TAXON 63 (5) • October 2014: 1103–1111 Ronse De Craene & al. • Floral formulae

METHODS AND TECHNIQUES

Understanding the structure of —The wonderful tool of floral formulae: A response to Prenner & al.

Louis Ronse De Craene,1 Akitoshi Iwamoto,2 Kester Bull-Hereñu,3,4 Patricia Dos Santos,1 Javier A. Luna1,5 & Jennifer Farrar1,5

1 Royal Botanic Garden Edinburgh, Edinburgh EH3 5LR, Scotland, U.K. 2 Department of Biology, Tokyo Gakugei University, Tokyo, Japan 3 Escuela de Pedagogía en Biología y Ciencias, Universidad Central de Chile, Santiago, Chile 4 Departamento de Ecología, Pontificia Universidad Católica de Chile, Santiago, Chile 5 Institute of Molecular Sciences, School of Biological Sciences, University of Edinburgh, Edinburgh, EH9 3JH, U.K. Author for correspondence: Louis Ronse De Craene, [email protected] ORCID: LRDC, http://orcid.org/0000-0002-8333-4596; AI, http://orcid.org/0000-0001-6492-495

DOI http://dx.doi.org/10.12705/635.35

Abstract This paper is a discussion and elaboration of a paper by Prenner & al. (2010), entitled “Floral formulae updated for routine inclusion in formal taxonomic descriptions”. The aim of the Prenner paper was to promote the use of floral formulae in and to reach a consensus among botanists for best practice. An important purpose of floral formulae is to induce users to observe and describe flowers accurately. It is proposed that additional information on anther, , style and should be included. Also, only visible organs should be included in a formula and theoretical speculations should be illustrated with floral diagrams, which are complementary to formulae, unless there is good reason to include absent organs. We propose a universal, standardized method to accurately shorthand a description of a . The level of detail given in the formula can be highly flexible and depends on the intentions of the user.

Keywords diagnostic characters; floral diagrams; floral formulae; floral

INTRODUCTION to describe flowers. Prenner & al. (2010: 241) also suggested that floral formulae should be an inherent component of plant Flowers are remarkably conservative in their numbers diagnoses and taxonomic descriptions, “functioning as a logi- of parts. Whichever flower of Erodium (Geraniaceae) a keen cal phenotypic counterpart to the DNA barcode”. It is clear that observer will look at (see Fig. 1A), he will always find the same floral formulae have a clear advantage over lengthy descrip- number of whorls and the same number of organs within each tions, as the structure of a flower can be presented in a suc- of the flower (five and , two whorls of five cinct way, accessible to any reader. As such floral formulae , one of which is sterile, and five carpels: Fig. 1A). are a powerful incentive for consistent observations and also Flower morphologists have sought for ways to succinctly a valuable teaching tool (Prenner & al., 2010). Prenner & al. describe flowers, and a perfect tool was conceived in the form of advocate the use of symbols that can be reproduced easily with floral formulae. The first floral formulae were created in the first a non-sophisticated computer, enabling its universal use. decades of the nineteenth century and have been progressively Floral diagrams, two-dimensional representations of the developed especially in Germany. For an overview of the history ground-plan of the flower, are the major counterpart to flo- of the use of floral formulae we refer to Prenner & al. (2010). ral formulae in floral descriptions. As discussed in Ronse De Although floral formulae (together with floral diagrams) Craene (2010), floral formulae have a disadvantage against have been integral part of the basic teaching in botany, the floral diagrams in that it is difficult to describe the position recent depreciation of botany at universities has not helped in of organs in the flower, and impossible to show any spatial the recognition of the value of this methodology. Few recent arrangements or any conspicuous special structures, such as papers or textbooks use floral formulae, often in a very simpli- nectaries, coronas or appendages. However, floral formulae fied and frequently inconsistent form, such as Judd & al. (2002), have the advantage that they do not necessitate any pictorial Stützel (2006), Tsou & Mori (2007), Leins & Erbar (2010), or means. They also open possibilities for comparative research Simpson (2010). and the understanding of trends in flower . The value of floral formulae has been highlighted by We agree with Prenner & al. (2010) that floral formulae Prenner & al. (2010) who proposed an updated list of symbols (as well as floral diagrams) are under-used tools in systematic

Received: 10 Mar 2014 | returned for first revision: 15 Apr 2014 | last revision received: 9 Jun 2014 | accepted: 9 Jun 2014 | published online ahead of inclusion in print and online issues: 24 Sep 2014 || © International Association for Plant (IAPT) 2014

Version of Record (identical to print version). 1103 Ronse De Craene & al. • Floral formulae 63 (5) • October 2014: 1103–1111

Fig. 1. Examples of flowers to illustrate different floral formulae. A, flower of Erodium corsicum Léman (Geraniaceae) with five equal petals. The abaxial side of the flower is below. B, flower of cotyledon (S.Watson) B.L.Rob. (). The flower consists of two median bracteoles and a variable number of (sepals), consisting of two outer lateral sepals (asterisks) and a variable number of median sepals. C, flower of Penstemon sp. (Plantaginaceae) laid open, showing the arrangement of different parts. The androecium consists of two pairs of sta- mens of different length. D, lateral view of flower of Blumenbachia hieronymi Urb. (Loasaceae). Note the white petals alternating with coloured complexes, which consist of five . E, large complex flower of Nymphaea sp. (Nymphaeaceae) showing an arrangement of alternating whorls that can be detected in the sepals (white numbers) and outer petals (black numbers); the inner whorls are less clear but androe- cium and carpels are arranged in ten orthostichies. F, asymmetric flower of Vigna linearis (Kunth) Maréchal & al. (Leguminosae). The abaxial (keel) petals are transformed into a coil that holds androecium and twisted style.

1104 Version of Record (identical to print version). TAXON 63 (5) • October 2014: 1103–1111 Ronse De Craene & al. • Floral formulae

research. Therefore, we believe that a broader use of these tools structures of unknown/unclear origin), K for calyx, Corolla, should be encouraged. Ronse De Craene (2010: 39) came up Perigon, Androecium, ), and a use of lower case with a list of abbreviations comparable to Prenner & al. (2010), letters for specific structures that are part of organs, such as although it was less elaborate. However, the advantages of flo- locules, (including their position on the placenta), ral formulae for educational purposes are clear and lecturers sacs, styles and stigmas. may adapt a personal approach (E. Smets, pers. comm.). As The inclusion of details of the flower, such as the number suggested by Prenner & al. (2010), their proposed system is of pollen sacs or locules forces the user to accurately observe open for discussion, and Taxon is probably a good medium to the flower to record as many details as possible. Not including start this discussion. The ultimate aim is to reach consensus these details in a consistent manner may lead to inconsisten- among botanists for the use of a balanced and universal system. cies in the recording of data, either by implying the presence of floral characters, when these have never been checked, or by assuming that the user will always know the relevant A REEVALUATION OF FLORAL FORMULAE characters. Several qualitative features can be shown through superscript symbols and Prenner & al. use this to great effect. The following points are a criticism of the paper by Prenner For example, C stands for the number of petals, while c stands & al. (2010) and suggestions for improvement. We also refer to for petaloidy, which can be used as a special feature of the Table 1 for clarity. perigon. A calyx can often be pigmented, as, e.g., in Fuchsia The floral formulae proposed by Prenner & al. (2010) are (Onagraceae) and this can be shown as Kc. Pc stands for a very detailed and contain more information than classical for- petaloid perigon, when it is not possible to differentiate sepals mulae, including extrafloral organs associated with flowers, from petals, as in tulips (Weberling, 1989). such as and bracteoles, and gynoecial characters, such Inclusion of bracts and bracteoles in floral formulas is use- as the and ovules. However, it is best to maintain ful, although strictly speaking they belong to the . a hierarchical arrangement with subcategories. Treating parts Bracteoles are generally paired in and single in mono- of organs (i.e., ovules) at the same level as the organs leads to cots (Ronse De Craene, 2010). However, in some cases with a confusion and makes the floral formulae appear too complex. multitude of bracts surrounding the flower a strict distinction A better approach includes a restriction of capital letters to between bracts and bracteoles is not possible (e.g., Reaumuria the main organs associated with the flower (see Table 1) , in Tamaricaceae: Ronse De Craene, 1990). Boundaries of flow- Bracteole, Epicalyx (i.e., agroup of structures below the flower ers can be vague, with no clear separation of bracts and sepals resembling an extra calyx; this can be a whorl of bracts or (e.g., Strasburgeria in Strasburgeriaceae; Clusia in Clusi­aceae),

Table 1. Proposal for an improved use of symbols in floral formulae (partly based on Prenner & al., 2010 and Ronse De Craene, 2010). Symmetry   Median monosymmetry   Transverse monosymmetry     Oblique monosymmetry  Disymmetry  Polysymmetry  Asymmetry  Spiral flower Organs B Bract Bt Bracteole E Epicalyx (generally of bract origin but occasionally stipular) K Calyx (number of sepals) (superscript = sepaloid organ) C Corolla (number of petals) (superscript = petaloid organ) P Perigon (number of tepals; no differentiation between calyx and corolla) A Androecium (number of stamens) (superscript ∞ or definite number refers to fascicles) Aobd Obdiplostemony Aobh Obhaplostemony G Gynoecium (number of carpels) G -G- G Position of : superior, half-inferior, inferior

Version of Record (identical to print version). 1105 Ronse De Craene & al. • Floral formulae TAXON 63 (5) • October 2014: 1103–1111

or with shifts of bracts within the limits of flowers in the form to be preferred. The symmetry symbol is placed before the epi- of an epicalyx (e.g., Hibiscus in Malv­aceae, Dipsacus in Cap- calyx members to show that the epicalyx is clearly connected rifoliaceae). In the case of Thunbergia () and to the flower (e.g., Lewisia, Potentilla: Fig. 1B; Table 2). This Calystegia (Convolvulaceae) the two large bracteoles behave is also shown for Meliosma, where bracteoles and sepals are as an epicalyx enclosing the flower but they are still distinc- impossible to differentiate from each other (Table 2; Wanntorp tive organs (Weberling, 1989). In some cases the epicalyx may & Ronse De Craene, 2007). even replace the original calyx (e.g., Thunbergia, Acanthaceae: Prenner & al. (2010) also include the placentation and Schönenberger & Endress, 1998). An epicalyx can have differ- ovules in the floral formulae. This information is generally ent origins (possibly stipular in Rosaceae and bract-derived missing from formulae, which are restricted to the organs and in Malvaceae: Eichler, 1878; Weberling, 1989), or the origin is their number, and in the case of carpels include their position debatable (e.g., Lythraceae: Cadet, 1954; Mayr, 1969). In cases (superior, G; half inferior, -G-; inferior, G ). Prenner & al. use a where the epicalyx is clearly developed, the use of the letter E is single symbol V for describing the number of ovules and refer

Table 1. Continued. Elements of organs lo Number of locules (allows for recognizing false septa) ov Ovule number per ovary ova Apical placentation ovb Basal placentation ovc Free-central placentation ovl Laminar placentation ovm Marginal placentation ovp Parietal placentation ovx Axile placentation ovx-p Both axile and parietal present in the same ovary ps Number of pollen sacs per anther psl Longitudinal dehiscence psp Poricidal dehiscence pst Transversal dehiscence psv Valvular dehiscence sty Number of styles stya Apical style(s) styg Gynobasic style(s) sti Stigma number (whether multilobed or single) stia Apical stigma (not possible to define its position relative to the carpels) stica Carinal stigma (opposite the carpels) stico Commissural stigma (alternating with the carpels) Additional information + Indicates when several whorls of the same organ can be distinguished : Indicates differences within an organ whorl Fusion of organs at three different levels: ( ), within the same whorl; [ ], involving two different whorls; { }, involving three differ- ( ) [ ] { } ent whorls. Symbols used for organs reflecting mainly congenital fusion or hypanthial growth ^ Represents a calyptra or fusion of pollen sacs and stigmas or styles; e.g., C^ ° Refers to organ reduction (can be used for the whole organ category, e.g. G°, or for part of the organs (e.g., A5° + 5) – Refers to a range of organ number. The most common number is shown in bold (e.g., K4–5–6) ∞ Many organs, when definite numbers cannot be counted or are unknown ♀ Female (pistillate) flower ♂ Male (staminate) flower ®

1106 Version of Record (identical to print version). TAXON 63 (5) • October 2014: 1103–1111 Ronse De Craene & al. • Floral formulae

Table 2. Examples of floral formulae of flowers of diverse families encompassing variable structures. Note that the floral formulae proposed in this table use a large number of possibilities for the description of the flower, but their ultimate application depends on the needs of users. name Family Reference Amborella trichopoda Amborellaceae ♂ B1 Q Bt2 Pc9–11 A12–21 psl2:2 G0 Buzgo & al., 2004; Ronse De Baill. ♀ B1 Q Bt2 Pc7–8 A°0–2 G4–5–6 lo1 sty0 stica1 ovb1 Craene, 2010

Nymphaea sp. Nymphaeaceae B1 Kc2+2 Cc4+8+4+4 A∞ psl2:2 -G-10–18 lo1 Fig. 1E; Ronse De Craene, 2010 sty0 stica1 ovl∞

Laurus nobilis L. Lauraceae ♂ B1 Pc2+2 A4+2+2+4 psv1:1 G0 L.P. Ronse De Craene, pers. obs.; ♀ B1 Pc2+2 A°2 G1 lo1 stya1 stica1 ova1 Watson & Dallwitz, 1992+

Canna indica L.* Cannaceae B1 Bt1 7 Kk(3) [Cc3 Ac1°–3°+11⁄2 :1⁄2°:2°] psl1/1°:0 Ronse De Craene, 2010; Almeida & G(3) lo3 stya1 stia1:1° ovx∞ al., 2013

Meliosma veitchiorum Sabiaceae B1 Bt2 Kk3 Cc2:3 A2:3° psv1:1 G(2) lo2 stya2 Wanntorp & Ronse De Craene, 2007; Hemsl. stica2 ovm2 Ronse De Craene & Wanntorp, 2008

Lewisia cotyledon Montiaceae B1 Bt2 /n E2 Kc8–13 A6–8–11 psl2:2 G(2–3) lo3 Fig. 1B; Dos Santos & Ronse De (S.Watson) B.L.Rob. stya1 stico3 ovfc2 Craene, in prep.

Potentilla fruticosa L. Rosaceae B1 Bt2 Ek5 Kk5 Cc5 A10+5+5+10 psl2:2 G∞ lo1 Lindenhofer & Weber, 2000 stya1 stica ovb1

Vigna linearis (Kunth) Papilionoid B1 Bt2 7 Kk(2:3)Cc2^7 :2:1 A(9:1) psl2:2 G1 lo1 Fig. 1F; Dwyer & al., 1980 Maréchal & al. Leguminosae stya7 1 stica1 ovx10

Chamaecrysta desvauxii Caesalpinioid B1 Bt2 7 Kc5 Cc4:1 A5+5 psl2:2 G1 lo1 stya1 stica1 Costa & al., 2007; A. Iwamoto, (Colladi) Killip Leguminosae ovx11–13 pers. obs.

Mimusops elengi Wight B1 Kc4+4 [Cc4+4 A8°+8] psl2:2 G(8) lo8 stya1 Kümpers & al., subm. stica8 ovx1

Fremontodendron califor- Malvaceae B1 Bt2 Kc5 C0 (A5°+52) pst2 G(5) lo5 stya1 stica1 Von Balthazar & al., 2006; Van Heel, nicum (Torr.) Coville** ovx∞ 1966

Blumenbachia hieronymi Loasaceae B1 Kk5 Cc5 A3°:2°:∞ psl2:2 G(3) lo1 stya1 stia1 Fig. 1D; Hufford, 1990 Urb. ovp∞

Ceratophyllum Ceratophyllaceae ♂7 Q Pk7–12 A∞ psl2:2 Iwamoto & al., 2003; in prep. demersum L.*** ♀ Pk5–9 G1 stya1 stia1 ova1

* The fertile stamen consists of one theca, half of which is petaloid and the other half is fertile; the other theca is aborted (Almeida & al., 2013). Therefore only one pollen sac is fertile. ** The sterile stamen is fused with two antepetalous stamens in a triplet resembling an ordinary, but highly fragmented stamen (von Balthazar & al., 2006). *** The of Ceratophyllum is questionably present and could represent a series of bracts.

Version of Record (identical to print version). 1107 Ronse De Craene & al. • Floral formulae TAXON 63 (5) • October 2014: 1103–1111

to the placentation as a secondary condition by an additional certain developmental stage but are absent in mature flowers lowercase character (e.g., Va, for apical placentation). They use (e.g., Choob, 1999), has meaning for developmental floral for- this symbol instead of the symbol O as they believe it avoids mulae (see below), but not for descriptions of mature flowers. confusion with a zero. However, the use of V for ovules is Reduced organs obviously show a progression in their reduc- counterintuitive. The use of lowercase and the symbol “ov” tion, as staminodes can be sterile stamens with aborted anthers, removes this difficulty. Instead of describing the placentation or small, barely developed stubs (Ronse De Craene & Smets, as a second lowercase character, we advocate the use of super- 2001; Ronse De Craene & al., 2002). If there is sufficient evi- script, such as ovp for parietal placentation. dence for a staminodial nature, any stubs should be referred The description of the flower is also incomplete and remains to as A°. of are best described as inconsistent by only describing ovules without mentioning the sterile stamens. In unisexual flowers the lost carpels or stamens anthers with pollen sacs. Although anthers are generally dithe- can be represented in the floral formulae, as done in Table 2. cal and tetrasporangiate in the majority of angiosperms with However, this decision lies with the user in the context of floral two pollen sacs in each theca (symbol “ps”, shown as ps 2:2), evolution. variation should be acknowledged, such as loss of one theca (ps A number of abbreviations in Prenner & al. (2010) are 2:0) or loss of one pollen sac in each theca (ps 1:1). Therefore for misleading or insufficiently clear (see Table 1). For example, the sake of consistency, we propose the use of “ps” for pollen oblique monosymmetry is shown by the symbol “ø”, but sacs as opposed to “ov” for ovules. Dehiscence of pollen sacs this does not convey any information about the orientation can also be shown, as this is an important diagnostic character: of monosymmetry. We advocate the use of different arrows psl for lateral dehiscence, psv for valvate dehiscence, pst for () reflecting the direction of oblique monosymmetry. transversal dehiscence, and psp for poricidal dehiscence. Stützel (2006) followed a similar approach. The same applies Although details provided by Prenner & al. (2010) enrich for disymmetry, using “  ” instead of “ + ”. Prenner & al. (2010) our understanding of the flower structure, the list is not com- use a separate symmetry symbol for each type of organ in the plete and there is room to add further details of style and flower. Symmetry obviously varies in the flower, but several stigma. Style number (symbol “sty”) can be variable reflect- reasons can be responsible for this, including gravity and loss ing the number of carpels or styles can be single. The vertical of organs (see Endress, 1999). For example, what can be done insertion of the style (apical or gynobasic) can be shown with with cases where stamens are inserted in a regular whorl, but the superscript characters stya and styg. While a style can hang down on one side of the flower? Is this to be described be either present or absent, stigmatic (symbol “sti”) is as a monosymmetric or a polysymmetric androecium as, e.g., most often developed. The horizontal insertion of styles and in several Amaryllidaceae? Flowers may also change their stigmas relative to the carpels can be carinal (i.e., opposite the symmetry by late developmental processes (Endress, 1999; carpels, shown with stica) or commissural (i.e., alternating Tucker, 1999). In the case of Vigna (Leguminosae) flowers with the carpels, shown with stico). When the stigma is apical become asymmetric due to a coiling of the keel petals and (i.e., terminal without clear association with the carpels), this style (Fig. 1F). In Chamaecrista and Senna (Leguminosae) can be shown as stia. the flower is asymmetric by unequal growth of petals and sta- Eichler (1875–1878) distinguished between empirical and mens and a twist in the ovary, starting early in development theoretical floral formulae; the former being a description of (Tucker, 1999). This regulates specific pollen deposition on what one actually sees in a flower and the latter can be a reflec- insects (Endress, 1999). A regular pentamerous flower often tion of an evolutionary hypothesis. Stützel (2006) and Prenner has a trimerous gynoecium with a single symmetry plane as & al. (2010) also used theoretical formulae in their distinction in Polemoniaceae and Caryophyllaceae. of organ reduction (with superscript r) and organ loss (with Describing a different monosymmetry for each whorl obvi- superscript 0). Ronse De Craene (2010) used the symbol “ ° ” ously adds to the complexity of the floral formula. We suggest for reduced or sterile organs, such as staminodes (A°) and car- restricting the symbol of symmetry to the whole flower, as we pellodes (G°). To include absent organs in a floral formula prefer to adhere to the overall perception of symmetry. How- makes sense only when there is clear evidence for this, such as ever, it should be open to the user if they should wish to add a absence of petals in a group where petals are generally present different symbol for each whorl, especially in cases where there (e.g., Fremontodendron californicum in Malvaceae: Table 2). is a clear switch in symmetry between different organs. There In the example of orchids used by Prenner & al., the androe- are several cases where the flower has a basically symmetrical cium is presented as A3°+2:1°, following the hypothesis that groundplan, but where the flower is weakly monosymmetric the androecium has been derived from the monocot ancestral (e.g., Rhododendron, Epilobium, Erodium). It is best to describe condition A3+3. While this is probably true, a medium such these flowers as polysymmetric. Any indications of monosym- as a is more appropriate to show organ losses, metry in different whorls can be shown by the use of a colon. where an asterisk shows the position of the lost organ. Depend- This is an important symbol, which shows differences in shape, ing on the context, we advocate to show only organs that are including partial sterility of the androecium and gynoecium. visible in the mature flower, even if they are present in early This was done independently by Ronse De Craene (2010) and developmental stages and are lost at maturity (e.g., the abaxial Prenner & al. (2010) and is particularly useful to show the of Melianthus: Ronse De Craene & al., 2001). Including orientation of petal lobes in a bilabiate flower. For example, so-called phantom organs, viz. organs that may appear at a some Erodium show distinctive spots on the two adaxial petals

1108 Version of Record (identical to print version). TAXON 63 (5) • October 2014: 1103–1111 Ronse De Craene & al. • Floral formulae

(E. leucanthemum Boiss., E. ×willkommianum Sünd.) or a dif- elements, although additional data complete the description of ference in size (E. foetidum (L.) L’Her.) and this can be shown the flower that can be used as a shortened diagnosis (cf. Prenner as C3:2 (L.P. Ronse De Craene, pers. obs.). Therefore C3:2 & al., 2010). The complexity of the floral formulae depends indicates an arrangement of a lower lip with three petal lobes on the needs of the user. We refer to Table 2 for examples of and an upper lip with 2 lobes (e.g., Penstemon, Fig. 1C). flowers with occasionally difficult structures. Contrary to Donoghue & al. (1998) or Prenner & al. (2010) we do not visualize the sequence from the adaxial (dorsal, pos- terior) to the abaxial (ventral, anterior) side of the flower, but ADDITIONAL COMPLICATIONS FOR from the abaxial to the adaxial, as this is the logical direction FLORAL FORMULAE for a approaching the flower and also reflects the sequence of initiation of the petals in many monosymmetric Floral formulae are not static and could be utilized for flowers (e.g., Tucker, 1999 for Fabaceae; Endress, 1998, 1999 for developmental studies. Changes in the morphology of flowers ). “Anterior” linguistically also reflects a precedence during development occur in several groups of , especially in time and space. A partly sterile androecial whorl inserted in the symmetry of the flower (e.g., Endress, 1999; Tucker, 1999; on different levels can be shown as A2:2:1° as is a common Ronse De Craene & Smets, 2001; Ronse De Craene & al., 2001; condition in Lamiales (e.g., Penstemon, Fig. 1C). Olson, 2003; Iwamoto & al., 2003; Patchell & al., 2011). This Pseudomonomery reflects the presence of a single fertile can be highlighted in floral formulae, as has been done for the carpel out of multiple carpels and refers to an incompletely floral diagram of Tropaeolum by Ronse De Craene & Smets fertile ovary (see Eckardt, 1937; Ronse De Craene & Smets, (2001). Tsou & Mori (2007) used floral formulae to describe 1998). This is an important diagnostic feature of some families different stages in the development of the complex flowers of such as Anacardiaceae expressed as G(2°:1), where only a single Lecythidaceae. Leins & Erbar (2010) used symbols to show the carpel out of three is fertile, while two locules are empty (e.g., direction of stamen development in multistaminate flowers. Bachelier & Endress, 2009). A full gradation is possible, rang- Indeed, the floral formulae are a useful tool to describe flow- ing from three fully fertile carpels to a single carpel without ers at different stages of their development and the sequence any traces of the other two. Apart from Anacardiaceae, this is of initiation of organs, although they should be used for spe- characteristic for Restionaceae (Ronse De Craene & al., 2002). cific comparative purposes, not involving mature flowers. For The representation of the position of stamens in the flower example, a whorled centrifugal sequence of stamen initiation is a feature of great systematic importance. In flowers with can be presented and read in a reversed order in a floral for- two stamen whorls a distinction can be made between dip- mula (e.g., Capparis: Ronse De Craene & Smets, 1997). The lostemony (the outer whorl inserted opposite the calyx) and gynoecium can also arise before the androecium, as described obdiplostemony (the outer whorl inserted in antepetalous or in a number of families (e.g., Ge & al., 2007; Ronse De Craene, alternisepalous position). In case of a single stamen whorl, it is 2011; Wanntorp & al., 2011). Studies in floral development are either inserted opposite the calyx (haplostemony) or opposite useful in understanding the structure of the flower, which can the corolla (obhaplostemony; alternisepalous). Obhaplostemony be obscured at maturity. Especially with large flowers having and obdiplostemony also have strong phylogenetic significance numerous organs an early developmental sequence can help as characters (see Ronse De Craene & Smets, 1995; Ronse De in understanding the number and arrangement of organs with Craene, 2010). Although we agree with Prenner & al. (2010) that complex phyllotaxis, facilitating a depiction with floral formu- it is important to show the anomalous condition of obdiploste- lae (e.g., Nymphaea in Nymphaeaceae: Fig. 1E; Anaxagorea mony in the flower, a left-right arrow (n) does not clearly show in Annonaceae: Endress & Armstrong, 2011; Corbichonia in the number of whorls and can be confused with the symbol of Lophiocarpaceae: Ronse De Craene, 2010; Brockington & al., disymmetry; in addition, if other organs are arranged as a series 2013). of whorls with “+”, this makes little sense. It is better to use Another important issue with constructing floral formu- “OBH” for obhaplostemony and “OBD” for obdiplostemony in lae is the interpretation of visually striking flower structures, superscript after the symbol for the stamens (Aobh and Aobd). such as receptacular cups or hypanthia. There is an abundant This allows for the use of stamen position to reconstruct the literature dealing with the controversy of floral cups being the spatial understanding of flowers, even in the absence of petals result of fusions of organs or the result of zonal growth lifting or sepals. organs into a common structure (see Weberling, 1989; Leins An extreme case of a complex flower is Canna (Cann­ & Erbar, 2010). The distinction between congenital fusion and aceae, see Ronse De Craene, 2010; Glinos & Cocucci, 2011; postgenital fusion of organs is intimately linked to this dis- Almeida & al., 2013) with an androecium derived from two cussion (see Sattler, 1978; Leins & Erbar, 2010). As hypanthia whorls (A3+3): the outer whorl is restricted to a single sta- or receptacular outgrowths are highly variable in flowers, it minode and the inner whorl consists of two staminodes with is preferable to limit the use of floral formulae to what can the fertile part of the androecium restricted to half of a theca: actually be observed in the flower as a connection between A1°+11⁄2 :1⁄2°:2°. different organs. The use of three kinds of brackets allows for In the case of Erodium (Fig. 1A), a complete floral formula additional detail although it does not necessarily reflect fusion would be as follows: B1 Bt1 Kk5 Cc5 Aobd(5+5°) psl2:2 G(5) (see Table 1). Postgenital fusion can occasionally be shown lo5 stya1 stica5 ovx5. It is not necessary to include all possible by the symbol “ ^ ”, as the connivent anthers in Asteraceae

Version of Record (identical to print version). 1109 Ronse De Craene & al. • Floral formulae TAXON 63 (5) • October 2014: 1103–1111

and Asclepias (Asclepiadaceae) (A5^) or the development of a powerful tool to accurately describe and compare floral mor- calyptrate corolla in Vitaceae (C5^). Prenner & al. (2010) use the phologies. Presenting exact floral formulae requires a detailed brackets in subscript and superscript to show the difference in investigation of the flower by the user and is an excellent incen- the level of fusion, as in Asteraceae with the common stamen- tive to observe flowers in the field and in the classroom. ( ) petal tube shown by [ ] and the fusion of the anthers by . Although there are several parallel systems for building Nectaries or other appendages (corona appendages, obvi- floral formulae (e.g., the American versus European system), ous concentrations of tufts of ) are generally omitted we emphasize the need of a universal language that is clear and from floral formulae, although they can be highly important concise. A consensus among botanists is a necessary require- visual cues in flowers (e.g., the corona of : Waters ment to make the use of floral formulae really worthwhile. We & al., 2013). Unfortunately, nectaries or coronas are too variable believe that this paper is a step in the right direction. in position and shape to be incorporated in purely descriptive floral formulae, as they would add complexity to descriptions. The use of floral diagrams is complementary in this way. ACKNOWLEDGEMENTS Another problem is the occasional loss of one of two peri- anth whorls, leaving a single petaloid perianth, which can either We thank RBGE for providing facilities to organize a Floral Mor- be derived from a calyx (e.g., in Daphne of Thymelaeaceae) phology Discussion Group, which led to the development of this paper. or from a corolla (as in Osyris of Santalaceae). This difficulty can be solved in the context of the knowledge of the phylogeny c c of plants in question, where K 4 C0 and K0 C 4 can be used LITERATURE CITED respectively, or a neutral approach can be advocated by using c the symbol P for tepals. We advocate the use of superscripts Almeida, A.M.R., Brown, A. & Specht, C.A. 2013. Tracking the devel- c and k on a general basis, even if the organs are obviously opment of the petaloid fertile stamen in : Insights into -like or petal-like. With a broader use of floral formulae the origin of androecial petaloidy in the . AoB Plants in descriptions or diagnoses, this information is important. 5. http://dx.doi.org/10.1093/aobpla/plt009 Prenner & al. (2010) also advocate writing out the name of Bachelier, J.B. & Endress, P.K. 2009. Comparative floral morphology and anatomy of Anacardiaceae and Burseraceae (Sapindales), with a particular structure in full when it is highly distinctive in a a special focus on gynoecium structure and evolution. Bot. J. Linn. flower (e.g., labellum, pappus). This makes the floral formulae Soc. 159: 499–571. http://dx.doi.org/10.1111/j.1095-8339.2009.00959.x longer and more complex and could be inconsistent due to the Brockington, S.F., Dos Santos, P., Glover, B. & Ronse De Craene, great diversity of floral forms. It will be of little value to the L.P. 2013. Evolution of the androecium in : Insights user if he is not aware of the structure and different authors from a paraphyletic Molluginaceae. Amer. J. Bot. 100: 1757–1778. sometimes use different names for the same structure; at the http://dx.doi.org/10.3732/ajb.1300083 Buzgo, M., Soltis, P.S. & Soltis, D.E. 2004. Floral developmental same time users familiar with the structure would find this morphology of Amborella trichopoda (Amborellaceae). Int. J .Pl. information superfluous. Sci. 165: 925–947. http://www.jstor.org/stable/10.1086/424024 Other complications arise when two types of flowers are Cadet, C. 1954. Recherches sur la valeur morphologique du calicule present (e.g., left and right-oriented). This would necessitate chez les Lythracées. Bull. Sci. Bourgogne 15: 53–83. the use of two different arrows, adding to the confusion. Also, Choob, V.V. 1999. Phantom leaves: A new look to the old problem of exaggerated contortions of organs cannot be shown fully, as for branching in (Amaryllidaceae). Syst. & Geogr. Pl. 68: 67–72. http://dx.doi.org/10.2307/3668591 example in Vigna (Fig. 1F) and Chamaecrista where the ovary Costa, C.B.N., Lambert, S.M., Borba, E.L. & De Queiroz, L.P. 2007. is curved for about 45 degrees. Post-zygotic reproductive isolation between sympatric taxa in the Chamaecrista desvauxii complex (Leguminosae–Caesalpini­ oideae). Ann. Bot. (Oxford) 99: 625–635. CONCLUSIONS http://dx.doi.org/10.1093/aob/mcm012 Dwyer, J.D. & collab. 1980. Leguminosae subfamily Papilionoideae (Conclusion). Pp. 523–818 in: Woodson, R.E., Jr., Schery, R.W. & We agree with Prenner & al. (2010) that floral formulae collab. (eds.), of Panama. Annals of the Missouri Botanical should be given greater weight in botanical research, as a Garden 67. St. Louis: Missouri Botanical Garden. valuable tool in comparative morphology for making concise http://biodiversitylibrary.org/item/89032#page/1/mode/1up descriptions, identification keys, and even comparisons of flo- Eckardt, T. 1937. Untersuchungen über Morphologie, Entwicklungs­ ral developmental sequences. However, floral formulae cannot geschichte und systematische Bedeutung des pseudomonomeren Gynoeceums. Nova Acta Leop., N.F., 5: 1–112. replace floral diagrams in the spatial depiction of flowers and Eichler, A.W. 1875–1878. Blüthendiagramme, 2 vols. Leipzig: Engel- in the description of special structures. At the same time floral mann. http://dx.doi.org/10.5962/bhl.title.12323 formulae must not be overcomplicated, as an accumulation Endress, P.K. 1998. Antirrhinum and Asteridae: Evolutionary changes of too much information can be counterproductive. The use of floral symmetry. Symp. Soc. Exp. Biol. 51: 133–140. of a hierarchical sequence in floral symbols allows for better Endress, P.K. 1999. Symmetry in flowers: Diversity and evolution. Int. clarity in conveying the information. Floral formulae can be J. Pl. Sci. 160(6 Suppl.): S3–S23. http://dx.doi.org/10.1086/314211 Endress, P.K. & Armstrong, J.E. 2011. Floral development and floral as detailed as the user wishes it to be. The descriptors that we phyllotaxis in Anaxagorea (Annonaceae). Ann. Bot. (Oxford) 108: propose here aim not to be dogmatic but to provide a clear 835–845. http://dx.doi.org/10.1093/aob/mcr201 and standardized system that is fit for multiple purposes as a Ge, L.-P., Lu, A.-M. & Gong, C.-R. 2007. Ontogeny of the fertile

1110 Version of Record (identical to print version). TAXON 63 (5) • October 2014: 1103–1111 Ronse De Craene & al. • Floral formulae

flower in Platycrater arguta (Hydrangeaceae). Int. J. Pl. Sci. 168: evidence for the systematic relationships of Tropaeolum (Tropae- 835–844. http://dx.doi.org/10.1086/518271 olaceae). Ann. Bot. (Oxford) 88: 879–892. Glinos, E. & Cocucci, A.A. 2011. biology of Canna indica http://dx.doi.org/10.1006/anbo.2001.1525 (Cannaceae) with particular reference to the functional morphol- Ronse De Craene, L.P. & Wantorp, L. 2008. Morphology and anatomy ogy of the style. Pl. Syst. Evol. 291: 49–58. of the flower of Meliosma (Sabiaceae): Implications for pollination http://dx.doi.org/10.1007/s00606-010-0379-x biology. Pl. Syst. Evol. 271: 79–91. Hufford, L.D. 1990. Androecial development and the problem of mono- http://dx.doi.org/10.1007/s00606-007-0618-y phyly of Loasaceae. Canad. J. Bot. 68: 402–419. Ronse De Craene, L.P., Linder, H.P., Dlamini, T. & Smets, E.F. 2001. http://dx.doi.org/10.1139/b90-052 Evolution and development of floral diversity of Melianthaceae, Iwamoto, A., Shimizu, A. & Ohba, H. 2003. Floral development and an enigmatic Southern African family. Int. J. Pl. Sci 162: 59–82. phyllotactic variation in Ceratophyllum demersum (Ceratophyll­ http://dx.doi.org/10.1086/317911 aceae). Amer. J. Bot. 90: 1124–1130. Ronse De Craene, L.P., Linder, H.P. & Smets, E.F. 2002. Ontogeny http://dx.doi.org/10.3732/ajb.90.8.1124 and evolution of the flower of South African Restionaceae with Judd, W.S., Campbell, C.S., Kellogg, E.A., Stevens, P.F. & Donoghue, special emphasis on the gynoecium. Pl. Syst. Evol. 231: 225–258. M.J. 2002. Plant systematics: A phylogenetic approach, 2nd ed. http://dx.doi.org/10.1007/s006060200021 Sunderland, Massachusetts: Sinauer. Sattler, R. 1978. “Fusion” and “Continuity” in floral morphology. Notes Leins, P. & Erbar, C. 2010. Flower and : Morphology, ontogeny, Roy. Bot. Gard. Edinburgh 36: 397–405. phylogeny, function and ecology, 2nd ed. Stuttgart: Schweizerbart Schönenberger, J. & Endress, P.K. 1998. Structure and develop- Science Publishers. ment of the flowers in Mendoncia, Pseudocalyx, and Thunbergia Lindenhofer, A. & Weber, A. 2000. Structural and developmental (Acanth­aceae) and their systematic implications. Int. J. Pl. Sci. 159: diversity of the androecium of Rosoideae (Rosaceae). Bot. Jahrb. 446–465. http://dx.doi.org/10.1086/297563 Syst. 122: 63–91. Simpson, M.G. 2010. Plant systematics, 2nd ed. Amsterdam: Elsevier. Mayr, B. 1969. Ontogenetische Studien an Myrtales-Blüten. Bot. Jahrb. Stützel, T. 2006. Botanische Bestimmungsübungen, 2nd ed. Stuttgart: Syst. 89: 210–271. Ulmer. Olson, M.E. 2003. Ontogenetic origins of floral bilateral symmetry in Tsou, C.-H. & Mori , S.A. 2007. Floral organogenesis and floral evolu- Moringaceae (Brassicales). Amer. J. Bot. 90: 49–71. tion of the Lecythidoideae. Amer. J. Bot. 94: 716–736. http://dx.doi.org/10.3732/ajb.90.1.49 http://dx.doi.org/10.3732/ajb.94.5.716 Patchell, M.J., Bolton, M.C., Mankowski, P. & Hall, J.C. 2011. Com- Tucker, S.C. 1999. Evolutionary lability of symmetry in early floral parative floral development in Cleomaceae reveals two distinct development. Int. J. Pl. Sci. 160(6 Suppl.): S25–S39 pathways leading to monosymmetry. Int. J. Pl. Sci. 172: 352–365. http://dx.doi.org/10.1086/314212 http://dx.doi.org/10.1086/658158 Van Heel, W.A. 1966. Morphology of the androecium in Malvales. Prenner, G., Bateman, R.M. & Rudall, P.J. 2010. Floral formulae Blumea 13: 177–394. updated for routine inclusion in formal taxonomic descriptions. Von Balthazar, M., Schönenberger, J., Alverson, W.S., Janka, H., Taxon 59: 241–250. http://www.jstor.org/stable/27757066 Bayer, C. & Baum, D.A. 2006. Structure and evolution of the Ronse De Craene, L.P. 1990. Morphological studies in Tamaricales I: androecium in the Malvatheca clade (Malvaceae s.l.) and impli- Floral ontogeny and anatomy of Reaumuria vermiculata L. Beitr. cations for Malvaceae and Malvales. Pl. Syst. Evol. 260: 171–197. Biol. Pflanzen 65: 181–203. http://dx.doi.org/10.1007/s00606-006-0442-9 Ronse De Craene, L.P. 2010. Floral diagrams: An aid to understanding Wanntrop, L. & Ronse De Craene, L.P. 2007. Flower development of flower morphology and evolution. Cambridge: Cambridge Univer- Meliosma (Sabiaceae) – Evidence for multiple origins of pentamery sity Press. http://dx.doi.org/10.1017/CBO9780511806711 in the eudicots. Amer. J. Bot. 94: 1828–1836. Ronse De Craene, L.P. 2011. Floral development of Napoleonaea http://dx.doi.org/10.3732/ajb.94.11.1828 (Lecythidaceae), a deceptively complex flower. Pp. 279–29 in: Wanntorp, L., Puglisi, C., Penneys, D. & Ronse De Craene, L.P. 2011. Wanntorp, L. & Ronse De Craene, L.P. (eds.), Flowers on the Multiplications of floral organs in flowers: A case study in Cono- of life. Systematics Association Special Volume Series 80. Cam- stegia (Melastomataceae, Myrtales). Pp. 218–235 in: Wanntorp, bridge: Cambridge University Press. L. & Ronse De Craene, L.P. (eds.), Flowers on the Tree of Life. Ronse De Craene, L.P. & Smets, E. 1995. The distribution and sys- Systematics Association Special Volume Series 80. Cambridge: tematic relevance of the androecial character oligomery. Bot. J. Cambridge University Press. Linn. Soc. 118: 193–247. Waters, M.T., Tiley, A.M.M., Kramer, E.M., Meerow, A.W., Lang- http://dx.doi.org/10.1111/j.1095-8339.1995.tb00469.x dale, J.A. & Scotland, R.W. 2013. The corona of the daffodil Ronse De Craene, L.P. & Smets, E.F. 1997. A floral ontogenetic study Narcissus bulbocodium shares stamen-like identity and is distinct of some species of Capparis and Boscia, with special emphasis on from the orthodox floral whorls. Plant J. 74: 615–625. the androecium. Bot. Jahrb. Syst. 119: 231–255. http://dx.doi.org/10.1111/tpj.12150 Ronse De Craene, L.P. & Smets, E.F. 1998. Meristic changes in gynoe- Watson, L. & Dallwitz, M.J. 1992+. The families of flowering plants: cium morphology, exemplified by floral ontogeny and anatomy. Pp. descriptions, illustrations, identification, and information retrieval. 85–112 in: Owens, S.J. & Rudall, P.J. (eds.), Reproductive biology Version 14 Dec 2000. http://www.biologie.uni-hamburg.de/b-on in systematics, conservation and economic botany. Richmond: line/delta/angio/ (accessed 19 Aug 2014). Royal Botanic Gardens, Kew. Weberling, F. 1989. Morphology of flowers and . Cam- Ronse De Craene, L.P. & Smets, E.F. 2001. Floral developmental bridge: Cambridge University Press.

Version of Record (identical to print version). 1111