Journal of Chromatography a Flavylium Chromophores As Species Markers for Dragon's Blood Resins from Dracaena and Daemonorops

Journal of Chromatography A, 1209 (2008) 153–161

Contents lists available at ScienceDirect

Journal of Chromatography A

journal homepage: www.elsevier.com/locate/chroma

Flavylium chromophores as species markers for dragon’s blood resins from Dracaena and Daemonorops trees

Micaela M. Sousa a,b , Maria J. Melo a,b,∗ , A. Jorge Parola b , J. Sérgio Seixas de Melo c , Fernando Catarino d , Fernando Pina b, Frances E.M. Cook e, Monique S.J. Simmonds e, João A. Lopes f a Department of Conservation and Restoration, Faculty of Sciences and Technology, New University of Lisbon, 2829-516 Monte da Caparica, Portugal b REQUIMTE, CQFB, Chemistry Department, Faculty of Sciences and Technology, New University of Lisbon, 2829-516 Monte da Caparica, Portugal c Department of Chemistry, University of Coimbra, P3004-535 Coimbra, Portugal d Botanical Garden, University of Lisbon, Lisbon, Portugal e Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, UK f REQUIMTE, Servic¸ o de Química-Física, Faculdade de Farmácia, Universidade do Porto, Rua Aníbal Cunha 164, 4099-030 Porto, Portugal article info abstract

Article history: A simple and rapid liquid chromatographic method with diode-array UV–vis spectrophotometric detec- Received 20 May 2008 tion has been developed for the authentication of dragon’s blood resins from Dracaena and Daemonorops Received in revised form 28 August 2008 trees. Using this method it was discovered that the flavylium chromophores, which contribute to the red Accepted 3 September 2008 colour of these resins, differ among the species and could be used as markers to differentiate among Available online 7 September 2008 species. A study of parameters, such as time of extraction, proportion of MeOH and pH, was undertaken to optimise the extraction of the flavyliums. This method was then used to make extracts from samples Keywords: of dragon’s blood resin obtained from material of known provenance. From the samples analysed 7,6- Dragon’s blood dihydroxy-5-methoxyflavylium (dracorhodin), 7,4-dihydroxy-5-methoxyflavylium (dracoflavylium) and Red dyes Flavylium chromophores 7,4 -dihydroxyflavylium were selected as species markers for Daemonorops spp., Dracaena draco and Dra- Dracaena caena cinnabari, respectively. The chromatograms from these samples were used to build an HPLC-DAD Daemonorops database. The ability to discriminate among species of dragon’s blood using the single marker compounds Species markers was compared with a principal components analysis of the chromatograms in the HPLC-DAD database. The results from the HPLC-DAD method based on the presence of these flavylium markers was unequivocal. The HPLC-DAD method was subsequently applied to 37 samples of dragon blood resins from the historical samples in the Economic Botany Collection, Royal Botanic Gardens, Kew. The method identified anomalies in how samples in this collection had been labelled. It is clear that the method can be used to evaluate the provenance of samples used in different areas of cultural heritage. It also could be used to monitor the trade of endangered species of dragon’s blood and the species being used in complex formulations of traditional Chinese medicine. © 2008 Elsevier B.V. All rights reserved.

1. Introduction the over exploitation of Dracaena, species of Daemonorops, Cro- ton and Pterocarpus were used as substitutes [5]. The red resin Dragon’s blood is a red resin obtained from species of Dra- is collected from natural exudates that appear in injured areas caena (Dracaenaceae) [1,2], Daemonorops (Palmae) [3,4], Croton on the stem and branches of Dracaena spp. or from a brittle (Euphorbiaceae) and Pterocarpus (Fabaceae) [5]. It has been used red layer formed outside the scaly fruits of Daemonorops spp. for centuries for medicinal [6] and artistic purposes [7].Itis [8]. thought that dragon’s blood was originally produced from species Species of Dracaena used as dragon’s blood included D. draco of Dracaena, especially the dragon tree Dracaena draco. But due to (L.) L. from Madeira, Canary, Cape Verde archipelagos [9], Dracaena cinnabari Balf.f. endemic to Socotra [10], Dracaena ombet Kotschy & Peyr. (synonym Dracaena schizantha Baker) from North-East

∗ tropical Africa and western Arabian peninsula [11] and Dracaena Corresponding author at: Department of Conservation and Restoration, Faculty tamaranae Marrero Rodd., R.S. Almeira & M. Gonzales-Martin of Sciences and Technology, New University of Lisbon, Quinta da Torre, 2829-516 Monte da Caparica, Portugal. Fax: +351 212948322. [12] from Canary Islands. Of these sources, the most frequently E-mail address: [email protected] (M.J. Melo). traded samples of dragon’s blood in Europe were obtained from

Table 1 [22,23–27]. In this paper when reference is made to the flavylium Structures for the flavylium compounds responsible for the red colour in dragon’s chromophore in dragon’s blood resin, it is considered to be present blood resins from species of Daemonorops and Dracaena as the red quinoid base [22]. Because compounds with a flavylium chromophore have not been found in the red resins or exudates from the species of Croton and Pterocarpus used as dragon’s blood Dracorhodin [5] they were not studied in this paper. The fact that different flavylium chromophores can be found in samples of Daemonorops [21] and Dracaena draco [23] suggests that these chromophores could be used as markers to differentiate among species. In order to test this hypothesis 47 samples, rep- resentative of three of the most commonly traded dragon’s blood sources in Europe, Dracaena draco, D. cinnabari and Daemonorops Dracorubin draco, were characterised by HPLC-DAD and selected flavylium chromophores were tested as species markers. The results were subsequently applied to 37 samples of dragon’s blood labelled as Daemonorops draco, Daemonorops sp., Dracaena cinnabari, D. draco, D. schizantha and Dracaena sp. from the Economic Botany Collec- tion, Royal Botanic Gardens, Kew (EBC). The EBC contains perhaps the largest and most reliably identified assemblage of dragon’s Dracoflavylium blood resins dating from the 19th century, that were donated by Sir Isaac Bailey Balfour, the Pharmaceutical Society of Great Britain and others [8]. This is the first report of flavylium compounds being used as markers to identify the species origin of dragon’s blood resins. 7,4 -Dihydroxy-flavylium

2. Experimental The chemical structures correspond to the quinoid bases. 2.1. Extraction methods

Samples of commercial dragon’s blood resins (Zecchi-Colori e D. cinnabari and D. draco [8,5]. Because of over exploitation both Belli Arti, Florence, Italy) and Dracaena draco collected in Madeira, these species are currently regarded as Vulnerable in the IUCN Portugal were extracted using different extraction parameters. The Red List of Threatened Species [13]. Of the species of Daemonorops parameters studied were: (i) extraction time with 2, 5, 10, 60 and [14] that are sources of dragon’s blood, Daemonorops draco (Willd.) more than 60 min; (ii) solvent polarity using water:methanol solu- Blume (synonym Daemonorops propinqua Becc.) is now grown com- tions, with MeOH in the following percentages (v/v) 100, 75, 50 and mercially for the production of dragon’s blood [5] and is distributed 0%; (iii) pH with 1, 6–7 and >7. After the optimization of a selected from Thailand, to Sumatra and Borneo [15]. Dragon’s blood from parameter, the best result obtained for that parameter was applied Daemonorops draco is imported into China from South-East Asia to the samples. for use in traditional Chinese medicine [16–18], but sometimes Each sample weighed circa 0.2 mg. The samples were placed local species are used including Dracaena cochinchinensis (Lour.) in 1.5 mL Eppendorf tubes (Nirco, Spain) and 400 ␮L of extrac- S.C. Chen (China, S.W. Guangxi to S. Yunnan, to Indochina), Dracaena tion solution was added; the tubes were capped and kept at 25 ◦C, angustifolia Roxb (tropical and subtropical Asia to North Australia mechanical stirring, when necessary. Finally the samples were including Taiwan, Guangdong and Yunnan) and Dracaena cambodi- filtered with 0.45 ␮m Acrodisc syringe filters (Macherey-Nagel, ana Pierre ex Gagnep. (South Hainan to Indochina) [19]. Species Düren, Germany) and analysed by HPLC-DAD. Once the optimal of Croton and Pterocarpus were not usually used historically as extraction method had been developed then it was applied to the dragon’s blood in Europe and Asia, although the use of dragon’s experimental samples. blood from species of Croton has become common in the Americas [5] and could be entering the trade in Europe. Compounds in dragon’s blood resins that could be associ- 2.2. Resin samples ated with the red colour were studied by Brockmann and Junge [20], who attributed the colour to a flavylium, dracorhodin The initial tests to evaluate the HPLC method were undertaken (Table 1), which could be considered as belonging to the group on (i) 33 samples of Dracaena draco obtained from the Jardim of compounds known as anthocyanins. Dracorhodin (7-hydroxy-5- Botânico da Ajuda (Lisbon, Portugal); Jardim Botânico de Lisboa methoxy-6-methylflavylium) was isolated and characterised from (Lisbon, Portugal); Jardim Botânico da Madeira (Madeira, Portu- a commercial source of powdered dragon’s blood resin that was gal); Núcleo de dragoeiros das Neves (Madeira, Portugal); other probably obtained from a species of Daemonorops [21]. More places/gardens in Lisbon and Madeira (Portugal) and from Cape recently another flavylium, called dracoflavylium (7,4-dihydroxy- Verde (Santo Antão, Cape Verde). (ii) Twelve samples of Dracaena 5-methoxyflavylium) (Table 1) was isolated and characterized from cinnabari of which 11 from Socotra were made available by Jindrich D. draco [22]. Pavlis and one was purchased from Kremer (Aichstetten, Germany). The red colour of dragon’s blood resin obtained from species (iii) Two commercial samples of Daemonorops draco were analysed, of Dracaena and Daemonorops is associated with the red quinoid one from Zecchi (Florence, Italy; source Sumatra and Borneo) and bases of the respective yellow flavylium cations (Table 1) [22]. the other from Healing Waters & Sacred Spaces (Portland, USA; In solution, both forms are connected, and can undergo multiple source Indonesia). The resin samples of Dracaena draco collected structural transformations, in what can be described as a multistate in Portugal were obtained directly from botanically verified trees system, reversibly interconverted by external stimuli, such as pH of different ages (from 10 to circa 200/350 years old), from three M.M. Sousa et al. / J. Chromatogr. A 1209 (2008) 153–161 155 different injured areas of the stem and branches. When possible, the basis of the principal components retaining the major part of the samples were collected in different seasons of the year (summer original chromatogram data variance. Since principal components and winter) and the results were consistent. The samples sent by represent the original chromatograms in a smaller dimension, Pavlis were collected from trees older than 100 years. space scatter plots can be used to visualize the original data. PCA A further 37 samples were analysed from the EBC. These samples calculations were carried out using Matlab version 6.5 release 13 were labelled as Daemonorops draco (5), Daemonorops propinqua (a (MathWorks, Natick, MA). The algorithm for PCA was written using synonym of Daemonorops draco) (3), Daemonorops sp. (1), Dracaena the method described in Naes et al. [29]. It is based on the singu- cinnabari (15), Dracaena draco (7), Dracaena schizantha (a synonym lar value decomposition of the chromatographic data covariance of D. ombet) (1) and Dracaena sp. (5). A sample of D. ombet (voucher matrix. Each row in the chromatographic data matrix corresponds number 1979-4242, labelled as D. schizantha sourced from Ethiopia) to a chromatogram of a dragon’s blood resin sample (signal inten- was also taken from the living collection of plants growing at the sity over time). Model scores and loadings were obtained from the Royal Botanic Gardens, Kew. covariance matrix eigenvectors [29]. Similarity between the dragon’s blood resin samples was 2.3. Extraction of the dragon’s blood dye chromophores, isolation assessed with the chromatogram data. A preliminary analysis was and identification of flavylium markers made for samples belonging to the same species and it was found that retention times were consistent (no retention time shifts were 2.3.1. Extraction observed). Therefore, no retention time correction was adopted The red colorants of the resin samples were extracted with as a pre-processing step. The PCA models were estimated using 100% methanol (Panreac, Barcelona, Spain) acidified with perchlo- the chromatographic data (signal intensity) obtained from 15.3 to ric acid (Riedel-de Haën, Seelze, Germany) in water (Millipore 28.9 min (retention time) since all peaks were found to be within Simplicity Simpak 2, R = 18.2 M cm, USA), pH 1, at 25 ◦C for circa this region. For each chromatogram, 816 points were available for 5 min. The samples were filtered with 0.45 ␮m Acrodisc syringe fil- the selected retention time region. Prior to PCA modelling, all chro- ters (Macherey-Nagel, Germany) and analysed by HPLC, with DAD matograms were pre-processed using the standard normal variate detector (Thermofinnigan, Surveyor PDA 5), using a RP-18 Nucle- method and subjected to mean centering. Given the multivari- osil column (Macherey-Nagel) with 5 ␮m particle size column ate nature of the resin samples’ chromatograms, multivariate data (250 mm × 4.6 mm). The column was kept at controlled temper- analysis was required in order to analyse samples. PCA was selected ature (35 ◦C). The samples were injected onto the column via a to perform a similarity analysis [29]. Rheodyne injector with a 25 ␮L loop. A solvent gradient of A-pure methanol and B-0.3% (v/v) aqueous perchloric acid was used at a 3. Results and discussion flow rate of 1.7 mL/min; 0–2 min 7A:93B isocratic, 8 min 15A:85B linear, 25 min 75A:25B linear, 27 min 80A:20B linear, 29–40 min 3.1. Extraction methods 100A isocratic [18]. All the resin samples were injected at least three times by HPLC-DAD, with exception of a few small samples from the Extraction with 100% water did not remove any red chro- EBC, when less than 0.2 mg of resin was used for one HPLC-DAD mophores and even when 50% MeOH was used, only slightly yellow analysis. or colourless compounds were extracted. However, when the proportion of MeOH was greater than 50% the flavylium chromophores 2.3.2. NMR and mass characterization were detected. In order to obtain a high percentage of the flavylium The identification of the isolated flavylium markers was made chromophores it was important to keep the pH around 1 during on the basis of mass spectrometry and 1H NMR, although extraction. At pH 1 the main species present in the equilibrium was the complete structure confirmation of the 7,4-dihydroxy-5- the flavylium cation, by increasing the pH, flavylium cation was methoxyflavylium and the 7,4-dihydroxyflavylium required their converted partially to other coloured forms as the quinoid base synthesis, and were prepared according to the published proce- and chalcones [22–27], and several chromatographic peaks were dures [22,28]. The high-resolution (HR) mass spectra were obtained eluted. As these peaks were eluted in a range of less than 2 min, by laser desorption/ionization (LDI) with a Finnigan FT/MS 2001- the identification and separation of the flavylium chromophores DT Fourier transform ion cyclotron resonance mass spectrometry became difficult. The time taken to extract the material also influ- (FTICR-MS) system (Finnigan, USA), equipped with a 3 Tesla super- enced the profile of flavyliums extracted, with low levels being conducting magnet and coupled to a Spectra-Physics Quanta-Ray detected after longer extraction times. Good results were obtained GCR-11 Nd:YAG laser operated at the fundamental wavelength with circa 5 min. The most efficient method for the extraction of (1064 nm). the dragon’s blood flavylium chromophores was a 5 min extrac- The NMR spectra in CD3OD (Panreac) at 298.0 K were obtained tion with 100% MeOH, acidified at pH 1. With these extraction either on a Bruker AMX400 (Bruker, USA) operating at 400.13 MHz parameters, a 0.01 mg resin samples was enough to extract the (1H) and 100 MHz (13 C) or on a Bruker Avance 600, operating at flavyliums. Nevertheless, at least 0.2 mg of sample was used for 600.13 Hz (1H) and 150.91 Hz (13 C). Proton assignment was done on most samples to allow a better quantification of the dragon’s blood the basis of chemical shifts and COSY spectra. Carbon assignments chromophores. were made on the basis of chemical shifts, HSQC or HMQC, and HMBC NMR spectra. 3.2. Analysis of samples

2.4. Statistical analysis of data The results of analysing the verified samples of Dracaena draco, D. cinnabari and Daemonorops draco were used to build-up the 2.4.1. Principal components analysis (PCA) HPLC-DAD database. The data show that the three flavylium com- Given the multivariate nature of the resin samples’ chro- pounds, dracoflavylium, 7,4-dihydroxyflavylium and dracorhodin matograms, multivariate data analysis was required in order to elute at different times and vary in their UV spectra (Table 2). The analyse samples. Principal components analysis (PCA) was selected data also show that the compounds vary in their distribution among to perform a similarity analysis [29]. PCA results were analysed on the three species of dragon’s blood. 156 M.M. Sousa et al. / J. Chromatogr. A 1209 (2008) 153–161

Table 2 HPLC chromatogram proﬁles at 462 nm for Dracaena cinnabari, Dracaena draco, Daemonorops draco; retention times, absorption maxima and UV–vis absorption spectra for the ﬂavylium markers

Species Chromatogram Flavylium marker tr, Flavylium marker UV–vis spectrum max

Dracaena cinnabaria (1) 7,4- dihydroxyﬂavylium

tr = 18.03 ± 0.15 min, max = 462 nm

Dracaena dracob (2) Dracoﬂavylium 7,4 -dihydroxy-5- methoxyﬂavylium

tr = 20.51 ± 0.12 min, max = 476 nm

Daemonorops dracoc (3) Dracorhodin 7,6-dihydroxy-5- methoxyﬂavylium

tr =21.76± 0.07 min, max = 438 nm

a Collected in Socotra Island; older than 100 years. b Collected in the Natural Park of Madeira; age: circa 200 years old. c Acquired from Zecchi, Colori e Belle Arti, Florence, Italy; age unknown.

3.2.1. Dracaena draco (dracoflavylium) total amount of the red colorants. The colour of this resin is due to a It was observed that dracoflavylium was always present in the complex mixture of red compounds, 7,4-dihydroxyflavylium being 33 resin samples of dragon’s blood from D. draco. However, the con- one of the major chromophores (Table 2). centration varied among samples. In samples from trees circa 200 years old, dracoflavylium was the major red compound (circa 32%; 3.2.3. Daemonorops draco (dracorhodin) Fig. 1), whereas in other samples this flavylium was present as a In the two commercial sources of Daemonorops draco resin, dra- minor product of the total amount of the red colourants of resin, corhodin was the major red compound. This supports data already ranging from 1 to 10% (relative area). published that dracorhodin occurs in Daemonorops spp. [20,30–34].

3.2.2. Dracaena cinnabari (7,4-dihydroxyflavylium) 3.2.4. The flavylium markers In the 12 samples of resin from D. cinnabari, the quantity of 7,4- The data show the extracts of Dracaena draco, D. cinnabari dihydroxyflavylium varied from 5 to 15% of the relative area for the and Daemonorops draco each contain a characteristic flavylium: M.M. Sousa et al. / J. Chromatogr. A 1209 (2008) 153–161 157

Table 3 Samples labelled as Daemonorops from EBC, with the percentage of the ﬂavylium markers obtained by HPLC-DAD and respective species attribution

EBCa EBC classiﬁcation Date of donation Provenance, donor Observations % ﬂavylium markerb

35487 Daemonorops draco c India, India Museum Mixture of resin, bark and powder Dracaena cinnabari (6% 7,4 -dihydroxyﬂavylium) 35489 Daemonorops draco 1851 Singapore, A.S. Hill & Son Mostly resin. Labelled as “Calamus Daemonorops draco (65% dracorhodin) draco, in lumps, colour of powder, brick red. Contains about 12 per cent of insoluble matter”. Paler and duller than other samples 35490 Daemonorops draco 1851 India, Calcutta, Royal Mostly resin. Labelled as “Reed Daemonorops draco (55% dracorhodin) Commonwealth dragon’s blood” Exhibition

35495 Daemonorops draco cd, British Museum Mixture of resin and powder. The Daemonorops draco (65% dracorhodin) (Natural History) sample appears to be small pieces from a reed resin

35499 Daemonorops draco c Sumatra, e Resin attached to fruit scales Daemonorops sp. (?) (65% unknown

compound; tr = 21.03 min, max = 453 nm) 35500 Daemonorops propinqua 1896 Sumatra, e Resin attached to fruit scales Daemonorops draco (65% dracorhodin) 35526 Daemonorops propinqua 1890 d, A Hill & Son Mixture of resin, powder and Daemonorops draco (65% dracorhodin) contaminants. It looks paler and much less resinous than some other samples of Lump dragon’s blood 35527 Daemonorops propinqua cd, Savory & Co. Mixture of resin and powder. Lump Dracaena cinnabari (16% dragons’s blood 7,4 -dihydroxyﬂavylium) 35505 Daemonorops sp. 1851 Sumatra, Royal Mostly resin. Similar to Lump dragon’s Daemonorops draco (55% dracorhodin) Commonwealth blood Exhibition

a Voucher number for samples from the Economic Botany Collection, Royal Botanic Gardens, Kew. b The relative peak areas were calculated with the chromatographic program ChromQuest 4.1 at the maximum wavelength absorption for each flavylium marker selected: 438 nm for dracorhodin (Daemonorops sp.), 462 nm for 7,4-dihydroxyflavylium (Dracaena cinnabari and D. schizantha) and 476 nm for 7,4-dihydroxy-5-methoxyflavylium (Dracaena draco). c Donation date unrecorded, likely 19th century specimen. d Provenance unknown. e Donor unrecorded. dracoflavylium, 7,4-dihydroxyflavylium or dracorhodin, respec- in the PCA figures can be analysed according to the correspond- tively. As depicted in Table 1, these compounds have the ing loadings (each sample PCA score is the inner product between 2-phenyl-1-benzopyrylium core in common, but a different sub- the sample chromatogram and the loading corresponding to a stitution pattern, consequently each exhibit characteristic UV–vis given principal component). It was observed that the loading for spectra and retention times (Table 2). This in turn, enables a the first score, PC 1, contains strong positive peaks at 20.5 (dra- straightforward identification of these flavylium chromophores by coflavylium), 27.4 and 27.9 min and strong negative peaks at 18.0 HPLC-DAD, and the identification of which of the three species (7,4-dihydroxyflavylium), 18.7 and 23.3 min. The former corre- could be the source of the dragon’s blood. spond to Dracaena draco elution peaks while the latter correspond The species discrimination was further tested by PCA applied to to D. cinnabari peaks. Therefore, the first PCA component, PC 1, is the chromatograms acquired at the wavelengths for the detection able to discriminate between these two species of Dracaena (pos- of red compounds (462 nm). The principal components represented itive scores for D. draco and negative scores for D. cinnabari in the first component axis). The third score, PC 3, exhibits two strong positive peaks at 21.1 and 21.7 min (dracorhodin). These peaks correspond to peaks observed in chromatograms from Daemonorops draco. No other relevant peaks were observed in the third loading, which means that this component captures only information from the Daemonorops draco samples (strong positive scores in the third score). Thus, it is possible to acquire information on the dragon’s blood source using the single flavylium markers as well as PCA applied to all red chromophores present (Table 2, Fig. 1).

3.3. Samples of dragon’s blood from the Economic Botany Collection (EBC), Royal Botanic Gardens, Kew

Extracts of the EBC samples were then analysed and compared with the HPLC-DAD data library, using the ﬂavylium markers and Fig. 1. PCA scores for reference samples of Dracaena draco (squares), Dracaena the PCA of the chromatograms acquired at 462 nm. The data are cinnabari (circles) and Daemonorops draco (stars) representing 58.4% of total data variance. The model was obtained from mean centered HPLC chromatograms also, when possible, compared to the results of previous analysis of acquired at 462 nm between 15.3 and 28.9 min. the EBC samples by Raman spectroscopy [35,36]. 158 M.M. Sousa et al. / J. Chromatogr. A 1209 (2008) 153–161

3.3.1. Daemonorops draco (synonym D. propinqua) and rently identified as being from Daemonorops may not be accurately Daemonorops sp. named. Some of these naming errors may have arisen in the cura- Of the nine EBC samples labelled as Daemonorops draco, Dae- tion of the samples when Latin scientific names have been assumed monorops propinqua or Daemonorops sp. the HPLC-DAD analysis from the vernacular name. These errors are more likely to occur shows that six samples contained dracorhodin. The relative level of from samples, such as EBC 35497, that was sourced from India as dracorhodin in these six samples (EBC 35489, 35490, 35495, 35500, both Daemonorops draco, and Dracaena cinnabari were traded there 35526 and 35505) was between 55 and 65% of the total area of red as dragon’s blood. Analysis of further samples within the EBC using chromophores (Table 3). This is similar to that found in the com- the HPLC-DAD technique would be profitable. mercial resins sourced from Daemonorops. It is worth mentioning that two other EBC samples labelled as D. propinqua (EBC 35500, 3.3.2. Dracaena cinnabari, D. ombet (synonym D. schizantha) 35526) contained dracorhodin as the major compound and PCA and Dracaena sp. analysis of the chromatograms obtained for these two samples were All 15 samples from the EBC labelled as Dracaena cinnabari very similar and similar to those of the commercial samples of Dae- contained 7,4-dihydroxyflavylium (5–20%) the flavylium that is monorops. A recent publication [37] announced D. propinqua to be a marker for this species (Table 4). These samples were mostly synonymous with Daemonorops draco (Willd.) Blume, the now cur- acquired directly or indirectly, via market products, from Socotra rently accepted botanical name. In contrast, two samples labelled as where the species is endemic. The EBC 36816 sample labelled as Daemonorops draco (EBC 35487) and Daemonorops propinqua (EBC D. schizantha from Zanzibar contained 7,4-dihydroxyflavylium in 35527) contained 7,4-dihydroxyflavylium and no dracorhodin. This 15% of the total of the red chromophores. It also had a similar chro- suggests that these samples were from resins of Dracaena cinnabari matographic elution profile to resins of D. cinnabari (Fig. 2A and B). and had been incorrectly labelled. Another sample labelled as Consequently, it is possible that this resin comes from D. cinnabari. Daemonorops draco (EBC 35499) contained a compound with an The original label for specimen EBC 36816 only referred to dragon’s unknown red chromophore (Table 3). The botanical source of this blood and not to the botanical name which was attributed to the resin is unclear. specimen some time later. The sample was donated to Kew in These results indicate that the botanical names attributed to 1871 when there was an established trade route for the commer- some of the samples of dragon’s blood resins in the EBC at Kew cur- cial supply of D. cinnabari between Socotra and Zanzibar. Another

Table 4 Samples labelled as Dracaena cinnabari, Dracaena schizantha and Dracaena sp. from EBC, with the percentage of the ﬂavylium markers obtained by HPLC-DAD and respective species attribution

EBCa EBC classiﬁ- Date of Provenance Observations % ﬂavylium markerb cation donation

36489 D. cinnabari de, labelled Socotra Dragon’s blood Mixture of resin, bark and powder D. cinnabari (7% 7,4-DHAc) from Allen & Co. 36542 D. cinnabari 1881 e, labelled Socotra Dragon’s blood, Mixture of resin, bark and powder D. cinnabari (8% 7,4-DHA) donated by IB Balfour 36543 D. cinnabari 1875 e, labelled Socotra Dragon’s blood, Resin D. cinnabari (5% 7,4-DHA) donated by Dr. Vaughan 36545 D. cinnabari 1899 e, labelled Zanzibar Dragon’s blood, Resin. Labelled as “Extra ﬁne Zanzibar leas” D. cinnabari (5% 7,4-DHA) purchased by Mather at Ripley Roberts Drug sale, 3 Mincing lane 36557 D. cinnabari 1899 Zanzibar, labelled Socotra Dragon’s Heterogeneous resin D. cinnabari (15% 7,4-DHA) blood, purchased by Mather 36563 D. cinnabari de, labelled Socotra dragon’s blood Powder D. cinnabari (15% 7,4-DHA) 36580 D. cinnabari 1899 e, Kurachi, labelled Kurrachi dragon’s Resin. Labelled as “Fine marbles of dragon’s blood” D. cinnabari (5% 7,4-DHA) blood, purchased by Mather 36599 D. cinnabari 1881 Socotra Mixture of pigment and resin. Labelled as “dam el D. cinnabari (17% 7,4-DHA) akhuwen” 36611 D. cinnabari 1880 e, labelled as Socotra dragon’s blood Tears of resin D. cinnabari (15% 7,4-DHA) presented by JB Balfour 36622 D. cinnabari de, labelled as Socotra dragon’s blood Tears of resin D. cinnabari (5% 7,4-DHA) 36773 D. cinnabari 1880 Socotra. Donated by IB Balfour Tears of resin. Labelled as “Edah Amsellah” D. cinnabari (22% 7,4-DHA) 36808 D. cinnabari d Socotra. Donated by IB Balfour Resin wrapped in bark D. cinnabari (20% 7,4-DHA) 36809 D. cinnabari 1880 Socotra. Donated by IB Balfour Mixture of resin, pigment and bark. Labelled as D. cinnabari (19% 7,4-DHA) “Edah-Muck-Dehar” “prepared from the boiled dust” 36823 D. cinnabari 1881 Socotra. Donated by IB Balfour Powder. Labelled as “Edah Dukkah” “consisting of D. cinnabari (20% 7,4-DHA) small fragments broken tears of Dragons blood” 79745 D. cinnabari d Socotra Tears of resin D. cinnabari (5% 7,4-DHA) 36816 D. schizantha 1871 Zanzibar Resin D. cinnabari?(15%7,4-DHA) 36819 Dracaena sp. d Socotra Mixture of resin, pigment and bark D. cinnabari (5% 7,4-DHA) 36820 Dracaena sp. de Mixture of resin, pigment and bark D. cinnabari (12% 7,4-DHA) 36821 Dracaena sp. 1906 London Drug market, Zanzibar Mixture of resin, pigment and bark D. cinnabari (13% 7,4-DHA) 36822 Dracaena sp. de, donated by East India Company Mixture of resin, pigment and bark D. cinnabari (5% 7,4-DHA) 75793 Dracaena sp. d Socotra Mixture of resin, pigment and bark D. cinnabari (15% 7,4-DHA)

Fig. 2. HPLC profile for (A) EBC 36809, labelled Dracaena cinnabari; (B) EBC 36816, labelled Dracaena schizantha. sample of D. cinnabari (EBC 36557) from Zanzibar would support the trade of this species. As no species of Dracaena grow natu- rally in Zanzibar, the only other likely source of dragon’s blood from N.E. African is D. ombet (synonym for D. schizantha) but this species was not frequently traded [39]. A sample of resin from D. ombet growing at Kew was analysed and the trace of this resin differed from that of sample EBC 36816, which sup- Fig. 3. HPLC profiles of two resin samples labelled as Dracaena draco from the “Great dragon tree” of Tenerife-EBC, K collection. (A) EBC 26421 displaying dracoflavylium ports the conclusion that the name on the label for EBC 36816 is (2); (B) EBC 36825 displaying ellagic acid (E). The chromatogram in (B) suggests this incorrect. sample is not from a Dracaena draco resin, see text for more details. The five EBC samples that had been labelled as Dracaena sp. (EBC 36819, 36820, 36821, 36822, 75793) contained 7,4- dihydroxyflavylium and the PCA analysis of the chromatographic sp. [35]. However, the presence of 7,4-dihydroxyflavylium in the elution profiles of these samples showed that they were similar sample would not support this conclusion. Pearson and Prender- to those of D. cinnabari. This suggests that these samples of resin gast [8] discussed that frequently in the past resins traded out of are all from D. cinnabari. In a previous study using Raman spec- Bombay were assumed to be sourced from Daemonorops draco, but trometry data the sample EBC 36822, a sample labelled Dracaena the East India Company also had connections with Socotra and East sp. from the East India Company, had been tentatively identified as Africa so both species are represented in dragon’s blood sourced being sourced from either Daemonorops sp. or a degraded Croton from India.

Table 5 Samples labelled as Dracaena draco from EBC, with the percentage of the ﬂavylium markers obtained by HPLC-DAD and respective species attribution

EBCa EBC Date of Provenance, donor Observations % ﬂavylium markerb classiﬁcation donation

26397 D. draco 1867 Kew, Palm House Red resin fragments. Dracaena draco (2% dracoflavylium) 26421 D. draco 1871 Tenerife, Canary Is, e Red wood. Labelled as “Celebrated Dragon tree Dracaena draco (33% of Tenerife” dracoflavylium) 36516 D. draco c “Socotra?”, e Mostly red resin. Labelled as “Resin wrapped in Daemonorops sp. (42% dracohodin) leaves” 36653 D. draco c Madeira, e Red resin. Dracaena draco + Dracaena ombet/cinnabari (9% dracoflavylium and 2% of 7,4-dihydroxyflavylium) 36824 D. draco c Lisbon, Botanic Garden Mostly red resin. Labelled as “Resin wrapped in Dracaena draco (5% dracoflavylium) leaves” 36825 D. draco c Tenerife, Canary Is, e Brown resin. Labelled as “Gum-resin exuded Pterocarpus or Croton sp. (?) (37% from the great Dragon tree of Tenerife” ellagic acid) 78811 D. draco 2004 d, Adelaide, Botanic Garden Red resin. Dracaena draco (3% dracoflavylium)

this and a D. draco sample from the Botanical Garden of Lisbon [36].

4. Conclusion

The flavylium chromophores that contribute to the red colour of dragon’s blood resins can be used as markers to differentiate among resins from Daemonorops draco, Dracaena draco and D. cinnabari.AsDracaena draco and D. cinnabari are endangered the HPLC-DAD methods developed in this paper could be used to evaluate whether these species are being substituted for Daemonorops draco the species that is being commercially cultivated to meet the increased interest in dragon’s blood, especially for use in traditional Chinese medicine. Although these markers can differentiate Fig. 4. PCA scores for reference (open symbols) and EBC, K (solid symbols) samples between Dracaena draco and D. cinnabari further research needs of Dracaena draco (squares), D. cinnabari (circles) and Daemonorops draco (stars). to be undertaken to identify the compounds that could be used to The model was obtained only from reference samples mean centered HPLC chro- discriminate between D. cinnabari and D. ombet, as they appear to matograms acquired at 462 nm between 15.3 and 28.9 min. EBC, K samples were share the same marker, 7,4-dihydroxyflavylium, albeit in different projected onto the model for validation purposes. concentrations. Principal component analysis was applied to the chromatogram representing all the red compounds present in the resins analysed and the results were in full agreement with the 3.3.3. D. draco conclusions obtained using a single molecule marker. Of the seven EBC samples labelled as D. draco, four samples (EBC Once the HPLC-DAD method for the flavylium species mark- 26387, 26421, 36824 and 78811) from botanic gardens contained ers had been successfully developed on the resins from accurately dracoflavylium the marker for this species (Table 5). This includes provenanced samples of Dracaena draco and D. cinnabari and the a sample of wood (EBC 26421) donated to Kew in 1871 from the trade samples of Daemonorops it was used to analyse 37 samples “celebrated Dragon tree of Tenerife” that contained a high concen- from the (mainly) 19th century dragon’s blood samples in the EBC, tration (33%) of dracoflavylium (Fig. 3A), similar to the levels of Royal Botanic Garden, Kew. It was possible to confirm the inven- dracoflavylium observed in the samples of resin (Fig. 4) from the toried sources of 25 samples of these resins, identify species for old (circa 200 years old) Dracaena draco trees growing in Portugal 5 samples where previously only genus was known (EBC 36819, that were used to build the HPLC-DAD data library. 36820, 36821, 36822 and 75793) or where the species was only In contrast, dracoflavylium was not detected in another sam- tentatively assigned (EBC 36816), and clarify or discuss incorrect ple (EBC 36825) that is labelled as Dracaena draco and is reported attributions in the inventory of EBC (four samples with incorrect to be “gum resin exuded from the great Dragon tree of Tenerife”. genus, EBC 35487, 35527, 36516), one sample correct to genus but This sample contained hydrolysable polyphenols “tannins” includ- incorrect to species (EBC 35499) and one mixed collection (EBC ing ellagic acid (Fig. 3B) a group of compounds not observed in 36653). One sample could not be attributed to any genus or species the HPLC-DAD of the other verified samples of Dracaena draco. (EBC 36825). These results suggest that other samples in the EBC The occurrence of “tannins” in this specimen of dragon’s blood may benefit from re-examination using these techniques, espe- resin was reported in 1895 by H. Trimble and he concluded that cially items labelled “Lump dragon’s blood” that came from India. the specimen was very similar to Pterocarpus draco L. or Cro- Finally, it is possible to conclude that, the use of single flavylium ton draco Schltdl. [38]. The results from an earlier analysis using compounds as species markers was clearly validated in this work. Raman spectrometry also showed this sample was very different This HPLC-DAD method constitutes a breakthrough in the analysis from other samples of Dracaena draco [35]. This suggests that in of complex samples containing dragon’s blood resin, including aged future the chemical profile of this sample should be compared samples, and it is anticipated that it could be used to monitor the with extracts of dragon’s blood from species of Pterocarpus and trade in endangered species of dragon’s blood, validate the species Croton. being used in traditional Chinese medicinal formulations as well Dracoflavylium was not detected in another sample (EBC as in the field of cultural heritage to establish which species were 36516) labelled as being from Dracaena draco, this sample con- used as dyes. tained a high concentration of dracorhodin (42%), the marker for Daemonorops sp. The authentication of this sample had been previously challenged using Raman spectrometry and Edwards et Acknowledgements al. [35] suggested this sample could be from Daemonorops sp. Interestingly, although the original EBC label is no longer with We are grateful to POCI (POCI/QUI/55672/2004 and the sample the database at Kew records “?Socotra” as the ori- PTDC/EAT/65445/2006), FCT and FEDER for further funding. gin, suggesting that the writing on the label or the document We would like to thank to botanical garden of Lisbon (Portugal), that accompanied the sample was difficult to decipher. Suma- botanical garden of Ajuda (Portugal), to Roberto Jardim, director of tra (the major origin of dragon’s blood from Daemonorops draco) botanical garden of Madeira (Portugal), to Natural Park of Madeira is not unlike Socotra if written illegibly. Thus curatorial errors (Portugal) for the Dracaena draco samples and J. Pavlis for the D. could have contributed to the confusion of the source of this cinnabari samples (Mendel University of Agriculture and Forestry, sample. Czech Republic). We also would like to thank to Dr. Anita Quye Finally sample EBC 36653 recorded as being from Madeira con- (National Museums of Scotland, UK) for dragon’s blood samples tained both 7,4-dihydroxy-5-methoxyflavylium (dracoflavylium) and Ms. H. Chantre (University of Coimbra, Portugal) for the and 7,4-dihydroxyflavylium. This suggests it was a mixture of Cape Verde species. Finally we would like to thank Prof. Joaquim resins from Dracaena draco and D. ombet or D. cinnabari. Edwards Març alo (Instituto Tecnológico Nuclear, Portugal) for help in the et al. detected in their Raman spectra some differences between MS measurements. M.M. Sousa et al. / J. Chromatogr. A 1209 (2008) 153–161 161

References [18] W.J. Gong, Y.H. Cao, Y. Wang, Chromatographia 66 (2007) 767. [19] D. Bensky, S. Clavey, E. Stöger (translators), Chinese Herbal Medicine: Materia [1] G. Mwachala, P.K. Mbugua, Flora of Tropical East Africa, Dracaenaceae, Royal Medica, Eastland Press, Seattle, WA, 2004. Botanic Gardens, Kew, 2007. [20] H. Brockmann, H. Junge, Ber. Dtsch. Chem. Ges. B 76 (1943) 751. [2] J.J. Bos, Agric. Univ. Wageningen Pap. 84 (1984) 1. [21] H. Brockmann, R. Haase, Bernoulli 69 (1936) 1950. [3] N.W. Uhl, J. Dransfield, Genera Palmarum. A Classification of Palms Based on [22] M.J. Melo, M.M. Sousa, A.J. Parola, J.S. Seixas de Melo, F. Catarino, J. Març alo, F. the Work of Harold E. Moore Jr., L.H. Bailey Hortorium and the International Pina, Chem. Eur. J. 13 (2007) 1417. Palm Society, Lawrence, 1987. [23] F. Pina, J. Chem. Soc., Faraday Trans. 94 (1998) 2109. [4] H. Rustiami, Gardens’ Bull. Singapore 54 (2002) 199. [24] F. Pina, M. Maestri, V. Balzani, in: H.S. Nalwa (Ed.), Handbook of Photochemistry [5] D. Gupta, B. Bleakley, R.K. Gupta, J. Ethnopharmacol. 115 (2008) 361. and Photobiology, vol. 3, Supramolecular Photochemistry, American Scientific [6] Y. Zhu, P. Zhang, H. Yu, J. Li, M. Wang, W. Zhao, J. Nat. Prod. 70 (2007) 1570. Publishers, 2003, p. p. 411. [7] Cennino Cennini, Il Libbro dell’Arte, Neri Pozza Editore, Vicenza, 1982. [25] F. Pina, M.J. Melo, A.J. Parola, M. Maestri, V. Balzani, Chem. Eur. J. 4 (1998) 2001. [8] J. Pearson, H.D.V. Prendergast, Econ. Bot. 55 (2001) 474. [26] F. Pina, M. Maestri, V. Balzani, Chem. Commun. (1999) 107. [9] R. Govaerts, S.A. Zona, World Checklist of Asparagaceae, The Board of Trustees [27] R.A. McClelland, S. Gedge, J. Am. Chem. Soc. 102 (1980) 5838. of the Royal Botanic Gardens, Kew; http://www.kew.org/wcsp (accessed April [28] C. Michaelis, R. Wizinger, Helv. Chim. Acta 24 (1951) 1170. 30, 2008). [29] T. Naes, T. Isaksson, T. Fearn, T. Davies, Multivariate Calibration and Classifica- [10] R. Adolt, J. Pavlis, Trees 18 (2004) 43. tion, NIR Publications, Chichester, 2004. [11] M. Thulin, Flora of Somalia, vol. 4, Royal Botanic Gardens, Kew, 1995. [30] A. Robertson, W.B. Whalley, J. Yates, J. Chem. Soc. (1950) 3117. [12] A. Marrero, R.S. Almeida, M. González-Martin, Bot. J. Linn. Soc. 128 (1998) 291. [31] H. Brockmann, R. Haase, Bernoulli 70 (1937) 1733. [13] IUCN. 2007 IUCN Red List of Threatened Species. IUCN: http://www. [32] A. Robertson, W.B. Whalley, J. Chem. Soc. (1950) 1882. iucinredlist.org (accessed 16 September 2008). [33] E.O.P. Agbakwuru, W.B. Whalley, J. Chem. Soc., Perkin I (1976) 1392. [14] H. Rustiami, Phenetic Study on Dragon’s Blood Species of Daemonorops Section [34] G. Cardillo, L. Merlini, G. Nasini, J. Chem. Soc. (1971) 3967. Piptospatha (Family Arecaceae), Thesis (MSc), University of reading, 1999. [35] H.G.M. Edwards, L.F.C. Oliveira, H.D.V. Prendergast, Analyst 129 (2004) 134. [15] R. Govaerts, J. Dransfield, World Checklist of Palms, The Board of [36] H.G.M. Edwards, L.F.C. Oliveira, A. Quye, Spectrochim. Acta A 57 (2001) 2831. Trustees of the Royal Botanic Gardens, Kew. Published on the Internet; [37] H. Rustiami, F.M. Setyowati, K. Kartawinata, J. Trop. Ethnobiol. 1–2 (2004) http://www.kew.org/wcsp (accessed April 30, 2008). 65. [16] Q.-A. Zheng, H.-Z. Li, Y.-J. Zhang, C.-R. Yang, Helv. Chim. Acta 87 (2004) 1167. [38] H. Trimble, Am. J. Pharm. 67 (1895) 516. [17] C. Shen, S. Tsai, S. Wei, S. Wang, V. Shieh, C. Chen, J. Nat. Prod. 21 (2007) 377. [39] J. Vaughan, Pharm. J. 12 (1852–1853) 385.