A Terrestrial Furbearer Estimator Based on Probability Sampling

1. What they did

Large furbearers such as lynx, wolverine, wolf and bobcat live at low densities, are secretive and often nocturnal, so they are hard to count. Earlier methods (mark-recapture, howling surveys, track count indexes, wolf pack counts, mail surveys of trappers and harvest reports) had proven difficult or gave unsatisfactory results, and most gave no measure of precision. The author used probability sampling to develop a new way to estimate furbearer numbers from tracks in fresh snow.

Both methods need good snow: fresh snow deep enough that tracks made before and after a snowstorm can be told apart. Survey lines (transects) are laid out in a repeated systematic design across the study area, and every animal track that crosses a transect must be seen. The author presented two designs:

• Following tracks. Each track that crosses a transect is followed both to where the animal is now and back to where they waited out the snowstorm. From this, the distance the animal travelled across the study area is measured. The farther an animal travelled, the more likely their tracks were to cross a transect, and the estimator corrects for this. Animals travelling together can be counted as a group.
• Radio-collared animals. Where tracks cannot all be followed, observers count the number of different animals whose tracks cross the transects. Movement data from a random sample of radio-collared animals then give the average distance an animal travels, and the two are combined in a ratio.

He tested each design with a worked example in southcentral Alaska:

• Wolverines (18 March 1988): three pilot and biologist teams in light aircraft (PA-18 Super Cubs) flew four systematic samples of three 32.25 km transects across a 1,870.5 km² area of the Chugach Mountains, 12 to 18 hours after a snowstorm. They followed the wolverine tracks they found. Four transects were flown again by different teams to check for missed tracks.
• Lynx (22 January 1987): an aerial survey of a 285 km² area on the Kenai Peninsula, about 24 hours after a snowstorm, with four systematic samples of three transects. Movement data for two radio-collared lynx came from radio telemetry and ground tracking at the end of the survey.

2. What they found

Wolverines

• Six wolverine tracks were seen and followed. Two wolverines had waited out the snowstorm in different places, travelled to the same carrion and then gone their separate ways, so they were treated as a group.
• The estimate was 9.69 wolverines, with a standard error (SE) of 1.97.
• No extra tracks were found when four transects were flown again. Three tracks that crossed two transects were seen both times, and one track that crossed five transects was seen every time.
• All wolverines with observed tracks were found, except two whose tracks passed bare dark rocks that may have hidden them. Repeated circling where the tracks disappeared made it unlikely the tracks reappeared unseen.

Lynx

• Twelve individual lynx tracks were counted on the transects.
• Five lynx had been radio-collared, but two left the area in early January and a third was illegally trapped during the survey, leaving two (one female, one male) for movement data. They had travelled 7.35 km and 4.38 km across the study area.
• The estimate was 15.09 lynx, with a standard error of 4.34. The estimated bias was 0.97 lynx, or 6.88% of the estimate.
• Both collared lynx crossed a transect and were seen during the survey.

3. What they concluded

• Wolverines and lynx differed in movement patterns and habitat use, yet both designs gave apparently reasonable population estimates.
• The rule that at least half of an animal’s movement must fall inside the study area is needed for accurate estimates. Without it, the method counts the animals that used the area between the snowstorm and the survey, rather than giving a snapshot of how many were there at one time.
• When tracking from the air, pilots and observers should be highly qualified at seeing and identifying tracks from an airplane or helicopter. Unqualified observers will probably miss tracks and underestimate the population by an unknown amount. The author suggests using the aerial method only in fairly open habitats, possibly using helicopters, and searching areas between transects afterwards, with ground sampling or radio-collared animals to check that no tracks were missed.
• Unlike an earlier method by Hayashi, this one lets the chance of detecting each animal differ between individuals, which matters for species such as wolverine where movements differ by individual and sex.
• The biggest problem with the radio-collar design is logistical: it is not easy to catch furbearers and keep a sample of them collared.
• Precision improves when the chance of each animal’s tracks crossing a transect is higher, for example by orienting the transects across major travel routes such as valleys, or by waiting longer after a snowstorm. But the longer the wait, the harder it becomes to meet the assumptions, as tracks degrade or are covered by other animals’ tracks.
• Site conditions and the species’ movement behaviour will decide which design fits, and the degree of access and availability of trained observers will decide which method is preferable.

4. How tracking was or was not used

Tracking in snow is the basis of the method. Observers had to recognise the tracks of the study species, tell fresh tracks from those made before the snowstorm, and see every track that crossed a transect. In the wolverine survey, teams followed each track from the air both forwards to the animal and back to where the animal waited out the snowstorm, which measured how far each animal had travelled and showed when two animals had joined at carrion. In the lynx survey, observers counted individual tracks crossing transects, and the movement of the radio-collared lynx was measured partly by ground tracking. Missed tracks were checked by reflying transects with different teams and by circling where tracks disappeared.

5. Tracker or observer certification

Reported but not validated. The pilots are described as having “extensive aerial tracking experience” and all the biologists as “experienced aerial observers”, but their skill was not tested. Reflying four transects with different teams checked whether tracks had been missed, which supports the wolverine survey’s track detection but is not a test of the observers. The paper does not say who did the ground tracking of the collared lynx. The author stresses that the accuracy of the estimates depends on highly qualified track observers.

6. Other technologies, skills or innovations

• Light aircraft (PA-18 Super Cubs) for aerial tracking along transects.
• Radio collars and radio telemetry to measure lynx movement.
• A new estimator that lets each animal’s chance of detection differ, and a way to count animals travelling together as a group.

7. Implications for conservation (from the paper)

• Growing trapping pressure and habitat loss have increased the demand to monitor furbearer populations, and older methods were hard to use or gave unsatisfactory results.
• The snow-track estimators give population estimates with a measure of precision for rare, secretive species where mark-recapture is not appropriate because populations are small and capture rates are low.