JXB Advance Access originally published online on September 27, 2006
Journal of Experimental Botany 2006 57(13):3517-3530; doi:10.1093/jxb/erl126
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Evolution of Flowers and Inflorescences |
Flower-like terminal structures in racemose inflorescences: a tool in morphogenetic and evolutionary research
1Higher Plants Department, Biological Faculty, Moscow State University, 119992, Moscow, Russia
2Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3DS, UK
* To whom correspondence should be addressed. E-mail: remizowa{at}herba.msu.ru
Received 11 March 2006; Accepted 6 July 2006
 |
Abstract |
|---|
Top
Abstract
IntroductionTerminal flower-like structures (TFLS) occur in many angiosperms that possess indeterminate inflorescences such as spikes, racemes, or spadices. We describe and review TFLS in early-divergent angiosperms, especially the magnoliid order Piperales and the monocot order Alismatales, in which floral interpretation is controversial. Essentially similar TFLS occur in a wide range of taxa. Among magnoliids, they occur in some Piperales (Saururaceae and a few Piperaceae), but are absent from Chloranthaceae. Among monocots, they occur in some early-divergent families such as Acoraceae, Aponogetonaceae, Juncaginaceae, Potamogetonaceae, and Ruppiaceae. Similar TFLS with obscure organ identity are recorded in mutants of Arabidopsis. TFLS can often be interpreted as pseudanthia (close aggregations of reduced flowers), but in some cases the entire terminal pseudanthium is very similar to a true flower. In some cases, elaborated TFLS could therefore have given rise to what are normally termed ‘true’ (i.e. euanthial) flowers. Data presented here on terminal pseudanthia in Potamogeton and Ruppia support a pseudanthial evolutionary origin of reproductive units in the alismatid families Zannichelliaceae and Cymodoceaceae. Furthermore, in some alismatid species, either the entire inflorescence apex or an individual primordium at or near the inflorescence tip can be transformed into a filamentous or tubular (or intermediate) structure. A tubular structure enclosing stamens and carpels is described in Piper. This indicates that pseudanthium formation can provoke morphological novelties, perhaps due to new patterns of overlap between expression zones of regulatory genes and/or new spatial constraints.
Key words: Alismatales, development, flower, inflorescence, organ identity, Piperales, pseudanthium, regulatory genes
|
Introduction |
|---|
The spontaneous occurrence of atypical terminal flower-like structures (TFLS) in inflorescences that are otherwise indeterminate has long been considered important for understanding the morphogenesis and evolution of the inflorescence and flower. Interest in such structures has increased following discovery of the genes responsible for terminal flower formation in laboratory-induced mutants of model organisms such as Arabidopsis and Antirrhinum (Coen and Nugent, 1994; Bradley et al., 1996, 1997; Cremer et al., 2001; Ezhova and Penin, 2001; Penin et al., 2004; see also Ordidge et al., 2005). Terminal peloria are polysymmetric terminal flowers occurring in plants that normally possess only lateral monosymmetric flowers (reviewed by Rudall and Bateman, 2003). For example, they are common in Lamiales (the Antirrhinum clade), in which flowers are typically monosymmetric. However, they are absent from many other monosymmetric-flowered groups, such as Orchidaceae. Atypical terminal structures also occur in some plants that normally possess polysymmetric flowers; these can be interpreted either as peloria or sometimes as pseudanthia (aggregations of several flowers integrated together). The degree of distinction between terminal peloria and terminal pseudanthia is ambiguous (Rohweder and Treu-Koene, 1971; Rudall and Bateman, 2003). Both types of TFLS result in a determinate inflorescence in a species that typically possesses indeterminate inflorescences. Even indeterminate inflorescences eventually become exhausted, and sometimes a terminal protrusion of the transformed apex (apical residuum) is present.
Evaluation of comparative empirical observations on spontaneous variability in inflorescence tip structure in a phylogenetic context can enhance our understanding of floral evolution. This is a complementary approach to research in developmental genetics. In particular, examination of TFLS in early-divergent angiosperms, in which many morphogenetic patterns are relatively labile, will ultimately allow questions to be addressed about the origin of the flower. Reproductive units of several early-divergent species (including some basal monocots) have been interpreted as pseudanthia (Kunth, 1841; Wettstein, 1907; Uhl, 1947; Prokhanov, 1974; Burger, 1977; Hickey and Taylor, 1996; see also Rudall, 2003) although this has been challenged by others (Singh, 1965; Posluszny and Sattler 1973, 1974a, b; Endress, 1995; Takhtajan, 1997; Mavrodiev and Sokoloff, 1998). Some authors (Nozeran, 1955; Meeuse, 1975; Kusnetzova, 1986) have postulated that some terminal structures on inflorescences (e.g. in some Saururaceae: Anemopsis and Houttuynia) represent a ‘pre-floral state’, implying a primitive condition in angiosperms. However, it is currently widely believed that the presence of flowers rather than a ‘pre-floral state’ is plesiomorphic in angiosperms, based on data from extant species (Endress, 2001) or early Cretaceous fossil angiosperms (Friis et al., 2000, 2003). Here, terminal structures in racemose inflorescences (spikes, racemes, or spadices) of early-divergent angiosperms, especially the magnoliid order Piperales and the early-divergent monocot order Alismatales, are reviewed and compared. Although not closely related, these both represent early-divergent angiosperm lineages. Both orders have long been the subject of controversy regarding interpretation of their reproductive structures. The present report contributes to this ongoing discussion and examines whether studies of terminal pseudanthia and associated phenomena (including anomalous filamentous and tubular structures) can help in understanding evolutionary patterns and processes.
Structures that are usually termed pseudanthia fall into two distinct types that are not necessarily evolutionarily related to each other. In the first type (reviewed by Claßen Bockhoff, 1990, 1991), the pseudanthium contains a number of closely associated and sometimes even fused flowers, but individual flowers are recognizable units despite their reduced nature and/or connation with other flowers. Capitula of Asteraceae can be referred to this type. In the second group, loss of flower individuality occurs. In extreme cases, certain organs (e.g. stamens, carpels) cannot be assigned to individual flowers. All the terminal pseudanthia discussed herein are of the second type.
|
Materials and methods |
|---|
The following species were examined: Piper arboreum Aubl. (Piperaceae), cultivated in the Princess of Wales Conservatory at the Royal Botanic Gardens, Kew and herbarium material (K); Potamogeton natans L. and P. lucens L. (Potamogetonaceae), collected in 2003–2005 by DD Sokoloff, MV Remizowa, and students of the Biological Faculty at Zvenigorod Biological Station of Moscow University, Moscow province, Russia; Ruppia maritima L. (Ruppiaceae), Triglochin maritima L., and T. palustris L. (Juncaginaceae), collected in July 2003 and July 2004 by DD Sokoloff and MV Remizowa, at the White Sea Biological Station of Moscow State University in N Karelia, Russia; Triglochin bulbosa L., collected in March 2004 by DD Sokoloff and AP Seregin, Akrotiri, Cyprus; Tofieldia and other genera of Tofieldiaceae, from all available herbarium material at Kew, Komarov Botanical Institute, St Petersburg, and Moscow State University, Moscow, plus living material listed in Remizowa et al. (2006).
Living material was fixed in either FAA or 70% ethanol. For scanning electron microscope (SEM) examination, material was dissected in 70% ethanol and dehydrated in an ethanol series. Dehydrated material was then critical-point dried, mounted onto SEM stubs using double-sided adhesive tape, coated with platinum or gold using a sputter coater, and examined using a Hitachi cold field emission SEM S-4700-II (2 kV) and a Hitachi SEM S-405A (15 kV).
|
Taxonomic distribution of terminal structures in early-divergent angiosperms |
|---|
Among magnoliids, spikes and spadices are typical for both perianthless Piperales (Saururaceae and Piperaceae) and Chloranthaceae (Chloranthales). However, TFLS are absent from Chloranthaceae (Endress, 1987; Kong et al., 2002), although Leroy (1983) interpreted the entire spike of male flowers in Hedyosmum as a flower-like structure. Anemopsis and Houttuynia (Saururaceae) possess showy bracts below the spicate inflorescence, resulting in a flower-like appearance; in addition, there is frequently a peculiar terminal structure. The TFLS of Houttuynia can include a terminal gynoecium surrounded by stamens and reduced male flowers; in other cases, the gynoecium is absent (Nozeran, 1955; Rohweder and Treu-Koene, 1971; Tucker, 1981; Liang and Tucker, 1990). Apical residua are present rather than TFLS in some Saururaceae and most Piperaceae (Tucker, 1979, 1980, 1982; Liang and Tucker, 1989). Thus far, TFLS are known in only two species of Piperaceae. In Peperomia fraseri, one or two carpels often develop at the tip of the spadix (Remizowa et al., 2005). Among plants of Piper arboreum in the Kew living collections, we found one individual that produced TFLS in all inflorescences (Fig. 1), whereas other individuals produce only apical residua.
|
Among monocots, Buzgo and Endres (2000) and Buzgo et al. (2004) described TFLS in Acorus, and noted a similarity with Anemopsis and Houttuynia. The taxonomically isolated genus Acorus is putatively sister to all other monocots, with Alismatales (Araceae, Tofieldiaceae, plus 12 families of aquatic alismatids) as sister to all monocots except Acorus. Terminal structures occur in many members of Alismatales. Either an apical residuum or a terminal flower is present in the racemose inflorescences of Tofieldiaceae (unpublished data). TFLS are mostly absent from Araceae (Buzgo, 2001), but present in Calla palustris as a cluster of stamens that cannot be assigned to individual flowers (Lehmann and Sattler, 1992). However, it is the aquatic alismatids (formerly Helobieae) that display the widest range of TFLS. Terminal flowers (often peloric) or pairs of united flowers are present in spadices of at least some Aponogetonaceae. In Juncaginaceae, terminal or subterminal male flowers similar to the lateral male flowers have been described for Lilaea (Posluszny et al., 1986). Bisexual terminal flowers are common in Triglochin (Buzgo et al., 2006). Other types of terminal structure also occur in T. maritima (Fig. 2; see also Charlton, 1980). TFLS are extremely diverse in Potamogeton (Figs 3, 4; see also Charlton and Posluszny, 1991), although they do not occur in every inflorescence. Ruppia (Ruppiaceae) possesses two-flowered spikes, but the flowers are sometimes united (Fig. 5).
|
|
|
|
|
Morphological nature of terminal structures |
|---|
Three basic types of terminal structure are recognized, although intermediates exist: (i) a true terminal flower; (ii) a morphologically lateral flower shifted into a pseudoterminal position; and (iii) a pseudanthium formed by united uppermost lateral flowers. (A similar phenomenon, of atypical ‘double’ flowers occurring in the lower or middle part of a racemose inflorescence, has also been recorded. However, united flowers in a terminal position are of particular interest because they could be mistaken for or give rise to true terminal flowers.)
There are several criteria that can be used to distinguish between these three types. (i) A true terminal flower opens before other flowers in an inflorescence, at least before the upper lateral flowers, which may be reduced (Goebel, 1931; Weberling, 1989). This criterion is problematic for inflorescences in which flowers open simultaneously, and inapplicable in cases where the inflorescence is reduced to a single flower. (ii) The internode of the inflorescence axis below a terminal flower (termed the ‘final internode’ by Weberling, 1989) is often different in length from other internodes (either longer or shorter). This criterion in particular suggests that the terminal structures of Triglochin (Buzgo et al., 2006) and P. arboreum are true flowers. (iii) The presence of more than one bract surrounding a terminal structure indicates a pseudanthium, as in some material of Potamogeton (Fig. 3C, K–M, O) and Piper (Fig. 1, in which three bract homologues are united into a tubular structure). In contrast, the presence of only one bract indicates a pseudoterminal flower, as in some inflorescences of Potamogeton lucens. However, a complicating factor is that many angiosperms possess sterile bracts below a true terminal flower; these are possibly derived from suppressed uppermost lateral flowers (Weberling, 1989; Kusnetzova, 1995). Furthermore, some inflorescences lack bracts entirely (as in Triglochin, Ruppia). [When lateral flowers possess bracteole(s), or prophyll(s)—which is not the case in perianthless Piperales and tepaloid Alismatales—these structures are absent from the terminal flower by definition, for example, Goebel (1931). The presence of a terminal flower in the dense bractless capitula of Asteraceae is discussed by Teeri et al. (2006).] (iv) Based on data both from mutants of model organisms, and a review of various angiosperms, Penin et al. (2004) concluded that perianth-bearing true terminal flowers cannot occur in abracteate inflorescences. If so, perianth-bearing terminal structures in abracteate inflorescences must be either pseudoterminal flowers or pseudanthia. This is the case in Triglochin (Fig. 2A–E, J–K). (v) Terminal flowers often differ in merosity or symmetry from lateral ones (Kusnetzova, 1985; Rudall and Bateman, 2003; Penin et al., 2004). However, the same should be expected from terminal pseudanthia. (vi) Fedorova (2004) found that in compact inflorescences with unisexual or partly unisexual flowers, a terminal flower (if present) is either bisexual or female. If the terminal flower is absent, then the uppermost lateral flowers are either male or bisexual. This suggests that the uppermost flower of Lilaea (Posluszny et al., 1986) (which is male) is pseudoterminal rather than terminal. (vii) Preliminary (unpublished) observations on vascular anatomy of terminal structures in Potamogeton natans indicate that vasculature could represent an additional criterion, since the vascular supply of TFLS is different from what could be expected for just three united uppermost flowers. Rudall (2003) also discussed vasculature in relation to pseudanthia in Triuridaceae, in which vascular traces depart from the axial vascular cylinder in a manner similar to a condensed branching axis with very short internodes. (viii) The presence of an apical residuum at the top of the inflorescence proves absence of a terminal flower. However, an apical residuum is sometimes entirely absent even if all flowers are lateral. Also, an apical residuum may be present in true flowers (e.g. in Nymphaea).
Most data suggest a pseudanthial nature for the terminal flower-like structures of Potamogeton and Ruppia (Figs 3, 5). However, interpretation of these structures in Peperomia fraseri (Remizowa et al., 2005), Piper arboreum (Fig. 1), and Triglochin (Fig. 2) is more problematic as the different criteria provide conflicting evidence. It is possible that there is a series of intermediate forms in nature (at least in certain plant groups) between a true terminal flower and a terminal pseudanthium (see also Charlton, 1980). In evolutionary terms, this may mean that an elaborated terminal pseudanthium can give rise to what is normally termed a true flower. A further possibility—for example, in polymerous TFLS of Triglochin—is loss of flower identity. Since possession of ‘true’ flowers probably represents the primitive condition in angiosperms, loss of flower identity is most probably a derived condition that occurred many times during the course of evolution of terminal structures in various species. Similar phenomena have been described in mutants of different model organisms.
All the terminal pseudanthia discussed here represent a phenomenon that is similar to fasciation, with the important distinction that fasciation is normally believed to be derived from a discrete structure (e.g. flower or vegetative shoot) that has incompletely multiplied. For example, Worsdell (1915) defined fasciation as ‘multiple forking of an axis’. Thus, the pseudanthia here represent the opposite phenomenon, assuming their homology with several independent flowers. However, a broader definition of fasciation which includes both cases is also possible, although perhaps less useful. For example, Masters (1869) cited cases in which buds are pressed together and become fused during development.
|
Unusual floral organs |
|---|
An apparently novel observation from this investigation is that both tubular and filamentous structures sometimes occur at the inflorescence apex in wild-source material in species of at least two alismatid monocot families, Potamogetonaceae and Juncaginaceae (Figs 2E–G, 3C–K). There are also intermediate forms between the two structures, indicating that they represent different manifestations of the same morphogenetic phenomenon. The tubular structures in Potamogeton are essentially similar in morphology and position to the club-shaped organs described in Acorus (Buzgo and Endress, 2000). Filamentous structures also occur in wild-source material of another alismatid monocot, Scheuchzeria (Scheuchzeriaceae: M Remizowa and D Sokoloff, unpublished data).
Unusual filamentous structures are of special interest because they have been described in inflorescences of some Arabidopsis mutants, especially in fil (Sawa et al., 1999) and ufo (Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). Although filamentous structures are more common, tubular structures that are apparently homologous to filamentous ones have also been described for Arabidopsis mutants such as lfy-6 fl54. Levin and Meyerowitz (1995) searched for precedents for the natural formation of filamentous structures in plants other than Arabidopsis. They listed such examples as filiform leaves of aquatic plants as well as spines and tendrils.
Filamentous and tubular structures of both Arabidopsis mutants and alismatid monocots often develop in the position of flowers. Since flower-subtending bracts are absent in wild-type Arabidopsis, it is problematic to decide whether filamentous structures represent reduced, modified flowers or partially de-suppressed bracts (with flowers completely suppressed). Levin and Meyerowitz (1995) observed that a filamentous structure sometimes subtends the lowermost flower(s) in inflorescences of Arabidopsis ufo mutants, and concluded a likely bract homology for filamentous structures, although a flower homology could not be excluded. Sawa et al. (1999) suggested that filamentous structures of fil mutants represent underdeveloped flowers that form a pedicel but lack a receptacle and floral organs. Watanabe and Okada (2003) showed that FIL plays a role in specifying the abaxial side tissues in the development of lateral organs such as cotyledons, leaves, young flower buds, and flower organs. In Potamogeton, apparently normal flower-subtending bracts often subtend tubular and filamentous structures; thus, they are apparently of floral nature. The developmental mechanism responsible for unusual structure formation may not be the same in Arabidopsis mutants and Alismatales.
Tubular structures described here in Piper arboreum, a perianthless member of Piperales, differ significantly from the tubular structures of alismatid monocots. In some inflorescences, the final-whorl bracts are almost free (Fig. 1G), indicating that the tubular structure in P. arboreum is probably homologous with three congenitally united flower-subtending bracts of the final whorl. Furthermore, it often contains reproductive organs (Fig. 1H–K); these are often sterile, although stamens are occasionally fertile. Interestingly, some terminal structures of Piper resemble flowers of Aristolochiaceae, a perianth-bearing family of Piperales in which flowers are in a terminal position. The three-lobed tubular structure of Piper resembles a perianth of Aristolochiaceae that is often composed of three united members. Reproductive organs inside the tubular structure often show a trimerous pattern, which is also the case in Aristolochiaceae.
|
Use of terminal structures in understanding inflorescence and flower morphogenesis |
|---|
We propose some preliminary hypotheses regarding regulation of development in Potamogeton natans. Elongation of the inflorescence apex and initiation of lateral flowers are apparently controlled by two independent developmental programmes. Elongation of the inflorescence apex commences first, but proceeds more slowly than the formation of lateral flower primordia, which occurs in an acropetal pattern. The flower primordia are usually initiated in regular alternating whorls (of two, three, or four) up to the tip of the developing inflorescence (see also Posluszny and Charlton, 1983). Sometimes the flower primordia of the final whorl have (by chance) just sufficient space for their formation at the inflorescence apex, which is then completely exhausted. However, this is often not the case, because the inflorescence apex may be longer or shorter than is necessary for insertion of the final-whorl primordia. Another factor is the change in inflorescence shape from cylindrical throughout its length to hemispherical at the tip; this may also cause problems with proper patterning of the final-whorl primordia. If considerable space remains on the apex after formation of the final whorl of free flower primordia, a compound floral primordium is initiated at the inflorescence apex (Fig. 3B, C). Depending on the space available, this compound terminal primordium gives rise to what could easily be interpreted as a pseudanthium (Fig. 3G) or an almost completely flower-like structure (Fig. 3M). In contrast, when the remaining space at the apex is smaller than the size of a typical flower primordium (or the shape of the space is atypical), unusual filamentous or tubular structures are initiated at the inflorescence apex. Thus, it seems likely that, in Potamogeton, abnormalities are due to spatial problems. Despite the fact that there is insufficient space to initiate a flower of typical structure, expression of genes responsible for flower initiation still occurs. In this case, unusual patterns of overlap between expression zones of regulatory genes may result in the formation of unusual organs (e.g. tubular structures). Studies of terminal structures in Potamogeton may therefore help in understanding the ‘mass effect’ in flower formation (Tucker, 1979). There is apparently a minimum size of flower primordium that can produce flower-bearing structures of ‘normal’ organ identity (tepals, stamens, carpels). If the floral primordium size is smaller than this threshold, flower morphogenesis is shifted entirely, and a tubular or filamentous structure is formed. These unusual structures do not merely represent suppressed flowers, but result from activity of an entirely different developmental programme. In mature inflorescences, these structures are often much longer than normal flowers.
We propose a similar pattern for the terminal structures of Peperomia fraseri, a species of perianthless Piperales (Remizowa et al., 2005). However, among Piperales, inflorescence morphogenesis in Piper is quite different from that of Potamogeton. In wild-type P. arboreum, as well as in many other Piperaceae (Tucker, 1980, 1982) and other Piperales such as Saururus (Tucker, 1979), an apical residuum occurs at the inflorescence tip. A critical point is that the overall size of this residuum is larger than the space necessary for flower initiation. Thus, cessation of growth is an active process, and genes responsible for flower initiation are no longer active here at the appropriate time. This could represent an adaptation to prevent formation of organs with unusual identity and to unify floral structure in the inflorescence. In an anomalous plant of P. arboreum that bears terminal structures, pattern formation of the terminal structures probably occurs prior to patterning of the upper lateral floral primordia. There is often sufficient space on the inflorescence meristem to initiate more flowers immediately below the terminal structure. The fact that these flowers are not initiated must be due to active inhibition, as in apical residua of most Piperaceae and Saururus. Interestingly, hairs are formed instead of flowers in both P. arboreum (Fig. 1A, B, D, F, L) and Saururus (Tucker, 1979).
In summary, it is important to emphasize the differences between elaborated apical residua (as in some perianthless Piperales) and terminal filamentous structures (in some inflorescences of Potamogeton and Triglochin). The former reflect canalization and stability in inflorescence morphogenesis, whereas the latter result from unstable morphogenesis of the inflorescence tip. Apical residua have been described in a wide range of angiosperms (Troll, 1964). These observations may need re-investigation in the light of this new evidence.
|
Pseudanthial origin of reproductive units in Zannichelliaceae and Cymodoceaceae |
|---|
Relationships within Alismatales have long been problematic. The submerged aquatic families Zannichelliaceae and Cymodoceaceae have traditionally been considered closely related (Dahlgren et al., 1985; Takhtajan, 1997) or even treated as a single family (Eckardt, 1964), but molecular phylogenetic analyses indicate that Zannichelliaceae are closely related to (and could be synonymized with) Potamogetonaceae s.str. (Les et al., 1997; APG, 2003; Chen et al., 2004) whereas Cymodoceaceae are allied with Ruppiaceae (Les et al., 1997). Apart from features obviously connected with submerged growth and underwater pollination, Zannichelliaceae and Cymodoceaceae share a cymose pattern of flower arrangement, although inflorescences of many Cymodoceaceae are reduced and sometimes single flowered (Irmisch, 1851; Eckardt, 1964; Kay, 1971; Posluszny and Sattler, 1976; Posluszny and Tomlinson, 1977; Tomlinson and Posluszny, 1978; Dahlgren et al., 1985; Takhtajan, 1997; Mavrodiev and Sokoloff, 1998). In contrast, Potamogetonaceae and Ruppiaceae possess flowers arranged in open spikes.
Mavrodiev and Sokoloff (1998) drew attention to the essential similarity of general inflorescence architecture between Zannichelliaceae, Ruppiaceae, and Potamogetonaceae (see Supplementary Figs 1 and 2 available at JXB online), but with the important difference that the basic inflorescence units of Zannichelliaceae are flowers and those of Ruppiaceae and Potamogetonaceae are open spikes of flowers. They hypothesized that entire spikes of Potamogeton and Ruppia may be homologous to individual flowers of Zannichelliaceae. They emphasized that one of two observations could prove this hypothesis: either flowers of Zannichelliaceae are morphologically lateral but shifted to a terminal position, or spikes of Potamogeton sometimes bear a terminal flower. However, at that time, they did not observe any of these phenomena. Now we have found considerable diversity of flower-like terminal structures in Potamogeton, which should be termed terminal pseudanthia rather than true terminal flowers. We postulate that reproductive units of Zannichelliaceae could be derived from such terminal pseudanthia. One possible objection is that terminal pseudanthia (and terminal peloria) are mostly present at the ends of long, many-flowered racemes or spikes, whereas the present hypothesis implies that the entire spike in Zannichelliaceae was taken up by terminal pseudanthium formation. However, we have also observed this phenomenon in two-flowered spikes of Ruppia (Fig. 5), which demonstrates the possibility of use of the entire spike for formation of a terminal flower-like structure.
With a few exceptions (e.g. Syringodium filiforme), inflorescences of Cymodoceaceae (see Supplementary Fig. 3 available at JXB online) are mostly reduced and few-flowered, but the basic pattern is cymose (Tomlinson and Posluszny, 1978). The inflorescence of S. filiforme is essentially similar to that of Ruppia, except that the basic unit of Syringodium is commonly described as a flower whereas Ruppia possesses two-flowered spikes. Since amalgamation between flowers has been observed in Ruppia, we suggest a pseudanthial origin of reproductive structures in Cymodoceaceae.
Authors who have interpreted the female units of Zannichelliaceae and Cymodoceaceae as pseudanthial (Prillieux, 1864; Uhl, 1947; Ducker and Knox, 1976) have regarded each carpel as a naked female flower. However, although we agree with a pseudanthial origin of the terminal structures of these plants, the end-product of amalgamation of two or three flowers can be quite an integrated structure, and it is futile to search for parts that precisely correspond to those of the initial flowers. This is especially the case for the gynoecium. For example, if the terminal structure of Potamogeton illustrated in Fig. 3M is a pseudanthium, each of the three carpels belongs to the entire pseudanthium, not to particular individual flowers. Flowers in two-flowered inflorescences of Ruppia (Kaul, 1993) and Potamogeton densus (Posluszny and Sattler, 1973) are in the median plane of the two subopposite leaves situated below the spike (see Supplementary Fig. 4 available at JXB online), thus continuing the distichy of the vegetative part of the shoot into the inflorescence. In Zannichelliaceae (Posluszny and Sattler, 1976) and Cymodoceaceae (Tomlinson and Posluszny, 1978), when female units are bicarpellate, carpels are in the transverse plane of distichy of the vegetative part of the shoot (see Supplementary Fig. 4 available at JXB online). These carpels cannot be assigned to individual flowers.
There are interesting morphological parallels between some terminal structures of Potamogeton and Ruppia and reproductive units of Cymodoceaceae and Zannichelliaceae. In all Cymodoceaceae and Zannichelliaceae, the male unit contains a single anther that frequently possesses more than four microsporangia and has hence been interpreted as a fusion product of different stamens. Terminal structures of Ruppia and Potamogeton also show a tendency for stamen connation. The structure of P. lucens illustrated in Fig. 3R (with four united thecae) resembles male flowers of some Zannichelliaceae. Interestingly, it possesses an apical appendage, a feature common in both Cymodoceaceae and Zannichelliaceae but not present in normal flowers of P. lucens. This probably illustrates how morphological novelty can arise during pseudanthium formation. In contrast to stamens, united carpels have not been observed in terminal structures of Potamogeton and Ruppia. This is also the case in reproductive structures of Zannichelliaceae and Cymodoceaceae. Thus, there is apparently a constraint against carpel connation in all four families, as they are among very few monocots bearing free carpels (Igersheim et al., 2001). In Zannichellia, there is a tubular envelope surrounding the female unit. Despite extensive discussion in the literature, its origin remains unclear. However, it is morphologically comparable with the terminal tubular structures that we observed in Piper arboreum (Fig. 1H–K); the two species are not closely related, but the example of Piper demonstrates how such a morphological novelty can arise.
|
Terminalization versus truncation in evolution of flower arrangement in angiosperms |
|---|
Evolutionary relationships between monotelic and polytelic types of flower arrangement (i.e. with and without a terminal flower; Weberling, 1989) or, more precisely, between types of flower arrangement with different axiality, have long been debated. Axiality can be defined as minimum branch order (within an architectural unit of branching system) that can be terminated by a flower (reviewed in Notov and Kusnetzova, 2004). For example, plants with terminal monotelic synflorescences are monoaxial (e.g. Ranunculus and Zannichelliaceae), plants with terminal open racemes, spikes, or umbels are diaxial (e.g. Arabidopsis, Antirrhinum, Ruppia, and Potamogeton), and plants with lateral spikes or racemes are triaxial (e.g. Pisum and many other legumes).
Many authors have emphasized the importance of truncation (Sell, 1969; Kusnetzova, 1988, 1998; Weberling, 1989), or evolutionary loss of a terminal flower. This transformation results in an evolutionary transition from monotelic to polytelic synflorescences, or (in more general terms) in increased axiality. When putatively primitive members of a natural group bear clearly polytelic synflorescences while putatively advanced members bear what appears to be a terminal flower, closer inspection usually shows that it is either not terminal (i.e. a lateral flower shifted into a pseudoterminal position) or a pseudanthium rather than a flower. In Zannichelliaceae and Cymodoceaceae, we have demonstrated the pseudanthial origin of terminal reproductive units. This raises the question of whether an elaborated terminal pseudanthium could give rise to a ‘true’ flower. This is probably so in cases such as Cymodoceaceae and Zannichelliaceae, in which the terminal structure is derived from two or more united flowers but cannot be subdivided into parts corresponding to putative individual flowers. If it is accepted that the reproductive units of these two families are flowers derived from pseudanthia, then an evolutionary change that is the opposite to truncation could be proposed. This phenomenon, which we term ‘terminalization’, can result in an evolutionary shift from polytelic to monetelic synflorescences and in decreased axiality.
Terminalization does not necessarily occur via formation of terminal pseudanthia, but could arise by saltation. As mentioned above, several well-known mutants differ from the wild type in the consistent presence of terminal flowers, although the question remains of whether such an apparently simple mechanism could occur during the course of evolution (see also Singer et al., 1999). An example of terminalization that does not necessarily involve pseudanthium formation was described by Endress (1970, 1978) in Hamamelidaceae. In perianth-bearing Hamamelidaceae (e.g. Parrotia), flowers are organized into spikes that are themselves arranged in panicles. Spikes lack either a terminal flower or a TFLS. In perianthless Hamamelidaceae (e.g. Distylium), solitary terminal flowers occur in paniculate inflorescences instead of spikes. There is robust evidence that the ramification pattern in perianthless Hamamelidaceae represents a derived condition (Endress, 1970, 1978).
Similarities between the conditions in Alismatales and Hamamelidaceae include the following (for Hamamelidaceae, see Endress, 1970, 1978; Li et al., 1999). (i) Terminalization is the most likely evolutionary explanation of recent diversity in flower arrangement. (ii) In both groups terminalization is associated with loss of a perianth. (iii) Terminal flowers in derived taxa differ from lateral flowers in primitive taxa in the structure of the androecium and gynoecium (an unusual Bauplan of terminal structures is also typical for mutants of model organisms such as Arabidopsis). (iv) In both Alismatales and Hamamelidaceae, species with monotelic synflorescences have previously often been considered as forming natural groups, but in both cases molecular analyses have indicated polyphyly, which suggests non-unique terminalization events.
|
Conclusions and perspective |
|---|
The present study highlights the considerable difficulties that exist in distinguishing between some elaborated terminal pseudanthia (i.e. closely associated and integrated uppermost lateral flowers) and terminal flowers. These difficulties are linked with a more general question on the homologies between lateral flowers and (true) terminal flowers, including both serial and historical homologies (Rutishauser and Moline, 2005). In turn, this homology issue is at least partially comparable with Meeuse's (1975) ideas on the differences between (holo)anthocormoids (essentially terminal and two-axial) and anthoids (essentially lateral and uniaxial) (see also Krassilov, 1989).
Application of traditional homology criteria could help to compare terminal and lateral flowers, though even these can be problematic. Remane's (1956) three basic homology criteria (reviewed by Mamkaev, 2001; Rutishauser and Moline, 2005) are: (i) position; (ii) special quality of structures (including genetic control); and (iii) linkage by intermediate forms. The position criterion (i) is usually considered the most important, but it does not work when comparing terminal and lateral flowers.
Criterion (ii) is also unlikely to help in establishing flower identity, since male, female, perianthless, and even sterile flowers exist among various flowering plants and are commonly accepted as flowers. It is relatively straightforward to list features that occur only in pseudanthia (e.g. presence of more than one gynoecium, carpels facing each other via their dorsal sides) but it is almost impossible to indicate any feature that exclusively characterizes an individual flower. Furthermore, studies of genes responsible for regulation of flower development can be of limited help in establishing morphological homologies. For example, flowers of Althenia and Lepilaena (both Zannichelliaceae) possess a tripartite ‘perianth’ around the male and female reproductive structures. Expression of A-class genes in these structures would test their possible bract homology. However, evidence of tepal identity does not refute either the euanthial or pseudanthial interpretations. In addition, expression of some floral organ identity genes can occur in non-floral regions, even in leaves (see Buzgo et al., 2006, for expression of AP3 in Triglochin). There are several examples in which orthologous genes show a similar expression pattern and key morphogenetic function in structures that are clearly non-homologous. Converse examples show that morphogenesis of homologous structures can be controlled by different genes in different taxa (Nielsen and Martinez, 2003; Theißen and Becker, 2004; Rutishauser and Moline, 2005; see also Rautian, 1988; Wagner 1989; Baum and Donoghue, 2002). Finally, an elaborated end-product of amalgamation between individual flowers could be as similar to a single true flower at the molecular level as it is sometimes similar morphologically.
The occurrence of intermediate forms (criterion iii) is probably the only criterion that can give substantive grounds for homologizing terminal and lateral flowers, although even this does not guarantee a correct conclusion. Series of intermediate structures could be due to amalgamation of developmental pathways leading to ‘developmental mosaics’ between structures that are normally assumed to have different identities (Rutishauser and Moline, 2005; see also Lodkina, 1983; Sattler, 1988; Baum and Donoghue, 2002). In particular, structures with long-debated pseudanthial or euanthial interpretation in various angiosperm groups (e.g. Piperales and Alismatales) could have evolved as a result of overlap between the developmental programmes of a typical flower and inflorescence (Posluszny and Charlton, 1993). Some terminal structures described here should also be described in terms of loss of flower identity (e.g. TFLS of Triglochin with multiple carpels).
Within Alismatales, future research on flower and inflorescence evolution would benefit from a more explicitly phylogenetic approach. An important test of our hypothesis will lie in more extensive comparative and developmental studies of a wider range of species. More broadly, studies of terminal structures offer an opportunity to understand the possible morphological consequences of loss of flower identity. An important conclusion from the present investigation is that pseudanthium formation following loss of floral identity can provoke morphological novelties (including unusual tubular and filamentous structures, and unusual patterns of connation). These could be the result of either new patterns of overlap between expression zones of regulatory genes, or new spatial constraints in developing reproductive structures.
|
Supplementary data |
|---|
Supplementary data can be found at JXB online.
|
Acknowledgements |
|---|
We thank Peter Endress and Rolf Rutishauser for highly stimulating reviews, Richard Bateman, Matyas Buzgo, Vladimir V Choob, Tatiana A Fedorova, Maria D Logacheva, Evgeny V Mavrodiev, Alexey A Penin, and Alexander C Timonin for helpful discussion, and Alexei P Seregin for help during collecting Triglochin bulbosa. DS and MR gratefully acknowledge the Royal Society and Systematic Research Fund for support for their visits to Kew in 2004, and grants of the President of Russia (MD-1200.2005.4) and RFBR (06-04-48113) to support their work in Russia.
|
Abbreviations |
|---|
SEM, scanning electron microscope; TFLS, terminal flower-like structure.
|
References |
|---|
APG II. (2003) An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants. Botanical Journal of the Linnean Society 141:399–436.[CrossRef][ISI]
Baum DA and Donoghue MJ. (2002) Transference of function, heterotopy and the evolution of plant development. In Cronk QCB, Bateman RM, Hawkins JA (Eds.). Developmental genetics and plant evolution(Taylor & Francis, London) pp. 52–69.
Bradley D, Carpenter R, Copsey L, Vincent C, Rothstein S, Coen E. (1996) Control of inflorescence architecture in Antirrhinum.. Nature 379:791–797.[CrossRef][Medline]
Bradley D, Ratcliffe O, Vincent C, Carpenter R, Coen E. (1997) Inflorescence commitment and architecture in Arabidopsis.. Science 275:80–83.
Burger WC. (1977) The Piperales and the monocots: alternative hypotheses for the origin of monocotyledonous flowers. Botanical Review 43:345–393.
Buzgo M. (2001) Flower structure and development of Araceae compared with alismatids and Acoraceae. Botanical Journal of the Linnean Society 136:393–425.
Buzgo M and Endress PK. (2000) Floral structure and development of Acoraceae and its systematic relationships with basal angiosperms. International Journal of Plant Sciences 161:23–41.[CrossRef][ISI][Medline]
Buzgo M, Soltis DE, Soltis PS, Ma H. (2004) Towards a comprehensive integration of morphological and genetic studies of floral development. Trends in Plant Science 9:164–173.[CrossRef][ISI][Medline]
Buzgo M, Soltis DE, Soltis PS, Ma H, Hauser BA, Leebens-Mack J, Johansen B. (2006) Perianth development in the basal monocot Triglochin maritima. Aliso (in press).
Charlton WA. (1980) Features of inflorescences of Triglochin maritima. Canadian Journal of Botany 59:2108–2115.
Charlton WA and Posluszny U. (1991) Meristic variation in Potamogeton flowers. Botanical Journal of the Linnean Society 106:265–293.
Chen J-M, Chen D, Gituru WR, Wang Q-F, Guo Y-H. (2004) Evolution of apocarpy in Alismatidae using phylogenetic evidence from chloroplast rbcL gene sequence data. Botanical Bulletin of Academia Sinica 45:33–40.
Claßen Bockhoff R. (1990) Pattern analysis in pseudanthia. Plant Systematics and Evolution 171:57–88.[CrossRef][ISI]
Claßen Bockhoff R. (1991) Anthodien, Pseudanthien und Infloreszenzblumen. Beiträge zur Biologie der Pflanzen 66:221–240.
Coen ES and Nugent JM. (1994) Evolution of flowers and inflorescences. Development 120:107–116.
Cremer F, Lonnig WE, Saedler H, Huijser P. (2001) The delayed terminal flower phenotype is caused by a conditional mutation in the Centroradialis gene of snapdragon. Plant Physiology 126:1031–1041.
Dahlgren RM, Clifford HT, Yeo PF. (1985) The families of the monocotyledons(Springer-Verlag, Berlin).
Ducker SC and Knox RB. (1976) Submarine pollination in seagrasses. Nature 263:705–706.
Eckardt T. (1964) Reihe Helobiae. In Melchior H and Herausg A (Eds.). Engler's Syllabus der Pflanzenfamilien(Borntraeger, Berlin) pp. 499–511.
Endress PK. (1970) Die Inflorescencen der apetalen Hamamelidaceen, ihre grundsätzliche morphologische und systematische Bedeutung. Botanische Jahrbücher für Systematik 90:1–54.
Endress PK. (1978) Blütenontogenese, Blütenabgrenzung und systematische Stellung der perianthlosen Hamamelidoideae. Botanische Jahrbücher für Systematik 100:249–317.
Endress PK. (1987) The Chloranthaceae: reproductive structures and phylogenetic position. Botanische Jahrbücher für Systematik 109:153–226.
Endress PK. (1995) Major traits of monocot flowers. In Rudall PJ, Cribb PJ, Cutler DF, Humphries CJ (Eds.). Monocotyledons: systematics and evolution(Royal Botanic Gardens, Kew, London) pp. 43–79.
Endress PK. (2001) The flowers in extant basal angiosperms and inferences on ancestral flowers. International Journal of Plant Sciences 162:1111–1140.[CrossRef][ISI]
Ezhova TA and Penin AA. (2001) A new BRACTEA (BRA) gene controlling the formation of an indeterminate bractless inflorescence in Arabidopsis thaliana.. Russian Journal of Genetics 37:772–775.[CrossRef]
Fedorova TA. (2004) Compact inflorescences: regularities of the origin and details of distribution of sexual types of flowers as a basis of dioecism origin. In Savinykh NP (Ed.). Materials of the tenth school of theoretical plant morphology: construction units in plant morphology(Vyatka State University of Humanities, Kirov) pp. 110–115.
Friis EM, Doyle JA, Endress PK, Leng Q. (2003) Archaefructus: angiosperm precursor or specialized early angiosperm? Trends in Plant Science 8:369–373.[CrossRef][ISI][Medline]
Friis EM, Pedersen KR, Crane PR. (2000) Reproductive structure and organization of basal angiosperms from the Early Cretaceous (Barremian or Aptian) of western Portugal. International Journal of Plant Sciences 161:S69–S182.
Goebel K. (1931) Blütenbildung und Sprossgestaltung (Anthoklad