基于机器学习算法的小麦遗传群体持绿性研究  

Study on Greening of Wheat Genetic Population Based on Machine Learning Algorithm

作　　者：王多霞 李雷 费帅鹏 付雪丽 肖永贵[2] 孙海艳 陶志强[2] 孟亚雄[1] WANG Duoxia;LI Lei;FEI Shuaipeng;FU Xueli;XIAO Yonggui;SUN Haiyan;TAO Zhiqiang;MENG Yaxiong(College of Agronomy,Gansu Agricultural University,Lanzhou,Gansu 730070;Institute of Crop Sciences,Chinese Academy of Agricultural Sciences,Beijing 100081;Zhongnongfa Seed Industry Group Co.,Ltd,Beijing 100052)

机构地区：[1]甘肃农业大学农学院,甘肃兰州730070 [2]中国农业科学院作物科学研究所,北京100081 [3]中农发种业集团股份有限公司,北京100052

出　　处：《核农学报》2024年第8期1476-1486,共11页Journal of Nuclear Agricultural Sciences

基　　金：科技创新2030—“新一代人工智能”重大项目(2022ZD0115703);国家重点研发计划小麦优异种质资源精准鉴定(2021YFD1200600);现代农业产业技术体系Supported by the earmarked fund for China Agriculture Research System(CARS-03)。

摘　　要：精准预测小麦旗叶中的叶绿素含量(SPAD值)是衡量品种持绿和高光效高产特性的重要方式。为提高小麦SPAD值的预测精度,本试验以中麦578/济麦22重组近交系(RIL)-F8代群体为试验材料,设置正常灌溉和节水灌溉两种水分处理,结合无人机多源传感器融合以及5种机器学习算法和集成方法,预测不同家系在灌浆中后期(花后28 d)的SPAD值,并通过筛选持绿和非持绿性家系,探究灌浆中后期SPAD值与籽粒产量的关系。结果表明,花后28 d旗叶SPAD值和籽粒产量呈正相关(相关系数为0.59);在节水灌溉处理下,小麦旗叶SPAD值和籽粒产量均显著降低。基于多源数据融合的集成方法预测旗叶SPAD值精度优于单一模型。结合不同水分处理下的家系籽粒产量数据发现,受水分胁迫不敏感的10个持绿高产家系中ZJ-16家系持绿、高产效应明显。该研究为田间快速无损监测小麦叶绿素含量提供了一种方便、灵活和有效的方法,为小麦抗旱、高光效育种提供了理论依据和技术支撑。Accurately predicting the chlorophyll content(SPAD value)in wheat flag leaves is crucial for assessing the variety’s greenness,photosynthetic efficiency,and productivity.This study utilize multi-source data from drones to enhance the prediction accuracy of wheat SPAD values.This improvement offers effective technical support for accelerating the breeding of wheat varieties with greenness,high yield,and broad adaptability during the late filling stage.The study used the recombinant inbred line(RIL)population of Zhongmai 578/Jimai 22 as experimental material,with two water treatments:normal irrigation and water-saving irrigation.By combining multi-source sensor fusion from drones with five machine learning algorithms and integration methods,this study predicted the SPAD values of different genotypes in the late filling stage(28 days after flowering).It also investigated the relationship between SPAD values in the late filling stage and grain yield by selecting genotypes with green and non-green traits.The findings revealed a positive correlation(correlation coefficient of 0.59)between the SPAD value of flag leaves at 28 days after flowering and grain yield.Under water-saving irrigation,both the SPAD value of wheat flag leaves and grain yield decreased significantly.The integrated method,which is based on multi-source data fusion,demonstrated better accuracy in predicting flag leaf SPAD values compared to a single model.By combining grain yield data from different genotypes under varous water treatments,the study identified 10 green and high-yielding genotypes that were insensitive to water stress.Among them,the ZJ-16 genotype exhibited noticeable greenness and high-yield effects.This study provides a convenient,flexible,and effective approach for rapid and non-destructive monitoring of wheat chlorophyll content in the field,providing theoretical basis and technical assistance for breeding wheat varieties with drought resistance and high photosynthetic efficiency.

关 键 词：小麦 叶绿素含量 多源数据融合 机器学习 

分 类 号：S512.1[农业科学—作物学]