Replacement of Replications with Additional Locations for Grain Sorghum Cultivar Evaluation J. J. Johnson,* J. R. Alldredge, S. E. Ullrich, and O. Dangi ABSTRACT Crop failure and low yields due to insufficient rainfall have led to undernutrition and starvation in West and Central Africa in recent years. Fanners have not adopted superior cultivars recommended from research station trials, necessitating on-farm testing of new cul- tivars under farmers' conditions. The number of test plots possible within a farmer-managed on-farm test is constrained, but the number of tests is flexible. More new cultivars could be tested in farmers' fields if replications within locations were replaced by additional lo- cations. Improved and local cultivars of sorghum [Sorghum bicolor (L.) Moench] were tested in North Cameroon in replicated on-farm tests in 23 locations in 1987 and 60 locations in 1988. The required number of locations necessary to replace replications was examined under different probabilities of making Type I, II, and III errors and various true differences between cultivar yields for each year. Tra- ditionally, only Type I error risk has been considered when deriving the optimum allocation of plots among replications and locations. Type III error, leading to choice of inferior cultivars, is considered the most serious risk. Results indicated that simultaneous control of Type I and Type III errors would require two to three times the number of lo- cations as required for control of Type I error alone. Three-replicate tests can be replaced by 42 to 65% additional single-replicate locations and triple the number of cultivars tested annually in farmers' fields. J.J. Johnson and S.E. Ullrich, Dep. of Crop and Soil Sciences, and J.R. Alldredge, Program in Statistics, Washington State Univ. Pullman, WA 99164; and O. Dangi, Inst. of Agronomic Research, P.O. Box 33, Maroua, Cameroon. WSU Departmental Paper no. 9001-48. Received 2 Nov. 1990. "Corresponding author. Published in Crop Sci. 32:43^6 (1992). S PORADIC DROUGHT and crop failure during the last 20 yr have caused serious deficits in food crop production across West and Central Africa from Sen- egal to Chad. Grain sorghum is the major food cereal of the semiarid region of subsaharan Africa. Plant breeding programs have been initiated or reinforced in hopes that subsistence levels of food production might be regained by rapid selection of improved cul- tivars. However, new grain sorghum cultivars rec- ommended on the basis of superior research station performance have not maintained their superiority when subjected to variable soil, climatic, and management conditions typically found in farmers' fields. Cultivar evaluation under farm conditions is recommended when a discrepancy is found between farmers' field data and research station data (5). On-farm testing of improved grain sorghum culti- vars in semiarid North Cameroon arose from the de- sire of both research and extension organizations to improve cultivar recommendations and to accelerate their adoption. In the Northcentral region, a sorghum production area of 250 000 ha and encompassing = 210 000 farm families, at least a hundred farmers are willing to conduct on-farm sorghum variety tests in a single season. However, nonexperimental error becomes substantial within locations when the number of plots, from replications or cultivars, exceeds 8 to Abbreviations: E, environment; G, genotype. Published January, 1992