miles) of NH 115 in the towns of Jefferson and Randolph, White Mountain Region, New Hampshire, United States. Valley bottoms at about 300 to 500 m elevation, mostly forested (deciduous, coniferous and mixed forest), with pasture, hay fields, a golf course, low-density homes and protected lands. Three sub-areas: Route 2 west (13.5 km, most open and most developed), NH 115 (9.8 km) and Route 2 east (17.9 km, most deciduous forest and steepest slopes). Unfenced, at-grade two- and three-lane highways carrying about 3000 to 5000 vehicles a day.
- **Topics:** road ecology; wildlife highway crossings; crossing hotspots; animal-vehicle collisions (AVC) and moose-vehicle collisions (MVC) as indicators of crossing; habitat connectivity and linkages; guardrail ends; roadside wetlands; roadside as habitat (red fox); crossing structures and highway planning; landscape, local and roadside scales; animal density index from sweep surveys.
- **Statistics or models:** Descriptive statistics; chi-squared test across sub-areas; Pearson correlation and two-sample t-tests to screen variables; backwards stepwise multiple regression (Systat 11) on 52 overlapping 500 m buffers around 700 m road segments; log and inverse transformations; diversity indices for cover, slope and parcel density. Final models: moose (guardrail ends plus mixed forest, r = 0.650), deer (open cover, r = 0.479), wide-ranging predators (cover diversity, r = 0.488). Inconsistency: in Table 5 the Rt2E and Rt2W column headings appear swapped. The table shows the most moose and fewest deer on Rt2W, but the text says moose and predators were recorded far less often than expected on Rt2W, deer far more often, and that more than twice as many deer crossed Rt2W as the other two sub-areas together. Other small slips: the cover page title reads “ad Indicators or” instead of “as Indicators of”; the moose model variable is “MIXED” in the text but “MIX” in Table 1; the high parcel density variable is “PDMXbin” in Table 1 and “MXPDbin” in the text; the variable screening threshold is printed as r > 0.0300. The paper is written in the first person singular (“I”) although it has three authors.
- **Data type:** New field data: roadside track surveys, December 2005 to May 2006 (most data collected on snow, December 2005 to 26 March 2006), plus 19 sweep transects in nearby habitat. Existing data: moose-vehicle collisions reported to NH Fish and Game, 1984 to 2004; National Land Cover Data, Digital Elevation Models, town parcel maps and New Hampshire Department of Transportation (NHDOT) road data. Roadside-scale analysis was not completed for this report.

## Synopsis

### 1. What they did

Highway planners need to know where wild animals cross roads, so they can protect habitat links with crossing structures or better road design. Most studies have used animal-vehicle collision (AVC) records for this, but these are imprecise and mostly record deer, elk and moose, since collisions are usually reported only when the vehicle is disabled. A collision site may also be where animals come to the road for food, or simply a dangerous spot, rather than where they prefer to cross. The authors argue that tracking can record many more species, and their behaviour, at the roadside.

- Tracking the roadside (December 2005 to May 2006). A two-person crew searched 41 km of US Route 2 and NH 115 in New Hampshire for tracks. On snow they drove slowly (10 mph or less) and could cover the whole study area in a day; after the snow melted they walked it, which took three days, looking mainly for larger tracks in mud, sand and short vegetation. For each track they logged the location with a handheld Global Positioning System (GPS) data logger, the species, number of animals, what the animal did (crossed, entered, exited, turned away, or moved parallel to the road), the age of the track and sometimes the sex.

- Sweep surveys. To compare the roadside with the surrounding habitat, an observer walked 19 transects away from the road and recorded all tracks.

- Mapping habitat and road features. Using a Geographic Information System (GIS), they measured land cover, slope, housing (parcel) density, pavement width, and the locations of guardrail ends, bridges and moose-vehicle collisions.

- Analysis. They grouped animals into moose, deer and “wide-ranging predators” (coyote, fisher and bobcat). They compared the three road sub-areas, then used multiple regression on 52 overlapping circles (500 m radius) along the road to find which habitat and road features went with more crossings.

### 2. What they found

Tracks recorded

- 7111 track records of 22 species along the roadside. Moose left the most (2709), then red fox (1939), white-tailed deer (1081) and coyote (488). Fisher (119) and bobcat (30) were also recorded, and one Canada lynx.

- 4414 records were road crossings; the rest were animals entering, exiting, moving along or turning away from the road.

- Species differed in how much they used the roadside compared with nearby habitat. Red fox made up 27% of roadside records but only 7% of sweep records, and seemed to use the roadside as their main habitat, so they were left out of further analysis. Snowshoe hare, bobcat and fisher were much more common in the sweeps than at the road.

Landscape scale

- No species group was spread along the road as expected from road length alone. Deer were concentrated along the more open Route 2 west, while moose and predators were recorded there less often than expected.

- Deer crossed the road most times they approached it. Moose crossed only just over half the time along NH 115 and Route 2 east, where forest borders the road, and coyotes were also less likely to cross there. The authors suggest the roadside offers these animals resources in those settings: local experts and residents say moose often use wet areas beside the road, some created when the road was built.

Local scale

- Moose: more crossings where guardrails end and where there was more mixed forest. Guardrails usually sit along steep embankments, which moose may avoid, and their ends may coincide with small roadside wetlands too small to show in the land cover data.

- Moose and collisions: moose crossing rates were not related to moose-vehicle collision locations. The collision records (1984 to 2004) were reported by many different people relative to landmarks, and lacked the precision of the tracking data.

- Deer: more crossings with more open cover, a sign of the edge habitat deer favour. This may reflect the deer’s preference for the open Route 2 west area as a whole.

- Wide-ranging predators: more crossings where cover was more varied, driven mainly by coniferous cover. Crossings tended to be lower where housing density was highest, though this was not significant.

### 3. What they concluded

- Tracking can provide a wealth of data on the presence and behaviour of wild animals at the roadside, and an excellent way to locate crossing hotspots that matter for habitat connectivity.

- The features linked to crossing differed for moose, deer and predators, so planners who want to keep connectivity for all species should consider each group’s needs separately.

- Many places were used by two species groups, and a few by all three. Planners could consider crossing areas suited to single species as well as to several, to achieve landscape-scale connectivity.

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

Tracking was the main method. The crew identified tracks to species along the road shoulders, first from a slowly moving vehicle on snow and later on foot in mud, sand and short vegetation, and recorded each animal’s behaviour at the road from their trail: crossing, entering, exiting, moving parallel or turning away. They also aged tracks and sometimes sexed them. Sweep transects in nearby habitat recorded all tracks to give an index of each species’ abundance away from the road. No attempt was made to tell individuals apart, so the same animal may have been counted more than once. By mid-May, roadside vegetation made most of the area too hard to track efficiently.

### 5. Tracker or observer certification

Reported but not validated. The two-person crew is described as “highly skilled”, but they are not named and no test of their skill is reported. The authors’ biographies state that co-author Mark Elbroch had earned a Senior Tracker Certificate in Kruger National Park, South Africa, but the paper does not say whether he was part of the field crew.

### 6. Other technologies, skills or innovations

- Trimble Geo Explorer III handheld GPS data logger, with data exported to Excel and ArcView.

- ArcView GIS with the Spatial Analyst extension and automated scripts to count track records and measure habitat inside each buffer.

- National Land Cover Data reclassified into seven cover types, Digital Elevation Models for slope, and town parcel maps for housing density.

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

- Identifying where animals cross highways is key to environmentally sensitive highway planning: designing roads to reduce their barrier effect, placing crossing structures, or avoiding building or upgrading roads in sensitive places.

- Collision-based studies miss many of the species that most need connectivity, especially predators, which are rarely reported in collisions and are quickly removed from roads as roadkill.

- Crossing hotspots should be planned for each species group as well as for places used by several groups.

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

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

- Roadside tracking on snow recorded over 7000 track records of 22 species in one season, including carnivores that collision records almost never capture. This shows how tracks and signs can fill gaps left by other data sources.

- Recording what each animal did at the road (crossing, turning away, moving along it) adds behaviour that collision or roadkill records cannot give.

- The paper notes the crew was highly skilled, but the reliability of identifications was not tested, so the strength of the findings depends on observer skill that is not documented.

## Cite this summary

Lawrence, K. 2026. Summary of: Habitat, Highway Features, and Animal-Vehicle Collision Locations as Indicators of Wildlife Crossing Hotspots (Barnum and Rinehart et al., 2007). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2007-barnum-rinehart-habitat-highway-features-animal/