We consulted the specimen records of Z. nitidum (including two varieties) available in the Chinese Virtual Herbarium (CVH, http://www.cvh.ac.cn/) and also examined all specimens of this species deposited in GXMI of Guangxi Institute of Chinese Medicine and Pharmaceutical Sciences and IBSC of South China Botanical Garden, CAS. Then, we conducted extensive sampling across its whole distribution range in China according to the consulted specimen records to collect enough samples for genetic analyses. All sampled individuals of the same population were spaced more than 50 m apart. Fresh leaves were collected for DNA sequencing and immediately placed in plastic sealed bags with sufficient silica gel to avoid DNA degradation. We also took photos of the trunks, leaves, branchlets and inflorescence of each individual for initial identification. We identified samples of this monophyletic species based on three taxonomic criteria (Figs 1, 2). All voucher specimens and dried leaves for DNA extraction were deposited in GXMI. Geographical information for each location was recorded using a HOLUX M-241 GPS Track instrument (Technology, Inc., Taiwan, China). Finally, 420 individuals from 38 natural populations across the entire distribution in China were collected for genetic analyses (Table 1).

Figure 1. 

Geographical distribution of 38 Z. nitidum populations in China and its taxonomic scenarios currently used based on the presence of hairy or nearly glabrous leaf and rachis in Zhang et al. (2008) (a), the description of the Editorial Committee of Flora (1997) (b), and trunk appearance in Qin et al. (2020) (c).

The genomic DNA of all samples was extracted from approximately 15 mg of silica gel-dried leaves using the CTAB method (Doyle and Doyle 1990). The quality and concentration of genomic DNA were measured using a NanoDrop 1000 spectrophotometer. Qualified DNA was defined as having values of both 260/280 and 260/230 located in the range 1.6–2.1. Then, each DNA solution was diluted or concentrated to approximately 50 ng/µL for PCR amplification.

Figure 2. 

Maximum likelihood (ML) tree of Zanthoxylum species based on nuclear internal transcribed spacer (ITS) sequences.

Owing to its high discrimination power, nuclear internal transcribed spacer (ITS) sequence fragments are frequently used for plant classification (CBOL Plant Working Group 2011; Pang et al. 2011; Lu et al. 2016), and have been proposed as core barcodes (CBOL Plant Working Group 2011). As core chloroplast barcodes (CBOL Plant Working Group 2009), matK and rbcL are widely used for most seed plants (Levin et al. 2003; Kress and Erickson 2007; Fazekas et al. 2008; Kress et al. 2009; Su et al. 2015; Amandita et al. 2019). Other chloroplast fragments (e.g., ycf6-trnC and petA-psbJ) have also been used for phylogenetic studies (Shaw et al. 2007; Liu et al. 2012; Sokoloff and Gillespie 2012). Therefore, we used universal primers for the nuclear ITS fragments and four chloroplast DNA fragments (matK, rbcL, ycf6-trnC, and petA-psbJ) to amplify and sequence each of the qualified DNA samples (Table 2). The primers were selected based on information from previous studies (White et al. 1990; Kress and Erickson 2007; Shaw et al. 2007; Kress et al. 2009; Liu et al. 2012; Amandita et al. 2019). PCR amplification was performed in a 50 µL reaction system with 2 µL of the prepared DNA, 5 µL of 10 × PCR buffer, 0.8 µL of dNTPs (2.5 mmol/ml), 4 µL of the primer, 0.4 µL of rTaq polymerase, and 41.5 µL of ddH₂O. The reaction was programed according to the following procedure: initial template denaturation at 94 °C for 4 min, followed by 38 cycles of 94 °C for 40 s, 48 °C (petA-psbJ), 52 °C (matK), 55 °C (rbcL), 59 °C (ycf6-trnC) or 60 °C (ITS) for 45 s, and 72 °C for 1.5 min, and final extension at 72 °C for 10 min.

All PCR products were examined by agarose gel electrophoresis and photographed using a UV transilluminator. The products were purified using a TIANquick Midi Purification Kit (TIANGEN, Beijing, China) and the reaction mixtures were sequenced on an ABI 3130xl automated sequencer (Applied Biosystems, Foster City, California, USA). For DNA products that failed direct sequencing, we repeated the extraction, amplification, or sequencing experiment at least three times.

We aligned and manually corrected nuclear ITS sequences and chloroplast sequences using MEGA 5 (Tamura et al. 2011). We first analyzed the haplotype relationship using NETWORK 10.0, based on the Median-Join method, for single-barcode and four-chloroplast fragments combined (Bandelt et al. 1999). Owing to the presence of some heterozygous sites in the nuclear ITS dataset, we phased these sequences using DnaSP v5 with default parameters (Librado and Rozas 2009). In addition, we constructed an ML tree of the Zanthoxylum species based on ITS sequences downloaded from NCBI (https://www.ncbi.nlm.nih.gov/), using MEGA 5 (Tamura et al. 2011) to further demonstrate the distinctiveness of Z. nitidum to the congeneric species. We further identified operational taxonomic units (OTUs) based on DNA barcodes using the Automatic Barcode Gap Discovery (ABGD) method (Puillandre et al. 2012). Indels or heterozygous sites of each fragment were deleted. The ABGD analysis was conducted based on information available online (https://bioinfo.mnhn.fr/abi/public/asap/asapweb.html). We used the Kimura (K80) ts/tv model and set X = 1.5 as the relative barcode gap width. The remaining parameters were retained as the default parameters.

After genetic classification, we used the barcodes of each nuclear cluster or chloroplast cluster as references for subsequent molecular identification. The Liang-Mian-Zhen used by China Resources Sanjiu was treated as the standard reference because it has a rapid growth rate, a high level of nitidine chloride (Han et al. 2013; Qin et al. 2019), and is widely cultivated. We then visited 10 indigenous individuals who sold Liang-Mian-Zhen in different folk markets in Hainan, Guangxi, and Guangdong, with 1–4 samples obtained from each. We collected only a small number of broken leaves and voucher specimens without the trunk from both China Resources Sanjiu and the indigenous individuals, thereby making visual identification difficult. In total, 20 Liang-Mian-Zhen samples, including 3 standard samples and 17 samples of folk medicinal materials, were used for molecular classification and identification (Table 3). Total DNA was extracted, amplified, and sequenced according to the methods described above. For rapid identification, we directly used the aligned consistent sequences to construct ML trees using MEGA 5 (Tamura et al. 2011). The closely related species Zanthoxylum armatum was chosen as an outgroup for phylogenetic analyses within Z. nitidum (Accession numbers: MW362848 and MN080708; Chen et al. 2013; Appelhans et al. 2018). Finally, we classified and identified the samples using the nuclear ITS barcode and four chloroplast barcodes.

We obtained 269 ITS sequences from 420 individuals across 38 natural populations of Z. nitidum in China (Fig. 1, Table 1). We recovered 31 different ITS types, and the aligned sequences were 658 bp in length (Table 1). Four genetic clusters (A, B, C, and D) were detected based on the network analysis of these ITS sequences (Fig. 3a). Cluster D was widely distributed across Guangxi, Guangdong, Zhejiang, Fujian, and Taiwan. Cluster C was distributed in North and Southwest Guangxi. Cluster B was narrowly distributed in northeastern Guangxi. Cluster A was widely distributed across Yunnan, Guangxi, Hainan, and Guangdong, and its distribution partially overlapped with that of clusters C and D (Figs 1c, 3a). These genetic classification results are consistent with the third taxonomic scenario based on trunk appearance (Fig. 1c). Phylogenetic analysis confirmed that Liang-Mian-Zhen formed a clade distinct to congeners as the downloaded ITS sequences covered approximately a quarter of the Zanthoxylum species (Fig. 2).

Figure 3. 

Network analysis of each barcode. Circle size is proportional to the haplotype frequency. Different colors indicate different clusters. The four clusters correspond to four morphological clusters in Fig. 1c.

After sequencing all four chloroplast barcodes (matK, rbcL, ycf6-trnC, and petA-psbJ), the aligned sequences of these barcodes were 703, 533, 600, and 1040 bp in length, respectively. In total, 33 substitutions were detected across the 4 chloroplast barcodes. Only 8 cpDNA haplotypes were recovered from the 420 individuals across 38 natural populations, and these haplotypes were divided into 5 clusters (A1, A2, B, C, and D; Fig. 4, Table 1). All five chloroplast clusters were distributed across different geographical regions. Chloroplast clusters B, C, and D corresponded to the three nuclear clusters B, C, and D, respectively. However, two different chloroplast clusters (A1 and A2) were found within A. A1 was mainly found in Hainan and partially extended to southern Guangdong. In addition, it was also found in a preserved individual (coded as pop 1 in table 1) in the Xishuangbanna Tropical Botanical Garden (XTBG; Menglun, Yunnan). A2 was found in the remaining populations within A. We did not find distinct morphological differences between A1 and A2 from voucher specimens based on leaf shape and size, trunk appearance, and flowers. Accession numbers of all nuclear and chloroplast sequences are from MZ769312 to MZ769362 and from MZ773282 to MZ773393, respectively.

Figure 4. 

Distribution and network of five chloroplast lineages recovered from Zanthoxylum nitidum in China. Dotted lines of different colors indicate different nuclear clusters. Circle size is proportional to the haplotype frequency. Different colored circles indicate different chloroplast clusters.

Additionally, the four OTUs recognized by the ABGD analysis were consistent with four clusters based on the network analysis (Figs 3a, 5), the results of which were highly inconsistent with the two past taxonomic scenarios (Fig. 1a, b) but highly consistent with the recent taxonomic scenario based on trunk appearance (Fig. 1c). For the combined four-chloroplast fragments, the five unveiled chloroplast clusters based on the network analysis were also revealed by the ABGD method (Figs 4, 6). The single chloroplast barcode, rbcL, produced the same result as the four fragments combined (Figs 3b, 4). The matK gene only distinguished C from a group comprising the remaining nuclear or chloroplast clusters. Additionally, the other two chloroplast barcodes (ycf6-trnC and petA-psbJ) similarly classified this species into two groups: one represented by B, and the other comprising the remaining clusters.

Figure 5. 

Genetic classification based on nuclear internal transcribed spacer (ITS) sequences using the Automatic Barcode Gap Discovery (ABGD) method.

Figure 6. 

Genetic classification based on four chloroplast barcodes combined using the Automatic Barcode Gap Discovery (ABGD) method.

Molecular identification was determined via phylogenetic analyses. During identification, each sample was delimited into the closest cluster. Four nuclear clusters (A, B, C, and D) and five chloroplast clusters (A1, A2, B, C, and D) were strongly supported by phylogenetic analyses (Fig. 7). The three standard Liang-Mian-Zhen samples from China Resources Sanjiu were identified as cluster A based on the nuclear ITS barcode and as A2 based on the four chloroplast fragments. However, for the 17 folk medicinal samples, three clusters (A, B, and D) were identified based on the nuclear ITS barcode. Four chloroplast clusters (A1, A2, B, and D) were identified based on the four chloroplast fragments. Interestingly, the molecular identification results for the folk medicinal samples were related to the natural distribution of Z. nitidum based on nuclear and chloroplast barcodes (Figs 1, 7, Table 1). For example, two folk Liang-Mian-Zhen samples from Jinxiu (Guangxi) were identified as cluster B based on their nuclear ITS sequences and cluster B based on four chloroplast fragments. Similarly, folk Liang-Mian-Zhen samples collected from Guangdong were assigned to nuclear clusters A or D and chloroplast clusters A2 or D. Folk samples collected from southern Guangxi were identified as A or D and A2 or D, and folk samples from Hainan were identified as clusters A and A1. In addition, all folk Liang-Mian-Zhen samples shared chloroplast haplotypes with nearby natural populations (Table 1 and Table 3).

Figure 7. 

Molecular identification of the Liang-Mian-Zhen samples using Maximum likelihood (ML) trees based on nuclear internal transcribed spacer (ITS) barcodes (a) and four chloroplast fragments combined (b). Red stars indicate the standard samples obtained from China Resources Sanjiu and black dots indicate folk medicinal samples collected from indigenous individuals in different geographical regions.

In this study, we examined the intraspecific delimitation of the commercial medicinal plant Z. nitidum using 5 DNA barcodes based on samples from 38 natural populations (Fig. 1). The nuclear ITS sequences classified the species into four distinct genetic clusters (A, B, C, and D; Figs 3a, 5). This finding is consistent with a recent taxonomic scenario based on trunk appearance (Editorial Committee of Flora 1997; Zhang et al. 2008; Qin et al. 2019). However, chloroplast delimitation based on 4 fragments classified the 38 natural populations of this species into 5 clusters (A1, A2, B, C, and D; Figs 4, 6). The chloroplast clusters B, C, and D correspond one-to-one to the three nuclear clusters B, C, and D. In addition, they also correspond to three different geographical groups. Previous empirical studies have suggested that morphological taxonomy can be supported by molecular evidence (Su et al. 2015; Lu et al. 2021). According to the integrative taxonomic method (Su et al. 2015; Lu et al. 2021; Parker et al. 2021), taxonomic boundaries for clusters B, C, and D were demarcated based on the above evidence and form three well-delimited clusters. However, cluster A identified by nuclear barcode is split into two distinct chloroplast clusters, A1 and A2 (Fig. 4).

Chloroplast cluster A2 is phylogenetically and geographically closer to D, which is a possible indicator for introgression from D to A (Arnold 1997; 2006; Abbott 2017; Lu et al. 2021). However, morphological differences between A1 and A2 were not recognized based on leaf shape and size, trunk appearance, and flowers. Thus, the conflict between nuclear and chloroplast barcodes may be the consequence of cryptic evolutionary units and/or historical chloroplast introgression. Nevertheless, incomplete lineage sorting cannot be completely ruled out as an explanation for this phenomenon (Hollingsworth et al. 2011; Lu et al. 2016; Hu et al. 2019). In fact, the three well-delimited clusters (B, C, and D) can be directly distinguished from each other and from A based on the trunk appearance, while the other two morphological taxonomic scenarios failed in doing this (Fig. 1). However, none of these three scenarios could recognize the two chloroplast clusters (A1 and A2) within A. Hence, the third taxonomic scenario based on the trunk appearance (Fig. 1c) is better than the other two (Fig. 1a, b).

Usually, hybrid introgression and incomplete lineage sorting can induce a low discrimination power for DNA barcodes (Su et al. 2015; Lu et al. 2016, 2021). Nuclear ITS barcodes have shown strong discrimination power in many empirical studies (Pang et al. 2011; Wang et al. 2011; Su et al. 2015; Lu et al. 2016). In this study, both the nuclear ITS barcodes and the four chloroplast fragments combined showed high intraspecific discrimination power, although historical introgression between strains A and D might have occurred (Figs 3, 4, 7). However, a large discrepancy in the discrimination power of different chloroplast fragments was observed (Fig. 3b). For example, only rbcL could distinguish all chloroplast clusters, while the other three barcodes could only distinguish one from the remaining chloroplast clusters. Unsurprisingly, using DNA fragments with low discrimination power (e.g., matK, ycf6-trnC, and petA-psbJ) to delimit clusters within this species did not yield accurate identification.

In consideration of the large differentiations based on both trunk appearance and molecular markers, whether the clusters in Fig. 1c and Fig. 7a should be treated as separate species or subspecies of Liang-Mian-Zhen remains to be evaluated with more substantial evidence based on genomic data, phenotypic traits, and crossing experiments. Regardless, ITS sequences and rbcL are proposed as barcodes for rapid and accurate identification of the different Liang-Mian-Zhen lineages in China.

Discrepancies in the safety and quality of medicinal materials from different species (or genetic clusters) have been demonstrated in many empirical studies (Bansal et al. 2018; Xin et al. 2018). In the present study, folk Liang-Mian-Zhen samples were shown to comprise several nuclear or chloroplast clusters with regional differences (Figs 3, 4, 7, Tables 1, 3), whereas cultivated Liang-Mian-Zhen material from a state-owned holding pharmaceutical listed company belong to just one of the clusters, suggesting the need to scientifically evaluate the safety and quality for different Liang-Mian-Zhen clusters or lineages. This study is the first to successfully reconcile nuclear molecular evidence (chloroplast conflict exists merely within cluster A) with the morphological taxonomy at the population level and to present accurate resource classification and molecular identification of Liang-Mian-Zhen in China. In summary, there are three major reasons for our accurate molecular identification within this species. First, barcodes with a high level of discrimination power were used to conduct taxonomic delimitation, and to carry out molecular identification. Second, extensive sampling helped to account for overlooked morphological variations and promoted the accuracy of taxonomic delimitation (Linck et al. 2019; Lu et al. 2021). Third, and most importantly, morphological classification or initial identification based on trunk appearance was further supported by integrating evidence from DNA barcodes and geographical distribution. In summary, accurate taxonomic delimitation is the basis for accurate and rapid molecular identification of closely related medicinal plants.