Determination of Geographical Origin and Protein Content of Acacia Gums Using Infrared Spectroscopy and Chemometrics

ZOU Xiaobo1, Alaa Kamal Mohmmed KHAIR KHOGLY1, SHI Jiyong1, Mel HOLMES2

(1. School of Food and Biological Engineering, Jiangsu University, Zhenjiang 212013, China;2. School of Food Science and Nutrition, the University of Leeds, Leeds LS1 3AB, United Kingdom)

Abstract: The feasibility of using Fourier transform infrared (FTIR) spectroscopy and chemometrics as a rapid and noninvasive technique to determine the geographical origin and protein content of Sudanese Acacia gums was investigated.Seventy-two samples of Acacia gums were collected from six different regions (12 samples from each region). Linear discriminant analysis (LDA) was used to discriminate the geographical origin of Acacia gums, and backward in terval partial least squares (Bi-PLS) was applied to build a prediction model for the protein content of Acacia gums. The results showed that the recognition rates of LDA for calibration set (48 samples) and prediction set (24 samples) were both 100% when the fi rst 6 principal components were used. In addition, Bi-PLS yielded a good prediction model (RP= 0.937 3 and RMSEP =0.173%) for protein content by using the optimal combination of 4 out of 20 spectral intervals. Hence, FTIR spectroscopy coupled with chemometrics can be considered as a valid approach for the determination of the geographical origin and protein content of Acacia gums.

Key words: Acacia gums; Fourier transform infrared (FTIR) spectroscopy; geographical origin; protein content;linear discriminant analysis (LDA); backward interval partial least squares (Bi-PLS)

Acacia gums, also called gums Arabic, are the dried exudates extracted from the stem and branches of Acacia trees. They are often used as non-digestible food ingredients in the foodstuffs, such as an emulsifier in the manufacture of soft drinks and a stabilizer in fermentation milk[1]. The chemical composition of gum Arabic is complex and consists of a group of macromolecules characterized by a high proportion of carbohydrates (-97%), which are predominantly composed of D-galactose and L-arabinose units and a low proportion of proteins (<3%)[2]. Acacia gums from different Acacia species exhibited different characteristics, and even gums in the same species bear some difference because of their geographical origin. For example, Acacia senegal(A. senegal) has a higher degree of branching and better emulsifying properties than Acacia seyal (A. seyal), gum does[3]. This is because their biochemical composition and molecular characteristics can vary, depending on internal and external factors (Acacia specie, tree location, or weather conditions) and postharvesting processes (storage conditions,filtration, spray drying, irradiation, or heat treatments)[4].Hence, recognition of these Acacia gums is of great signifi cance for the market and the end use.

Traditionally, the identifi cation of the speices or quality of Acacia gums can be conducted by sensory analysis, such as observing the color and smelling, which are unfortunately impeded the subjectivity of human. Other physicochemical methods, such as detection of the sugar and nitrogen content[5], and component analysis by using combined gel permeation chromatography and multi-angle laser light scattering (GPC-MALLS)[6], are either destructive to samples or time-consuming. Nuclear magnetic resonance(NMR) was also regarded as an effective technology to identify A. senegal and A. seyal, but it may be inaccessible for industrial high throughput screening purposes[7]. As a result, a rapid, non-invasive and economical method for distinguishing the speices or the quality of Acacia gums is highly desired.

Fourier transform infrared (FTIR) spectroscopy, which keeps the advantages of being robust, cost-effective and nondestructive to sample, has been widely used as an effective identifi cation and quantitative analysis technique[8]. In recent years, many studies have shown the feasibility of using FTIR spectroscopy in discriminating similar foodstuffs and medicines, such as walnut oil[9], Cortex Eucommiae[10],minced beef[11], and detecting specific components, such as free acids, invertase, moisture, hydroxymethyl furfural,polyphenol oxidase and electrical conductivity[12-14].Quantification of gums is another interesting issue for determining the quality of the gums. For A. senegal and A. seyal gums, one of the most signifi cant difference is their protein content (-2.5% for A. senegal, -1% for A. seyal gums)[15]. Accordingly, the protein content can be used as an imporatant quality index of these Acacia gums.

Sudan is a dominant leader in Acacia gum production in the world[16]. The aim of the present work is discriminating the A. senegal and A. seyal gums collected from Sudan,and predicting their total protein content by using FTIR spectroscopy and chemometrics.

1 Materials and Methods

1.1 Materials

Two categories of Acacia gum samples (n=72) were collected from different regions in Sudan, namely A. senegal(n=36) (Group (A), n=12∶ West Bara Locality in North Kordofan State; Group (B), n=12∶ Ennuhud Locality in West Kordofan State; Group (D), n=12∶ Acacia gum Ltd., Sudan)and Acacia sayal (n=24) (Group (C), n=12∶ South Kordofan State; Group (E), n=12∶ Acacia gum Ltd., Sudan). One other Acacia gum sample was obtained from China (Group (F),n=12∶ Siyuan Biology Ltd., Henan, China). All other reagents are of analytical grade purchased from Sinopharm Chemical Reagent Co. Ltd., Shanghai, China.

1.2 Methods

1.2.1 Determination of total protein content

Firstly, all raw Acacia gum samples were dried ina forced draught oven (Shanghai Yi-Heng Machine Co.,Shanghai, China) at 50 ℃ for 24 h and then mashed into powder by using a cyclone mill. The total protein contents of the gums were measured by using the Kjeldahl method[17].

1.2.2 FTIR spectroscopy

FTIR spectra of gum powders were collected by using a FTIR spectrometer (Thermo Scientifi c Nicolet iS50, Thermo Fisher, USA). The spectra were obtained at reflectance mode from 650–4 000 cm‒1at a resolution of 4 cm‒1and the total number of scans was 32. Three spectra were recorded for each sample and the average spectrum was used for subsequent data analysis.

1.2.3 Chemometrics

Pri nciple component analysis (PCA) is a statistical procedure which is able to reduce the dimensions of the data set by using an orthogonal transformation[18]. After PCA, the useful information can be extracted from the original spectral data by eliminating overlapping information and the remained dimensions are defined by principal components (PCs).It is able to provide the visual graphical information for determining differences within and between cluster trends[19].

Lin ear discriminant analysis (LDA) is a c lassical statistical approach for feature extraction and dimension reduction[20]. LDA can classify the objects into groups by estimating the distance between each observation form all group centers[21]. It manages to find out the optimal transformation (projection) that minimizes the intraclass distance and maximizes the inter-class distance simultaneously in order to achieve the maximum discrimination.

Partial least squares (PLS) regression is one of the multiple linear regression methods[22]. PLS is useful when number of predictors (i.e. spectral peaks) is much higher than number of samples in data set[23]. Backward interval partial least squares (Bi-PLS) regression is an extension of PLS. The basic principle of Bi-PLS is as follows[24]∶ The fullspectrum region is split into a number of equidistant spectral subintervals. Then, PLS models are calculated with each subinterval left out. The fi rst left out interval is the one that when it gives the poorest performing model with respect to the root mean square error of cross-validation (RMSECV).

A cross-validation process was used in model validation with leave-one-out method. The performance of the regression models was evaluated according to the correlation coefficient of calibration set (Rc) and prediction set (RP), RMSECV and root mean standard error of prediction set (RMSEP). Gen erally, a good model should have high correlation coefficients along with low RMSECV and RMSEP[25].

1.3 Data analysis

Matlab V7.0 (MathWorks, USA) was used for data processing under Windows 7.

2 Results and Analysis

2.1 The total protein content of Aca cia gums

All 72 samples were randomly separated into two subsets, including the calibration set (48 samples) used to build models and the prediction set (24 samples) used to test the robustness of models. The total protein content in Acacia gums is shown in Table 1. It can be seen that the total protein contents of A. senegal (Groups (A), (B), and (D)) and A. sayal (Groups (C) and (E)) were in the range of 1.53%–2.22% and 0.56%–1.87%, respectively. These results are close to the Sud anese Standards and Metrology Organization (SSMO) in which the protein contents for A. senegal and A. sayal gums are recorded to be in the range of 1.50%–2.70% and 0.7%–1.0%, respectively. In comparison,the Acacia gums obtained from Group (D) had the highest protein content, while the Acacia gums obtained from China had the lowest protein content. These difference were closely associated with the species of the Acacia gums and also with the gum origin, age, storage conditions, and so forth[26].

Table 1 Protein content of Acacia gums

Note∶ SD. Standard deviation.

2.2 Spectra investigation

The FTIR spectra for the Acacia gum samples acquired in the range of 650–4 000 cm–1are shown in Fig. 1. In order to avoid the strong absorption of water, especially in the amide I band region (1 720–1 580 cm–1)[17], Acacia gums were fully dried before spectra collection. For all of the Acacia gums, there are four main characteristic spectral bands. No significant difference can be observed between the FTIRspectra of different groups. This is because all of them consist of similar kinds of chemical components. Table 2 shows the wavenumbers and attribution of the characteristic spectral bands of these Acacia gums.

Fig. 1 FTIR spectra of Acacia gums

Table 2 Wavenumbers and ttribution of the characteristic spectral bands in FTIR spectra of Acacia gums

2.3 Principle component analysis of FTIR spectra of Acacia gum

Prior to regression analysis, an exploratory analysis was performed in order to investigate any trend of discrimination among the Acacia gums from different regions. PCA is the most widely applied linear projection method for unsupervised exploratory multivariate data analysis to visualize the similarities and differences between the spectra[29]. Before PCA, the raw mean spectra were preprocessed by using standard normal variate (SNV). Fig. 2a shows the score plot of the two-dimensional component space of six categories of Acacia gum samples. The accumulated variance contribution rates were 99.33% for the top two PCs.The classifi cation trend of these six categories of Acacia gum samples can be also observed from the score plot. The Acacia gum samples obtained from China were clearly separated from that obtained from Sudan, indicating the obvious difference of the chemical constituents between the Acacia gum from different geographical origins. In order to furtherly visualize the difference between Acacia gum collected from Sudan, PCA was conducted for the Acacia gum regardless of Group (F) and the score plot of the two-dimensional was shown in Fig. 2b. It can be seen that the five categories of Acacia gum from Sudan are clearly distinguished from each other, demonstrating their intrinsic difference. Nevertheless,PCA is not able to defi ne the boundaries of the six categories to discriminate the Acacia gum samples so that further analysis was carried out in the next sections.

Fig. 2 S core cluster plot of top two principal components (PCs) for Acacia gum samples

2.4 Determination of geographical of Ac acia gum by LDA

LDA focus on finding the optimal boundaries between the classes[30]. The number of PCs is crucial to the performance of the LDA discrimination model. The discrimination rates by cross-validation were used to optimize the number of PCs. Fig. 3 shows the discrimination rates of LDA model according to different PCs by cross-validation.The optimal LDA models were achieved when the number of PCs was 6, and meanwhile the discrimination rate was up to 100% for both calibration set and prediction set. These results indicated that an ideal separation can be achieved for these six categories of Acacia gum. On the basis of LDA,a cluster analysis was conducted. Samples were grouped in clusters based on their nearness or similarity. Fig. 4 shows the dendrogram of Acacia gum samples. All the Acacia gums in the same group gathered together and were distinguished from other groups. In addition, when all the Acacia gum were clustered into two big groups, Group (E) and Group (F) were defi ned into the same group. This may be due to the reason that Group (E) and (F) had similar protein contents, which were much lower than those of other groups.

Fig. 3 Rec ognition rates of LDA models

Fig. 4 Dendrogram of Acacia gum samples

2.5 Prediction of the total protein content in Acacia gum by Bi-PLS

Fig. 5 Spectral interval selected by Bi-PLS (a) and actual versus predicted protein contents (b) in calibration set and prediction set

In this study, the Bi-PLS approach was applied for the quantification of the total protein contents in Acacia gums.Bi-PLS aimed to select the most important subintervals in the whole data matrix. The spectra data set was split into some subintervals, and then PLS models were calculated when each subinterval left out. When the whole spectrum region was split into 20 subintervals, the optimal Bi-PLS model for the total protein content was obtained with the combination of four subintervals, resulting in the lowest RMSECV of 0.155%.As shown in Fig. 5a, the optimal subintervals are [1 4 7 19],which correspond to 650–817, 1 152–1 319, 1 652–1 819 and 3 666–3 833 cm–1, respectively. Fig. 5b presents the performance of the Bi-PLS model on the calibration and prediction set. For the prediction set, RPwas 0.937 3 and RMSEP was 0.173%,demonstrating that the Bi-PLS model can be used to predict the total protein content of Acacia gums.

3 Conclusion

FTIR spectroscopy technique combined with chemometrics was successfully established and employed to distinguish the geographical origins and predict the total protein contents of Acacia gums. The results showed that the optimal LDA model was achieved in determining the geographic origins with 100% of discrimination rate for both the calibration and prediction set when the number of PCs was 6. The Bi-PLS model showed a good performance in predicting the total protein content (RP= 0.937 3 and RMSEP = 0.173%) by using the optimal combination of 4 spectral subintervals among 20 subintervals. FTIR spectroscopy coupled with chemometrics can be considered as a valid approach for determination of the geographical origin and protein content of Acacia gums.

[1] DA UQAN E, ABDULLAH A. Utiliz ation of gum Arabic for industries and human health[J]. American Journal of Applied Sciences, 2013,10(10)∶ 1270-1279. DOI∶10.3844/ajassp.2013.1270.1279.

[2] IS LAM A M, PHILLIPS G O, SLJIVO A, et al. A review of recent developments on the regulatory, structural and functional aspects of gum arabic[J]. Food Hydrocolloids, 1997, 11(4)∶ 493-505.DOI∶10.1016/S0268-005X(97)80048-3.

[3] EL MANAN M, AL-ASSAF S, PHILLIPS G O, et al. Studies on Acacia exudate gums∶ Part Ⅵ. Interfacial rheology of Acacia senegal and Acacia seyal[J]. Food Hydrocolloids, 2008, 22(4)∶ 682-689.DOI∶10.1016/j.foodhyd.2007.02.008.

[4] LO PEZ-TORREZ L, NIGEN M, WILLIAMS P, et al. Acacia senegal vs. Acacia seyal gums-Part 1∶ composition and structure of hyperbranched plant exudates[J]. Food Hydrocolloids, 2015, 51∶41-53. DOI∶10.1016/j.foodhyd.2015.04.019.

[5] DA OUB R M A, ELMUBARAK A H, MISRAN M, et al.Characterization and functional properties of some natural Acacia gums[J]. Journal of the Saudi Society of Agricultural Sciences, 2016.DOI∶10.1016/j.jssas.2016.05.002.

[6] AL ASSAF S, PHILLIPS G, WILLIAMS P. Studies on Acacia exudate gums∶ part Ⅱ. Molecular weight comparison of the Vulgares and Gummiferae series of Acacia gums[J]. Food Hydrocolloids, 2005,19(4)∶ 661-667. DOI∶10.1016/j.foodhyd.2004.09.003.

[7] NIE S P, WANG C, CUI S W, et al. The core carbohydrate structure of Acacia seyal var. seyal (gum Arabic)[J]. Food Hydrocolloids, 2013,32(2)∶ 221-227. DOI∶10.1016/j.foodhyd.2012.12.027.

[8] SI NELLI N, SPINARDI A, EGIDIO V D, et al. Evaluation of quality and nutraceutical content of blueberries (Vaccinium corymbosum L.)by near and mid-infrared spectroscopy[J]. Postharvest Biology and Technology, 2008, 50(1)∶ 31-36. DOI∶10.1016/j.postharvbio.2008.03.013.

[9] LI B, WANG H, ZHAO Q, et al. Rapid detection of authenticity and adulteration of walnut oil by FTIR and fluorescence spectroscopy∶a comparative study[J]. Food Chemistry, 2015, 181∶ 25-30.DOI∶10.1016/j.foodchem.2015.02.079.

[10] W ANG X, SHENG D, ZHU Z, et al. Identification of Cortex Eucommiae from different producing areas by FTIR microspectroscopy[J].Spectrochimica Acta Part A Molecular & Biomolecular Spectroscopy,2015, 141∶ 94-98. DOI∶10.1016/j.lwt.2013.01.027.

[11] A LAMPRESE C, CASALE M, SINELLI N, et al. Detection of minced beef adulteration with turkey meat by UV-vis, NIR and MIR spectroscopy[J]. LWT-Food Science and Technology, 2013, 53(1)∶225-232. DOI∶10.1016/j.lwt.2013.01.027.

[12] P RADO B M, KIM S, ÖZEN B F, et al. Differentiation of carbohydrate gums and mixtures using Fourier transform infrared spectroscopy and chemometrics[J]. Journal of Agricultural and Food Chemistry, 2005, 53(8)∶ 2823-2829. DOI∶10.1021/jf0485537.

[13] KO NG J, YU S. Fourier transform infrared spectroscopic analysis of protein secondary structures[J]. Acta Biochimica et Biophysica Sinica,2007, 39(8)∶ 549-559. DOI∶10.1111/j.1745-7270.2007.00320.x.

[14] BALTACIOGLU H, BAYINDIRLI A, SEVERCAN F. Secondary structure and conformational change of mushroom polyphenol oxidase during thermosonication treatment by using FTIR spectroscopy[J]. Food Chemistry, 2017, 214∶ 507-514. DOI∶10.1016/j.foodchem.2016.07.021.

[15] GA SHUA I B, WILLIAMS P A, BALDWIN T C. Molecular characteristics, association and interfacial properties of gum Arabic harvested from both Acacia senegal and Acacia seyal[J]. Food Hydrocolloids, 2016, 61∶ 514-522. DOI∶10.1016/j.foodhyd.2016.06.005.

[16] OL AWALE A B H, OKEGBILE J, MOHAMMED A, et al. Effect of starch and gum arabic binders in the combustion characteristics of briquette prepared from sawdust[J]. International Journal of Scientifi c &Engineering Research, 2014, 5(3)∶ 1005-1009.

[17] RE NARD D, LAVENANT-GOURGEON L, RALET M C, et al. Acacia senegal∶ continuum of molecular species differing by their protein to sugar ratio, molecular weight, and charges[J].Biomacromolecules, 2006, 7(9)∶ 2637-2649. DOI∶10.1021/bm060145j.

[18] TAN ASKOVIĆ I, GOLOBOCANIN D, MILJEVIĆ N. Multivariate statistical analysis of hydrochemical and radiological data of Serbian spa waters[J]. Journal of Geochemical Exploration, 2012, 112(1)∶ 226-234.DOI∶10.1016/j.gexplo.2011.08.014.

[19] TEYE E, HUANG X, DAI H, et al. Rapid differentiation of Ghana cocoa beans by FT-NIR spectroscopy coupled with multivariate classifi cation[J].Spectrochimica Acta Part A∶ Molecular and Biomolecular Spectroscopy,2013, 114(10)∶ 183-189. DOI∶10.1016/j.saa.2013.05.063.

[20] JIA S Q, YANG L G, AN D, et al. Feasibility of analyzing frostdamaged and non-viable maize kernels based on near infrared spectroscopy and chemometrics[J]. Journal of Cereal Science, 2016,69∶ 145-150. DOI∶10.1016/j.jcs.2016.02.018.

[21] MARQU ES A S, CASTRO J N, COSTA F J, et al. Near-infrared spectroscopy and variable selection techniques to discriminate Pseudomonas aeruginosa strains in clinical samples[J]. Microchemical Journal, 2016, 124∶ 306-310. DOI∶10.1016/j.microc.2015.09.006.

[22] WOLD S, SJOSTROM M, ERIKSSON L. PLS-regression∶ a basic tool of chemometrics[J]. Chemometrics and Intelligent Laboratory Systems, 2001, 58(2)∶ 109-130. DOI∶10.1016/S0169-7439(01)00155-1.

[23] OLESZ KO A, HARTWICH J, WOJTOWICZ A, et al. Comparison of FTIR-ATR and Raman spectroscopy in determination of VLDL triglycerides in blood serum with PLS regression[J]. Spectrochimica Acta Part A Molecular & Biomolecular Spectroscopy, 2017, 183∶239-246. DOI∶10.1016/j.saa.2017.04.020.

[24] ZOU X B, ZHAO J W, LI Y X. Selection of the effi cient wavelength regions in FT-NIR spectroscopy for determination of SSC of ‘Fuji’apple based on BiPLS and FiPLS models[J]. Vibrational Spectroscopy,2007, 44(2)∶ 220-227. DOI∶10.1016/j.vibspec.2006.11.005.

[25] WU D, SUN D W, HE Y. Application of long-wave near infrared hyperspectral imaging for measurement of color distribution in salmon fillet[J]. Innovative Food Science & Emerging Technologies, 2012,16(39)∶ 361-372. DOI∶10.1016/j.ifset.2012.08.003.

[26] ALASS AF S, ANDRESBRULL M, CIRRE J, et al. Structural changes following industrial processing of Acacia gums[M]//KENNEDY J F, PHILLIPS G O, WILLIAMS P A. Gum Arabic.Special publication∶ Royal Society of Chemistry, 2011∶ 153-168.DOI∶10.1039/9781849733106-00153.

[27] BHUSHETTE P R, ANNAPURE U S. Comparative study of Acacia nilotica exudate gum and Acacia gum[J]. International Journal of Biological Macromolecules, 2017, 102∶ 266-271. DOI∶10.1016/j.ijbiomac.2017.03.178.

[28] ADERIBIGBE B A, VARAPRASAD K, SADIKU E R, et al.Kinetic release studies of nitrogen-containing bisphosphonate from gum acacia crosslinked hydrogels[J]. International Journal of Biological Macromolecules, 2015, 73∶ 115-123. DOI∶10.1016/j.ijbiomac.2014.10.064.

[29] DONG Y J, SRENSEN K M, HE S, et al. Gum Arabic authentication and mixture quantification by near infrared spectroscopy[J]. Food Control, 2017, 78∶ 144-149. DOI∶10.1016/j.foodcont.2017.02.002.

[30] CHEN Q S, ZHAO J W, LIN H. Study on discrimination of Roast green tea (Camellia sinensis L.) according to geographical origin by FT-NIR spectroscopy and supervised pattern recognition[J].Spectrochimica Acta Part A∶ Molecular and Biomolecular Spectroscopy, 2009, 72(4)∶ 845-850. DOI∶10.1016/j.saa.2008.12.002.

应用红外光谱结合化学计量学方法检测阿拉伯胶产地和蛋白质含量

邹小波1,Alaa Kamal Mohmmed KHAIR KHOGLY1,石吉勇1,Mel HOLMES2
(1.江苏大学食品与生物工程学院,江苏 镇江 212013;2.利兹大学食品科学与营养学院,英国 利兹 LS1 3AB)

摘 要:研究利用傅里叶红外光谱结合化学计量学方法来实现对苏丹阿拉伯胶的产地和蛋白质含量的快速无损检测的可行性。采集自6 个不同的产地,每个产地12 个,总计72 个阿拉伯胶样本,作为研究对象,运用线性判别分析(linear discriminant analysis,LDA)和反向区间偏最小二乘(backward interval partial least squares,Bi-PLS)法分别实现对苏丹阿拉伯胶的产地区分和蛋白质含量检测。结果表明,当主成分数为6时,LDA对样本的训练集(48 个样本)和预测集(24 个样本)的识别率都为100%。Bi-PLS法回归联合20 个光谱子区间中的4 个子区间得到最佳的蛋白质预测模型,其预测集相关系数为0.937 3,均方根误差为0.173%。因此,利用傅里叶红外光谱结合化学计量学方法可实现对苏丹阿拉伯胶的产地以及蛋白质的含量的快速无损检测。

关键词:阿拉伯胶;傅里叶红外光谱;产地;蛋白质含量;线性判别分析;反向区间偏最小二乘法

中图分类号:TS201.1

文献标志码:A

文章编号:1002-6630(2017)20-0229-06

References:nces:

收稿日期:2017-05-08

基金项目:“十二五”国家科技支撑计划项目(2015BAD17B04);“十三五”国家重点研发计划重点专项(2016YFD0401104);

作者简介:邹小波(1974—),男,教授,博士,研究方向为食品与农产品无损检测。E-mail:zou_xiaobo@ujs.edu.cn

引文格式:

国家自然科学基金面上项目(31671844);国家自然科学基金青年科学基金项目(31601543);

中国博士后科学基金项目(2014T70483;2016M590422);江苏省自然科学基金项目(BK20160506;BE2016306);

江苏省国际科技合作项目(BZ2016013);苏州市科技发展计划项目(SNG201503);

江苏高校优势学科建设工程资助项目(PAPD)

ZOU Xiaobo, KHAIR KHOGLY A K M, SHI Jiyong, et al. Determination of geographical origin and protein content of Acacia gums using infrared spectroscopy and chemometrics[J]. 食品科学, 2017, 38(20)∶ 229-234. DOI∶10.7506/spkx1002-6630-201720033. http∶//www.spkx.net.cn

ZOU Xiaobo, KHAIR KHOGLY A K M, SHI Jiyong, et al. Determination of geographical origin and protein content of Acacia gums using infrared spectroscopy and chemometrics[J]. Food Science, 2017, 38(20)∶ 229-234. DOI∶10.7506/spkx1002-6630-201720033. http∶//www.spkx.net.cn

DOI∶10.7506/spkx1002-6630-201720033