It is long believed that some spices may help protect against certain chronic conditions. Spices are usually parts of plants that have been powdered into small pieces. Have you ever wondered what the curry powder in your dish is made of? The aim of this work was to develop an appropriate DNA-based method for assessment of spice identity. Selecting the best marker for species recognition in the Zingiberaceae family. Six DNA regions were investigated in silico, including ITS, matK, rbcL, rpoC, trnH-psbA and trnL. Then, only four regions (ITS, matK, rbcL and trnH-psbA) were included in the simulated HRM (High-resolution Melting) analysis as the results from previous analysis showed that rpoC and trnL may not be suitable to be used to identify Zingiberaceae species in HRM analysis based on both the percentage of nucleotide variation and GC content. Simulated HRM analysis was performed to test the feasibility of Bar-HRM. We found that ITS2 is the most effective region to be used for identification of the studied species and thus was used in laboratory HRM analysis. All seven tested Zingiberaceae plants were then able to be distinguished using the ITS2 primers in laboratory HRM. Most importantly the melting curves gained from fresh and dried tissue overlapped, which is a crucial outcome for the applicability of the analysis. The method could be used in an authentication test for dried products. In the authentication test, only one of seven store-sold Zingiberaceae products that were tested contained the species listed on their labels, while we found substitution/contamination of the tested purchased products in the rest.
Citation: Osathanunkul M, Ounjai S, Osathanunkul R, Madesis P (2017) Evaluation of a DNA-based method for spice/herb authentication, so you do not have to worry about what is in your curry, buon appetito! PLoS ONE 12(10): e0186283. https://doi.org/10.1371/journal.pone.0186283
Editor: Diego Breviario, Istituto di Biologia e Biotecnologia Agraria Consiglio Nazionale delle Ricerche, ITALY
Received: June 18, 2017; Accepted: September 28, 2017; Published: October 11, 2017
Copyright: © 2017 Osathanunkul et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This work was supported by the National Research Council of Thailand.
Competing interests: The authors declare that no competing interests exist.
Spices are primarily used for flavouring, colouring or preserving food, and they usually come from seeds, fruit, roots, bark, berries, and buds of plants. In addition to culinary use, spices are sometimes used in medicine, cosmetics or perfume productions and in religious rituals. Modern science has now shown that many of them do indeed carry remarkable health benefits. There has been great expansion of spicy products in the Thailand market as the number of products increased rapidly, particularly spices in the ginger or Zingiberaceae family, which includes several important spice producers. There are many plants in the Zingiberaceae family that have been deemed as economic crops that can generate national income, such as Zingiber officinale, Curcuma longa, Curcuma zedoaria, Kaempferia parviflora, Boesenbergia rotunda, Zingiber montanum, Alpinia galangal and Amomum uliginosum . The external appearance of plants in the family is very similar, and identification based on morphological characters is sometimes difficult. Moreover, plants sold as spices are commonly powdered into small pieces, which also hampers morphological identification. Most spices are distributed in a processed form and it is very difficult to prove and identify their original species, but authentication of spice products is vital for both manufacturers and consumers.
Molecular identification through DNA barcoding is an efficient technique for the identification of plant species. A considerable amount of literature has actually been published, and it shows that DNA barcoding can be used to distinguish closely related medicinal plants [2–4]. The increasing and diverse applications of medicinal plant barcoding have been highlighted in several scientific reviews [5,6]. In addition, DNA barcoding is now recognized as a suitable molecular technology to greatly improve the traceability of food and drinks from producers to dining tables. Currently, this method continues to be largely exploited for genetic identification and adulteration detection in many plant species along with their food products . However, DNA barcoding in plants does have limitations, including difficulty in DNA amplification due to degraded DNA in the processed samples , mismatches at binding sites such as matK [3,9,10], and low rates of discrimination capabilities, such as rbcL [4,11]. Developing and validating sequencing-free methods that are reliable and faster than DNA barcoding is challenging. Recent development of a DNA-based identification approach combines the advantages of two techniques, i.e., barcoding and high-resolution melting (HRM) analysis, called Bar-HRM [12–15]. This rapid growing sensitivity is likely to generate an increase in the speed and accuracy of species identification compared with morphological-based identification or using DNA barcoding alone.
Bar-HRM has been used in a number of comparable applications. Bar-HRM is revolutionising the way we can trace plant species in herbal, agricultural and food products; however, the approach faces challenges that can hinder our ability to produce robust results. Recently, a comprehensive evaluation of the use of Bar-HRM in plant species identification addressed the choice of markers used, which could lead to success in different plant groups . Different primer sets in HRM analysis were found to be suitable for different plant groups. Here, we emphasize choosing the best marker for spice species in the Zingiberaceae family. The success of this study will highlight the potential of Bar-HRM as a rapid, sensitive, economical, high-throughput and taxonomical expertise-free technique for routine identification of spice identity as a quality assurance approach. This will lead to creating standards for raw material inspection and spice processing manufacturing standards for consumer confidence, as well as quality assurance of Thai spicy products for export. Spices can then be developed as significant economic crops to generate national income.
Materials and methods
DNA mining of barcode regions
To address the most suitable markers for identification of spices in Zingiberaceae based on the Bar-HRM technique, a few datasets were constructed to conduct a sequence profile analysis. The first dataset was used to evaluate the sequence profile of ITS, matK, trnH-psbA spacer region, rbcL, rpoC and trnL retrieved from GenBank, which included sequences from the entire family of Zingiberaceae (in total of 3,377 sequences from 62 genus) (Fig 1). With the second dataset, we deepened our scope to evaluate the sequence profile of amplicons generated by the Bar-HRM primers with only seven spice species in Zingiberaceae (Table 1).
Numbers of sequences from each genus are indicated.
Sequences of ITS, matK, trnH-psbA spacer region, rbcL, rpoC and trnL were extracted from GenBank using the keyword ‘Zingiberaceae + (region title)’ at the end of September 2015. Some of the sequences retrieved from public databases, including GenBank, were found to be low quality with no known associated herbarium vouchers. For this reason, all of the sequences were subjected to critical evaluation and any low-quality sequences were removed. Criteria used to filter the sequences were as follows: (1) sequences were not ‘unverified’ without a species name, (2) sequences contained <3% ambiguous base ‘N’, and (3) a maximum of 3 samples (sequences) were included from a species. After processing, multiple alignments were made from the selected sequences using MEGA6 , and the sequence length (bp), conserved sites (%), variable sites (%), and GC content (%) of each data set were recorded.
Plant material and DNA isolation
In laboratory analyses, seven Zingiberaceae spices with economic significance, including Alpinia galangal, Amomum uliginosum, Boesenbergia rotunda, Curcuma longa, Kaempferia parviflora, Zingiber montanum and Zingiber officinale, were investigated. Dried and fresh plant tissues for DNA extraction were kindly provided by the Queen Sirikit Botanic Garden (QSBG) and the faculty of pharmacy, Chiang Mai University, respectively (Table 1). The plant material was ground with liquid nitrogen, and 100 mg of fine powder was then used for DNA extraction with the Nucleospin Plant II kit (Macherey-Nagel, Germany) following the manufacturer’s instructions. The DNA concentration was estimated by standard spectrophotometric methods at 260 nm and 280 nm UV length by BioDrop (UK) and adjusted to a final concentration of 30 ng/μL. The DNA was stored at −20°C for further use.
Simulated High Resolution Melting (HRM) analysis
Eight DNA sequences or fragments were selected based on their availability in all studied regions to test the feasibility of Bar-HRM. To determine the melting profile for each region, simulated HRM analyses were performed using melting curve predictions software called uMeltSM  following the user guide (https://www.dna.utah.edu/umelt/docs/uMeltUserGuide.pdf). Species included in the assay were as follows: 1) Amomum krervanh 2) Boesenbergia rotunda 3) Curcuma longa 4) Curcuma zedoaria 5) Kaempferia parviflora 6) Zingiber montanum 7) Zingiber montanum and 8) Zingiber zerumbet (Table 1).
Real-time PCR amplification and HRM analysis
To acquire the characteristic melting temperature (Tm) that was capable of distinguishing the seven selected species of Zingiberaceae, PCR amplification using real-time PCR and DNA was performed using a Rotor-Gene Q 5plex HRM system (Qiagen, Australia) in a total volume of 20 μl on a plate. The reaction mixture for the real-time PCR and HRM analysis contained 10 μl of MeltDoctor HRM Master Mix (Applied Biosystems, USA), 1.2 μl of 10 mM forward primer, 1.2 μl of 10 mM reverse primer, 1 μl of 30 ng DNA and 6.6 μl of ddH2O. The primer pair was derived from the ITS2 sequence data retrieved from an online database (GenBank) (Forward 5’- CGCCTGCTTGGGCGTCATGGC -3’ and Reverse 5’- GGGCCTCGCCTGACTTGGGGCC -3’). Fluorescence dye was used to monitor both the accumulation of the amplified product and the high-resolution melting process in order to derive the Tm value during PCR.
The real-time PCR reaction conditions were as follows: an initial denaturing step at 95°C for 10 min, followed by 35 cycles of 95°C for 30 s, 57°C for 30 s, 72°C for 30 s, then a final extension step of 72°C for 5 min. The fluorescent data were acquired at the end of each extension step during PCR cycles. For the HRM experiments, fluorescence data were collected every 0.1°C. Rotor-Gene Q Series Software (version 2.3.1) was used to analyse the Tm and melting profile.
Authentication of spice products
To further investigate the feasibility of Bar-HRM in species identification of Zingiberaceae spice products, the protocol was similar to the ‘Real-time PCR amplification and High-resolution Melting (HRM) analysis. Seven commercial products of each species, Zingiber officinale (P1.0-P1.6), Curcuma longa (P2.0-P2.6), Kaempferia parviflora (P3.0-P3.6), Boesenbergia rotunda (P4.0-P4.6), Zingiber montanum (P5.0-P5.6), and four of Alpinia galangal (P6.0-P6.1, P6.3-P6.4), two of Amomum uliginosum (P7.0-P7.1), were included in the analyses (Table 2).
Results and discussion
In silico analyses
The Zingiberaceae sequences from the selected barcode regions were extracted from GenBank, and the variable characters and average %GC content were calculated for all samples using MEGA6. HRM analysis is a method that is able to discriminate between DNA amplicons based on their, length, actual sequence and GC content. Sequence data were available for all selected markers. In total, 879 sequences of matK, 425 of rbcL, 60 of rpoC, 647 of trnH-psbA spacer region, 436 of trnL and 930 of ITS were retrieved, of which 242, 323, 60, 87, 102 and 252 sequences of each barcoding region were considered as being useful for further analysis (Table 3 and S1 Table). Most of the Zingiberaceae sequences found in GenBank belonged to genus Curcuma followed by Alpinia, Amomum, Globba, Zingiber and Boesenbergia, respectively (Fig 1). Five hundred and thirty-six variable sites (69.07%) were found inside the analysed ITS fragment (Table 4). The ITS sequences were found to be the most polymorphic, having the highest nucleotide differences, and the rank of the different DNA barcoding regions used in terms of nucleotide variation was found to be as follows: ITS > matK > trnL > rbcL > trnH-psbA > rpoC. The nucleotide variation within sequences is one of the key factors in HRM analysis as it influences the dissociation energy of the base pairs and result in different Tm values.
The average %GC content of all barcoding region amplicons was calculated, and thus, the predicted possible variation in the melting curves for the different markers was used. The matK region had the lowest average %GC content (29.3%), followed by trnH-psbA, trnL, rpoC, rbcL and ITS, with 29.5%, 33.1%, 38.9%, 44.0% and 55.8%, respectively. Based on the results above, it was estimated that the ITS would be a suitable barcoding marker for the HRM analyses for the target species. ITS has also been used for plant species identification with excellent results [2,5,19,20].
Evaluation of markers for the discrimination of closely related Zingiberaceae species
The DNA-fragments or sequences of selected Zinbearaceae species were analysed using a web-based application called uMeltSM to define the discriminatory power via prediction of fluorescent high-resolution DNA melting curves of PCR products. Eight species (Table 1) were chosen for the analysis. Only four regions (ITS, matK, rbcL and trnH-psbA) were included in the simulated HRM analysis as the results from previous analysis showed that rpoC and trnL may not be suitable to be used to identify Zingiberaceae plants in HRM analysis based on both the percentage of nucleotide variation and the GC content. However, only seven sequences were used in the rbcL assay, as there is no sequence of B. rotunda that could be retrieved from the database (GenBank). Although we expected eight different melting curves from the eight tested species, as seen from Fig 2, none of the selected regions met our expectations. However, ITS and trnH-psbA showed the highest number of distinctive curves (seven), which made them the most promising regions for our study. Thus, we chose only ITS and trnH-psbA for further analyses.
Simulated prediction of fluorescent high-resolution DNA melting curves of PCR products from (A) rbcL (B) matK (C) trnH-psbA and (D) ITS. Sequences or fragments of eight different plant species in Zingiberaceae were included in the analyses of three regions (matK, trnH-psbA and ITS), but only seven plant species of rbcL were included.
HRM analyses to identify Zingiberaceae species
Finding the most suitable marker.
Three primer pairs were developed based on the retrieved sequences from GenBank, trnH-psbA, ITS1 and ITS2. To indicate the most suitable pair, HRM analysis was performed in triplicate on each of the three selected taxa (Zingiber officinale, Curcuma longa and Boesenbergia rotunda) to establish the Tm and melting profiles for each primer set. The DNA of each tested sample was isolated in triplicate and used in the HRM analysis. The melting profiles of all studied amplicons are illustrated in Fig 3A–3F. The mean of the melting temperatures obtained from each primer pair, along with the melting curve shape, was used to measure the discriminatory power of each region. Considering both Tm values and melting curves, all three primer pairs, trnH-psbA (Fig 3A and 3B), ITS1 (Fig 3C and 3D) and ITS2 (Fig 3E and 3F), could clearly differentiate the three tested plants species. The similar meting profiles of each tested species were obtained, thus indicating the feasibility of the method. Notably, both trnH-psbA and ITS are non-coding regions that normally contain high nucleotide variables. The trnH-psbA intergenic spacer is located in the chloroplast genome. This spacer has a high level of variation in the nucleotide sequences, and it was suggested as the barcode for plants by several research groups [21,22]. ITS, or internal transcribed spacer, is the non-coding region found in ribosomal DNA (18S-5.8S-26S rDNA) in the nuclear genome, which can be divided into ITS1 and ITS2. ITS1 is located between 18S rDNA and 5.8S rDNA, and ITS2 is located between 5.8S rDNA and 26S rDNA. Recently, ITS has been widely taken to be used in phylogenetic analysis. Moreover, a high level of nucleotide variation could be found in the ITS region, and it is also recommended for species identification in various plant groups [2,5,23].
The melting profiles of amplicons generated from trnH-psbA (A-B), ITS1 (C-D) and ITS2 (E-F). The HRM analyses were performed in triplicate on each of the three taxa included Zingiber officinale, Curcuma longa and Boesenbergia rotunda. Fresh samples are represented with a thin line and the dried samples with a dotted line.
Although the three tested primer pairs were efficient enough to differentiate the three selected species, the trnH-psbA region was reported to have high indels, and thus, there was inconstancy in the lengths of nucleotide sequences in the area of trnH-psbA in each group of plants (between 200 and over 1,000 base pairs) . As amplicon length is one of major factors to be considered for HRM analysis, we did not choose trnH-psbA for the next analysis. When considering the differences in Tm values, melting profile and nucleotides variation (data not shown) of both ITS1 and ITS2, we only chose ITS2 for further experiments.
ITS2 and identification of seven popular Zingiberaceae spices.
We used ITS2 primers in the HRM analysis in order to confirm its ability to differentiate the seven selected Zingiberaceae plants before using it to identify the species origin in spice products. The melting profiles of ITS2 amplicons from all tested species are shown in Fig 4A. The ITS2 primers produced unique fluorescence melting curves for each Zingiberaceae species. Similar melting curves were achieved from the same species regardless of whether the DNA template was extracted from fresh or dried tissues (Fig 4A). This is an important finding in view of the applicability of the analysis. Thus, there would be no problem in using the approach for identification of species in spice powders or to verify spice products sold in the markets.
Melting profiles of ITS2 amplicons in HRM analyses, (A) of seven tested Zingiberaceae species. Fresh samples are represented with a thin line, and the dried samples with a dotted line. Average Tm values from three replicates of each plant species were presented. The fluorescent melting curves from ITS2 amplicons of spice products; (B) Zingiber officinale, (C) Curcuma longa, (D) Kaempferia parviflora, (E) Boesenbergia rotunda, (F) Zingiber montanum, (G) Alpinia galangal, and (H) Amomum uliginosum in an authentication assay. The coloured thin line indicates the reference sample of each plant species. The dotted lines with the colour as same as the reference samples indicate products containing the plant species as labelled. The grey lines indicate products that did not contain the plant species as labelled.
The ITS2 primer pair was then used in HRM analysis for authentication to see whether they were consistent with the species indicated by the sellers. The tested spice products were purchased from various stores or producers in northern Thailand. HRM analysis was performed in triplicate on each of the tested products to establish the melting profiles. The shapes of the melting curves were analysed using a Rotor-Gene Q Series Software (version 2.3.1) to indicate plant species contained in the products. The melting profiles of all product amplicons are illustrated in Fig 4B–4H. HRM can detect differences among samples of as little as one base pair, however, in some cases, intra-specific and geographical variations within species exist in some plant groups. An arbitrary confidence interval cut-off of 90% was then chosen based on previous molecular studies. The results of the analyses shown in Table 5 reveal that more than half of the tested samples (24 of 41 products) may be contaminated or substituted with other species. Notably, only one of each C. longa and Z. montanum products (P2.0 and P5.0) presented similar melting curves to the reference species and thus implied that the products contained the indicated species. In addition, only one store (of seven) in this study, sold all tested spices that were identified as the same species indicated by the seller or on the labels (P1.0, P2.0, P3.0, P4.0, P5.0 P6.0 and P7.0).
Spices were the main focus of our study as they were not only used for medicinal purposes but also serve as aromatherapy products, natural pest repellents, skin soothers, and even weapons (pepper spray). Several studies have shown that substitution of plant species occurs in herbal medicines. This could pose a serious problem in quality control in production, especially for those with high consumption and economic value. Only 41.46% of the tested spices here were found to be consistent with the species indicated by the sellers or labels. Whether intentional or not, substitution, contamination and adulteration of species should not happen. As most of the spices are sold in powdered form, it is hard to tell whether a product is actually the indicated species based on visual inspection. In addition, the Bar-HRM was previously reported to be able to discriminate an herbal mixture at various concentrations (1%–50%) . Thus, the Bar-HRM has great potential for the identification of mixed spices.
Several scientific research reports have noted that spices have health benefits, such as boosting the immune system, and they may help protect against certain chronic conditions, such as cancer, diabetes, and heart disease. Many traditional remedies have hinged on the use of spices as well, so they have always been in high demand. Before selling, most spices are powdered into small pieces, so it is difficult to identify the species by using morphological approaches. For detecting substitution and adulteration in spice products using molecular techniques, the Bar-HRM, has a number of advantages: (1) the HRM analysis method is highly sensitive for detecting a 1%–0.1% presence of an adulterated sample; (2) it is a high throughput technique that is capable of analysing multiple samples at the same time; and (3) no post-PCR processes are needed, and thus, cross-contamination can be avoided.
We thank The Botanical Garden Organization, Ministry of Natural Resources and Environment, Thailand) and Medicinal Plant Garden, Faculty of Pharmacy, Chiang Mai University) for providing, identifying and collecting the samples.
- 1. Larsen K (2002) The identity of Boesenbergia acuminata (Zingiberaceae). Nordic Journal of Botany 22: 39–39.
- 2. Chen SL, Yao H, Han JP, Liu C, Song JY, Shi L, et al. (2010) Validation of the ITS2 Region as a Novel DNA Barcode for Identifying Medicinal Plant Species. Plos One 5: e8613. pmid:20062805
- 3. Kool A, de Boer HJ, Kruger A, Rydberg A, Abbad A, Björk L, et al. (2012) Molecular Identification of Commercialized Medicinal Plants in Southern Morocco. Plos One 7.
- 4. Newmaster SG, Grguric M, Shanmughanandhan D, Ramalingam S, Ragupathy S (2013) DNA barcoding detects contamination and substitution in North American herbal products. Bmc Medicine 11.
- 5. China Plant BOL Group, Li DZ, Gao LM, Li HT, Wang H, Ge XJ, et al. (2011) Comparative analysis of a large dataset indicates that internal transcribed spacer (ITS) should be incorporated into the core barcode for seed plants. Proceedings of the National Academy of Sciences of the United States of America 108: 19641–19646. pmid:22100737
- 6. Techen N, Parveen I, Pan ZQ, Khan IA (2014) DNA barcoding of medicinal plant material for identification. Current Opinion in Biotechnology 25: 103–110. pmid:24484887
- 7. Barcaccia G, Volpato M, Gentili R, Abeli T, Galla G, Orsenigo S, et al. (2016) Genetic identity of common buckwheat (Fagopyrum esculentum Moench) landraces locally cultivated in the Alps. Genetic Resources and Crop Evolution 63: 639–651.
- 8. Sarkinen T, Staats M, Richardson JE, Cowan RS, Bakker FT (2012) How to Open the Treasure Chest? Optimising DNA Extraction from Herbarium Specimens. Plos One 7.
- 9. Piredda R, Simeone MC, Attimonelli M, Bellarosa R, Schirone B (2011) Prospects of barcoding the Italian wild dendroflora: oaks reveal severe limitations to tracking species identity. Molecular Ecology Resources 11: 72–83. pmid:21429102
- 10. Sass C, Little DP, Stevenson DW, Specht CD (2007) DNA Barcoding in the Cycadales: Testing the Potential of Proposed Barcoding Markers for Species Identification of Cycads. Plos One 2.
- 11. Stoeckle MY, Gamble CC, Kirpekar R, Young G, Ahmed S, Little DP (2011) Commercial Teas Highlight Plant DNA Barcode Identification Successes and Obstacles. Scientific Reports 1.
- 12. Ganopoulos I, Madesis P, Darzentas N, Argiriou A, Tsaftaris A (2012) Barcode High Resolution Melting (Bar-HRM) analysis for detection and quantification of PDO “Fava Santorinis” (Lathyrus clymenum) adulterants. Food Chem 133.
- 13. Madesis P, Ganopoulos I, Anagnostis A, Tsaftaris A (2012) The application of Bar-HRM (Barcode DNA-High Resolution Melting) analysis for authenticity testing and quantitative detection of bean crops (Leguminosae) without prior DNA purification. Food Control 25: 576–582.
- 14. Osathanunkul M, Madesis P, de Boer H (2015) Bar-HRM for Authentication of Plant-Based Medicines: Evaluation of Three Medicinal Products Derived from Acanthaceae Species. PLoS ONE 10: e0128476. pmid:26011474
- 15. Singtonat S, Osathanunkul M (2015) Fast and reliable detection of toxic Crotalaria spectabilis Roth. in Thunbergia laurifolia Lindl. herbal products using DNA barcoding coupled with HRM analysis. BMC Complementary and Alternative Medicine 15: 1–8.
- 16. Osathanunkul M, Suwannapoom C, Osathanunkul K, Madesis P, de Boer H (2016) Evaluation of DNA barcoding coupled high resolution melting for discrimination of closely related species in phytopharmaceuticals. Phytomedicine 23: 156–165. pmid:26926177
- 17. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S (2013) MEGA6: Molecular Evolutionary Genetics Analysis Version 6.0. Mol Biol Evol 30.
- 18. Dwight Z, Palais R, Wittwer CT (2011) uMELT: prediction of high-resolution melting curves and dynamic melting profiles of PCR products in a rich web application. Bioinformatics 27: 1019–1020. pmid:21300699
- 19. de Boer HJ, Ouarghidi A, Martin G, Abbad A, Kool A (2014) DNA Barcoding Reveals Limited Accuracy of Identifications Based on Folk Taxonomy. PLoS ONE 9: e84291. pmid:24416210
- 20. Osathanunkul M, Suwannapoom C, Ounjai S, Rora JA, Madesis P, de Boer HJ (2015) Refining DNA Barcoding Coupled High Resolution Melting for Discrimination of 12 Closely Related Croton Species. PLoS ONE 10: e0138888. pmid:26406615
- 21. Kress WJ, Erickson DL (2007) A Two-Locus Global DNA Barcode for Land Plants: The Coding rbcL Gene Complements the Non-Coding trnH-psbASpacer Region. PLoS ONE 2: e508. pmid:17551588
- 22. Kress WJ, Wurdack KJ, Zimmer EA, Weigt LA, Janzen DH (2005) Use of DNA barcodes to identify flowering plants. PNAS 102.
- 23. Gao T, Yao H, Song J, Liu C, Zhu Y, Ma X, et al. (2010) Identification of medicinal plants in the family Fabaceae using a potential DNA barcode ITS2. J Ethnopharmacol 130: 116–121. pmid:20435122