R.S. Reid, S. Serneels, M. Nyabenge and J. Hanson
All eastern Africa is in the tropics, but its grasslands cover a very wide range of altitudes. Extensive grassland s are mostly in arid and semi -arid zones. The area is subject to droughts and a high degree of pastoral risk. Potential vegetation is largely desert and semi-desert, bush and woodland, with only a small area of pure grassland, but the grass -dominated herbaceous layer of the other formations is very important for wildlife and livestock; 75 percent of eastern Africa is dominated by grasslands, often with a varying amount of woody vegetation. The grasslands have been grazed by livestock and game for millennia. Eastern Africa is a centre of genetic diversity for grasses. Six to eleven main grassland zones have been described. Grasslands are either under government control, are open access or are common property resources. Access to resources are under national laws but frequently traditional land use rights are granted by local communities. National land tenure systems are unrelated to traditional ones. Governments supported cropping and reduction of communal grazing land; contraction of pastoral systems reduces the scale of resource use by pastoral peoples. The population is very varied - pastoral groups tend to be of different ethnicities from agricultural or agropastoral groups. Most pastoral systems are in the semi-arid areas, with small areas in hyper-arid and subhumid zones. Traditionally, livestock and their products were for subsistence and wealth, but now many are marketed. Grasslands are increasingly being integrated into farming as pastoral systems evolve. Sown forages are widely used in agricultural areas. Cattle, like people, are mostly in the non-pas-toral areas (70 percent), except in countries with little high-potential land. Cattle, camels, sheep, goats and donkeys are the main livestock kept by the pastoralists for subsistence; most herds are mixed. Indigenous breeds are the majority, although exotic cattle are kept for dairying in high altitude zones. Wildlife are widespread in the grazing lands and are important for tourism. Agricultural development along watercourses limits access by wildlife and pastoral stock.
This chapter focuses on the grazing lands or rangelands of Burundi, Eritrea, Ethiopia, Kenya, Rwanda, Somalia, the Sudan, the United Republic of Tanzania (Tanzania) and Uganda (Figure 2.1). These comprise extensive semi arid to arid grasslands, savannah, bushlands and woodlands, and also cover the natural grazing areas of the extensive highland areas of the region. These are also the pastoral rangelands that Holechek, Pieper and Herbel (1989) defined as uncultivated land that will support grazing or browsing animals.
Figure 2.1 - Countries in eastern Africa as defined for this chapter.
Pastoral management systems in eastern Africa have developed over the last three to four thousand years by the indigenous groups of pastoral peoples living in the region, whose livelihoods depend on livestock. These traditional and often sustainable ways are now being threatened by agricultural development, the need to produce more food from marginal lands, population growth and global climate change. Fluctuations in rainfall and drought are recurring problems in the rangelands of the region and 70 million people in the Horn of Africa, many of whom are pastoralists, suffer from long-term chronic food insecurity (FAO, 2000). Poverty levels are high, with more than half of the people in the region surviving on less than US$ 1 per day (Thornton et al., 2002). The population of the region has doubled since 1974, and it is predicted to increase another 40 percent by 2015 (FAO, 2000). Against this background, the traditional ways of pastoralists continue to change, and many are settling (or are settled) and diversifying their income-generating activities into crop production, wage labour and other activities, while other family members continue to herd the family stock and move to follow the availability of forage.
This chapter examines the changes in pastoral rangeland systems in eastern Africa over recent years and estimates future changes in the rangelands of the region due to global climate change, human population growth and market opportunities.
The productive potential of the eastern Africa n region varies enormously from place to place, as shown by the differences in the growing season across the region (Figure 2.2; Fischer, Velthuizen and Nachtergaele, 2000). On this map, areas coloured brown and yellow have less than 60 growing days and thus rarely support crops (= arid, according to White, 1998); areas adequate for short-season crops with 60-120 growing days are shown in light green (= semi -arid); areas with 121-180 days, shaded in medium green, can support longer-season crops (= dry subhumid); and areas with >180 growing days are in dark green, and have few production constraints (= wet subhumid). Over the region, about 37 percent of the land surface (or 2.3×106 km2) is only agriculturally suitable for grazing by wildlife and livestock (= arid and semi-arid areas), while the other 63 percent (3.9×106 km2) is additionally suitable for crop cultivation, forestry and other types of land use. Of these arid and semi-arid areas principally suitable for grazing, about 1.6×106 km2 (or about 70 percent of the grazing land) is arid and completely unsuitable for crop production (zero growing days) and thus is probably only available for grazing during the rare high rainfall years or during a few weeks or months in normal or low rainfall years. Significant drylands cover northern Sudan, eastern Ethiopia, much of Eritrea and Somalia and northern Kenya, while most of Tanzania, Rwanda, Burundi and Uganda are relatively wet. These four high-rainfall countries and southern Kenya, the highlands of Ethiopia and southern Sudan have the highest potential for intensive crop-livestock production. Much of this is now already under cropland, with the exception of southern Sudan (for cropland, see Figure 2.7).
Figure 2.2 - Length of growing period (days) with sufficient soils and water to grow crops. Reclassified from Fischer, Velthuizen and Nachtergaele, 2000.
Figure 2.3 - Potential vegetation in eastern Africa. Re-classified from White, 1983.
The potential vegetation of eastern Africa is largely desert and semi desert (26 percent of the land surface), bushland (33 percent) and woodland (21 percent) (from Figure 2.3; White, 1983). Only 12 percent of the region is naturally forested, and even less is pure grassland (7 percent). Afromontane vegetation, much of it potential grazing land, is rare (0.5 percent) and mostly restricted to Ethiopia, with very small amounts on volcanic mountains in Kenya, Uganda, Rwanda and Tanzania. Although pure grassland is found only in central Harpachne schimperi - Athi plains, Kenya. and south-eastern Sudan, northern and western Tanzania and northwest Kenya, the herbaceous layer of semi-deserts, bushlands and woodlands are dominated by grasses, so they are included here as part of the grass -dominated areas of eastern Africa because of their importance for livestock and wildlife. This means that 75 percent of eastern Africa is dominated by either pure grasslands or grasslands with varying amounts of woody vegetation within or above the grass layer. Significant woodlands exist only in southern Sudan, Tanzania and Eritrea, and in northern Uganda and western Ethiopia.
Figure 2.4 - Density of large mammal species in eastern Africa, based on data from IEA, 1998.
Predator harvesting game. Cheetah among
Eastern Africa is renowned for the diversity and number of its large grazing and browsing wildlife (Plates 2.1, 2.2 and 2.3). A map of the density (number per km2) of species of medium and large mammals in eastern Africa was developed by a simultaneous overlay of 281 individual species distribution maps (see Figure 2.4, developed by Reid et al. (1998) based on analysis of databases from IEA (1998)). The highest diversity of medium to large mammal species is found in two large, contiguous patches: one in the Rift Valley of south-central Kenya and central Tanzania, and the other in and east of the Ruwenzori Mountains in southwestern Uganda and northern Rwanda. This is the richest diversity of mammals of this size in all of Africa (Reid et al., 1998) and probably the world. Most of Burundi, Kenya, Rwanda, Tanzania and Uganda support diverse groups of large mammals, with fewer in most of Djibouti, Eritrea, Ethiopia and Somalia. This map does not account for the rarity or endemism of large mammals, which can be distributed quite differently from overall diversity.
Large non-ruminant herbivores - zebra herd - Athi plains, Kenya.
Gerenuk - dry area browsers - Tsavo East, Kenya.
As expected, most of the people in eastern Africa live in the wetter and highland areas (Figure 2.5; Deichmann, 1996; Thornton et al., 2002). High population levels are found in the Ethiopian highlands, the Lake Victoria Basin and the southern Tanzanian highlands. Significant clusters of people et al., 2002. live in areas marginally suitable for cultivation in Eritrea around Asmara, in central Sudan and along the coasts of Kenya, Tanzania and Somalia. The only places where many people live in drylands are along the Nile in northern Sudan, around Mogadishu in Somalia and in western Somaliland of northern Somalia. Few people live in most of the drylands of eastern Africa and in the wetter areas of the Sudd in southern Sudan, in the tsetse belts of Tanzania and in protected areas.
Figure 2.5 - Human population density in eastern Africa in 2001. From Deichmann, 1996; Thornton
Figure 2.6 - Cattle population densities in eastern Africa in the late 1990s, from Kruska, 2002.
Cattle are largely distributed in a pattern similar to the human population distribution in eastern Africa (Figure 2.6; Kruska, 2002), with high concentrations around Lake Victoria and in the Ethiopian highlands. Few cattle are found in the driest areas of northern Sudan, eastern and northern Ethiopia, Eritrea and northeastern Somalia. There are also few cattle in wet, southern Sudan (the Sudd) and northern Uganda, and in the subhumid, miombo woodland regions of southern Tanzania. Most of the cattle are in non-pastoral areas across the region: 70 percent are in cropland and urban areas, while 30 percent are in pastoral lands. These proportions vary strongly from country to country, partly because of differences in amounts of high-potential land. For example, about 35 percent of Kenya is high potential and 80 percent of the nation's cattle herd resides there. In contrast, there is very little high-potential land in Somalia and Djibouti and thus all the cattle live in drylands in those countries.
A previous global analysis of pastoral systems (from Reid et al., 2003) has been used to estimate the extent of grass -dominated pastoral systems in eastern Africa. This pastoral systems map (Figure 2.7) was created using four Geographical Information System (GIS) data layers: land cover (USGS/EDC, 1999; Loveland et al., 2000), length of growing period (Fischer, Velthuizen and Nachtergaele, 2000), rainfall (IWMI, 2001; Jones and Thornton, 2003) and human population density for Africa (Deichmann, 1996).
Initially, land cover, length of growing period and human population maps were used to establish the location of all cultivatable land (>60 growing days), all land cover currently under crops in the USGS coverage (dryland cropland and pasture; irrigated cropland and pasture; mixed dryland and irrigated cropland and pasture; cropland and grassland mosaic; and cropland and woodland mosaic) and any other areas with sufficient human population (>20 people/ km2) to exclude extensive rangeland use (for details, see Reid et al., 2000a; Thornton et al., 2002). This classification thus joined all but the most extensive agropastoral systems with cropland, and maps about 9 percent more cropland than is in the USGS database. Urban included all areas with more than 450 people/km2. The remaining areas (not cultivatable, low human population density) were discriminated into pastoral system classes by mean annual rainfall as follows: areas receiving less than 50 mm of rainfall were classified as hyperarid; areas with 51-300 mm were arid; and areas with 301-600 mm were semi arid. Highland areas were those with temperatures of more than 5°C but less than 20°C during the growing season, or less than 20°C for one month a year.
Most of eastern Africa's pastoral systems are semi -arid (34 percent), with much smaller areas of arid (12 percent), hyper-arid (8 percent), humid to sub-humid (9 percent), and temperate and highland (1 percent) pastoral systems (Figure 2.7). Cropland and urban areas cover 27 percent of the region. Only Sudan has the driest (hyper-arid) pastoral systems, while eastern Eritrea, northern Ethiopia, Djibouti, Somalia and northern Kenya support extensive arid pastoral systems. The most common land cover type in Kenya, Somalia, Ethiopia and Sudan is semi-arid rangeland. Tanzania, Uganda and Sudan have the most extensive wet pastoral systems.
By comparing potential vegetation (Figure 2.3) and pastoral and cropland systems (Figure 2.7), we can see what types of vegetation farmers have preferred to use for cropland. On average, 27 percent of the region is cropped, but this is disproportionately found in afro montane vegetation (74 percent converted to cropland), forest (62 percent converted), woodland (34 percent converted) and bushland (31 percent converted). Farmers have ploughed lesser areas of pure grassland (23 percent converted), semi -deserts (3 percent) and deserts (1 percent). These land use choices have pushed pastoral use from the wetter to the drier areas in eastern Africa over time.
Figure 2.7 - Pastoral system areas and cropland and urban areas of eastern Africa in 2001, based on Thornton et al. (2002) and Reid et al. (2003).
The grasslands of eastern Africa are very diverse, with a range of dominant species dependent on rainfall, soil type and management or grazing system. Eastern Africa is renowned as a centre of genetic diversity of tropical grasses and the centre of greatest diversity of cultivated grass species (Boonman, 1993). Over 90 percent of the major cultivated forage grasses have their centre of origin in sub-Saharan Africa and are indigenous to the extensive grasslands of eastern Africa. There are an estimated 1 000 species of grass indigenous to the region, with more than 600 species found in Kenya alone (Boonman, 1993). The wide distribution and adaptability of many of these species across a range of environments and management systems indicates the presence of considerable genetic diversity within the region. This diversity has been exploited to select superior ecotypes for use in many other parts of the world. Brachiaria species, originating from eastern Africa, are the most widely planted forage grass, with estimates of areas under Brachiaria pastures in Brazil ranging from 30 to 70 million hectares in 1996 (Fisher and Kerridge, 1996).
To aid description and study of the rangeland, many attempts have been made to classify the vegetation into types that cover large areas of the region. Rattray (1960) identified 12 types of grassland in eastern Africa, based on the genera of the dominant grass in the grassland. These include Aristida, Chloris, Cenchrus, Chrysopogon, Exotheca, Hyparrhenia, Heteropogon, Loudetia, Pennisetum, Panicum, Setaria and Themeda. Pratt and Gwynne (1977) described six eco-climatic zones based on climate, vegetation and land use. These are described as the afro-alpine area of upland grasslands; the equatorial humid to dry sub-humid area of forests and bushlands (Plates 2.4 and 2.5); the dry subhumid to semi -arid area of savannah, shrub and woodland; the semi-arid areas of dry woodland and savannah (such as the Acacia-Themeda association); the arid area of Commiphora, Acacia, Cenchrus ciliaris and Chloris roxburghiana; and the very arid area of dwarf shrub grassland of Chrysopogon. A more recent classification, based primarily on the dominant grass, is described as vegetation type or region by Herlocker (1999). He described eleven vegetation regions in eastern Africa as Pennisetum mid-grass; Pennisetum giant grass; Panicum-Hyparrhenia tall-grass; Hyparrhenia tall-grass; Hyparrhenia-Hyperthelia tall-grass; Themeda mid-grass; Chrysopogon mid-grass; Leptothrium mid-grass; Cenchrus-Schoenefeldia annual mid-grass; Panicum-annual; Aristida mid- and short-grass region; and Aristida short-grass region.
Themeda triandra (Plate 2.6) is one of the most widespread grass species in sub-Saharan Africa but it is only the dominant grassland type in central and northern Tanzania. The species is very variable and shows wide adaptation to growth in both the highland regions and the lowland savannahs. Themeda, Bothriochloa, Digitaria and Heteropogon mixtures are common in the open dry savannah areas of Tanzania, such as the Serengeti plains. Short tufted ecotypes of Themeda triandra are found at high altitudes and taller more woody types are found in the open lowland savannahs (Rattray, 1960). These vary in palatability, but all types quickly lose palatability with age. Themeda triandra can tolerate light to moderate grazing, and productivity can reach 400 kg/ha/day in the wet season in the Serengeti plains, making them among the most productive grasslands in the world (Herlocker, 1999). Plant biomass, quality and species numbers decline in the absence of grazing, are at a peak under moderate to high grazing (McNaughton, 1976, 1979, 1984) and can decline under very high grazing. In the Mara region in Kenya, to the north, which is a continuation of the grassland ecosystem of the Serengeti Plains, Themeda makes up about 50 percent of the grass cover in lightly to moderately grazed sites, dropping to 1-5 percent cover near settlements where Maasai corral their livestock each night (Vuorio, Muchiru and Reid, in prep.).
Acacia bushlands cover much of the rich volcanic soils of eastern Africa.
Farmers use many of the trees in bushlands and woodlands to manufacture charcoal for market.
Maasai sheep grazing in a Themeda grassland, southwestern Kenya.
The dominant grass species in the drylands of eastern Africa include Aristida, Cenchrus, Chrysopogon and Heteropogon. These are often found growing as an association, the dominant species determined by the environment and soil type. Aristida grassland is widely distributed in the dry pastoral areas of Kenya, Ethiopia and the Sudan. Although many species are tough and have low palatability, they have wide adaptability to a broad range of environments. Cenchrus grassland is often found associated with Aristida, or in Somalia with Leptothrium (Herlocker, 1999), and has higher palatability and better adaptation to hot dry areas with high evapotranspiration. Cenchrus is one of the few grass genera that has been characterized for agronomic attributes. Over 300 ecotypes, mostly collected from Tanzania and Kenya, were characterized for 12 agronomic attributes (Pengelly, Hacker and Eagles, 1992). The ecotypes showed wide variability in their agronomic traits and were clustered into six groups (Pengelly, Hacker and Eagles, 1992). The annual C. biflorus, which is adapted to dryland areas, is also found in eastern Africa associated with Schoenefeldia sp. and is typical of one dry area south of the Sahara in western Eritrea (Herlocker, 1999).
Chrysopogon plumulosus is the most widespread species found in the semi desert grasslands and bushlands of the Horn of Africa (Herlocker, 1999) and is avidly grazed, especially in Somalia and Sudan, where it is burnt to stimulate regrowth for grazing. Chrysopogon is very sensitive to grazing. Overgrazing results in elimination of the species and a change in species composition to annuals such as Aristida spp. (Herlocker, 1999). This harsh management regime in low rainfall areas has resulted in reduced stands of this grassland in recent years (IBPGR, 1984). Herlocker (1999) recognized three zones in the Chrysopogon region according to the associated woody vegetation. These include Commiphora-Acacia bushland and Acacia etbaica open woodland, which occur across the region, and the Acacia bussei open woodland in Somalia and Ethiopia. He also recognized two subregions: the Cenchrus-Chloris subregion in the wetter areas and the Sporobolus subregion in the drier areas. Rattray (1960) recognized the Chloris areas as a vegetation type in its own right, and included the Sporobolus as an associated grass in a Chrysopogon vegetation type in very dry semi-desert areas of Somalia and Ethiopia.
Although not a vegetation type recognized by Herlocker (1999), Heteropogon grassland is found in open woodland or grassland in the semi -arid and arid rangelands in Somalia (Box, 1968), Kenya and Ethiopia. It is represented mostly by H. contortus, which is commonly called spear grass due to its awns and needle-sharp tips on the grass florets. It is a persistent species, which is indigenous to the region, spreads rapidly through seed and grows in lowland or middle altitudes with poor, stony, well drained soils. It is commonly found with annual species of Aristida and Digitaria (Rattray, 1960). The species does not have good palatability and is only useful when young.
Chloris roxburghiana is a dominant species in dryland areas of Kenya, Ethiopia, Tanzania, Somalia and Uganda, and is usually found growing in association with Chrysopogon aucheri and Cenchrus ciliaris in Commiphora and Acacia woodland (Rattray, 1960). Despite its wide distribution, Herlocker (1999) treats this vegetation type as a subtype of the Chrysopogon mid-grass region. Chloris roxburghiana is widespread throughout the entire region and is an important species for livestock and wildlife. This species contributes up to 50 percent of the diet of wild herbivores in eastern Kenya (IBPGR, 1984) but is in danger of disappearing due to overgrazing and land degradation. The species is very variable. A recent study using random amplified polymorphic DNA (RAPD) markers to study diversity among four populations from ecologically distinct sites in eastern Kenya showed significant variation among the populations (W.N. Mnene, KARI, Nairobi, pers. comm.).
Chloris gayana is an important native species and a component of the Hyparrhenia type of grassland (Rattray, 1960) in open steppe and wooded grassland vegetation or flooded valleys in the higher rainfall areas of Kenya, Ethiopia, Tanzania, Somalia and Uganda. Herlocker (1999) considers this vegetation type part of the Hyparrhenia-Hyperthelia tall-grass region of miombo woodland. The miombo woodland is an important vegetation type covering the southern two thirds of Tanzania. Chloris gayana, or Rhodes grass, is not an important grass ecologically in the vegetation of the region, but is important commercially as a forage grass. It shows wide adaptability with high palatability, and is a fast-growing, persistent, frost- and drought -tolerant species valued for grazing (Skerman and Riveros, 1990). Commercial cultivars of Rhodes grass have been developed from genotypes collected in Kenya and grown in the region since the 1930s (Boonman, 1997). An analysis of genetic diversity in Chloris gayana using amplified fragment length polymorphisms (AFLPs) revealed considerable variation between the diploid and tetraploid cultivars, with genetic similarity ranging between 66 and 89 percent in the diploids and 63 and 87 percent in the tetraploids (Ubi, Komatsu and Fujimori, 2000).
Hyparrhenia is one of the most widespread grassland types in eastern Africa, and this grassland region, which is characterized by woodlands and wooded grasslands dominated by H. rufa, covers parts of Uganda, Kenya and Ethiopia (Herlocker, 1999). Several other species of Hyparrhenia are found in the region, of which the most important are H. hirta, H. diplandra and H. filipendula. These tough perennial grasses are usually found growing in combination with other grasses in woodland or open grassland, from the lowland to mid-altitude areas. They are fast growing, and grazed while young, but become tough and unpalatable as they mature and lose nutritive value (Skerman and Riveros, 1990). Crude protein levels of H. dissoluta in Kenya can decrease from over 14 percent to less than 3 percent after flowering (Dougall, 1960). After flowering, these grasses are much valued and used as thatching for traditional rural housing, and mature grasses have commercial value, being sold as standing grass to be cut for roofing in some rural areas. This and burning ensure young regrowth with higher value for grazing in many areas. Grazing is important to encourage growth of other more palatable and valuable forage grasses, such as Cynodon dactylon, Panicum maximum and Setaria sphacelata (Herlocker, 1999).
Loudetia species are often found mixed with Hyparrhenia spp. and Themeda triandra in open grassland on shallow, rocky, sandy soils. They provide late-season grazing for livestock (Rattray, 1960) but have low palatability (Skerman and Riveros, 1990). Although Herlocker (1999) did not consider this a vegetation type per se, and Rattray (1960) only considered this as a grassland type for Uganda, Loudetia is widely distributed in rangeland ecosystems in Tanzania, Kenya and Ethiopia, but is never the dominant species. The most common species in the region is Loudetia simplex, which shows considerable variability in morphology in Ethiopia (Phillips, 1995). However, the genus has not been widely studied due to its low economic importance.
The highland areas of eastern Africa cover about 80 million hectares of Ethiopia, Kenya and Uganda. Exotheca abyssinica grassland is common on poor waterlogged soils in high altitude areas of eastern Africa, especially on the seasonally waterlogged vertisols, of which there are 12.6 million hectares alone in Ethiopia (Srivastava et al., 1993). This species is closely related to Hyparrhenia and is often found growing in association with Themeda triandra. E. abyssinica has tough leaves and low nutritive value (Dougall, 1960), providing good grazing while young but quickly becoming tough and unpalatable. Setaria incrassata and S. sphacelata are also common grassland species found in Acacia woodland up to 2 600 m altitude on the vertisols of Uganda, Sudan and Ethiopia (Rattray, 1960). S. incrassata is a very variable species, with morphotypes varying in plant robustness, bristles, and number and density of spikelets (Phillips, 1995). It is closely related to S. sphacelata, which is also a very variable species, allowing selection of a range of cultivars from Kenyan ecotypes that vary in frost tolerance, maturity, pigmentation and nutritive value (Skerman and Riveros, 1990). Both S. sphacelata and S. incrassata are palatable grasses that withstand heavy grazing.
Pennisetum grassland areas can be classified as two types: high altitude grasslands of P. clandestinum and savannah grassland of P. purpureum (Rattray, 1960; Herlocker, 1999). Although belonging to the same genus, these species are morphologically and ecotypically very distinct, and have very different distribution and ecological niches. Both species are indigenous to eastern Africa, with high economic importance, and are cultivated in many other parts of the world.
P. clandestinum is a prostrate stoloniferous perennial that is widely distributed in areas from 1 400 m to over 3 000 m in Kenya, Ethiopia, Tanzania and Uganda. Its common name, Kikuyu grass, derives from the highlands of Kenya, where it is abundant, being named after the Kikuyu ethnic group of central Kenya. It shows wide adaptability to drought, waterlogging and occasional frosts (Skerman and Riveros, 1990). It is highly digestible, palatable, persistent and withstands severe defoliation and grazing. It is the dominant species in natural pastures in many parts of the eastern Africa n highlands. It is an invasive secondary species, which can quickly colonize disturbed soil in cropping areas and fallow land, spreading by seeds or stolons, and may become a serious weed in cropland (Boonman, 1993). It shows wide variability, with three distinct ecotypes classified on leaf width and length, stolon size and floral structure (Skerman and Riveros, 1990). Several ecotypes have been selected as commercial cultivars, which have been widely introduced into tropical highland and subtropical areas. It is now widely grown outside its native distribution and is commonly cultivated in the Americas. Studies in the USA using starch gel electrophoresis to describe the distribution of genetic variation within and among introduced populations found a relatively high proportion of polymorphic loci across populations, indicating fixed heterozygosity due to polyploidy (Wilen et al., 1995). The highland grazing areas of P. clandestinum are often mixed with P. sphacelatum and Eleusine floccifolia. These two grasses are frequent in overgrazed pastures in the highlands and mid-altitudes in the Rift Valley, but are not palatable (Sisay and Baars, 2002) and are important for traditional basket making. Cattle avoid these grasses, which have the potential to become major weeds on upland pastures unless collected for basket making. Basket making is an important activity and source of income for rural women and collection of these weedy grasses also maintains the quality of the communal grazing areas and grasslands in the highlands.
Pennisetum purpureum is a tall, erect, vigorous perennial species that grows in damp grasslands and forest areas up to 2 400 m in Kenya, Tanzania, Uganda and Sudan. Herlocker (1999) recognized this as a vegetation region in Kenya and Uganda, around the shores of Lake Victoria. Pennisetum purpureum is widely distributed through sub-Saharan Africa and is commonly called elephant grass or Napier grass, named after Colonel Napier of Bulawayo in Zimbabwe, who promoted its use at the start of the century. It is now widely used for cut-and-carry (where grass is collected by hand and carried to stall-fed cattle) for the smallholder dairy industry in eastern Africa and frequently produces up to 10-12 t/ha dry matter in rainfed conditions (Boonman, 1993). Elephant grass is palatable when young and leafy. It is fast growing and should be cut often to avoid its becoming tough and unpalatable with a high proportion of stem. Due to its importance in the region, considerable research has been done on elephant grass, including studies on its diversity. Tcacenco and Lance (1992) studied 89 morphological characters on 9 genotypes of elephant grass to determine which characters were most useful for description of the variation in the species, and concluded that variation existed from plant to plant, even within the same accession. A larger collection of 53 accessions was characterized for 20 morphological and 8 agronomic characters (Van de Wouw, Hanson and Leuthi, 1999). Again the germplasm was found to be very variable, but accessions could be clustered into six groups with similar morphology. More recently, molecular techniques using RAPD markers were applied to study the genetic diversity in the same collection, and also among farm clones in Kenya (Lowe et al., 2003). This technique was able to separate out hybrids between P. purpureum and P. glaucum from pure elephant grass accessions. Despite being clonally propagated, genetic diversity (Magguran, 1988) across all accessions was found to be fairly high, with a Shannon's diversity index of 0.306.
Panicum maximum is another tall, fast growing species that is often found associated with Pennisetum in eastern Africa n grasslands or associated with Cenchrus and Bothriochloa in Acacia woodland in the dry savannah areas (Rattray, 1960). Herlocker (1999) recognized the Panicum-Hyparrhenia region along the coast northwards from Tanzania, through Kenya into Somalia. Panicum maximum is more widely distributed in Kenya, Ethiopia and Tanzania and is typical of shady places in the foothills of mountain ranges up to 2 000 m. P. maximum is a pioneer grass that comes in after clearing and cultivation of the lowland forest. There is a wide variation in plant habit, robustness of culms and pubescence (Phillips, 1995), and ecotypes with good agronomic characters have been selected as commercial cultivars. P. maximum is fast growing and palatable, and its wide adaptation and variability make it an excellent grazing species in the savannahs. A collection of 426 ecotypes of P. maximum collected from Tanzania and Kenya were evaluated for morphological and agronomic traits in Brazil (Jank et al., 1997). Twenty-one morphological descriptors were found to discriminate among accessions and were used to cluster the collection. Considerable variation was found among the ecotypes and some with wide adaptation were selected for establishment of a breeding programme. Other locally well -adapted ecotypes are also being developed for use within the region of adaptation.
Most dry grasslands of eastern Africa are characterized by frequent droughts and high levels of risk of production for pastoral peoples (Little, ). Livestock are one of the few ways to convert sunlight into nutritious food in these drylands (wildlife are also important). Pastoralists traditionally manage risk by moving their livestock on a daily and seasonal basis to follow changes in the quality and quantity of pasture (IFAD, 1995). Cattle, camels, sheep, goats and donkeys are the main livestock species and are kept by the pastoralists for subsistence for their milk, meat and traction. Most herds are mixed as a means of adaptation to a changing environment, to supply food for the family and to act as a cash reserve in times of shortage, during droughts or disease-pandemics (Niamir, 1991).
Although sale of livestock is a major source of income for pastoralists today, widespread sale (or commoditization) of livestock only became common in the last century, with colonialism (Hodgson, 2000). Settled crop-livestock farmers are particularly oriented toward marketing: selling animals, milk and hides regularly. Herds are managed in a way that minimizes sales because of the traditional social and economic functions of livestock other than income generation (Coppock, 1994). In most pastoral areas, livestock are used as a social safety net, with livestock exchange cementing mutual obligations to help each other in times of need. Like many other pastoral areas, cattle are also of particular significance in the Borana area of Ethiopia as a symbol of wealth and prestige, and owners are reluctant to sell. Sheep and goats are usually sold to raise cash for household needs. Although marketing of livestock products (milk, meat, hides) in pastoral systems is a relatively new phenomenon, pastoral peoples who live near markets and roads are increasingly selling products.
Traditionally, herders consume a large part of the milk produced; any surplus is shared with neighbours, exchanged in barter or sold in urban areas. In Somalia, a commercial milk chain through a cooperative has been established by the pastoralists for marketing camel milk in Mogadishu as a source of income to buy sugar, clothes and medicines (Herren, 1990). An EU-funded project, Strengthening food security through decentralized cooperation, active from 1996 to 2002, also supported establishment of a small processing plant for pasteurizing camel milk and marketing the resulting products in suitable packaging for the Somali market (EC, 2000). The 2001-2 drought had a considerable effect on camel calving intervals and milk sales. In some parts of Somalia, there was virtually no income from milk sales following the drought. Milk formerly provided approximately 40 percent of a household's income and the return on livestock sales, which typically provide an additional 40 percent of income, was halved after the drought (FSAU, 2003). Maasai in Kenya and Tanzania living close to main roads or towns sell fresh milk, butter or fermented milk. The Borana in southern Ethiopia sour cow's milk and process it into butter for sale in local markets or for transport to large cities (Holden and Coppock, 1992). Distance to market, season and wealth of the household (which is directly related to the number of livestock owned) influence marketing of dairy products in the southern rangelands of Ethiopia (Coppock, 1994).
Most of the extensive grasslands in the region are either under the control of the government and designated as wildlife and conservation areas for national parks (about 10 percent of the land area) or are open access or common property resources. Access to these resources and the conditions under which they can be used are under national laws, but frequently traditional land use rights are granted by local communities. Traditionally, long-term sustainability of these rangelands has been ensured by agreed management norms, but these are increasingly breaking down as lands privatize, crop farmers migrate to pastoral areas and human needs grow. Governments are also reducing support to pastoral peoples, who are often marginalized in national affairs (IFAD, 1995). Options for income generation and alternative land uses for extensive grasslands for pastoralists are limited and can lead to overutilization and land degradation if none of the users take responsibility for the management and sustainability of the system.
Common property and traditional access regimes with sustainable range management institutions and resource sharing arrangements were practiced in the region until the colonial era (IFAD, 1995) and continue in some areas today. These were and are based on a transhumance grazing system developed over many years to exploit the ecological heterogeneity and make optimal use of the scarce resources of grazing and water throughout the year. These traditional management practices include grazing rotation strategies and establishment of grazing preserves for the dry season. Drought is the most serious challenge facing pastoralists in the region and access to land and water are often the cause of conflict between pastoralists, ranchers and crop-livestock farmers (Mkutu, 2001). Traditional systems of access to water are common in most countries in the region. The pastoralists of northern Somalia and southern Ethiopia also have a complex and well -regulated system of well management to regulate water use, as well as traditional informal and formal social controls on use of common property and open property resources to ensure sustainable use of the grassland and water resources (Niamir, 1991). This is exemplified by herder response to drought and conflict in southern Somalia, where herders move camels and cattle great distances to good pastures in times of drought, while they graze small stock closer to home (Little, ).
Over the last century, these indigenous range management institutions have been weakened by demographic, political and social change in the region. The greatest threat to the traditional pastoralist system comes from the rapid population growth of the last twenty years and conversion of communal grassland to open access state property or private land, which has led to more grassland being used for smallholder crop-livestock farming. Policies have constrained the movement of pastoralists and promoted sedentarization and many permanent settlements have been established in the rangelands; with many pastoralists choosing to shift their production systems to include crop-livestock farming (Galaty, 1994; Campbell et al., 2000). In S.E. Kajiado District, Kenya, land use conflict reflects ongoing competition over access to scarce land and water resources between herders, farmers and wildlife - competition that has intensified strongly over the last 40 years, after the district became open to outside migrants.
Today, farming extends into the wetter margins of the rangelands, along rivers and around swamps. This has reduced the area available for grazing and the ease of access to water for both domestic stock and wildlife. Political alliances have emerged among land managers to gain or maintain control of critical land and water resources and to influence policy on agriculture, wildlife and tourism and land tenure (Campbell et al., 2000). Another well -documented example of this is from the Beja pastoralists in northeastern Sudan, who, as a result of drought, are changing their nomadic way of life as camel and smallstock herders to more settled, smallholder farming and rearing of small ruminants. Like other pastoralists in the region, they find that small ruminants are easy to manage near the homestead, cost less, are more easily sold and breed more quickly than camels (Pantuliano, 2002). Government policies have supported cropping and reduction of communal grazing land and, more recently, mobility patterns and access to key resources have been constrained by conflict and civil insecurity. Many Beja now move very little or not at all, reducing their capacity to make effective use of the rangeland from the perspective of livestock production. As Beja settle, vegetation around settlements has changed, with the disappearance of seven palatable species and an increase in unpalatable species (Pantuliano, 2002). These changes are typical of those faced by pastoralists across the region. Even so, many families (or parts of families) still send the younger family members for transhumance in the dry season while the women and older family members remain on the farm to take care of the crops and smallstock.
The national land tenure systems of the region are unrelated to the traditional land tenure and access regimes of the pastoralist groups. In Ethiopia, the Sudan and Somalia, all land is state owned and cropping land can be leased from or allocated by the government. In Somalia, land tenure is under a mixture of traditional and modern legal systems (Amadi, 1997). The 1975 Land Reform Act of Somalia gave land for state enterprises and mechanized agriculture (Unruh, 1995); pastoralists only had rights as part of government-sponsored cooperatives and associations, and were forced to move from their traditional lands to more marginal lands with open access. All land belonged to the state and 50year leases were provided to users, although many enclosures were not legally leased and ownership was respected by local communities under traditional systems (Amadi, 1997). Following the conflict and the absence of a central government, the deregulation of land tenure and unauthorized enclosure of pastoral land for grass production by entrepreneurs for export livestock production to Kenya left poor herders and agropastoralists with little livelihood security (de Waal, 1996). For the Sudan, the government recognizes rights of possession over land but also reserves the right to acquire land from local owners for the state (Amadi, 1997). In Ethiopia, land is allocated through the land administration, and redistribution occurs, so people do not have secure rights over their land (EEA/EEPRI, 2002), resulting in inter-ethnic and inter-com-munal conflict over resources. In neighbouring Eritrea, land is owned by the community, and land tenure is governed by traditional laws and administered under traditional village administrative bodies (Amadi, 1997).
Land tenure in Uganda is very complex, reflecting the rich history of the country. Mailo tenure is particular to the Buganda area of the country and dates back to 1900 when the king (kabaka) of the Buganda people shared land among the chiefs to own in perpetuity. In 1975, the Land Reform Decree made all land public with title vested in the Uganda Land Commission, and allowed leasehold tenure (Busingye, 2002). Although the mailo system was officially abolished, it continued until the late 1990s, when the 1995 Constitution and the Land Act of 1998 were implemented. Freehold tenure was also granted by the state and later by the Land Commission, mostly to institutions for religious and educational purposes (Busingye, 2002). The 1995 Constitution and 1998 Land Act also identified a new land tenure system called customary tenure. The land is held, used and disposed of following the customary regulation of the community, and people using the land have some rights. Customary tenure is the most common system in the rangelands (Amadi, 1997). The emphasis is on use, which is controlled by the family, who distribute land to male family members for their use rather than ownership. Customary tenure also includes the communal land, where users have rights to grazing, farming, fuelwood, access to water and land for traditional uses and burial grounds (Busingye, 2002). Ownership is through the family or community, and there are no individual ownership rights. Traditional authorities allocate the land and resolve disputes. In addition some land was declared Crown Lands in 1900, and areas are still held by the state under the Uganda Land Commission as protected areas, some of which are now open access.
The land tenure system in the United Republic of Tanzania is a legacy of colonial rule, with all lands being public land and remain vested in the President as a trustee for and on behalf of all citizens of Tanzania (Nyongeza, 1995; Shivji, 1999). The state grants rights of occupancy and tolerates customary occupation and use of land. All public land is categorized under three types: General, Reserved or Village land, which are each managed and administered by ministry officials. The Commissioner for Lands has the power to allocate land on the general, and even reserved, lands. When a village registers its land, the title deeds are held in trust for the whole village by the Village Chairman and Council. Numerous land-related conflicts exist in Tanzania, partly caused by conflicting land use policies. The Villagization Programme (1974-76) concentrated people together, displacing some and allocating them land that was taken from others. Some of the villages were relocated into reserved land, thus creating pockets of habitation and cultivation in protected areas. With the economic liberalization in the mid 1980s, large-scale land alienation occurred, in particularly in the Arusha region, where vast parts of rangelands were leased out to large-scale farmers (Igoe and Brockington, 1999). Village land can also be allocated by the government, if it is not registered or its use can not be demonstrated. To secure their title deeds, many pastoralists started cultivating. Much of the rangeland areas in Tanzania have been categorized as reserved lands, having been set aside as national parks, game reserves or game controlled areas, thus making them inaccessible for herders and their livestock (Brockington, 2002).
Land tenure in Kenyan pastoral systems has evolved rapidly over the last half century. About the 1940s, Kenyan colonial authorities introduced an entirely new type of land use to rangeland ecosystems: wildlife -only protected land. In subsequent years, the Kenya Game Department transferred the management of game reserves in Maasailand to local District Councils. After independence in 1963, these reserves were designated County Council Reserves (Lamprey and Waller, 1990). Most of these conservation areas were established in the dry season grazing reserve for pastoral people, livestock and wildlife. This change in land tenure appropriated these critical resources for use by wildlife alone for the first time.
Also in the mid-1960s, the Kenya Government gave pastoral groups title deeds to large tracts of grazing land that they had used traditionally over a long period (Lawrance Report, 1966). Each member shared ownership of the entire ranch under the Group Representatives Act, 1968, but the livestock were owned by individual members (Lamprey and Waller, 1990). Although these ranches were large (Koyake Group Ranch in the Mara area is 971 km2) and group ranch boundaries were relatively porous to livestock and wildlife movement, these group ranches started to circumscribe who could live where in the ecosystem. The group ranch system was instituted more strongly in the wetter rangelands in the south and just north of Mt. Kenya; arid rangelands further northwest and northeast were largely unaffected by this change in tenure.
Since the early 1980s, group ranches have been adjudicated and are becoming privatized (Galaty, 1994). Areas near towns and roads were the first to be privatized. For example, the rangeland nearest to Nairobi was privatized in the early 1980s, while other group ranches in drier areas are currently undergoing subdivision. Pastoral land owners are struggling to balance the trade-offs of private tenure: even though secure ownership is a boon, lack of access to wider grazing lands and loss of wildlife are not. Groups and families are trying to address these problems with reciprocal grazing arrangements and establishment of community wildlife reserves. This process has partly been driven by pastoral peoples throughout Kenya beginning to settle permanently to have access to schools, health care and other business opportunities in the higher potential areas. At the same time, pastoral people want to secure their ownership rights as they see large tracts of communal land leased to outsiders for mechanized agriculture.
Privatization of land in pastoral areas robs pastoral peoples of one of their greatest assets: communal access to land. In the 1960s, Hardin (1968) decried communal access to land, describing it as the tragedy of the commons , assuming that communal access meant free and unregulated access leading to overuse. This has been used as an argument in favour of privatization. However, most communal access to pastoral land and water is not unregulated, rather it is governed by traditional rules of access controlling who uses the land and water, where and when. These rules were designed to sustain grassland productivity for the use of all in communally shared lands. Privatization of land is now causing the tragedy of privatization, where pastoral people are impoverished because land holdings are too small to support their livelihoods in dry grazing lands. This is what Rutten (1992) nicely coined as selling land to buy poverty. The overgrazing issue is discussed below, applicable to both communal and privatized land.
As pastoral systems evolve and herders avoid drought and disasters through diversification and risk management, sedentarization and settlement to improve income-earning capacity is occurring in northern Kenya and southern Ethiopia (Little et al., 2001). There continues to be an expansion of cropping in areas where agriculture is feasible, to allow herders to better manage risk and respond to drought (Little et al., 2001). As cropping expands into the rangelands of the region, grasslands have become an integral part of crop-livestock systems.
Nearly all grassland areas in developing countries are grazed (CAST, 1999). One viable alternative for settled crop-livestock farmers in the region is to use cultivated forage grasses to support livestock production and reduce the pressure on the natural grassland. Cultivated forages have received less attention from breeders than other crops (CAST, 1999). However, recent expansion in dairying, especially around urban areas in eastern Africa, and the anticipated increased demand for livestock production proposed by Delgado et al. (1999) has led smallholder farmers to pay more attention to increased use of cultivated grasses. Inclusion of grasses into a crop-livestock system can also have positive environmental benefits. Vegetation cover can be improved through transfer of seeds and trampling and breaking soil crusts and fertility improved by manure deposited during grazing (Steinfeld, de Haan and Blackburn, 1997). Fallow and grassland rotations improve soil fertility and minimize soil erosion, while reduced nutrient losses from manure from livestock fed on grasses in a cut-and-carry system double the effective availability of nitrogen and phosphorus and can be put back into the system to maintain nutrient balances (de Haan, Steinfeld and Blackburn, 1997).
Rhodes grass and elephant grass are among the earliest tropical grasses grown in eastern Africa, since the start of the twentieth century. They have been widely planted for livestock production in Kenya and Uganda since the 1930s (Boonman, 1993) and are an important part of crop-livestock systems in higher-potential areas. Grass rotations and fallowing of crop lands were common practices to provide soil cover and restore organic matter some 50 years ago, but this practice has reduced due to increasing population pressure and demand for crop land (Boonman, 1993). Due to scarcity of land, most dairy farmers in the heavily populated highlands of eastern Africa now practice a cut-and-carry zero grazing system. Currently, elephant grass is the most important forage crop in dairy systems in the Central Kenya Highlands (Staal et al., 1997) and has been shown to constitute between 40 to 80 percent of the forage for the smallholder dairy farms. In Kenya alone, more than 0.3 million smallholder dairy producers (53 percent) rely on elephant grass as a major source of feed. The demand is so high that landless farmers plant along highway verges and on communal land to cut and sell to stock owners.
Rhodes grass has also been widely used for improved pastures due to its wide adaptation and vigorous root system, which confers reasonable tolerance to drought and persistence under grazing and makes it suitable for erosion control, and of value for hay making (Boonman, 1993). It shows some cold tolerance, and several commercial varieties have been developed in Kenya. It ranks second only to elephant grass in yield and drought tolerance, producing up to 18 t DM/ha in suitable environments (Boonman, 1993).
Another cultivated grass with wide adaptability that is being grown in eastern Africa is setaria (Setaria sphacelata). Herbage yield can equal Rhodes grass and it is more persistent at higher altitudes, up to about 3 000 m above sea level, and can tolerate frost and seasonal waterlogging (Boonman, 1993). However, it is not as drought tolerant as Rhodes grass and has a tendency to invade agricultural land, and can become weedy and difficult to eradicate. Although its use reduced in Kenya during the 1980s, it is still a useful grass in wetter and higher-altitude areas, and it is now gaining importance for use in soil stabilization and erosion control along bunds in Tanzania and central Kenya (Boonman, 1993). Unfortunately, none of these options for improved forage production are available to settled pastoralists across the vast dryland areas of the region.
Expansion of cropland, intensification of livestock production and changes in land tenure are common forces for change in pastoral systems around the world (Niamir-Fuller, 1999; Blench, 2000). Across Africa, colonial and postcolonial policies favoured crop cultivation over livestock production, thus giving agriculturalists the economic upper hand compared to pastoralists (Niamir-Fuller, 1999). As described earlier, pastoralists are thus either pushed onto more marginal lands for grazing or they begin to take up crop agriculture themselves, becoming agropastoralists (vide Campbell et al., 2000). In most cases, customary political and management systems are becoming weaker (Niamir-Fuller, 1999). Livestock development projects are also driving change in pastoral lands by opening up remote pastures with the spread of borehole technology and fragmentation of rangelands by veterinary cordon fences; this is true in eastern Africa, but particularly in southern Africa. Conflicts have resulted in changes in land tenure, with restricted access to traditional grazing lands as well as reduced mobility of pastoralists in insecure areas (Mkutu, 2001).
This contraction of pastoral grazing systems reduces the scale of resource use by pastoral peoples. Pastoral success depends largely on tracking patchy resources through time. In most traditional systems, this requires an opportunistic strategy of movement from daily and weekly changes in grazing orbits, to seasonal migrations over large landscapes. Many of the forces driving change in pastoral systems curtail the ability of pastoralists to move: sedentarization limits the maximal grazing distance achievable from a fixed homestead; privatization of land tenure limits access to many pastures; and gazetting of protected areas prevents pastoralists from reaching some pastures.
Maasailand in southern Kenya and northern Tanzania has been subject to considerable vegetation changes since the beginning of the twentieth century. Over the past century, the area has passed through successive stages of transformation as the result of the interaction between four distinct, and probably cyclical, processes of change: change in vegetation; climate; tsetse and tick infection; and pastoral occupation and management. At the end of the nineteenth century, Maasai pastoralists had access to extensive grasslands (Waller, 1990). During and following the great rinderpest epidemic of 1890, cattle populations in eastern Africa succumbed rapidly: by 1892, 95 percent had died. Famine and epidemics of endemic diseases such as smallpox reduced human populations to negligible numbers in Maasailand. Wild ruminants also died in great numbers due to rinderpest, but gradually developed immunity. By 1910, wildlife numbers rose, with the exception of wildebeest and buffalo, whose numbers were kept low from yearling mortality. These natural disasters disrupted the grazing patterns and reduced intensity. Dense woodlands and thickets established in the Mara Plains and northern Serengeti (Dublin, 1995) because fires were less frequent, since population decreased with the famine and there were fewer people to light fires, so fuel loads grew with less grazing offtake. This dense, woody vegetation was a habitat for tsetse flies, which fed on the abundant wildlife and prevented significant human re-settlement. Until the 1950s, Maasai chose to settle and graze away from the Mara Plains (Waller, 1990). At that time, the human population in the area was rapidly increasing and Maasai herdsmen used fire to improve grazing pastures (Plate 2.7) and to clear tsetse-infested bush. Increased elephant densities further maintained the woodland decline in the Maasai Mara and Serengeti as the animals moved to the protected areas from the surrounding, more densely inhabited areas. Between 1957 and 1973, woodlands in the Mara decreased from about 30 percent to about 5 percent cover (Lamprey and Waller, 1990). By the mid-1970s the wildebeest population had increased to about 1.5 million, and currently fluctuates around 1 million (Dublin, 1995).
Large areas of eastern African grasslands burn every year, providing short green regrowth for many species of livestock and wildlife in these ecosystems.
Over the past 25 years, considerable changes in land cover and land use have taken place in the Serengeti-Mara ecosystem and in the rangelands surrounding the protected core of the ecosystem (Serneels, Said and Lambin, 2001). The ecosystem is made up of protected land (Serengeti National Park, Ngorongoro Conservation Area (NCA) and several Game Controlled Areas in Tanzania, and Maasai Mara National Reserve in Kenya), surrounded by semi -arid rangelands that are largely inhabited by Maasai agropastoralists. Land cover changes leading to a contraction of the rangelands were most pronounced in the Kenyan part of the ecosystem, surrounding the Maasai Mara. About 45 000 ha of rangelands were converted to large-scale mechanized farming after 1975. Expansion of the wheat farms reached a maximum extent in 1997-8, at 60 000 ha. By 2000, about half of the wheat fields had been abandoned, mostly because the yields in the drier areas were too uncertain to make cultivation viable. The abandoned areas once more became available to livestock and wildlife. Permanent settlements have spread from the north to the south in the last 50 years, with significant settlement areas now on the northern border of the Mara Reserve (Lamprey and Waller, 1990). In the rangelands, most attempts at subsistence cultivation were abandoned after a few years, due to crop destruction by wildlife and highly variable yields linked with climate variability. In the Tanzanian part of the ecosystem, land cover changes were less pronounced. No conversion for large-scale farming occurred; most land cover changes were either expansion of smallholder cultivation or natural succession in rangelands. Extensive areas of cultivated land (subsistence to medium-scale agriculture) were found in the unprotected lands, right up to the border with the protected areas west of Serengeti and southeast of NCA. In the NCA and the Loliondo Game Controlled Area, about 2 percent of land cover changes were attributed to smallholder impact over the past 20 years. In the NCA, cultivation is regulated: only hand-hoe cultivation is allowed and fields are small and scattered. In the Loliondo, no such restrictions are in place, but the area is very inaccessible, so the lack of opportunities to export the crops outside the area effectively controls the extent of cultivation.
The conversion of rangelands to agriculture has had a serious impact on the wildebeest population in the Kenyan part of the Serengeti-Mara ecosystem. The population declined drastically over the past twenty years and is currently fluctuating around an estimated population of 31 300 animals, which is about 25 percent of the population size at the end of the 1970s. Fluctuations in the wildebeest population in the Kenyan part of the Serengeti-Mara ecosystem, over the last decades, have been correlated strongly with the availability of forage during the dry and the wet seasons (Serneels and Lambin, 2001). Expansion of large-scale mechanized wheat farming in Kenya since the early 1980s has drastically reduced the wildebeest wet-season range, forcing the wildebeest population to use drier rangelands or to move to areas where competition with cattle is greater. The expansion of the farming area has not influenced the size of the total cattle population in the Kenyan part of the study area, nor its spatial distribution. The much larger migratory wildebeest population of the Serengeti, in Tanzania, did not decline at the same time as the Kenyan population but is also regulated by food supply in the dry season (Mduma, Sinclair and Hilborn, 1999). Around the Serengeti, in Tanzania, land use changes are much less widespread, occur at a lower rate and affect a much smaller area compared with the Kenyan part of the ecosystem. Moreover, land use changes around the Serengeti have taken place away from the main migration routes of wildebeest.
Savannah ecosystems are well represented in African protected area networks (Davis, Heywood and Hamilton, 1994). In Tanzania, very large tracts of savannah have been set aside for conservation, partly because these rangelands support the most diverse assemblage of migrating ungulates on earth (Sinclair, 1995). However, there are few resources to manage these conservation areas effectively and the rural populations surrounding them are among the poorest in the world. Thus, conflict and complementarity between conservation and development have become major issues in Ngorongoro (Homewood and Rodgers, 1991), Mkomazi (Rogers et al., 1999), Selous (Neumann, 1997) and Tarangire (Igoe and Brockington, 1999).
Mkomazi Game Reserve in Northern Tanzania is a 3 200 km2 savannah area stretching from the Kenya-Tanzania border to the northeastern slopes of the Pare and Usambara mountains. Mkomazi lies within the Somali-Maasai regional centre of endemism (RCE) (White, 1983), where the dominant vegetation is Acacia -Commiphora bush, woodland and wooded grassland. Mkomazi borders the Afromontane RCE, with the lowland and montane forests of the Usambaras recognized as an outstanding centre of plant diversity (Davis et al., 1994), an endemic bird area (Stattersfield et al., 1998) and a centre of endemism for many other taxa (Rodgers and Homewood, 1982). This dry border ecotone position means that Mkomazi species richness may be enhanced not only by the presence of species primarily associated with the adjacent ecosystems, but also by divergent selection driving the evolution of new forms (cf. Smith et al., 1997). This diversity makes Mkomazi particularly valuable to opportunistic land users like pastoralists, but also for conservation of its rich species and landscape diversity. Based on the perceived species richness and concerns by the conservationists about the impacts on Mkomazi's vegetation of large numbers of cattle grazing in the western part of the reserve and large mammal populations, the resident pastoralists were evicted from the park in 1988 and use of its resources by the neighbouring communities was prohibited.
Mkomazi has been widely presented as undergoing ecological degradation prior to the 1988 evictions and recovery since then (e.g. Mangubuli, 1991; Watson, 1991). Data to confirm or refute that claim are as yet unavailable (Homewood and Brockington, 1999), but eviction was viewed as a risk -averse decision from a conservation point of view. However, from a pastoral point of view, the eviction did have serious impacts on the livelihoods of those who were evicted. Besides pastoral people, a large number of non-pastoral people also depended on the reserve for their livelihoods and used the reserve for beekeeping, collection of wild foods to supplement their diets or for sale at the local markets, and collection of fuelwood. Since the eviction, an estimated 25 percent of the livestock population have been restricted to a narrow and insufficient grazing area between Mkomazi reserve and the mountains bordering it to the south. Others have moved away from the reserve onto the increasingly crowded rangelands. Options for long-distance migration were greatly reduced, as the evictions occurred three years after the proliferation of large-scale commercial agriculture in northeastern Tanzania (Igoe and Brockington, 1999).
The impact of the evictions from Tarangire National Park, north-central Tanzania, shortly after its creation in 1968 was not felt immediately. There was no large-scale farming in the region at that time and pastoral Maasai were able to develop alternative, if less optimal, subsistence strategies. The effects became visible more than 20 years later, during the 1993/4 drought. By this time, some of the best wet-season pastures in Simanjiro District had been lost to large-scale commercial agriculture and more livestock were forced onto the dry -season grazing grounds in the early grazing season, depleting the season's grass growth sooner. The Maasai of Simanjiro found previous drought-coping strategies precluded by loss of access to drought reserve areas which had been enclosed inside the Tarangire National Park or allocated to large-scale commercial farms (Igoe and Brockington, 1999).
The examples of Mkomazi and Tarangire clearly point to costs and benefits in conservation decisions, with conflicts likely to intensify as human needs grow.
Wetter grasslands and woodlands have also evolved rapidly in the last century. One cause of that change is the control of trypanosomiasis, the disease transmitted to livestock and people by the tsetse fly, which allowed farmers to use animal traction more extensively (greater numbers and more healthy oxen) and thus to expand the amount of land they cultivated at the household level (Jordan, 1986). Despite the logic of this progression, Ghibe Valley in Ethiopia is one of the few places in Africa where these changes have been seen clearly (Reid, 1999; Bourn et al., 2001)
Ghibe Valley is located about 180 km to the southwest of Addis Ababa, where the main road to Jimma descends from the Ethiopian highland massif. Tsetse flies were first controlled in this area in 1991, using pesticide-drenched targets and pesticide poured on the cattle themselves. Within this landscape in 1993, just after the control, the majority of the arable land was wooded grassland used by wildlife and the few livestock herded by agropastoral peoples. Smallholder farms covered about a quarter of the arable land, while large-holder farms covered less than 1 percent (Reid et al., 1997). About 90 percent of this landscape supports soils that are moderately to highly suitable for agriculture. Smallholder farmers grow a diversity of crop types, including maize, sorghum, tef, noug or niger seed (Guizotia abyssinica), false banana (Ensete ventricosum), groundnuts, wheat, beans and hot peppers, while large-holders grow a number of crops for market (citrus, onions, maize, spices). People use the large uncultivated tracts of grassland and woodland for settlements, hunting, wild plant gathering, beekeeping, livestock grazing, firewood collection, charcoal making and woodlot cultivation.
Rapid land use and land cover change was caused by the combined effects of drought and migration, changes in settlement and land tenure policy, and changes in the severity of trypanosomiasis (Reid et al., 2000b; Reid, Thornton and Kruska, 2001). Each cause affected the location and pattern of land use and land cover in different ways. Previous to the control, a strong increase in the severity of the trypanosomiasis caused massive loss of livestock, farmers were unable to plough as effectively and the area of cropland contracted by 25 percent. Changes after tsetse control were slow to appear on the land itself, with nearly a five-year delay in impact on land use, although there was a more immediate impact on livestock health and populations. Changes were bi-direc-tional and varied in speed, with both intensification and dis-intensification (Conelly, 1994; Snyder, 1996) occurring within the same landscape, sometimes slowly and sometimes rapidly.
These changes in land use caused profound changes in ecological properties and the structure of the valley's ecosystems (Reid et al., 2000b). When land use expanded, large areas of woodland were cleared for cultivation and firewood became more scarce. As human populations grew, plants with medicinal value became more rare and the large herds of grazing herbivores were decimated. Most of the biodiversity in the valley is limited to the narrow ribbons of woodland along the rivers; it is these rich woodlands that farmers began to clear after successful tsetse control (Reid et al., 1997; Wilson et al., 1997).
Management regimes for the grasslands of eastern Africa generally fall into three types: (1) state-managed for tourism and ranching; (2) commercial use for livestock or crop production; or (3) traditional management by pastoral and agropastoral groups. Livestock production, particularly cattle, is the major use for rangelands, with over 100 million head of livestock in the rangelands of eastern Africa (Herlocker, 1999). There is also a growing market for meat from wildlife, which is being met through commercial ranching and culling in the region. Grassland management is linked to use by livestock and wildlife, and there is often conflict between their exploitation for commercial income generation and the more sustainable management regimes of traditional groups. Wildlife -based tourism is of particular importance for generation of state, private and community income in the rangelands of Kenya (Plate 2.8) and Tanzania (Myers, 1972) and to a lesser extent in Uganda and Ethiopia. Recent efforts to privatize land and introduce more livestock are also changing the way people interact with wildlife.
Government development projects have focused on improving the productivity of the rangelands and increased livestock production from common property resources. The World Bank has sponsored several projects on rangeland management, including in Somalia, Kenya and Ethiopia. Earlier projects focused on increasing the productivity of rangelands for livestock production and several included formation of pastoral associations, which dealt with grazing rights and policies. These projects had disappointing results due to the parastatal organizational form, inappropriate technologies and poor appreciation of traditional systems and people. Only recently have issues of integrated natural resource management and full involvement of stakeholders been given attention, although there remain problems in reaching the local people through public sector organizations (de Haan and Gilles, 1994).
Acacia bushland near Nakuru, Kenya, supports endangered Rothchild's giraffe.
Traditional management systems by pastoralists recognized the need for controlled access to conserve the biodiversity and allow the rangeland to recover. Traditional grazing systems are more effective for sustainable resource use and maintenance of rangeland condition (Pratt and Gwynne, 1977). However, the traditional systems are under threat from increased livestock populations and decreased grazing lands, resulting in increased grazing pressure. This is already being recognized by Boran pastoralists in Ethiopia, who perceive that the condition of the rangelands is poor compared to 30 to 40 years ago (Angassa and Beyene, 2003) and consider the rangelands degraded and their livestock production declining.
Annual variation in amount and distribution of rainfall, together with grazing, fire and human activities, results in wide variation in grassland productivity (Walker, 1993). Rangeland ecosystems are very resilient and recover well when there is sufficient rainfall and controlled use of the resources. Range condition is dependent on both the grazing system, considered as timing and frequency of grazing, and grazing intensity, defined as the cumulative effects grazing animals have on rangelands during a particular period (Holechek et al., 1998). Grazing intensity is closely associated with livestock productivity, trends in ecological conditions, forage production, catchment status and soil stability. It is considered as a primary tool in range management, and flexibility of grazing intensity is critical to rangeland ecosystem health. Grazing intensity has a major impact on range condition (Plate 2.9). In a recent study in Ethiopia, range condition, based on the herbaceous layer, basal cover, litter cover, relative number of seedlings, age distribution of grasses, soil erosion and soil compaction, was higher in lightly grazed areas in the Rift Valley than in the heavily grazed communal lands (Sisay and Baars, 2002). In Serengeti and Maasai Mara, grazing was found to stimulate net primary productivity at most locations, with maximum stimulation at intermediate grazing intensities and declines at high levels of grazing. Stimulation was dependent upon soil moisture status at the time of grazing (McNaughton, 1985).
Heavily grazed grassland in the highlands near Bule, Ethiopia.
There are many examples in the literature of the impact of management on species composition and diversity (Herlocker, 1999). In a study in the rangelands of southern Ethiopia, perennial grasses were relatively resilient in terms of cover and productivity in response to grazing, while continuous grazing encouraged forbs with lower grazing value for cattle (Coppock, 1994). Grazing affects species diversity and richness in grasslands (Oba, Vetaas and Stenseth, 2001). Optimal conservation of plant species richness was found at intermediate levels of biomass production and was found to decline if biomass increased in ungrazed areas of arid -zone grazing lands in northern Kenya. These intermediate levels can be achieved by manipulating management and grazing pressure, although there will be seasonal fluctuations due to environment. In the Serengeti Plains in Tanzania, elimination of grazing led to dominance by tall vegetatively propagated grass species, while the short sexually reproduced species disappeared (Belsky, 1986). This implies that even though the greatest number of species are found at intermediate grazing intensities, some species are always lost when ungrazed pastures are grazed. Although there are more species at intermediate levels of grazing, it is possible that any grazing negatively affects rare plant species that are sensitive to grazing.
One of the most controversial and debated aspects of research about pastoral systems is the existence and extent of overgrazing, desertification and land degradation in pastoral lands, particularly in Africa. Global assessments of drylands maintain that much of the earth's land surface is degraded (GLASOD, 1990) and that livestock are the principal global cause of desertification (Mabbutt, 1984). Analysts suggest that African pastures are 50 percent more degraded than those in Asia or Latin America (GLASOD, 1990). However, other analyses show that livestock numbers only exceed likely carrying capacities of arid and semi -arid rangelands in about 3-19 percent of Africa (Ellis et al., 1999). In addition, there is no sustained evidence for a reduction in productivity, as measured by no change in the water-use efficiency of the Sahelian vegetation over 16 years, suggesting that the extent of the Sahara is more strongly influenced by drought than grazing (Tucker, Dregne and Newcomb, 1991; Nicholson, Tucker and Ba, 1998).
However, these broad assessments are only correlative and can not assess cause and effect rigorously, and can not measure the relative impacts of different causative agents. Certainly, at more local scales, livestock impacts are highly visible and persistent around towns, water points and along cattle tracks (e.g. Georgiadis, 1987; Hiernaux, 1996). More illuminating - but much more difficult to acquire - are two types of evidence: (1) remote sensing studies at the landscape scale, tracking the extent of degradation or desertification and of sufficient duration to cover drought and non-drought periods; and (2) either long-term experiments or observational studies at the pasture scale that assess the relative impacts of drought, livestock and other agents on ecosystem dynamics. Most of the research on land degradation and desertification in Africa has been focused on the Sahel, which has been intensively studied since the first droughts in the 1970s. Some landscape-scale studies based on a combination of fine-resolution satellite data and field measurements have been carried out in different parts of Africa and demonstrated the existence of local-scale degradation of rangelands in Ferlo, Senegal (Diouf and Lambin, 2001) and Turkana, Kenya, where highly affected areas covered only 5 percent of the land surface (Reid and Ellis, 1995). The importance of studying land degradation and desertification problems over a sufficiently long period is illustrated by studies conducted in Burkina Faso by Lindqvist and Tengberg (1993) and later by Rasmussen, Fog and Madsen (2001). The first group of scientists studied the amount of woody vegetation cover in three sites in northern Burkina Faso (1955-89). They found that important loss in woody vegetation cover occurred during the first of a series of droughts that started in the late 1960s, when large areas of bare soil developed. The authors found little evidence of vegetation recovery until 1989, despite increasing rainfall since 1985. Rasmussen, Fog and Madsen (2001) revisited the area, adding 10 years of satellite data. They found a decrease in albedo and thus increase in vegetation cover over the period 1986-1996, which was confirmed by fieldwork, thus showing the recovery of vegetation after drought. Interviews with local people indicated that the species composition of the regenerating herbaceous vegetation has changed considerably since the late 1970s. Land degradation caused by heavy grazing pressure was mostly found in the proximity of important water resources, which probably cover only a small proportion of the landscape in total. Schlesinger and Gramenopoulos (1996) also used the amount of woody vegetation cover as an indicator of desertification in western Sudan, but studied sites that were devoid of human use over the period 1943-1994. They analysed a time series of aerial photographs and Corona satellite images and did not find a significant decline in woody vegetation for the study period, despite several droughts having occurred during that period. Thus, at least in this area, they showed that the Sahara is not expanding and that drought had little effect on woody vegetation. In another part of the Sudan, the concentration of livestock around water points and settlements led to local loss of vegetative cover and accelerated erosion (Ayoub, 1998). In contrast, others have found that heavy livestock grazing around pastoral settlements in arid areas had minor impacts on woody vegetation and biodiversity, with impacts confined within the settlements themselves (Sullivan, 1999). Woody vegetation can replace palatable grass species in heavily grazed areas, caused by grazing pressure rather than climate (Skarpe, 1990; Perkins, 1991).
At the pasture or field level, the picture is more complex. Generally, livestock grazing, browsing and trampling causes loss of vegetation, competition with wildlife and sometimes a change in soils when use is prolonged and heavy. The impacts depend to some degree on the level and variability of rainfall (Ellis and Swift, 1988). In areas of southern Ethiopia with more and reliable rainfall supporting perennial vegetation (an equilibrium grazing system), the impacts of grazing in one season can reduce vegetative cover and production in the next (Coppock, 1994). In systems on the edge of perennial grass production, heavy grazing, especially in combination with drought, can reduce vegetative cover and production, even during subsequent wet years when more lightly grazed areas recover fully (de Queiroz, 1993). In systems with low and erratic rainfall (non-equilibrium systems), heavy grazing may (Milchunas and Laurenroth, 1993) or may not (Hiernaux, 1996) strongly influence production in subsequent seasons. Heavy grazing in annual grasslands changes the species composition of grassland vegetation, with more species in areas protected from grazing and fewer in heavily grazed areas (Hiernaux, 1998).
The nature of interrelationships and thresholds between biophysical, socio-economic, institutional and policy factors at different spatial scales and temporal dimensions influencing land degradation and desertification are still poorly understood. A recent initiative on land degradation assessment in drylands (LADA project), executed by FAO, responds to the need for an accurate assessment of land degradation in drylands at a flexible scale and to strengthen support to plan actions and investments to reverse land degradation, improve socio-economic livelihoods, conserve dryland ecosystems and their unique biological diversity (see: http:/www.fao.org/ag/agl/agll/lada/home. stm). Besides developing a set of tools and methods to assess and quantify the nature, extent, severity and impacts of land degradation on ecosystems, catchments, river basins and carbon storage in drylands at a range of spatial and temporal scales, the project also aims to build national, regional and global assessment capacities to enable the design and planning of interventions to mitigate land degradation and establish sustainable land use and management practices (Nachtergaele, 2002).
Another useful livestock evaluation tool to enhance early warning systems to detect changes in livestock condition is being developed under the USAID Global Livestock Collaborative Research Support Program (CRSP) by Texas A&M University (Corbett et al., 1998). The Livestock Early Warning System (LEWS) integrates advanced crop and grazing models, based on empirical relationships between weather, vegetation, regrowth potential, soil and climate dynamics, with near infra-red spectroscopy (NIRS) for faecal analysis to detect changes in diet of free ranging livestock. These changes are linked to changes in vegetation patterns and can be used to predict drought and feed shortages for livestock some 6 to 8 weeks before pastoralists begin to see changes in the condition of the rangelands and their livestock. This allows them to better prepare for the coming feed shortages and nutritional crises in a timely manner by transhumance, as well as avoiding overgrazing of the rangeland resources.
 Growing days are defined as
the period (in days) during the year when precipitation (P) exceeds half
the potential evapotranspiration (PET) plus a period required to evapotranspire
up to 100 mm of water from excess precipitation assumed stored in the soil
profile (FAO, 1978). The mean daily temperature during the growing period
has to exceed 5°C (Fischer, Velthuizen and Nachtergaele,
 In this chapter, we consider degraded land to be land that due to natural processes or human activity is no longer able to sustain an economic function or the original ecological function, or both (GLASOD, 1990).
 Intitially established in the late 1940s as National Reserves under Kenya Royal National Parks, these areas were once again redesignated as national reserves under the 1976 Wildlife (Conservation and Management) Act. However, they continued to be managed by county councils.
 Definitions of desertification, overgrazing and degradation are controversial and problematic. We use the term desertification here to refer to the concepts used by the cited global assessment (GLASOD, 1990). We use degradation to mean an irreversible change in ecosystem state or function. We agree with de Queiroz (1993) that degradation has been defined relative to human management objectives and thus is relative; for example, a change from grassland to bushland is degradation to a cattle -keeper but may be aggradation from the perspective of carbon sequestration. We would prefer a set of quantitative measures that can assess the state of a particular piece of land and be used by land managers to assess the desirability of different changes from the context of their own management objectives. We define overgrazing here as any level of herbivore grazing that induces either temporary or permanent changes in the species composition or function of a grassland.
 Carrying capacity is exceeded in 19% of areas receiving 0-200 mm rainfall per annum, 15% of the 200-400 mm zone, 3% of the 400-600 mm zone and 8.5% of the 600-800 mm zone (Ellis et al., 1999).