Physiol Mol Biol Plants (March 2021) 27(3):605–617

https://doi.org/10.1007/s12298-021-00965-9

RESEARCH ARTICLE

The efficacy of machine learning algorithm for raw drug

authentication in Coscinium fenestratum (Gaertn.) Colebr.
employing a DNA barcode database
Remya Unnikrishnan1,4 • M. Sumod2 • R. Jayaraj3 • P. Sujanapal2 • Suma Arun Dev1

Received: 3 January 2021 / Revised: 19 February 2021 / Accepted: 2 March 2021 / Published online: 15 March 2021
Ó Prof. H.S. Srivastava Foundation for Science and Society 2021

Abstract Medicinal plants are a valuable resource for Knowledge Analysis was employed to prove efficacy of the
traditional as well as modern medicine. Consequently huge method. The automated species identification technique,
demand has exerted a heavy strain on the existing natural Artificial Intelligence uses the ability of computers to build
resources. Due to over exploitation and unscientific col- models that can receive the input data and then conduct
lection most of the commercially traded ayurvedic plants statistical analyses which significantly reduces the human
are in the phase of depletion. Adulteration of expensive labour. Concurrently, scientific management, restoration,
raw drugs with inferior taxa has become a common prac- cultivation and conservation measures should be given
tice to meet the annual demand of the ayurvedic industry. utmost priority to reduce the depletion of wild resources as
Although there are several recommended methods for well as to meet the rapidly increasing demand of the herbal
proper identification varying from the traditional taxo- industries.
nomic to organoleptic and physiochemical, it is difficult to
authenticate ayurvedic raw drugs available in extremely Keywords Machine learning algorithm  Artificial
dried, powdered or shredded forms. In this regard, the intelligence  DNA barcoding  Threatened species 
study addresses proper authentication and illicit trade in Medicinal plants
Coscinium fenestratum (Gaertn.) Colebr. using CBOL
recommended standard barcode regions viz. nuclear ribo-
somal–Internally Transcribed Spacer (nrDNA- ITS), mat- Introduction
urase K (matK), ribulose-1,5-bisphosphate carboxylase/
oxygenase large subunit (rbcL), and psbA-trnH spacer Coscinium fenestratum (Gaertn.) Colebr. a dioecious
regions. Further, an integrated analytical approach woody climber of Menispermaceae family is one of the
employing Maximum Likelihood phylogenetic tree and commercially traded medicinal plant species in tropical
Machine Learning Approach, Waikato Environment for forests of India, endemic to Kerala, Tamil Nadu and Kar-
nataka (Ved et al. 2015). This liana has been listed as
threatened by International Union for Conservation of
Nature (IUCN) and Convention on International Trade in
& Suma Arun Dev Endangered Species of Wild Fauna and Flora (CITES)
sumadev@kfri.res.in
(Sarvalingam and Rajendran 2016). In India, distribution of
1
Forest Genetics and Biotechnology Division, Kerala Forest the species is restricted to the Western Ghats, mostly in wet
Research Institute, Peechi, Thrissur, Kerala 680653, India evergreen, moist evergreen, semi evergreen and semi
2
Sustainable Forest Management Division, Kerala Forest deciduous forests within an altitudinal range of 500–750
Research Institute, Peechi, Thrissur, Kerala 680653, India mean sea level (msl) (Mohanan and Sivadasan 2002). The
3
Forest Ecology and Biodiversity Division, Kerala Forest stems and roots of the plants are extensively used in diverse
Research Institute, Peechi, Thrissur, Kerala 680653, India systems of medicine, such as Ayurveda, Folk, Tibetan,
4
Cochin University of Science and Technology, Kochi, Siddha, and Thai because of the presence of major alkaloid,
Kerala, India berberine (Tran et al. 2003; Thriveni et al. 2015). C.

123
606 Physiol Mol Biol Plants (March 2021) 27(3):605–617

fenestratum is the major ingredient of 62 Ayurvedic for- organoleptic and physiochemical to authenticate raw drugs
mulations and credited with anti-diabetic, anti-inflamma- in the ayurvedic formulations that are manufactured, dis-
tion, anti-proliferative, anti-hepatotoxic, anti-plasmodial, tributed and sold by licensed manufacturers in India. Tra-
CNS depressant and analgesic properties (Ueda et al. 2002; ditional plant identification such as microscopic and
Tran et al. 2003; Punitha et al. 2005; Udayan et al. 2005; macroscopic methods are cost effective and capable of
Rai et al. 2012). Recent studies have shown that the plant is dealing with the multi-herbal formulations and impurities;
also exploited as a source of berberine which acts effec- however, it has low efficiency in highly processed samples
tively against chronic respiratory inflammatory diseases (Upton et al. 2019; Ichim et al. 2020). Chemical methods
and multiple myeloma (Chunming et al. 2020; Tew et al. also had wide applicability in the quality assurance of
2020). An average annual collection of 2000 tons of dried herbal medicines, nevertheless, studies have reported the
stem to meet industrial demand has negatively affected discrepancy of chemical markers in delineating medicinal
survival of the species (Thriveni et al. 2015). Further, plants, due to variation with respect to age and environ-
removal of this liana from wild even before reaching its mental heterogeneity of plant species (Liu et al. 2011; Kaur
reproductive maturity of fifteen years is another setback et al. 2016; Moustafa et al. 2016; Cao et al. 2017). A
affecting long term survival of the species. Unscientific stable system for checking the authenticity of herbal drugs
collection along with intrinsic properties of the liana such and safety of consumers are thus envisaged for ensuring
as habitat specificity, dioecious nature, poor seed set and safe usage of herbal dugs (Mishra et al. 2016). Many
lack of regeneration had labelled the species as critically researchers have diverted their focus towards the multi-
endangered (Ramasubbu et al. 2012; Thriveni et al. 2015). tiered or integrated approach, by amalgamating already
Consequently, an export ban of the species was introduced existing methods which are yet to be fine-tuned with
by the Ministry of Commerce, Government of India, molecular tools (Palhares et al. 2014,2015; Urumarudappa
accentuating the demand (vide notification No. 47 (PN)/ et al. 2016; Seethapathy et al. 2018).
92–97 dated 30 March 1994). This further indicates that, Recently, DNA barcode based plant identification has
the current demand for C. fenestratum has been met been considered as a powerful and potential method in
probably by wide spread adulteration/substitution of the herbal pharmacovigilance research for quality assurance
species with other plant materials (Tushar et al. 2008). along with hyphenated methods (De Boer et al. 2015;
Recently, raw drug adulteration has become a burning Palhares et al. 2015; Urumarudappa et al. 2016; Seethap-
problem in herbal industries wherein the quality of for- athy et al. 2018; Ichim et al. 2019). DNA barcode analysis
mulations is compromised with look-alike plants of inferior has revealed wide-range of discrepancies in the claimed
properties (Srirama et al. 2017). Consequences of herbal composition of various North American herbal products
drug adulteration were reported from the countries like viz. in saw palmetto dietary supplements, ginkgo products,
Australia, Japan, Taiwan and China, where chronic use of black cohosh, herbal teas and in ginseng (Stoeckle et al.
Artistolochia fangchi adulterated products led to death of 2011; Baker et al. 2012; Wallace 2012; Little and Jeanson
patients due to renal failure (Michl et al. 2013; Jadot et al. et al. 2013; Newmaster et al. 2013; Ichim and de boer
2017). Earlier, US Food and Drug Administration also 2020). Though DNA barcoding can be used specifically for
banned Piper methysticum containing products in Ger- unprocessed plant materials, it has practical limitations to
many, Switzerland, France, Canada and UK owing to authenticate highly processed multi-herbal formulations
health issues related to hepatitis, cirrhosis and liver failure (Raclariu et al. 2017b). The post harvesting process or
(U.S. Food and Drug Administration 2002). Adulteration manufacturing of herbal materials affects the quality of
and contamination of herbal medicinal products resulted in DNA which further affects DNA barcoding as well as meta
agranulocytosis, meningitis, multi-organ failure, perinatal barcoding (de Boer et al. 2015).
stroke, arsenic, lead or mercury poisoning, malignancies or The present study addresses the extent of adulteration
carcinomas, hepatic encephalopathy, hepato-renal syn- and illicit trade of C. fenestratum in south India using
drome, nephrotoxicity, rhabdomyolysis, metabolic acido- integrated approach of DNA barcoding and HPTLC (High
sis, renal or liver failure, cerebral edema, coma, Performance Thin Layer Chromatography) fingerprinting.
intracerebral haemorrhage, and even death (Posadzki et al. The study also analyses the efficiency of approaches like
2012). In this regard, WHO pharmacopoeia highlights the Maximum Likelihood phylogenetic tree and Machine
importance of ensuring the quality of raw drugs through Learning Algorithm (WEKA) for molecular sequence data
proper identification of recommended species for the pur- and hierarchical clustering for chemical fingerprints to
pose (WHO 2011). Accordingly, Indian Pharmacopoeia ascertain the extent of adulteration. This is for the first time
recommends methods such as traditional taxonomic to MLA is used in herbal drug authentication.

123
Physiol Mol Biol Plants (March 2021) 27(3):605–617 607

Materials and methods extraction. DNA was quantified using Nanodrop pho-
tometer programmed through CFR 21 software (Implen,
Collection of authenticated biological reference Germany), and quality was analyzed by agarose gel elec-
material trophoresis. Total genomic DNA was extracted from
commercially traded samples using Qiagen Plant DNeasy
The common market adulterants of C. fenestratum were Mini Kit (Qiagen Inc., USA) with slight modifications
identified through a preliminary market and industrial (samples were finely powdered using blender followed by
survey in south India as part of a previous research project. 3-h incubation at 65° C). Working concentration of geno-
Accordingly, four species with similar features (woody mic DNA was prepared by diluting the stock solution at a
stem with berberine) were selected. Mature stem, leaf and concentration 25 ng/lL. PCR reaction mixture comprised
flower samples of C. fenestratum and its market adulterants of 2.5 lL PCR buffer at 1X (supplied with10X concen-
viz. Anamirta cocculus (L.) Wight & Arn., Diploclisia tration), 1 lL each of forward and reverse primers
glaucescens (Blume) Diels, Morinda pubescens Sm. were (5 pmol), 2.5 lL of dNTPs from 10 mM stock, 2 U/25 lL
collected from different geographic locations of south India of Taq-polymerase, 1 lL template DNA with the concen-
and Berberis aristata DC. from north India (Fig. 1). GPS tration of 25 ng/lL, and the final volume of the PCR
coordinates of the collected locations were also recorded reaction mixture was made up to 25 lL with sterile distilled
(Supplementary Table S1). Taxonomically confirmed water. Total genomic DNA of BRM was used for the
samples at the Institute were then assigned a specific amplification of CBOL recommended barcode regions viz.
voucher number and deposited in the herbarium at KFRI, ITS, rbcL, psbA-trnH and matK (Table 1). The eluted PCR
known by the acronym KFRI by Index Herbariorum products were subjected to Sanger’s dideoxy sequencing
(Taxon 37: 503. 1988). This taxonomically authenticated using Applied Biosystems DNA Sequencer (ABI) and ABI
Biological Reference Materials (BRM) were used to DNA Sequencing Analysis Software v5.1 (Perkin-Elmer,
develop species specific barcodes for four of the standard made in USA).
barcode gene regions (rbcL, matK, ITS and psbA-trnH).
The generated DNA barcode database was further analysed HPTLC
for validating the authenticity of market raw drugs.
CAMAG Linomat 5 with twin plate chamber and CAMAG
Collection of commercially traded samples TLC scanner instrument programmed through Win CATS
software (Merck, made in Germany) was used for HPTLC
Commercially traded samples were purchased from the fingerprinting. Matured stem of BRM (n = 10, with 2
selected authorised dealers of ayurvedic raw drugs and individuals in each species) was powdered coarsely. Ten
major ayurvedic industries in south India (Kerala and gram powdered samples from each stem was accurately
Tamil Nadu). Thirty raw drug samples of C. fenestratum weighed (10 g) and extracted using 300 mL methanol.
sold under common vernacular name maraman- Extracts were filtered and concentrated under reduced
jal/daruharidra were purchased from Kerala and Tamil pressure and made up to 10 mL in standard flasks sepa-
Nadu. About 100 g of raw drug sample (available in rately. Stationary phase of aluminium TLC plates pre-
extremely dried and shredded form) was collected from coated with Silica gel 60 F254 of 0.2 mm thickness and
each shop to check adulteration. Each of the collected mobile phase of toluene: ethylaceate: acetic acid (5:1:0.6
samples was given a Herbal Authentication Service Code v/v) were used for the analysis. Samples were visualised in
(HAS) with details of location (Supplementary Table S2). 366 nm and 254 nm (Supplementary Figs. 1 and 2). The
To avoid the chances of mixing up, strict protocol was plate was derivatized with anisaldehyde sulphuric acid
followed from collection to final data analysis. Most of reagent at 366 nm for band visualization of phenolic
collected samples had not retained any of the morpholog- components in the plant parts used.
ical features of the original plant. HPTLC banding profile of BRM was documented under
366 nm UV light. Chemical profile of each sample was
Genomic DNA extraction, PCR amplification analysed according to their RF values (Retention factor).
and Sequencing Dendrogram was constructed by SPSS v.16.0 (SPSS Inc
2007) using nearest neighbour adopting euclidean distance,
Total genomic DNA was extracted from BRM leaf samples which revealed the relation between each species according
(n = 25 with 5 individuals for each species) using Qiagen to their phytochemical constituents. Owing to variations in
Plant DNeasy Mini Kit (Qiagen Inc., USA). 100 mg tissue HPTLC fingerprints between geographical accessions of a
from silica gel dried leaf was used for genomic DNA species, the technique cannot be further considered for
market sample validation.

123
608 Physiol Mol Biol Plants (March 2021) 27(3):605–617

Fig. 1 Map showing the Biological Reference Material (BRM) collected from multiple locations

123
Physiol Mol Biol Plants (March 2021) 27(3):605–617 609

Table 1 Details on primers and PCR reaction conditions

Barcode Primer name Sequence 50 -30 PCR profile (35 Cycles)
loci

ITS ITS 1 TCCGTAGGTGAACCTGCGG 95 °C 5 min, 94 °C 1 min, 60 °C 1 min, 72° 1 min,

ITS 2 TCCTCCGCTTATTGATATGC 72° C 10 min
psbA- psbA GTWATGCAYGAACGTAATGCTC 95 °C 5 min, 94 °C 1 min, 58 °C 1 min, 72° 1 min,
trnH trnH CGCGCATGGTGGATTCACAATCC 72° C 10 min
matK matk 427 F CCCRTYCATCTGGAAATCTTGGTT 95 °C 5 min, 94 °C 1 min, 50 °C 1 min, 72° 1 min,
matK 1248 R GCTRTRATAATGAGAAAGATTTCTGC 72° C 10 min
rbcL rbcL 1 F ATGTCACCACAAACAGAAAC 95 °C 5 min, 94 °C 1 min, 60 °C 1 min, 72° 1 min,
rbcL 724 R TCGCATGTACCTGCAGTAGC 72° C 10 min

DNA barcode development from BRM samples reference database. The BLAST analysis was performed
with rbcL sequences as queries to identify similar nucleo-
Raw chromatograms were edited and trimmed using tide reference sequences in the GenBank database with
BioEdit software (Hall 1999). Edited sequences were highest homology, maximum query coverage, and maxi-
aligned using Clustal W (Thompson et al. 1994) and sub- mum score to assign the identity of the species. Maximum
mitted to GenBank (http://www.ncbi.nlm.nih.gov/genbank/) Likelihood tree was constructed using developed barcode
(Supplementary Table S3). Homology searches of all the sequences of commercially traded samples and BRM, with
sequences obtained from BRM barcode library were per- 1000 bootstrap using MEGA v.7. 0 adopting Kimura 2
formed using BLAST (http://blast.ncbi.nlm.nih.gov/Blast. model with partial deletion (Kumar et al. 2016).
cgi), to confirm the identity of sequences. For pair-wise
genetic distance (PWG) method, genetic distances of BRM Machine learning approach (MLA)
samples (interspecific as well as intraspecific distances)
were determined by MEGA v.7.0 using Kimura two-pa- In MLA, DNA barcoding analysis can be performed with a
rameter distance model (K2P) adopting complete deletion reference data set composed of DNA sequence of known
(Tamura et al. 2013). The interspecific divergence among species (BRM) and query data set with the sequence of
species was calculated using three parameters; (1) average unknown species (market samples). In the adopted algo-
interspecific distance, (2) average theta prime and (3) rithm namely, Waikato Environment for Knowledge
minimum interspecific distances. Intraspecific parameters; Analysis (WEKA), the function-based method Support
(4) average intraspecific distance, (5) theta (h) and (6) Vector Machines (SMO) (Suykens and Vandewalle 1999),
coalescent depth were also calculated to characterize the rule-based RIPPER (Jrip) (Shahzad et al. 2013), the
intraspecific divergences (Meyer and Paulay 2005). DNA decision tree C4.5 (J48) (Quinlan 1996) and the Bayesian-
barcoding gaps were calculated by comparing intra and based method Naive Bayes (Lewis 1998) were tested on
interspecific genetic distances (Kress and Erickson 2007). DNA barcodes with tenfold cross validation. The ‘‘.fasta’’
An ideal barcode region with higher interspecific but low files of barcode sequences were converted to ‘‘.arff’’ format
intraspecific divergence was further used to differentiate using ‘‘Fasta2Weka’’ programme for analysis in WEKA
the market samples. (Weitschek et al. 2014). All four classification methods in
WEKA were run with four barcode primer sequences of
Data analysis and validation of commercial sample BRM. Best classifier was selected according to their effi-
ciency in species discrimination. Using the best classifier,
Total genomic DNA from 30 commercially traded samples rbcL sequences from commercially traded market samples
was initially amplified with CBOL recommended barcode were further analysed along with BRM sequence database.
regions viz. ITS, rbcL, psbA-trnH and matK. The genomic
DNA extracted from extremely dried samples was slightly
degraded and three barcode regions viz. ITS, psbA-trnH Result
and matK failed to give amplification consistently. rbcL
barcode region successfully amplified even from degraded DNA barcode analysis
specimens, was thus considered for the validation analysis.
The eluted PCR products were subjected to Sanger’s All the analysed barcode regions (ITS, psbA-trnH, matK
dideoxy sequencing and sequences obtained from traded and rbcL) amplified successfully with 100 percent PCR
samples were analysed along with the BRM barcode efficiency. Among these barcode regions, psbA-trnH spacer

123
610 Physiol Mol Biol Plants (March 2021) 27(3):605–617

Table 2 Sequence characteristics of four barcode regions and their Table 4 Species identification rates as percentage (correctly identi-
combinations fied/misidentified/not identified) for the BRM using Machine Learn-
ing Algorithm (WEKA)
Comparison rbcL matK psbA-trnH ITS
Barcode loci Naı̈ve Bayes J48 JRip SMO
Sequence length 590 780 648 423
Conserved sites 378 580 188 257 WEKA classifiers
Variable site 212 200 460 166 rbcL 50/50/0 100/0/0 91/9/0 100/0/0
Informative site 375 287 455 166 matK 77/23/0 100/0/0 90/10/0 100/0/0
Singleton site 3 0 15 0 ITS 44/56/0 100/0/0 72/28/0 100/0/0
psbA-trnH 77/23/0 94/6/0 55/45/0 100/0/0

region showed highest nucleotide variation (460/648),

Similar results were also generated in the MLA analysis.
followed by rbcL (212/590), matK (200/780) and ITS (166/
All four classification methods were run in WEKA with
423) regions, respectively (Table 3). Intra and Inter specific
ten-fold cross validation (Table 5). Among four machine
genetic divergences of these sequences analysed from the
learning algorithms, Naive bayes and JRip failed to identify
four barcode regions showed only interspecific divergences
the sequences of reference data set as well as the test
and none of them displayed any intraspecific divergences
sequence database. SMO and J48 showed species identifi-
(Supplementary Fig. 3). Among these four barcode regions,
cation in BRM samples with 100 percent discriminatory
psbA-trnH and rbcL showed highest interspecific diver-
power (Supplementary Figs. 5–8). These two machine
gence in the Biological Reference Materials (Table 4).
learning algorithms were subsequently employed for
Based on intra and inter specific distances, barcode gap was
authentication of sequences of unknown samples. When
also estimated. The barcode regions viz. ITS, psbA-trnH,
the test data of unknown commercially traded samples with
matK and rbcL showed distinct barcode gap of 0.0276,
large variations were analysed, the J48 classifier could
0.175, 0.0234 and 0.0632 respectively. Wilcoxon’s signed
identify only 35 percent of the species. Best performance
rank test performed to test the significance of interspecific
was shown by SMO with 100 % accuracy in authenticating
divergences in barcode regions (psbA-trnH, matK, rbcL,
the test data of thirty market samples (Fig. 3). Test data set
and ITS) showed significant values. Based on barcode gap
of thirty market samples showed similarity with the refer-
analysis, psbA-trnH and rbcL gene regions can be consid-
ence data set (BRM sequence database). Twenty com-
ered as effective barcodes (Supplementary Fig. 4). Further,
mercially traded samples showed similarity with B.
rbcL barcode region alone was adopted for authentication
aristata, and ten with C. fenestratum provided in the ref-
of commercially traded samples successfully owing to
erence data set, which again substantiated the dendrogram
difficulty in amplification of other barcode regions from
based sequence analysis in MEGA.
degraded DNA. The barcode sequences developed from
these samples were clustered into separate clades corre-
HPTLC analysis
sponding to the respective sequences of BRM samples. The
Maximum Likelihood tree with 1000 bootstraps using
Dendrogram generated using RF values was used to anal-
MEGA v.7. 0 adopting Kimura 2 model with partial
yse the banding pattern of BRM samples (Fig. 4). Each
deletion generated showed a clear clustering of thirty
species showed specific banding pattern with intra species
commercially traded samples with those of BRM (Fig. 2).
variation (Supplementary Fig. 1) belonging to different
Thus, among the thirty samples, twenty samples could be
geographical locations were grouped in different clades. C.
identified as B. aristata and remaining ten samples were
fenestratum from Aralam, south India and B. aristata from
authenticated as C. fenestratum.
Himachal Pradesh, north India were grouped together,

Table 3 Evaluation of the four DNA barcode regions used in this study
Parameters rbcL matK PsbA-trnH ITS

Average intraspecific distance 0 0 0 0

Average interspecific divergence 0.0632 ± 0.0015 0.0234 ± 0.0064 0.175 ± 0.0880 0.0276 ± 0.0064

123
Physiol Mol Biol Plants (March 2021) 27(3):605–617 611

Fig. 2 Maximum Likelihood

tree (ML) of market samples
along with BRM using rbcL
barcode

123
612 Physiol Mol Biol Plants (March 2021) 27(3):605–617

Fig. 3 The confusion matrix showing identification rate of market samples along with BRM based on rbcL using J48 and SMO

while C. fenestratum from Sholayar, south India showed conventional methods more complicated. Since raw drugs
similarity with Morinda pubescens. Likewise, B. aristata are available in extremely dried, shredded or powdered
showed similarity with Anamirta cocculus and Diploclisia form, species identification using microscopic and macro-
glaucescens. Therefore, HPTLC fingerprinting could not be scopic methods may not always be feasible. In many cases,
further used as a species authentication tool because of the chemical fingerprinting/chemical methods are utilized for
inconsistencies in the fingerprints generated. the qualitative and quantitative analysis of herbal drugs.
However, chemical fingerprints are controlled by external
environmental factors such as plant age, storage conditions,
Discussion post-harvest technology, type of plant parts used (Kaur
et al. 2016; Liu et al. 2011) and show discrepancy in the
Adulteration/substitution is recently a critical issue in results. But DNA barcode based plant identification
ayurvedic industries and hence authentication constitutes a method has been comparatively considered as most pow-
pertinent role in assuring quality, safety and efficacy of erful and potential method in herbal pharmacovigilance
ayurvedic formulations. A vast array of techniques such as research (De Boer et al. 2015; Raclariu et al. 2017b) owing
physical, chemical (analytical), anatomical, organoleptic, to its consistency. As compared to other traditional meth-
and recently emerged DNA based molecular methods are ods, DNA barcode database once developed can be made
widely used for plant species authentication (Mishra et al. available in a public domain for species identification
2016; Raclariu et al. 2017a). Conventional taxonomic without involving much technical expertise. In India, spe-
method generally emphasize on morphology and anatomy cies adulteration in commercially traded samples of
of the species. However, recent studies have shown the Aswagandha, Cinnamomum verum, Cassia, Myristica fra-
impact of ecological and climatic variables in morphology grans, Phyllanthus, Sida, Santalum album, Saraca asoca,
and anatomy of species (Li et al. 2019, 2020), which makes Senna and Garcina was analysed using DNA barcodes
species discrimination, a tedious job. The lack of expertise, (Srirama et al. 2010; Dev et al. 2013; Seethapathy et al.
phenotypic plasticity and species delimitation make the 2015,2018; Urumarudappa et al. 2016).

123
Physiol Mol Biol Plants (March 2021) 27(3):605–617 613

Fig. 4 Dendrogram constructed using HPTLC banding pattern (RF value)

In our study, CBOL recommended barcode primers authenticate the congeneric plants as well as distantly
rbcL, matK, psbA-trnH and ITS were used to create BRM related species (Newmaster et al. 2006; Chase et al. 2007).
library in which psbA-trnH and rbcL barcode regions The present study demonstrated the efficiency of dif-
showed higher interspecific divergence. High level of ferent approaches for DNA barcode sequence data analysis
interspecific variation, difficulty of sequence alignments to validate the occurrence of adulterants in commercially
due to high number of indels and lack of high quality traded samples of C. fenestratum. The competence of rbcL
bidirectional sequences are major limitations of psbA-trnH barcode sequence data to authenticate the market samples
(Chase et al. 2007; Hollingsworth et al. 2009; CBOL 2009) was analysed employing Maximum Likelihood (ML) based
and hence was not further utilised in this study. Conse- phylogenetic tree as well as Machine learning algorithm,
quently, rbcL database was used to authenticate the market WEKA. ML tree was the most efficient and statistical
samples. PCR amplification was inconsistent and difficult suitable method to understand the process of sequence
to achieve for other barcode regions due to degradation of evolution compared to Maximum Parsimony (MP) and
DNA in extremely dried market samples. The poor storage Neighbour Joining (NJ). ML tree also has a rich repertoire
conditions (results in fungal and humus contamination) and of sophisticated evolutionary models and computer stimu-
inappropriate post-harvest methods along with degraded lations (Kuhner and Felsenstein 1994; Guindon and Olivier
DNA might have further hindered the primer annealing and 2003; Ziheng et al. 2012). Moreover, previous studies also
subsequent amplification in the traded samples (Newmaster reported the use of ML tree to detect raw drug adulteration
et al. 2013; Santhosh Kumar et al. 2018). Prior studies also (Seethapathy et al. 2018). Compared to classical barcode
showed potentiality of rbcL barcode gene region to sequence data analysis methods such as Maximum

123
614 Physiol Mol Biol Plants (March 2021) 27(3):605–617

Likelihood (ML), BLAST and nearest-neighbour, recent ranging from herb gatherers, local middlemen, urban tra-
studies have shown the higher efficacy of Machine ders, wholesalers, manufacturers, exporters and herbal
Learning Algorithm (MLA) for species authentication healers in the medicinal plants trade sector (Goraya and
(Hartvig et al. 2015). Present study could demonstrate the Ved 2017). In India, there is no proper guidelines to
effectiveness of both sequence based phylogenetic tree as coordinate and maintain the information related to collec-
well as WEKA analyses to authenticate the identity of tion, supply, trade and consumption of botanicals (Kala
unknown samples based on the BRM reference databases et al. 2006; Goraya and Ved 2017). Recently, National
created. Machine Learning Approaches were earlier per- Medicinal Plant Board (NMPB), Government of India, has
formed for authentication of timber species like Pterocar- launched an online platform e-charak, to create transparent
pus and Dalbergia (He et al. 2018, 2019). Large data trade linkage among primary collectors to end users of
analysis ability being one of the major advantages of medicinal plant sector (http://www.nmpb.nic.in).
Artificial Intelligence (AI), provides quite a large platform B. aristata is a commonly available species in north
for rapid analysis of raw drug samples, which significantly India and according to current market price list of National
reduces the human labour (He et al. 2018). AI utilises the Medicinal Plant Board (NMPB), B. aristata belongs to low
ability of computers to build models that can receive input price category of medicinal plants compared to C. fenes-
data and then conduct statistical analyses. Even though, tratum (http://www.nmpb.nic.in). However, there is no
MLA recommends minimum of four specimens per species codified market price for the raw drugs and the price of
for the BRM library, it provides higher precision, greater collected raw drugs often tend to vary from shop to shop. In
cost-effectiveness, time efficiency, and a lower threshold India, there is neither a centralised agency to monitor the
than other analytical methods. proper collection, processing, production and sale of herbal
Most of the market samples showed similarity with B. medicine nor an effective regulatory framework to evaluate
aristata when compared to the authentic species C. fenes- the quality, authenticity and safety of raw drugs or drug
tartum. Similar instances of adulteration were earlier formulations available in the market (Sahoo et al. 2013).
reported wherein Saraca asoca was adulterated with Prior studies have also suggested B. aristata as substitute of
Polyalthia longifolia, Piper nigrum with Capsicum C. fenestratum, mainly due to the presence of potent
annuum and Hypericum perforatum with Senna alexan- alkaloid berberine (Tamilselvi et al. 2014). However, it is
drina, among others (Urumarudappa et al. 2016; Parvathy not recommended as a substitute by Pharmacopoeia
et al. 2014). This adulteration of C. fenestratum with B. Commission for Indian Medicine & Homoeopathy
aristata may be due to the fact that B. aristata, a common (PCIM&H) and there is no scientific study available
species of north Indian states, are easily available in huge comparing the phytochemical constituents in B. aristata
quantities in the wild and that they share the same ver- and C. fenestratum.
nacular name and also has the potent alkaloid berberine. Recently, the combination of two or more diverse
However, C. fenestratum and B. aristata belong to two techniques were employed to authenticate the species with
different families with dissimilar anatomical, morphologi- high precision. The efficiency of multitier approach com-
cal and sexual features. Unavailability of C. fenestratum in prising DNA barcoding in combination with HPTLC for
south Indian markets point to the overexploitation and the identification of popular species such as Hamamelis
consequent loss of species in the wild. A recent survey virginiana, Matricaria recutita, Maytenus ilicifolia, Mika-
conducted by KFRI indicates that the total number of nia glomerata, Panax ginseng, Passiflora incarnata, Peu-
mature individuals in Kerala is around 110 and the repre- mus boldus and Valeriana officinalis (Palhares et al. 2014),
sentation of female plants was only 45. Moreover, the with NMR for finding out adulteration in Garcinia species
unscientific collection and extensive consumption to meet and Sarca asoca (Kumar et al. 2016; Seethapathy et al.
the huge industrial demand exerts a heavy strain on exist- 2018), with TLC and HPLC for authentication of Mars-
ing resources of the endangered populations of C. fenes- denia (Yu et al. 2018) were reported to authenticate species
tratum. Adulteration with inferior taxa is reported to with high precision. However, studies have also reported
significantly alter the therapeutic properties of ayurvedic the inconsistency of chemical markers in delineating
formulations (Srirama et al. 2010). Most herbal drugs medicinal plants, owing to variation with age of plant and
available in the market are sourced from the wild through environmental heterogeneity (Liu et al. 2011; Kaur et al.
supply chains in the formal or informal sectors (Larsen 2016; Moustafa et al. 2016; Cao et al. 2017). In the present
et al. 2006; Goraya and Ved 2017). In India, major herbal analysis, for developing HPTLC fingerprint, C. fenestratum
trade occurs through conventional collection centres and accessions were collected from environmentally heteroge-
wholesale markets. Along with government agencies like neous areas and of unknown age and sex, which must have
Forest Department, tribal cooperative society and Vana contributed to the discrepancies in HPTLC profiles. In
Samrakshana samiti, there are number of stakeholders

123
Physiol Mol Biol Plants (March 2021) 27(3):605–617 615

Tinosporia cordifolia, sex specific disparity in chemical Ministry of AYUSH, Government of India (No. F.No. Z.18017/187/
constituents was earlier reported (Bajpai et al. 2017). CSS/R&D/KE-01/2016-17-NMPB-IV A) is gratefully acknowledged.
C. fenestratum is hermaphrodite in nature with only very
Author contribution Suma Arun Dev, Jayaraj R: Fund acquisition,
few female individuals at reproductive maturity. An conceptualization, methodology & designing the experiments, writ-
imbalance in the sex ratio has affected the population ing, editing and finalization of manuscript. Remya Unnikrishnan:
viability and regeneration efficiency, which negatively Conducted wet lab experiments as part of doctoral research, data
analysis & manuscript writing. Sujanapal P, Sumod M: Sample col-
affected the long term survival of the available scattered
lection and manuscript writing.
populations of this endangered species in the Western
Ghats. This warrants an immediate action to be taken to Declaration
sustain the available resources augmented through recov-
ery, assisted regeneration, restoration and sustainable Conflict of interests The authors declare that they have no conflict of
interests.
management of the available resources of wild C. fenes-
tratum (Sumod et al 2016). DNA barcode tool can monitor
and regulate the illicit trade in their raw drug sector and
subsequently can protect and conserve the available References
germplasm in situ. Recently, KFRI has initiated the in situ
conservation of C. fenestratum in Kerala, south India Bajpai V, Kumar S, Singh A, Singh J, Negi MPS, Bag SK, Kumar B
through aided natural regeneration (ANR) in its natural (2017) Chemometric based identification and validation of
microhabitats. As part of this programme, 14,500 hardened specific chemical Markers for Geographical, Seasonal and
Gender Variations in Tinospora cordifolia stem using HPLC-
seedlings were transplanted through National Afforestation ESI-QTOF-MS Analysis. Phytochem Anal 28:277–288. https://
and Eco-Development Board (NAEB), Ministry of Envi- doi.org/10.1155/2012/18010310.1002/pca.2673
ronment & Climate Change (MOEF &CC), Government of Baker DA, Stevenson DW, Little DP (2012) DNA barcode identifi-
India. cation of black cohosh herbal dietary supplements. J AOAC Int
95:1023–1034. https://doi.org/10.1155/2012/1801310.5740/jaoa
cint.11-261
Cao Y, Fang S, Yin Z, Fu X, Shang X, Yang W, Yang H (2017)
Conclusion Chemical fingerprint and multicomponent quantitative analysis
for the quality evaluation of Cyclocarya paliurus leaves by
HPLC–Q–TOF–MS. Molecules 22:1927. https://doi.org/10.
Adulteration of ayurvedic drugs is quite rampant to meet 1155/2012/18010310.3390/molecules22111927
the industrial demand and strict regulation is therefore CBOL Plant Working Group (2009) A DNA barcode for land plants.
warranted to monitor the quality of herbal products by Proc Natl Acad Sci USA 106:12794–12797. https://doi.org/10.
authenticating right from the time of collection to the 1073/pnas.0905845106
CFSAN March 25 (2002) Centre for Food Safety and Applied
formulation of herbal drugs. A national level certification Nutrition(CFSAN): Kava-containing dietary supplements may
agency for authentication and chain of custody validation is be associated with severe liver injury. U.S. Department of Health
vital in India to ensure safety and quality of ayurvedic and Human Services, Food and Drug Administration; Rockville,
formulations. Our study could demonstrate the effective- Maryland
Chase MW, Cowan RS, Hollingsworth PM, Van den Berg C,
ness of DNA barcode tool together with artificial intelli- Madrinan S, Petersen G, Seberg O, Jorgsensen T, Cameron KM,
gence (AI) based automated species identification Carine M, Pedersen N, Hedderson TAJ, Conrad F, Salazar GA,
technique to detect the extent of adulteration in commer- Richardson JE, Hollingsworth ML, Barraclough TG, Kelly L,
cially traded raw drugs. Therefore, including DNA bar- Wilkinson M (2007) A proposal for a standardised protocol to
barcode all land plants. Taxon 56:295–299
coding as one among the standard techniques in Indian Chunming G, Zhao Y, Hong N, Yanjun L, Juhua Y et al (2020)
Ayurvedic Pharmacopoeia to validate the authenticity of Identification of berberine as a novel drug for the treatment of
raw drugs is highly recommended. Simultaneously, con- multiple myeloma via targeting UHRF1. BMC Biol 18(1):33.
servation and restoration measures through in situ and ex https://doi.org/10.1186/s12915-020-00766-8
De Boer HJ, Ichim MC, Newmaster SG (2015) DNA barcoding and
situ means of the available resources in commercially pharmacovigilance of herbal medicines. Drug Saf 38:611–620.
traded medicinal plants is recommended to replenish and https://doi.org/10.1007/s40264-015-0306-8
enrich the rapidly declining resources. Dev SA, Muralidharan EM, Sujanapal Balasundaran P (2013)
Identification of market adulterants in East Indian sandalwood
Supplementary Information The online version contains using DNA barcoding. Ann For Sci 71:517–522. https://doi.org/
supplementary material available at https://doi.org/10.1007/s12298- 10.1007/s13595-013-0354-0
021-00965-9. Goraya GS, Ved DK (2017) Medicinal plants in India: an assessment
of their demand and supply, national medicinal plants board,
Acknowledgements We are thankful to the State Forest Department ministry of AYUSH, Government of India. New Delhi and
of Kerala, for their permission to collect wild samples. The financial Indian Council of Forestry Research & Education, Dehradun
support received from National Medicinal Plant Board (NMPB),

123
616 Physiol Mol Biol Plants (March 2021) 27(3):605–617

Guindon C, Olivier G (2003) A simple, fast, and accurate algorithm to Li Y, Li S, Lu X, Wang Q, Han H, Zhang X, Gan X (2020) Leaf
estimate large phylogenies by Maximum Likelihood. Syst Biol phenotypic variation of endangered plant Tetracentron sinense
52(5):696–704. https://doi.org/10.1080/10635150390235520 Oliv. and infuence of geographical and climatic factors. J For
Hall TA (1999) BioEdit: a user-friendly biological sequence align- Res. https://doi.org/10.1007/s11676-020-01124-8
ment editor and analysis. Department of Microbiology, North Little DP, Jeanson ML (2013) DNA barcode authentication of saw
Carolina State University palmetto herbal dietary supplements. Sci Rep. https://doi.org/10.
Hartvig I, Czako M, Kjær ED, Nielsen LR, Theilade I (2015) The Use 1038/srep03518
of DNA barcoding in identification and conservation of Liu Y, Reich PB, Li G, Sun S (2011) Shifting phenology and
rosewood (Dalbergia spp). PLoS ONE. https://doi.org/10.1371/ abundance under experimental warming alters trophic relation-
journal.pone.0138231 ships and plant reproductive capacity. Ecology 92:1201–1207.
He T, Jiao L, Wiedenhoeft AC, Yin Y (2019) Machine learning https://doi.org/10.1890/10-2060.1
approaches outperform distance- and tree-based methods for Meyer CP, Paulay G (2005) DNA barcoding: error rates based on
DNA barcoding of Pterocarpus wood. Planta. https://doi.org/10. comprehensive sampling. PLoS ONE 3:e422
1007/s00425-019-03116-3 Michl J, Jennings HM, Kite GC, Ingrouille MJ, Simmonds MSJ,
He T, Jiao L, Yu M, Guo J, Jiang X, Yin Y (2018) DNA barcoding Heinrich M (2013) Is aristolochic acid nephropathy a widespread
authentication for the wood of eight endangered Dalbergia problem in developing countries? A case study of Aristolochia
timber species using machine learning approaches. Holz- indica L. in Bangladesh using an ethnobotanical-phytochemical
forschung. https://doi.org/10.1515/hf-2018-0076 approach. J Ethnopharmacol 149:235–244
Hollingsworth ML, Andra clark A, Forrest LL, Richardson J, Mishra P, Kumar A, Nagireddy A, Mani DN, Shukla AK, Tiwari R,
Pennington RT, Long DG, Hollingsworth PM (2009) Selecting Sundaresan V (2016) DNA barcoding: an efficient tool to
barcoding loci for plants: evaluation of seven candidate loci with overcome authentication challenges in the herbal market. J Plant
species-level sampling in three divergent groups of land plants. Biotechnol 1:8–21. https://doi.org/10.1111/pbi.12419
Mol Ecol Resour 9:439–457. https://doi.org/10.1111/j.1755- Mohanan N, Sivadasan M (2002) Flora of Agathymala. Bishen singh
0998.2008.02439.x Mahendra Pal Singh, Dehra Dun India, p 65
Ichim MC (2019) The DNA-based authentication of commercial Moustafa MF, Hesham AEL, Quraishi MS, Alrumman SA (2016)
herbal products reveals their globally widespread adulteration. Variations in genetic and chemical constituents of Ziziphus
Front Pharmacol. https://doi.org/10.3389/fphar.2019.01227 spina-christi L. populations grown at various altitudinal zonation
Ichim MC, Haser A, Nick P (2020) Microscopic authentication of up to 2227m height. J Genet Eng Biotechnol 14:349–362
commercial herbal products in the globalized market: potential Newmaster SG, Grguric M, Shanmughanandhan D (2013) DNA
and limitations Front. Pharmacol 11:876. https://doi.org/10. barcoding detects contamination and substitution in North
3389/fphar.2020.00876 American herbal products. BMC Med 11:222. https://doi.org/
Ichim MC, de Boer H (2020) A review of authenticity and 10.1186/1741-7015-11-222
authentication of commercial ginseng herbal medicines and Newmaster SG, Fazekas AJ, Ragupathy S (2006) DNA barcoding in
food supplements Front. Pharmacol. https://doi.org/10.3389/ land plants: evaluation of rbcL in a multigene tiered approach.
fphar.2020.612071 Can J Bot 84:335–341
Jadot I, Decleves A, Nortier J, Caron N (2017) An integrated view of Palhares RM, Gonçalves Drummond M, Brasil AF, Cosenza GP, Lins
aristolochc acid nephropathy: update of the literature. Int J Mol Brandao MDG, Oliveira G (2015) Medicinal plants recom-
Sci 18:297 mended by the World Health Organization: DNA barcode
Kala C, Dhyani P, Sajwan B (2006) Developing the medicinal plants identification associated with chemical analyses guarantees their
sector in northern India: challenges and opportunities. J Ethno- quality. PLoS ONE 10:e0127866. https://doi.org/10.1371/jour
biol Ethnomed 2:32. https://doi.org/10.1186/1746-4269-2-32 nal.pone.0127866
Kaur T, Bhat R, Vyas D (2016) Effect of contrasting climates on Palhares RM, Drummond MG, Brasil BS, Krettli AU, Oliveira GC,
antioxidant and bioactive constituents in five medicinal herbs in Brandao MG (2014) The use of an integrated molecular-
Western Himalayas. JMS 13:484–492. https://doi.org/10.1007/ chemical- and biological-based approach for promoting the
s11629-014-3380-y better use and conservation of medicinal species: a case study of
Kress WJ, Erickson DL (2007) A two-locus global DNA barcode for Brazilian quinas. J Ethnopharmacol 155:815–822
land plants: the coding rbcL gene complements the non-coding Parvathy VA, Swetha VP, Sheeja TE, Leela NK, Chempakam B,
trnH-psbA spacer region. PLoS ONE 2:e508 Sasikumar B (2014) DNA barcoding to detect chilli adulteration
Kumar S, Stecher G, Tamura K (2016) MEGA7: molecular evolu- in traded black pepper powder. Food Biotechnol 28:25–40.
tionary genetics analysis version 7.0 for bigger datasets. Mol https://doi.org/10.1080/08905436.2013.870078
Biol Evol 33:1870–1874. https://doi.org/10.1093/molbev/ Posadzki P, Watson L, Ernst E (2012) Contamination and adulteration
msw054 of herbal medicinal products (HMPs): an overview of systematic
Kuhner MK, Felsenstein J (1994) A simulation comparison of reviews. Europ J Clin Pharmacol 69(3):295–307. https://doi.org/
phylogeny algorithms under equal and unequal evolutionary 10.1007/s00228-012-1353-z
rates. Biol. Evol, Mol. https://doi.org/10.1093/oxfordjournals. Punitha ISR, Rajendran K, Shirwaikar A, Shirwaikar A (2005)
molbev.a040126 Alcoholic stem extract of Coscinium fenestratum regulates
Larsen HO, Olsen CS (2006) Unsustainable collection and unfair carbohydrate metabolism and improves antioxidant status in
trade? Uncovering and assessing assumptions regarding Central streptozotocin-nicotinamide induced diabetic rats. Evid Based
Himalayan medicinal plant conservation. Biodivers Conserv Complement Altern Med 2:375–381. https://doi.org/10.1093/
16:1679–1697. https://doi.org/10.1007/s10531-006-9039-4 ecam/neh099
Lewis DD (1998) Naı̈ve (Bayes) at forty: the independence assump- Quinlan JR (1996) Improved use of continuous attributes in C4.5.
tion in information retrieval. Mach Learn 98:4–15 JAIR 4:77–90
Li J, Wu J, Peng K, Fan G, Yu H, Wang W, He Y (2019) Simulating Raclariu AC, Heinrich M, Ichim MC, de Boer H (2017a) Benefits and
the effects of climate change across the geographical distribution limitations of DNA barcoding and metabarcoding in herbal
of two medicinal plants in the genus. Nardostachys PeerJ product authentication. Phytochem Anal 29(2):123–128. https://
7:e6370. https://doi.org/10.7717/peerj.6730 doi.org/10.1002/pca.2732

123
Physiol Mol Biol Plants (March 2021) 27(3):605–617 617

Raclariu AC, Mocan A, Popa MO, Vlase L, Ichim MC, Crisan G Tamura K, Stecher G, Peterson D, Filipski A, Kumar S (2013)
(2017b) Veronica officinalis product authentication using DNA MEGA6: molecular evolutionary genetics analysis version 6.0.
metabarcoding and HPLC-MS reveals widespread adulteration Mol Biol Evol 30:2725–2729
with Veronica chamaedrys. Front Pharmacol 8:378. https://doi. Tewa XN, Laua NJX, Chellappanb DK, Madheswaranc T, Zeeshanc
org/10.3389/fphar.2017.00378 TMM et al (2020) Immunological axis of berberine in managing
Rai RV, Rajesh PS, Kim HM (2012) Medicinal use of Coscinium inflammation underlying chronic respiratory inflammatory dis-
fenestratum (Gaertn.) Colebr.: an short review. Oriental Pharmacy eases. Chem Biol Interact 1:1. https://doi.org/10.1016/j.cbi.2020.
Exp Med 13:1–9. https://doi.org/10.1007/s13596-012-0094-y 108947
Ramasubbu R, Chandra Prabha A, Kumuthakalavalli R (2012) Seed Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W:
biology of Coscinium fenestratum (Ga ertn.) Colebr.—a criti- improving the sensitivity of progressive multiple sequence
cally endangered medicinal plant of Western Ghats. J Med Plant alignment through sequence weighting, position specific gap
Res 6:1094–1096 penalties and weight matrix choice. Nucleic Acids Res
Sahoo N, Manchikanti P (2013) Herbal drug regulation and 22:4673–4680
commercialization: an Indian industry perspective. J Altern Thriveni HN, Srikanth V, Gunaga RHN, Vasudeva R (2015)
Complement Med 19:957–963. https://doi.org/10.1089/acm. Ecological niche modeling, population status and regeneration
2012.0275 of Coscinium fenestratum colebr. (Menispermaceae): a medic-
Santhosh Kumar JU, Krishna V, Seethapathy GS, Ganesan R, inally important liana of the central Western Ghats. Trop Ecol
Ravikanth G, Uma Shaanker R (2018) Assessment of adulter- 56:101–110
ation in raw herbal trade of important medicinal plants of India Tran QL, Tezuka Y, Ueda JY, Nguyen NT, Maruyama Y, Begum K,
using DNA barcoding. 3 Biotech 8(3):135. https://doi.org/10. Kadota S (2003) In vitro antiplasmodial activity of antimalarial
1007/s13205-018-1169-3 medicinal plants used in Vietnamese traditional medicine.
Sarvalingam A, Rajendran A (2016) Rare, endangered and threatened J Ethnopharmacol 86:249–252. https://doi.org/10.1016/s0378-
(RET) climbers of aouthern Western Ghats, India. Rev Chil de 8741(03)00045-x
Hist Nat 89:2–5. https://doi.org/10.1186/s40693-016-0058-6 Tushar KV, George S, Remashree AB, Balachandran I (2008)
Seethapathy GS, Ganesh D, Kumar JUS, Senthilkumar U, Newmaster Coscinium fenestratum (Gaertn.) Colebr.—a review on this rare,
SG, Ragupathy S (2015) Assessing product adulteration in critically endangered and highly traded medicinal species.
natural health products for laxative yielding plants, Cassia, J Plant Sci 3:133–145
Senna, and Chamaecrista in Southern India using DNA barcod- Udayan PS, Satheesh George KV, Thushar IB (2005) Ethanomedic-
ing. Int J Legal Med 129:693–700. https://doi.org/10.1007/ inal plant used by the Oorali tribe of Idukki district, Kerala state,
s00414-014-1120-z India. J Econ Taxon Bot 29:219–220
Seethapathy GS, Tadesse M, Urumarudappa SKJV, Gunaga S, Ueda J, Tezuka Y, Banskota AH, Le Tran Q, Tran QK, Harimaya Y,
Vasudeva R, Malterud KE, Wangensteen H (2018) Authentica- Kadota S (2002) Antiproliferative activity of Vietnamese
tion of Garcinia fruits and food supplements using DNA medicinal plants. Biol Pharm Bull 25:753–760
barcoding and NMR spectroscopy. Sci Rep 8:10561. https:// Upton R, David B, Gafner S et al (2019) Botanical ingredient
doi.org/10.1038/s41598-018-28635 identification and quality assessment: strengths and limitations of
Shahzad W, Asad S, Khan MA (2013) Feature subset selection using analytical techniques. Phytochem Rev 19:1157–1177. https://doi.
association rule mining and JRip classifier. Int J Phys Sci org/10.1007/s11101-019-09625-z
8:885–896 Urumarudappa SK, Gogna N, Newmaster SG, Venkatarangaiah K,
SPSS Inc (2007) Released. SPSS for Windows, Version 16.0. Subramanyam R, Saroja SG, Gudasalamani R, Dorai K,
Chicago, SPSS Inc Ramanan US (2016) DNA barcoding and NMR spectroscopy-
Srirama R, Senthilkumar U, Sreejayan N, Ravikanth G, Gurumurthy based assessment of species adulteration in the raw herbal trade
BR, Shivanna MB, Uma Shaanker R (2010) Assessing species of Saraca asoca (Roxb) Willd. An important medicinal plant. Int
admixtures in raw drug trade of Phyllanthus, a hepato-protective J Legal Med 130:1457–1470
plant using molecular tools. J Ethnopharmacol 130:208–215. Ved D, Saha D, Ravikumar K, Haridasan K (2015) Coscinium
https://doi.org/10.1016/j.jep.2010.04.042 fenestratum. The IUCN Red List of Threatened Species.
Srirama R, Santhosh Kumar JU, Newmaster SGS, Ragupathy S, e.T50126585A50131325
Ganeshaiah KN, Ravikanth G (2017) Species adulteration in the Wallace LJ, Boilard SMAL, Eagle SHC, Spall JL, Shokralla S,
herbal trade: causes, consequences and mitigation. Drug Saf Hajibabaei M (2012) DNA barcodes for everyday life: routine
40:651–661. https://doi.org/10.1007/s40264-017-0527-0 authentication of natural health products. Food Res Int
Stoeckle MY, Gamble CC, Kirpekar R, Young G, Ahmed S, Little DP 49:446–452. https://doi.org/10.1016/j.foodres.2012.07.048
(2011) Commercial teas highlight plant DNA barcode identifi- Weitschek E, Fiscon G, Felici G (2014) Supervised DNA Barcodes
cation; successes and obstacles. Sci Rep. https://doi.org/10.1038/ species classification: analysis, comparisons and results. Biodata
srep00042 Minining 7:4
Sumod M, Sujanapal P, Robi AJ, Dantus KJ, Arun KR, Sathiyakumar WHO (2011) The World medicines situation—traditional medicines:
K (2016) Population status and distribution of Coscinium global situation, issues and challenges. WHO Press, Geneva
fenestratum Colebr. (Menispermaceae) in northern Kerala. In: Yu N, Wei Y, Zhu Y, Zhu N, Wang Y, Zhang H, Sun A (2018)
Proceedings of 26th Kerala Science Congress 2499–2507, Govt. Integrated approach for identifying and evaluating the quality of
of Kerala Marsdenia tenacissima in the medicine market. PLoS ONE
Suykens JAK, Vandewalle J (1999) Least squares support vector 13:e0195240
machine classifiers. Neural Process Lett 9:293–300 Ziheng Y, Bruce R (2012) Molecular phylogenetics: principles and
Tamilselvi S, Venkatasubramanian P, Vasanthi NS (2014) Physico- practice. Nat Rev Genet 13:303–314. https://doi.org/10.1038/
Chemical characterization and anti-inflammatory activity of nrg3186
stem extracts of Berberis aristata DC and Cosinium fenestratum
Linn in Carrageenan induced wistar rats. Pharmacogn J 6:72–77. Publisher’s Note Springer Nature remains neutral with regard to
https://doi.org/10.5530/pj.2014.4.11 jurisdictional claims in published maps and institutional affiliations.

123