Environmental change in the Kalahari: Integrated land degradation studies for non equilibrium dryland environments
Andrew J. Dougill*, David S.G. Thomasy1 and A.Louise Heathwaitey
* - The Environment Centre, University of Leeds, Leeds, LS2 9TJ, UK.
y - Sheffield Centre for International Drylands Research (SCIDR), Department of Geography, University of Sheffield, Sheffield, S10 2TN, UK.
1- author for correspondence
Recent decades have seen major intensification of cattle-based agricultural production in semiarid savanna ecosystems. In the Kalahari of Botswana, cattle production now occurs on privatized and fenced ranches. Patterns of ecological change, notably increased bush dominance, have been linked to increased cattle grazing intensity, but it remains contentious whether these changes represent land degradation. Uncertainty in ecological understanding stems from the dynamic, "non equilibrium" functioning of semiarid ecosystems. Given the inherent ecological variability of drylands, it is argued that degradation assessments should be based not on ecological observations alone, but on the study of long-term changes in pastoral production figures and on changes in the ecologically-determining factors of soil water and soil nutrient availability. Provided here is a framework incorporating soil and ecological changes at a range of scales that can enable us to distinguish drought-induced fluctuations from long-term ecological state changes. The results demonstrate that increased cattle use and associated ecological changes have not been caused by, nor are associated with, changes in soil water and nutrient availability. A model of ecosystem dynamics is presented that displays bush encroachment not as a definite form of land degradation. Encroachment may also be curtailed by resilience mechanisms found in protected ecological niches and also by the resilience of the nutrient poor sandy soils. Key words: Land degradation, Dryland pastoralism, Ecological modeling, Resilience, Bush encroachment, Soil hydrochemistry, Kalahari
Issues of contemporary environmental change in dryland areas, and in particular their association with desertification, remain controversial in many respects due to their multi-faceted social and environmental dimensions. Desertification does, however, remain high on the global environmental agenda, as witnessed through the ratification of the United Nations Convention to Combat Desertification (CCD), and continued global efforts to assess the extent of desertification (Middleton and Thomas 1997). The CCD defines desertification as "land degradation in arid, semiarid and dry subhumid areas resulting from various factors, including climatic variations and human activities" (Middleton and Thomas 1997:viii). In this context, land includes the soil and water resources, as well as local vegetation cover upon which rural agricultural livelihoods are dependent. Environmental studies therefore need to be wide-ranging in their remit to produce a clear understanding of the nature and extent of desertification problems. The term degradation implies a reduction of the resource potential by one or a number of processes operating on the land (UNEP 1997). These definitions provide the context for contemporary environmental studies in drylands, implying that changes in the natural resource base of vegetation and soil systems can only be classified as desertification if they reduce the land’s value to human systems. One study has defined desertification in this context as "an effectively permanent decline in the rate at which land yields agricultural products under a given management system" (Abel and Blaikie 1989: 113).
Conventional media images showing desertification as extensive wind or water erosion of soils serve to over-simplify and often confuse the most important environmental and social issues. Much academic debate has focused on how these images have exaggerated both the physical extent and the social impacts of dryland degradation, leading to the alternative view that desertification is an institutional myth (Warren and Agnew 1988, Hellden 1991, Thomas and Middleton 1994). Less debate, concern, and publicity has been afforded to changes in the dryland ecological resource-base (the savanna vegetation communities) which are less overt, and yet are responsible for much of the reduced agricultural potential, especially in areas used for intensive cattle ranching. The most notable example of this, and that investigated here for the Kalahari of southern Africa, is that of bush encroachment, the ecological transition to a state of increased bush dominance. This transition reduces grass cover in the landscape and can reduce economic outputs from grazing-based farming systems (de Ridder and Breman 1993, Quan et al. 1994), as both the total fodder availability and its heterogeneity are reduced (Scoones 1995).
Bush encroachment has now become recognized as a threat to sustainable pastoralism in all major savanna regions throughout the globe. Indeed, Warren and Agnew (1988: 6) state "it has often been maintained that the most widespread problem on (dryland) pastures is not their complete degradation but their invasion by thorn scrub." Previous environmental studies of this phenomenon have implicitly linked it to the intensity of cattle grazing, with bush encroachment being associated with areas of intensive cattle use (e.g., Buffington and Herbel 1965, van Vegten 1983, Andrew 1988, Perkins and Thomas 1993a,1993b). Increased awareness of the existence of bush encroachment has not, however, been paralleled by advancements of our understanding of the underlying processes of change in relation to the controlling factors of rainfall variability, changes in fire regimes, the direct impacts of cattle grazing, nor the interrelationships between these factors. Yet without such environmental understanding, preventative and control measures aimed at reducing bush encroachment will remain difficult to provide.
A number of different theoretical models have been proposed to explain the transition to greater bush dominance in savannas (e.g., Walker and Noy-Meir 1982, Westoby et al. 1989, Schlesinger et al. 1990). This paper starts by outlining differences in these ecological change models before assessing their applicability to, and implications for, privately owned cattle ranches on the flat extensive sandy plains of the Kalahari. The wider significance of these studies to other semiarid regions with similar environmental and land use characteristics, such as in Australia (Andrew 1988, Pickup 1995), Sahelian Africa (Warren and Khoghali 1992), and the southwest USA (Schlesinger et al. 1990, Archer 1990) are highlighted to demonstrate the wider applicability and complexity of current geographical debates associated with changes in the dryland ecological resource base. These debates are extended to examine the future role, and suggested methodologies for, integrated environmental monitoring of changes in soil and vegetation systems aimed at land degradation assessments, and for improved incorporation of environmental studies into the development of sustainable land use practices.
Ecological change models
Theoretical discussions about the causes of bush encroachment have developed as part of the rapid evolution of dryland research and the recent paradigm shift recognizing the "non-equilibrium" nature of ecological functioning (Behnke et al. 1993, Scoones 1995, Warren 1995). Due to the inherent rainfall variability and patchy fire regimes of drylands, there is now a recognized need to place short-term environmental studies within the context of the natural environmental variability that characterizes such areas. The environmental paradigm shift has important implications for determining what constitutes degradation and for models attempting to explain the interrelated causes of ecological transitions. Ecological change models require adaptation to incorporate the effects of variability in rainfall and fire regimes, factors which now form the focus of dryland ecological studies. In this section we discuss current understandings of the causes of bush encroachment in drylands, and detail the more comprehensive integrated studies needed to adapt environmental studies to the dynamic, highly variable nature of non-equilibrium theories.
Conventional models derived to explain the processes leading to bush encroachment can be divided into two main groups: first, those that suggest bush encroachment occurs solely due to changes in grazing and fire regimes acting independently from soil factors (e.g., Westoby et al. 1989), and second, those that act through changes in the soil resource (e.g., Walker and Noy-Meir 1982, Schlesinger et al. 1990). Given the ability for agricultural land use practices to modify grazing and fire regimes in relation to the environmental conditions experienced at a given time and place, this distinction has important implications for the provision of sustainable natural resource management strategies. Such adaptations have been termed "opportunistic management" (Westoby et al. 1989) and have been heralded as a flexible and sustainable way to manage natural resources through coordination between land use planners and local land managers (Scoones 1995). Adaptations such as rapid destocking in times of drought to prevent grass mortality (Mott et al. 1992) and the use of wet season fires to reduce bush dominance (Hodgkinson et al. 1984) will only be successful in controlling bush encroachment if it is these pressures alone that are the causal agents.
The aim and importance of soil-based studies, on the other hand, is to determine whether bush encroachment occurs at least partly because of changes in the soil resource, including the depth profile and spatial distributions of soil water and nutrient availability (Walker and Noy-Meir 1982, Schlesinger et al. 1990). If the ecological distribution of bushes has a soil hydrochemical causation independent of changes in the grazing and fire regimes, it suggests that the ecological dominance of bushes can remain in the landscape, then bush encroachment represents a permanent decline in the resource base for pastoralism. This is especially true given the findings of longer term ecological studies showing that bush species can recover rapidly from disturbances that cause die-back (Whitford et al. 1995). Consequently, the recent shift in emphasis of dryland degradation studies to consider the impact of management strategies on soil factors (e.g., Parr et al. 1990, Oldeman et al. 1990, Stocking 1995, Middleton and Thomas 1997) is important, and provides a context for the research reported here. This article builds on previous studies by moving away from investigating visible attributes of soil degradation to consider patterns of soil water and nutrient availability. Importantly, it is these factors that are recognized as the primary controls on the productivity and vegetation community structure (i.e., the ratio of bush to grass cover) within dryland savanna ecosystems (Scholes and Walker 1993). In this respect the "two-layer model of environmental change” (Walker and Noy-Meir 1982) has been widely used as the explaination of bush encroachment in the Kalahari (Skarpe 1990, 1991, Perkins and Thomas 1993a, 1993b) and in other regions. This model is now described, before moving to consider recent ecological developments that are important in land degradation debates.
The two layer model
In this model, the ecological balance between bush and grass production is determined by the relative availability of soil water and key nutrients (principally nitrogen (N) and phosphorus (P)) in different rooting zones in the soil profile. Ecologically, grasses out-compete bush species for water and nutrients in the topsoil (0 - 50 cm depth), while bush species have the competitive advantage in the subsoil below 50 cm (Walker et al. 1981, Belsky 1990). According to the two layer model, cattle grazing effects the bush:grass ratio by suppressing grass growth, and therefore topsoil plant water uptake, promoting soil water movement into the subsoil. At the same time, the increased mineralization of organic nitrogen inputs into the water soluble inorganic form of nitrate (NO3--N), caused by the fact that the surface inputs of cattle dung are more readily decomposed than residual plant litter, enhances leaching of this vital plant nutrient into the subsoil. Consequently, the two layer model predicts that areas of intensive grazing experience significant increases in moisture content and "plant-available" inorganic nutrient concentrations in their subsoil layers (Figure 1), with bush encroachment being the ecological consequence. The model implies that bush encroachment is linked to major soil hydrochemical changes that favor bush dominance into the future, irrespective of changes in grazing and fire regimes. However, changes in the nutrient mineralization and leaching processes that the model is dependent upon have not been observed in controlled laboratory or field process-based studies on Kalahari sand soils (Dougill et al. 1998a) leaving uncertainty about the actual extent of the soil-to-ecological-change links proposed by the two-layer model. Research in this paper examines the links between patterns of ecological change and soil profile hydrochemical factors through ranch scale measurements at sites experiencing different grazing intensities and ecological community structures.
New ecological theories and land degradation assessments
"Non-equilibrium" ecological theories are founded on the identification of the inherent dynamism of dryland ecosystems, where rainfall plays a greater role in plant growth than variations in grazing regimes (Holling 1973, Wiens 1977, Sandford 1983, Ellis and Swift 1988, Friedel et al. 1993). Vegetation changes are often reversible, such that even systems classified as severely degraded demonstrate rapid recovery characteristics, especially following the return of good rainfall after drought years (Tucker et al. 1991, Perkins and Thomas 1993a). Such studies suggest that semiarid ecosystems are well adapted to cope with, and to respond to, disturbance, due partly to the continuous natural (rainfall and fire) disturbances upon which their original community composition is dependent. This recent view of dryland ecosystems as resilient systems directly opposes the traditional representation of semiarid ecosystems as fragile "degradation prone" environments (Stoddart et al. 1975, Kharin 1990). Rather than grazing disturbances being automatically seen as necessarily increasing system instability and enhancing the risk of desertification, more integrated understanding of the causes and permanence of transitions between vegetation states is required. In particular, environmental monitoring of changes in vegetation and soil resources is essential to distinguish between drought-induced ecological fluctuations and permanent changes in vegetation community state.
Non-equilibrium theory clearly dismisses the applicability to dryland savannas of traditional "climax and equilibrium" ecosystem models (Behnke et al. 1993, Warren 1995), necessitating alternative models of ecosystem functioning for these areas. Westoby et al. (1989) have formulated these, as "state and transition" models. These models suggest that dryland ecosystem dynamics can be described by a set of discrete states of the vegetation community, and a series of distinct transitions between states. State and transition models imply that environmental changes can occur very rapidly and may be triggered either by management actions, such as increased cattle stocking levels or managed burning, and/or by natural events including rainfall variability and fire. Such models therefore better represent the natural complexity of interrelations experienced in drylands, but they pose difficulties for the environmental scientist attempting to maintain clarity and simplicity in degradation debates, as well as for land managers seeking to intervene in the land use decision making process (Thomas 1997). It is, however, essential that the exact conditions and processes leading to bush encroachment in particular environments be fully understood so as to prevent declines in agricultural productivity.
The acceptance of non-equilibrium ecological theory has a number of important implications for limited time-scale and small spatial scale studies that typify environmental investigations used to assess the existence of land degradation in dryland areas (Middleton and Thomas 1997). The main implications are summarized in Table 1, which details the implications of theories in terms of the need to move away from degradation assessments based solely on vegetation community changes, and also the specific implications for the findings of previous Kalahari-based environmental studies. Non equilibrium concepts need to influence the manner in which contemporary environmental studies are designed, with there being an increased focus on understanding the interrelated effects of rainfall, grazing, and fire regimes in affecting the ecological and soil resource bases. As Table 1 describes, this can be achieved by extending the time frame of ecological studies, by incorporating clear recognition of different spatial patterns of ecological change in relation to patterns in the driving forces of grazing, rainfall, and fire regimes, and/or by assessing the impact of grazing on soil hydrochemical factors. This paper represents a case study detailing how these implications have been incorporated into integrated environmental studies, in this case for eastern Kalahari ranches in Botswana.
The Kalahari desert, thirstland, or sandveld, as it has variously been termed, can be defined in a number of ways using a range of criteria (Thomas and Shaw 1991). Here it is designated as the area covered by deep Kalahari Sand deposits, centered upon western and central Botswana but extending into neighboring countries (Figure 2). It is characterized by semiarid climatic conditions (Bhalotra 1987) and typified by very limited relief. Geologically, the Kalahari is an extensive basin infilled with nutrient deficient aeolian sediments (Thomas and Shaw 1990). Kalahari soils are therefore predominantly deep, structureless, fine sands, low in essential plant nutrients (Skarpe and Bergström 1986, Buckley et al. 1987a, 1987b) and with a low organic matter content (Perkins and Thomas 1993b). The high infiltration rates afforded by the sandy sediments, together with the lack of perennial rivers, make the Kalahari a location where natural water availability is limited to small short-lived accumulations in occasional pan depressions. It is this lack of surface water, not the climatic characteristics, that lead to its portrayal as a desert.
Despite the absence of surface water, the mixed grass and shrub savanna (Weare and Yalala 1971) have resulted in the view that the Kalahari is an untapped grazing resource (Debenham 1952). Pastoral utilization of this ecological resource base has intensified since the 1950s as a result of the large-scale sinking of deep boreholes to exploit ancient groundwater reserves in the underlying sandstone aquifers (de Vries and von Hoyer 1988). The nature of this intensifying development program has been clearly linked to a series of national policy-driven structural changes in the agricultural economy.
Botswanan agricultural development
Since independence in 1966, the government of Botswana, supported by a range of international development aid and trade agreements, has voraciously pursued policies of expanding privatized "ranch-style" livestock production in the Kalahari. Successive agricultural policies have aimed to relieve pressure on the country’s more densely populated eastern hardveld, where land degradation caused by high livestock numbers is perceived to have caused environmental damage (see Abel and Blaikie 1989 and Biot 1993 for discussion). The first Livestock Development Project (LDP1), begun in 1970, established forty ranches in the Kalahari as a means of encouraging herd owners to commercialize their activities (Odell 1980). In 1975, the LDP1 was subsumed under the wide-ranging Tribal Grazing Lands Policy (TGLP), which aimed to enhance national economic and social development by reorganizing Botswana's livestock industry. In so doing, it was hoped to both improve the livelihoods of rural dwellers and to reduce the environmental degradation attributed to "overstocking" under the traditional communal-grazing systems in the eastern hardveld.
Development policies in Botswana have matched those in many other parts of dryland Africa (Lane 1998) by using the well publicized principle of the "Tragedy of the Commons" (Hardin 1968) to fence off and privatize large portions of once communal land. The TGLP aimed to expand cattle production in the Kalahari through the creation of fenced leasehold ranches of a uniform size, each supplied with groundwater via a borehole. Ranches were to be leased from the state by cattle owners or syndicates who owned over 400 animals. Lessees would gain sole water rights for the groundwater boreholes and therefore de facto control of grazing and ecological resources within a ranch. It was hoped that by moving large livestock owners off communal lands, overgrazing would be reduced, giving the smaller farmers who remained the opportunity to improve their livelihoods. Overall, therefore, the TGLP was intended to alleviate environmental and social problems in one part of the country by encouraging development elsewhere. The social applicability of these policies have now been widely criticized (Campbell et al. 1991, Tsimako 1991, Peters 1995, Thomas and Sporton 1997), due largely to the inapplicability of assumptions underlying the TGLP. These included the view that the Kalahari was an uninhabited wilderness and that, with the allocation of private ranches, large cattle owners would no longer graze their cattle on communal lands. In reality, many designated areas were on lands used by rural peoples for seasonal grazing for over 200 years (Campbell et al. 1991, Moyo et al. 1993). In addition, ranch lessees have continued to use their dual grazing rights on both private and communal lands, thereby exacerbating social inequalities in terms of access to grazing land and livestock (Good 1992, Peters 1995). The social implications of the TGLP therefore include reduced status and increased poverty among rural dwellers (Thomas and Sporton 1997). What remains less clear is the extent and permanence of ecological changes on TGLP ranch blocks and consequently the long term environmental sustainability of the ranching practices promoted. It is these factors which form the focus of the integrated environmental study reported here.
The framework for expansion of privatized ranches provided by the TGLP was strengthened further by the 1991 National Policy on Agricultural Development (Republic of Botswana 1991). This called for further increases in livestock productivity through improved management, and continued commercialization, the establishment of additional ranches and, significantly, the fencing of all communal lands. Given the uncertain environmental consequences of this move to complete privatization, it is vitally important to assess the environmental impacts of the 25 years of intensive grazing experienced on TGLP ranches. The changes seen on these ranches will act as an indicator of the threats to sustainable natural resource management faced in the future, further privatized, Kalahari.
Study site: The Makoba ranches
One of the original blocks of leasehold ranches, established by the TGLP in 1975, is the Makoba ranch block, in the eastern sandveld of the Central District (Figure 2). This area has a mean annual rainfall of approximately 400 mm. The inter-annual variability of over 35 percent (Bhalotra 1987) greatly regulates ecological changes. At the time the ranches were designated, vegetation communities consisted of mixed savanna with grassy plains dominated by tufted perennial grasses, including Eragrostis, Pogontharia and Stipagrostis species, and low shrubs, commonly of Acacia, Lonchocarpus, Grewia, Rhigozum, and Terminalia species (Perkins and Thomas 1993b). Each ranch covers a standard area of 6400 hectares and was initially served by a single, diesel-pumped, borehole obtaining groundwater from over 60 meters depth. Up until the early 1990s, the ranches were indistinguishable in their operation from traditional “cattleposts,” with the absence of any paddocks or perimeter fencing, implying cattle were free to roam throughout the rangeland surrounding the borehole. Therefore, ecological studies that began on the Makoba ranches in 1989 (Perkins and Thomas 1993a, 1993b) initially assessed the impacts of the shift to a more intensively grazed cattle-based system, in which the single borehole is effectively "the herder" (Jerve 1982). Given the increasing available grazing area with distance from a point water source grazing intensity therefore declines exponentially in all directions away from the borehole. This pattern of grazing intensity decline enables direct study of the relationship between grazing intensity and changes to soil or vegetation resource bases, a technique termed the "piosphere" approach (Georgiadis 1987, Andrew 1988).
Ongoing studies throughout the 1990s have taken place through a period of further structural changes, in line with the aims of the National Policy on Agricultural Development in the Makoba ranches. This has seen further boreholes sunk on ranches and the partitioning of ranches into paddocks that are then used for rotational grazing in a manner similar to conventional western-world ranching schemes. One such ranch, Uwe Aboo, has been the site of detailed environmental studies throughout the 1990s. Uwe Aboo has seen two new boreholes drilled and the ranch has been partitioned into five paddocks (Figure 3) since the initial vegetation surveys of Perkins and Thomas (1993a). These additional development pressures and the spatial and temporal intensification of cattle grazing they cause need to be considered throughout our discussion of the results from our environmental studyof TGLP ranches.