A new method for estimating visitation rates of cryptic animals via repeated surveys of indirect signs

1. What they did

Knowing how often animals visit a particular place (a fish pond, a nest site, a farm) matters for managing conflict with people, predation and disease. For shy, hard-to-see animals, researchers often check the place again and again for signs such as tracks or droppings. The usual method simply counts how many checks found signs. It ignores how old the signs are.

The authors developed a new way to estimate visitation rate that uses the age of the signs. At each check, signs are recorded and removed, and each is classed as fresh (from the previous night) or aged (from the nights before). For Eurasian otters, fresh spraints (droppings) are wet and soft, while aged ones have a dry surface and solid consistency.

• Building the estimators. They worked out the maths for four estimators: fresh signs only, aged signs only, all signs without telling them apart (the method used in the literature), and fresh and aged signs together (the new method). They showed the older methods are special cases of the new one, and that the new one also works when checks are not evenly spaced.
• Testing with simulated data. They simulated weekly checks for 399 combinations of survey length (2 to 42 weeks) and daily visitation probability (0.05 to 0.95), 100 times each, and compared bias, precision and overall accuracy.
• Planning surveys. They simulated different intervals between checks to find the best interval for different visitation levels.
• A field example. They applied the method to otter spraints at a commercial fish pond in Upper Lusatia, Germany, checked weekly for 12 weeks.

2. What they found

Simulations

• The new estimator (fresh plus aged signs) was on average more than three times more accurate than the next best, the fresh-signs-only estimator (mean relative accuracy 0.077 against 0.250).
• The fresh-signs-only estimator was the only unbiased one, but it was imprecise, especially when visits were rare. The new estimator had only slight bias (3%).
• The aged-signs-only and combined estimators overestimated visitation (by about 36% and 44%). The combined estimator, the one used in earlier studies, performed worst, and did well only with more than 30 checks or very high visitation.
• The new estimator was inaccurate only with very few checks (fewer than about 4) and very low visitation probability (below about 0.1). In those cases none of the others did better.

Best interval between checks

The rarer the visits, the longer the best interval: about 30 days when the daily visitation probability is below 0.05, 10 to 20 days at 0.05 to 0.15, 5 to 10 days at 0.15 to 0.25, 4 to 5 days at 0.25 to 0.55, 3 days at 0.55 to 0.8, and every other day above 0.8 (otherwise aged signs are almost always present). The number of checks had only a minor effect on the best interval.

Otters at the fish pond

• Fresh spraints were found in 3 of the 12 weekly checks and aged spraints in 10.
• All four methods gave a daily visitation probability of about 0.25 to 0.26 (a visitation rate of about 0.3 visits per day). The new method gave the narrowest confidence interval: probability 0.256 (95% confidence interval 0.144 to 0.405).
• With the new method, 10 checks gave about the same precision as 12 checks with the other methods.
• The estimate was very similar to one from an earlier radio-tracking study of otters.

3. What they concluded

• Using both fresh and aged signs gives precise, almost unbiased estimates of visitation across the widest range of conditions. Wherever possible, fresh and aged tracks and signs should be told apart and used together.
• The commonly used estimator is heavily biased and imprecise, which the authors call alarming.
• Signs should be aged as accurately as possible, but the authors report that the method still works when ageing is uncertain, as long as errors are not systematic.
• The method assumes visits arrive independently at a constant rate. This does not need each animal to move at random, only that the animals act independently and are not too few. Where this assumption is strongly broken, estimates may be best used only to rank sites against each other.
• The method could work for any species whose signs can be dated, for example eaten fruits, leaves, carcasses or burrows. It could be extended to more than two age classes, longer “fresh” periods, or counts of individuals at each check.

4. How tracking was or was not used

The method rests on tracks and signs: repeated surveys for signs at a site, with each sign recorded, aged and removed. In the field example the signs were otter spraints, aged as fresh (wet and soft, from the previous night) or aged (dry surface, solid). The authors note that ageing can vary slightly with weather and time of day, that otter spraints are typically obliterated after about 7 weeks, and that checks must be frequent enough for aged signs not to disappear between visits. No trailing was reported.

5. Tracker or observer certification

Not reported. The paper does not say who surveyed the fish pond. It states in general that experienced researchers can age spraints correctly, and recommends a pilot study with daily sampling. The accuracy of sign ageing matters to the results, but no test of the observers’ ageing skill is described.

6. Other technologies, skills or innovations

• “DoubleTrack”, a free Excel workbook with macros, calculates visitation rates, visitation probabilities and confidence intervals for regular and irregular intervals between checks (supplementary material and authors’ website).
• A table of the best interval between checks for different visitation levels, to help design surveys.

7. Implications for conservation (from the paper)

• Visitation rates matter for many conservation and management problems: the conflict between otters and fish-pond owners, predator visits to nests, and the spread of disease (such as tuberculosis by badgers or rabies by foxes).
• Better estimates of how often otters visit commercial fish ponds can help assess that conflict, and the new method is widely applicable in ecology and conservation biology.