Jing REN, Gang LIU, Xingxiang ZHAO, Shuaiqun ZHAO, Quanhong OU, Juan XU, Jianfei HU

School of Physics and Electronic Information, Yunnan Normal University, Kunming 650500, China

Rhizome crops mainly refer to the terrestrial crops with edible roots or underground stems.The edible parts usually contain large amounts of starch and sugar, and can be used as vegetables, food supplies,feed and materials for producing starch and alcohol.Rhizome crops not only have high nutritional value, but also have extensive medicinal value.They are also potential biomass crops.The rhizome crops studied in this paper included root crop (fragrant taro),tube crops (taro, Jerusalem artichoke,konjak), bulb crops (arrowhead, water chestnut) and root-tube crops (ginger,galangal).

The most-commonly used traditional methods for rhizome crops include biochemical method, chromatography, etc. The glucomannan and total sugar (including starch)contents in konjac are usually determined with colorimetry, mannose hydrazone gravimetric method and high performance liquid (gas)phase chromatography;the inulin is usually extracted by microwave-assisted method and salting out separation method; the ginger essential oil is extracted by steam distillation, solvent extraction, compression,ultrasonic method and supercritical CO2extraction method; the watersoluble polysaccharide in arrowhead is extracted by hot water extraction method and water-extraction and alcohol-precipitation method; the starch in purple yam is extracted by aqueous ammonia, centrifugation method and citric acid method. Although these methods are accurate, they all need experiences, skilled operation skills,long time, as well as large numbers of samples and chemical reagents. Even worse, these methods may pollute the environment[1-5].

Fourier transform infrared (FTIR)spectroscopy can reflect the vibra-tional mode of molecular functional groups. It has fingerprint features and does not require separation and extraction processes of samples. Moreover, FTIR is characterized by simple operation,fastness,low costs and non pollution. It has been successfully applied to quality testing of agricultural products and identification of traditional Chinese medicines. Currently,FTIR has been applied to the studies on konjac. Wu et al.[6]quickly distinguished the konjac flour blended with hydroxypropyl starch by mid-infrared spectroscopy. Li et al.[7]studied the synthesis and biological activity of graft copolymer of konjak glucomannan using FTIR techniques.There have been no reports on application of FTIR in studies on other rhizome crops. In this paper, in order to provide a reference for the identification of different rhizome crops and their nutritional values, 8 kinds of rhizome crops were identified using Fourier transform infrared spectroscopy combined with wavelet transform.

Materials and Methods

Apparatus

Frontier Fourier transform infrared spectrometer(Perkin Elmer),equipped with DTGS detector, was used to record spectrum. The spectral range was 4 000-400 cm-1with a resolution of 4 cm-1and an accumulation of 16 scans.

Sample preparation and pretreatment of spectral data

Konjac is a perennial herb, and it is rich in nutrients, including starch,proteins,a variety of vitamins,potassium, phosphorus, selenium and other minerals, as well as konjac polysaccharides that are needed by human bodies.Konjac is characterized by low calorie,low fat and high fiber[8].

Jerusalem artichoke (Asteraceae:Helianthus) is a perennial herb. Its underground tubes are rich in starch, inulin and other fructose polymers.Jerusalem artichoke is edible and can also be used as raw material for feed,starch and alcohol. Its tubers or stems and leaves have medicinal benefits,such as diuresis dehumidification,clearing heat and cooling blood, and benefiting stomach[9].

Water chestnut (Cyperaceae:Heleocharis) is a perennial herb of shallow water. It has purple-black skin and sweet, juicy, crisp and delicious white pulp. Water chestnut can be used as both fruit and vegetable that is popular among people[10].

Ginger (Zingiberaceae: Zingiber)is a perennial herb. The rhizome, peel and leaves of ginger all can be used as medicines. In traditional Chinese medicine, ginger root has functions of diverging, alleviating vomiting and alleviating cough[11].

Galangal is also a perennial herb.Its roots are tuberous. Galangal is solitary or clustered and pale green,and it has aromatic flavor. The rhizome of galangal can be used as medicine, and has functions of warming dampness, promoting the circulation of qi and relieving pain. Galangal can be used to treat acute gastroenteritis, indigestion, stomach pain,toothache, rheumatic joint pain and traumatic injury[12].

Arrowhead belongs to Alismataceae Sagittaria. Its fruit is edible. Arrowhead has a high nutritional value and is rich in starch,proteins,vitamins,ion, calcium, zinc, phosphorus, boron and other trace elements. It can regulate and promote muscle function, and has good medicinal value[13].

Taro belongs to Araceae Colocasia. Its leaves are peltate and petioles are long and corpus. The globular underground stems (tubers) are edible and can be used as medicine.Taro is nutritious and contains a lot of starch, minerals and vitamins. It is not only vegetable but also food.Taro can be cooked, dried or processed into flour[14].

Fragrant taro (Leguminosae:Dioscorea) is a perennial rampant herb. Its tubes, with diameter of 2-7.5 cm, are spherical, and the shape is same with that of small potatoes. Fragrant taro has abundant nutritional value and medicinal value, and is rich in proteins, fats, carbohydrates, calcium,phosphorus and other nutrients[15].

The used 8 kinds of rhizome crops were all collected from the local market in Kunming.The samples were cleaned, sliced and dried in an oven.Subsequently, an appropriate amount of each sample was mixed with potassium bromide and ground to fine powder and then pressed into a tablet.The sample tablet was placed in the sample compartment and FTIR spectra were recorded. The wavelet analysis,principal component analysis and hierarchical cluster analysis were all performed using MATLAB R2010a.

Results and Analysis

Characteristic analysis of infrared spectra of the 8 kinds of rhizome crops

Fig.1 showed the FTIR spectra of the 8 kinds of rhizome crops. As shown in Fig.1, there were some typical common peaks among the spectra.Among these peaks, the broad intensity peak at 3 400 cm-1was assigned as the stretching vibration of O-H;

the peak at 2 930 cm-1was assigned as the stretching vibration of asymmetric stretching vibration of CH2;the range of 1 500-1 200 cm-1is the mixed vibration absorption region of proteins and polysaccharides; the peaks in the range of 1 440-1 317 cm-1were due to the symmetric bending vibration of methyl and methylene and the scissoring vibration absorption ofCH3; the peak near 1 412 cm-1was due to the stretching vibration of CH2and C-O-O in starch; the range of 1 153-950 cm-1was the mixed vibration absorption region of polysaccharides; the peak near 1150 cm-1was assigned as the stretching vibration of C-O and C-C in starch;the peak near 1 024 cm-1was due to the stretching vibration of C-O and the bending vibration of C-OH in starch; the peak at about 930 cm-1was due to the skeleton vibration of α-1, 4 glycosidic bond in starch; the peak at about 575 cm-1was ascribed to the skeleton vibration of starch[16-17].

The absorption peaks at 3 414,2 927,1 650,1 460,1 420,1 372,1 160,1 061,1 016,987,928,859,764,710,607, 575, 527 and 439 cm-1were the typical absorption peaks of starch[17].Fig.1 showed that the 8 kinds of rhizome crops all contained starch.

Wavelet analysis of Fourier transform infrared spectra

After the derivative processing,the resolution of the Fourier transform infrared spectra will be improved. At the same time,noise will be also introduced. So denoising processing measures should be adopted.Even worse,the derivative order and width are difficult to be directly determined, so continuous optimization and selection are needed. The smoothing of spectra may also lead to the lost of important information. Wavelet transform can solve the above problems.It can flexibly decompose the singles into different-frequency and different-scale parts according to the actual needs. It can also be focused to any part of the signals. So wavelet transform is known as“mathematical microscope”.

For the continuous wavelet transform of Fourier transform infrared spectra, different wavelets and different decomposition scales should be chosen according to the actual situation. The requirements by continuous wavelet transform are to highlight the original characteristic peaks and to choose the wavelets with good smoothness. Due to more concentrated frequency domain energy, narrow pass band,weak effect of aliasing and sporadic domain symmetry and linear phase of Morlet wavelet, the transformed signals will not be distorted[18].The infrared spectra of the 8 kinds of rhizome crops differed significantly within the range of 1 800-700 cm-1.The Morlet wavelet was chosen as the mother wavelet, and continuous wavelet transform with 30 scales as one dimension was performed for the infrared spectra in this range. The results showed that the differences in the 15th-level decomposition coefficients of continuous wavelet transform were most significant(Fig.2).

The wavelets from the Db’s family have good scalability, and they can better solve the border problem brought by improving energy concentration. The 4-order Daubechies (Db4)wavelet was chosen for the 5 times of compression (i.e., the decomposition levels were 5). The 5th-level approximation and detail coefficients were shown in Fig.3. As shown in Fig.3, the approximation coefficients decomposed by discrete wavelet transform were all very close, and the main differences were concentrated in detail coefficients.

Principal component analysis(PCA)

The projection of the principal component analysis can visualize thesample classification situation. The 15th-level decomposition coefficients of CWT were selected for principal component analysis. The variances of the first two principal components were 54.17%and 27.11%,respectively,and the total variance of the first three principal components was 89.78%. The projections of the first and second principal components were shown in Fig.4.

The 5th-level detail coefficients of DWT were chosen for principal component analysis. The variances of the first and second principal components were 45.38%and 26.62%,respectively;and the total variance of the first three principal components was 93.12%. The projections of the first and second principal components were shown in Fig.5.

The results showed that the clustering accuracies of the data processed by CWT and DWT all reached 100%.

Hierarchical cluster analysis(HCA)

Cluster analysis is a classification method in accordance with certain criteria according to characteristics of objects[17]. The 15th-level decomposition coefficients of CWT and the 5thlevel detail coefficients of DWT were chosen for hierarchical cluster analysis.As shown in Fig.6 and Fig.7,cluster analysis could correctly distinguish the 8 kinds of samples. However, the clustering effect of the data processed with DWT was better than that pro-cessed with CWT.

Conclusions and Discussion

The 40 samples of the 8 kinds of rhizome crops were studied by Fourier transform infrared spectroscopy combined with wavelet transform, principal component analysis and hierarchical cluster analysis. The spectral data in the range of 1 800-700 cm-1of the original infrared spectra were processed by wavelet transform.The 15thlevel decomposition coefficients of CWT and the 5th-level detail coefficients of DWT were chosen for principal component analysis and hierarchical cluster analysis. The analysis results showed that the clustering accuracies of principal component analysis and hierarchical cluster analysis all reached 100%; the data processed by DWT was more suitable for principal component analysis and hierarchical cluster analysis compared with those processed by CWT. The final results showed that Fourier transform infrared spectroscopy combined with wavelet transform has characteristics of rapidness,no destruction,easiness and accuracy in identifying rhizome crops,and it is a new potential method for variety identification of rhizome crops.

In this study,Fourier transform infrared spectroscopy showed that the basic ingredients in the 8 kinds of rhizome crops are the same, but their contents are different, so their values in the food processing and industrial production are different. Our study show that Fourier transform infrared spectroscopy combined with wavelet transform can distinguish the 8 kinds of rhizome crops. However, the sample number in this study is limited. So the results can only be treated as a reference. To draw general conclusions, a large number of samples are needed for further study. In addition,the samples of different origins and different varieties taste different. The quantitative analysis of infrared spectra can also determine the proportions of basic components in samples,which still needs further study.

[1]HU X(胡曉),W JP(吳金平),DIAO Y(刁英), et al. Determination of dasheen mosaic virus in Amorphophallus sinesis using RT-PCR (魔芋中芋花葉病毒的RT-PCR 檢測)[J]. Hubei Agricultural Sciences (湖北農(nóng)業(yè)科學(xué)),2010,47(1):9-11.

[2]LIU X(劉旭),ZHOU ZG(周忠光),AN BS(安柏松),et al.Research progresses in pharmacology of water chestnut(荸薺藥理學(xué)研究進展)[J]. Information on Traditional Chinese Medicine(中醫(yī)藥信息),2010,27(6):106-108.

[3]LI Y(李嬡),QIAO XG(喬旭光). Properties of ginger starch(生姜淀粉的基本性質(zhì))[J].Food Science (食品科學(xué)),2011,32(13):131-135.

[4]XU XQ(徐秀泉),SUN XQ(孫雪琴),QIU JB (邱晶波), et al. Optimization of extraction conditions of essential oil from Rhizoma zingiberis by response surface methodology(響應(yīng)面分析法優(yōu)化生姜揮發(fā)油提取工藝)[J].Applied Chemical Industry (應(yīng)用化工),2011,40 (6):1035-1044.

[5]QIU YX(丘苑新),WU XJ(吳雪君),YU X(于新), et al. Research on extraction and partial properties of taro starch (香芋淀粉提取工藝及部分特性研究)[J].Journal of Changjiang Vegetables(長江蔬菜),2010,4(009):32-34.

[6]WU YJ (吳月蛟), DONG G (董剛),ZHANG SL(張盛林),et al.Rapid identification of konjac flour with adulterated hydroxypropyl starch based on mid-infrared spectroscopy(中紅外光譜法快速鑒別摻雜羥丙基淀粉的魔芋粉)[J].Food Science(食品科學(xué)),2013,34(24):140-143.

[7]LI HL(李慧靈),LEI WX(雷萬學(xué)),GAO X(高霞),et al.Synthesis and biologic activity of graft copolymer of konjac glucomanman and methacryloxyethyl-octyldimethyl ammonium bromide(魔芋葡甘聚糖季銨鹽接枝共聚物的合成及生物學(xué)活性)[J].Journal of Henan University of Technology (河南工業(yè)大學(xué)學(xué)報),2010,31(1):44-48.

[8]WANG LX,WEN CR,WU J,et al.Studies on the molecular chain conformation stability of aminated konjac glucomannan[J]. Chinese Journal of Structural Chemistry,2013,32(12):1845-1853.

[9]IRENE AR,ETHEL EP,DIEGO B, et al.In vitro prebiotic activity of inulin-rich carbohydrates extracted from Jerusalem artichoke(Helianthus tuberosus L.)tubers at different storage times by Lactobacillus paracasei [J]. Food Research International, 2014, 3 (62): 59-65.

[10]SUN Q(孫奇),LIAO Y(廖艷),LIN Y(林殷), et al. Textual research on Wuyu,arrowhead and water chestnut (烏芋、慈姑、 荸薺考)[J]. World Chinese Medicine(世界中醫(yī)藥),2013,8(1):81-84.

[11]LIU CY(劉春葉),ZHANG J(張劍),YOU J(尤靜),et al.Detection of curcumin in ginger by HPLC-MS (生姜中姜黃素的高效液相色譜- 質(zhì)譜法檢測)[J].Lishizhen Med Mater Med Res(時珍國醫(yī)國藥),2011,22(7):1637-1639.

[12]LING YZ(凌育趙),LIU JL(劉經(jīng)亮).Microwave assisted hydro-distillation extraction of essential oil from of Kaempferia galangal L.(微波輔助水蒸汽蒸餾提取沙姜揮發(fā)油的研究)[J].China Condiment (中國調(diào)味品),2010,35(6):57-65.

[13]LUO YH(羅永會),ZHANG CX(張翠香),XU CP ( 徐春萍), et al. Study on polysaccharides extraction from sagittaria(慈姑多糖的提取研究)[J].Tropical Agricultural Engineering (熱帶農(nóng)業(yè)工程),2012,36(3):5-8.

[14]CHANG YZ(常銀子), WANG LX(王麗霞), ZHONG SM (仲山民), et al. Extraction protein from taro by enzymatic method (酶法提取芋艿蛋白質(zhì)工藝)[J].Food Research and Development (食品研究與開發(fā)),2011,32(3):19-22.

[15]CHEN ZY (陳致印),HU YH (胡一鴻),JIN CZ(金晨鐘),et al.Extracting technologies for soluble dietary fibers from ube and its properties(香芋可溶性膳食纖維提取工藝及其性能研究)[J]. Hunan Agricultural Sciences(湖南農(nóng)業(yè)科學(xué)),2012,09(03):89-91.

[16]WENG SF (翁詩甫).Fourier transform infrared spectroscopy(傅里葉變換紅外光譜分析)[M]. Beijing: Chemical Industry Press (北京: 化學(xué)工業(yè)出版社),2010.

[17]SUN SQ (孫素琴), ZHOU Q (周群),CHEN JB (陳建波). Infrared spectroscopic analysis and identification of Chinese medicine (中藥紅外光譜分析與鑒定)[M].Beijing:Chemical Industry Press(北京: 化學(xué)工業(yè)出版社),2010.

[18]SHI YG(史永剛),SU B(粟斌),TIAN GY(田高友),et al.Chemometric methods and MATLAB realization (化學(xué)計量學(xué)方法及MATLAB 實現(xiàn))[M]. Beijing:China Petrochemical Press(北京: 中國石化出版社),2010.