Can trackers count free-ranging wildlife as effectively and efficiently as conventional aerial survey and distance sampling? Implications for citizen science in the Kalahari, Botswana

1. What they did

Botswana's wildlife authority counts wildlife mainly by aerial survey, sometimes adding ground line transects. Both are costly, and aerial surveys now happen less often. The authors asked whether Kalahari trackers could collect equivalent information, and do it as efficiently.

• Transects: the 13 transects that the aerial survey flies across the KD2 Wildlife Management Area, about 10.2 km apart and 648.4 km in total. All three surveys were done within nine days (25 October to 2 November 2015).
• Aerial survey: a Cessna 206 at 91.5 m and 167 km per hour, with observers counting animals within marked strips on each side, following standard Department of Wildlife and National Parks procedures.
• Ground line transects: five teams, each with a driver, a data recorder and two trackers seated over the front of a 4 × 4 vehicle, drove the same lines at 15–25 km per hour and measured the distance to every animal or group seen with a laser rangefinder.
• Track counts: the same teams then drove back along each transect. The trackers counted every large animal's track that crossed between or onto the tyre marks from the first pass, recording only tracks they judged to be less than 24 hours old. Every crossing was counted, whether or not the trackers thought it was the same animal. Large herds were counted with handheld tally counters. To keep up speed, the trackers looked only for large tracks, including large carnivores, not small antelope.
• Analysis: each method was analysed blindly by a different co-author. Track counts were converted to density with the Formozov–Malyshev–Pereleshin (FMP) formula. Day ranges came from body-mass scaling rules with the Kalahari correction factor for most antelope, from earlier trailing for gemsbok, from GPS collars for wildebeest, and from published studies of habituated and collared animals for large carnivores.
• Efficiency: they compared encounter rates and costs per kilometre, using survey data pooled from the region between 2001 and 2015, and drew density-distribution maps for each method in the same format as Botswana's aerial survey maps.

2. What they found

Similar estimates

• Ground transects stayed on average 23 m from the flight lines, so all three methods sampled the same populations.
• For six large grazers, raw counts were highly correlated among all three methods, population estimates showed no significant differences (Wilcoxon tests), and 95% confidence intervals overlapped broadly.
• Precision was similar: the average coefficient of variation for seven large antelopes was 46.0% by air, 50.1% by distance sampling and 38.4% by tracks.
• Aerial estimates averaged 93% of line transect estimates, apart from wildebeest, where the aerial count was four times higher.

Browsers undercounted from the air

• The aerial survey returned 60%, 29% and 5% of the line transect estimates for kudu, duiker and steenbok, and 35% of the track-based kudu estimate.
• The authors trust the kudu track counts because kudu tracks are easily told apart from similar-sized grazers. They note that steenbok are sand-coloured, shade-loving and usually do not move in response to aircraft.

Eland

• Track counts may have underestimated eland, which moved in herds sometimes exceeding 1000 animals; hooves at the back of a herd can erase the tracks of those in front.

Far more detections

• Trackers counted 19,584 track sets, compared with 2137 animals seen from the air and 2921 from the ground.
• Line transects reached the recommended minimum of 40 observations for only one-third of antelope species.
• Track counts gave estimates for all six large carnivores in the region, which the direct methods saw too rarely to estimate.
• For example, seeing one aardvark by night-time spotlighting takes an expected 442 km of survey ($3171), while detecting an aardvark's tracks from the previous night takes 3.9 km ($28).

Maps

• Tracks detected each species in more grid cells: on average 3.3 times more than the aerial survey and 4 times more than line transects (for kudu, 6.8 and 8.5 times more). Many empty cells on aerial survey maps are probably false absences.

Costs

• The aerial survey cost $7794 ($12.02 per km). Each ground survey cost $4652 ($7.17 per km), so the track survey cost 40% less than the aerial survey.
• Trackers on horseback, recording their own data, could cut this to about $1400: a threefold drop from using vehicles, and less than 20% of the aerial survey cost.

3. What they concluded

• The main finding: Kalahari trackers' data "rivals" data collected by conventional methods. The authors found little evidence that direct sightings are better than tracks for precision, number of species, distribution mapping or cost; the track survey showed advantages in all of these.
• Their recommendation: they urge Botswana to consider large-scale track-based wildlife counts led by citizen scientists, to complement its aerial survey.
• A first: to their knowledge, this was the first time a community in Botswana had not only taken part in, but successfully led, a wildlife count within its Wildlife Management Area.
• Advantages (Box 1): similar estimates and precision for equal effort; many species assessed at once; more complete distribution maps; lower costs, so more frequent surveys; participatory conservation and much-needed employment.
• Limitations (Box 1): day ranges at the time of the survey must be reliable (although accuracy depends much more on track counts than on day range); counting errors with large herds; declining traditional tracking skills.
• Skill required: track surveys need observers who can identify tracks to species with more than 95% consistency and reliably age tracks to within 24 hours. The authors note that tests have shown even experienced wildlife professionals often fall far below this standard.
• Culture and future: tracking skills are declining with the loss of traditional hunting. The authors argue that involving trackers in conservation could reverse this, and that tracking is an intangible cultural heritage worth safeguarding.
• Caveat: results apply to this environment in the late dry season and could differ after leaf flush or elsewhere in Botswana. Counting animals by their tracks is limited to areas with suitable substrates.

4. How tracking was or was not used

• Track counts: trackers identified and counted every large mammal track crossing the transect, aged each track and recorded only those under 24 hours old, including repeat crossings by the same animals, using tally counters for large herds.
• Line transects: trackers spotted most of the animals. When animals fled before they could be measured, trackers walked to and interpreted the place where they had been standing, so an accurate distance could be measured.
• Day ranges: no trailing was done in this survey. The gemsbok day range came from earlier trailing of daily movement paths (Keeping and Pelletier 2014).
• Future: the authors suggest trackers could survey on horseback (or camel) and record data themselves with CyberTracker software on smartphones.

5. Tracker or observer certification

• Reported but not validated. Most trackers lived in the remote village of Zutshwa and had hunted with subsistence game permits on their traditional territories in KD2 before 1997. The paper calls them "local expert observers" but reports no certification or test of their skill.
• The authors cite Stander et al. 1997, a different study of Kalahari trackers, for near-perfect accuracy.
• Many of the trackers could not read or write; tally counters let them focus entirely on the tracks.

6. Other technologies, skills or innovations

• Tyre marks as a sampling frame: the first pass along each transect left tyre tracks that marked the strip for the track count.
• Tally counters: these freed trackers from keeping a running count in their heads.
• GPS and ArcGIS: used to navigate, record each observation and measure how closely the ground transects followed the flight lines.
• Day ranges: from body-mass scaling rules, earlier trailing with GPS loggers, GPS collars and published studies.
• Suggested for the future: CyberTracker software on smartphones so trackers can upload data directly, and horses or camels instead of vehicles.

7. Implications for conservation (from the paper)

• Cost and frequency: aerial surveys are increasingly expensive and infrequent. For equal investment, KD2 could be surveyed more often from the ground than from the air.
• Rare species: track-based detection may be the only practical way to monitor rare and cryptic species over large areas, such as ground pangolin, which is data-deficient and heavily trafficked. Tracker data also supports anti-poaching work.
• Land-use decisions: aerial survey maps inform land-use decisions, but many of their empty cells are probably false absences.
• Community involvement: communities have long been dissatisfied at being left out of wildlife counts. Local involvement in monitoring could bring benefits beyond the data itself.
• International commitments: as a signatory to the UNESCO Convention for the Safeguarding of the Intangible Cultural Heritage and the Convention on Biological Diversity, Botswana could meet both commitments through tracker-led monitoring.