Spatial ecology of cheetahs on north-central Namibian farmlands


  • L. L. Marker,

    1. Cheetah Conservation Fund, Otjiwarongo, Namibia
    2. Wildlife Conservation Research Unit, Department of Zoology, Oxford, UK
    Search for more papers by this author
  • A. J. Dickman,

    1. Cheetah Conservation Fund, Otjiwarongo, Namibia
    2. Wildlife Conservation Research Unit, Department of Zoology, Oxford, UK
    Search for more papers by this author
    • *Current address: Nuffield Building, Institute of Zoology, Zoological Society of London, Regents Park, London NW1 4RY, UK.

  • M. G. L. Mills,

    1. South African National Parks, Endangered Wildlife Trust and Mammal Research Institute, University of Pretoria, Skukuza, South Africa
    Search for more papers by this author
    • Current address: Kgalagadi Cheetah Project, P. Bag X5890, Upington 8800, South Africa.

  • R. M. Jeo,

    1. Cheetah Conservation Fund, Otjiwarongo, Namibia
    Search for more papers by this author
    • Current address: The Nature Conservancy, 4245 North Fairfax Drive, Arlington, VA 22203 USA.

  • D. W. Macdonald

    1. Wildlife Conservation Research Unit, Department of Zoology, Oxford, UK
    Search for more papers by this author

Laurie L. Marker, Cheetah Conservation Fund, PO Box 1755, Otjiwarongo, Namibia.
Fax: 264 67 306247


Knowledge of a species' ranging behaviour is both fundamental to understanding its behavioural ecology and a prerequisite to planning its management. Few data exist on the spatial ecology of cheetahs Acinonyx jubatus outside protected areas, but such areas are particularly important to their conservation. Cheetahs on Namibian farmlands occupied exceptionally large home ranges, averaging 1651 km2 (±1594 km2), with no detectable effect of sex, social grouping or seasonality. Despite such large ranges, cheetahs tended to utilize intensively only a small fraction of that area: 50% of the fixes were located within an average of 13.9±5.3% of the home range. Ranges were not exclusive, overlapping on average by 15.8±17.0%, with male cheetahs showing more intra-sexual range overlap than did females. Coalitions of males appeared to select for a dense, prey-rich habitat, but this preference was not apparent for other social groupings. Conflict with humans is an important contributor to the species' decline, and these large, overlapping cheetah home ranges result in the movements of each individual cheetah encompassing many farms (21 based on the average home-range size). Consequently, many cheetahs may be exposed to a minority of farmers attempting to kill them, and also that many farmers may see the same cheetahs, thereby gaining an exaggerated impression of their abundance. Conservation priorities for cheetahs outside protected areas are the development of techniques for conflict resolution, as well as the maintenance and restoration of suitable habitat and promotion of land-management practices compatible with the continued existence of large carnivores.


Although the creation and maintenance of a connected, representative protected-area network is important for both large carnivore persistence and biodiversity conservation (Margules & Pressey, 2000), the inadequate size of many protected areas means that management of surrounding lands may be equally important for conservation (Newmark, 1996). In Namibia, the majority of wildlife populations exist outside of protected areas (Barnard, 1998), with most of the country's ungulates occurring on commercial farmland (Richardson, 1998). This abundance of prey, coupled with the provision of artificial waterpoints and the widespread extirpation of large competitors, including lions Panthera leo and spotted hyaenas Crocuta crocuta, make commercial farmland a favourable refuge for cheetahs Acinonyx jubatus (Marker-Kraus et al., 1996). Approximately 90% of Namibia's estimated 3000 cheetahs are found on 275 000 km2 of farmland in the north-central region of the country (Morsbach, 1987; Marker-Kraus & Kraus, 1990).

This distribution has caused conflict with farmers, who perceive cheetahs as a significant threat to both livestock and ranched wildlife (Marker, Mills & Macdonald, 2003c). This conflict led to large numbers of cheetahs being killed or taken into captivity, with an estimated halving of the population size during the 1980s (Morsbach, 1987), and also meant that farmers who captured cheetahs and reported them to the Cheetah Conservation Fund (CCF) were often unwilling to have animals re-released onto their land. Therefore, cheetahs were often moved substantial distances, which could have a marked effect on their spatial ecology in this system. Although conservation efforts seem to have reduced this conflict (Marker et al., 2003c), understanding cheetah ecology in this landscape is crucial for developing effective management strategies.

Our goal was to describe cheetah ranging behaviour on Namibian farmlands. Inter-sexual differences in spatial ecology are widely reported (Wilson, 1975; Caro, 1994), and so we examined home-range size in relation to sex, social group composition, age and season. Various aspects of cheetah spatial ecology have been documented previously (Caro, 1994; Durant, 1998; Broomhall, Mills & du Toit, 2003), but the only detailed, long-term study to date has been conducted in Serengeti National Park. This study provides the first long-term information regarding cheetah ranging behaviour outside of protected areas.

Study area

Radio tracking was conducted within an 18 000 km2 area in north-central Namibia. The study area primarily consisted of commercial farmland, but also encompassed the Waterberg Plateau (a 48 × 16 km protected area), communal farmland and several fenced game-farms (Fig. 1). The area received an average of 472 mm rainfall annually, with 93% rain falling in the wet season (15th September to 14th April) and 7% in the dry season (15th April to 14th September). The area was generally flat, with slow rainfall run-off, no permanent rivers and numerous man-made semi-permanent water reservoirs.

Figure 1.

 Radio tracking study area, in the north-central Namibian commercial farmland, used to track radio collared cheetahs Acinonyx jubatus between 1993 and 2000.

Land use in the area was primarily commercial cattle and wildlife farming, with a low human population density, averaging 2.3 people  km2 (CIA, 2003). The majority of commercial farms were individually owned and ranged in size from 50 to 200 km2, with a mean of 80 km2 (Marker-Kraus et al., 1996). The area was situated in the Thornbush Savannah vegetation zone (Geiss, 1971), with Acacia, Dichrostachys, Grewia, Terminalia and Boscia being the dominant woody plant genera.


Between 1993 and 2000, radio collars were fitted onto 41 wild caught cheetahs on Namibian commercial farmland. Cheetahs were captured opportunistically, mainly by local farmers but also by CCF researchers, using box traps (see Marker et al., 2003a for details). Cheetahs were classified as being in one of the following social groupings: male coalition, single male or female (with or without cubs) for analyses. If one cheetah was caught, further traps were placed beside it until there was confidence, from lack of spoor or other signs of other cheetahs nearby, that the animal was a singleton or that all members of the social group had been captured.

Cheetahs were examined either at the capture site or at CCF headquarters. Immobilization was achieved using Telazol® (tiletamine-HCl and zolazepam-HCl; Warner Lambert, Ann Arbor, MI, USA), administered at 4 mg/kg intramuscularly. Cheetahs were fitted with a neoprene radio tracking collar with an external antenna (Advanced Telemetry Systems, Isanti, MN, USA), which had a life expectancy of >36 months. Radio collars weighed 280 g, equivalent to 0.56% of body mass for a 50 kg male and 0.76% for a 37 kg female, well below the 3% limit recommended by Kenward (2001). We found no evidence of collars affecting survival or behaviour, and the same design has been used previously without evidence of significant adverse effects (Laurenson & Caro, 1994).

Age classification was based on experience with captive cheetahs and upon information from previous studies (Burney, 1980; Caro, 1994) and derived from weight, body measurements, tooth condition, gum recession, pelage condition, reproductive condition and social groupings of animals (Marker & Dickman, 2003). We categorized adult cheetahs at the time of collaring as: newly independent (>18–30 months), young adults (>30–48 months), prime adults (>48–96 months) and old adults (>96–144 months). None of the cheetahs radio collared was estimated to be over 144 months old. To give confidence to the above process, a cementum ageing model, as described in Matson (1981), was used on a subsample of individuals recovered after death, as well as from known-age animals, which revealed a good correlation between estimated and actual ages (Matson's Laboratory, LLC, Milltown, MT, USA; Marker et al., 2003a). Full details of the criteria used to assign cheetahs to age classes are provided in Marker & Dickman (2003).

Depending on the landowner who had caught the animal, cheetahs were either released at the capture site, or on other farms where ranchers had given permission. The distance from capture to release site ranged from 0 to 600 km (Table 1). Only one cheetah per social group was radio collared, and cheetahs were released in the grouping with which they had been captured. Young, dependent cheetahs were only released if they were captured with their mothers, and animals were not collared unless they were fully grown. Wherever possible, radio collars were retrieved at the end of the project, but this was not always feasible due to the opportunistic nature of cheetah capture, and occurred in 63% of the cases.

Table 1.   Information regarding all cheetahs Acinonyx jubatus radio-tracked on the Namibian farmlands during this study
Cheetah ID#Social
Distance (km)
from capture
site to release site
age group at
No. of
Total no.
of fixes
% flights
Overall HRSCore HRS
(50% kernel)
Wet season
Dry season
No. of
No. of
  1. Social group categories are as follows: SM, single male; CM, coalition male; F, female. Cheetahs were assigned to one of the following age groups at time of radio collaring: 1=newly independent (>18–30 months), 2=young adult (>30–48 months), 3=prime adult (>48–96 months) and 4=old adult (>96–144 months). None of the radio collared cheetahs was estimated to be >144 months old.

  2. MCP, minimum convex polygon; HRS, home-range size.

821SM0343.5June 93–June 94136973.4272596.138.831193.338715.3
842SM50350.0May 93–November 941912493.91146.5774.2254.140940.2841009.6
867SM75240.0July 93–January 9475084.81763.71343.0171.4221474.8281178.2
868SM75242.0July 94–December 963013277.21723.84221.5268.057909.5752607.3
881SM0346.0August 93–April 952112089.6281.6419.116.965260.755362.7
952SM0348.5June 95–April 96115696.6119.6266.316.828117.628210.4
977SM0248.0June 95–July 952880.0
985SM0443.0October 95–March 97187184.53938.12514.9833.3433281.5283864.4
1025SM0355.0October 96–March 98186490.1574.1989.547.640390.124599.8
1043SM350350.0June 97–February 9892575.8176244.8
1061SM200354.0July 99–December 00185898.31297.9917.257.1311331.5271540.4
1062SM200347.0September 97–October 972360.0
1105SM50343.0June 98–May 99123071.42205.45658.2345.6181298.5126045.1
1158SM0242.0February 99–October 00218391.21227.7382.0135.528900.8331431.6
1163SM200240.5April 99–October 00195998.33333.12292.3450.9262519.6333125.6
All single males (n=15)Mean8046.214.763.584.31490.31697.9219.735.81134.837.12225.8
831CM50243.0April 93–April 952414085.41406.43402.8140.9631156.3771828.2
835CM0245.0October 95–November 952666.7
865CM0148.0July 93–August 952617294.0710.7824.678.872452.2100937.3
869CM75246.0October 93–April 9474493.6644.1385.398.041621.3
937CM150238.5September 94–October 942787.5
974CM0248.0June 95–June 994921696.41076.21371.6138.5101904.61151324.0
979CM0351.0August 95–December 96178097.6680.4712.874.640807.540552.3
990CM0354.5December 95–April 984111595.8544.52437.649.163312.6521129.3
1063CM200150.0September 971342.90
1164CM0240.0April 99–July 99412100.0
1167CM0142.0September 97–May 9893083.34347.62124.2872.3202837.7108496.4
All coalition males (n=11)Mean43.246.016.575.085.71344.31608.4207.557.11013.265.72377.9
All males (n=26)Mean64.446.115.568.384.91436.51664.9215.243.61090.046.12273.8
878F450135.0September 93–October 932981.8
948F260134.0October 94–December 007925786.94024.76353.7473.41293617.91283920.9
967F04February 95–August 994321892.01029.31445.8157.7100206.8118619.8
978F300346.0October 95–March 97186798.5999.91190.8217.6451473.222147.8
984F275130.0October 95–December 006322094.41282.51041.7221.51191357.51011232.0
986F0334.0October 95–March 9662996.7281219.2
992F350342.0January 96–February 962861.5
1026F50331.0November 96–July 9793382.5553.9324.067.222479.811430.7
1084F0340.0January 98–May 9851995.015957.3
1100F600336.0June 98–December 99194894.17063.33012.11795.3192686.2296654.4
1107F125125.0September 98–October 9823100.0
1144F0235.0December 98–March 9941083.3
1154F0235.0March 99–September 00195698.21705.61016.0126.7281369.7282090.8
1177F600239.0December 9911100.0
1184F0134.0October 99–December 00144797.9626.3306.6122.730683.817444.0
All females (n=15)Mean200.735.419.168.390.92160.71836.3397.853.51405.156.81942.6
All cheetahs (n=41)Mean114.342.416.868.387.11651.11715.7269.347.01198.749.32175.7

Radio tracking

Following release, radio-collared cheetahs were tracked from a Cessna 172 aeroplane, utilizing a dual antenna procedure, with the animal's location determined using a portable global positioning system. Between May 1993 and May 1996, aerial tracking was conducted twice a week, while from June 1996 to December 2000 it was conducted once a week.

Home-range area and overlap calculations

Data were plotted and analysed using ArcView GIS (version 3.2, ESRI, Redlands, CA, USA) and the Animal Movement extension (Hooge, Eichenlaub & Soloman, 1999). Latitude and longitude recordings were used to calculate 95% minimum convex polygon (MCP) home ranges (White & Garrott, 1990), as well as 95 and 50% adaptive kernel home-range estimates (Worton, 1989; Seaman et al., 1999). Four estimates of home-range size were calculated: (1) overall (the entire length of time a cheetah was tracked); (2) annual (based on 12-month periods from the time of collaring); (3) dry season; (4) wet season home-range size. Analysis was restricted to cheetahs with enough fixes to reach an asymptotic level, as determined using Ranges V (Kenward & Hodder, 1996), and was set at ≥30 fixes within a year for overall and annual home range and ≥15 fixes within a season for seasonal home range. When the effect of age on range size was being investigated, analyses were restricted to the first year after collaring, to improve the likelihood that cheetahs remained in the age group in which they had originally been classified.

Core home-range size was defined as the 50% probability kernel, and was determined for all cheetahs whose overall home range had been calculated. The degree of home-range overlap between cheetahs tracked concomitantly was calculated for each year of the study using the dynamitic interaction analysis as described in Ranges V (Kenward & Hodder, 1996). This analysis uses Jacob's Index (Jacobs, 1974), which compares the observed and possible distances between each range and provides a single index for each pair of animals.

Habitat selection

The habitat type in which each radio-telemetry fix occurred was visually classified from the air, and categorized in terms of bush density, namely sparse (<30% canopy), medium (30–75% canopy) and thick bush (>75% canopy). We also flew stratified random transects 20 km apart over the entire study area, and every 5 km, bush density was visually classified as above. This allowed estimation of the relative proportions of each habitat type available to cheetahs in the study area, providing a baseline for assessing habitat selection.

Prey density in different habitat types was calculated by driving strip counts across 70 km2 of the study area, while the availability of different habitat types across that area was assessed using aerial photographs and ground observations. A minimum of three counts, over a standardized 50 km route, was conducted each month. The program DISTANCE (Thomas et al., 1998) was used to estimate prey density, and the strip was classified by habitat type, as above. The relative utilization of habitat types by prey species was examined alongside cheetah habitat selection, to investigate whether cheetahs' favoured habitat that was selected for by ungulates.

Data analysis

Analyses were conducted using the statistical package SPSS version 12.0.1 (SPSS Inc. Chicago, IL, USA). The statistical tests used depended on the distribution of the data and included χ2, t-test, z-test, Kruskal–Wallis (KW) χ2, analysis of variance and Spearman's and Pearson's correlations, with P<0.05 considered to be significant.


Forty-one cheetahs (26 males, 15 females) were radio-tracked between April 1993 and December 2000 (Table 1). Cheetahs were located on 87.4% (±12.6) of the flights, during which they were searched, and overall, annual and seasonal home ranges were determined from a mean of 68.34 (±68.14) fixes per cheetah, and 7.4 (±2.8) days between fixes.

Because cheetahs were opportunistically caught, we examined their ages in order to reveal potential biases in sampling before comparing social groupings and sexes (Table 1). There was no significant difference between sexes regarding age category at collaring (z=−0.562, P=0.602); however, there was between social groupings (KW χ2=6.549, P=0.038), as single males were significantly older at the time of collaring (z=−2.73, P=0.006) when compared with other groupings. There were no significant differences between the mean number of fixes (F=0.087, P=0.917), percentage of flights on which they were located (F=1.070, P=0.353) or the number of months tracked (KW χ2=0.472, P=0.790).

Overall home-range size

Twenty-seven cheetahs yielded sufficient fixes to estimate overall home-range size, which revealed a mean overall range size of 1651.1 km2 (±1594.2 km2) and a median overall range size of 1146.5 km2 (Table 1). Estimates of overall home-range size produced using the 95% kernel method did not vary significantly from those using the 95% MCP, for any of the social groupings (single males: z=−0.115, P=0.908; coalition males: z=−0.831, P=406; females: z=−0.315, P=0.753). Therefore, the 95% kernel method was used for further estimates of home-range size.

No statistically significant differences were detected in home-range size between sexes (t=−1.081, P=0.290) or social groupings (F=0.580, P=0.586). For all social groupings, the overall home-range sizes estimated here were significantly larger than those reported elsewhere (single males: t=−5.15, P=0.036; coalition males: t=−18.9, P=0.001; females: t=−15.8, P<0.001: Table 2). Despite the general trend for vast ranges, some individuals still managed to exist within relatively small areas. Three single males, with a mean of 46 months of age, each occupied overall ranges of <300 km2, which might be considered to be large for cheetahs elsewhere, but tiny compared with the averages found here. In addition, ranges shifted between years (Fig. 2), and some home ranges were shown to increase when a male coalition was reduced to a single male (Fig. 3).

Table 2.   Estimates of cheetah Acinonyx jubatus home ranges sizes, and methods of estimation, reported in this study and elsewhere in Africa
AreaMethod of HRS
FemaleCoalition maleSingle maleMale
  1. Figures in parentheses indicate sample size, that is the number of single males, single females or coalitions of males radio collared.

  2. MCP, minimum convex polygon; HRS, home-range size.

  3. *Male ranges given for the Serengeti refer to resident and non-resident males rather than coalitions and singletons respectively.

Namibia (north-central farmlands)95% kernel2161 (15)1344 (11)1490 (15)This study
Namibia (north-central farmlands)95% MCP1836 (15)1608 (11)1698 (15)This study
Serengeti Plains*Minimum polygon833 (19)37 (22)777 (9)Caro (1994)
Kalahari Gemsbok/Gemsbok National ParksSightings320 (4)125 (3)Mills (1998)
Kruger National Park95% MCP161 (2)126 (1)195 (1)Broomhall et al. (2003)
Kruger National Park95% kernel212 (2)188 (1)250 (1)Broomhall (2001)
Kruger National ParkConvex polygon160 (2)332 (2) Mills (1998)
Figure 2.

 Home range of a male coalition group (cheetah ID# 974) during the dry season of 1995 (a) and 1996 (b) showing a major shift in home range between the 2 years.

Figure 3.

 One of the largest home ranges during the 10-year study was of a single male cheetah Acinonyx jubatus (ID# 985), which, after losing his coalition member, continued to shift his movement patterns and increase his home-range size.

Annual home-range size

Annual home-range sizes could be calculated for 23 cheetahs (Table 3). The mean annual range sizes did not differ significantly from the overall home-range size for any of the social groupings, and did not vary significantly between age groups, sexes or social groupings.

Table 3.   Mean annual home-range size (HRS) (estimated using the 95% kernel method) for all cheetahs Acinonyx jubatus radio-tracked during the study for at least 12 months, split by sex and social group by age
Cheetah ID#Social groupingEstimated age group
at collaring
Years radio trackedMean annual
HRS (km2)
985Single maleOld adult964191.5
821Single malePrime adult93–94272.0
831Single malePrime adult94519.9
842Single malePrime adult93–94907.5
881Single malePrime adult94246.3
952Single malePrime adult95–96119.6
1025Single malePrime adult97706.7
1061Single malePrime adult20001203.0
1105Single malePrime adult98–992205.4
868Single maleYoung adult95–96722.5
1158Single maleYoung adult99–2000736.6
1163Single maleYoung adult99–20001524.9
All single males (n=12)Mean  1113.0
sd  1134.1
974Coalition malePrime adult96–98514.5
979Coalition malePrime adult96541.7
990Coalition malePrime adult96–97862.5
865Coalition maleNewly independent93–95615.4
All coalition males (n=4)Mean  633.5
sd  158.5
All males (n=16)Mean  993.1
sd  997.1
967FemaleOld adult96–98984.3
978FemalePrime adult96561.3
1100FemalePrime adult992323.9
1154FemaleYoung adult99–20001000.7
948FemaleNewly independent95–20001678.7
984FemaleNewly independent96–2000905.7
1184FemaleNewly independent2000637.9
All females (n=7)Mean  1156.1
sd  629.2
All cheetahs (n=23)Mean  1042.7
sd  889.8

Seasonal home-range size

Neither wet nor dry season home-range size differed significantly between sexes, social groupings or age groups (Table 1). There was no significant difference between the size of wet and dry season home ranges for any of the social groupings: their boundaries shifted over time but did not differ significantly in terms of overall size.

Core home-range size

No significant seasonal variation was observed regarding core home-range area for any of the social groupings. Core areas comprised a significantly smaller percentage of single males' overall home range in the wet season (11.3±5.0%) than in the dry season (14.5±2.9%; z=−2.13, P=0.034), but no significant difference was found for the other social groupings.

The sizes of coalition males' core home ranges were significantly smaller than those of females (z=−2.19, P=0.028), but there were no detectable differences between other social groupings. Core areas comprised on average 13.9±5.3% of the home-range size, with no significant difference between social groupings.

Effect of distance moved between capture and release sites

The distance moved from capture to release site was positively correlated to the overall home-range size (Table 4), but this may be an artefact of social groupings moved: coalition males, which tended to have smaller home ranges, accounted for 35% of those released at the capture site, but only 19% of animals moved. When examined separately, there was no significant relationship between distance moved and overall home-range size for any of the social groupings. Distance moved had some effect on spatial ecology, however, releasing a cheetah further from the capture site was linked to larger core home ranges for females, and to ranging further in the first year for single males (Table 4). There did not seem to be a marked effect of cheetahs' rapidly travelling long distances to return to original locations: cheetahs released away from their capture site did not range over larger areas in the first year than in subsequent years (t=0.576, P=0.577).

Table 4.   Relationship between distance moved from capture to release site, and overall, core, first season and first annual home-range size, for cheetahs Acinonyx jubatus of various social groupings radio-tracked on the Namibian farmlands
Social groupVariable being examinedCorrelation with distance moved
  • *


  • **


All cheetahsOverall home-range size0.440280.019*
Core home-range size0.482280.009**
Home-range size over the first season0.181300.339
Home-range size over the first year0.394220.069
Coalition malesOverall home-range size−0.08970.849
Core home-range size0.26770.562
Home-range size over the first season−0.43960.383
Home-range size over the first year0.70750.182
FemalesOverall home-range size0.2790.483
Core home-range size0.71490.031*
Home-range size over the first season−0.524100.12
Home-range size over the first year−0.21360.686
Single malesOverall home-range size0.541120.07
Core home-range size0.466120.127
Home-range size over the first season0.671120.017*
Home-range size over the first year0.553110.008

Home-range overlap

The large sample size of cheetahs collared, and the close proximity of their capture sites (Fig. 1) allowed range overlap to be calculated for cheetahs utilizing the same area in the same year. Mean home-range overlap averaged 15.8±17.0% (minimum=0%, maximum=74%, median=9.83%) across all collared cheetahs (Fig. 4), with males exhibiting a significantly greater intra-sexual range overlap than females (z=−2.23, P=0.026). However, more males than females were tracked in the study area, which could lead to underestimation of the true extent of female intra-sexual overlap. Single males overlapped with one another significantly more than females did (z=−2.10, P=0.036), but there were no significant differences when other social groupings were compared. There are caveats to interpreting these data, however, they do not take account of overlap with cheetahs that were not radio-tracked, and extensive movements by animals released far from their capture sites (particularly single males in the first year) may increase the range overlap that those cheetahs exhibited with others, at least in the short term.

Figure 4.

 Degree of home-range overlap for cheetahs Acinonyx jubatus tracked concomitantly on the commercial farmland, shown for each year of the study. The mean overlap for all cheetahs tracked that year is shown at the top of each map, while the figure in parentheses denotes the standard deviation.

Habitat selection

Overall, radio-collared cheetahs appeared to utilize the three habitat types approximately in proportion to their availability, but when examined by social grouping, female cheetahs seemed to use thickly bushed areas less than their availability, preferring medium bush (χ2=7.37, P=0.025). Conversely, coalition males appeared to select for areas of thick bush while utilizing medium bush areas far less than expected (χ2=23.9, P<0.001). Single males showed no evident selection, utilizing all habitat types in approximate proportion to availability.

The habitats in which radio-tracked cheetahs were recorded were not closely correlated with those in which ungulates were detected. The majority of ungulate species, particularly large prey, showed a preference for dense bush (Fig. 5a and b), whereas overall, radio-collared cheetahs were recorded in dense bush in proportion to its availability.

Figure 5.

 (a) Relative habitat utilization scores for radio-collared cheetahs Acinonyx jubatus and for the seven key prey species monitored during the study. Scores above 1.0 indicate utilization of a habitat proportionally more than its availability in the study area, while scores below 1.0 indicate lower usage than would be expected from the amount of that habitat available. The horizontal line indicates parity, that is the level (1.0) at which habitat is used in direct proportion to its availability. (b) Relative habitat utilization scores for radio-collared cheetahs of various social groupings; namely, coalition males, single males and females, and for prey species monitored during the study. The prey species are classified as either large (oryx, kudu, red hartebeest and eland) or as small (duiker, steenbok and warthog).

When examined by social group, however, coalition males did show some preference for the dense bush preferred by ungulates (r=1.00, P=0.019), and particularly for habitat preferred by large prey (r=1.00, P=0.009). Single males showed no relationship between their habitat preferences and those of prey species, while females actually seemed to avoid those habitats preferred by ungulates (r=−1.00, P=0.046).


Understanding the factors influencing cheetah spatial ecology in Namibia is fundamental to developing effective and appropriate regional conservation strategies for the species. The most striking result of this study was the very large home-range sizes of cheetahs, found across all sexes, social groupings, age groups and seasons. The long-distance movements of cheetahs elsewhere, particularly of females in the Serengeti, have been attributed to seeking out migratory prey after ephemeral rains (Caro, 1994), and Broomhall (2001) also found in Kruger National Park that dependence upon migratory prey was linked to large ranges. This explanation was inappropriate for the large ranges revealed for Namibian farmland cheetahs, where both sexes and all social groupings ranged extensively while reliant upon sedentary prey.

It is well established that intra-guild hostility from larger carnivores can lead to high levels of juvenile mortality and kleptoparasitism among cheetahs (Caro, 1994; Laurenson, 1994), which may therefore attempt to avoid competitors to minimize such risks (Durant, 2000). Such avoidance may result in cheetahs ranging across larger areas than would be predicted on the basis of prey availability. However, both lions and spotted hyaenas have been widely eradicated on Namibian farmlands (Marker, 2002), but cheetah ranges were nonetheless much larger than those recorded where cheetahs coexist alongside larger carnivores. Furthermore, in the Serengeti consistent differences between the sexes and social groupings have been reported, with females occupying large, overlapping ranges and territorial males maintaining much smaller areas (Caro, 1994). In this study, we did not have the behavioural data needed to assess territoriality, but found no significant differences between sexes or social groupings in range size, with all utilizing very large areas. However, the smallest home ranges were of single males of prime breeding age, the age group most frequently removed by people from the farmlands (Marker et al., 2003b). These cheetahs may be seen more regularly by farmers, and therefore are removed. Namibian cheetahs were found to be as highly mobile and often closely related to others in their home-range areas (Marker et al., accepted), possibly contributing to the lack of exclusivity. We found that most home ranges shifted over time, which may be affected by fluctuating drought cycles; for instance, during the worst drought of the past two decades in 1996, one of two radio-collared groups moved into the other's territory and were killed by the other coalition (L. L. Marker, pers. obs.).

An explanation, therefore, for the ubiquitously large ranges recorded here could be low annual rainfall, averaging 468 mm in Namibia (Marker, 2002), compared with 600–700 mm on the Serengeti plains (Maddox, 2002). Precipitation is often a corollary of prey biomass, with most expansive ranges in the driest areas (Stander et al., 1997), but this effect may be diminished on farmland with man-made waterpoints. Moreover, range sizes in Namibia did not differ significantly between the wet and dry seasons, indicating that another factor aside from rainfall is influencing the need for such large areas. Most cheetahs also did not appear to select for areas based simply on prey density, suggesting again that some other variable is affecting home-range size and configuration.

Understanding cheetah spatial ecology here may require examination at a finer scale: Muntifering et al. (2006) used these radio-telemetry data to define ‘high-use’ areas that cheetahs occupied relatively frequently – these were relatively small habitat patches with good visibility and grass cover, which may be advantageous for stalking. Observations of cheetah hunting have revealed the use of edges of dense habitat patches to provide cover for stalking, and so cheetahs may configure range use to include a matrix of these different habitat patches (Frame & Frame, 1980; Caro, 1994). However, it should be noted that using weekly radio-telemetry fixes may not be the best technique for assessing habitat importance, and these data will be influenced by the time of day that the fix was obtained – more accurate information on daily habitat use would require finer-resolution studies such as the use of collars with frequent fixes. Despite this caveat, differences between ‘high-use’ and ‘low-use’ areas (Muntifering et al., 2006) suggest that there may be important small-scale variation in habitat and vegetation determining which parts of their ranges cheetahs are likely to utilize more intensely, which could have important implications for understanding and resolving conflicts on farmlands.

Despite having such large home ranges, a finer-scale examination of core areas revealed that cheetahs tended to utilize intensively only a small fraction of their overall ranges, with very large ‘peripheral’ areas used less intensely.

The reason for this is unclear, but the size of the overall ranges and the use of such large peripheral areas could feasibly be linked to the serious perturbation recently suffered by this population due to removals from the farmlands (Marker et al., 2003b). One of the largest home ranges was from a male cheetah (ID# 985) that had lost its male coalition partner, which was killed by a farmer. Such perturbation has been shown to affect a species' ecology and behaviour (Tuyttens & Macdonald, 2000), and high levels of human intervention are likely to have had a substantial impact on Namibian cheetahs, and so the consequences of disturbance could be an important factor affecting cheetah spatial ecology in this region. Further work on the features of intensively utilized patches compared with peripheral range areas, the spatial stability of ‘core’ areas across seasons and years and the impact of human disturbance on cheetah movements would provide useful information regarding the determinants of cheetah range use and habitat selection on the Namibian farmlands, which could in turn have important implications for cheetah management and conflict resolution.

Conservation implications

The range sizes found for cheetahs on Namibian farmland were unexpectedly large, considering that the region is typified by widely available water and a sedentary prey base. These very large, overlapping home ranges have important conservation implications, in that each cheetah will likely range over many farms, and so even a few hostile farmers could have a significant effect on the local cheetah population with these farms acting as population sinks. Given the average farm size of 80 km2 in the study area, the mean cheetah home-range size revealed here (1651 km2) shows that each home range could well incorporate 21 different farms during the course of the year. Assuming an average group size of 2.5, this would lead to an estimate of 53 ‘cheetah-farm encounters’ in one home range alone. Bearing in mind the degree of range overlap (16% here, which is almost certainly conservative, as it does not incorporate non-collared cheetahs in the study area), this would rise to an estimated 60 ‘cheetah-farm encounters’ within one home range annually. Sightings of cheetahs on multiple, discontinuous farms could mistakenly be taken as evidence for many cheetahs in an area, whereas it could easily be the same cheetah or group of cheetahs simply moving through its range. It would be easy for farmers to overestimate local cheetah population numbers based on encounter rates: just one of those farmers killing cheetahs could have impacts over an area of more than 1500 km2. In Kenya, research found that the actions of just one rancher, on a 180 km2 ranch, had direct impacts on lion populations over 2000 km2 (Woodroffe & Frank, 2005). Given that adult females, potentially the most valuable component of the population in terms of long-term viability, range over the largest areas, the size of these ranges is of particular significance for cheetah conservation in Namibia. It is important to implement effective conservation strategies over very large areas of unprotected rangeland to achieve country-wide cheetah conservation in Namibia, and reduce human-cheetah conflict on the farmlands. Fundamental to this is developing economic advantages to maintaining carnivores on private land (Sillero-Zubiri & Laurenson, 2001), such as through ecotourism, trophy hunting and incentives for predator-friendly farming (Archabald, 2000; Lindsey et al., 2005; Woodroffe, Thirgood & Rabinowitz, 2005). Additionally, if cheetah are to survive on farms without depending on domestic stock, then creating conservancies where wild prey survive sustainably will be essential (Marker, 2002). Such strategies, integrating conflict resolution with more effective land and wildlife management, will be critical to conservation outside protected areas, both for cheetahs and for other threatened large carnivores.


We would like to thank the Namibian farmers, in particular those from the Waterberg Conservancy and Osondjache Farming, and the Namibian Government, particularly the Ministry of Environment and Tourism, without whose support this research would not have been possible. We are grateful to the CCF staff, especially Bonnie Schumann, Daniel Kraus and Matti Nghikembua and volunteers, and to our CCF pilots, Jacques Imbert and Arthur Bagot-Smith. We are indebted to Bruce Brewer, Tim Caro, Sarah Durant, Paul Honess, Paul Johnson and to anonymous reviewers for their valuable input. Partial funding for this research was provided by the Bay Foundation, CCF-USA, Earthwatch Institute, the Tapeats Foundation, the Weeden Foundation, the WILD Foundation and the WWF SA Green Trust, and thanks are due to Fort Dodge, Iowa, for the donation of Telazol, as well as to ESRI for assistance with training in GIS techniques and donation of Arc-View software.