Papilionoid inflorescences revisited (Leguminosae-Papilionoideae)

Abstract

Background and Aims

The inflorescence structure determines the spatiotemporal arrangement of the flowers during anthesis and is therefore vital for reproductive success. The Leguminosae are among the largest angiosperm plant families and they include some important crop plants. In papilionoid legumes, the raceme is the most common type of inflorescence. However, a range of other inflorescence types have evolved via various developmental processes. A (re-)investigation of inflorescences in Swainsona formosa, Cicer arietinum, Abrus precatorius, Hardenbergia violacea and Kennedia nigricans leads to new insights into reduction mechanisms and to a new hypothesis on the evolution of the papilionoid pseudoraceme.

Methods

Inflorescence morphology and ontogeny were studied using scanning electron microscopy (SEM).

Key Results

The inflorescence in S. formosa is an umbel with a rare type of pendulum symmetry which may be triggered by the subtending leaf. Inflorescences in C. arietinum are reduced to a single flower. An early formed adaxial bulge is the sterile apex of the inflorescence (i.e. the inflorescence is open and not terminated by a flower). In partial inflorescences of A. precatorius, the axis is reduced and its meristem is relocated towards the main inflorescence. Flower initiation follows a peculiar pendulum pattern. Partial inflorescences in H. violacea and in K. nigricans show reduction tendencies. In both taxa, initiated but early reduced bracteoles are present.

Conclusions

Pendulum symmetry in S. formosa is probably associated with distichous phyllotaxis. In C. arietinum, strong reduction tendencies are revealed. Based on studies of A. precatorius, the papilionoid pseudoraceme is reinterpreted as a compound raceme with condensed lateral axes. From an Abrus-like inflorescence, other types can be derived via reduction of flower number and synchronization of flower development. A plea is made for uniform usage of inflorescence terminology.

INTRODUCTION

The form and structure of floral arrangement along a common axis, which is widely referred to as the angiosperm inflorescence, is vital for the reproductive success of flowering plants. Different kinds of inflorescence result in spatiotemporal differences in the presentation of anthetic flowers and their receptive stigmas and/or fertile pollen to suitable pollen vectors (e.g. Wyatt, 1982; Friedman and Harder, 2004; Harder et al., 2004; Iwata et al., 2012). Inflorescence structure is also frequently used as an important systematic character (cf. Briggs and Johnson, 1979; Classen-Bockhoff, 2000; Vrijdaghs et al., 2005).

According to Weberling (1989), the basic branching pattern in legume inflorescences is racemose (Fig. 1A–C), but systems with cymose branching do occur (e.g. Tucker, 1998). Racemes are common in Papilionoideae, which is the largest subfamily of Leguminosae. Among the best known papilionoid examples are cultivated ornamentals such as common laburnum (Laburnum anagyroides), Wisteria spp., black locust (Robinia pseudacacia), lupins (Lupinus spp.; also important crop plants), and the pagoda tree (Styphnolobium japonicum), which flowers late in the year with highly branched multiple racemes. However besides the ‘simple’ raceme there are special cases such as the dorsiventral partial inflorescences in Vicieae, some Trifolieae, Galegeae (Naghiloo et al., 2012) and Loteae (Fig. 1D–F; cf. Sokoloff et al., 2007).

Fig. 1.

Racemes in Adenocarpus decorticans and Ebenus cretica; bilateral umbels in Lotus corniculatus, Lotus berthelotii × maculatus and Dorycnium germanicum. Inflorescence meristems are highlighted with yellow dots. (A, B) Adenocarpus decorticans. Frontal view of racemes with spiral flower formation. Flowers are initiated acropetally and in a helical direction; either anticlockwise (A) or clockwise (B). Acropetal maturation of flowers, with the oldest flowers positioned proximally and flower primordia distally. (C) Ebenus cretica. Side view of a raceme with spiral flower formation showing characteristic parastichies highlighted with white and yellow lines. (D) Lotus corniculatus. Flower formation starts on the right side and proceeds in a zig-zag pattern towards the left side (1–6). Distinct hairs along the borders of neighbouring flowers are formed (modified after Prenner, 2003). (E) Lotus berthelotii × maculatus. Inflorescences are reduced to three flowers per umbel. The flowers are formed in rapid sequence (1–3) and hairs arise between neighbouring flowers. (F) Dorycnium germanicum. A second set of flowers (coloured in blue) is formed in the centre of the somewhat broadened inflorescence base. Flower formation and maturation is unidirectional from right to left. Scale bars: (A)=400 µm; (B)=500 µm; (C)=200 µm; (D)=50 µm; (E, F)=100 µm.

Inflorescences with strong reduction tendencies and pseudoracemes are other peculiarities on which the present study focuses. The pseudoraceme was highlighted as a special type of inflorescence in papilionoid legumes (Lackey, 1981; Tucker, 1987a, b, 2003, 2006). Tucker (2003, p. 920) mentioned that ‘pseudoracemes differ from racemes in that two to several flowers are initiated in each bract axil rather than just one as in a raceme’ (see also Tucker, 1987a).

This study deals with the following questions. (1) Does distichous phyllotaxis have an impact on inflorescence morphology and symmetry in Swainsona formosa (Galegeae)? (2) How can the unifloral inflorescences in Cicer arietinum (Cicereae) be interpreted and is there any evidence for reduction? (3) What is the exact pattern of flower formation in partial inflorescences of Abrus precatorius (Abreae) and how can this be related to ‘pseudoracemes’ of Hardenbergia violacea and Kennedia nigirans (both Phaseoleae-Kennediinae)? (4) Is the papilionoid pseudoraceme just a compound raceme with reduced lateral partial inflorescences?

MATERIALS AND METHODS

Material

Scanning electron microscopy (SEM) micrographs are shown for Abrus precatorius L.: Teppner s.n., cult. Austria, Botanic Garden Graz (BGG), Institute of Plant Sciences, Karl-Franzens-University Graz; Adenocarpus decorticans Boiss.: Prenner 783, UK, cult. Royal Botanic Gardens, Kew (K 1995-2238); Cicer arietinum L.: Prenner 443, cult. BGG; Dorycnium germanicum Rouy: Prenner 288, cult. BGG; Ebenus cretica L.: Prenner 440, Greece, Crete, Psiloritis; Hardenbergia violacea (Schneev.) Stearn: Prenner 658, cult. BGG; Kennedia nigricans Lindl.: Prenner 491, cult. BGG; Lotus berthelotii Masf. × maculatus Breitfeld: Prenner 404, cult. BGG; Lotus corniculatus L.: Prenner 445, Austria, Graz, Schubertstrasse 21; and Swainsona formosa (G.Don) J.Thomson, Prenner 784, Australia, New South Wales, Calder Hwy, N of Mildura. Voucher specimens have been deposited in the herbarium of the Institute of Botany, Karl-Franzens-University Graz (GZU) and/or liquid preserved collections are held by the author. Specimens were determined by the author, except for A. precatorius which was determined by H. Teppner (Institute of Plant Sciences, Karl-Franzens-University Graz, Austria).

Methods

Material was collected and immediately fixed in FAA (90 parts 70 % ethanol, 5 parts 100 % acetic acid, 5 parts formalin) and stored in 70 % ethanol. The specimens were critical point dried in an Autosamdri-815B critical-point dryer at the Royal Botanic Gardens (RBG), Kew (UK) or in a Polaron 7010 CPD at the Institute of Plant Sciences, Karl-Franzens-University Graz (Austria). Dried material was mounted onto specimen stubs using clear nail polish and coated with platinum using an Emitech K550 sputter coater at RBG, Kew or coated with gold using an Agar sputter coater in Graz. SEM studies were made with a Hitachi cold field emission SEM S-4700-II at RBG, Kew or a Philips XL 30 ESEM at the Institute of Plant Science in Graz (Austria). Between two and ten inflorescences were studied from each examined species, and between 40 and 140 SEM images were analysed per species.

RESULTS

Swainsona formosa (Galegeae; Fig. 2A–G)

Fig. 2.

Inflorescence development in Swainsona formosa (SEM). (A) Distichous phyllotaxis of inflorescences (1–4; inflorescence 1 and 2 removed). Subtending leaves are highlighted in green and their direction in relation to the inflorescences is highlighted with arrows. The shoot apical meristem (SAM) in the centre is partially obscured by the third inflorescence. (B) Detail of (A). A young inflorescence in the axil of a frondose leaf (L) with two lateral stipules (Si) is visible. Flower initiation started furthest away from the subtending leaf and continues in an anticlockwise direction (1–6). (C) Young inflorescence (IM) subtended by a frondose leaf (L) with two large stipules (Si). The subtending leaf points to the right, and flower formation starts furthest away from this and continues in a clockwise direction. Another inflorescence (IM') in the background appears in the axil of a leaf (L') flanked by two lateral already well developed stipules (Si'). The SAM is visible in the centre. (D) A similar ontogenetic stage to that shown in (C) with the first formed inflorescence removed. In contrast to (C), the leaf subtending the inflorescence is pointing to the left. (E) Young inflorescence (IM') formed in the axil of a leaf (L') and flanked by two large stipules (Si'). When the leaf is bent to the right, the inflorescence meristem is bent to the left. The SAM is again bent to the right. (F) A somewhat older developmental stage showing the onset of flower formation (asterisk) furthest away from the subtending leaf. The formation of the subtending leaf (L'') becomes just visible on the SAM. (G) Young inflorescence showing developing flowers which were formed in rapid succession in a clockwise direction (1–6). In the centre, there is a massive sterile tip of the inflorescence meristem. IM, IM', inflorescence meristem; L, L', L'', leaf subtending first, second and third inflorescence; SAM, shoot apical meristem; Si, Si', stipule. Scale bars: (A)=1 mm; (B)=200 µm; (C, D)=400 µm; (E, F)=200 µm; (G)=500 µm.

In S. formosa, racemose inflorescences are formed in a distichous phyllotaxis in the axil of frondose leaves, each with two large stipules at the base (Fig. 2A, B). The position of the subtending leaf changes at each node (pendulum symmetry; Fig. 2A–F). During the earliest developmental stages, the subtending leaf points towards one side of the subtended inflorescence meristem. Depending on the position of the subtending leaf, the inflorescence meristem shows a characteristic asymmetric shape and is broader either to its left or to its right side (Fig. 2C, E). Flower formation always starts in the position furthest away from the subtending leaf (Fig. 2B, C, F). Flowers are formed in an acropetal spiral, in either an anticlockwise (Fig. 2B) or a clockwise direction (Fig. 2C, G). The inflorescence axis does not elongate significantly and the inflorescence stays flat and dense, resulting in a five- to eight-flowered umbel. The inflorescence is open and terminates with a prominent vegetative tip in its centre (Fig. 2G).

Cicer arietinum (Cicereae; Fig. 3A–D)

Fig. 3.

Inflorescence development in Cicer arietinum (SEM). (A) A reduced inflorescence is formed in the axil of a pinnate leaf (L) with basal stipules (Si). Flower (F) subtended by a bract (B) and preceded by bracteoles, of which only the left one is visible (arrow). In the adaxial position, the inflorescence meristem is found (IM). In the background, a second inflorescence meristem (IM') associated with a young leaf (L') can be seen. To the right of this complex, the shoot apical meristem (SAM) can be found. (B) Side view of (A). The bracteole of the first formed flower and the bract which subtends the flower are visible. Another undifferentiated inflorescence meristem is formed to the left in the axil of a leaf, with a stipule forming at its base. (C) Abaxial view of a flower primordium subtended by a bract. The flower (i.e. reduced inflorescence) sits between the subtending leaf and its stipule. (D) Young flower with the formation of the abaxial sepal immediately followed by the two lateral sepals (unidirectional sepal formation). The inflorescence meristem which does not produce any more flowers can be seen in the upper/adaxial position (arrow). Abbreviations: B, bract; Bl, bracteole; F, flower; IM, IM', inflorescence meristem; L, L', frondose leaf; s1, s2, sepals; Si, Si', stipule. Scale bars=50 µm throughout.

In C. arietinum, inflorescences are reduced to a single flower which is formed in the axil of a frondose leaf (Fig. 3A). The leaf which subtends the reduced inflorescence is pinnate and early forms two stipules at its base. The inflorescence meristem is positioned between the leaf and the stipule. Later in development, the inflorescence meristem forms a single floral primordium which is subtended by an abaxial bract and preceded by two bracteoles (Fig. 3A–C). When the flower is still at the primordial stage, a meristematic bulge becomes visible in the adaxial position opposite the subtending bract. This can be interpreted as the sterile ending of the reduced inflorescence (Fig. 3A, B, D). At maturity, this sterile ending is still visible as an approx. 1 mm long appendix opposite the bract (not shown).

Abrus precatorius (Abreae; Fig. 4A–F)

Fig. 4.

Inflorescence in Abrus precatorius (SEM). (A) Young partial inflorescences with initiation of three flowers (1–3) and three flower subtending bracts (4–6). No flowers or bracts are formed in an adaxial position (asterisk); inflorescence meristem is highlighted with an orange dot. (B) Partial inflorescence with nodose axis and acropetal formation of individual flowers (1–15). The partial inflorescence meristem (highlighted with an orange dot) is located off-centre close to the main inflorescence axis (asterisk). (C) Detail of (B). Complete numbering of flowers (4–10), flower primordia (11–15) and flower subtending bracts (16–21). Flowers and flower primordia are highlighted in red. No flowers are formed in an adaxial position (asterisk). (D) Partial inflorescence with a nodose axis and acropetal formation of individual flowers (1–18). The partial inflorescence meristem (highlighted with an orange dot) is located off-centre close to the main inflorescence axis (asterisk). (E) Detail of (D). Complete numbering of flowers (7–10), flower primordia (11–15) and flower subtending bracts (16–21). Flowers and flower primordia are highlighted in red. No flowers are formed in the adaxial position (asterisk); partial inflorescence meristem highlighted with an orange dot. (F) Same as (D). The pendulum direction of flower formation is highlighted with blue (to the right) and red (to the left) arrows. Scale bars: (A)=100 µm; (C)=200 µm; (E)=300 µm; (B, D, F)=500 µm.

In A. precatorius, partial inflorescences are formed in acropetal direction along a common inflorescence axis (not shown). The axis of each partial inflorescence is nodose, and the inflorescence meristem (orange dot in Fig. 4A–F) is positioned off-centre, towards the main inflorescence axis (asterisk in Fig. 4A–E). Individual flowers are formed acropetally (Fig. 4A–E) and in a characteristic pendulum pattern, switching between right (blue arrows in Fig. 4F) and left (red arrows in Fig. 4F). Each flower is subtended by an abaxial bract and preceded by two bracteoles (Fig. 4C, E). Within each partial inflorescence, a complete ontogenetic series from fully developed flowers (e.g. flower number 1 in Fig. 4D) to floral primordia (e.g. 15 in Fig. 4C, E) can be found.

Hardenbergia violacea (Phaseoleae-Kennediinae; Fig. 5A–C)

Fig. 5.

Inflorescences in Hardenbergia vilacea and Kennedia nigricans (SEM). (A–C) Hardenbergia violacea. (A) Inflorescence with acropetal and helical formation of three-flowered partial inflorescences (1–11). One of the three-flowered partial inflorescences is coloured in red and the inflorescence meristem is highlighted with a yellow dot. (B) Young partial inflorescence with two lateral flowers initiated (F1), each subtended by a bract (B'). The bract subtending a third flower is formed in the abaxial position and the partial inflorescence axis is visible (orange dot). (C) Three-flowered partial inflorescence. Two lateral flowers are formed first and show very similar developmental stages. At the base of these flowers, bracteoles can be distinguished (arrowheads). The third flower is formed later. The orange dot marks the sterile ending of the partial inflorescence (i.e. the partial inflorescence is open). (D–F) Kennedia nigricans. (D) Inflorescence with acropetal and helical formation of two-flowered partial inflorescences (1–9). One of the two-flowered partial inflorescences is coloured in red and the inflorescence meristem is highlighted with a yellow dot. (E) Young partial inflorescence with two flowers simultaneously initiated. The orange dot highlights the meristem of the partial inflorescence. (F) Partial inflorescence with two simultaneously developing flowers with abaxial and lateral sepals. Bracteoles (arrowheads) and a remnant of inflorescence meristem (orange dot) are visible. Scale bars: (A, C, D)=100 µm; (B, E, F)=50 µm.

In H. violacea, three-flowered partial inflorescences are formed in an acropetal spiral (Fig. 5A). Within each partial inflorescence, two lateral flowers are formed first, followed by a single distal flower (Fig. 5B, C). The meristem of the partial inflorescence is (re-)located towards the main inflorescence axis which terminates with a sterile tip (i.e. the partial inflorescence is open; Fig. 5C). Each flower is subtended by an abaxial bract and preceded by two bracteoles which are visible at early developmental stages but which cease to grow early (i.e. initiated but reduced bracteoles; Fig. 5C).

Kennedia nigricans (Phaseoleae-Kennediinae; Fig. 5D–F)

In K. nigricans, two-flowered partial inflorescences are initiated in an acropetal spiral (Fig. 5D). Within each partial inflorescence, flowers are formed simultaneously (Fig. 5E) and their further development is also simultaneous (Fig. 5F). The meristem of the partial inflorescence is used up gradually in the course of flower formation. While there is a small residuum of the meristem visible in early ontogeny (Fig. 5E, F) there is no remnant left in later stages (not shown). Each flower is subtended by an abaxial bract and preceded by two bracteoles which stop growing early and deteriorate subsequently (i.e. initiated but reduced bracteoles; Fig. 5F).

DISCUSSION

Inflorescence ontogeny in Swainsona formosa shows a rare form of pendulum symmetry

The consecutively formed inflorescences in Swainsona formosa show a pronounced pendulum symmetry which seems to be influenced by the position of the subtending leaf. The following morphological characters can be determined in relation to the position of the subtending leaf: (1) the asymmetric shape of the early inflorescence meristem; (2) the position of the first formed flower, always furthest away from the subtending leaf; and (3) a switch between clockwise (with the subtending leaf pointing to the right) and anticlockwise (with the subtending leaf pointing to the left) formation of flowers in consecutive inflorescences (Fig. 6A).

Fig. 6.

Pendulum symmetry in Swainsona formosa and strong reduction tendencies in Cicer arietinum. (A) Two consecutive inflorescences in S. formosa along a common axis (asterisk). The position of the subtending leaves pre-determines the position of the first formed flower (always formed furthest away from the subtending leaf) and the direction of the helical flower formation, which is either anticlockwise (right inflorescence) or clockwise (left inflorescence). (B–D) Cicer arietinum. The unifloral inflorescence in C. arietinum can be deduced from a multiflowered raceme (B) via a strong reduction to a single flower which is subtended by a foliage leaf, the floral bract and two bracteoles (C, D). The sterile tip in the adaxial position (green dot in D) is the remnant of the reduced inflorescence (arrows in B, C). Only one (of several) reduced partial inflorescences is shown. The asterisk marks the main axis.

The term Pendelsymmetrie, or pendulum symmetry, was coined by Goebel (1928) for left/right oscillation of symmetry in plants. Pendulum symmetry is more often studied and apparently more common in vegetative shoots (see Charlton, 1998). In Ulmus, distichous pendulum symmetry of the leaves was studied in detail by Charlton (1993, 1998), who found that leaf primordia are asymmetrical from their inception.

Pendulum symmetry is also found in flowers of Centranthus (Valerianaceae), where the chirality changes in a regular pattern in the monochasial partial inflorescence, or in flower pairs of Marantaceae (Kirchoff, 1983; Endress, 1999). In flowers of Oxalis (Oxalidaceae) and Malvaceae, the direction of petal contortion changes with every node in monochasial partial inflorescences (Endress, 1999, 2001). In Leguminosae, Tucker (1996) described the phenomenon in detail for the caesalpinioid Chamaecrista fasciculata. In this species, flowers on successive nodes have the enlarged keel petal either on the right or on the left side. Phyllotaxis in this species is also distichous, similar to Swainsona (Dormer, 1945). While in these examples the floral symmetry changes in a pendulum pattern, it is the inflorescence which is affected in Swainsona.

The exact mechanisms that lead to pendulum symmetry are still unknown. Based on the existing evidence, it seems plausible that in S. formosa phyllotaxis influences shoot morphology. In this context, the symmetry of both the shoot apical meristem (SAM) and its products is of special interest. In Swainsona, spatial constraints may act on the shape of the SAM which might be shaped by the adjacent developing inflorescence (Fig. 2C–F). This in turn could lead to the location of the leaf that subtends the inflorescence on one particular side of the SAM. An inhibition factor may then be responsible for the formation of the first-formed flower furthest away from the subtending leaf.

Pendulum symmetry was not mentioned in a recent study that focused on floral ontogeny in S. formosa (Tapingkae et al., 2007). The interpretation of the earliest generative stages by Tapingkae et al. (2007) deviates from the current study in some details. In Tapingkae et al., fig. 2A (p. 646) is reminiscent of a vegetative apex rather than an ‘early transition stage’ as stated by the authors, and the depicted floral primordium seems to resemble a leaf primordium more closely. Furthermore, based on my results, the structure labelled as a floral primordium in fig. 2B in Tapingkae et al. (2007, p. 646) is more likely to be a leaf primordium with two lateral stipules, because there is no sign of a bract which would be expected to subtend the floral primordium. One stipule at the right side of the leaf primordium is visible (correctly labelled ‘st’), while the left-hand stipule was removed during dissection but remains visible as a scar. The second structure opposite the leaf primordium (erroneously labelled as stipule) is more likely to be the SAM, and the unlabelled structure in the centre of their fig. 2B corresponds to the inflorescence primordium, as shown in the present study (see Fig. 2C–F in this study).

Reduction tendencies in Cicer arietinum and how they can be traced and evaluated

The genus Cicer comprises 44 species which were recently reclassified by Davies et al. (2007). Its flowers are either solitary or organized in two- to five-flowered axillary racemes (van der Maesen, 1972). In C. arietinum, which has unifloral inflorescences, very early in ontogeny a sterile tip is evident on the adaxial side of the flower. This can be interpreted as a sterile ending of the open and unifloral inflorescence which is most probably derived via reduction from an ancestor with more than one flower per raceme (Fig. 6B–D). Note that Weberling (1989) treated unifloral units of C. arietinum as ‘coflorescences’ of a more complex ‘synflorescence’ (see also Weberling and Troll, 1998). However, this does not affect the above interpretation significantly.

Besides this, an abaxial bract and two bracteoles can be found at the base of the young flower. This observation is in contrast to those of Hutchinson (1964, p. 452) who quoted the absence of bracteoles in Cicer, and Tucker (1987a) who cited Cicereae as having bracteoles converted to spines. It seems more likely that the spines occurring in some species of Cicer are the sterile endings of reduced inflorescences, rather than converted bracteoles (cf. Wydler, 1860).

A new interpretation of the papilionoid pseudoraceme

The present study provides strong evidence that in partial inflorescences of Abrus, flower initiation follows a very consistent pendulum pattern. A single flower is formed per node and the inflorescence axis is strongly condensed to a nodose structure. All flowers are formed in a continuous ontogenetic series as found in ‘normal’ papilionoid racemes (cf. Fig. 1A–C) and no evidence was found for a temporary suppression of flowers as mentioned by Tucker (1987b). Therefore, the inflorescence of Abrus can be interpreted as a compound raceme (sensu Prenner et al., 2009) with condensed partial inflorescences (Fig. 7A, B). In the course of this condensation, the meristem of partial inflorescences was relocated towards the main inflorescence axis. The reason for this relocation might be an inhibitory field acting from the inflorescence which prevents flower formation on the adaxial side of the partial inflorescence. The relocation of the meristem might in turn disturb spiral flower formation; instead, flower formation oscillates in a pendulum pattern around the inflorescence axis. It is important to highlight that this pattern is not related to the pendulum symmetry of Swainsona because the direction of flower formation among individual partial inflorescences appears not to follow a fixed pattern, which would be expected for pendulum symmetry.

Fig. 7.

Reinterpretation and evolution of the papilionoid ‘pseudoraceme’. (A, B) Modified after Lackey (1981). Bracteoles are omitted in (C–E). (A) Double raceme composed of three partial inflorescences, one of which is encircled. (B) Inflorescence in which the axes of the lateral racemes are reduced, as found in Abrus precatorius. The axis of the partial inflorescences is nodose and the inflorescence meristem (small arrows) is positioned close to the main inflorescence axis. (C) Diagram of a partial inflorescence of A. precatorius. Three consecutively formed flowers are numbered and encircled (1–3). The inflorescence meristem (green dot) is positioned off-centre and close to the main inflorescence axis (asterisk). (D) Three-flowered partial inflorescences, as found in Hardenbergia violacea, can be deduced via reduction of flower number and synchronization of flower formation. The inflorescence meristem (green dot) is positioned close to the main inflorescence axis (asterisk). (E) Two-flowered units, as found in, for example, Kennedia nigricans, can be deduced via further reduction of the central flower. The inflorescence meristem is used up completely. The main inflorescence axis is on top (asterisk). Bracetoles and initiated but reduced bracteoles are not shown in (C–E).

‘Pseudoracemes’ are found in Abreae (this study; Tucker, 1987b), Desmodieae (Prenner, 2004a), Millettieae (Teixeira et al., 2009), Psoraleeae and Phaseoleae (cf. Lackey, 1981; Tucker, 2006). The phylogenetic position of the tribe Abreae is rather isolated at the base of the ‘core Millettioids’ sensu Hu et al. (2000), and various elements which were excluded from Phaseoleae sensu lato (see also Wojciechowski et al., 2004; Lewis et al., 2005). Taking this into account, one scenario is that more reduced inflorescence types could have evolved from an inflorescence similar to that of Abrus. The three-flowered partial inflorescences of H. violacea may have evolved via a reduction of the number of flowers per partial inflorescence and a synchronization of the flower initiation within partial inflorescences (Fig. 7C, D). Moreover, the two-flowered partial inflorescences in K. nigricans can be derived from a Hardenbergia-like inflorescence via further reduction in flower number (Fig. 7D, E). Other types such as those found in Apios, Vigna, Erythrina, Desmodium, Dioclea and Pongamia (all depicted by Tucker 1987b, fig. 60, p. 200) can be deduced accordingly. However, the phylogenetic relationships among these taxa are not fully resolved and understood (see Wojciechowski et al., 2004; Lewis et al., 2005) and, therefore, it is not possible to draw final conclusions. Based on the available phylogenetic data, it seems most probable that a condensation of partial inflorescences took place independently several times. An alternative hypothesis for the evolution of compound racemes with condensed lateral inflorescences is via the reduction of ‘normal’ radially symmetric partial inflorescences. Butea monosperma could provide evidence for such a scenario. According to Tucker (1987b), in this species, a triad of flowers is formed first, subsequently followed by helical formation of flowers.

It is important to highlight that compound racemes are also sometimes referred to as ‘panicles’ (e.g. Lackey, 1981; Tucker, 1987b) which might be one of the reasons why the pseudoraceme was long considered as a special type of inflorescence. However, from recent attempts to clarify conflicting definitions, it is clear that the panicle is a special case and not to be confused with compound (double or triple) racemes (cf. Prenner et al., 2009; Endress, 2010). It is therefore timely to emphasize that in legumes we frequently find compound racemes (sensu fig. 1b in Prenner et al., 2009; fig. 3 in Endress, 2010), but there is currently no solid proof for the existence of true panicles in the group.

The situation found in Abrus demonstrates the need for careful (re-)investigation of flowers in groups or ‘fascicles’ across angiosperms. Flower fascicles can result either from the condensation of racemose systems as shown in the present study or via condensation of cymose systems. These two scenarios represent two significantly different evolutionary histories. The euphorb cyathium is a classic example of such a condensed cymose system (cf. Prenner and Rudall, 2007; Prenner et al., 2008; Horn et al., 2012).

The role of bracts and bracteoles as key markers in inflorescence architecture

Bracts and bracteoles are key markers of inflorescence architecture (Prenner et al., 2009). Therefore, it is important that these structures are characterized correctly by carefully investigating their (relative) position and presence or absence. In the present study, I have shown that bracts or pherophylls are also involved in the symmetry of inflorescences in S. formosa, and that in C. arietinum a sterile protrusion opposite the flower subtending bract represents the sterile ending of the inflorescence axis and therefore should not be confused with a bract.

As shown earlier (Prenner, 2004b), bracteoles may also be discernible only at the earliest developmental stages. They stop growing at a primordial stage and are considered to be lost or resorbed in the anthetic flower. According to Lackey (1981), bracteoles are missing in Phaseoleae subtribe Kennediinae which includes the genera Hardenbergia, Kennedia and Vandasina. However, Prenner (2004b) found initiated but reduced bracteoles in K. nigricans, and further evidence for this was shown in the present study. Initiated but reduced bracteoles are also found in Kennedia beckxiana F. Muell. and K. rubicunda Vent. (G. Prenner, unpubl. res.). Furthermore, the present study shows that bracteoles are also formed in H. violacea (Fig. 5C). This character was not mentioned by Tucker (2006) who labelled similar structures as pedicels (cf. fig. 2J in Tucker, 2006). It remains to be studied whether initiated but early resorbed bracteoles also occur in Vandasina and whether this character is a synapomorphy for the subtribe Kennediinae.

A plea for detailed ontogenetic studies and uniform usage of terminology

Detailed ontogenetic studies are vital for the evaluation of complex and/or highly reduced morphological structures such as found in some inflorescences. Such studies are also invaluable in the context of determining characters and character states and/or in the context of further homology assessments. To avoid misinterpretations and to obtain a conclusive result, it is particularly important to generate and compare as many developmental stages as possible. This can be achieved by collecting plant material over a longer period of time and/or by dissecting samples of various developmental stages from single collections.

Another important aspect is the uniform usage of terminology, and a major objective for the future should be to find a broad consensus on a widely accepted terminology of inflorescences (see recent attempts by Prenner et al., 2009; Endress 2010; Bull-Herenũ and Classen-Bockhoff, 2011). This is especially critical at the interface of ‘classical’ plant morphology and more recent scientific fields such as developmental genetics or plant modelling where often different terms are used for the same structure, or the same terms are used for different structures, which are prone to create confusion.

Conclusions

It seems likely that in S. formosa, distichous phyllotaxis has an impact on the morphology and symmetry of inflorescences. It will be interesting to study other taxa with distichous phyllotaxis and to test the hypothesis that spatial constraints via developing inflorescences act on the morphology and symmetry of the SAM (and in this way induce pendulum symmetry). The results found in C. arietinum lead on to the question of how often and where in the phylogeny of the genus inflorescences were reduced and if the ontogenetic pathways are the same within the genus. Clarification of the genetic background of this reduction could prove important for the cultivation of multiflowered taxa with a potential increased crop yield. The redefinition of the pseudoraceme as a compound raceme with condensed partial inflorescences is a good starting point for a reinvestigation of the evolution of such inflorescences among papilionoid legumes. It is most probable that such reductions have taken place several times independently. The advantages or disadvantages of inflorescences with condensed lateral inflorescences await further investigation.