Ecological implications of food and predation risk for herbivores in the Serengeti Hopcraft, John Grant Charles

Transcription

1 University of Groningen Ecological implications of food and predation risk for herbivores in the Serengeti Hopcraft, John Grant Charles IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2010 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Hopcraft, J. G. C. (2010). Ecological implications of food and predation risk for herbivores in the Serengeti Groningen: s.n. Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date:

2 Chapter 1 Chapter 1 General Introduction. Author: J. Grant C. Hopcraft 3

3 SECTION I: Background and Theory. THE CENTRAL QUESTION IN A GLOBAL CONTEXT Unprecedented increases in human activity and land use change will undoubtedly have huge impacts on how ecosystems function in the future (Sala et al., 2000). One of the major concerns for conservation is that the increasing isolation of protected areas means natural processes such as population regulation and dispersal will become progressively more disconnected from regional trends. Species and communities in the future might find it difficult (if not impossible) to move along resource gradients to more suitable locations because these areas might already be modified by humans and therefore unavailable. For example, shifts in the vegetation community caused by changes in regional precipitation patterns (as predicted by global climate change) might necessitate a species to disperse up or down the rainfall gradient. However, if the species is confined to a delimited protected area and unable to disperse to better habitats, the population might decline. The relevance for conservation is that the location of our current protected areas may no longer host the preferred niches of many rare, endangered or economically important species in the future, particularly if the global climate changes significantly. Therefore understanding the factors that regulate herbivores and determine their distributions is critical to the long-term management and conservation of ecosystems. In sub-sahara Africa, the primary consequence of global climate change is a shift in precipitation patterns (Hulme, 2000). There still remains much debate as to which areas will receive more rainfall and which areas will receive less rainfall, however all the evidence unanimously agrees that there will be some change either way. For African savanna ecosystems an increase of annual rainfall would lead to a greater abundance of grass and woody vegetation (Sankaran et al., 2005). Because grass becomes more lignified under wetter conditions, more rainfall would also result in a decline of its nutritional quality. Furthermore, denser vegetation caused by increased rainfall could simultaneously facilitate predators to capture their prey because thick vegetation conceals stalking carnivores. Less rainfall would have the opposite effects on the quality and abundance of grass and would decrease predators ability to capture prey. Therefore, changes in precipitation will change how food and predation regulate grazers, and thereby alter the dynamics of African savanna ecosystems in the future. In this thesis I ask how grazers of different body sizes choose their habitat with respect to food abundance, food quality and predation risk, and how these factors might influence the migration patterns of wildebeest (Connochaetes taurinus) and zebra (Equus burchelli) in the Serengeti ecosystem. The analysis of long-term aerial wildlife counts from the Serengeti, data from GPS radio collars, and ground and satellite measurements of vegetation and soils are presented in the following chapters of this thesis. The results illustrate how forage quality, forage abundance and predation risk affect the choices grazers make about the habitats they occupy and how they move between these habitats. Specifically, the areas in which different species occur and the choices they make depend on both the size of the species and their digestive physiology, because body size and physiology simultaneously influence how grazers extract energy from plant material, and which predators are capable of killing them. STUDY AREA 4 The Greater Serengeti-Mara Ecosystem The Greater Serengeti-Mara Ecosystem is approximately 25,000 km2 situated in northeastern Tanzania and crossing into southeastern Kenya (34º to 36º E and 1º to 3º 30 S) (Figure 1). From a biological perspective, the Greater Serengeti-Mara represents a complete and functioning ecosystem defined by the extent of the wildebeest migration. The elevation of the Greater Serengeti-Mara Ecosystem ranges between 1100m on the shores of Lake Victoria to 3000m at the top of the Ngorongoro highlands, with temperatures

4 General Introduction. between 15º to 35ºC. There are two rainy seasons; the long rains generally occur from late February until mid May and the short rains are typically from November to December (Figure 2). Rainfall typically varies more during the wet season resulting in these two rainfall periods occasionally fusing into one long rainy season or failing altogether (Norton-Griffiths et al. 1975, Sinclair 1995b). The average yearly rainfall for the southeastern section of the park is roughly 500mm and increases to the west and north. The northeast section of the ecosystem has the highest annual rainfall (1200mm) with the lowest monthly variation (Figure 3) (Norton-Griffiths et al.1975). Chapter 1 Figure 1. Broad vegetation types of the Greater Serengeti-Mara Ecosystem with surrounding land use types (Vegetation modified from (Reed et al., 2008): Land Use from UNEP). 5

5 SECTION I: Background and Theory. Figure 2. The average monthly rainfall (mm) for the Serengeti-Mara ecosystem from 1990 to The southern extent of the park is predominantly open grassland interspersed with large rocky outcrops called kopjes. The soils on the southern plains are pyroclastic and originate from ash plumes expelled from the volcanoes to the east of the ecosystem along the rift valley. A shallow calcareous hard-pan just below the surface of the soil on the plains impedes the growth of any deep-rooted vegetation such as trees. The sandy and loamy soils in the north originate from granitic parent material and form deep nutrient poor swales along drainage lines and with well drained ridge tops. There are moderately dense Acacia and Commiphora woodlands with occasional grass openings (termed mbugas ) in the north (Figure 1). The riparian areas where water accumulates generally have the most herbaceous and woody vegetation. The west is primarily composed on alluvial soil carried by two major rivers (the Grumeti and the Ambalageti rivers) which are separated by a line of hills made of Precambrian banded ironstone (Sinclair, 1995). The management and conservation of the Serengeti ecosystem falls under several different jurisdictions and varies in its levels of protection (Thirgood et al., 2008). The core of the ecosystem is fully protected (Serengeti National Park in Tanzania and Masai Mara Reserve in Kenya) while the perimeters of the ecosystems are multiple use areas each with some degree of allowable resource extraction (Figure 1). To the south, west, and the northeast the Game Reserves are managed as trophy hunting concessions operated by private companies under the jurisdiction of the Tanzanian Wildlife Division. On the north and east side of the core protected area the Mara Group Ranches and the Ngorongoro Conservation Area are managed as communal pasture land reserved for Maasai pastoralists, several of which run community based conservation schemes. The remaining areas beyond the buffer zones are open agricultural lands where few wild animals occur. These open agricultural areas are the primary sources of illegal resource extraction (poaching), however the degree of illegal activity varies across the landscape depending on accessibility by local communities and the levels of law enforcement (Hilborn et al., 2006; Hofer et al., 2000; Loibooki et al., 2002). 6

6 General Introduction. Chapter 1 Figure 3. Average rainfall isohyets for the Greater Serengeti-Mara Ecosystem. The south-east section of the ecosystem is driest and the north-west is consistently the wettest with the lowest variation between months. OVERVIEW OF MIGRATIONS The Serengeti ecosystem is best known for the large scale migrations of wildebeest and zebra. Therefore, it is necessary to provide the reader with some background information about migrations so as to set the stage for the rest of the thesis. Defining migrations versus dispersal Migrations are the synchronous large-scale return movement of a population of animals over large distances where as dispersal is the movement of animals between locations whose direction and periodicity is not necessarily recurrent. There are several attributes that are consistent in all migratory systems regardless of the distance traveled or life cycle of the animal (Baker, 1978). (1) The resource must be sufficiently attractive to warrant moving large distances. Either the food must be exceptionally nutritious, or the chances of survival of progeny must be significantly elevated. (2) There is always a returning movement which is usually initiated by the decline of a periodic resource. (3) There are no physical barriers blocking the migratory path (Harris et al., 2009) and that (4) individuals must have sufficient physiological capacity to move between distant resource patches. This includes the perception and comprehension of navigational cues (such as olfactory, geomagnetic, polarized light, or 7

7 SECTION I: Background and Theory. 8 celestial cues) as well as adequate energy reserves to carry the animal through lean periods. (5) The resource should cycle through periods of plenty and paucity (usually seasonally), or alternatively the animal only requires the resource at certain times of its lifecycle. (6) The cycle of available resources must be predictable, often occurring along abiotic gradients such as temperature or rainfall over altitudinal or latitudinal gradients. If cycles become unpredictable the animal risks expending large amounts of energy for no returns and will either suffer or die as a consequence. (7) In some scenarios, animals tend to migrate as a function of density dependence, moving larger distances as populations increase (Fryxell, Greever & Sinclair, 1988a). (8) There is consistent direction of movement during migrations by virtually the entire population (that is, animals do not simply disperse). The direction is determined by the availability of resources and, therefore, movement is often groupwise. Migrations: Maximizing rewards and minimizing costs The simultaneous movement of thousands of animals is stunning and yet ironically paradoxical. Why should animals travel such large distances when moving can be both energetically costly and potentially very risky? By traveling hundreds of kilometers, animals potentially have no prior knowledge as to the quality of the new foraging grounds they will arrive at, and therefore expose themselves to the risks of energy and nutrient deficits. In theory an animal should only move if the new area is better than the current location, however traveling over exceptionally large distances precludes the knowledge of future conditions. Species that maximize their reproductive fitness while minimizing their energy expenditure are expected to be the most successful, therefore we might expect a resident strategy to be more advantageous since it minimizes the risk of uncertain food supply and predation. However, contrary to this expectation, migratory behavior is a convergent trait expressed in many unrelated taxa, such as wildebeest and zebra, which implies that migration is evolutionarily advantageous behavior. A migratory strategy (as opposed to remaining resident) is expected to prevail in populations where the costs associated with migrating in terms of energy expenditure and individual risk do not out-weigh the benefits gained by accessing new resources (Houston, McNamara & Hutchinson, 1993). The movement of animals seeking to maximize individual foraging rewards has been viewed as the common driver of migrations (Alerstam, Hedenstrom & Akesson, 2003; Fryxell & Sinclair, 1988b; Holdo, Holt & Fryxell, 2009). Migrations as a means of avoiding risk have been described in only a few systems (Frair et al., 2005; Gilg & Yoccoz, 2010; Lauridsen et al., 1999; Neill, 1990). How animals move in relation to both resources and risk is the central theme in this thesis, and therefore makes the work novel. Common attributes of African migratory systems: Seasonal resource gradients Africa is an ancient continent weathered by billions of years of erosion that has leached the soils of minerals which are more common on other continents (Reader, 1999). Acquiring sufficient nutrients throughout the year when seasonal deficits are frequent is a problem faced by many African herbivores. Historically, migrations were common across Africa which still hosts 6 of the 9 remaining herbivore migrations worldwide (Harris et al., 2009). The distribution of current and historic African herbivore migrations overlaps closely with relatively steep rainfall gradients (Figure 4). These environmental gradients provide a key insight into the causes of herbivore migrations, namely predictable seasonal gradients of resources. By sequentially moving between the best available food patches, migrants are in essence increasing the total available biomass of highest quality food without exhausting the overall supply (Fryxell et al., 1988b; Hopcraft, Olff & Sinclair, 2010). Therefore, migrations across local resource gradients enable a population to escape the local limitation of low forage biomass or food quality.

8 General Introduction. Chapter 1 Figure 4. The distribution of current and historic herbivore migrations in Africa closely overlaps with strong local rainfall gradients (migration data from (Harris et al., 2009), J. Ogutu pers. comm., and R. Fynn pers. comm.). 9

9 SECTION I: Background and Theory. FOCAL SPECIES This thesis focuses primarily on two migratory species in the Serengeti ecosystem, wildebeest (Connochaetes taurinus) and zebra (Equus burchelli). Both species have converged onto essentially sympatric annual migration routes, while originating from different taxonomic lineages (bovids versus equids) and with different resource requirements. In this section I review some basic biology for wildebeest and zebra as an introduction to more detailed comparisons in the following chapters. The comparison of distribution and movement patterns between sympatric species is a central tenant in this thesis, because it provides insights into the role of resources and risk in shaping herbivore assemblages. Wildebeest Wildebeest are a member of the Alcelaphinae subfamily (family Bovidae) which is at least 2 million years old (Marean & Gifford-Gonzalez, 1991; Peters et al., 2008; Vrba, 1979). Evidence suggests that wildebeest have been part of the Serengeti ecosystem at least since the last glaciation about 12,000 years ago and probably as far back as one million years ago (Georgiadis, 1995; Marean et al., 1991), Wildebeest are obligate grazers whose diet is composed of 90% C4 grasses (Casebeer & Koss, 1970; Cerling et al., 1997; Codron et al., 2007; Tieszen & Imbamba, 1980). The intake rate of wildebeest changes seasonally with the quality of grass. Wildebeest tend to consume more when the grass quality is poor and relatively scarce during the dry season, suggesting their daily nutrient requirements are determined by the quality of the food rather than by the quantity. Stomach contents from a random group of sampled animals by Watson (1967) were significantly different between seasons; 11.9% of total body mass during the wet season as opposed to 15.9% during the dry season. Regular access to drinking water is critical for all alcelaphines particularly if the forage is dry (Ambrose & Deniro, 1986). In general, most large herbivores (except buffalo) drink at locations where there is good visibility at the water s edge with little vegetative cover and gentle banks, and often where a river is crossable (Jarman & Mmari, 1971), which is often promoted by hippo (Kanga, in preparation). Experiments with captive animals suggest that wildebeest are able to detect and reject water with concentrations of as little as 0.625% calcium carbonate (Watson 1967). The salinity of surface water in the Serengeti increases with the progression of the dry season due to evaporation and this imposes serious restrictions on access to drinking water for all grazers (Gereta & Wolanski, 1998). Female wildebeest are either pregnant or lactating (or both) for the entire year (see Chapter 2). Reproductively active females are the most nutritionally constrained members of the population and seldom have more than 50 grams of kidney fat. In contrast, non-reproducing females can exceed 250 grams of kidney fat (see Chapter 2). Furthermore, bone marrow fat (which is the final fat reserve in the body, and mobilized only when other reserves are exhausted) is rarely used by non-reproducing females (Sinclair, 1977a). Females with calves seasonally deplete these critical bone marrow reserves to 15% of their potential (see Chapter 2), which illustrates the extreme nutritional demands associated with reproduction. This thesis focuses on the movement of female wildebeest because they are the most energetically challenged part of the population and therefore the most revealing to study in terms of migration of the species. The rut occurs during the first full moon in late May or early June (Sinclair, 1977b) as the wildebeest are moving off the plains. During the rut bulls fracture the randomly mixed herds into smaller regular pods. A bull has an average harem size of 28.4 (SE 19.8) females. There appears to be an optimal number of females per harem above which a bull is unable to defend from rival males and below which females become too anxious of being separated from 10

10 General Introduction. the main herd (generally not less than 16) (Watson 1967). Calves are only 5 months old at the time of the rut and are often accidentally separated from their mothers by persistent bulls. Reproductively active males advertise with continuous bellows and grunts, actively defending their moving harem of females from neighboring competitors. Aggression is directed at usurpers, otherwise bulls display their prowess with high headed struts, horning vegetation, and rubbing their pre-orbital glands in exposed soil (Estes & Wilson, 1992; Estes, Raghunathan & Van Vleck, 2008). Copulation is preceded with stiff-leg approaches by the male and a flehmen response in which the male assesses the females reproductive status by examining her urine, however these rituals are ignored during the peak of the rut. Copulation is unceremonious and typically lasts only a few seconds (Estes et al., 1992). The gestation period for wildebeest is 240 days (approximately 8 months) and occurs during the dry season (Clay et al., ; Estes et al., 1992). Generally the last month of gestation and the first 2 months of lactation are the most nutritionally stressful for cows (Sinclair, 1977a; Sinclair & Arcese, 1995) because the calves are growing quickly (Figure 5). Therefore, early green flushes and good grazing in December and January are critical. However, the length of gestation has been shown to be somewhat flexible in many ruminants depending on age and nutritional status (Berger, 1992). Gestation and the onset of calving can be delayed for up to a week in the case of extreme droughts or floods, which often results in very poor calf survival (Estes, 1976; Talbot & Talbot, 1963). Wildebeest give birth in early to mid February usually on the short grass plains, if the rains permit. Female wildebeest often appear completely unperturbed during labor and will continue grazing in a large herd even with the forelegs of the calf exposed. Approximately 84% of wildebeest calves are born between 6am and 12 noon, with fewer in the afternoons, and virtually none overnight (Estes & Estes, 1979). If the female is disturbed during labor (as by a predator) she can voluntarily delay parturition for up to an hour. However, once contractions Chapter 1 Figure 5. The fetus growth rate of wildebeest. Wildebeest gestation is approximately 8 months (240 days ± 11.7 days) (Clay et al.) and at time of partum the calves weigh between 15.6 to 16.6 kg (data from (Watson 1967)). The final stages of pregnancy at the end of the dry season and early stages of lactation are the most demanding for female wildebeest. Dark shading represents wet season months, light shading represents dry season months. 11

11 SECTION I: Background and Theory. 12 begin females generally deliver quite rapidly, often lying on their sides in an alert head-up posture (Estes et al., 1979). Once the calf is born it is quickly licked dry and watched intensely as it attempts to stand and nurse. Neonates often attract the attention of other females in the vicinity which builds the security of the crèche. Mothers do not assist their calves to stand. The mother and new born calf generally move a short distance away from the birthing site, presumably to reduce the smells that might attract predators. Expulsion of the placenta is often delayed by 3 hours post-partum after which time the calf is fully mobile and can escape most predators scavenging the afterbirth. Wildebeest calves are precocial at birth, probably as a means of avoiding predation as well as a critical adaptation to a migratory lifestyle in herds. Calves range in mass between 15.6 to 16.6 kg (Figure 5) with no differences between sexes (Watson 1967). The average weight of calves born during favorable years with good rainfall is larger than those born during droughts. In extremely poor years, aborted fetuses are more common. Calves generally loose weight in the first 5 days after birth (approximately 1kg), after which they grow rapidly. By May when calves are 4 months old and the migrants are moving off the plains the calves are approximately 60 kg. Lactation stops within 2 weeks of losing a calf (Watson 1967). Male wildebeest attain a larger body size than females (160 kg versus 120 kg, respectively) (Watson 1967). Both sexes reach their peak weight between 4 to 6 years, after which they tend to decline slowly soon after sexual maturity. Figure 6 illustrates the body growth of migratory wildebeest in Serengeti. Female wildebeest reach sexually maturity by three years old. Post-mortem data collected by Watson showed 37% of cows aged 2 or below were pregnant, while 94.5% of cows aged 3 and over were pregnant or showed signs of recent pregnancy. Females do not reach their adult weight before 3 (Figure 6), which suggests they are not completely sexually mature before this age. There is no indication of female menopause in wildebeest, although reproductive success tends to decline in older animals (Watson 1967). Male wildebeest have live spermatozoa in the tubules starting from 2 years old. However, they generally do not become sexually active until their fourth and fifth year once they are cable of competing in the rut (Watson 1967). Bulls are capable of mating year round and experience no clear seasonal physiological limitations. Furthermore, there is no evidence that wildebeest bulls stop breeding beyond a certain age, however the intense competition of the rut might behaviorally exclude older weaker males from mating. From histological examination of wildebeest testes, Watson found 23% (n=44) of males had degenerating seminiferous tubules or tubular abscesses leading to orchitis and sterility (Watson 1967). This remarkably high sterility rate did not seem to hamper recruitment in the population which was growing very fast during this time, however it does open the question as to the causes (Watson proposed brucellosis or pasteurella may be agents). The average life span for adult wildebeest is about 8 years for males and approximately 7 years for females. Both male and female wildebeest rarely live beyond 17 years old (Watson 1967). The sex ratio of adults over 2 years old is also not different from 1:1 (Mduma, Sinclair & Hilborn, 1999), however resident herds might be female biased (Ndibalema, 2009). Survivorship improves beyond the age of 2 (see Chapter 2). For instance, males younger than 2 have approximately 27% mortality where as males between the ages of 2 to 8 have 12% mortality. There is a similar trend for females but not as drastic; 18% mortality before the age of 2, and 14% from the ages of 2 to 7. The greater mortality and shorter average life span for female wildebeest is presumably a consequence of resource investment in offspring (Sinclair et al., 1995). Lion and hyena are the main predators of grazers in Serengeti. On average a lion consumes about 32 large prey animals per year while a hyena consumes approximately 10

12 General Introduction. a Chapter 1 Weight (kg) y = 162.3(1 e 0.142x ) b Age (years) Weight (kg) y = 121.7(1 e x ) Age (years) Figure 6. The body growth rate of (a) male and (b) female wildebeest (weights are without the contents of reticulum, rumen and blood; data from (Watson 1967)). animals per year of which 20% to 40% are wildebeest (Schaller, 1972). Currently there are approximately 2500 lion and 7000 hyena in the ecosystem which sums to between 16,000 and 32,000 wildebeest consumed by lions annually, and 14,000 to 28,000 are consumed by hyena. Together lions and hyena account for 30,000 to 60,000 wildebeest kills per year or 2.5% to 5% of the population, which therefore accounts for a quarter to half the annual mortality since the maximum instantaneous rate of increase for wildebeest is 10% per annum. Periodic calamities are common especially while crossing rivers or lakes but have very little consequence for the entire population. Victims tend to be calves and weaker individuals. For instance, from 101 carcasses inspected on 2 occasions when wildebeest were killed en 13

13 SECTION I: Background and Theory. 14 masse while crossing rivers, 78 were calves and 23 were adults (Watson 1967). Large drowning events - such as the 2007 crossing of the Mara River when up to 10,000 wildebeest drowned in 3 days (Brian Heath, pers. comm.) - tend to occur where the river banks are steep and incised or where animals get stuck in thick mud such as on the shores of Lake Lagarja. Estimates of poaching pressure from Mduma (1999) indicate that approximately 40,000 wildebeest / year are being illegally harvested from the Serengeti (~3.5% of the population). Population models suggest that an increase to 80,000 wildebeest / year (~7%) would lead to a population collapse, whereas no harvest could lead to a population recovery to approximately 1.6 million (Mduma et al., 1999). Rinderpest was eradicated from Serengeti wildebeest populations by 1963 as a result of a large vaccination campaign between 1957 and 1959 combined with a concerted effort to reduce livestock-wildlife interactions. Cattle were removed from the Moru Kopjes area by 1959 (Watson 1967) and serological evidence indicates the disease was exterminated in wildebeest, buffalo, eland, impala and warthog within 4 years (Plowright & McCulloch, 1967; Sinclair, 1977a; Taylor & Watson 1967). This dispelled the prevailing theory at the time that wildlife were the reservoir for the disease and were responsible for its prevalence in livestock. Rinderpest immunity is passed from wildebeest mothers to newborns, however immunity disappears beyond 4 months old which means the adult population could be re-infected if an outbreak were to reoccur. Rinderpest epidemics continue to occur in wild ruminants across East Africa but only as a consequence of direct interaction with infected livestock (Kock et al., 1999; Rossiter et al., 1987). In the Serengeti ecosystem, livestock and wildebeest still overlap in the Ngorongoro Conservation Area and in the group ranches of Kenya (during the wet and dry season, respectively) and is the most likely route for any re-infection. The current policy of mandatory vaccination for livestock reduces the chance of rinderpest reoccurring in the Serengeti, although rinderpest still erupts occasionally in livestock around the ecosystem (Cleaveland et al., 2008; Dobson, 1995; Rossiter et al., 1987; Taylor et al., 2002). Malignant catarrhal fever (MCF) is transmitted both via the after-birth and calves under the age of 3 months (Cleaveland et al., 2008; Plowright, Ferris & Scott, 1960). MCF has no significant effect on wildebeest survival (Estes et al., 1979), however it can be lethal for cattle and is a source of significant economic loss for local livestock owners (Bedelian, Nkedianye & Herrero, 2007; Cleaveland et al., 2008). Currently, pastoralists keep their livestock separated from the wildebeest during the calving period because of transmission of MCF which also reduces the chances of cross-species transmission of many other diseases. Recent advances suggest promising results for a nasally administered MCF vaccine for cattle which is currently being trialed (S. Cleaveland, pers. comm.) and which might change the disease dynamics between wildebeest and livestock. Brucellosis and trypanosomiasis have also been detected in wildebeest populations. Watson (1967) reports that approximately 11% of the migrant population and 28% of the Kirawira resident population tested positive for brucellosis; an estimate supported also by Sachs (Sachs & Staak, 1966). Similarly, wildebeest are hosts to Trypanosoma brucei which is transmitted by the tsetse fly. An intensive trypanosomiasis survey in 1971 indicated that as many as 12% of the wildebeest population were exposed (Bertram, 1973), however wildebeest appear to be asymptomatic. Similarly, localized anthrax epidemics occur periodically in Serengeti (Hampson et al., submitted; Lembo et al., submitted) however they do not appear to influence the wildebeest population. Zebra Zebra belong to the order Persissodactyla (the odd-toed ungulates) which includes the tapirs and rhinoceros. Africa is home to 4 of the 7 remaining wild equid species plus one

14 General Introduction. recently extinct equid from southern Africa called the quagga (Equus quagga). The plains zebra (E. burchelli) is the only extant equid in the Serengeti ecosystem, although fossil evidence suggests several equid progentors may have occurred in the ecosystem during the Pleistocene (5Ma) (Marean et al., 1991; Peters et al., 2008) and perhaps previous to this in the upper Miocene (8Ma) (equids co-evolved in Africa with the expansion of C4 grasses during this time) (Janis, Gordon & Illius, 1994; Segalen, Lee-Thorp & Cerling, 2007). The plains zebra is a phylogenetically complex group and contains several races distinguished by variations of their stripe pattern, and are distributed from Ethiopia through to southern Africa (Nowak, 1999). Genetic evidence suggests that plains zebra should be reconsidered as a race of quagga which should also include the extinct giant Cape zebra, E. capensis (Orlando et al., 2009). The Grevy s zebra (E. grevyi) occur in the arid ranges of northern Kenya, Sudan and Ethiopia, and the mountain zebra (E. zebra) occur only in southern Africa (Nowak, 1999). There are both migrant and resident populations of zebra found in the Serengeti. There is little is known about the three resident populations, other than they occur in the Western Corridor, Loliondo and the Masai Mara (Gogan, 1973; Skoog, 1969). Zebra migrations in Serengeti have been noted from the earliest records (Darling, 1960; Grzimek & Grzimek, 1960; Watson 1967). It has been suggested that zebra migrate with the wildebeest to reduce their exposure to predators (Sinclair, 1985), in addition to meeting their nutritional requirements. The zebra population in Serengeti has remained remarkably stable since the 1960 s despite large changes in the abundance of other grazer species (approximately 200,000). Previous authors suggest that zebra in the Serengeti are limited by predation especially on juvenile age classes (Grange et al., 2004; Senzota, 1988; Sinclair & Norton Griffiths, 1982), and not by food availability. As much as 59% to 74% of the mortality in the zebra population is due to predation (Sinclair et al., 1982), as compared to less than 25% for wildebeest (Mduma et al., 1999). Data from a long-term lion studies in the Serengeti indicate that zebra compose about 18% of kills where as the expected based on their abundance in the ecosystem is about 8% (by comparison, wildebeest compose 25% of lion kills but based on their abundance the expected is about 60%) (Hopcraft, 2002). Zebra also reduce the amount of time they spend near waterholes which are often associated with predation (Hopcraft, Sinclair & Packer, 2005; Valeix et al., 2009a), as well as selecting more open habitats when predators are present (Valeix et al., 2009b). Furthermore, taller grass associated with more rainfall also leads to greater seasonal mortality in zebra (Owen-Smith, 2008). All of this evidence suggests that predation has an unusually large impact on the zebra population. The role of density dependent interactions in regulating zebra abundance has only been shown in Laikipia (northern Kenya) as a function of rainfall (Georgiadis, Hack & Turpin, 2003). It is likely that droughts have a greater impact on the survival of sub-adult zebra than on adults (Ogutu et al., 2008; Owen-Smith, Mason & Ogutu, 2005) because adults tend to have more fat reserves to cover them through periods of food scarcity. Zebra diet is 90% grass with some browse (Codron et al., 2007; Tieszen et al., 1980). There is some evidence that zebra select patches with the highest grass nitrogen and phosphorous content (Ben-Shahar & Coe, 1992), however they tend to be less selective for specific grass species than either hartebeest or wildebeest (Casebeer et al., 1970). Digestion takes place in an enlarged caecum and colon in the hind-gut, which is more efficient at digesting poor quality food but less efficient at digesting moderate quality food (Duncan, 1992; Duncan et al., 1990; Foose, 1982; Maloiy & Clemens, 1991) (Figure 7). Zebra compensate for the poor digestive efficiency by eating more grass biomass and processing it faster than wildebeest (Bell, 1971; Bell, 1970; Gwynne & Bell, 1968; Illius & Gordon, 1992). Zebra will continue grazing for up to 15 hours per day and will often graze into the night (Gogan, 1973) which might expose them to greater predation risk (Duncan et al., 1990). Therefore, zebra have many of the characteristics of a ruminant that is disproportionately larger than themselves (Bell, 1971), but without the added benefit of out growing predators to which they are very susceptible. 15 Chapter 1

15 SECTION I: Background and Theory. The annual rates of reproduction are much slower in zebra than wildebeest. Although females are polyestrous and may begin their next estrus cycle within 10 days post-partum, the gestation is between 360 to 396 days (Klingel, 1975). Females give birth to a single precocial foal weighing about 32 kg that is capable of running within an hour. Foals are generally weaned by 11 months old, which means the inter-birth interval for zebra is between 1 and 3 years. The long gestation period and inter-birth interval means that zebra cannot synchronize their reproduction with the seasonal cycles of food supply, whereas by comparison wildebeest gestation is only 8 months and their annual reproduction coincides with the peak amount of food (Sinclair, Mduma & Arcese, 2000). As a result, zebra foals are born throughout the entire year with a slight increase in December and January, which means they can not swamp predators with a single cohort of thousands of foals (Sinclair et al., 2000). Zebra foals probably suffer very high predation rates (Mills & Shenk, 1992), although this rate has not yet been quantified. Young zebra stallions build a harem of females once they are about 5 years old and able to compete with dominant stallions. Females remain in their father s harem until their first estrus (about 2 years old) at which point competing stallions and eligible bachelors challenge the father until one proves strong enough to successfully abduct the young mare from her father s harem (Estes et al., 1992). Breeding mares will remain with the same stallion for their lifetime, unless the stallion is killed or seriously injured in which case they will switch. Harems usually consist of 1 to 6 breeding mares with their offspring and have a rank hierarchy. Stallions actively protect their harem and offspring from predators and other threats. Male foals leave the protection of the harem after 2 years to join bachelor groups, where they often form strong social bonds with other males for many years (Estes et al., 1992; Nowak, 1999). WILDEBEEST AND ZEBRA MIGRATIONS IN THE SERENGETI ECOSYSTEM The wildebeest and zebra migration in the Serengeti ecosystem is one of the last large scale terrestrial mammal migrations in the world (Harris et al., 2009). The presence of wildebeest in the ecosystem dates at least to the time of the last glaciation period (12,000 years ago) (Georgiadis, 1995; Marean et al., 1991), and probably as far back as a million years ago, however the deep history of their migrations in unknown. Previous to this, the evidence from the fossil record and the unique geomorphology of the area suggest that wildebeest progenitors such as Megalotragus kattwinkeli and other related smaller alcelaphines probably migrated intermittently in the Serengeti ecosystem up to 2 million years before present (Marean et al., 1991; Peters et al., 2008; Vrba, 1979). However, drier and wetter phases across Africa during this time drastically changed the Serengeti vegetation from savannas to thick low-land Congo forests and back again which caused the extinction of several species that were once plentiful. Today, there are approximately 1.2 million wildebeest and 250,000 zebra (Conservation Information Monitoring Unit, Tanzania Wildlife Research Institute) that move in a distinctive clock-wise fashion around the ecosystem; a direct distance of about 650km per year. 16 Gradients of food quality, food abundance and predation in Serengeti The Serengeti has a distinct southeast-to-northwest rainfall gradient which is matched by an equally clear soil fertility gradient but in the opposite direction (Figure 8a and 8b). The lowest rainfall occurs on the south-western plains (approximately 450mm / annum) where the volcanically derived soils are most fertile. The north has the greatest amount of rain (approximately 1200mm / annum) but the older granite derived soils tend to be heavily leached and infertile (de Wit, 1978; Epp, 1978; Gerresheim, 1974; Jager, 1982; Norton-Griffiths, Herlocker & Pennycuick, 1975). The counter-gradient of rainfall and soil fertility in Serengeti influences the abundance and digestive quality of grasses which are available to herbivores (Figure 8c). Under high rainfall conditions plants invest more energy in structural support and protection against herbivory (e.g. stems, lignified tissues, and secondary compounds) (McNaughton et al., 1985).

16 General Introduction. a Effectiveness in extracting nutrients R um inant H indgut fermenter Chapter 1 Low fibrosity High quality High fibrosity Low quality F av o rs hindgut ferm enters Favors ruminants Favors hindgut ferm enters b Favors Hindgut Fermenters Quantity & Hindgut Both Ruminants Fermenters Favors Rum inants Low fibrosity High fibrosity Figure 7. The extended fermentation time in ruminants enables them (a) to extract more energy from vegetation in general, except when the grass quality is very poor or when it is very rich. Hind-gut fermenters, such as zebra, can extract sufficient energy from very low quality grasses because their relatively short through-put times allows them to process greater quantities of forage (Foose, 1982). As a result, (b) ruminants excel when high quality food is present in low quantities (otherwise they bloat), while equids are more competitive when low quality forage is abundant or when it is very fibrous. Furthermore, there is a compositional shift in wetter areas towards taller grass species with higher carbon to nitrogen ratios, which lowers the concentration of nitrogen and phosphorous available to herbivores (Anderson et al., 2007). Therefore, the digestibility of forage from a herbivore s perspective is inversely related to rainfall. Wet areas tend to have high grass biomass but low grass quality, while dry areas have low biomass but higher quality (Botkin, Mellilo & Wu, 1981; Fritz & Duncan, 1994; Olff, Ritchie & Prins, 2002). The concentration of sodium, nitrogen, phosphorous, and calcium in the grasses is also partially dependent on the 17

17 SECTION I: Background and Theory. properties of the soil (Anderson et al., 2007; Kreulen, 1975). Therefore the gradient of grass quality (Figure 8c) in the Serengeti can be partially explained by rainfall and soil properties (Figure 8a and 8b). The grasses on the southern plains of the Serengeti are more nutritious than those in the north or the west. However the soils on the plains are also shallow due to the presence of a calcareous hardpan, and the grasses quickly senesce and dry after a few days without rain (McNaughton, 1985). Therefore, these high quality grasses on the short-grass plains are only available for a brief period of time during the wet season. The western and northern parts of the ecosystem have deeper soils and higher rainfall, and retain green growth well into the dry season, but are lower in quality. In essence, migrants move seasonally up and down this grass quality gradient to maximize their intake of the highest quality grasses. The short vegetation on the southern plains also reduces the cover available that conceals stalking predators, which makes it a relatively safe area. The gradient of predation risk might also partially explain the seasonal movement of many herbivores in Serengeti. a b c Figure 8 The (a) mean yearly rainfall and (b) soil fertility (soil nitrogen) in Serengeti tend occur as a counter-gradient, which results in a distinctive (c) grass quality gradient (regression kriged data from 148 field measurements of soil and grass nitrogen). Migrants follow the seasonal access to nutrient rich grasses of the southeastern plains during the wet season. Proposed drivers of the Serengeti migration There is little doubt that the dynamics of the Serengeti ecosystem are rainfall dependent. Therefore, rainfall is the premise that underlies the following explanations of drivers of the migration since rainfall is intrinsically tied to everything else (Holdo et al., 2009). Evolutionary modeling of the wildebeest movement in relation to grass greenness and new growth (both of which respond to rainfall) shows that the north-south movement of the wildebeest can be accurately predicted, but the east-west movement is less predictable by rainfall patterns (Figure 9) (Boone, Thirgood & Hopcraft, 2006). This suggests that although rainfall may be an important driver of wildebeest and zebra migrations there may be additional variables particularly between the east and west parts of the ecosystem. Grass protein (carbon to nitrogen ratios) The total wet season biomass production between grazed patches in the northern long grass plains and the southern short grass plains does not differ (Augustine & McNaughton, 18

18 Chapter 1 General Introduction. a b c Figure 9. Movement paths of (a) 5 VHF collared wildebeest ( ), (b) 5 GPS collared wildebeest ( ), and (c) 5 theoretical wildebeest paths based on evolutionary modeling where wildebeest survival depends on maximizing NDVI and grass greenness (Boone et al., 2006). 2006; Murray, 1995), which negates the explanation that wildebeest and zebra are simply moving to the short grass plains to take advantage of an abundance of fresh grazing. However, the young short grass in the south has higher crude protein concentrations (Braun, 1973; Kreulen, 1975; Murray, 1995), so by moving to areas with the most recent growth on the southern plains, wildebeest and zebra could maximize their total protein consumption (Holdo et al., 2009). As the rains subside, the long grass dries more slowly and tends to retain crude protein for longer periods into the dry season than the short grass (Braun, 1973). The closest available patches of long green grass after wildebeest leave is in the Western Corridor. If migrants select patches based on protein, this might account for a triangular migration path that detours to the Western Corridor, rather than a straight linear northward movement as predicted by rainfall. Compensatory vegetation production The compensatory vegetation production of grass has been interpreted as a positive feed back function between high densities of grazers and the growth of nutrient rich grass during the wet season (McNaughton, 1976). Areas that are heavily grazed such as the shortgrass plains, tend to produce more net grass biomass than ungrazed patches, which in turn 19

19 SECTION I: Background and Theory. 20 supports a greater number of herbivores. This repeated rotational grazing system increases the leaf biomass, induces tillering, and reduces the stem biomass which makes the grass more digestible and increases it quality (Braun, 1973; McNaughton, 1976). Generally, compensatory vegetation production cannot occur in areas that have less than 5 months of rainfall because plants become water limited (Augustine et al., 2006). Furthermore, the net yield declines in over-grazed areas suggesting there is some optimal grazing return interval that maximizes net grass production (Braun, 1973). By repeatedly grazing the same areas, wildebeest and zebra might increase the crude protein production of the grass (Braun, 1973; Murray, 1995). Also the grass has time to recover when grazers rotate frequently between patches (Fryxell, 1995). Therefore, migrations of large densities of wildebeest and zebra could facilitate themselves by increasing the seasonal availability of high quality grass; in essence, wildebeest and zebra are maintaining very large grazing lawns (i.e. over 2,500km2). Ultimately, the populations are regulated by the total consumable biomass in the system (Mduma et al., 1999). The ratio of calcium to phosphorous in the grass Phosphorus deficiencies are common in livestock and other mammalian grazers. The absorption of phosphorus through the intestinal wall is dependent on calcium concentrations (ideally Ca:P ratios in the forage should be between 1 and 2). In situations where the forage contains insufficient calcium concentrations, animals will continue to absorb phosphorous by mobilizing calcium from their bones. Therefore, if the ratios of Ca:P vary below 1 or above 2, wildebeest and zebra could become calcium or phosphorous stressed. There is a physiological limit beyond which animals can no longer buffer these deficiencies before they need to replenish their stores of calcium. This physiological limit might be the bottle-neck which partially accounts for the seasonal movement of wildebeest and zebra. The concentration of calcium in northern grasslands during the wet season only just meets the minimum requirements for lactation in wildebeest (Figure 10a). Calcium concentrations decrease even more in the dry season which would result in a deficit if wildebeest remained in the north all year round. Furthermore, the phosphorous concentration in the north is well below the requirements for wildebeest (Figure 10b) (Murray, 1995). The Ca:P ratio is commonly above 2 in the north, which probably exacerbates any phosphorous deficiencies experienced by wildebeest (McNaughton, 1990). Physiological restrictions therefore make it impossible for 1.2 million wildebeest and 250,000 zebra to remain in the north, and therefore necessitates wildebeest to move further afield in search of vegetation with sufficient calcium and phosphorous concentrations. In the Serengeti ecosystem calcium and phosphorous occur in abundance and in suitable ratios only on the short grass plains (Figure 10c), and especially in patches that are and kept in a productive growing phase by repetitive grazing (Kreulen, 1975; Murray, 1995). These elements are most available to grazers during the wet season (McNaughton, 1990). This seasonal flux of calcium and phosphorous is probably an important aspect of southern return journey by wildebeest. Furthermore, the synchronous reproduction of the wildebeest means that peak lactation when females are most nutritionally challenged coincides with an abundance of phosphorous and calcium on the plains. This also provides females an opportunity to replenish their reserves before becoming pregnant again in June (Murray, 1995). Sodium concentrations of the grass Lactating wildebeest require nearly twice the amount of sodium compared to non-lactating females which adds a large nutritional strain on individual animals (Murray, 1995). The grasses in northern woodlands are especially poor in sodium concentrations (Figure 11) and do not meet the requirements for wildebeest, which is another reason that makes it impossible for large herds of wildebeest to remain in the northern woodlands year round. Although grazers