A Possible Method for Estimating River Otter, Lutra canadensis, Populations Using Snow Tracks

1. What they did

River Otters are trapped for fur across much of North America, yet they are hard to count: they spend much of their time in water and usually live at low densities. Earlier ideas, such as scent stations with chalk-dusted track boards, or labelling otter scats with a radioactive tracer, had problems or were untested in the wild. The authors developed and field-tested a way to estimate otter numbers from their tracks in snow, as part of a long-term otter study in northeastern Alberta, Canada.

• Designing the survey. Radio-telemetry and years of watching otter sign showed that, once lakes freeze, otters must come up at the shore to breathe and move about. All winter dens were within 10 m of water, and otters crossing land took straight-line routes between water bodies. So the authors treated otter habitat as a line of shoreline.
• Dividing the shoreline. They split all 250 km of stream and lake shore on 1:50,000 maps into 500 numbered segments, each 500 m long. Judging from earlier winters, 500 m held most of a day’s activity of one otter or group, with little chance of a second group on the same segment the same day.
• Choosing the timing. The survey needs good snow and weather mild enough that every otter comes above ground each day. In this area that window is short: usually the last week of November and the first ten days of December. They checked the second condition with radio-collared otters: in December 1983, all 11 moved above ground during the survey period.
• Surveying. Random samples of 60 segments (1 to 6 December 1983) and 55 segments (5 to 11 December 1984) were searched on foot. Only sign judged less than 24 hours old was counted. Observers also recorded the age of tracks and the size and make-up of otter groups.
• Estimating. They counted otters per segment, fitted the counts to a Poisson distribution (a model for rare, random events), and multiplied the mean by all 500 segments. They also worked out how many segments would be needed for more precise estimates.

2. What they found

Otters per segment

• Most segments had no otter tracks, and none had tracks of more than two otters.
• 1983: 56 segments with no otters, 2 with one, 2 with two; mean 0.100 otters per segment; estimate 50 otters (95% confidence limits 18 to 109).
• 1984: 51 with none, 4 with one, none with two; mean 0.073; estimate 36 otters (9 to 93).
• Both years together: mean 0.087; estimate 44 otters in 250 km of shoreline (20 to 80).
• The two years did not differ statistically, and the counts fitted a Poisson distribution.

Precision and effort

• The estimate of 44 had a very wide confidence interval, even though the combined sample covered 23% of all segments.
• To be within 20% of the true number with 95% confidence would take 364 segments (72.8% of the area); within 10% would take 454 segments (90.8%).
• Two observers each surveyed about five segments a day, a sampling intensity of 10 to 15%.
• The estimate could not be checked directly, but from four years of watching otter sign and radio-collared otters, the authors believed it was reasonable.

Possible errors

• Misreading tracks. It can be hard to tell how many otters used a well-worn trail from a den to a latrine, or a single furrow made by otters travelling in line, or otters going different ways along the same route.
• The 24-hour window. Otters that left the segment after making tracks would still be counted, giving an overestimate.
• Otters with no above-ground sign. Otters that reached a den through open water or under the ice leave no tracks, giving an underestimate. This may get worse in very cold weather and after beaver ponds drop and leave air spaces under the ice.
• Double counting. Sign on nearby segments may come from the same otters.

The second and third errors tend to cancel each other, but taken together the errors most likely lead to an underestimate.

Comparison with other methods (Table 3)

• Snow tracks: no handling of animals, very little equipment, heavy labour but only one visit; limited to the northern part of the otter’s range.
• Scent stations with track boards: no handling, elaborate equipment (chalkboards and scent), heavy labour with repeated visits.
• Radioactive labelling: animals must be handled, elaborate equipment (a scintillation counter), heavy labour with repeated visits.

3. What they concluded

• Snow tracks can give an absolute estimate of otter numbers, while methods based on scats or scent stations need much more work on how age, sex and season affect the sign before they can.
• It is the least expensive of the methods compared, because little equipment is needed and observers visit only once. But it is labour-intensive, and reasonable precision needs very intensive sampling, the central problem with any low-density population.
• Surveys must be done in a short period (about a week) in early winter. The exact timing will vary with latitude and the start of winter.
• Results should be reported both per length of shoreline and per area of catchment, along with the lengths of each shoreline type (streams of each order, lakes, ponds), so that study areas can be compared.
• Otters probably prefer certain water bodies or shoreline types in different seasons, so stratifying the sample by shoreline type would likely improve accuracy and reduce effort.

4. How tracking was or was not used

Tracking was the core of the method. Observers searched each 500 m segment on foot for otter tracks and other sign in the snow and:

• Counted otters on each segment from their tracks, including group size and composition.
• Aged the sign, judging whether it was less or more than 24 hours old from snow and hoarfrost in the tracks and whether scats had frozen. Only fresh sign was counted.
• Followed tracks through neighbouring, unsampled segments, beyond the segment limits, to check that the same otters were not counted twice and to find where trails led to water access.
• Told individuals apart by measuring and comparing the widths of the furrows otters leave in snow, on nearby segments and where otters travelled in line.

The authors point out that misreading tracks could lead to either an underestimate or an overestimate, and that observers must look carefully and follow tracks to avoid these errors.

5. Tracker or observer certification

Not reported. The paper says only that two observers did the surveys. The estimate depends directly on their ability to count otters from tracks, age the sign and tell individuals apart by furrow width, and the authors name misreading tracks as a source of error, but no training, experience or test of the observers’ skill is described.

6. Other technologies, skills or innovations

• Radio-telemetry of live-captured otters, used to learn how otters use winter habitat and to check that all collared otters came above ground during the survey window.
• A simple system for dividing all stream and lake shores on topographic maps into numbered 500 m segments for random sampling.
• Sample-size calculations to show how much survey effort a given precision would need.

7. Implications for conservation (from the paper)

• Otters are trapped in nearly all jurisdictions, but their numbers are still estimated from fur-harvest records, which have built-in biases. Snow-track surveys offer a direct, low-cost way to estimate numbers without handling animals.
• Because otters live at low densities, managers need very intensive sampling for precise estimates. Stratifying by shoreline type could make surveys more efficient.
• Reporting density per shoreline length and per catchment area, with shoreline types described, would let managers compare otter populations between areas.