棉籽蛋白质和油分形成的模拟模型

基于不同熟性棉花品种的异地分期播种和施氮量试验，综合量化品种特性、主要气象条件(温度、太阳辐射)和栽培措施(施氮量)对棉籽蛋白质和油分的影响，基于棉籽氮素积累和油分合成的“库限制”假设，结合棉花铃期与棉籽干物质积累模型，建立基于过程的棉籽蛋白质和油分形成模拟模型。利用不同生态点不同品种、播期和施氮量的田间试验资料对模型进行检验的结果表明，供试品种科棉1号和美棉33B棉籽的蛋白质含量模拟值与实测值的根均方差(RMSE)分别为2.05%和2.33%，其油分含量模拟值与实测值的RMSE分别为2.45%和2.95%。模型以主要气象资料(日平均温度、日太阳辐射量)和栽培措施(施氮量)作为模型输入，以棉花铃期与棉籽干物质积累模拟模型为基础，实现了较广泛生态条件下对不同品种棉花棉籽蛋白质和油分形成过程的模拟及其含量的动态预测。模型预测精度高，广适性强。

Simulation Model of Cottonseed Protein and Oil Formation

The simulation of cotton (Gossypium hirsutum L.) seed growth is an area of great uncertainty, especially in the process of cottonseed quality formation. To simulate the formation of cottonseed protein and oil under different environmental conditions, we developed a simple process-based model driven by the inputs of cultivar parameters, weather information, and crop management variable (precisely N supply). A set of field experiments were conducted with Kemian 1 and NuCOTN 33B in the lower reaches of Yangtze River Valley (Nanjing, Huai’an) and the Yellow River Valley (Xuzhou, Anyang) in 2005. Two sowing dates and three N rates were set in the trials. According to the data collected in Nanjing, the response functions of cottonseed protein and oil accumulation to weather conditions (temperature, solar radiation), crop management (variable N supply) and boll positions were all developed and involved in the model. The subtending leaf N concentration of cotton boll obtained from a semi-empirical equation was made as a direct indicator of the N nutrition affecting cottonseed quality formation. The model was based on the hypothesis that nitrogen accumulation and oil synthesis in cottonseed are mainly sink-determined, and was integrated with the cotton boll maturation period model and cottonseed biomass accumulation model. The parameters in the model were calibrated using the field data obtained in Nanjing. The model was tested using the field data obtained in Huaian, Xuzhou and Anyang. The root mean square error (RMSE) of the simulated and measured cottonseed protein contents was 2.05% for Kemian 1 and 2.33% for NuCOTN 33B. The RMSE of the simulated and measured cottonseed oil content was 2.45% for Kemian 1 and 2.95% for NuCOTN 33B. Driven by the inputs of data including weather conditions (daily maximum, minimum and average temperatures and daily solar radiation), management variable (precisely N supply), the present model accurately predicted cottonseed protein content and oil content under diverse environmental conditions. This model is a necessary component of cotton growth model, and provides a good platform for further study on modeling cottonseed protein and oil yield.