Testing GIS-generated Least-cost Path Predictions for Martes pennanti (Fisher) and its Application for Identifying Mammalian Road-crossings in Northern New Hampshire

1. What they did

Roads can cut animals off from the places they need, split populations, and kill animals in collisions. To decide where to put wildlife bridges, tunnels or overpasses, biologists often use computer maps to predict the “least-cost path” (LCP): the route an animal would find easiest to travel. But few studies have checked these predictions against what animals actually do.

The authors tested an LCP model for the fisher, a wide-ranging mesocarnivore that is sensitive to habitat fragmentation, along two roads in northern New Hampshire:

• Field data. Certified track and sign observers surveyed the roads for tracks on 51 days from December 2005 to May 2006, recording 7151 places where wildlife crossed or approached the roads.
• Building the model. They compared the land cover within 125 m of the 117 places where fishers crossed the road with the land cover available along the roads as a whole. Each land-cover type was scored by how much fishers seemed to prefer or avoid it, guided also by the authors’ expert opinion and published studies of fisher habitat. A Geographic Information System (GIS) then drew the easiest routes for fishers between three blocks of habitat on either side of the roads.
• Testing the model. They compared tracks on 11 road sections inside the predicted corridors with tracks on 11 randomly placed road sections outside them. They asked whether fishers, bobcats and American black bears were found more often inside the corridors, and whether more species used them.

2. What they found

Fisher land-cover preferences

• Fishers crossed more often where the roadside had conifers (10.3% of use, 6.1% available) or mixed forest (13.4% used, 8.7% available), and less often where it had hay or pasture (2.9% used, 11.7% available).
• Hardwoods and other cleared land were used in proportion to what was available. Wetlands were used slightly more than available.

Tracks recorded

• The analysis used 7099 tracks of 21 mammal species. The most common were moose (2590), red fox (1862), white-tailed deer (1157) and coyote (662). Fishers made 117 crossings, bobcats 32 and black bears 42.
• 1366 tracks (19%) fell within the corridors, including 35 of the fisher tracks (30%) and 14 of the bobcat tracks (44%).

Inside versus outside the corridors

• Fishers were found on 8 of the 11 road sections inside the corridors, and on 2 of the 11 outside them.
• Bobcats were found on 6 of the 11 sections inside, and on none outside.
• Black bears were found on 2 sections inside and 2 outside: no difference.
• Road sections inside the corridors had more species (an average of 7.36) than those outside (4.45).
• 15 of the 21 mammal species were found within the corridors.

3. What they concluded

• The fisher corridor model, tested with field tracking data, successfully identified roadside corridors for fishers and bobcats, but not for black bears.
• Corridors designed for the fisher also picked out places used by many other mammals. This may be because the fisher is sensitive to habitat fragmentation, or because the roadside cover fishers choose marks movement routes for a range of wildlife. Fishers and other mammals may also travel in conifers in winter, where snow is shallower and its crust harder, and bobcats may follow prey there.
• Black bears were dormant for most of the study and were sampled only briefly in spring. They may use the corridors more in other seasons, and the authors cannot say whether the patterns hold year-round.
• Uncertainties remain: the 2001 land-cover map may not match the land cover of 2005 and 2006; land cover was the only factor in the model; and terrain, which other studies use, should be explored.
• Multi-species corridors may prove better than single-species ones, but this study shows single-species LCP models are still a valuable tool in conservation planning.

4. How tracking was or was not used

Tracking was the basis of both building and testing the model. Observers drove slowly (under 16 km/h, 10 mph) along the roads, stopping often to record the species whenever tracks were within 10 m of the road. Surveys were done 48 to 72 hours after new snow, and the same observers repeated them, so old tracks were not counted twice. Bridges and culverts were checked every session. In spring, after the snow had gone, observers walked the whole road system twice a week to record black bear and moose tracks, erasing each track once recorded. No trailing is reported. The authors note that, unlike radio-telemetry, their snow-tracking method recorded fishers at the predicted corridor locations and also gave information on the other species crossing the roads.

5. Tracker or observer certification

The track and sign observers were tested in the field and certified by CyberTracker Conservation (citing Elbroch et al., then in press, and Evans et al. 2009). The level of certification and the number of observers are not given. Rose Graves is thanked for help with data collection.

6. Other technologies, skills or innovations

• CyberTracker data collection software on Global Positioning System (GPS) data loggers (TDS Recon and Trimble GeoExplorer) to record where tracks were found.
• The CorridorDesigner extension for ArcGIS 9.2 to draw least-cost corridors from a land-cover-based habitat suitability map.

7. Implications for conservation (from the paper)

• LCP analysis is widely used to choose places for road crossing structures, but its predictions need testing with real data. Here roadside track data confirmed the model for fishers and bobcats.
• Roadside habitats chosen by fishers may help identify and protect movement corridors for many species.
• The model could be tested further by predicting corridors elsewhere in northern New England and checking them with track data on several species.