To plan improved agriculture, or even to attempt to halt or slow down the effect of soil degradation, we must know the facts which determine what is possible and what is practical, what the physical resources allow, and which of these options can be made to work in human terms.
It is remarkable that so many countries, whose future development is going to depend mainly on agricultural production, have not taken the important step of compiling an inventory of their available natural resources, and a classification into agro-ecological zones of different potentials. Such surveys can be done using methods according to the scale. The FAO programme of regional agro-ecological studies uses massive computa- tional capacity to combine the 1:5 000 000 World Soil Map with climatic data to give maps of agricultural potentials. The same method can be applied to countries, with mapping at 1:1 000 000. Studies at this scale are useful for very broad strategic planning but need to be followed up by more detailed studies (FAO 1978).
Another style of resource inventory is the series of natural resources studies by the Land Resources Development Centre (LRDC) which combine air-photo interpretation and ground survey with existing country data on geology, soils, and climate, and may be at a scale suitable for the whole of Northern Nigeria (LRDC 1968), or at 1:250 000 for the tiny country of Lesotho (LRDC 1968). An effective system has been developed in Zimbabwe which could serve as a model for most developing countries. The agro-eco- logical survey of Zimbabwe was completed and mapped at 1:1 000 000 in the nineteen-fifties, and has been progressively refined and improved to the present "natural regions and farming areas" which is an essential tool for the government strategic planning of agricultural policy for Zimbabwe. Australia provides good examples of detailed surveys at catchment level (Aldrick et al. 1978). The French service for overseas research (ORSTOM) has developed in Tunisia an ecological classification with 22 "ecological systems", suitable for Mediterranean semi-arid zones in North Africa (Floret 1981).
Erosion also needs to be surveyed and recorded, at different scales according to the degree of detail required. Landsat and the constantly improving satellite imagery, can be used for broad assessments (Pickup and Nelson 1984), or air-photo interpretation for more detail (Keech 1985), in both cases with field work to establish ground truth. As well as studying present erosion we should collect information on soil conservation. What has been done in the past, even if it has not been successful, can provide useful information on mistakes to be avoided. Also we need to study conservation methods in case they can also be used in other situations.
The fact that semi-arid regions have not been intensively developed has led to a generally low level of information on soils and climate. Rainfall measurements tend to be few and far between whereas the strong spacial variability means that the available data may be not representa- tive. Similarly, soil surveys and studies of the physical and chemical characteristics of soils are in general much weaker than in areas of more reliable rainfall.
Because of the wide variation of conditions in semi-arid regions, the results of research and development in other countries need to be interpreted with caution, but in recent years there has been a greater interest in farming drier areas and so any review of existing information should include information from research programmes, particularly current and recent programmes where the results may not have been widely dissemi- nated. In the most common situation of existing knowledge and information being inadequate, programmes of research and development must include and bring together both component research within disciplines, and research on farming systems both on-station and on-farm.
There is little doubt that in the past much competent component research has not been successfully transferred and adopted by the farmers because it did not fit into the farming system. It is increasingly recog- nized that a complete understanding is required of all the social and economic ramifications of the farming system in order to guide component research into topics and techniques which will be applicable. Both col- lecting the information and understanding it is particularly difficult in the case of farming systems in semi-arid regions. The usual difficulty in collecting data from farmers, such as scepticism of official intentions, and reluctance to disclose financial information, are compounded by the complexities of the system. For example, in commercial farming for profit, it is fairly straightforward to find out the process of taking management decisions and who the decisions are taken by. In tribal or village commu- nities, taking decisons can be a very lengthy and complicated operation. The facts which are being sought are complicated and may be unfamiliar to the people conducting the survey. Rights of tenure and rights of access, e.g. to water supplies, to grazing, to fuel, or cultivation rights or gathering rights are usually complicated. In nomadic or semi-nomadic live- stock management even ownership may not be straightforward; for example, individual ownership of animals within a communal herd, or ownership of the animal but not its offspring, or conditional ownership with reversion if a marriage breaks up, and many other complications unfamiliar to western agriculture. There can be no doubt that studying the whole of the agricul- tural system is fundamental to any consideration of the possibilties for agricultural development. Techniques for obtaining the required information have been greatly improved in recent years, by Mann (1974), Chambers (1981), and Collinson (1981).
Many of the strategies for improved cropping and methods to make more effective use of the rainfall, discussed in Chapters 5 and 6, are most likely to be applicable to individual land holdings or to small groups. This probably accounts for the popularity and success of small development schemes operated by Non Government Organizations (NGOs), for example the Oxfam project in Burkina Faso, or the project by War on Want in Kenya, which are discussed in Section 5.2.4. The classic phrase of Schumaker "small is beautiful" certainly seems to apply to most development projects in semi-arid areas. There are of course some examples of large-scale run-off farming operations, such as the ancient systems in the Negev or the inundation tanks or khadins of northern India. One such tank near Bharatpur has an impounding bund 19 km long, inundates 4100 hectares, and irrigates another 4800. These schemes are described in Section 5.3.3.
The slow rate of adoption of what appears to be a wholly satisfac- tory new method was discussed against the background of extension in India in Section 2.1.4, but there are also examples of speedy adoption, such as the recent rapid spread in Botswana of the use of medium horsepower tractors and borehole syndicates. A study of why new ideas sometimes move slowly and in other cases take off rapidly would be an interesting field of study. One problem of the adoption of methods which attempt to change the rainfall/run-off/crop usage pattern is that the rainfall varies too much to form the base of a reliable strategy. The water collection device which is required in a dry year, is counterproductive in a wet year when drains are required, and vice versa.
In the long term there are opportunities for improving the physio- logy of plants for use in semi-arid areas. The selection and breeding of short-maturing varieties is well established and has resulted in some significant successes, and there are other aspects not so well developed such as increasing the transpiration efficiency. The question of more effective use of available water is discussed further in Section 4.2.4.