A review of methods to estimate the abundance of terrestrial carnivores using field signs and observation

1. What they did

Carnivores are hard to count. Many are secretive, active at night and range over large areas, and they live in very different social systems, from solitary animals to pairs to large groups. Trapping animals often enough to estimate numbers can be impractical, labour-intensive and very expensive.

The authors reviewed field methods for estimating and monitoring the abundance of terrestrial carnivores without capturing them. They grouped the methods into two families:

• Field signs: track counts along transects, track stations (with and without scent lures), counts of droppings, DNA from droppings and hair, howling and other calls, and counts of dens, setts and holts.
• Observation: direct counts, call-in (playback) surveys, aerial surveys, spotlight counts at night, thermal imaging, and remote cameras and video.

For each, they described how it works, gave case studies, and discussed its strengths, weaknesses and costs.

2. What they found

Field signs in general

• Sign-based methods assume that the amount of sign relates in some way to the number of animals. They are mostly used as indices, to compare places or follow trends over time, rather than to give actual numbers.
• Results depend on the skill of the fieldworkers. The authors point to Stander et al. (1997) as showing the value of field craft, and say such skills are likely to raise detection and interpretation.
• Season, habitat, weather, substrate, and the sex and age of animals can all change how much sign is found, separately from how many animals there are. These effects are easier to control when following one place over time than when comparing places.

Tracks

• Track counts are a relatively cheap way to follow trends, and several species can be monitored at once.
• Where individual footprints can be told apart, by highly skilled fieldworkers (Stander et al. 1997) or by measurements from photographs of tracks, actual numbers can be estimated.
• Some track indices matched known populations well: road-transect track counts for lion, leopard and hunting dog in Namibia were a function of true density, and snow-track counts of coyotes and lynx in the Yukon mirrored other population estimates. An earlier lynx study with fewer surveys and no covariates found a poor match.
• Mountain lion tracks were easier to detect on roads with good dust cover, and roadside habitat changed fox track counts regardless of density. Choosing transects with ecology in mind more than doubled the chance of finding tracks in one study.
• Scent stations gave mixed results when checked against known populations, and lures may attract some animals (by age, sex or status) more than others. For dingoes, track plots without lures gave a more reliable index.

Droppings and DNA

• Counts of droppings can give relative abundance; actual numbers need defaecation and decay rates, and the authors call this an impractical goal for most carnivores. Fox scat counts dropped after rain, and otter spraint numbers swung widely while populations were stable, probably because of seasonal and habitat changes in scent marking.
• DNA from droppings and hair can identify individuals. Coyote numbers from faecal DNA compared well with trapping, and brown bear numbers in the Pyrenees from faecal and hair DNA closely matched an established monitoring scheme based on identifying tracks. Fresh samples, good storage and checks for laboratory errors are essential.

Calls and dens

• Wolf howling surveys can count packs but give only a crude population estimate, and one survey overestimated pack density.
• Den counts work when the number of animals per den is known; counts of otter holts in Shetland were linearly related to the number of resident males. Badger group size varies so much that converting sett counts into badger numbers carries a large potential error.

Observation

• Playback of prey distress calls and lion calls gave a lion estimate within 1% of a photographic count, though responses varied with season and prey abundance.
• From the air, wolf tracks in fresh snow were followed until the pack was seen and counted. This needs experienced pilot-observer teams.
• Spotlighting sometimes missed carnivores in areas where track plots showed they were present.
• Thermal imaging from an aircraft found all the radio-collared mountain lions known to be in a Florida study area.
• Camera traps can estimate numbers where individuals are recognisable, such as tigers by their stripes, but placing cameras on animal trails can bias results.

3. What they concluded

• Many methods exist; the choice depends on the study’s aims and scale, the resources available, the ecology of the target species, and the accuracy needed.
• Where animals can be seen, especially as individuals, observation is likely to be preferred. Where that is impossible or too costly, field signs may be necessary.
• Sign-based indices are probably more reliable for following trends in one place over time than for comparing places.
• Methods should, where possible, be validated against a population of known size or a formal estimate; in practice this is seldom done. Where calibration is not possible, using two or more methods together adds confidence.
• Methods based on field signs are especially prone to differences between observers. Skilled fieldworkers are a highly valuable resource, while large surveys with many inexperienced staff should be kept simple enough for consistent recording.

4. How tracking was or was not used

The authors did no fieldwork; tracking is a main subject of the review. They discuss studies that used footprints along dirt roads and in sand and snow, prepared track stations, scats and spraints, hair, trails and paths, and dens, setts and holts. Trailing appears in the aerial wolf surveys, where tracks in snow were followed from the air until the pack was found. They note that identifying individual animals from their tracks is sometimes possible with highly skilled fieldworkers, citing Stander et al. (1997), and that a Pyrenean brown bear monitoring scheme was based on identifying tracks.

6. Other technologies, skills or innovations

• Genetic fingerprinting (microsatellite DNA) from droppings and hair, with rarefaction and capture-recapture analysis
• Glue-plate and baited hair snares
• Track measurements from photographs to identify individual mountain lions
• Playback of recorded calls to draw lions and hyenas into view
• Spotlights, night-vision and aerial thermal imaging
• Camera traps triggered by trip lines, plates, photic cells or infra-red beams, and remote video

7. Implications for conservation (from the paper)

• Reliable abundance estimates are basic to conservation, research and management, whether the goal is to increase an endangered population or reduce a pest one.
• Survey design should take account of season, habitat and weather, either by standardising surveys or by recording these as covariates.
• Large increases in survey effort often give only small gains in the power to detect change, so effort should be planned, for example with power analysis.