Establishment of Element Fingerprints and Application to Geographical Origin Identification of Chinese Fenghuangdancong Tea by ICP-MS

Abstract

The purpose of this study is to examine the feasibility of element fingerprints in identifying the geographical origin of Fenghuangdancong and other Chinese tea. A total of 45 elements were analyzed by inductively coupled plasma mass spectrometry (ICP-MS). Element fingerprints were established by transferring element composition into graphs. Similarity degrees of element fingerprints and linear discriminant analysis were performed in order to identify tea samples according to their origin. The analysis revealed a satisfactory separation of tea samples according to their origin with 93.3 % classifications rate of similarity degrees and 96.6 % predictive ability of the cross-validation of linear discriminant analysis. There is small difference between each other. The analysis indicated element fingerprints could be used to identify the authenticity of Fenghuangdancong tea.

Introduction

With the rapid development of social economy and the improvement of people's living standard, food quality and security issues have aroused big concern. Food with high quality has become the indispensable choice for most consumers. Since tea is rich in vitamins, mineral elements, tea polyphenols and other substances that are beneficial to human's health (Peluso and Serafini, 2017; Salahinejad and Aflaki, 2010), it has gained great popularity in recent years. Fenghuangdancong (FHDC) is a Chinese traditional tea grown in Phoenix Mountain in Guangdong province, and its history can date back to Tang Dynasty. The unique geographical location and climate conditions make FHDC brim with special tea aroma (Dai et al., 1998). As a geographical indication product with high quality, the tea produced in Phoenix Mountain is more expensive than those produced in ordinary regions, which causes the market filled with many kinds of fake tea named FHDC. Some manufacturers often process the tea from ordinary regions or even the tea of other varieties into FHDC to get a higher price. Besides, FHDC is similar to some kinds of black tea in appearance and aroma, which makes some avaricious sellers fraudulently label their black tea as FHDC. To protect the interests of consumers and crack down on the illegal acts of those who produce fake tea, the methods to identify the geographical origin of tea should be established immediately.

Compared with organic components, the elements in agriculture products are affected less by store time and processing technology (Brzezichacirocka et al., 2016). Many studies have used inorganic elements to trace the origin of agricultural products, such as honey (Uršulintrstenjak et al., 2015), milk (Osorio et al., 2015), grape wine (Geana et al., 2013; Coetzee et al., 2014), rice (Li et al., 2016), etc. The accumulation of inorganic elements in tea is closely related with their concentrations in soils (Zhao et al., 2017), so it is effective to identify the geographical origin of tea through inorganic elements. At present, the origin trace of tea is mostly based on element concentrations, which are combined with pattern recognition methods, such as variance analysis, cluster analysis, linear discriminant analysis and principal component analysis (Ma et al., 2016; Fernández-Cáceres et al., 2001; Moreda-Piñeiro et al., 2003; Hu and Yin 2017). Because of the difference among methods, the effective indexes of judgmental elements are usually diverse, and there isn't a consistent standard of judgment. Furthermore, the pattern recognition methods usually involve a few elements, which are far less than the kinds of elements contained in tea leaves, so they can't reflect the integral element information well. Fingerprints have the attributes of specificity and stability, which emphasize integrity and fuzziness (Zeng et al., 2004; Cao et al., 2004), and they have been widely used to control the quality or identify the authenticity of traditional Chinese herbal medicine (Liu et al., 2017; Guo et al., 2017). However, fingerprints are limited to organic component research and the application of element fingerprints is rarely reported.

In the present study, standard element fingerprint of FHDC was established. Then similarity degrees and linear discriminant analysis were performed to distinguish the tea samples in terms of their geographical origin.

Materials and Methods

Samples and reagents    A total of 59 tea samples harvested from different regions in the spring of 2016 are shown in Table 1. FHDC tea samples were provided by Tea Research Institute, Guangdong Academy of Agricultural Sciences. Others were purchased from farm producers in each region. In this study, 29 FHDC tea samples were used to establish the standard element fingerprint and the other 30 tea samples were used to verify the accuracy of origin identification by element fingerprints.

Table 1. The information of tea samples studied in present work

Geographical location	Code	Number of samples	Variety	Taxon
Phoenix Mountain, Guangdong province	C	39	Fenghuangdancong	Oolong tea
Wuyi Mountain, Fujian province	W	6	Wuyi rock tea	Oolong tea
Wuyi Mountain, Fujian province	W	4	Zhengshanxiaozhong	Black tea
Hangzhou, Zhejiang province	H	10	Jiuquhongmei	Black tea

Ultra-pure water (18.2 MΩ·cm) was obtained from a Milli-Qsystem (Millipore, USA); 68 % nitric acid of ultrapure grade was purchased from Suzhou Crystal Clear Chemical Co., Ltd (Suzhou, China); multi-element standard solution (100 µg/mL) of Ca, K and Mg was obtained from National Center for Analysis and Testing of Nonferrous Metals and Electronic Materials (Beijing, China); multi-element standard solution (100 µg/mL) of Al, As, B, Ba, Be, Bi, Cd, Co, Cr, Cu, Fe, Ga, Li, Mn, Ni, Pb, Sb, Sn, Sr, Ti, Tl, V and Zn was purchased from National Center for Analysis and Testing of Nonferrous Metals and Electronic Materials (Beijing, China); multi-element standard solution (100 µg/mL) of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Y was obtained from National Center for Analysis and Testing of Nonferrous Metals and Electronic Materials (Beijing, China); single-element standard solution (1000 µg/mL) of Ge was obtained from National Center for Analysis and Testing of Nonferrous Metals and Electronic Materials (Beijing, China); single-element standard solutions (1000 µg/mL) of Sc, Hg, Se, P, Rh, In, Re were obtained from National Iron and Steel Testing Center (Beijing, China), respectively; the certified reference material (CRM) of tea (GBW10083) was purchased from National Institute of Metrology (Beijing, China). Rh, In, Re and Ge were used as internal standard elements.

Instruments    Agilent 7700X ICP-MS (Agilent Technologies, USA) was used to determine 45 elements in tea. WX-8000 microwave digestion instrument (PreeKem Scientific Instruments Co., Ltd. Shanghai, China) was used to digest tea samples. HT-200 experiment electric heating board was used to heat the solution to remove nitric acid (Guangdong Institute of Analysis, Guangzhou, China).

Analysis by ICP-MS    An aliquot of 0.3 g tea sample grinded was accurately weighed into Teflon digestion vessel, and dissolved in 6 mL nitric acid. Then it was digested by microwave digestion instrument according to the digestion procedures shown in Table 2. Then the solution was heated for 20 min to remove nitric acid and transferred to colorimetric tube. It was then made up to 25 mL with ultra-pure water. After that, the solution in colorimetric tube was diluted 20-fold. The previous solution and diluted solution were stored for ICP-MS.

Table 2. Procedures of microwave digestion

Procedures	Temperature/°C	Holding time/min	Pressure/MPa
1	120	2	3.5
2	150	3	3.5
3	180	25	3.5

Data analysis    The statistical methods include independent t-test, one-way analysis of variance, linear discriminant analysis and pearson correlation analysis, all of which were carried out by SPSS for Windows (release 20.0., SPSS, 2011). Independent t-test was used to assess whether there was a significant difference (P < 0.05) between oolong tea and black tea from the same region or not. One-way analysis of variance was used to select elements of significant difference (P < 0.05) in tea from different regions. Linear discriminant analysis was applied to establish tea groups. Pearson correlation analysis was carried out to compute similarity degrees of element fingerprints. The principles of similarity degrees calculation are as follows:   

  

Above, (X₁, X₂, X₃, X₄, … X_N) is a N dimensional vector, which is represented by X; |X| and |Y| are the modules of vectors; besides, X·Y is the inner product of X and Y; cosθ represents the cosine between the X and Y, the closer cosθ to 1 is, the more similar the two vectors are.

Results and Discussion

Validation of quality control method    The analytical results of the certified reference material of tea (GBW10083) are summarized in Table S1 in Supplementary Material. It can be seen that the element concentrations are in agreement with certified concentrations in the CRM. Due to the limited varieties of elements in the certified reference material, a tea sample was analyzed six times to determine repeatability of the method. Table S2 shows the relative standard deviations (RSDs %) for the studied elements are less than 9 % and the limit of detections (LODs) for all the elements range from 0.00850 µg·kg⁻¹ (Ho: the lowest) to 0.0335 g·kg⁻¹ (K: the highest). From Table S3, the recovery of elements is between 88.6 % and 110 %. The results offer confidence to determine other tea samples.

Table S1. Element concentrations for CRM

Element	Certified value	Measured value	RSD/%
Ce	0.457±0.022	0.436	6.54
P	3.02±0.07	2.95	2.84
K	17.3±0.4	17.1	2.71
Mg	2.03±0.05	2.01	2.93
Ca	3.53±0.08	3.48	2.9
Al	991±22	1006	3.33
Mn	552±12	560	1.72
Fe	126±4	127	4.00
Zn	19.0±0.4	19.3	3.14
Cu	5.69±0.13	5.69	1.95
Ba	3.93±0.12	3.87	3.43
Ni	3.61±0.14	3.54	3.93
Pb	1.43±0.06	1.41	8.54
Cd	0.026±0.003	0.023	4.50

For GBW10083, the units of P, K, Mg and Ca were mg·g⁻¹; those of the other elements were µg·g⁻¹, respectively.

Table S2. The RSDs and LODs for this method

Element	1	2	3	4	5	6	Mean	RSD%	LOD
Li	35.3	37.6	38.2	38.3	31.1	32.8	35.6	8.52	14.0
Be	3.49	3.10	3.75	3.77	3.19	3.25	3.43	8.5	9.33
B	9.11	8.39	8.82	8.19	8.37	8.45	8.55	3.97	0.0560
Mg	1.92	1.96	1.91	1.95	1.96	1.96	1.94	1.14	0.00154
Al	430	425	416	424	429	429	425	1.19	0.763
P	2.99	2.95	2.97	2.96	2.92	2.94	2.96	0.87	0.00630
K	19.6	19.4	19.0	19.1	19.4	19.0	19.2	1.21	0.0335
Ca	2.34	2.38	2.38	2.42	2.36	2.42	2.39	1.31	0.0059
Sc	50.0	49.4	49.2	50.0	50.4	49.4	49.7	0.88	0.925
Ti	2.16	2.24	2.37	2.41	2.16	2.59	2.32	7.22	0.00690
V	82.3	76.9	81.7	81.5	79.6	78.8	80.1	2.59	0.708
Cr	267	254	256	266	262	251	259	2.53	1.99
Mn	804	785	809	816	798	799	802	1.32	0.0755
Fe	104	104	96.7	97.8	94.0	95.6	98.8	4.50	0.978
Co	224	221	220	233	222	221	224	2.21	2.12
Ni	2.03	2.03	2.03	2.05	2.09	2.01	2.04	1.34	0.000942
Cu	11.0	11.0	10.9	10.9	10.9	11.0	10.9	0.19	0.00314
Zn	27.7	27.2	25.7	26.5	26.6	26.5	26.7	2.54	0.0187
Ga	824	815	847	834	845	811	829	1.79	0.850
As	11.8	10.1	10.6	10.7	10.6	12.3	11.0	7.63	1.10
Se	113	116	122	109	120	120	117	4.19	10.8
Sr	13.4	13.1	13.1	13.4	13.8	13.6	13.4	1.90	0.000725
Y	59.3	56.5	59.7	56.2	58.0	57.1	57.8	2.50	0.117
Cd	28.3	27.4	28.7	28.5	28.0	28.8	28.3	1.87	0.183
Sn	35.7	35.7	37.7	38.9	32.9	37.4	36.4	5.75	1.86
Sb	1.75	1.72	1.70	1.63	1.99	1.86	1.78	7.28	0.576
Ba	2.95	2.87	2.87	2.89	2.97	2.95	2.92	1.54	0.00613
La	62.6	66.2	64.1	63.2	61.5	60.5	63.0	3.17	0.435
Ce	489	489	489	495	492	499	492	0.78	0.0342
Pr	15.6	14.9	15.0	15.2	14.6	14.5	15.0	2.82	0.0841
Nd	59.4	62.6	63.8	59.2	58.1	60.2	60.6	3.62	0.371
Sm	12.1	12.3	12.6	12.8	12.4	12.6	12.5	1.97	0.152
Eu	2.53	2.32	2.58	2.47	2.59	2.52	2.5	3.96	0.127
Gd	15.6	15.6	14.1	14.4	15.2	15.1	15.0	4.15	0.152
Tb	1.75	1.74	1.71	1.76	1.73	1.73	1.74	1.01	0.0167
Dy	9.83	9.64	9.8	9.48	9.39	9.43	9.59	1.99	0.0427
Ho	3.79	3.69	3.59	3.70	4.31	3.72	3.80	6.77	0.00850
Er	7.28	7.25	7.73	7.33	7.59	7.22	7.40	2.85	0.0502
Tm	1.38	1.34	1.43	1.42	1.38	1.35	1.38	2.64	0.0510
Yb	10.5	10.0	11.7	10.6	10.9	10.7	10.7	5.26	0.142
Lu	1.72	1.80	1.77	1.91	1.91	1.81	1.82	4.12	0.0686
Hg	6.49	6.52	6.07	6.47	6.16	6.88	6.43	4.53	2.62
Tl	45.2	45.8	45.8	45.4	47.7	47.1	46.2	2.14	0.169
Pb	162	148	134	148	159	144	149	6.88	0.933
Bi	-	-	-	-	-	-	-	-	3.00

The units of Mg, P, K and Ca were g·kg⁻¹; the units of B, Al, Ti, Mn, Fe, Ni, Cu, Zn, Sr and Ba were mg·kg⁻¹; and those of other elements were µg·kg⁻¹, respectively.

Table S3. Recoveries percentage for this method

Element	Added determined value	Measure value	Concentration added	Recovery/%
Li	541	37.0	500	101
Be	503	3.45	500	99.9
B	14.1	8.77	5.00	107
Sc	153	49.6	100	103
Ti	7.56	2.26	5.00	106
V	627	80.3	500	109
Cr	798	259	500	108
Co	756	222	500	107
Ga	1.92	0.829	1.00	109
As	491	10.8	500	96.0
Se	308	117	200	95.5
Sr	18.7	13.2	5.00	110
Y	154	58.5	100	95.5
Cd	493	28.1	500	93.0
Sn	531	36.4	500	98.9
Sb	482	1.73	500	96.1
La	152	64.3	100	88.6
Ce	581	489	100	92.0
Pr	114	15.2	100	98.8
Nd	164	61.9	100	102
Sm	113	12.4	100	101
Eu	99.9	2.48	100	97.4
Gd	113	15.1	100	97.9
Tb	101	1.73	100	99.3
Dy	108	9.76	100	98.2
Ho	103	3.69	100	99.3
Er	108	7.42	100	101
Tm	102	1.38	100	101
Yb	109	10.7	100	98.3
Lu	104	1.76	100	102
Hg	98.6	6.36	100	92.2
Tl	553	45.6	500	101
Pb	629	143	500	97.2
Bi	472	-	500	94.4

The units of B, Ti, Ga and Sr were mg·kg⁻¹; and those of other elements were µg·kg⁻¹, respectively.

Element concentrations in tea and the establishment of element fingerprints of FHDC    The concentrations of 45 elements in tea were determined by using ICP-MS. As is shown in Table 3, the level of most elements in tea is significantly different. Elements in tea can be divided into four categories in terms of their mean element concentrations. The first category is K, whose concentration is greater than 10000 mg·kg⁻¹. The second includes Mg, Al, P, Ca, Mn and Fe, the concentrations of them range from 50 to 10000 mg·kg⁻¹. The third is 7–50 mg·kg⁻¹, such as B, Cu, Zn, Sr and Ba. And the last are those elements, whose concentrations are less than 7 mg·kg⁻¹, including Li, Be, Cr, Co, Ni, Ti, V, Ga, As, Se, Cd, Sn, Sb, Hg, Tl, Pb, Bi and 16 kinds of rare earth elements (Sc, Y, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu).

Table 3. The mean element concentrations and standard deviations of tea

Element	Fenghuangdancong	Wuyi rock tea	Zhengshanxiaozhong	Jiuquhongmei
Li	38.7±16.3	87.3±40.3	51.9±4.25	145±96.5
Be	6.01±2.67	16.0±12.7	7.81±11.8	5.28±7.35
B	12.2±1.90	12.6±5.25	14.2±4.72	15.7±3.62
Mg	2.70±0.895	1.93±0.374	1.95±0.534	1.77±0.152
Al	534±189	647±239	616±251	432±104
P	2.80±0.408	2.57±0.523	3.01±0.347	3.67±0.441
K	29.6±9.68	17.7±2.86	18.6±0.723	18.4±1.26
Ca	3.63±1.39	3.58±1.02	3.27±0.745	2.85±0.413
Sc	31.4±13.6	37.3±30.3	39.5±6.69	33.0±10.1
Ti	2.58±0.759	2.59±2.00	2.73±1.25	4.04±1.71
V	87.8±29.5	110±97.1	91.0±72.8	149±70.4
Cr	230±164	581±445	305±106	773±432
Mn	655±351	524±287	534±76.9	878±247
Fe	191±345	114±89.9	77.2±10.7	182±87.2
Co	0.260±0.262	1.05±1.87	0.741±0.323	0.390±0.214
Ni	2.02±0.421	3.95±2.61	2.92±0.843	6.07±0.824
Cu	13.2±3.01	9.20±2.11	11.5±1.91	26.2±16.8
Zn	18.9±3.21	29.1±7.76	33.8±3.45	37.9±13.2
Ga	2.08±1.07	4.27±1.29	6.57±0.84	2.31±0.955
As	31.0±11.1	48.4±15.2	52.5±7.70	89.3±71.9
Se	96.4±39.3	58.4±36.4	71.2±27.2	81.8±20.7
Sr	15.2±8.59	10.7±2.27	13.8±3.17	9.45±4.75
Y	135±83.7	160±105	149±24.8	123±60.3
Cd	22.9±8.28	39.6±20.2	36.4±5.55	37.6±14.2
Sn	143±125	20.4±23.9	24.6±13.9	240±154
Sb	54.5±139	26.4±11.1	17.1±6.24	36.6±36.2
Ba	7.17±3.79	20.2±6.33	27.7±3.99	17.3±7.03
La	137±96.1	263±161	275±50.7	181±91.3
Ce	408±182	421±212	380±72.9	233±79.8
Pr	29.9±21.1	40.8±24.9	36.1±6.94	29.7±14.5
Nd	115±82.7	139±92.1	135±26.7	110±53.8
Sm	21.4±14.7	27.6±18.5	22.2±2.25	19.7±8.25
Eu	4.50±3.01	11.1±3.75	12.9±1.19	6.67±3.99
Gd	23.0±12.7	26.3±17.7	20.4±3.17	21.8±10.1
Tb	3.03±1.90	3.42±2.75	3.32±0.504	2.12±1.35
Dy	18.8±11.2	20.2±5.62	22.9±2.60	15.5±7.58
Ho	3.86±2.45	4.20±3.04	4.55±2.12	2.41±1.60
Er	12.9±8.17	12.6±9.85	14.3±2.03	8.81±4.95
Tm	2.06±1.37	2.08±1.69	1.96±2.54	1.76±0.796
Yb	15.7±10.2	12.3±3.10	13.3±2.05	8.16±5.00
Lu	2.51±1.73	2.92±1.82	1.94±1.33	1.77±0.887
Hg	6.96±9.62	1.46±1.21	2.12±1.24	4.47±4.56
Tl	38.9±27.2	35.7±4.34	38.2±2.62	12.3±4.66
Pb	312±185	443±173	560±76.5	751±503
Bi	-	63.7±40.2	56.7±28.5	-

The units of Mg, P, K and Ca were g·kg⁻¹; the units of B, Al, Mn, Fe, Ni, Cu, Zn, Ga, Ba, Sr, Ti and Co were mg·kg⁻¹; and those of other elements were µg·kg⁻¹, respectively.

In this study, the content distribution curves of elements were defined as the element fingerprints. And 45 elements were used to establish the element fingerprints of tea in the order of their atomic number. Elements were then narrowed in term of their categories divided above to take both macro-element and micro-element into account. K was narrowed 10000-fold. Mg, Al, P, Ca, Mn and Fe were narrowed 1000-fold. B, Cu, Zn, Sr and Ba were narrowed 10-fold. Finally, the relationship curves between element types and element concentrations were drawn with smooth lines, such as Fig. 1. In the plot, the concentration of each element always fluctuates within a certain range so that all FHDC tea samples have a similar peak shape. It indicates elements in tea from the same region are similar.

Fig. 1.

The element fingerprints of 29 FHDC

The establishment of standard element fingerprint of FHDC    To eliminate the influence of extreme samples on the determination results, the standard element concentrations of FHDC were represented by the mean element concentrations of 29 FHDC tea samples in Table 3. Fig. 2 is the standard element fingerprint of FHDC, which is established in accordance with its mean element concentrations. And to measure the stability of standard element fingerprint, twenty-nine tea samples were regarded as spatial vectors, which were composed of element types and element concentrations. The similarity degrees between element fingerprints and standard element fingerprint were calculated by Eq.1. In this case, element concentrations were narrowed according to their categories divided above to take the differences of macro-element and micro-element into account. From Table 4, the similarity degrees between element fingerprints of the twenty-nine tea samples and standard element fingerprint of FHDC range from 0.9234 to 0.9906.

Fig. 2.

The standard element fingerprint of FHDC

Table 4. The similarity degrees between element fingerprints and the standard element fingerprint of FHDC

No.	Similarity degree	No.	Similarity degree	No.	Similarity degree
C1	0.9234	C11	0.9564	C21	0.9596
C2	0.9649	C12	0.9451	C22	0.9458
C3	0.9615	C13	0.9665	C23	0.9452
C4	0.9698	C14	0.9697	C24	0.9710
C5	0.9679	C15	0.9484	C25	0.9793
C6	0.9244	C16	0.9906	C26	0.9808
C7	0.9585	C17	0.9876	C27	0.9640
C8	0.9551	C18	0.9777	C28	0.9637
C9	0.9890	C19	0.9843	C29	0.9548
C10	0.9889	C20	0.9851	Mean	1.000

The minimum similarity degree between the element fingerprint and standard element fingerprint of FHDC was defined as similarity threshold, so the similarity threshold of FHDC was determined as 0.92. To identify whether the origin of tea is Phoenix Mountain or not, the similarity degree between the element fingerprint of unknown origin tea and the standard element fingerprint of FHDC is calculated. If the similarity degree is less than 0.92, the origin of unknown tea won't be regarded as Phoenix Mountain, otherwise, the origin of unknown tea will be Phoenix Mountain.

Verification of origin identification by element fingerprints    To verify the feasibility of standard element fingerprint for tea origin identification, 30 tea samples were randomly selected from Phoenix Mountain, Wuyi Mountain and Hangzhou as unknown origin tea samples, none of which was used for standard element fingerprint. Then the similarity degrees between element fingerprints of these tea samples and the standard element fingerprint of FHDC were calculated. From Table 5, it can be clearly seen that the similarity degrees between the element fingerprints of the first eight FHDC tea samples and the standard element fingerprint of FHDC are all more than 0.92, while the similarity degrees of the other two FHDC tea samples are less than 0.92. The reason might be that the tea in a certain region was not included when standard element fingerprint was established. In addition, there are low similarity degrees between the element fingerprints of these tea from Hangzhou and Wuyi Mountain and the standard element fingerprint of FHDC, indicating that elements in tea from Hangzhou and Wuyi Mountain are different from Phoenix Mountain. Similarity degrees verification results illustrate that the standard element fingerprint of FHDC can distinguish tea from Phoenix Mountain and other regions with 93.3 % classifications rate.

Table 5. Similarity degree verification results of the tea samples

No.	Similarity degree	Results	Actual place of origin	Taxon	Accuracy (%)
1	0.9334	C	C	Oolong tea	93.3
2	0.9852	C	C	Oolong tea
3	0.9655	C	C	Oolong tea
4	0.9457	C	C	Oolong tea
5	0.9651	C	C	Oolong tea
6	0.9704	C	C	Oolong tea
7	0.9467	C	C	Oolong tea
8	0.9729	C	C	Oolong tea
9	0.9033	Not C	C	Oolong tea
10	0.8902	Not C	C	Oolong tea
11	0.8757	Not C	H	Black tea
12	0.8752	Not C	H	Black tea
13	0.8186	Not C	H	Black tea
14	0.8676	Not C	H	Black tea
15	0.8717	Not C	H	Black tea
16	0.8753	Not C	H	Black tea
17	0.8574	Not C	H	Black tea
18	0.8598	Not C	H	Black tea
19	0.8557	Not C	H	Black tea
20	0.8154	Not C	H	Black tea
21	0.8616	Not C	W	Oolong tea
22	0.8099	Not C	W	Oolong tea
23	0.8136	Not C	W	Oolong tea
24	0.8119	Not C	W	Oolong tea
25	0.8863	Not C	W	Oolong tea
26	0.8785	Not C	W	Oolong tea
27	0.9045	Not C	W	Black tea
28	0.8153	Not C	W	Black tea
29	0.8622	Not C	W	Black tea
30	0.8321	Not C	W	Black tea

Classification by linear discriminant analysis    Independent t-test was carried out to assess the difference between oolong tea and black tea from Wuyi Mountain. And results showed that only Ga was significantly different (p < 0.05) while other elements had no significant difference (p > 0.05), which indicated there was small difference between oolong tea and black tea from Wuyi Mountain. In the classification of agricultural products according to their geographical origin, the most common method is the combination of element concentrations and statistical analysis. The tea from Phoenix Mountain, Wuyi Mountain and Hangzhou was processed by one-way analysis of variance. Results showed there were no significant difference (p > 0.05) in the concentrations of B, Al, Ca, Sc, Mn, Fe, Co, Sr, Y, Sb, Pr, Nd, Sm, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu and Hg, while other 23 elements in tea samples (Li, Be, Mg, P, K, Ti, V, Cr, Ni, Cu, Zn, Ga, As, Se, Cd, Sn, Ba, La, Ce, Eu, Tl, Pb and Bi) were obviously different among these three regions. In order to evaluate the difference of tea from these three regions, linear discriminant analysis was performed on the basis of the concentrations of 23 elements with significant difference (p < 0.05). As a result, six elements (Li, Be, Ni, Ga, Sn and Ba) were selected as effective discriminant indicators, and the units of these elements concentrations were all µg·kg⁻¹. Two canonical discriminant functions were constructed, which were shown as follows:   

  

The separation of tea from Phoenix Mountain, Wuyi Mountain and Hangzhou was checked by plotting the two canonical discriminant functions' scores. According to Fig. 3, tea from different regions is separated. The classification results of linear discriminant analysis based on elements above are summarized in Table 6. 100 % of the original groups are correctly classified. To evaluate the predictive capacity, the model was then validated by the cross-validation method. The predictive ability of the cross-validation of the model was 96.6 %, in which one sample from Hangzhou was incorrectly assigned as Wuyi Mountain and one sample from Wuyi Mountain was misjudged as Phoenix Mountain. Fisher's linear discrimination functions for Phoenix Mountain, Wuyi Mountain and Hangzhou were as follows:   

  

  

Fig. 3.

Scatter plot of tea from different regions

Table 6. Classification of tea in different regions by linear discriminant analysis

			Predicted group membership
			Phoenix Mountain	Hangzhou	Wuyi Mountain	Total
Original	Count	Phoenix Mountain	39	0	0	39
		Hangzhou	0	10	0	10
		Wuyi Mountain	0	0	10	10
	%		100	100	100	100
Cross-validated	Count	Phoenix Mountain	39	0	0	39
		Hangzhou	0	9	1	10
		Wuyi Mountain	1	0	9	10
	%		100	90	90	96.6

Conclusions

In this study, a simple and effective approach was developed to identify the geographical origin of tea on the basis of multielement analysis by ICP-MS. The classifications rate of tea from Wuyi Mountain, Hangzhou and Phoenix Mountain was 93.3 % with similarity degrees of element fingerprints. Besides, linear discriminant analysis showed the predictive ability of the cross-validation of the model reached 96.6 %, which had small difference with the verification results of similarity degrees. However, linear discriminant analysis requires a large number of tea samples from at least two known regions to construct discriminant functions, while standard element fingerprint can be used as a standard of elements in tea from one region, so it is more convenient than linear discriminant analysis. The results showed that element fingerprint was a reliable and promising tool to identify the authenticity of tea. In the present study, the number of tea samples from other regions is not sufficient to assess the influence of natural variability on tea samples. A large number of tea samples will be collected for further investigation. In addition, the influence of cultivar and annual variation also need further exploration.

Acknowledgments    This work was supported by the Scientific and Technology Plan Project of Guangdong province (NO: 2015A030401065). We are grateful for Tea Research Institute, Guangdong Academy of Agricultural Sciences for the supply of FHDC tea samples in this study.

References

•  Brzezichacirocka,  J.,  Grembecka,  M., and  Szefer,  P. (2016). Monitoring of essential and heavy metals in green tea from different geographical origins. Environ. Monit. Assess., 188, 183.
•  Cao,  J.,  Xu,  Y.,  Zhang,  Y. Z.,  Wang,  Y. M., and  Luo,  G. A. (2004). Integrated Fingerprint Chromatogram Analysis for Gardenia. Chinese J. Anal. Chem., 32, 875-878.
•  Coetzee,  P. P.,  van Jaarsveld,  F. P., and  Vanhaecke,  F. (2014). Intraregional classification of wine via ICP-MS elemental fingerprinting. Food Chem., 164, 485-492.
•  Dai,  S. X.,  Xie,  C. J.,  Li,  Q. N.,  Chen,  D., and  Zheng,  R. Q. (1998). Analysis on Aromatic Constituents of Five Well-known Strains from Fenghuang Dancong Tea Cultivar. J. Tea. Sci., 18, 39-46.
•  Fernández-Cáceres,  P. L.,  Martín,  M. J.,  Pablos,  F., and  González,  A. G. (2001). Differentiation of tea (Camellia sinensis) varieties and their geographical origin according to their metal content. J. Agric. Food Chem., 49, 4775-4779.
•  Geana,  I.,  Iordache,  A.,  Ionete,  R.,  Marinescu,  A.,  Ranca,  A., and  Culea,  M. (2013). Geographical origin identification of Romanian wines by ICP-MS elemental analysis. Food Chem., 138, 1125-1134.
•  Guo,  Y. J.,  Sun,  L. Q.,  Yu,  B. Y., and  Qi,  J. (2017). An integrated antioxidant activity fingerprint for commercial teas based on their capacities to scavenge reactive oxygen species. Food Chem., 237, 645-653.
•  Hu,  L. Q. and  Yin,  C. L. (2017). Development of a New Three-Dimensional Fluorescence Spectroscopy Method Coupling with Multilinear Pattern Recognition to Discriminate the Variety and Grade of Green Tea. Food Anal. Method., 10, 2281-2292.
•  Li,  Y. L.,  Zheng,  Y. J.,  Tang,  L.,  Su,  Z. Y., and  Xiong,  C. (2016). Study on the Identification of Geographical Indication Wuchang Rice Based on the Content of Inorganic Elements. Spectrosc. Spect. Anal., 36, 834-837.
•  Liu,  Z. J.,  Shi,  Z. L.,  Wang,  J. B.,  Zhang,  H. Z.,  Li,  C. Y.,  Niu,  M.,  He,  Q.,  Zhen,  H. S., and  Xiao,  X. H. (2017). Quality Evaluation of Salivia Miltiorrhiza Bge. Based on Chemical Profile and Bioassay of Antiplatelet Aggregetion. Chinese J. Anal. Chem., 45, 693-699.
•  Ma,  G. C.,  Zhang,  Y. B.,  Zhang,  J. Y.,  Wang,  G. Q.,  Chen,  L. Y.,  Zhang,  M. L.,  Liu,  T.,  Liu,  X., and  Lu,  C. Y. (2016). Determining the geographical origin of Chinese green tea by linear discriminant analysis of trace metals and rare earth elements: Taking Dongting Biluochun as an example. Food Control, 59, 714-720.
•  Moreda-Piñeiro,  A.,  Fisher,  A., and  Hill,  S. J. (2003). The classification of tea according to region of origin using pattern recognition techniques and trace metal data. J. Food Compos. Anal., 16, 195-211.
•  Osorio,  M. T.,  Koidis,  A., and  Papademas,  P. (2015). Major and trace elements in milk and Halloumi cheese as markers for authentication of goat feeding regimes and geographical origin. Int. J. Dairy Technol., 68, 573-581.
•  Peluso,  I. and  Serafini,  M. (2017). Antioxidants from black and green tea: from dietary modulation of oxidative stress to pharmacological mechanisms. Br. J. Pharmacol., 174, 1195-1208.
•  Salahinejad,  M. and  Aflaki,  F. (2010) Toxic and essential mineral elements content of black tea leaves and their tea infusions consumed in Iran. Biol. Trace. Elem. Res., 134, 109-117.
•  Uršulintrstenjak,  N.,  Levanic,  D.,  Primorac,  L.,  Bošnir,  J.,  Vahčić,  N., and  Šarić,  G. (2015). Mineral profile of Croatian honey and differences due to its geographical origin. Czech J. Food. Sci., 33, 156-164.
•  Zeng,  Z.,  Yang,  D. H.,  Tao,  J. Q.,  Yang,  T.,  Yuan,  M., and  Zeng,  H. P. (2004). Application of High Performance Liquid Chromatographic Fingerprint to Phragmites Communis Trin., a Traditional Chinese Medicine. Chinese J. Anal. Chem., 32, 1035-1038.
•  Zhao,  H. Y.,  Zhang,  S. L., and  Zhang,  Z. W. (2017). Relationship between multi-element composition in tea leaves and in provenance soils for geographical traceability. Food Control, 76, 82-87.