CULTIVAR, NITROGEN, AND MOISTURE EFFECTS ON A RICE-WHEAT SEQUENCE - EXPERIMENTATION AND SIMULATION

Rice (Oryza sativa L.) followed by wheat (Triticum aestivum L.) is a d ominant cropping sequence under a range of management regimes in South Asia, with variable productivity. Simulation models can be used to ex trapolate experimental results to other sites of interest in variable environments. Experiments were conducted to quantify the responses of two rice cultivars to three N regimes and two soil moisture regimes, f ollowed by the responses of the succeeding two wheat cultivars to thre e N regimes, two soil moisture regimes, and two dates of planting, on a Haplaquept noncalcareous brown floodplain soil at the Wheat Research Center, Nashipur, Bangladesh. CERES-Rice and CERES-Wheat models were validated using the experimental data set, and then were used to estim ate the grain yields at two other sites in Bangladesh. In rainfed plot s, grain yields of 'BR14' rice were reduced by 48.8, 43.4, and 39.3% r elative to irrigated plots, and those of 'BR11' were reduced by 49.1, 43.8, and 42.2% (P < 0.05), for 0, 90, and 135 kg N ha(-1), respective ly. The optimum N rate was between 90 and 135 kg ha(-1). In the succee ding wheat crop, at 0 N, there were no significant differences between the two moisture regimes but, at high N (180 kg ha(-1)), irrigation p romoted greater yield. Relative to the high-N treatment, gain yields o f irrigated 'Kanchan' wheat at ON were reduced by 73.0 and 71.6%, and of 'Sowgat' by 75.2 and 73.5%, for early and late plantings, respectiv ely. Simulation investigated year-to-year variation in grain yields of both crops under their several N regimes. Rice yields tended to incre ase over years, but wheat yields had no definite trend. The simulation results indicated that, without N fertilizer application, N will be l imiting across sites and long-term weather conditions for both rainfed and irrigated crops. Field experimentation quantified actual crop res ponses to a range of management practices and allowed validating the m odels for the rice-wheat sequence. The models were then used to extrap olate and generalize the results for two other sites in Bangladesh.