Field screening for drought tolerance - principles and illustrationsF RBidingerauthorEstablishing a screening procedure for genetic differences in drought tolerance involves 1) practical decisions on the objectives of such a screening program, 2) the selection of environment(s) and stress occurrence(s) to be targeted in the program, and 3) the design and operation of field physical facilities and experimental methods to apply a uniform, repeatable drought stress. This paper considers these points from a conceptual and a practical viewpoint. Drought tolerance can be approached on various plant organizational levels, from crop yield stability under stress, through responses to stress indicative of tolerance, to the biological mechanisms that underlie these responses, to the genes and alleles governing the presence or expression of the responses/mechanisms. Defining stress tolerance at each level has specific advantages and disadvantages for designing a field-screening program. Work on pearl millet has mainly focused on the crop tolerance response level, targeting the relative ability of genotypes to maintain grain numbers per panicle and seed filling in terminal stress environments. Target environments and target stress occurrences for a screening program must be established from the analysis of historical climate data. Water budgeting is probably the minimum level, but opportunities to use crop simulation modeling for this purpose are improving. Establishing screening systems with environmental conditions representative of the target environment, is difficult, involving a major tradeoff between providing representative daylength, vapor pressure, and temperature conditions, and easily managing soil water/rainfall. In contrast, duplicating target environment moisture patterns in non-target environments is easier, but G x E effects can be a problem. The effectiveness of a drought screening procedure is best measured by the genetic heritabilities achieved for target traits, whether the focus is nurseries therefore requires careful analysis of likely sources of nongenetic variation among plots, replications, and repeated experiments, and seeing that these are minimized. These include 1) the choice of site for screening, 2) the physical management of both water-related and non water-related sources of variation in crop growth within and across experiments, 3) the choice of experimental design and the effective use of blocking to remove expected sources of nonmanageable variation, and 4) the efficient collection and management of data. These considerations are illustrated here with examples from the pearl millet drought screening system used at ICRISAT.Agriculture-Farming, Production, Technology, Economics2002International Crops Research Institute for the Semi-Arid Tropics Book Section