Migration of farmers, for example due to human population growth, increases the pressure on often arid or semi­arid lands. In particular these migrant farmers attempt to maintain their familiar farming systems, which are often ill­suited to the possibilities of these marginal lands.

Many have stressed that changes in the prevalent farming systems and their adaptation to the specific environments is a highly important strategy to improve yields and food security. For example, the replacement of maize by sorghum and millets, as well as the acceptance of cassava as a crop for human consumption, have been propagated in Eastern and Southern Africa. As a complementary activity, the breeding of more drought tolerant varieties of preferred staple crops, such as maize, wheat and rice, should be undertaken. Drought tolerance is generally defined as the property of a given cultivar to show a relatively small yield reduction upon exposure to drought. This implies that breeding for stress adaptation is at the expense of yield potential. Drought tolerance has appeared to be a difficult research aim in breeding programmes due to several technical limitations:

• Definition of drought. Drought stress is highly heterogeneous in time (over the seasons and years) and space (between and within sites), and is extremely unpredictable. This makes it very difficult to identify a representative drought stress condition.
• Environmental influences. Since the phenotype is the product of genotype and environment, assessment of the desired genotype is highly dependent on the proper environmental conditions. The unpredictable and variable forms in which drought stress will manifests itself, makes selection of promising individual plants and breeding for drought tolerance extremely difficult.
• Multiple genes involved. Drought tolerance has been shown to be a highly complex trait, influenced by many different genes. In fact, drought tolerance should not be regarded as a unique heritable trait, but as a complex of often fully unrelated plant properties.
• Interrelation with other stresses. Drought can hardly be separated from other important abiotic stresses such as temperature and salinity. Due to these interrelations, no single mechanism exists by which multiple stresses are alleviated.

• The ability of a plant to escape periods of drought, in particular during the most sensitive periods of its development. One breeding strategy is to shorten the life cycle of a crop to enable it to mature safely during a rainfall period. For example, in the Sahel very short season cowpeas now avoid drought by maturing before any substantial stress develops, in less than 65 days.
• The ability of a plant to endure or withstand a dry period by maintaining a favourable internal water balance under drought condition. Osmotic adjustment, in which the plant increases the concentration of organic molecules in the cell water solution to 'bind' water is one example. A thicker layer of waxy material at the plant surface and a more extensive and deeper rooting are others.
• The ability of a plant to recover from a dry period by producing new leaves from buds that were able to survive the dry spell, is commonly considered less interesting from the breeder's point of view. For example, the alternately branched forms of groundnuts bear large numbers of terminal vegetative buds, which may help them to produce new leaves more rapidly after damaging dry periods.

Current research at international institutes
For several crops marker­assisted breeding for drought tolerance is now expanding. The efforts of the international research institutes funded by the Consultative Group on International Agricultural Research (CGIAR) are explicitly directed at improved genetic material for agriculture in developing countries and rely in their breeding programmes on germplasm from these countries. The status and quality of the breeding programmes will, to a large extent, determine whether and over which time period the use of molecular markers will advance the selection of more drought­tolerant varieties.
In general, worldwide research on drought tolerance breeding has focused on cereals. In developing countries, the crops that receive attention for drought tolerance include maize, upland rice, wheat, sorghum, cowpea, pigeonpea and Phaseolus bean.
For maize, rice and soya bean, and to some extent for wheat and Phaseolus, genomic maps have been developed and molecular markers are available. Whereas breeding research in maize and soya bean has been largely an affair of private US companies, the International Rice Research Institute (IRRI, Philippines), with the support of the Rockefeller Foundation, has played a central role in rice research. The International Maize and Wheat Improvement Center (CIMMYT, Mexico) has developed marker­assisted breeding programmes in maize, on drought tolerance (anthesis­silk interval) and on insect resistance. In wheat, CIMMYT carries out research on disease resistance and drought tolerance.
At the Centre for Tropical Agriculture (CIAT, Colombia) transfer of drought tolerance from tepary to Phaseolus bean (genetic exchange between different species!) assisted by molecular markers is part of the bean improvement programme. The International Crops Research Institute for the Semi­Arid Tropics (ICRISAT, India), in collaboration with several international partners, has developed a genomic map for sorghum using maize markers, and mapping for drought tolerance in sorghum and pearl millet has been identified as a priority. Similar research in chickpea, pigeon pea and groundnut is in a preliminary stage. In addition, considerable public research efforts into conventional and/or marker­assisted breeding for drought­tolerance have been undertaken in drought­prone countries, such as Australia, USA, Israel and India.
Bert Visser (DGIS, Ministry of Foreign Affairs, The Netherlands)