Building Near-infrared Reflectance Spectroscopy Models Of Glucosinolate Contents In Brassica Napus L.

Brassica napus L. is an important oil crop in our country, and its planting area and total production are primacy in China. Its rapeseed cake contains of rich proteins, they are excellent protein resources. But the degradation of glucosinolates in rapeseed cake can be toxic, which affecting the nutritional value of feed. Reducing the glucosinolate contents in rapeseed is an important task of rapeseed quality breeding. To achieve this goal, the precise determination of glucosinolate contents is very significant.At present, determination of glucosinolates contents generally using high performance liquid chromatography. But filtering large numbers of samples by this method is not Feasible. In recent years, near infrared spectroscopy technology as an important means of selection in crop breeding is widely used in filtering large quantities of samples. To explore the feasibility of rapidly detecting glucosinolates constituent content by near infrared reflectance spectroscopy technology, we selected266representative samples and analysed glucosinolate contents by HPLC (High Performance Liquid Chromatography), processing the data by using near infrared spectroscopy analyzer and WinISIⅢ software. In the end, we built near infrared reflectance spectroscopy inspection models of glucosinolate contents in Brassica napus L.The main contents and results are as follows:1、In this study, glucosinolate constituent contents of266samples in Brassica napus L. analysis results: The average content of Desulfoprogoitrin(DSPRO) was the highest and its range was maximum, the mean of DSPRO was51.19μmol·g-1, the range was0.60-122.21μmol·g-1; Secondly, the mean of Desulfogluconapin (DSGNA) was34.70μmol·g-1, the range was0.66-108.49μmol·g-1; The mean of Desulfo-4-hydroxyglucobrassicin(DS4-OHGBS) was5.20μmol·g-1, the range was0.51-24.13μmol·g-1; The mean of Desulfoglucobrassicanapin(DSGBN) was3.35μmol·g-1, the range was0.25-17.33μmol·g-1; The mean of Desulfogluconasturtiin (DSGST) was3.38μmol·g-1, the range was0.69-11.23μmol·g-1; The last three glucosinolates constituent content was low,and with little change.2、This research building near-infrared reflectance spectroscopy(NIRS) models of glucosinolate contents in Brassica napus L is based on modified partial least squares method(MPLS), and compared these models with the models by partial least squares method(PLS) and principal component regression analysis(PCR). The results showed: PCR and PLS regression analysis methods had respective advantages, both were effective methods to solving multiple regression problem; NIRS models of MPLS is better than those models by PLS and PCR.3、In glucosinolates constituent content NIRS models, the models of DSPRO and DSGNA had good accuracy. The results showed that the cross-validation correlation coefficient(1-VR) of Near Infrared Spectroscopy models of DSPRO and DSGNA were0.9372and0.9009; The external validation correlation coefficient(RSQ) were0.962and0.937, which suggests that the models were satisfactory and could be used for the rapid screening of rapeseed varieties.However, the R squared (RSQ) of DS4-OHGBS, DSGBN and DSGST were0.147,0.219and0.419;1-VR were0.175,0.128and0.304; which were unsatisfactory and needed further research.