Can parks protect migratory ungulates? The case of the Serengeti wildebeest

Transcription

1 Animal Conservation (2004) 7, C 2004 The Zoological Society of London. Printed in the United Kingdom DOI: /S Can parks protect migratory ungulates? The case of the Serengeti wildebeest Simon Thirgood 1, Anna Mosser 2, Sebastian Tham 1, Grant Hopcraft 1, Ephraim Mwangomo 3, Titus Mlengeya 3, Morris Kilewo 3, John Fryxell 4,A.R.E.Sinclair 5 and Markus Borner 1 1 Frankfurt Zoological Society, PO Box 14935, Arusha, Tanzania 2 Department of Ecology, Evolution & Behavior, University of Minnesota, St. Paul, Minnesota 55108, USA 3 Tanzania ational Parks, PO Box 3434, Arusha, Tanzania 4 Department of Zoology, University of Guelph, Ontario 1G 2W1, Canada 5 Centre for Biodiversity Research, University of British Columbia, Vancouver V6T 1Z4, Canada (Received 28 August 2003; accepted 12 January 2004) Abstract The conservation of migratory species can be problematic because of their requirements for large protected areas. We investigated this issue by examining the annual movements of the migratory wildebeest, Connochaetes taurinus, in the km 2 Serengeti-Mara Ecosystem of Tanzania and Kenya. We used Global Positioning System telemetry to track eight wildebeest during in relation to protected area status in different parts of the ecosystem. The collared wildebeest spent 90% of their time within well-protected core areas. However, two sections of the wildebeest migration route the Ikoma Open Area and the Mara Group Ranches currently receive limited protection and are threatened by poaching or agriculture. Comparison of current wildebeest migration routes to those recorded during indicates that the western buffer zones appear to be used more extensively than in the past. This tentative conclusion has important repercussions for management and needs further study. The current development of community-run Wildlife Management Areas as additional buffer zones around the Serengeti represents an important step in the conservation of this UESCO World Heritage Site. This study demonstrates that detailed knowledge of movement of migratory species is required to plan effective conservation action. ITRODUCTIO Tanzania has one of the world s most extensive networks of protected areas with about 25% of the country managed for conservation. Some of these areas are very large for example the Selous Game Reserve, covering km 2, is the largest protected area in Africa. Despite these vast areas, concerns have been expressed over the ability of the protected area network to conserve biodiversity. These include the location of protected areas in relation to biodiversity hotspots (Lombard, 1995), the unsustainable use of natural resources (Campbell & Hofer, 1995) and the encroachment of agriculture into protected areas (Homewood et al., 2001). Concerns have also been raised that protected areas are too small for wide-ranging species that come into conflict with people (Woodroffe & Ginsberg, 1998). Large protected areas are also important for migratory species and it is notable that declines have occurred in some populations of migratory ungulates in Africa and Asia (e.g. wildebeest, Connochaetes taurinus, All correspondence to: Simon Thirgood. Tel: ; Fax: ; simonthirgood@fzs.org in Botswana (Williamson & Williamson, 1984), whiteeared kob, Kobus kob leucotis, in Sudan (Fryxell & Sinclair, 1988) and saiga, Saiga tatarica, in Kazackhstan (Bekenov et al., 1998)). The Serengeti-Mara Ecosystem (SME) covers km 2 on the border of Tanzania and Kenya (Fig. 1). The core area comprises the Serengeti ational Park (SP) and Masai Mara ational Reserve (MMR) where wildlife tourism is the only permitted land use. The core areas are surrounded by buffer zones: Maswa Game Reserve (MGR), Grumeti Game Reserve (GGR) and Ikorongo Game Reserve (IGR) to the southwest, gorongoro Conservation Area (CA) and Loliondo Game Controlled Area (LGCA) to the southeast and the Mara Ranches (MR) to the north. The Game Reserves allow tourism and hunting, the CA allows tourism, settlement, livestock and cultivation, the LGCA allows all of the above and hunting, whilst the MR allow tourism, settlement, livestock and cultivation but no hunting. Outside the buffer zones there are few restrictions on the use of resources. Hard edges where the core areas abut unprotected land occur in three places in Tanzania: the Western Corridor of SP, the Ikoma Open Area (IOA) between GGR and IGR and the northwest SP abutting

2 114 S. THIRGOOD ET AL. Grumeti Game Reserve The Serengeti-Mara Ecosystem Ikorongo Game Reserve Ikoma WMA Serengeti ational Park Maswa Game Reserve Masai Mara ational Reserve Kilometres Mara Inner Ranch Loliondo WMA Mara Outer Ranch Mara Inner Ranch Loliondo GCA gorongoro Conservation Area KEYA TAZAIA Fig. 1. Map of the Serengeti-Mara Ecosystem showing the location and status of the protected areas. GCA, game controlled area; WMA, wildlife management area. Kenya. Human density is high on the western boundary of the ecosystem and poaching is a management problem (Arcese et al., 1995). The SME is defined by the movements of the migratory wildebeest. The migration of 1.3 million wildebeest and 0.6 million zebra, Equus burchelli, and gazelle, Gazella thomsoni, is the best-known phenomena of the ecosystem (Grzimek & Grzimek, 1960; Talbot & Talbot, 1963; Watson, 1967; Pennycuick, 1975; Maddock, 1979; Sinclair, 1995). The wildebeest concentrate on the southern grasslands of the SP and CA during the wet season (December May). In May or June they move northwest and concentrate in the northern woodlands of the SP and MMR during the dry season (August ovember). The wildebeest return to the southern grasslands with the onset of rain in December. The underlying causes of the migration are not yet fully understood (Sinclair, 1995). The main factor influencing the northern migration is thought to be rainfall through its effect on food or salinity (Maddock, 1979; Wilmshurst et al., 1999; Wolanski et al., 1999). Current explanations for the southern migration include energy maximization and nutrient availability (Fryxell, 1995; Murray, 1995). Aerial surveys of the Serengeti wildebeest have been conducted since the early 1960s. Following the eradication of rinderpest, the migratory wildebeest population erupted from in 1961 to 1.3 million in 1977 (Sinclair, 1979). The population stabilized after 1977 and the estimate for 2000 was (± ) (Sinclair, 1995; TWCM, 2000). The migratory wildebeest are regulated by food availability in the dry season and although losses to predators and poachers are high this has little impact on population dynamics (Mduma et al., 1998, 1999). Smaller resident wildebeest populations are found in the Western Corridor, CA and LGCA and a separate, smaller, migratory population moves between the MMR in the dry season and the MR in the wet season. This latter population has declined from to in the last 20 years due to agricultural intensification in the MR (Ottichilo et al., 2001; Serneels & Lambin, 2001a,b,c). The SME faces anthropomorphic threats despite the large size of the protected area. Sinclair & Arcese (1995) estimated that half the ecosystem has been lost to agriculture and that a large proportion of the existing protected area has been modified by poaching. Comparison of the population trajectories of wildebeest in Tanzania and Kenya indicate that agricultural intensification is the main threat and has focused attention on the northern part of the ecosystem (Homewood et al., 2001). However, human density continues to increase around the protected area, with particular pressure in the west (Campbell & Hofer, 1995). The northern migration of the Serengeti wildebeest encompasses the western part of the ecosystem and is susceptible to land-use change (Maddock, 1979). There has been no systematic study of the migration since the 1970s and this period has coincided with the greatest change in land-use outside the protected area (Sinclair & Arcese, 1995). In this paper we assess the ranging patterns of migratory wildebeest in the SME in relation to the protected area. We use Global Positioning System (GPS) telemetry on eight wildebeest to quantify the time spent in different parts of the SME with different levels of protection. We also compare these data collected in to data collected in by VHF telemetry to assess whether the migration track has changed over the intervening 30 years. We discuss these results in the context of current efforts to conserve the Serengeti-Mara Ecosystem. METHODS Eight adult wildebeest (seven females and one male) were anaesthetized and fitted with GPS collars during February 1999 and May 2000 in the central SP. The anaesthetic (Etorphine) was delivered intramuscularly to the rump using a dart gun. The anaesthesia was reversed intramuscularly (Diprenorphine) after fitting the GPS collar. The wildebeest were visually monitored for 1 h after recovery and were located within a week using VHF telemetry. There were no apparent adverse effects of handling and all wildebeest resumed normal activity within 10 min of reversal. The GPS collars (Lotek, Canada) weighed 1.6 kg and consisted of a GPS unit and VHF beacon. Wildebeest were located at monthly intervals using the VHF beacon and multidirectional antennae on a Cessna 182 aircraft. Data were downloaded remotely from the GPS unit whilst airborne. The GPS units were programmed to provide fixes every 6 h. The

3 Conservation of migratory wildebeest 115 longevity of the GPS collars was variable with the longest two data sets consisting of 1000 fixes over 10 months and the shortest being 50 fixes over 13 days. The GPS data were projected into UTM coordinates in ArcInfo (v8.0) and then imported to ArcView (v3.2) for spatial analysis. The digitized boundary map of the SME was assembled from maps provided by the Frankfurt Zoological Society and the International Livestock Research Institute. To quantify the time spent in different parts of the SME we characterized fixes by protected area. We grouped these protected areas by the level of resource protection: well protected (SP, MMR, CA), less protected (MGR, GGR, IGR) and unprotected (LGCA, IOA, MR). We first calculated the proportion of each day spent in each protected area for each individual wildebeest. This removed any bias due to multiple fixes of the same individual on the same day. For each day, averaging across years and individuals, we then calculated what proportion of the day was spent where. At least one fix was available for each day of the year. The data were averaged first within each year and then across years, in order to reduce bias due to multiple fixes on the same date. The proportions for each day were then summed to produce the overall number of days per year that were spent within each protected area. As it is known that poaching is high on the boundaries of the protected areas we also investigated how much time was spent within 10 km of a less protected or unprotected area and within 10 km of a village. Data on ranging patterns of the migratory wildebeest during were available from VHF telemetry studies conducted by A. R. E. Sinclair and partially published by Inglis (1976). Up to 10 wildebeest were collared at any time and new individuals were collared as radios ceased functioning, typically after 9 months. Wildebeest were tracked using a Piper Super Cub aircraft with antennae attached to the wings. Aerial searches were conducted every 3 days for 3 h at a time. The detection range was about 15 km and most collared animals were found in the search time of a given day. When a signal was detected, the herd in which the collared wildebeest occurred was identified by triangulation and then confirmed by flying over the herd. When the aircraft was directly above a collared animal the signal disappeared for a brief interval and this interruption was used to mark the location fix. Fixes were marked on a 1: map accurate to 0.5 km. The co-ordinates were subsequently converted to UTM format in ArcInfo (v8.0) and then imported to ArcView (v3.2) for spatial analysis as above. Table 1. Time spent by wildebeest in each of the protected areas of the SME in relation to levels of resource protection Protection level Days per year Percentage of year Well-protected Serengeti Masai Mara gorongoro Less-protected Maswa 0 0 Grumeti 4 1 Ikorongo 5 1 Unprotected Loliondo 10 3 Mara ranches 11 3 Ikoma 7 2 SME, Serengeti-Mara Ecosystem. into the Western Corridor in May and June, moved north through GGR, IOA and IGR during June and July, reaching the northern SP and MMR during July and August. Most individuals remained in or adjacent to the MMR during August and September before returning south in October and ovember. Data for the southern migration were limited to two wildebeest who both moved south through IGR, IOA and GGR before returning to the CA in December. Time in protected areas The collared wildebeest spent 90% of the year during in the well-protected areas of SP, MMR and CA (Table 1). MGR was not used by any of the collared wildebeest in these 2 years. The less-protected GGR and IGR were traversed for a few days (2.5% of the year) during both northern and southern migrations. Particularly notable is the 7.5% of the year spent in the unprotected areas of the MR, LGCA and IOA. The importance of the IOA to the northern migration in June July is highlighted by the migration track of five wildebeest for which we have data in this period (Fig. 4). The IOA may also be important during the southern migration with two wildebeest traversing through or within 5 km of the area. The collared wildebeest spent 121 days (33% of the year) within 10 km of a less-protected or unprotected area. This time was primarily spent in the western SP and MMR. An average of 12 days (3% of the year) was spent within 10 km of a village along the western border of the SME. RESULTS Ranging patterns in All wildebeest fixes for are shown in Fig. 2 and the seasonal migration is shown in the series of maps in Fig. 3. The general pattern of movement indicates that the wildebeest were concentrated in the southern SP and CA during March and April, moved west and north Ranging patterns in The seasonal track of the collared wildebeest during is shown in the series of distribution maps (Fig. 5). The general pattern of movement is similar to that described for , however, there appear to be two important differences. First, the northern migrations in May July of

4 116 S. THIRGOOD ET AL Kilometres Fig. 2. Individual wildebeest location fixes in the Serengeti-Mara Ecosystem during as determined by Global Positioning System (GPS) telemetry. Individual wildebeest are identified by colour appear less concentrated in the western buffer zones of the SME and more widely dispersed through the central SP. Second, during the dry seasons of July October in the wildebeest were widely dispersed throughout the Western Corridor and north-west of the SP rather than concentrated in the MMR as at present. DISCUSSIO The key finding of this study was that the migration routes used by the Serengeti wildebeest were only partially protected despite the large size of the protected area. Threats to the wildebeest from agricultural intensification to the north of the MMR have been well-documented (Homewood et al., 2001; Serneels & Lambin, 2001a,b,c). Our study has identified additional threats to the migration in the unprotected IOA and the adjacent GGR and IGR. Furthermore, although our collared wildebeest spent only 10% of the year outside the well-protected SP, MMR and CA, 33% of the year was spent within 10 km of an unprotected or less-protected area. These areas were primarily in the west and north of the SME where poaching is a major concern (Arcese et al., 1995; Campbell & Hofer, 1995; Mduma, 1996). The migratory wildebeest population is currently stable at 1.3 million having recovered from a 40% decline due to drought in 1993 (TWCM, 2000). We can tentatively conclude that the existing protected area can maintain the migratory population with current poaching levels (Mduma et al., 1998, 1999). evertheless, the increase in human population density around the SME and the associated potential increase in poaching is a cause for concern. Comparative analyses demonstrate that there is a complex interaction between reserve size and the density of adjacent human populations in determining the persistence of large mammal populations in protected areas (Brashares et al., 2001; Harcourt et al., 2001; Parks & Harcourt, 2002). The wildebeest spent, on average, only 7 days in the unprotected IOA and 9 days in the less-protected GGR and IGR. However, these areas appeared to be critical to the migration. These wildebeest did not use the central SP during their migration, although other observations during

5 Conservation of migratory wildebeest 117 January February March April May June July August September October ovember December Kilometres Fig. 3. Seasonal track of the wildebeest migration in the Serengeti-Mara Ecosystem during as determined by Global Positioning System (GPS) telemetry. Individual wildebeest fixes have been combined together to provide a monthly pattern of movement showed that a proportion of the wildebeest population did migrate through the central woodlands (A. R. E. Sinclair, per. obs). Our observations of collared wildebeest indicate that a proportion of the population moves through the unprotected IOA to access their dry season range in MMR. While they are outside the protected area they are exposed to poachers (Mduma, 1996). Protection of the IOA is of imminent concern, along with increased protection of the GGR and IGR. A loss of part of the migration route, even if only used for short periods, could leave the population at risk. Poaching also occurs at high levels along the entire western boundary of the SP. Mduma et al. (1998, 1999) have shown that the poaching in the early 1990s removed up to wildebeest per year from the SME and was sustainable. However, poaching levels are directly related to human density around the west of the park and these populations are increasing at 5% p.a. (Campbell & Hofer, 1995).

6 118 S. THIRGOOD ET AL. (a) orthern migration (b) Southern migration Kilometres Fig. 4. Detailed map of the Ikoma Open Area and the adjacent Grumeti and Ikorongo Game Reserves showing the migration trajectory of individual wildebeest during (a) northern and (b) southern migration. Models suggest that the wildebeest population would be over-harvested if poaching were to double from the levels of the early 1990s (Mduma et al., 1998, 1999). The second area of concern for the Serengeti wildebeest is agricultural intensification in the MR to the north of the MMR. The Serengeti wildebeest use the MMR and the southern MR during the dry season. Here they meet the Mara migratory wildebeest who also use this area at this time. The expansion of mechanized agriculture in the northern MR during the last 20 years has reduced the wet season range of the Mara wildebeest, forcing this population to use the rangelands to the south (Serneels et al., 2001a). The Mara wildebeest have declined by 75%, from in 1977 to in 1997, with agricultural intensification identified as the driving force (Homewood et al., 2001; Serneels et al., 2001a,b,c). Although mechanized agriculture is currently restricted to the northern MR, further intensification would threaten the Serengeti wildebeest because the area available in the dry season determines their equilibrium population size (Mduma et al., 1998, 1999). How do the migration patterns observed during compare to those recorded during ? Comparison of the range maps of collared wildebeest suggest that there was a tendency for increased use of the western buffer zones during and a reduction in use of the central woodlands. However, caution needs to be attached to this suggestion, since the sample sizes of collared wildebeest were small in relation to population size during both the and studies. As noted above, other observations during indicated that some wildebeest did migrate through the central SP. Earlier accounts in the literature also indicate that the western buffer zones were used by wildebeest (Maddock, 1979; Sinclair, 1995) and this was the primary reason for the establishment of the Game Reserves. These earlier accounts also indicated considerable annual variation in the pattern of migration due to variation in rainfall and increases in the wildebeest population. The second notable difference between the two sets of telemetry data was the continued use of the Western Corridor and central SP during the dry seasons of and the apparent lack of use of the MMR in comparison to The 1971 and 1972 dry seasons were characterized by sporadic rainfall and many wildebeest remained in the SP (Maddock, 1979). However, this period also coincided with political instability in Uganda and the use of Kenyan airspace for telemetry was restricted (A. R. E. Sinclair, pers. obs.). Other differences between our observations and accounts of the migration in the literature (Maddock, 1979; Sinclair, 1995) include the limited use of MGR during the wet season and of LGCA during the northern migration. It is clear from the current study and other descriptions (Maddock, 1979; Sinclair, 1995) that the Serengeti wildebeest migration takes a proportion of the population outside the protected area. Earlier studies have also indicated that the wildebeest are subject to high levels of poaching (Arcese et al., 1995; Campbell & Hofer, 1995; Mduma, 1996), although this does not limit the population (Mduma et al., 1998, 1999). There is, however, little room for complacency, given human population growth rates in the west of the SME. How can the wildebeest be protected whilst ensuring that communities adjacent to parks benefit from the use of natural resources (Mbano et al., 1995)? The recent development of Wildlife Management Areas (WMAs) represents an important step for community-based conservation in Tanzania. WMAs have been designated in the IOA identified by our study as critical to the migration and in the LGCA. The WMAs will empower local communities to manage natural resources on village land including the harvesting of migratory and resident ungulates. The replacement of unregulated poaching with sustainable harvesting should effectively increase the size of the

7 Conservation of migratory wildebeest 119 January February March April May June July August September October ovember December Kilometres Fig. 5. Seasonal track of the wildebeest migration in the Serengeti-Mara Ecosystem during as determined by VHF telemetry. Individual wildebeest fixes have been combined together to provide a monthly pattern of movement. protected area to include the entire range of the Serengeti wildebeest within Tanzania. Whilst concerns remain about increases in human population density around the SME, the increase in protected area size should help to secure the future of one of the world s last intact ecosystems that supports large-scale wildlife migrations.

8 120 S. THIRGOOD ET AL. Acknowledgements We are grateful to Tanzania ational Parks for permission to conduct this research and the Frankfurt Zoological Society for funding it. Robin Reed kindly provided digital maps and Simon Mduma, Karen Laurenson, Craig Packer, John Reynolds and two referees improved the manuscript with their comments. REFERECES Arcese, P., Hando, J. & Campbell, K. (1995). Historical and present-day anti-poaching efforts in Serengeti. In Serengeti II: Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: Chicago University Press. Bekenov, A. B., Grackhev, I. A. & Milner-Gulland, E. J. (1998). The ecology and management of the saiga antelope in Kazakhstan. Mamm. Rev. 28: Brashares, J. S., Arcese, P. & Sam, M. K. (2001). Human demography and reserve size predict wildlife extinction in West Africa. Proc. Roy. Soc. Lond. Ser. B 268: Campbell, K. & Hofer, H. (1995). People and wildlife: spatial dynamics and zones of interaction. In Serengeti II: Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: Chicago University Press. Fryxell, J. M. (1995). Aggregation and migration by grazing ungulates in relation to resources and predators. In Serengeti II: Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: Chicago University Press. Fryxell, J. M. & Sinclair, A. R. E. (1988). Causes and consequence of migration by large herbivores. Trends Ecol. Evol. 3: Grzimek, M. & Grzimek, B. (1960). Census of plains animals in the Serengeti. J. Wildl. Mgmt. 24: Harcourt, A. H., Parks, S. A. & Woodroffe, R. (2001). Human density as an influence on species/area relationships: double jeopardy for small African reserves? Biodiv. Conserv. 10: Homewood, K., Lambin, E., Coast, E., Kariuki, A., Kivelia, J., Said, M., Serneels, S. & Thompson, M. (2001). Long-term changes in Serengeti-Mara wildebeest and land cover: pastoralism, population, or policies? Proc. atl. Acad. Sci. USA 98: Inglis, J. M. (1976). Wet season movements of individual wildebeests of the Serengeti migratory herd. E. Afr. Wildl. J. 14: Lombard, A. T. (1995). The problems with multi-species conservation: do hotspots, ideal reserves and existing reserves coincide. S. Afr. J. Zool. 30: Maddock, L. (1979). The migration and grazing succession. In Serengeti: Sinclair, A. R. E. & orton-griffiths, M. (Eds). Chicago: Chicago University Press. Mbano, B..., Malpas, R. C., Maige, M. K. E., Symonds, P. A. K. & Thompson, D. M. (1995). The Serengeti regional conservation strategy. In Serengeti II: Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: Chicago University Press. Mduma, S. A. R. (1996). Serengeti wildebeest population dynamics: regulation, limitation and implications for harvesting. PhDThesis: University of British Columbia. Mduma, S., Hilborn, R. & Sinclair, A. R. E. (1998). Limits to exploitation of Serengeti wildebeest and implications for its management. In Dynamics of Tropical Communities: ewbery, D. M., Brown,. & Prins, H. H. T. (Eds). London: Blackwell Science. Mduma, S. A. R., Sinclair, A. R. E. & Hilborn, R. (1999). Food regulates the Serengeti wildebeest: a 40-year record. J. Anim. Ecol. 68: Murray, M. G. (1995). Specific nutrient requirements and migration of wildebeest. In Serengeti II: Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: Chicago University Press. Ottichilo, W. K., de Leeuw, J. & Prins, H. H. T. (2001). Population trends of resident wildebeest and factors influencing them in the Masai Mara ecosystem, Kenya. Biol. Conserv. 97: Parks, S. A. & Harcourt, A. H. (2002). Reserve size, local human density, and mammalian extinctions in US protected areas. Conserv. Biol. 16: Pennycuick, L. (1975). Movements of the migratory wildebeest population in the Serengeti between 1960 and E. Afr. Wildl. J. 13: Serneels, S. & Lambin, E. (2001a). Impact of land-use change on the wildebeest migration in the northern part of the Serengeti-Mara ecosystem. J. Biogeogr. 28: Serneels, S. & Lambin, E. (2001b). Land cover changes around a major African wildlife reserve: the Mara Ecosystem (Kenya). Int. J. Remote Sensing 22: Serneels, S. & Lambin, E. (2001c). Proximate causes of land-use change in arok District, Kenya: a spatial statistical model. Agric. Ecosys. Environ. 85: Sinclair, A. R. E. (1979). The eruption of the ruminants. In Serengeti: Sinclair, A. R. E. & orton-griffiths, M. (Eds). Chicago: Chicago University Press. Sinclair, A. R. E. (1995). Serengeti past and present. In Serengeti II: Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: Chicago University Press. Sinclair, A. R. E. & Arcese, P. (1995). Serengeti in the context of worldwide conservation. In Serengeti II: Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: Chicago University Press. Talbot, L. M. & Talbot, M. H. (1963). The wildebeest in Western Masailand, East Africa. Wildl. Mongr. 12: TWCM (2000). Status and Trend of the Migratory Wildebeest in the Serengeti Ecosystem. Arusha: Tanzania Wildlife Conservation Monitoring/Frankfurt Zoological Society. Watson, R. M. (1967). The population ecology of the wildebeest in the Serengeti. PhD Thesis: University of Cambridge. Williamson, D. & Williamson, J. (1984). Botswana s fences and the depletion of Kalahari wildlife. Oryx 18: Wilmshurst, J. F., Fryxell, J. M., Farm, B. P., Sinclair, A. R. E. & Henschel, C. P. (1999). Spatial distribution of Serengeti wildebeest in relation to resources. Can. J. Zool. 77: Wolanski, E., Gereta, E., Borner, M. & Mduma, S. (1999). Water, migration and the Serengeti Ecosystem. Am. Sci. 87: Woodroffe, R. & Ginsberg, J. (1998). Edge effects and the extinction of populations inside protected areas. Science 280: