Study area
This study was conducted in Olgulului- Ololorashi group ranch located in the southern part of Kenya (Fig. 1) in Kajiando County, which lies between longitudes 36° 5′ and 37° 5′ east and latitudes 1° 0′ and 3° 0′ (Wayumba and Mwenda 2006). The ranch is bordered by Eselenkei group ranch to the north, Mbirikani group ranch to the northeast, and former Kimana group ranch to the southeast and the Kenyan-Tanzanian international border to the south. Amboseli National Park is almost completely surrounded by Olgulului-Ololorashi group ranch, thus making it an important wildlife dispersal area for the park. Overall, the ranch covers an area of 1232 km2 and surrounds 90% of Amboseli National Park (Okello and Kioko 2010). Amboseli National Park which lies contiguous to Olgulului-Ololorashi group ranch covers an area of 392 km2 and forms part of the larger Amboseli ecosystem whose area is approximately 5000 km2 (Ntiati 2002). The group ranch was started in 1975 and had 3418 members by 2001. The members increased to 11,500 in 2008 representing 236% increase in a span of 8 years (Campbell et al. 2003).
The annual rainfall in the Amboseli ecosystem is strongly influenced by mountains, hills, and the rift valley (Norton 1977). High rainfall in the Loitokitok Sub-county occurs around the slopes of Mt. Kilimanjaro and the Chyulu Hills (Norton 1977). Other areas especially the lower rangelands are characterized by lower rainfall; these include the Amboseli basin especially in Mbirikani, Olugulului, and Eselesnkei group ranches (Ntiati 2002). During the wet season, large carnivores including lions and wild herbivores are dispersed throughout much of the Amboseli ecosystem. However, during dry seasons, most of the wildlife especially water-dependent species concentrates in the park, particularly in the swamps whose water is replenished by underground springs originating from melt water on the nearby Mt. Kilimanjaro (Western 1975; Campbell et al. 2003). In the dry season, the Maasai access the park for salt licks and water for their livestock and as such predation of livestock by lions and other carnivores is frequent both inside and outside the park.
OGR receives low rainfall of 500 mm or less that follows a seasonal pattern of short rains between October and December and long rains between March and May. The rainfall levels make OGR a semiarid to arid land suitable for nomadic pastoralism which has been the traditional land use in the region (Ntiati 2002). The majority of top soils within the Amboseli ecosystem are shallow and unproductive due to recent volcanic activity and therefore unsuitable for agriculture (Katampoi et al. 1990). The recent trends in land use, however, indicate an expansion in rain-fed agriculture in the group ranch and increased permanent settlement (Campbell et al. 2003; Okello and Kioko 2010). The vegetation communities of the Amboseli ecosystem are dominated by bushland, woodland, and open grassland. The most common species include Acacia - commiphora, Acacia tortilis, Acacia xanthophloea, Azima tetracantha, and Suaeda monoica (Githaga et al. 2003).
Methods
A total of 229 people were interviewed including 199 questionnaire respondents and 30 key respondents drawn from three KWS officials (senior warden, community warden, and ranger), three focus group discussions with five members each, two officials of provincial administration (chief and subchief), two group ranch officials (chairman and treasurer), four village elders, and four verification officers from Amboseli Trust for Elephants and Predator Compensation Fund. The three focus group discussions consisted of elders from different Manyattas, Morans, and young pre-Moran men in each group.
Olgulului group ranch covers a very large area and the Maasai villages are distributed as clusters, and therefore, cluster sampling method was used to select villages included in the study to ensure the community views were well represented (Zar 1999). The western side of the GR consisted of 26 villages (a village may contain more than one manyatta depending on the settlement density of a particular area) in total from which 11 villages with a population of approximately 2000 were selected using cluster sampling. The household was used as the sampling unit, since manyattas often consist of more than one household. Using systematic sampling, every tenth household in each village was visited until a total of 199 respondents were interviewed.
Three people from the group ranch were recruited as research assistants to administer the household questionnaire among the residents of Olgulului group ranch. With assistance of the research assistants, 199 respondents were interviewed within a period of 4 weeks in June 2009. Since the interviews were done during the day, the researcher and her team had an opportunity to interview a few women and elderly men as they had been left in the bomas when young morans went out to herd livestock. The research team was able to interview morans at grazing sites as the assistants knew which boma each moran belonged to. The questionnaire survey was done using a semistructured questionnaire, which included both closed and open-ended questions. The semistructured questionnaire allowed for a much wider range in statistical comparisons, and also allowed interviewees to add their own personal experiences and elucidate subject matters that may have been too restricted in highly structured questionnaire format (Hazzah 2006). The questionnaire was divided into four sections. The first section focused on demographic information (gender, age, level of education, employment, and livelihood strategy). The second section dealt with various aspects of land use and livestock production, including the type and number of livestock kept per household. The section also dealt with other economic activities such as agropastroralism and crop farming. Questions on losses of livestock to wildlife including when and where attacks on livestock occurred and livestock lost to other causes of death were covered in the third section. The final section of the questionnaire dealt with interventions aimed at reducing livestock losses due to wildlife.
In addition to the household surveys, three focus group discussions with five members each were held with members of the community consisting of young pre-moran and morans who were livestock herders and elderly men from different villages. The discussions were done with the aid of the three trained research assistants using semistructured interview sessions, and new questions were adopted as the discussions progressed. Efforts to hold similar discussions with women were fruitless as getting women in a group was hard because they were busy doing household chores and taking care of children. Men were easily found in groups drinking beer or at slaughter sites where they were interviewed on the history of livestock predation, the major causes of livestock deaths, the species that attacked livestock most frequently, attitudes to livestock predation by lions and potential solutions, benefits received from wildlife conservation, and views about the predator compensation fund in the group ranch.
Key informers were interviewed using semistructured questionnaires to gather information not captured in other surveys described above. Among the key informants were Kenya Wildlife Service officers, local leaders, group ranch officials, Amboseli Trust for Elephants and Predator Compensation Fund officials. Issues discussed during the interviews included the role of the KWS in reducing human wildlife conflict, collaboration between compensation organizations and KWS, administration of compensation schemes, whether depredation had increased in recent years, and how benefits from wildlife conservation were shared among group ranch members. Key informants also provided insights on sensitive information like the methods used in killing offending wildlife by the locals.
There are some potential disadvantages of relying solely on questionnaires that might have influenced livestock loss data. Firstly, as Rasmussen (1999) has documented, livestock holders may wrongly attribute stock that has died of natural causes to predation. This may happen through sheer neglect or prejudices towards specific carnivore species. Secondly, livestock keepers might also have had an interest in overestimating the rate of loss, expecting compensation in return. This may have been influenced by the presence of a predator compensation scheme that began operating in OGR in the year 2008.
To avoid this misconception, respondents were informed that the research was purely academic and that no benefits were to be expected. Respondents often bias their recollection of past events in favor of larger species, especially when sampling from multiple years (Kruuk 1980). This was reduced by collecting data for 2008 up to June 2009, instead of using a longer time period. Despite these limitations, several studies show that the livestock keeper’s perception of livestock depredation gives a relatively reliable index of livestock predation (Kruuk 1980; Woodroffe et al. 2005b).
Data from the questionnaires was analyzed using the Statistical Package for Social Scientists (SPSS) version 12.0 (SPSS Inc., Chicago, USA) computer software. ANOVA (Zar 1999) was used to test the statistical difference in means of variables inclusive of the total number of livestock lost to various wildlife species and economic costs associated with the losses of livestock to different wildlife species. The level of significance used for all statistical tests is 0.05. The information collected from focus group discussions and key informant interviews was used to validate and supplement the perceptions of the household respondents.
Average prices of different livestock types (cattle US$332.33, shoats (sheep and goats) US$77.73, and donkeys US$58.36) were computed using different market prices that livestock were sold at as given by respondents and used to calculate the cost of livestock lost to various wildlife species. Values of economic losses are discussed in US$, using the exchange rate of 1 US$ = 78 Kshs at that time.