Acta Theriologica

, Volume 57, Issue 3, pp 225–231

The response of lions (Panthera leo) to changes in prey abundance on an enclosed reserve in South Africa


    • Wildlife and Reserve Management Research Group, Department of Zoology and EntomologyRhodes University
    • Kwandwe Private Game Reserve
  • Ric T. F. Bernard
    • Wildlife and Reserve Management Research Group, Department of Zoology and EntomologyRhodes University
  • Daniel M. Parker
    • Wildlife and Reserve Management Research Group, Department of Zoology and EntomologyRhodes University
Original Paper

DOI: 10.1007/s13364-011-0071-8

Cite this article as:
Bissett, C., Bernard, R.T.F. & Parker, D.M. Acta Theriol (2012) 57: 225. doi:10.1007/s13364-011-0071-8


Large mammalian carnivores place significant pressure on their prey populations and this is exacerbated within the fenced reserves of Africa. However, foraging theory predicts that diet switching by predators may mitigate this pressure. In this study, we use data collected between 2003 and 2007 from an enclosed system in the Eastern Cape Province of South Africa to examine the response of lions Panthera leo to changes in the abundance of two important prey species — kudu Tragelaphus strepsiceros and warthog Phacochoerus africanus. As the relative abundance of warthogs increased, the number of kudu kills decreased significantly, whereas warthog kills became significantly more frequent. A similar pattern was observed for lion prey preference and the switch from kudu to warthog was also reflected in a significant decrease in the mean prey mass. Our results suggest that a diet shift occurs in lions and that the change in diet is primarily in response to an increase in warthog numbers. Prey switching may promote the persistence of predator–prey systems, which is particularly important for fenced systems where natural immigration of prey is not possible. However, continued collection and analysis of long-term observational data from the multipredator, multiprey systems of Africa is required to facilitate a full understanding of predator–prey dynamics.


Diet switchingLarge predatorsPredator–prey interactionPrey preferences


The diet of predators in multiple prey species systems is affected by a range of factors including abundance of different prey species and their vulnerability, anti-predator behavior and defence, age and sex, and the presence of and interference from other predators (Perry and Pianka 1997; Garrott et al. 2007). Vulnerability is not solely a species-specific characteristic and will vary with body condition and therefore with age, reproductive status and environmental factors such as drought (Owen-Smith and Mills 2008). Thus diet can vary in space and time. The optimal foraging theory (MacArthur and Pianka 1966) provides a theoretical framework against which feeding behavior can be understood and predicts that the diet of predators, such as lions Panthera leo, will vary as the relative abundance of one or more of a range of alternative prey species varies (Pyke et al. 1977). This dietary variation may be seasonal as in some mustelids (Carss et al. 1998; Begg et al. 2003) and Geoffroy’s cat Leopardus geoffroyi (Canepuccia et al. 2007), where seasonal climate variation drives seasonal changes in food abundance. Alternatively, it may coincide with longer population cycles such as that of the snowshoe hare Lepus americanus where lynx Lynx canadensis switch to preying on red squirrels Tamiasciurus hudsonicus during periods of least hare abundance (O’Donoghue et al. 1998). Finally, it may be driven by longer cycles of climate change, such as drought, where one species may be more susceptible to drought than others and that species then becomes more vulnerable (Owen-Smith and Mills 2008).

The influence of generalist mammalian predators on prey population dynamics has been well studied (Korpimäki and Krebs 1996; Hanski et al. 2001; Dell’Arte et al. 2007). Much of this work has focussed on the effect of predators on the well-known population cycles of rodents in the relatively simple systems of the northern hemisphere, and a central theme has been that of predator switching which is thought to be responsible for controlling some population cycles (Křivan 1996; van Baalen et al. 2001; Ma et al. 2003; Sundell et al. 2003; Murrell 2005). However, evidence from the southern Hemisphere (where ecosystems are generally more complex) suggests that other factors (e.g., disease) are also important in shaping the population cycles of prey animals (Pech et al. 1992). While many of these studies have used an experimental approach, with invertebrate or small vertebrate models, it is the large mammalian predators that may be expected to place the greatest pressure on their prey populations because of their high daily energy requirements (Williams et al. 2004). However, comparatively little data exists for Africa and the southern hemisphere in general (Viljoen 1993; Höner et al. 2002; Owen-Smith and Mills 2008). This is because direct experimentation using large mammalian predators is difficult and data can typically only be collected through careful, long-term observation of natural systems (Estes 1994; Radloff and du Toit 2004; Owen-Smith and Mills 2008; Randa et al. 2009).

The reintroduction of lions to enclosed conservation areas in the Eastern Cape Province of South Africa and elsewhere has created a range of conditions which we have used to study aspects of the feeding behavior of predators including the response of predators to changes in prey abundance. At one of the sites of lion reintroduction (Kwandwe Private Game Reserve), two prey species (kudu Tragelaphus strepsiceros and warthog Phacochoerus africanus) from a total of 21 species killed, comprised more than 55% of all kills each year (Bissett 2007). No other prey species comprised more than 10% of kills in any year. On Kwandwe, the numbers of kudu and warthogs changed substantially over a relatively short period, and here, we use data from this site to examine the response of lions to the changes in the abundance of two principal prey species. We hypothesized that as the abundance of the primary prey species (i.e., kudu) declined, so significantly more alternative prey (i.e., warthog) would be consumed. In view of the reported relationship between rainfall, prey vulnerability and lion diet (Owen-Smith and Mills 2008), we also investigated the effect of rainfall.


Study sites

The study was carried out between 2003 and 2007 on Kwandwe Private Game Reserve (Kwandwe; 185 km2; ca. 33°09ʹS, 26°37ʹE) in the Eastern Cape Province, South Africa. The reserve has a warm temperate climate with an average annual rainfall of 410 mm, and the vegetation is a mosaic of open savanna-like areas and more thickly wooded areas (Mucina and Rutherford 2006). Rainfall data were recorded at one site in the center of the reserve on a monthly basis.

Data collection

Annual changes in lion numbers

The lions were located daily and thus the total number of animals was known. The total number of lions is expressed in Female Equivalent Units (FEQs), where 1FEQ is equivalent to the mass of an adult female (Bertram 1973). An adult male is 1.5FEQs, subadult lions 1FEQ, large cubs (1–2 years) 0.75FEQ and small cubs (less than one year old) 0.3FEQ (Schaller 1972; van Orsdol 1982; Packer et al. 1990). We calculated lion FEQs in July and December of each year and used the average value in our analyses.

Prey abundance

Annual, helicopter-based game counts were used to estimate the abundance of all prey species. Aerial counts were conducted over two to three consecutive days (weather dependent) in July/August of each year (i.e., the period after most ungulates had given birth) and covered the entire reserve (Reilly and Emslie 1998; Reilly 2002; Bothma and du Toit 2010).

Carnivore diet

We collected data for three small prides of lions (1–4 adults and 0–8 juveniles per pride) at Kwandwe between 2003 and 2007. At least one member of each pride was equipped with either a very high frequency (VHF) radio collar or an implanted VHF transmitter (Africa Wildlife Tracking, Rietfontein, Gauteng, South Africa) which incorporated Telonics high-power transmitters (Telonics, Mesa, AZ, USA). All animals were located by radiotelemetry using a Telonics TR-4 receiver and Telonics RA-2A directional antenna. All capture and immobilization was done by a qualified veterinarian. Radio collars were not removed at the end of this project to allow ongoing studies of the feeding and spatial ecologies of the animals. Information on the diet was collected in three ways: opportunistic (ad hoc) observations of kills made during the daily location of the lions; kills recorded during six intensive, 2-week-long periods of continuous observations (Bissett 2007), and kills recorded by field guides on their daily game drives, which is considered a viable method for monitoring the diet of carnivores (Radloff and Du Toit 2004; Bissett and Bernard 2007). For all kills, the identity of the predator, the date, species, sex, and, where possible, age (juvenile, subadult, adult) of the prey were recorded. The corrected mass of each kill was estimated by accounting for the sex and age of the animal (Radloff and du Toit 2004). Adult body mass was taken from Meissner (1982), Bothma (2002), and Skinner and Chimimba (2005). The mean prey mass for the lions was calculated using the corrected mass of every kill.

Analysis of prey preference

Jacobs’ index, D) (Jacobs 1974) was calculated for each prey species as follows:
$$ D = \frac{{r - p}}{{r + p - 2rp}} $$
where r is the number of kills of a prey species as a proportion of kills made by lions, and p is the proportional availability of the prey species. Proportional availability was based on data from annual game counts and was the number of a particular species as a proportion of the total number of all species preyed upon by the lions. Hayward et al. (2007b) further derived an equation for predicting the number of kills of a particular species:
$$ {R_i} = \frac{{{D_i}{p_i} + {p_i}}}{{1 - {D_i} + 2{D_i}{p_i}}} \times \sum K $$
where Ri is the predicted number of kills of species i when a total of ∑K kills are observed, Di represents the Jacob’s index value of species i, and pi represents the proportional abundance of prey species i at a site (Hayward et al. 2007b; Meena et al. 2011). We used this equation to calculate the predicted number of warthog and kudu kills in each year.

Data analyses

Dependent variables were not normally distributed. Thus, Spearman’s rank correlations were used to analyze annual changes in the abundance of lion FEQs, in the numbers of kudu and warthogs, in the relative abundance of warthogs (Nwarthog/Nkudu), in kills of warthogs and kudu as a proportion of all kills in that year, and in mean annual prey mass. Kills were expressed as a proportion of all kills in that year to overcome any bias introduced by changes in annual observation effort and total numbers of kills recorded. Chi-square tests were used to compare the observed and predicted kills for warthogs and kudu over the study period. The relationship between total annual rainfall and time was tested using a Spearman’s rank correlation. All statistical analyses were completed using Statistica (StatSoft, Tulsa, OK, USA).


Annual changes in the abundance of lions, kudu, and warthogs and changes in rainfall

The number of lion FEQs increased through the study (Table 1), and there was a significant correlation between year and FEQs (r = 0.90; n = 5; P < 0.05). There was no change in the abundance of subadult lions over the study period (Table 1; r = 0.44; n = 5; P > 0.05). The abundance of kudu and warthogs changed substantially during the study period (Table 1; Fig. 1A), while the abundance of all prey species killed by lions declined by about 8% (Table 1). The number of kudu declined by 23% from 2003 to 2005 and then increased slightly to 2007. Over the 5 years, the decline was not statistically significant (r = -0.60; n = 5; P > 0.05). Over the same period, the number of warthogs increased significantly (Fig. 1A; r = 0.98; n = 5; P < 0.05). The relative abundance of warthogs (Nwarthog/Nkudu) increased significantly through the study (Table 1; r = 0.99; n = 5; P < 0.05).
Table 1

Annual changes in the abundance of lions, warthogs, and kudu and the number of warthogs and kudu killed by lions at Kwandwe Private Game Reserve. The total abundance is for all the prey species consumed by lions and total observed kills is the total of all kills made by lions in that year. Predicted warthog and kudu kills were calculated using the equation derived by Hayward et al. (2007b)







Lion FEQs






Sub-adult lions


















Total abundance






Relative abundance (Nwarthog/Nkudu)






























Total observed kills





Fig. 1

Annual changes in total rainfall (grey bars) and numbers of kudu (solid circles) and warthogs (open circles) (A), and kudu and warthog kills as a proportion of all kills (B)

Rainfall varied throughout the study (Fig. 1A) and was above the 10-year average of 410 mm in 2002 and 2006 and below average in the other years. There was no significant linear trend in rainfall over time (r = 0.40; n = 5; P > 0.05).

Annual changes in the diet of lions

Kudu comprised 42% of all lion kills recorded at the beginning of the study, and from 2003 to 2007, there was a significant decrease in kudu kills (Table 1; Fig. 1B; r = -0.96; n = 5; P < 0.05). Over the same period, there was a significant increase in warthog kills (Table 1; Fig. 1B; r = 0.99; n = 5; P < 0.005). Although the observed numbers of warthogs and kudu killed were similar to the predicted values in 2003 and 2004, there were substantial differences between observed and predicted kills from 2005 to 2007 (Table 1). Significantly fewer kudu were killed between 2003 and 2007 than predicted (Table 1; χ2 = 36.6; df = 4; P < 0.0001), and significantly more warthogs than predicted were killed over the same period (Table 1; χ2 = 21.4; df = 4; P < 0.0001).

The switch from kudu to warthog was further reflected in a change in the mean prey mass (Table 2) which decreased significantly through the study (r = -0.98; n = 5; P < 0.005).
Table 2

Annual changes in the mean prey mass (kg ± SD) of all kills made by lions and in the relative abundance of warthogs on Kwandwe Private Game Reserve








Mean prey mass

182 ± 169.6

150 ± 91.9

141 ± 136.7

130 ± 135.2

108 ± 75.5

Relative abundance of warthogs






The effect of changing prey base on the prey preference of lions

At low relative warthog abundance (0.3–0.5 = two to three times as many kudu as warthogs), the preference of lions for kudu and warthog was similar and stable (D = 0.2; Fig. 2). As the relative abundance of warthogs increased from 0.5, Jacobs’ index for kudu declined to a negative value indicating an avoidance of kudu as prey (r = - 0.93; n = 5; P < 0.05). By contrast, the Jacobs’ index for warthogs was greater than zero at all times (Fig. 2), and as the relative abundance of warthogs increased above 0.6 so did the Jacobs’ index increase (r = 0.94; n = 5; P < 0.05).
Fig. 2

The relationship between the relative abundance of warthogs and Jacob’s index for warthogs (open circles) and kudu (solid circles)


The results suggest that as warthog numbers increased, the lions killed more warthogs, in absolute and relative terms, and this was matched by a decline in the number of kudu killed. The increase in the proportion of warthogs killed resulted in an increase in prey preference for warthogs and a decline in preference for kudu that was driven primarily by the change in warthog numbers.

The development of Kwandwe as an ecotourism reserve began in 1999. Prior to this, most of the indigenous large mammalian fauna had been locally extirpated and surviving, resident populations of kudu and warthog were heavily hunted. Between 1999 and 2001, over 2,000 ungulates were reintroduced, but this did not include kudu or warthogs. In October 2001, two adult male and two adult female lions were introduced, and in 2002, the first cubs were born (Hayward et al. 2007a). In addition, reintroductions of other large carnivores (cheetahs Acinonyx jubatus and African wild dogs Lycaon pictus) occurred. In the early years after reintroduction, the predators all killed kudu (Bissett 2007) while only the lions killed small numbers of warthogs. We suggest that a combination of the initial release from hunting pressure and the subsequent predation pressure on kudu and, to a much lesser extent, on warthogs resulted in the increase in warthog numbers reported in this study. On another enclosed reserve in South Africa, the impala (Aepyceros melampus) population responded to the reintroduction of lions by increasing in a way that was similar to the response of warthogs on Kwandwe (Power 2002).

Dietary flexibility and prey switching have been reported previously for large carnivores and may be driven by changes in the abundance or vulnerability of prey species. In the Chobe National Park, Botswana, large prey (buffalo Syncerus caffer and zebra Equus burchelli) comprise the majority of lion diet but, in the dry season, when they migrate out of the area, warthogs become an important prey species (Viljoen 1993, 1997). Similarly, in the Ngorongoro Crater, Tanzania, spotted hyaenas (Crocuta crocuta) included a significantly greater proportion of buffalo in their diet after a significant increase in buffalo numbers (Höner et al. 2002). This was not only a response to an increase in the abundance of buffalo but also to an increase in the abundance of more vulnerable juvenile animals (Höner et al. 2002). In the Kruger National Park (KNP), selection for alternative prey by lions, including warthogs, is affected by changes in the relative abundance and vulnerability of the three principal prey species (wildebeest Connochaetes taurinus, zebra and buffalo; Owen-Smith and Mills 2008). Selection for buffalo increased after a severe drought which increased their vulnerability, while wildebeest and zebra appeared to be less susceptible to predation under conditions of low rainfall (Owen-Smith and Mills 2008). The latter examples clearly illustrate how changes in predator diet are not simply driven by changes in prey numbers and that changes in vulnerability, which may result from climatic variability or be a characteristic of a particular age or sex, play an important role in shaping diet. Interestingly, a recent meta-analysis by Hayward (2011) suggested that the lions of sub-Saharan Africa tend to select their preferred prey less frequently when the abundance of these species increases. By contrast, when nonpreferred prey species become more abundant, they are preyed upon more regularly by lions (Hayward 2011). These findings support those of the present study where the diet shift was due to an increase in the abundance of a preferred but secondary prey species, the warthog.

Changes in the diet of large carnivores, as reported in the present and previous studies, need not be associated with a change in prey preference. In the present study, the Jacobs’ index for kudu was positive when kudu were relatively abundant and negative when they were less common thus meeting the criteria for prey switching (Murdoch 1969; Garrott et al. 2007). The results for warthogs were ambiguous in that the Jacobs’ index was always positive but the pattern, where preference increased with increasing relative abundance, again supports the occurrence of a prey shift. Similar trends were observed for spotted hyaenas in the Ngorongoro Crater, Tanzania, where preference for adult buffaloes remained the same despite a significant increase in relative abundance (Höner et al. 2002).

There are three possible explanations for the dietary shift observed here. Firstly, it is possible that as the lion population increased during the study, there was an increase in the number of subadult lions not yet capable of hunting larger prey such as kudu (a pride demography response; Hayward et al. 2007b). However, our results indicate that there was no significant change in subadult lion numbers during the study period, making this explanation unlikely. Secondly, one or more of the adult lions could have learned to effectively hunt warthogs, allowing the rest of the pride(s) to acquire the technique, and ultimately resulting in an increase in the number of warthogs killed (a learned response). Thirdly, changes in the relative abundance and vulnerability of warthogs and kudu may have influenced lion diet (prey density response). We propose that the mechanism driving the suggested dietary shift was probably a combination of the prey density and learned responses in conjunction with the small pride size, which will increase the risk associated with killing large prey species (Packer et al. 1990; Funston et al. 2001), the relative vulnerability of warthogs and kudu and the energetic returns from the two prey species. We suggest that as the abundance of warthogs increased, the encounter rate increased and the search time for warthogs decreased. In response, the small prides of lions killed more of the more vulnerable warthogs (prey density response). Lions became more efficient at catching and killing warthogs (learned response) and relatively more warthogs and fewer kudu were killed. While our data do not support an important role for pride age structure, it may well play a role at other study sites.

Our results contradict those of some other studies which have suggested that lions prefer large prey regardless of abundance (Hunter 1998) and that prey availability has no significant effect on the diet of lions (Hayward and Kerley 2005). However, if lions are opportunistic predators (Schaller 1972), then it should be expected that their diet will be influenced by prey abundance.

True prey switching is likely to promote the persistence of predator–prey systems (van Baalen et al. 2001) as predator pressure on a declining prey species will decrease, providing the opportunity for that species to recover (O’Donoghue et al. 1998). This will be particularly important on relatively small fenced conservation areas where opportunities to escape predation are few and population renewal through immigration is not possible. Thus, an understanding of the factors that influence prey selection is important for the management of enclosed conservation areas. However, a full understanding of the dynamics of multipredator, multiprey systems, such as those that characterize the African savanna, is dependent on the careful analysis of data from long-term observational studies.


Our sincere thanks go to Angus Sholto-Douglas of Kwandwe Private Game Reserve for providing logistical and financial support for this project. We acknowledge the staff of Kwandwe for providing assistance and information. All experiments complied with the ethical protocols of Rhodes University and the laws of South Africa. Matt Hayward and two anonymous referees are thanked for constructive comments on earlier drafts of the manuscript.

Copyright information

© Mammal Research Institute, Polish Academy of Sciences, Białowieża, Poland 2012