Commuting, territoriality and variation in group and territory size in a black-backed jackal population reliant on a clumped, abundant food resource in Namibia


  • N. Jenner,

    1. Institute of Zoology, Zoological Society of London, Regent's Park, London, UK
    2. Durrell Institute of Conservation and Ecology, School of Anthropology and Conservation, University of Kent, Canterbury, Kent, UK
    Search for more papers by this author
  • J. Groombridge,

    1. Durrell Institute of Conservation and Ecology, School of Anthropology and Conservation, University of Kent, Canterbury, Kent, UK
    Search for more papers by this author
  • S. M. Funk

    1. Institute of Zoology, Zoological Society of London, Regent's Park, London, UK
    2. Nature Heritage, St Lawrence, Jersey, UK
    Search for more papers by this author

  • Editor: Andrew Kitchener

Nicky Jenner, Fauna and Flora International, Jupiter House, Station Road, Cambridge CB1 2JD, UK. Tel: +44 012 2343 1961


This study examines the effect of a clumped, non-defendable and abundant year-round food resource (Cape fur seals) for black-backed jackal Canis mesomelas social structure and spatial organization at Cape Cross Seal Reserve and the National West Coast Recreation Area in Namibia during the jackals' denning period in 2004 and 2005. Geo-referenced observations of behaviour and space-use were used to test for territoriality, and to assess commuting distances, territory size, group size and within-territory density on the Namibian coast. Jackals displayed behaviour indicative of territoriality to within 50 m of the fur seal colony. In accordance with optimal foraging theory, jackals commuted between 0.45 and 20.03 km from their territory (low prey availability) to the seal colony (high prey availability). The observed within-population variation in group size (two to eight adults), territory size (0.20–11.11 km2) and within-territory density (0.31–9.80 jackals km−2) was unprecedented and strongly associated with distance from the food resource. Group and territory size increased, while within-territory density declined with increasing distance from the fur seal colony. We discuss the relative importance of the food resource and other factors in determining jackal social and spatial organization.


Among carnivores there is considerable variation in social and spatial organization between species, within species and within populations over space and time. An increasing body of evidence points to resource-based explanations (Macdonald, 1983; Geffen et al., 1996). For example, studies of social carnivores have linked territory size to the dispersion of resources (typically food) within their range, and group size to resource availability (Macdonald, 1983). Meanwhile, theoretical models predict territoriality breaks down when food resources become very abundant either because the amount of food exceeds intruder pressure such that competition ceases, or competitors are so numerous that excluding them would require more energy than is warranted by territory defence (Maher & Lott, 2000).

However, field studies suggest that traditional models fail to capture the complexity and flexibility observed in wild populations. For example, studies of spotted hyaena Crocuta crocuta, showed group and territory size become decoupled from measures of resources in the territory when individuals undertake regular extra-territorial foraging excursions, while territoriality persists in the presence of a super abundance of food (Hofer & East, 1993a,b). Investigating the impacts of resource-use in other species where populations are reliant on clumped and abundant food resources offers an opportunity to further elucidate the relative importance of resource-based explanations for understanding variation in carnivore social and spatial organization. To date such studies are rare, either because this scenario rarely occurs as a stable system in the wild or because it is difficult to observe when it does.

At Cape Cross Seal Reserve (CCSR) in Namibia a black-backed jackal Canis mesomelas population meets the unusual criteria of being reliant on a food resource that is clumped, abundant and available year-round. The black-backed jackal is a highly adaptive, medium-sized canid that occurs in various habitats, acting as predator and scavenger, and adopting an omnivorous diet that varies with food availability (Loveridge & Nel, 2004). At CCSR, jackals feed on Cape fur seals Arctocephalus pusillus which provide an abundant year-round food source (Nel & Loutit, 1986; Hiscocks & Perrin, 1987). The availability of alternative prey in the gravel desert that flanks the coastline is extremely low (Loveridge & Nel, 2004) thus jackals rely on fur seals as their primary food source and ‘commute’ from inland and along the coast to the colony to feed (Hiscocks & Perrin, 1988).

As facultative cooperative breeders, mated pairs are sometimes joined by subordinates that may help raise the current litter (Moehlman, 1983). At CCSR, the jackals' breeding season is highly synchronized. Pups are born during September–October in underground den-sites. Offspring emerge 3 weeks later (Moehlman, 1979), coinciding with the fur seal pupping season. Alloparental care may offset costly trade-offs between offspring care at the den and food acquisition away from the den. We therefore predict larger group sizes further from the food resource where costs of acquiring food (time and energy) are greatest.

Studies across the jackals' range report that the species is territorial, with each mated pair defending a shared territory (Loveridge & Nel, 2004). However, previous radio-telemetry studies at CCSR have suggested territoriality breaks down (Hiscocks & Perrin, 1988; Gowtage-Sequeira, 2005) based on large home-range overlap, high foraging densities and lack of territorial behaviour, though observational study was limited to opportunistic sightings. While there have been many interpretations of territoriality in the literature (see Maher & Lott, 2000 for review) it is generally accepted that territoriality is the maintenance and defence of an area through self-advertisement and aggressive/threat behaviour. A territory is the area actively maintained and defended. By definition, observations of self-advertisement and aggressive/threat behaviour are thus needed to detect territoriality.

In this paper, we challenge through behavioural study the conclusion of past studies that territoriality breaks down and we investigate the jackals' social structure (group size, presence of subordinates) and spatial organization (territory size and commuting system) in relation to the fur seal colony.

Materials and methods

Study area

Black-backed jackal groups were located within a 250 km2 area of the National West Coast Recreation Area, which encompasses the 60 km2 CCSR (21°46′S/14°00′E). This coastal area receives 0–50 mm rainfall annually (Barnard, 1998) with most moisture originating from coastal fog banks. The coastline comprises sandy beaches with hummock vegetation (e.g. Zygophyllum clavatum, Psilocaulon kuntzei) while salt flats, gravel plains and schist mountains inland support lichen fields and sparse vegetation in ephemeral riverbeds. The only other terrestrial carnivore is a small transient population of brown hyena Hyaena brunnea.

The area is uninhabited except for a ranger post, salt mine and lodge. Consequently, jackals are subject to limited human disturbance and active during the day. There are two artificial freshwater holes (c. 1 m diameter) located by the CCSR ranger post and Cape Cross Lodge. A salt road connects Swakopmund town and the Skeleton Coast National Park. Off-road driving is prohibited, with vehicles restricted to the few existing tracks. Thus our study concentrated 20 km north along the coast and 12 km inland.

Resource distribution and jackal diet

CCSR was established to protect one of Namibia's largest permanent Cape fur seal breeding colonies, estimated at 187 000 pups, cows and bulls (Gowtage-Sequeira, 2005). The colony stretched 5.24 km along the coastline during our study (Fig. 1). The fur seals form an abundant, clumped and spatially predictable food resource exploited by jackals year round (Nel & Loutit, 1986; pers. obs.). Jackals forage opportunistically on fur seal carcasses but also kill small adults (pers. obs.). The fur seal pupping season in November/December provides a glut of easily accessible food (live/dead pups, placental remains). Inland from CCSR, densities of rodents and other potential jackal prey are very low (Nel & Loutit, 1986). Fur seals constitute 86–95% of the jackals' diet even during the winter months when the colony is most reduced in size, and virtually no terrestrial prey is taken (Nel & Loutit, 1986; Hiscocks & Perrin, 1987).

Figure 1.

 Map of the study site illustrating defended territories, fur seal colony location (SC) and extent (white line), and the coastal road (black line). For three groups territories were mapped in both years of study. For one group, territory location and size changed substantially between 2004 (T1) and 2005 (T2). For two groups there was substantial overlap in territory location and size and therefore only territory boundaries in 2005 are shown.

Data collection

Data were collected during October 2004 to February 2005 and October to December 2005, in accordance with research permits issued by Namibia's Ministry of Environment and Tourism (No. 795/2005, 888/2005). This timeframe when jackals were constrained by having pups at a den, was selected because groups could be repeatedly located and observed with minimal disturbance, and territorial behaviour was expected to be more pronounced (Wolff & Peterson, 1998).

Individual identification and group size

Through a broader research programme, 56 jackals had been immobilized, sampled and ear-tagged (2002–2004), as described in Gowtage-Sequeira (2005). A unique combination of coloured ear tags facilitated identification of some individuals. Jackals were also individually identified using a digital photographic database. Sex determination was conducted using morphology and posture during urination; verified with molecular techniques (Jenner, 2008). Individuals were assigned to a group if repeatedly located in close proximity to the active den and/or within the same area. A suite of morphological and behavioural characteristics were used to identify the dominant pair and subordinates (Jenner, 2008). Group size was assessed by direct enumeration, and presence/absence of subordinates recorded as a binary response [0=none, 1=subordinate(s) present].

Commuting distance and spatial distribution of highways

We measured the distance each group lived from the fur seal colony by tracking individuals on foot from their den or resting place to the closest point of the colony, at a minimum distance of 25 m and following a 4-week habituation period. We recorded point locations using global positioning system (GPS) that were imported into ArcGIS v9.0 (ESRI, Redlands, CA, USA) and converted into continuous lines. Distance (km) was calculated using Hawth's analysis tools (Beyer, 2004). As route starting position varied over time (e.g. jackals moved dens; rested at different locations), we calculated average distance for each group, each season.

We quantified density of jackal ‘highways’, defined as well-trodden routes with individual tracks no longer distinguishable (Fig. 2), along a south–north gradient. There was no possibility for misidentification of jackal highways because brown hyaena tracks are considerably larger and game species absent. Highways were counted along 11 transects, each 1 km long and located within a 60 km stretch adjacent to the coast road. Transects were located between 6.20 and 34.95 km from the colony.

Figure 2.

 Jackal highways and a scent-marking rock demarcate the boundary of a territory in the National West Coast Recreation Area, Namibia.

Territorial behaviour, territory sizes and within-territory density

To establish whether jackals were territorial we observed and tracked individuals from 12 focal groups, three in both years of the study, during daylight hours (06:00–19:30 h) over a 6-week period from early November to mid-December in 2004 and 2005. Observations were conducted using established activity conventions (P. Moehlman pers. comm.) to identify agonistic encounters (including fighting, chasing, aggressive body postures) and self-advertisement, in the form of scent-marking and vocalizations by the dominant pair. The locations of behavioural observations were recorded using GPS.

To distinguish territorial scent-markings we considered only raised-leg urinations, scratching and rubbing performed by the dominant pair in tandem (i.e. male and female scent-marked the same site sequentially). In most cases we could track pairs on foot and during border patrols, and record exact scent-marking locations. During border patrols the dominant pair would trot or walk along territory boundaries, frequently sniffing and tandem scent-marking, and occasionally emitting loud vocalizations. In contrast, when jackals commuted to the colony, territory holders typically travelled at a fast trot, did not scent-mark frequently and would not always travel together. At the fur seal colony communication through advertisement or defensive/aggressive behaviour was usually associated with food and thus differentiated from territorial behaviour during border patrols and when individuals were located within their territory.

Following Höner et al. (2005), geo-referenced observations of agonistic encounters (N=164) and self-advertisement behaviour (N=1447) were recorded, along with locations of active dens with offspring (N=60) and potential dens where repeated digging by the dominant pair was observed (N=54). Territory size was calculated using the minimum convex polygon method (Harris et al. 1990) with Hawth's analysis tools (Beyer, 2004) in ArcGIS v9.0 (ESRI).

Within-territory density was calculated as group size divided by territory size. To aid interpretation of data on territorial behaviour and territory size we recorded number of different dens pairs utilized while pups were 0–12 weeks old and/or no longer den-dependent.

Statistical analysis

We performed statistical analyses in spss (release 16.0). In examining the effects of distance on group size, highway density and within-territory density, linear regression analysis was used. Distance was log transformed for all statistical analyses to aid visual interpretation of data. To test the effect of group size, presence of subordinates, number of dens and distance on territory size a generalized linear model (GLM) with normal error structure was used. Group size was square route transformed to stabilize variance for analysis. A GLM with binomial error structure was used to compare the effect of distance on the presence or absence of subordinates within a group.


Group size

We identified 26 groups; nine were monitored in both years. Group size ranged from two to eight (mean±se=3.14±0.30, N=35) with 15 (42.86%) of 35 pairs accompanied by one to six subordinates. There were no inconsistencies between methods of identification (ear-tag vs. digital photographs). Group size increased significantly (F(1,33)=7.860, R2=0.165, P=0.008) (Fig. 3) and subordinates were more likely to be present (Wald χ2=6.995, P=0.008) further from the fur seal colony.

Figure 3.

 Relationship between group size and distance from the fur seal colony.

Commuting distance and spatial distribution of highways

Visual tracking and re-sightings of individually identifiable jackals confirmed that all groups travelled to the colony. Commuting distance varied from 0.45 to 20.03 km (mean±se=5.70 km±0.96, N=35).

We recorded 39 highways in eleven 1 km transects (mean±se=3.50 highways km−1±1.17). Highway density declined significantly with distance from the colony (F(1,9)=13.626, R2=0.602, P=0.005), (Fig. 4).

Figure 4.

 Number of highways recorded per 1 km transect in relation to the distance from the fur seal colony.

Territorial behaviour

We recorded agonistic and self-advertisement behaviour by all 12 dominant pairs. Agonistic behaviours involved chasing (N=86) of non-group, often same-sex, individuals at a run or trot with ears pinned back, head down, back straight or arched and tail straight down or swishing. Intruders responded by moving away immediately or following display of submissive behaviour. Physical contact through fighting or biting was not recorded as part of territorial defence during the 6 week focal observations. We recorded 78 boundary displays in which neighbouring pairs would pace along their boundary in parallel, scent-marking frequently and emitting loud vocalizations.

Self-advertisement through vocalizations (N=38) and tandem scent-marking (raised-leg urination, typically accompanied by scratching and rubbing) by the dominant pair (N=1419) was observed within or on the border of their territory. Additional, opportunistic observations during October to February confirmed self-advertisement and defence behaviour persisted outside of the 6-week focal observation periods.

Territory size

Territory size was calculated for 12 different groups and for three in both years. Territory size ranged from 0.20 to 11.11 km2 (mean±se=3.12 km2±0.98, N=15), increased significantly with distance from the colony (Wald χ2=46.140, P=<0.001) (Figs 1 and 5), but was independent of group size (Wald χ2=1.180, P=0.292) and presence of subordinates (Wald χ2=1.392, P=0.238). Among groups that were successful in breeding there was no relationship between number of dens and territory size (Wald χ2=0.624, P=0.430).

Figure 5.

 Territory size in relation to distance from the fur seal colony.

For two of three groups for which territory size was estimated in both years, no substantial changes in territory size were observed. For one group, territory size declined by 64% in the second year of study (Fig. 1). Notably this group failed to breed in 2005. Although territory boundaries shifted over time there was consistency in space use by these three pairs between years, indicating site fidelity. Observations of another nine groups relocated in 2005, but for whom territory size was not measured, revealed that pairs exhibited between breeding season tenure and utilized many of the same den sites between years.

Within-territory density

Within-territory density varied between 0.31 and 9.80 jackals km−2 (mean±se=3.50 jackals km−2±0.80, N=15) and decreased further from the colony (F(1,13)=20.270, R2=0.568, P=0.001), (Fig. 6). Within-territory density did not equate to the inverse of territory size because the number of adults within a territory varied.

Figure 6.

 Within-territory jackal density in relation to distance from the fur seal colony.


On the Namibian coast, a Cape fur seal colony forms a clumped and abundant year-round food resource that promotes a commuter system among the resident black-backed jackal population. All jackals in this study travelled to the colony to feed. Outside the colony, jackals displayed behaviour indicative of territoriality and, during the jackals' denning period, dramatic within-population variation in social and spatial organization was observed, with territory and group size increasing and within-territory density declining further from the colony.

Commuting and territoriality

Jackals commuted up to 20 km from their den/resting site to the fur seal colony to feed; a system consistent with theoretical studies that describe an optimal distribution of animals between different areas offering varying levels of profitability (Fretwell & Lucas, 1970). Jackals leave areas of low prey availability to forage at the fur seal colony because the benefits of such trips outweigh the costs (Höner et al., 2005).

The costs to commuting jackals of aggressive physical contact with resident pairs may be low. Fighting between resident pairs and commuters was not observed during the study and is thus expected to be rare, with commuters avoiding territory holders, active dens and utilizing networks of common routes. Vocalizations by residents may play a role in reducing the probability of encounters (Sillero-Zubiri & Macdonald, 1998) and enable commuters to adopt appropriate, submissive behaviours should contact with territory holders be unavoidable.

The benefits of travelling to the colony to forage are high. Marine material provides a protein and energy-rich food resource (Rose & Polis, 1998) and jackals can consume large quantities with minimal disturbance. Consequently, energetic returns may outweigh costs in energy expended by jackals living up to 20 km away. It is important to highlight, however, that the population at CCSR is not isolated. Jackals move up and down the coast and we have evidence, in the form of fur seal hair and teeth in jackal faecal deposits, that jackals over 30 km inland forage at the fur seal colony. As prey availability becomes increasingly scarce, one would expect that with increasing distance in any direction from the colony a point would be reached when commuting becomes unprofitable. Further research along this coastal, and coastal-inland gradient is required to identify optimal and maximum distances at which jackals sustain commuting behaviour.

Previous studies have suggested that territoriality breaks down in the CCSR jackal population (Hiscocks & Perrin, 1988). However, we observed behaviour indicative of territoriality during the denning period: Pairs advertized territory ownership through tandem scent-marking and vocalizations, and boundaries were defended against intruders, in some cases to within 50 m of the colony, through aggression and display behaviour.

That previous studies have not detected territoriality may reflect the limited scope of observations, which failed to capture defence and self-advertisement behaviour, coupled with their focus on radio-telemetry and MCP analysis of foraging tactics, which are potentially problematic for detecting defended parts of an animal's home range. Radio-tracking is subject to error and MCPs are severely affected by outliers which can result in exaggerated home-range sizes and reporting of greater range overlap between individuals than actually occurs (Burt, 1943). Given the small size of some territories in our study (minimum 0.20 km2) it is plausible that these defended areas were masked by exaggerated estimates of home-range size (3.1–24.9 km2) and range overlap (Hiscocks & Perrin, 1988; Gowtage-Sequeira, 2005).

Traditional models of territoriality state that individuals defend territories to gain exclusive access to critical limiting resources such as food, shelter or mates (Burt, 1943). Jackals in this study exhibited territorial behaviour and defended areas that were ‘unprofitable’ in terms of food while suitable locations for den construction, whether for breeding or shelter to avoid low effective temperature (Dreyer & Nel, 1990), did not appear limited. Jackals are also physiologically able to survive without fresh water (Loveridge & Nel, 2004) and the two watering holes were not competed for. So what is being defended? We suggest it is the need for exclusive space to breed and raise offspring to independence that underlies existence of territoriality at CCSR. In support of this, records of infanticide at CCSR imply that defence of exclusive areas may confer benefits for offspring survival (Jenner, 2008). Furthermore, studies demonstrate that territoriality increases during mating (Loveridge & Nel, 2004) and may intensify during offspring rearing (Wolff & Peterson, 1998).

While lack of comparative data outside the denning season means we cannot assume year-round territoriality, several lines of evidence suggest that jackals may hold territories throughout the year. First, observations conducted ‘ad hoc’ during April to September (outside the denning season) confirmed presence of pairs within the area of their breeding territory. Second, we observed between-breeding season tenure: pairs observed in both years of the study exhibited site fidelity and re-used many of the same dens. If jackals are not territorial year round, re-establishment of territories and fresh allocation of dens would be required each year and one would expect that territories will not be held by the same pairs in subsequent breeding seasons. Thus, our observations of between-breeding season tenure coupled with the fact that offspring remain on their natal territory and do not disperse until sexual maturity at 11 months (Ferguson, Nel & de Wet, 1983; P. Moehlman pers. comm.), suggests that offspring protection could play a role in determining territorial behaviour throughout the year. Longer time-series data are needed to investigate this further and test for year-round territoriality.

Territory size, group size and population density

Our estimates of territory size are conservative and temporally sensitive, owing to the restricted timeframe of data collection when space use by parents was most constrained by having pups at a den. Average territory size (2.9 km2) at the study site was, however, comparable with findings elsewhere in the species' range (Loveridge & Nel, 2004). We would expect undefended home ranges to be considerably larger than defended territories, especially for jackals further from the colony, owing to the commuter system.

We observed unprecedented levels of within-population variation, with territory size varying by a factor of 55, increasing further from the colony. As territory holders did not appear limited by food, water or shelter within their territory, why should territory size vary so dramatically in relation to the colony? One hypothetical explanation is that jackals operate as ‘expansionists’ (Kruuk & Macdonald, 1985), with territory holders occupying available space and extending existing territorial boundaries until neighbouring dominant animals are encountered; a process affected by population density.

Linear density is reported to be high (7.0–32.0 jackals km−2) in and around the Cape Cross fur seal colony and is associated with heightened levels of intra-specific competition and greater intrusion pressure. This may increase defence costs at territory boundaries and lead to smaller territory size (Fretwell & Lucas, 1970). Linear density declines to 0.1–0.53 jackals km−2 along the coast (Loveridge & Nel, 2004), with similar trends expected inland. As breeding pairs become more dispersed, intra-specific competition for space will be reduced and territory holders may extend territorial boundaries to incorporate vacant areas and defend an area larger than would be required to sustain the group. This process of territory expansion has been documented in red foxes following removal of neighbouring groups and was not associated with changes in food availability, group size or relinquishment of existing space (Baker et al., 2000).

Defending a larger territory is likely associated with some costs, such as increased time and energy expended in producing and depositing scent-marks and patrolling territory boundaries. To offset such costs, some benefit must be gained. Expansionism is generally explained by the advantages accruing to membership of larger groups (e.g. alloparental care, cooperative defence, group hunting) outweighing costs of defending the large territory required to sustain them. However, we found no relationship between group and territory size, a finding we attribute to the carrying capacity of territories not being limited by availability of food, water or shelter. In one striking example, group size increased from a single pair in 2004 to a group of seven in 2005 while territory size declined by almost two-thirds (Fig. 1). Notably, this pair failed to breed in 2005. It is plausible that maintaining a larger territory confers benefits for reproductive success by increasing the distance between offspring and neighbouring conspecifics and reducing infanticide risk. This could outweigh costs of defending a larger area by pairs with offspring. Advantages of maintaining congruent territories may also accrue through elimination of interstitial areas between groups which can serve as a settling point for dispersing or itinerant individuals which may then seek to expand these small areas and establish their own territory at the expense of resident territory holders (Baker et al., 2000). One would expect that as breeding pairs become more dispersed, a point would be reached where costs of defending additional vacant areas outweigh any benefits and render an expansionist strategy untenable.

Black-backed jackal social organization varied with 43% of dominant pairs accompanied by one to six subordinates. Group size increased, and subordinates were more likely to be present, further from the colony. During the denning period, jackals face challenging trade-offs between the need to nurse, provision and protect offspring at the den and food acquisition and territory defence away from the den. Having additional group members that contribute towards offspring care (e.g. provisioning) may help offset this trade-off for breeders living further from the colony while subordinates gain direct and indirect benefits through group living and helping (Jennions & Macdonald, 1994).

An alternative, not mutually exclusive, explanation is that variation in territory size, population density and within-territory density influences dispersal behaviour of subordinates. During mating subordinates may be excluded from their natal territory (Loveridge & Nel, 2004). Whether subordinates ‘float’ in territory edges or disperse will depend on the balance of costs and benefits. Close to the colony, where territories are small, population density and within-territory density are high and there is high intraspecific competition for space, dispersal may be the favoured strategy. In contrast, further from the colony where territories are large and density is low, jackals may adopt a ‘floating’ strategy and later return to their natal (or other) group, assuming the benefits of not dispersing outweigh the costs.


Commuting systems have been described in other social carnivores reliant on clumped and unpredictable food resources (Hofer & East, 1993b; Höner et al., 2005). However, there are no records of jackals operating a commuting system elsewhere in their range, highlighting the flexibility of this adaptive species. Defence and self-advertisement behaviour revealed jackals were territorial during the denning season and we suggest that the need for exclusive space to breed and raise offspring underlies territoriality in this population while expansionist behaviour may underpin within-population variation in territory size. Long-term research is needed to assess the stability of behaviours and trends documented in our study during different times in the jackals' annual cycle, along coastal and coastal-inland gradients, and to elucidate relationships between territoriality, territory size and reproductive success.


Research was supported by the Zoological Society of London's Institute of Zoology, States of Jersey Education Department, and Nature Heritage and would have been impossible without assistance of: Amy Gander, Andy Temple, Chris Elvidge, Clare Marsden, Cristina Garcia, James Howard, Krystyna Golabek, Leo Hughes, Niall McCann, Peggy Poncelet, Phillippa Morrison, Rob Pickles, Sarah Brooke. We thank the Ministry of Environment and Tourism, Desert Research Foundation of Namibia, Gobabeb Training and Research Centre for research permissions and support. Special thanks to Anna Amukugo, Joh Henschel, Simone Hertzog, Job Kamati, Gerry Maritz, Felix Mettler, Hartmut Winterbach for logistic support and hospitality; Joh Henschel, Rod Braby, Patricia Moehlman, John Fa for insightful discussions; Trent Garner, Richard Pettifor and two anonymous reviewers for helpful comments on the manuscript.