Field Detection Of Citrus Huanglongbing By Near-infrared Spectroscopy And The Effection Of Pesticides

Citrus Huanglongbing(HLB)is one of the bacterial diseases that seriously endanger the world citrus industry.At present,there is not any report about the special effect pesticide for curing HLB.Therefore,the development of real-time detection technology for HLB is of particularly important for preventing and controlling HLB and ensuring the healthy development of the citrus industry.There are many disadvantages of traditional molecular detection technology,such as high cost,long cycle,and destruction of plant tissue.On the contrary,near-infrared spectroscopy detection technology has many advantages,for example,real-time,non-destructive,and suitable for large-scale production,so that near-infrared spectroscopy detection technology is popular with HLB detection.This study collected near-infrared spectra of citrus leaves under the basis of previous studies.Using near-infrared spectroscopy analysis technology combined with chemometric analysis methods,establishing models of HLB of four varieties of citrus.Discussed the portable of Micro NIR 1700 used for the qualitative identification of HLB,analyzing the effects of the pesticide of leaf surface.The research results will provide an application reference for the near-infrared detection technology in the field of accurate HLB discrimination.The main research results are as follows:1.Establishming the PLS-DA model of Tiancheng,Nianju,Shatangju,Chuntianju.The near-infrared spectrum(905～1650nm)collected by Micro NIR 1700 was used to establish PLS-DA models with different pretreatment methods.The results prove that the best pretreatment methods used by different varieties are different.The spectrum of Tiancheng is pretreated with first-derivative,RMSEV=0.13159,R_v~2=0.92135,which makes the model satisfy the R_v~2close to 1 and the RMSEV close to 0.By the way,the best pretreatment method for spectrum of Shatangju is the first-derivative,RMSEV=0.05129,R_v~2=0.98926.The best pretreatment method for Chuntianju spectrum is the first-derivative,RMSEV=0.16325,R_v~2=0.89361.The best pretreatment method for Nianju is the first-derivative combined with multivariate scattering correction,RMSEV=0.09045,R_v~2=0.96538.2.Field detection show that the model has a correct recognition rate of HLB higher than 97%for citrus leaves of the same variety as the calibration,which meets the field detection requirements.The model of Tiancheng,Nianju,Shatangju have strong specificity and can only be used for the same variety as the calibration.The Chuntianju model can be used not only for the Chuntianju leaves,but also Tiancheng leaves,which correct recognition rate is higher than 79%.The chi-square test result(P>0.05),which proves that the variety is one of the factors that determine the accuracy of the model prediction.3.Effects of chemicals on near-infrared spectrum of the leaf surface.In this study,Bifenthrin,Imidacloprid,Spirotetramat-enol and Chlorpyriphos were selected.Micro NIR1700 for quick identification of leaf surface agents.The concentration of 100 mg/L,1000mg/L,10000 mg/L,Micro-NIR 1700 is used to collect the spectral data of the citrus leaf sprayed with the pesticide,The component analysis shows that after spraying different types and concentrations of chemicals,the near-infrared spectrum of the leaves is different.4.Analysising the influence of pesticides on the prediction accuracy of the model.The Nianju model is used to predict the correct recognition rate of HLB,which leaves sprayed with four Bifenthrin,Imidacloprid,Spirotetramat,and Chlorpyrifos.The results prove that the types and concentrations of pesticides will affect the accuracy of the model.Among them,the pesticide concentration is inversely proportional to the correct recognition rate of the model.By the way,spraying pesticides of Bifenthrin,Chlorpyrifos,and Spirotetramat,the accuracy of the Nianju model decreased significantly,and more false negative samples appeared.Although the prediction accuracy of the Nianju model decreased after spraying Imidacloprid,but it can still reach more than 88%,which can meet the needs of field testing.