No abstract

In many savanna ecosystems worldwide, livestock share the landscape and its resources with wildlife. The nature of interactions between livestock and wildlife is a subject of considerable interest and speculation, yet little controlled experimental research has been carried out. Since 1995, we have been manipulating the presence and absence of cattle and large mammalian herbivore wildlife in a Kenyan savanna in order to better understand how different herbivore guilds influence habitat use by specific wildlife species. Using dung counts as a relative assay of herbivore use of the different experimental plots, we found that cattle had a range of effects, mostly negative, on common mesoherbivore species, including both grazers and mixed feeders, but did not have significant effects on megaherbivores. The effect of cattle on most of the mesoherbivore species was contingent on both the presence of megaherbivores and rainfall. In the absence of megaherbivores, wild mesoherbivore dung density was 36% lower in plots that they shared with cattle than in plots they used exclusively, whereas in the presence of megaherbivores, wild mesoherbivore dung density was only 9% lower in plots shared with cattle than plots used exclusively. Cattle appeared to have a positive effect on habitat use by zebra (a grazer) and steinbuck (a browser) during wetter periods of the year but a negative effect during drier periods. Plots to which cattle had access had lower grass and forb cover than plots from which they were excluded, while plots to which megaherbivores had access had more grass cover but less forb cover. Grass cover was positively correlated with zebra and oryx dung density while forb cover was positively correlated with eland dung density. Overall these results suggest that interactions between livestock and wildlife are contingent on rainfall and herbivore assemblage and represent a more richly nuanced set of interactions than the longstanding assertion that cattle simply compete with (grazing) wildlife. Specifically, rainfall and megaherbivores seemed to moderate the negative effects of cattle on some mesoherbivore species. Even if cattle tend to reduce wildlife use of the landscape, managing simultaneously for livestock production (at moderate levels) and biodiversity conservation is possible.

The impact of elephants on the woody plant community through debarking was investigated in Samburu and Buffalo Springs National Reserves, Kenya. Acacia elatior Brenan, the most abundant tree species in the riverine zone, accounted for 68% (n = 1375) of woody plants. A. tortilis (Forsskal) Hayne dominated plots away from the river. Debarking incidences were significantly higher for A. elatior than for other species indicating selective utilization. The riverine zone by virtue of having more trees of the preferred species, A. elatior, had the highest debarking incidences. Presence of very few saplings along the river is attributed to both elephant trampling and herbivory by other species. An estimated 38.5% and 22.5% of the riverine A. elatior and A. tortilis trees respectively, were bound to die within the next 4-5 years because of severe debarking, ‡75% of bark circumference. Debarking was positively correlated with stem circumference; the mediumsized trees being the worst affected by the elephants' selective debarking behaviour. Intense debarking incidences were recorded during the dry season. Through the elephants' selective debarking, the riverine habitat is bound to open up gradually, leading to considerable habitat change in the near future. Elephant impact on vegetation is less away from the river and increases with their densities.Key words: Elephants, Samburu, selective debarking, woody plants condamné à s'ouvrir progressivement, ce qui entraînera un changement considérable dans un proche avenir. L'impact des éléphants sur la végétation est moindre lorsqu'on s'éloigne de la rivière, et il augmente en fonction de leur densité.

The magnitude of debarking by elephants was investigated in Samburu and Buffalo Springs National Reserves. About 1617 plants were monitored for debarking intensities for 6 months spanning through dry and wet seasons. Debarking indices ranged from no debarking at all during the wet months to complete stem girding at the height of the dry season. A negative correlation was found between rainfall and debarking indices. It was hypothesized that nutrient content of the bark influenced the magnitude to which trees were debarked. Bark samples were collected from least, moderate and intensely debarked plants throughout the 6 months. These were analysed for calcium (Ca), sodium (Na), phosphorus (P), magnesium (Mg), potassium (K), nitrogen (N), iron (Fe), copper (Cu), manganese (Mn) and zinc (Zn). Significant positive correlations were found between debarking intensity and each of the nutrients N [crude protein (CP)], P, K and Zn. Bark was found to be richest in CP and Calcium. Neutral detergent fibre content was on average 67%. Monthly variations in nutrient composition were minimal. Acacia elatior, the most preferred species had significantly higher quantities of each of the four elements N, P, K and Zn than Acacia tortilis, the second most preferred woody species.

This study assesses the ecological pressure exerted by changing land use and tenure on the Kitenden wildlife corridor, a critical cross‐border link between the Amboseli and Kilimanjaro national parks. The implications on viability of the two high‐value protected areas and their respective dispersal areas are both negative and serious. The extent of land use change and its impacts were assessed through household and vegetation surveys, while wildlife abundance was measured using transect walks. Approximately 30% of the study area has shifted from community to private land ownership over the last two decades. Except for baboon and vervet monkey, most wildlife avoided the cultivated area. Vegetation composition on the noncultivated area has been greatly altered by intense wildlife and livestock use, where mean herbaceous vegetation cover differed significantly among range‐plant categories (F3, 524 = 29.015, p < 0.05). The frequency of increaser I (21.4%) differs greatly from that of decreaser and forbs, at 8.3% and 7.4%, respectively Tree recruitment was low, with a significant difference in mean density among age classes (F2, 110 = 3.663, p < 0.05). Only through land leasing agreements between landowners and conservation organisations, and a widely supported land use plan, can the spread of cultivation be controlled and complete cessation of wildlife movement be prevented.