Comparison of Proposed Survey Procedures for Detection of Forest Carnivores

1. What they did

American marten, fisher, wolverine and lynx are midsized forest carnivores thought to be threatened by the loss of old forests in the northwestern United States. Years of surveys had used many different methods, so results could not be compared across the region. In 1995 the U.S. Forest Service proposed standard survey procedures using remote cameras, sooted track plates and snowtracking. This study was the first critical test of those procedures.

• Study design. In the Bitterroot Mountains of Montana, the authors set up eight sample units of 10.36 km² each, two in each of four drainages, and tested each method in every unit.
• Remote cameras (winter). Two dual-sensor cameras (triggered by motion and heat) per unit, about 0.8 km apart, baited with deer quarters hung out of reach between trees and scented with trapping lure. Cameras ran from 30 November 1994 to 28 March 1995 and were checked every 4 to 7 days.
• Track plates (spring). Three open and three covered track plates per unit, alternating, about 0.54 km apart, baited with chicken and scented with lure. Plates were sooted on site with a portable acetylene torch and checked every other day for at least 12 days, from 24 April to 5 June 1995. The authors used a sturdier roof than the protocol suggested.
• Telling marten from fisher. They measured three toe pad dimensions from the original track prints (not photocopies) under a microscope with digital calipers, and applied a published formula (a discriminant function) that assigns each print to fisher or marten.
• Comparing methods. They ranked the two methods on five categories: species detected, time to first detection (latency to detection, LTD), ease of identification, effort to run, and cost.

They followed the protocol’s timing (cameras in winter, track plates in spring), so the two methods were not run at the same time.

2. What they found

Remote cameras

• Cameras recorded three of the four target species. Marten were photographed in all 8 units (average 13.5 days to first detection), fisher in 2 units (9.0 days) and a wolverine in 1 unit (13 days). No lynx were detected.
• Marten took much longer to show up in the first survey period (24.0 days) than in the second (3.0 days). The authors put this down to unusually deep, soft snow limiting marten movement.
• Equipment failed often. Battery failure alone caused 147 days (49%) of the down time. On average 18 extra days were needed to complete the survey, and cameras were working only 62% of the time.
• Most visits by target species were at night: 72% overall, 73% for marten and 100% for fisher; the wolverine came by day.

Track plates

• Track plates detected two species: marten in 6 of 8 units (75%) and fisher in 3 of 8 (38%).
• Marten were detected faster at track plates than at cameras (about 2 to 3 days), though in the 6 units where both methods found marten the difference was not significant.
• Heavy rain made open plates useless on 99 of 152 checks (65%). When they worked, open plates detected marten 26 times out of 53 (49%) and fisher twice (4%). Covered plates were not affected by rain and detected marten 28 times (18%) and fisher 14 times (9%) out of 152.
• Some detections came from other signs: five from marten scat left on plates, one from fresh marten tracks in snow with the bait gone, and one fisher from tracks near the plate with the bait gone.
• In both units where track plates missed marten, cameras had photographed marten; in one unit cameras photographed fishers that track plates missed.

Reading the tracks

• Only 22 of 26 marten prints, and 16 of 18 fisher print sets, were clear enough to measure. Overall, the authors found that 87% of tracks collected in the field were not good enough for a confident identification.
• One large fisher could be recognized as an individual by scars on two pads.
• Left and right feet of the same fisher differed so much in size that they could be mistaken for different animals.

Marten behaviour at covered plates

• After 2 days, 67% of open plates had been visited by marten, but only 11% of covered plates. By the end, the figures were 92% and 75%.
• On 36% of first visits to covered plates, marten left no print on the contact paper: they left scat at the entrance or stepped only on the sooted plate.
• No marten were recorded at covered plates where fishers were detected, although cameras photographed marten in those units before and after.

Overall ranking

Cameras ranked better for detection, identification and effort to run. Track plates ranked better for time to detection and cost. Overall, cameras ranked first.

3. What they concluded

• Cameras detected more species, made identification simpler (nearly 100% confident), and needed less effort, so they generally performed better. But the best method depends on each project’s goals, and track plates performed well for marten and fisher.
• Because the methods were run in different seasons, and marten and fisher behaviour may change between seasons, the results may not hold if both were run at the same time.
• The difference in latency between methods may be smaller than these results suggest.
• Marten are hesitant to enter covered plates on their first visit, and may avoid plates where fishers have been, which could reduce detection.
• The published method for telling marten and fisher tracks apart was tested on clear prints from adults; in the field, many prints are poor and young fishers overlap in size with marten, so some error is likely and its true size is unknown.
• Recommendations: use lighter, more reliable cameras (Trailmaster TM500, 0.8 kg, rather than the 13.6 kg Manley set-up); run camera surveys for 28 days with weekly checks, fresh batteries every visit and new film when half a roll is used; set cameras up for night visits; use open plates when the weather allows and both types together when it does not; measure original tracks rather than photocopies; and follow the standard protocols with these changes so results can be compared across regions.

4. How tracking was or was not used

Track plates are a way of collecting footprints: animals walk over a sooted plate and leave prints on contact paper (covered plates) or in the soot (open plates). The authors identified species from these prints, measured toe pads to tell marten from fisher with a published formula, and recognized one fisher as an individual from scars on its pads. They also counted scat left on the plates, and in one case fresh marten tracks in the snow, as detections. Snowtracking is one of the three protocol methods and the introduction says it was compared, but no snowtracking survey or results are reported.

5. Tracker or observer certification

Not reported. The paper does not say who identified and measured the tracks or how skilled they were. The authors note that measuring tracks and applying the formula takes more skill than reading photographs, and that track-plate identification needs some training.

6. Other technologies, skills or innovations

• Dual-sensor remote cameras (Manley systems, triggered by motion and heat) with Yashika AW-mini cameras; the authors recommend the lighter Trailmaster TM500.
• A portable acetylene welding torch to soot plates on site, and a heavier 3.30 mm PVC roof on covered plates.
• Digital calipers under a stereomicroscope, measuring to 0.01 mm.
• Permutation statistics suited to small samples.

7. Implications for conservation (from the paper)

• To manage forests for these species, managers need effective ways to detect them. The five categories used here give biologists a guide for choosing the method that best fits their project.
• Following the standard protocols, with the suggested changes, across wide regions would give a better understanding of where these rare species live.