| Cotton is one of the most important economic crops in the world, which is also a special good for national economy and people’s livelihood. Cottonseed, more than 10 million tons of cottonseed and 6 million tons of cottonseed cake annually in China, is an important by-product of cotton production with high contents of protein and oil. Inorganic elements are necessary for plant growth and development, while it could not be synthesized in plants and must be absorbed from soil. The content of inorganic elements has important influence on cotton’s growth. Cottonseeds are used as livestock feel and edible oil. Inorganic elements, especially trace elements have vital effect on the health of animals. A series of disease to poultry would be caused when the ratio of inorganic elements are imbalanced, and would bring serious damage to animal husbandry. Therefore, measuring the content of inorganic elements has important meaning not only for cotton production, but also to the development of animal husbandry. The present work is to establish an accurate and stable calibration model for inorganic elements in cottonseed meal. The main results are as follows:(1) A total of 288 cottonseed samples were collected from different environments, their contents of 20 inorganic elements were determined by inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma optical emission spectrometry (ICP-OES).(2) Standard normal variate (SNV) and first derivative were applied to spectra preprocessing, and partial least squares (PLS) and least-squares support vector machine (LSSVM) regression algorithm were used to develop the calibration model for determination of 20 inorganic elements in present work. In addition, Monte Carlo uninformation variables elimination (MCUVE) as variable selection method were employed to improve the performance of models.(3) The performance of MC-UVE-LS-SVM regression algorithm was validated to be the best calibration model for contents of Mn and Cr. The coefficient of determination in calibration set (R2c) for Mn and Cr was 0.9763 and 0.9671, respectively. The coefficient of determination in prediction set (R2p) for Mn and Crwas 0.9486 and 0.9343, respectively. and the residual predictive deviation (RPD) for Mn and Cr was 4.3354 and 03.7872, respectively. the root-mean square error of coss-validation (RMSECV), the root-mean square error of calibration (RMSEC), and the root-mean square error of prediction (RMSEP) for Mn were 2.1013,1.2420 and 2.0108, respectively, and those of Cr were 0.5465,0.2495 and 0.2661 respectively. It showed that the calibration model for Mn and Cr in cottonseed meal were achieved with good accuracy and robustness (R>0.9, RPD>3), which could replace the traditional method to determine their contents.(4) The performance of MC-UVE-PLS regression algorithm was the best calibration model for K content, with the value of R2C, R2P, RPD, RMSECV, RMSEC, and RMSEP were 0.9394,0.8114,2.2224,0.4946,0.3116, and 0.5616, respectively. This model was accuracy and robustness as well for determination of K. content.(5) The best model for P and S was PLS regression algorithm, and the best model for Mg was the MC-UVE-LS-SVM model. The value of R2C, R2P, RPD, RMSECV, RMSEC, and RMSEP for P were 0.8621,0.6769,1.6701,0.4155,0.2643 and 0.3408, respectively. Those of S were 0.8731,0.7153,1.8438,0.1501,0.1128, and 0.1482, respectively, and those of Mg were 0.8668,0.6606,1.6888,0.1417,0.1009, and 0.1295, respectively. The models for P, S, and K were not good enough to substitute the traditional analytical method, but it could be used to screen a large number of samples.(6) The value of R2C in PLS model for Ca, Ni, Se, and Sr were 0.8619,0.8179, 0.8117, and 0.8069, respectively, their Revalue were 0.3988,0.4353,0.3615, and 0.3596, respectively. It could be improved by increasing of sample number.(7) The PLS model for Cu, Na, Fe, Zn, Al, B, Co, Cd, As, and Pb had bad prediction ability, and the R c value for those inorganic elements were 0.5275,0.6840, 0.2501,0.5542,0.4047,0.2281,0.5537,0.7421,0.2274, and 0.3756, respectively. Their Revalue were 0.4634,0.3541,0.1694,0.3409,0.4133,0.2123,0.3523,0.4132, 0.2047, and 0.2589, respectively. With the lower value of R2C and R2P, it was difficult to develop calibration model for those inorganic elements in present work.As a rapid, non-destructive, pretreatment simple, low cost, effective, and reagent-free analytical method, near-infrared spectroscopy could be used to accurately measure some inorganic elements in cottonseed meal. In cotton breeding, which could be used for screening of cotton germplasm with high or lower inorganic elements, with the benefits such as:shorten the time of breeding, improve the efficiency of breeding, and save money and time. |
PDF Full Text Request