# Estimating Carnivore Densities for Conservation Purposes: Indirect Methods Compared to Baseline Demographic Data

> How many cheetahs live in a park, and how would you know without years of fieldwork? Researchers compared four quick ways of estimating cheetah numbers in Kenya and Tanzania against counts of individually known animals. Simply asking park staff, scientists, tourists and farmers about their sightings came out most accurate, reaching 75 to 100% of the true numbers.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/1996-gros-kelly-estimating-carnivore-densities-conservation/

## Citation and access

- **Authors:** Paule M. Gros, Marcella J. Kelly and T. M. Caro
- **Year published:** 1996
- **Journal or source:** Oikos
- **Full citation:** Gros, P. M., Kelly, M. J. and Caro, T. M. 1996. Estimating carnivore densities for conservation purposes: indirect methods compared to baseline demographic data. Oikos 77(2): 197-206.
- **DOI:** 10.2307/3546058
- **Access status:** Publisher link only
- **Where to find it:** https://doi.org/10.2307/3546058

## Tags

- **Species:** Mammals > Cheetah
- **Region:** Africa > Kenya; Africa > Tanzania
- **Behaviour and ecology:** Behaviour; Home range
- **Methods and technology:** Geographic Information System (GIS); Interviews; Linear regression; Population estimation; Radio telemetry; Wilcoxon matched pairs test
- **Data type:** Existing data; New field data
- **Observer qualification:** Not needed
- **Discipline:** Behaviour and ecology
- **Year published:** 1996

## Detailed keywords

- **Species:** Studied: cheetah (Acinonyx jubatus), used as a model for large carnivores. Prey used in the prey biomass method: herbivores weighing 15 to 60 kg (species not listed individually); hares mentioned as smaller prey. Other carnivores mentioned only as examples from the literature: lion (Panthera leo), African wild dog (Lycaon pictus), spotted hyaena (Crocuta crocuta), wolf (Canis lupus), black bear (Ursus americanus), tiger (Panthera tigris), European badger (Meles meles), red fox (Vulpes vulpes), cougar (Felis concolor), coyote (Canis latrans) and Iberian lynx (Felis pardina).
- **Location and habitat:** Interview study areas in Kenya: Amboseli, Meru and Nairobi National Parks; Lake Bogoria, Masai Mara and Samburu National Reserves; Taita Hills Game Sanctuary; cattle ranches in Laikipia District. In Tanzania: Serengeti and Tarangire National Parks. Home range model built on the central Serengeti Plains (1738 km²), Tanzania. Published density and prey data from protected areas in Eastern and Southern Africa, including Chobe (Botswana), Etosha (Namibia), Kruger, Kalahari Gemsbok, Itala, Mkuzi and Umfolozi-Hluhluwe (South Africa), Hwange and Gonarezhou (Zimbabwe), Mkomazi and Ngorongoro (Tanzania) and Narok District (Kenya). Savanna, plains and woodland; the Masai Mara is described as having relatively closed vegetation.
- **Topics:** carnivore density estimation; calibrating indirect census methods; interviews and sighting reports; average density from published estimates; prey biomass regression (preferred prey of 15 to 60 kg); home range and demographic model; individual recognition by spot and tail-band patterns; female home range overlap; group living in cheetahs (male coalitions, family groups); detectability in open versus closed vegetation; conservation planning and rapid assessment.
- **Statistics or models:** linear regression of cheetah biomass on prey biomass (y = 0.002x + 0.21, r² = 0.616, p = 0.0071, n = 10); log-log regression between interview and prey biomass estimates (r² = 0.702, p = 0.0185); Wilcoxon matched pairs signed rank tests with an experimentwise error rate of 0.1 (each test at 0.02); minimum convex polygon home ranges; 1 by 1 km grid overlap counts (30 random squares) in Arc Info GIS; a demographic equation (population = adult females × 3.54). Inconsistencies in the paper: the text says three methods “consistently underestimated prey density”, which reads as a slip for cheetah density; the home range average deviation (51%) is based on two areas only, as Serengeti was not tested; the text gives “nine study areas” for the home range method while Table 3 lists ten (Serengeti included); Samburu is a National Reserve in the text but “NP” in Table 3, and Masai Mara is called both National Reserve and Game Reserve.
- **Data type:** New field data: interviews by P.M.G. in Kenya (July and September 1990) and Tanzania (September 1993 to May 1994). Existing data: Serengeti female home ranges and demography from 1988 (collected by Karen Laurenson), long-term individual recognition studies in Nairobi National Park (22 months), Masai Mara (17 months) and Serengeti (20 years), and published cheetah and prey densities.

## Synopsis

### 1. What they did

Protecting large carnivores needs reliable numbers, but carnivores are secretive, often active at night, and live at low densities. The most direct way to count them is to recognise every individual, which takes years of fieldwork. So managers often rely on quicker, indirect methods. The authors asked how accurate these indirect methods are, using the cheetah as a model.

They tested four indirect methods:

- Interviews. Biologists, park staff, tourists and farmers in ten study areas in Kenya and Tanzania reported their cheetah sightings (date, place, number, age class and sex). Sightings of groups with the same make-up were counted as the same animals, to stay on the cautious side.

- Average density. An average of reported densities from 13 protected areas in Eastern and Southern Africa (0.021 cheetahs per km²), multiplied by the size of each area.

- Prey biomass. A regression linking cheetah biomass to the biomass of their preferred prey (herbivores of 15 to 60 kg) in ten areas.

- Home range. A model built from the home ranges of female cheetahs on the Serengeti Plains in 1988: how big they were (414 km² on average) and how much they overlapped (4.82 at any point), then scaled up to the whole population with a demographic equation.

They compared each method with “reference” densities from three long-term studies where every cheetah was known by their spot and tail-band patterns: Nairobi National Park, Masai Mara National Reserve and Serengeti National Park. They also compared the four methods with each other across all ten study areas.

### 2. What they found

Accuracy against the reference densities

- Interviews were the most accurate: 75% of the reference density in the Masai Mara, 100% in Nairobi and 90% in the Serengeti. Their average deviation from the reference was 12%.

- Prey biomass came next (average deviation 37%), then the home range model (51%) and the average density method (53%).

- Interviews, prey biomass and average density all underestimated cheetah numbers. The home range model showed no consistent bias: it overestimated in the Masai Mara (146%) and underestimated in Nairobi (44%).

- With only three reference areas, the authors could not test whether these differences were statistically significant.

Agreement between methods

- Prey biomass estimates were significantly higher than home range estimates. No other pair differed significantly.

- After log transformation, interview and prey biomass estimates were correlated.

Why the methods went wrong

- Interviews did best in small, much-visited parks with open country. In the Masai Mara, cheetahs may be harder to spot in the more closed vegetation. In the Serengeti, cheetahs gather seasonally while following migrating prey, which may let people see most of them.

- Prey biomass may fall short because aerial counts often miss small and medium herbivores, and because cheetahs also take the young of larger migrating herbivores that the model leaves out.

- The home range model applied to Nairobi failed mostly in its demographic part: cheetahs there have unusually large litters and subadult groups.

- The average density method underestimated where cheetahs are dense, as in Nairobi and the Serengeti.

- Even the reference counts may be low: new adult males were still being found in the Serengeti after five years, and shy animals that avoid vehicles may never be photographed or drawn.

### 3. What they concluded

- Each indirect method works best only in certain conditions.

- Interviews give accurate, inexpensive estimates in small, much-visited protected areas, and could be run by reserve staff. The method only suits species that people recognise and remember, and whose individuals live in separate ranges or stable groups so they can be told apart.

- Prey biomass gives a good first estimate if local prey preferences are known and a reliable prey count exists.

- Average density only works across areas with similar ecology and human pressure; densities from cheetah strongholds overestimate numbers in poorer habitat.

- Home range is very “park-specific”, so it is best used to back up interview or average density estimates.

- Interviews and prey biomass can be used together, because the relationship between their estimates is known. Prey biomass and home range estimates should not be combined to monitor changes in numbers.

- The authors encourage researchers and managers to calibrate their methods and, where possible, to use more than one.

### 6. Other technologies, skills or innovations

- Individual recognition of cheetahs from photographs or drawings of spot and tail-band patterns, used to build the reference counts.

- Radio-collared females (located more than 10 times in 1988) gave the home range sizes.

- Sighting locations of all 51 known adult females on the Serengeti Plains were mapped in Arc Info Geographic Information System (GIS), with home ranges drawn as minimum convex polygons.

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

- Conservation planning often needs quick estimates of carnivore numbers, so indirect methods are widely used, yet they have rarely been checked against real numbers.

- Estimates from different methods can disagree, which makes it hard for planners to combine them into one strategy.

- At least one of these methods is available for many carnivore species, and the findings apply to monitoring carnivores generally, not only cheetahs.

## Cite this summary

Lawrence, K. 2026. Summary of: Estimating Carnivore Densities for Conservation Purposes: Indirect Methods Compared to Baseline Demographic Data (Gros and Kelly et al., 1996). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/1996-gros-kelly-estimating-carnivore-densities-conservation/