Rapid assessment of wildlife abundance: estimating animal density with track counts using body mass–day range scaling rules

1. What they did

The Formozov–Malyshev–Pereleshin (FMP) formula turns track counts into density, but it needs each species’ day range: the average distance an animal travels in 24 hours. Day ranges are rarely measured. There is, however, a consistent relationship between an animal’s body mass and its day range. Keeping tested whether day ranges predicted from body mass could be used in the formula to estimate the density of a whole wildlife community quickly, using the southern Kalahari of Botswana as an example.

Track counts

• He laid out 24 spatially separated track transects, totalling 634 km, along 4 × 4 trails and firebreaks: 16 on trails and 8 on firebreaks.
• Most were 20 km long, about the most that could be scrutinized in a day when recording every species over 0.2 kg.
• Transects within about 20 km of settlements were excluded, because people affected wildlife up to that distance.
• Each transect was sampled once, between 15 October and 15 April.
• Clearing the transect: a heavy steel beam was first dragged behind a vehicle to wipe out old tracks, so only fresh tracks from the next 24 hours were counted.
• Counting: early the next morning, two observers on seats at the front of a vehicle moving at 6–8 km per hour recorded every track crossing by species and number, with a GPS location. One was an expert local tracker, and the other was the author, “also a competent tracker”. No assumptions were made about whether crossings belonged to the same animal, so recrossings were counted too.
• An analysis of variance checked whether counts differed between trails and firebreaks.

Day ranges

• For every species, day range was predicted from body mass, using separate scaling rules for three groups (even-toed hoofed animals, carnivores and rodents) built from data on 200 mammal species worldwide.
• Measured (“empirical”) day ranges were available for 12 mammal species (and for the tortoises) from the study area or similar habitat. Most came from radio tracking and from following habituated animals in published and unpublished studies. Wildebeest day range came from GPS collars.
• The author measured gemsbok, steenbok and aardvark day ranges himself, by following their tracks over 24 hours with handheld GPS loggers taking a fix every second.

Estimates and checks

• Density for each species was estimated with the FMP formula, and its uncertainty with bootstrapping.
• The estimates were compared with independent estimates: an aerial survey for large herbivores, distance sampling from a vehicle for smaller species, and published spoor counts calibrated against known densities for large carnivores.

2. What they found

Firebreaks

• Four small species (hare, porcupine, springhare and steenbok) had much higher counts on firebreaks than on trails, especially on recently cleared firebreaks. They appeared to be attracted to these open strips.
• Their firebreak counts were left out of the analysis. For most other species, firebreaks and trails sampled equally well.

Day ranges were longer than predicted

• All 12 mammals with measured day ranges travelled, on average, 140% farther each day than predicted from their body mass. This held across taxonomic groups.
• Examples of measured day ranges: brown hyena 31.1 km, spotted hyena 26.5 km, lion 19.4 km, honey badger 10.8 km, leopard 9.7 km, cheetah 9.0 km, aardvark 8.6 km, gemsbok 5.9 km, steenbok 4.5 km, wildebeest 4.3 km and yellow mongoose 3.2 km.
• The author suggests this is because the Kalahari has nutrient-poor sandy soils and lower productivity than the environments the scaling rules came from.

The correction factor

• Because predicted day ranges were too short, densities based on them were generally too high compared with the independent estimates.
• He derived a correction factor from the relationship between densities calculated with measured and with predicted day ranges, for the 12 species (slope 0.393, r² = 0.94).
• After correction, the differences from independent estimates balanced around zero across 22 species. For carnivores, 9 of 11 species moved closer to their independent estimates. Lion and spotted hyena did not, possibly because of small samples.
• Overall, there was good correspondence between the corrected and independent estimates.

Density estimates

• This gave the first comprehensive density estimates for the southern Kalahari mammal community over 0.2 kg, including 11 carnivores, 3 insectivores and 2 herbivores that other multispecies methods could not estimate.
• Examples per 100 km²: hares 376.1, steenbok 222.6, gemsbok 141.7, black-backed jackal 63.9, brown hyena 2.3, leopard 1.0, cheetah 0.9, lion 0.6 and spotted hyena 0.1.
• Converting track counts to density changed how species ranked. Animals with short daily movements, such as tortoises, can be more numerous than their track counts suggest, and wide-ranging ones, such as brown hyena, less numerous.

3. What they concluded

• What the approach can do: day ranges predicted from body mass are a coarse input to the FMP formula. But they at least give species’ abundances relative to one another, which raw track counts do not, and they make rough density estimates possible for species no other multispecies method can count.
• Local day ranges matter: general scaling rules can overestimate density where productivity is low, or underestimate it where productivity is high. If accuracy is important, measuring day range locally for a subset of species can support or correct the predictions for the others.
• Day range changes: it can vary with season, rainfall, food, temperature, moonlight, migration, mating and dispersal, and possibly with density. So day range should be measured at the same time of year as the track survey, and reassessed if density is monitored over time.
• Counting rules: estimates are most accurate if transects do not influence animal movements and if every crossing is counted, however many times an animal recrosses.
• Linear features: these can bias results. Some small herbivores are drawn to firebreaks, and many carnivores travel along roads and trails. Surveys that use linear features should be interpreted cautiously.
• Animals that enter and leave without crossing: the author suggests counting animals that enter the track bed and leave on the same side, because their movements brought them at least to the centre line.
• Comparison with other methods: in the Kalahari, aerial surveys are irregular and limited to large antelope, and distance sampling adds only some common species. Camera trapping is limited by cost, theft and failures in a vast, harsh area. Tracking “supersedes the limitations of other methods”. With ideal substrate all year and skilled local trackers, southern Africa is especially well placed for track-based counts, and for building a library of day ranges by putting trackers on animal trails.
• Limitations: independent estimates are of unknown quality themselves. Missed tracks may be slightly more common for small, light species. Springhare may bound over transects without leaving tracks, or their numbers may have changed between surveys.

4. How tracking was or was not used

• Track counts: two trackers identified and counted every track crossing of every species over 0.2 kg on 24 pre-cleared transects, including recrossings by the same animal. The lead tracker only rarely missed recrossings, mostly where tracks became highly tortuous over short distances.
• Trailing: the author followed the 24-hour tracks of gemsbok, steenbok and aardvark with GPS loggers to measure their day ranges very accurately. Some of the other empirical day ranges came from following habituated animals.

5. Tracker or observer certification

• Reported but not validated. The surveys were done by “an expert local tracker” and the author, “also a competent tracker”. No certification or test of their skill is reported.
• The paper notes that tracks of all species were readily visible and could be interpreted by more than one observer.
• The acknowledgements thank the author’s tracking companions, Panana Sebati and Nicolas Molese, of Ngwatle and Zutshwa.

6. Other technologies, skills or innovations

• Track drag: a heavy steel beam dragged behind a vehicle cleared the transects, so each count covered exactly one 24-hour period.
• GPS: handheld GPS loggers taking a fix every second recorded animals’ full 24-hour trails. The author contrasts this with infrequent telemetry fixes, which can severely underestimate day range.
• Body mass–day range scaling rules allow density to be estimated for species whose day range has never been measured.
• Existing tracking data: the author suggests that archives of animal tracking data, such as Movebank, could provide more day ranges.

7. Implications for conservation (from the paper)

• Why it matters: limited conservation resources call for efficient ways to assess wildlife abundance. Most population estimates of threatened species rely on direct counts, which leaves gaps.
• What track surveys offer: they accumulate unbiased observations quickly where tracks are readily visible, so this approach has potential for rapid, comprehensive “snapshot” assessments of wildlife communities in remote areas where other methods are unfeasible.
• Comparing species: differences in how far species move mean that raw track indices cannot be compared between species.
• Southern Africa is especially well placed to benefit, given its substrate and skilled local trackers.