| Nitrogen is very important for the growth and development of citrus.Real-time and non-destructive monitoring of nitrogen nutrition status of citrus is of great significance for accurate management of nitrogen nutrients.The nitrogen in plants can be divided into three categories according to different forms:assimilable nitrogen（AN）,structural nitrogen（SN）and functional nitrogen（FN）.The three forms of nitrogen are in dynamic changes.The content and distribution of each component in the leaves have a certain indication on the physiological and biochemical reactions of plant leaves,and show different trends with the growth period.In recent years,with the application of visible/near-infrared spectroscopy in crop growth monitoring,its continuous and fine characteristics provide new means and approaches for plant nitrogen nutrition diagnosis.In this study,the changes of different forms of nitrogen content in citrus leaves at different growth periods and their correlation with fruit yield and quality were analyzed.The visible/near-infrared spectral characteristics of citrus leaves and their quantitative relationship with FN content in leaves under different nitrogen treatments during fruit swelling period and fruit coloring period were analyzed.A non-destructive monitoring model of FN content in citrus leaves based on visible/near-infrared spectroscopy was constructed.The non-destructive monitoring model was further used to estimate the FN content of leaves at fruit swelling period and fruit coloring period under different nitrogen application treatments.The recommended nitrogen topdressing formula was used to calculate the amount of nitrogen topdressing under the regulation of nitrogen application,so as to realize the non-destructive diagnosis of citrus nitrogen nutrition and the regulation of nitrogen application.The main results are as follows:1.The contents of three forms of nitrogen in leaves of citrus at different growth periods under different nitrogen application treatments were determined.The results showed that the content of FN in citrus leaves was the highest,with an average of 17.06g·kg-1,accounting for about 60%of total nitrogen,followed by SN(8.03 g·kg-1)and AN(3.06 g·kg-1).Among them,the leaf AN and SN had little change with the growth periods under different nitrogen application treatments(the difference was 1.75 g·kg-1and 2.61 g·kg-1),and the difference between different nitrogen application treatments was not significant.The variation of leaf FN in different nitrogen application treatments was small from fruit mature period to summer shoot development period（the difference was 19%）,and the difference was not significant,while the variation was large at fruit swelling period and fruit coloring period(the difference was 5.3 g·kg-1,accounting for 27%),and the difference was significant among nitrogen application treatments.Fruit yield and quality of adult citrus increased first and then stabilized with the increase of nitrogen application rate.The correlation between three kinds of nitrogen forms in citrus leaves at different growth periods and fruit yield and quality was analyzed.The results showed that the correlation between leaf FN content and fruit yield was the highest at fruit swelling period,with the correlation coefficient of 0.75.The correlation between leaf FN content and fruit quality was the highest at fruit coloring period,with the correlation coefficient of 0.79.The FN content of citrus leaves during fruit swelling period and fruit coloring period has certain guiding significance for understanding the nitrogen nutritional status of citrus and recommending nitrogen application.2.The reflectance spectra of citrus leaves under different nitrogen application treatments were measured by visible/near-infrared spectrometer at fruit swelling period and fruit coloring period,and the FN content of citrus leaves was determined.The spectral curve characteristics of citrus leaves under different nitrogen treatments were analyzed.In the visible region of 350-700 nm,the leaf spectral reflectance decreased with the increase of nitrogen application rate.However,in the near infrared region of700-1350 nm,the leaf spectral reflectance increased with the increase of nitrogen application rate.The correlation analysis between leaf FN content and leaf original spectrum and the first derivative spectrum showed that the correlation between leaf FN content and leaf original spectrum and the first derivative spectrum was different at different growth periods of citrus.The maximum correlation coefficient of fruit swelling period was 0.8,and the maximum correlation coefficient of fruit coloring period was 0.6.The bands with correlation coefficients greater than 0.6 at fruit swelling period and fruit coloring period were selected as sensitive bands for subsequent modeling analysis.Based on the spectral full-band and sensitive-band,three types of dual-band vegetation indices（DVI,RVI,NDVI）,partial least squares regression and three machine learning methods（back propagation neural network,random forest,support vector machine）were used to construct the non-destructive monitoring model of FN content in citrus leaves,and the modeling accuracy was compared.The results showed that the full-band back propagation neural network（BPNN）model had the best prediction effect on the FN content of citrus leaves during the fruit swelling period and fruit coloring period.The R2 of validation set was 0.78 and 0.74,and the RMSE was 0.83 g·kg-1 and 1.19g·kg-1,respectively.In the fruit swelling period,the random forest（RF）model constructed by the sensitive band also had a good prediction effect on the FN content of citrus leaves.The R2 of validation set was 0.74 and the RMSE was 0.83 g·kg-1.However,in the fruit coloring period,the accuracy of FN content model constructed by the sensitive band was poor.The effects of three spectral variants and six spectral pretreating methods on the modeling accuracy were compared and analyzed.The results showed that the modeling accuracy of the full-band BPNN model could be improved by the standard normal variant transformation pretreatment of the original spectrum at the fruit swelling period and the fruit coloring period.The R2 of validation set was 0.78 and 0.76,and the RMSE of the validation set was 0.82 g·kg-1 and 1.13g·kg-1,respectively.3.The non-destructive monitoring model of FN content in citrus leaves was used to invert the FN content in citrus leaves at fruit swelling period and fruit coloring period under different nitrogen application levels,and the recommended nitrogen application amount was calculated by the formula of nitrogen topdressing amount.The regulatory nitrogen application technology based on the FN content in citrus leaves was preliminarily established and its effect was evaluated.The results showed that compared with the control nitrogen application treatment,the yield of citrus fruits under the regulatory nitrogen application treatment increased,and the yields of Nr1 and Nr3 increased by 5.49 kg/tree and 4.43 kg/tree compared with N1 and N3,respectively（increasing by 48%and 40%）.Compared with N1 treatment,the soluble solids content of citrus fruit in Nr1 treatment was significantly increased.The effect of nitrogen application regulation on the horizontal and vertical diameters of citrus fruits was not significant.Compared with the control nitrogen treatments N1 and N3,the nitrogen partial factor productivity of Nr1 was reduced by 10%,and that of Nr3 was increased by123%.The agronomic efficiency of Nr1 and Nr3 increased by 290%and 364%,respectively.These results indicated that the regulation of nitrogen application based on the non-destructive monitoring model of FN content in citrus leaves could alleviate the effects of insufficient or excessive nitrogen application on yield and quality of citrus fruits to a certain extent. |
