The Research On Cistanche Species

Najibeh Ataeia,b, Gerald M. Schneeweissc,⁎, Miguel Angel Garcíad,e,1, Michael Kruger, Marcus Lehnerta, Jafar Valizadehf, Dietmar Quandta,⁎

a Nees Institute for Biodiversity of Plants, Rheinische Friedrich-Wilhelms-Universität Bonn, Meckenheimer Allee 170, 53115 Bonn, Germany

b General Directorate for Agricultural Research Institute of Afghanistan (ARIA), Ministry of Agriculture, Irrigation and Livestock, Badam Bagh, Kabul, Afghanistan

c Department of Botany and Biodiversity Research, University of Vienna, Rennweg 14, A-1030 Vienna, Austria

d Department of Biology, the University of Toronto at Mississauga, 3359 Mississauga, Canada

e Real Jardín Botánico, CSIC, Plaza de Murillo 2, 28014 Madrid, Spain f Department of Biology, University of Sistan and Baluchistan, Zahedan, Iran

Contact: joanna.jia@wecistanche.com / WhatsApp: 008618081934791

Abstract

Phylogenetic relationships of and within non-photosynthetic parasitic lineages are notoriously poorly known, which negatively affects our understanding of parasitic plants. This is also the case for Cistanche (Orobanchaceae), an Old World genus with about two dozen species, whose relationships have not yet been addressed using molecular phylogenetic approaches. Here we infer phylogenetic relationships within the genus, employing a taxonomically and geographically broad sampling covering all previously distinguished infrageneric groups and most of the currently recognized species. A combined matrix of three plastid markers (trnL-trnF, including the trnL intron and the intergenic spacer (IGS), trnS-trnfM IGS, and psbA-trnH IGS) and one nuclear marker (ITS) was analyzed using maximum parsimony, maximum likelihood, and Bayesian inference. Cistanche falls into four well-supported and geographically differentiated clades: East Asian Clade, Northwest African Clade, Southwest Asian Clade, and widespread Clade. Of those, only the East Asian Clade corresponds to a previously recognized taxonomic section, whereas the others either contain members of two or three sections (Widespread Clade and Southwest Asian Clade, respectively) or have not been taxonomically recognized so far (Northwest African Clade). Whereas the SouthwestAsian Clade exhibits strong phylogenetic structure among and partly within species (the East Asian Clade and the northwest African Clade are monospecific), phylogenetic resolution within the Widespread Clade is often low and hampered by discrepancies between nuclear and plastid markers. Both molecular and morphological data indicate that species diversity in Cistanche is currently underestimated.

Keywords: Cistanche Host, Orobanchaceae, Parasite, Phylogeny, Systematics, Taxonomy

Cistanche herba Extract Powder from Chengdu Wecistanche

1. Introduction

The Orobanchaceae are an excellent model system for studying the evolution of parasitism in plants and the underlying phenotypic and genetic modifications (Westwood et al., 2010, Wickett et al., 2011). They also are an example of how molecular data have successfully improved our understanding of phylogenetic relationships. Based on molecular studies, the family circumscription has considerably changed and intergeneric relationships have been modified (Wolfe et al., 2005; Bennett and Mathews, 2006; Park et al., 2008; McNeal et al., 2013; Fu et al., 2017; Li et al., 2019). The broadest and most comprehensive phylogenetic analyses of the Orobanchaceae to date is by McNeal et al. (2013), who used a combined data set of five markers (nuclear: ITS, PHYA, PHYB; plastid: matK and rps2) comprising more than 50 genera of the family. Despite this progress, many genera have not been included in molecular studies and remain unplaced (Schneeweiss, 2013). Additionally, phylogenetic relationships within genera, especially in taxonomically difficult non-photosynthetic parasitic groups, remain poorly explored.

A group in need of a more thorough phylogenetic investigation is the non-photosynthetic root-parasitic genus Cistanche. It is a potentially rewarding object for studying the evolution of large genomes (much larger chromosomes and correspondingly much larger genome size than in closely related lineages: Schneeweiss et al., 2004b; Weiss-Schneeweiss et al., 2006; Wicke, 2013), reductive evolution of plastid genomes (Li et al., 2013; Wicke et al., 2013, 2016; Liu et al., 2020), or species diversity and biogeography in arid regions, but any of these research avenues is currently hampered by poor understanding of species relationships due to the lack of thorough phylogenetic data. The few studies that include Cistanche focus on the evolution of the plastid genome (Li et al., 2013; Wicke et al., 2013, 2016) or the relationships of species used in traditional Chinese medicine (Tomari et al., 2002, 2003; Han et al., 2010; Sun et al., 2012; Zheng et al., 2014), and are taxonomically very restricted. Likewise, phylogenetic studies directed at the family level or focusing on related genera, such as Orobanche, include only one or few representatives of Cistanche (Young et al., 1999; Schneeweiss et al., 2004a; Wolfe et al., 2005; Park et al., 2008; McNeal et al., 2013; Schneeweiss, 2013; Li et al., 2017, 2019). Their results show that Cistanche is closely related to other non-photosynthetic parasitic genera, such as Orobanche, Phelipanche, Conopholis, or Epifagus (clade III of McNeal et al., 2013; Orobanche-Phelipanche clade of Schneeweiss, 2013), but its precise phylogenetic placement has not been fully settled (Schneeweiss, 2013).

The latest monograph of the entire genus is by Beck-Mannagetta (1930). Based on calyx shape and bracteole number, he distinguished four sections: (i) C. sect. Cistanchella with the single species C. ridgewayana; (ii) C. sect. A substance with the single species Cistanche Sinensis; (iii) C. sect. Heterocalyx, with three species (C. fissa, C. ambigua, and C. rosea); (iv) C. sect. Cistanche (nomenclatural incorrectly named sect. Eucistanche by Beck Mannagetta, 1930) contains the remaining 12 species. Later taxonomic treatments, usually in the context of national floras, add a few new species and partly modify circumscriptions of often morphologically variable and thus taxonomically complex species already recognized by BeckMannagetta (1930) so that currently about 26 species are accepted (The Plant List, http://www.theplantlist.org, assessed on 6. Dec. 2017). The genus is widely distributed in arid regions of the Old World from the Macaronesian islands and western Africa to central and eastern Asia, with the center of highest species diversity in southwestern Asia and the Middle East (Beck-Mannagetta, 1930). Like other non-photosynthetic genera of Orobanchaceae, Cistanche is characterized by morphological reduction, especially of vegetative characters (Rodrigues et al., 2011; Schneeweiss, 2013), so that most diagnostic characters are from the inflorescence and flowers (shape and indumentum of floral bracts and bracteoles, structure and indumentum of the calyx, flower color, shape and indumentum of the anther), some of which are poorly preserved on herbarium specimens. Lack of a comprehensive taxonomic treatment covering all currently recognized species, poor representation in collections especially from less explored areas, and paucity of taxonomically useful morphological characters contribute to the insufficiently known and unconsolidated taxonomy of Cistanche species.

In this study, we establish phylogenetic relationships within Cistanche as a basis for a phylogenetically predictive taxonomic system. To this end, we collect sequence data from rapidly evolving and well-established plastid markers as well as nuclear ITS sequences from a taxonomically and geographically comprehensive sampling and analyze those using maximum parsimony, maximum likelihood, and Bayesian methods. Specifically, we want (i) to test hypotheses of relationships implied by the classification of Beck-Mannagetta (1930), i.e. whether his sections constitute monophyletic groups, and (ii) to test whether morphologically and taxonomically complex species like C. phelypaea and C. tubulosa form natural groups.

2. Materials and methods

2.1. Plant material

One hundred-eighty nine samples (newly collected, herbarium material or sequences from GenBank) were included, corresponding to 17 previously identified Cistanche species plus seven outgroup taxa (one accession of Conopholis Americana, Phelipanche cf. Iberica, two accessions of Orobanche cernua, one each of O. anatolica, O. densiflora and O. Transcaucasia). A sampling of Cistanche aimed at a broad geographic coverage for each species in the genus. Species identification was based on the taxonomy used in monographic treatments and in floras (BeckMannagetta, 1930; Zhang and Tzvelev, 1998), designating morphologically deviating types as species affinis (“aff. species name”) or, where the description of a new subspecies is anticipated (a taxonomic treatment of the entire genus is in preparation), as “subsp. nov.”; for samples from GenBank, for which we could not check the herbarium vouchers, the taxonomic designation used by the original authors was maintained. The outgroup taxa were selected in accordance with our current knowledge on relationships within Orobanchaceae (Schneeweiss et al., 2004a; McNeal et al., 2013; Schneeweiss, 2013). Locality and voucher information including GenBank accession numbers are given in Supplementary Table S1.

2.2. DNA isolation and sequencing

Extraction of genomic DNA from newly collected silica-gel dried corolla tissue followed the CTAB-protocol (Doyle and Doyle, 1987). The majority of herbarium material was isolated either using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) or the NucleoSpin Plant II (Macherey-Nagel, Düren, Germany). Approximately 50 mg of dried material were homogenized using a mixer mill (Retsch MM200, Haan, Germany) at 30 Hz for 3 min followed by incubation in 700 μl extraction buffer for at least one hour at 65 °C. Thereafter, the CTABprotocol or the protocols supplied by the kit manufacturers were used.

Genomic DNA was stored at −80 °C and dilutions were used for later amplification. Three plastid markers (trnL-F, including the trnL intron and the intergenic spacer (IGS), trnS-FM IGS and psbA-trnH IGS), located in the large single-copy region (LSC) of the plastid genome, and the nuclear internal transcribed spacers (ITS1 and ITS2 plus the intervening 5.8S rDNA gene) were selected for phylogenetic inferences. The plastid markers have been repeatedly shown to be well-suited for phylogenetic studies at the species level (Borsch and Quandt, 2009). Additionally, psbA-trnH has been already used for DNA barcoding of Chinese Cistanche taxa (Han et al., 2010; Sun et al., 2012; Zheng et al., 2014). The nuclear ITS region was chosen as a technically readily amenable nuclear marker that despite several potential problems (Álvarez and Wendel, 2003) has been successfully applied in numerous phylogenetic studies with a focus on species relationships (Baldwin et al., 1995; Álvarez and Wendel, 2003; Bailey et al., 2003) including Orobanche and related genera (e.g., Schneeweiss et al., 2004a; McNeal et al., 2013). Furthermore, ITS2 has also been used for DNA barcoding in Cistanche (Han et al., 2010; Sun et al., 2012; Zheng et al., 2014; Wang et al., 2018). The trnL-F region was amplified using the primers C and F (Taberlet et al. 1991). In some cases, such as for old and supposedly degraded material, the region was amplified in two separate fragments, i.e. using primer pair C and D (Taberlet et al., 1991) and primer pair trnL460F (Worberg et al., 2007) and F (Taberlet et al., 1991), respectively. The trnS-trnfM IGS was amplified using trnS(UAG)-pF1 (5′- ACTATACCGGTTTTCAAGGCCG-3′) and trnfM(CAU)-pR1 (5′-TAG AGC AGTTTGGTAGCTCGCA-3′; S. Wicke, Münster, pers. comm.), the psbAtrnH IGS using the primers of Kress et al. (2005). The PCR profile for the plastid markers included an initial denaturation step of 5 min at 94 °C, followed by 30 cycles each with 1 min at 94 °C, 1 min at 55 °C, 90 s at 72 °C, and a final elongation step of 7 min at 72 °C. The ITS region was amplified using the primers ITS4 and ITS5 (White et al., 1990) with an amplification profile of 5 min at 94 °C followed by 40 cycles each with 1 min at 94 °C, 1 min at 48 °C with a time-increment of + 4 s per cycle, 45 s at 68 °C, and a final extension step of 7 min at 68 °C. Unsuccessful PCRs were repeated using internal primers 5.8S106-R (5′- AGGCGCA ACTTGCGTTCAAA -3′) and 5.8S32-F (5′-GCATCGATGAAGAACGT AGC-3′; D. l. Nickrent, Southern Illinois University, USA, pers. comm.) in combination with the respective external primers ITS5 and ITS4. PCR reactions were performed in a volume of 25 μl and included 1.5 UGoTaq Flexi DNA polymerase (Promega, Madison, USA), 0–0.2 M betaine monohydrate, 0.4 µM of each forward and reverse primer,0.15 mM dNTPs (Carl Roth, Karlsruhe, Germany), 1 mM MgCl2 in 1xGoTaq Flexi buffer, 1 μl genomic DNA of unknown concentration, and water. For some highly degraded herbarium material Ready-To-Go PCRBeads (Amersham-Pharmacia Biotech, Amersham, UK) were used following the manufacturer’s instructions. As commonly required for DNA-isolates of herbarium material, PCR additives such as 1 μl PVP-40(10–40%) and/or 5 μl enhancer solution P (5×; PeqLab, Erlangen, Germany) were added to the reactions at the expense of water. Amplification products were gel-purified on a 1% agarose gel using the thePeqLab PCR purification kit (Peqlab) or the quick PCR purification kit (Qiagen). For nested ITS-PCR products of < 300 bp (amplification from herbarium material, which often is contaminated with fungi), a higher gel concentration (1.4%) and longer run times were chosen. Cleaned PCR products were sequenced by Macrogen (Seoul, Korea)with the amplification primers and the additional internal primers mentioned above, where needed.

cistanche wirkung

2.3. Sequence alignment, indel coding, and phylogenetic analyses

DNA sequences were assembled and aligned using PhyDE 0.97 (Müller et al., 2006). Following Olsson et al. (2009), regions of uncertain homology (mutational hotspots) were annotated in PhyDE and removed from the analyses. Inversions were positionally isolated in the alignment and included as reverse-complement as suggested previously (Quandt et al., 2003; Borsch and Quandt; 2009). Indels were coded using simple indel coding (SIC; Simmons and Ochoterena, 2000) as implemented in SeqState 1.25 (Müller, 2005) and added as an additional data partition. In all data matrices, sequence gaps were treated as missing data and aligned positions were treated as equally weighted.

Analyses were conducted on three sequence data sets corresponding to concatenated plastid markers, ITS, and combined data (plastid and nuclear markers combined), each with or without indels, resulting in a total of six data sets being analyzed. Prior to the combined analyses, single marker data sets were screened for incongruences among markers via separate analysis of every single locus in MrBayes using the default settings (data not shown). Maximum parsimony analyses (MP) of the concatenated nucleotide matrix with and without the appended indel matrix were performed in PAUP* 4.0b10 (Swofford, 1999) using the parsimony ratchet (Nixon, 1999) via the command files generated by PRAP2 (Müller, 2004). The following ratchet settings were used: 200 iterations with 25% of the positions randomly up-weighted (weight = 2) in each replicate and 10 random addition cycles. Maximum likelihood (ML) analyses of the concatenated nucleotide matrix

with and without the appended indel matrix, were conducted using RAxML 8 (Stamatakis, 2014) employing the GTRCAT model. For both MP and ML analyses support was estimated via bootstrapping (Felsenstein, 1985) using 10,000 replicates. Bayesian inferences (BI), conducted on all six data sets (single and concatenated markers, with and without indels coded) were performed with MrBayes 3.2.2 (Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003) employing the GTR + Γ + I model for the nucleotide partition (two unlinked models in case of the combined data set) and the restriction site model (F81-like; Felsenstein, 1981) for the indel partition with default priors. Six runs with 10 million generations each and four chains each (one cold and three heated chains using the default heating) were run in parallel, sampling every 1000th generation. The first 25% were discarded as burn-in, which was well after chains had reached stationarity, as identified in Tracer v.1.7.1 (Rambaut et al., 2018). Trees were edited using TreeGraph2 (Stöver and Müller, 2010).

3. Results

Characteristics of the full data set, where no positions were removed, and of the data set, where hotspots (for details on those see Supplementary Table S2) have been removed (mononucleotide repeats) or reverted (hairpin associated inversions), are given in Table 1. Most indels in the dataset were repeats of adjacent fragments (i.e., simple sequence repeats [SSR]), ranging in length from 2 to 20 nucleotides (data not shown). Additionally, in the psbA-trnH spacer, a deletion of 1148 bp in the aligned matrix was observed in C. aff. fissa 2, corresponding to a reduction in sequence length to 20% of the average length.

We observed no significant incongruences among phylogenetic relationships inferred from single plastid markers (data not shown), but between concatenated plastid markers and the nuclear marker (Supplementary Figs. S1–S2). These were mostly found in the clade including, among others, C. phelypaea and Cistanche tubulosa (i.e., the Widespread Clade: see next paragraph), which was characterized by an overall shallow level of divergence (see phylogram in Fig. 1). There were no significant discrepancies between trees inferred from data sets with or without coding indels; however, using indel information increased the overall topological resolution, whereas its impact on support values was ambiguous (Fig. 1, Supplementary Figs. S1–S3).

The phylogenetic trees obtained from ML and BI based on combined plastid and nuclear markers, both with indels (ML log-likelihood score of −21272.719; harmonic mean of log-likelihood scores from BI of −21432.751) and without indels (ML log-likelihood score of −17135.239; harmonic mean of log-likelihood scores from BI of −17654.075), were strongly congruent and yielded a well-resolved tree (Fig. 1). In the following, we will focus on results from the combined analyses, referring to single-marker analyses (concatenated plastid versus nuclear) only in case of significant discrepancies. The combined analyses identified Cistanche as a monophyletic group, although only weakly supported (ML 61/ < 50; BI 0.77/0.91; support values from analyses with coded indels/without coded indels). In MP analysis, the position of C. Sinensis with respect to other Cistanche species and the outgroup taxa was unresolved (Supplementary Fig. S3). Cistanche fell into four major well-supported clades (Fig. 1, Supplementary Fig. S3), henceforth referred to as East Asian Clade, Southwest Asian Clade, Northwest African Clade, and Widespread Clade, respectively. The east-Asian Clade, containing only C. Sinensis (MP 100/100; ML 100/100; BI1.00/1.00) from China and Mongolia, was inferred as sister to the remainder of Cistanche (MP 100/100; ML 100/100; BI 1.00/1.00). TheSouthwest Asian Clade (MP 100/100; ML 100/100; BI 1.00/1.00), comprising several species from southwestern to central Asia, was sister to the clade (MP 100/100; ML 100/100; BI 1.00/1.00) including the northwest African Clade and the Widespread Clade. Whereas the northwest African Clade (MP 100/100; ML 100/100; BI 1.00/1.00)contained only one species restricted to northwestern Africa, the widespread Clade (MP 100/100; ML 100/100; BI 1.00/1.00) included about ten species jointly having a wide distribution from the Atlantic coasts of Europe and Africa to central Asia.

In the Southwest Asian Clade, the phylogenetic structure was pronounced (Fig. 1, Supplementary Fig. S3). Clade A (MP 98/100; ML 100/ 100; BI 1.00/1.00) included only C. ambigua, which fell into two geographically separate subclades, one (MP 79/100; ML 75/ < 50; BI 1.00/ < 50) containing exclusively accessions from northeastern Iran, the second (MP 75/51; ML 82/ < 50; BI 1.00/0.51) with some accessions from northern Iran as well as two accessions from southwestern Afghanistan. Clade A was sister (MP 70/52; ML 69/ < 50; BI 1.00/0.93) to C. ridgewayana from Afghanistan. Subsequent sister-group (MP 98/ 99; ML 100/100; BI 1.00/1.00) was clade B (MP 62/68; ML 75/79; BI 0.97/1.00) comprising the accessions of C. aff. ridgewayana. Within clade B phylogenetic structure corresponded to geography, because accessions from northwestern and central Iran (aff. ridgewayana 2) formed a subclade nested within a grade of accessions from mainly central and southern Iran (aff. ridgewayana 1). Relationships among the accessions of C. ridgewayana s. l. (C. ridgewayana and C. aff. ridgewayana) differed, however, among markers: Whereas C. ridgewayana s. l. was inferred as monophyletic without any supported internal structure by ITS data (Supplementary Fig. S2), it was inferred as paraphyletic by plastid data with, compared to the combined analysis, switched positions of C. aff. ridgewayana 1 and C. aff. ridgewayana 2. The subsequent sister group to the clade includes C. ambigua and C. ridgewayana s. l. (MP 74/60; ML 96/93; BI 1.00/1.00) was clade C (MP 100/100; ML 100/100; BI 1.00/1.00) containing C. aff. fissa 1 from Afghanistan. The subsequent sister groups were (i) the single accession of C. aff. fissa 2 from Azerbaijan (MP 57/100; ML 84/85; BI 0.88/0.57), (ii) clade D (MP 100/100; ML 100/100; BI 1.00/1.00) comprising C. salsa from southwestern Asia and from China (MP 100/100; ML 80/90; BI 0.99/0.99), and (iii) clade E (MP 100/100; ML 100/100; BI 1.00/ 1.00) containing Cistanche deserticola from China and Mongolia (MP 100/100; ML 100/100; BI 1.00/1.00).

Fig. 2. Characteristic flower and calyx shapes for selected members of the East Asian Clade (J), the Southwest Asian Clade (F–I), the Northwest African Clade (K), and the Widespread Clade (A–E, L–M) of Cistanche. (A) C. senegalensis, (B) C. flava, (C) C. tubulosa subsp. tubulosa, (D) C. violacea, (E) C. phelypaea subsp. nov., (F) C. fissa,(G) C. salsa, (H) C. ridgewayana, (I) C. ambigua, (J) C. Sinensis, (K) C. spec. nov., (L) C. rosea, (M) C. laxiflora.

In the Widespread Clade relationships were occasionally ambiguous due to low resolution of the nuclear data as well as discrepancies between plastid and nuclear markers (Supplementary Figs. S1–S2). These concerned not only the position of single accessions (e.g., C. Senegalese ED1096, Cistanche tubulosa subsp. tubulosa ED891) but also the circumscription of larger subclades. For instance, whereas plastid data inferred C. Rosea as a well-supported sister to the remaining species (BI 1.00; Supplementary Fig. S1), nuclear data suggested it as a member of a clade (BI 1.00) including all species except C. phelypaea, C. violacea, and most accessions of C. lutea (Supplementary Fig. S2). Consequently, relationships among major clades (clades F–J) inferred from the combined analyses were often poorly resolved and poorly supported (Fig. 1). The only exception was clade F (MP 100/100; ML 100/100; BI 1.00/1.00), containing C. rosea from the Arabian Peninsula (in the ITS analysis, the corresponding and only weakly supported clade additionally included accession ED1096 of C. senegalensis: Supplementary Fig. S2), whose sister-group relationship to the clade (MP 99/94; ML 98/99; PP 1.00/ 0.99) comprising the remaining species (clades G–J; Fig. 1) was well supported.

herba cistanches

Clade G (MP 63/99; ML 84/54; BI 1.00/ < 50) contained C. laxiflorafrom Iran and Afghanistan together with two accessions of C. tubules from China (GB1 and GB2). Within clade G, C. laxiflora was paraphyletic, because the subclade of the two C. tubulosa accessions (MP 98/65; ML 99/73; BI 1.00/0.77) grouped in a clade (MP 54/ < 50; ML 60/91; BI 1.00/1.00) with accessions of C. laxiflora from northern and southern Iran as well as Afghanistan (MP 85/ < 50; ML < 50/50; BI < 0.50/0.74) to the exclusion of accessions of C. laxiflora from central Iran forming a separate clade (MP 99/95; ML 99/95; BI 1.00/1.00).

Clade H (MP unsupported; ML 97/50; PP 1.00/ < 50) contained C. senegalensis (eastern Africa) and, nested therein, a monophyletic, yet unsupported, C. Flava (southwestern Asia) and a subclade (MP 90/73; ML 86/69; BI 0.79/0.86) comprising accessions of C. aff. tubulosa from the southern Arabian Peninsula (Oman, Yemen). Evidence for clade H came essentially from plastid data alone because in ITS C. Flava was inferred as closely related to C. laxiflora and relationships among C. senegalensis and C. tubulosa remained unresolved (Supplementary Fig. S2). In clade H, not only relationships of C. tubulosa and C. Flava to C. senegalensis were poorly resolved and poorly supported, but also those among lineages within C. senegalensis and within C. Flava, the only exception being a subclade of four C. Flava accessions (Flava subsp. nov.) from northern Iran (MP 100/97; ML 100/99; BI 1.00/1.00).

Clade I (MP unsupported; ML 62/54; PP 1.00/ < 0.50) contained exclusively accessions of C. tubulosa, covering a wide geographic range from the Cape Verde Islands and Mali via the Arabian Peninsula and Iran to Pakistan and India. Evidence for clade H also came essentially from plastid data alone, because in ITS this group was inferred as paraphyletic and often unresolved with respect to C. Rosea (clade F), C.laxiflora (clade G) as well as C. senegalensis and C. Flava (clade H). Phylogenetic structure in clade I was generally weak and showed no clear geographic pattern.

Clade J (MP 92/53; ML 97/87; PP 1.00/0.99) contained C. lutea s. l. (i.e., including C. lutea, C. aff. lutea 1 and C. aff. lutea 2; Macaronesian Islands and the Mediterranean) and, nested therein, C. Brunner (western Africa), C. phelypaea (Macaronesian Islands, Atlantic and Mediterranean coastal regions of Africa and southwestern Europe), and C. violacea (northern Africa). Cistanche phelypaea (MP 100/98; ML 100/ 100; BI 1.00/1.00) formed, together with a grade of C. lutea s. l. (accessions from all over this species’ distribution range) plus two accessions of C. violaceus (ED807 and ED1012 from Saudi Arabia and Jordan, respectively) and the hybrid of C. aff. lutea 1 and C. violacea, a weakly supported clade (MP 55/-; ML 76/67; BI 0.64/ < 0.50) that was a sister (MP 90/53; ML 97/92; BI 0.96/0.89) to a clade (MP 68/58; ML 93/82; BI 0.90/0.63) including all but the two previously mentioned accessions of C. violacea. In contrast to C. violacea, where a geographic separation into a widely distributed clade (MP 53/70; ML 84/85; BI 0.92/ 0.59) and an exclusively Moroccon clade (MP 92/93; ML 99/99; BI 1.00/0.95) was found, phylogenetic and geographic structure in C. lutea s. l. and C. phelypaea were limited. The remaining accessions of C. lutea (all from western Africa) plus C. brunneri (only a single accession included) constituted a grade at the base of clade J due to the position of the C. lutea accession from Niger (ED726) outside the clade (MP 94/60, ML 95/69; BI 1.00/0.99) formed by the other accessions. Inference of this C. lutea grade was likely due to discrepancies between plastid and ITS data. Specifically, from plastid data, these C. lutea accessions plus C. brunneri were inferred as closely related to C. violacea and other C. lutea accessions (Supplementary Fig. S1), i.e., as in the combined data, whereas ITS data placed those, in an unresolved position, in a clade with C. senegalensis and C. tubulosa to the exclusion of other members of clade J (Supplementary Fig. S2).

4. Discussion

In the latest monograph of the genus Cistanche, Beck-Mannagetta (1930) distinguished four groups (as taxonomic sections) differing in calyx features and the number of bracteoles. With the exception of the monospecific C. sect. Substance (East Asian Clade), none of Beck Mannagetta’s sections are supported as monophyletic. Instead, species of C. sect. Cistanche and C. sect. Heterocalyx is intermixed (in the southwest Asian Clade and the Widespread Clade) and the single species of C. sect. Cistanchella is paraphyletic and nested in the SouthwestAsian Clade. A fourth major lineage (Northwest African Clade), represented here by a species still to be described that is morphologically close to C. mauritanica, traditionally placed in the C. sect. Cistanche has not been identified in any of the previous classifications

The four major lineages identified here can also be characterized morphologically. Specifically, the sole species of the East Asian Clade, C. Sinensis, is the only Cistanche species with a deeply incised quadripartite calyx, whereas all others have (at least) five mostly connate calyx lobes (Fig. 2). Species of the Widespread Clade have glabrous stems, bracts, and calyces, whereas those of the Southwest Asian and the Northwest African Clade are at least partially hairy (woolly or lanuginose versus arachnoid-lanuginose, respectively) (Fig. 2). Finally, members of the Northwest African Clade have a densely arachnoid-lanuginose indumentum and broadly rhomboid bracts not found in the Southwest Asian Clade (Fig. 2). Previously emphasized characters, i.e., the number of bracteoles (one or none in C. sect. Cistanchella versus two in all other sections) and the shape of the calyx lobes (more or less deeply separated and unequal lobes in C. sect. Heterocalyx versus shallowly separated and equal lobes in C. sect. Cistanche), do not reflect deeper splits and, at least in case of the shape of the calyx lobes, appear to have evolved at least twice independently.

Several of the currently recognized species are not supported as monophyletic (Fig. 1), which is not restricted to taxa known to be taxonomically difficult, such as C. lutea or C. tubulosa, but also to taxa considered taxonomically unproblematic, such as C. ridgewayana s. l. or Cistanche fissa s. l. (here represented by C. aff. fissa 1 and 2). The discrepancy between taxonomic and phylogenetic units may be due to several, not mutually exclusive factors, including species misidentification, lack of taxonomic recognition of morphologically and/or geographically differentiated entities, hybridization, and shallow divergence (rapid radiation). Species misidentification is likely the cause for the position of two accessions of C. tubulosa (GB1 and GB2) nested in C. laxiflora (Fig. 1). The vouchers of these accessions were unavailable to us for revision, but, as judged from web images, plants from adjacent areas in the central deserts of China morphologically agree with C. laxiflora subsp. laxiflora (white corolla tubes with light purple lobes instead of pale to deep yellow in C. tubulosa). This species is not reported from China, where plants pertaining to this taxon are either listed under C. tubulosa (Zhang, 1990) or, more recently, under C. mongolica (Zhang and Tzvelev, 1998). Lack of taxonomic recognition of morphologically and/or geographically differentiated lineages appears to apply both to paraphyletic species, like C. ridgewayana s. l. and C. aff. fissa, as well as to monophyletic species with strong phylogenetic structure, such as C. Flava. In each of these cases, phylogenetically circumscribed groups differ morphologically and geographically and, therefore, deserve taxonomic recognition (a detailed taxonomic treatment will be given in a subsequent publication). Both, species misidentification and insufficient taxonomic resolution are common problems in holoparasitic Orobanchaceae, in particular in Orobanche and related genera (Manen et al., 2004; Schneeweiss et al., 2004a; Schneeweiss et al., 2009; Schneeweiss, 2013).

Hybridization is a common phenomenon in flowering plants (Rieseberg and Carney, 1998; Payseur and Rieseberg, 2016) and has also been reported for Cistanche. Specifically, Beck-Mannagetta (1930) described hybrids between C. lutea (as C. tinctoria f. lutea) and C. violacea as nothospecies, C. hybrid. These hybrids may be morphologically intermediate, as is the case for accession ED686, or may resemble one of the parents, as is the case for accession ED1012; this latter accession has been originally determined as C. violacea, but on re-examination, provoked by its phylogenetic position within C. lutea s. l., turned out to be a hybrid as well. Assuming maternal inheritance of the plastid genome (as shown for the Orobanchaceae Rhinanthus angustifolius: Vrancken and Wesselingh, 2010) in Cistanche, the presence of the maternal ITS ribotype in the hybrid accession ED686 (it's not available for accession ED1012) would be consistent with the formation of later generation hybrids and/or backcrosses, resulting in plastid capture. Plastid capture could also explain discrepancies between nuclear and plastid markers. For instance, C. aff. tubulosa from the Arabian Peninsula, morphologically differing in several characteristics from widespread C. tubulosa, formed a clade close to C. senegalensis in the ITS data set but was separated into two different clades in the plastid and the combined data set grouping with C. senegalensis (clade H) or accessions of C. tubulosa subsp. tubulosa (clade I; both C. senegalensis andC. tubulosa subsp. tubulosa also occur on the Arabian Peninsula).

An alternative to introgression for explaining discrepancies between phylogenetic positions inferred from nuclear versus plastid data is incomplete lineage sorting, which will affect nuclear and plastid genomes differently due to differences in their effective population sizes (in plastid genomes only half as large as in nuclear genomes in a monoecious group such as Cistanche). Incomplete lineage sorting is expected to be particularly relevant in the case of shallow divergence and rapid radiation (Maddison and Knowles, 2006), which apparently is the case in the Widespread Clade, where the clade comprising Clades G to J is characterized by short branch lengths (Fig. 1). Incomplete lineage sorting may be responsible for the lack of molecular differentiation and/or coherence of morphologically differentiated lineages within C. lutea s. l. (i.e., C. aff. lutea 1 and C. aff. lutea 2). Disentangling incomplete lineage sorting from introgression will require coalescent-based methods (Blanco-Pastor et al., 2012) ideally involving nuclear phylogenomic data (Bravo et al., 2019), which in the absence of solid taxonomic framework and sufficiently extensive data are not possible yet.

Host specialization is an important evolutionary force in holoparasitic Orobanchaceae (Schneider et al., 2016). Cistanche, however, appears to be an exception, as many species have been found in members of several plant families, most frequently Amaranthaceae (incl. Chenopodiaceae) and Polygonaceae, but also Fabaceae, Zygophyllaceae, Tamaricaceae, Rosaceae, Nitrariaceae, and Salvadoraceae. As in Orobanche (Manen et al., 2004), the same host may be shared by several unrelated Cistanche species (e.g., Haloxylon ammodendron is the host for C. deserticola from the Southwest Asian Clade and for C. phelypaea from the Widespread Clade). Cases of putative host specialization (e.g., C. senegalensis mostly on Acacia/Fabaceae; mainly Fabaceae as hosts for C. Flava subsp. nov. versus mainly Calligonum bungei/Polygonaceae for C. Flava subsp. Flava) may reflect the abundance of suitable host species rather than genuine host specialization. Testing any role of host specialization for the diversification of Cistanche will require a well-resolved species phylogeny and sufficiently detailed host data.5. Conclusions

This is the first comprehensive molecular phylogenetic analysis of the holoparasitic plant genus Cistanche, widely distributed in arid regions of the Old World. Four major clades within Cistanche have been identified that only partially correspond to traditionally recognized sections (BeckMannagetta, 1930) and generally show a strong geographic component. Whereas the Southwest Asian Clade exhibits strong phylogenetic structure among and partly within species (the East Asian and the Northwest African Clade are monospecific), phylogenetic resolution within the Widespread Clade is often low and hampered by discrepancies between nuclear and plastid markers, which at least partially is due to hybridization/introgression and/or incomplete lineage sorting. Molecular phylogenetic evidence and results of a morphological re-evaluation of Cistanche species indicate that species diversity in Cistanche is currently underestimated (see taxa indicated as sp. nov. or subsp. nov. in the text). Although some species still have to be included in any molecular phylogenetic study (e.g., C. mauritanica) and additional markers will be needed to resolve all species relationships, the sound phylogenetic hypotheses presented here provide a valuable basis for ongoing cytogenetic, taxonomic, and biogeographic research in this genus.

cistanche bienfaits

Funding

This work has been partially supported by SYNTHESYS financed by the European Community Research Infrastructure Action under the FP7 “Capacities” Program at the Real Jardín Botánico (ES-TAF-1663). We would like to thank the University of Bonn, the OeAD (Österreichische Austauschdienst) and DAAD (Deutscher Akademischer Austauschdienst) for financial support.

CRediT authorship contribution statement

Najibeh Ataei: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Project administration, Visualization, Writing - original draft, Writing - review & editing.Gerald M. Schneeweiss: Conceptualization, Supervision, Writing -review & editing. Miguel Angel García: Resources. Michael Krug: Formal analysis. Marcus Lehnert: Writing - original draft. JafarValizadeh: Resources. Dietmar Quandt: Conceptualization, Formal analysis, Project administration, Resources, Supervision, Writing - review & editing

Acknowledgments

Thanks go to Edit Korpinos, who provided help at the Jodrell lab in Kew and to the herbarium curators (TARI, BM, IRAN, BONN, USB, P, TUH, E, W, KAS, MSB, K, B, G, PEY, M, UG, BR, GUH, MA, WU) for sending out loans and photos. We thank Susann Wicke for primer design and lab support. We appreciate Dr. Hossein Akhani at TehranUniversity and Mrs. Robabeh Shahi Shavvon at Gilan University for providing some DNA material from Iran. We are grateful to Dr. FedericoLuebert and Juliana Chacon at Nees Institute for their helpful comments.

References:

Álvarez, I., Wendel, J.F., 2003. Ribosomal ITS sequences and plant phylogenetic inference. Mol. Phylogenet. Evol. 29, 417–434. https://doi.org/10.1016/S1055-7903(03)00208-2.
Bailey, C.D., Carr, T.G., Harris, S.A., Hughes, C.E., 2003. Characterization of angiosperm nrDNA polymorphism, paralogy, and pseudogenes. Mol. Phylogenet. Evol. 29, 435–455.
Baldwin, B.G., Sanderson, M.J., Porter, J.M., Wojciechowski, M.F., Campbell, C.S., Donoghue, M.J., 1995. The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Ann. Missouri Bot. Gard. 82, 247–277.
Beck-Mannagetta, G., 1930. Orobanchaceae, in: Engler, A. (Ed.), Das Pflanzenreich IV. 261. Wilhelm Engelmann, Leipzig, pp. 1–348.
Bennett, J.R., Mathews, S., 2006. Phylogeny of the parasitic plant family Orobanchaceae inferred from phytochrome A. Am. J. Bot. 93, 1039–1051.
Blanco-Pastor, J.L., Vargas, P., Pfeil, B.E., 2012. Coalescent simulations reveal hybridization and incomplete lineage sorting in Mediterranean Linaria. PLoS ONE 7 (6), e39089.
Borsch, T., Quandt, D., 2009. Mutational dynamics and phylogenetic utility of noncoding chloroplast DNA. Plant Syst. Evol. 282, 169–199.

Bravo, G.A., Antonelli, A., Bacon, C.D., Bartoszek, K., Blom, M.P.K., Huynh, S., Jones, G., Knowles, L.L., Lamichhaney, S., Marcussen, T., Morlon, H., Nakhleh, L.K., Oxelman,

B., Pfeil, B., Schliep, A., Wahlberg, N., Werneck, F.P., Wiedenhoeft, J., WillowsMunro, S., Edwards, S.V., 2019. Embracing heterogeneity: coalescing the Tree of Life and the future of phylogenomics. PeerJ 7, e6399.

Doyle, J.J., Doyle, J., 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19, 11–15.
N. Ataei, et al. Molecular Phylogenetics and Evolution 151 (2020) 106898 8
Felsenstein, J., 1981. Evolutionary trees from DNA sequences: a maximum likelihood approach. J. Mol. Evol. 17, 368–376.
Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783–791.
Fu, W., Liu, X., Zhang, N., Song, Z., Zhang, W., Yang, J., Wang, Y., 2017. Testing the hypothesis of multiple origins of holoparasitic in Orobanchaceae: phylogenetic evidence from the last two unplaced holoparasitic genera, Gleadovia and
Phacellanthus. Front. Plant Sci. 8, 1380.
Han, J.-P., Song, J.-Y., Liu, C., Chen, J., Qian, J., Zhu, Y.-J., Shi, L.-C., Yao, H., Chen, S.-L., 2010. Identification of Cistanche species (Orobanchaceae) based on sequences of the plastid psbA-trnH intergenic region. Acta Pharmaceut. Sin. 45, 126–130.
Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754–755.
Kress, W.J., Wurdack, K.J., Zimmer, E.A., Weigt, L.A., Janzen, D.H., 2005. Use of DNA barcodes to identify flowering plants. Proc. Natl. Acad. Sci. U.S.A. 102, 8369–8374.

Li, X., Feng, T., Randle, C., Schneeweiss, G.M., 2019. Phylogenetic relationships in Orobanchaceae inferred from low-copy nuclear genes: consolidation of major clades and identification of a novel position of the non-photosynthetic Orobanche clade sister to all other parasitic Orobanchaceae. Front. Plant Sci. 10, 902.

Li, X., Jang, T.-S., Temsch, E.M., Kato, H., Takayama, K., Schneeweiss, G.M., 2017. Molecular and karyological data confirm that the enigmatic genus Platypholis from Bonin-Islands (SE Japan) is phylogenetically nested within Orobanche
(Orobanchaceae). J. Plant. Res. 130, 273–280.
Li, X., Zhang, T.-C., Qiao, Q., Ren, Z., Zhao, J., Yonezawa, T., Hasegawa, M., Crabbe, M.J.C., Li, J., Zhong, Y., 2013. Complete chloroplast genome sequence of holoparasite Cistanche deserticola (Orobanchaceae) reveals gene loss and horizontal gene
transfer from its host Haloxylon ammodendron (Chenopodiaceae). PLoS ONE 8, e58747.
Liu, X., Fu, W., Tang, Y., Zhang, W., Song, Z., Li, L., Yang, J., Ma, H., Yang, J., Zhou, C.,
Davis, C.C., Wang, Y., 2020. Diverse trajectories of plastome degradation in holoparasitic Cistanche and genomic location of the lost plastid genes. J. Exp. Bot. 71, 877–892.
Maddison, W., Knowles, L.L., 2006. Inferring phylogeny despite incomplete lineage sorting. Syst. Biol. 55, 21–30.
Manen, J.-F., Habashi, C., Jeanmonod, D., Park, J.-M., Schneeweiss, G.M., 2004. Phylogeny and intraspecific variability of holoparasitic Orobanche (Orobanchaceae) inferred from plastid rbcL sequences. Mol. Phylogenet. Evol. 33, 482–500.
McNeal, J.R., Bennett, J.R., Wolfe, A.D., Mathews, S., 2013. Phylogeny and origins of holoparasitic in Orobanchaceae. Am. J. Bot. 100, 971–983.
Müller, J., Müller, K.F., Neinhuis, C., Quandt, D., 2006. PhyDE – Phylogenetic Data Editor. http://www.phyde.de (accessed 2. April 2019).
Müller, K., 2004. PRAP-computation of Bremer support for large data sets. Mol. Phylogenet. Evol. 31, 780–782.
Müller, K., 2005. SeqState: primer design and sequence statistics for phylogenetic DNA datasets. Appl. Bioinform. 4, 65–69.
Nixon, K.C., 1999. The parsimony ratchet is a new method for rapid parsimony analysis. Cladistics 15, 407–414.
Olsson, S., Buchbender, V., Enroth, J., Hedenäs, L., Huttunen, S., Quandt, D., 2009. Phylogenetic analyses reveal high levels of polyphyly among pleurocarpous lineages as well as novel clades. Bryologist 112, 447–466.
Park, J.-M., Manen, J.-F., Colwell, A.E., Schneeweiss, G.M., 2008. A plastid gene phylogeny of the non-photosynthetic parasitic Orobanche (Orobanchaceae) and related genera. J. Plant Res. 121, 365–376.
Payseur, B.A., Rieseberg, L.H., 2016. A genomic perspective on hybridization and speciation. Mol. Ecol. 25, 2337–2360.
Quandt, D., Müller, K., Huttunen, S., 2003. Characterization of the chloroplast DNA psbTH region and the influence of dyad symmetrical elements on phylogenetic reconstructions. Plant Biol. 5, 400–410. https://doi.org/10.1055/s-2003-42715.
Rambaut, A., Drummond, A.J., Xie, D., Baele, G., Suchard, M.A., 2018. Posterior summarisation in Bayesian phylogenetics using Tracer 1.7. Syst. Biol. 67, 901–904.

Rieseberg, L.H., Carney, S.E., 1998. Plant hybridization. New Phytol. 140, 599–624.

Rodrigues, A.G., Colwell, A.E.L., Stefanovic, S., 2011. Molecular systematics of the parasitic genus Conopholis (Orobanchaceae) inferred from plastid and nuclear sequences. Am. J. Bot. 98, 896–908.
Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572–1574.
Schneeweiss, G.M., 2013. Phylogenetic relationships and evolutionary trends in Orobanchaceae, in Joel, D.M., Gressel, J., Musselman, L.J. (Eds.), Parasitic Orobanchaceae. Springer, Berlin Heidelberg, pp. 243–265.
Schneeweiss, G.M., Colwell, A., Park, J.-M., Jang, C.-G., Stuessy, T.F., 2004a. Phylogeny of holoparasitic Orobanche (Orobanchaceae) inferred from nuclear ITS sequences. Mol. Phylogenet. Evol. 30, 465–478.
Schneeweiss, G.M., Frajman, B., Dakskobler, I., 2009. Orobanche lycoctoni Rhiner (Orobanchaceae), a poorly known species of the Central European flora. Candollea 64, 91–99.
Schneeweiss, G.M., Palomeque, T., Colwell, A.E., Weiss-Schneeweiss, H., 2004b. Chromosome numbers and karyotype evolution in holoparasitic Orobanche (Orobanchaceae) and related genera. Am. J. Bot. 91, 439–448.
Schneider, A.C., Colwell, A.E.L., Schneeweiss, G.M., Baldwin, B.G., 2016. Cryptic host-specific diversity among western hemisphere broomrapes (Orobanche s.l., Orobanchaceae). Ann. Bot. 118, 1101–1111.
Simmons, M.P., Ochoterena, H., 2000. Gaps as characters in sequence-based phylogenetic analyses. Syst. Biol. 49, 369–381.
Stamatakis, A., 2014. RAxML Version 8: A tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313.
Stöver, B.C., Müller, K.F., 2010. TreeGraph 2: combining and visualizing evidence from different phylogenetic analyses. BMC Bioinf. 11, 7.
Sun, Z.Y., Song, J.Y., Yao, H., Han, J.P., 2012. Molecular identification of Cistanches Herba and its adulterants based on nrITS2 sequence. J. Med. Plant Res. 6, 1041–1045.

Swofford, D.L., 1999. PAUP*4.0 Beta for Macintosh: Phylogenetic Analysis Using Parsimony. Sinauer Associates, Sunderland, MA.

Taberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Mol. Biol. 17, 1105–1109.Tomari, N., Ishizuka, Y., Moriya, A., Kojima, S., Deyama, T., Coskun, M., Tu, P.,

Mizukami, H., 2003. Pharmacognostical studies of Cistanchis Herba (IV) phylogenetic relationship of the Cistanche plants based on plastid rps2 gene and rpl16-rpl14 intergenic spacer region sequence. Natur. Med. 57, 233–237.
Tomari, N., Ishizuka, Y., Moriya, A., Kojima, S., Deyama, T., Mizukami, H., Tu, P., 2002. Pharmacognostical studies of Cistanchis Herba (III) phylogenetic relationship of the Cistanche plants based on plastid rps2 gene and rpl16-rpl14 intergenic spacer sequences. Biol. Pharm. Bull. 25, 218–222.
Vrancken, J., Wesselingh, R.A., 2010. Inheritance of the chloroplast genome in Rhinanthus angustifolius (Orobanchaceae). Plant Ecol. Evol. 143, 239–242.
Wang, X.-Y., Xu, R., Chen, J., Song, J.-Y., Newmaster, S.G., Han, J.-P., Zhang, Z., Chen, S.-L., 2018. Detection of Cistanches Herba (Rou Cong Rong) medicinal products using species-specific nucleotide signatures. Front. Plant Sci. 9, 1643.

Weiss-Schneeweiss, H., Greilhuber, J., Schneeweiss, G.M., 2006. Genome size evolution in holoparasitic Orobanche (Orobanchaceae) and related genera. Am. J. Bot. 93, 148–156.

Westwood, J.H., Yoder, J.I., Timko, M.P., dePamphilis, C.W., 2010. The evolution of parasitism in plants. Trends Plant Sci. 15, 227–235.

White, T., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis, M., Gelfand, D., Shinsky, J., White, T. (Eds.), PCR Protocols: A Guide to Methods and Applications. Academic

Press, London, pp. 315–322.
Wicke, S., 2013. Genomic evolution in Orobanchaceae. In: Joel, D.M., Gressel, J., Musselman, L.J. (Eds.), Parasitic Orobanchaceae. Springer, Berlin Heidelberg, pp. 267–286.
Wicke, S., Müller, K.F., dePamphilis, C.W., Quandt, D., Wickett, N.J., Zhang, Y., Renner,
S.S., Schneeweiss, G.M., 2013. Mechanisms of functional and physical genome reduction in photosynthetic and nonphotosynthetic parasitic plants of the broomrape family. Plant Cell 25, 3711–3725.
Wicke, S., Müller, K.F., dePamphilis, C.W., Quandt, D., Bellot, S., Schneeweiss, G.M., 2016. Mechanistic model of evolutionary rate variation en route to a nonphotosynthetic lifestyle in plants. Proc. Natl. Acad. Sci. U.S.A. 113, 9045–9050.

Wickett, N.J., Honaas, L.A., Wafula, E.K., Das, M., Huang, K., Wu, B., Landherr, L., Timko, M.P., Yoder, J., Westwood, J.H., dePamphilis, C.W., 2011. Transcriptomes of the parasitic plant family Orobanchaceae reveal surprising conservation of chlorophyll

synthesis. Curr. Biol. 21, 2098–2104.
Wolfe, A.D., Randle, C.P., Liu, L., Steiner, K.E., 2005. Phylogeny and biogeography of Orobanchaceae. Folia Geobot. 40, 115–134.
Worberg, A., Quandt, D., Barniske, A.-M., Löhne, C., Hilu, K.W., Borsch, T., 2007. Phylogeny of basal eudicots: Insights from non-coding and rapidly evolving DNA. Organ. Divers. Evol. 7, 55–77.
Young, N.D., Steiner, K.E., dePamphilis, C.W., 1999. The evolution of parasitism in Scrophulariaceae/Orobanchaceae: plastid gene sequences refute an evolutionary transition series. Ann. Missouri Bot. Gard. 86, 876–893.
Zhang, Z.Y., 1990. Orobanchaceae, in: Wang, W. (ed.), Flora Reipublicae Popularis Sinica, vol. 69. Bejing, Science Press, pp. 69–124.
Zhang, Z.Y., Tzvelev, N.N., 1998. Orobanchaceae, in: Wu, Z.Y., Raven, P.H. (Eds.), Flora of China, vol. 18 (Scrophulariaceae through Gesneriaceae). Science Press, Beijing, and Missouri Botanical Garden Press, St. Louis, MI, pp. 229–243.
Zheng, S., Jiang, X., Wu, L., Wang, Z., Huang, L., 2014. Chemical and genetic discrimination of Cistanches Herba based on UPLC-QTOF/MS and DNA barcoding. PLoS ONE 9 (5), e98061.