Simulation analysis of rice plant traits based on the nutrient area of Gompertz model

【Objective】This study systematically investigates the influence of nutrient area on the aboveground dry matter accumulation and yield of rice plants using the Gompertz model, aiming to provide theoretical basis for the measurement method of the non-uniformity of rice pit distribution and the precise cultivation of high-yield rice. 【Method】Field experiments were conducted using 2 rice cultivars: the conventional late-season variety Nongxiang 42 and the hybrid late-season variety Taiyou 398. Rice was transplanted using a double-seedling casting method under 9 planting density levels（140, 120, 100, 60, 40, 28, 20, 12 and 8 hill/m²） to create locally heterogeneous plant distribution patterns. At the greening stage of 6-12 d after transplantion, orthophotos of the experimental plots were acquired using unmanned aerial vehicles（UAV） aerial rice plants image and stitching software. The rice plant identification algorithm, the rice plant position calibration algorithm and Voronoi diagram were applied to divide the nutrient area of the rice plants and calculate the area of the rice plants nutrient area. Then, the Gompertz model was used to simulate the trend of the nutrient area affecting trait indicators such as panicle weight, effective panicle number and aboveground dry matter. 【Result】The trends of rice trait responsed to nutrient area were well described by the Gompertz function. The coefficients of determination（R²） for the fitted models exceeded 0.8000 for all traits in both varieties, with the exception of effective panicle number in Nongxiang 42（R²=0.7873）. The responses of rice traits to nutrient area followed an S-shaped curve. The optimal nutrient area for Nongxiang 42 ranged from 429.88 to 445.47 cm², with an effective nutrient area below 900 cm². For Taiyou 398, the optimal nutrient area ranged from 464.67 to 484.43 cm², and the effective nutrient area below 1000 cm². The analysis of model fit goodness indicated that the Gompertz model had a good explanatory power for the changes in the nutrient area in response to the yield traits of rice plants. The model fitting accuracy between the measured values of the traits of Taiyou 398 rice plants and the predicted values of the Gompertz curve was higher than that of Nongxiang 42. 【Conclusion】The Gompertz model is used to simulate the quantitative relationship between the nutrient area of rice plants and trait indicators, and the optimal nutrient area and effective nutrient area of rice plants are obtained. The trait indicators of rice plants change in an S-shaped curve with the increase of the nutrient area, and the yield per unit area of each trait indicator shows the law of first increasing, then decreasing, and finally tending to stabilize.