Using the Formozov–Malyshev–Pereleshin formula to convert mammal spoor counts into density estimates for long-term community-level monitoring

1. What they did

Namibia’s communal conservancies must count their wildlife every year, and spoor (track) counts are a routine and efficient way for communities to do it. On their own, spoor counts are an index, not a density. The Formozov–Malyshev–Pereleshin (FMP) formula, used for decades in Russia, converts spoor counts into density using each species’ day range: how far an animal travels in a day. The more an animal moves, the more tracks one animal leaves, so the formula divides by day range. Accurate day ranges are often missing, so the authors tested whether estimates from body size could fill the gap.

• Spoor surveys. In Salambala Conservancy, north-eastern Namibia, teams of three or four surveyors walked 16 transects (215 km in all) over 2 days each September from 2010 to 2016, starting at 6:30 a.m. They recorded the location, species and number of spoor for all hoofed mammals and large carnivores, counting only spoor judged to be from different individuals and made within the last 24 hours. Spoor of animals they saw were left out, because sightings went to a separate distance sampling count.
• Day ranges. For African buffalo and plains zebra they calculated day ranges from Global Positioning System (GPS) collar data; for elephant, giraffe and spotted hyena they used published values. For the other species, they predicted day range from body mass, using the known relationship across 22 hoofed species.
• Density and trends. They combined spoor counts and day ranges in the FMP formula, used bootstrapping to get confidence intervals, and fitted generalised additive models (GAM) with rainfall to look for trends from 2010 to 2016.
• Checks. They simulated what would happen if all sighted animals had also been counted as spoor, and compared their densities with the conservancy’s distance sampling estimates.

2. What they found

Densities

• Seventeen mammal species were recorded over the seven years. Densities could be estimated for 10; the other seven (buffalo, jackal, steenbok, leopard, reedbuck, roan and waterbuck) had very few spoor or appeared in only a few years.
• The highest average densities (per 100 km²) were common impala (34.5), blue wildebeest (23.8), common duiker (16.1) and greater kudu (15.2). The lowest were eland (5.4), giraffe (3.9) and spotted hyena (0.2).

Trends

• No trend was statistically significant: the confidence intervals for 2010 and 2016 overlapped for every species.
• The paper describes the trends differently in different places (see the detailed keywords). In the Discussion, five species tended to increase (blue wildebeest, impala, duiker, giraffe and kudu) and four to decrease (eland, elephant, zebra and warthog).
• Adding all sightings as spoor gave similar trends; only zebra and blue wildebeest differed.

Day ranges from body size

• Day ranges predicted from body mass varied much less than real GPS data. For buffalo, the GPS day ranges varied more than twice as much (standard deviation 3.73 km against 1.57 km). This made the density estimates look more precise than they really were.

Spoor counts compared with distance sampling

• The two methods mostly agreed, with the largest differences for impala, zebra and warthog.
• Precision was similar: distance sampling was more precise in 15 of 29 comparisons, FMP in 13, with one tie. Distance sampling needed three years of sightings pooled together for each estimate.
• FMP densities were lower than distance sampling in 25 of 29 comparisons.
• FMP gave estimates for 10 species; distance sampling for eight, and for only two of them across the whole 2010 to 2016 period.

Undercounting

• The survey rules ask trackers to record each animal’s spoor only once and to ignore spoor of animals they saw. The FMP formula needs every crossing of a transect counted. For zebra alone, up to 10,174 spoor may have gone unrecorded over the seven years.

3. What they concluded

• Applied to routine community spoor counts, the FMP formula can give density estimates for more species than the conservancy’s distance surveys, including elusive species and those at low density.
• Densities were probably underestimated, because of the recording rules, and day ranges from body size or from GPS fixes an hour or more apart add error. The study also assumed day ranges stayed the same every year, although rainfall varied.
• The authors recommend recording all spoor, whatever the individual and whether or not the animal was also seen, and measuring day ranges over several years, for example by local experienced trackers trailing animals. They also recommend double sampling with line transects run at the same time.
• With these changes, the FMP method could be an invaluable tool for monitoring many species together.

4. How tracking was or was not used

The whole study rests on spoor counts. Surveyors on foot identified the species of each spoor crossing the transect, judged whether spoor were made within the last 24 hours, and judged whether spoor came from different individuals, discarding duplicates. No trailing was done in this study, but the authors note that in many places where the FMP can be used, trackers can follow trails with handheld GPS units to measure day ranges, and they recommend local experienced trackers trailing animals to fill this gap.

5. Tracker or observer certification

Reported but not validated. Surveyors were Ministry of Environment and Tourism (MET) rangers, local community game guards and trained community members; the paper says the community game guards “were all expert trackers”. No certification or test of their skill is reported. The authors cite Stander et al. (1997) to say indigenous trackers have proven highly accurate under different conditions, but that study did not test these surveyors. Many of the same community members took part across the seven years.

6. Other technologies, skills or innovations

• Allometric scaling: predicting a species’ day range from its body mass, with bootstrapped regression to carry the uncertainty into the density estimates.
• GPS collar data (four fixes a day for buffalo, one an hour for zebra) to measure day ranges. The authors note that fixes an hour or more apart miss fine-scale movement and underestimate day range, which inflates density.
• Generalised additive models to smooth year-to-year ups and downs when looking for trends.

7. Implications for conservation (from the paper)

• Communities already collect spoor data routinely; turning them into densities and trends makes them more useful for decisions such as hunting quotas.
• Distance sampling estimates pooled over three years can hide sudden changes and would need three times the current effort for robust yearly estimates.
• Possible declines in eland, elephant, warthog and zebra deserve attention. The authors suggest causes: eland were reintroduced in 2012 and are sensitive to disturbance and hunted for meat; warthog may be killed in retaliation for crop damage or poached; elephants are declining across Africa, with poaching a possible driver in Namibia; and zebra declines may be linked to rainfall.