Genotypic performance under fertilised and non-fertilised conditions in rainfed lowland rice

The rate of fertiliser application to rainfed lowland rice is commonly low while selection in breeding programs is conducted under well-fertilised conditions. The appropriateness of applying fertiliser in a rainfed lowland rice breeding program at a rate above the average farmers' rate was examined in Laos. Experiments were conducted at three locations in 2 years using 72-60 genotypes most of which were typical materials used in the Lao breeding program. The fertiliser rates were nil and 60N : 13P : 16K kg ha(-1). The effect of fertiliser application was significant in most experiments, but the response to fertiliser depended on location, There was often significant genotype-by-environment interaction for grain yield in individual experiments. There was also significant correlation for grain yield between fertilised and non-fertilised conditions, with coefficients of determination (R-2) of 0.45 and 0.54 in the 2 years, for combined results of three locations. Combined analyses of grain yield from three locations indicate that for either year the variance component for the genotype-by-fertiliser interaction was smaller than for genotype. The coefficient of variation for grain yield was often smaller under fertilised compared with non-fertilised conditions. These findings as well as the rapidly increasing use of fertiliser justify the level of fertiliser currently applied in the Lao breeding program, The significant genotype-by-location interaction requires that for an effective breeding program, locations be selected to represent farmers' fields. (C) 2000 Elsevier Science B.V. All rights reserved.