A Comparison of Noninvasive Techniques to Survey Carnivore Communities in Northeastern North America

1. What they did

Carnivores are hard to survey because they are rare across the landscape and wary of people. Noninvasive methods avoid catching and handling animals, but little had been written about how well each method works for each species. The authors compared five commonly used methods, using data from two studies of carnivore communities in New York State.

• Adirondack State Park. At 54 sites they marked a 5 km transect along trails and unpaved roads. Along each they ran 3 baited camera traps for about a month and 6 baited track-plates for about two weeks. A track-plate is a soot-covered aluminium plate inside a lightweight plastic housing, with contact paper at the baited end to collect footprints. They also cleared each transect of scats, walked it once a month for 3 summer months to collect all suspected carnivore scats (except bear), and identified the species from the DNA in 472 scats.
• Albany Pine Bush Preserve. In summer 2001 they set scent stations: a scent tablet in the middle of a 1 m radius circle of raked sand, each watched by a camera, run for one night at each of 22 forest sites. In winter, a team of 2 to 3 biologists trained in snowtracking searched a 1 hectare plot at each of 21 sites the morning after a full night of fresh snowfall, 4 times over 2 winters.
• Comparing methods. They used two measures: latency to initial detection (LTD), the number of survey nights before a species was first detected at a site, and probability of detection (POD), the chance that a method detects a species where it is present.

2. What they found

Adirondacks: cameras and track-plates

• Together the two methods detected 10 species. Fisher was found at the most sites (61%) and gray fox at the fewest (2%). Bobcat, mink and otter, known in the region, were not detected by either.
• Of the 5 most common species detected by both methods, 4 were recorded on more transects by track-plates than by cameras, although the differences were not significant.
• Cameras detected species sooner. Weasels and marten were photographed after about 9 to 15 survey nights, while coyotes took the longest (40.8 survey nights). For track-plates there was a long lag before fishers and raccoons stepped on the plates.
• Given enough time, track-plates had a higher POD than cameras for fishers and raccoons. Marten and weasels were photographed too rarely to calculate a POD, which suggests track-plates work better for these small species.
• Cameras detected black bears quickly and at 61% of sites. Coyotes had the lowest camera POD of any species. Coyotes and bears were too large for the enclosed track-plates.

Adirondacks: scats and DNA

• DNA confirmed the species for 377 of 472 scats (79.9%); 88.9% were coyote.
• Coyotes were confirmed by DNA at 28 of 35 sites, but cameras detected them at only 12 of these.
• Scat DNA also revealed red fox, marten, mink, fisher, gray fox, raccoon and domestic dog. Cameras and track-plates rarely or never detected mink, gray fox or red fox.
• The largest non-coyote scat was a 22 mm red fox scat. Treating scats over 22 mm as coyote would have misidentified only one scat and missed coyotes on only one trail. By this rule, 86.4% of the 54 sites had coyotes.
• The number of coyote scats on a transect was not related to how quickly or whether cameras detected coyotes.

Albany Pine Bush: snowtracking, cameras and scent stations

• The methods detected 9 species, including domestic cats and dogs.
• Snowtracking found coyotes, fishers and gray foxes at more sites; this was significant only for coyotes (55% of sites by snowtracking, 5% by cameras or scent stations).
• Raccoons and skunks were detected only in summer.
• Where cameras watched the scent stations, tracks were missing in 17.0% of carnivore detections and photos in only 1.9%. Animals came close enough to trigger the camera without stepping on the sand. A sand circle of only 50 cm radius would have missed 12.5% of raccoons, 4.5% of cats and the single coyote track.
• Snowtracking had a high POD for coyotes, fishers and domestic cats.

3. What they concluded

• No single method is ideal for all carnivore species. Cameras detected the most species but did poorly for coyotes and small species; animals may approach baited stations but be wary of stepping on tracking surfaces.
• Coyotes: survey with scat surveys, ideally backed by DNA, or with snowtracking. In their region a 22 mm size rule could work without DNA, but probably only for the large northeastern coyotes.
• Black bears: cameras worked well.
• Fishers, marten, raccoons, weasels, cats and opossums: cameras for quick surveys (under a week); track-plates may be slightly more efficient when there is more time.
• Foxes and mink: cameras and track-plates are inefficient for small canids here; scat DNA and snowtracking picked up foxes where other methods failed.
• Surveys of about 2 weeks detect most species, about a month is needed for a thorough inventory, and surveys under a week are not advised.
• Track-plates cost under 5% of a camera (cameras were about US$200 each in 2005) but must be checked every 2 to 3 days.
• Latency to detection might index relative abundance for some species, but not for coyotes. Snowtrack counts on small plots measure local activity rather than density.
• Snowtracking is robust for species active in winter, but snow may be suitable for only a few days, so crews must be ready at short notice, and weather affects how much animals move.
• They recommend using several independent methods and testing their relative value.

4. How tracking was or was not used

Tracking was central to four of the five methods:

• Track-plates: footprints of visiting animals were recorded on sooted plates and contact paper and identified to species. Ermine and long-tailed weasel tracks could not always be told apart, so they were combined. Marten and weasel tracks were compared with tracks from another track-plate survey to rule out mink.
• Scent stations: footprints in a circle of raked sandy soil, with the distance of tracks from the centre measured to test how big the tracking surface needs to be.
• Snowtracking: trained biologists searched 1 hectare plots for tracks after fresh snowfall.
• Scat surveys: scats were collected along trails and their species identified by DNA. This also tested identification by scat size (the 22 mm rule).

No trailing was reported. The authors also saw bobcat tracks along the trails, although no survey method detected bobcats.

5. Tracker or observer certification

Reported but not validated. The snowtracking was done by “a team of 2–3 biologists trained in snowtracking”; their training and skill were not described or tested. Nothing is said about who identified tracks on the track-plates and scent stations.

6. Other technologies, skills or innovations

• Infrared, motion-triggered still cameras (mainly Camtracker II) with bait and a skunk-scented lure.
• Enclosed track-plates in corrugated plastic (Coroplast) housing: lightweight, cheap, foldable and more durable than plywood boxes. Soot was applied with a kerosene lantern made from a paint can.
• Fatty acid scent tablets on raked sand.
• Fecal DNA: mitochondrial DNA primers to identify coyote scats, with other scats sequenced and matched to GenBank or museum tissue samples.
• The program PRESENCE for probability of detection.

7. Implications for conservation (from the paper)

• Cameras, track-plates and snowtracking all have potential for surveying whole carnivore communities, but no single method surveys all species at once.
• Cameras probably miss smaller species and wary larger ones such as coyotes; track-plates suit only small and medium species and also miss some wary species; snowtracking overcomes wariness but only works for species active in winter and needs care to compare sites.
• As noninvasive surveys become more common, it is more important to document their limits and biases, and to combine several independent methods.