Identification of species within Tetrastigma (Miq.) Planch. (Vitaceae) based on DNA barcoding techniques


  • Yuan-Miao FU,

    Corresponding author
    1. (The Key Laboratory of Conservation Biology for Endangered Wildlife of the Ministry of Education, College of Life Sciences, Zhejiang University, Hangzhou 310058, China)
    2. (Laboratory of Systematic and Evolutionary Botany and Biodiversity, Institute of Plant Sciences and Conservation Center for Gene Resources of Endangered Wildlife, Zhejiang University, Hangzhou 310058, China)
    Search for more papers by this author
  • Wei-Mei JIANG,

    Corresponding author
    1. (The Key Laboratory of Conservation Biology for Endangered Wildlife of the Ministry of Education, College of Life Sciences, Zhejiang University, Hangzhou 310058, China)
    2. (Laboratory of Systematic and Evolutionary Botany and Biodiversity, Institute of Plant Sciences and Conservation Center for Gene Resources of Endangered Wildlife, Zhejiang University, Hangzhou 310058, China)
    Search for more papers by this author
  • Cheng-Xin FU

    Corresponding author
    1. (The Key Laboratory of Conservation Biology for Endangered Wildlife of the Ministry of Education, College of Life Sciences, Zhejiang University, Hangzhou 310058, China)
    2. (Laboratory of Systematic and Evolutionary Botany and Biodiversity, Institute of Plant Sciences and Conservation Center for Gene Resources of Endangered Wildlife, Zhejiang University, Hangzhou 310058, China)
    Search for more papers by this author

 These authors contributed equally to this work.

 Author for correspondence. E-mail:; Tel.: 86-571-88206607; Fax: 86-571-86432273.


Abstract  Many species of Tetrastigma (Miq.) Planch. (Vitaceae) have long been used as medicinal plants in China, and some are endangered due to overexploitation. Although adulterants are often added to traditional Chinese medicines, there is no reliable or practical method for identifying them. In this study, we used four markers (rbcL, matK, trnH-psbA, and internal transcribed spacer [ITS]) as DNA barcodes to test their ability to distinguish species of Tetrastigma. The results indicated that the best barcode was ITS, which showed significant inter-specific genetic variability, and thus its potential as a DNA barcode for identifying Tetrastigma. Multiple loci provided a greater ability to distinguish species than single loci. We recommend using the combined rbcL+matK+ITS barcode for the genus. Phylogenetic trees from each barcode were compared. Analyses using the unweighted pair group method with arithmetic mean discriminated an equal or greater percentage of resolvable species than did neighbor joining, maximum likelihood, or maximum parsimony analyses. Additionally, five medicinal species of Tetrastigma, especially T. hemsleyanum, could be identified precisely using DNA barcoding.

Tetrastigma (Miq.) Planch. (Vitaceae) is a widespread genus of approximately 100 species occurring from Asia to Oceania. Forty-four species (24 endemic), including some important medicinal herbs, are found in China (Ren & Wen, 2007). The tubers of Tetrastigma hemsleyanum Diels & Gilg ex Diels are used as traditional Chinese medicine against inflammatory disorders, as an analgesic and antipyretic, to dispel phlegm and to improve circulation (Liu et al., 2002), and to strengthen liver function. It is also reported to have antiviral, and even antitumor properties (Yang et al., 1989; Xu et al., 2008). Tetrastigma serrulatum (Roxb.) Planch., T. obovatum (Laws.) Gagnep., T. obtectum (Wall.) Planch., T. formosanum (Hemsl.) Gagnep., T. lenticellatum C.Y. Wu ex W.T. Wang, T. triphyllum (Gagnep.) W.T. Wang, and T. planicaule (Hook.) Gagnep. are also used as medicinal herbs (State Administration of Traditional Chinese Medicine, 2004). In traditional Chinese medicine markets, the tubers of T. hemsleyanum are mainly from wild populations, although some are now from plants cultivated in small farms. Along with these are other natural products reported to be tubers of Tetrastigma. It is difficult to authenticate tubers of Tetrastigma by traditional morphological methods. Although Li & Wu (1995) carried out a taxonomic study of Tetrastigma, and phylogenies of the Vitaceae (Rossetto et al., 2001; Soejima & Wen, 2006; Rossetto et al., 2007; Wen et al., 2007) have been proposed, none of them provide information on distinguishing the tubers of Tetrastigma from imitations. Accordingly, it is necessary to develop a simple method for authenticating the species of Tetrastigma.

DNA barcoding is a method that has been proposed for the identification of species based on unique sequences in specific regions of DNA (Hebert et al., 2003a). DNA barcoding has shown powerful potential in taxonomic research. In animals, the mitochondrial gene CO1 (cytochrome c oxidase subunit 1) has been used as a DNA barcode. Numerous studies have shown that CO1 can provide efficient and accurate identification of most animal species (Hebert et al., 2003b; Hebert et al., 2004; Hajibabaei et al., 2006; Yoo et al., 2006; Yancy et al., 2008). Mitochondrial genes in plants, however, do not work as well because of the lower rate of change in gene sequences. DNA barcoding in plants has proven to be more difficult than in animals (Chase et al., 2005; Pennisi, 2007; Fazekas et al., 2009). Although the difficulties of plant DNA barcoding have been debated, many detailed studies have shown that DNA barcoding is an effective tool for plant identification (Kress & Erickson, 2007; Lahaye et al., 2008; Newmaster et al., 2008). Recently, at the Third International Barcode of Life Conference in Mexico, the Plant Working Group of the Consortium for the Barcode of Life (CBOL) designated rbcL and matK as core barcodes and trnH-psbA or internal transcribed spacer (ITS) regions as supplemental loci for differentiating plant species.

In this study, we used four candidate DNA barcodes (rbcL, matK, trnH-psbA, and ITS) to identify the species of Tetrastigma and to differentiate them from other medicinal plants. Our goal was to examine the efficiency of the barcodes individually and in combination to determine their suitability for distinguishing species from closely related species and from adulterants.

1 Material and methods

1.1 Plant materials and barcode primer selection

Plant samples (listed in Table 1) from 93 individuals representing 23 species, including five medicinal species, were collected from different localities in China and Thailand over a period of three years. Fresh leaves were dried in silica gel upon collection. Vouchers of the collections were deposited in the Herbarium of Zhejiang University (HZU) and in the Herbarium of the Kunming Institute of Botany, Chinese Academy of Sciences (KUN). Yuan-Miao FU and Cheng-Xin FU identified the samples. The barcodes and primers selected in this study are listed in Table 2.

Table 1. Tetrastigma taxa sampled, collection, and voucher specimen information
SpeciesLocality of collectionVoucher no.CollectorsSamples used
  1. †Medicinal plants.

T. caudatum Merr. & ChunQixianling, Hainan, ChinaF0902034Yuan-Miao FU1
Wuzhishan, Hainan, ChinaF0902005Yuan-Miao FU1
Wuzhishan, Hainan, ChinaF0902008Yuan-Miao FU1
Wuzhishan, Hainan, ChinaF0902003Yuan-Miao FU4
Wuzhishan, Hainan, ChinaF0902009Yuan-Miao FU2
Wuzhishan, Hainan, ChinaF0902007Yuan-Miao FU1
Wuzhishan, Hainan, ChinaF0902006Yuan-Miao FU1
T. planicaule (Hook.) Gagnep.Wuzhishan, Hainan, ChinaF0902170Yuan-Miao FU2
Guilin, Guangxi, ChinaF0804112Yuan-Miao FU2
Nanning, Guangxi, ChinaF0906010Yuan-Miao FU2
T. hemsleyanum Diels & Gilg ex DielsNingbo, Zhejiang, ChinaJ07004Wei-Mei JIANG1
Sanming, Fujian, ChinaF08001Yuan-Miao FU1
Zhangjiajie, Hunan, ChinaS08003Shu-Qing SUN1
Xingshan, Hubei, ChinaF0810018Yuan-Miao FU1
Jinggangshan, Jiangxi, China0808023Yun-Peng ZHAO1
Wuyi, Zhejiang, ChinaF07008Yuan-Miao FU1
Meitan, Guizhou, ChinaF07007Yuan-Miao FU1
Jianyang, Fujian, ChinaF0812001Yuan-Miao FU1
Shixing, Guangdong, ChinaF07005Yuan-Miao FU1
Jinfoshan, Chongqing, ChinaS08004Shu-Ting YANG1
T. serrulatum (Roxb.) Planch.Pingbian, Yunnan, ChinaLP0904395Pan LI3
T. erubescens Planch.Xishuangbanna, Yunnan, ChinaH0807003Pan LI1
Dinghushan, Guangdong, ChinaF2010038Cheng-Xin FU4
Jianfengling, Hainan, ChinaF0902041Yuan-Miao FU2
Wuzhishan, Hainan, ChinaF0902004Yuan-Miao FU1
Jianfengling, Hainan, ChinaF0902047Yuan-Miao FU3
T. cruciatum Craib & Gagnep.Xishuangbanna, Yunnan, ChinaH0807001Pan LI1
Jinghong, Yunnan, ChinaBQ1005354Zhe-Chen QI1
Jinghong, Yunnan, ChinaBQ1005359Zhe-Chen QI1
Chiang Mai, ThailandF09073Cheng-Xin FU1
T. funingense C. L. LiFuning, Yunnan, ChinaS08112Pan LI2
T. pachyphyllum (Hemsl.) ChunQixianling, Hainan, ChinaF0902038Yuan-Miao FU1
Qixianling, Hainan, ChinaF0902030Yuan-Miao FU1
Qixianling, Hainan, ChinaF0902036Yuan-Miao FU1
T. erubescens Planch. var. monophyllum Gagnep.Malipo, Yunnan, ChianGBOWS466Zhe-Chen QI1
T. napaulense (DC.) C. L. LiChiang Mai, ThailandF09054Cheng-Xin FU1
Malipo, Yunnan, ChianLP0904400Pan LI1
Xichou, Yunnan, ChinaLP0904414Pan LI1
T. cauliflorum Merr.Nanning, Guangxi, ChinaF0906004Yuan-Miao FU2
Nanning, Guangxi, ChinaF0906002Yuan-Miao FU2
Nanning, Guangxi, ChinaF0906003Yuan-Miao FU2
T. obovatum (Laws.) Gagnep.Hekou, Yunnan, ChinaGBOWS1229Zhe-Chen QI2
T. tonkinense Gagnep.Nanning, Guangxi, ChinaF0906001Yuan-Miao FU2
T. pseudocruciatum C. L. LiQixianling, Hainan, ChinaF0902029Yuan-Miao FU1
Qixianling, Hainan, ChinaF0902031Yuan-Miao FU1
Qixianling, Hainan, ChinaF0902028Yuan-Miao FU1
Qixianling, Hainan, ChinaF0902035Yuan-Miao FU1
T. delavayi Gagnep.Malipo, Yunnan, ChinaLP0904401Pan LI1
T. obtectum (Wall.) Planch. var. pilosum Gagnep.Xingshan, Hubei, ChinaF0810117Yuan-Miao FU2
Jinfoshan, Chongqing, ChinaS08001Shu-Qing SUN4
T. papillatum (Hance) C. Y. WuQixianling, Hainan, ChinaF0902022Yuan-Miao FU1
Qixianling, Hainan, ChinaF0902037Yuan-Miao FU1
Qixianling, Hainan, ChinaF0902023Yuan-Miao FU1
Longzhou, Guangxi, ChinaQ0903082Zhe-Chen QI1
T. henryi Gagnep.Mengla, Yunnan, ChinaBQ1005362Zhe-Chen QI1
T. retinervium Planch. var. pubescens C. L. LiXichou, Yunnan, ChinaLP0904410Pan LI1
T. pubinerve Merr. & ChunXishuangbanna, Yunnan, ChinaH0807002Pan LI1
T. rumicispermum (Laws.) Planch.Chiang Mai, ThailandF09055–1Cheng-Xin FU1
Malipo, Yunnan, ChinaGBOWS483Zhe-Chen QI1
Malipo, Yunnan, ChinaGBOWS504Zhe-Chen QI2
T. kwangsiense C. L. LiMalipo, Yunnan, ChinaGBOWS669Zhe-Chen QI3
Hekou, Yunnan, ChinaGBOWS758Zhe-Chen QI1
T. subtetragonum C. L. LiHekou, Yunnan, ChinaGBOWS1244Zhe-Chen QI1
Hekou, Yunnan, ChinaGBOWS1267Zhe-Chen QI2
Table 2.  Polymerase chain reaction primers and conditions used for amplification of barcode sequences
Gene and regionName of primerPrimer sequence 5′–3′Reaction condition
  1. ITS = internal transcribed spacer.

ATGTCACCACAAACAGAAAC TCGCATGTACCTGCAGTAGC (Fay et al., 1997)94°C 5 min 94°C 45 s, 55°C 45 s, 72°C 60 s, 35 cycles 72°C 10 min
CGATCTATTCATTCAATATTTC TCTAGCACACGAAAGTCGAAGT (Cuénoud et al., 2002)94°C 5 min 94°C 60 s, 54°C 60 s, 72°C 90 s, 35 cycles 72°C 10 min
94°C 5 min 94°C 45 s, 56°C 45 s, 72°C 50 s, 35 cycles 72°C 10 min
AGAAGTCGTAACAAGGTTTCCGTAGG TCCTCCGCTTATTGATATGC (White et al., 1990)94°C 5 min 94°C 45 s, 58°C 45 s, 72°C 60 s, 35 cycles 72°C 10 min

1.2 DNA extraction, amplification, and sequencing

Genomic DNA was isolated from approximately 15 mg of each leaf sample following the CTAB protocol of Doyle (1991). Polymerase chain reaction (PCR) amplification of the four candidate DNA barcodes was carried out in a PTC-100 PCR DNA Thermal Cycler (Bio-Rad Laboratories, Hercules, CA, USA) using approximately 40 ng genomic DNA as a template in a 50 μL reaction mixture containing 5 μL 10 × buffer (Mg2+ free), 2.5 mmol/L MgCl2, 0.2 mmol/L each dNTP, 0.3 μmol/L each primer, 10 μg bovine serum albumin, and 2.0 U Taq DNA polymerase. Primers and reaction conditions used in this study are listed in Table 2. The PCR products were run in a 1.5% agarose gel, stained with ethidium bromide, and visualized under UV light. The available PCR products were purified using a PCR purification kit (Shanghai Bio-engineering, Shanghai, China) and were directly sequenced by Shanghai Bio-engineering.

1.3 Date analyses

Sequences were edited and assembled by using Sequencher v4.0.5 (Gene Codes, Ann Arbor, MI, USA). Sequence alignments for each region were initially carried out in ClustalX version 2.0 (Larkin et al., 2007), then modified artificially. Multiple sequence alignments were concatenated into a single file using Geneious Pro version 4.8.5 (Drummond et al., 2009). Genetic distances among barcode sequences were quantified using the Kimura 2-parameter (K2P) distance model through MEGA version 4.0 (Tamura et al., 2007). The significance of divergence was then assessed by Median and Wilcox on two-sample tests ( The barcoding gaps were graphed by comparing the distributions of intra- and inter-specific divergences of each candidate barcodes. To evaluate whether species were recovered as monophyletic with each barcode, we used standard phylogenetic techniques: neighbor joining (NJ) and unweighted pair group method with arithmetic mean (UPGMA) with MEGA version 4.0 (Tamura et al., 2007); and maximum parsimony (MP) and maximum likelihood (ML) with PAUP version 4.0b10 (Swofford, 2002). NJ and UPGMA analysis used K2P model and ML analyses used best fit model. The best fit model as given by applying ModelTest 3.7 (Posada & Crandall, 1998). Branch support for all analyses was assessed with 500 bootstrap replicates.

2 Results

2.1 Character analysis of each barcode

We produced 358 DNA barcode sequences, representing 23 species of Tetrastigma. The rbcL, matK, and trnH-psbA of all species could be sequenced, but the ITS region of 4.3% of the species could not be sequenced (Table 3). For individual regions, the aligned sequence length was 436 bp for trnH-psbA, 704 bp for rbcL, 724 bp for ITS, and 790 bp for matK (Table 3). Among the DNA barcodes, the ITS region showed the greatest number of variable sites (236) and mean inter-specific distance (0.0745) (Table 3). Among the plastid barcodes, the non-coding region trnH-psbA had 64 variable sites, which were approximately 1.6 and 3.5 times more variable than matK and rbcL, respectively (Table 3). matK had 39 variable sites, approximately 2.1 times more variable than rbcL (18 variable sites).

Table 3.  DNA barcoding utility of single and combination barcodes for Tetrastigma
Potential barcodeAligned length (bp)Variable sitesMean intra-specific distanceMean inter-specific distanceTotal individualsSequenced individuals (%)Sequenced species (%)Species with ≥ 2 individuals sequenced (%)
  1. Species with only one individual did not join the calculation of intra-specific distances.

rbcL 704 180.00010.00639393 (100)23 (100) 18 (78.3)
matK 790 390.00020.00499387 (93.5)23 (100.0)18 (78.3)
trnH-psbA 436 640.00060.018493 93 (100.0)23 (100.0)18 (78.3)
ITS 7242360.00360.07459385 (91.4)22 (95.7) 16 (72.7)
ITS+rbcL14282540.00180.03739385 (91.4)22 (95.7) 16 (72.7)
ITS+matK15142740.00180.03309380 (86.0)22 (95.7) 16 (72.7)
ITS+trnH-psbA11602940.00240.05159385 (91.4)22 (95.7) 16 (72.7)
matK+rbcL1494 570.00020.00549387 (93.5)23 (100.0)18 (78.3)
matK+trnH-psbA12261030.00030.00909387 (93.5)23 (100.0)18 (78.3)
rbcL+trnH-psbA1140 820.00030.010493 93 (100.0)23 (100.0)18 (78.3)
rbcL+matK+trnH-psbA19301200.00030.00789387 (93.5)23 (100.0)18 (78.3)
rbcL+matK+ITS22182930.00120.02389380 (86.0)22 (95.7) 16 (72.7)
rbcL+matK+trnH-psbA+ITS26543570.00110.02259380 (86.0)22 (95.7) 16 (72.7)

2.2 Monophyletic test based on phylogenetic trees

The identification of single and combined barcodes was determined by evaluating the percentage of monophyly of each species using NJ, UPGMA, MP, and ML trees. In all cases, UPGMA analyses discriminated an equal or greater percentage of resolvable species than did NJ analyses, NJ analyses equal or greater than ML, and ML analyses equal or greater than MP. The highest species monophyly value (93.8%) was achieved using ITS or combined barcodes when ITS was involved (such as rbcL+matK+ITS and rbcL+matK+trnH-psbA+ITS), and the next best performance (88.9%) was provided by rbcL+matK+trnH-psbA (Table 4). The rbcL, matK, and trnH-psbA had species identification rates of 22.2%, 61.1%, and 66.7%, respectively (Table 4). The combined barcodes, matK+rbcL, matK+trnH-psbA, and rbcL+trnH-psbA had the same species identification rates (83.8%), and the rbcL+matK+trnH-psbA rate was 88.9%. Amplification and sequencing of the ITS region was not successful in all tested species (Table 3). Accordingly, after taking the sequencing success rates, ITS analysis resolved 89.8% of the species. Additionally, the five tested medicinal species (T. planicaule, T. hemsleyanum, T. serrulatum, T. obovatum, and T. obtectum var. pilosum) were consistently identified at bootstrap values above 95% (Fig. 1).

Table 4.  Percent of Tetrastigma species recovered as monophyletic based on phylogenetic trees for each barcode
Potential barcodeNJUPGMAMPML
  1. Resolved species with bootstrap >70% are in brackets. ITS, internal transcribed spacer; ML, maximum likelihood; MP, maximum parsimony; NJ, neighbor joining; UPGMA, unweighted pair group method with arithmetic mean.

rbcL22.2 (11.1)22.2 (11.1)16.7 (11.1)16.7 (5.5) 
matK55.6 (22.2)61.1 (33.3)55.6 (27.8)  50 (16.7)
trnH-psbA55.6 (44.4)66.7 (38.9)33.3 (16.7)55.6 (33.3)
ITS93.8 (93.8)93.8 (93.8)81.3 (81.3)87.5 (87.5)
ITS+rbcL93.8 (93.8)93.8 (93.8)87.5 (81.3)87.5 (87.5)
ITS+matK93.8 (93.8)93.8 (93.8)87.5 (87.5)87.5 (87.5)
ITS+trnH-psbA93.8 (93.8)93.8 (93.8)87.5 (87.5)93.8 (93.8)
matK+rbcL72.2 (33.3)83.3 (55.6)66.7 (38.9)67.7 (27.8)
matK+trnH-psbA72.2 (61.1)83.8 (72.2)66.7 (55.6)72.2 (61.1)
rbcL+trnH-psbA77.8 (50.0)83.3 (38.9)44.4 (27.8)72.2 (50.0)
rbcL+matK+trnH-psbA83.3 (72.2)88.9 (66.7)72.2 (55.6)83.3 (72.2)
rbcL+matK+ITS93.8 (93.8)93.8 (93.8)87.5 (87.5)87.5 (87.5)
rbcL+matK+trnH-psbA+ITS93.8 (93.8)93.8 (93.8)87.5 (87.5)87.5 (87.5)
Figure 1.

A neighbor joining tree of Kimura 2-parameter distances of internal transcribed spacer sequence data for Tetrastigma species. Numbers at nodes represent bootstrap values with 500 replicates (only values greater than 50 are shown). Numbers adjacent to taxon names are voucher numbers (see Table 1).

2.3 Barcoding gap test

The barcoding gap between intra- and inter-specific distance was determined by graphing the distribution of variation in K2P distances for rbcL, matK, trnH-psbA, and ITS (Fig. 2). The significance between intra- and inter-specific distances was evaluated by applying median and Wilcoxon two-sample tests for each barcode. Our results showed that the intra-specific distance was lower than inter-specific distance, with the highest significances found in trnH-psbA (Wilcoxon two-sample test, P≤ 3.316e-143), followed by ITS (Wilcoxon two-sample test, P≤ 2.187e-133) (Table 5). We did not find any large barcoding gaps as typical of CO1 in animals (Meyer & Paulay, 2005), although in the ITS matrix the distribution of intra- versus inter-specific distances were relatively well separated (Fig. 2).

Figure 2.

Distribution of intra- and inter-specific Kimura 2-parameter (K2P) distances for four candidate barcodes. The X-axis relates to the K2P distances, and the Y-axis corresponds to the number of occurrences.

Table 5.  Median test and Wilcoxon two-sample test based on inter-specific versus intra-specific Kimura 2-parameter distances of each barcode
RegionMedian testWilcoxon two-sample test
  1. #A, number of inter-specific distances; #B, number of intra-specific distances; ITS, internal transcribed spacer; W, Wilcoxon two-sample test.

ITS#A = 3340 #B = 218, Median = 0.0728, P≤ 3.050e-51#A = 3340 #B = 218, W = 26697, P≤ 2.187e-133
rbcL#A = 4021 #B = 253, Median = 0.071, P≤ 3.554e-37#A = 4021 #B = 253, W = 105095, P≤ 6.979e-116
matK#A = 3519 #B = 222, Median = 0.051, P≤ 5.266e-32#A = 3519 #B = 222, W = 54663.5, P≤ 4.049e-118
trnH-psbA#A = 4033 #B = 245, Median = 0.0134, P≤ 2.024e-49#A = 4033 #B = 245, W = 45833, P≤ 3.316e-143

3 Discussion

DNA barcoding has been proposed as a rapid tool for identifying species. Since the concept was advanced, however, some experts have expressed doubts as to its utility (Hebert et al., 2003a; Mallet & Willmott, 2003; Will & Rubinoff, 2004; Meyer & Paulay, 2005). To date, there are no standard barcodes for plants. A great deal of effort is still needed before DNA barcoding of plants can be applied and accepted widely.

In our study, as in other barcoding studies (Pang et al., 2010; Zuo et al., 2010), we found the ITS region to be the most valuable and powerful among the tested regions. Polymerase chain reaction and sequencing of the ITS region failed or was carried out with difficulty in some species (e.g., T. rumicispermum (Laws.) Planch. and T. papillatum (Hance) C.Y. Wu). Difficulties with PCR and sequencing of the ITS regions was also a problem in other barcoding studies (Sass et al., 2007; Song et al., 2009) and remains one of the main constraints for using ITS regions as a standard barcode.

The Plant Working Group CBOL (2009) recommended rbcL and matK genes as the core plant barcodes. In Tetrastigma, however, rbcL showed little variation and was powerless in differentiating species (Tables 3, 4), implying that rbcL is unsuitable for distinguish closely related species. Clerc-Blain et al. (2010) and Zuo et al. (2010) reported similar results in their studies. matK, as the most widely accepted potential barcode, also has a problem with primer universality (Chase et al., 2007; Kress & Erickson, 2008; Lahaye et al., 2008). The matK gene was amplified successfully in almost all tested species with the current primers in this study. Only a few individuals failed to amplify, and it was therefore much more powerful than rbcL in resolving the species of Tetrastigma. The non-coding spacer trnH-psbA provided more variable sites than the coding regions (rbcL and matK), and showed better discrimination than matK in the UPGMA analysis, but many indels were found in the trnH-psbA spacer. Kress et al. (2005) and Logacheva et al. (2008) considered that indels of the non-coding spacer trnH-psbA induced variation in sequence length among different species, as in our study results, suggesting that non-coding regions are not suitable in distinguishing species of Tetrastigma.

Multi-locus barcodes always show better powers of discrimination than single regions in species identification (Plant Working Group CBOL, 2009). In this study, when the two core barcodes (rbcL and matK) were combined for further analysis, 72.2% of the species of Tetrastigma could be resolved, which was much higher than using a single barcode (rbcL 22.2%, matK 61.1%). We also tested the validity of rbcL+matK+X (ITS or trnH-psbA), which was proposed by the Chinese Plants Barcoding program. The results showed the identification rate of rbcL+matK+ITS and rbcL+matK+trnH-psbA to be 93.8% and 88.9%, respectively (Table 4). Thus the three combined barcodes were better than the two coding cpDNA barcodes, although the ITS region alone was able to discriminate species with a similar rate of success.

Neighbor joining, UPGMA, MP, and ML analyses were used to test the percent of monophyly of the species in this study. Among the analyses, UPGMA always yielded good results in differentiating more species than the other methods tested. Similar results were obtained by Lahaye et al. (2008). Neighbor joining analysis was considered to be the simplest and most widely used method in barcoding research (Frezal & Leblois, 2008). Neighbor joining tree reconstructions based on K2P distances in this study revealed that T. subtetragonum C.L. Li, T. obtectum var. pilosum, and T. rumicispermum were consistently resolved in all single- and multi-locus analyses. These species are taxonomically distributed in different sections of Tetrastigma, and are not considered to be closely related (Li & Wu, 1995), which is consistent with the observed genetic distance between them. The results show that DNA barcoding has no problem in distinguishing species with clear morphological boundaries. Tetrastigma erubescens Planch., however, showed polyphyly in all analysis. Similar results were also found in animal DNA barcoding studies (Steinke et al., 2009). According to the Flora of China (Ren & Wen, 2007), T. erubescens has two varieties (var. erubescens and var. monophyllum). Our results showed that different samples of T. erubescens from Yunnan, Guangdong, and Hainan provinces were clearly different, suggesting that polyphyly in T. erubescens might be caused by intraspecific genetic variation and geographical differentiation. For some taxa (T. pubinerve Merr. & Chun, T. delavayi Gagnep., T. henryi Gagnep., and T. retinervium Planch. var. pubescens C.L. Li), we sampled only one individual. They formed independent clades and were differentiated from related species in ITS and also in some combined barcode analyses (such as rbcL+matK+ITS, rbcL+matK+trnH-psbA+ITS) (Fig. 1; Appendix S1). Further sampling and research on these taxa is necessary.

Many important medicinal plants have been reported in Tetrastigma. In this study, we tested five species reported to have medicinal properties (T. planicaule, T. hemsleyanum, T. serrulatum, T. obovatum, and T. obtectum var. pilosum), and10 individuals of T. hemsleyanum. It will be of great significance if these species could be accurately identified. Our results indicated that T. hemsleyanum is a distinct species, and all five species were separable from related species by ITS (Fig. 1) and some combined barcode analyses (such as rbcL+matK+trnH-psbA, rbcL+matK+ITS, rbcL+matK+trnH-psbA+ITS) (Appendix S1), which provides evidence for the feasibility of using DNA barcoding to identify the species of Tetrastigma.

Meyer & Paulay (2005) and Lahaye et al. (2008) pointed out that an ideal barcode can be detected by survival of a barcoding gap (i.e., inter-specific genetic distance is significantly bigger than intra-specific genetic distance and forms an obvious spacer). Although we did not find clear barcoding gaps in the distribution of the four tested barcodes (Fig. 2), such as co1 in animals (Meyer & Paulay, 2005), the intra- and inter-specific divergence in each barcode was significantly different (Table 5).

In conclusion, suitable barcodes (rbcL+matK+ ITS or ITS alone) for Tetrastigma were established in this study. We found that the medicinal species, T. hemsleyanum, can be accurately identified using DNA barcodes. It remains necessary to carry out further research on the more than 100 species of Tetrastigma.


Acknowledgements  We thank Pan LI, Zhe-Chen QI, Yun-Peng ZHAO, Shu-Ting YANG, and Shu-Qing SUN for their help in collecting specimens. We also thank Yi SUN for her assistance with the experiments and data analyses, and thank David BOUFFORD of Harvard University for his revision in English. This study was supported by the Research Fund for the Large-scale Scientific Facilities of the Chinese Academy of Sciences (Grant No. 2009-LSF-GBOWS-01), by the National Key Project of Scientific and Technical Supporting Programs Funded by the Ministry of Science and Technology of China (Grant No. 2008BAC39B05), and by the Key Program of Traditional Chinese Medicine Modernization of Zhejiang (Grant No. 2006C13077).