Definitions and key concepts
Apart from desertification itself, several important concepts are mentioned in this review. For example: "First of all it is important to recognise that dryland ecosystems are inherently non-equilibrium systems and ecosystem dynamics are essentially event-triggered. Most disturbances, such as rainfall variability and fire, are incorporated in dryland ecosystems during their evolution. However, some disturbances are new or not yet incorporated and may drive the system to qualitatively different new states along irreversible trajectories" (Puigdefábregas, 1998). These concepts are explained here. Also the concept of land degradation and the extent of drylands are discussed briefly here.
Desertification
Over a hundred formal definitions of desertification have been proposed, covering many spatial and temporal scales and representing different viewpoints (Thomas, 1997; Reynolds and Stafford Smith, 2002). The term desertification was first used in West Africa in 1949 by a French forester to describe the way in which it was perceived that the Sahara Desert was expanding to encompass desert-marginal savannah grasslands (UNEP, 1997). Afterwards it was realised that desertification was not only happening in Africa but in dryland areas (see 1.2.3) worldwide (UNEP, 1997). In 1992, UNEP defined desertification as 'land degradation in arid, semi-arid, and dry sub-humid areas resulting mainly from adverse human impact' (UNEP, 1992). It was recognised that such changes should be "effectively permanent" (Abel and Blaikie, 1989), distinguishing it from short-term, reversible changes such as drought. It should be noted that while many forms of environmental change are theoretically reversible over short time-frames (e.g. thorny bush encroachment that out-competes more productive forage), socio-economic constraints may render the change effectively permanent (e.g. if land users do not have the capacity to remove bushes and exclude livestock to facilitate recovery). In 1994, the UNCCD broadened the definition by adding climatic fluctuations. This latter became the most widely used definition, which is the one used in this review as well: 'land degradation (see 1.2.2) in arid, semi-arid and dry sub-humid areas resulting form various factors, including climatic fluctuations and human activities' (UNCCD, 2006). Land in this context includes, according to the UNEP (1997), soil and local water resources, land surface and vegetation including crops. Thornes (1996) describes this as the bio-productive system comprising soil, vegetation, other biota and the ecological and hydrological processes that operate within the system.
The definition of the UNCCD includes a wide range of conditions and processes which ultimately lead to the onset of desert conditions (Wainwright, 2004). It has been criticised as being too vague and ambiguous (Juntti and Wilson, 2005). These authors state that although human activities are mentioned, the emphasis is on biological processes and technical aspects, thereby sidelining political, economic and socio-cultural dimensions. Other authors and projects have included specific processes in the definition, e.g. Project DM2E specifies that the term desertification refers to the combination of economic, social and climatic processes that cause an imbalance in ecosystems and the reduction or the destruction of the biological potential of soils (Wainwright, 2004). Thornes (2002) states that, if anything, it would be useful to incorporate at least the rural depopulation, especially in the European context, as abandonment of rural areas is a pivotal problem in this region. The EFEDA project includes in the list of causes of desertification, instead of various factors, water erosion, salinization, alkalinization, elimination of plant cover, soil structure degradation, over-exploitation of water resources, cessation of traditional soil conservation techniques and improper land-use planning (Wainwright, 2004). The MEDIMONT project includes the concepts of non-reversibility and the alteration of key components of the soil, vegetation and water system (Wainwright, 2004). They conclude that desertification is a complex phenomenon involving both degradation and recovery processes. Puigdefábregas and Mendizabal (1998), state that desertification is a well-defined process, triggered by changes in climate and socio-economic boundary conditions of affected dryland systems. These changes cause the system to enter an irreversible positive feedback loop of overexploitation of land of which the final outcomes are land degradation and disruption of local economies. They add that desertification is an acute process that occurs at rates several orders of magnitude faster than purely climate-driven land responses.
As a concluding remark to these (slightly) different definitions of desertification, Juntti and Wilson (2005) state that while it is clear that the difference in emphasis in the definition can lead to very different ways of conceptualising and diagnosing the problem and, consequently, to the adoption of different remediation techniques, different emphasis can also be used to serve different interests.
Land degradation
Many academic definitions of land degradation refer to a loss of the biological and/or economic resilience (see 1.2.4) and adaptive capacity of the land system (Holling, 1986; Dean et al., 1995; Kasperson et al., 1995; Holling, 2001; IPCC, 2001). This approach emphasises the maintenance of basic system functions that may (or may not) include human uses. Building on this, it is argued that land degradation can only be determined in relation to the goals of the management system at the time of investigation (Abel and Blaikie, 1989; Turner and Benjamin, 1993), and in the context of a specific time frame, spatial scale, economy, environment and culture (Warren, 2002). In this context, Kasperson et al. (1995) define land degradation as "a decrease in the capacity of the environment as managed to meet its user demands". This resonates with UN definitions emphasising the "resource potential" and "productive capacity" of the land (UNEP, 1992; UNEP, 1997). As such, the extent and severity of land degradation may vary between land users with different management goals in different places at different times and in different socio-economic, environmental and technological contexts.
Land degradation and environmental sustainability are mirror images of the same process (Warren and Agnew, 1988; Warren, 2002). Environmental sustainability depends on the inherent stability and resilience (see 1.2.4) of the resources being used, their sensitivity to change and the system's capacity to adapt to change. For example, a sustainable land use system can either regain its productive potential after a perturbation (e.g. rapid and full recovery after drought) or provide alternative ways to support the livelihoods of those who depend on it (e.g. exploitation of bush encroachment by smallstock). By its definition, land degradation occurs when the resilience and adaptive capacity of the land is compromised.
Despite ongoing political and academic debate over the definition of land degradation, it is possible to distil a number of key components from this discussion. Land degradation: 1. is a human-induced phenomenon that cannot be caused by natural processes alone; 2. decreases the capacity of the land system as managed to meet its user demands; and 3. threatens the long-term biological and/or economic resilience and adaptive capacity of the ecosystem.
Drylands
The "arid, semi-arid and dry sub-humid areas" of the UNEP (1992) definition of desertification are collectively referred to as 'susceptible drylands' (UNEP, 1997; see Fig. 1.1). Hyperarid zones, the true deserts, are not included as they are not considered prone to desertification because of their naturally very low biological productivity.
Fig. 1.1: Dryland systems. From: Millenium Ecosystem Assessment (2005).
Ecological dynamics: resilience, non-equilibrium and multiple-stable states
Currently, resilience is defined as the capacity of a system to absorb disturbance and re-organize while undergoing change so as to still retain essentially the same function, structure, identity and feedbacks (Walker et al., 2004). The resilience perspective emerged from ecology in the 1960s and early 1970s (Folke, 2006). It was introduced by Holling (1973) as the capacity to persist within a domain in the face of change and as a measure of the ability of the system to absorb changes of state variables, driving variables and parameters and still persist. The useful measure of resilience was the amount of disturbance a system can take before its controls shift to another set of variables and relationships that dominate another stability region (Folke, 2006).
It is stated by Wiens (1984) that under natural conditions, disturbances are so frequent that there is rarely enough time between them for plant and animal communities to reach stable equilibria. It has been argued that ecosystems characterised by frequent disturbance, such as drought-prone semi-arid systems, therefore never reach equilibrium (e.g. Behnke et al., 1993; Scoones, 1995). Various authors have argued that for this reason, conceptions of equilibrium ecological dynamics are not relevant for semi-arid systems (e.g. de Angelis and Waterhouse, 1987; Ellis and Swift, 1988; Westoby et al., 1989). Such authors argue that these systems display "non-equilibrium" behaviour. For example, frequent droughts prevent livestock populations ever growing large enough to reach or exceed equilibrium with their fodder resources due to drought-induced mortality in cattle herds (Mace, 1991).
Alternatively, Gunderson and Holling's (2002) framework captures equilibrium ecosystem dynamics within a broader framework of episodic ecosystem collapse and re-organisation. The concept views rangelands as complex systems capable of reaching stable equilibria, or ecological climax and yet vulnerable to collapse in response to perturbations (fire or a combination of grazing and drought in semi-arid rangelands) and able to re-organise to form potentially new species assemblages that become increasingly rich, connected and rigid as they build towards new equilibria (Gunderson and Holling, 2002; Walker and Abel, 2002).
