Total genomic DNAs were extracted from herbarium material (V. aurea) and silica gel-dried material (V. furcata) using the CTAB-based method (Doyle and Doyle, 1987) with the following modifications: chloroform extraction was performed twice. DNA of V. furcata was additionally extracted using the DiamondDNA kit (DiamondDNA, Russia) for clarification of the borders of the inverted repeat (IR) and single copy (SC) regions. For library preparation, we used NEBNext Ultra II DNA sample preparation kit for Illumina (New England Biolabs, USA). Before processing, DNA was sheared using Covaris S220 sonicator (Covaris, USA) with the following settings: time 40 s, peak power 175 W, duty cycle 10%. Libraries were sequenced using Hiseq2000 (V. aurea) or Nextseq (V. furcata) instruments (Illumina, USA).

We failed to combine a complete plastome of V. aurea as only short non-overlapping plastid contigs were assembled; nevertheless, these data were useful for employment in the phylogenetic analyses.

For the plastome of V. furcata, de novo assembly was performed using a CLC Genomics Workbench and IDBA version 1.1.3 (Peng et al., 2012). Contigs showing similarity to plastid genomes were joined by overlapping ends. To check the accuracy of assembly, trimmed paired reads were mapped onto the assembled plastome sequence and the mapping was examined in order to check that there are no regions with gaps in coverage (see Supplementary Figure 1 for the borders of inverted repeats); also, PCR and Sanger sequencing were used for the verification of IR-SC borders. The following primers flanking the IR-SC borders were used: Vf35037F: ATTTCGATTAGGGTCGTATTCTATGG, Vf35269R: CACGGCAATACATTTATACAAAACTTC; Vf35970F: TTCGTGGATCAATTTTAATTCAGTGG, Vf36190R: ATGAAAATATTCGCGATACTTGGTTG. PCR was run on T100 Thermal Cycler (Bio-Rad, USA) using Encyclo PCR kit (Evrogen, Russia) under the following program: initial denaturation for 3 min at 95°C, followed by 35 cycles each comprising 15 s at 95°C, 25 s at 58°C and 40 s at 72°C. PCR products were visualized on agarose gel, cleaned using AMPure beads (Beckman Coulter, USA) and submitted for sequencing to “Genome” sequencing facility (Engelhardt Institute of Molecular Biology of the Russian Academy of Sciences). Sequencing reaction was performed using BigDye Terminator v. 3.1 kit (Thermo Fisher Scientific, USA) and run on a sequencing instrument Applied Biosystems 3730 DNA Analyzer (Thermo Fisher Scientific, USA).

Annotation of plastid genes was performed using the GeSeq web tool (Tillich et al., 2017) with land plant plastid sequences as a reference set.

Colinearity of the sampled plastomes was estimated using the Mauve program (Darling et al., 2004).

Dispersed repeat content was explored using the repeat finder module in the Unipro UGENE package version 37.0 (Okonechnikov et al., 2012) with minimal length restricted to 20 bases for direct, inverted and palindrome repeats.

Estimation and comparison of synonymous and nonsynonymous substitution rates in V. furcata and other orchid lineages were performed using the CodeML program from PAML package (Yang, 1997, 2007) with EasyCodeML interface (Gao et al., 2019). The tree inferred from the phylogenetic analysis of the 29-gene set was used as the input tree. A hypothesis that natural selection acting on the plastid proteins of V. furcata (25 of which were revealed in this study, see below) differs from those of other orchids was tested using two branch models, assuming a single omega value (ω, nonsynonymous to synonymous substitution rate ratio) for all branches versus different values of omega for the V. furcata branch (foreground) and the rest of the branches (background). Besides, the branch-site test was performed to detect signs of possible positive selection affecting a few sites of a protein in the V. furcata branch; Bonferroni correction was applied in both tests. Additionally, a similar gene-wide test for positive selection implemented in the BUSTED web tool (Murrell et al., 2015) was performed.

Sequences for Vietorchis aurea ( Figures 1A, B ) and V. furcata ( Figures 1C, D ) were generated de novo, representing the first DNA data obtained for this genus. In order to get rough estimates of the phylogenetic affinities of Vietorchis, the obtained ITS1–2 regions of both species were searched against the NCBI database using the BLAST tool. The analyses indicated high similarity with Sirindhornia H.A.Pedersen & Suksathan (93–95%) and the related genera from Orchidinae. Thus, in further phylogenetic analysis we focused on this subtribe. Additionally, sequences for Sirindhornia monophylla (Collett & Hemsl.) H.A.Pedersen & Suksathan ( Figure 2 ) were generated for the first time. The rest of the sequences were obtained from GenBank. The outgroup taxa were selected based on Jin et al. (2017).

Our datasets employing selected nuclear and plastid DNA regions were based on the dataset for the subtribe Orchidinae used by Jin et al. (2017). We compiled a sampling of accessions following the idea that the generic diversity should be represented the best, as well as the basal lineages of the main clades. Fifty five specimens were employed in total (including those used by Jin et al., 2017 and studied here for the first time). These specimens belong to 54 species and 24 genera. Of them, Disperis sp., Goodyera schlechtendaliana Rchb.f. and Spiranthes sinensis (Pers.) Ames were included as outgroup taxa.

For the dataset based on the entire plastid genomes, all available published information (by March 30, 2023) on plastome structure of the species belonging to the subtribe Orchidinae was used. Goodyera pubescens (Willd.) R.Br. and Spiranthes sinensis were included as outgroup taxa. Altogether, this dataset covered 29 species (each represented by a single accession) and 13 genera.

Information on all the studied specimens is summarized in Appendices A1 and A2 . General taxonomy of Orchidaceae, including the generic placement of the studied species, follows Govaerts et al. (2022). In the Appendices, taxon names used in the cited papers are indicated in brackets for clarity, in case they are heterotypic synonyms of the currently accepted names, and also for the specimen of Herminium tibeticum X.H.Jin, Schuit. & Raskoti that was re-identified by Jin et al. (2017). The subtribe Orchidinae is accepted here in a wide sense, following Chase et al. (2015) and Jin et al. (2017), i.e. including the genera of the traditionally recognized subtribes Habenariinae and Satyriinae (e.g. Dressler, 1993; Bateman et al., 2003; Jin et al., 2012, Jin et al., 2014; Tang et al., 2015).

Low-coverage genome sequencing that we used for the assembly of the plastid genes/genome of Vietorchis species also allowed us to assemble the nuclear ribosomal RNA operon. Its genes and spacers, especially the 18S gene and the spacers ITS1 and ITS2, are valuable as phylogenetic markers.

Within the datasets employing selected DNA regions, we used four markers: the nuclear ribosomal ITS1–2 region (including internal transcribed spacer 1, the 5.8S rRNA gene and internal transcribed spacer 2; together referred to as ITS), a part of the nuclear Xdh gene, a part of the plastid matK gene and the plastid psbA-trnH intergenic spacer region (including the rps19 gene). These regions have previously been successfully used for phylogenetic analysis of Orchidinae (Jin et al., 2014; Tang et al., 2015; Jin et al., 2017).

The plastid dataset was composed of rRNA and protein-coding genes (29 genes in total) found in the plastome of V. furcata (accD, clpP, infA, matK, rpl14, rpl16, rpl2, rpl20, rpl22, rpl23, rpl32, rpl36, rps11, rps12, rps14, rps16, rps18, rps19, rps2, rps3, rps4, rps7, rps8, rrn16, rrn23, rrn4.5, rrn5, ycf1, and ycf2). Sequences of the corresponding genes were manually extracted from 27 complete plastomes retrieved from Genbank. For V. aurea, partial sequences of 11 of these genes (accD, clpP, infA, matK, rpl14, rpl16, rpl2, rps2, rps3, ycf1, ycf2) were available from the obtained plastid contigs.

For phylogenetic purposes, the sequences of Sirindhornia monophylla were generated. Total genomic DNA was extracted from silica gel-dried leaves using a modified CTAB method (Smith et al., 1991). The primers used for amplification of the ITS, Xdh and matK regions are listed in Table 1 . The psbA-trnH region was not investigated. For the PCR, we used MightyAmp DNA polymerase (Takara Bio Inc., Japan) with corresponding buffer in a 30 mkl reaction mix. The PCR program consisted of 38–42 cycles, with each cycle as follows: denaturation for 20–30 s at 98°C, primer annealing for 30 s at 53°C, and elongation for 60–120 s at 68°C, with an initial denaturation for 3 min at 98°C and the final extension for 7 min at 68°C. The PCR products were run on 1.5% agarose gels to check the quality of amplified DNA. Commercial service of purification and Sanger sequencing was provided by the Invitrogen (China). Both forward and reverse sequences were edited and assembled using DNASTAR (http://www.dnastar.com/).

For the specimens of Vietorchis, sequences of the regions ITS, matK and psbA-trnH were obtained from the results of the high-throughput sequencing (HTS, described above), but we were unable to get the Xdh sequences from the HTS data. We therefore used Sanger sequencing to obtain the Xdh sequence of one of the species, V. furcata. We used the following PCR primers: X502F (TGTGATGTCGATGTATGC), X1599R (G(AT)GAGAGAAA(CT)TGGAGCAAC), X551F (GAAGAGCAGATTGAAGA(AT)(AT)GCC) and X1591R (AA(CT)TGGAGCAACTCCACCA) (Górniak et al., 2010). The PCR program followed Górniak et al. (2010) and Jin et al. (2017). The PCR product was sent to the Majorbio Company (www.majorbio.com, China) for Sanger sequencing.

Sequences were aligned using MAFFT version 7.471 (Katoh et al., 2002; Katoh and Standley, 2013) and corrected manually in BioEdit (Hall, 1999). Regions where positional homology could not be firmly determined were excluded along with the gap-rich positions.

Phylogenetic reconstructions were performed for the concatenated alignments of the two nuclear markers (ITS+Xdh), for the concatenated alignments of the two plastid markers (matK+psbA-trnH), for the alignments of these four markers separately, and for the concatenated alignments of the 29 plastid genes.

The Bayesian phylogenetic reconstruction was performed by MrBayes v.3.2.7 (Ronquist et al., 2012) using four simultaneous runs of 20 million generations and four chains sampling every 1000th generation. The GTR+Γ model of nucleotide substitutions was selected for the data matrices (except for HKY+Γ for the Xdh set) as the most appropriate one according to the Akaike information criterion (Akaike, 1974) in PAUP version 4.0a (Swofford, 2003). The first million generations were discarded as burn-in, and the remaining trees were combined in a majority-rule consensus tree. Effective sample sizes were evaluated using Tracer v.1.7.1 (Rambaut et al., 2018). The effective sample sizes were > 200 for all statistics in all datasets, suggesting that the run length was adequate.

The maximum likelihood (ML) phylogenetic reconstruction was performed by IQ-tree (Minh et al., 2020). Internal branch support was assessed with the ultrafast bootstrap approximation (Hoang et al., 2018) using 10 thousand replications.

The main characteristics of the alignments are listed in Table 2 . The Bayesian and ML approaches for each dataset resulted in generally congruent tree topologies ( Figures 3 – 5 ; Supplementary Figures 2–12 ).

The analyses of four separate markers (ITS, Xdh, matK, psbA-trnH; Supplementary Figures 5–12 ) lack any well-supported incongruence with the two-marker trees ( Figures 3 , 4 ; Supplementary Figures 2, 3 ), but only show a lower resolution.

In all the two-marker trees (i.e., the nuclear ITS+Xdh trees and the plastid matK+psbA-trnH trees), the genera belonging to the subtribe Orchidinae (as defined by Chase et al., 2015) form a well-supported clade [posterior probability (PP) 1.00 for both datasets, ultrafast bootstrap percentage in the maximum likelihood analysis (BP_ML) 98 and 100, respectively]. The genus Satyrium occupies a basal position within Orchidinae (the monophyly of the rest of Orchidinae is supported with PP 0.95, BP_ML 84 for the nuclear dataset and PP 1.00, BP_ML 98 for the plastid dataset). The rest of the accessions form two clades sister to each other. One of them (PP 1.00, BP_ML 100 for both datasets) comprises all the species of Habenaria studied, intermixed with the species of Diplomeris, Gennaria, Herminium, Hsenhsua, Pecteilis, and Peristylus. The second clade (PP 1.00 for both datasets, BP_ML 99 and 100, respectively) is subdivided into three subclades. One of them is Brachycorythis (PP 1.00, BP_ML 100 for both datasets; Brachycorythis occupies a sister position to the rest of the clade in the nuclear trees, and forms a polytomy with the two other subclades in the plastid trees). The second subclade (PP 1.00 for both datasets, BP_ML 100 and 92, respectively) contains species of the mostly extra-tropical genera (Anacamptis, Dactylorhiza, Galearis, Gymnadenia, Ophrys, Orchis, Platanthera). In the third subclade (PP 0.99, BP_ML 95 for the nuclear dataset and PP 0.73, BP_ML 89 for the plastid dataset), a clade comprising Sirindhornia and Vietorchis (PP 1.00, BP_ML 100 for both datasets) is sister to a clade containing the species of Hemipilia together with Ponerorchis limprichtii and Tsaiorchis keiskeoides (PP 1.00, BP_ML 100 for both datasets), i.e. corresponding to Hemipilia sensu latissimo as accepted by Tang et al. (2015) and Yang et al. (2022). The clade comprising Sirindhornia and Vietorchis shows a tritomy in the nuclear trees with the following branches: V. aurea, V. furcata and the monophyletic Sirindhornia (S. monophylla + S. pulchella; PP 1.00, BP_ML 100). In the plastid trees, both Sirindhornia (PP 1.00, BP_ML 100) and Vietorchis (PP 0.99, BP_ML 97) are monophyletic.

The trees based on the 29-gene plastid dataset ( Figure 5 , Supplementary Figure 4 ) are highly congruent to the two-marker trees (ITS+Xdh and matK+psbA-trnH), despite it employed a smaller sampling that was quite different at the species level. In the 29-gene trees, the monophyly of the subtribe Orchidinae is supported (PP 1.00, BP_ML 100). Satyrium is the basalmost taxon within Orchidinae, and the other accessions are distributed within two clades. One of these clades comprises the species of Habenaria intermixed with Pecteilis and Peristylus (PP 1.00, BP_ML 100). Within the second clade (PP 1.00, BP_ML 100), the subclade containing species of the mostly extra-tropical genera (Dactylorhiza, Galearis, Gymnadenia, Ophrys, Platanthera; PP 1.00, BP_ML 100) is sister to a subclade (PP 1.00, BP_ML 98) with the following topology: (Hemipilia gracilis + Hemipilia yajiangensis; PP 1.00, BP_ML 100) + (Vietorchis aurea + Vietorchis furcata; PP 1.00, BP_ML 100).

The newly assembled complete plastid genome of Vietorchis furcata is 65,969 base pairs (bp) in length with the typical quadripartite structure containing a large (LSC) and a small single copy (SSC) region (35,123 and 29,056 bp, respectively) separated with inverted repeats (IR) of 895 bp ( Figure 6 ). The overall GC content was 33.58%.

Forty five unique genes were revealed in the plastome ( Table 3 ), including those of 4 rRNAs, 16 tRNAs, 11 proteins of small ribosomal subunit, 8 proteins of large ribosomal subunit, and 6 other proteins. Of these genes, 31 were situated in the LSC, 11 were in the SSC, and the trans-spliced rps12 was spread in both single copy regions. The inverted repeats contained two genes (rpl23 and trnI-CAU) and partial sequences of rpl2 and ycf2 genes (with their other parts being in the LSC and SSC, respectively). Fraction of the coding DNA was 44.01%.

No intact photosynthesis-related genes were found, but 17 pseudogenes were annotated in the plastome: atpB, atpE, atpH, cemA, rpoA, rpoB, rpoC2, ndhB, ndhF, petA, petD, psaB, psaJ, psbA, psbI, psbK, and rps15 ( Figure 6 ); all of them except for rps15 are derived from the photosynthesis-related genes. All of the pseudogenes contained multiple internal termination codons. Of the six intron-containing genes, five were protein-coding sequences (rps16, rpl16, rpl2, rps12, clpP; the last one contained two introns) and one was the tRNA gene (trnA-UGC).

The genes and pseudogenes in the plastome of V. furcata retained the same relative position as in the phylogenetically closest known plastomes, i.e. those of the photosynthetic Hemipilia gracilis and Hemipilia yajiangensis ( Supplementary Figure 13 ).

A total of 13 dispersed repeats were found in the plastome of V. furcata, with the longest repeat being 68 bases long ( Supplementary Figure 14 ). The numbers of direct, inverted and palindrome repeats were almost equal: 4, 4 and 5, respectively. Comparison of the plastomes of Vietorchis furcata and Hemipilia yajiangensis showed similar relative amount of dispersed repeats (~0.20 vs. ~0.19 repeats per 1000 bases) and their relative length (0.50% vs. 0.47% of the plastome length).

The likelihood ratio tests (LRTs) based on the branch models (and performed using CodeML) showed that the “single ω model” was preferable for 23 (of 25) protein-coding genes, whereas the difference in ω values between the foreground and background branches can be considered statistically significant (implying the alternative model) for the two remaining genes (rps2 and rps11, Table 4 ). However, for all the genes, an averaged omega ratio was below 1, suggesting that all of them undergo negative selection. The performed branch-site tests also did not reveal any signs of positive selection acting on the amino acid residues ( Supplementary Table 1 ). Similarly, the LRTs performed using BUSTED showed no evidence of episodic diversifying selection in any of the analyzed genes in the plastome of V. furcata ( Table 4 ).

In our reconstructions, Vietorchis is deeply nested within the subfamily Orchidoideae, and thus the earlier discussed possibility of the placement of Vietorchis outside Orchidoideae is now decisively refuted. Taking into account the implied close relationship between Vietorchis and Silvorchis, we extrapolate this conclusion to the latter genus. The reason of doubts regarding the subfamily placement of these two genera was the structure of their underground organs, i.e. the “Epipogium-Cyrtosia-like” plagiotropic thick fleshy branching rootless tuber-like rhizomes (in S. colorata and in the genus Vietorchis; Averyanov et al., 2013) or “Gastrodia-like” rhizome-like tubers (in S. vietnamica; Averyanov et al., 2018). It is noteworthy that Smith (1907) described the underground organs of S. colorata as short fleshy rhizome bearing 0.6 cm thick roots. This was reproduced in subsequent accounts (e.g. Pridgeon et al., 2005) since no material was available to clarify the morphology. Similarly, Averyanov (2010) initially described V. aurea as having tuber-like roots due to the lack of appropriate material at that time. Based on the newly obtained material of Silvorchis and Vietorchis, as well as on the drawing from the protologue of S. colorata (Smith, 1907), Averyanov et al. (2013, 2018) concluded that the two genera share fleshy rootless underground stems ranging from thick rhizomes to somewhat elongate tubers. There is also a possibility that the underground plant part in Silvorchis and Vietorchis is a root-stem tuberoid (see also Averyanov et al., 2023), i.e. a storage root with its basal portion surrounding a core of stem tissue with a bud (as defined by Dressler, 1993), but anyway these genera lack morphologically distinct roots of typical shape and structure.

Thus, our molecular phylogenetic data support the idea that the similarities in the rhizome morphology between Vietorchis (and the supposedly related Silvorchis) and the genera from the subfamilies Epidendroideae and Vanilloideae are examples of a convergence, and are likely caused by the mycoheterotrophic lifestyle shared by these genera. At the same time, the rhizomes of Vietorchis are markedly different from those of its closest photosynthetic relatives within Orchidoideae. In particular, Sirindhornia, the closest genus to Vietorchis in our reconstructions based on the datasets employing selected nuclear and plastid regions, is characterized by root-bearing tubers/tuberoids (Pedersen et al., 2003; Chen et al., 2009; Pedersen, 2011).

Our reconstructions based on the nuclear dataset and two plastid datasets demonstrate that Vietorchis is placed within the subtribe Orchidinae, where it (together with Sirindhornia, when included in analysis) forms a sister group to Hemipilia s.l. (including Ponerorchis and Tsaiorchis). Molecular phylogenetic evidence therefore supports the synonymization of the subtribe Vietorchidinae with Orchidinae suggested by Olędrzyńska et al. (2016). Since the latter authors have not indicated the reasons of their decision, our study is the first one to provide a basis for such a taxonomic rearrangement.

In the two-marker trees (i.e. those based on ITS+Xdh and matK+psbA-trnH datasets), which employed a broader sampling with respect to the trees based on the plastomes, Vietorchis is most close to Sirindhornia. Vietorchis is therefore inferred here to be a part of the clade XVIII recognized by Jin et al. (2017). All the three species of Sirindhornia are restricted to limestone mountains (Pedersen et al., 2003; Pedersen, 2011), and one species of Vietorchis (V. aurea) also inhabits limestone karsts. With this evidence, our phylogenetic results suggest that association with limestone is possibly a plesiomorphic condition for Vietorchis. This assumption is to be tested after establishment of the phylogenetic position of Silvorchis, a putative closest relative of Vietorchis known exclusively in non-limestone areas.

Morphologically, Vietorchis is unique among Orchidinae (and among the entire tribe Orchideae) in having a lip with a massive callus and a raised longitudinal keel (Averyanov, 2013; Averyanov et al., 2013). In addition, Vietorchis and Silvorchis are remarkable in their spurless lips, which is a very rare feature in Orchidinae (Dressler, 1993), known, for example, in Ophrys and Serapias L. No morphological proximity between Vietorchis and Sirindhornia has ever been proposed. Indeed, with respect to the flower structure, Vietorchis (together with Silvorchis, for which the molecular phylogenetic data are still lacking) is equally close to a number of genera traditionally associated with Orchis as Orchidinae s.s., of which the ones inhabiting tropical mainland Asia being Sirindhornia and Hemipilia s.l. (including Ponerorchis, Tsaiorchis, the formerly recognized Amitostigma Schltr. and Neottianthe (Rchb.) Schltr. and possibly Apetalanthe Aver. & Vuong, the latter genus not yet included in a phylogenetic analysis). All genera of this alliance share such characters as small plant habit, fleshy tuber-like roots, erect anther with closely spaced almost parallel thecae, thecae with short bases supported by small rostellum having no rostellar arms, large hemispheric or conoid auricles on the sides of the anther, clavate pollinia with long caudicles terminated by closely spaced viscidia, viscidia completely separated or united into a single body, viscidia naked or covered by bursiculum (or bursicula), entire concave stigma without any extensions. Except for the roots, this set of characters is also found in Vietorchis and Silvorchis.

Thus, flower morphology is concordant with the molecular data in placing Vietorchis into Orchidinae, but appears to be only moderately instructive in determination of its phylogenetic relationships within the subtribe. Morphology strongly supports the assignment of Vietorchis to the second major clade (the one containing Orchis, etc), but fails to guide its affinities within the clade. At the same time, the two subclades of this clade are correlated with the geographical evidence, and Vietorchis fits this pattern as it occupies the predominantly tropical subclade. Therefore, characteristics of flower morphology combined with the geographical distribution exhibit considerable phylogenetic signal.

While the typical plastomes of autotrophic angiosperms are 120–170 kb in length and encode 120–130 genes (Ruhlman and Jansen, 2021), the plastome of the non-photosynthetic Vietorchis furcata is substantially reduced in both length and gene content, in line with the tendency observed in heterotrophic plants (Barrett and Davis, 2012; Barrett et al., 2014; Wicke et al., 2016; Graham et al., 2017). Vietorchis furcata possesses one of the most reduced plastomes in the subfamily Orchidoideae (65969 bp), the other ones being those of the two other fully mycoheterotrophic species, Rhizanthella gardneri R.S.Rogers (59190 bp: Delannoy et al., 2011) and Corybas cryptanthus Hatch (69300 bp: Murray, 2019). The plastome of Vietorchis furcata lacks intact photosynthesis-related genes and contains some genes of ribosome components and transfer RNAs in addition to several other “housekeeping” genes (matK, accD, clpP, infA, ycf1, ycf2). The retained protein-coding genes in Vietorchis furcata appear to undergo the same negative selection as in its photosynthetic relatives from the subfamily Orchidoideae, although the stabilizing constraints are likely to be relaxed in the rps2 and rps11 sequences in V. furcata.

In terms of functional gene content, the plastome of Vietorchis furcata seems to be at one of the last stages of degradation (sensu Barrett et al., 2014), because it has lost several genes of tRNA and the rps15 ribosomal protein has been pseudogenized. Similar gene content is characteristic of the most reduced plastomes in Orchidaceae, e.g., in Epipogium (Schelkunov et al., 2015), Gastrodia (Wen et al., 2022), Pogoniopsis (Klimpert et al., 2022), Rhizanthella (Delannoy et al., 2011). However, in spite of their similar gene content, the plastomes of Epipogium, Gastrodia and Pogoniopsis are considerably shorter than those of Rhizanthella and Vietorchis. This is due to a larger fraction of non-coding DNA in the plastomes of Rhizanthella and Vietorchis, and particularly, the presence of multiple pseudogenes in the latter. Pseudogenes are usually rapidly purged from the plastomes of heterotrophic plants, and their abundant presence is associated with very recent transition to heterotrophy (see e.g. Barrett et al., 2014; Samigullin et al., 2016). The other types of non-coding DNA (intergenic spacers and introns) are also scarce in highly reduced plastomes (Schelkunov et al., 2015, 2019; Su et al., 2019). Thus, the plastome of Vietorchis represents an interesting example of heterochrony: it has a highly reduced gene set (with the reduction affecting even the ribosomal protein genes, which is typical for the last steps of plastome degradation, following the model of Barrett and Davis, 2012) but keeps a large amount of pseudogenes and non-coding DNA of the other types (which is typical for the earliest steps of degradation). It should also be noted that though ribosome is a necessary component of virtually any cell (and plastids and mitochondria as well, since they are derivatives of a bacterial cell), the essentiality of different ribosomal proteins is considerably unequal (see Nikolaeva et al., 2021). It might be beneficial to update the model of plastome reduction in heterotrophic plants taking this idea into account.

In the phylogenetic trees based on the plastid datasets, both studied species of Vietorchis occupy long terminal branches, which assume a substantial overall elevation of substitution rates. The phenomenon of long branches is characteristic of plastomes of many heterotrophic lineages (Lam et al., 2018), and the elevated substitution rates in plastomes of fully heterotrophic plants seem to be frequent (see a review by Sanchez-Puerta et al., 2023), although not obligatory (e.g., Cephalanthera humilis X.H.Jin: Lam et al., 2018, Petrosavia stellaris Becc.: Logacheva et al., 2014).

The small length of the inverted repeats in Vietorchis furcata (895 bp) is a remarkable feature, which is shared by this species with Epipogium roseum Lindl. (about 250–300 bp in different accessions: Schelkunov et al., 2015) and Pogoniopsis schenckii Cogn. (1509 bp: Klimpert et al., 2022). The IR reduction is the only apparent structural alteration in the plastome of Vietorchis furcata with respect to the typical plastomes of photosynthetic angiosperms. Both large contractions and expansions of the IRs are often documented in heterotrophic plant lineages (Wicke et al., 2013; Feng et al., 2016; Logacheva et al., 2016; Kim et al., 2020b; Yudina et al., 2021), sometimes even within a single genus (e.g., Epipogium: Schelkunov et al., 2015, Neottia Guett.: Feng et al., 2016, Thismia Griff.: Yudina et al., 2021).

The retention of plastome gene order revealed in Vietorchis furcata is rather typical for the species with recent transition to heterotrophy and low degree of plastome reduction (e.g. Barrett et al., 2014; Samigullin et al., 2016). And vice versa, numerous deeply reduced plastomes of mycoheterotrophic (e.g., Schelkunov et al., 2015; Lim et al., 2016; Li et al., 2019; Yudina et al., 2021; Wen et al., 2022) as well as holoparasitic (e.g., Schelkunov et al., 2019; Su et al., 2019) species show a highly altered gene order, although the retained colinearity was also demonstrated for some of such taxa (e.g., Logacheva et al., 2011; Lam et al., 2015; Klimpert et al., 2022). Thus, the plastome of Vietorchis furcata, being colinear to the plastomes of the autotrophic genus Hemipilia (which are the most close phylogenetically to Vietorchis among the available orchid plastomes) again demonstrates the same heterochronic pattern as outlined above. It should be noted, however, that the major determinant of the rearrangements (inversions and translocations) in plant plastomes are the dispersed repeats, and therefore the number of such rearrangements relates largely to the repeat richness, and not to the nutrition type. Indeed, numerous autotrophic plant groups with highly rearranged plastomes are known (e.g. within Campanulaceae: Haberle et al., 2008, Ericaceae: Logacheva et al., 2016, Geraniaceae: Guisinger et al., 2011, Oleaceae: Lee et al., 2007); all of these plastomes have high fraction of repetitive DNA, which is supposed in the cited studies to relieve rearrangements. The presence of abundant dispersed repeats seems to be one of the necessary conditions for the occurrence of rearrangements; in line with this idea, the unrearranged plastome of Vietorchis furcata has low abundance of repeats.