Monitoring Rice Under Cadmium Stress In Soil Based On Chlorophyll Fluorescence Reflectance Spectroscopy Characteristics

Leaf is an important indicator of rice response to environmental stress.Chlorophyll fluorescence can quickly and sensitively respond to changes in crop habitat under heavy metal stress,and leaf characteristic spectra can better characterize heavy metal stress information in crop populations.The non-destructive perception of cadmium(Cd)content in rice canopy leaves through chlorophyll fluorescence reflectance spectroscopy characteristics can directly diagnose the degree of Cd stress in leaves and indirectly estimate the Cd pollution status in rice.This has important scientific significance and application value for early non-destructive monitoring of Cd polluted rice and ensuring food security.This study is based on rice potted soil culture and hydroponic Cd stress experiments to obtain information on chlorophyll content,chlorophyll fluorescence parameters,and reflectance spectra of leaves at different growth stages and different leaf positions.The differences in response and sensitive spectral characteristic parameters of leaves at different leaf positions to Cd stress are clarified,and a monitoring model that can estimate the Cd pollution status of rice is constructed.(1)The distribution characteristics of Cd content and chlorophyll content in rice canopy leaves under Cd stress were studied.The results showed that there were significant vertical distribution characteristics of Cd and chlorophyll contents in rice canopy leaves.The correlation between the Cd content in the top 1 leaf of tillering stage and the Cd content in mature brown rice is the highest,with R~2=0.898.It is the best period and leaf for monitoring Cd content and brown rice Cd content.(2)A monitoring model for Cd content in rice canopy leaves was constructed based on chlorophyll fluorescence parameters.The results showed that as the Cd concentration increased,the maximum photochemical efficiency(Fv/Fm)gradually decreased,while the initial fluorescence(Fo)and non photochemical quenching coefficient(NPQ)gradually increased.The Fv/Fm value decreases with the decrease of leaf position and can be used as a potential sensitive parameter to characterize the vertical distribution of leaves.The results of the stepwise regression equation show that the accuracy of the water culture experiment is better than that of the soil culture experiment,with R~2=0.690.(3)A monitoring model for Cd stress in rice at canopy and single leaf scales was constructed based on spectral vegetation index and machine learning methods.The results show that compared with the accuracy of modeling and verification,the single leaf scale is superior to the canopy scale in both methods.The soil cultivation experiment based on the Difference Vegetation Index(DSI)had the highest monitoring accuracy of Cd content in the top 1 leaf(R~2=0.818),while the water cultivation experiment based on DSI had the highest monitoring accuracy of Cd content in the top 2 leaf(R~2=0.848).Based on machine learning methods,the top 1leaf of soil culture experiment based on partial least squares(PLS)has the highest accuracy(R~2=0.738),while the top 1 leaf of soil culture experiment based on BP neural network(BP)has the highest accuracy(R~2=0.817).In summary,the Cd content monitoring model constructed based on the top 1 and top 2 leaves has the best prediction and validation effect,and is the best leaf for monitoring Cd stress,providing technical support for rapid and non-destructive monitoring of Cd stress in rice.