|Home | About | Journals | Submit | Contact Us | Français|
The nearly 30 000 species of orchids produce flowers of unprecedented diversity. However, whether specific genetic mechanisms contributed to this diversity is a neglected topic and remains speculative. We recently published a theory, the ‘orchid code’, maintaining that the identity of the different perianth organs is specified by the combinatorial interaction of four DEF-like MADS-box genes with other floral homeotic genes.
Here the developmental and evolutionary implications of our theory are explored. Specifically, it is shown that all frequent floral terata, including all peloric types, can be explained by monogenic gain- or-loss-of-function mutants, changing either expression of a DEF-like or CYC-like gene. Supposed dominance or recessiveness of mutant alleles is correlated with the frequency of terata in both cultivation and nature. Our findings suggest that changes in DEF- and CYC-like genes not only underlie terata but also the natural diversity of orchid species. We argue, however, that true changes in organ identity are rare events in the evolution of orchid flowers, even though we review some likely cases.
The four DEF paralogues shaped floral diversity in orchids in a dramatic way by modularizing the floral perianth based on a complex series of sub- and neo-functionalization events. These genes may have eliminated constraints, so that different kinds of perianth organs could then evolve individually and thus often in dramatically different ways in response to selection by pollinators or by genetic drift. We therefore argue that floral diversity in orchids may be the result of an unprecedented developmental genetic predisposition that originated early in orchid evolution.
The typical flower of a petaloid monocot consists of five fundamentally 3-fold whorls or derivatives thereof. However, in contrast with other petaloid monocots, such as lilies or tulips, orchids have flowers of breathtaking morphological diversity (Fig. 1A). This diversity is mainly brought about by variation on a relatively simple scheme (Fig. 1B–D). Like flowers of lilies and tulips, those of orchids comprise two whorls of petaloid organs termed tepals surrounding the reproductive organs. In orchids, reproductive organs are special in that they constitute a gynostemium or column, a compound structure formed by adnation of male and female organs (Dressler, 1993; Rudall and Bateman, 2002). Species-specific variation in the size and shape of the column, together with the presence of appendages, the conformation, position and number of the anthers, as well as characteristics of the pollinia and other structures, make the column a remarkably complex organ. Evolution of the orchid column is worthy of detailed study, and there is relatively detailed information on the molecular basis of its development (Yu and Goh, 2000; Johansen and Frederiksen, 2002; Tsai et al., 2004, 2005; Skipper et al., 2006; Song et al., 2006; Xu et al., 2006; Kim et al., 2007); however, here the focus will be on development and evolution of the perianth organs.
Although the perianth of most petaloid monocot families consists of (almost) identical organs, three kinds of ‘organ identity’ have been distinguished in the perianth of orchids: three outer tepals (T1–T3; often also termed ‘sepals’) in the first floral whorl, and two lateral inner tepals (t1 and t2; ‘petals’) and a median inner tepal (t3) termed the labellum or lip in the second whorl (Rudall and Bateman, 2002; Mondragón-Palomino and Theißen, 2008; Fig. 1B–D). Although most descriptions of the orchid perianth consider the outer tepals (T1–T3) unlobed organs without adornments, they can be similar to the lateral inner tepals (Fig. 1A) or form nectar spurs (e.g. Plectophora, Oncidiinae). The lip is, with few exceptions, always different from the other perianth organs and elaborately adorned with calli, spurs, glands and a distinctive colour pattern (Fig. 1A). Although the lip is probably homologous to the adaxial tepal of other monocots and hence it should be the uppermost one, it is often the lowest one due to resupination (180° developmental rotation in floral orientation; Arditti, 2002). The abaxial orientation of the resupinate lip and its location in direct opposition to the fertile anther suggest that its strong degree of morphological elaboration resulted from adaptations to specific pollinators.
Questions concerning causes of evolution are generally difficult since they can address fundamentally different things. One can distinguish between ‘ultimate causes’ such as selection or drift and ‘proximate causes’ such as the molecular developmental genetic mechanisms facilitating or constraining evolutionary change. From Darwin onwards, orchid biology has focused on the ultimate causes of morphological diversity and species richness (Darwin, 1862; Cozzolino and Widmer, 2005; Schlüter and Schiestl, 2008). Specifically, the impressive floral diversity of orchids has been attributed to adaptation to specific pollinators (e.g. Johnson et al., 1998). Although in Orchidaceae there is a wide range of specificity in plant–pollinator interactions, it has been estimated that about 60 % of orchid species have only one recorded pollinator species (Tremblay, 1992), and this specificity has been considered an important ethological, prezygotic mechanism of reproductive isolation (reviewed in Cozzolino and Widmer, 2005). Phenomena such as the attraction of pollinators by mimicking food or mating partners without providing a reward (termed food and sexual deception, respectively) have fascinated many researchers and have been studied intensively (Jersáková et al., 2006).
However, increased understanding of the ultimate causes of orchid evolution tells us little about the reasons why orchids, but not, for example, Liliaceae or Hypoxidaceae, exhibit extreme floral diversification. One may hypothesize, for example, that special genetic and developmental properties could have contributed to orchid diversity. In contrast to the great interest in the ultimate causes of orchid floral diversity, possible proximate causes are a severely neglected topic and remain speculative. Exceptionally, Bateman and Rudall (Bateman, 1985; Rudall and Bateman, 2002, 2003; Bateman and Rudall, 2006) have integrated current knowledge about the genetics of flower development in model species such as Arabidopsis thaliana to predict the genetic basis of organogenesis in wild-type and peloric orchid flowers. These contributions, together with recent knowledge on floral developmental control genes in orchids, enable an improved understanding of orchid evolution.
Recently, we proposed a theory on the developmental determination and evolution of organ identity in the orchid perianth (Mondragón-Palomino and Theißen, 2008). The underlying developmental-genetic code for organ identity in the orchid perianth (here termed the ‘orchid code’) assumes that identity of the different organs in orchid flowers is specified by the combinatorial expression of orthologues of organ identity genes known from eudicot model plants such as Arabidopsis thaliana and Antirrhinum majus. These include DEF-like (or AP3-like) and GLO-like (or PI-like) genes specifying stamen and petal identity, collectively termed class B floral homeotic genes. Although identity of petals in eudicot model plants involves interaction and function of one DEF-like and one GLO-like gene, our theory proposes that the identity of different petaloid tepals in the perianth of orchid flowers is specified by the combinatorial interaction of four paralogous DEF-like genes with one GLO-like gene. Phylogenetic reconstructions have indicated that these DEF-like genes fall into four distinct clades, termed PeMADS2-like (clade 1), OMADS3-like (clade 2), PeMADS3-like (clade 3) and PeMADS4-like (clade 4) (Mondragón-Palomino and Theißen, 2008). These genes show highly conserved, clade-specific expression patterns. Based on these findings, our theory maintains that in the primordia of the first floral whorl, the combined expression of clade 1 and clade 2 genes determines formation of outer tepals (T1–T3). In the second whorl, identity of lateral inner tepals (t1 and t2) is determined by the combined action of clade 1, clade 2 and clade 3 genes. Identity of the lip (t3) is specified by the organ-specific expression of a clade 4 gene in addition to expression of all other DEF-like genes (Fig. 2). Our theory implies that differential expression of clade 3 genes distinguishes between inner and outer tepals, whereas differential expression of clade 4 genes distinguishes between identities of the lateral inner tepals and lip (Fig. 2).
Clade 1 and clade 2 DEF-like genes represent orchid-specific sister clades, and the same is true for clade 3 and clade 4 genes; with these and some other findings concerning the phylogeny of orchid DEF-like genes, it was possible to establish a relationship between molecular evolution of these genes and morphological differentiation of the orchid perianth (Mondragón-Palomino and Theißen, 2008). We hypothesized that gene duplications played a fundamental role in perianth differentiation in the common ancestor of all Orchidaceae. The most recent common ancestor of orchids and the rest of Asparagales, probably had an actinomorphic perianth composed of six almost identical tepals in which an ancestral DEF-like gene was uniformly expressed. According to our theory, the first duplication of the DEF-like gene gave rise to the ancestor of clade 1 and clade 2 genes and the ancestor of clade 3 and clade 4 genes. Evolution of differential expression of the precursor of clade 3 and clade 4 genes may have led to the establishment of different organ identities for outer (gene expression ‘off’) and inner (gene expression ‘on’) tepals. Similarly, another gene duplication gave rise to clade 3 and clade 4 genes, and differential expression of clade 4 genes led to distinction between lateral inner tepals (gene expression ‘off’) and the lip (gene expression ‘on’; Fig. 2). Based on these considerations, it is safe to assume that different cis-regulatory elements evolved in the four clades of DEF-like genes after their origin from a common ancestor gene, and these are now responsible for the differential expression of these genes. Moreover, it also appears plausible that these paralogous DEF-like genes must respond to positional cues in different ways to obtain their characteristic expression patterns. We hypothesize that at least some of their cis-regulatory elements respond to positional cues within floral primordia, such as a basipetal–acropetal gradient in the case of clade 3 genes and an adaxial–abaxial (dorsiventral) gradient in the case of clade 4 genes. Two ‘clines’ of gene expression in the orchid flower were previously inferred by Bateman and Rudall (2006).
Dorsiventral ‘pre-patterns’ existing throughout floral primordia due to dorsal expression of CYCLOIDEA-like transcription factors, which are members of the TCP family, appear to be good candidates for providing positional information for clade 4 gene expression (Mondragón-Palomino and Theißen, 2008). Although there is no experimental evidence yet linking TCP-type transcription factors to dorsiventral differentiation of the orchid perianth, studies in model species have already suggested genetic mechanisms responsible for this process. Specifically, early in flower development of eudicot Antirrhinum majus, expression of CYCLOIDEA (CYC) and DICHOTOMA (DICH) in whorls 1, 2 and 3 determines differential growth and number of dorsal organ primordia in the floral meristem (Luo et al., 1996, 1999). Later, CYC expression consolidates the zygomorphic configuration of whorls 2 and 3 by influencing differential growth and shape of dorsal petals and arrest of the dorsal stamen (Luo et al., 1996). In A. majus, this last expression of CYC depends on transcription of the class B floral organ identity gene DEF (Clark and Coen, 2002). These findings make it conceivable that rather than being upstream of the DEF-like genes, TCP-type genes in orchids function as direct or indirect target genes of DEF-like genes or even both up- and downstream of DEF-like genes. If TCP-type genes in orchids function in a way similar to those in A. majus, they would be expected to influence differential development of the lateral outer tepals (T2 and T3) as compared with the median tepal (T1) and especially of the lip (t3) as compared with the lateral inner tepals (t1 and t2; as exemplified in Fig. 1C), as well as being involved in developmental arrest and eventual suppression of dorsal stamens in Orchidaceae. Research on CYC and its orthologue TCP1 from the distantly related and actinomorphic eudicot Arabidopsis thaliana showed that these genes are dorsally expressed in both floral meristems and axillary shoots (Cubas et al., 2001). Because axillary shoots do not have dorsiventral symmetry, this suggests that CYC and TCP1 respond to a pre-pattern in shoots and floral meristems, whereas effects are only obvious in floral development (Cubas et al., 2001). The actinomorphic symmetry of A. thaliana is possibly the outcome of TCP1 being only transiently expressed in early stages of floral development.
So far, the genetic basis of dorsiventral pre-patterning in the floral meristem of orchids remains unknown. However, existence of a dorsiventral gradient of positional cues influencing specification of distinct organ identities in the orchid perianth is corroborated by the developmental gradient observed in the floral apex of monandrous orchids. Specifically, organ initiation follows a dorsiventral sequence, with primordia of the outer lateral tepals (T2 and T3) emerging first, followed by primordia of the lip (t3), lateral inner tepals (t1 and t2) and finally the median outer tepal (T1) (Kurzweil, 1987a, b, 1988).
As mentioned in the previous section, the phenomenon of resupination changes the orientation of many orchid flowers during anthesis so the median inner tepal or lip, although initially in an adaxial/dorsal position, becomes the lowermost perianth organ. This process also takes place in some other species with zygomorphic flowers such as Lobelia (Campanulaceae) and Orchidantha (Lowiaceae), but it is clearly more widespread in Orchidaceae. Assuming that this character facilitates pollination by presenting the lip as a landing platform and nectar guide, it is reasonable to assume that it evolved after the flower became zygomorphic. This is well exemplified by Apostasioideae, the orchid subfamily sister to the other four, in which resupination occurs in the zygomorphic genus Neuwiedia but not in the actinomorphic Apostasia (Kocyan and Endress, 2001). This scenario begs the question of molecular mechanisms that associated zygomorphy with the subsequent torsion of the pedicellate ovary. Experimental removal of column and pollinia indicate that resupination occurs as a response of the flower to gravity (Nyman et al., 1985), which is mediated by auxin and other hormones produced by orchid pollinia and associated with gravitropic phenomena in plants (Nyman et al., 1985; Nair and Arditti, 1991). The link between pollinia and resupination is further supported by observations in some species of Catasetum, in which the female flowers are non-resupinate, whereas the male ones are resupinate even when they occur in the same inflorescence (Dressler, 1993). This suggests that pollinia are somehow associated with the process of resupination.
Although orientation of the lip in a way that suits pollinator behaviour provides a convincing ultimate explanation for the evolution of resupination, we wonder why a similar morphological change did not evolve by a much simpler mechanism, i.e. evolution of the abaxial/ventral median outer tepal (rather than the median inner tepal) into a lip, since this organ would already have been in the correct position and orientation. Our theory about the evolution of organ identity in the orchid flower offers a proximate cause: the distinction between the lip and other tepals depends on the unique expression of clade 4 DEF-like genes, which may have evolved under the control of an adaxial/dorsal positional cue. Shortly after origin of the clade 4 gene by duplication from a clade 3 and 4 precursor, all inner tepals probably had the same structure, but soon the lip may have become different from the two other inner tepals to serve special functions. An important new function was pollinator attraction, and resupination originated as a secondary mechanism to make the lip also function as a landing platform. Admittedly, this hypothesis is extremely speculative. However, to the best of our knowledge it provides the first evolutionary developmental genetic explanation for resupination and could be experimentally tested once more is known about the cis-regulatory elements of clade 4 DEF-like genes and their trans-acting factors. A similar logic could be employed to approach the evolution and development of the perianth of flowers in Orchidantha (Lowiaceae), which similarly to orchid flowers are resupinate and have a median inner tepal modified into a lip (Kirchoff and Kunze, 1995).
Previously, Bateman and Rudall categorized naturally occurring floral terata in Orchidaceae by systematically distinguishing six classes of peloria and pseudopeloria (Bateman, 1985; Rudall and Bateman, 2002, 2003; Bateman and Rudall, 2006). With the exception of type A pseudopeloria (assumed to originate by heterochrony), all terata involve homeotic transformation of one or more tepals. Although in peloric flowers the wild-type zygomorphic symmetry is completely lost due to development of an actinomorphic perianth, in pseudopeloric forms it is only reduced. The authors distinguished three categories each of peloric and pseudopeloric orchids by invoking homeotic substitutions within the second whorl or between the first and second whorl (Bateman, 1985; Rudall and Bateman, 2002, 2003; Bateman and Rudall, 2006).
To generate hypotheses about developmental and genetic bases of teratological flowers in orchids, we inferred some simple and plausible rules that may govern function of DEF-like genes. One assumes that in any given organ, clade 3 gene function requires expression of clade 1 and clade 2 genes, and clade 4 gene function requires expression of clade 1, clade 2 and clade 3 genes. Another rule assumes that loss of DEF-like gene function is recessive, whereas ectopic expression of clade 3 or clade 4 genes within the perianth leads to a dominant gain-of-function. Here it is shown that on these grounds, all categories of floral terata considered by Rudall and Bateman (Rudall and Bateman, 2002; Bateman and Rudall, 2006) plus another one not considered by these authors can be explained by monogenic gain- or-loss-of-function mutants of clade 3 or clade 4 DEF-like genes. Intriguingly, supposed dominance or recessivity of mutant alleles is correlated with the frequency of the terata in both cultivation and nature.
In type A peloria the lateral inner tepals (‘petals’; t1 and t2) are transformed into lip-like structures (t3; Fig. 1B). This is a frequent kind of peloria in cultivation (see, for example, Chen et al., 2005; and our own observations), where it can result from somaclonal variation after tissue culture. Besides full transformants, ‘semi-peloric’ variants in which the lateral inner tepals are partially transformed into lip-like structures are frequent and commercially available (Wallbrunn, 1987; Chen and Chen, 2007). Type A peloria may also represent the most common category of natural terata affecting perianths of orchids in nature; for example, it has been recorded in about 25 % of British native orchid species, including several species of Dactylorhiza and Ophrys (Bateman and Rudall, 2006).
According to our theory on specification of organ identity in orchid flowers, type A peloria results from ectopic expression of a clade 4 gene in the lateral inner tepals (Fig. 3B). Our hypothesis is based on work by Tsai et al. (2004) in Phalaenopsis equestris, in which a variant of type A peloria shows this kind of ectopic gene expression. However, this teratological orchid shows a mutant change in the putative promoter region of its clade 2 gene that may have caused loss-of-function of that gene (Tsai et al. 2004). This may indicate that clade 2 DEF-like genes regulate clade 4 genes such that they prevent their ectopic expression in the lateral inner tepals, but this inference certainly requires further investigation.
According to the combinatorial rules of the ‘orchid code’, type B peloria may result from loss-of-function of clade 4 genes, leaving only expression of clade 1, 2 and 3 genes in the median inner tepal that, therefore, acquires the same identity as the lateral inner tepals (Fig. 3C). The phenotype of a type B peloria is here illustrated with Phragmipedium lindenii, a possible teratological form of Phragmipedium caudatum (Figs 3C and and1A,1A, respectively) (Hurst, 1925; Bateman and Rudall, 2006; Mondragón-Palomino and Theißen, 2008). Another well-known case in point is Calochilus imberbis, which has been considered a ‘hopeful monster’ (Burns-Balogh and Bernhardt, 1986) that like C. robertsonii may proliferate autogamously (Tonelli, 1999) (Fig. (Fig.4).4). Overall, examples of type B peloria are relatively frequent, but less common than examples of type A peloria (Bateman and Rudall, 2006).
According to the ‘orchid code’, type C peloria may result from lack of function of clade 3 DEF-like genes (Fig. 3D). Since clade 4 gene function is assumed to depend on clade 3 genes, only clade 1 and clade 2 gene activity is left in all tepals that, therefore, adopt the identity of outer tepals (Fig. 3D). Loss of clade 3 DEF-like gene function leading to type C peloria (Fig. 3D) is possibly behind the independent emergence of rare actinomorphic genera within zygomorphic groups of subfamily Orchidoideae. Examples of this are the Australasian genus Thelymitra (tribe Diurideae; Fig. 3D) or the monospecific genera from tribe Neottieae Diplandorchis, Tangtsinia, Sinorchis and Holopogon (Komarov, 1935; Chen, 1965, 1978, 1979). More specifically, Dressler (1993) and Rudall and Bateman (2002) argued that Diplandorchis and Holopogon are peloric forms of Neottia, whereas Tangtsinia and Sinorchis may be actinomorphic variants of Cephalanthera. Possibly, autogamy and cleistogamy aided these putative ‘hopeful monsters’ to form constant populations (Chen, 1965, 1979). Remarkably, Thelymitra diversified to form a new genus that evolved more complex pollination systems including floral mimicry (Burns-Balogh and Bernhardt, 1986; Tonelli, 1999). A less parsimonious alternative to explain this phenotype would be to consider that all perianth organs have lateral inner tepal identity (yellow). This unclassified peloric form would require both ectopic expression of the clade 3 DEF-like gene in all perianth organs and complete disruption of the clade 4 DEF-like gene (not illustrated).
Pseudopeloric orchid variants can also be explained by gain or loss of clade 3 gene function (Fig. 3E–G). In type B pseudopeloria, the lip adopts outer tepal identity by loss-of-function of the clade 3 gene only in the lip, but not in the lateral inner tepals, most likely by restriction of the expression domain to the lateral inner tepals (Fig. 3E). Although the clade 4 DEF-like gene may still be expressed in the median inner tepal, lip identity does not develop because this would require the combinatorial expression of all DEF-like genes (Fig. 3E).
In type C pseudopeloria, lateral inner tepals adopt outer tepal identity by loss-of-function of the clade 3 gene only in the lateral inner tepals, but not in the lip, most likely by the restriction of the expression domain to the lip (Fig. 3F).
Although it is not clear whether type C pseudopeloria sensu Bateman and Rudall (2006) is the result of homeotic conversion of inner lateral tepals into outer tepals or vice versa, our theory on specification of organ identity in the orchid flower distinguishes between these two possibilities. This distinction is a first step to testing these possibilities by comparing expression patterns of DEF-like genes. Thus, in addition to defining type C pseudopeloria as described above we propose a type D pseudopeloria in which all outer tepals are transformed into organs that adopt lateral inner tepal identity, possibly by ectopic expression of clade 3 genes in the outer tepals (Fig. 3G).
We reason that type C and type D pseudopeloria, in which all inner and outer tepals except the lip are highly similar to each other (Fig. 3F, G), may be more common in Orchidaceae than previously recognized. For example, in the group of Brazilian Cattleya species (formerly Laelia), the flowers of section Parviflorae have a perianth in which all tepals except the lip have similar shape, size and colour (Fig. 3G), giving the perianth a slight degree of actinomorphy. This morphology is in clear contrast to the related sections Hadrolaelia and Cattleyodes (e.g. Cattleya tenebrosa in Fig. 1A) in which all species (van der Berg et al., 2000) have distinct lateral inner and outer tepals. In this and other examples, the molecular phylogeny of the groups involved and pattern of DEF-like gene expression would help to clarify whether apparent morphological differences between these groups resulted from homeotic transitions (Fig. 3F, G) or from changes in downstream targets, as previously discussed (Mondragón-Palomino and Theißen, 2008).
Our hypothesis on the developmental genetic basis of organ identity in the orchid perianth explains, at least in part, the frequency of orchid terata. Assuming three classes of organ identity (‘outer tepal’-like, ‘lateral inner tepal’-like, and ‘lip’-like) on three kinds of positions (in the wild type occupied by three outer tepals, two lateral inner tepals and a lip, respectively) would allow for 27 (= 33) types of flowers; by definition, 26 (= 33 – 1) are teratological configurations. Only a few of these categories are frequent, and some others are extremely rare. We assume that their relative frequency is largely explained by their mutant origin rather than their maintenance by selection. This certainly applies in cultivation, but probably also in nature, where the vast majority of terata appears sporadically and briefly and may presumably be eliminated due to lack of pollinators, if they are not autogamous.
It is thus probably not by chance that terata reported by Bateman and Rudall (2006) are all predicted to be based on monogenic changes (loss- or gain-of-function of a single clade 3 or clade 4 DEF-like gene), as outlined above. Note that terata requiring changes in more than one gene have not been considered here. Moreover, terata supposed to be based on dominant gain-of-function mutants, such as type A peloria, are more frequent than terata supposed to be based on recessive loss-of-function mutants, such as type B peloria, possibly because a mutant phenotype develops only in homozygous plants.
Despite the explanatory power of the ‘orchid code’, some orchids have variant flowers that do not easily fit into the scheme of organ identity discussed above (Bateman and Rudall, 2006). Specifically, the lateral outer tepals (T2 and T3) are sometimes more similar to the lip (t3) than to the median outer tepal (T1); likewise, the median outer tepal (T1) may resemble closely the lateral inner tepals (t1 and t2). Flowers of that type exist at the species level (e.g. Psychopsis papilio; Fig. 1A) and at the level of mutants, such as Habenaria radiata ‘Hishou’ (Fig. 5A), a horticultural variant of H. radiata ‘Ginga’ (Fig. 1A). According to our ‘orchid code’ theory, in the case of H. radiata ‘Hishou’ both lateral outer tepals and the lip should express DEF-like genes of all four clades, whereas other tepals should express just the genes of clades 1, 2 and 3 (Fig. 5B). This would require both the ectopic expression of a clade 3 gene in outer tepals and of a clade 4 gene in lateral outer tepals (Fig. 5B).
Kim et al. (2007) showed that clade 3 DEF-like genes are ectopically expressed in the first floral whorl of ‘Hishou’, but more data are not presently available. In principle, parallel changes in both clade 3 and 4 genes could underlie the floral phenotype, but a single change in the common upstream control of both clade 3 and 4 genes appears more likely. A likely candidate would be any gene that translates the primary adaxial–abaxial (dorsal–ventral) positional cue into a gradient ‘felt’ by clade 3 and 4 genes. As discussed in the previous section, a feasible mechanism to form such a gradient would be differential expression of CYC-like TCP-type genes in dorsal (strong expression) and ventral regions (low expression) of the floral meristem (Fig. 5C). We hypothesize that deviant flowers such as those of Habenaria radiata ‘Hishou’ are ‘dorsalized’ by overexpressing the material basis of the dorsal–ventral positional cue; i.e. they may show stronger TCP-type gene expression, especially in the dorsal part of the floral meristem (Fig. 5C).
Combining the dorsiventral patterning system with the organ identity system (the ‘orchid code’) may lead to a refined system that distinguishes four different types of organs. Dorsally, the lip is defined by clade 4 gene expression, whereas the lateral outer tepals are defined by clade 1 and 2 expression and both types of dorsal organs by high TCP-type gene concentration. Ventrally, the lateral inner tepals are defined by clade 3 gene expression and the median outer tepal is defined by clade 1 and 2 gene expression, as well as by low TCP-type concentration affecting both types of ventral genes (Fig. 5C).
Throughout this discussion, we have dealt with the genetic basis of homeotic transitions affecting perianth organs. However, DEF-like genes, together with GLO-like genes, also determine stamen identity in angiosperms. Thus, one might expect that changes in the expression of DEF-like genes associated with development of peloric flowers may also affect male reproductive structures. Contrary to this general expectation, in the previously discussed examples of confirmed and likely homeotic transformations (Figs 3 and and5),5), male reproductive structures do not seem to be equally affected. For instance, in early developmental stages of type A peloria of Phalaenopsis equestris analysed with scanning microscopy by Tsai et al. (2004), lack of stamen and staminode development and fusion of adaxial carpels were observed. Dissection of adult flowers from other Phalaenopsis hybrids classified them as peloric or pseudopeloric, depending on the degree of identity between the modified lateral inner tepals and lip. This analysis showed that pseudopeloric flowers have normal stigmas and anthers, whereas fully peloric flowers lack stigmas and anther tissues and are thus sterile (Wallbrunn, 1987), suggesting that the underlying genetic causes may have different degrees of phenotypic penetrance. Although it is clear that in all peloric or pseudopeloric orchids the structure of the column remains zygomorphic, there are in each type of terata a few informative examples that suggest an association between developmental changes in the perianth and stamens. For instance, Phragmipedium lindenii (Fig. 3C), in which the lip is replaced by a lateral inner tepal (type B peloria), has a third fertile anther, in contrast to all other members of Cypripedioideae that have only two (Hurst, 1925). Furthermore, the presumed type C peloric Diplandorchis sinica has two fertile median stamens from the first and second floral whorl growing on the end of the stigma, directly opposite the dorsal tepal and lip (Chen, 1979). In contrast, Tangtsinia nanchuanica, another actinomorphic species discussed above, has five staminodial projections on the column that may represent three stamens of the inner whorl and two of the outer whorl (Chen, 1965). As the name of the potential type C pseudopeloric indicates, Prosthechea cochleata var. triandra has three anthers instead of the one normally found in this species (Sauleda et al., 1985). Nevertheless, this general relationship between peloria in the perianth and changes in stamen structure might not necessarily hold for Thelymitra, the most species-rich candidate case of type C peloria. It is not yet clear whether the distinctive and elaborated hood-like structure formed by the posterior anther lobe (the mitra) on the tip of the column represents modified stamens or staminodes. Besides this, the reproductive organs in Thelymitra do not have any apparent modification in number or configuration (Tonelli, 1999).
Further systematic description of different degrees of perianth modification and associated variation in reproductive structures is needed for more types of orchid peloria and pseudopeloria. Certainly, the fact that the cases previously discussed (Fig. 3) include cases of autogamy, cleistogamy and animal pollination suggests that peloric changes of reproductive organs do not necessarily result in sterility.
Considering the incredible diversity of orchid flowers, it is clear that nature provided us with a lot of material to test the ‘orchid code’ theory by determining organ-specificity of expression of different DEF-like and CYC-like genes via, for example, northern hybridization, in situ hybridization or quantitative RT-PCR.
As outlined above, homeotic mutations probably contributed to diversification of orchid flowers during evolution. It is assumed that morphological changes are based on changes in expression of DEF-like genes or genes controlling their expression. These increases or decreases in domains or intensities of expression of developmental control genes led to changes in organ identity, such as transformation of lateral inner tepals into lip-like organs or vice versa. However, even though homeotic mutants occur sporadically in many (if not all) populations of orchids, they tend to be ephemeral, presumably because they are often not pollinated (Bateman and Rudall, 2006). For example, even though occurring most frequently, type A peloria does not result in new species. Even the famous naturally occurring Cattleya intermedia var. aquinii is considered only a variety. Type B peloria is also frequent, but it too only rarely causes speciation, a remarkable exception being Phragmipedium lindenii (Hurst, 1925; Bateman and Rudall, 2006; Mondragón-Palomino and Theißen, 2008). Thus, generation of homeotic variants is not the most efficient mechanism for diversification of orchid flowers, and true changes in organ identity are probably rare events in orchid evolution. This is not to say, however, that they have not been crucial for the origin of some other plant groups (Ronse de Craene, 2003; Theißen, 2006).
Considering the floral diversity of orchids (incompletely captured in Fig. 1A), it is likely that differential elaboration of specific perianth organs without change of organ identity has been a much more important mechanism of floral diversification. For example, although still recognizable as a lip, the median inner tepal can develop in different ways. At maturity, it may be similar to the lateral inner tepals or outer tepals, but it can also appear dramatically different (Fig. 1A). The lip can be much larger or much smaller than other tepals, have a different colour or the same, have different patterns of coloration and ornamentation or may or may not bear nectar spurs. It may develop into an almost flat organ like the other tepals, but it may also form elaborate structures such as a tube or a ‘lady's-slipper’. In essence, it is obviously the potential of different kinds of petals, especially the lip, to develop independently from other types of organs that has made a major contribution to evolutionary diversification of orchid flowers. Assuming that our theory on development of organ identity in the orchid perianth (the ‘orchid code’) is correct, it appears not too far-fetched that it was the origin of four paralogous classes of DEF-like genes that enabled the orchid flower to ‘address’ the different types of tepals individually and hence allowed them to evolve independently. Although all tepals of flowers such as lilies and tulips (like the petals of eudicots) are probably under developmental control of one and the same set of floral homeotic genes, including DEF-like and GLO-like genes (Kanno et al., 2003), the outer tepals, lateral inner tepals and lip of orchids are controlled by different sets of genes with nested expression domains (Fig. 2). Thus, mutational changes in the orchid perianth can easily be restricted to the inner tepals by means of clade 3 gene mutation, or targets thereof, or to the lip by means of clade 4 gene mutation, or targets thereof. This may provide a proximate explanation for why the lip is the most diverse organ. A similar scenario is unlikely in petaloid monocots with identical tepals, such as lilies and lily-like species. Here, mutant changes in one tepal are likely accompanied by the same (pleiotropic) changes in all other tepals because of common developmental genetic control. Thus, evolution of the four classes of paralogous DEF-like genes ‘modularized’ the orchid perianth in such a way that the inner tepals could evolve semi-independently of the outer ones and the lip semi-independently of the lateral inner tepals. In this way, evolution of the paralogous DEF-like genes may have ‘deconstrained’ a lily-like floral perianth that was limited in its evolutionary potential by the pleiotropic interdependence of tepals. Once these constraints were reduced by modularization, the different classes of tepals thus generated were capable of evolving in a semi-independent way, and an almost explosive morphological diversification occurred, largely driven by adaptation of orchid flowers to specific classes of pollinators. Independent evolution was probably still restricted by the fact, however, that some developmental control genes (such as clade 1 and clade 2 DEF-like genes and GLO-like genes) are required for all petals to develop.
Origin and evolution of the four classes of paralogous DEF-like genes were probably not simple processes (Mondragón-Palomino and Theißen, 2008). We assume that the four DEF paralogues were subject to a complex series of sub-functionalization events that mainly affected cis-regulatory elements and led to changes in gene expression domains, as well as neo-functionalization events that could also involve the coding region. During these events, some upstream regulators and the target genes of DEF-like genes may have changed.
For instance, imagine an orchid with white tepals (e.g. some Phalaenopsis) or greenish tepals (e.g. Vanilla imperialis in Fig. 1A) in which only the lip is coloured purple by anthocyanins. Work on model plants such as Antirrhinum majus and Zea mays (Grotewold, 2006), and also on orchids (Chiou and Yeh, 2008), revealed that anthocyanin production depends on the expression of key enzymes including chalcone synthase, chalcone isomerase (CHI), flavanone 3-hydroxylase, dihydroflavonol 4-reductase (DFR) and others. Expression of the corresponding genes is under control of some transcription factors of the bHLH and MYB families; the latter includes OgMYB1 of the orchid hybrid Oncidium Gower Ramsey, which activates CHI and DFR (Chiou and Yeh, 2008). Anthocyanins can be produced in different floral organs and other parts of the plant, depending on expression of the appropriate enzymes, so in cases in which only the lip produces anthocyanins, clade 4 DEF-like genes are probably required and sufficient to activate the whole anthocyanin pathway, possibly by activating the regulatory bHLH and MYB genes as shown for other petal-specific genes in Antirrhinum majus (Perez-Rodriguez et al., 2005). In cases where all inner tepals, or even all tepals, produce anthocyanins (a frequent situation in orchids), clade 3 genes or even clade 1 or 2 genes might be sufficient to activate the anthocyanin pathway. This simple example would explain how evolutionary changes in the link between paralogous DEF-like organ identity genes and their target genes may have contributed to diversification of orchid flowers.
We have outlined the hypothesis that four paralogues of DEF-like class B floral organ identity genes have modularized the perianth of orchid flowers by a complex series of sub- and neo-functionalization events. These genes may have eliminated constraints, so that the different kinds of tepals could evolve individually and often in dramatically different ways. We thus argue that a developmental genetic predisposition unique to orchids may have played an important role in floral diversification in this family.
It is an important insight provided by evolutionary developmental biology (‘evo-devo’) that the internal organization of organisms, especially their developmental genetic systems, can influence the tempo and direction of evolutionary change (e.g. Brakefield, 2003). Generally, gene duplications increase the mutational robustness of organisms and thereby facilitate evolutionary innovations (Wagner, 2008). However, assuming that our hypothesis about the ‘orchid code’ is correct, we are not aware of many other systems that show such a close association between gene duplications and origin of evolutionary novelties directly involved in speciation. We are therefore convinced that the orchid flower is an extremely well-suited system for future evo-devo studies aimed at a better understanding of the relationship between developmental gene evolution and changes in morphology leading to speciation, especially once methods for validation of orchid gene function, such as transformation and virus-induced gene silencing, are optimized (Lu et al., 2007). Developing a diploid orchid with a rapid life cycle and a small genome size into a tractable model system is another urgent goal for the near future, which is already underway (Mark Chase, Royal Botanic Gardens, Kew, UK and B. Gravendeel, National Herbarium of the Netherlands, ‘pers. comm.’ November 2007). The great number of orchids with peloric or pseudopeloric flowers, however, can already be used to test the ‘orchid code’ hypothesis and the impact of this system on evolution of orchid flowers.
Many thanks to Mark Chase and Michael Fay (Royal Botanic Gardens, Kew) and the Linnean Society for inviting M.M.-P. to participate in the Symposium ‘Orchid evolutionary biology and conservation: From Linnaeus to the 21st century’, at which this paper was presented. We thank Mark Chase, Paula Rudall, Richard Bateman and Rainer Melzer for helpful comments on a previous version of this manuscript, Pia Nutt for producing an orchid line drawing and Thomas Wolf for assistance with photography. We thank the following colleagues for providing permission to use their valuable photographic material: A. Kocyan (Apostasia wallichii), A. Kanno and S. Y. Kim (Habenaria radiata in Figs Figs11 and and4),4), T. Kusibab (Phragmipedium caudatum and Phragmipedium lindenii in Figs Figs11 and and3),3), H. Schildauer and W. Schraut (Vanilla imperialis, Ophrys apifera, Aerangis fastuosa, Telipogon intis, Cattleya tenebrosa, Epidendrum pseudoepidendrum and Cattleya alvaroana in Figs Figs11 and and3),3), Hans Wapstra for the picture of type B peloric Calochilus robertsonii (from the collection of the late Les Rubach), James Wood (wild-type and type A peloric Calochilus robertsonii), Michael Pratt (Thelymitra formosa) and Richard Bateman and Robin Bush (peloric Platanthera chlorantha). This work was supported by the VolkswagenStiftung (I/81 901 to M.M.P. and G.T.).