Flower-like heads from flower-like meristems: pseudanthium development in Davidia involucrata (Nyssaceae)
- 221 Downloads
Flower-like inflorescences (pseudanthia) have fascinated botanists for a long time. They are explained as condensed inflorescences implying that the pseudanthium develops from an inflorescence meristem (IM). However, recent developmental studies identified a new form of reproductive meristem, the floral unit meristem (FUM). It differs from IMs by lacking acropetal growth and shares fractionation, expansion and autonomous space filling with flower meristems (FM). The similarity among FUMs and FMs raises the question how far flower-like heads originate from flower-like meristems. In the present paper, pseudanthium development in Davidia involucrata is investigated using scanning electron microscopy. D. involucrata has pincushion-shaped heads composed of densely aggregated, perianthless flowers and associated with two large showy bracts. Early developmental stages show a huge naked FUM. The FMs appear almost simultaneously and lack subtending bracts. With ongoing FUM expansion new space is generated which is immediately used by further FM fractionation. The heads have only staminate flowers or are andromonoecious with staminate and a single perfect flower in oblique position. All FMs lack perianth structures and fractionate a variable number of stamen primordia. The perfect FM is much larger than the staminate FMs and forms a syncarpous gynoecium with inferior ovary. Pseudanthium development in D. involucrata confirms the morphogenetic similarity to FMs as to acropetal growth limitation, meristem expansion and fractionation. It thus should not be interpreted as a condensed inflorescence, but as a flower equivalent. Furthermore as the FUM develops inside a bud, its development is considered to be influenced by mechanical pressure. The oblique position of the perfect flower, the developmental delay of the proximal flowers, and the variable number of stamens which were observed in the pseudanthium development, can be caused by mechanical pressure. Next to the Asteraceae, D. involucrata offers a further example of a pseudanthium originating from a FUM. More knowledge on FUMs is still needed to understand diversification and evolution of flower-like inflorescences.
KeywordsExtrafloral bracts Floral unit meristem (FUM) Flower meristem (FM) Inflorescence meristem (IM) Mechanical pressure Utilization of space
One of the most fascinating tendencies in flowering plants is the formation of flower-like patterns, i.e. pseudanthia (e.g. Claßen-Bockhoff 1990, 1991; Froebe and Ulrich 1978; Good 1956; Johow 1884; Leppik 1969; Troll 1928). The well-known similarity of the daisies and everlastings in the Asteraceae to radial flowers depends on the formation of a sexual center composed of small, densely packed flowers and the presence of a showy periphery formed by ray flowers or showy bracts.
Pseudanthia have gained increasing interest in the last years as they share characters of the vegetative plant body (branching patterns, bracts, phyllotaxis) with reproductive functions (flower protection, attractiveness, symmetry). The central questions concerning the conditions in the transitional zone of the shoot system, where meristems merge from the vegetative to the reproductive stage, are investigated on the molecular and morphogenetic level. While molecular studies aim to elucidate how far floral gene expression is transferred to extrafloral structures (e.g. Albert et al. 1998; Claßen-Bockhoff et al. 2013; Vekemans et al. 2012) and how subordinate symmetry is established and regulated in pseudanthia (e.g. Bello et al. 2017; Broholm et al. 2014; Mizzotti et al. 2015; Owens et al. 2016; Prenner et al. 2011; Zhao et al. 2016), morphogenetic studies deal with the flower-inflorescence boundary and with meristem conditions (e.g. Claßen-Bockhoff and Bull-Hereňu 2013; Naghiloo and Claßen-Bockhoff 2017; Prenner et al. 2008; Rudall 2003; Rudall and Bateman 2010).
Inflorescence (IM) and floral unit meristems (FUM)
An exiting finding in recent morphogenetic studies was the discovery of three different reproductive meristems, i.e. inflorescence (IM), flower (FM) and floral unit meristems (FUM, Bull-Hereñu and Claßen-Bockhoff 2011a, b; Claßen-Bockhoff and Bull-Hereñu 2013).
The shoot apical meristem (SAM) is characterized by apical growth based on continuous stem cell production (Bäurle and Laux 2003). Lateral primordia arise by the process of segregation, i.e. the meristem tip remains active while lateral parts are continuously segregated merging into leaf primordia. In consequence, the shoot is composed of nodes and internodes.
The IM shares the apical growth with the SAM. In Arabidopsis thaliana, it has a central zone whose cell division activity is controlled by a genetic feedback loop maintaining open growth (Kwiatkowska 2008 and references herein). The process of segregation characterizes all inflorescences originating from an IM in which bracts and lateral flowers are segregated from the meristem tip constituting nodes (Claßen-Bockhoff and Bull-Hereñu 2013). However, in contrast to SAMs, the activity of the meristem tip is limited in IMs. After flower formation (which rarely last more than one season), the IM ceases growth and merges into a flower (closed inflorescence) or parenchyma cells (open inflorescence).
The FUM, in contrast, is more similar to a flower meristem (FM) lacking a central zone (Bull-Hereñu and Claßen-Bockhoff 2011a; Claßen-Bockhoff 2016; Kwiatkowska 2008; Tucker and Grimes 1999). The meristem has lost the capability to segregate lateral primordia and to form nodes and internodes by this process. Instead, it is able to expand considerably forming a naked stage at the beginning and generating new space during development. It is completely used by the process of fractionation. This means, the meristem splits into floral organs in the case of FMs and into sub-meristems in the case of FUMs. Based on the concept of directed auxin flow (Reinhardt et al. 2003), the process of fractionation runs autonomously densely filling all available space (Claßen-Bockhoff and Meyer 2016; Douady and Couder 1996; Runions et al. 2014).
The concept of FUMs was introduced in a generalized way with the main focus on different meristem conditions (Claßen-Bockhoff and Bull-Hereñu 2013). Among the floral units identified so far, simple and secondary heads in Asteraceae are the most prominent examples (Claßen-Bockhoff 2016). However, as also further taxa with pseudanthia are proved or assumed to have FUMs, it is hypothesized that a high number of flower-like heads originate from flower-like FUMs. This hypothesis is currently tested in a comprehensive project aiming to explain the peculiarities of FUMs and to understand evolution and diversification of inflorescences and pseudanthia in more detail (Claßen-Bockhoff et al. 2013; Naghiloo and Claßen-Bockhoff 2017). The present paper continues this series investigating pseudanthium development in Davidia involucrata Baill.
Davidia involucrata is an endangered species endemic to China. It belongs to the paleo-tropical flora of the Tertiary period (You et al. 2014) and is the only extant representative of a genus widely spread in the past (Manchester 2003). Today, it is distributed in several provinces of China ranging from Gansu in the north to Yunnan in the south and Hunan in the east (You et al. 2013). Its trivial names “dove tree” and “handkerchief tree” refer to the two large white bracts associated with each of the pendent heads (Fig. 1j). The whole canopy is full of white flags when flowering looking like doves sitting in the tree or like handkerchiefs hanging down from the branches (Fig. 1a).
The systematic position of Davidia was discussed for a long time, but recent molecular data confirm its close relationship to Camptotheca Decne and Nyssa L. within the Nyssaceae (Xiang et al. 2011 and references herein). The three genera are monotypic or poor in species, distributed in E-Asia, SE-Asia and N-America and share a woody growth form (Eyde 1963; Kubitzki 2004). Inflorescences are racemose to globose in the polygamodioecious genus Nyssa (8 ssp.) and capitate with dyads instead of flowers in the monoecious genus Camptotheca (Jahnke 1986; Moser 1968).
Flowering heads are globose and look like pincushions due to the numerous exposed stamens. They include many reduced flowers in dense arrangement, thus forming a pseudanthium. The attractiveness of the head is based on the large showy bracts, which are unequal in size (Fig. 1j). These bracts are greenish in the beginning (Fig. 1g) and lose their chloroplasts when expanding. The white color is based on flavonoids (Hu et al. 2007), which weakly absorb UV-light (Burr and Barthlott 1993). Experiments have shown that the bracts take over sepal and petal functions protecting pollen grains from rain and attracting pollinators (Sun et al. 2008). This transfer of function goes along with the expression of a subset of stamen identity genes in the bracts (Vekemans et al. 2012).
The heads either consist of staminate flowers or are andromonoecious with a single perfect flower surrounded by many staminate ones. As nectaries are lacking, pollen is the only reward (pollen blossom). The extremely high pollen/ovule ratio may compensate for pollen loss by pollen-collecting bees and pollen-eating beetles (Sun et al. 2008).
The pseudanthium of D. involucrata is usually named a head (Moser 1968; Vekemans et al. 2012), synflorescence (Jahnke 1986) or capitulum (Sun et al. 2008) implying that it is homologous to a condensed raceme. However, there are some peculiarities, which are not easy to interpret. The perfect flower has an oblique position raising the question whether it is terminal and secondarily shifted towards the side of the larger bract or whether it is obliquely arranged from the beginning (Jahnke 1986; Moser 1968). Moser (1968) noted an irregular initiation of the FMs and discussed this as a possible indication of an underlying cymose structure. Jerominek et al. (2014) confirmed the unusual initiation sequence. They recorded the development of the pseudanthium by video documentation following a living meristem for 30 days. They found that the proximal FMs were initiated later but develop faster than the equatorial FMs. A first analysis by scanning electron microscopy (SEM) was presented by Vekemans et al. (2012), but neither these authors nor the previous ones discussed the development of the D. involucrata head with reference to inflorescence structure and meristem conditions.
Aims of the paper
The pseudanthium of D. involucrata was already identified as a floral unit (Claßen-Bockhoff and Bull-Hereñu 2013; Jerominek et al. 2014), but its development was not yet shown in detail. The first aim of the present paper is to explain the peculiarities of the FUM considering the processes of meristem expansion and fractionation. The second aim is to understand the unusual FM initiation at the FUM. It has to be tested whether the initiation pattern is irregular, i.e. variable and unpredictable, or whether it is always basipetal as shown by Jerominek et al. (2014). If it is ‘only’ unusual, but regular, it may indicate a complex structure as assumed by Moser (1968) and discussed for some Asteraceae (Bello et al. 2013; Pozner et al. 2012; Zhao et al. 2016). If it is unpredictable, space and mechanical pressure may play a role (Hamant et al. 2008). The latter is possible because head development starts in the previous year (Moser 1968) and young stages overwinter inside the tight package of winter buds.
Materials and methods
Plant material was provided by the Botanical Garden at Mainz University, Germany, where three trees are cultivated. The trees are 10–12 m high and more than 60 years old. The seasonal cycle of flowering branches from sprouting in spring to flowering and fruiting was recorded by photographical documentation. In parallel, 10 buds from different branches were collected and fixed in 70% EtOH every two weeks to document head development throughout the year.
In total, 118 buds were dissected and dehydrated in an alcohol-acetone series. The probes were critical point-dried (BAL-TEC CPD030), mounted and sputter coated with gold (BAL-TEC SCD005). Finally, they were observed and analyzed using the scanning electron microscope (ESEM XL-30 Philips). All steps were conducted according to the manufacturer’s protocol.
To illustrate meristem expansion, samples were photographed in a standardized way, i.e. from side and top views. Maximal height and width were recorded in 44 samples including all developmental stages. As far as possible, SEM pictures are presented in the same magnification visualizing meristem expansion during development.
To ease reading, some key terms are recapitulated (Claßen-Bockhoff and Bull-Hereñu 2013):
The term primordium refers to a young organ (leaf, floral organ), while the term meristem is used for a tissue originating primordia.
The process of segregation means the separation of a lateral part of the shoot apical meristem that undergoes differentiation while the meristem itself continues acropetal growth. The process of fractionation, in contrast, means the subdivision and complete use of an existing or expanding meristem lacking the capability of acropetal growth.
A reproductive meristem showing segregation (as in a raceme) is called an inflorescence meristem (IM); one showing fractionation is either a flower meristem (FM) or, in case of repeated fractionation (as in a head), a floral unit meristem (FUM).
Seasonal life stages in D. involucrata
Heads rarely appear at the end of long shoots. Instead, they often terminate lateral short shoots (Fig. 1b, j). These short shoots usually produce four spirally arranged leaves per season which are densely aggregated. The pseudanthium is stalked presenting the two bracts several centimeters above the leaves and in a small distance to the flower head (Fig. 1j). After fruit set, the stalk breaks off and the short shoot continues growth from the bud located in the axil of the distal leaf. The short shoot, thus, has an acrotonous, sympodial, extremely monochasial construction.
Davidia involucrata is a spring flowering tree. At the study area, the well protected winter buds start to grow up end of March (Fig. 1b–e). The outermost scales (tegments) are pushed aside by young leaves already preformed in the bud. Now, development proceeds very fast. Together with the unfolding of the leaves (Fig. 1f) young heads appear (Fig. 1g). They enlarge the bracts within one week and flower from mid to end of April (Fig. 1i, j). Fruit development starts end of April (Fig. 1k) and is completed in August (Fig. 1l). The large oval-shaped, very hard fruits contain only few seeds and persist until October, when they drop down together with the leaves. Single fruits can be still seen on the tree in winter.
Young leaves within the bud are very hairy on both sides. When they unfold, they loose the hairs and only remain a bit silky on the lower side. Each leaf elongates its petiole and unrolls a lamina with teethed margins (Fig. 1f). The two bracts are greenish in the beginning (Fig. 1g) and resemble the leaves in shape and vascularization. However, in contrast to them, they do not form a petiole and become white when expanding up to 20 cm (Fig. 1j). They considerably differ in size, one being almost double as large as the other one. After internode elongation, they stand a bit apart from the flower head with usually the larger one in a slightly lower position.
The young heads are red due to the colored anthers. In the case of andromonoecy, they include a single perfect flower presenting 6–8 (rarely more) greenish stigma rays (Fig. 1h). Only now, i.e. immediately before flowering, filaments elongate and stigma rays mature. Staminate heads consist of 80–100 staminate flowers with (1–)6–8 stamens (Fig. 1j), while andromonoecious heads have less staminate flowers and a single perfect flower with more than 20 stamens (Fig. 1i). The perfect flower has an oblique position and an inferior ovary presenting the stamens on its top (0.75 cm high). Interestingly, the filaments of these stamens are much shorter (0.25 cm) than those in the staminate flowers (1 cm), thus compensating for their higher level of insertion. As a result, all anthers are exposed equally on the surface of the pincushion-shaped head (Fig. 1k).
Heads flowering in April are already initiated in June of the previous year, thus, only 6 weeks after the end of flowering of the current year’s heads. Development is almost completed in August. For 8 months, the young heads rest in bud stage. Only shortly before flowering, when the bud scales drop down and space is made available, filaments and stigma rays elongate.
Bud development passes through seven steps from the vegetative state to the completion of flower development.
Stage 1: vegetative meristem (VM)
Vegetative meristems are small and flat (Table 1). They segregate leaf primordia in a spiral sequence (Fig. 2a: 1–6). Epidermal hairs appear very early at both sides of the erect standing leaf primordia (Fig. 2b). The hairs become rather long and finally cover the whole surface of the young leaves (Fig. 2c). In the bud, the young leaves are already petiolate and densely pressed together by the surrounding leaves (Fig. 2c).
Meristem size changes during head development
332 ± 56
501 ± 73
557 ± 81
932 ± 237
1192 ± 179
1813 ± 158
342 ± 70
499 ± 77
679 ± 44
979 ± 182
1280 ± 251
1988 ± 190
Stage 2: formation of the floral unit meristem (FUM)
Head development starts when the VM merges into the huge, globose FUM (Fig. 2d–i). The latter is almost threefold larger than the VM (Table 1; Fig. 3: stage 2) and completely naked, i.e. the tissue has expanded before primordia formation. The shape of the FUM is irregular. The stripe pattern on the meristem results from the imprints of the leaf hairs (Fig. 2g–i). At the base of the globose meristem, the two primordia of the showy bracts appear following the spiral phylloaxis of the leaves. They are arranged in almost opposite position and are unequal in size. The larger one is formed first (Fig. 2d–g), but due to the inclined base of the FUM and continuous growth activity, it may be elevated above the inserting line of the smaller one (Fig. 2i). The larger bract has sometimes its tip bent backwards obviously hindered in growth by the dense package of the surrounding leaves (Figs. 2g, h, 4c). It curves inwards and covers part of the meristem (Fig. 2i). The bracts do not form a petiole, neither in bud stage nor later on.
Stage 3: first fractionation forming flower meristems (FMs)
End of June, the FUMs fractionate FMs (Fig. 4). These sub-meristems appear as bulges of different size and shape. The general picture is that the FMs originate very fast and almost simultaneously (Fig. 4a, b, e). The equatorial FMs are larger than the remaining ones, a situation easily misinterpreted as divergent initiation. Sometimes, parastichy-like patterns appear (Fig. 4b, d), but they are not regular. In andromonoecious heads a large meristem part is occupied by the single meristem of the perfect flower (Fig. 4c, d: *). In all FUMs, further FMs appear with delay in basipetal sequence (Fig. 4b, c).
The staminate FMs appear as naked fractions of the FUM without any subtending bracts. They are striped through leaf pressure. The oval shape of some fractions and the basipetal direction of fractionation reflect meristem growth. The FUM is still expanding and reaches an average value of about 500 µm (Table 1; Fig. 3: stage 3), i.e. the one and a half-fold size of the average FUM in stage 2.
The perfect FM is almost radial, flat and a bit higher than the staminate FMs. It also lacks a subtending leaf. It has an oblique position above the equator and is usually directed towards the larger bract (Fig. 4d), but exceptions are possible (Fig. 4c).
Stage 4: expansion of staminate and perfect FMs
Until mid of July, all FMs expand considerably.
The staminate FMs expand from about 110 µm (Fig. 4c, e) to 150 µm (Figs. 5a, 6a) getting a more and more irregular shape. Again, expansion starts mainly at the equatorial level and continues in a divergent sequence (Fig. 5a, b). The irregular shape indicates begin of the second fractionation originating stamen primordia.
The perfect FM expands from about 280 µm (Fig. 4c) to 400 µm (Fig. 6a). It is still radial and flat, but forms a peripheral bulge with increasing size. This bulge will generate the stamen primordia (Fig. 6b).
Stage 5: second fractionation forming floral organ primordia
The second fractionation gives rise to floral organs. This process proceeds almost simultaneously in all FMs and goes along with further FM expansion (Fig. 5a–c). It begins mid of July and lasts only about one week.
In staminate FMs, six stamen primordia arise on average at the meristem margin (Fig. 5b, c). They appear almost simultaneously, but differ in size indicating a sequential order. Most FMs have an oval shape and then, the largest primordia appear at the poles. However, it is difficult to identify a clear pattern of organ initiation. Small FMs at the base of the FUM produce fewer stamens, and large FMs at the equator form additional stamens with increasing space (Fig. 5c: arrows).
The perfect FM also lacks any perianth structures. Instead, the peripheral bulge directly fractionates more than 20 stamen primordia. These appear in almost a single row on the top of the flat meristem but differ in size (Fig. 6b). As the perfect FM continues expansion from about 400 µm (Fig 6b) to 700 µm (Fig. 6c), new space is made available at the peripheral ring, which is immediately used by additional stamen primordia (black arrows). At the same time, the gynoecium starts to develop (Fig. 6c, d). Inside the stamen ring, a huge undulated wall grows out of the naked part of the perfect FM. It corresponds to 6–8(–9) congenitally fused carpel primordia. The number of carpels is given by the number of free tips arising between the inward growing septum corresponding to the congenitally fused flanks of the respective carpels.
Stage 6: floral organ development
Stamen primordia are globose when initiated (Fig. 5c). They differentiate anther structures within the next month. Young anthers look like dumbbells when the two thecae emerge and like typical bisporangiate anthers after pollen-sac formation (Fig. 5d, e). Mid of August, six weeks after FUM formation, male flower development is almost completed.
The gynoecial wall elevates together with the central part of the perfect FM generating cavities in front of the carpel tips (Fig. 6c, d). Very early, i.e. when the septum is still short and the ovary not closed at all, ovules are initiated. One ovule per cavity arises from the central plug (white arrows). Gynoecium development proceeds now also including the proximal zone of the perfect FM below the stamens. This zone forms the synascidiate part of the inferior ovary of the gynoecium (Fig. 7e), which is chambered and characterized by the elongated plug in its center. The distal zone above the stamens forms the symplicate zone in which the carpel walls get into contact, resulting in loculament formation and closure of the gynoecium (Fig. 6e–g). It is followed by the asymplicate zone including the young stylar branches and prospective stigmatic surfaces.
Stage 7: winter bud formation and completion of flower development
Floral organ formation is almost completed in early to mid of August. The young head is densely enveloped and protected by the two bract primordia, which on their side are closely pressed together by the outer leaf primordia and scales. At this time, a bud appears in the axil of the distal leaf primordia (Fig. 7a). It will repeat the circle and develop a head in the next season.
During autumn and winter, the young head rests in the bud. The head shows pressure marks caused by the outer bud structures and the anthers are so densely pressed against each other that the borders among the flowers are completely masked. The different size of the anthers indicates that stamens are either produced successively or do not have enough space to develop.
In March, the buds enlarge and start to open. Shortly before anthesis, when the pressure of the surrounding leaves and scales decreases and space for extension is made available, pseudanthium development is completed by filament elongation, stigma ray formation, internode elongation below the head and bract enlargement.
The flower-like head in D. involucrata meets all criteria of a pseudanthium, i.e. figural subdivision into a showy periphery and a central sexual field composed of de-individualized flowers (Claßen-Bockhoff 1990). In addition, it presents a flower-like organ arrangement with a single ovary (perfect flower) surrounded by many stamens (perfect and staminate flowers). In this respect, it is functionally similar to the cyathium in Euphorbia (Euphorbiaceae) in which the terminal flower is carpellate and the lateral ones staminate (Prenner and Rudall 2007).
Space dependent fractionation in FUM
FUMs differ from IMs in the lack of open growth (Bull-Hereñu and Claßen-Bockhoff 2011a; Claßen-Bockhoff 2016; Claßen-Bockhoff and Bull-Hereñu 2013). Instead, they are able to expand considerably forming huge naked meristems as seen in D. involucrata. The two simultaneous processes, cessation of apical growth and meristem expansion, essentially change the physical conditions of the meristem directly influencing the process of primordia formation (Prusinkiewicz and Barbier de Reuille 2010; Prusinkiewicz and Runions 2012).
Fractionation is usually a fast process. In D. involucrata, FMs are almost simultaneously initiated. They lack subtending bracts, a characteristic which rarely appears in IMs (e.g. Arabidopsis), but is quite common in FUMs (e.g. many Asteraceae and Apiaceae). The reason for lacking bracts may be found in the different meristem qualities. In IMs, bracts and flowers originate from the process of segregation. In contrast, FUMs lack acropetal growth activity and immediately respond to changing geometries. This is clearly shown by the basipetal fractionation of further FMs in D. involucrata, which appear automatically when new space is generated. This process of FM initiation is little understood and resembles the formation of additional structures in FMs (Claßen-Bockhoff 2016; Claßen-Bockhoff and Meyer 2016). In fact, molecular data obtained from Asteraceae (Broholm et al. 2014; Zhao et al. 2016) and results from modelling (Prusinkiewicz and Runions 2012) indicate that FUMs and FMs may be regulated in a similar way. This means, that flowers and pseudanthia (with exceptions, e.g. Actinodium, Claßen-Bockhoff et al. 2013) may not only correspond in phenotype and morphogenesis, but also at the molecular level. These pseudanthia should not longer be derived from inflorescences (e.g. condensed racemes), but taken as flower equivalents with flower-like developmental conditions.
Meristem expansion and variation in FMs
Davidia involucrata varies in number and size of FMs, in number of stamens and in the staminate vs. perfect organization of the FMs. This high degree of variation can be attributed to the spatial conditions of the FUM.
Ongoing meristem expansion either passively widens already existing meristems or generates new space, which can be immediately used by new fractions (Naghiloo and Claßen-Bockhoff 2017). The first concerns the more or less roundish FMs, which become oval-shaped and even deformed by meristem expansion. The latter can be observed in both steps of fractionation resulting, first, in small FMs at the base and top of the meristem and, second, in flowers with a reduced or increased number of stamens depending on position and time of initiation.
The perfect FM is much larger than the staminate FM. It also lacks any rudimentary perianth elements. The finding of perianthless flowers is in agreement with the early observation of Horne (1909) and the SEM investigation of Vekemans et al. (2012). However, it is in conflict with Eyde (1963) and the general pattern in the Cornales. As SEM pictures clearly show the early initiation pattern, it is likely that the assumed presence of perianth structures in staminate flowers is based on a misinterpretation of inhibited stamen primordia as perianth elements (Fig. 5c: arrows).
Perfect FMs only appear in some FUMs and the question is: why? Quantitative morphogenetic studies in Daucus carota L. (Apiaceae; Bull-Hereňu and Claßen-Bockhoff 2010) revealed that umbellets with terminal flowers had larger meristems than those without a terminal flower. Based on this finding, it is likely that also in Davidia involucrata size and shape of the FUM control perfect FM initiation. One may assume that all FUMs are able to form a large, perfect FM but that only those become andromonoecious whose relative proportions allow for its initiation (see also Sokoloff et al. 2006). This hypothesis corresponds to the underlying assumption that fractionation is a self-organizing process (Douady and Couder 1996; Runions et al. 2014) using space completely (Prusinkiewicz and Barbier de Reuille 2010).
Mechanical pressure inside the bud
The heads of D. involucrata develop in summer and overwinter in bud stage. Young leaves and scales surround the expanding FUM pressing all organs closely together. This mechanical pressure leaves stripes on the FUM, hinders the development of the large bract, and influences the number of FMs and floral organs. However, does it also contribute to the oblique position of the perfect flower and the remarkable FM initiation sequence?
SEM pictures clearly illustrate that the FUM has a variable shape and suffers from mechanical pressure. The side opposite to the larger bract grows faster rendering the FUM asymmetrical. On this side, more FMs are fractionated (Figs. 4e, 6b) and already existing FMs get passively stretched (Fig. 4c). It is assumed that the FUM is pressed against the larger bract by the innermost leaf primordium, which stands on the opposite side and is already well developed in the bud.
The perfect FM has an oblique position from the early beginning (Fig. 4c, see also Vekemans et al. 2012). Based on the concept of fractionation, this could be the mere outcome of FUM geometry. However, Moser (1968) observed that the procambial strands end in the perfect flower and concluded that the perfect flower should be interpreted as the terminal flower being secondarily dislocated by unilateral meristem growth. This view is in accordance with the general finding that sexual flower morphs occupy defined positions within an inflorescence (e.g. Diggle 2003; Reuther and Claßen-Bockhoff 2010). Moser (1968) also mentioned that the perfect FM appears the first, a feature often found in terminal flowers, but not confirmed in the present study.
The most interesting feature in D. involucrata is the irregular sequence of fractionation. Both scanning electron microscopy (SEM) and live imaging based on epi-illumination light microscopy (ELM, Jerominek et al. 2014) show that the FMs appear very fast, but that the proximal ones are later initiated than the equatorial ones. While SEM pictures are based on many single meristems not allowing for a clear comparison of the spatio-temporal correlations in FU formation, video documentation of a fractionation in a single living FUM directly illustrates that new space is generated with increasing FUM expansion. This space is immediately used by new FM formation confirming the hypothesis that fractionation is an autonomous, space filling process (Prusinkiewicz and Barbier de Reuille 2010). Interestingly, the proximal primordia in the living meristem, free from all enveloping bud structures, develop faster than the equatorial ones. This conflicts with the observation based on SEM pictures where the proximal FMs also develop with delay and may remain poor in stamen number. Obviously, basipetal FM initiation results from FUM expansion, while the developmental delay of the proximal FMs may be the effect of mechanical pressure in the bud.
Basipetal (or centrifugal) development is rare in FMs (reviewed by Rudall 2010) and FUMs. In flowers, it is often associated with meristem expansion, e.g. in flowers with secondary polyandry (Ronse de Craene and Smets 1992). In FUMs, it appears in Asteraceae, in which ray flowers develop with delay compared to tubular flowers (Bello et al. 2013; Harris 1995; Harris et al. 1991; Zhao et al. 2016). The bidirectional initiation sequence is presently interpreted as the remnant of an underlying cymose branching pattern (Pozner et al. 2012; Zhao et al. 2016). However, it would be worth to discuss the phenomenon in view of the specific FUM conditions and the possible influence of mechanical pressure to the outermost flowers by involucral bracts. As to Davidia involucrata, Moser (1968) assumed that the irregular FM initiation could indicate a cymose origin of the head. However, the present study does not support this assumption.
Petaloid bracts: intermediate structures at intermediate positions
Petaloid bracts are widely distributed in angiosperms. They contribute to pseudanthia formation in about 30 families (Claßen-Bockhoff 1990). Despite their diverse numbers, shapes and colors, they all share the same position in the transitional zone between the vegetative part of the plant and the flowers.
The transitional zone is often characterized by bracts, i.e. by leaf organs which differ in size, shape and/or color from the green leaves of the plants. In Davidia involucrata, the bracts are of similar shape and size, but differ from leaves by their whitish color. Interestingly, young bracts are greenish and lose their chlorophyll only when they rapidly expand. A similar phenomenon is known from plants, which unfold their leaves very fast and present ’juvenile’ colors before chlorophyll synthesis starts and covers these pigments. However, the temporarily yellow bracts in Smyrnium perfoliatum L. (Apiaceae) and some Euphorbia species, which are green before and after anthesis, indicate that inhibition/delay of chlorophyll synthesis is not necessarily connected with rapid growth.
It has been hypothesized that the formation of petaloid bracts is physiologically controlled by basipetal extension of the ‘reproductive program’ (Claßen-Bockhoff 1996; Hempel and Feldman 1994), i.e. by the heterotopic expression of genes controlling petal development (Albert et al. 1998; van Tunen et al. 1993). However, there are only few studies showing floral gene expression in extrafloral structures. CYC-like gene expression was detected in the short shoots mimicking ray flowers in the pseudanthium of Actinodium cunninghamii Schau. (Myrtaceae; Claßen-Bockhoff et al. 2013) and B-class gene expression in some Cornus species (Feng et al. 2012). In Davidia involucrata, Vekemans et al. (2012) identified B- and C-class gene representatives in early stages of bract development. Activity of these genes is more associated with stamen than with petal development, but, nevertheless, it indicates heterotopic floral organ identity gene expression in extrafloral elements. As the bracts also differ from leaves in their specific epidermal cell pattern (Vekemans et al. 2012), they represent structures receiving developmental signals from both the vegetative and flowering parts of the plant.
FUMs are meristems, which share more developmental characteristics with FMs than with IMs. Their discovery has an essential influence on the interpretation of heads and pseudanthia as the constricting corset of inflorescence morphology does not apply any longer. Instead, meristem expansion, fractionation and usage of space play the dominant role.
Recognizing that part of the structures traditionally called inflorescences are floral units allows explaining bidirectional and basipetal FM initiation, presence or absence of subtending bracts and terminal flowers as response to space and mechanical pressure. Accepting fractionation as a self-organizing process facilitates understanding the formation of additional structures in flowers and floral units.
Although FUMs are already described in at least 10 angiosperm families (Claßen-Bockhoff and Bull-Hereňu 2013; Stützel and Trovó 2013), knowledge about their existence, development, genetic regulation and evolution is almost lacking. Fortunately, molecular studies in Gerbera mutants have recently shown that ectopic expression of GhUFO converts the head into a flower and that GhLFY appears to be involved in the establishment of the naked stage of FUMs (Zhao et al. 2016). Both findings support the view that FUMs are different from IMs and close to FMs.
The most important consequence of realizing the existence of FUMs is the option to interpret the evolution of inflorescences in an alternative way. Heads should not be interpreted as condensed racemes without morphogenetic confirmation, bidirectional initiation patterns do not necessarily indicate an underlying cymose branching pattern and absence of terminal flowers can also be explained by spatial conditions. All these alternative interpretations, even if they question a morphological paradigm, should be critically tested to contribute to a deep understanding of inflorescence diversity and diversification.
We thank Akitoshi Iwamoto (Tokyo) and Kester Bull-Hereñu (Santiago de Chile) for initiating this special issue on flowers. The first author is very grateful to the Botanical Society of Japan for the invitation and financial support to participate at its annual congress at Okinawa 2016. We thank Bernd Mengel (BG Mainz) for collecting Davidia buds throughout the year, Madeleine Junginger (Mainz) for measuring the primordia at the SEM pictures and Maria Geyer (Mainz) for working over the illustrations. Special thanks go to Margerita Remizowa (Moscow) and an unknown reviewer for stimulating discussion.
- Albert VA, Gustafson MHG, Di Laurenzio L (1998) Ontogenetic systematics, molecular developmental genetics, and the angiosperm petal. In: Soltis DE, Soltis PS, Doyle JJ (eds) Molecular systematics of plants. II. DNA sequencing. Kluwer, Boston, pp 1–25Google Scholar
- Bello MA, Cubas P, Álvarez I, Sanjuanbenito G, Fuertes-Aguilar J (2017) Evolution and expression patterns of CYC/TB1 genes in Anacyclus: Phylogenetic insights for floral symmetry genes in Asteraceae. Front Plant Sci 8:589. https://doi.org/10.3389/fpls.2017.00589 CrossRefPubMedPubMedCentralGoogle Scholar
- Burr B, Barthlott W (1993) Untersuchungen zur Ultraviolettreflexion von Angiospermenblüten.II. Magnoliidae, Ranunculidae, Hamamelididae, Caryophyllidae, Rosidae. Trop subtrop Pflanzenwelt 87, Akad Wiss Lit Mainz. Steiner, StuttgartGoogle Scholar
- Claßen Bockhoff R (1996) A survey of flower-like inflorescences in the Rubiaceae. Opera Bot Belg 7:329–367Google Scholar
- Claßen-Bockhoff R (1991) Anthodien, Pseudanthien und Infloreszenzblumen. Beitr Biol Pfl 66:221–224Google Scholar
- Diggle PK (2003) Architectural effects on floral form and function: a review. In: Stuessy T, Hörandl E, Mayer V (eds) Deep morphology: toward a renaissance of morphology in plant systematics. Koeltz, Konigstein, pp 63–80Google Scholar
- Douady S, Couder Y (1996) Phyllotaxis as a dynamic self organizing process. J Theor Biol 139:178–312Google Scholar
- Eyde RH (1963) Morphological and paleobotanical studies in Nyssaceae. J Arnold Arboretum 44:1–54Google Scholar
- Fang WP, Chang CY (1983) Flora Republicae Popularis Sinicae, vol 52. Science, BejingGoogle Scholar
- Froebe HA, Ulrich G (1978) Pseudanthien bei Umbelliferen. Beitr Biol Pfl 54:175–206Google Scholar
- Good R (1956) Features of evolution in the flowering plants. Longman, Green & Co, LondonGoogle Scholar
- Horne AS (1909) The structure and affinities of Davidia involucrata Baill. Trans Linn Soc Bot 2 Ser 7:303.367Google Scholar
- Jahnke C (1986) Der Infloreszenzbau der Cornaceen sensu lato und seine systematischen Konsequenzen. Trop subtrop Pflanzenwelt 57, Akad Wiss Lit Mainz. Steiner, StuttgartGoogle Scholar
- Johow F (1884) Zur Biologie der floralen und extrafloralen Schauapparate. Jb Kgl Bot Gart Berlin 3:47–68Google Scholar
- Kubitzki K (2004) Cornaceae. In: Kubitzki K (ed) The families and genera of vascular plants. VI. Flowering plants dicotyledons. Celastrales, Oxalidales, Rosales, Cornales, Ericales. Heidelberg. Springer, Heidelberg, pp 82–90Google Scholar
- Leppik EE (1969) Morphogenetic classification of flower types. Phytomorphology 18:451–466Google Scholar
- Manchester SR (2003) Leaves and fruits of Davidia (Cornales) from the Paleocene of North America. Syst Bot 27:368–382Google Scholar
- Moser V (1968) Der Blütenbau der angeblich verwandten Gattungen Davidia und Camptotheca. Vierteljahrsschr Nat forsch Ges Zür 113:157–184Google Scholar
- Troll W (1928) Organisation und Gestalt im Bereich der Blüte. Springer, BerlinGoogle Scholar
- van Tunen AJ, Eikelboom W, Angenent GC (1993) Floral organogenesis in Tulipa. Flower Newsl 116:33–38Google Scholar
- You H, Fujiwara K, Liu Y (2014) A preliminary vegetation-ecological study of Davidia involucrata forest. Nat Sci 6:1012–1029Google Scholar