Molecular phylogenetic study of the tribe Tropidieae (Orchidaceae, Epidendroideae) with taxonomic and evolutionary implications

Abstract The orchid tribe Tropidieae comprises three genera, Tropidia, Corymborkis and Kalimantanorchis. There are three fully mycoheterotrophic species within Tropidieae: Tropidia saprophytica, T. connata and Kalimantanorchis nagamasui. A previous phylogenetic study of K. nagamasui, based only on plastid matK data, placed K. nagamasui outside the clade of Tropidia and Corymborkis without support. In this study, we performed phylogenetic analyses using a nuclear ribosomal DNA spacer (ITS1-5.8S-ITS2), a low-copy nuclear coding gene (Xdh) and a mitochondrial intron (nad1b-c intron) to study the phylogenetic relationships within Tropidieae. We included six photosynthetic and all three fully mycoheterotrophic Tropidieae species. The resulting phylogenetic trees placed these fully mycoheterotrophic species inside the Tropidia clade with high support. In our trees, these three species do not form a monophyletic group together, because the photosynthetic T. graminea is nested amongst them. Our results also suggest that the loss of photosynthetic ability occurred at least twice in Tropidia.


Introduction
The tribe Tropidieae is one of the earliest diverging clades in subfamily Epidendroideae (Orchidaceae) and currently contains three genera: Corymborkis Thouars, Tropidia Lindl. and Kalimantanorchis Tsukaya, M.Nakaj. & H.Okada (Tsukaya et al. 2011;Freudenstein and Chase 2015;Koch et al. 2016). Corymborkis comprises eight species and has a pantropical distribution (Rasmussen 1977;Govaerts et al. 2019). About 30 species are recognised in Tropidia and most of the species occur in tropical Asia and Australasia; the Neotropical T. polystachya (Sw.) Ames is the only species outside Asia and Australasia (Jones and Bolger 1988;Kumar et al. 2015;Koch et al. 2016;Ormerod 2018). Kalimantanorchis is the most recently established genus in the tribe with the single fully mycoheterotrophic species K. nagamasui Tsukaya, M.Nakaj. & H.Okada from Sabah and West Kalimantan in Borneo (Tsukaya et al. 2011;Suetsugu et al. 2017a). Two other fully mycoheterotrophic species are reported in Tropidieae; both belong to Tropidia: T. saprophytica J.J.Sm. and T. connata J.J.Wood & A.L.Lamb. Tropidia saprophytica has been recorded from Sabah and Sarawak and T. connata from Sabah and West Kalimantan (Wood 1988;Kikuchi and Tsukaya 2017). Wood (1984) established the fully mycoheterotrophic genus Muluorchis J.J.Wood together with the description of Muluorchis ramosa J.J.Wood. He later found that the species is conspecific with T. saprophytica and synonymised Muluorchis with Tropidia (Wood 1988). Wood and Lamb later described another fully mycoheterotrophic species, T. connata (Wood and Cribb 1994). Tropidia connata is easily distinguished from T. saprophytica by the zigzag inflorescence, connate lateral sepals and shortly spurred lip (Wood and Cribb 1994). More recently, Kalimantanorchis was established as a new genus, based on its morphological characters and molecular phylogenetic data (Tsukaya et al. 2011). The subterranean tuberous structure of Kalimantanorchis had never been observed in the other two fully mycoheterotrophic Tropidia species, although we now know that a tuberous structure occurs at least in T. connata as well (Tsukaya et al. 2011;Kikuchi and Tsukaya 2017). Kikuchi and Tsukaya (2017) suggested that the tuberous structures of the fully mycoheterotrophic Tropidieae species may be equivalent to the tuberlike nodules of other leafy Tropidia (Yeh et al. 2009, Chun-Kuei et al. 2013). On the other hand, Tsukaya et al. (2011) provided a molecular phylogenetic tree based only on plastid matK data, in which Kalimantanorchis was placed outside the clade comprising Tropidia and Corymborkis.
The study of mycoheterotrophy is hampered by two major obstacles (Merckx et al. 2013a). One is the rarity of many fully mycoheterotrophic species: several have been found only once or twice in the field. Due to the scarcity of adequate material, for some of the species, information is lacking even at the basic level of morphology and anatomy. The other major obstacle is the elevated plastid DNA substitution rates usually occurring in fully mycoheterotrophic plants. In these plants, the leaves are normally reduced to scales or sheaths and they are achlorophyllous (Tsukaya 2018). Most of the plastid genes of fully mycoheterotrophic species are, supposedly, not functioning anymore as the plants have lost the ability to conduct photosynthesis and often experienced higher mutation rates and deletions in these genes (Freudenstein and Senyo 2008;Delannoy et al. 2011). Although some fully mycoheterotrophic species still retain amplifiable plastid DNA regions, they can be difficult to align and analyse due to these elevated substitution rates (Merckx et al. 2013a). Since most of the phylogenies of the photosynthetic, presumably autotrophic or partially mycoheterotrophic relatives are at least partly based on plastid regions, it is sometimes impossible to relate the fully mycoheterotrophic species properly to other lineages. Therefore, the phylogenetic placement of fully mycoheterotrophic plants has often been difficult. These problems also apply to the fully mycoheterotrophic Tropidieae species.
In 2011 and 2012, multiple specimens of Tropidia connata were collected by the last author (HT) in Borneo. These specimens included underground tuberous structures, which were not described in the original description (Wood and Cribb 1994) due to the incompleteness of the examined material. The examination of tuberous structures in T. connata clearly indicates that the tuber is a shared character between Kalimantanorchis nagamasui and T. connata, suggesting that Kalimantanorchis may not be distinct from Tropidia (Kikuchi and Tsukaya 2017) (Fig. 1). As already mentioned, Tsukaya et al. (2011) provided a phylogenetic analysis based on a plastid region (matK), which placed Kalimantanorchis outside the Tropidia + Corymborkis clade. However, this result had low support, possibly because of a highly elevated substitution rate of matK in Kalimantanorchis. Based only on the morphological findings of Kikuchi and Tsukaya (2017), Ormerod and Juswara (2019) synonymised Kalimantanorchis with Tropidia; however, they did not provide molecular evidence to counter the findings of Tsukaya et al. (2011). In this study, we were able to include all three fully mycoheterotrophic species within a broader sampling of Tropidieae compared to the previous phylogenetic analysis of Tsukaya et al. (2011) and conducted molecular phylogenetic analyses using a nuclear ribosomal DNA spacer (ITS1-5.8S-ITS2), a low-copy nuclear coding gene (Xdh) and a mitochondrial intron (nad1b-c intron) to clarify the placement of the fully mycoheterotrophic species.

Taxon sampling and plant material
Nine species of Tropidieae were included in this study (six species of Tropidia, one of Kalimantanorchis and two of Corymborkis). We sampled all three fully mycoheterotrophic species. Apart from Tropidia polystachya occurring in the Neotropics, all sampled Tropidia species are native to Asia. We selected Sobralia rosea Poepp. & Endl.
(Epidendroideae: Sobralieae) as an outgroup taxon for the phylogenetic analyses considering that Sobralieae is sister to the rest of Epidendroideae, except for Neottieae (Freudenstein and Chase 2015;Givnish et al. 2015). We obtained three sequences of Sobralia rosea from GenBank. We used five DNA samples from the DNA Bank of the Royal Botanic Gardens, Kew. Four samples were collected in the field and extracted in this study. We provided the sample and voucher information in Table 1. DNA extraction, PCR amplification and sequencing DNA extraction was conducted for the four samples obtained in the field using FTA cards (Whatman, Tokyo, Japan) and DNeasy kit (Qiagen, Hilden, Germany). Amplification for the nuclear ITS, Xdh and mitochondrial nad1b-c intron, was performed for nine samples using TaKaRa ExTaq polymerase (TaKaRa Bio, Shiga, Japan) and DreamTaq DNA Polymerases (Thermo Fisher, Epsom, UK). The following primer sets were used: ITS1 and ITS4 for ITS1-5.8S-ITS2 (White et al. 1990), X502 and X1599R for Xdh (Górniak et al. 2010) and nad1 exon B and nad1 exon C for nad1 b-c intron (Demesure et al. 1995). For ITS, the thermal cycling protocol began with 3 min initial denaturation at 94°C, followed by 35 amplification cycles, each with 15 s at 94°C, 30 s at 44°C and 40 s at 72°C, which was concluded by a final extension at 72°C for 7 min. For Xdh, the thermal cycles were as follows: initial denaturation at 94°C for 2 min, 6 touchdown cycles of 45 s at 94°C, 45 s at 53°C and 90 s at 72°C, reducing one degree per cycle, which was followed by 28 cycles of 45 s at 94°C, 45 s at 47°C and 90 s at 72°C and a final extension at 72°C for 5 min. For nad1 b-c intron, the thermal cycles were set as follows: pre-melting for 3 min at 94°C, followed by 35 cycles of 15 s at 94°C, 30 s at 50°C and 2 min at 72°C and a final extension at 72°C for 7 min. Cycle sequencing was conducted using Big Dye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) with the same primers used for PCR reactions. We additionally used nad1M as an internal primer for sequencing nad1 b-c intron (see Freudenstein and Chase 2001). Newly generated 27 sequences were deposited in GenBank and their accession numbers are listed in Table 1 with the accession numbers of the three sequences of Sobralia, which were obtained from GenBank.

Phylogenetic analyses
In total, 30 sequences were edited and assembled using Geneious Prime 2019.2.1 (https://www.geneious.com) (Kearse et al. 2012). Sequence alignments for each region were generated with the MUSCLE alignment tool (Edgar 2004), installed in Geneious Pro using the default setting. Phylogenetic relationships were inferred for the concatenated alignments of the three markers (ITS, nad1b-c intron, Xdh). The best partitioning scheme was selected by the Bayesian Information Criterion (BIC) using Partition-Finder 2.1.1 (Guindon et al. 2010;Lanfear et al. 2016). The best partitioning and substitution models for each partition were estimated as follows: ITS (K80+G), nad1 b-c intron (K81UF), Xdh first and second codon (K80), Xdh third codon (HKY). The Maximum Likelihood (ML) analysis was conducted using IQ-TREE 1.6.12 (Nguyen et al. 2015). We obtained branch supports by the ultrafast bootstrap with 1000 replicates (Hoang et al. 2018). The Bayesian Inference (BI) analysis was performed using MrBayes 3.2.7a (Ronquist et al. 2012). The BI analysis was run for 50000 generations. Trees were sampled every 100 generations of the MCMC chain. By default, MrBayes discarded the first 25% samples from the cold chain. The average standard deviation of split was checked at the end of the run. Using the sump command, the model parameters were also checked for the convergence diagnostic, the Potential Scale Reduction Factor (PSRF). All the PSRF values were close to 1.0. A 50% majority rule consensus tree was estimated using the sumt command to obtain the posterior probabilities for each clade. Generated trees were visualised using FigTree 1.4.3 (Rambaut 2017). Clades with over 85% bootstrap (BS) value or 0.95 Bayesian Posterior Probability (PP) were considered strongly supported. Clades with over 75% BS or 0.90 PP support were considered moderately supported. Clades with lower support values were not regarded as reliably supported clades.

Results
The aligned sequence lengths for each region were 669 bp (ITS), 1389 bp (nad1b-c intron) and 843 bp (Xdh). We separately conducted ML analyses for the three regions and none of the incongruences in topology was significantly supported between the trees. Therefore, the combined usage of the three markers was justified.
The positions of all the taxa were consistent in both trees resulting from the ML and BI analyses for the combined alignments. As the topology was consistent with strong support, we added the bootstrap values obtained from the BI analysis to the ML tree, which is shown in Figure 2.
In both the ML and BI trees, Corymborkis species form a monophyletic clade (BS = 100%, PP = 1). Tropidia species also form a monophyletic clade with the Neotropical T. polystachya as the first diverging species with a moderately high PP support (0.91) but with a low BS support (65%) (Fig. 2). In our trees, T. polystachya is sister to all Asian Tropidia species. In this Asian Tropidia clade, the East Asian Tropidia nipponica var. hachijoensis F.Maek. & Yokota is the first diverging species, followed by the Sri Lankan Tropidia bambusifolia (Thwaites) Trimen. The three fully mycoheterotrophic Tropidieae species form a clade (clade Myco, hereafter) with the photosynthetic Tropidia graminea Blume. In clade Myco, the clade which includes K. nagamasui and T. saprophytica (clade A, hereafter) is sister to the clade formed by T. connata and T. graminea (clade B, hereafter) with high support (BS = 95%, PP = 1).

Monophyly of Corymborkis and Tropidia
Corymborkis and Tropidia were each inferred as monophyletic in our study. In the comparative morphological study of Corymborkis by Rasmussen (1977), Tropidia polystachya was mentioned as an exceptional Tropidia species because of the type of inflo-rescence feature, which is shared with Corymborkis: branched inflorescence when fully developed. As T. polystachya is placed as the first diverging species of Tropidia in our trees, we suggest that the branched inflorescence may be a symplesiomorphic character in Corymborkis and T. polystachya. As the morphology of these orchids is still little studied, it would be desirable to analyse character evolution in more detail using a broader sampling of Tropidieae.

Evolution of non-photosynthetic feature
The evolutionary pathway to full mycoheterotrophy is hypothesised to be irreversible (Barrett et al. 2014). Once the ability to perform photosynthesis is lost, the photosynthetic genes are not expected to be regained (Lam 2016). In our results, Tropidia graminea, a photosynthetic species with well-developed leaves, is placed as a sister taxon to fully mycoheterotrophic T. connata in clade B, which is sister to clade A. Granted that the evolution of the non-photosynthetic feature is irreversible, we hypothesise that the ancestral character of clade A and clade B is photosynthetic. Therefore, this result indicates that non-photosynthetic features independently evolved at least twice from photosynthetic ancestors in the Tropidia clade (Fig. 2). However, the branch lengths of the basal nodes of clade A and clade B are short and it might be the case that long branch attraction or the incongruences amongst datasets influenced this topology. We can only test this if we increase the number of sampled species from Asian Tropidia and include more markers to conduct phylogenetic analyses. Flowering plant genera with multiple transitions to full mycoheterotrophy have rarely been reported to date. Hexalectris Raf. (Orchidaceae) and Burmannia L. (Burmanniaceae) are two of the few genera known to have multiple transitions to full mycoheterotrophy (Merckx et al. 2008;Barrett et al. 2019).
In Orchidaceae, more than 230 species in 43 genera are assumed to be non-photosynthetic and thus fully mycoheterotrophic (Merckx et al. 2013b). It is estimated that within Orchidaceae, full mycoheterotrophy has evolved about 30 times independently, possibly from partially mycoheterotrophic ancestors (Freudenstein and Barrett 2010). Partially mycoheterotrophic species still retain the ability to photosynthesise at the same time gaining carbon also from the mycorrhizal fungal partners and this nutritional mode of life has been suggested to be more widespread in Orchidaceae than previously assumed (Gebauer et al. 2016). Partial mycoheterotrophy has been hypothesised to be an intermediate evolutionary step from initial mycoheterotrophy towards full mycoheterotrophy in Orchidaceae (Merckx et al. 2013a). For instance, it is likely that in Cymbidium Sw. fully mycoheterotrophic species evolved from partial mycoheterotrophic ancestors (Motomura et al. 2010). Partial mycoheterotrophy has been confirmed for many photosynthetic orchid species, mainly in subfamily Epidendroideae through isotopic analyses (Julou et al. 2005;Cameron et al. 2009;Suetsugu et al. 2017b).
So far, the occurrence of partial mycoheterotrophy in Tropidia has not been examined. However, all species of Tropidia usually grow in deep shade in the forest understorey, which supports the hypothesis that they might be partially mycoheterotrophic (Rasmussen 2005). In order to elucidate the evolutionary pathway from photosynthetic plants to fully mycoheterotrophic organisms in Tropidia, isotopic analyses on photosynthetic Tropidia species would be indispensable.

Kalimantanorchis, a synonym of Tropidia
Kalimantanorchis is nested in the Asian Tropidia clade with high support in our study. This suggests that the previous molecular phylogenetic analysis, based only on plastid matK (Tsukaya et al. 2011), failed to show the phylogenetic relationships. The matK sequence of Kalimantanorchis used in that study was extremely short (501 bp) and we suspect that it also had a particularly high substitution rate, which possibly made the alignment problematic and thus caused the unsupported placement of Kalimantanorchis outside the Tropidia + Corymborkis clade. According to our results and the finding of the tuberous underground structures being a shared character with T. connata (Kikuchi and Tsukaya 2017), we cannot support the recognition of Kalimantanorchis as a distinct genus.

Conclusions
According to morphological data by Kikuchi and Tsukaya (2017) and the highly supported positions of Kalimantanorchis inside Tropidia in our phylogenetic trees, we reject both grounds (i.e. the unique subterranean tuberous structure of Kalimantanorchis and the result of the molecular phylogenetic analysis based on matK) for the establishment Kalimantanorchis as a distinct genus suggested by Tsukaya et al. (2011). We support the synonymisation of Kalimantanorchis with Tropidia, as proposed by Ormerod and Juswara (2019).
Our results suggest that the loss of photosynthetic ability occurred at least twice in Tropidia. Genera with multiple transitions to non-photosynthetic full mycoheterotrophy have rarely been reported. These genera may play an important role in future studies of the evolution of mycoheterotrophy, as they may provide insights into the drivers of such transitions.