The difficulty of estimating population densities of nocturnal forest mammals from transect counts of animals

1. What they did

Many small to medium-sized nocturnal forest mammals, especially in the tropics, can only be surveyed by seeing or hearing each animal. The author explains why: signs such as footprints, dung and scratch marks are often not identifiable to species, because similar species often live together; villagers’ identifications are rarely clear enough for research; and the origin of animals sold in markets can rarely be pinned down.

Such animals are usually surveyed by an observer walking slowly with a headtorch, searching all levels of the vegetation for reflecting eyes (eye-shine) or body shapes, and listening for calls, rustling and dropping fruit. Results may be given as survey time divided by the number of contacts, and are sometimes turned into population densities. The author asks whether the assumptions of distance sampling, a widely used way of estimating density from transect counts (Buckland et al. 1993), hold for these mammals. The author draws on the literature and on night surveys of their own in Ethiopia, Laos and Borneo.

2. What they found

Animals on the line are missed

Distance sampling assumes that every animal on the transect line itself is detected. If not, densities are underestimated. The author shows that many nocturnal forest mammals are missed for reasons that have nothing to do with their distance from the line. The author sorts them into four groups by how easy they are to detect:

• Partly nocturnal primates with weak eye-shine that alarm call (baboons, colobus monkeys, brown lemur): very hard to see even from directly below; found almost only by their calls.
• Quiet mammals with bright eye-shine that rarely call (lorises, some lemurs, common palm civet, colugo): many are hidden by vegetation. Some are only seen after the observer has passed and looks back; when the observer stops, more animals turn up several minutes later; and a second observer walking alongside always finds animals the first one missed. A slow loris watched in a leafless tree for 55 minutes was invisible for several minutes at a time simply because they were not looking towards the torch.
• Noisy mammals with bright eye-shine (kinkajou, small-toothed palm civet, bushbabies): the two carnivores are usually found by sound. In Ethiopia, bushbabies in low acacia bushland were usually found by eye-shine, but those in nearby dense forest were heard first. At a site in Laos, 4 pygmy lorises and 9 small-toothed palm civets were found; only one civet was found by eye-shine, which suggests many pygmy lorises were passed by unseen.
• Quiet mammals with no eye-shine (sloths, neotropical porcupines): sightings are rare, yet dung counts show sloths are among the commonest forest mammals.

Detection, even on the line, also changes with activity and calling through the night and the seasons, the moon, the fruiting and leafing of plants, and hunting pressure, which makes animals shy of torches.

Sample sizes

Density estimates need at least 60 to 80 contacts. Over much of South East Asia, mammals as a whole are rarely found more than once an hour, and even common species rarely more than once every 6 to 9 hours. Thousands of hours per site would be needed.

Roads and repeated walking

• Survey lines should be placed at random, which is unlikely when roads are walked. But roads give good visibility and quiet walking, while walking inside the forest gives few sightings because the observer sees little and flushes animals with noise.
• Because roads are scarce, the same line is often walked many times. A line walked three nights in a row often gives mostly different species, and different individuals, each night, because encounters depend so much on chance. This is acceptable for contact rates but is at odds with the near-complete coverage that density calculations assume.
• The author suggests dividing the survey area into sections, walking each whole section the same number of times, and combining the encounter rates of each section weighted by its length.

3. What they concluded

• For most nocturnal forest mammals, transect sightings cannot be used to calculate population densities, because major assumptions of distance sampling are not met.
• For many purposes, a rough idea of a species’ status is what is most needed, and comparable data from more areas may be more useful for conservation than density estimates.
• Until methods deal with animals missed on the line, results are best given as contact frequencies, with this context:
• Contacts per unit of time rather than distance, because a stationary observer keeps finding animals. Walking speed also changes which species are recorded: in Borneo, walking faster gave more mouse-deer and fewer canopy flying squirrels.
• Calculations and statistical tests based on the number of contacts (a group counts as one contact), with the total number of animals also given.
• Visibility in the habitat and typical sighting distances.
• The main way each species was detected.
• Hunting pressure in the area.
• Whether observations were made from roads, paths or at random across the habitat, with results given separately for each.

6. Other technologies, skills or innovations

• Headtorch spotlighting for eye-shine, combined with listening for calls, rustling, displaced vegetation and dropping fruit.
• The paper draws on observer skill: stopping to scan, looking back after passing, and the value of a second observer.

7. Implications for conservation (from the paper)

• Quick assessments of animal status are increasingly important for conservation, and a repeatable field method for nocturnal mammals is needed to compare sites.
• Population estimates may not be the best use of limited survey time; comparable sighting data from more areas often help more in setting conservation priorities for species and areas.
• Raw data and methods should be published, so that others can try alternative calculations.