The genus Anoectochilus Blume (Goodyerinae, Cranichideae, Orchidaceae) consists of about 40 reported species in the world, distributed mainly from south and southeast Asia to Australia and the southwest Pacific islands (Jin et al. 2021; POWO 2023). To date, a total of 20 species (including 11 endemics) have been recorded from China (Tian and Xing 2008; Tian et al. 2014; Jin et al. 2021; POWO 2023). These Anoectochilus species, along with certain members of Goodyera and some related genera, are famous as Jewel Orchids in English (Cavestro 1994), and known as “Jin Xian Lian” in Chinese for their “golden lines” on leaves, with both medicinal and edible values.

In the course of our comprehensive resource survey of Anoectochilus species in China from 2015 to 2021, we have reported three new record species: A. elatus (Zheng et al. 2018), A. papillosus (Zheng et al. 2019) and A. brevilabris (Wu et al. 2017), mainly based on their plant morphology. Chemical diversity could also provide some taxonomic evidence for the quality control and authentication of Anoectochilus species using kinsenoside and goodyeroside A (a pair of diastereoisomers) as promising chemotaxonomic markers (Wu et al. 2020). However, species delimitation in this genus is still a big challenge and interspecific relationships among Anoectochilus species remain unclear. Molecular markers have therefore been more and more explored and utilized to solve those phylogenetic issues. ITS2 region as DNA barcode has been commonly adopted for distinguishing those related species, and also for the rapid discrimination of A. roxburghii and its counterfeits (Lv et al. 2015). Our recent study also indicated that both ITS2 and psbA-trnH sequences can be used to distinguish A. roxburghii from its related species with larger genetic distances (Wu et al. 2022b). By analyzing 20 species containing 58 samples, Han et al found that the combination of chloroplast gene fragments (rbcL-matK-trnL-F) was more helpful in exploring the infrageneric relationships among Anoectochilus species than any of those three individual DNA sequences (Han 2019). A similar strategy has been more recently adopted to identify a new Anoectochilus species, A. medogensis, from Tibet, China (Jin et al. 2021).

During our plant resource investigation in Zhongshan, Guangxi Province, China in Aug 2020, an Anoectochilus species was found to be difficult to identify, which was finally clarified as a new species on the basis of detailed morphological, molecular and chemical studies, and described below as A. zhongshanensis C.J. Zheng & Y.B. Wu. In this study, we deciphered the morphological differences between the new species and its nearest congener, A. zhejiangensis Z. Wei & Y. B. Chang. We also used three combined cpDNA sequences (matK, trnL-F and rbcL) to infer the phylogenetic relationships and substantiate the systematic status of A. zhongshanensis as a new species in Anoectochilus. In addition, potential chemotaxonomic markers, kinsenoside and goodyeroside A, were also employed to disclose the chemical difference between the new species and its nearest congeners, A. zhejiangensis and A. malipoensis.

Voucher specimens of A. zhongshanensis were collected in Zhongshan, Guangxi Province, and preserved at the herbarium of Fujian Agriculture and Forestry University (FAFU!). Fresh leaves were washed and dried with filter paper in the field, and then stored in a plastic bag with silica gel for molecular experiments. The living plants of A. zhongshanensis were carefully observed for detailed morphological description and local observation of the plant′s small parts was performed using a stereo microscope (SZ61). Tissue cultures of all collected Anoectochilus species in our lab have been successfully established for resource protection and further chemical and biological studies.

A total of 42 samples representing 18 Anoectochilus species were included for molecular analysis, and all sequences used for constructing the phylogenetic tree were downloaded from GenBank (Suppl. material 6) except those of the new species. Zeuxinella vietnamica was selected as anoutgroup according to previous phylogenic studies (Averyanov 1988; Averyanov and Averyanova 2003). To investigate the phylogenetic status of the new species within the genus Anoectochilus, four DNA markers, including internal transcribed spacer (ITS) and three plastid DNA regions (matK, rbcL and trnL-F), were selected to reconstruct the phylogenetic tree based on previous studies (Jin et al. 2021; Tong et al. 2022).

Ezup column plant tissue genomic DNA extraction kit (Sangon B518261) was used to extract the total genomic DNA from silica gel-dried leaves of the new species. The concentration of DNA samples used in this study were ≥ 20 ng/µL, and the working DNA was stocked in refrigerator at 4 °C for use. Polymerase Chain Reaction (PCR) amplification was performed on a Veriti 96-well thermal cycler (Verity, ABI, USA) using a 25 μL reaction system containing 2.5 μL 10× Taq Buffer (with MgCl₂), 0.2 μL Taq enzyme, 1.0 μL Dntp (mix), 1.0 μL forward and reverse primers, 1.0 μL target DNA template and 18.3 μL ddH₂O. Information on primers and amplification protocols for each DNA region is listed in Suppl. material 7.

DNA quality was detected by electrophoresis using 1% agarose and 1× TAE buffer solution (voltage 120–180V). The concentration and purity were detected by spectrophotometer, and gel imager FR-980A (Shanghai Furi Technology Co., LTD.) was used to record and take photos. The qualified PCR products were sequenced bi-directionally on a 3730XL sequencer (ABI, USA) after purified by a SanPrep column DNA gel extraction kit (Sangon B518131). Sequences were first assembled and edited with SEQMAN (DNA STAR package, USA), followed by sequence alignment with MEGA11 to trim the irregular bases at both ends of the aligned sequences. Bayesian inference (BI), maximum likelihood (ML) and maximum parsimony (MP) methods were used to construct the dataset of multi-gene tandem (rbcL-matK-trnL-F). All characters are considered as unordered and equally weighted, while the indels were processed as missing data after sequence alignment.

MP analysis was performed using MEGA11 (Felsenstein 1985; Tamura et al. 2021). Bootstrap values were generated with 1000 bootstrap replicates with the Subtree-Pruning-Regrafting (SPR) algorithm, with a search level of 1 in which the initial trees were obtained by the random addition of 10 sequences and a limit of 1000 trees, and branch lengths were calculated using the average pathway method. ML analysis of concatenated cpDNA by IQ-TREE-1.6.2 runs with aligned partitions and allows ModelFinder (Kalyaanamoorthy et al. 2017) to identify the best model for each partition (Table 1). Node support was estimated using 1,000 bootstrap iterations and other parameters were set as default for the searches. The BI analysis was performed using MRBAYES-3.2.7-WIN (Ronquist and Huelsenbeck 2003) with best-fit evolutionary models selected under the Bayesian Information Criterion (BIC) also using ModelFinder (Table 1), and the analysis consisted of 2,000,000 generations of four simultaneous Monte Carlo Markov chains. We increased the number of generations (typically 1,000,000) until the average standard deviation of split frequencies falls below 0.01 (3,000,000 generations in total). Phylogenetic trees were sampled every 1000 generations. When the average standard deviation of split frequencies consistently fails to reach the ideal values, the *run.1.p file can be viewed by Tracer v.1.6. The effective sample size (ESS) of all parameters were > 200. Posterior probability (PP) ≥ 0.95 or bootstrap values (BS_ML, BS_MP) ≥ 85 indicates strong support, whereas 0.95 > PP ≥ 0.85 or 85 > BS_ML, BS_MP ≥ 70 suggests moderate support and weak support otherwise.

HPLC-ELSD was performed using EasySep®-3030 HPLC system (Shanghai Tongwei Analytical Technology Co., LTD., China) equipped with an AQ-C18 chromatographic column (3 µm, 4.6 × 250 mm) and an ELSD detector. The mobile phase was ultrapure water (100%) and the flow rate was set at 0.5 mL/ min. The column temperature and ELSD spray chamber temperature were 30 °C and 70 °C, respectively, while the nitrogen flow rate was 2.5 mL/min (Wu et al. 2020). Tissue cultured plant samples of A. zhongshanensis, A. zhejiangensis and A. malipoensis were dried and ground into powder. 0.1 g powder of each sample was accurately weighed and ultrasonically extracted with 20 mL distilled water for 45 min. The extract was subsequently passed through a 0.22 µm PTFE syringe filter for HPLC-ELSD analysis.

Since ITS2 sequence alone could not confirm the systematic status of the new species with convincing infrageneric relationships in the tested Anoectochilus species (Suppl. materials 1–3), combined cpDNA sequences (matK, trnL-F and rbcL) were used to construct datasets that may provide more discriminative information in phylogenetic analysis. By comparing the phylogenetic trees of BI, MP and ML, it was found that the topologies of ML and BI analysis were almost the same, and the topology of MP tree was also congruent with those of ML and BI, only with slight differences in the largest clade including the new species (Suppl. material 4). Here, the topology diagram generated by BI is displayed as Fig. 1, with most of the major clades receiving relatively strong support (PP_BI ≥ 0.95). This BI phylogenetic tree revealed that two endemic species of Hainan Island, A. hainanensis and A. baotingensis, split off first and formed a branch as the sister clade of the other 16 species from mainland China with relatively strong support (PP = 0.99, BS_ML = 61, BS_MP = 86). The other 16 species separated into two large clades. Our two samples of A. zhongshanensis clustered into a well-supported subclade (PP_BI = 1, BS_MP = 94, BS_ML = 99), which is nested in the clade consisting of seven other Anoectochilus species, including A. brevilabris, A. zhejiangensis, A. longilobus, A. roxburghii, A. malipoensis, A. nandanensis, and A. formosanus. These phylogenetic analyses therefore supported the recognition of A. zhongshanensis as a new species of the genus Anoectochilus.

In addition, HPLC-ELSD analysis displayed interesting chemotaxonomic difference that the principle and characteristic lactone glycoside in A. zhongshanensis was kinsenoside rather than its diastereoisomer, goodyeroside A, a major glycoside in A. zhejiangensis and A. malipoensis (Fig. 2). Though morphologically similar, these three species are genetically and chemically different. Especially, those HPLC profiles revealed that A. malipoensis probably only contained goodyeroside A or with trace amounts of kinsenoside that cannot be detected, whereas the other two species have a mixture of goodyeroside A and kinsenoside but with inverse proportion.

Anoectochilus zhongshanensis (Fig. 3), collected from Guangxi, China, possesses typical features of Anoectochilus species, characterized by golden-vein foliage, conical spur, separated stigma-lobes, and the pair of lamellae on column. This new species shows close morphological similarity to A. zhejiangensis and A. malipoensis (Figs 4, 5), but can be distinguished by the hastate or scalpel-shaped lobes of epichile, forward curved and pinnately divided cristate lobes at both sides of the mesochile, and inverted triangle column wings. Molecular studies based on cpDNA (rbcL-matK-trnl-F) combined with chemical analysis on characteristic lactone glycoside further supported the uniqueness of this new species. The key for identifications of 20 already reported species of genus Anoectochilus in China has been recently and systematically established by Jin et al (Jin et al. 2021), and A. zhongshanensis is thus the 21^st member that can be distinguished from its nearest congeners, A. zhejiangensis and A. malipoensis.

The single use of nrDNA (ITS) (Suppl. materials 1–3) or concatenation of both nrDNA and cpDNA (ITS, rbcL, matK, trnl-F) (Suppl. material 5) failed to establish distinguishable relationships among those tested Anoectochilus species. However, only combined cpDNA resulted in a more resolved topology with a reliable phylogenetic result consistent with previous studies (Han 2019; Jin et al. 2021). These inconsistences can be explained by the varied evolution between nuclear ribosomal DNA and chloroplast DNA (Clegg et al. 1994; Pérez-Escobar et al. 2020), which makes nrDNA ITS much more suitable for genetic diversity analysis due to its faster evolutionary rate and cpDNA better for genetic relationship investigation in virtue of their prolific hereditary information. Based on a series of new species (Tian and Xing 2008; Chen and Shui 2010; Tian et al. 2014; Qu et al. 2015) and new records for China (Hu et al. 2012; Wu et al. 2017; Zheng et al. 2018; Jin et al. 2019; Zheng et al. 2019; Jin et al. 2021), there are currently 21 Anoectochilus species known from China. And more distinguishable DNA molecular markers are needed to be developed to provide convincing species delimitation in the Anoectochilus genus.