Objective  To improve efficiency and reduce waste on rice fertilization in the fields while protecting the environment, the upper limits on N, P and K applications for the farming of early rice, late rice, single cropping rice, rice after vegetable, and rice after tobacco in Fujian were studied.

  Method  A total of 832 field experiments were conducted on (1) the early rice, late rice and single cropping rice in mountainous and coastal areas, (2) the after-tobacco cropping in mountain areas only, and (3) the after-vegetable cropping in either mountainous or coastal areas. System clustering analysis and significance test of difference were employed to classify the data generated. A ternary non-structural NPK fertilizer response model was then constructed for each category prior to field testing.

  Result  Seven optimum fertilization categories for the early rice in the mountainous areas, 6 in the coastal areas, and 7 in the entire province were established. For the late rice, 6 in the mountainous areas, 4 in the coastal areas, and 7 in the entire province were obtained. For the single cropping rice, 6 in the mountainous areas, 4 in the coastal areas, and 6 in the entire province were resulted. And, for the after-tobacco cropping, 4 in the mountainous areas; while, for the after-vegetable cropping, 3 categories became available for analysis. Based upon these 60 models on the NPK fertilization limits set according to the agronomic effects on the rice plants, 234 field tests were carried out for final selection. It was verified that the formulations produced 4.0% to 12.5% increases on yield with an averaged net income increase from 875-2 616 Yuan·hm^-2. In comparison to the common practice by the farmers, the recommended fertilizations reduced 19.4% on N usage and 23.9% on P but increased 27.3% on K₂O.

  Conclusion  The fertilization recommendations would optimize the fertilizer application with significantly improved efficiency for all types of rice cultivation currently practiced in the province.