A model explaining genotypic and environmental variation of rice spikelet number per unit area measured by cross-locational experiments in Asia

Rice (Oryza sativa L.) yield is closely related to the number of spikelets per unit area which varies with the genotype and environment. The objectives were to identify plant factors associated with genotypic and environmental variations in spikelet number and to develop a model that may explain the variations based on plant factors. Two-year cross-location experiments were conducted with nine widely different rice genotypes at eight locations in Asia, which covered a wide climate range from the cool temperate zone in northern Japan to the tropics in Thailand. We determined the spikelet density, crop biomass and plant nitrogen (N) content at different development stages. The spikelet density of the genotypes grown at the eight locations ranged from 5130 to 71,480 m(-2). Spikelet density was related (R-2 = 0.58) to N accumulation in plants up to the late spikelet differentiation stage, i.e., 2 weeks before heading. The spikelet production per unit crop-N content, however, was under the additional influence of crop growth rate (CGR) during 2-week period preceding heading. On the basis of these results and previous studies, we modeled the spikelet number per unit area as a function of plant N accumulation up to 2 weeks before heading and CGR during the 2-week period preceding heading. Approximately half of the data sets were used for parameter estimation by a SIMPLEX method and the rest for model validation. The observed genotypic and environmental variation in spikelet number was well explained (R-2 = 0.85) by the model with only one genotype-dependent parameter to represent the number of juvenile spikelets produced per unit plant N. The other three parameters were genotype-independent. Quite a large genotypic difference was recognized in the spikelet number produced per unit plant N; the efficiency was higher in improved indica genotypes, lower in unimproved indica and javanica genotypes, and intermediate in japonica genotypes. These results suggested that the primary plant factors for increased spikelet number per unit area in rice are: (1) larger accumulation of plant N up to 2 weeks before heading to increase spikelet differentiation, (2) higher CGR during 2-week period preceding heading to suppress degeneration of juvenile spikelets and (3) high spikelet production efficiency per unit N of genotypes. (c) 2005 Elsevier B.V. All rights reserved.