Nutritional Components Of Bothriochloa Ischaemum And Lespedeza Davurica By Near Infrared Spectroscopy (NIRS) Analysis Technology

The experiment predicted the contents of conventional nutrients in Bothriochloa ischaemum and Lespedeza davurica by using the Fourier Transform Infrared Spectroscopy and analyzed the factors that might affect the predictive results of models.With 100 and 35 samples for B.ischaemum and L.Michx.respectively,the experiment built near infrared prediction models of CP,EE,Ash,NDF,ADF,Ca and P for each species and used internal and cross validation to understand the predictive effects of models.The results showed:(1)according to the statistics,the contents of CP,EE,Ash,NDF,ADF,Ca,P in Bothriochloa ischaemum were 8.25%,1.79%,9.25%,65.89%,40.94%,0.40%,0.23%respectively,and the contents of CP,EE,Ash,NDF,ADF,Ca,P in Lespedeza Michx.were 23.98%,2.22%,11.78%,51.46%,26.86%,0.27%,0.18%respectively.(2)Near infrared diffuse reflection spectroscopy was applied to detect the contents of conventional nutrients in Bothriochloa ischaemum,with effective information spectrum range of 4000cm-1-10000cm-1.After first derivative processing to original spectrum and the establishment of calibration model through PLS,we can clearly see that the prediction of calibration models of NDF,ADF,Ca and P performed well.R values of them were 0.9313,0.9760,0.9645,0.9605 apiece,RMSEC values of them were 0.648,0.640,0.0110,0.0120 apiece,and the coefficient of association of cross validation all reached above 0.95,at the same time,the RMSECV ranged from 0.910^0.0117 which was minimum.In terms of the contents of CP,EE,Ash,Calibration models showed that it got 0.9384,0.9403,0.954 of R value apiece simply and 0.183,0.127,0.268 of RMSEC apiece.However the R value of internal validation reached 0.9617,0.9562,0.9541 respectively and the RMSEP were relatively small with 0.193,0.193,0.286 apiece,which indicated that these three nutrients got good near infrared prediction models with high degree of fitting.(3)Near infrared diffuse reflection spectroscopy was applied to detect the contents of conventional nutrients in Lespedeza Michx.,with effective information spectrum range of 4000cm-1-10000cm-1.After first derivative processing to original spectrum and the establishment of calibration model through PLS,we can clearly see that the prediction of Calibration models of CP,EE,Ash,NDF,ADF,Ca and P performed well.R values of them were 0.9998,1.0000,0.9827,0.9827,0.9940,0.9999,0.9667 apiece,RMSEC values of them were 0.0301,0.00345,0.185,0.9827,0.9827,0.9999,0.9667 apiece,and the coefficient of association of internal validation were 1.0000,0.9821,0.9459,1.0000,1.0000,0.9997,0.0390 apiece.RMSEP were 0.00143,0.9821,0.286,0.00143,0.00143,0.0741,0.0116 apiece.However,the standard deviation of prediction set of P which was 0.0390 showed that validation set was too concentrated and not evenly distributed so that a cross validation was needed.R value(the coefficient of association of cross validation)ranged from 0.9743-0.9999 and RMSEV value fluctuated from 0.00266-0.753,which implied an evenly distributed Scaling chemical values featuring good capacity of prediction for samples within the calibration range.The coefficient of association of all indicators reached above 0.95,which showed a good fitting between the true values of models and the prediction values of the near infrared spectroscopy and high capacity to predict the contents of conventional nutrients in Lespedeza Michx..All in all,the near infrared spectrum calibration models of contents of conventional nutrients in Bothriochloa ischaemum performed well.The true values of models for CP,EE,Ash,NDF,ADF,Ca,P were close to their prediction values of the near infrared spectroscopy,so the models can well predicted the contents of conventional nutrients in Bothriochloa ischaemum.In term of Lespedeza Michx.,the Scaling chemical values of CP,EE,Ash,NDF,ADF,Ca,P distributed evenly which was good for the prediction of samples within calibration range,and models with high degree of fitting showed strong capacity in predicting contents of conventional nutrients in Lespedeza Michx..By comparing coefficient of association and standard deviation as well as doing internal and cross validation,we can clearly see that the contents of conventional nutrients and the overall prediction efficiency of near infrared models of Lespedeza Michx.were better than that of Bothriochloa ischaemum.