Introduction
The charisma of carnivores and their connection with the human psyche, coupled with their important functional roles in ecosystems, often make carnivores suitable flagship species for conservation (Gittleman et al., Reference Gittleman, Funk, MacDonald and Wayne2001; Dalerum et al., Reference Dalerum, Somers, Kunkel and Cameron2008; Ducarme et al., Reference Ducarme, Luque and Courchamp2013; Ripple et al., Reference Ripple, Estes, Beschta, Wilmers, Ritchie and Hebblewhite2014). The tiger Panthera tigris, lion Panthera leo, grey wolf Canis lupus and cheetah Acinonyx jubatus have been the face of projects focused on habitat conservation, ecosystem revival and conservation introductions (Bangs & Fritts, Reference Bangs and Fritts1996; Jhala et al., Reference Jhala, Banerjee, Chakrabarti, Basu, Singh, Dave and Gogoi2019; Packer, Reference Packer2019; Jhala et al., Reference Jhala, Gopal, Mathur, Ghosh, Negi and Narain2021). Although such conservation policies currently aim to restore carnivores through protection and/or augmentation, carnivore recovery can lead to increased conflicts with people (e.g. with brown bears Ursus arctos in the Pyrenees mountains; Piedallu et al., Reference Piedallu, Quenette, Mounet, Lescureux, Borelli-Massines and Dubarry2016) and reduction of threatened prey species (e.g. predation of bighorn sheep Ovis canadensis by pumas Puma concolor; Johnson et al., Reference Johnson, Hebblewhite, Stephenson, German, Pierce and Bleich2012). The depletion of threatened or rare prey through predation is particularly important in systems where predators have abundant primary prey. As a result of this, predator densities in such systems can remain high and are not restricted by the availability of rare or secondary prey. Such high predator densities can, however, result in high predation on rare prey, trapping them in a predator pit that can lead to further population declines (Allee et al., Reference Allee, Park, Emerson, Park and Schmidt1949; Messier, Reference Messier1994; Sinclair et al., Reference Sinclair, Pech, Dickman, Hik, Mahon and Newsome1998; Schmidt, Reference Schmidt2004; Johnson et al., Reference Johnson, Hebblewhite, Stephenson, German, Pierce and Bleich2012). Because of such layered effects mediated by carnivores, knowledge of their ecology and behaviour remains key for effective management of ecosystems (Boitani & Powell, Reference Boitani and Powell2012).
Lions are iconic for their connection to human culture and their apex position in ecosystems, often making them symbols of conservation (Roemer et al., Reference Roemer, Gompper and Van Valkenburgh2009). However, lions are also one of the top species involved in conflict, severely threatening human lives and livelihoods (Ray et al., Reference Ray, Hunter and Zigouris2005). Direct persecution and detrimental habitat alteration have reduced their numbers by > 75% since 1970, with a range contraction of > 90% (Ripple et al., Reference Ripple, Estes, Beschta, Wilmers, Ritchie and Hebblewhite2014; Loveridge et al., Reference Loveridge, Sousa, Cushman, Kaszta and Macdonald2022). The Tsavo Conservation Area (hereafter Tsavo) in south-east Kenya is a semi-arid, drought-prone landscape that hosts one of the largest populations of lions in East Africa (Henschel et al., Reference Henschel, Petracca, Ferreira, Ekwanga, Ryan and Frank2020). This landscape is also an important connection between two major African biomes: mesic grassland savannah to the south and semi-arid bush savannah to the north (Henschel et al., Reference Henschel, Petracca, Ferreira, Ekwanga, Ryan and Frank2020). Nearly 450 lions live in the Tsavo landscape (covering both East and West Tsavo National Parks; Elliot et al., Reference Elliot, Broekhuis, Omondi, Ngene, Kariuki and Sankan2021), yet lion ecology has been understudied in this landscape (barring mane growth, Kays & Patterson, Reference Kays and Patterson2002, and the human-eating tendencies of two infamous male lions, DeSantis & Patterson, Reference DeSantis and Patterson2017). This contrasts with other well-studied East African lion populations in Serengeti and Ngorongoro (Packer et al., Reference Packer2019).
Tsavo is also home to one of the last remaining populations of the Critically Endangered hirola Beatragus hunteri. The hirola is restricted to a small area of its natural range on the Kenyan–Somali border, but was introduced to Tsavo East National Park in 1962 and 1996 as insurance against extinction in its natural range (Probert et al., Reference Probert, Evans, Andanje, Kock and Amin2014). The hirola population in Tsavo has remained small, and such small herbivore populations can be prone to the detrimental effects of predator mediated apparent competition when primary prey is abundant (Johnson et al., Reference Johnson, Hebblewhite, Stephenson, German, Pierce and Bleich2012). The Endangered Grevy's zebra Equus grevyi was also introduced to Tsavo, in 1964 and 1977, but the population has also remained small (Githiru, Reference Githiru2017). Multiple factors can impede recovery of introduced herbivores, and predation is a plausible cause for the lack of population growth in these two species of conservation concern in Tsavo (Evans, Reference Evans2011; Probert et al., Reference Probert, Evans, Andanje, Kock and Amin2014).
To examine the ecology of this important population of lions and their potential predation impacts on the hirola and Grevy's zebra, we present information on lion diet and preferences in Tsavo, determined from scat analysis using biomass models. Our findings have important implications for predation ecology of threatened herbivores in an important conservation landscape.
Study area
We conducted the study in the southern part of Tsavo East National Park, south of the Galana River in Kenya (Fig. 1). Our study area spans > 4,065 km2 and is bordered by Tsavo West National Park to the west, Taita ranches and settlements to the south and south-west, and Galana and Kulalu ranches to the east (Lala et al., Reference Lala, Chiyo, Kanga, Omondi, Ngene and Severud2021). Our study area has a high density of lions and herbivores compared to elsewhere in Tsavo (Elliot et al., Reference Elliot, Broekhuis, Omondi, Ngene, Kariuki and Sankan2021), as the Galana River to the north forms a natural barrier for animal movement. It is a semi-arid landscape with a dry season from January to early March and a cool season from June to October. Annual temperature is 20–30 °C and annual precipitation 200–700 mm (Lala et al., Reference Elliot, Broekhuis, Omondi, Ngene, Kariuki and Sankan2021).
Fig. 1 The location of Tsavo Conservation Area in southern Kenya, with the locations of the 74 lion scats collected in the southern part of Tsavo East National Park, south of the Galana River during 2019–2023.
Much of the landscape is open and shrubby grasslands, with a mosaic of riparian habitats along the rivers. In addition to the hirola and Grevy's zebra, common herbivores include the savannah elephant Loxodonta africana, giraffe Giraffa camelopardalis, plains zebra Equus quagga and Cape buffalo Syncerus caffer. The lion, cheetah, leopard Panthera pardus, African wild dog Lycaon pictus and spotted hyena Crocuta crocuta are common carnivores. A list of the mammalian and avian species of Tsavo can be found in Lack et al. (Reference Lack, Leuthold and Smeenk1980), Lepage (Reference Lepage2004) and Tóth et al. (Reference Tóth, Lyons and Behrensmeyer2014). Tsavo is a keystone habitat that with Amboseli National Park forms the Tsavo–Amboseli ecosystem.
Methods
Scat collection and processing
We opportunistically collected 74 whole lion scats during November 2019–February 2023, with the majority of the scats collected during the dry season (Fig. 1). All scats were georeferenced in the field, and were identified based on their size and the presence of lion tracks and scrapes. We relied on the ecological knowledge of experienced field personnel from the Wildlife Research and Training Institute, and Kenya Wildlife Service rangers, who are familiar with the area. We only analysed scats that could be unambiguously identified as those of lions.
Each collected scat was sun dried before being washed through fine mesh (1–3 mm) sieves. We separated prey hairs from other remains such as bones, hooves and teeth, and randomly selected 20 prey hairs from each washed sample. Each hair was mounted on a slide with a drop of DPX mountant and a coverslip, and examined under a compound microscope at × 400 magnification. We cross-referenced the cortex and medulla patterns, and the thickness and proportion of the medullary layer in each hair with a key for mammalian prey hair from the area (Perrin & Campbell, Reference Perrin and Campbell1980) and with a digital hair library at the Kenya Wildlife Service (Evans, Reference Evans2011), facilitating identification of each prey hair to species. The reference library at the Kenya Wildlife Service (compiled using hairs collected from the Kenya Wildlife Service animal orphanage and from known predation events) was required as Perrin & Cambell's (Reference Perrin and Campbell1980) key did not include all prey species occurring in the study area. From each scat sample of 20 randomly selected hairs, we determined the number belonging to a particular prey species, and converted it into a proportion or scat equivalence (Chakrabarti et al., Reference Chakrabarti, Jhala, Dutta, Qureshi, Kadivar and Rana2016), where:
Scat equivalence of prey species A in sample i = number of hair/s of species A in sample i / 20
We added the scat equivalence for each prey species across all 74 scats to obtain the whole scat equivalence for that species (Chakrabarti et al., Reference Chakrabarti, Jhala, Dutta, Qureshi, Kadivar and Rana2016). We used the whole scat equivalence to compute biomass consumption for each prey species instead of the more commonly used frequency of occurrence because it provides a more accurate representation of consumption (Chakrabarti et al., Reference Chakrabarti, Jhala, Dutta, Qureshi, Kadivar and Rana2016).
Prey consumption
We calculated the total biomass consumption of each prey species found in the scats using a lion specific biomass model (Chakrabarti et al., Reference Chakrabarti, Jhala, Dutta, Qureshi, Kadivar and Rana2016): y = 4.105 − 3.116 exp −0.032x, where x is the mean prey weight and y is the prey biomass consumed per scat. The biomass consumed per scat for each prey species was then multiplied with its whole scat equivalence to estimate total biomass consumption of that particular prey in our collected samples. For prey weights, we used 75% of the mean adult female body weight, following Schaller (Reference Schaller1972) and Hayward & Kerley (Reference Hayward and Kerley2005), because it accounts for the weights of juveniles and subadults in the population and is a better representation of the mean weight of a prey population available to a predator. Prey weights were obtained from Stewart (Reference Stewart1963), Sachs (Reference Sachs1967), Myers et al. (Reference Myers, Citino and Mitchell2008), Pinto (Reference Pinto2018) and Hayward & Kerley (Reference Hayward and Kerley2005). To examine the importance of each prey species in lion diet, we calculated the relative biomass consumed for each prey as a per cent of total consumption.
Prey selection
We used recent herbivore aerial census data from Tsavo (Waweru et al., Reference Waweru, Omondi, Ngene, Mukeka, Wanyonyi and Ngoru2021) to estimate study site specific abundance/individuals observed for lion prey species. We apportioned herbivore sightings to our study area by clipping all sightings by the boundary of the study area. Although post hoc, this delineation ensured that prey consumption, based on scats, and prey availability were matched geographically. We subsequently calculated the relative abundance of each prey species as an index of availability, in which the abundance of each prey species was converted as a per cent of the total abundance of all prey species occurring in the lion scats in our study area. We could not include dikdik (Madoqua sp.) and fringe-eared oryx Oryx beisa callotis in this prey selection analysis because of the lack of sightings in our study area during the aerial census. Our analysis indicates that consumption of these two species by lions was low in Tsavo, and hence we are confident that their exclusion did not significantly alter the prey selection indices. We also excluded elephants from this part of our analysis as we found elephant hair in only one scat. From the relative biomass consumed and relative prey density, we computed Jacobs' selectivity index for each prey species occurring in the lion scats, using the dietR package (Borstein, Reference Borstein2020) in R 4.2.1 (R Core Team, 2021). Jacob's selectivity index (D) ranges from +1 to −1, indicating maximum preference and maximum avoidance respectively, and a value close to 0 indicates random selection of a prey species (Jacobs, Reference Jacobs1974).
Results
The cumulative species occurrence plot indicated our sampling was adequate to characterize lion diet in Tsavo East as no new species were detected after 20 scats (Fig. 2). We identified 16 prey species from the scats; African elephant being the largest and dikdik being the smallest (Table 1).
Fig. 2 Calculation of sampling adequacy for the 74 lion scats collected in Tsavo East National Park, Kenya (Fig. 1), during 2019–2023 for the characterization of diet, with no additional prey identified with the analysis of > 20 scats.
Table 1 The 16 prey species detected in 74 lion Panthera leo scats collected in Tsavo East National Park during 2019–2023, with their body weight (75% of mean adult female body weight), whole scat equivalence (WSE; see text for details), biomass consumed per scat (y; see text for details), total biomass consumed (WSE * y), relative biomass consumed (per cent of total consumption), total abundance (total number of individuals of a species observed within our study area), and relative abundance (per cent of the total abundance of all prey species). Species are sorted by relative biomass consumed, and those in bold indicate preferred prey items, with a threshold of preference/avoidance set at D ± 0.3 (see Fig. 3 for details). Dikdik and fringe-eared oryx could not be included in the prey preference analysis as a population estimate for the species is not available, and the elephant was excluded as hair was found in only one lion scat.
Fig. 3 Jacobs’ selectivity indices (D) for prey species (excluding dikdik, elephant, and fringe-eared oryx) occurring in 74 lion scats from Tsavo East National Park, Kenya, collected during 2019–2023. Values < -0.3 indicate avoidance and values > 0.3 preference for the corresponding prey species, and values ±0.3 indicate that the corresponding species were predated randomly or in proportion to their availability. Waterbuck, Grevy's zebra, warthog, Cape buffalo and hirola were the most preferred prey items.
The majority of lion diet comprised of herbivores that weigh > 150 kg, with zebra, giraffe, Cape buffalo and waterbuck Kobus ellipsiprymnus accounting for > 70% of biomass intake. Small antelopes, such as gerenuk Litocranius walleri and Grant's gazelle Nanger granti were rarely consumed (2% of the total biomass intake). Approximately 5% of the diet comprised the Endangered Grevy's zebra and Critically Endangered hirola.
Based on prey availability, seven of the 16 species detected were preferred by lions (Grevy's zebra, hirola, waterbuck, warthog Phacochoerus africanus, Cape buffalo, gerenuk and giraffe), and five species were avoided (Grant's gazelle, hartebeest Alcelaphus buselaphus, plains zebra, impala Aepyceros melampus and lesser kudu Tragelaphus imberbis). The eland Tragelaphus oryx was consumed according to its availability (Table 1, Fig. 3).
Discussion
Comprehensive characterization of diet is critical for understanding animal ecology and is of particular importance for apex carnivores as their predation can have cascading implications for the entire ecosystem. We investigated diet and prey selection of lions south of the Galana River in Tsavo East National Park, Kenya, using scat analysis through a lion-specific biomass conversion model and estimation of prey abundance. Earlier studies of lion diet have either relied on direct observations or frequency of occurrence of prey remains in scats (Hayward & Kerley, Reference Hayward and Kerley2005), which have the limitations of overestimating large and small prey, respectively (Chakrabarti et al., Reference Chakrabarti, Jhala, Dutta, Qureshi, Kadivar and Rana2016). Biomass conversion models account for differential prey digestibility and hence provide a more accurate representation of carnivore diet from scats (Ackerman et al., Reference Ackerman, Lindzey and Hemker1984; Chakrabarti et al., Reference Chakrabarti, Jhala, Dutta, Qureshi, Kadivar and Rana2016).
Our findings reveal that lions in this area include a wide size range of mammalian prey species in their diet, from the dikdik to the African savannah elephant, but they typically rely on large ungulates (Fig. 4). Our findings also reflect similar patterns in lion diet across the species’ global range (Hayward & Kerley, Reference Hayward and Kerley2005), as well as in the adjacent population in Amboseli (Courtois, Reference Courtois2015). Lions in Tsavo seem to consume large and dangerous prey species such as Cape buffalo and giraffe, substantially (Table 1). Although lions often hunt alone (Packer, Reference Packer, Rubenstein and Wrangham1986; Packer et al., Reference Packer, Scheel and Pusey1990), killing large and dangerous prey may require cooperation. When lions hunt as a group, the simplest tactics of cooperation, which include either similarity in actions or synchrony, or a combination of both, can be sufficient to hunt such prey (Boesch & Boesch, Reference Boesch and Boesch1989; Palmer et al., Reference Palmer, Packer Borrego, Packer, Srinivasan and Würsig2023). Complex cooperative hunting that requires coordination and collaboration has been documented only in Etosha National Park in Namibia (Stander, Reference Stander1992; Palmer et al., Reference Palmer, Packer Borrego, Packer, Srinivasan and Würsig2023) and recently in Gir National Park in India (Chakrabarti et al., Reference Morris, Smith, Chakrabarti, Lala, Nyaga and Bump2023). Lions in Etosha cooperatively hunt large and dangerous herbivores such as giraffes (Stander, Reference Stander1992), and cooperative hunting by lions in Gir involves killing livestock within human settlements, requiring individual lions to navigate threats from herders and other people. A substantial reliance on large and dangerous prey among Tsavo lions perhaps indicates that lions in this population employs cooperative hunting techniques, and observational studies are needed to examine this. Tsavo lions are also known to scavenge on elephant carcasses that become available as a result of drought or poaching (Morris et al., Reference Morris, Smith, Chakrabarti, Lala, Nyaga and Bump2023). However, we found elephant hair in only one scat. This could be because our study did not coincide with a major drought event and/or because scavenging from a large carcass mainly involves consumption of flesh and organs rather than skin and hair, and is thus not detectable in scats without DNA analysis (Chakrabarti et al., Reference Chakrabarti, Jhala, Dutta, Qureshi, Kadivar and Rana2016).
Fig. 4 Relative biomass consumption of lion prey (Table 1) segregated into weight categories (small, < 50 kg; medium, 50–150 kg; large, > 150 kg). Large prey comprise the majority (> 75%) of lion diet. Prey weight is 75% of mean adult female body weight, following Schaller (Reference Schaller1972) and Hayward & Kerley (Reference Hayward and Kerley2005).
Although large ungulates such as the Cape buffalo and waterbuck were among the most consumed and preferred prey of Tsavo lions, we also found considerable consumption of small ungulates such as the warthog (Table 1). Warthogs comprised c. 5% of lion diet and were highly preferred. The warthog is a preferred prey species of the lion across its range as a result of its spatio-temporal sympatry with lions and its low evasion speed (Hayward & Kerley, Reference Hayward and Kerley2005). Consumption of the threatened hirola and Grevy's zebra by lions in Tsavo was high, cumulatively accounting for c. 5% of their diet even though they comprised only c. 1% of the relative abundance of available prey. The introduced hirola population in Tsavo is of critical importance because of the relatively low numbers of this species in its natural range; the Tsavo population comprises nearly 15% of the species’ global abundance (Probert et al., Reference Probert, Evans, Andanje, Kock and Amin2014; Waweru et al., Reference Waweru, Omondi, Ngene, Mukeka, Wanyonyi and Ngoru2021). The hirola population in Tsavo has remained small since its introduction, and its recovery is a management concern. Similarly, the Grevy's zebra population in Tsavo has remained small since its introduction. Little is known about the potential stressors that have kept these populations low (Jowers et al., Reference Jowers, Queirós, Resende Pinto, Ali, Mutinda and Angelone2020), although an earlier study of lion scats in Tsavo also reported relatively high predation of hirola (Evans, Reference Evans2011).
Small prey populations can be disproportionately vulnerable to predation when primary prey is abundantly available. This is because a high primary prey density can sustain high, stable predator densities, which in turn can cause high and persistent offtake of the small, secondary prey population. A high predation pressure can reduce population growth and potentially trap the small prey population in a predator pit where the population size of the secondary prey species declines (Sinclair et al., Reference Sinclair, Pech, Dickman, Hik, Mahon and Newsome1998; Johnson et al., Reference Johnson, Hebblewhite, Stephenson, German, Pierce and Bleich2012; Ng'weno et al., Reference Ng'weno, Buskirk, Georgiadis, Gituku, Kibungei and Porensky2019). Our analysis indicates that both the hirola and Grevy's zebra are consumed by lions relatively more than their proportional availability in Tsavo, perhaps because of abundant primary/alternative prey species that sustain high lion numbers. High predation can be detrimental to population recruitment and can potentially cause Allee effects or inverse-density dependent declines among these threatened and small herbivore populations (Allee et al., Reference Allee, Park, Emerson, Park and Schmidt1949; Messier, Reference Messier1994; Sinclair et al., Reference Sinclair, Pech, Dickman, Hik, Mahon and Newsome1998; Schmidt, Reference Schmidt2004; Johnson et al., Reference Johnson, Hebblewhite, Stephenson, German, Pierce and Bleich2012). Based on our diet analysis, we recommend a telemetry-based investigation of predation ecology of lions and other carnivores in this area, to examine age- and sex-class specific offtake of the hirola and Grevy's zebra. This will facilitate an accurate population viability analysis of these two threatened herbivores, and guide management decisions.
Author contributions
Study conception and design: SC, FL, JKB; field and laboratory work: SN, GW, PIC, JK, FL; data analysis: EK, SC; writing: EK, SC, with inputs from JKB, FL, PO, RM, GW; funding acquisition: JKB, RM, SC; revision: all authors; project supervision: SC, JKB, FL, PO.
Acknowledgements
We thank the Wildlife Research & Training Institute, Kenya, for granting permission to conduct the research; all the staff of the Wildlife Research & Training Institute, Tsavo East, Kenya Wildlife Service rangers, and Tsavo Trust staff and personnel for assisting with fieldwork and data collection; and two anonymous reviewers and the editor for their constructive comments. SC thanks Harshini Jhala for her help with Fig. 1.
Funding
The project was funded by the United States National Science Foundation, Grant/Award Number: NSF ID#1545611 and NSF ID#1556676 to JKB, Louis Frenzel Jr Endowed Grant of Macalester College to SC to fund summer research of EK, and additional funding for fieldwork by Tsavo Trust.
Conflicts of interest
None.
Ethical standards
This research abided by the Oryx guidelines on ethical standards. Fieldwork was conducted as part of regular monitoring by the Wildlife Research & Training Institute and the Kenya Wildlife Service. The data involved no experimentation, collection or monitoring of live animals.
Data availability
The scat analysis is in Table 1. Scat locations, which correspond to lion locations, have not been included because of their sensitive nature, and can be made available on reasonable request to the corresponding author. Prey abundance estimates are from published reports.
Open access
