Contributions to the systematic position of Hydrolea(Hydroleaceae) based on floral development

C. Erbar1, S. Porembski

2, and P. Leins

1

1Universitat Heidelberg, HIP - Biodiversitat und Pflanzensystematik, Heidelberg, Germany2Universitat Rostock, Institut fur Biodiversitatsforschung: Allgemeine und Spezielle Botanik, Rostock,Germany

Received June 29, 2004; accepted October 27, 2004Published online: March 24, 2005� Springer-Verlag 2005

Abstract. In the flower of Hydrolea palustris,unusually orientated with one sepal abaxially,organogenesis starts in following sequence: fivesepals (2/5 sequence), five simultaneously initiatedalternating petals, five episepalous stamens, two(seldom three) carpels forming a coenocarpousseptate gynoecium. The two carpels are orientatedrather in the diagonal floral plane than in themedian one. Petal primordia fuse very late byforming interprimordial bridges (late sympetaly!).Many ovules develop on considerably widenedplacentas. On the very basis of the superior ovary afive-humped nectary disk is formed.

Within Solanales (APG II 2003) late sympetaly,an intrastaminal disk and a 2-carpellate, septate,superior ovary are found in Hydroleaceae, Con-volvulaceae, and Solanaceae. Enlarged axile pla-centas characterize Hydrolea, Solanaceae, andSphenocleaceae but Sphenocleaceae differ consid-erably by early sympetaly. Montiniaceae differ byhaving a choripetalous corolla. Nearly diagonalorientation of the carpels seems to relate Hydroleaclose to Solanaceae, but the orientation of the calyxis different.

Key words: Flower morphology, flower ontogeny,Hydrolea, Solanales.

Introduction

The genus Hydrolea, comprising 11 hygro-phytic species, is pantropically distributed. Itwas described by Linne (1762) and assigned to‘‘Ordo Convolvuli’’ by De Jussieu (1789).Brown (1818; cited after Ferguson 1999)removed Hydrolea from ‘‘Ordo Convolvuli’’and Hydrophyllum from ‘‘Ordo Boragineae’’creating two new families ‘‘Hydroleae’’ and‘‘Hydrophylleae’’, close to ‘‘Boragineae’’. Thisseparation of family Hydroleaceae was main-tained by Choisy (1833 a, b) in the firstmonograph of Hydrolea. It was Gray (1875)who merged Hydroleaceae and Hydrophylla-ceae, recognizing a single family Hydrophyll-aceae (with different tribes). His classificationwas adopted throughout the 20th century (e.g.Peter 1897, Brand 1913, Cronquist 1981,Takhtajan 1997). Constance (1963), however,questioned whether Hydrophyllaceae in thebroad sense are a natural group, and specu-lated that Hydrolea should be maintained as aseparate family (as in 1818 to 1875). Indeed,due to molecular data, Hydrolea has now notonly been separated from Hydrophyllaceae,

Plant Syst. Evol. 252: 71–83 (2005)DOI 10.1007/s00606-004-0263-7

but the two groups are no longer very closerelatives (Cosner et al. 1994, Ferguson 1999,Savolainen et al. 2000, Soltis et al. 2000).Thorne (2001) treated Hydroleaceae as alargely aquatic family adjacent to the Hydro-phyllaceae. However, in the recent classifica-tions of APG (1998, 2003), based on variousmostly molecular phylogenetic analyses, thefamily Hydroleaceae is part of the orderSolanales, whereas the rest of the Hydrophyll-aceae is found within the Boraginaceae (notassigned to any order because there is so far noclear placement of the family).

The intention of this study is to contributeto the systematic position of Hydrolea by floralmorphological and developmental characterssince floral ontogenetical studies on Hydroleawere totally lacking until now. We haveinvestigated Hydrolea palustris which is re-stricted to tropical Africa from Guinea toCongo and Madagascar (Davenport 1988).

Material and methods

Flowers and flower buds of Hydrolea palustris(Aublet) Raeuschel (= H. glabra Schumach. &Thonn.¼H. guineensisChoisy; seeDavenport 1988)have been collected and fixed in 70% ethanol byStefan Porembski at a natural site: Cote d’Ivoire, inpool on rock outcrop, 9 km north of Lolobo (6�58’N5�25’W). A voucher of Hydrolea palustris is depos-ited in the herbarium of the University of Rostock(ROST; S. Porembski No. 2651).

For scanning electron microscopy, the budswere dehydrated in dimethoxymethane, criticalpoint dried using liquid CO2, mounted on stubs,and coated with gold. They were viewed with aLeitz (AMR 1200B) scanning electron micro-scope.

For histological sections, the buds were dehy-drated in an alcohol series and embedded in amethacrylate resin (Kulzer Technovit 7100). Theresin blocks were cut at 6 lm and the tissue sectionsstained with toluidine blue.

For calculating the P/O ratio all pollen grainsof each anther in ten flowers were counted inaniline blue-stained anther squashes under the lightmicroscope. Also the number of ovules wascounted in each ovary of the flowers used for thepollen counts.

Observations

The blue flowers of Hydrolea palustris areborne on short branches in axillary clusters ofmostly three or sometimes four or five (Figs. 1–2). The tetracyclic flowers are only weaklysynsepalous, have a short rotate to campanu-late five-lobed corolla tube, five alternatingstamens and a short stamen-corolla tube(Fig. 13). The bilocular (seldom trilocular)ovary is superior and bears numerous ovuleson thickened placentae. The fruit is a capsule.

The flowers have two prophylls (= bracte-oles) in whose axils additional flowers mayarise. The early floral development could onlybe studied in flowers that are formed in the axilof the prophylls of the terminal flowers of acluster. Flower buds sometimes appear some-what moved away from the median plane ofthe bract. After prophyll (bracteole) forma-tion, five sepal primordia are initiated in a 2/5-sequence (Fig. 3a–b). The first sepal to arise isadaxially in the flower diagonal plane. Thesecond sepal primordium originates abaxially.Thus calyx and flower orientation respectivelydiffer from that usual in most asterids (Figs.3a, 4a, 16a).

Alternating to the sepal primordia fivepetal primordia are formed simultaneouslyon the rim of the floral apex (Fig 3b). Theinterprimordial gaps between the petal primor-dia are relatively large, even when the stamenprimordia have been initiated alternating tothem (Fig. 4b). The petal primordia are stillseparate from each other when the gynoecialprimordium is formed (Fig. 5a–b). Only whenthe petals have enlarged considerably, curvedhyponastically and show an imbricate aestiva-tion at the top of the corolla, the shape of theinterprimordial areas changes: The petalsextend laterally at their bases and their mar-ginal meristems fuse more or less abruptlybehind the stamen bases (Fig. 6a–c). Equalgrowth in the petal bases and interprimordialridges (Fig. 7) results in a short corolla tube(Fig. 13b).

Before the corolla lobes enclose the floralapex, two (seldom three) carpel primordia are

72 C. Erbar et al.: Floral development in Hydrolea palustris

Figs. 1–4. Hydrolea palustris. 1–2, Part of inflorescence. 3 Floral primordium with sepals initiated in a 2/5-sequence (as labelled), petal primordia arise simultaneously and separately from each other. 3a Flower bud inside view. 3b Same flower bud seen from above. 4 Five stamen primordia originate alternating to the petalprimordia; calyx partly removed. 4a Flower bud seen from above. 4b Same flower bud magnified and in sideview. B bract, P petal primordium, Pr prophyll (bracteole), S sepal primordium

C. Erbar et al.: Floral development in Hydrolea palustris 73

Figs. 5–7. Hydrolea palustris. 5a Calyx, corolla, and stamens are established, formation of the gynoecium(arrows); calyx partly removed. 5b Same flower bud in side view showing the still separate petals. 6 Formation ofsympetaly: the petals enlarge laterally at their bases and fuse behind the dorsal flanks of the stamens.6b-cMagnificationsof 6a; identical petals are labelled.7Part of olderflowerbud: petals connectedbya small ridge.B bract, P petal primordium, Pr prophyll (bracteole), S sepal primordium, St stamen primordium

74 C. Erbar et al.: Floral development in Hydrolea palustris

Figs. 8–12. Hydrolea palustris. 8–9 Early stages of gynoecium development. 10 Differentiation of five nectarybulges (arrows) around the ovary. 11 Older, tricarpellate ovary with three styles and annular, five-humpednectary disk. 11b Close up. Part of the nectary disk with nectary slits (arrows). 12 Basis of an older flower bud,corolla and stamen-corolla tube partly removed.A anther, C carpel primordium, CT corolla tube, F filament,Nnectary, O ovary, P petal, St stamen, St-CT stamen-corolla tube, Sty style

C. Erbar et al.: Floral development in Hydrolea palustris 75

initiated which from the beginning are fusedlaterally forming a ring (Figs. 5a, 8). Soon theybecome horseshoe-shaped and the lateralflanks run in a common protuberance(Fig. 9) that together with the ovary growsup forming the high septum. The gynoecium issynascidiate (in the sense of Leinfellner 1950)for most of its length and symplicate in itsupper region (Figs. 14, 16). The carpels (if two)are not exactly orientated in the median planeof the flower, but rather in the diagonal plane(Fig. 16a). The carpel tips develop into formost of their length separate styles (Figs. 10,11a). At maturity the styles with their terminalcapitate stigmas bend inwards (Figs. 13, 14).Sometimes three carpels are initiated and thusthree styles top the ovary (Fig. 11a). At thevery base of the ovary five bulges alternatingwith the stamens arise (Fig. 10) which developinto an annular, five-humped nectary disk(Figs. 11a, 12, 15a–b). On the upper side ofthis disk nectary slits can be observed (Figs.11b, 15a–b). Nectary tissue contains cells withdense plasma, embedded in more vacuolizedtissue.

The transversal to diagonal septum pro-duces a thick, U-shaped placenta in each locule(Figs. 14, 16a–d) nearly filling up the latter.Each placenta bears numerous ovules (200–300; on average 478 in the whole ovary,Table 1).

In the adult flower, under the insertion areaof filaments and corolla tube a stamen-corollatube has formed (Figs. 11a, 12, 13b) which hasabout the same length as the short corollatube. The free petal lobes, however, surpassboth these tubes considerably in length.

The stamens with broad deltoid bases(Fig. 13b) have basifixed calcarate anthersdue to basal prolongation of the pollen sacs(Figs. 13a–b). Since the anther is delimitedfrom the filament by a thin joint, the anthersare versatile bending their tips toward thecentre of the flower. Anther dehiscence – eachtheca opens slowly by a longitudinal slit –takes place nearly simultaneously with theoccurrence of a stigmatic exudate (perhapsslightly ahead of the female phase).

The five stamens of a flower produce onaverage 18886 pollen grains so that the P/Oratio is about 40 (Table 1).

Discussion

On the systematic affinities of Hydrolea-

ceae. The split of Hydrophyllaceae, with Hy-drolea assigned to Convolvulaceae and Hydro-phyllum and other genera to Boraginaceae, wasfirst proposed by De Jussieu (1789), but waslater dismissed (Gray 1875). Characters sepa-rating Hydrolea from Hydrophyllaceae are thebilocular ovaries with axile placentation versusunilocular with parietal placentation in Hyd-rophyllaceae. All molecular studies (e.g. Chaseet al. 1993, Cosner et al. 1994, Ferguson 1999,Savolainen et al. 2000, Soltis et al. 2000,Albach et al. 2001) have supported that Hyd-rophyllaceae (excluding Hydrolea) are nestedwithin Boraginaceae (position of Boragina-ceae-Hydrophyllaceae among Gentianales, So-lanales, and Lamiales is not completelyresolved), whereas Hydrolea falls outside theBoraginaceae. Chase et al. (1993) mentionedthat the rbcL-sequence of Hydrolea is highlydivergent from all other taxa and the positionof the genus is very different as other taxa areadded or removed. In most of the molecularstudies Hydrolea occupies a position nearConvolvulaceae and Solanaceae (Albach et al.2001, Cosner at al. 1994, Soltis et al. 2000,Savolainen et al. 2000). In a phylogeneticanalysis of the rbcL gene sequence in nearlyall eudicot families the Solanales are paraphy-letic to the Lamiales (Savolainen et al. 2000).Based on a combined atpB/rbcL/18S rDNAanalysis the Solanales received a better support(Soltis et al. 2000).

According to APG (1998, 2003) Solanalesconsist of five families, namely Convolvulaceae,Hydroleaceae, Montiniaceae, Solanaceae, andSphenocleaceae. For Sphenocleaceae andMon-tiniaceae this is an unexpected position. Mon-tinia appears as sister to Sphenoclea andHydrolea (Cosner et al. 1994, Fay et al. 1998)or as sister to Hydrolea if Sphenoclea is notincluded in the study (Chase et al. 1993, Soltis

76 C. Erbar et al.: Floral development in Hydrolea palustris

Table 1. Calculation of P/O ratio

Flower Anther Pollen/Th P/Anther P/Flower O/Flower P/O P/O

1 1 2184 4368 18464 508 36,35 mean 39,66

1 2 2240 4480 se 8,951 3 1296 2592 max 58,491 4 1292 2584 min 29,481 5 2220 44402 1 2240 4480 25176 536 46,97

2 2 2160 43202 3 2728 54562 4 2772 55442 5 2688 53763 1 2600 5200 23280 398 58,49

3 2 2520 50403 3 1872 37443 4 2088 41763 5 2560 51204 1 1904 3808 18016 458 39,34

4 2 1824 36484 3 1296 25924 4 1664 33284 5 2320 46405 1 1584 3168 15536 527 29,48

5 2 1280 25605 3 1560 31205 4 1472 29445 5 1872 37446 1 2112 4224 21968 603 36,43

6 2 2352 47046 3 2000 40006 4 2040 40806 5 2480 49607 1 2808 5616 22128 444 49,84

7 2 2592 51847 3 1848 36967 4 2280 45607 5 1536 30728 1 1040 2080 12520 409 30,61

8 2 748 14968 3 1320 26408 4 1792 35848 5 1360 27209 1 832 1664 11912 388 30,70

9 2 1156 23129 3 1216 24329 4 1520 30409 5 1232 246410 1 2048 4096 19856 517 38,41

10 2 2040 408010 3 1792 358410 4 1728 345610 5 2320 4640

C. Erbar et al.: Floral development in Hydrolea palustris 77

et al. 2000, Albach et al. 2001) thus forming asister clade to the Solanales.

We will try to contribute to the phyloge-netic discussion with flower morphological andnew ontogenetical data. The 3/4-inferior ovaryand the circumsessile capsule separate Sphe-noclea from Hydrolea and the Solanales (Erbar1995). Morphological characters like intrasta-minal disk and 2-carpellate, septate, superiorovary with numerous ovules on thickened

placentas are congruent in Hydrolea, Convol-vulaceae and Solanaceae. A rather diagonalthan median orientation of the two-carpellategynoecium brings Hydrolea near Solanaceae,but it differs considerably in its whole floralorientation from this family. Whereas inSolanaceae and in most asterids one sepal liesin the upper median plane of the flower, inHydrolea palustris the upper median planepasses more or less between two adaxial sepals.

Figs. 13–14. Hydrolea palustris. 13 Open flower in different views, corolla partly removed. 13a Nearly top view.13b Side view, showing the deltoid dilated filament bases. The arrow (like in Fig. 13a) indicates one of the fivecanalized accesses to the nectar. 13c Top view of the flower, note receptive stigmas and open anthers. 14 Lateralview into the ovary (ovary wall partly removed), ovules formed on a U-shaped placenta. A anther, CT corollatube, F filament, O ovary, P petal, St-CT stamen-corolla tube, Sty style

78 C. Erbar et al.: Floral development in Hydrolea palustris

We can find the same orientation only in theasterid families Cyphiaceae and Lobeliaceae,members of Campanulales (Leins and Erbar2003, 2005). Certainly the deviating floralorientation in Hydrolea has evolved indepen-dently. Concerning floral orientation of Hy-drolea spinosa illustrated by Eichler (1875)there is no agreement with our results. How-ever, Eichler stated that he often found thecarpels of Hydrolea spinosa orientated obli-quely to the median plane of the flower.

Montinia (variously treated as genus orsubfamily or family) has usually been alliedwith members of Saxifragaceae s.l. (Schulze-Menz 1964, Cronquist 1981) or Hydrangeales(Takhtajan 1997, Thorne 2001). Due to thepresence of iridoid compounds the familyMontiniaceae has been placed within broadlycircumscribed Cornales (Dahlgren et al. 1981,Dahlgren 1983). Nowadays, as already men-tioned, part of the Solanales (APG 2003) thefamily Montiniaceae differs from the Solanalesin important morphological characters asinferior ovary, intrusive parietal (subaxile)placentation and drupes (Schulze-Menz 1964,Stevens 2003).

In having numerous ovules on thickenedplacentae (Figs. 12, 18–19 in Erbar 1995)Sphenoclea resembles Hydrolea and Solana-ceae. In contrast to the position near theSolanales with Hydrolea as sister group assuggested on the basis of the rbcL data (Cosneret al. 1994, Gustafsson and Bremer 1995,Gustafsson et al. 1996, Savolainen et al.2000; in the combined analysis of Soltis et al.2000 Sphenoclea was not involved), Sphenocleausually has been treated as a genus of theCampanulaceae or as a closely related distinctfamily (e.g. Cronquist 1981; Thorne 1983,2001; Takhtajan 1997). A chemical synapo-morphy of the Campanulales-Asterales is thepossession of inulin. Therefore inulin occur-rence in Sphenoclea supports a position withinthe Campanulales-Asterales (Porembski andKoch 1999), despite the fact that inulin alsooccurs in perennial Boraginaceae (Pollard andAmuti 1981). It is not known if inulin ispresent or absent in Hydrolea.

The floral developmental character ‘‘onto-geny of sympetaly’’ proved to be valuable forsystematic considerations (Erbar 1991, Erbarand Leins 1996). Solanales belong to theasterid I clade (Chase et al. 1993) which ispredominantly characterized by late sympetaly(petal primordia arise separately, and becomeconnected later on the dorsal side of thestamens by interprimordial protrusions). TheAsterid II clade, in which the Campanulales-Asterales fall, is characterized by early sym-petaly throughout (petal primordia arise on aring primordium and the stamen primordiaoriginate in front of the interprimordial con-nections).

The monogeneric family Sphenocleaceaestands for an example in which molecular andfloral developmental data show no congruence(Erbar 1995): Hydrolea palustris and Spheno-clea zeylanica, both tropical rice weeds, aresister groups in molecular based cladogramsbut show a quite different mode of corolla tubeformation: Sphenoclea is early sympetalous,whereas Hydrolea is late sympetalous. In itslate sympetaly Hydrolea resembles the otherSolanales (see below).

The family Montiniaceae is another exam-ple in which a relationship with Solanales,indicated by molecular data, is not supportedby floral ontogenetical and anatomical studies(Ronse Decraene et al. 2000). As regardscorolla development, the petal primordia ariseand remain separately from each other (RonseDecraene et al. 2000) so that Montinia ischoripetalous during development as well asin adult stages. Chase et al. (1993) suggestedthat the sympetalous corolla of Montinia hasbeen secondarily lost. Ontogeny, however,cannot give support for this idea.

The remaining Solanales – Convolvulaceaeand Solanaceae – are late sympetalous.Whereas in Convolvulaceae (e.g. Ipomoeacoccinea, Figs. 14–14 in Erbar and Leins1996) the initially free petals become connectedby the formation of a bridge like in Boragin-aceae (Figs 22–30 in Erbar 1991, Figs. 12–13 inErbar and Leins 1996) and Hydrophyllaceae(Fig. 35 in Erbar 1991, Figs. 16–17 in Erbar

C. Erbar et al.: Floral development in Hydrolea palustris 79

and Leins 1996), in Solanaceae (Figs. 22–23 inErbar and Leins 1996) the mode of corollatube development is transitional betweenbridge formation and successive extension of

the petal bases. This latter mode is also foundin Hydrolea palustris.

Thus morphological and floral ontogene-tical features do not contradict a position of

80 C. Erbar et al.: Floral development in Hydrolea palustris

Hydroleaceae near Solanaceae and Convolvul-aceae and an inclusion of the family in theSolanales. A position of Sphenocleaceae andMontiniaceae near Hydroleaceae, however,does not find any support in floral morpho-logical data till now.

Some aspects of floral ecology. Our inves-tigations have shown that Hydrolea palustrishas a well-developed disk secreting somenectar through slits (Fig. 11), a fact that hasbeen reported (Hydrolea spinosa: Willis 1895)as well as negated (Davenport 1988, Stevens2003). In floras (e.g. Baker and Brown 1906)the ‘‘hypogynous’’ disk in Hydrolea guineensis(= H. palustris) has been described as incon-spicuous.

Hydrolea is mentioned to be autogamous(Brand 1913, Mabberley 1997). This maydepend on the observations by Willis (1895)who described Hydrolea spinosa as – at leastwhen cultivated in England – excellentlyadapted to self-fertilization (stamens areusually found touching the stigmas whenthe flower opens). Every flower sets seed,even in the absence of insects. Nevertheless,he stressed the chance given of cross-fertil-ization. Our observations on the fixed mate-rial of Hydrolea palustris are congruent withthat of Willis (1895). The open versatileanthers are in close position to the receptivestigmas, and self-pollination seems possible,

even mediated by the action of visitinginsects. The P/O ratio of 40 is only slightlyhigher than the ratio of 30 postulated byCruden (1977) for obligate autogamy ingeneral. Our preliminary results on the pol-lination of Hydrolea palustris predict a fac-ultative autogamy: Autogamy may occur, butthe showy flowers presenting some amountof nectar may attract pollinators. The broaddeltoid filament bases by which the access tothe nectar is restricted to five small entrances(arrows in Fig. 13a–b) can also be inter-preted as an adaptation to insect pollination.The flower of Hydrolea palustris appears tofunction as a revolver flower with its cana-lized access to the nectar.

References

Albach D. C., Soltis P. S., Soltis D. E., OlmsteadR. G. (2001) Phylogenetic analysis of the asteridsbased on sequences of four genes. Ann. MissouriBot. Gard. 88: 163–212.

APG (Angiosperm Phylogeny Group) (1998) Anordinal classification for the families of floweringplants. Ann. Missouri Bot. Gard. 85:531–553.

APG II (Angiosperm Phylogeny Group) (2003) Anupdate of the Angiosperm Phylogeny Groupclassification for the orders and families of flower-ing plants:APGII. Bot. J. Linn. Soc. 141: 399–436.

Baker J. G., Brown N. E. (1906) Order LXXXVIII:Hydrophyllaceae. In: Thiselton-Dyer W. T. (ed.)Floral of Tropical Africa IV.2: 1–5. Reeve &Co., Ashford, Kent.

Brand A. (1913) Hydrophyllaceae. In: Engler A.(ed.) Das Pflanzenreich IV. 251: 1–210. Engel-mann, Leipzig.

Brown R. (1818) Observations, systematical andgeographical, on Professor Christian Smith’scollection of plants from the vicinity of the riverCongo. In: Tuckey J. K. (ed.) Narrative of anexpedition to explore the river Zaire. JohnMurray, London, pp. 420–485.

Chase M. W., Soltis D. E., Olmstead R. G.,Morgan D., Les D. H., Mishler B. D., DuvallM. R., Price R. A., Hills H. G., Qiu Y. L.,Kron K. A., Rettig J. H., Conti E., Palmer J.D., Manhart J. R., Sytsma K. J., Michaels H.J., Kress W. J., Karol K. G., Clark W. D.,Hedren M., Gaut B. S., Jansen R. K.,

Figs. 15–16. Hydrolea palustris. 15a Longitudinalsection through the base of a flower bud showing theannular nectary around the ovary. 15b Close up ofthe nectary with a nectary slit (arrow). 16 Crosssections through the flower at different levels. 16aFlower cross section; note the imbricate aestivation ofthe corolla and the orientation of the flower; theasterisk indicates the upper median plane of theflower. 16b Ovary of 16a at higher magnification:symplicate zone of the ovary with placenta on thelateral carpel flanks. 16c At the level of the septumwith still bilobed placenta in each locule. 16c At thelevel of the synascidiate zone with uniform placentain each locule. C carpel, N nectary, O ovary, P petal,Pl placenta, S sepal, Se septum, St stamen; bar¼ 200lm in Figs. 15a, 15b, 16a, 100lm in 16b, 16c, 16d

b

C. Erbar et al.: Floral development in Hydrolea palustris 81

Kim K. J., Wimpee C. F., Smith J. F.,Furnier G. R., Strauss S. H., Xiang Q. F.,Plunkett G. M., Soltis P. S., Swensen S. M.,Williams S. E., Gadek P. A., Quinn C. J.,Equiarte L. E., Golenberg E., Learn Jr. G. H.,Graham S. W., Barrett S. C. H., DayanandanS., Albert V. A. (1993) Phylogenetics of seedplants: an analysis of nucleotide sequencesfrom the plastid gene rbcL. Ann. MissouriBot. Gard. 80: 528–580.

Choisy J. D. (1833a) Description des Hydroleacees.Mem. Soc. Phys. Geneve 6: 95–122.

Choisy J. D. (1833b) Description des Hydroleacees.Ann. Sci. nat. 30: 225–251.

Constance L. (1963) Chromosome numbers andclassification in Hydrophyllaceae. Brittonia 15:273–285.

Cosner M. E., Jansen R. K., Lammers T. G. (1994)Phylogenetic relationships in the Campanulalesbased on rbcL sequences. Plant Syst. Evol. 190:79–95.

Cronquist A. (1981) An integrated system ofclassification of flowering plants. ColumbiaUniv. Press, New York.

Cruden R. W. (1977) Pollen-ovule ratios: a conser-vative indicator of breeding systems in floweringplants. Evolution 31: 32–46.

Dahlgren R. (1983) General aspects of angiospermevolution and macrosystematics. Nord. J. Bot. 3:119–149.

Dahlgren R., Jensen S. R., Nielsen B. J. (1981) Arevised classification of the angiosperms withcomments on correlation between chemical andother characters. In: Young D. A., Seigler D. S.(eds.) Phytochemistry and angiosperm phylog-eny. Praeger, New York, pp. 151–204.

Davenport L. J. (1988) A monograph of Hydrolea(Hydrophyllaceae). Rhodora 90: 169–208.

Eichler A. W. (1875) Bluthendiagramme I. Engel-mann, Leipzig.

Erbar C. (1991) Sympetaly - a systematic character?Bot. Jahrb. Syst. 112: 417–451.

Erbar C. (1995) On the floral development ofSphenoclea zeylanica (Sphenocleaceae, Campa-nulales)-SEM-investigations on herbarium mate-rial. Bot. Jahrb. Syst. 117: 469–483.

Erbar C., Leins P. (1996) Distribution ofthe character states ‘‘early’’ and ‘‘late sympet-aly’’ within the ‘‘Sympetalae Tetracyclicae’’ andpresumably related groups. Bot Acta 109: 427–440.

Fay M. F., Olmstead R. G., Richardson J. E.,Santiago E., Prance G. T., Chase M. W. (1998)Molecular data support the inclusion of Duck-eodendron cestroides in Solanaceae. Kew Bull.53: 203–212.

Ferguson D. M. (1999) Phylogenetic analysis andrelationships in Hydrophyllaceae based on ndhFsequence data. Syst. Bot. 23: 253–268.

Gray A. (1875) A conspectus of the North Amer-ican Hydrophyllaceae. Proc. Amer. Acad. 10:312–32.

Gustafsson M. H. G., Backlund A., Bremer B.(1996) Phylogeny of the Asterales sensu latobased on rbcL sequences with particular refer-ence to the Goodeniaceae. Plant Syst. Evol. 199:217–242.

Gustafsson M. H. G., Bremer, K. (1995) Morphol-ogy and phylogenetic interrelationships of theAsteraceae, Calyceraceae, Campanulaceae, Goo-deniaceae, and related families (Asterales).Amer. J. Bot. 82: 250–265.

Jussieu A. L. de (1789) Genera plantarum secun-dum ordines naturales disposita 129. Herissantand Barrois, Paris.

Leinfellner W. (1950) Der Bauplan des synkarpenGynozeums. Osterr. Bot. Z. 97: 403–436.

Leins P., Erbar C. (2003) The pollen box in Cyphia(Cyphiaceae). Int. J. Plant Sci. 164 (5 Suppl.):S321–S328.

Leins P., Erbar C. (2005) Floral morphologicalstudies in the South African Cyphia stenopetalaDiels (Cyphiaceae). Int. J. Plant Sci. (in press).

Linne C. von (1762) Species plantarum 1: 328, 2nd

edn. Stockholm.Mabberley D. J. (1997) The Plant-Book, 2nd edn.

Univ. Press, Cambridge.Peter A. (1897) Hydrophyllaceae. In: Engler A.,

Prantl K. (eds.) Die naturlichen Pflanzenfamilien… IV, 3a: 54–71. Engelmann, Leipzig.

Pollard C. J., Amuti K. S. (1981) Fructose oligo-saccharides: possible markers of phylogeneticrelationships among dicotyledonous plant fam-ilies. Biochem. Syst. Evol. 9: 69–78.

Porembski S., Koch M. (1999) Inulin occurrence inSphenoclea (Sphenocleaceae). Plant Biol. 1: 288–289.

Ronse Decraene L. P., Linder H. P., Smets E. F.(2000) The questionable relationship of Montinia(Montiniaceae): evidence from a floral ontoge-netic and anatomical study. Amer. J. Bot. 87:1408–1424.

82 C. Erbar et al.: Floral development in Hydrolea palustris

Savolainen V., Fay M. F., Albach D. C.,Backlund A., van der Bank M., Cameron K.M., Johnson S. A., Lledo M. D., Pintaud J.-C.,Powell M., Sheahan M. C., Soltis D. E.,Soltis P. S., Weston P., Whitten W. M.,Wurdack K. J., Chase M. W. (2000) Phylogenyof the eudicots: a nearly complete familialanalysis based on rbcL gene sequences. KewBull. 55: 257–309

Schulze-Menz G. K. (1964) Rosales. In: MelchiorH. (ed.) A. Engler’s Syllabus der Pflanzenfami-lien II. Gebr. Borntraeger, Berlin-Nikolassee,pp. 193–242.

Soltis D. E., Soltis P. S., Chase M. W., Mort M.E., Albach D. C., Zanis M., Savolainen V.,Hahn W. H., Hoot S. B., Fay M. F., AxtellM., Swensen S. M., Prince L. M., Kress W. J.,Nixon K. C., Farris J. S. (2000) Angiospermphylogeny inferred from 18S rDNA, rbcL, andatpB sequences. Bot. J. Linn. Soc. 133: 381–461.

Stevens P. E. (2003) Angiosperm phylogeny web-site. Version 4, May 2003. ttp://www.mobot.org/MOBOT/research/APweb/.

Takhtajan A. (1997) diversity and classification offlowering plants. Columbia Univ. Press, NewYork.

Thorne R. F. (1983) Proposed new realignments inthe angiosperms. Nord. J. Bot. 3: 85–117.

Thorne R. F. (2001) The classification and geogra-phy of the flowering plants: dicotyledons of theclass Angiospermae. Bot. Rev. 66: 441–647

Willis J. C. (1895) Contributions to the naturalhistory of the flower. Part II. Fertilizationmethods of various flowers (Hydrolea spinosa).Bot. J. Linn. Soc. 30: 293–294.

Addresses of the authors: Prof. Dr. ClaudiaErbar (e-mail: ERLE@uni-heidelberg.de), Prof.Dr. Peter Leins, Universitat Heidelberg, HIP -Biodiversitat und Pflanzensystematik, ImNeuenhei-mer Feld 345, 69120Heidelberg, Germany. Prof.Dr.Stefan Porembski (e-mail: stefan.porembski@uni-rostock.de), Universitat Rostock, Institut fur Bio-diversitatsforschung: Allgemeine und Spezielle Bot-anik, Wismarsche Straße 8, 18051 Rostock,Germany.

C. Erbar et al.: Floral development in Hydrolea palustris 83