Modeling for recommending panicle nitrogen topdressing rates for yield and milled-rice protein content

This experiment was conducted to calibrate models for recommending panicle N fertilizer rate for target grain yield and milled-rice protein content based on not only rice growth and N nutrition status at panicle initiation stage (PIS) but also on weather conditions. Five rice cultivars; Chucheongbyeo, Daeanbyeo, Hwasungbyeo, Surabyeo, and Juanbyeo, were grown under various N fertilizer application rates. Shoot N accumulation at PIS (Pnup) was measured before panicle N application at PIS and at harvest. Grain yield, yield component, and milled-rice protein content were measured at harvest. Three models for prescribing panicle N fertilizer rate were calibrated by stepwise multiple linear regression (SMLR) analysis. The model for shoot N accumulation from PIS to harvest (PHnup) was calibrated using Pnup, panicle N application rate, and weather variables like air temperature and solar radiation. Models for grain yield and milled-rice protein content were calibrated using Pnup, PHnup, and weather variables. The models for PHnup, grain yield, and milled-rice protein content showed acceptable accuracy and precision with R² of 0.78, 0.85, and 0.77, respectively. The stability of the models was tested through the comparison of slopes between the observed and the predicted values in different conditions of temperature, radiation, and cultivar. The models for PHnup and grain yield showed homogeneity of slopes between the observed and the predicted values regardless of different temperature, radiation conditions, and cultivars. The model for milled-rice protein content showed homogeneity of slopes between the observed and the predicted values across low and high temperature and radiation conditions while the slopes were significantly different among cultivars. The oldest cultivar “Chucheongbyeo” being significantly different from the other four cultivars recently improved. In conclusion, the three models were precise and accurate enough to be used effectively for prescribing panicle N topdressing rate if Pnup could be measured timely with a cost-effective method.