What can be learned from a study of desertification? Can this phenomenon be analysed to increase our understanding of arid ecosystems? Whatever strategy is employed, the task will not be easy because the accounts of desertification suffer from being too closely identified with the human dimensions of the process. The word itself is mostly used in a dramatic, emotional context to maximize attention and focus on the human component to the exclusion of the landscape (e.g. Ahmad and Kassas, 1987).

   Desertification is often defined in dynamic terms such as: 'the expansion of desert-like conditions and landscapes to areas where they should not occur climatically', e.g. the 'desertization' concept of Le Houerou and Gillet (1986). While the word desertification implies a process, this term has largely been used to describe the endpoint, i.e. the desertified or degraded landscape.

   A simpler and more useful definition is that of Dregne (1983): 'desertification is the impoverishment of terrestrial ecosystems under the impact of man!' Thus desertification is just one form of land degradation that is associated with semi-arid or arid landscapes. Implicit in the above definition is a continuum of ecosystem modification, from slight to severe as the result of the degradation process, and that is the direct result of human activity.

   Dregne (1983) does not restrict desertification to the semi-arid and arid lands, citing similar outcomes of the degradation process in irrigated and cultivated land. However, in this chapter the concept of desertification is restricted to the semi-arid and arid ecosystems (collectively called 'arid ecosystems') that are grazed by domestic and wild herbivores.

   Reports dealing with desertification are numerous, particularly since the UN Conference on Desertification (UNCOD) held in Nairobi in 1977 (United Nations, 1980; Spooner and Mann, 1982; El-Baz and Hassan, 1986). In spite of this abundance, the major difficulty in analysing the phenomenon of desertification is the virtual absence of ecological detail in most reports. In the published work most attention has been paid to describing the higher levels of ecosystem functioning, e.g. the changes in the number of livestock carried, with considerably less emphasis on the reaction of soils, plants, and related processes. Even Sahelian Africa, the site of the most modern epoch of desertification, has not been a source of definitive ecological research on desertification as such. Because international concern has been focused on the human misery, much of the reporting of ecosystem condition and change has been after the event, broad-scale, sweeping, and without the benefit of measurements or assessments of the actual changes that took place.

   Substantial ecosystem change has been claimed, but the quantitative evidence for the exact nature and extent of that change is largely lacking. Indeed several authors dispute that the decline in the productivity of some of the Sahel has been permanent or in any way influenced by the state of degradation of the landscape (Hellden, 1986; Olsson, 1985). The application of satellite data to provide objective quantitative estimates of change, e.g. Jacobberger(1987), has little been used.

   Given the limitations of the reported evidence, the question of how to use these observations on desertification to better understand ecosystem structure and function becomes difficult. A useful starting point is the word itself. Much of what is known about the modern epoch of desertification in Sahelian Africa comes from the observations of ecologists who have spent their professional lives in the region; see, for example, the comments of Le Houerou and Gillet, 1986. Desertification has been defined in many ways. Verstraete(1986) provides an historical summary of the evolution and use of the word, and indicates how the emphasis on the causal factors (man vs. climate) and the importance of the impacts (soils and vegetation vs. economic or social) varies greatly between authors. There is debate whether the process is irreversible or not, progressing or not, and since there is uncertainty about the causes, there is disagreement as to what can be done about it.

   The simplest interpretation of the disagreement and disparity between the authors reporting desertification is that the phenomenon is not a simple process. The course and outcome of desertification cannot be unambiguously identified or forecast at any point. Given the very simple nature of the pastoral system, these observations are in keeping with the behaviour of a system wherein the components are very loosely coupled, and that perturbations at one level are amplified by positive feedback at another .

   Desertification is derived from 'desert' and the suffix 'fication', which together should be interpreted as the 'making of a desert' or the 'production of desert-like conditions'. This interpretation is of limited ecological utility because degradation of some semi-arid lands, by the encroachment of woody, unpalatable shrubs, results in a wooded landscape very different from that of a typical desert. Also, degradation of arid landscapes may not always result in typical desert landforms, e.g. mobile sand dunes. Lastly, it is difficult to unequivocally substantiate the claim that degradation of arid or semi-arid landscapes always resulted in a 'xerification or desertization' by a permanent reduction in rainfall or rainfall effectiveness. Links between surface conditions (albedo and temperature) and the probability of rainfall have been postulated (Otterman, 1974; Charney et al.,1977) and partly substantiated by modelling (Rasool, 1984; Laval, 1986), but they are disputed by measurement (e.g. Vukovich et al. , 1987). Idso(1981) suggests that there is an inadequate understanding of this feedback loop, both of the processes involved and the scaling of their importance from the land surface to the top of the troposphere.

   Nonetheless, however varied the definitions of desertification are, the underlying theme is that, depending upon location and initial starting conditions, some of the ecological, rather than the climatological, characteristics of deserts can be imposed upon the landscape. Ecosystems, through the cumulative impact of man's activities, can approach the structure and functioning of systems that are characteristic of climatically far more arid areas. This is the key issue of this chapter.