Plastid phylogenomics reveals evolutionary relationships in the mycoheterotrophic orchid genus Dipodium and provides insights into plastid gene degeneration

The orchid genus Dipodium R.Br. (Epidendroideae) comprises leafy autotrophic and leafless mycoheterotrophic species, the latter confined to sect. Dipodium. This study examined plastome degeneration in Dipodium in a phylogenomic and temporal context. Whole plastomes were reconstructed and annotated for 24 Dipodium samples representing 14 species and two putatively new species, encompassing over 80% of species diversity in sect. Dipodium. Phylogenomic analysis based on 68 plastid loci including a broad outgroup sampling across Orchidaceae found sect. Leopardanthus as sister lineage to sect. Dipodium. Dipodium ensifolium, the only leafy autotrophic species in sect. Dipodium was found sister to all leafless, mycoheterotrophic species, supporting a single evolutionary origin of mycoheterotrophy in the genus. Divergence time estimations found that Dipodium arose ca. 33.3 Ma near the lower boundary of the Oligocene and crown diversification commenced in the late Miocene, ca. 11.3 Ma. Mycoheterotrophy in the genus was estimated to have evolved in the late Miocene, ca. 7.3 Ma, in sect. Dipodium. The comparative assessment of plastome structure and gene degradation in Dipodium revealed that plastid ndh genes were pseudogenised or physically lost in all Dipodium species, including in leafy autotrophic species of both Dipodium sections. Levels of plastid ndh gene degradation were found to vary among species as well as within species, providing evidence of relaxed selection for retention of the NADH dehydrogenase complex within the genus. Dipodium exhibits an early stage of plastid genome degradation as all species were found to have retained a full set of functional photosynthesis-related genes and housekeeping genes. This study provides important insights into plastid genome degradation along the transition from autotrophy to mycoheterotrophy in a phylogenomic and temporal context.


Introduction
Heterotrophic plants -plants that rely on other organisms for energy and nutrients -are remarkable survivors, exhibiting often curious morphological, physical, or genomic modifications, reflecting evolutionary relaxed selective pressure on photosynthetic function (Graham et al., 2017;Barrett et al., 2019).Advances in next generation sequencing and bioinformatic pipelines have vastly accelerated the characterisation of plastid genomes (plastomes), including of heterotrophic plants, providing new insights into plastome evolution.
Orchidaceae, one of the two largest flowering plant families, has undergone a greater number of independent transitions from autotrophy to heterotrophy than any other land plant lineage (Merckx 2013;Christenhusz and Byng 2016;Jacquemyn and Merckx 2019).The family comprises several heterotrophic orchid lineages which rely to some extent on mycorrhizal fungi for carbon and other nutrients i.e., initial, partial, or full mycoheterotrophy (Merckx 2013).
So far, most examined mycoheterotrophic orchid plastomes exhibited degradation patterns similar to those found in heterotrophic plastomes of other plants.These include a reduction in genome size, decrease in guanine-cytosine (GC) content, rearrangements, pseudogenisations and gene losses (e.g., Delannoy et al., 2011;Barrett et al., 2018;Lallemand et al., 2019;Barrett et al., 2019;Wen et al., 2022).Moreover, whole plastome sequencing has revealed patterns of plastid gene degradation for various heterotrophic plastomes which led to the development of conceptual models to predict the evolutionary transition from autotrophy to heterotrophy of the plastid organelle (e.g., Graham et al., 2017;Barrett et al., 2019).Several studies in mycoheterotrophic orchid lineages found support for these models which predict a progression from losses of the chloroplast ndh genes to genes encoding complexes which are directly involved in photosynthesis (e.g., psa, psb) to more general 'housekeeping' genes (e.g., accD, matK) (Wicke and Naumann 2018;Barrett et al., 2018;Barrett et al., 2019;Kim et al., 2020;Kim et al., 2023).
Interestingly, degraded ndh genes were also found in some autotrophic orchids (e.g., Kim et al., 2015;Niu et al., 2017;Kim and Chase 2017;Lallemand et al., 2019;Kim et al., 2023).This appears curious, as the ndh genes encode proteins of the NADH dehydrogenase complex (NDH complex) which is assumed to play a role in cyclic electron flow and thus fine-tunes photosynthesis (Yamori et al., 2015;Peltier et al., 2016).Degradation of ndh genes is hypothesised to have led to additional structural changes of the plastome (Kim et al. 2015).In particular, ndhF gene loss was correlated with shifts in the position of the junction of the inverted repeat/small single copy (IR/SSC) region in Orchidaceae and other plants (Kim et al., 2015;Niu et al., 2017;Dong et al., 2018;Roma et al., 2018;Thode and Lohmann 2019;Li et al., 2021;Könyves et al., 2021).However, within Orchidaceae, degradation of ndh genes was found to vary even among closely related species (e.g., Kim et al., 2015;Feng et al., 2016;Kim and Chase, 2017;Barrett et al., 2018;Barrett et al., 2019) which suggests the genes for the NDH complex may be under relaxed selective pressure in several orchid lineages (Kim and Chase, 2017).Moreover, previous studies found that ndh degradation patterns vary considerably and have been independently degraded among orchids (Kim et al., 2015;Niu et al., 2017;Kim and Chase 2017;Lallemand et al., 2019).
Dipodium are generally assumed to be fully mycoheterotrophic (O 'Byrne, 2014).However, one Australian species of sect.Dipodium, D. ensifolium F.Muell., stands out as a leafy terrestrial (Figure 1, A).The aims of this study were to: 1. sequence and assemble plastid genomes for species of Dipodium to elucidate patterns of plastid genome modification (e.g., rearrangement, structural variation, pseudogenisation, gene loss) across autotrophs and mycoheterotrophs within the genus and examine gene degradation in context of current models of plastome degradation in heterotrophic plants.
3. estimate divergence times of Dipodium to assess the origin of mycoheterotrophy within the genus and elucidate over which evolutionary timeframes plastid gene degradation and losses has taken place within Dipodium.

2
Material and methods

DNA extraction, library preparation, and sequencing
Standard plant DNA extractions were carried out from 5-20 mg of silica dried plant tissue from field collections or herbarium material (Table 1) at the National Research Collections Australia (NRCA, CSIRO) in Canberra.The Invisorb DNA Plant HTS96 kit (Stratec, Birkenfeld, Germany) was used following the manufacturer's protocol, with a final elution of 60 ml.
DNA of Dipodium samples (Table 1) was sonicated to an average target length of ca.200 bp using a LE220 sonicator (Covaris, Bankstown, Australia).After sonication, DNA length and concentration were quantified on Fragment Analyzer (Agilent Technologies, California, USA) using the Agilent high-sensitivity genomic DNA kit.
DNA libraries were prepared using the QiaSeq UltraLow Input library kit (Qiagen, Germantown, Australia) using custom dual-indexed adapters.Final libraries were size-selected on Fragment Analyzer using the high-sensitivity Genomic Fragment Analyzer Kit (Agilent, Santa Clara, USA), quantified using the Fluoroskan plate fluorometer (Thermo Fisher Massachusetts, USA) and the Quant-iT HS dsDNA kit (Invitrogen, California, USA) following the manufacturer's instructions.Samples were pooled equimolarly and sequenced using 150 bp paired end reads on a NovaSeq S1 flowcell (Illumina, California, USA) at the Biomolecular Resource Facility within the John Curtin School of Medical Research, Australian National University (Canberra, Australia).

Data processing and whole plastid genome assembly
We carried out both de novo and reference-guided assemblies for the Dipodium data set.
Trimming and assembly of de novo contigs were carried out as described in Nargar et al.
Reference-guided assemblies were performed with paired, merged reads and the recently published and closely related plastome of Dipodium roseum D.L.Jones and M.A.Clem.
(KP205432, Kim et al., 2015) was included as an additional reference sequence to ensure that regions which already showed degradation in some plastid genes in the plastome of D. roseum (e.g., all ndh genes) (MN200386, Kim et al., 2020) and which may still be present in other Dipodium species could be assembled as the plastome of M. coccinea has a full set of functional plastid genes (Kim et al., 2015).
Reference-guided assemblies were carried out using the plugin 'map to reference' in Geneious Prime (Version 2022.0.2,Biomatters Ltd, www.geneious.com)with default settings.To obtain complete plastome assemblies, consensus sequences for each sample were extracted (threshold 60%, reading depth > 10), aligned using MAFFT v7.388 (Katoh and Standley 2013) in Geneious, manually checked and compared.Reference-guided assemblies were visually inspected and in cases of misassembled regions due to potential mismatches between the sample and the reference de novo assemblies were consulted, and were quality allowed the region extracted from the de novo assembly.The prediction and finding of gene annotations for complete plastome assemblies were performed with the Geneious plugin 'predict annotation' (similarity: 90% and best match with D. roseum (MN200386)).Open reading frames (ORFs) were manually checked and verified by identifying the start and stop codons.
In cases of remaining ambiguities, BLAST searches were conducted for reading-frame verification (Altschul et al. 1990; National Center for Biotechnology Information; Available from: https://blast.ncbi.nlm.nih.gov/Blast.cgi[cited: 08 Sept 2023]).The inverted repeat (IR) boundaries were identified using the 'repeat finder' plugin in Geneious with default settings.
In total, 24 complete Dipodium plastomes were assembled in this study.The graphical representation of each plastome and divergent regions with annotations were created in OrganellarGenomeDRAW (OGDRAW, version 1.3.1, Greiner et al., 2019).

Phylogenetic analyses
To elucidate phylogenetic relationships within Dipodium and to assess the phylogenetic position of Dipodium within Cymbidieae we performed a phylogenetic analysis with DNA sequences of 33 newly sequenced plastomes from this study (Table 1) and an extended outgroup sampling for 115 samples from published plastid data (Supplementary Material 1).
Coding regions of respective genes of 33 samples were extracted with the 'extract' function in Geneious Prime.Where mutations had led to frame shifts with internal stop codons, the affected sequences were excluded from phylogenetic analyses.
Each extracted coding region of in total 68 plastid loci from 33 samples (including the intron regions) and from 115 published plastomes (excluding intron regions) were aligned using MAFFT (v7.388;Katoh et al., 2002;Katoh and Standley 2013) Geneious prime plugin with default settings, checked manually and subsequently concatenated to an alignment of 69,335 bp (Supplementary Material 2).

Divergence-time analysis
For divergence-time estimations of Dipodium, the alignments were reduced to the 30 most parsimony informative loci due to computational limitations.The 30 plastid loci were selected based on their most parsimony informative (Pi) sites estimated with MEGA (Molecular Evolutionary genetics Analysis;ver. 11.0.11, Tamura et al., 2021) and presence of loci across the dataset (Supplementary Material 2).For taxa represented by more than one sample, duplicates were removed from alignments as recommended for divergence time estimation.
Alignments of 30 plastid loci from 134 taxa were concatenated yielding a total alignment length of 27,934 bp using MAFFT (v7.388;Katoh et al., 2002;Katoh and Standley 2013) implemented in Geneious Prime (Supplementary Material 2).Absolute node ages and phylogenetic relationships were jointly estimated in BEAST (ver. 2.7.4;Bouckaert et al., 2019, Bouckaert et al. 2014) applying the best fit partition scheme and substitution model as determined by IQ-TREE's ModelFinder (GTR+F+I+I+R4).Four different models were tested: a Bayesian optimised relaxed and a strict molecular clock with uncorrelated lognormal rates with each a Yule and a Birth-death tree prior on the speciation process (Douglas et al., 2021;Gernhard et al., 2008;Zuckerkandl and Pauling, 1965;Yule, 1925).

Structural variation in Dipodium plastomes
To examine structural variation among the plastomes of Dipodium, whole plastome alignments were generated using MAFFT (v7.388;Katoh et al., 2002;Katoh and Standley 2013) implemented in Geneious Prime with full annotations.Alignments were manually checked, in cases of divergent regions e.g., the operon region of ndhC, ndhK, and ndhJ genes or junctions between the large single copy (LSC)/ inverted repeat B (IRB)/ small single copy (SSC)/ inverted repeat A (IRA) regions, and respective regions (including annotations) were extracted in Geneious Prime, separately aligned, proofread, and subsequently visualised using OGDRAW (ver. 1.3.1, Greiner et al., 2019).

Functional genes, pseudogenes, and physical gene loss
To classify the level of degradation of plastid genes in Dipodium, we used the following categories: (1) functional -the reading frame was intact and less than 10% of the open reading frame was disrupted by small indels; (2) moderately pseudogenised -less than 10% of the open reading frame was disrupted by internal stop codons or indels causing non-triplet frame shifts; (3) severely pseudogenised -more than 10% of the open reading frame was disrupted by either internal stop codons, large deletions (> 10%), and non-triplet frame shifts (based on Barrett et al., 2019), or (4) lost -the gene was not identified in the annotation process of the de-novo assembly (e.g., Joyce et al., 2018) and/or was not detectable within the reference-guided assembly.A gene was considered as not detectable within the reference-guided mapping process if at least 70% of the gene sequence could not be identified for calculation of the consensus sequence within the Geneious mapping process.The coded matrix of gene degradation was plotted against the maximum likelihood phylogenetic tree of Dipodium.

Phylogenetic placement of Dipodium in tribe Cymbidieae and infrageneric relationships within the genus
The maximum likelihood analysis based on 68 plastid loci and 148 samples yielded highly resolved and well-supported tree topologies for the phylogenomic relationships within Orchidaceae (Supplementary Material 3).Within Epidendroideae, Cymbidiinae was

Divergence-time estimations
Absolute times of divergence under strict and optimised relaxed clocks for Orchidaceae based on 30 plastid loci and 134 taxa showed similar results.Strict clock models consistently yielded slightly older age estimates than the analyses based on the relaxed clock models (Supplementary Material 4).Model comparison using AICM (Fabozzi et al., 2014) identified the relaxed clock model under the birth-death speciation model as the best fit models for the dataset (Supplementary Material 4).
The Bayesian relaxed clock tree topology and the maximum likelihood phylogeny agreed overall in major relationships within Orchidaceae and the placement of species within

Characterisation of Dipodium plastomes
Complete plastome assemblies and annotations were successfully carried out for 24 Dipodium samples, representing all Australian species of section Dipodium including two recently discovered species of section Dipodium (D. ammolithum and D. basalticum), two putatively new species of section Dipodium (D. aff.roseum, D. aff.stenocheilum) and one species of section Leopardanthus (D. pandanum) (Table 2).Plastome assemblies for D. pandanum 2 and D. aff.stenocheilum showed an insufficient mean coverage (<30) for non-coding regions which caused unsolved gaps and ambiguous bases which could not be reliably resolved.The number of paired-end, trimmed reads for the successfully assembled complete plastomes ranged from 332,604 (D. pandanum 1) to 27,999,734 (D. pardalinum 2) and the mean coverage ranged from 31x to 627x (Supplementary Material 6).
The plastid genes of each plastome were rated as functional; moderately to severely pseudogenised; or physically lost.The total number of functional genes in Dipodium plastomes ranged slightly from 119 to 121 including a total of 73 or 74 functional protein-coding sequence regions (CDS) (68 or 69 unique CDS), 37 to 39 functional tRNA genes (30 or 31 unique tRNA genes) and 8 rRNA genes (4 unique rRNA genes) (Table 2).
The IR region was largely conserved among all examined Dipodium plastomes.All species showed six duplicated coding regions in the IRs (i.e., rpl2, rpl23, rps7, rps12, rps19, ycf2) and all four rRNA genes (  3, Figure 5).The rps12 gene was transspliced with the 5' end located in the LSC region and 3' end was duplicated in the IRs in all studied plastomes (Figure 5, Supplementary Material 7).
The SSC region was found to vary the most among the examined samples.All plastomes showed a contraction of the SSC with a reduction of 20-40% compared to the average size of to 1,072 bp (D. aff.roseum 3) (Figure 5).
In contrast to the instability of the IR/SSC boundaries, IR/LSC boundaries were found to be relatively stable.For all studied plastomes, the LSC/IRA boundaries were located near the 3' end of psbA (Figure 5).Variations within the LSC regions were limited to the operon which contained ndhC, J, K (Figure 5, A) and the independent pseudogenisation of cemA in the plastome of D. aff.roseum 4 and trnD-GUC in the plastome of D. campanulatum (Table 3, Supplementary Material 7).

ndh gene degradation and loss in Dipodium
All ndh genes exhibited varying degrees of putative loss or pseudogenisation; not a single ndh gene remained functional in the examined Dipodium plastomes (Table 3, Figure 5 & 6).
The most severe ndh gene loss occurred in the plastome of D. variegatum, with ndhA, ndhB, ndhD and ndhK severely pseudogenised and ndhJ moderately pseudogenised.
The greatest degradation processes within Dipodium occurred for the ndhG gene, which was putatively lost in almost all plastomes, except D. stenocheilum 1 which retained a severely pseudogenised ndhG gene (Figure 6).This was followed by ndhK, which was lost in 19 out of  O 'Byrne, 2017;Jones, 2021), lending support to the diagnostic value of vegetative traits (i.e., the presence or absence of adventitious roots) in infrageneric classification of Dipodium.
Section Leopardanthus is characterised by leafy species which possess adventitious roots, such as Dipodium pandanum.In contrast, sect.Dipodium comprises species without adventitious roots and includes all leafless species, the leafy species D. ensifolium, and the morphologically similar D. gracile from Sulawesi, the latter being only known from the type (destroyed) (O 'Byrne, 2017).Our phylogenomic study supported the placement of the D. ensifolium in sect.Dipodium, resolved as sister to all leafless species in the section.However, further molecular study is warranted to ascertain the monophyly of the two sections based on an expanded sampling of sect.Leopardanthus.
Our phylogenomic study is the first to shed light on evolutionary relationships within sect.
Dipodium, which was found to comprise six main lineages.The phylogenomic framework now allows assessment of useful diagnostic morphological traits to characterise main lineages within the section.For example, the yellow stem and flower colour of species of the D.
hamiltonianum complex easily distinguishes this clade from other mycoheterotrophic orchids within sect.Dipodium (Figure 1; Jones, 2021).Stems of remaining mycoheterotrophic species of sect.Dipodium are mostly greenish to dark reddish or purplish, whereas flowers vary in color from pale white, pinkish to purplish (Figure 1, Barrett et al., 2022;Jones, 2021).Also, sepal and petal characters were found to differ among clades: for example, species of clade A, comprising the D. hamiltonianum and D. stenocheilum complexes, possess sepals and lateral petals that are markedly narrower compared to species of clade B (comprising the D. punctatum and D. roseum complex) and D. ensifolium, the first diverging lineage within the sect.
Phylogenetic divergence within the two species complexes in clade B, i.e., the D. punctatum and the D. roseum complexes, was shallow overall and thus interspecific relationships in these two groups remained largely unclear (Figure 3).Previous morphological studies highlighted difficulties in species delimitation within the D. punctatum complex, in particular between D.
pulchellum and D. punctatum (Jones, 2021).While D. pulchellum is morphologically very similar to D. punctatum, the two species are differentiated by the intensity of their flower colours, which are richer in D. pulchellum and paler in D. punctatum (Jones and Clements, 1987).However, a morphological study by Jones (2021)  atropurpureum possesses dark pinkish-purple to dark reddish-purple flowers with spots and blotches, and the flowers of D. pardalinum are pale pink to white with large reddish spots and blotches (Figure 1) (Jones, 2021).Taken together, the overlapping distribution, similar appearance, and very shallow genetic divergence found in the present study among species in the D. roseum complex suggest that D. atropurpureum and D. pardalinum may be colour variations of D. roseum.Further molecular study with more highly resolving molecular techniques such as genotyping-by-sequencing is required to rigorously assess species delimitation within Dipodium.

Divergence-time estimations
Our divergence time estimations yielded results comparable to previous studies regarding the temporal diversification of major orchid clades (e.g., Givnish et al., 2015, Givnish et al., 2018;Kim et al., 2020;Serna-Sánchez et al., 2021;Zhang et al., 2023).Within Epidendroideae, this study confirmed that Cymbidieae was one of the most recently diverged tribes in Orchidaceae, consistent with previous studies (e.g., Givnish et al., 2015;Serna-Sánchez et al., 2021;Zhang et al., 2023).Our study is the first to elucidate phylogenetic relationships and divergence times within Dipodium.Previously, only two studies included a representative of Dipodium (D. roseum, MN200368) in divergence-time estimations for Orchidaceae (Kim et al., 2020;Serna-Sánchez et al., 2021).These studies estimated the origin of Dipodium to ca. 17 pattern of floristic exchange -the Sunda-Sahul Floristic Exchange -which was initiated as early as c. 30 Ma (Crayn et al., 2015;Joyce et al., 2021b).However, the data are insufficient at present to resolve the ancestral area of Dipodium and its main lineages.Further research is needed including an increased sampling to shed light on range evolution of Dipodium through ancestral range reconstruction.
Our results indicate that the Australian leafy species D. ensifolium diverged from the remainder of section Dipodium approximately 8.1 Ma (late Miocene) (Figure 4).The remainder of the sect.Dipodium clade, which includes all leafless, putatively fully mycoheterotrophic species, emerged ca.7.3 Ma (late Miocene) followed by rapid diversification from ca. 4.3 Ma onwards (early Pliocene) (Figure 4).Thus, mycoheterotrophy has most likely evolved only once within Dipodium, on the Australian continent during the late Miocene-early Pliocene.
From the late Miocene-early Pliocene (ca. 5 Ma) climatic conditions in Australia became increasingly arid, leading to a decline of rainforest vegetation and expansion of open sclerophyllous forests (Quilty, 1994, Gallagher et al., 2003, Martin, 2006, He and Wang, 2021).
By the end of the Pliocene Australia's landscape was similar to the present day, with much of the continent a mosaic of open woody vegetation dominated by Eucalyptus, Acacia and Casuarinaceae (e.g., Martin 2006).The Pleistocene (ca.2.58 -0.012 Ma) was characterised by climatic oscillations which led to repeated forest expansion and contraction (Byrne, 2008).The

Plastome structural features and variations
In this study, whole plastome assemblies were generated for 24 Dipodium samples, including representatives of all leafless, putatively full mycoheterotrophs of sect.Dipodium found in Australia, one leafy photosynthetic species of sect.Dipodium (D. ensifolium) and one leafy photosynthetic species of sect.Leopardanthus (D. pandanum).The overall organisation and the plastid gene content is generally conserved in most examined Dipodium plastomes (Figure 5, Table 2 and 3).All examined plastomes showed the typical quadripartite structure of angiosperms (Ruhlman and Jansen 2014).However, some genomic features among several Dipodium plastomes were not conserved, including 1) differences in total genome length; 2)  ) and all fell into the range of typical angiosperm plastomes (ca.30-40%) (Table 2).

Patterns of ndh gene degradation within Dipodium
In orchids, ndh gene losses and pseudogenisations which occurred in both autotrophic and heterotrophic species have been documented in various genera (e.g., Kim et al., 2015;Feng et al., 2016;Niu et al., 2017;Barrett et al., 2018, Barrett et al., 2019;Roma et al., 2018;Lallemand et al., 2019;Kim et al., 2020;Peng et al., 2022;Kim et al., 2023).This study is in line with these general findings in that ndh gene degradation was also observed within the orchid genus Dipodium.All chloroplast ndh genes in Dipodium plastomes exhibited varying degrees of putative pseudogenisation and loss, not a single ndh gene remained functional among the examined chloroplast genomes (Table 3, Figure 5, Figure 6).These findings include all plastomes of leafless putatively fully mycoheterotrophic species and of two autotrophic leafy species (D. pandanum and D. ensifolium) and thus suggest that all examined species, independently of their nutritional status, have lost the functionality of the plastid NADH dehydrogenase complex.Hence, the last common ancestor of extant Dipodium is likely to have lacked a functional NDH complex.Previous studies in Cymbidiinae, the first diverging lineage in Cymbideae, found that all species studied so far exhibited at least one degraded ndh gene (e.g., Yang et al., 2013;Kim and Chase 2017).As the next diverging lineage in Cymbidieae is Dipodium, this suggests that the degradation of ndh genes in Cymbidieae was likely a dynamic process from functional to non-functional.However, further research is needed e.g., ancestral state reconstructions of gene degradation with increased taxonomic sampling.The inclusion of more species among sect.Leopardanthus is warranted to clarify if some ndh genes have remained functional in some autotrophic species of sect.Leopardanthus.
Previous studies examined ndh gene loss at genus level and revealed an independent loss of function of the NADH dehydrogenase complex for several genera (e.g., Lin et al., 2015, Kim et al., 2015).However, comparative whole plastome studies examining gene degradation and loss among closely related mycoheterotrophic species are still scarce.For a better understanding of ndh gene degradation patterns this study investigated the degree of ndh gene degradation among closely related orchid species (Figure 6).Greatest degradation within Dipodium were found for ndhG which is putatively lost in almost all examined plastomes, except D. stenocheilum 1 which retained a putative severely pseudogenised ndhG (Figure 6).
The ndhG gene is located within the SSC region.In general, it is well established that genes in the SSC region experience higher substitution rates compared to genes located within IR regions (Ruhlman and Jansen 2014).The latter is the case for ndhB which is located in the IRs and structurally more conserved in Dipodium compared to most ndh genes located in the SSC.
The greatest degree of ndh gene degradation occurred in D. variegatum which putatively lost ndhC and ndhE-ndhI.All other plastomes putatively lost at least one to three ndh genes and showed different levels of degradation (Figure 6).
Interestingly, the level of ndh gene degradation varied even among closely related species within species complexes.For example, D. stenocheilum 1 independently lost ndhI and ndhF, whereas all other studied samples of the D. stenocheilum complex retained those two genes as moderately or severely pseudogenised (Figure 6).Different levels of gene degradation and loss were even found within the same species.For example, D. atropurpureum 1 lost ndhC whereas D. atropurpureum 2 retained a severely pseudogenised ndhC (Figure 6).Moreover, the study of Kim et al., (2020)  Overall, some patterns of ndh gene degradation found in this study in Dipodium are similar, however many were unique for each individual examined.Hence, this suggests that sect.
Dipodium has undergone a recent and active ndh gene degradation which strongly implies a relaxed evolutionary selective pressure for the retention of the NDH complex.

IR/SSC junctions and IR instability
Orchidaceae plastomes frequently show an expansion/shift of the IR towards the SSC region (e.g., Kim et al., 2020).This instability of the IR/SSC junction is assumed to correlate with the deletion of ndhF and has resulted in a reduction of the SSC, as observed in several Orchidaceae plastomes (e.g., Kim et al., 2015;Niu et al., 2017;Dong et al., 2018;Roma et al., 2018) and in other land plant plastomes (e.g., Amaryllidaceae, Bignoniaceae, Orobanchaceae) (Thode and Lohmann 2019;Li et al., 2021;Könyves et al., 2021).This study revealed reduced SSC regions for most examined plastomes which correlated with the degradation of the ndh gene suite located in the SSC.Compared to typical SSC regions found in angiosperms (ca.20 kb, Ruhlman and Jansen 2014), the smallest SSC region was reduced by ca.7,900 bp (D. variegatum) and the largest SSC region was reduced by ca.4,700 bp (D. ensifolium) (Table 2, Figure 5).However, a large expansion of the IR such as found in Vanilla and Paphiopedilum plastomes (Kim et al., 2015) was not found in Dipodium (IR sizes ranging between 24,436-26,817 bp, Table 2).
In angiosperms, the ycf1 gene usually occupies ca.1,000 bp in the IR (Sun et al., 2017, Kim et al., 2015).Dipodium plastomes in this study displayed varying positions of ycf1 within the IR.
In plastomes in which the ndhF gene was completely lost or severely truncated, the portion of ycf1 within the IRA was mostly shorter compared to plastomes which contained moderately truncated ndhF genes (Figure 5).These results are similar with findings of Kim et al., (2015), the IR (M.archiducis-nicolai and M. ruthenica) which were linked to forward and tandem repeats.Interestingly, Wu et al. (2021) findings support the hypothesis that repetitive sequences lead to genomic rearrangements and thus affect plastome stability.This may also apply for some Dipodium plastomes.However, to rule out any technical issues throughout the NGS process and to validate findings of duplicated tRNAs (and above-mentioned boundaries of IR/SC regions), PCR amplification of affected regions should be carried out in future studies.
However, in strong support of tRNA duplication is their independent presence within the IR of five plastomes among individuals of the same species complexes (D. stenocheilum complex and D. hamiltonianum complex).However, an increased sampling is necessary to better understand the impacts of genomic rearrangements due to repetitive sequences and thus plastome instability in Dipodium.

Evolution of mycoheterotrophy and associated plastome degradation in Dipodium
Heterotrophic plants are remarkable survivors, exhibiting often curious morphological, physical, or genomic modifications.Multiple heterotrophs were found to have suffered plastid genome degradations due to relaxed pressure on photosynthetic function.In recent years, evidence has accumulated that plastid genomes have undergone gene degradation in the evolutionary transition from autotrophy to heterotrophy (e.g., Graham et al., 2017;Barrett et al., 2019;Wicke et al., 2016).Among these, the first stage is the loss and pseudogenisation of genes involved in encoding the NDH complex.Interestingly, all examined plastomes of Dipodium have lost or pseudogenised all 11 ndh genes regardless of their nutritional status (Figure 6).Two photosynthetic species with green leaves were included in this study, D. pandanum (sect.Leopardanthus) and D. ensifolium (sect.Dipodium).Degradation in ndh genes among photosynthetic species is not surprising and was frequently reported in previous plastome studies in land plants.The large-scale study on Orchidaceae plastomes of Kim et al., (2020) observed ndh gene pseudogenisation and losses among species in many epiphytes and several terrestrials which have retained their photosynthetic capacity.The NDH complex is thought to mediate the Photosystem I cyclic electron transport, fine-tunes photosynthetic processes and alleviates photooxidative stress (e.g., Yamori et al., 2015;Peltier et al., 2016;Sabater 2021).D. pandanum is a terrestrial or climbing epiphytic orchid and highly localised in rainforest habitats, whereas the terrestrial D. ensifolium grows in open forests and woodlands (Jones, 2021), thus both species seem to prefer shaded understory habitats.For epiphytic or terrestrial plants living in low-light habitats it has been proposed that the NDH complex may not be essential anymore (e.g., Barrett et al., 2019).One reason for this may be that they are less exposed to photooxidative stress (e.g., Feng et al., 2016;Barrett et al., 2019).However, the NDH complex is composed of 11 chloroplast encoded subunits and additional subunits encoded by the nucleus (e.g., Peltier et al., 2016).It has been established that genomic material was repeatedly exchanged between the nucleus, mitochondrion, and chloroplast in the evolutionary course of endosymbiosis.Thus, previous studies examined whether genes were transferred from the chloroplast to the nucleus and/or mitochondrion genome or whether nuclear genes for the NDH complex suffered under degradation.Indeed, Lin et al., (2015) reported ndh fragments within the mitochondrial genomes of orchids, however no copies were found in the nuclear orchid genomes.Similar findings were reported from the orchid genus Cymbidium (Kim and Chase et al., 2017).However, further studies are needed to determine whether ndh gene transfer into the nucleus or mitochondrion may play a role within Dipodium.
The proposed subsequent next steps toward (myco-) heterotrophy is the functional loss of photosynthetic genes (e.g., psa, psb, pet, rbcL or rpo) followed by genes for the chloroplast ATP synthase and genes with other function such as housekeeping genes (e.g., matK, rpl, rnn (e.g., Graham et al., 2017;Barrett et al., 2019).Most examined Dipodium plastomes displayed no additional plastid gene degradation besides ndh gene degradation, except in D. aff.roseum 4 where cemA was pseudogenised and in D. campanulatum 1 where the trnD-GUC gene was pseudogenised (Table 3).The cemA gene encodes the chloroplast envelope membrane protein and was found to be non-essential for photosynthesis, however cemA-lacking mutants of the green alga Chlamydomonas were found to have a severely affected carbon uptake (Rolland et al., 1997) and may therefore be classified as directly involved in photosynthesis.Transfer RNA genes (trn) are involved in the translation process and categorised as 'housekeeping' genes (e.g., Graham et al., 2017;Wicke and Naumann 2018;Barrett et al., 2019).Moreover, similar gene degradation patterns were found in the plastomes of D. roseum (MN200386, Kim et al., 2020 andOQ885084, Mclay et al., 2023) and D. ensifolium (OQ885084, Mclay et al., 2023), which functionally lost all ndh genes.However, most photosynthesis related genes in the plastomes of Dipodium were found to be functional.Thus, mycoheterotrophic species of Dipodium display evidence of being at the beginning of plastid gene degradation, in contrast with the majority of fully mycoheterotrophic orchids which are in more advanced stages of degradation, e.g.Cyrtosia septentrionalis (Kim et al., 2019), Epipogium (Schelkunov et al., 2015), and Rhizanthella (Delannoy et al., 2011).On the other hand, mycoheterotrophs such as Corallorhiza trifida (Barrett et al., 2018), Cymbidium macrorhizon (Kim et al., 2017), Hexalectris grandiflora (Barrett et al., 2019) and Limodorum abortivum (Lallemand et al., 2019) display functionally losses within the plastid ndh genes only and some species among them additionally lost one or two other genes, similar to findings in Dipodium.Interestingly, most of these species are leafless, but considered putatively partially mycoheterotrophic.Suetsugu et al. (2018) demonstrated that the leafless green orchid Cymbidium macrorhizon contains chlorophyll and can fix significant quantities of carbon during the fruit and seed production phase and thus, is photosynthetically active.Chlorophyll is present in Corallorhiza trifida also, but this green, leafless coralroot is an inefficient photosynthesiser (Barrett et al., 2014).Some species among leafless orchids within sect.psbA, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ, psbK, psbL, psbM, psbN, psbT,

Figure 2 :
Figure 2: Phylogenetic relationships among major orchid lineages and placement of subtribe Dipodiinae in Cymbidieae.Maximum likelihood tree of 148 taxa based on 68 plastid loci.Support values are shown above each branch, SHaLRT followed by UFBoot values.Scale bar represents branch length, along which 0.02 per-site substitutions are expected.Detailed phylogeny provided in Supplementary Material 3.

Figure 3 :
Figure 3: Phylogenetic relationships in Dipodium.Maximum likelihood tree based on 68 plastid loci and 148 taxa (outgroups not shown).Support values are given above each branch, SHaLRT is followed

Figure 4 :
Figure 4: Chronogram of Cymbidieae.Maximum-clade-credibility tree from Bayesian divergence-time estimation in BEAST2 based on 30 plastid loci and an optimised lognormal molecular clock model under the birth-death prior (outgroups not shown).Divergence times (million years ago) are shown at each node, together with 95% highest posterior density (HDP) values indicated by grey bars and values in parentheses.A and B refers to the two main lineages within sect.Dipodium.Cy: Cymbidiinae, Ct: Catasetinae, Cr: Cyrtopodiinae, Eu: Eulophiinae, On: Oncidiinae, St: Stanhopeinae, Ma: Maxillariinae, the angiosperm SSC regions (ca.20 kb)(Ruhlman and Jansen 2014).Three plastomes (D. pandanum 1, D. stenocheilum 1, D. variegatum) lost the ndhF gene.This complete loss of the ndhF gene resulted in the ycf1 fragment being located in the vicinity of the rpl32 (Figure5, D, b-d) and caused a boundary shift of the IRB/SSC region located at the 3' end of the ycf1 fragment and spacer region of rpl32 (Figure5).While all other plastomes exhibited a severely truncated ndhF gene but did not exhibit an IRB/ SSC boundary shift (Figure5, Supplementary Material 7).The IRA/SSC junction in all examined plastomes was located within the 5' portion of the functional ycf1 gene, ranging from 97 bp (D. pandanum 1)

Figure 5 :
Figure 5: Plastome map and boundary shifts in Dipodium.The plastome of D. atropurpureum 2 is illustrated as representative and shown as a circular gene map with the smallest and the largest Dipodium plastome of this study.Genes outside the circle are transcribed in a clockwise direction, those inside the circle are transcribed in a counterclockwise direction.The dark grey inner circle corresponds to the G/C content, and the lighter grey to the A/C content.The major distinct regions of complete Dipodium plastomes are compared in each detailed enlarged box (A-D).(A) Note that each representative block (a-d) has pseudogenised or lost either ndhJ, ndhK or ndhC genes.(B, C).Duplication of trnV-GAC in the Inverted Repeat regions of D. interaneum (IRB), D. hamiltonianum (IRA), D. elegantulum (IRB), D. stenocheilum 2 (IRA), D. ammolithum (IRA).(D) Each block (a. as representative D. roseum 2; b.D. pandanum 1; c.D. stenocheilum 1; d.D. variegatum) shows differences in the length (bp) of the SSC region caused through loss or pseudogenization of either ndhF, ndhD, ndhE, ndhG, ndhI, ndhA or ndhH, note the boundary shift of the IRs/SSC region caused through the loss/ pseudogenisation of ndhF and the inclusion of the functional ycf1 and the ycf1-fragment (grey, dashed line) into the IRs.SSC: Small Single Copy; LSC: Large Single Copy: IRA/B: Inverted Repeat A/B.

Figure 6 :
Figure 6: Pattern of putative ndh gene degradation in Dipodium.Gene degradation plotted against the maximum likelihood tree with focus on 24 fully assembled plastomes.(outgroups not shown).Support values (SHaLRT/ UFboot) are shown on each branch.ps = pseudogenisation; D. ham.= D. hamiltonianum revealed that the strong floral coloration of D. pulchellum flowers was likely due to differences in environmental factors (i.e., soil type and rainfall regime) of growing sites and thus Jones (2021) proposed to synonymise D. pulchellum with D. punctatum.Similar challenges in taxonomic delimitation based on flower colours are also evident within the D. roseum complex.The distribution of the more widespread species D. roseum largely overlaps with the distributions of D. atropurpureum and D. pardalinum (ALA, 2023).Besides a very similar growing habit, the flowers of the three species are very similar in shape and vary only slightly in coloration: D. roseum has bright, rosy flowers with small darker spots, D.
Stem and crown diversification of Cymbidieae were estimated to have commenced at ca. 42.2 Ma and 38.0 Ma respectively, which is similar to the estimates of Serna-Sánchez et al. (2021) and slightly younger than those of Zhang et al. (2023) (Figure 4, Supplementary Material 4 and 5).
Ma and ca. 31 Ma, respectively.Our study placed the divergence of Dipodium from the other subtribes in Cymbidieae to ca. 33.3 Ma in the early Oligocene which is closer to the findings of Serna-Sanchez et al. (2021).O'Byrne (2014) hypothesised that lineage divergence into sect.Dipodium and sect.Leopardanthus resulted from vicariance in conjunction with the break-up of Pangaea, in particular the separation of the Indian and Australian continental plates (O'Byrne, 2014).However, our divergence-time estimations show that Dipodium is far too young (< 33 Ma) to have been influenced by the break-up of Pangaea, which occurred from the early Jurassic and onwards.Lineage divergence of sect.Dipodium and sect.Leopardanthus were estimated to ca. 11.3 Ma in the late Miocene (Figure 4), when Australia had already assumed, approximately, its present geographical position.Rather, Dipodium is likely to have achieved its current distribution through range expansion between Australia and Southeast Asia across the Sunda-Sahul Convergence Zone (Joyce et al. 2021a), consistent with a general evolution of mycoheterotrophy and the subsequent radiation of sect.Dipodium may have been facilitated by two factors: aridification in Australia favouring the reduction of leaf area to decrease water loss (O'Byrne, 2014), and the expansion of sclerophyll taxa and their mycorrhizal partners.Mycoheterotrophic Dipodium are assumed to share mycorrhizal fungi with Myrtaceae trees, especially Eucalyptus, (Bougoure and Dearnaley, 2005; Dearnaley and Le Brocque, 2006; Jones, 2021) which explosively diversified and came to dominate most Australian forests and presumably led to an increased diversity and abundance of suitable mycorrhizal partners for Dipodium.The rapid diversification of Dipodium from the Pleistocene onwards (ca.3.2-0.3Ma) (Figure 4) may have been driven by cycles of population fragmentation and coalescence in response to climatic oscillations.
study which compared the locations of the IR/single-copy region junctions among 37 orchid plastomes and closely related taxa in Asparagales.In at least three plastomes (D. pandanum 1, D. stenocheilum 1, D. variegatum) ndhF was independently lost, the SSC/IRB junction was shifted into the spacer region near the rpl32 gene in direct adjacency to the partially duplicated ycf1 fragment (Figure 5, D, b-d).These findings suggest the deletion of ndhF correlated with the shift of the SSC/IRB junction.Interestingly, the boundaries between SSC and IR regions were found to be variable even among closely related species e.g., in Cymbidium.Some species in Cymbidium showed similar patterns of IR/SSC shifts (Kim and Chase 2017) as found in Dipodium.In at least five plastomes (D. ammolithum, D. elegantulum, D. hamiltonianum, D. interaneum, D. stenocheilum 2) the trnV-GAC gene was triplicated (i.e., duplicated trnV-GAC version in close proximity to each other either in IRA or IRB) (Figure 5, B, C; Table 3).To the best of our knowledge, similar tRNA duplication patterns within the IR regions have not yet been found in any other Orchidaceae plastome.However, a recent study on plastomes of the angiosperm genus Medicago (Wu et al., 2021) yielded similar patterns.Wu et al. (2021) have found three copies of the trnV-GAC gene in the plastomes of two closely related species within

Table 3 )
. Most plastomes showed eight duplicated tRNA genes in the IR regions with exception of the plastomes of D. interaneum and D. elegantulum which comprised a duplicated trnV-GAC within the IRB and the plastomes of D. ammolithum, D.

Table 1 .
(Kim et al. 2020;McLay et al. 2023)   stenocheilum, D. variegatum)appear green on stems (Figure1,Jones 2021), which suggests they may contain some chlorophyll and be able to photosynthesise.Coupled with relatively mild plastid gene degradation compared to other fully mycohetrotrophic orchids, this suggests some leafless species among sect.Dipodium may be partially mycoheterotrophic rather than fully mycoheterotrophic as has been hypothesised for D. roseum(Kim et al. 2020;McLay et al. 2023).However, no studies so far have examined whether leafless species among sect.Dipodium diversified in the late Miocene ca.11 Ma, and the mycoheterotrophic lineage divergent from the autotrophic lineage ca.8.1 Ma which is slightly younger compared to Corallorhiza.Hence, time of divergence may play a role in the degree of degradation of Dipodium plastomes which show an early stage of plastome degradation Dipodium, corresponding to the morphologically defined sect.Dipodium and sect.Leopardanthus.Phylogenetic analysis resolved the leafy autotroph D. ensifolium as being part of sect.Dipodium and found to be in sister group position to all leafless species in sect.Dipodium.Divergence-time estimations placed the divergence of the leafy species D. ensifolium from the remainder of section Dipodium in the late Miocene.Shortly after, the remaining clade including all leafless, putatively full mycoheterotrophic species within sect.Dipodium emerged ca.7.3 Ma in the late Miocene followed by rapid species diversification from ca. 4.3 Ma onwards in the early Pliocene.Thus, this study indicates that mycoheterotrophy has most likely evolved only once on the Australian continent within Dipodium during the late Miocene, and that the ancestors of putatively full mycoheterotrophic species may have had green leaves.Among the examined plastomes, all plastid ndh genes were pseudogenised or physically lost, regardless of the individual's nutrition strategy (i.e., autotroph versus mycoheterotroph).Thus, this study provides molecular evidence of relaxed evolutionary selective pressure on the retention of the NADH dehydrogenase complex.Dipodium based on physiological data such as from the analysis of chlorophyll quantities and the ratio of photosynthetic carbon to fungal carbon for Dipodium.The Australian orchid flora Plant material used in this study inclusive voucher details and provenances with botanical districts.Taxonomy according to the Australian Plant Census (APC, 2023).CANB = Australian National Herbarium, CNS = Australian Tropical Herbarium.AU = Australia, PG= Papua New Guinea.ACT = Australian Capital Territory, NT = Northern Territory, NSW = New South Wales, SA = South Australia, QLD = Queensland, WA = Western Australia, Vic = Victoria.
Dipodium contain chlorophyll and whether they are capable to carry out photosynthesis at sufficient rates.Therefore, more research is needed to assess the trophic status, including analysis of chlorophyll quantities and the ratio of photosynthetic carbon to fungal carbon forDipodium.Compared with recently published studies on mycoheterotrophic orchids such as Corallorhiza and Hexalectris(Barret et al. 2018; Barret et al. 2019) which incorporated divergence time estimations, plastomes of Dipodium showed the least degradation.Hexalectris crown age was estimated to ca. 24 Ma and plastomes of mycoheterotophs were more degraded compared to mycoheterotrophic plastomes of Corallorhiza which diversified ca. 9 Ma onwards(Barret et    al. 2018; Barret et al. 2019).5 ConclusionThis molecular phylogenomic comparative study clarified evolutionary relationships and divergence times of the genus Dipodium and provided support for two main lineages within Mycoheterotrophic species among sect.Dipodium retained a full set of other functional photosynthesis-related genes and exhibited an early stage of plastid genome degradation.Hence, leafless species of sect.Dipodium may potentially be rather partially mycoheterotrophic than fully mycoheterotrophic.To further disentangle evolutionary relationships in Dipodium, future studies based on nuclear data such as derived from target capture sequencing and with a denser sampling at population level are warranted.Moreover, the inclusion of a denser sampling of sect.Leopardanthus is warranted to clarify if some ndh genes may have remained functional in some of the autotrophic species of sect.Leopardanthus.To obtain further insights into the nutritional strategies in Dipodium, future studies should assess the trophic status of mycoheterotrophic species in

Table 2 .
Comparison of plastome features in Dipodium.