Africa Wild

Discussion

Lions in the Addo ecosystem make the majority of their kills in areas of dense vegetation compared to resting or unsuccessful hunting sites. These conditions provide cover for stalking lions and suggest an ambush style of attack, in keeping with findings for lions elsewhere [19,22]. Although vegetation structure has previously been recognized as an important component of lion hunting success [16,19,36,37], it has been considered less influential relative to other environmental variables [23]. In Addo, however, vegetation density, as measured by viewsheds, was the most important environmental predictor of where lions make kills. However, although lions killed in denser vegetation than where they rested, this was not the densest vegetation in the Addo ecosystem. Many of the random sites were located in even denser vegetation, with limited access to open areas. This suggests an upper limit on vegetation density that provides assistance to hunting lions, with very dense areas not used. The thicket vegetation in Addo can form virtually impenetrable stands, especially in the absence of elephants [42], and lions would struggle to move through the vegetation or locate prey within such areas, rendering them unsuitable for hunting, and possibly providing refuge for some prey species [45].

The finding that viewshed (vegetation structure) exerted the strongest influence over kill site locations, and that it did so equally across lion sexes, could be considered somewhat surprising given the prominence of other factors elsewhere (e.g. [23]) and the strong sexual dimorphism displayed by lions, with sex-based differences in kill locations reported previously [16,23]. However, because our study was not based on direct observations of lion hunting, we cannot definitively conclude that our male lion kill sites were the result of males killing, as opposed to scavenging from females, although our results for male kill (carcass) sites do match results from elsewhere when the same methods were used [16]. To understand the drivers of these differences in lion kill sites, the nature of the Addo ecosystem needs to be considered. Almost all previous lion hunting studies have been conducted in savanna landscapes, and so our understanding of their hunting habits is derived mostly from these ecosystems. In Addo, the succulent thicket vegetation is much denser than typical savanna vegetation, making it feasible for lions to almost always select dense areas as hunting habitats, relying less on other environmental variables such as darker nights. Female lion social groups in Addo also differ from elsewhere. Whereas male groups are comparable to other areas, generally forming coalitions of two or three males, female groups mostly consist of either solitary individuals or females with cubs (C.J. Tambling, pers. obs.), in contrast to pride sociality found in most savanna areas [35]. Our findings for male lion kills are similar to savanna lions (although we might not be recording male lion hunting), but female reliance on vegetation structure is unusual [16]. In savannas, female lion hunting success generally increases with group size [23,36,37], and cooperative hunting is most effective in open areas where predators benefit by encircling prey [23]. In Addo, the cover provided by the dense vegetation negates the need for female lions to form large social groups when hunting. Moreover, hunting success decreases when cubs or sub-adults are present [23], and since the majority of female lion groups larger than singletons in Addo are comprised of mothers with cubs, there is no advantage to hunting in the open, even with increased group size. Female lions in savanna ecosystems also rely on other pride members to safeguard young cubs during group hunts [35]. However, the dense vegetation in Addo likely enables single mothers to hide their cubs more effectively and so be less reliant on forming crèches and therefore groups. Based on our findings, we predict that solitary lionesses will also show such reliance on vegetation structure in savannas, indicating an additional dimension, beyond predator sex, that drives predator behavior. These concepts of varying functional behavior with predator sex and group size should also be assessed for other predators that display sexual dimorphism and varying social structures.

After viewshed, wind speed was the next most important variable for predicting kill sites relative to resting or unsuccessful hunting sites. The dense vegetation lions utilize when hunting in Addo could make increased wind speeds particularly useful because of increased noise during windy conditions (as opposed to more open savanna environments where wind speed has been considered less influential [23] or only important for select prey species [37]). Furthermore, given the coastal location of Addo, strong wind is fairly predictable on a daily basis, being stronger in the afternoon. Such conditions might result in wind speed being more important here than other ecosystems where wind is less prevalent or predictable. High wind speeds were, however, more important for kills of male prey (Fig 5C), possibly because of increased vigilance by female prey with young [40]. How such behavior might affect vulnerability and habitat preference across ungulate sexes will influence the functional roles predators play in ecosystems, with differences in the magnitude of their effect between prey sexes (see [56]).

Although the probability of a site being a kill site and not a resting site increased slightly when the GPS clusters originated during darker periods (less moonlight), moonlight did not feature as an important variable compared to viewsheds or wind speed. Although found to be important elsewhere [23,36], it is likely less influential in Addo because of the dense vegetation. In Kruger National Park, moonlight was considered the most important environmental variable affecting hunting success, but not during impala, Aepyceros melampus, hunts [23]. Impala were more often encountered by lions in dense bush [52], where darkness was considered less important because lions could rely on the cover provided by the vegetation [23]. A possible constraint in our methods, however, is that moonlight was taken at the start of the GPS cluster, and we do not know the exact time of the kill and how much moonlight was available then. Because of this, the resolution of our moonlight data is also fairly coarse (moon present or absent), and our results regarding the importance of moonlight should be considered with some caution. However, given our results and the similarity between the different moonlight metrics tested, it is doubtful that moonlight plays a substantive role in lion hunting in Addo.

Not only do the hunting habitats of Addo lions differ from savanna lions, but so do the prey species selected. Whereas buffalo are a preferred prey species in several savanna ecosystems [21,57,58], they constituted less than 5% of kills found during our study, and less than 1% of male kills (Table 5), despite male lions specifically targeting buffalo elsewhere [52]. Although buffalo were initially selected by lions in Addo, they later formed larger, more defensive herds and successfully fended off the majority of lion attacks [15]. The small lion group sizes in Addo possibly prevent successful hunting of large groups of buffalo, which generally requires larger groups of lions [23]. Instead of typical target species such as buffalo and zebra [57], half the kills made in Addo consisted of kudu and ostrich, species which are usually killed in proportion to their abundance (kudu) or actively avoided (ostrich) [57]. Kill sites of these species were located in denser vegetation compared to other prey species (Fig 5A) and lions are possibly selecting them because they are easier to locate (kudu frequent densely vegetated areas) or catch (ostrich are less agile in thick vegetation) in dense vegetation, which is more common in Addo relative to savanna areas. The method we used to detect kill sites (GPS telemetry) does, however, have important biases. Carcass remains likely persist for longer periods in areas of dense vegetation where scavengers might take longer to locate them, and we expect that the remains of small prey species will likely go undetected because of the shorter time required for lions to consume the carcass, diminishing the probability of a GPS cluster forming.