Orchid phylogenetics and evolution: history, current status and prospects

Abstract

Background

Orchidaceae are one of the two largest families of angiosperms; they exhibit a host of changes – morphological, ecological and molecular – that make them excellent candidates for evolutionary study. Such studies are most effectively performed in a phylogenetic context, which provides direction to character change. Understanding of orchid relationships began in the pre-evolutionary classification systems of the 1800s, which were based solely on morphology, and now is largely based on genomic analysis. The resulting patterns have been used to update family classification and to test many evolutionary hypotheses in the family.

Scope

Recent analyses with dense sampling and large numbers of nuclear loci have yielded well-supported trees that have confirmed many longstanding hypotheses and overturned others. They are being used to understand evolutionary change and diversification in the family. These include dating the origination of the family, analysis of change in ecological habit (from terrestrial to epiphytic and back again in some cases), revealing significant plastid genome change in leafless holomycotrophs, studying biogeographic patterns in various parts of the world, and interpreting patterns of fungal associations with orchids.

Conclusions

Understanding of orchid relationships has progressed significantly in recent decades, especially since DNA sequence data have been available. These data have contributed to an increasingly refined classification of orchids and the pattern has facilitated many studies on character evolution and diversification in the family. Whole-genome studies of the family are just beginning and promise to reveal fine-level details underlying structure and function in these plants, and, when set in a phylogenetic context, provide a much richer understanding of how the family has been so successful in diversification.

Orchidaceae, with over 31 000 species (Govaerts, 2024), are one of two families (Asteraceae being the other) contending to be the largest in angiosperms. They are popular not only as research subjects for understanding diversification in a cosmopolitan and morphologically specialized group, but more broadly as a commercially important group of plants for aesthetics and flavouring (vanilla). The family is a member of Asparagales and appears to be sister to the remainder of that order (Graham et al., 2006; Zuntini et al., 2024), which includes asparagus, irises, agaves, onions and hyacinths, among many others. Phylogenetic analyses indicate that the terrestrial habit is plesiomorphic for orchids (Chomicki et al., 2014), although most species (~70%) are epiphytic (Zotz, 2016) and tropical. Orchids are also represented in temperate regions, where they are primarily terrestrial. With respect to our understanding of evolutionary pattern and classification, the family provides a microcosm reflecting developments in plant systematics over the past 200 years. This includes progress from classifications early in the 19th century that were ‘natural’ in the sense of employing multiple characters to reveal groups, to the more phylogenetically informed yet still fairly intuitive classifications of the 20th century. From explicit phylogenetic analyses of morphological characters and single molecular loci to today’s broad genomic sampling, advancing our understanding of orchid relationships and evolutionary patterns is still very much in process.

Although orchid flowers appear to be very diverse in morphology, much of this is based on differences in size, shape, colour, and fusion in the perianth; the basic floral structure of the monandrous (single-anthered) orchids, comprising some 99% of species, is quite conserved. All these species have, in addition to the single anther, pollen occurring in masses (pollinia), androecium and gynoecium fused into a gynostemium (column), and zygomorphic symmetry with the median petal (labellum) usually distinct in appearance from the other petals (Fig. 1). It is in the details and modifications of this basic plan that orchid floral diversity emerges.

Fig. 1.

Examples of orchids from the five subfamilies. (A) Apostasia wallichii (Apostasioideae); photograph by Kim Kristiansen. (B) Vanilla dilloniana (Vanilloideae); photograph by Erich DeLin. (C) Cypripedium parviflorum (Cypripedioideae). (D) Galearis spectabilis (Orchidoideae). (E) Spiranthes ochroleuca (Orchidoideae). (F) Cattleya maxima (Epidendroideae). (G) Restrepia cf. sanguinea (Epidendroideae). (H) Zygopetalum maculatum (Epidendroideae), showing the labellum, column and anther. Scale bars = 1 cm. Photographs by J. Freudenstein unless specified otherwise.

Orchid life history is highly specialized and directly connected to floral structure and function. Seeds of orchids are structurally minimal, normally comprising no embryo or endosperm, only a mass of undifferentiated cells protected by a single cell-layered testa (Fig. 2G; Rasmussen, 1995). Large numbers of seeds are produced, from thousands to perhaps several million in larger capsules (Vij et al., 1992). Having orchid pollen condensed into pollinia facilitates this mass development of seeds, such that in most orchid flowers either large numbers of seeds are produced or none. It is only in some groups that have a preponderance of plesiomorphic character states (such as apostasioids) where more granular pollen is found, or in groups that have their pollinia organized into subpackages (Fig. 2D) known as sectile pollinia), that fertilization of only a subset of the ovules in an ovary will likely result. Given the lack of resources in the seed, assistance in germination is required and this is provided by a fungus that, as far as is known, represents a parasitism or predation of the plant on the fungus, since no clear benefit to the fungus has been demonstrated (Fig. 2F; Rasmussen and Rasmussen, 2009). After a period of development, most orchids transition to autotrophy, producing leaves, stems and roots; however, many lineages have transitioned to full reliance on the fungus for fixed carbon (known as holomycotrophy (Rasmussen, 1995) or full mycoheterotrophy;Fig. 3). An unknown number of species, perhaps mainly terrestrials, continue to derive some carbon from fungi although they are photosynthetic; these are termed mixotrophs. Hence, the orchid life history is highly integrated, requiring large numbers of pollen grains that are efficiently delivered in masses to produce the large numbers of low-investment seeds needed to maximize the chance that some will find the appropriate fungus and germinate.

Fig. 2.

Orchid pollination, pollinia and seeds. (A) Calopogon tuberosus. The labellum is uppermost (non-resupinate), bearing false stamens. Scale bar = 1 cm. (B) Phragmipedium lindleyanum. Scale bar = 1 cm. Insects enter the slipper-shaped labellum and emerge at the anthers (one shown by the arrow). (C) Pollinarium of Stanhopea wardii, showing yellow pollinia attached to a cellular stalk (stipe). Scale bar = 2 mm. (D) Pollinium of Pogonia ophioglossoides, which is weakly sectile. Scale bar = 1 mm. (E) Ophrys insectifera, one of the pseudocopulatory species. Scale bar = 1 cm. (F) Fungal coils (pelotons) in cortical cells of Neuwiedia veratrifolia roots; photograph by Kim Kristiansen. Scale bar = 50 µm. (G) Seed of Corallorhiza odontorhiza. Scale bar = 0.2 mm. Photographs by J. Freudenstein unless specified otherwise.

Fig. 3.

Orchid habits and mycoheterophic diversity. (A) Govenia superba, terrestrial. Scale bar = 40 cm. (B) Pleurothallis cardiophylla, epiphytic. Scale bar = 1 cm. (C) Bletia (Hexalectris) spicata, holomycotrophic. Scale bar = 15 cm. (D) Cephalanthera damasonium, mixotrophic. Scale bar = 5 cm. (E) Cephalanthera austiniae, holomycotrophic. Scale bar = 3 cm. (F) Corallorhiza trifida, mixotrophic. Scale bar = 6 cm. Photographs by J. Freudenstein.

Orchids display a range of plant–pollinator relationships from highly specialized, such as the famous Darwin’s orchid (Angraecum sesquipedale) with its ~33-cm nectar spur that is pollinated by a moth with a similarly long proboscis (Wasserthal, 1997), to others that are more generalized in their requirements. Although a diversity of pollinator rewards are offered depending on species, from nectar to floral scent compounds, a surprising number are deceit pollinated (as many as 50 %; Shrestha et al., 2020; Ackerman et al., 2023), either mimicking flowers of another nearby species that offers a reward or producing structures and/or compounds that trigger an insect to expect a reward (Fig. 2A). The most extreme of the latter strategies is pseudocopulatory pollination, in which the flower mimics key features of a female insect, attracting males that attempt to mate with the flower (Fig. 2E). This strategy is limited to orchids with the exception of rare instances in Asteraceae and Iridaceae (Ellis and Johnson, 2010; Vereecken et al., 2012); even among orchids it is rare as far as currently documented cases attest (~340 species; Gaskett, 2011).

This short summary of orchid features illustrates many of the evolutionarily significant aspects of orchid biology that can be investigated. Such studies nearly always require a phylogenetic context to be meaningful, thus giving an impetus beyond just family classification for efforts to understand their relationships.

HISTORICAL PERSPECTIVE: THE CHALLENGES OF ORCHID RELATIONSHIPS

As with most angiosperm groups, the origin of our current understanding of orchid phylogenetic patterns lies in the systems of classification that were published from the late 18th to the 20th century. These classifications were the starting point for understanding hypotheses of orchid relationships, even in a time of pre-evolutionary thinking. Although Linnaeus (1753) described the 62 species known to him in eight genera, Swartz (1800) is usually considered to be the first real classification of orchids because of his use of supra-generic groups. He included only 25 genera in his system, but distinguished between groups that we would today recognize as Cypripedioideae, Orchidoideae and Epidendroideae, using primarily anther characters. This was followed by Lindley’s systems (1826, 1840), and those of Bentham (1883), Pfitzer (1887), Schlechter (1926), Dressler and Dodson (1960) and Garay (1960, 1972), among others. The systems through Schlechter were not explicitly phylogenetic in the sense of being depicted as a tree but were often described concisely in hierarchic fashion by contrasting character states as in an identification key. Rolfe (1890) may have been the first to depict relationships among major groups of the orchids as a tree (and as a Venn diagram). For more details on the early history of orchid classification, see Schweinfurth (1959) and Rasmussen (1999).

The majority of important morphological characters used in orchid classification are derived from the anther, although classifications increasingly benefitted from the rich tradition of broader morphological and anatomical study under way in north-western Europe in the 19th century (Solereder and Meyer, 1930; Stern, 2014); Pfitzer (1887) in particular emphasized vegetative features in his system. Unlike many other lilioid monocots, the orchid anther is a rich source of characters – the number and position of fertile anthers (one to three; Fig. 4), the position of the anther (erect, incumbent), numbers of pollinia and their orientation and substructure, as well as the nature of any associated pollinium stalks (Fig. 2C). Most of this variation is related to the animal (usually insect) pollination systems that are important to orchids and to the fundamentally zygomorphic nature of orchid flowers in all but the earliest-diverging and relatively plesiomorphic group (Apostasioideae, and even there zygomorphy has its beginnings; Johnson and Edwards, 2000).

Fig. 4.

Relationships among subfamilies of orchids. Symbols to the right of each subfamily name depict anther states for the subfamily; black dots indicate functional anthers and white dots are staminodes.

Resolution of higher-level orchid groups and their relationships based only on morphological characters was challenging, as was the case in many plant groups. At least since the time of Lindley (1826), a core group of species that possessed incumbent (inflexed or bent forward) anthers, coherent pollinia, and frequently advanced types of cellular pollinium stalks was evident and accorded status in classifications (the ‘epidendroids’). Also distinct were the ‘orchidoids’, today recognized as Orchideae, because of their distinctive sectile pollinia, caudicular (secreted) pollinium stalks and the basal positioning of their viscidia (adhesive structures); these were also recognized in Lindley’s (1826) classification. Finally, those species with two fertile anthers and a slipper-shaped labellum (Fig. 2B) were segregated as ‘cypripedioids’. During the remainder of the 19th and early 20th centuries, large numbers of new species were described; some of them fit easily into these distinct groups, but many had few apomorphic characters that could link them to these, which resulted in their being placed together in an ill-defined group known as the ‘neottioid’ orchids. These include groups now resolved as the Vanilloideae, early-diverging lineages in Epidendroideae, and members of the currently recognized Diurideae, Goodyerinae, Spiranthinae and Cranichidinae. Most of these species have soft pollinia, a more or less erect anther and poorly developed pollinium stalks, if any. Such a higher-level group persisted in most systems of the later 19th and 20th centuries (e.g. Pfitzer, 1887; Schlechter, 1926; Dressler and Dodson, 1960) until explicit phylogenetic analyses that treated the entire family were performed.

In addition to the neottioids, holomycotrophs (see below) and some genera with unusual morphology have been particularly difficult to place. Dressler (1980), for example, called Cryptarrhena an ‘orphan’ and Dressler (1990a, 1993) listed Arundinae, Collabiinae, Pogoniinae, Claderia, Eriopsis, Thaia and Xerorchis as problematic taxa because he found it difficult to relate these clearly to his more well-defined epidendroid groups.

TREE-BASED ANALYSES

It was in the 1980s that the first explicit phylogenetic analyses of orchids were performed, where well-defined character matrices, tree rooting and monophyly of groups were emphasized. Some analyses of morphological matrices with explicit optimality criteria were performed (e.g. Linder, 1981; Herschelia), while others employed a phylogenetic argumentation approach without quantitative optimality analyses. The latter include analyses by Rasmussen (1982, 1985), Linder (1986) and Dressler (1986, 1989, 1990a, b, c, 1993), the last of whom incorporated results from this work in continuing refinements of his system. These were followed by other studies of smaller parts of the family, including Romero (1990; Catasetinae), Linder and Kurzweil (1990; Disinae), Kurzweil et al. (1991; Pterygodium–Corycium), Linder and Kurzweil (1994; Diseae), Albert (1994; Cypripedioideae), Freudenstein (1994; Corallorhizinae); Neyland et al. (1995; Pleurothallidinae), Linder and Kurzweil (1996; Brownleea), Morris et al. (1996; Dendrobiinae), and Repetur et al. (1997; Bromheadia).

Phylogenetic studies at the family level based on morphology were also being conducted. In addition to the traditionally used morphological characters from the anther and vegetative structures, a number of new broadly sampled morphological/anatomical character sets were developed in the 1980s and 1990s that were useful in morphological phylogenetics. These include pollen surface structure and texture (Williams and Broome, 1976), seed surface structure and sculpturing (Ziegler, 1981), the presence and nature of specialized cells associated with vascular tissue (stegmata; Møller and Rasmussen, 1984), the nature and details of pollinium stalks (Rasmussen, 1982, 1986), structural details of the root velamen layers and their cells (Porembski and Barthlott, 1988), stigma morphology (Dannenbaum et al., 1989) and the details of endothecial thickenings in the anther (Freudenstein, 1991). The developmental studies of Kurzweil (1987a, b, 1988, 1993) were important in clarifying homologies in some cases. Dressler (1993) employed many of these characters as he worked towards refining orchid relationships in his later system. Szlachetko (1995) proposed a classification system for the family that used phylogenetic argumentation and included many of these characters.

Rasmussen (1985) presented a phylogenetic diagram that depicted his assessment of major clades of orchids, recognizing separate Apostasiaceae, Cypripediaceae and Orchidaceae. He viewed Apostasiaceae, Neottioideae and Diurideae as paraphyletic, given that no clear synapomorphies define them. Burns-Balogh and Funk (1986) conducted the first comprehensive morphological phylogenetic analysis of the family, which included a considerable amount of character analysis. They resolved Neuwiedioideae separate from Apostasioideae (which was sister to Cypripedioideae), Spiranthoideae, Neottioideae, Orchidoideae and Epidendroideae. Details of the optimality procedure were not presented, however, making it difficult to reproduce the tree from their data. Freudenstein and Rasmussen (1999) provided an explicit phylogenetic analysis of morphological and anatomical data that revealed some insights but also the limitations of morphological data for resolving structure within the family.

Molecular phylogenetic analysis of orchids got under way in the early 1990s. As with morphological analyses, many of the first molecular analyses focused on smaller orchid groups. These include Yukawa et al. (1993; Dendrobiinae), Albert (1994; Cypripedioideae), Freudenstein and Doyle (1994a, b; Corallorhiza), Bateman et al. (1997; Orchidinae), Caputa et al. (1997; Orchis), Hapeman and Inoue (1997; Platanthera) and Cozzolino et al. (1998; Orchis). Additional studies of tribal or subtribal groups followed, including of groups within Maxillarieae (now part of Cymbidieae; Whitten et al., 2000; Williams et al., 2001), Laeliinae (van den Berg et al., 2000), Arethuseae (Goldman et al., 2001), Coelogyninae (Gravendeel et al., 2001), Diurideae (Kores et al., 2001) and Pleurothallidinae (Pridgeon et al., 2001). Many more studies on various tribal and subtribal groups were done in the subsequent two decades.

Molecular studies of subfamilies include Neyland and Urbatsch (1996b), van den Berg et al. (2005) and Freudenstein and Chase (2015) for Epidendroideae, Kores et al. (1997) for Orchidoideae, Albert (1994) and Cox et al. (1997) for Cypripedioideae, Cameron (2009) for Vanilloideae and Kocyan et al. (2004) for Apostasioideae. These were limited to plastid and nuclear ribosomal sequences, the loci commonly used at that time. Freudenstein and Chase (2001) introduced the mitochondrial DNA locus nad1 for use across the family and Cameron (2009) used mitochondrial atpA and nad1 in an analysis of Vanilloideae. Górniak et al., (2010) used the single-copy nuclear Xdh in their study of the family at a time when new nuclear loci were being developed for use in many groups.

Chase et al. (1994) conducted the first molecular analysis at the family level; it comprised plastid rbcL sequences for 33 orchid species and a broader context of monocots that served to root the tree. The analysis resolved the same subfamily groupings that we recognize today but differed in their relationships, the tree being poorly supported along its backbone. Neyland and Urbatsch (1995) employed sequences from part of another plastid gene, ndhF, and included 30 orchid species and two outgroups. That study found the same subfamily groupings that we recognize today (where they included more than one species per subfamily) but also showed a somewhat different pattern of relationships among them; their focus was on determining if the terrestrial state was plesiomorphic for the family, which it was in their study. Their subsequent study (Neyland and Urbatsch, 1996a) used the same locus and sampling trimmed to just six orchid species to trace evolution in the number of anthers, concluding that cypripedioid diandry was probably derived from a monandrous ancestor. Cameron et al. (1999) provided a much-expanded sampling (158 orchids plus 13 outgroups) for rbcL sequences, again resolving the subfamilies that we recognize today and emphasizing the distinctness of the vanilloids as a group apart from epidendroids and orchidoids, although the single gene did not provide enough data to confidently place the subfamilies with respect to each other.

As for the problematic neottioids, the morphological phylogenetic analyses of Burns-Balogh and Funk (1986) and Freudenstein and Rasmussen (1999) began to sort them into their rightful places among the well-defined subfamilies, but it was not until DNA sequence data became available that their placement and the relationships of orchid subfamilies to each other really became clear. Most neottioid genera found their place as early-diverging lineages in Orchidoideae and Epidendroideae, while others formed the distinct and early-branching vanilloid lineage. Resolution of the neottioids was a major development in our understanding of orchid relationships, while the placement of Vanilloideae as a distinct early-branching lineage in the family clarified the convergent nature of the incumbent anther between vanilloids and epidendroids. These analyses resulted in the five-subfamily pattern that today comprises Apostasioideae, Vanilloideae, Cypripedioideae, Orchidoideae and Epidendroideae and that has been supported by multiple family-wide studies (Figs 1 and 4; e.g. Górniak et al., 2010; Givnish et al., 2015; Li et al., 2019a; Pérez-Escobar et al., 2021, 2024; Serna-Sanchez et al., 2021; Zhang et al., 2023).

Recent studies have often focused on bringing genomic-level nuclear data to bear on our understanding of family-wide orchid relationships, although genomic sampling has also been used even at the level of a species complex (e.g. Bogarín et al., 2018). Deng et al. (2015) used transcriptome data for ten representatives of the five orchid subfamilies, assembling data for 315 orthologues. The subfamily-level pattern and those within subfamilies were strongly supported and agreed with the now standard understanding of relationships. Zhang et al. (2017) performed a phylogenetic analysis of the family based on 132 orthologues from 12 orchids plus a broader sampling of angiosperms, again confirming the now standard pattern of subfamily relationships. Unruh et al. (2018) assembled transcriptomes for 13 species of orchids, focusing especially on cypripedioids and on understanding the pattern of whole-genome duplications (WGDs) in the part of Asparagales near the orchids. Eserman et al. (2021) developed a target-capture set (the ‘Orchid963’ set) and provided a skeletal analysis of 28 species that again confirmed the standard pattern of subfamilies. Pérez-Escobar et al. (2021) used the Angiosperm353 bait set (Johnson et al., 2019) and included 75 species, while Wong and Peakall (2022) analysed publicly available transcriptome data for 69 orchids and 633 loci, both finding patterns largely in agreement with each other. Zhang et al. (2023) used 1450 nuclear genes for 610 species in their family-wide analysis, yielding the most data-rich analysis to date. Pérez-Escobar et al. (2024) extended their sampling by combining multi-locus nuclear data (448 orchid species with Angiosperm353 genes) with 2060 species for which ITS + matK were available, providing the most broadly sampled molecular analysis to date. The latter two analyses largely agreed in the relationships they depicted, suggesting that we are moving towards stability in our understanding of many orchid relationships.

In parallel to the nuclear, multi-orthologue analyses, many analyses have been performed on whole plastid genomes. These include Luo et al. (2014), Givnish et al. (2015), Kim et al. (2015), Dong et al. (2018), Li et al. (2019), Kim et al. (2020) and Serna-Sanchez et al. (2021). These analyses also have recovered the standard subfamily pattern and have increasingly added sampling such that tribal and subtribal relationships have been revealed. These analyses are data-rich in the number of nucleotide base characters that contribute to the analysis and often have produced trees with strong branch support, but, in the sense that they comprise single non-recombining units, they still represent single-locus analyses (Doyle, 1992, 2022; but see Gonçalves et al., 2019).

PLACEMENT OF ANOMALOUS GENERA AND THE CHALLENGE OF HOLOMYCOTROPHY

Some phylogenetic studies have focused specifically on determining the placement of especially challenging genera, including those highlighted by Dressler (1993). Xiang et al. (2012), for example, resolved the previously somewhat enigmatic genera Thaia and Tangtsinia among the early-diverging epidendroids. Dressler had provisionally placed Thaia in Neottieae but this analysis, based on three plastid loci, suggests that it should be a distinct lineage. Moreover, it had originally been described as leafless but recent collections show the clear presence of leaves. Tangtsinia was shown to be part of Cephalanthera and the hypothesis that it is a peloric derivative within that genus remains a likely explanation for its nearly actinomorphic symmetry.

At many points in the orchid tree, primarily in Epidendroideae, a leafless, holomycotrophic condition has evolved, at least in some cases by retaining and expanding the mycorhizome that can form after germination (Rasmussen, 1995). Such plants are only evident when in flower (Fig. 3). Estimates of the number of times that this shift has occurred among orchids vary and depend especially upon the resolution of some clades of holomycotrophs among the early-diverging lineages of the epidendroids, but it may be as many as 30–31 or more (Freudenstein and Barrett, 2010) and no family has as many holomycotroph species as Orchidaceae (Merckx et al., 2013; Imhof, 2024). Such plants pose a particular challenge in orchid phylogenetics for a number of reasons. First, many of them are rarely seen and difficult to obtain. Second, they are not in cultivation because of their specialized fungal relationships. Third, various morphological characters are absent because of the loss of leaves and the often autogamous nature of the flowers. Lastly, plastid genomes are reduced and the remaining loci experience accelerated changes that frequently make phylogenetic reconstruction difficult due to the presence of long branches.

Perhaps the most problematic region of the orchid tree with respect to holomycotrophs is the base of Epidendroideae. Morphological features and molecular data suggest that genera such as Gastrodia, Stereosandra, Epipogium and Wullschlaegelia belong there but detailed resolution of relationships remains elusive. The grouping of such leafless genera together in a clade as seen by Zhao et al. (2024) could be real or could be due to long branch effects, given that the analysis was based on plastid genome sequences for these species that have experienced accelerated rates of evolution and reduced genome sizes. Access to good quality material for DNA isolation remains a challenge for many of the species in this part of the tree.

Pogoniopsis is a case in point of a holomycotroph that has been difficult to place, but that is now being resolved. Its name suggests similarity to Pogonia and thus a likely placement in Vanilloideae. However, its membership in that subfamily was not clear; Cameron and van den Berg (2017) and Klimpert et al. (2022) found that it is in fact an epidendroid, probably related to Neottieae. Another holomycotroph genus that may have been misplaced is Risleya. It has traditionally been placed in Malaxideae because of its raceme of very small flowers with apparently no pollinium stalks (e.g. Dressler, 1981, 1993), although Xiang et al. (2014) reported that the pollinia are attached to a viscidium. Xiang et al. (2014) placed it in Collabieae based on plastid DNA sequences and subsequently Li et al. (2019, 2020) suggested that it belongs in Calypsoinae based on mitochondrial and plastid sequence analyses.

The patterns revealed by phylogenetic analyses at all levels have been used periodically to update the formal classification of orchids (Table 1; Chase et al., 2003, 2015) and informed the comprehensive Genera Orchidacearum series (Pridgeon et al., 1999–2009). They have been used as the basis for adjusting taxonomy to reflect new understanding of species relationships in many cases (e.g. Martos et al., 2014; Bone et al., 2015, Eulophia; Li et al., 2015, Arethuseae; Chase et al., 2021, Coelogyne). Although refining our classification of orchids is an important task, of perhaps greater interest to the broader scientific community are the uses of orchid trees for many additional purposes where the direction, degree and relative diversification of groups in the tree are important, as in studies of many aspects of orchid diversification and evolution.

Table 1.

Classification of Orchidaceae to the tribal level, based on Chase et al. (2015) with slight modification to accommodate recent phylogenetic results.

Apostasioideae	Epidendroideae (continued)
Vanilloideae	Triphoreae
Pogonieae	Xerorchideae
Vanilleae	Wullschlaegelieae
Cypripedioideae	Gastrodieae
Orchidoideae	Nervilieae
Codonorchideae	Arethuseae
Cranichideae	Malaxideae
Diurideae	Cymbidieae
Orchideae	Epidendreae
Epidendroideae	Collabieae
Neottieae	Podochileae
Sobralieae	Vandeae
Tropidieae

DATING

With the availability of increasingly detailed and well-supported phylogenetic trees has come the desire to estimate dates for nodes on them. Such dating is important because it allows relating phylogenetic patterns for one group to those in other groups in time and to events in Earth history, thereby facilitating more concrete hypotheses about evolution in specific groups to be proposed. However, such dating is not trivial, given the rate variation in genomic data present across the tree in almost every studied group and the paucity of solid time-calibration points (usually fossils) available in many groups. This has been a distinct challenge in orchids due to the scarcity of unambiguous fossils. Given their herbaceous nature and primarily tropical distribution, orchids are not likely to form fossils, even given the large numbers of species in the group. Macrofossils such as Protorchis (Massalongo, 1858) that were originally described as orchids have largely been discounted because of their ambiguous nature (reviewed in Schmid and Schmid, 1977). Although vegetative fossils of orchids are very rare, Conran et al. (2009) described leaf compression fossils from New Zealand that they assigned to Dendrobium and Earina based on distinctive morphological features (see also Iles et al., 2015). The most convincing fossils described are those of pollinaria, especially those preserved in amber.

The first study to affix a date to the origin of Orchidaceae was that of Janssen and Bremer (2004), who estimated a crown age of 111 MY (million years) for the family based on an rbcL tree for monocots dated at the split between Acorus and the rest of the monocots at 134 MY (Bremer, 2000; Table 2); this was not based on any orchid fossils but did include sampling of 145 orchid sequences. The first use of an orchid fossil to date the family was by Ramirez et al. (2007), based on a pollinarium attached to an insect preserved in Dominican amber, dated to early-mid Miocene (15–20 MY; named Meliorchis). Because the pollinarium was sectile, and based on its similarity to other pollinaria in the subtribe, it was assigned to Goodyerinae (Orchidoideae) and provided a minimum age for that subtribe. They used that fossil along with dates for Liliacidites pollen from North America/New Zealand, believed to be the oldest asparagalean fossils (Dettmann, 1973; Walker and Walker, 1984), to calibrate a tree of 55 orchid species spanning the family based on rbcL and matK and found a crown age for orchids of 76–84 MY (numbers for ages rounded). Gustafsson et al. (2010) used a tree constructed from matK and rbcL and also dated with Meliorchis and the asparagalean Liliacidites, as well as the Dendrobium and Earina leaf fossils of Conran et al. (2009) and the oldest monocot fossil (reported in Friis et al., 2004), and found a date for orchids of (63)–77–(92) MY. Subsequent analyses have increased taxon sampling and genome sampling, to the point that the most recent analyses (Zhang et al., 2023; Pérez-Escobar et al., 2024) have included hundreds of species and nuclear loci (Table 2). The analyses employing dating vary in genomic source (plastome versus nuclear or both) and in approach (sequencing all taxa for the same loci or adding taxa for which only a small number of loci have been sequenced). Most of the studies, including the most recent, have found a crown age for Orchidaceae of ~80–90 MY (although often with broad confidence intervals). The Zhang et al. (2023) study is an outlier with its older 101 MY date.

Table 2.

Studies that have estimated crown age for Orchidaceae.

Analysis	Data set size and type	Calibrations	Crown age estimate of Orchidaceae (MY)
Janssen and Bremer (2004)	878 species from across monocots (145 orchids); rbcL sequences	Age of monocots from Bremer (2000)	111
Ramirez et al. (2007)	55 orchids; rbcL + matK sequences	Meliorchis fossil; Liliacidites fossil	76–84
Gustafsson et al. (2010)	61 orchids; rbcL + matK	Meliorchis fossil; Liliacidites fossil; Dendrobium and Earina fossils; Friis et al. (2004) earliest fossil monocot pollen	(63)–77–(92)
Chomicki et al. (2014)	335 orchids; matK + rbcL + trnL-trnF + ITS	Meliorchis fossil; Dendrobium and Earina fossils; Liliacidites pollen	(75)–94–(121)
Givnish et al. (2015)	39 orchid plastomes + 162 orchids for atpB + psaB + rbcL	Meliorchis fossil; Earina fossil; 15 other monocot and dicot calibrations	90
Givnish et al. (2018)	53 orchids, 514 other monocots and dicots; 77 plastome coding loci	13 fossil and 7 secondary points within monocots and dicots, no orchid fossils included	78
Li et al. (2019)	74 orchids; 76 plastome coding loci + 38 mitochondrial coding loci	Meliorchis fossil, Dendrobium fossil, Givnish et al. (2015) orchid date used as prior	90
Kim et al. (2020)	124 orchids; 83 plastome coding loci	Meliorchis, Dendrobium fossil, Liliacidites pollen	(60)–80–(99)
Serna-Sanchez-et al. (2021)	264 orchids; 78 plastome coding loci	Dendrobium fossil, secondary Orchidaceae crown age from Givnish et al. (2015)	(81)–90–(96)
Zhang et al. (2023)	610 orchids; 1450 nuclear genes	Meliorchis, Dendrobium and Earina fossils, Liliacidites pollen, other monocot calibration points	(97)–101 Codonorchidea (102)
Perez-Escobar et al. (2024)	448 orchids for Angiosperm353 loci + 1940 orchids for matK + ITS	Secondary calibrations: Most Recent Common Ancestor (MRCA) of Orchidaceae + Dioscoreales + Liliales, MRCA of Orchidaceae, MRCA of Goodyerinae from Givnish et al. (2015)	83 (± 10)

Dated trees have also been used increasingly at lower levels within Orchidaceae, usually based on the dates for the family determined by the family-level studies. Li et al. (2022), for example, found a crown age for the tropical–subtropical Asian Gastrochilus of 8.13 MY and Gamish and Comes (2019) studied Bulbophyllum, finding an early-to-mid Miocene origin in the Pacific followed by broad dispersal. Gravendeel et al. (2005), however, dated the origin of the Coelogyninae using a matK tree and a different approach that did not rely on fossils. They used geological calibration points corresponding to geological events in south-east Asia that distinguish clades of species in that subtribe.

Additional orchid fossils continue to come to light that show promise for further helping calibrate the increasingly densely sampled trees. Poinar and Rasmussen (2017) described a Baltic amber pollinarium associated with a fungus gnat (Succinanthera) and placed it in Epidendroideae but were not able to assign it further to a specific tribe. Given the quality of preservation in amber and the distinctive morphology of orchid pollinaria, amber from deposits in various places in the world should be examined carefully for pollinaria because these remain the best hope of extending the fossil record for Orchidaceae.

BIOGEOGRAPHY

The biogeography of orchids has been studied increasingly in recent years as more well-supported and detailed phylogenetic trees have appeared that make tracing such patterns possible. One objective of many biogeographic studies is to locate the ancestral area for the group (formerly known as the ‘centre of origin’) and it has been a goal for most family-level biogeographic analyses of orchids. Given the preponderance of tropical species in the family, and the presence of the highly plesiomorphic apostasioid group in tropical east Asia, it has long been tempting to suggest this area as the ‘cradle’ of the orchids, as did Garay (1972), placing its origin in the phytogeographic area of ‘Malaysia’ (probably corresponding to Malesia, which comprises the Philippines, New Guinea, Indonesia and Peninsular Malaysia). Givnish et al. (2016) used their dated analysis based on plastid loci to examine patterns across the family, concluding that the family likely originated in Australia. In contrast, Pérez-Escobar et al. (2024) used their highly sampled genomic-level analysis to reanalyse area patterns and concluded that the area of origin for the family was likely Laurasian.

Classic biogeographic patterns such as the North American–east Asian vicariance have been described in various groups of orchids (Chen, 1983), but most biogeographic studies in the family have focused on particular genera or suprageneric groups in a phylogenetic context. Repetur et al. (1997) examined biogeographic patterns of Bromheadia using a morphologically based phylogenetic analysis and analysed the relationship of the south-east Asian regions in which that genus is distributed, suggesting that Borneo is the ancestral area and that sympatric speciation may have predominated. Cameron and Chase (1999) studied Pogoniinae, which includes an example of the North American–east Asian vicariance (Pogonia), explaining the long-distance disjunctions in that genus by dispersal and migration. Micheneau et al. (2008) used four plastid regions for 138 angraecoid orchids (Angraecinae) and outgroups, showing that the Neotropical members of the subtribe comprise a single clade and hypothesizing Madagascar as the most probable ancestral area for the subtribe. Fan et al. (2009) used ITS and two plastid sequences to produce a tree for Holcoglossum from which they drew conclusions about shifts from the south towards the Qinghai-Tibetan Plateau in China. Guo et al. (2012) concluded that the ancestral area for cypripedioids was likely the Nearctic + Neotropics based on analysis of plastid and low-copy nuclear genes. Tsai et al. (2015) analysed patterns within a species complex in Phalaenopsis, showing correlation with classic south-east Asian biogeographic boundaries (Huxley’s, Wallace’s and Lydekker’s Lines), hypothesizing vicariance and dispersal to account for current distributions. Xiang et al. (2016) performed a detailed analysis of Dendrobium in south-east Asia, resolving a northern Asian clade sister to an Australasian clade that they estimate split during the lower Oligocene. Pérez-Escobar et al. (2017) studied the biogeography of Pleurothallidinae in South America, concluding that the group probably originated in the Amazonian lowlands and experienced its impressive diversification in concert with the Andean orogeny. Nauheimer et al. (2018) examined biogeographic patterns in Thelymitra, finding that early diversification took place mainly in south-western Australia, with a number of eastward long-distance dispersal events that may be explained by prevailing westerly winds on the continent. Li et al. (2022) examined Gastrochilus and explained its distribution by a mix of vicariance and dispersal events associated with global cooling in the Pliocene–Pleistocene. Carnevali et al. (2023) analysed the biogeography of Encyclia and concluded that its origins lay in Mexico and adjacent countries in Central America, apparently dispersing outward and diversifying in South America and the Caribbean.

CHARACTER ASSOCIATION WITH DIVERSIFICATION

Phylogenetic patterns within orchids have formed the basis for evolutionary conclusions at all levels. By mapping character states on trees, we can trace patterns of change in characters both simple and complex, as in the study by Neubig et al. (2009), in which they showed that the unusual setose fruit of Dichaea had several losses and gains within the genus. Perez-Escobar et al. (2016) traced the evolution of floral sexual systems in Catasetinae, showing that protandry evolved once in the core group of genera followed by three separate originations of environmental sex determination (ESD) among these, suggesting that protandry is a necessary antecedent to ESD. Mosquera-Mosquera et al. (2019) studied changes in states of pollinaria across the epidendroid tree. Mauad et al. (2022) traced the evolution of floral structures associated with pollination in Catasetum, focusing on transformations in column structures (‘antennae’). Connecting such character patterns with diversification shifts on the tree is an especially powerful approach to understanding the origins of diversity across the tree. Breitkopf et al. (2015) used trees, dating and ancestral area analysis of Ophrys to show that evolutionary patterns in these sexually deceptive orchids were characterized by shifts to different groups of bees followed by bursts of diversification. Other studies have attempted to understand the characters associated with diversification in orchids more broadly. Gravendeel et al. (2004) tested for association between habit (terrestrial versus epiphytic; Fig. 3) and diversification, examining habit numbers within a selection of genera, as well as plotting the shift on two clades within the family. They found that epiphytism is positively correlated with diversification in these groups (especially twig epiphytism). Of the characters that they tested in Epidendroideae, Freudenstein and Chase (2015) found that epiphytism had the strongest correlation with diversification in that subfamily, which comprises the majority of orchid epiphytes. Givnish et al. (2015) identified several characters associated with diversification across the family, including epiphytism, appearance of pollinia, CAM photosynthesis, tropical distribution, and lepidopteran or euglossine pollination. Collobert et al. (2023) studied evolution of epiphytism among epidendroids, concluding that it probably evolved three times, and examined its association with stem and leaf succulence. Reversions to the terrestrial habit have also been documented, as in the study of Galeandra by Martins et al. (2018). Thompson et al. (2023) studied the correlation of various factors with diversification in a dated tree of Orchidoideae and concluded that global cooling patterns were most important (more than morphological or habit features) for explaining diversification. Zhang et al. (2023) examined diversification across the family and detected upshifts in diversity in Epidendroideae that are not correlated with the evolution of epiphytism, suggesting that epiphytism alone is not responsible for diversification in this subfamily.

PATTERNS OF FUNGAL ASSOCIATION IN A PHYLOGENETIC CONTEXT

The association of orchids with fungi and its evolution through time provide an additional dimension for study of orchids. Orchids predominantly associate with Basidiomycota, although Ascomycota are sometimes involved (Selosse et al., 2022). Those orchids that eventually become autotrophic typically use Rhizoctonia-type Basidiomycota for germination. Rhizoctonia is a form genus used for non-sporing (anamorphic) forms of these fungi, the orchid-associated ones of which belong to the teleomorphic genera Ceratobasidium, Thanetephorus, Oliveonia, Serendipita and Tulasnella. Holomycotrophs may also use these fungi for germination, but at maturity are usually associated with ectomycorrhizal (ECM) fungi, presumably due to their greater ability to supply fixed carbon (Ogura-Tsujita et al., 2012). Specificity of the interaction is an inherently phylogenetic parameter that is of particular interest in these studies, given its conservation implications for rare species. Specificity appears to be somewhat variable among orchids that use Rhizoctonia groups, from narrow to somewhat broader. As early as 1909, Bernard argued that some degree of specificity exists based on his morphological assessment of fungal identity; contrasting patterns of specificity were documented by Harvais and Hadley (1967) and Warcup (1971) in British and Australian terrestrials. However, morphological identification of species of Rhizoctonia fungi is challenging given that the sporing stage is required for precise identification and this is rarely observed in many cases. The introduction of molecular methods (e.g. White et al., 1990; Gardes and Bruns, 1993) greatly facilitated this work. Taylor and Bruns (1997) used phylogenetic analysis of RFLP data from orchid endophytes and sporocarps of potentially associated fungi to show that Corallorhiza maculata and Cephalanthera austiniae used different families of ECM fungi. For holomycotrophs using ECM fungi, specificity is typically quite narrow and may involve a single species of fungus (although exceptions to this pattern exist; cf. Roy et al., 2009). In a study conducted within a species, Taylor et al. (2004) showed that genetic races in C. maculata used distinct fungal groups.

Phylogenetic trees of the plant and fungal associates can be constructed and compared in order to test hypotheses of phylogenetic tracking of the orchids on fungal lineages and to reveal patterns of host shifts. In many cases one can see some evidence for both patterns, although one or the other may predominate. There is a significant body of literature on analysing host–parasite patterns and methods that can be applied to studying orchid systems. Barrett et al. (2010), for example, related clades of Corallorhiza striata accessions from across their range to the tree of their Tomentella associates, revealing a mix of tracking and host-shifting events between the two. Shefferson et al. (2010) traced shifts in host breadth on a tree of species of Goodyera from Asia and North America, revealing sequential expansions and contractions similar to what they had seen in Cypripedium (Shefferson et al., 2007). Motomura et al. (2010) and Ogura-Tsujita et al. (2012) traced the shifts of fungi used by Cymbidium in a clade in which holomycotrophy has evolved, revealing the shift to wholescale dependence on ECM fungi. Jacquemyn et al. (2011) used trees of hosts and fungi to show broad specificity in many cases for a group of 16 Orchis species. Kennedy et al. (2011) showed varying degrees of specificity within Hexalectris (now Bletia) for Sebacinaceae hosts. Tupac Otero et al. (2011) found a mix of host-tracking and host-shifts among Pterostylidinae and their Ceratobasidium hosts. Freudenstein and Barrett (2014) investigated fungal associates of Corallorhiza wisteriana across the USA and found different fungal families involved in the eastern and western portions. They concluded that this feature, along with morphological and habitat differences, could be the basis for recognizing two species in this group. Yagame et al. (2016) depicted the shift in Neottia from non-ECM Sebacinales to ECM Sebacinales corresponding with the transition from leafy to leafless species.

GENE AND GENOME EVOLUTION

Orchidaceae have been the subject of studies of genome and gene evolution for over two decades. On the genome level, the first studies worked with DNA amount (C-value) as assessed by flow cytometry and sometimes considered chromosome numbers. The study by Cox et al. (1998) examined variation in DNA amount and chromosome number in cypripedioids. Leitch et al. (2009) broadened the analysis to cover patterns across the family and Jersakova et al. (2013) broadened this to include apostasioids, revealing that the unusually large genomes present in some members of cypripedioids and vanilloids appear to have arisen within these groups and are not the primitive state for the family. Moraes et al. (2012) examined genome size variation in Maxillariinae and found no correlation with change in chromosome number. Unruh et al. (2018) investigated genome evolution in Cypripedioideae, many of which are known to have large genomes. They asked whether polyploidy (WGD) could be responsible for genome size increase in this subfamily and did not find evidence for it there but did find evidence for WGD at the base of the family. Chumová et al. (2021) examined the pattern of repetitive DNA, polyploidy and type of endoreplication in a phylogenetic context in Pleurothallidinae, finding that genome size was not correlated with chromosome number but was strongly correlated with repetitive DNA content and degree of endoreplication. Moraes et al. (2022) examined correlations between chromosome number and genome size and ecological parameters in Neotropical orchids broadly; they found an intriguing association between small genome size and epiphytism (also seen by Chase et al. (2005) in Oncidiinae), which could be due to the fluctuating moisture levels that epiphytes can experience. Such constraints on genome size in arid environments have been observed in palms (Schley et al., 2022).

Plastomes are a subject of great interest with respect to molecular evolution. In general, among angiosperms they are highly conserved with respect to gene order and composition. Li et al. (2019b), for example, investigated plastome evolution in the autotrophic Holcoglossum and identified five mutation hotspots for the 12 species but, other than some variation in plastome length, found conserved structure. Many of the most revealing investigations with respect to plastomes in orchids are among the holomycotrophs, however. Their unusual trophic nature makes them interesting because with reduced selection on genes associated with photosynthesis, gene loss and high rates of change occur. Freudenstein and Doyle (1994a), using restriction fragment mapping, depicted changes in the large single-copy region of the plastid genome in the leafless Corallorhiza in a phylogenetic context. Barrett and Freudenstein (2008) documented the plastome gene degradation that occurs in leafless mycoheterotrophs, using rbcL sequences in Corallorhiza as an example. Delannoy et al. (2011) characterized the plastome of Rhizanthella gardneri, the ‘underground orchid’ from Australia, and found that it comprised only 59 kb. Barrett et al. (2014) examined whole plastomes in Corallorhiza and traced patterns of gene loss and pseudogenization among species in that genus. Barrett et al. (2019) found an unexpected four to five parallel losses of photosynthesis within the small genus Hexalectris (Bletia). Murray (2019) investigated the plastomes of New Zealand Orchidoideae, including Danhatchia, Corybas and Microtis, documenting plastome reduction in the holomycotrophic Danhatchia but also in the two autotrophic genera. Li et al. (2020) traced patterns of plastid genome evolution in Calypsoinae, the subtribe that contains Corallorhiza as well as the leafless Danxiaorchis and possibly Risleya. Kim et al. (2020) showed patterns of plastome change among orchids and confirmed previously observed patterns of gene loss, noting 104 gene rearrangements in the 106 plastomes they analysed, many near the IRa/SSC border. Wen et al. (2022) investigated the genomes of a number of 11 species of Gastrodia and of Didymoplexis pallens. Gastrodia exhibited plastome sizes of 29–36 kb, while Didymoplexis was 51 kb. The species diversity in Gastrodia has long been unclear; the large number of species described in recent years suggests that it could be near 100, making it the largest holomycotroph genus. Barrett et al. (2023) studied the plastome of the orchidoid Degranvillea dermaptera and found a reduction to 46 kb. Barrett et al. (2024) similarly recovered highly reduced plastomes for six accessions of the early-divergent yet phylogenetically unplaced epidendroid Wullschlaegelia calcarata, which were ~37 kb in size. Zhao et al. (2024) included Epipogium in their plastome analysis, in which they also assembled its mitochondrial genome (414 kb in 26 circular chromosomes), concluding that ten regions had been transferred from the plastome to the mitochondrial genome.

The plastomes of sufficient orchid species have been examined that it has been possible to characterize the stages of plastome change in these plants and produce a model for the progression of gene loss (Barrett and Davis, 2012; Graham et al., 2017). Not surprisingly, genes explicitly related to photosynthesis are first to be lost and those associated with basic ‘housekeeping’ features of the plastome are last. Currently known plastome sizes from presumed autotrophs range from 127 kb (Microtis unifolia) to 234 kb (Cypripedium lichiangense), although most range from 145 to 160 kb. This is in contrast to holomycotrophs, in which plastomes range from 14 kb (Pogoniopsis schenkii) to 151 kb (Corallorhiza macrantha). We know very little about changes in nuclear genomes of holomycotrophs, but in one of the first studies to examine transcriptomes from nuclear genes among holomycotrophs (including two orchids), Timilsena et al. (2023) found a group of 1375 standard orthologues normally expressed in green plants that could not be detected in the holomycotrophs.

Mitochondrial genomes provide another opportunity for phylogenetically useful characters and for evolutionary insights but they have been much less studied in orchids than plastomes. Sinn and Barrett (2020) showed that two mitochondrial fragments from Ustilaginaceae-type fungi have been transferred to orchid mitochondrial genomes – one in the ancestor to all orchids and the second in an epidendroid ancestor. Valencia-D. et al. (2023) found the same pattern in their analysis. Much potential for more discovery lies in further investigation of these genomes.

A phylogenetic perspective is also essential for understanding the homology issues that are revealed in evolutionary development work in orchids. The morphological evolution of floral structure in the family has synorganization as a principal theme, given the fusion of structures resulting in the column and long-standing suspicions that the labellum could also be a hybrid structure including both petal and stamen elements, at least in some circumstances (Brown, 1830; Darwin, 1862). Much has been learned about the basis of floral organ identity and especially the role of MADS-box genes in model systems such as Arabidopsis. It is in groups with specialized floral morphology such as orchids that these approaches hold particular promise for understanding how such flowers have arisen. Studies such as those summarized in Mondragón-Palomino (2013) have leveraged this comparative approach to begin to understand the evolution of these floral identity gene families in orchids and to connect these patterns with phenotypes. Hsu et al. (2015) focused on the factors distinguishing labellum morphology from that of other petals, identifying protein complex variants derived from MADS-box genes that appear to specify the difference. Lin et al. (2016) investigated MADS-box loci in Erycina transcriptomes, identifying 28 that were connected to putative function in floral development. Further molecular developmental work on Erycina pusilla floral structures suggests that the labellum callus and stelidia may be staminal in origin (Dirks-Mulder et al., 2017) and provide insights into the evolution of fruit dehiscence (Pramanik et al., 2023). Pramanik et al. (2020) also found a staminal and partly staminal origin for the stelidia and callus, respectively, in Phalaenopsis, and a sepaloid–petaloid–staminal origin for the mentum in that genus. Clearly, there is more potential for exploration of the origin of structures resulting from synorganization using developmental genetic approaches.

As sequencing technology has improved and costs have decreased, and with concomitant developments in bioinformatic methods, it has become possible to sequence and compare entire nuclear genomes of orchids in addition to the much smaller organellar genomes. It is unclear whether even more data from whole genomes beyond an already highly sampled reduced representation of these genomes will produce improved phylogenetic hypotheses, however, except perhaps in areas of the tree where branch lengths remain especially short. Although our trees may have maximal branch support on their nodes, they may still display significant lack of concordance in the phylogenetic signal derived from different genes or regions of the genome, and additional data may not resolve those discordances. Nonetheless, there is great potential for what we can learn from whole genomes with respect to the basis for phenotypic changes across the orchid tree and thus our understanding of evolutionary change and its implications in the group.

The first steps towards the realization of genome-level study have been taken with the sequencing of several orchid nuclear genomes. Cai et al. (2015) were the first to publish an analysis of one of these genomes (Phalaenopsis equestris). They found 29 431 predicted protein-coding genes and compared the genome with those of other plants, detecting evidence of a WGD event in the orchid lineage, as well as gene family expansions in MADS-box loci that the authors suggested could be responsible for the floral diversity seen in orchids. They also detected some expansion in CAM-related genes, hypothesizing that this could have provided the basis for increased efficiency via dosage effects. Additional genomes have appeared since then, including Dendrobium catenatum (Zhang et al., 2016), Apostasia shenzhenica (Zhang et al., 2017), Cymbidium sinense (Yang et al. (2021) and the leafless Gastrodia menghaiensis (Jiang et al., 2022). The addition of each genome sequence allows greater resolution of changes that have occurred in the evolution of orchids when placed in a phylogenetic context. As of this writing, genome assemblies for ~17 species and three hybrids of orchids have been published and/or are available in public databases, including Apostasia (two species), Bletilla, Cymbidium (three species), Dactylorhiza, Dendrobium (five species and one hybrid), Gastrodia (two), a Papilionanthe hybrid, Phalaenopsis (two species and one hybrid), Platanthera (two species) and Vanilla.

PROSPECTS—WHAT WE NEED TO KNOW

Given the impressive amount that has been learned about orchid relationships and the evolutionary studies that have used this framework to draw conclusions, we might be tempted to think that much of the work is finished. However, in other ways the work has just begun. Given that many species of orchids are still being described each year, it is clear there is significant diversity yet to be discovered at a time when human-induced change in high-quality habitats continues to rise, raising the stakes for documenting diversity, as is the case in so many organismal groups.

Although we now have a reasonably supported structure for the family at higher levels, there is still significant work to be done to understand relationships near the tips. Subtribes characterized by small flowers such as Pleurothallidinae and large genera such as Bulbophyllum undoubtedly harbour secrets yet to be discovered. Additional rare and easily overlooked holomycotrophs are certainly yet to be found, especially in tropical areas. Learning as much as possible about them in situ, including pollination biology, and collecting material for plant and fungal isolation when they are found, is essential to integrate them into growing data sets. Overall, there remains much phenotypic character information to be collected for many species, since these are the features on which selection is presumably operating and thus will enable connecting the phylogenetic pattern generated by molecular data to features important in their ecological function.

Genome-level analysis of orchids is in its infancy. There is enormous potential to increase our understanding of character changes across the family as we begin to discover groups of loci that are responsible for specifying structures, chemistry and developmental patterns of the plants. Putting these into context by comparison with asparagalean relatives will be important as we begin to flesh out the reasons why orchids as a group have been so successful in diversification. Analysing the interactions of gene products from orchids and their associated fungi during the initiation and development of the parasitism also promises to shed light on evolution of the orchids and will aid in understanding how such an interaction began, likely through co-opting genes already present that are normally involved with arbuscular mycorrhizal relationships (Radhakrishnan et al., 2020). Much remains to be learned about orchids’ relationships with fungi in general and the diversity of those that are used by orchids, and this is best understood in a phylogenetic context. This information not only advances our understanding of orchid evolution and ecology but has practical application in the conservation of orchids, many of which are rare species.

Lastly, new fossils, especially in amber, undoubtedly remain to be discovered. Given the poor preservation of orchid parts as fossils in general, special situations such as amber nodules and sites where charcoalified fossils are present should be focused on for further fossil discovery that may provide information on character combinations that do not exist among extant species and for chronological information that will allow continued refinement of tree dating. With respect to amber, not only is there potential for finding new orchid fossils in the many deposits of this material across the globe, but collections of this material that already exist may well harbour orchid fragments just waiting to be discovered.

Note: The review paper by Wang et al. (2024) came to the author’s attention after the original submission of this paper.