Objectives The early version of Nutrient Expert for Rice did not differentiate rice types, reducing the precision of fertilizer recommendation at times. We optimized the version by establishing Nutrient Expert for Rice targeted to specific rice type.

Methods The yield and nutrient uptake data were collected and summarized from multi-year and -site rice field experiments from International Plant Nutrient Institute (IPNI) China Program, our research group and relevant published articles. QUEFTS model was used to analyze the nutrient uptake of different rice types, to establish the intrinsic relationship between rice yield response and agronomic efficiency, construct a fertilizer application model and develop Nutrient Expert system for rice. A total of 452 field experiments were conducted in eight provinces in the main rice producing areas from 2013 to 2020 to calibrate and improve the system. Each experiment included six treatments: fertilizer recommendation based on Nutrient Expert (NE), farmers’ practices (FP), conventional recommendation based on soil testing (ST), and N omission, P omission and K omission based on the NE treatment. Rice yield, economic benefit, nutrient use efficiency and fertilizer agronomic efficiency were investigated.

Results The result of the relationship between rice yield and nutrient uptake based on QUEFTS model showed that the N, P and K requirements at above-ground dry matter to produce 1 ton of grain yield were 14.2, 3.9 and 14.3 kg for single-season rice, respectively, and 16.9, 3.3 and 20.9 kg for early, middle, late rice, respectively. The yield responses to N, P and K fertilizer application were 3.1, 1.1 and 1.0 t/hm² for single-season rice, 2.3, 1.0 and 1.0 t/hm² for early rice, 2.3, 0.9 and 1.0 t/hm² for middle rice, 1.9, 0.8 and 0.8 t/hm² for late rice, respectively. The agronomic efficiency of N, P and K fertilizer were 17.6, 15.3 and 9.8 kg/kg for single-season rice, 14.6, 15.3 and 10.1 kg/kg for early rice, 11.4, 12.4 and 9.1 kg/kg for middle rice, and 11.5, 14.8 and 9.2 kg/kg for late rice, respectively. The NE treatment applied 12.6%, 9.6% and 21.4% less N, P and K fertilizer than FP treatment, and 7.6% less N and 15.6% less K fertilizer than ST treatment. Compared with FP and ST treatments, the NE treatment increased yield by 0.5 and 0.3 t/hm², and net profit by 1627 and 893 yuan/hm²; increased N recovery rate by 13.1 and 9.3 percentage points, and increase agronomic efficiency by 4.1 and 2.7 kg/kg; increased P recovery rate by 7.8 and 5.0 percentage points; increased agronomic efficiency by 6.6 and 3.3 kg/kg; increased K recovery rate by 14.6 and 9.5 percentage points; and increased agronomic efficiency by 5.6 and 4.9 kg/kg, respectively.

Conclusions The newly constructed Nutrient Expert for Rice, based on the intrinsic relationship of yield response, agronomic efficiency and relative yield of different rice types further optimazed fertilizer rate, improved rice yield and fertilizer use efficiency, and increased farmers’ income. The nutrient expert for rice is a feasible fertilizer application method for small farmers in China.