# Cost and efficiency of large mammal census techniques: comparison of methods for a participatory approach in a communal area, Zimbabwe

> Is an aircraft or a bicycle the better way to count elephants? In a communal area of Zimbabwe’s Zambezi Valley, local agents on bicycles and on foot saw far more animals for far less money than cars or aircraft: about US$1 per elephant seen, against US$18 from the air and over US$38 from a car. Quiet, slow observers got closer to wary wildlife in thick bush.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2006-gaidet-drapier-fritz-cost-efficiency-large-mammal/

## Citation and access

- **Authors:** Nicolas Gaidet-Drapier, Hervé Fritz, Mathieu Bourgarel, Pierre-Cyril Renaud, Pierre Poilecot, Philippe Chardonnet, Craig Coid, Denys Poulet and Sébastien Le Bel
- **Year published:** 2006
- **Journal or source:** Biodiversity and Conservation
- **Full citation:** Gaidet-Drapier, N., Fritz, H., Bourgarel, M., Renaud, P.-C., Poilecot, P., Chardonnet, P., Coid, C., Poulet, D. and Le Bel, S. 2006. Cost and efficiency of large mammal census techniques: comparison of methods for a participatory approach in a communal area, Zimbabwe. Biodiversity and Conservation 15(2): 735–754.
- **DOI:** 10.1007/s10531-004-1063-7
- **Access status:** Hosted (freely available elsewhere)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/2006_Gaidet-Drapier_and_Fritz_etal_cost-and-efficiency-of-large-mammal-census-techniques.pdf

## Tags

- **Species:** Mammals > African buffalo; Mammals > African elephant; Mammals > Common duiker; Mammals > Greater kudu; Mammals > Impala
- **Region:** Africa > Zimbabwe
- **Methods and technology:** Aerial surveys; Descriptive statistics; Direct observation; Distance sampling; Road transects; Survey costs
- **Data type:** Existing data; New field data
- **Observer qualification:** Not needed
- **Discipline:** Just interesting; Population trends
- **Year published:** 2006

## Detailed keywords

- **Species:** Studied: all medium to large mammals over 200 g, 27 species in total, from genet (given as Genetta tigrina) to elephant (given as “Loxondota africana”, a misspelling of Loxodonta africana). Most often seen: common duiker (Sylvicapra grimmia), impala (Aepyceros melampus) and kudu (Tragelaphus strepsiceros). Elephant and African buffalo (Syncerus caffer) were used to compare ground and aerial counts. The paper does not list the other species.
- **Location and habitat:** Wards 2, 3 and 4 of the Dande Communal Area, Guruve District, middle Zambezi Valley, Zimbabwe (between 30° and 31° E and 15°30’ and 16°20’ S). Communal land of 2044 km², about 400 m altitude, former Zambezi floodplains with three main rivers. About 13,000 people live along the rivers, farming cotton and maize; about 80% of the area is uninhabited. Deciduous dry savanna dominated by mopane (Colophospermum mopane), a mosaic of woodland and shrubland 4 to 18 m tall. Comparison sites: an open shrub woodland and a thick wooded shrubland. Rainfall averages 750 mm a year; rainy season December to March. Comparison data mentioned from Hwange National Park, Zimbabwe.
- **Topics:** wildlife census and monitoring methods; cost effectiveness and efficiency of surveys; aerial census; day and night car counts with spotlights; bicycle counts; foot counts; water point counts from tree blinds; visibility and effective strip width; detection efficiency; community-based wildlife management; CAMPFIRE (Communal Areas Management Programme for Indigenous Resources); local monitoring agents (Anti-Poaching Units, Natural Resources Monitors, “Barefoot Biologists”); wildlife wariness outside protected areas; vegetation and visibility bias; cost per elephant seen.
- **Statistics or models:** Three indices calculated per transect and weighted by transect length, each with a mean and coefficient of variation (%CV): a sampling effort index (SEI, human-hours per km²), a sampling cost index (SCI, US dollars per km²) and a detection efficiency index (DEI, sightings per km² per replicate), plus elephant and buffalo versions (individuals per km²) and a site-specific version (DEISS) for two habitats. Effective strip width for bicycle counts from DISTANCE software (line transect analysis); visible distance for car and foot counts measured with a rangefinder. No hypothesis tests. Small inconsistencies: the text cites “Reilly and Hoskins 1999”, while the reference list has Reilly and Haskins 1999; the discussion says car and foot counts had similar “protocols, season and staff”, although Table 1 shows car counts in 1997 and foot counts in 1999. The 1999-only aerial figures for elephant (0.56 per km²) and buffalo (0.15 per km²) differ a lot from the two-year aerial means in Table 3 (0.25 and 0.60). This is possible across years, but worth checking.
- **Data type:** New field data: car counts June to October 1997; water point counts May to October 1997; bicycle counts September to December 1999; foot counts June to November 1999. Existing data: World Wildlife Fund (WWF) Zimbabwe aerial censuses, September 1997 and November 1999.

## Synopsis

### 1. What they did

Wildlife counts are usually compared by how precise they are, but projects run by rural communities on small budgets also need to know what each method costs. In the Dande Communal Area of Zimbabwe’s Zambezi Valley, where people and wildlife share the land, a project supporting the CAMPFIRE programme (Communal Areas Management Programme for Indigenous Resources) needed a way for local people to monitor large mammals.

The researchers compared six methods of counting animals by sight (no tracks or signs), all for mammals over 200 g:

- Aerial census by the World Wildlife Fund (WWF), from a light aircraft, in 1997 and 1999.

- Day car counts from the back of a pick-up truck, with binoculars and a rangefinder, on 12 road transects (1997).

- Night car counts on the same roads, with a 1000-watt spotlight to pick up eyeshine (1997).

- Bicycle counts, one observer riding slowly along 10 road transects, about 30 times each (1999).

- Foot counts, one observer walking eighteen 1.8 km transects cut into the bush, once a month (1999).

- Water point counts, two observers in a tree blind watching for 24 hours around full moon, at 27 water points (1997).

Ground observers were local agents: Anti-Poaching Unit staff, the District Council’s Natural Resources Monitors and locally trained “Barefoot Biologists”. For each method, the researchers worked out how many human-hours and how many US dollars it took to survey one square kilometre, and how many animal sightings it produced per square kilometre covered.

### 2. What they found

Overall

- 10,368 animals of 27 species were counted. Common duiker, impala and kudu were seen most often.

Effort and cost

- Ground methods moved at very different speeds: 11.6 km/h by car, 5.5 km/h by bicycle and 1.3 km/h on foot.

- How far observers could see also differed: 46 m for night car counts, 56 m for day car counts, 79 m (effective strip width) for bicycles and 91 m on foot.

- Effort to survey one km² ranged from 0.07 human-hours (aerial) to 4.80 (foot). Among the ground transect methods, bicycle counts needed the least effort (1.32).

- Cost per km² ranged from US$0.18 (water points) to US$6.35 (night car counts). Bicycle counts cost US$1.46 and foot counts US$2.74, against US$4.54 for the aerial census and US$4.58 for day car counts.

- Water point counts were the cheapest, but only if one assumes the watched water points cover the whole area, which the authors call a very constraining assumption.

Detection

- Sightings per km² rose from day car counts (0.39) to night car counts (2.09), bicycle counts (2.79) and foot counts (5.83).

- At water points, observers had about one sighting every 10 hours, against one to two sightings an hour on bicycle, foot and night car counts.

- On the same roads at the same time of year, car counts did better in open woodland than in dense shrubland; foot counts did better in dense shrubland.

- Elephants were detected far more often by bicycle (1.15 per km²) and foot counts (1.96) than from the air (0.25). Buffalo showed the same pattern but varied more.

- Seeing one elephant cost about US$1 on bicycle and foot counts, against US$8 at water points, US$18 from the air, US$38 on night car counts and US$44 on day car counts.

### 3. What they concluded

- Simple, slow, cheap ground methods, especially bicycle counts and foot counts, gave the best value in this communal area, with its low animal numbers, thick bush and hunted, wary wildlife.

- Observers on foot or bicycle approach quietly, so they can get close to wary animals before they flee. Faster methods (cars, aircraft) miss animals in thick bush.

- Bicycle counts need good, safe roads; foot transects can go anywhere but need paths cut and marked. Both cover only a small area per observer, so they need many observers counting at the same time and may bring in observer bias. The authors suggest these methods suit small management areas.

- Where animal density is high, methods that cover large areas (aircraft and cars) may be better.

- The authors did not test how precise each method is from year to year, or how well each detects population change. Sampling designs differed between methods, so the cost comparisons are rough.

### 6. Other technologies, skills or innovations

- Bicycles as a cheap survey vehicle, with step counts to measure distance to animals.

- A simple way to measure how far observers could see: a person walked away from the road until out of sight, and the distance was measured with a rangefinder. At night, the person held a reflector to imitate eyeshine.

- The ground counts built on the local agents’ everyday patrols for anti-poaching and human–animal conflict. Observers needed only a few days of training to collect data on their own, and some later trained new observers.

### 7. Implications for conservation (from the paper)

- Cost matters as much as precision when choosing a monitoring method, especially for community-based programmes with little money. The extra cost of a better count should not outweigh what the better count is worth to management.

- A monitoring method in a community-based programme must be affordable, efficient and supported by local communities. Building counts into the duties of local agents is a step towards the community running its own monitoring over the long term.

- Census methods should be adapted to local conditions and facilities rather than following one standard method.

## Cite this summary

Lawrence, K. 2026. Summary of: Cost and efficiency of large mammal census techniques: comparison of methods for a participatory approach in a communal area, Zimbabwe (Gaidet-Drapier and Fritz et al., 2006). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2006-gaidet-drapier-fritz-cost-efficiency-large-mammal/