Perianth development in the basal monocot Triglochin maritima (Juncaginaceae)

development in the basal monocot Triglochin maritima (Juncaginaceae). ABSTRACT Basal monocots exhibit considerable variation in inﬂorescence and ﬂoral structure. In some cases, such as Triglochin maritima, it is not clear whether the lateral and terminal structures of the inﬂorescence are ﬂowers or pseudanthia, or where the limits between ﬂowers and inﬂorescence lie. To address these questions, morphological studies were carried out, and the results show that in T. maritima both terminal and lateral structures are ﬂowers, not pseudanthia. The terminal ﬂower of T. maritima develops from the apical inﬂorescence meristem, suggesting that the apical meristem identity changes from ‘‘inﬂorescence’’ to ‘‘ﬂower’’ during inﬂorescence development. In addition, distal ﬂowers of T. maritima are reduced, and there is no distinct ﬂower-subtending bract; instead, the perianth develops unidirectionally, resulting in an abaxial-median bract-like tepal and bilaterally symmetrical ﬂowers, similar to those of other basal monocots, such as Aponogeton and Acorus. It is possible that the leaf primordium changes its positional homology from ‘‘ﬂower-subtending bract’’ to ‘‘tepal.’’ Therefore, in some basal angiosperms with abbreviated development of lateral ﬂowers the demarcation of the ﬂower vs. the inﬂorescence is ontogenetically ambiguous. In situ hybridization experiments show that a putative ortholog of the B-class gene APETALA3 / DEFICIENS is expressed in developing stamens and carpels, and may also be expressed in the shoot axis of the very young inﬂorescence. This expression pattern seems to be consistent with the gradual transition between inﬂorescence and ﬂower that was observed morphologically.


INTRODUCTION
Recent molecular phylogenetic investigations of flowering plants have revealed that the position of monocots remains uncertain; the monocots, Chloranthaceae, and magnoliid clade form a grade between the basalmost clades (Amborellaceae, Nymphaeaceae, and Austrobaileyales) and eudicots (e.g., Qiu et a!. 2000;Borsch et a!. 2003;Hilu et a!. 2003).Within the monocots, the monogeneric Acoraceae, with three or four species, are sister to all other extant monocots in most analyses (Chase et al. 1993(Chase et al. , 2000(Chase et al. , 2006;;Duvall et al. 1993a, b; Qiu et al. 1993Qiu et al. , 2000;;Nadot et al. 1995;Savolainen et al. 2000;Soltis et al. 2000;Borsch et al. 2003;Hilu et al. 2003), but occasionally placed within alismatids (Qiu et al. 2001; some trees in Zanis et al. 2002).Acoraceae exhibit little variation in floral morphology (Chen et al. 2002) and are characterized by a single, cone-shaped inflorescence with numerous small flowers in a dense arrangement (spadix), and the inflorescence elevated on a stalk together with a foliar leaf (spathe).The trimerous flowers consist of two whorls of inconspicuous, scale-shaped tepals (the outer median tepal is on the abaxial side of the flower and bract-like), two whorls of stamens with introrse anther dehiscence, and a synascidiate-symplicate trimerous gynoecium that lacks septal nectaries (Buzgo and Endress 2000;Buzgo 2001).
Scheuchzeria L. (Scheuchzeriaceae, Fig. 1, iii), the sister to other members of clade iii in Alismatales (Les et al. 1997;Chase et al. 2006), also possesses expanded inflorescences, but flowers are few, small, and inconspicuous with small perianth organs.In Scheuchzeria, flower-subtending leaves are the main protective organ for the flower (Uhl 1947;Posluszny 1983;Haynes et al. 1998c;Gupta et al. 1998).The inflorescence morphology in the remaining members of clade iii (Aponogetonaceae, Juncaginaceae, Lilaeaceae [now included in Juncaginaceae, APG II 2003], Potamogetonaceae, Zannichelliaceae, Zosteraceae, Cymodoceaceae, Ruppiaceae, Posidoniaceae; Les et al. 1997, Chase et al. 2006) provides a strong contrast to that of the second clade.Typical for this third clade are small, inconspicuous floral units (term including flowers and pseudanthia, definition below) in a dense arrangement, often sessile, sub-sessile, or on a swollen inflorescence axis (spadix) and with a reduced perianth (Eber 1934;Uhl 1947;Hutchinson 1973;Sattler 1965Sattler , 1973;;Haynes et al. 1998b).The reduction of the perianth in members of clade iii sometimes blurs the distinction between the flower-subtending bract and tepals (e.g., in Apanagetan L. f., Juncaginaceae, some Potamogetonaceae), similar to Acarus L. (Buzgo and Endress 2000) and the calyculus in Tofieldiaceae (Remizova and Sokoloff 2003;Remizowa et al. 2006).In contrast, in some other species of Potamogetonaceae, as well as taxa outside this clade with similarly reduced flowers in dense inflorescences (e.g., Araceae), the flowersubtending bract is not included in the perianth, although the bract may be suppressed.Correlated with this different pattern of floral reduction, flower-like terminal structures are absent from Potamogetonaceae and the fourth clade, Araceae (Fig. 1, iv; Buzgo 2001).
The compact inflorescence and reduced perianth in some members of Alismatales make it difficult to ascertain the identities of particular structures.In their reviews of floral morphology in basal angiosperms, Eames (1961), ClaBen-Bockhoff (1990), and Hay and Mabberley (1991) suggested a gradual transition of organ identities in some taxa.Although species of Triglachin L. (Juncaginaceae) have been well studied for their floral morphology (Cordemoy 1862;Uhl 1947;Eames 1961;Aston 1973Aston , 1993a, b;, b;Robb and Ladiges 1981;Ford and Ball 1988;Cooke and Davies 1990;Harden 1993;Endress 1995;Haynes et al. 1998b;Igersheim et al. 2001), interpretations of the "floral units" (definition below) are controversial.The "floral units" have been considered to represent either flowers (Hill 1900;Arber 1940;Eckardt 1957;Singh 1973;Serbanescu-Jitariu 1973;Lieu 1979;Charlton 1981;Endress 1995;Igersheim et al. 2001) or pseudanthia (Miki 1937;Uhl 1947;Eames 1961;Burger 1977).The definition of a pseudanthium, however, also differs among authors.According to Rudall and Bateman (2003) it is a structure that is neither a true flower nor a true inflorescence.This differs from the traditional definition of a pseudanthium as an inflorescence that imitates a flower, as a result of the aggregation of flowers (Eames 1961;ClaBen-Bockhoff 1990;Endress 1994).This second definition neither implies nor excludes the loss of the distinction of meristem identity between flower and inflorescence.We follow this second, more commonly used terminology.For the structure that resembles a flower (actual flower or pseudanthium), ClaBen-Bockhoff (1990) uses the term pollination unit, or blossom, whereas Rudall and Bateman (2003) use "floral unit."In this study we apply floral unit (Rudall and Bateman 2003), which includes flowers and pseudanthia.The term does not imply a function in animal pollination (as Patamageton L. and Triglachin are both probably wind-pollinated), although most authors use "flower" in reference to Triglachin (Hill 1900;Lieu 1979;Charlton 1981;discussion below).
In angiosperms, lateral shoots (including lateral flowers) typically are subtended by a leaf.The subtending leaf is thereby considered an appendage of the main shoot (Troll 1937;Esau 1977;Hagemann 1963Hagemann , 1970Hagemann , 1984)).Consequently, the flower-subtending leaf is considered extrafloral.In many species the flower-subtending leaf is reduced to a scale-shaped flower-subtending bract.Bracts and tepals are often similar and difficult to distinguish, especially in basal angiosperms (von Balthazar and Endress 2002;Buzgo et al. 2004a, b).Many taxa have no visible flower-subtending leaves (e.g., Arabidopsis Heynh.), and in these cases, the flower-subtending leaf or bract is not a universal morphological marker for an extrafloral position.
Although most authors do not explicitly differentiate between lateral and terminal floral units in Triglachin, they apparently refer to the lateral floral units (Miki 1937;Uhl 1947;Eames 1961;Rudall and Bateman 2003).In this study, we examine these two positions in the inflorescence separately: (i) to determine whether the floral units are flowers or pseudanthia, and (ii) to identify the limits between inflorescence and flower.We hypothesize (Hypothesis 1) that the lateral structures in Triglochin are pseudanthia (Miki 1937;Uhl 1947;Eames 1961;Rudall 2003).We predict that the answer is not absolute, but that the transition from inflorescence to flower is gradual.The following hypotheses specify those characteristics of a flower that concern the loss of flower delimitation.
Hypothesis 2: The primordium in the position of the flower-subtending leaf is not always extrafloral, but is sometimes involved in the perianth.The concept of the flower-subtending leaf as a marker for an extrafloral position is challenged by studies of some basal monocots (Burger 1977;Buzgo and Endress 2000;Remizova and Sokoloff 2003;Rudall 2003;Remizowa et al. 2006) and magnoliids (Tucker 1979(Tucker , 1981(Tucker , 1985;;Liang andTucker 1989, 1990;Tucker et al. 1993;Tucker and Douglas 1996).Some taxa possess reduced flowers that develop unidirectionally (from abaxial to adaxial), in which the first organ of a lateral flower is on the abaxial side of the lateral shoot and could therefore be considered either a flower-subtending bract or a first abaxial tepa!.Such situations occur in Saururaceae and Acarus (Tucker 1979(Tucker , 1981(Tucker , 1985;;Liang andTucker 1989, 1990;Tucker et al. 1993;Tucker and Douglas 1996;Buzgo and Endress 2000).
Here we discuss a similar phenomenon in Triglachin maritima.
Hypothesis 3: In Triglachin maritima, the terminal structure is composed of organs corresponding to several flower primordia, and therefore is a pseudanthium.This hypothesis corresponds to statements regarding (i) floral units in Triglochin in general (for lateral flowers; Miki 1937; Uhl 1947;  Eames 1961; Rudall 2003) and (ii) terminal flower-like structures (Greek pelor for "monster") in some taxa (Buzgo and Endress 2000;Buzgo 2001;Rudall and Bateman 2003).However, this hypothesis contradicts Miki (1937), Uhl (1947), andCharlton (1981), who considered the inflorescence to be indeterminate.Among basal monocots and magnoliids with dense inflorescences, unidirectional flower development, reduction of the adaxial floral organs, and formation of peloria at the apex of the inflorescence appear to be correlated (Buzgo and Endress 2000;Buzgo 2001).Strong initial floral bilateral symmetry and reduction on the adaxial side of the flower can result in flowers represented by only a single organ (Burger 1977;Dahlgren et al. 1985;Lilaea Bonpl., Arber 1940;Posluszny et al. 1986), and ultimately the formation of a terminal pseudanthium.
Hypothesis 4: Genes that are considered strictly floral are transcribed in the inflorescence axis.That is, genetically, the inflorescence of T. maritima has features that are typically exclusive to the flower.The MADS-box orthologs DEFI-CIENS (DEF) and APETALA3 (AP3) take part in B-class function, which is responsible for stamen and petal-like features in Antirrhinum majus L. and Arabidopsis thaliana (L.) Heynh., respectively (Coen and Meyerowitz 1991).DEFI AP 3 orthologs are regulated by genes that also control the induction of floral meristem identity (see Discussion for citations).Therefore, the presence of B-class mRNA is strong evidence for floral identity.Further, floral MADS-box genes have been intensively studied, offering a large literature for comparison of sequences and mRNA localization profiles.The MADS-domain is well conserved and suitable for screening for genes in a total RNA extraction.The C-terminal sequence is highly variable, which allows easy identification of different members of the MADS family.In addition, the C-terminal sequence can be used to construct RNA probes that are sufficiently specific to target genes exclusive to the AP3 clade.

Morphological Studies
Plants of Triglochin marztzma were collected in March 2001 and January 2002 near Copenhagen, Denmark (Buzgo collection numbers: 1068(Buzgo collection numbers: , 1072(Buzgo collection numbers: , 1073)); other taxa were collected at various times and locations (Table 1).Buds of T. maritima were removed by dissection and either used for RNA extraction (below) or fixed in FAA, involving a short application of vacuum (about 7 min) until no more bubbles appeared, and incubated for approximately 6 hr at 4°C, then transferred to 70% ethanol (RNase free), and dehydrated along an ethanol series.For scanning electron microscopy (SEM), samples were critical point-dehydrated, gold-sputtered, and observed in a Hitachi S-4000 FE-SEM at the University of Florida Biotechnology Program.For microtome sections, the samples were transferred to xylene and embedded in Paraplast, sectioned using a rotary microtome (10 J.Lm thick), and placed onto Fisherbrand SuperFrost/Plus microscope slides (Fisher Scientific, Pittsburgh, Pennsylvania, USA).Mounting was in Cytoseal 280 (Richard Allen Scientific, Kalamazoo, Michigan, USA).Observations were made using a Leica MZ12-5 dissection microscope and a Carl Zeiss compound microscope with transmitted light.Photographs were taken with a Nikon Coolpix 995 digital camera.Image editing included linear adjustment of contrast, color-temperature, frame, and resolution, using Adobe Photoshop vers.7.0.

Isolation and Sequence Analysis of eDNA Clones for AP3 Homologs
Total RNA extraction from T. marztlma was carried out using FastPrepl20 (Bio101 Savant, Qbiogene, Irvine, California, USA) tissue homogenizer and the FastRNA Green kit (Bio 101).Total RNA concentration was estimated by 1% agarose gels and by spectrometry with an Eppendorf Bio Photometer.Reverse transcription was conducted using GeneAMP In Situ Core Kit (Perkin-Elmer Applied Biosystems, Wellesley, Massachusetts, USA), adding RNAguard RNase inhibitor (Human Placenta, Amersham Biotech, Piscataway, New Jersey, USA), MLV-M Reverse Transcriptase with Buffer II (Applied Biosystems, Foster City, California, USA), and a T( 161 -primer with an adapter (T( 161 -CCGAGA-GTCGATCAGCTGC).The polymerase chain reaction (PCR) was carried out with Amplitaq Gold Polymerase (Applied Biosystems) and Pfu DNA polymerase (Promega, Madison, Wisconsin, USA), using intron-spanning primers for AP3 homologs based on alignments of B-class MADSbox genes (Kramer et al. 1998 The eDNA PCR products were cloned and selected using PCR-Script AMP Cloning Kit (Stratagene) andre-amplified by PCR using primers for T7 and T3 promoters in the vector according to the Stratagene PCR-script instruction manual.For full eDNA sequences, SMART RACE eDNA Amplification Kit (Clontech, Palo Alto, California, USA) was employed, with the internal primers and the adapter (above).Sequencing used the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit with Amplitaq DNA Polymerase, FS (Applied Biosystems).Sequence analysis was performed on an ABI 377 Prism DNA Sequencer (Applied Biosystems) with associated software.The sequences were analyzed for continuous open reading frames (GenDoc vers.2.6).Blast searches (Blastp) were performed against GenBank, and sequences were aligned with genes annotated as MADS-box genes using GenDoc vers.2.6 and manually.The sequences Tr.ma.AP3-l and Tr.ma.AP3-2 were deposited in GenBank as accession number AY956349 and AY956348, respectively.To reconfirm the sequence homology of our probe templates with AP3-annotated amino acids, a preliminary maximum parsimony analysis and a bootstrap analysis were carried out, involving 427 nucleotide sequences representing all major clades of MADS-box genes (sequences from Becker and TheiBen 2003, combined with sequences from Johansen et al. 2002), using PAUP* vers.4.0b10 for 32-bit for Windows (Pentium 4 CPU 2.4 GHz PC, Win XP) and for unix (on a Dual 2 GHz PowerPC G5, OS X) (Swofford 1998).The specifications of the maximum parsimony analyses included simple taxon addition, using tree bisection-reconnection (TBR) branch swapping, and saving 100 most parsimonious trees.The bootstrap analysis included 100 replicates, using TBR branch swapping and saving I 00 trees.

In Situ Hybridization
RNA probes were generated from the DNA insert representing the sequence 3' from position 191 (Tr.ma.AP3-191F forward primer) comprising the more specific K and C regions of the AP3 homolog.PCR-amplified sequences were inserted into pGEM vector (Riboprobe in vitro Transcription Systems, Promega, Technical Manual No. 016), and inserts of selected clones were sequenced for determination of the insert direction.Clones representing two insert directions were chosen: antisense as probe, and sense as negative control.From each construct, plasmid DNAs were prepared using E.Z.N.A. EaZy Nucleic Acid Isolation-Kit (Omega Biotek, Inc., Doraville, Georgia, USA).Plasmid DNAs were digested with Hind III (Promega), purified by phenol-chloroform extraction and a sodium-acetate ethanol precipitation, and checked on 1% agarose gel.Probe synthesis was by Riboprobe System-SP6 (Promega) transcription kit, including Bohringer-Mannheim DIG RNA Labeling Mix (Roche Applied Science, Indianapolis, Indiana, USA), followed by a DNase digest (Promega), according to the transcription kit protocol.Hydrolysis of the RNA probe was in a mix of Na 2 C0 3 (60 mM) and NaHC0 3 (40 mM), fragment length was evaluated on agarose gel, and the hydrolysate was precipitated in ethanol containing sodium-acetate, tRNA, and dithiothreitol (DTT).
Hybridization followed a modification of the protocol of the Meyerowitz lab (http://www.its.caltech.edul~plantlab/protocols/insitu.html[Jan 2005]).Microtome slides with sections of T. maritima were deparaffinized and hydrated in a xylene-ethanol series, followed by a digest with Proteinase K (Promega), and an acetylation reaction.Hybridization was at 55°C.For background suppression, slides were incubated in RNAse A (Sigma-Aldrich, St. Louis, Missouri, USA, not boiled), then washed twice in 0.2X SSC in a gyratory agitator for one hour at 55°C, and pre-blocked in phosphate buffered saline buffer (PBS) with 1% BSA-c (New England BioLabs, Inc., Beverly, Massachusetts, USA, purified BSA #B90015).Sections were incubated with Bohringer-Mannheim Anti-Digoxigenin Fab Fragment solution according to the manufacturer's instructions.Signal detection was by alkaline phosphatase reaction with NBT/BCIP Tablets (Roche) according to the manufacturer's instructions.The signal was monitored in the dissection microscope and photographed as described above.

Morphological Development
The inflorescence of Triglochin maritima is initiated as a large coherent meristem at the apex of the shoot, which becomes conical as flowers are initiated (Fig. 2).The diameter of floral primordia is small in relation to the inflorescence, allowing several primordia to appear at one level around the inflorescence.Floral primordia appear in several parastichies (Fig. 2).Lateral meristems are initiated acropetally in fast succession along the inflorescence, above a short basal peduncle.As the distal diameter of the shoot apical meristem (SAM) of the inflorescence is reduced, the number of floral primordia progressively decreases, whereas the size of primordia at initiation is not reduced significantly.At the time when the inner tepals initiate, the inflorescence becomes deformed between the last foliar leaf of the main shoot and the prophyll of the continuation shoot.The side of the inflorescence facing the continuation shoot is flatter, while the side toward the last foliar leaf maintains its rounded surface.The two sides are separated by two rims longitudinally on the inflorescence, corresponding to the limit where the inflorescence touches the prophyll (Fig. 3).
At initiation, some "floral" primordia first exhibit a slight enlargement of the abaxial side (Fig. 4), but no distinct abaxial organ develops earlier than the rest of the floral meristem (Fig. 5).The abaxial median tepa! and lateral outer tepals are initiated almost simultaneously, with a larger fraction of the floral meristem dedicated to the median abaxial tepal.As a result, most flowers develop with a slight bilateral symmetry: the abaxial tepa! is slightly larger, the outer tepals do not form an isometric triangle (60°), and instead the lateral tepals slant toward the transverse orientation (Fig. 6).A flower-subtending bract is not initiated (Fig. 5, 6).The size of the inflorescence SAM reduces gradually as floral primordia emerge from it (Fig. 2).Finally, a short lag occurs after which the remainder of the SAM gives rise to a terminal structure.The lateral primordia across the lag abruptly change from flower to single floral organs, and the terminal structure is identical to a flower (below; Fig. 8-10).That is, the terminal flower is the last one to be initiated, and the inflorescence is a determinate raceme (Troll 1964;Weberling 1989).
In the flowers, the three inner tepals develop almost synchronously, followed shortly by two trimerous alternate whorls of stamens (Fig. 6-9).At this stage the outer abaxial tepa! grows faster in most flowers and increasingly appears bract-like (Fig. 6).Distally in the inflorescence, the adaxial organs develop to a smaller size, and the position of the lateral outer tepals slants toward the adaxial side.Just below the apical flower, this adaxial inhibition affects even the median inner tepals and stamens; in some flowers these organs are absent (Fig. 8).However, on the two longitudinal rims of the inflorescence that meet in the terminal flower (Fig. 3), most flowers appear radially symmetric with equal outer tepals (Fig. 7, 9).At this stage, the constriction below the first abaxial organ elongates: the pedicel is formed, and the first abaxial organ is clearly identified as a floral organ (tepa!).
During organ initiation, the floral center remains prominently convex: cell division at the floral SAM exceeds the formation of organ primordia, and a lateral expansion of the receptacle (below the outer tepals) is not observed (Fig. 6-9; however, it expands above the tepals.When the hemispherical carpel primordia initiate, the floral apex has risen above the stamens (Fig. 6, 7, 10).As a result of this meristem expansion, the carpels are initiated on the slope of the floral SAM and have a tilted base (Fig. 11, 12).The outer carpels alternate with the inner stamens and arise after them following a lag; the inner carpels appear after the outer carpels following a short plastochron (Fig. 6, 7, 11); that is, the plastochron between the two whorls of carpels is shorter than that between the inner stamens and outer carpels.At initiation of the inner carpels, each outer carpel develops a rim around a depression.The rim appears more like a torus than a horseshoe (as is typical for many other Alismatales; e.g., Sattler 1973;Sattler andSingh 1973, 1978), correlating with the meristem expansion of the adaxial carpel side and the elongation of the floral apex (Fig. 12).Within the three inner carpels, the apex of the flower remains plane to slightly convex (Fig. 6, 12).
In later development, the tepals elongate and overlap (Fig. 11, 13).Normally, the abaxial median tepa! overlaps all oth-  ~ er organs.However, the outer abaxial tepal can be covered by the lateral tepals, because the inflorescence continues to be deformed as it grows between the continuation shoot and the last foliar leaf (Fig. 13).Along the longitudinal rim, more space is available on the lateral side of each flower than on the median side.This causes the lateral tepals to be lifted away from the flower above the abaxial tepal (Fig. 13), resulting in an asymmetric appearance.However, this is a secondary effect, and the flower is actually bilaterally symmetric.After all organs are initiated, the terminal structure is identical to a "flower" -completely radially symmetrical whorls of tepals, stamens, and carpels.Until anthesis, the terminal structure remains the largest "flower" of all on a prominent base lifted above the adjacent subterminal flowers.
Before anthesis, the inflorescence emerges from the foliar leaf sheaths, by elongation of the basal inflorescence axis (Fig. 14).The internodes between the flowers elongate later, separating the flowers from each other before anthesis (Fig. 15).The flowers are protogynous.

Morphological Studies of Triglochin procera, T. striata, and Maundia
For comparison, flowers of Triglochin procera (s.l., including T. multifructa and T. microtuberosa; Fig. 16-19), T. striata , and Maundia triglochinoides (Juncaginaceae; Fig. 23) were examined.Whereas, T. maritima grows best above water level or only temporarily submerged, we found that the rhizome of the Australian T. procera complex is almost always submersed.Triglochin procera differs from T. maritima in having a much more robust growth form, with an inflorescence that can reach more than 1 m in length (instead of 40 em in T. maritima), bearing flowers on its distal 25 em.Flowers of T. procera are correspondingly larger, up to 1 em in diameter (compared with 3-4 mm in T. maritima). Theflowers of both T. procera and T. maritima are trimerous-hexacyclic, but the stigma of T. procera is more spreading and star-shaped; additionally, carpels are only basally fused and sometimes twisted.Because of the larger size of flowers of T. procera, we expected them to be more radially symmetric than those of T. maritima.Indeed, we found fewer indications of flower reduction, though some reduced flowers occur apically.We also had difficulty in distinguishing the terminal flower from lateral flowers.Nevertheless, flowers of T. procera also exhibit bilateral symmetry (Fig. 18, 19) and lack a subtending bract.Instead, the outer median tepal is abaxial and slightly prominent (Fig. 19), as in T. maritima.The inner median tepal on the adaxial side is smaller than the other inner tepals, but expands above the outer lateral tepals (Fig. 18, 19).
Triglochin striata from Australia was only observed in cultivation.It differs from both T. procera and T. maritima by being much smaller.The distal portions of the leaves are round in transverse section, and the entire slender inflorescences of T. striata reach only 30 em (Fig. 20), with flowers of about 3 mm in diameter with only one whorl of carpels .Associated with smaller flower size, flower reduction within the inflorescence is much more frequent (we never found all whorls to consist of three organs).Particularly at the base Fi g .2-1 0.-Early Horal develo pment of Triglochin maritima, all but Fi g .3 are S EMs.-2.Yo ung inH orescence, s ide view, ini tiating latera l Howers, do me-shaped SAM (x), prophy ll (p), fo Jj ar leaf (f) of the co ntinu ati on shoot.-3.Young inHorescence, s ide view showing the side toward the prophy ll (*) a nd the lo ng itudina l rim (arrows).-4.C lose up of Hower primordia at initiati on, apical view, d1e abaxial side is more pro no unced ( I*) than the adaxia l side, and there is some space between the primo rdi a.-5.C lose up o f Hower primord ia after initiati on, api cal view, the abax ial s ide ( I*) is equ al to the adax ial side as compared with F ig. 4, a nd the re is a lmost no space between the primo rdi a.-6.Yo ung Howe r a lo ng d1e side o f the inH orescence, apica l vi ew, outer tepa ls ( l ), the o uter, abax ia l medi an tepa! is larger ( I *), whereas inne r tepa ls (2), outer and inner sta me ns (3, 4), outer and inner carpe ls (5, 6) a ll develop equa lly.-7 .Young Hower on longitudinal rim of the inH ore cence, api cal-abax ia l view, outer tepa ls ( I) are equa l, inc lud ing the abax ia l te pa! ( 1 *), the inner carpe ls ( 6) are e levated on the Howe r center.-8.Young Howers, oblique-api cal view, apex of terminal Hower (x), abax ial medi an te pa! enl arged in late ra l Howers Young lateral fl ower shortly after initiation of inner carpels (6), SEM , side view.-12.Young gynoecium, SEM, oblique-apical view, carpels developing an adaxial cross meri stem (k), outer carpels forming an ovary depression (arrow).-13.Flowers on the longitudinal rim of the inflorescence, SEM, apical view, the median abaxial tepals are overlapped by smaller lateral tepals ( 1), in flowers besides the rim , the abaxial median tepa! ( I *) overlaps the lateral te pals, as expected fo r unidirectional, bilaterally sy mmetrical development.-14.Youn g infl orescence, side view, the Rowers are still densely arranged.-15.Inflorescence, side view, female stage of an thesis, stigma papillae exposed, internodes between Rowers elongate.Outer tepa Is (l , I* abaxial median tepa!), carpels (5, 6), carpel cross meristem (k).Bars in Fig. 11, 12 = 0.1 mm, in Fig. 13 = l mm, in Fig. 14, 15 = I 0 mm. of the inflorescence, flowers appear iJTegular in symmetry a nd merosity.At mid-leve l of the inflorescence, the merosity of flowers may be reduced , resulting in apparently three tetramerous whorls (tricyclic) rather than s ix trimerous whorls, as in the two larger species describe d above (Fig. 2 1).No bracts were observed and the median abaxial tepa! is prominent thmughout the inflorescence, appearing bract-Like (Fig. 2 1).Distally in the inflorescence, the adaxial s ide of the fl ower can be reduced to such an extent that the median a baxial tepa! is the only sizable perianth organ a nd appears bract-like (Fig. 22).Terminal flowers were not observed in T. striata, becau se the s lender inflorescences tended to abort at the tip.
Maundia triglochinoides, a monotypic Australian aquatic, appears similar to T. procera in gross morphology.The two reported differences between the species are the formation of sto lons in Maundia E Muell., a nd the merosity of the flower.Maundia has only two tepa ls, laterally on the abaxial side of the flower (Fig. 23), simil ar to the peri anth in some Aponogetonaceae.In addition, the androecium consists of four to six stamens; the gy noec ium of Maundia consists of four carpels (so metimes three or two distally in the inflorescence) with a prominent, plicate apex a nd is similar to female flowers of Tetroncium Willd.(Juncaginaceae, two or three tepals, three or four conically-elongate carpels with a large plicate proportion; pers.obs.).Due to the lack of material, terminal flowers and floral development could not be studied in detail in Maundia and Tetroncium..  Flower, SEM, apical view, be fore antbesis, the inner medi an tepa! (2*) is sma ll er tban the inne r late ral tepa ls, and overl aps one o f the outer lateral te pa ls.-19.Fl ower, SEM , side view, before a nthesis, tbe inner medi an te pa! (2*) is sma ll er th an tbe inner lateral tepa ls, and overl aps tbe outer lateral te pals, the outer median tepa! ( I*) is promi nent.Bars in Fig. 16, 17 = I e m, in F ig. 18, 19 = 0.2 mm.-20.Infl orescence, side view, be fore anthesis.-21. Flower, SEM , side view, fe ma le stage of anthesis, tbe outer medi an tepal is pro minent, tbe lateral tepals are transverse ( I), inner and outer whorl are not distinct (reduced perianth); a lso showing stamens (3), sti gma papillae on top of carpe ls (5-6), and pedicel (*) without subte nding leaf.-22.Flower, SEM , side view, before an thesis, tbe outer medi a n tepa! is much larger than tbe latera l one.Bars in Fig. 20 = 5 mm, in F ig. 2 1, 22 = 0.2 mm .-23 .SEM , apical view, fe male stage of anthesis, the re are onl y two abax ia l lateral tepa Is (2).Outer tepa Is ( I, I* abax ial med ian tepa!), inner tepa Is (2), outer stamen (3), carpe ls (5 , 6).Bar in Fig. 23 = 0.2 mm .

Identification of A PETALA 3 eDNA Sequence and In Situ Hybridization
The A P3 ho molog seque nces recovered (Tr.ma.A PJ-1, Tr.ma.A P3-2) are nearl y ide ntical to each other and lack six amino acids at the 5 '-end .The most simil ar DNA sequence found (Bl as tn) annotated for AP3 was fro m Lauraceae (A PJlike of Lindera eryth.rocarpaMakino), not mo nocots.The best hits to monocots (Oryza sativa L. and Asparagus officinalis L. ) have sig ni ficantl y lower bl as t scores, as do hits to mode l organi sms (e.g., DEFICIENS A of Antirrh.inummajus) .The most simil ar amino acid sequences (Bl astp) are from two mo nocots, Asparagus officinalis and a Hemerocallis L. hybrid culti var; however, these two seque nces are only annotated as MADS -box genes, not A P3-orth ologs.The hig hest score for a n APJ -ann otated prote in is fro m Chloranthus spicatus M aki no of C hloranthaceae, a fami ly that with monocots and magno liids fo rms part of a po lyto my after the basal grade of Amborella Ba il!. , N ymphaeaceae, and Austrobai leyale (e.g., Solti s et aJ.2000).The A rabidopsis thaliana AP3 prote in has a substantially lower score than the monocot and the Chloranthaceae sequences.
The maxi mum parsimony anal ysis included a total of 854 alig ned amino acids, 494 of which were parsimony-informati ve.The strict consensus of the 100 most parsimonious trees tha t were reta ine d pl aced Tr. ma.A P 3-l and Tr.ma.A P3-2 in a clade of A P3 seque nces, separate from a clade of PI ho mo logs.The bootstrap support for the clade exclus ively including a ll DEF-A P3 transcription facto rs and Tr.ma.A P3-l and Tr.ma.A P3-2 was 89% .These results support that Tr.ma.AP3-J and Tr.ma.A P3-2 are ortho logs of the DEF-A P3 transc1iptio n factors.
Using A P3 probes , mRNA localizati o n was determined by in situ hybridi zati on in inflorescences of two stages.In the younger stage exami ned (corresponding to stame n initi ati on; Fig. 6-10), AP3 m.RN A was detected throughout the entire infl orescence, as well as in leaves (Fig. 24 , 26, 27).It is unlikel y that thi s signal re fl ects nonspec ific hybridi zation with mRNAs in young tissues, because the negative control (sense) yie lded much lower levels of background (Fig. 25).The older stage (corresponding to the initi ation of carpels) shows a c lear differe ntiation of signal .AP 3 sig nal is strongest in newly initiating thecae, carpels (Fig. 29, 31), procambia1 ti ssue (Fig. 30), and tepa! tips.Weak expressio n was detected in future inflorescence parenchyma, epidermal cells, tepal bases, and in the center of the flower (the terminati ng apex, rather than the carpels).

Bilateral Symmetry and Flexibility of the Bract
We suggest that every lateral shoot starts with an inherent bil ateral symmetry due to the subtending leaf, and that the putative fl ower-subtending bract is not always extra floral , but is someti mes involved in the perianth (Hypothesis 2) .These issues are c losely linked .The leaf and its axillary shoot develo p fro m a commo n meri stem (Fig. 32, 33;Troll 1937;Hagemann 1963Hagemann , 1970Hagemann , 1984;;Esau 1977); thus, both sy mmetry a nd the production of a flowe r-subtending bract depend o n how abrupt the transition is between leaf and ax ill ary shoot at th e base of both organs.If the transition is gradual, then the uppe r (adax ial) side of the leaf and the lower (abaxial) side of the ax illary shoot may mutually affect one another.For example, the meri stem dedicated to the subtending leaf is absent on the abaxial side of the lateral SAM .As a result, the first leaf of the lateral shoot initiates on the adaxial side, opposite the subtending leaf.Indeed, in monocots, the first leaf on the axillary shoot is a single prophyll on the adaxial side of the axill ary shoot, alternating with the subtending leaf, corresponding to a di stichous phyllotaxy resulting from the abaxial inhibition by the subtending leaf (Fig. 34).Inhibition could be due to the lack of auxin, which was proposed to affect the radial position and size of lateral organs in tomato and Arabidopsis (Reinhardt et al. 2000).Although lateral shoots initiate with an inherent bilateral symmetry, this bilateral symmetry is lost as the lateral shoot grows.In a lateral flower with a significant pedicel , a prophyll , and possi bly additional bracts, the SAM of the latera l shoot has time to equalize its sides: abaxial inhibition by the subte nding leaf is countered by inhibition by the prophyll , the SAM becomes radially symmetrical, and whorled fl oral organs develop simultaneously (Fig. 35).
If fl oral development is abbreviated, no intermedi ate bracts are formed a long the flora l shoot, and the fi~s t organs initiated are already part of the peri anth.Nonethe less, due to abaxi al inhibition, the first floral organs still develop on th e adax ial side, in the position of the prophyll.This results in unidirectional flower development from adaxial to abaxial (Fig. 36), as in Neuwiedia Blume (Kocyan and Endress 2001).A flower-subtending leaf might be suppressed, as suggested for Nymphaeaceae (Cutter 1957a, b;Moseley 1972), the basal monocot family Araceae, and some Potamogetonaceae (Eber 1934;Posluszny andSattler 1973, 1974;Tomlinson 1974;Posluszny 1981;Buzgo 2001).In Araceae and some Potamogetonaceae no median organ develops on the abaxial side or the outermost whorl, as if there was still inhibition by the flower-suppressed leaf (Fig. 37).Suppression of the flower-subtending bract has also been shown in Arabidapsis and other Brassicaceae (Saunders 1923;Troll 1937;Hagemann 1963Hagemann , 1970Hagemann , 1984;;Esau 1977;Shu et al. 2000;Heisler et al. 2005).However, in Arabidapsis, the abaxial median sepal is larger during early flower development (Smyth et al. 1990), and later adjusts its growth to equal the size of the other three sepals.The result is similar to those cases in which the subtending bract is involved in the perianth (Triglachin, Acarus; Fig. 38).In the inflorescence of Triglachin there is no distinct flower-subtending bract, but the flower initiates with an abaxial organ that shares features of both the subtending bract and the tepal.This is a frequent phenomenon, especially in flowers that are small and initiate in fast succession, as has been discussed for Acarus (Buzgo and Endress 2000;Buzgo 2001).If floral shoot development is abbreviated even further, then the lateral meristem comprising subtending bract and axillary shoot does not subdivide before the meristem identity for the flower is determined.The result is a lack of inhibition by an extrafloral flower-subtending bract and a direct transition of the lateral shoot into the perianth zone, without forming any bracts (Fig. 38).
Meristem identity of the flower is based on the expression of specific genes (Coen et al. 1990;Schwarz-Sommer et al. 1990, 1992;Coen and Carpenter 1992;Huala and Sussex 1992;Singer et al. 1992;Weigel et al. 1992;Weigel and Nilsson 1995;Blazquez et al. 1997;Hempel et al. 1997;Lee et al. 1997;Ma 1997Ma , 1998;;Parcy et al. 1998;Weigel 1998;Wagner et al. 1999;Berleth et al. 2000;Ferriindiz et al. 2000;Frohlich and Parker 2000;Yu et al. 2000;Araki 2001;Coen and Langdale 2001;Pena et al. 2001;Soltis et al. 2002).By slightly altering gene expression levels, the first abaxial organ of the lateral structure (leaf and axillary shoot) might be turned into a floral organ (bract-like tepal).This would result in unidirectional development from abaxial to adaxial, as observed in Acarus, Apanagetan, and Triglachin (Fig. 37).Intercalary elongation within the common base of subtending leaf and axillary shoot results in a recaulescence of both organs: by intercalary growth, the subtending leaf is lifted away from the main shoot, along with the axillary shoot.This occurs in Triglachin and Arabidapsis, where a distinct pedicel is present.In Triglachin maritima, this feature is intermediate between the situation in Arabidapsis and Acarus.In Arabidapsis, the abaxial sepal is not much larger than the other sepals in later development.By contrast, in Acarus the bract-like appearance persists throughout development.In Lilaea scillaides (Poir.)Hauman (Juncaginaceae, sensu APG II [2003]), the perianth is reduced to a single median bract-like organ (Posluszny et al. 1986; but a bract according to Uhl 1947), similar to that of Saururaceae (see below).Strong reduction is also found in Aponogetonaceae.
The Australian species Apanagetan hexatepalus H. Bruggen has two trimerous perianth whorls.Most other species of Apanagetan have only one trimerous perianth whorl (representing the inner whorl) with an adaxially median organ (Singh and Sattler 1977b;van Bruggen 1985van Bruggen , 1990van Bruggen , 1998;;Hellquist and Jacobs 1998) that is often reduced, resulting in a flower like that of Maundia.Finally, Apanagetan distachyus L. f. possesses only one bract-like organ.As a result, the flower-subtending bract can appear as the abaxial median tepal of lateral flowers.This reflects a change of organ identity and of corresponding shoot order (from being an attribute of the flower as lateral shoot to an attribute of the inflorescence as main shoot).
In the magnoliid family Saururaceae, a bract-like leaf appears at the abaxial side of the otherwise perianthless flower (Tucker 1975(Tucker , 1979)).In some genera, this leaf is conspicuously petaloid (Hauttuynia Thunb., Anemapsis Hook.& Arn.), whereas in others it is on a common stalk and shares vasculature with the rest of the flower at the pedicel (Tucker 1979(Tucker , 1981(Tucker , 1985;;Liang andTucker 1989, 1990;Tucker et al. 1993;Tucker and Douglas 1996); no axillary shoots have been reported in association with this abnormal leaf, apart from the flower.Therefore, this median abaxial leaf meets the expectations of a perianth organ (sterile phyllome on a floral shoot, position on a receptacle, with short subsequent internodes, no axillary meristem; Buzgo et al. 2004a, b).Its interpretation as a flower-subtending bract lacks developmental support, and is historically based on earlier studies of the closely related family Piperaceae, which possess a more scale-like median abaxial organ inserted strictly on the inflorescence main axis (Tucker 1979(Tucker , 1981(Tucker , 1985;;Liang andTucker 1989, 1990;Tucker et al. 1993;Tucker and Douglas 1996).
We, therefore, conclude that in some taxa with dense inflorescences, the delimitation between inflorescence and flower is less clear than classical morphology implies.The data indicate that in some taxa the organ initiated in the position of an extrafloral flower-subtending bract may become involved in the perianth as the median abaxial tepal.
Is There a Pseudanthium in Triglochin maritima?Miki (1937) proposed a link between flowers of Potamogetonaceae to those of Pandanales based on: (i) the position of the tepals ("bracts" associated with stamens) on a common elevation with the stamens in Potamogetonaceae, and (ii) the assumption that floral reduction from Alismatales-like flowers is "not probable."No feature was given by Miki (1937) to differentiate "bracts" from tepals (axillary shoots, phyllotaxy) or to indicate that the floral units of Patamagetan were composed of several flowers, instead of representing single flowers lacking a perianth.In addition, no developmental data were provided.Only Najas L. and Patamagetan were considered by Miki (1937).The most significant data are provided by Uhl (1947), who concluded that the floral units of Scheuchzeriaceae, Aponogetonaceae, Juncaginaceae, and Potamogetonaceae were composed of radial "staminate units," and one to several central pistillate flowers.The staminate units consisted of a single stamen representing an entire reduced flower subtended by a bract (the tepal, in this study).The "floral unit" of all of these taxa was considered to be composed of highly reduced inflorescences (staminate units) and therefore to represent a pseudanthium in the commonly used sense (see Introduction).This pseudanthial concept (Uhl 1947) is based on three observations: ( 1) the vasculature of the "staminate unit" leaves the rest of the floral vasculature as one strand, which then divides into two; (2) the staminate unit is often supported by a common elevated base (Potamogeton), or in some taxa (Triglochin subgen.Cycnogeton (Endl.)Buchenau & Hieron., Scheuchzeria; Uhl 1947) the inner whorl of staminate units inserts distally of the stamens of the outer whorl (see also Rudall 2003) and are shed as a unit (stamen and tepal together as "staminate unit" in Triglochin; Uhl 1947); and (3) reductions of flowers often involve merosity of all whorls (sectors consisting of tepal, stamen, and carpel).The study by Uhl (1947) included a diverse array of taxa, and its conclusions were based almost entirely on vasculature of mature stages.However, no developmental data were provided, and series of organ initiation were not presented.Uhl (1947) did not consider the possibility of unequal intercalary growth or unidirectional flower development.
In Potamogeton and Triglochin the initiation sequence of the organs on the floral units corresponds perfectly with that of flowers consisting of whorls of outer tepals, inner tepals, outer stamens, inner stamens, outer carpels, inner carpels (Charlton 1981;Posluszny 1981;this study).Any position of outer tepals seemingly distal from outer stamens can be explained by unequal intercalary elongation and unilateral flower development, which also can confuse the recognition of whorls in other taxa (Tucker 1979(Tucker , 1981(Tucker , 1985;;Liang andTucker 1989, 1990;Tucker et al. 1993;Tucker and Douglas 1996;Buzgo and Endress 2000).

Terminal Peloria and Pseudanthia
In T. maritima and other species of Triglochin, a flowerlike terminal structure occurs, which is considered a terminal flower by most authors (Hill 1900;Aston 1973Aston , 1993a, b;, b;Lieu 1979;Posluszny et al. 1986;Harden 1993), but this structure is considered absent by Uhl (1947) and Charlton (1981).The terminal structure is larger than lateral flowers, probably because it is formed by a larger primordium (the inflorescence SAM) than lateral flowers.The terminal structure is radially symmetrical (this study), but not aberrant, and therefore the term peloria may be inaccurate.For example, the terminal structure is initiated with a distinct lag in development after the lateral distalmost flowers of the inflorescence, causing a gap between the insertions of lateral flowers and the first organs of the terminal structure.Consequently, there is an abrupt transition from lateral floral primordia to floral organs toward the apex, although the subapical flowers show reduction on the adaxial side, as in Houttuynia andAcorus (Tucker 1979(Tucker , 1981(Tucker , 1985;;Liang andTucker 1989, 1990;Tucker et al. 1993;Tucker and Douglas 1996;Buzgo and Endress 2000).
Our observations support a correlation between smaller inflorescences, proportionally stronger reduction of the adaxial organs in distal flowers, and the formation of terminal flowers that differ from the lateral ones, as suggested previously by Buzgo and Endress (2000) and Buzgo (2001) for Acorus.Members of the Triglochin procera group (Aston 1973(Aston , 1993a, b) , b) grow vigorously, forming inflorescences in which the flower-bearing portion is up to 30 em long, with flowers more than 8 mm in diameter with distinct pedicels, and the terminal flower resembles the lateral flowers.Triglochin palustris and T. striata have much smaller inflorescences than T. maritima.Flowers of T. striata possess distinct pedicels.However, in many cases not all floral organ whorls are trimerous, and whorls are sometimes difficult to distinguish.In the distal portion of the inflorescence, flowers are strongly reduced on the adaxial side (T.striata), sometimes leaving only one median tepal, which is bract-like.
The Australian group of annual species (T.turrifera Ewart, T. centrocarpum Hook., T. hexagona J. M. Black, T. calcitrapum Hook.) (Aston 1973;Harden 1993;K. Meney pers. comm.) has been reported to have extremely small inflorescences.In at least some of these species the lateral flowers are unisexual, with only the terminal flower being bisexual.This "completeness" of the terminal flower may result from a larger meristem as compared with the lateral primordia (as in T. maritima), and thus represents a distinct difference between terminal and lateral flowers, similar to that of the larger peloria in Acorus and Saururaceae.All Juncaginaceae and Aponogetonaceae may be affected by a convergent tendency of adaxial flower reduction, leading to similar transitions between bracts and tepals, between inflorescence and flower.Understanding the transition of inflorescence and flower in alismatids is crucial for elucidation of floral evolution in early monocots, and even for basal angiosperms, in general, because similar features also appear in magnoliids (Saururaceae; Tucker 1979Tucker , 1981Tucker , 1985;;Liang andTucker 1989, 1990;Tucker et al. 1993;Tucker and Douglas 1996) and basal eudicots (Buxaceae; von Balthazar and Endress 2002).
In Triglochin maritima, we can recognize the terminal flower.ClaBen-Bockhoff (1990) suggests a "paedomorphic trend," in which the progressive reduction of the inflorescence SAM results in the abbreviation (heterochrony) of the developmental process of lateral primordia, rendering them floral organs and resulting in an aberrant flower (peloria) at the inflorescence apex.This abbreviation reflects the "specific predisposition of the taxa concerned" required by ClaBen-Bockhoff (1990) for the convergent evolution of pseudanthia.This requirement is met in Acarus and some Saururaceae (above).However, in the clade comprising Juncaginaceae and Potamogetonaceae, this predisposition is only represented by the reduction of the lateral floral units (flowers); we find no signs or intermediate cases indicating the reduction and rearrangement of floral units to lateral pseudanthia.Yet, this extension of floral characteristics may be represented in the partial extension activity of genes responsible for the determination of flower meristem identity, i.e., upstream from B-class genes.

Molecular Genetic Perspective
The lateral flowers of Triglochin apparently are not defined by a flower-subtending bract.The inflorescence starts development as one large meristem and the apex of this meristem turns into a flower.How far does floral identity reach out into the inflorescence?When does the transition of the inflorescence apical meristem to a flower primordium occur?How far is the assumption of a homeotic change of flower features into the supporting inflorescence shoot supported by concepts or data of molecular development?A test for floral features in inflorescence development is provided by genes that are considered strictly floral (Hypothesis 4).The gene we used to test this hypothesis is an ortholog of Antirrhinum L. DEFICIENS (DEF) and Arabidopsis APETALA3 (AP3), a member of the B-class MADS-box gene family (e.g., Bowman et al. 1989;Sommer et al. 1990;Coen and Meyerowitz 1991;Soltis et al. 2002;Kramer et al. 2003;Stellari et al. 2004).Orthologs of AP3 are strictly regulated downstream of LEAFY and A-class genes, both of which are required for the conversion of a shoot into a flower (Coen et al. 1990;Schwarz-Sommer et al. 1990;Coen and Carpenter 1992;Huala and Sussex 1992;Singer et al. 1992;Weigel et al. 1992;Weigel and Nilsson 1995;Blazquez et al. 1997;Hempel et al. 1997;Lee et al. 1997;Ma 1997Ma , 1998;;Parcy et al. 1998;Weigel 1998;Wagner et al. 1999;Berleth et al. 2000;Fernindiz et al. 2000;Frohlich and Parker 2000;Yu et al. 2000;Araki 2001;Coen and Langdale 2001;Pena et al. 2001;Soltis et al. 2002).Because AP3 is only transcribed after a flower-specific developmental pathway has been activated, the significant occurrence of its mRNA is a conservative indicator of floral meristem identity.
The "sliding boundaries" concept of the ABC-class model (Kramer et al. 2003) predicts that in Triglochin maritima, B-class genes would be expressed only in stamens, but not in either whorl of sepaloid tepals, bracts, or inflorescence main shoot (although Kramer et al. 2003 specify that in Aquilegia L. one of the three copies of AP3 is the major factor for petaloid features, while the other two have expression patterns that are less correlated with petaloid features).For older developmental stages of T. maritima, our results generally correspond to this concept, although AP3 is also weakly transcribed in the tips of tepals, very young carpels, and vascular strands.These expression patterns are in greater agreement with the concept of "fading borders" of gene expression described for basal angiosperms (Buzgo et al. 2004)."Fading borders" suggests that in basal angiosperms the functions of floral transcription factors are not restricted to only one zone or whorl of organs, but exhibit a gradual transition from the periphery to the center of the flower.Corresponding to the often spiral or irregular floral phyllotaxy in basal angiosperms (instead of a few distinct whorls of floral parts, as in eudicots), "fading borders" explains the gradual transition of morphological features, such as features commonly associated with stamens or petals (e.g., papillae, thickening, secretion, color).The concept does not specify how the gradual transition in gene function is achieved (duration of gradual expression, diversified function of gene copies [Stellari et al. 2004 ], transcription rate, post-transcriptional modification, or protein-affinities).Although "fading borders" was developed with a focus on B-class genes, other genes may exhibit a similar transition in expression pattern.The hypothesis of "fading borders" is supported by studies employing relative-quantitative gene expression (Kim et al. 2003(Kim et al. , 2005)).In particular, B-class genes are expressed in tepals, stamens, and carpels of several basal angiosperms that exhibit gradual transitions between adjacent floral organs.
For very young inflorescences, our expression results are puzzling in that the mRNA of AP3 appears to be present not only in stamens, but also throughout the entire inflorescence (and even in leaves).The absence of signal from the negative controls (sense probes) supports the interpretation that the apparent expression is a true signal.One explanation could be that B-class genes are expressed in other meristems as well, for example, in procambial strands.B-class gene transcripts have been reported from procambial strands in other studies (e.g., Skipper 2002) and also occur in the procambial strands of older inflorescences of Triglochin maritima (this study).However, the future parenchyma of the leaves and inflorescence also stains strongly in leaves, even at a stage where the intercellular spaces have begun to form.Based on our results, it appears as if AP3 is more widely expressed in the inflorescence of T. maritima than in other plants examined to date.Because of upstream regulation by floral meristem identity genes (see above), this broad expression of AP3 suggests that at early stages of development the axis of the inflorescence may share some identity with that of a flower.This is in accordance with the transition of the inflorescence SAM into a flower: the identity of the entire young inflorescence is "floral" and the restriction of this identity to lateral meristems only occurs later.This pattern is consistent with reports of transcription of SEPALLATA in inflorescences of Oryza sativa (Malcomber and Kellogg 2004) and could explain similar phenomena in other monocots and basal angiosperms.If our interpretation of this pattern of AP3 expression is correct, our results would expand the concept of "fading borders" beyond the limits of the flower to the inflorescence.