sign in sign up
<<
Home , Agriculture in Gabon, Cameroon, Panda oleosa, Uganda

The social dimensions of human- conflict in : A literature review and case studies from and

Lisa Naughton*, Robert Rose* and Adrian Treves†

*Department of Geography University of Wisconsin, Madison 550 N. Park Street Madison, WI 53706 [email protected]; [email protected]

†Department of Zoology University of Wisconsin, Madison 250 N. Mills Street Madison, WI 53706 [email protected]

A Report to the African Elephant Specialist , Human-Elephant Task Conflict Task Force, of IUCN, Glands, .

December 1999 Acknowledgments:

Several individuals and organizations provided support:

In Cameroon, Robert Rose’s fieldwork was funded by the Wildlife Conservation Society, through a grant by Dutch Foreign Aid. The Cameroon case study was aided greatly by the expertise of WCS field staff, particularly Anthony Nchanji, Roger Fotso, and Bryan Curran. Walters Arrey was responsible for monitoring crop damage during July-October 1999.

In Uganda, Lisa Naughton and Adrian Treves’ fieldwork was funded by the Wildlife Conservation Society, National Geographic, University Biological Field Station, NSF, and Fulbright-Hays. Pascal Baguma and Patrick Katuramu provided first-rate assistance with data collection and interviews around . Patrick Ilukol and Erica Cochrane generously shared their knowledge of elephant movement and raiding behavior at Kibale.

In Madison, Karen Archabald provided comments on draft excerpts of this report. Erin Olson- Dedjoe and Nora Alvarez helped with data entry.

Finally, Richard Hoare deserves special thanks for his expert counsel and assistance throughout the study.

2 Table of Contents

Acknowledgements…………………………………………………………………………………….2 Table of Contents………………………………………………………………………………………3 List of Tables and Figures……………………………………………………………………………...4

I. INTRODUCTION A. Overview………………………………………………………………………………..…6

B. Physical and social factors intensifying human-elephant conflict in Africa………………8

II. LITERATURE SURVEY A. Estimating crop loss to pests in developing countries………………………………….…11

B. The social significance of crop pests. Collective versus individual coping strategies ….12

C. Comparing to other wildlife ‘pests’ in African and …….…....14

III. CASE STUDY: LOCAL RESPONSE TO CROP DAMAGE BY ELEPHANTS AND OTHER WILDLIFE AROUND KIBALE NATIONAL PARK, UGANDA A. Introduction and historical background……………………………………………….…21

B. Results of previous research on crop raiding at Kibale………………….………………26

C. 1999 Research……………………………………………………………………………28

IV. DAMAGE PATTERNS BY ELEPHANTS AND OTHER WILDLIFE AROUND BANYANG- MBO WILDLIFE SANCTUARY, CAMEROON A. Introduction………………………………………………………………………………41

B. Background……………...………………………………………………………………..43

C. Field Research on Human-Elephant Conflicts at Banyang-Mbo…………………………47

D. Discussion and Management Implications………………………………………………..65

V. CONCLUSIONS……………………………………………………………………………..……68

APPENDIX 1. Study methods for human-elephant conflict research A. Recommendations for research design and methodology………………………………...70

B. Comments on proposed HETF data collection protocol…………………………………..74

APPENDIX 2. References. …………………………………………………………………………...77

3 List of Tables and Figures

II. INTRODUCTION Table II.1. Estimates of crop damage by elephants in Africa……………………………………..…..15

Table II.2. Ranking elephants and other wildlife pests in Africa…………………………...……..17-18

Figure II.1. Schematic of factors influencing local tolerance for wildlife pests………..……….……20

III. CASE STUDY: LOCAL RESPONSE TO CROP DAMAGE BY ELEPHANTS AND OTHER WILDLIFE AROUND KIBALE NATIONAL PARK, UGANDA

Figure III.1. Map of Kibale National Park, Uganda. …………………………………………..……..22

Figure III.2. Elephant Control in Uganda, 1925-1984………………………………………………..25

Figure III.3. Study sites and zones of chronic elephant conflict around Kibale National Park, Uganda. (1992-1999)…… ………………………………………………………..……27

Figure III.4. Distribution of farm size in 5 villages around Kibale National Park, Uganda……….…30

Table III.1. Crop damage by animals in farms neighboring Kibale National Park, Feb-Aug 1999…..32

Figure III.5. Area damaged by wildlife and livestock in 3 villages around Kibale National Park, Feb-Aug 1999………………………………..………………………………………….33

Figure III.6. Direct costs of crop damage by the worst 5 animals on farms neighboring Kibale National Park, Feb-Aug 1999………………………………..……………………….…35

Figure III.7. Total direct costs of crop damage in 51 farms neighboring Kibale National Park Feb-Aug 1999………………………………..………………………………………….36

Figure III.8. Frequency of elephant forays as a function of distance from the boundary of Kibale National Park, 1992-4, 1999……………………………………………………..….…..37

Figure III.9. Percentile plot comparing size of active and abandoned farms……………………..…..39

IV. CASE STUDY: DAMAGE PATTERNS BY ELEPHANTS AND OTHER WILDLIFE AROUND BANYANG-MBO WILDLIFE SANCTUARY, CAMEROON

Figure IV.1 Map of Banyang-Mbo Wildlife Sanctuary, South-West Cameroon. ……………………42

Figure IV.2 1986 Landsat MSS image of the Banyang-Mbo Wildlife Sanctuary, Cameroon. ………45

Figure IV.3 Distribution of agricultural lands between villages and the Banyang-Mbo Wildlife Sanctuary, Cameroon. ………………………………………………………………….……48

Figure IV.4 Villages with chronic elephant damage re: Nchanji and Lawson (1998)………..50

4 List of Tables and Figures (cont.)

Figure IV.5 Locations of elephant and buffalo damage reported during the pilot study (February – April, 1999). ……………………………………………………...………….…52

Figure IV.6 Villages selected for the long-term crop monitoring study ( – October, 1999). ……58

Table IV.1: Crop damage by wildlife around Banyang-Mbo Wildlife Sanctuary, June – October 1999…………………………………………………………………………..……60

Table IV.2: Amount of damage by crop type, Banyang-Mbo Wildlife Sanctuary, June – October 1999……………………………………………………………………………..…61

Figure IV.7 Monthly pattern of wildlife damage to crops at Banyang-Mbo Wildlife Sanctuary, Cameroon (June – October 1999). …………………………………………………………..61

Table IV.3: Crop damage according to field condition, Banyang-Mbo Wildlife Sanctuary, June – October 1999…………………………………………………………………62

Table IV.4: Crop damage by buffalo gathered during the opportunistic data collection (June 1999 – October 1999). …………………………………………………………...……62

Figure IV.8 Crop damage by cane rats and other wildlife as a function of distance from villages. …………………………………………………………………………..……63

Figure IV.9 Locations of buffalo damage reported during the long-term crop monitoring study (June – October, 1999). ……………………………………………………………...………64

5 I. INTRODUCTION

A. Overview Few animals elicit such drastically different human emotions as do elephants. Elephants capture the imagination and unswerving affection of people worldwide, but inspire animosity and fear among those sharing their land with these huge animals. Field reports from across Africa describe local antipathy to elephants beyond that expressed for any other wildlife. People living in central African forests “fear and detest” elephants (Barnes 1996:77). Farmers in display “ingrained hostility” to elephants who are the “focus of all local animosity toward wildlife” (Wunder 1997:314,316). Rural Ugandans complain bitterly about elephants, except where they have been eradicated (Hill 1998; Naughton-Treves 1997), and in , farmers still dread elephants years after the country’s last were killed (A. Weber, WCS, pers. comm.). This animosity is an ominous sign for future elephant survival, particularly given the trend toward decentralized wildlife management throughout Africa. Under current conditions, most local farmers would eliminate elephants from their environment if given the choice. Conservationists must find ways to raise public tolerance of elephants, and this requires a better understanding of elephants as ‘pests’. Do local complaints match the economic impact of elephants on agricultural communities? Why does human-elephant conflict appear to be intensifying even though elephant numbers have declined dramatically across the continent? How can we protect vulnerable individuals from the costs of wildlife while maintaining elephants for regional and global benefits?

In this report, we analyze the social and physical factors that shape local attitudes to African elephants. Our aim is to provide a broader view of the sociopolitical and ecological dimensions of human-elephant conflict. We confirm that elephants pose a serious threat to some members of farming communities, but that in most cases, elephants’ regional economic impact on agriculture is negligible relative to other vertebrate and invertebrate pests. Nonetheless, human-wildlife conflict is a major obstacle to community support for conservation, and the hostility of a vocal minority can undermine regional conservation initiatives (De Boer and Baquete 1998; Gillingham and Lee 1999; Naughton-Treves 1997; Nchanji and Lawson 1998; Newmark et al. 1994).

Our analysis is organized around an extensive literature review coupled with two in-depth case studies. First, we survey reports from 15 African countries to explore the physical and social factors that intensify human-elephant conflict. We also draw from the general literature on pests and risk in African peasant agriculture to better understand why some communities be unable or unwilling

6 to tolerate crop losses. Then we compare elephants to other wildlife pests and examine the spatial and temporal patterns of crop damage by elephants.

In our literature review we find that the database on crop damage amounts and patterns is poor and burdened by ill-defined methods that limit comparisons between species and between sites. Too often, researchers exaggerate impacts by extrapolating results from crop-raiding ‘ spots’ to entire regions, and rarely do they compare farmers’ reports to systematic field measurements. For this reason, we place special emphasis on research design and methods throughout this report. Finally, we conclude that much of the best work is not easily available - e.g., many illuminating reports on wildlife ‘pests’ are tucked away in local Game Department or Ministry of Agriculture archives. As more managers and researchers turn their attention to human-wildlife conflict, we must disseminate our results and experiences. On that note, the Human-Elephant Task Force (HETF) of IUCN’s African Elephant Specialty Group deserves special commendation for their recent efforts to distribute the results of pioneering research (e.g., the special issue of Pachyderm, vol. 19, 1995).

Our review of the literature sets the stage for a more detailed examination of human-elephant conflict at two sites. The first, at Kibale National Park in western Uganda, illustrates the socioecological dynamics of conflict at a ‘hard-edge’ (sensu Hoare 1995), i.e., where densely settled agriculture abuts a park boundary. The second case comes from Banyang-Mbo Wildlife Sanctuary in western Cameroon where elephants and agriculturalists coexist in a mosaic of planted fields, fallows, secondary forests and protected forests. Both case studies show that elephants create distinctive, highly localized crop damage patterns that are cataclysmic for the affected individual farmers, but insignificant to the regional farming economy. At both sites, traditional, collective coping strategies were all but absent and crop losses were absorbed by individual households. The Kibale case also reveals that households with certain economic endowments (namely large landholdings) cope more effectively with elephant raids than do their neighbors. Managers could ameliorate human-elephant conflict if they could better predict high-risk zones for elephant raiding. We suggest hypotheses for the spatial pattern of elephant raiding, and call for better monitoring, particularly in Africa’s forested regions.

Finally, we offer conclusions on the human and ecological dimensions of wildlife conflict in Africa and make research recommendations to improve the database on human-elephant conflict. The need for this research to be cost-effective and participatory is especially challenging given the complexity of most agroecosystems and variability of wildlife and farmer behavior over time and space. We

7 identify basic design principles for sound research on the issue, and discuss trade-offs in sampling intensity and duration of study. Specifically, we evaluate the HETF protocol for monitoring human- elephant conflict.

B. Physical and social conditions intensifying human-elephant conflict in Africa Human-elephant conflict is not a new problem. Precolonial and early 19th c. historians describe areas in Africa where elephant crop depredations caused food shortages and displaced settlements (Vansina 1990; Schweitzer in Barnes 1996:68). dwellers in precolonial northern lived at the “brink of starvation”, because their food supply was regularly devastated by elephants (Barnes 1996:68). In parts of western Uganda early this century, high elephant populations prevented agriculture altogether (GDA 1924; Osmaston 1959). While some observers blame for ruining traditionally harmonious relations between wildlife and local people (Adams and McShane 1992; Martyn 1991), others believe human-elephant conflict is as old as agriculture in Africa (Bell 1987; Naughton-Treves 1999).

Across most of Africa, habitat loss and the local extirpation of big game has reduced the geographical range of human-elephant contact (Hoare 1995). Yet only one out of more than 30 studies published during the and 1990s described a local decline in elephant crop raiding (Bell 1987). Why is human-elephant conflict apparently intensifying? The answer lies in the fact that where elephants persist, contemporary physical conditions draw them into close contact with humans, and contemporary social conditions lower human tolerance of their presence. These conditions include: Land use change: Ø Human-elephant conflict is intensified by the spread of agriculture into previously unoccupied wildlife habitat. This process may be driven by human , voluntary or state-sponsored settlements, or a shift to farming by pastoral communities. The end result is that elephants and farmers compete directly for scarce land (Barnes 1996; Campbell et al. 1999; Gachago and Waithaka 1995; Graham 1973; Hill 1997; Kiiru 1995; Tchamba 1996; Thouless 1994; Thouless and Sakwa 1995; Western 1997). Ø Elephants may be ‘packed’ into protected areas by habitat loss and . Elephants isolated in parks at high densities inevitably raid surrounding farms (Barnes et al. 1995; Gachago and Waithaka 1995; Mwathe 1992; Naughton-Treves 1998; Thouless and Sakwa 1995). Ø In other areas (e.g., sites within ) declining human population density and farm abandonment yields extensive fallow and secondary vegetation (Houghton 1994).

8 Remaining farms may be isolated amidst ‘bush’, and more vulnerable to raiding. Similarly, localized soil degradation forces people to fields in scattered patterns at ever greater distances from villages where they are more vulnerable to elephant raids (Lahm 1996; Mascarenhas 1971). Ø Human activities in African forests (e.g., logging) create abundant secondary vegetation that attracts elephants and brings them closer to human settlements (Barnes et al. 1991; Lahm 1996). Ø Artificially maintained water sources attract elephants to human settlements during droughts (Thouless 1994). Ø The construction of canals, power installations and cattle fences can cut off traditional migration routes and lead to unusually aggressive elephant behavior and conflict (Kangwana 1995; Kothari 1996; Lahm 1994). Changes of elephant behavior and socioecology due to human intervention Ø Following the CITES listing and improved protection against poaching, elephant numbers have increased within many parks and reserves and some elephants have lost their fear of people (Gachago and Waithaka 1995; Kangwana 1995; Naughton-Treves 1998; Tchamba 1996). Ø Elephants may be displaced by war and turn to crop raiding to survive in a resource-poor habitat (Tchamba 1995). Ø Elephants subject to intense hunting or culling form large groups and cause greater damage to local crops and vegetation (Southwood 1977). Changes in social relationships in rural communities Ø Centralized, state ownership of wildlife and prohibitions on hunting lower local tolerance of elephant raiding (Naughton-Treves 1997; Western 1997b). Ø The trend toward privatized land ownership erodes traditional farming strategies based on joint property and kin networks, and focuses the impact of crop loss on individuals rather than communities (Agrawal 1997; Bell 1984; Lahm 1996). Similarly, at many sites farmers have abandoned communal hunting, planting and guarding activities that once reduced crop loss (Lahm 1996; Rose, unpub. data; Mubalama, 1996). Ø Households in much of rural Africa can no longer rely on men to guard fields from elephants because they have moved to cities, seeking employment (Lahm 1996). Increased educational opportunities also releases children from their traditional role as guards and sentinels against raiding wildlife (Goldman 1996).

9 Ø Politicians now pay closer attention to local citizens who complain loudly against elephants, and this raises public awareness of the conflict (Dublin et al. in Barnes 1996; anon. 1994; Hoare 1995; Kangwana 1995).

Together, these social and physical conditions exacerbate conflict that always existed between elephants and agriculturalists. It is also important to recognize that no single factor or condition explains human-elephant conflict across the continent. Elephants and agriculture meet and mix in numerous ways with varying consequences (Hoare 1995). For example, human population growth may heighten conflict with elephants in , Uganda and , while declines in human population heighten conflict in Gabon and Congo. However, one can generalize that edges of protected areas are the focal point of conflict throughout Africa. Farmers residing at these edges typically demand protection or compensation from the government, or they retaliate and kill elephants. Such protests can undermine regional conservation initiatives and turn local conflicts into national political issues (anon. 1994; Tchamba 1995).

Despite growing attention to human-elephant conflict around protected areas, uncertainty persists about the actual magnitude of the problem. Technical experts claim that farmers universally exaggerate crop damage to wildlife (Bell 1984a; Roper et al. 1995; Wakeley and Mitchell 1981). Other studies suggest that elephants and other megafauna are unjustly blamed for damage, and that smaller animals, such as rodents or primates, cause much greater losses over time (Gesicho 1991; Gillingham and Lee 1999; Hawkes 1991; Mascarenhas 1971). The high variability of human-elephant conflict and inadequate database hamper efforts to address this highly charged political issue (Kiiru 1995). To understand farmers’ complaints, the spatial distribution, frequency and extent of crop loss must be examined, as must the socioecological factors that shape local coping strategies and perception of risk. Precise measurement is needed given that leading conservationists now identify human-elephant conflict as a primary threat to elephant survival throughout Africa (Barnes 1996; Hoare 1995; Kangwana 1995; Tchamba 1995; Western 1997)

II. LITERATURE SURVEY

10 A. Estimates of crop loss to pests in the tropics Definitive comparisons of the economic impact of elephants versus other pests1 in African peasant agricultural systems are difficult due to scarce data and extreme variability in crop yields and losses across farms, communities and regions (Porter and Sheppard 1998; Yudelman et al. 1998). However, the general literature on ‘pests’ provides rough estimates for the magnitude of non-wildlife losses, and also reveals important factors shaping local tolerance of pests.

Pest management is a vital issue for tropical farmers. Farmers in tropical environments are exposed to a greater variety of pests than are temperate farmers, although the density of any given pest species is usually lower (Porter and Sheppard 1998). Tropical farmers also tend to be exposed to elevated and chronic levels of loss, versus the periodic outbreaks of single pests in temperate agroecosystems (Oerke et al. 1995; Yudelman et al. 1998). In one study, 60% of farmers (n=916) in rated pests as their primary economic problem, above low crop prices, lack of transport, failed and poor soils (Porter 1976). In Zimbabwe, local farmers ranked pests (including wildlife) first among 30 obstacles to improved quality of life (Wunder 1997). While there is general consensus that pests reduce agricultural productivity significantly in developing countries, losses are rarely measured precisely, particularly in peasant agricultural systems. Estimates for the tropics range from 10 to 50% of total crop production, with an average estimate of 30% loss (Ceres and Howe in Porter and Sheppard 1998; Yudelman et al. 1998). A recent survey estimated even higher losses for African farmers; roughly 51% of production was lost due to insects (15%), pathogens (13%), weeds (13%), and other pests (10%) during the agricultural cycle (Oerke et al. 1995). Other studies have placed greater blame on rodents as major tropical pests (e.g., Wagle in Hill 1997). In the tropics, , and potatoes suffer the greatest losses; all are important food crops for the poor in Africa (Yudelman et al. 1998). These data may lack precision, but they suggest the general order of magnitude of losses.

Analysts pose various reasons why pest loss in peasant agriculture is poorly understood. For one, in many developing countries, priority goes to funding for research on agro-industrial export production. Also, the agronomic research methods imported from temperate zones may be inappropriate to the complexity and scale of peasant agriculture (Goldman 1996; Yudelman et al. 1998). Others emphasize methodological challenges in estimating pest losses (summarized in Yudelman et al. 1998). These challenges are familiar to anyone who has attempted to quantify crop damage by

1 The term ‘pests’ is typically defined as any animal, bird or insect that consumes crops during any stage of the agricultural cycle, from planting to post-harvest storage (Porter and Sheppard 1998). Some definitions also include pathogens and weeds (Yudelman et al. 1998).

11 elephants. Crop yields and losses are difficult to measure and compare because farmers typically plant complex polycultures in fields of ill-defined acreage. Planting densities vary greatly within and between fields. Pest infestations happen sporadically and often coincide with changes in climatic conditions. Few farmers maintain written records. Given the spatial and temporal complexity of peasant agricultural systems, calculating average pest losses is not only difficult, it may be misleading. One farmer may easily tolerate a 15% loss in maize, while her neighbor cannot (Goldman 1996). A 28% loss during a drought may cause a subsistence crisis, but not during a good planting season (Scott 1976). In sum, explaining local tolerance or intolerance to elephants via percent losses is inadequate. One must also address the socioeconomic factors that influence local capacity and willingness to cope with elephants or other wildlife pests.

B. The social significance of crop pests. Collective vs. individual strategies for coping with risk The social significance of crop loss to elephants and wildlife may best be understood in terms of vulnerability and risk2, here both are broadly defined as the potential for loss (sensu Cutter 1996). Vulnerability varies with environmental, technological and social conditions (Carter 1997; Liverman 1990). For example, a farmer might face high levels of risk because he crops in an frequented by ; another might be vulnerable because she lacks kinship ties with her neighbors and has no other source of income. Carter (1997) describes risk as a ‘mechanism of differentiation’, meaning that communities are internally differentiated by individual exposure to risk and individual capacity to cope with risk, and that risk in turn can further differentiate members of communities. In the next two sections we review two factors that commonly influence individual vulnerability to crop raiding by elephants: insurance and wealth.

The vulnerability of smallholder farmers to elephant crop raiding can be mitigated by two insurance strategies: 1) individualist self-insurance (e.g., field scattering, crop diversification, employment of guards on individual property), and 2) social reciprocity between households (e.g., voluntarily sharing public spaces and labor, and aiding less fortunate neighbors) (Carter 1997; Scott 1976). Individualist self-insurance strategies depend heavily on individual access to land, labor, etc. By contrast, social reciprocity insurance depends on traditions of sharing, close community relations and communal land management. Of course there is overlap between individual and social insurance strategies, and farmers may participate in both. However, given the shift toward private landholding and markets, and the decline of social sharing and communal property regimes, the tendency in rural Africa today

2 Carter (1997) used the term ‘risk’ throughout his analysis. Here we use risk and vulnerability interchangeably, although we recognize that theoreticians debate the distinctions between the two terms.

12 is toward greater reliance on individualist self-insurance. This suggests a trend toward individualization of risk (Carter 1997). For example, Bell (1984) demonstrated that large extended families on traditional farms in suffered 80% less crop damage to wildlife than smaller families on small plots in government settlements. Thus a community that once may have collectively coped with the risk of elephant crop raiding is now more likely to face catastrophic damage as individuals or single households.

The capacity of individuals or households to absorb risk, depends largely on wealth (social and physical endowments) and political influence. In peasant agriculture, farm size is an index of wealth and may be the most important endowment for coping with risk. A case study from southern Africa showed that only 10% of individuals in the upper quartile of landholding size suffered food scarcity during drought, while 85% of the bottom quartile suffered food scarcity (Carter 1997). Land availability is also an important predictor of farmers’ capacity to cope with crop losses in Kenya (Goldman 1996). As long as farmers had sufficient access to land, they continued to tolerate 15% losses of their maize yields to invertebrate pests. As land became scarce, individuals bought or changed to another crop (Goldman 1996). Wealth can also be measured in access to capital or labor. Capital permits smallholder farmers to hire guards or build barriers. But the poorest households face compounding vulnerability (Carter 1997; Naughton-Treves 1997). Without large landholdings they cannot buffer themselves from wildlife conflict, nor can they hire additional labor. For example, widows and invalids often suffer the greatest damage within communities and are least able to cope (Bell 1984; Naughton, L. unpubl. data). In short, subsistence farmers with minimal endowments (i.e., access to kinship or community labor and resources) are the most vulnerable (Porter 1979; Scott 1976).

In sum, the general pest literature reveals that: Ø Pest losses in African agriculture are significant, but difficult to measure. Average annual losses to invertebrates, weeds, pathogens and vertebrates combined range from 30 to 50% in African . Ø Individuals’ capacity to cope with crop loss is influenced by environmental, social and technological factors. Wealthier individuals, particularly those with large landholdings, are best able to manage losses, while poor individuals may suffer ‘compounding vulnerability’ (i.e., they live in risky areas and have few resources to cope with losses).

13 Ø Pest losses alone rarely cause farmers to abandon their land. It is when pest problems coincide with other problems (e.g., drought, illness), that their impact is severe enough to cause farm abandonment or other drastic responses. Ø Many farming communities are shifting away from traditional crop loss insurance strategies based on social reciprocity, toward strategies based on individual self-insurance. As risk in individualized, wealth (measured in social and physical endowments) plays a more powerful role in shaping local capacity to cope with crop losses.

C. Comparing elephants to other wildlife ‘pests’ in African forests and savannas 1. Elephant damage: Magnitude and pattern Large vertebrates (>2 kg) are rarely mentioned in the general literature on pests in tropical agriculture, except as localized problems associated with game parks and reserves (Goldman 1996; Goldman 1986; Southwood 1977). The only large vertebrate pest commonly mentioned in nationwide agronomic surveys is the (Potamochoerus sp.). For example, were ranked as a significant problem by over 60% of Tanzanian farmers (only 4% complained of elephants) (n=916, 1972 Agro-climatological survey of Tanzania in Porter and Sheppard 1998). Elephants are seldom mentioned in the traditional agricultural literature. Not surprisingly, precise data on average crop loss to elephants is scarce.

We surveyed the wildlife management literature and found 16 studies that quantified crop damage by elephants (Table II.1). Average losses ranged from 0.2% () to 61% (Gabon) of planted fields. Estimates of annual costs of elephant raids ranged from $60 (Uganda) to $510 (Cameroon) per affected farmer. Elephants were recorded consuming over 20 different crops, with maize ranking first. Nearly all the researchers commented on the irregular, patchy distribution of damage. Certain farms and/or communities were disproportionately damaged due to their proximity to a forest edge, a water source or migration route. Similarly, many researchers commented on variability in the timing of raiding, particularly for forest elephants (Loxodonta africanus cyclopis). Despite their recognition of the highly irregular and localized pattern of elephant damage, few researchers assessed regional levels of damage. Rather, they disproportionately sampled farms hit by elephants and measured damage at these sites. Therefore, these estimates provide valuable information on the experience of individuals suffering losses to elephants, but say little about how this experience is (see Kibale Case Study below). Extrapolating data on average losses to an entire park or region is inappropriate without data on the proportion of farmers affected by elephant raids.

14 15 2. Ranking elephants and other wildlife pests Another method to ascertain the relative cost of elephant crop raiding is to compare their damage to that for other animals. To compare ranking of elephants versus other wildlife pests, we tabulated the results of 25 published and unpublished studies of wildlife pests in Africa (Table II.2). We selected only studies that explicitly ranked problem animals by species or group, and those from sites or regions where elephants are now present or were during the lives of resident farmers (Table II.2). These 25 studies come from 13 countries and include both and forest sites. They also include examples of each major type of human-wildlife interface (Hoare 1995): hard edges (e.g., Uganda, , Kenya and Ghana cases), mosaics (e.g., Gabon, Cameroon) and isolated settlements (e.g., Congo). The table is dominated by S. Lahm’s tour de force study of crop raiding across Gabon. In analyses, we used only her results at the national level. Out of 38 types of animals ranked as problem animals, the five most frequently mentioned were: elephants (32 cases), monkeys (including baboons) (30), rodents (19), bushpigs (18) and antelopes (11). The animal most frequently described as ‘worst animal’ was elephant (8), monkey (including baboon) (8), bushpig (5), (2) and buffalo (2). Elephants’ mean rank was 2.5 +/- 1.5 (n=33), and there was no significant difference between rankings at savanna versus forest sites (Mann-Whitney U test, n=14 savanna and 14 forest sites). We also tested whether farmers and researchers ranked problem animals differently. In 18 studies, the ranks were derived from farmers’ responses to interviews, and in 6, researchers themselves did the ranking based on field observations. Again, there was no difference in the relative ranking of elephants as pests. The only significant discrepancy was between the ranking of elephants at local levels and provincial or national levels. Elephants were not ranked ‘worst pest’ in any of the 5 nation-level assessments (Kenya, Gabon, Tanzania; se also Deodatus 1993on Malawi), and only in 2 of the 15 provincial or district-level rankings. In contrast, 6 of 16 studies conducted on the borders of protected areas ranked elephants worst. This confirms the view that elephants tend to be a significant pest at the local or possibly provincial level, but not at the national level (Dudley et al. 1992).

Comparing “worst pest” rankings between studies is problematic. For one, some studies focused specifically on elephants, and may have biased results accordingly. Also, the scale of analysis varied from single villages to nations. Methods were often poorly defined. Many studies ranked animals by interviewing local farmers and tabulating results by frequency of total complaints or by the frequency that an animal was rated worst. This is a valuable approach for learning about local attitudes and it may reflect past damage. But several researchers have reported that individuals often hope for

16 17 18 compensation and thus may inflate damage reports, particularly for large game species (De Boer and Baquete 1998; Gesicho 1991; Mascarenhas 1971). Other studies ranked animals by the relative amount or frequency of their damage. This approach may avoid the problems of inflated complaints, but it introduces other problems. For example, given the unpredictable nature of raiding by wildlife (particularly elephants), results from a single season or even a single year may miss key events. (These problems are discussed in greater detail in Appendix 1 below). Kiiru (1995) underscores how a lack of quantitative data on crop damage, frequency, patterns and seasonality hinders comparison and ranking of sites. As a result, we consider the data in Table II.2 preliminary, and they should be used and interpreted with caution. 3. Factors shaping local attitudes toward, and capacity to cope with wildlife pests To better understand farmers’ attitudes to various wildlife species, and to explain their apparent intolerance of elephants, we reviewed several field studies done on wildlife pests. Thus we identified key factors shaping local tolerance of crop loss (Figure II.1). Some of these factors are obvious and reflect basic human values and economic rationale. For example, any animal taking human lives is intolerable (Aboud 1986). Livestock losses to wildlife are considered worse than crop losses. Tolerance is shaped more by amounts of crop loss than by frequency of raids (Naughton-Treves 1997). Animals highly prized as game by the local population are generally tolerated despite significant costs. For example, each year, white-tailed deer (Odocoileus virginianus) in the state of Wisconsin cause >$34 million in crop damage and $92 million in damage to vehicles (38,000 deer- car collisions each year) (WDNR 1994). Yet there is widespread support for maintaining a population of >1.2 million deer due to the profitable and popular 9-day annual hunt (670,000 hunters participate and generate $255m in sales) (WDNR 1994). Other influential factors are more subtle. For example, some studies conclude that farmers least tolerate damage to high-value cash crops, while others suggest that animals like bushpigs that target ‘’ crops like cause greater resentment (Mascarenhas 1971). Based on the factors in Figure II.3, one would expect elephants to rank high in lists of worst animals, due to their large size and physical threat, nocturnal raiding habits, and catholic diet. Local intolerance for wildlife may also be amplified by institutional constraints on coping strategies. Farmers feel especially vulnerable to large animals, such as elephants and bushpigs, which inflict localized, infrequent and potentially catastrophic losses. The perceptions of farmers often reflect rare, extreme-damage events rather than persistent, small losses that cumulatively may be greater (Naughton-Treves 1997). The complex interplay of actual risk and the effectiveness of each farmer’s coping strategies is filtered through a cultural and socioeconomic perspective. When asked “which

19 Figure II.1: Factors influencing local tolerance for wildlife pests.

20 animal is worst?” or “how severe are your losses to wildlife?”, a farmer’s answer is shaped not only by his previous experiences with wildlife pests, but also by his perceived status with respect to the park, conservation authorities, and the researcher herself (Naughton-Treves 1997). The following case study from Kibale National Park explores the relationship between damage amounts, perceptions and local response to elephants and other wildlife.

III. CASE STUDY: LOCAL RESPONSE TO CROP DAMAGE BY ELEPHANTS AND OTHER ANIMALS AT KIBALE NATIONAL PARK, UGANDA A. Introduction and historical background3 Kibale National Park is a 760 km2 forest remnant located in western Uganda (in , known formerly as ‘Toro’) (Figure III.1). Kibale is rich in primates and other species (Struhsaker 1997), including those notorious for crop raiding, such as olive baboons (Papio cynocephalus), redtail monkeys (Cercopithecus ascanius), elephants and an unidentified species of bushpig (Potamochoerus sp.). Currently, 54% of the land within 1 km of Kibale’s boundary is used for small holder agriculture (Mugisha 1994). Farmers in the area belong to two predominant ethnic groups, the long-present Batoro, and the immigrant Bakiga, who came to Kibale by the tens of thousands from southwestern Uganda during the 1950s and 1960s (Turyahikayo-Rugyema 1974). Toro chiefs traditionally allocated land to immigrants on the outskirts of their settlements, in part to buffer Toro farmers from crop damage by wildlife (Aluma et al. 1989). Today, both groups plant more than 30 species of subsistence and cash crops: , maize, , yams, and cassava cover the greatest area. In both groups, women generally assume responsibility for food crops, whereas men tend cash crops, such as brewing bananas. Farm sizes are small—averaging 1.4 ha-- and population density is high-272 individuals per km2 4. Within this diverse farming system, various wildlife species forage on crops. Several of these species enjoy legal protection because they are threatened with extinction. As a result, many local cultivators are frustrated and resent the park. To understand the national significance of present-day conflict at Kibale’s boundaries, a brief history of human-wildlife conflict and problem animal control in western Uganda is helpful.

Western Uganda was once famous for its extensive forests and wild animals of extraordinary diversity and density (Hamilton 1984; Wing and Buss 1970). Traditionally, agriculturalists tried to

3 Included in the introduction is material from (Naughton-Treves 1997, 1998. 1999; Naughton-Treves et al. 1998). For further details on Kibale’s local ecology, settlement history and farming strategies, see (Chapman and Chapman 1996; Chapman and Chapman 1997; Edmunds 1997; Struhsaker 1996). 4 Farm-size and population-density figures refer only to smallholdings. More extensive land uses within 1.5 km of Kibale’s boundary include forest fragments, estates and (Mugisha 1994).

21 Figure III.1 Map of Kibale National Park, Uganda.

22 balance crop loss to mammals with bush-meat gains by trapping in and around their fields (Koch 1968; Vansina 1990). Other coping strategies included planting widely dispersed fields in rotation and guarding crops near harvest. Nonetheless, crop damage by wildlife, particularly elephants, prevented the cultivation of some arable land (Osmaston 1959). Isolated agricultural settlements, diminished already by war and rinderpest outbreaks, were especially vulnerable (Osmaston 1959). At some Toro sites early this century, the forest expanded when people abandoned agriculture, reportedly due to ‘the depredations of game that the few remaining inhabitants were no longer able to repel’ (Osmaston 1959). A district warden visiting Toro in 1916 described ‘wanton destruction’ and ‘complete devastation’ of farms by elephants such that ‘the [human] populace has been forced to move elsewhere’ (GDA 1924).

This small is still literally over-run with elephants – big, dangerous, destructive beasts – if we are to afford the inhabitants of this fertile land the measure of protection which is their due, an annual slaughter of elephants on a large scale is not only necessary but imperative (GDA 1924).

In 1906, British colonial authorities enacted a game ordinance that designated wildlife as crown property and greatly restricted so-called ‘native’ hunting (Graham 1973; Naughton-Treves 1999). Claiming all Ugandan animals as crown property legitimized the capture of enormous revenue from ivory and other animal products. According to colonial notions of property, it also obliged the government to assume responsibility for any damage to life or property caused by ‘their game’. First established as the Elephant Control Department (1923) and later renamed, The Ugandan Game Department set out to partition Uganda into wildlife and agricultural areas, or in their terms, ‘elephant and no-elephant areas’ (GDA, 1926,1927, 1931). Great emphasis was placed on controlling and confining wildlife to parks. Several species were subject to control operations, but none received more attention than elephants. ‘Defensive fronts’ were set up, chiefly at agricultural frontiers where elephants ‘swarmed’ (Brooks and Buss 1962). Elephants reportedly responded with ‘evil-intentioned sagacity’ and ‘truculence’, and maintained ‘strongholds’ (GDA, 1928).

During the 1920s, elephants ranged across 75% of Uganda and numbered between 20,000 to 30,000 (GDA 1924). Between 1925 and 1958, the Game Department killed over 1000 elephants a year, for a total of 31,966 (Brooks and Buss 1962). Elephant control campaigns were particularly intense in Toro District, accounting for 29% of the total (GDA, 1957). Toro was considered the most difficult district for elephant control in Uganda, particularly western Toro, where Batoro farmers had newly settled.

23 The Toro district is the most difficult of the control areas and will be hard work for many years to come. There are some thirty to forty herds of elephant totaling fully 2000 animals, the majority of which live in close proximity to settlements and cultivation. (GDA, 1934).

The elephant frontier persisted in Toro and long after elephants were exterminated elsewhere (Figure III.2). But eventually, decades of control shooting packed elephant herds into parks and reserves. Thus the Game Department achieved its goal of confining elephants and other large game to government land. Once elephants were exterminated, land was cleared and crops planted. In Toro, a herd of 2000 elephants sought refuge in the Kibale Forest Reserve where they caused considerable vegetation damage and damaged crops along the forest boundaries (Wing and Buss 1970). In sum, control operations lowered crop loss to large wildlife on a regional basis, and intensified human-wildlife conflict at the edges of parks and reserves.

By the mid 20th century, the central government’s ability to control access to wildlife in parks was threatened by poaching, population growth and civil unrest. In 1971, Uganda plunged into a brutal war lasting over 15 years. During this period of immense human suffering, the Ugandan government lost control of wildlife and parks entirely (Hamilton 1984). War and the de facto removal of all property rights devastated wildlife. For example, rhinoceros (Diceros bicornis) were completely extirpated from Uganda (NEMA 1996). In Bunyoro, elephants dropped from 14,500 in 1969 to 1,420 by 1980 (NEMA 1996). In Toro, Kibale’s herds of elephants were similarly decimated.

With peace in 1987, Ugandan civil society began to rebuild and the national government endorsed biodiversity conservation (NEMA 1996). In 1993, the Kibale Forest Reserve was regazetted as a 760 km2 national park, and more than 30,000 residents of the adjacent Kibale Game Corridor were forcibly evicted and resettled elsewhere (NEMA 1997). Systematic data on wildlife recovery are missing, but Kibale’s surviving herd of 100-300 elephants is growing, and now moves the length of the park (Cochrane, E., unpubl. data).

24 Figure III.2 Elephant Control in Uganda, 1925 – 1984

25 The social and physical landscape of Toro has profoundly changed this century. Where there were once isolated agricultural settlements amidst wildlife habitat, today there are islands and corridors of wildlife habitat embedded in agriculture. Natural habitat continues to shrink outside of Kibale National Park. Edge species persist in Toro agroecosystems (e.g., bushpigs, baboons, and cane rats), but large or interior forest species are mainly confined to the park (Chapman and Onderdonk 1998). Despite regional declines in wildlife populations, farmers living within 1 km of Kibale complain bitterly about crop loss to animals. Anger about crop loss to wildlife is expressed most intensely during group discussions. People ask, “Why should we starve so that baboons may eat?”

B. Results of previous research on crop raiding at Kibale During 23 months spanning 1992-1994, crop damage to animals was monitored weekly in 6 villages in 93 farms lying within 500 m of Kibale’s boundary (Naughton-Treves 1998)(Figure III.3). We also conducted several community-level meetings and 145 interviews to appraise local attitudes to wildlife and coping strategies (Naughton-Treves 1997). The results of this work can be summarized as follows: Ø There was marked variation in frequency and extent of damage within villages, between villages, and between species. The strongest predictor of damage was proximity to the forest boundary. Ninety percent of damage events occurred within 160 m of the forest boundary. Within this narrow band of farms, households lost 4-7% of crops per season on average, varying by village. The distribution of damage was highly skewed, such that maize and cassava fields were on occasion completely destroyed. Ø Five wildlife species accounted for 85% of the forays into fields: baboons, bushpigs, redtail monkeys, , and elephants. Pooling the data for 6 villages, redtail monkeys were the most frequent raiders, and baboons caused the most cumulative damage. Primate raiding on ever- available bananas decreased when fruit was abundant in the forest, but maize was raided at similar frequencies whenever it was available (Naughton-Treves et al. 1998). Ø Livestock caused considerable damage to crops (11% of total), but farmers seldom complained because they had institutionalized modes of restitution. Ø Elephants inflicted catastrophic damage to farms, but their forays were rare and highly localized. The mean crop damage amount per elephant foray was 874 m2, maximum 6510m2. Ø Farmers’ individual defensive strategies (e.g., hunting, strategic crop placement) diminished damage by some species (namely bushpigs), but did not appear to affect elephant raiding.

26 Figure III.3 Study sites and zones of chronic elephant conflict around Kibale National Park, Uganda (1992-1999)

27 Ø Local perception of risk reflected extreme damage, not average losses. All respondents complained vociferously about the severity of crop damage, even those who farmed as far as 1 km from the forest. Anger about wildlife was expressed most vehemently in group meetings. Ø Baboons were most often identified as the worst animals, followed by bushpigs and elephants. Assessments of ‘worst animal’ reflected the respondent’s social identity and economic status. Men complained more about elephants than did women. Owners of small farms more frequently cited damage by smaller animals (e.g., redtails) than did affluent farmers with large farms. Ø The majority of farmers (83%, n=145) believed that local people benefit from the park. But those who suffered elephant damage were significantly less likely to perceive local benefits. Only elephant raiding caused people to abandon farms. Where elephant raiding is frequent, the costs f living near Kibale outweigh its benefits.

A separate study on elephant raiding at Kibale conducted by Patrick Ilukol (1999) provides further insight on their behavior and damage patterns. During 13 months, Ilukol systematically monitored elephant raiding at 4 parishes abutting the forest, selected for their history of elephant raiding problems (Fig. III.3). He found that elephants raided in groups averaging 4.22 individuals (n=67 groups observed) and damaged between 19.6 to 38.4% of standing crops in the fields they raided. Elephants raided with equivalent frequency during wet and dry months, however the distribution and timing of raids apparently follows a cyclic pattern. Specifically, the north and western areas are raided during Jan-May, and eastern and southern areas during June-November.

Research by both Naughton and Ilukol confirms local accounts that elephant raiding is concentrated on certain villages, and often concentrated on certain farms within villages. Game department archives also reveal persistent elephant damage at certain key sites dating to 1951 (e.g., Sebitoli, Fig. III.3). Elephant foraging patterns are likely to be shaped by a complex set of factors operating at a large scale (e.g., forest disturbance by logging in the north of Kibale, heavy poaching in the south).

C. 1999 Research on crop damage and farmers’ responses 1. Objectives During Feb-August, 1999, we resumed monitoring damage to crops in 3 of the 6 original study villages (Fig. III.3). Again, the purpose of the 1999 research was to compare levels of damage between farms, villages and between raiding species. We also aimed to test the results of a lower cost, coarser sampling protocol with that used in the 1992-94 study. Finally, we explored local farmers’

28 long term response to raiding by returning to 5 of the original 6 villages and assessing land use changes on 85 farms.

2. Methods From February-July 1999, crop damage was monitored at 3 villages around Kibale National Park. The villages (Nyabubale, Kabucikire and Rurama) were among the 6 studied from 1992-1994 (Naughton-Treves 1998). These villages were chosen because they agreed to the study, together they faced the complete suite of wildlife crop raiders seen at Kibale (Naughton-Treves 1998), and they were within cycling distance of our assistants’ farms. Apart from these similarities, the three villages differed socially and physically (Naughton-Treves 1998, 1999). Of particular relevance to the present study, the villages differed in mean farm size (Kruskal-Wallis df=2, H=14.79, p=0.005), and in the distribution of farm sizes (Figure III.4). The smallest farms were found at Kabucikire, hence we sampled more farms at this site (N=30) than at Nyabubale (N=8) or Rurama (N=13). Each farm was assigned a row number as follows: farms abutting the park boundary were in row 1, while those behind them and further from the park boundary received row 2 or 3.

Crop damage was measured by our veteran field assistant, P. Baguma, and a new assistant, P. Katuramu, both of whom have farms of their own near Kibale. In the first week of each month, the two men would walk a transect (sweep) perpendicular to the boundary of Kibale. Each sweep was 300 m long and 30 m wide. Along the length of the sweep, crop type and maturity were recorded. Every trace of crop damage by vertebrates was noted and its extent measured by pacing area or counting stalks. Raiding animals were rarely seen, so evidence from dung, tracks, bite marks and patterns of damage were used to infer the identity of the responsible species. Interobserver reliability and damage measurement techniques are detailed in Naughton-Treves (1998). Also detailed there are techniques for identifying independent forays by animals. In brief, when adjacent sweeps crossed the same, large damaged area, only one event was noted (if the raiding species was the same). Similarly, if the same animal inflicted damage at multiple points along a monthly sweep, a single foray was recorded. These methods of determining independence do not inflate frequency estimates, particularly for animals that damage wide swathes of crops (e.g., elephants).

In addition to systematic monitoring of crop damage, we surveyed 5 of the 6 villages (Sebitoli and Kanyasohera plus the three described above) from the earlier study to document changes in land use, ownership, etc. that occurred in the intervening 5 years.

29 Figure III.4 Distribution of farm sizes in 5 villages around Kibale National Park, Uganda. Box plots: each box plot spans the second and third quartile. The central horizontal line indicates the mean, while the vertical error bars denote one standard deviation of the mean.

30 3. Results Overall pattern of crop damage in 1999 We measured crop damage on 51 farms of Nyabubale, Kabucikire and Kanyasohera. Every week, half of the farms suffered crop damage (combining livestock and wildlife). In total, we recorded 269 independent forays over the 6 months of this study. Combined, these forays inflicted crop damage totaling almost 2 ha in area (Table III.1). The mean for all animals was 9.4% of a field’s area damaged per foray.

Differences between animals We recorded damage by 12 varieties of animals (baboon, bird, bushbuck, bushpig, cattle, chicken, , civet, domestic pig, elephant, giant rat, goat, mouse, redtail monkey). Table III.1 presents the results for the 9 types of animals that caused damage more than once (mice and giant rats are pooled). Goats damaged crops most frequently, but elephants did the most damage per foray (mean and maximum). Livestock caused almost two-thirds of the damage, while wildlife caused one- third. Much of the livestock damage was done by animals from a neighboring farm (Table III.1). Among the wildlife, elephants accounted for the vast majority of area damaged (77.9%), but this was confined entirely to 6 farms at one of the three villages (Nyabubale). Baboons were the most frequent visitors.

Differences between villages Just as the villages differed socially and physically, they also differed in the crop damage they faced. Nyabubale, the northernmost, suffered damage in 56 of 168 sweeps (33%), while Kabucikire suffered damage in 95 of 546 sweeps (17.4%) and Rurama, the southernmost, suffered damage in 118 of 285 sweeps (41%). Differences between villages in the frequency of crop damage were mirrored by differences in the average amount of crop damage they suffered (Figure III.5). These data were analyzed with a factorial design ANOVA incorporating village and row as factors to predict the amount of damage in m2. For all animals (wildlife+livestock), the villages differed significantly (df=2, 982, F=12.4, p=0.0001). Divided by wildlife or livestock damage (Figure III.5), villages still differed in the amount of crop damage (wildlife: df=2, 971, F=7.4, p=0.0007; livestock: df=2, 971, F=8.2, p=0.0003). Nyabubale suffered the greatest amount of wildlife damage while Rurama suffered the greatest amount of livestock damage. Row did not predict the amount of damage suffered in any of the analyses described above.

31 32 Figure III.5 Area damaged by wildlife and livestock in 3 villages around Kibale National Park, Uganda (February – August 1999). The height of the bars measures the mean area of crop damage (+/- 1se) recorded in monthly sweeps.

33 Direct costs of crop raiding The direct, financial cost of crop-raiding can be estimated from the value of the crops per m2 multiplied by the area damaged (Figure III.6). Considering single forays, elephants inflicted the highest mean and maximum cost per farmer, but the overall cost of goat damage exceeded that of elephants and all other animals combined (Figure III.7). Indeed, two-thirds of the financial costs of crop damage were caused by livestock (goat, cattle, chicken and domestic pig combined).

Indirect costs of crop raiding Although monetary loss is a major component of the impact of crop raiding, there are other less direct costs of crop loss. Farmers might leave fields fallow or simply abandon them after repeated raids. In other cases, entire farms were abandoned. In 1999, we quantified these indirect costs by surveying 84 farms (in 4 villages) that had been studied from 1992-1994. In essence, we traced the fate of farms in relation to their history of crop raiding. Information was not available for every farm for every question, so sample sizes vary as shown.

In the intervening period between our studies (1994-1999), farmers abandoned 32 fields (N=58 farms, average=0.6±0.9 fields per farm) and left 30 fallow (N=59 farms, average=0.5±0.8 fields). By comparison, clearing of land led to the creation of 60 new fields or an average of 1.8±0.8 fields per farm (N=84, range 0-4). Hence, the clearing of new fields roughly equaled the combined abandonment and fallowing of old ones. There was a correlation between the number of fields cleared and the number fallowed (Spearman rho=0.348, Z=2.65, p=0.008), i.e., the same farmers that cleared new fields were the ones that fallowed older ones. However, there was no correlation between the number of fields cleared and the number abandoned (rho=0.192, Z=1.44, p=0.15).

There was good evidence that farmers abandoned fields because of wildlife crop raiding. Farmers (N=67 interviewed) stated that they abandoned fields because of baboons (36%), bushpigs (24%), weevils (15%), elephants (12%), poor soil (5%) or several rarer reasons: death, illness, chimpanzees and redtail monkeys (1.5% each). Abandoned fields averaged 52 m from the park boundary (se=10.1, n=43). At this distance, they were well within the zone of highest risk for crop raiding (Naughton-Treves 1998). This remained true in 1999, as data on elephant crop damage reveals (Figure III.8).

34 Figure III.6 Direct costs of crop damage by the worst 5 animals on farms neighboring Kibale National Park, Uganda (February – August 1999).

35 Figure III.7 Total Direct costs of crop damage in 51 farms neighboring Kibale National Park, Uganda (February – August 1999), split by type of animal causing damage.

36 Figure III.8. Frequency of elephant forays as a function of distance from the boundary of Kibale National Park, 1992-4, 1999

37 Our measurements of damage were consistent with the idea that field abandonment followed crop raiding. The three villages of the 1999 study differed significantly in the number of abandoned fields (Kruskal-Wallis H=10.82, p=0.0045), and this corresponded to measured crop damage by wildlife. Nyabubale suffered the most wildlife crop raiding (Figure III.5) and had the most abandoned fields (mean of 2.7, pairwise comparisons p<0.003 for each). The other two villages, Rurama and Kabucikire, did not differ significantly with means of 0.6 and 0.3 abandoned fields per farm respectively (p=0.31). Note that in the local context, ‘abandoning’ a field means to leave it without crops for more than 5 years. While ‘fallowing’ a field refers to letting it rest for 1 or 2 years. Of the 67 people who listed reasons for abandoning fields, 35% mentioned baboons, 24% bushpigs, 16% banana weevils (an invertebrate pest), 9% elephants and 7% poor soils.

In 21 cases, entire farms were abandoned. The 5 villages differed significantly in the proportion of farms abandoned (5%- 57%, df=4, X2=16.5, p=0.0024). Again, Nyabubale contained more abandoned farms (45.5%) than either Rurama or Kabucikire (12% and 5% respectively). Only 11 farmers could be interviewed about their reasons for abandoning their farms. They gave 17 reasons. Six farmers blamed elephants and baboons together, one blamed elephants alone, three blamed a death in the family, and one simply blamed .

We sought physical and social factors that might predict which farmers would abandon their farms (farm size, the row it was located in, the ethnicity of the farmer and whether or not a family member was employed by the national park in some capacity). Only the size of the farm predicted abandonment. We had data on the size of 75 farms. Overall, abandoned farms were the same size as farms that were active (Mann-Whitney U, Z=-0.52, p=0.598), however this result is strongly biased by the significant differences in farm size between villages (Figure III.4). To counter this bias, we compared the field size of abandoned and active farms within villages. The mean size of active farms was larger than the mean size of abandoned farms in every case (Wilcoxon signed-ranks test df=4, Z=2.02, p=0.043). Another way to view this data is with a percentile plot relating abandoned and active farms (Figure III.9). The larger farms were significantly less likely to be abandoned, and this effect emerged beyond a size of 1.8 ha.

It appeared that farmers with large land-holdings were less likely to abandon their land when faced by wildlife crop damage. This seems to reflect different land use practices and flexibility in field management. Larger farms contained more abandoned fields (rho=0.408, Z=2.85, p=0.0043), slightly Figure III.9. Percentile plot comparing size of active and abandoned farms

39 more fallowed fields (rho=0.289, Z=2.04, p=0.041), and many more newly cleared fields (rho=0.494, Z=4.22, p<0.0001). In effect, large farms were being maintained as small-scale swidden systems.

E. Discussion From an international perspective, an annual loss of 4-7% of planted fields along Kibale’s boundary equivalent to roughly US$6 per farmer, or US$100 per km of border, appears a trivial price for maintaining elephants and other threatened wildlife. Moreover, most of Kibale’s neighbors extract fuelwood and water from the park worth far more than $6/year (Naughton, unpubl. data). But the farmers who live on Kibale’s border are frustrated by crop loss to wildlife, particularly because they cannot legally use their full range of traditional defensive strategies. Moreover, estimates of average losses the great variation in amounts lost by different farmers and villages. The farmers suffering crop loss to elephants absorbed an average cost of $60 per year, a significant amount in an area where annual incomes average $200-300. A few individuals lost much more. To the farmer who has lost an entire year’s production in a single night to elephants, average losses are meaningless. In some cases elephant damage caused families to abandon their land, particularly those who owned <1.8ha. Although elephant raids are relatively rare, their severe potential impact shapes attitudes among Kibale’s neighbors.

The highly variable and localized nature of crop damage by elephants at Kibale accords with studies around other African parks, making it difficult to assess the socioeconomic impact for the entire park. But our field data from 1992-4 and 1999 and from Ilukol (1999) prove that elephant damage is tightly confined to <200 m of the park boundary. This concurs with others’ views that the best defense against elephants and other large game is to have an active farm between you and the park (Newmark 1997; Hill 1997). Even within this narrow band, only a few farms (on the order of 10%) suffered elephant damage. Obviously, the geographical range of human-elephant conflict in western Uganda has been drastically reduced from a century ago. Today, many more farmers around Kibale struggle with chronic losses to baboons, bushpigs, banana weevils or their neighbor’s goats than from elephants.

40 IV. DAMAGE PATTERNS BY ELEPHANTS AND OTHER WILDLIFE AROUND BANYANG- MBO WILDLIFE SANCTUARY, CAMEROON

A. Introduction Located in the South-West Province of Cameroon (Figure IV.1), the Banyang-Mbo forest is a biologically-rich mix of lowland forest to sub-montane savanna. It provides habitat for a relatively high density of forest elephants (Loxodonta africana cyclotis) and forest buffalo (Syncerus caffer nanus), as well as threatened species including chimpanzee (Pan troglodytes), Preuss’ monkey (Cercopithecus preussi) and ( leucophaeus) (Powell 1994). This region contains 70,000 of Cameroon’s estimated 20 million ha of remaining forest (Besong 1992), and is an important conservation area for both biological and cultural diversity. In March of 1996, Cameroon’s Ministry of Environment and Forest (MINEF) officially designated the Banyang-Mbo Forest as a wildlife sanctuary, the first of its kind in Cameroon (Nchanji and Lawson 1998). The “wildlife sanctuary” designation protects endangered species from hunting and restricts logging, but allows local communities to hunt non-endangered species, gather forest resources, and participate in the management of the sanctuary. This designation follows Cameroon’s national forest use policy to protect soils, habitats and the environment, and provide rural communities with economic benefits deriving from non-timber resource extraction (Besong 1992). In contrast, a “national park” designation would restrict all use by local communities. Following the designation, MINEF invited the Wildlife Conservation Society (WCS) to assist in the formation of a management plan for the sanctuary and to perform all relevant field research. Currently, teams of WCS staff are gathering social and ecological data to support a community-based management plan. They are surveying the flora and fauna within the sanctuary, measuring the impact of local resource use (including bush meat hunting), and assessing the impact of the creation of the sanctuary on local communities.

A key local concern in Banyang-Mbo’s management is human-wildlife conflicts, particularly between humans and elephants. To manage and ameliorate human-wildlife conflicts, field assessments of the pattern and amount of crop damage caused by wildlife are underway, with special emphasis on destruction caused by forest elephants. The present human-wildlife conflict study began with field visits and informal interviews conducted in seven villages around the sanctuary between and April 1999. Following this pilot survey, an intensive crop monitoring study was initiated in five of the seven villages. The study will extend through December 2000. The long-term study uses both opportunistic and systematic data collection schemes. The opportunistic data collection depends on farmers’ reports of recent elephant damage, while the systematic data collection regularly monitored three to four fields in each village for damage by any animal.

41 Figure IV.1 Banyang-Mbo Wildlife Sanctuary located in the South-West Province of Cameroon

42 The study at Banyang-Mbo, along with a previous crop raiding study by WCS field biologist Anthony Nchanji and Dwight Lawson (1998), show four compelling results about the pattern and amount of crop raiding by elephants. First, elephant damage is seasonal, occurring mainly during the rainy season from August to October. Second, elephant damage is concentrated on particular fields and villages due to their location and surrounding vegetation. For example, fields close to the sanctuary edge (and thereby close to secondary forests) or those surrounded by old fallow were more prone to elephant damage. Third, the data show that the pattern of land-use for farming is pushing new fields closer to the sanctuary edge, leading to more incidents of elephant crop raiding. Finally, while elephant damage on an individual field by may be quite high, and in some cases may force the farmer to abandon the field (Nchanji and Lawson 1998), the majority of crop destruction was caused by large rodents, such as cane rats (Thryonomys sp), domestic goats and grasshoppers. This result contradicts the views held by local farmers who complain most bitterly about elephants (Nchanji and Lawson 1998). Perceptions held by local farmers have resulted in elephant crop raiding becoming a highly politicized issue that threatens the relationship between conservation authorities and local communities.

The following case study draws on previous and current studies to present results and conclusions regarding elephant crop raiding. The background section includes a short historical review of the Banyang-Mbo Wildlife Sanctuary (BMWS), and a general description of the sanctuary and surrounding communities and farming systems. Next, the results of the current and previous crop raiding research projects are described followed by an examination of the human reaction to crop raiding around the sanctuary. Finally, management recommendations for the reduction of the impact of elephants on local communities are summarized. While this case study focuses predominantly on the problems facing the conservators at the BMWS in dealing with human/wildlife conflicts, it is also representative of many other situations in forested Central Africa where humans and wildlife attempt to coexist within a landscape of agriculture, secondary forest, and protected areas.

B. Background 1. Management History of the Banyang-Mbo Wildlife Sanctuary The Banyang–Mbo Wildlife Sanctuary (BMWS), located in the South West Province of Cameroon (Figure IV.1), dates back to 1936 when the British Colonial Administration created the Mbo Native Administrative Forest Reserve. Following recommendations from local Cameroonian authorities and the British forestry officers, the colonial administration gazetted an area of approximately 534 km2. The main objective was to protect timber rights for the colonial administration. Following

43 independence in 1960, the government renamed the reserve the Banyang-Mbo Native Authority Forest Reserve (BMNAFR) in 1961. As the name implies, management of the reserve fell under local authority with all resource benefits returning to the communities. However, in 1972, the Cameroon government centralized the management of all national resources, which included the BMNAFR. In essence, this left the reserve with no management, opening its borders to resource exploitation (Nchanji and Lawson 1998).

With increasing resource extraction, international concern focused on the declining wildlife in the BMNAFR and other biologically rich areas in this region. In an effort to protect these important areas, the Wildlife Conservation Society (WCS) established the Cameroon Biodiversity Project (CBP) in 1992 (Nchanji and Lawson 1998). Through this project, the WCS collaborated with the to provide “technical assistance and support in conservation and management of biodiversity and rain forest resources in Cameroon” (Lawson et al. 1996). This project included the Lobeke forest in southeast Cameroon and the Banyang-Mbo area. Following the initial surveys, the WCS recommended setting up a protected area, which became the Banyang-Mbo Wildlife Sanctuary (BMWS), designated by the Cameroon government in 1996. The BMWS includes the original BMNAFR plus an additional region to the south covering a total area of approximately 700km2 (Figure IV.2).

2. Biological Diversity Cameroon holds approximately 20 million ha of tropical moist forest, of which only 0.7 million ha are protected (Besong and Wencelius 1992). The BMWS stands as one of the few protected areas within Cameroon, and contains high densities of elephants, primates and other large mammals. Equally important are the high levels of plant endemism and species richness due to the local variations in elevation (120 to 1756 mnsl) (Curran and Fotso 1997). The BMWS is classified as a Guineo-Congolian wet evergreen lowland (Tonye et al. 1988), with an average rainfall of over 3000mm/yr (Tonye et al. 1988). Common tree species include: , Cola spp., , Uapaca spp., , Pentaclesthra mirophyllia, Mammea africana, , Poga oleosa, Massularia acuminata, Musanga cecropioides, Xylopia spp., Duboscia macrocarpus and Detarium marocarpum (Nchanji and Lawson 1998). Biological surveys within the BMWS have shown evidence of over 25 species of large mammals (Nchanji and Lawson 1998). Surveys have also shown a highly diverse population of herpetofauna and birds (Lawson et al. 1996).

44 Figure IV.2 Banyang-Mbo Wildlife Sanctuary overlayed on a 1986 Landsat MSS false-color infrared image (dark red indicates vegetated areas).

45 3. Local Land Use and Farming Systems Population Cameroon’s population is approximately 14 million (1997 estimate) with an annual growth rate of 2.9% and an overall population density of 27.8 people/km2 ( Intelligence Unit 1999). Most of Cameroon’s population currently live in rural areas (71%) (DeLancey 1989), with 63% of the population making their living through agriculture (The Economist Intelligence Unit 1999). These numbers are likely to decrease as rapid urbanization throughout Cameroon leads to increased emigration from rural areas (DeLancey 1989). In the Southwest Province, where the BMWS is located (Figure IV.1), an estimated 7.7% of Cameroon’s total population (1.1 million) lives in a mostly rural setting. The province has a population density of 38.2 people/km2 (Ndongko and Vivekananda 1989) and the majority of residents (approximately 75%) are agriculturists (1984 data) (Ministry of Agriculture 1986). Approximately 44 villages (average size of 200 people) are within close proximity to the BMWS. These villages include people from 4 different ethnic groups: Banyangi, Mbo, Bassosi, and Bakossi (Nchanji and Lawson 1998). With a pattern of dispersed agriculture settlements amidst forest, the Southwest Province typifies a region especially prone to human-wildlife conflicts.

Farming systems The Southwest province is located in a region with an estimated 330 day growing season (Schaefer- Kehnert 1988). The most common soils in this region are fertile Nitosols (Tonye et al. 1988). However, continuous agriculture has decreased soil fertility in many areas. Both cash crops and food crops are locally cultivated, each in distinct farming systems which significantly influence the pattern of elephant crop damage.

Small-scale cocoa () fields (< 5 ha), along with a small amount of coffee ( canephora), comprise the main cash crops in the region and are cultivated by men. Plantains and bananas (Musa spp.) can usually be found intercropped on cash fields and harvested as an additional source of food or income. Fields in the region are individually or family owned rather than community held. Individuals obtain land for new cash crops by requesting permission from the village traditional council. The acquired land is cleared of any forest or bush, then temporary food crops (cocoa yams (Colocasia spp.), maize (Zea mays), melons) are planted to establish the field. The following year cocoa planting begins, with new trees added each subsequent year until the cocoa field reaches the desired size (Nchanji and Lawson 1998). Once mature, the cocoa trees produce pods that are harvested and sold annually, usually late in the rainy season (October or November). Farmers

46 occasionally rely on other male family members or community farming assistance groups to perform large field projects such as clearing the field of weeds or removal of secondary growth or old fallow. These community farming groups meet once a week to assist in one member’s field on a rotational basis, while work continues on individual fields the remainder of the week. All benefits from the sale of the cocoa go back to individual households.

Women are primarily responsible for cultivating food crops, including cassava (Manihot esculenta), cocoa , maize, melons, groundnuts (Arachis hypogaea), and bananas/plantains. The planting season runs from April to May. The maize matures and is harvested in approximately three months. It is then replanted for a second harvest the same year. Cocoa yams and melons mature in approximately 12-16 months, and are harvested after the final maize harvest. After all harvests are complete, the field is fallowed for two to four years. Female members of the household usually maintain these small food fields as the main source of sustenance for the family. Similar to cocoa farmers, some women form community-farming groups to provide assistance on a rotational basis.

Typically, farming systems in the Southwest Province are characterized by a separation of cash and food fields (Schaefer-Kehnert 1988). In most villages around the sanctuary, food crops are grown near the village, and cash crops near the sanctuary border (Figure IV.3). This planting pattern may is partly due to higher soil fertility on newly cleared land closer to the sanctuary (Mbu Moses, pers. comm.). To obtain a new field or increase the size of an existing field, unused land within the village can be cleared and planted with the permission of the village council (Nchanji and Lawson 1998). This system has led to a highly fragmented mosaic of young and old fields, young and old fallow, and secondary forest.

C. Field Research on Human-Elephant Conflicts at Banyang-Mbo Wildlife Sanctuary 1. Previous Research WCS field biologists Anthony Nchanji and Dwight Lawson examined the amount and location of elephant crop damage as well as the social implications of these conflicts (Nchanji and Lawson 1998). They used a questionnaire and informal interviews to gather information on the history of crop raiding in the area (n=430 people in 44 villages). Following the interviews, researchers monitored elephant crop damage in five core villages using a participatory crop damage reporting system. The team relied on quick reports from farmers suffering elephant damage, which allowed them to visit any recently damaged field within four to five days. Crop damage data included amount

47 Figure IV.3 Distribution of agricultural lands between the villages and the BMWS

48 and nature of damage, field type, field age, field condition, distance of field from village, surrounding vegetation, and damage by other wildlife species.

Type and amount of crop damage by elephants Of all the wildlife, elephants caused damage to the greatest number of different crop species (38), while the African giant rat, bush pig, buffalo, cane rat, porcupine and squirrel caused damage to 14, 11, 10, 7, 6 and 3 different crop species respectively (Nchanji and Lawson 1998). Damaged crops included cassava, cocoa yam, banana and plantain, young cocoa trees, , pineapples, sweet yams, and maize (Nchanji and Lawson 1998). Of the crops damaged by elephants, maize suffered the most overall with an average loss of 66.15% per damaged field (Nchanji and Lawson 1998). Damage to cassava, cocoa yam, banana/plantain, and young cocoa fields (n=433) averaged 57.10%, 64.48%, 51.30%, and 51.75% respectively. These values reflect the average amount of damage occurring on all damaged fields within the core study villages. Overall, the average loss in the study villages was 60% and, in some cases, farmers were forced to abandon their fields entirely.

Damage at the village level ranged from no damage in some villages, to repeated, severe damage in others (Figure IV.4) (Nchanji and Lawson 1998). For example, villages like Bombe Konye and Ntale reportedly had serious problems (78 and 156 fields damaged by elephants in 1996) while others, such as Manyemen and Betock, had little to no problems (10 and 9 fields damaged in 1996). At the field level, the amount of damage also varied dramatically. Among those fields damaged by elephants (n=433), crop losses averaged 60% and ranged from 30 to 88% of the planted area (Nchanji and Lawson 1998). The results indicate a highly localized pattern of elephant crop damage at both the village and field level.

Factors influencing damage patterns Elephants raided certain fields and villages a number of times during the course of the study, suggesting they prefer raiding specific locations. This preference may be a result of: 1) the location of the field or village relative to the Sanctuary, and 2) the vegetation surrounding the field or village. Nchanji and Lawson (1998) found that 67% of the damaged fields were located more than three km from the village (Nchanji and Lawson 1998). Of 251 damaged fields surveyed, 178 (70%) were either at the secondary forest edge or totally enclosed within the secondary forest (forest enclave) (Nchanji and Lawson 1998).

49 Figure IV.4 Villages with chronic elephant damage reported during previous study (Nchanji and Lawson, 1998)

50 Nchanji and Lawson (1998) also found a relationship between the time of year and the amount of elephant damage. Out of 985 reported incidences, 830 (84%) occurred during the rainy season (November-May) (Nchanji and Lawson 1998). The authors suggest two explanations for the seasonality in raiding. First, mature crops and fruiting trees around the crops may have pulled elephants into fields during the rainy season; and second, the rainy season is believed to be a period of low food availability within the sanctuary (Nchanji and Lawson 1998).

The raiding pattern observed at Banyang-Mbo accords with studies of forest elephants elsewhere. Forest elephants apparently prefer secondary forest to other types of land cover due to increased forage availability and dense cover (Dudley et al. 1992; Merz 1986; Barnes et al. 1991). Also, forest elephants often avoid humans, especially in highly hunted areas (Barnes et al. 1997). Therefore, it is reasonable to conclude elephants at Banyang-Mbo are staying close to the sanctuary to avoid humans and to stay within the protective cover of secondary forest. Not surprisingly, fields near the sanctuary are most vulnerable to raiding.

2. Pilot Study (February 1999 – April 1999) Following Nchanji and Lawson’s study, R. Rose initiated field research at Banyang-Mbo with two goals: 1) To quantify and compare the amount of crop damage by elephants and other mammals, and 2) To test predictors for the spatial patterns of crop damage . In collaboration with other researchers, Rose visited six villages along the outside edge of the sanctuary (Tali I, Fotabe, Akiriba, Defang, Sumbe and Ntenmbang) and one village within the sanctuary (Barah) (Figure IV.2). During each three to four day visit, villagers guided the researchers on a thorough examination of many of the village fields. A handheld geographic position system (GPS) unit was used to locate fields with reported previous elephant and buffalo crop damage. These locations were later mapped within a geographic information system (GIS) (Figure IV.5). Along with the location of the field, the type and age of the field, amount of damage and the vegetation surrounding the field were recorded

In addition to the field surveys, informal interviews and community meetings with village farmers were conducted to discuss the problem of elephant crop-raiding and survey local reactions. The informal interviews usually took place while surveying fields and included two to three farmers (both men and women). Community meetings were held in the evening and attendance averaged 30 people per village.

51 Figure IV.5 Farms with previous elephant or buffalo damage reported during the pilot study (February – April, 1999)

52 Results of the pilot study show that: Ø cane rats and other small animals were hitting food crops, while elephants and buffalo caused damage to cocoa and other fields closer to the sanctuary edge, Ø farmers complain more about elephant damage than that of any other wildlife, Ø attitudes and responses towards crop damage varied at both the field and village level, Ø elephant damage was worse during the rainy season (June – November), Ø elephant and buffalo damage was worse on fields near or surrounded by secondary forest, and Ø fields >2km from villages were more likely to be damaged.

Local attitudes and responses to wildlife pests Due to crop losses to wildlife, farmers generally formed negative attitudes towards wildlife and local conservation programs. Village level responses ranged from complaining to local conservationists, to restructuring entire village farming systems. Even adjacent villages strongly differed in their reaction to elephant damage. For example, Sumbe and Ntenmbang are neighboring villages with opposite tolerances and reactions to elephant crop damage. Sumbe farmers, in an effort to reduce the impacts of elephants on food crops, rearranged their entire crop planting system. They now plant cocoa around each individual field of food crops. Ntenmbang still plants food fields close to the village with cocoa fields pushed out towards the edge of the sanctuary. Farmers in Sumbe reportedly now suffer very little from elephant (and other wildlife) damage while Ntenmbang residents continue to complain bitterly about elephants. The response of a village may be motivated by a number of factors, including previous amounts of elephant crop damage and current relationships with local conservationists.

Most villagers agreed that the Cameroon government and conservationists have not taken enough action to alleviate the problem of elephant crop damage. The first crop raiding team visited villages in 1992 and still no solutions have been proposed. The Cameroon government has also promised compensation for damage, but again the villagers were left waiting. In the end, five of six villages selected for the long-term study agreed to cooperate with the program and felt positive that the crop raiding research team was working to help them. The sixth village, Ntenmbang, would only accept the team if guns were brought to destroy the offending wildlife, and was therefore dropped from the study.

53 Despite the fact that cane rats collectively cause much higher crop damage, the strongest local complaints were usually about elephants. There are a number of reasons for this. First, elephants may destroy an entire field in one night’s foray (Nchanji and Lawson 1998). Therefore the impact an elephant may have on an individual farmer is potentially much greater than cane rats. Second, elephants can be dangerous and have been known to attack humans, leading to more animosity towards them. A final factor may be the resentment villagers feel towards the protected status of elephants. Since they are not allowed to shoot them, people feel helpless in protecting their crops from elephant raids. In essence, the farmers felt as though the government favored the lives of elephants over their own well being.

As a result of the protected status of elephants, many farmers also felt that it was the government’s responsibility to protect their fields from elephants and other wildlife and expected the government or conservationists to provide the protection. In one case, a guide in Akiriba village took the team to a field with a fence built around it to guard against cane rats. The fence was built with small holes in the bottom spaced about 5 feet apart. Within each hole was a snare to catch the cane rat as it tried to enter the field. The guide admitted that this method works well, but also stated that it was too much work for most people to do and felt that WCS should provide materials and labor for the fences (ENOW Rudolf, pers. comm.). This was typical of farmers’ attitudes towards protecting fields. These expectations of assistance led to further animosity between villages and the crop raiding team when villagers realized the team only came to discuss the problems and examine the current situation.

In a second case, a farmer who was clearing secondary forest for a cocoa field where elephant and buffalo already foraged felt that once his field was started, the elephants “should know not to come to my field” (MBU Bruno, pers. comm.). Evidently, he expected elephants and buffalo to recognize and respect the boundaries of a newly established field. While this may seem unrealistic, it reflects a common attitude held by communities that nothing should interfere with their farming.

In addition to variation at the village level, attitudes and responses to wildlife varied from farmer to farmer. Based on the informal interviews, these attitudes seem to be driven by three interrelated factors: 1) previous amounts of crop damage, 2) field susceptibility and 3) gender. The most vocal farmers at community meetings and informal interviews usually had experienced elephant damage to their fields. In particular, farmers with small cocoa fields at the edge of the sanctuary tended to complain more about elephants than any other farmer during group discussions and informal interviews. Finally, males usually complained more about elephant damage than

54 females. This was likely a result of males working the highly susceptible cocoa fields near the sanctuary and females working the less susceptible food fields closer to the village.

Tolerance and opinions also changed depending on the setting of the discussion. For example, during one community meeting, participants agreed that elephant crop damage was so severe that there was no option left but to shoot the elephants. However, the next day, during interviews, individuals admitted the problem was not as bad as previously described. In four of the six villages visited, complaints were more severe during large group discussions than in individual interviews. The presence of conservation authorities may have also influenced the intensity of complaint. Villagers finally had someone to listen to their complaints and hoped for significant compensation for damaged crops. Therefore, there was a motivation to make the impact seem worse than it really was, resulting in an inflated estimate of the amount of actual elephant crop damage.

In addition to complaining to authorities, farmers responded to crop damage in a variety of ways, including field guarding, and erecting barriers and hunting offending wildlife. Some went so far as hiring poachers to illegally kill elephants. Both legal and illegal hunting has been used to control crop-raiding elephants. Current laws allow problem elephants to be culled with government permission (Curran, B. pers. comm.). In the case of the legal culling, villagers informed WCS of the problem and, in a cooperative effort, WCS and the Cameroon government called in a professional hunter to kill the problem elephant. However, the majority of elephants are killed without permission. Nchanji and Lawson (1998) report that of 84 elephants killed between 1993 and 1996, with 83 were shot illegally. In the past, villages have invited poachers to destroy problem elephants. In these instances the poachers have killed other elephants leaving the problem elephant to continue raiding. When the elephant returns to the crops the village calls on the poacher again to try to eliminate the problem (Curran, B. pers. comm.). Nchanji and Lawson’s (1998) results show that destroying elephants did not reduce the problem, as elephants were found in the same village the following year. Therefore, hunting was not seen as an effective tool in reducing the amount of crop raiding around the sanctuary.

A second response to crop raiding involved protecting the field through guarding or erecting barriers. Overall, there was little sign of crop guarding except for an occasional magic totem or reports of attempts to scare elephants with loud noises. Farmers have used shotgun blasts to deter elephants, but no other villagers suggested that any guarding worked effectively. Guarding is also costly given that most villages return to their houses in the afternoon and are not near the fields during the nighttime

55 when elephants typically raid. Fear of elephants also motivated people to stay out of their fields at night. As with field guarding, villagers used barriers sporadically with limited success. Researchers observed a fence on only one field and this was designed to keep out cane rats. No elephant barriers were seen in any study village.

The most successful response to crop damage was a village-level change in planting strategies. In Akiriba village, farmers relocated food fields to the opposite side of the main road, away from the sanctuary in order to reduce the impacts of crop raiding. This has eliminated the problem of elephant crop raiding, and has moved fields closer to the village by opening up a new area for agriculture. By planting fields closer to the village, Akiriba village farmers also reduced the time and labor spent to reach distant fields. The previously described example of Sumbe village has also worked well as farmers reported that elephants do not visit Sumbe village fields anymore, with the exception of one field found within the secondary forest (AYUK Moses, pers. comm.).

Spatial pattern of raiding During the pilot study, researchers visited ten fields that were previously damaged by elephants or buffalo. Using a GPS to map the locations these fields, spatial patterns of elephant and buffalo crop damage were revealed. Based on a GIS analysis, 6 of the 10 fields surveyed were found farther than 2km from the village and all 10 fields were at least 1km from the village (Figure IV.5). Secondary forest was the common surrounding vegetation of most fields (60%) with previously reported elephant damage (Figure IV.5). The two spatial predictors of damage observed (distance from the village and amount of secondary vegetation) were intercorrelated. A field located farther from the village was more likely to be surrounded by secondary forest while fallow lands or other fields commonly surrounded a field near the village. In an effort to pull apart this relationship, Rose initiated systematic field monitoring to test the individual influence of these spatial variables on elephant and other wildlife crop damage.

3. Field Monitoring Program (June 1999 – October 1999) Objectives The underlying hypothesis of the monitoring program was that two factors led to higher incidences of elephant crop raiding: 1) field distance from the village, and 2) vegetation surrounding the field. Specifically, fields farther than 2km from the village and surrounded by secondary forest were expected to experience higher incidences of elephant and buffalo crop damage than those closer to the village or surrounded by other fields or young fallow areas. The goal of the monitoring is to validate

56 the influence of these two spatial variables and ultimately develop a predictive model of elephant crop raiding.

Methods The villages selected for this study were Tali I, Fotabe, Akiriba, Defang and Sumbe (Figure IV.6). All five villages were located within a three-hour drive from the WCS research station and could be reached throughout the rainy season. Farmers in these villages followed similar farming practices and all granted approval for the study during community meetings. Finally, these five villages were thought to be equally representative of the human-elephant conflicts around the sanctuary. This was based on results from the pilot study during which all villages complained of recent elephant damage and were able to guide researchers to previously damaged fields.

The research design depended on two types of data collection: systematic and opportunistic. The systematic data collection meanwhile used a matched-pair design to isolate important variables within a highly varying agroecosystem. In a farming system with numerous variables, the matched-pair sampling design helps control for the variability in the agroecosystem which otherwise might confound our analysis. Within each village, pairs of fields were selected that matched in age, type, and size, but differed in one factor, either surrounding vegetation or distance from village. For example, two fields of similar age, type and size are selected, one located close to the village and one located far from the village. The matched pair design enabled the use of a paired ANOVA to assess the effects of variables on crop damage.

In the systematic survey, three to four fields within each village were selected and monitored once a month for crop damage. Field selection was based on three criteria: 1) location from the village (near and far from village), 2) vegetation surrounding field (field/young fallow enclave and old fallow/secondary forest enclave), and 3) type of field (the main crops monitored were: maize, cocoa yams and melons.).

For each village, researchers selected two fields near the village and two fields far from the village (closer to the sanctuary). Researchers used a hand-held GPS to determine the distance from the field to the village. Of the two fields near the village, one was predominantly surrounded by other fields or young fallow (field/young fallow enclave), and the other was predominantly surrounded by old fallow and secondary forest (old fallow/secondary forest enclave). Similar to the fields close to the village,

57 Figure IV.6 Villages selected for long-term study

58 one distant field was a field/young fallow enclave and the other was an old fallow/secondary forest enclave. All fields selected for systematic monitoring were mixed food fields (cassava, cocoa yam, maize, and melon). There were two main reasons for the exclusion of cocoa fields from the systematic survey. First, due to the spatial distribution of cocoa fields, it would have been impossible to obtain four cocoa fields in each village that met the first two selection criteria. Second, due to the large, disorganized nature of cocoa fields, repeated systematic assessment of damage would have been inaccurate. Except for Akiriba, all village fields selected were located between the village and the sanctuary. Akiriba has taken some steps to reduce the impact of crop raiding by locating all food fields on the non-sanctuary side of the village. Therefore, the fields selected are located on the opposite side of the village and only three were selected due to a lack of secondary growth near the village.

For the nineteen fields (5 villages) selected for systematic monitoring: Ø field size ranged from 218.7m2 to 2808m2 with an average size of 1221m2, Ø all fields contained the same crops (maize, cocoa yam, melon), Ø fields varied by proximity to village (close fields averaged 631m from village, distant fields averaged 2380m from village) and, Ø fields varied by surrounding vegetation (field/young fallow enclave or old fallow/secondary forest enclave).

During the first season of systematic crop monitoring, these 19 fields were monitored on a monthly basis for crop damage by any mammal greater than 2 kg. A WCS field assistant (Arrey Walters) along with village guides assessed the amount of damage by counting the number of stems damaged per field. Field assistants used tracks and teeth marks to determine the wildlife species that caused the damage. The data were then analyzed according to wildlife species, crop species and field condition, with the results presented below.

Given the unpredictable nature of elephant crop raiding, the systematic sampling was supplemented by opportunistic observations. The opportunistic data collection was designed to collect data on any elephant or buffalo damage occurring in the five study villages. The data collection depended on a reporting system set up with each village in which any occurrences of elephant damage over the past month were reported to the village chief. Each month, when the researchers arrived in the village, a guide took them to the fields that suffered crop damage and measurements, such as GPS location,

59 type of field damaged, size of field, amount of damage and vegetation along the edge of the field, were made. The results were incorporated into a GIS for final analysis.

Phase One of this study collected data from June 1999 through October 1999. This time frame corresponded to the local rainy season which is the only time elephants reportedly enter crops around the sanctuary. The second phase is scheduled to occur during the rainy season of 2000. Preliminary results from the first season of crop monitoring are presented in the following section.

Results of the first season of crop monitoring Overall, 5 different wildlife species damaged crops. Cane rats damaged more stems on more fields than any other animal, followed by buffalo and porcupines. Elephants did not cause any damage to the fields selected for systematic sampling. This was unexpected, given farmers’ reports of repeated damage in every preceding year. Table IV.1 shows the amount of damage, the frequency of damage per month and the number of different fields damaged by each species. The total number of stems was determined by multiplying the average planting densities for the three main crops damaged (cocoa yam, melons and maize) by the area of each field and summing for all fields (N=19) in the study. The average frequency of damage represents the number of fields damaged by a given wildlife species divided by five months.

Table IV.1: Crop damage by wildlife around Banyang-Mbo Wildlife Sanctuary, June - October 1999

Total Stems Total Stems Percent Frequency of Number of Animal Damaged Available Damaged Damage Different (avg. # of fields/month) Fields Damaged Cane rat 1293 284488 0.455% 6.0 13 Buffalo 168 284488 0.059% 0.6 2 Porcupine 43 284488 0.015% 1.0 3 Antelope 20 284488 0.007% 0.6 2 Bush pig 6 284488 0.002% 0.2 1 Forest 0 284488 0.000% 0 0 elephant

The number of stems damaged per field for each month was recorded and the results were summed for all nineteen fields in the five-month period (Table IV.2). Although all crop types were planted at the same time, maize matures first and is usually planted a second time during the season. Therefore,

60 it was a more readily available target for crop raiders, especially early in the growing season (Figure IV.7).

Table IV.2: Amount of damage by crop type, Banyang-Mbo Wildlife Sanctuary, June 1999 – October 1999 Number of # of Fields Frequency Crop Stems % Damaged Damaged (avg. # of fields Damaged (N=19) damaged/month) Maize 1253 1.26% 13 3.4 Melons 200 0.54% 3 0.8 Cocoa yams 85 0.08% 5 1.4

Figure IV.7: Monthly pattern of wildlife damage to crops at Banyang-MboWildlife Sanctuary, Cameroon (June – October 1999).

11 10 9 8 7 Cocoa Yams 6 Melons 5 Cassava 4 Maize 3 2 1

Number of farms damaged by wildlife (N=19) 0 June July August September October

The data were grouped by field condition (field enclave close to village, secondary forest enclave close to village, field enclave far from village, secondary forest enclave field from village) with the results presented in Table IV.3.

61 Table IV.3: Crop damage according to field condition, Banyang-Mbo Sanctuary, June – October 1999 Fields Close to Village Fields Far from Village Field Enclave Secondary Forest Field Enclave Secondary Forest Enclave Enclave ANOVA # of stems damaged 411(0.51%) 406 (0.56%) 366 (0.50%) 200 (0.34%) ns # fields per month 2.6 1.8 1.8 1 ns Total Number of 4 (n=5) 4 (n=5) 4 (n=4) 2 (n=5) Fields Damaged

Using a factorial ANOVA test on both distance from village and surrounding vegetation, both variables were shown to be non-significant in predicting crop damage. However, there is a trend toward greater crop damage by cane rats close to villages, and greater crop damage by larger animals further from villages (Figure IV.8). Also, all buffalo damage occured in the “distant - field enclave” category. This result suggests that larger mammals may avoid areas of high human densities.

Turning to the opportunistic data, there were no reported incidences of elephant damage in any of the 5 villages over the 5 month study, but 9 events of buffalo damage were recorded in 3 villages (Table IV.4 and Figure IV.9). Of the nine events, 6 were found >2km from the village and 8 were surrounded by secondary forest. All damaged occurred on either cocoa or mixed cocoa/food fields. This confirms expectations regarding the spatial distribution of large mammal crop damage in that the majority of the fields damaged were distant cocoa fields surrounded by secondary forest. Without any data on elephants, it was impossible to draw the same conclusions about the patterns of elephant crop damage.

Table IV.4: Crop damage by buffalo gathered during the opportunistic data collection, Banyang-Mbo Wildlife Sanctuary, June – October 1999.

Village Animal Field Type Surrounding Amount of Damage Sumbe buffalo Cocoa Field 1 plantain tree Fotabe buffalo cocoa/plantain secondary forest 3 plantain trees Fotabe buffalo cocoa/plantain secondary forest 1 banana tree Fotabe buffalo cocoa/food secondary forest 25 cocoa yam stems Tali I buffalo Unknown secondary forest unknown Tali I buffalo Cocoa secondary forest 20 cocoa pods Tali I buffalo Cocoa secondary forest 2 cocoa pods Tali I buffalo cocoa/coffee secondary forest 52 cocoa pods Tali I buffalo Cocoa secondary forest 7 cocoa pods

62 Figure IV.8 Crop damage by cane rats and other wildlife as a function of distance from villages

63 Figure IV.9 Locations of buffalo damage reported during the opportunistic data collection (June – October 1999)

64 D. Discussion and Management Implications Crop damage at BMWS has led to a negative attitude towards wildlife among local communities, and farmers are increasingly calling for the culling of problem elephants. With increased land use and hunting and poaching pressure, the outlook for this small population (estimated at 100) of elephants is dire. If crop raiding continues to be a problem for local farmers, there is reason to believe that the remaining population of forest elephants will be hunted to extirpation.

The following sections compare the results from the crop monitoring study to those of the pilot study as well as the previous study by Nchanji and Lawson (1998). Potential management implications for ameliorating the impact of wildlife on local communities are discussed.

1. Pattern of damage and local complaint The results of the crop monitoring study led to the following conclusions: Ø cane rats did more damage to fields than any other animal, Ø all fields were subject to cane rat damage, although those closest to the village apparently were more frequently damaged, Ø buffalo damage occurred more commonly on cocoa fields located >2km from the village and surrounded by secondary forest, and Ø there was no elephant damage reported in any of the 5 study villages during the study period (June 1999 – October 1999).

These results generally confirm the findings of the pilot study and the previous study by Nchanji and Lawson (1998). The opportunistic data on buffalo also suggests confirmation of the pattern of large mammal (buffalo) crop damage. During both the pilot and crop monitoring study, locations of buffalo damage were consistently reported on fields farther from the village surrounded by secondary forest. However, the results of elephant damage were contrary to expectations. According to farmers’ reports, elephant damage had been so severe in the study villages it was forcing people off their fields. Yet no incidents were recorded during the predicted peak 5 months of raiding. This discrepancy is likely a result of the unpredictable nature of elephant raiding, as well as the tendency for local communities to inflate damage reports, and /or to feel heightened vulnerability to elephants due to other factors. Perceived vulnerability likely reflects the dangerous nature of elephants, the large amount of damage an elephant can cause in one foray, and promises of compensation for elephant damage made by local government officials.

65 2. Management Implications Any management plan proposed should recognize the importance of conserving the forest elephant population while respecting the needs of the local communities. Researchers hope that by reducing human-elephant conflicts, local communities will support conservation efforts in and around the sanctuary. In an effort to reduce the impact of elephants on local communities and protect the wildlife populations, conservators need to elaborate a management plan that decreases the extent of crop raiding while improving villagers’ perception of wildlife and local conservation efforts.

Methods utilized to reduce crop damage in other parts of Africa include building barriers, selective culling, and compensation (Thouless 1995; Hoare 1995). Unfortunately, these solutions may not work well around BMWS. Given the already small size of the sanctuary, an array of fences guarding crops would further reduce access to much needed land for the elephants. Culling at BMWS has already proved of limited success in deterring elephants (Nchanji and Lawson 1998). Compensation for damage may be an effective tool in the future but the current capacity of local institutions to manage such a scheme over a large area is questionable.

The most realistic option is to work with farmers to improve land use patterns so as to minimize large mammal damage. Any plan that moves fields away from high-risk zones should reduce the impact of crop raiding. High-risk zones include fields located near secondary forest and fields located far from the village. New fields are currently cleared in secondary forest near the sanctuary, while land close to the village lies fallow due to local access regimes, previous family claims to land and in come cases, soil infertility. A new land use plan would focus on returning to production old fallow lands near the village and designing a rotational system to allow current farmlands to regain their fertility. By moving fields closer to the village, a forest buffer zone around the sanctuary could be created where no secondary growth is cleared. The fact that two villages (Akiriba and Sumbe) voluntarily adopted such changes suggests other villages might as well. Finally, some areas may incorporate larger cooperative fields into the plan, which would reduce the impact any one farmer might suffer if an elephant does damage the field. The feasibility of cooperative farming is likely to vary greatly from village to village according to ethnicity and land availability.

Monetary compensation for crop damage by large mammals may be used in limited cases, but a set of conservation guidelines should be integrated with the compensation program. First, a buffer zone around the sanctuary should be observed, allowing the area to return to late successional forest. This

66 is dependent on a village’s willingness to alter current land distribution practices. Second, if a village grants access to the sanctuary for outside hunters or poachers, that village should lose all access to compensation for crop damage. Finally, a team of local, WCS and government officials will evaluate any damaged site to determine the nature and extent of the damage. If these guidelines are met, a fair payment, based on current market values and the amount of destruction, could be given to the farmer. An alternate form of compensation may be considered once the quantitative analysis of elephant crop raiding is finished. Using the true percentage of total crops destroyed per growing season, each village can be given a flat rate based on this percentage and fair market value. This comes down to paying the village to accept a small amount of elephant damage. Again, the previously stated requirements of a buffer zone and no access to hunters should be followed. These guidelines are recognized as part of the effort to conserve the population of elephants and are based on current threats to the elephants. They are suggested as a compromise between farmers and wildlife in order to reduce the impact of human/elephant conflicts and relieve the pressure placed on elephants.

67 V. CONCLUSIONS Banyang-Mbo Wildlife Sanctuary, Cameroon and Kibale National Park, Uganda are very different in farming intensity, settlement patterns, and local history of wildlife management. Despite these differences, our field research revealed common threads in human-elephant conflict. At Banyang- Mbo and Kibale, elephant raiding patterns were localized and variable, causing great hardship for some individuals, but less damage than that caused by smaller wildlife species or livestock striking the general community. At both sites, the risk of elephant damage was individualized, although two villages at Banyang-Mbo demonstrated collective crop defense strategies. Thus in the majority of cases, an individual’s landholding size and location were the most important factors shaping her vulnerability. At Kibale, we found empirical evidence that larger landholders coped better with elephant crop damage (i.e. they did not abandon their land after chronic raiding), while the same appeared true for settlements near Banyang-Mbo. But this does not mean that the larger landholders willingly tolerate elephants. Indeed, at Kibale, the wealthier, more powerful farmers were often the most vehement in their demands for compensation from the government. At both sites, hostility to elephants was intensified by general resentment to conservation authorities and the status of elephants as ‘property’ of the state. We also observed that, at both sites, the larger the assembled group of farmers, the louder the complaints and greater the estimates of elephant crop damages.

Human-elephant conflict serves as a vivid example of one of the greatest dilemmas in contemporary conservation: balancing global environmental goals with local residents’ concerns. Conserving elephants imposes risk on local farmers who vary in their capacity to cope with crop losses. To build local support for elephants, local communities must enjoy significant benefits if they are to absorb significant costs. But most community-based initiatives distribute benefits widely, while the costs of elephant crop raiding are felt by a narrow minority (Kangwana 1995; Ngure 1995). The mismatch between levels of conservation costs and benefits is further compounded by elephants’ regional movements. For example, during the dry season, elephants attract tourists to a park in Cameroon and generate local revenue (Tchamba 1995). Then, in the , the elephants move >100km and raid farms in another area without tourism (Tchamba 1995). This raises a dilemma. How should ‘the community’ be defined, given this pattern? Should those suffering the greatest losses to elephants, like the 6 targeted farmers at Nyabubale at Kibale, have special access to benefits from the park and/or greater authority over conservation decisions than other farmers? Similar dilemmas are found elsewhere. For example, citizens in a government settlement on Tsavo National Park’s boundary are subject to high elephant damage and are lobbying to have the park degazetted (Ngure 1995). While they suffer significant losses, their settlement constitutes less than 5% of the elephants’ range.

68 Another kind of dilemma is revealed at Kirinyaga District, Kenya, where local people graze their livestock in forest reserves and illegally fell trees (Gachago and Waithaka 1995). They also suffer high levels of crop damage by elephants (Gachago and Waithaka 1995). Should the Kenyan Wildlife Service invest in protecting or compensating these farms? These Kenyan examples are typical of the boundaries of protected areas in Africa. Often the people residing in the zones of greatest risk (i.e., on the park boundary) are poor and politically marginalized (Brandon and Wells 1992). Immigrants are often disproportionately represented at park boundaries, and they are unlikely to have the same ability to collectively manage risk as long-term resident communities (Gillingham and Lee 1999; Hill 1997; Kanuhi, S. pers. comm.; Naughton-Treves 1997; Porter 1976). Thus park edge communities are often especially vulnerable (both physically and socially) and likely to demand help from the government. Obviously, policies that promote resettlement of landless peasants along park boundaries will inevitably lead to human-wildlife conflict with costly consequences.

Beyond building better fences at park boundaries or planting appropriate buffer crops, one of the most important strategies to ameliorating human-elephant conflict is building local management institutions capable of balancing conservation objectives with the demands of local agriculturalists. Managing elephant crop raiding is inherently a communal or regional endeavor. To ameliorate the incidence and impact of elephant raids, farmers ideally would make collective land use decisions (e.g., plant crops together in large blocks, and/or plant large buffer strips), or draw on traditional insurance systems based on social reciprocity (e.g., share not just the benefits of elephants but the costs as well). More applied research is needed to test the viability of collective management of risk, and to identify political and institutional arrangements that foster community-level tolerance to elephant crop damage. Applied research is also needed to better predict the spatial pattern of elephant raids so that the costs and benefits of wildlife conservation are more equitably distributed. Unfortunately, the trend in much of rural Africa is toward individualized and private land management, making collective management difficult (Agrawal 1996). No doubt in situations where risk in entirely individualized among smallholder farmers, and wildlife is highly endangered, state agencies or conservation NGOs must compensate farmers for crop damage (Madden 1999). To avoid these situations, conservationists must lobby against land use policies that create high-conflict situations, e.g., smallholder settlements placed on park boundaries. And whenever possible, they should promote community-level management of elephants for tourism or hunting, building on promising examples from East and Southern Africa.

69 APPENDIX 1. METHODS FOR THE STUDY OF HUMAN-ELEPHANT CONFLICT A. Recommendations for research design and methodology Conservationists must work in a variety of habitats and social systems with varying amounts of time and money. Hence no single approach to studying human-elephant conflict will work for everyone, nor is one single approach the best. In this section we outline the basic challenges inherent to human- wildlife conflict research and suggest strategies for overcoming these challenges. We draw on our experiences at Kibale, Uganda and Banyang-Mbo, Cameroon and those of investigators elsewhere to offer suggestions regarding the role of researchers in conflict amelioration.

Much of the research on human-wildlife conflict falls into 3 broad categories: 1. Measurement of the timing, extent and distribution of wildlife damage to crops and livestock. 2. Surveys of local attitudes and response to wildlife conflicts. 3. Experimental tests of techniques to mitigate elephant damage.

All 3 types of research are essential to understanding human-elephant conflict. All are difficult, due to the spatial and temporal variability of elephant foraging behavior, often hostile attitudes of local populations suffering losses, and financial and time constraints on research. On the other hand, human-wildlife conflict research is an excellent opportunity for training host country-national students and park staff in research, monitoring and public outreach.

Specific suggestions: 1. Spatial and temporal variability. At Kibale Forest in Uganda, we found that villages separated by <1km suffered losses to different species at dramatically different intensities. This variability also existed at broader scales, requiring a nested sampling. To achieve this aim, we studied 6 villages that differed socially and ecologically and were located at some distance from each other. This design gave us confidence when generalizing about the park, and that our data represented the experiences of individual farmers and villages. Variability between sites and between species of wildlife demands careful attention to study design. A researcher who focuses on a site of known conflict will exaggerate the scale of the problem if he or she extrapolates outward. Similarly, selecting a single species based on local complaint may bias a study against more subtle--but perhaps ubiquitous--pests. Finally, the species of concern for a given site may vary with the observer and respondent. At Kibale, women were more likely to complain of bushpig and baboon damage to food crops, while men focused on elephant damage to cash crops (Naughton-Treves 1997). Individuals on small farms were more likely to complain about smaller pests (e.g.

70 guenons), than large landholders (Naughton-Treves 1997). A further challenge is the temporal variability of crop raiding behavior (Naughton-Treves et al. 1998). Similar caution is warranted in extrapolating results from one season or year to longer time periods. Systematic monitoring provides rigorous, quantitative data, but is restricted in time depth to the observers’ presence. Interviews with local farmers provide a deep time window but are vulnerable to exaggeration. Some combination of the two is probably best, a strategy we refer to as “survey and verify”.

2. Research scale The scale of your research (area covered, repetitions) should balance several factors. Usually, time and funding will set the upper limit on research scale, but the lower limit is set by conditions on the ground. (a) The variability as described above. The greater the temporal and spatial variation between sites (e.g., villages), the more sites one must study. As a rule of thumb, select two or more sites that differ in intensify of conflict. (b) The daily and annual ranging patterns of the species involved in conflict (e.g., elephants will require sampling across kilometers, guenons across hundreds of meters, rodents across tens of meters). (c) The size of local land management units. The researcher should explicitly relate their unit of analysis to the size of local farms and fields (e.g. if one measures damage to specific fields, be sure to indicate how many fields each farmer owns). (d) The relationship between sites of conflict and wildlife habitat. Researchers working in forests bounded by densely-settled agriculture generally concentrate on measuring damage within one or two km of the forest boundary (Barnes et al. 1995; Naughton-Treves 1998; Plumptre and Bizumuremyi 1996). In mosaics, circular study areas or areas that follow natural habitat are most sensible. We offer for comparison the two studies we completed at Kibale. Our 1992-1994 study employed 5 field assistants full time, took 23 months and cost approximately US$12,000 in labor alone. In this study, we mapped 93 farms in 6 villages for four seasons each. This provided us with weekly measurements of crop damage at a resolution of 10 x 10 m down to the number of fruits damaged on a banana tree. These data allow us to determine independence between events (Naughton-Treves 1997), predictors of field vulnerability, and temporal patterns of raiding in relation to forest food availability (Naughton-Treves 1998). We also interviewed 145 farmers and tested the relationship between amount of damage and complaint. By contrast, the 7 month study in 1999 employed two men part time at a cost of US$500 (in labor alone). No interviews were conducted, nor were farms mapped. We lost the ability to define the temporal pattern of raiding and had to rely on the previous study for determination of independence. Yet we could sample 3 villages suffering very different intensities of wildlife conflict. The two studies together

71 allowed us to trace the fate of dozens of farmers whose history of crop loss was at least partially known. At many sites, a detailed, intensive survey followed by coarse scale monitoring at regular intervals should suffice.

3. Measurement. In setting up a monitoring regime, a first step is deciding how to count damage events. Biologists will likely tally events from the animal’s perspective (Sukumar 1989), while social scientists will work from the farmer’s perspective (Hill 1997). This difference in perspective may strongly influence the judgement of the conflict’s intensity. For example, S. Kanuhi (unpub. data) reported >400 events of elephant damage in a single year at Aberdares National Park, based on repeated damage events to individual plots. At Kibale, we counted 34 forays by elephants outside the park in a year, even though 8 farms experienced several incidents of damage. Both approaches are valid. The problem is drawing comparison between Aberdares and Kibale based on these two studies. Some reports do not provide adequate detail about observation methods or the actual questions asked in surveys (e.g., Nchanji & Lawson 1998). Our understanding of human-wildlife conflict would be improved if researchers used more transparent research methodology.

4. GIS technologies in human-wildlife conflict research. Geographic Information Systems (GIS) are a very useful research tool in human-wildlife conflict research given that the spatial distribution of damage is a key management concern. At the simplest level, field data generated from intensive surveying of farms can be entered into a GIS to create layers of information on: incidents of damage, vegetation cover, population density, crop availability, etc. These layers of information can be used in statistical analyses to predict incidents of crop loss (Naughton-Treves 1998). Moving a small step up in technical sophistication, the use of a handheld geographic positioning system (GPS) (US$250-$1000) allows you to georeference your data (i.e., spatially locate or map your data with accuracies of 30-90m). Using a differential GPS (~US$3000) increases accuracy to less than 10 m and allows you to efficiently and accurately measure the size of fields, size of damaged areas, distances, etc. Georeferenced aerial images (air photos, satellite images) may also be incorporated to assist in analyzing land use and land cover patterns over large areas. The promise of GIS is to predict human-wildlife conflict, as shown by some recent work. In , Lewis (1995) used a GIS to help create a community-based wildlife management program. Using local knowledge to build the GIS database, simple maps were created to assist communities in making decisions regarding resource use around local game management areas (Lewis 1995). Harris et. al. (1995) predicted the location of potential conflict

72 by integrating a GIS model of human recreational use with a model of sheep habitat. Similarly, Mace and Waller (1996) predicted human-grizzly bear conflicts in western United States, using GIS modelling approaches.

5. Local participation. One of the most daunting aspects of human-wildlife conflict research is the level of anger or hostility in local communities suffering crop losses. Yet every effort should be made to involve local residents as active participants in the research. To achieve this goal, researchers must invest considerable time in listening to complaints and explaining the goals of his or her project. Naturally, people hope the researcher will provide compensation and/or final solutions to crop raiding. From the onset, it is important to dispel hopes for money or expensive interventions if they will not be possible. At Kibale, we explained that our project would call park managers' attention to the problem (it did), and give people numbers to use in their ongoing efforts to gain resource use rights from the park (it did, but only indirectly via NGO action). This explanation carried weight with some communities, who agreed to monitoring on their farms, attended workshops and are now participating in barrier experiments. Two communities refused to participate when they learned they would receive no compensation. Equally important is participation by the park or reserve staff. Ideally they should have input in research design and monitoring activities and participate in workshop discussions. However, the researcher is likely to be most effective in conflict resolution if local residents view him or her as a third party (see below). Meanwhile, systematic sampling revealed that damage by livestock was significant, yet this damage elicited no complaints directed to the government (see Kibale case study). Including livestock crop damage in monitoring schemes is a provocative way to call attention to levels of tolerance to various animals’ damage.

6. Distribution and application of results Heated controversy often precludes pragmatic discussion of managing human-wildlife conflict. When our research began at Kibale, several farmers complained that they were starving due to crop loss, while some park managers dismissed their complaints entirely. Systematic sampling of crop damage revealed that the actual severity of the problem lay somewhere between the two positions. Providing both parties with basic information on the amount, distribution and relative significance of damage to wildlife can foster productive discussion. Again, at Kibale, a workshop organized at the end of the research allowed local residents to compare conditions between communities, demand that authorities clarify hunting rules, and discuss possible collaborative measures. In the end, two communities volunteered land and labor, and the Field Station offered technical and material inputs for

73 buffer zone trials (the latter did not materialize until IUCN assumed responsibility for the initiative). During such workshops, presenting data in tabular form or as simple percentage may not be appropriate. Presenting maps of areas of high losses will likely be more easily understood. However any information has the potential to be misinterpreted or misused. At Kibale, the atmosphere during the workshop was highly charged due to a recent eviction of squatters from the park. Some feared that the zones of risk we described would become zones of further eviction. Representatives from communities where we had invested the greatest amount of time in pre- workshop discussions and debates were least likely to misinterpret data results, and they fared best in negotiations with Park authorities. At the end of the workshop, the Game Department (now UWA) announced that spearing baboons and bushpigs in farms outside Kibale would be permitted. Another outcome of the workshop was an announcement from the community outreach warden that the residents of Kibale’s edge who faced greatest losses would receive first priority for revenue sharing programs (to date these nascent efforts have distributed $3000 over two years). These deliberations and discussions are important and very time-consuming. Simply providing a report from a field project on crop loss is unlikely to have lasting impact. Finally, conservationists must be mindful that every research intervention has a ‘political cost’. Farmers suffer from ‘interview exhaustion’ around many parks and reserves; thus any research intervention will have an effect on future efforts. No research at all may be better than poorly planned research.

B. Comments on the proposed HETF data collection protocol Overall, this is a commendable effort to systematize and quantify the information coming from field sites. It has the potential to significantly advance our understanding and amelioration of human- elephant conflicts. The protocol best answers the need to quickly and efficiently monitor the intensity of elephant and other large game in zones of high-conflict. The data gathered will aid in the comparison of raiding intensity between high-conflict zones. However, our primary concern is that it will not provide accurate information on the significance of elephant damage for a protected area or a region, nor the relative importance of elephant versus other wildlife pests. This is due to the fact that it is set up to measure damage in recognized zones of conflict, rather than survey the frequency or pattern of incidence between communities at a broader scale. Without more background information on how the conflict site was selected for study and how representative it is of regional conditions, there is a risk that the data will be extrapolated to a regional level and elephant damage will be exaggerated. Beyond our concern with site selection criteria and representation, there are several

74 minor weaknesses that can be easily redressed and will make the database more precise and valuable. We have organized our comments according to the HETF Tables provided.

Table 1 Location (of a conflict zone) is ambiguous. How would location information on community- or province-wide surveys be included? We suggest that several separate lines of data be provided to help locate the study. These would be ordered hierarchically, e.g., country, province or district, locality and finally a longitude and latitude coordinate representing the precise point where human-elephant conflict occurred (if that is available).

Land tenure system, land use and human activities are coded using words like main or major. Perhaps these should read "predominant or most frequent" so that there is less subjectivity in their encoding. Also, under land tenure system, there is no category for corporate-owned land. This category may deserve more attention than it will receive in the 'other' category.

Human Activities -- this is very unclear, particularly if the protocol is to be used across Africa. Are these activities other than farming? What is meant by burning? What about husbandry? commerce?

Habitat poses serious problems of subjectivity in its present form. The researcher should first be asked for rainfall and altitudinal information, then an assessment of tree cover. Shrubland, grassland and semi- are sometimes hard to distinguish and these problems will multiply if non-English speakers are contributing data.

Water availability should specify for humans or for elephants.

Interface type -- again mosaic, shifting and isolated settlement are highly overlapping categories. Perhaps this could be simplified to capture whether the conflict arose in a community surrounded by natural vegetation or surrounded by agropastoral lands or by a mixture?

Conflict season -- for sites with 2 rainy and 2 dry seasons, we may need another option. Perhaps year- round should be used to distinguish intermittent conflict that can erupt at any time from conflict that happens frequently without regard to season?

75 Total raids -- these data are potentially the most inflammatory for political reasons, hence they deserve the greatest caution in collection and interpretation. In particular, we are concerned with independence of data points. If elephants cause a series of incidents in the course of one foray, these should not be counted as several incidents of conflict. The damage done can be summed but the frequency of conflict should not be inflated if multiple farms are affected. In short, we feel that the only reliable measure of frequency is one based on field verification by trained individuals.

Main group size -- some idea of sample size and standard error would help to evaluate these data.

Foodcrop damage and cashcrop damage -- we were under the impression that cassava and manioc were the same thing. In these and subsequent cells, it is not clear if one is intended to rank one sort of damage relative to others (i.e., coffee vs. cocoa) or rank damage relative to some conceptual benchmark (1=total ruin, 5=slight)?

Human death -- specify that these were due to elephants

Table 2 Population estimate -- the source of this estimate should be given. Area -- as above Conservation status is ambiguous -- perhaps the researcher could simply enter "Legal protection not enforced", "Legal protection enforced", "No legal protection", "Unknown", "Other"

Uncontrolled hunting pressure -- the distinction between categories may be too fine given the variety of sources and scales of investigation.

The simple index of crop damage seems an excellent idea. The fact that area damaged is weighted more heavily than the other two variables is appropriate. But 1. crop quality will always be difficult to describe, 2. six gradations are likely too fine, and 3. percentages of a field do not capture the magnitude of damage – e.g., 50% of a 1 ha field is a lot less than 25% of a 10 ha field. Could the index be changed to a four-numeral system? The first three numerals as you have them and the last one categorizing the size of the field or the total value of the crop?

76 APPENDIX 2. REFERENCES Aboud, A. A. 1986. Integrating wildlife resources with human needs in I. MacMillan, ed. Wildlife/Livestock Interfaces on Rangelands Conference. USAID, . Adams, J. S., and T. O. McShane. 1992. The Myth of . W.W. Norton & Co., New York. Agrawal, A. 1997. Community in conservation: beyond enchantment and disenchantment. Conservation and Development Forum, Gainesville, Florida. Aluma, J., C. Drennon, J. Kigula, S. W. Lawry, E. S. K. Muwanga-Zake, and J. Were. 1989. Settlement in forest reserves, game reserves, and national parks in Uganda. Makerere Institute of Social Research, . . 1994. Firing Line. Pages 3-10. The Weekly Review. Barnes, R., S. Azika, and B. Asamoah-Boateng. 1995. Timber, cocoa, and crop-raiding elephants: a Preliminary study from southern Ghana. Pachyderm 19:33-40. Barnes, R., K. Jensen, and A. Blom. 1991. Man determines the distribution of elephants in the rain forests of northeastern Gabon. African Journal of Ecology 29:54-63. Barnes, R. F. W. 1996. The conflict betweeen humans and elephants in the central African forests. Mammal Review 26(2/): 67=80. Barnes, R. F. W., et al. (1997). “Estimating Forest Elephant Numbers with Dung Counts and a Geographic Information System.” Journal of Wildlife Management 61(4): 1384-1393.

Bell, R. H. V. 1984a. The man-animal interface: An assessment of crop damage and wildlife control. Pages 387-416 in R. H. V. Bell and E. McShane-Caluzi, eds. Conservation and Wildlife Management in Africa. U.S. Office of Training and Program Support, Malawi. Bell, R. H. V. 1987. Conservation with a human face. Pages 79-101 in R. Anderson and R. Grove, eds. Conservation in Africa. Cambridge University Press, Cambridge, . Besong, J. B. (1992). New directions in national forestry policies - Cameroon. Conservation of West and Central African Rainforest. K. Cleaver, M. Munasinghe, M. Dysonet al. Washington, D.C., The Bank. 1: 353. Besong, J. B. and F. L. Wencelius (1992). Realistic strategies for conservation of biodiversity in the tropical moist forests of Africa: regional overview. Conservation of West and Central African Rainforest. K. Cleaver, M. Munasinghe, M. Dysonet al. Washington, D.C.,The . 1: 353 Booth, V. R., N. N. Kipuri, and J. M. Zonneveld. 1992. Environmental impact of the proposed fencing programme in Kenya. Elephant and Community Wildlife Programme, Kenya Wildlife Service, Nairobi. Brandon, K. E., and M. Wells. 1992. Planning for people and parks: Design dilemmas. World Development 20(4): 557-570. Brooks, A. C., and I. O. Buss. 1962. Past and present status of the elephant in Uganda. Journal of Wildlife Management 26(1): 38-50.

77 Campbell, D., H. Gichohi, A. Mwangi, L. Chege, and T. Sawin. 1999. Interactions between people and wildlife in SE Kajiado District, Kenya. Ford Foundation, Nairobi, Kenya. Carter, M. R. 1997. Environment, technology, and the social articulation of risk in West African agriculture. and Cultural Change 45(3): 557-590. Chapman, C. A., and L. J. Chapman. 1996. Mid-elevation forests: A history of disturbance and regeneration. Pages 385-400 in T. R. McClanahan and T. P. Young, eds. East African Ecosystems and their Conservation. Oxford University Press, New York. Chapman, C. A., and L. J. Chapman. 1997. Forest regeneration in logged and unlogged forests of Kibale National Park, Uganda. Biotropica 29:396-412. Chapman, C. A., and D. A. Onderdonk. 1998. Forests without primates: primate/plant codependency. American Journal of Primatology 45:127-141. Curran, B. and R. Fotso (1997). Cameroon Biodiversity Project - Banyang-Mbo Sanctuary Six Month Report of Activities (July - December 1997), Wildlife Conservation Society: 9. Cutter, S. L. 1996. Vulnerability to environmental hazards. Progress in Human Geography 20(4): 529-539. De Boer, W., and D. Baquete. 1998. Natural resource use, crop damage and attitudes of rural people in the vicinity of the Elephant Reserve, . Environmental Conservation 25(3): 208-218. DeLancey, M. W. (1989). Cameroon - Dependence and Independence. Boulder, CO, Westview Press Dudley, J. P., A. Y. Mensah-Ntiamoah, and D. G. Kpelle. 1992. Forest elephants in a rainforest fragment: preliminary findings from a wildlife conservation project in southern Ghana. African Journal of Ecology 30:116-126. Edmunds, D. 1997. Continuity and Change in the Resource Management Institutions of Communities Bordering the Kibale Forest Park, Uganda. Pages 327 pp. Geography Department. Clark University, Worcester, Massachusetts. Gachago, S., and J. Waithaka. 1995. Human-elephant conflict in Kiambu, Murang'a, Kirinyaga, Embu and Meru Districts. Kenya Wildlife Service, Nairobi, Kenya. GDA. 1924. Game Department Archives of Uganda. Property of the Ugandan Wildlife Authority, Ministry of Tourism and Wildlife, Government of Uganda, Kampala, Uganda. Gesicho, A. 1991. A survey of the Arabuko Sokoke elephant population. KWS Elephant Programme, Nairobi, Kenya. Gillingham, S., and P. Lee. 1999. The impact of wildlife-related benefits on the conservation attitudes of local people around the Selous Game Reserve, Tanzania. Environmental Conservation 26(3): 218-228. Goldman, A. 1996. Pest and disease hazards and sustainability in African agriculture. Experimental Agriculture 32:199-211. Goldman, A. C. 1986. Pest hazards and pest management by small scale farmers in Kenya. Clark University. Graham, A. D. 1973. The Gardeners of Eden. George Allen & Unwin Ltd., London. Hamilton, C. A. 1984. Deforestation in Uganda. Oxford University Press, Nairobi.

78 Hawkes, R. K. 1991. Crop and livestock losses to wild animals in the Bulilimamangwe Natural Resources Management Project Area. Center for Applied Social Sciences, . Harris, L., et al. (1995). “Multiple Use Management: Using a GIS Model to Understand Conflicts Between Recreationists and Sensitive Wildlife.” Society and Natural Resources 8: 559-572. Mace, R. D. and J. S. Waller (1996). “Grizzly Bear Distribution and Human Conflicts in Jewel Basin Hiking Area, Swan , Montane.” Wildlife Society Bulletin 24(3): 461-467. Hill, C. 1998. Conflicting attitudes towards elephants around the Budongo Forest Reserve, Uganda. Environmental Conservation 25(3): 244-250. Hill, C. M. 1997. Crop-raiding by wild animals: the farmers' perspective in an agricultural community in western Uganda. International Journal of Pest Management 43(1): 77-84. Hillman Smith, A. K. K., E. d. Merode, A. Nicholas, B. Buls, and A. Ndey. 1995. Factors affecting elephant distribution at and surrounding reserves, Zaire, with a focus on human- elephant conflict. Pachyderm 19:39-48. Hoare, R. 1995. Options for the control of elephants in conflict with people. Pachyderm 19:54-63. Houghton, R. A. 1994. The worldwide extent of land-use change. Bioscience 44:305-313. Ilukol, P. 1999. Elephant foraging and crop raiding on the border of Kibale National Park, Uganda. M.Sc. Thesis. Department of Zoology. Makerere University, Kampala Uganda. Kangwana, K. 1995. Human-elephant conflict: the challenge ahead. Pachyderm 19:11-14. Kanuhi, S. 1995. Elephant crop raiding around Aberdares National Park. , Nairobi, Kenya. KIFCON. 1992. Report on a survey to assess the damage caused by wild animals to farm-households adjacent to the Aberdares Forest Reserve. Kenya Indigenous Forest Conservation Project, Nairobi, Kenya. Kiiru, W. 1995. The current status of human-elephant conflict in Kenya. Pachyderm 19:15-18. Kiss, A. 1990. Living with Wildlife. Wildlife Resource Management and Local Participation in Africa. World Bank Technical Paper Number 130/Africa Technical Department Series, Washington, D.C. Koch, H. 1968. Magie et Chasse dans la Foret Camerounaise. Berger-Lerrault, Paris. Kothari, A. 1996. Toward joint management of 's protected areas. Workshop on Collaborative Management for Conservation, IUCN World Conservation , Montreal. Lahm, S. A. 1994. The impact of elephants on agriculture in Gabon. Pages 90. WWF Programme pour le Gabon, B.P. 9144, , Gabon. Lahm, S. A. 1996. A nationwide survey of crop-raiding by elephants and other species in Gabon. Pachyderm 21:69-77. Lawson, D., et al. (1996). Cameroon Biodiversity Project - Six Month Report of Activities, Wildlife Conservation Society: 10. Lewis, D. 1995. Importance of GIS to community-based management of wildlife: lessons from Zambia. Ecological Applications 5(4): 861-871. Liverman, D. 1990. Vulnerability to global environmental change. Pages 27-44 in R. E. Kasperson, ed. Understanding Global Environmental Change. The Earth Transformed Program, Worcester, MA.

79 Mace, R. D. and J. S. Waller (1996). “Grizzly Bear Distribution and Human Conflicts in Jewel Basin Hiking Area, Swan Mountains, Montane.” Wildlife Society Bulletin 24(3): 461-467. Madden, F. 1999. Human- conflict at Bwindi Impenetrable Forest, Uganda.Report to the International Gorilla Conservation Programme. , Uganda. Martyn, R. 1991. People-wildlife conflicts in Africa: Are there symbiotic alternatives? Wye College. University of London, London. Mascarenhas, A. 1971. Agricultural vermin in Tanzania. Pages 259-267 in S. H. Ominde, ed. Studies in East African Geography and Development. University of California press, Berkeley. Matzke, G. 1975. Large mammals, small settlements and big problems: A study of overlapping space preferences in Southern Tanzania. Pages 190. Geography. Syracuse University, Syracuse. Merz, G. (1986). “Counting elephants (Loxodonta africana cyclotis) in tropical rain forests with particular reference to the Tai National Park, .” African Journal of Ecology 24: 61-68. Ministry of Agriculture (1986). Cameroon Agriculture in Figures 1986, of Cameroon, Ministry of Agriculture: 28. Mugisha, S. 1994. Land cover/use around Kibale National Park. MUIENR, RS/GIS Lab, Kampala, Uganda, Kampala, Uganda. Mwathe, K. M. 1992. A preliminary report on elephant crop damage in areas bordering Shimba Hills National Reserve. Kenya Wildlife Service Elephant Program. Naughton-Treves, L. 1997. Farming the forest edge: Vulnerable places and people around Kibale National Park, Uganda. The Geographical Review 87(1): 27-47. Naughton-Treves, L. 1998. Predicting patterns of crop damage by wildlife around Kibale National Park, Uganda. Conservation Biology 12(1): 15 pages. Naughton-Treves, L. 1999. Whose Animals? A history of property rights to wildlife in Toro, western Uganda. Land Degradation and Development 10:311-328. Naughton-Treves, L., A. Treves, C. Chapman, and R. Wrangham. 1998. Temporal patterns of crop raiding by primates: Linking food availability in croplands and adjacent forest. Journal of Applied Ecology 36(4). Nchanji, A. C., and D. P. Lawson. 1998. A survey of elephant crop damage around the Banyang-Mbo wildlife sanctuary, 1993-1996. Pages 96. Cameroon Biodiversity Project and Wildlife Conservation Society, Yaounde, Cameroon. Ndongko, W. A. and F. Vivekananda (1989). Economic Development of Cameroon. Stockholm, Sweden, Bethany Books. NEMA. 1996. State of the Environment Report for Uganda. Pages 271. National Environment Management Authority, Kampala, Uganda. NEMA. 1997. Kabarole District Environment Profile. Pages 77. National Environment Management Authority, Kampala, Uganda.

80 Newmark, W. D., D. N. Manyanza, D.-G. M. Gamassa, and H. I. Sariko. 1994. The conflict between wildlife and local people living adjacent to protected areas in Tanzania: Human density as a predictor. Conservation Biology 8(1): 249-255. Ngure, N. 1995. People-elephant conflict management in Tsavo, Kenya. Pachyderm 19:20-25. Oerke, E. C., H. W. Dehne, F. Schohnbeck, and A. Weber. 1995. Crop production and crop protection: Estimated losses in major food and cash crops. Elsevier Press, Amsterdam. Osmaston, H. A. 1959. Working plan for the Kibale and Itwara Central Forest Reserves, Toro District, W. Province, Uganda. Forest Department, . Plumptre, A. J., and J. B. Bizumuremyi. 1996. Ungulates and hunting in the Parc National des Volcans, Rwanda. Wildlife Conservation Society, New York. Powell, J. A. (1994). Cameroon Biodiversity Project - Status Report January - June 1994, Wildlife Conservation Society: 8. Porter, P., and E. S. Sheppard. 1998. A World of Difference. The Guilford Press, New York. Porter, P. W. 1976. Agricultural Development and Agricultural Vermin in Tanzania. Pages 37. Geography of Economic Development. American Association for the Advancement of Science, Boston. Porter, P. W. 1979. Food and Development in the Semi-Arid Zone of . Maxwell School, Syracuse University, Syracuse, New York. Roper, T. J., S. R. Findlay, P. Lups, and D. J. Shepherdson. 1995. Damage by badgers Meles meles to wheat Triticum vulgare and barlery Hordeum sativum crops. Journal of Applied Ecology 32:720-726. Schaefer-Kehnert, H.C. (1988). Institutions and Technical Change in the Development of Smallholder Agriculture. Germany, Wissenschaftsverlag Vauk Kiel KG. The Economist Intelligence Unit (1999). The World of Information - Africa Review 1999. Essex, UK, The Economic and Business Report. Scott, J. C. 1976. The Moral Economy of the Peasant. Yale University Press, New Haven. Southwood, T. 1977. The relevance of population dynamic theory to pest status. Pages 36-54 in J. M. Cherret and G. R. Sagar, eds. Origins of Pest, Parasite, Disease and Weed Problems. Blackwell, Oxford, UK. Struhsaker, T. T. 1996. Ecology of an African Rain Forest. University Press of Florida, Gainesville, FL. Struhsaker, T. T. 1997. Ecology of an African Rain Forest. University Press of Florida, Gainesville, FL. Sukumar, R. 1989. The Asian Elephant. Cambridge University Press, Cambridge, U.K. Tchamba, M. 1995. The problem elephants of Kaele: A challenge for elephant conservation in Northern Cameroon. Pachyderm 19:26-31. Tchamba, M. N. 1996. History and present status of the human/elephant conflict in the Waza-Logone region, Cameroon, . Biological Conservation 75:35-41. Thouless, C. R. 1994. Conflict between humans and elephants on private land in northern Kenya. Oryx 28:119- 127. Thouless, C. R., and J. Sakwa. 1995. Shocking elephants: Fences and crop raiders in Laikipia District, Kenya. Biological Conservation 72:99-107.

81 Tonye, J., et al. (1988). Agroforestry Potential in the Humid Lowlands of Cameroon. Nairobi, The Cameroon Agroforestry Task force and The International Council for Research in Agroforestry: 98. Turyahikayo-Rugyema, B. 1974. The history of the Bakiga in southwestern Uganda and northern Rwanda, 1500-1930. University of Michigan, Ann Arbor, MI. Vansina, J. 1990. Paths in the . The University of Wisconsin Press, Madison, WI. Wakeley, J. S., and R. C. Mitchell. 1981. Blackbird damage to ripening field corn in Pennsylvania. Wildlife Society Bulletin 9:52-55. WDNR. 1994. Wisconsin's Deer Management Program. Pages 31. Wisconsin Department of Natural Resources, Madison, Wisconsin. Western, D. 1997a. In the Dust of Kilimanjaro. Island Press, Washington, D.C. Western, D. 1997b. Natural Connections. Island Press, Washington, D.C. Wing, L. D., and I. O. Buss. 1970. Elephants and forests. Wildlife Monographs 19:1-92. Wunder, M. B. 1997. Of elephant and men: Crop destruction, CAMPFIRE, and wildlife management in the Zambezi valley, Zimbabwe. Pages 360. Natural Resources and Environment. University of Michigan, Ann Arbor. Yudelman, M., A. Ratta, and D. Nygaard. 1998. Pest Management and Food Production. Pages 53. International Food Policy Research Institute, Washington, D.C.

82