Owing to the difficulty in positively identifying Nervilia species in the absence of correctly matched flowers and leaves, we included a mix of both named and unnamed accessions to account for as wide a cross section of the genus as possible and so permit an examination of patterns of genetic disparity among morphologically similar plants. Further, we included two or more accessions from different locations for some species in order to test vague or questionable species boundaries, particularly the taxonomically problematic ‘macrospecies’ N. infundibulifolia Blatt. & McCann, N. simplex, N. plicata and N. concolor, which have been subject to unstable and sometimes conflicting interpretation in different parts of their widespread geographic ranges. To help assess the merit of morphology-based assumptions within variable taxa, three names presently treated as synonyms by POWO (2024) were maintained for the purposes of this study: N. campestris (J.J.Sm.) Schltr. [now placed under N. holochila (F.Muell.) Schltr.], N. carinata (Roxb.) Schltr. (now placed under N. concolor) and N. prainiana (now placed under N. simplex).

In total, 96 Nervilia plants were sampled ( Table 1 ). These included 86 samples representing 45 named taxa, four samples that could only be doubtfully referred to a particular named species and were thus qualified with “conferatur” [viz N. cf. mackinnonii (Duthie) Schltr., N. cf. viridis S.W.Gale, Watthana & Suddee and N. cf. concolor], and a further six samples that could not be matched with any published species and were thus suspected to represent undescribed taxa. Our sampling covered all three sections of the genus, including several taxa with a flower number ‘atypical’ of the section in which they are placed ( Table 1 ). Also included was N. stolziana Schltr. which, on account of its spurred lip, was previously assigned (together with N. pectinata P.J.Cribb, not included in this study) to section Kyimbilaea, which has since been subsumed under section Linervia (Pettersson, 1991). In all, we included 11 taxa (13 samples) from Africa and Madagascar, 30 species (60 samples) from the seasonal Asian tropics, seven species (nine samples) from the moist Asian tropics, and seven species (14 samples) from Oceania. Based on the phylograms presented by Chase et al. (2015) and Freudenstein and Chase (2015), we included one sample each of Corymborkis veratrifolia (Reinw.) Blume (Tribe Tropidieae), Monophyllorchis maculata Garay (Tribe Triphoreae) and Gastrodia peichatieniana S.S.Ying (Tribe Gastrodieae) as outgroups from subfamily Epidendroideae for phylogenetic analysis, plus one sample of Habenaria dentata (Sw.) Schltr. (Tribe Orchideae) as an additional outgroup from subfamily Orchidoideae for biogeographic analysis. All samples were collected and transported with permission [CITES permits: ROP-008-2018, ROP-045-2019, PCIP-20-00094, CA-307/2012, 2014-TH006062/CA, 2021-TH010447/BE, JPHTN/PPP/BO-100-24/1(40), 003/16-01, 008/16-01].

Of the 100 samples used in this study, 63 were newly sequenced, either from fresh material (48 samples) or from well-preserved herbarium specimens (15 samples). Total DNA was extracted using a QIAGEN DNeasy^® plant DNA kit (Hilden, Germany) according to the manufacturer’s instructions. The internal transcribed spacer (ITS) region of nuclear ribosomal DNA was amplified using the primers of White et al. (1990) or Sun et al. (1994), the plastid maturase K gene (matK) region (including part of the flanking trnK introns) was amplified in three sections using the primers of Hidayat et al. (2005), and the entire trnL-F region (comprising the trnL intron and the trnL-F intergenic spacer) was amplified using the c and f primers of Taberlet et al. (1991). PCR was performed in a total reaction mixture of 25 µl containing 1 µl of template DNA (2–10 ng), 5 µl of 5 × Phire^® reaction buffer with MgCl₂, 0.5 µl 10 mM of dNTP mix, 0.5 µl of Phire^® hot start II DNA polymerase (Finnzymes, Finland) and 10 pmol of each primer (Beijing Genomics Institute). The thermal cycler programme consisted of an initial denaturation step at 98°C for 30 s, followed by 35 cycles of 5 s at 98°C, 5 s at 60°C for ITS, 10 s at 55°C for matK and 5 s at 55°C for the trnL-F region, 20 s at 72°C, and a final extension at 72°C for 1 min. Amplification products were purified using a DNA purification Kit (Beijing Genomics Institute). Purified PCR products were sequenced using an ABI 3730 DNA Sequencer (Applied Biosystems, Foster City, California). All sequences have been deposited in GenBank ( Table 1 ).

Alignments were constructed using the MAFFT multiple alignment plugin in Geneious v11.1.4 (Kearse et al., 2012), with subsequent adjustment by eye. We excluded two poly-A regions comprising 41 and 61 positions in the trnL–F and matK genes, respectively ( Supplementary File S1 ). An incongruence length difference (ILD) test (Farris et al., 1995) was performed in PAUP* v4.0b10 (Swofford, 2003) to assess whether the individual matK and trnL–F data sets, and the ITS and combined cpDNA data sets ( Supplementary File S1 ), reflect similar potential phylogenies; 1,000 replicates, each with 1,000 random addition sequence replicates and tree bisection-reconnection (TBR) branch swapping, were performed in each test, and a P value of <0.05 was considered significant (Sullivan, 1996; Darlu and Lecointre, 2002). A “hard” incongruence test was also performed by directly comparing respective topologies, as well as resolution, for each clade generated in the separate analyses, with bootstrap percentages (BP) of ≥85% (Chase et al., 2000) and posterior probabilities (PP) of ≥0.95 (Martínez-Azorín et al., 2011) being taken as evidence of strong support.

Both the homogeneity test for the matK and trnL-F data sets (P = 0.881) and visual node-by-node comparisons of trees generated for either region individually revealed no major topological disparities for nodes of BP ≥85% and PP ≥0.95, and so the two ptDNA regions were combined. Tree topologies generated for the individual ITS and ptDNA data sets using Bayesian inference (BI) were also largely congruent with those using maximum parsimony (MP; Supplementary File S2 ). However, the ILD test indicated significant incongruence between the ITS and ptDNA data sets (P = 0.001). Even so, visual comparison of the trees generated from the two data sets uncovered no topological disparities with nodes of BP ≥85% and PP ≥0.95, except for the position of a single clade containing four samples representing three species [N. bicarinata (Blume) Schltr., N. kotschyi and N. shirensis; Supplementary File S2 ]. Since Cunningham (1997) and Yoder et al. (2001) have argued that combined data sets improve phylogenetic accuracy regardless of incongruence, and numerous phylogenetic studies have found that trees generated from combined data sets with or without samples responsible for topological disparities remain highly consistent (e.g. Li et al., 2011; Kumar et al., 2022), we concatenated the ITS and ptDNA data sets and interpreted the resulting combined phylograms.

Phylogenetic analysis of individual and multilocus alignments were carried out using MP in PAUP* v4.0b10 and BI in MrBayes v3.2 (Huelsenbeck and Ronquist, 2003). For MP analyses, heuristic searches were conducted with 1,000 random addition replicates followed by TBR branch swapping. All characters were unordered and equally weighted with gaps (including unavailable sequences) treated as missing data. Topological robustness was assessed using 1,000 bootstrap replicates. For BI analyses, each DNA region was assigned its own model of nucleotide substitution, as determined by the Akaike information criterion (AIC) in Modeltest v3.06 (Posada and Crandall, 1998). Four simultaneous Monte Carlo Markov Chains (MCMC) were run, with sampling one tree every 1,000 generations for 30,000,000 generations, starting with a randomly generated tree. Majority rule (>50%) consensus trees were constructed after removing the first 25% of sampled trees as burn-in.

In constructing a dated phylogenetic tree, a single accession was selectively retained for each taxon represented by more than one sample ( Supplementary File S3 ). Divergence times were estimated using a Bayesian uncorrelated relaxed-clock model implemented in BEAST 2.7.6 (Bouckaert et al., 2019) with priors placed on the node for tribes Nervilieae and Gastrodieae (offset 34.93 Mya, mean:1, sigma:1) and the node for subfamilies Epidendroideae and Orchidoideae (offset 64 Mya, mean:1, sigma:1), based on results presented by Givnish et al. (2015); Li et al. (2019) and Li et al. (2022). MCMC searches were run for 50,000,000 generations and sampled every 5,000 generations, with convergence being monitored using Tracer 2.7.6 (Bouckaert et al., 2019). The effective sample sizes (ESSs) of all parameters were assessed as more than 200 and the maximum clade credibility tree was computed using treeAnnotator 2.7.6 (Bouckaert et al., 2019).

Four areas of endemism were defined for biogeographic analysis, reflecting the extant distribution of Nervilia demarcated by Pridgeon et al. (2005) as well as the climatic zones discernible within this range based primarily on seasonality, which is presumed to be of importance for the hysteranthous habit (Gale et al., 2021): (area 1) tropical Africa & Madagascar, (area 2) seasonal (monsoonal) tropical Asia, (area 3) aseasonal, moist tropical Asia, and (area 4) Oceania (encompassing Australasia, Micronesia, Melanesia and Polynesia). Ancestral area reconstruction was then performed using the package BioGeoBEARS (Matzke, 2016) in R 4.3.2 (R Core Team, 2023), applying the dispersal–extinction–cladogenesis (DEC) model (Ree and Smith, 2008), ML version of Dispersal Vicariance Analysis (DIVALIKE; Ronquist, 1997) and Bayesian biogeographical inference model (BAYAREALIKE; Landis et al., 2013) with the maximum range-size parameter set to three. We tested each of these models with and without founder-event speciation, which was incorporated with J-parameter modelling jump dispersal (Matzke, 2016). All six permutations were compared using likelihood values, and Akaike information criterion (AIC) was performed in BioGeoBEARS using the maximum clade credibility tree from the BEAST analyses described above. The best-fit model was selected based on lower corrected Akaike information criterion (AICc) values with larger weight (wAICc), representing relative support for each model (Burnham and Anderson, 2002). All underlying raw data used in the phylogenetic analyses and ancestral area reconstruction are available in the Dryad Digital Repository, DOI: 10.5061/dryad.tb2rbp0bn.

The genus Nervilia in its entirety was strongly supported as monophyletic, whether assessed using combined (BP 97%, PP 1.00; Figure 2 ) or individual ITS and ptDNA data sets ( Supplementary File S2 ). In the combined tree ( Figure 2 ), all ingroup taxa fell into two strongly supported clades, one comprising section Linervia (BP 98%, PP 1.00) and the other composed of sections Vinerlia and Nervilia together (BP 100%, PP 1.00). Represented here by N. plicata and N. platychila Schltr., section Vinerlia was also supported as monophyletic (BP 97%, PP 1.00), as was section Nervilia (BP 99%, PP 1.00). Section Nervilia itself comprised two well resolved sub-clades, one containing five African species [N. bicarinata, N. renschiana (Rchb.f) Schltr., N. shirensis, N. kotschyi and N. ballii; BP 100%, PP 1.00] and the other containing seven named species [N. campestris, N. holochila, N. concolor, N. gammieana (Hook.f.) Pfitzer, N. carinata, N. maculata (C.S.P.Parish & Rchb.f.) Schtlr. and N. fordii (Hance) Schtlr.] plus N. cf. concolor and N. sp. nov. 6 from Asia and Oceania (BP 100%, PP 1.00). Within the latter, N. campestris and N. holochila together formed a single, strongly supported lineage (BP 99%, PP 1.00) that was sister to all remaining members of this sub-clade (BP 75%, PP 1.00), which in turn partitioned the eight positively identified N. concolor accessions (BP 77%, PP 1.00) as separate from an unresolved grade of nine samples representing N. gammieana, N. carinata, N. maculata, N. fordii, N. cf. concolor and N. sp. nov. 6.

Section Linervia was also composed of two strongly supported sub-clades, one comprising N. stolziana, N. cumberlegei, N. petraea (Afzel. ex Sw.) Summerh. and the various samples belonging to the N. simplex complex, including N. prainiana and N. trichophylla Fukuy. (BP 95%, PP 1.00), and the other containing 24 other named species plus two unverified accessions (N. cf. viridis and N. cf. mackinnonii) and five putatively undescribed species (BP 100%, PP 1.00), all of which are referable to the ‘N. adolphi–punctata alliance’ on the basis of leaf and floral characters. Within this latter sub-clade, the two N. lilacea Jum. & H.Perrier accessions from Africa (BP 100%, PP 1.00) were strongly supported as sister to all other accessions from Asia and Oceania (BP 99%, PP 1.00). A derived internal sub-clade of the latter containing N. tahanshanensis T.P.Lin & W.M.Lin, N. nipponica Makino and N. taiwaniana S.S.Ying also received strong support (BP 100%, PP 1.00), as did a sister relationship between N. kasiensis S.W.Gale & Phaxays. and N. muratana S.W.Gale & S.K.Wu (BP 100%, PP 1.00). Similarly, the undescribed N. sp. nov. 2 from the Solomon Islands was strongly supported as sister to a clade containing the two Thai and Chinese N. viridis samples plus the two Thai N. cf. viridis accessions (BP 100%, PP 1.00), with the monophyly of the latter four also being strongly supported (BP 99%, PP 1.00). Further, the pair of samples included for each of N. nipponica, N. alishanensis T.C.Hsu, S.W.Chung & C.M.Kuo, N. futago S.W.Gale & T.Yukawa, N. khaoyaica Suddee, Watthana & S.W.Gale, N. umphangensis Suddee, Rueangr. & S.W.Gale, N. mackinnonii, N. marmorata S.W.Gale, Suddee & Duangjai and N. palawensis Schtlr. all formed strongly supported clades; but this was not the case for the pair of N. trangensis S.W.Gale, Suddee & Duangjai and N. taiwaniana samples. Meanwhile, the five N. infundibulifolia samples formed a clade inclusive of N. hemratii S.W.Gale, Tetsana & Suddee (BP 80%, PP 1.00), among which posterior probabilities hinted at a degree of internal structure.

Estimated divergence times for the genus encompassing 52 taxa are presented in Figure 3 . The results support a close relationship between the tribes Nervilieae (represented by Nervilia) and Gastrodieae (represented by Gastrodia R.Br.), with divergence between the two genera placed at around 31.01 Mya in the Early Oligocene, with a 95% highest posterior density (HPD) interval ranging from 18.5 to 38.45 Mya. The common ancestral age of Nervilia was inferred to be 24.4 Mya (95% HPD 14.77–34.93 Mya). Ancestral lineage Clade A diverged at 12.53 Mya (95% HPD 6.24–25.14 Mya), giving rise to two derived lineages, one representing section Vinerlia (Clade B) and the other representing section Nervilia (Clade C). Clade C subsequently split at 8.94 Mya (95% HPD 3.98–20.88 Mya), generating clades D and E, each of which underwent further differentiation at 4.01 Mya (95% HPD 1.52–11.08 Mya) and 3.66 Mya (95% HPD 1.16–11.31 Mya), respectively. Ancestral lineage Clade F, representing section Linervia, split at 19.11 Mya (95% HPD 10.31–30.76 Mya) into two derived clades, G and H, each of which underwent further divergence at 11.85 Mya (95% HPD 3.81–24.07 Mya) and 9.68 MYA (95% HPD 5.47–21.63 Mya), respectively. Diversification of all derived lineages within the genus occurred from the Late Miocene onwards but this process was concentrated in the Pliocene and appears to have persisted into the Pleistocene ( Figure 3 ).

BioGeoBEARS analysis identified the DEC (dispersal–extinction–cladogenesis) +J model as the best-fit model, recovering the lowest AICc value (138.5) and highest wAICc value (0.69) of the six models considered ( Table 3 ). This model suggests that the ancestral Nervilia population had a distribution encompassing either Africa & Madagascar (area 1) or seasonal tropical Asia (area 2), and that it diverged into two distinct lineages, Clades A and F, in the Late Oligocene ( Figure 3 ). Species derived from the earliest settlers of Clade A eventually gave rise to the common ancestor of Clades B and C in the Middle Miocene. Clade C appears to have originated in Africa & Madagascar, with subsequent divergence in the Late Miocene leading to diversification in that region during the Pliocene (Clade D), as well as contemporaneous dispersal to seasonal and moist tropical Asia (areas 2 and 3) initially, followed by Oceania (area 4) in the Pliocene (Clade E). Most diversification in the latter appears to have occurred in seasonal tropical Asia, but the rapid colonisation of seasonal and moist tropical Asia as well as Oceania by N. concolor over this timeframe is noteworthy. In contrast, diversification of Clade B occurred entirely outside of Africa & Madagascar, with its spread firstly across seasonal tropical Asia being followed by more recent colonisation of moist tropical Asia and Oceania during the Pleistocene.

The progenitor of Clade F was unequivocally of African & Madagascan origin, giving rise to two Clades, G and H, during the Early Miocene. Clade G underwent further differentiation primarily in the same region in the Middle to Late Miocene, but with the N. simplex lineage exhibiting enormous dispersal to seasonal tropical Asia and Oceania in the Late Miocene, mirroring the rapid spread of N. concolor. Clade H is also evidently of African & Madagascan origin and has the Afro-Malagasy N. lilacea at its base, but is otherwise characterised by dispersal to, and subsequent explosive diversification within, seasonal tropical Asia from the Late Miocene onwards, with several more recent, independent onward dispersals to both moist tropical Asia and Oceania. Clade J presents an interesting case of wide dispersal to and differentiation across the Pacific Islands, with subsequent colonisation of moist tropical Asia, from the late Pliocene onwards. Nervilia adolphi, in contrast, appears to represent an unusual dispersal back to Africa in the Late Miocene.

Overall, several long-distance dispersal events are revealed in the evolution of the genus. These include at least three independent migrations from Africa & Madagascar to seasonal tropical Asia in Clades C, G and H, and multiple colonisations of moist tropical Asia from seasonal tropical Asia in Clade I and possibly Clade E. Dispersal to Australasia and the Pacific appears to represent the most recent biogeographic step, having occurred both from seasonal tropical Asia in Clade I (and possibly Clade B) and from moist tropical Asia in Clade E. Moist tropical Asia was also colonised from Oceania in Clade J and, as already noted, Africa was ‘re-colonised’ from seasonal tropical Asia in Clade I.

Nervilia has been represented by up to just three samples in prior phylogenetic analyses of tribal- or genus-level relationships in the Orchidaceae (Rothacker, 2007; Li et al., 2019; Pérez-Escobar et al., 2021). Our findings corroborate a nested placement among the basal Epidendroids but, by virtue of much broader taxon sampling, we elaborate farther reaching hypotheses relating to the temporal and spatial scale of its evolution. Firstly, our analyses imply an origin in the Early Oligocene in either Africa & Madagascar or the seasonal Asian tropics, contrasting the one previous assertion by Pettersson (1991) that the genus arose in Asia, and most likely the wet Asian tropics. In fact, our results suggest that the genus was not present in that region until much more recently. Whilst further outgroup optimisation could yet alter this perspective, its affiliation with Gastrodia in the Gastrodieae on the one hand, and with Epipogium Borkh. and Stereosandra Blume in tribe Nervilieae itself on the other (Chase et al., 2015), might be expected to recover the same equivocal position, since all three genera are similarly widespread across the Old World tropics (Pridgeon et al., 2005). However, because Africa and Asia have never been geographically connected, it is necessary to discern which is the more likely ancestral area. In this regard, the prevalence of African branches at the more basal stem nodes (i.e. Clades C, F, G and H in Figure 3 ) lends weight to an African origin, all the more so for the generally eastward trajectory witnessed in the evolution of the genus as a whole: by and large, from Africa to seasonal tropical Asia, and from there onto the wet Asian tropics and Oceania. Movement from Africa to seasonal tropical Asia is also apparent in the transition from Clade H to Clade I in the Late Miocene (ca. 10 Mya onwards; Figure 3 ), a pattern probably repeated from Clade C to Clade E more or less contemporaneously and, given the absence of section Vinerlia in Africa (Pettersson, 1991), from Clade A to B over the same period, too.

The inferred timing of the origin of the genus is especially illuminating, since the Early Oligocene (from around 33.5 Mya onwards) was marked by the onset of an icehouse climate (Coxall and Pearson, 2007; Liu et al., 2009). This global transition is associated with significant sea-level drop, major aridification and a shift to more pronounced seasonality in rainfall as compared with the preceding warmer and more humid later Eocene (Berggren and Prothero, 1992; Miller et al., 2005; Guo et al., 2008). Although a more northerly inter-tropical convergence zone is thought to have delivered generally higher precipitation and thus wetter conditions to a band stretching across northern Africa and the Tethys oceans (Couvreur et al., 2021), the resulting expansion of savanna-like grassland in Africa and southern Eurasia, as well as subtropical woody savanna in central and southern China, is postulated to have led to fragmentation of closed tropical forest across these land masses (Morley, 2007; Guo et al., 2008). The advent of hysteranthy in the Nervilia lineage could thus be intrinsically linked to this period of increased seasonality in rainfall and lower mean temperatures, with the development of more open habitats at low to middle latitudes potentially offering distinct advantages for a terrestrial, seasonally dormant habit. Though the Early Oligocene is generally viewed as a time of widespread extinction of terrestrial biodiversity (Berggren and Prothero, 1992), compelling evidence for the first appearance of, and diversification within, numerous plant lineages at this time is accumulating (e.g. Zhou et al., 2012; Couvreur et al., 2021; Xue et al., 2024).

Progressive northward drift of the African plate through the Oligocene resulted in reconnection with Eurasia in the Middle Miocene (ca. 19–15 Mya) via formation of the Gomphotherium land bridge and eventual closure of the east Tethys Seaway ca. 14 Mya (Hamon et al., 2013; Couvreur et al., 2021). The Arabian plate, which had been contiguous with Africa throughout the Cenozoic and remained so at this juncture, is thought to have supported woodland and savanna ecosystems comprising warm and wet-adapted elements prior to undergoing aridification once in its modern position from the Late Miocene onwards (Steinthorsdottir et al., 2021). In light of this tectonic-cum-palaeoclimatic sequence and the phylogenetic chronology presented here, it seems reasonable to deduce that the very limited occurrence of Nervilia in the Arabian Peninsula today, with only the widespread Afro-Malagasy N. bicarinata being found in isolated parts of Yemen and Oman (Pettersson, 1991), is relictual and plausibly the result of climate-induced vicariance, as has been inferred in the biogeography of numerous sub-Saharan African lineages (Couvreur et al., 2021). The presence of Nervilia in seasonal tropical Asia within the last 10 Mya – apparently in the form of all three sections of the genus – might therefore be congruent with incremental spread through open woodland across the Gomphotherium land bridge to the Indian subcontinent, and from there to continental Southeast Asia, a pattern of incremental inter-continental migration that has been invoked in the dispersal of many ‘out-of-Africa’ palaeotropical taxa, including members of the disparate families Annonaceae, Asparagaceae and Hyacinthaceae (Zhou et al., 2012; Ali et al., 2013; Howard et al., 2022).

The prevailing occurrence of the African Nervilia species in deciduous and semi-deciduous forest, woodland savanna and grassland today (Pettersson, 1991) further hints towards an ancestral association with seasonally arid landscapes. In contrast, the species of topical Asia occur in a wider range of habitats, encompassing grassland and sparse forest types (Roxburgh, 1832; Su, 2000; Gale et al., 2014) but favouring closed-canopy communities, including semi-evergreen, mixed deciduous (or monsoon) and montane forest (e.g. Gale et al., 2015, 2018; Gale and Phaxaysombath, 2017), as well as true lowland rainforest (e.g. Smith, 1909, 1918). Expansion through continental and insular tropical Asia therefore appears to have gone hand-in-hand with extensive niche differentiation, including colonisation of the dark, moist, evergreen forest understorey. The recent discovery of partial mycoheterotrophy in N. nipponica, an Asian, forest-dwelling member of section Linervia in which reliance on fungal partners is most pronounced at lower light intensities (Nomura et al., 2013; Gale et al., 2021), invites closer scrutiny of the eco-physiological factors that could have facilitated this radiation, whether underpinned by vicariance or geodispersal. By combining carbon gain measurements with phylogenetic analysis of a cross-section of the genus representative of different habitat types in both Africa and Asia, it would be possible to address climate-linked landscape-scale patterns of divergence in light of the evolution of variable mixotrophy. But perhaps even more tellingly in this respect is the apparent loss of hysteranthy in a few derived Asiatic species – N. borneensis J.J.Sm., N. muratana and N. kasiensis – which produce successive, temporally overlapping flowering and leafing shoots along a persistent stolon (Smith, 1909; Gale and Wu, 2007; Gale and Phaxaysombath, 2017), implicating exceptional adaptive convergence that warrants finer phylogenetic reconstruction using next generation sequencing.

Whilst a comparatively ‘short hop’ overland from Africa to Arabia and onto seasonal tropical Asia via India therefore seems plausible and parsimonious in the palaeoclimatic contexts of the Middle to Late Miocene, our ancestral area analysis points to further, more complex patterns of migration thereafter. All three sections of the genus bear the same signature of recent arrival in moist tropical Asia and Oceania, as evidenced by the appearance within the past ca. 2.8 million years of section Vinerlia (represented here by N. plicata and N. platychila) in Malesia, Micronesia, New Caledonia and Fiji, section Nervilia (represented here by N. concolor and N. campestris/N. holochila) in Malesia, New Guinea, tropical Australia and the Southwest Pacific, and various species of section Linervia at various locations throughout this vast region. This timing broadly coincides with the Pliocene-Pleistocene boundary, a period of further global cooling, decreasing availability of growing season moisture and forest fragmentation (Couvreur et al., 2021; Steinthorsdottir et al., 2021). The geographic (and taxonomic) expansion of Nervilia across these land masses can probably be attributed at least in part to emergence of the Sunda shelf, since a terrestrial Sundaland was a consistent feature of the Cenozoic at least until the early Pliocene (5 Mya; Hall, 2009) with subsequent exposure occurring episodically through the Pleistocene (Voris, 2000; Sarr et al., 2019). However, permanent separation of Sundaland from both Wallacea and Oceania (Hall, 2009) implicates longer range onward dispersal in at least those lineages that gave rise to N. platychila, N. holochila, N. palawensis and, independently, two undescribed species (sp. nov. 2 and 3) both found on Malaita in the Solomon Islands, as well as in the lineage that gave rise to N. punctata, apparently through migration from Oceania ‘back’ to Malesia. Moreover, the surprising placement of the African N. adolphi within the overwhelmingly Asian ‘N. adolphi–punctata alliance’ of section Linervia is indicative of a somewhat deeper, long-range dispersal back to Africa, meriting further investigation of the origin and spread of the few other Afro-Malagasy members of this complex not included in this study (in particular, N. fuerstenbergiana Schltr. and N. subintegra Summerh.). Evidence of similar long-range dispersal from tropical Asia to Africa during the late Miocene has been uncovered in other plant groups with a marked Africa-Asia-Australasia disjunction (e.g. Li et al., 2009), though not yet, to our knowledge, in the Orchidaceae. We contend that the minute, mobile Orchidaceous dust seed could have been instrumental in facilitating both the stepwise spread and longer distance migrations uncovered here (McCormick and Jacquemyn, 2014; Givnish et al., 2016).

One the most striking features in the evolution of the genus, however, is the enormous taxonomic diversification that appears to have occurred from around 8 Mya, notably in section Linervia and particularly in seasonal tropical Asia. This timing and regionalisation coincide with the most active phase in the uplift of the Himalaya-Tibetan Plateau, which precipitated significant intensification of the Indian and East Asian monsoons (Zhisheng et al., 2001). Replacing Oligocene subtropical aridity (Guo et al., 2008), the evolution of this atmospheric system is tightly correlated with phased Himalayan orogeny through the Miocene, transforming the geography and biology of the continent through alternating circulations of moist, oceanic air during the summer and dry, inland air during the winter (Li et al., 2015; Nguyen et al., 2024). This process is believed to have reached its zenith by around 3.6–2.6 Mya, although the East Asia winter monsoon continued to strengthen thereafter (Zhisheng et al., 2001). Given that our results reveal both ongoing speciation and independent but broadly synchronous dispersal events between subtropical and tropical Asia and Oceania in all three sections of the genus from the upper Pliocene well into the Pleistocene, it is probable that monsoonal oscillations over tropical East Asia and Oceania played an important role in this dynamism, providing further evidence of the role of seasonality in the evolution of the genus as a whole. A similar explanation was proposed by Ji et al. (2024) in interpreting patterns of diversification within certain lineages of the orchid tribe Collabieae, and especially in the genus Calanthe R.Br.

The apparent link between rapid, recent diversification and the ubiquity of cryptic taxa across the genus warrants deeper examination. All three sections contain species complexes (Gale et al., 2016, 2018; Ketjarun et al., 2019) but the present study reveals the enormous geographic scale of their spread, mostly from the Pliocene onwards. Cursory appraisal of the multi-flowered N. concolor and allies in tropical Asia and Oceania has led to considerable taxonomic discord, with the name N. aragoana having been widely applied across Asia and the Pacific (e.g., Lewis and Cribb, 1989; Pearce and Cribb, 2002; Gale and Watthana, 2014) before being subsumed under the synonymy of the former without detailed analysis (POWO, 2024). Whilst our results support the recognition of a single though variable species ranging from southern Japan to Borneo and from the Western Ghats to the Society Islands, the inclusion also of N. carinata in its synonymy is unfounded, demanding critical review of its circumscription with respect to certain names not sampled in our study [e.g. N. scottii (Rchb.f.) Schltr. and N. tibetensis Rolfe; POWO, 2024]. Indeed, it is clear that this alliance harbours unrecognised diversity, given the presence of both morphologically distinct (N. sp. nov. 6) and anomalous (N. cf. concolor) entities here. Moreover, the placement of the strikingly different – and one-flowered – N. maculata ( Figures 1N, N´ ) in this clade underscores the need for caution before lumping grossly similar ‘floral types’ together without fully evaluating finer characters and phylogenetic distance in the light of ecological differentiation.

The same may apply to the macrospecies N. plicata, which is here found to comprise a grade of morphologically diverse, continental Southeast Asian and insular tropical Asian and Pacific elements, the latter including N. platychila, with a degree of structure suggested among some samples within each of these two vast regions hinting at a possible link between biogeographic history and taxonomic divergence (Ketjarun et al., 2019). But on an even more remarkable scale, species diversity within the N. adolphi–punctata alliance appears to have been generated across the entire generic range predominantly within the last ca. 8 million years, and much of it far more recently than that. Despite vegetative uniformity, this complex patently still conceals cryptic taxa, including the five undescribed species sampled here. In contrast, N. simplex exhibits little genetic discontinuity across its enormous range, supporting the incorporation of both the continental Southeast Asian N. prainiana and Micronesian N. trichophylla despite the morphological disparities that have been used to define them (e.g. Seidenfaden, 1978). Next generation sequencing, as well as analyses of polyploidy, reticulate evolution and possible hybridisation (Chennaveeraiah and Jorapur, 1966; Gale et al., 2015) and introgression, are recommended to further disentangle the evolutionary history of these taxonomically intractable lineages. In addition, since knowledge of pollination biology in Nervilia remains fragmentary (Pettersson, 1989; Gale, 2007), clarification of taxonomic, ecological and geographic biases in rewarding, deceptive and autogamous systems, for example, could further shed light on how floral divergence and pollinator shifts have shaped speciation and biogeographic spread (Ackerman et al., 2023).

That said, sufficient clarity is already achieved to draw several taxonomic conclusions. Firstly, though flower number is confirmed as an unreliable basis for defining a sectional classification of the genus, the three presently recognised sections are nevertheless clearly natural. Section Vinerlia occurs only in Asia and Oceania, not Africa and Madagascar, and is typified by the widespread N. plicata. Higher resolution, integrated phylogenetic and morphological research is needed to ascertain whether N. platychila can be maintained as distinct from that species, possibly reflecting a biogeographic split between insular tropical Asia plus Oceania on the one hand, versus inland, continental Asia on the other. Though both are two-flowered, the pubescent, reniform leaf and longitudinally folded labellum are diagnostic. Examination of other species that probably belong here, including N. ignobilis Tuyama and N. umenoi Fukuyama, is needed to better define the section. Pridgeon et al. (2005) list N. maculata as the type of section Vinerlia, but that species is here unequivocally placed in section Nervilia.

Section Nervilia occurs throughout the range of the genus and is highly variable not only in flower number, but also in terms of leaf shape and indumentum, as well as floral morphology. Pettersson (1991) used two labellum characters to define the section in Africa – the presence of nectar guides and a recurved mid-lobe – but these do not apply outside that continent. He also referred to a possible distinction in pollination ecology, with Eumenid wasps known to pollinate two African species (N. bicarinata and N. shirensis), but no pollination studies have yet been conducted on Asian or Australasian members of the section to either confirm or refute this as a reliable sectional trait. Therefore, though the African species appear to be monophyletic and sister to all remaining members, section Nervilia lacks a clear synapomorphy at present. As concluded elsewhere (POWO, 2024), we confirm that N. campestris is most likely conspecific with N. holochila, presenting an intriguing case of vicariance across Wallace’s Line, albeit highly localised to Java on the western side. Despite wide morphological variation throughout its enormous range, N. concolor is monophyletic and there is little evidence of internal genetic structure. However, N. carinata is not conspecific.

Section Linervia comprises two natural sub-groups: the fimbriate-lipped species typified by the extremely widespread N. simplex plus the spurred African species (represented here by N. stolziana) previously placed in section Kyimbilaea, and those species with an entire labellum mid-lobe that constitute the N. adolphi–punctata alliance. Within the latter, all four N. viridis samples included here were found to be monophyletic and almost certainly conspecific, even though Gale et al. (2018) refrained from combining the two “N. cf. viridis” samples from eastern Thailand on the grounds that Bayesian coalescence analysis resolved them as distinct. Intriguingly, this continental Asiatic species falls sister to an unnamed species from the Solomon Islands. The Himalayan N. macroglossa and southern Japanese N. futago present another interesting disjunction, potentially alluding to historic extinction of other closely related, geographically contiguous taxa, as suggested by the selected BioGeoBEARS model. The widespread continental Asian N. infundibulfolia exhibits considerable internal genetic structure worthy of further examination but is monophyletic only if N. hemratii is considered synonymous. Nervilia punctata is here placed in a clade with the Micronesian N. palawensis and an unnamed species from the Solomon Islands. Our analyses corroborate Gale et al. (2018) in determining N. punctata to be Malesian, with prior records of this entity from continental Southeast Asia (e.g. Seidenfaden, 1978; Gale and Watthana, 2014) probably amounting to misidentifications of N. mackinnonii or other members of this problematic complex. Though the section is overwhelmingly one-flowered, the two- or rarely three-flowered N. cumberlegei also belongs here, and thus the only synapomorphy for the section appears to be the elongating fruiting scape (Pettersson, 1991; Gale et al., 2006).