Phenotyping Of Arabidopsis Drought Stress Response Using Kinetic Chlorophyll Fluorescence And Multicolor Fluorescence Imaging

Drought stress has become one of the major challenges in global agricultural production and food security.Researchers have been interested in developing nondestructive,fast,and effcient plant phenotyping techniques,which could bridge the knowledge gap between the plant phenotype and the genotype.The chlorophyll fluorescence imaging can provide information related to the plant photosynthesis,and has become one of the useful tools in the current phenotyping techniques.This paper proved that a combination of kinetic chlorophyll fluorescence and multicolor fluorescence imaging could improve the possibility of detection of drought stress response of two genotypes of Arabidopsis.Meanwhile,machine learning were used to achieve the early classification of drought plants and the control ones,as well as visualization of three drought stress degrees.The main contents were as follows:(1)Optimization of the the water quantity and the dark adaptation duration were perfomed.As a result,the plants were watered 6 mL best within 0 to 10 mL every day.And the dark adaptation should last 20 min at least.(2)Morphological,physiological and fluorescence parameters provided qualitative information related to drought stress responses in different genotypes of WT and osca1.Representitive fluorescence parameters,such as ?PS?_Lss,NPQ_Lss,BF,GF,IrF and multicolor fluorescence ratio had significant differences between the drought stess groups and the control ones.In addition,MDA had high linear relationship with Fm_D2,NPQ L3,IrF,RF/IrF.These inferred that kinetic chlorophyll fluorescence and multicolor fluorescence imaging were useful for understanding the drought tolerance mechanism of Arabidopsis.(3)Early detection of drought stress could be achieved based on machine learning using kinetic chlorophyll fluorescence and multicolor fluorescence imaging.The fusion of two kinds of fluorescence parameters seleced by sequential forward selection were performed well by k-nearest neighbors model,resulting in good classification accuracies of 91%and 96%in WT and oscal for classifying the control plants from the drought-stressed as early as 3 days post drought stress,respectively.(4)Detection and visualization of drought stress degree in Arabidopsis were achieved based on LDC model combined with sequential forward selection using kinetic chlorophyll fluorescence imaging technique.In this study,the average accuracy of WT and oscal for the prediction of drought stress degree were 78.3%and 88.3%,respectively.The fusion image of WT is F=-9.54 Qn_D2+2.84-NPQ_L2-8.20 Qp_L4+5.75 Ql_L1,and oscal is F=-2.47-NPQ_L2-0.85 Qn_D2+2.22R·fd_Lss-6.13·Rfd_L2-1.54·Fv/Fm_D1+6.94—Rfd_L1+0.48·Fv_Lss,displaying strongest contrast among different degrees of drought stress in Arabidopsis.