pollen tube growth and the pollen-tube pathway of nymphaea odorata

Pollen tube growth and the pollen-tube pathway ofNymphaea odorata (Nymphaeaceae)boj_1039 581..593

JOSEPH H. WILLIAMS*, ROBERT T. MCNEILAGE†, MATTHEW T. LETTRE andMACKENZIE L. TAYLOR

Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN 37996, USA

Received 25 November 2009; accepted for publication 24 February 2010

An important aspect of the evolution of carpel closure, or angiospermy, is the relationship between pollen tubegrowth patterns and internalization of the pollen-tube pathway. True carpel closure, involving postgenital fusionof inner carpel margins, is inferred to have arisen once within the ancient order Nymphaeales, in the commonancestor of Nymphaeaceae. We studied pollen tube development, from pollination to fertilization, in a naturalpopulation of Nymphaea odorata, using hand pollinations and timed flower collections. Pollen germinates instigmatic secretions within 15 min and pollen tubes enter subdermal transmitting tissue within an hour, followingwide intercellular spaces towards the zone of postgenital fusion. At the zone of fusion they turn downwards to growin narrow spaces between interlocked cells and then wander freely to ovules within ovarian secretions. Thepollen-tube pathway is 2–6 mm long and upper ovules are first penetrated 2.5 h after pollination. Pollen tubes growat rates of approximately 1 mm/h whether in stigmatic fluid, transmitting tissues or ovarian secretions. Pollen-tubepathways are structurally diverse across Nymphaeales, yet their pollen tubes have similar morphologies and rapidgrowth rates. This pattern suggests pollen tube growth innovations preceded and were essential for the evolutionof complete carpel closure. © 2010 The Linnean Society of London, Botanical Journal of the Linnean Society, 2010,162, 581–593.

ADDITIONAL KEYWORDS: angiospermy – callose – carpel – character evolution – growth rate – hand-pollination – homoplasy – origin of angiosperms – Nymphaeales.

INTRODUCTION

Among seed plants, the unique life history stagebetween pollination and fertilization, the progamicphase, has been a focal period of reproductive inno-vation. In angiosperms a large number of structuraland developmental modifications to the fertilizationprocess are associated with the origin of an extremelyshort progamic phase relative to the plesiomorphicseed plant pattern (Williams, 2008). These includeboth abbreviation and acceleration of male gameto-phyte ontogeny and the origin and evolution of novelfemale tissues that could facilitate rapid pollen andpollen tube development. Understanding the origin ofthe closed carpel, the feature that gives angiosperms

their name and which gives rise to a novel internalpollen-tube pathway, also requires understanding theevolution of the pollen tube and its growth pattern(Williams, 2009). Fortunately for comparative biology,there is much variation in both pollen and carpeldevelopment among living early-divergent floweringplant lineages. Aquatics are especially interestingbecause they appear to have undergone the mostextreme accelerations of the fertilization process (Wil-liams, 2009).

Nymphaeales (Cabombaceae, Nymphaeaceae,Hydatellaceae; APG III, 2009) is an ancient aquaticlineage that in most recent analyses is sister to allangiosperms except Amborella trichopoda Baill.(Amborellaceae: Jansen et al., 2007; Moore et al., 2007;Saarela et al., 2007; see also Soltis et al., 2009). Studiesof the pollination to fertilization process are restrictedto Nymphaea capensis Thunb. var. zanzibariensis(Casp.) Verdc. (Orban & Bouharmont, 1995) and Bra-senia schreberi J.F.Gmel. and Cabomba caroliniana

*Corresponding author. E-mail: [email protected]†Current address: Department of Biochemistry, Molecular,Cellular and Developmental Biology, University of California,Davis, CA 95616, USA

Botanical Journal of the Linnean Society, 2010, 162, 581–593. With 6 figures

© 2010 The Linnean Society of London, Botanical Journal of the Linnean Society, 2010, 162, 581–593 581

A.Gray (Taylor & Williams, 2009). In these andNuphar polysepala Engelm. (Williams, 2008), the pro-gamic phase is strikingly short, less than 8 h induration. However, there is much variation in pollentube growth patterns and in the type or degree ofcarpel closure in the group (Igersheim & Endress,1998; Endress & Igersheim, 2000; Endress, 2005).

Nymphaeaceae are characterized by having open,ascidate carpels that become partially sealed late indevelopment by interlocking of the inner surfaces of ashort region of the upper carpel (‘partial postgenitalfusion at the periphery’, Igersheim & Endress, 1998;Endress & Igersheim, 2000). Such peripheral fusionrepresents an origin of true angiospermy (completecarpel closure) independent from its origin(s) withinthe major lineages of monocots, eudicots and eumag-noliids (Endress & Doyle, 2009). Most early-divergentangiosperms, including Amborellaceae, Cabombaceaeand Hydatellaceae, have completely open carpels thatare sealed by secretions and, hence, are not expectedto have solid pollen-tube pathways (Endress & Iger-sheim, 2000; Rudall et al., 2007). The consequence ofthe origin of true carpel closure in Nymphaeaceae isthat pollen tubes had to evolve, or to have alreadypossessed, the ability to grow through solid tissue,either in the zone of postgenital fusion or in carpelground tissue, to reach ovules.

There are numerous reports on carpel structurewithin Nymphaeaceae, but only a single study hasshown pollen tube growth (Orban & Bouharmont,1995). Detailed studies of the nature of the pollen-tube pathway, pollen tube growth patterns and espe-cially pollen tube growth rates for a number ofspecies are needed in order to answer the specificquestion of what role pollen tube growth patternsplayed in the origin of carpel closure in this ancientangiosperm lineage.

Nymphaea L. is the largest (45–50 spp. in fivesubgenera) and most widespread genus in Nymphae-aceae (Borsch et al., 2007). Recent studies show it tobe paraphyletic, with subgenus Nymphaea sister to aclade comprising the other four subgenera of Nym-phaea plus Ondinea Hartog. and the clade EuryaleSalisb./Victoria Lindl. (Löhne, Borsch & Wiersema,2007, 2009; Borsch, Löhne & Wiersema, 2008). Nym-phaea plus these three genera have been referred toas core Nymphaeaceae (Borsch et al., 2008). Morpho-logical analyses support the sister relationshipbetween subgenus Nymphaea and the rest of coreNymphaeaceae (Borsch et al., 2008; Taylor, 2008) andthere are differences in floral biology and probablyalso of breeding system (Wiersema, 1988). Little isknown of progamic phase biology in any of thesegroups apart from the greenhouse study of N. capen-sis within the pantropical subgenus Brachyceras(Orban & Bouharmont, 1995).

Nymphaea odorata Aiton is a perennial aquaticthat is widely distributed across North America(Wiersema & Hellquist, 1997) and placed within thenorthern temperate subgenus Nymphaea (Borschet al., 2007). The goal of this study is to characterizefunctional aspects of the carpel related to the pollen-tube pathway, pollen germination, pollen tube growthrates and fertilization timing in N. odorata in itsnative environment. We discuss such details in lightof recent studies of ancient flowering plant reproduc-tive biology and the origin of angiospermy.

MATERIAL AND METHODS

Experimental pollinations and collections of Nym-phaea were conducted in June 2008 at MontereyLake, Putman County, TN, USA (36°06′N 85°14′E;elevation 566 m). Female-phase flowers were coveredwith bridal veil fabric (gap diameter � 200 mm) inthe early morning prior to flower opening in order toprevent natural pollination. Stigmas of open female-phase flowers were artificially pollinated by brushinga dehiscent anther (collected from a distant male-phase flower) across its surface. Pollinated flowerswere tagged and re-bagged, then collected and fixedat 5, 10, 15, 30 or 60 min after pollination (map) orat 30-min intervals thereafter up to 6 h after polli-nation (hap). The gynoecium was removed and fixedfor 24 h in either 3:1 (95% ethanol:glacial aceticacid); FAA (40% formaldehyde, glacial acetic acid,95% ethanol) or Karnovsky’s fixative [50% gluteral-dehyde and 16% paraformaldehyde in 0.2 M phos-phate buffer (pH 7.4)]. Specimens fixed in 3:1 andFAA were rinsed and stored in 70% ethanol, whereasthose fixed with Karnovsky’s fixative were washedfour times in phosphate buffer, dehydrated in agraded ethanol series and then stored in 70%ethanol.

For analysis of pollen tube growth, carpels fixed in3:1 were hand sectioned and stained overnight withaniline blue (AB) (methods in Williams, 2009). Pollentube length was calculated as the distance from thepollen grain (or in some cases the proximal zone of thestigma) to the tip of the longest pollen tube (orleading clump of pollen tubes) within a carpel. Pollentube growth rates were only measured on leadingpollen tubes that could be followed for their entirelength. Growth rate was measured as length dividedby the time since pollination minus 15 min (theempirically determined time needed for sufficientpollen germination; see Results). Pollen tube nucleiwere visualized with 4′-6-diamidino-2-phenylindole(DAPI).

For histological analysis, carpels were dehydratedto 95% ethanol and then infiltrated and embeddedin JB-4 polymer (Polysciences Inc., Warrington, PA,

582 J. H. WILLIAMS ET AL.

© 2010 The Linnean Society of London, Botanical Journal of the Linnean Society, 2010, 162, 581–593

USA). Blocks were serial-sectioned with glass kniveson a Sorvall Dupont JB-4 microtome (Newtown, CT,USA). Serial sections (5 mm) were stained with tolui-dine blue (TBO) and viewed with a Zeiss Axioplan 2compound microscope with Axiocam digital camera ora Leitz Laborlux 11 compound microscope with aSony HDV 1080/mini DV digital camera.

RESULTSFLORAL BIOLOGY

Flowers of Nymphaea odorata are protogynous. Onthe first day of anthesis, flowers are partially openwith abundant stigmatic fluid and immature anthers(Fig. 1A). In second-day flowers, stigmatic fluid is

Figure 1. Floral biology of Nymphaea odorata. A, first-day of flower opening (female phase). Sepals (se) are reflexed,whereas inner petals (pe) and stamens form a protected chamber. Anthers (a) are non-dehiscent and the stigmatic cupis filled with stigmatic secretions. Bar, 1 cm. B, second-day flower (male phase) showing fully reflexed petals anddehiscence of inner, but not outer, anthers (a) (note foraging bee, Apis mellifera). Bar, 1 cm. C, hand section throughfemale-phase flower showing stamens (st) surrounding the stigmatic cup formed by the upper (stigmatic) surface of thesyncarpous gynoecium. The receptive surface of the stigma (between arrowheads) extends from the base of the carpellarappendages (‘stylar processes;’ sp), to the floral apex (asterisk) and receptacle (r). Each of the many separate ovariancavities (oc) contains many ovules. Bar, 1 cm. D, close-up of stigmatic cup of female-phase flower filled with stigmatic fluidup to bases of stylar processes. Radial slits (rs) mark lines of postgenital fusion of inner margins of each carpel, whereasthe base of each stylar process delineates a single carpel (between two asterisks); fa, floral apex. Bar, 1 mm.

POLLEN TUBE GROWTH IN NYMPHAEA (NYMPHAEACEAE) 583


absent (Schneider & Chaney, 1981; Wiersema, 1988)and the innermost anthers are dehiscent shortly afterflower opening, whereas outermost anthers do notopen until the third day (Fig. 1B). The upper, outermargin of each carpel extends to form a ‘stylarprocess,’ which is erect during the female phase(Fig. 1A, C) and then arches over the stigmatic cupduring the male phase (Fig. 1B). On each day ofanthesis, flowers open at approximately 09:00 h, thenclose and submerge between noon and 14:30 h. Male-phase flowers open slightly earlier than female-phaseflowers each day. Honey bees (Apis mellifera L.) werethe predominant pollinators present but were mostlyseen collecting pollen from second- and third-dayflowers (Fig. 1B). Flowers are often visited, sometimesby many bees at a time, and visitation is heaviestmid-morning.

The gynoecium is eusyncarpous; carpels of N.odorata are congenitally fused early in developmentand at maturity only a small intercarpellary gapdelineates adjacent carpels (Moseley, 1961). The con-joined carpels radiate from a central receptacle(Fig. 1C, D) and their upper walls form a cup-shapedplatform that on first-day flowers is filled with stig-matic fluid up to the bases of stigmatic processes(Fig. 1D). The entire surface of the stigmatic cupbeneath the fluid, except for the floral apex, is coveredby uniseriate, multicellular papillae (see Capperino &Schneider, 1985). The surface is marked by a series ofradial depressions and at the bottom of each is aradial slit (Fig. 1D) that marks the site of postgenitalfusion of upper, inner carpel margins (Moseley, 1961;Igersheim & Endress, 1998; Schneider, Tucker & Wil-liamson, 2003; Hu et al., 2009).

POLLEN TUBE GROWTH AND POLLEN-TUBE PATHWAY

Pollen grains are bicellular in dehiscent anthers andupon arriving on the stigma (Fig. 2A). Pollen of Nym-phaea is zonosulcate and has a large aperture area.At germination, the inner wall of the pollen grain isstrongly thickened by callose deposits which extendout from the aperture to the callosic inner wall of thepollen tube (Fig. 2A, B). Pollen germinates both onand beneath the surface of the stigmatic fluid. Lowlevels of pollen germination were observed at 5 and10 min after pollination, 50% of maximum germina-tion was reached 20–30 map and maximum germina-tion (c. 90%) first occurred between 45 and 60 map(Fig. 3).

Pollen tubes grow freely in stigmatic fluid, notalong papillar or epidermal cell surfaces. Multicellu-lar papillae emerge from an epidermal layer that isunderlain by a few distinct layers of periclinally ori-ented cells with wide intercellular spaces (Fig. 2C).Pollen tubes grow downward between papillae to

enter the subdermal region (Fig. 2D). The subdermalregion is a substigmatic transmitting tissue. It ismuch thinner in distal regions of the stigma (nearadjacent carpel; Fig. 2C) than in proximal regions(near the radial slit; Fig. 2E). Pollen tubes enteringdistal regions sometimes fail to continue develop-ment, but generally pollen tubes entering the substig-matic transmitting tissue turn sharply towards thezone of postgenital fusion, growing in wide intercel-lular spaces (Fig. 2E). This substigmatic portion ofthe transmitting tract (Orban & Bouharmont, 1995;Igersheim & Endress, 1998) is composed of small andlong, densely cytoplasmic cells that seem to mechani-cally direct pollen tubes toward the zone of postgeni-tal fusion.

At the bottom of each radial depression the papil-late carpel margins meet to form the radial slit andhave become developmentally interlocked such thatthe ovarian cavity is sealed off by a solid carpel wall(Fig. 4A, B). However, within the zone of postgenitalfusion cells are small and densely cytoplasmic andthere is evidence of small intercellular spaces(Fig. 4B, C). While doing hand sections, carpels couldbe easily pulled open along this plane of fusion, indi-cating that cells are only weakly interlocked. Pollentubes turn abruptly downward when they reach thezone of postgenital fusion, following the contours ofindividual cells within narrow intercellular spaces(Fig. 4B, C).

Near the bottom of the zone of postgenital fusion,the inner margins of the carpel gradually becomeapparent as two appressed, single-layered epidermalsurfaces that separate to form a secretion-filled ‘stylarcanal’ (Fig. 4D). This thin upper extension of theovarian cavity is evident from near the receptacle tothe outer carpel margin (Fig. 5). In most carpels,pollen tubes wander freely within secretions of thestylar canal and ovarian cavity until nearing a micro-pyle of an ovule, then turn sharply to enter (Fig. 4A).In a few carpels, pollen tubes in the stylar canalfollowed the epidermal surface of carpel wall closelybefore turning to grow freely through secretions inthe upper ovarian cavity.

The pollen-tube pathway ranges in length from 2.04to 6.30 mm, depending on the width of the stigmaticsurface and the depth of the ovary (Fig. 5). Pollen tubesare first seen entering substigmatic transmittingtissue at 1 hap (Fig. 6). After passing through thestylar canal, pollen tubes are first observed in theovarian cavity at 2 hap, entering upper ovules at2.5 hap and reaching ovules in the bottom of the ovaryat 3 hap (Figs 5, 6). In some carpels in which an upperovule had been penetrated, other pollen tubes could beseen at the bottom of the ovarian cavity without havingpenetrated ovules. By 5 hap, the majority of ovulesthroughout the ovary are penetrated.



Fig

ure

2.E

arly

poll

entu

begr

owth

ofN

ymph

aea

odor

ata.

A,

bice

llu

lar

poll

engr

ain

onst

igm

atic

surf

ace

show

ing

tube

nu

cleu

s(t

n)

and

gen

erat

ive

nu

cleu

s(g

n).

Upp

erh

alf

ofzo

nos

ulc

ate

poll

engr

ain

wit

hec

hin

ate

exin

e(e

)ex

posi

ng

the

call

osic

inn

erw

all

(bri

ght

yell

ow)

ofin

ner

mar

gin

ofap

ertu

re(b

otto

mh

alf

ofpo

llen

grai

nou

tof

view

).D

oubl

est

ain

edw

ith

AB

and

DA

PI.

Bar

,10

mm.B

,pol

len

tube

emis

sion

,30

min

afte

rpo

llin

atio

n(d

iffe

ren

tial

inte

rfer

ence

con

tras

t).C

lear

area

exte

ndi

ng

out

ofap

ertu

rein

topo

llen

tube

isa

call

ose

thic

ken

ing

(c),

asve

rifi

edby

AB

stai

n(n

otsh

own

).B

ar,1

0mm

.C,t

ange

nti

alse

ctio

nth

rou

gha

dist

alpo

rtio

nof

stig

mat

icsu

rfac

e(f

arfr

omra

dial

slit

)sh

owin

gba

ses

ofu

nis

eria

te,

mu

ltic

ellu

lar

papi

llae

(p)

mee

tin

gth

inla

yer

oflo

osel

ypa

cked

subs

tigm

atic

cell

sw

hic

hov

erli

eca

rpel

wal

lgr

oun

dti

ssu

e(g

)(4

.5h

ap;T

BO

stai

n).

Bar

,25

mm.D

,tan

gen

tial

sect

ion

thro

ugh

apr

oxim

alpo

rtio

nof

stig

ma

(adj

acen

tto

radi

alsl

it)

show

ing

poll

entu

be(a

rrow

)en

teri

ng

subs

tigm

atic

tiss

ue

bygr

owin

gw

ith

inse

cret

ion

sbe

twee

npa

pill

aean

dth

enbe

twee

nlo

osel

ypa

cked

subs

tigm

atic

cell

s(1

2h

ap;T

BO

stai

n).

Bar

,25

mm.

E,

poll

entu

bes

(arr

ows)

insu

bsti

gmat

ictr

ansm

itti

ng

tiss

ue

grow

ing

para

llel

tost

igm

atic

surf

ace

tow

ard

the

zon

eof

post

gen

ital

fusi

on(t

ange

nti

alse

ctio

n,

prox

imal

port

ion

ofst

igm

a).

Pol

len

tube

sgr

owon

lyw

ith

inse

cret

ion

sbe

twee

nde

nse

lycy

topl

asm

icce

lls

ofth

etr

ansm

itti

ng

trac

t(t

t)an

dn

otbe

nea

thth

isla

yer

wit

hin

the

carp

elw

all

grou

nd

tiss

ue

(12

hap

;T

BO

stai

n).

Bar

,50

mm.



Mitosis of the generative nucleus to produce twosperm nuclei occurs while pollen tubes grow withinthe substigmatic or postgenitally fused areas, from 2to 2.5 hap (Fig. 4E). Callose plugs are first observed1.5–2.5 hap and are clearly evident by 3.5 hap(Fig. 4F). Although pollen tubes often reach the upperovules before callose plugs first form, a plug is almostalways present at the sharp bend that most tubesmake as they enter the micropyle of an ovule(Fig. 4A). Plugs are much more common in tubesgrowing in secretions of the ovarian cavity than intubes growing within transmitting tissues or alongthe inner surface of the carpel wall.

Mean (± SD) growth rate of the leading pollen tube(where entire pollen tube was observed) was1066 ± 548 mm/h (range: 332–3007 mm/h; N = 78).Mean (± SD) growth rates of leading pollen tubeswithin different portions of the pathway (see Fig. 6)were 951 ± 583 mm/h in the stigmatic region (0.4–1 hap; N = 35), 972 ± 470 mm/h in the transmitting

tissue/zone of postgenital fusion (1.5–2.5 hap; N = 20)and 1323 ± 486 mm/h in the ovarian cavity (3–6 hap;N = 23).

DISCUSSIONPOLLINATION AND FLORAL BIOLOGY

The sequence and timing of female and male flower-ing phases reported here are consistent with thestudy of N. odorata in Texas by Schneider & Chaney(1981). There is no overlap of male and female phasesin N. odorata and this also seems to be true for othertemperate zone Nymphaea spp. (Capperino &Schneider, 1985; Wiersema, 1988). Male phase beginswith dehiscence of inner anthers on second-day N.odorata flowers. In contrast, in at least two tropicalspecies of Nymphaea the outer anthers open first andthey open on first-day flowers (Prance & Anderson,1976; Orban & Bouharmont, 1995; Endress, 2001).

The honeybee, Apis mellifera, was by far the mostcommon insect visitor to flowers, but Schneider &Chaney (1981) viewed this non-native bee as a rela-tively ineffective pollinator of N. odorata as it prima-rily visited male-phase flowers to collect pollen.However, transfers from male- to female-phaseflowers by the honeybee were frequent in the relatedN. mexicana Zucc. (Capperino & Schneider, 1985).

Pollen is bicellular in N. odorata, as also reportedfor N. alba L., N. colorata Peter, N. heudelotiiPlanch., N. rosea Sweet and N. tuberosa Paine(Schnarf, 1929; Corriveau & Coleman, 1988; Van Mie-groet & Dujardin, 1992; Zhang, Liu & Sodmergen,2003). Both bicellular and tricellular pollen have beenreported in N. stellata Willd. (Brewbaker, 1967),whereas tricellular pollen has been reported in anumber of core Nymphaeaceae (see Endress & Doyle,2009: 59). It is not clear whether such reports repre-sent independent evolutionary origins of tricellularpollen or are attributable to developmental variation(Lora, Herrero & Hormaza, 2009) or to errors ofinterpretation. Brewbaker (1967) suggested that tri-cellular pollen was advantageous in aquatic plants

Figure 3. Pollen germination over time on first-day Nym-phaea odorata stigmas. The percentage of pollen grainsgerminated on stigmas (mean N = 5.4 independent handpollinations per time point).

�Figure 4. Late pollen tube growth of Nymphaea odorata. A, tangential hand section through the gynoecium, aboutmidway between floral apex and carpel margin (5.5 hap; AB stain). There are no lines of postgenital fusion evidentbetween carpels, but within each carpel there is a zone of postgenital fusion (zf) where the inner carpel walls (e.g. c1) meetand become developmentally interlocked. To reach the ovarian cavity (oc), pollen tubes grow through the zone of fusion– they do not grow within ground tissue of carpel walls. Bar, 300 mm. B, tangential section of pollen-tube pathway(4.5 hap; TBO stain). Pollen tubes (arrows) can be seen among stigmatic papillae (p), between loosely packed cellsunderlying the stigmatic surface (tt) and within zone of postgenital fusion; g, ground tissue. Bar, 100 mm. C, close-up ofzone of postgenital fusion (from fig. 4B) showing pollen tube growing in spaces between cells. Bar, 20 mm. D, close-up ofopen stylar canal showing pollen tubes growing entirely within secretions (4.5 hap). Note single layer of epidermal cells(ep) of inner margin of carpel wall (TBO stain). Bar, 100 mm. E, pollen tube with two sperm nuclei (sn) (2 hap; DAPI stain).Bar, 10 mm. F, callose plug (2.5 hap; AB stain). Bar, 10 mm.



Figure 5. Pollen-tube pathway of Nymphaea odorata. View of tangential hand section facing away from floral apex, aboutmidway to the base of the stylar process (sp) of the middle carpel in the figure (AB stain). Limits of pollen-tube pathwaylength were measured from proximal (p) and distal (d) regions of the stigma to first (apical) and last (basal) ovules,indicated by asterisks. Arrowheads indicate intercarpellary gaps; st, base of stamen. Bar, 1 mm.



because of its rapid germination. However, rapid ger-mination also depends strongly on pollen hydrationstatus. For example, the highly hydrated pollen ofaquatic Cabombaceae is bicellular and germinateswith the same rapidity as that of N. odorata (Taylor& Williams, 2009). Although we did not measurehydration status directly, callose inner walls of pollengrains at dispersal (Fig. 2A, B) are one indicator ofrapidly germinating, well-hydrated pollen (Franchiet al., 2002).

THE CLOSED CARPEL AND POLLEN-TUBE PATHWAY

The internal pollen tube transmitting tract, fromstigma to ovule, is a functional innovation generatedby angiospermy, or closure of the carpel. Among early-divergent angiosperms, true closure of the carpeloccurs by postgenital fusion of inner epidermal sur-faces; it originated once in the common ancestor ofNymphaeaceae and one or more times within ancientmonocots, eudicots and eumagnoliids (Endress &Doyle, 2009). Syncarpy, the joining of adjacent carpelsto each other, also originated in the common ancestorof Nymphaeaceae (Endress & Doyle, 2009). Syncarpyin N. odorata is by lateral fusion of adjacent carpelprimordia early in development (i.e. congenital, notpostgenital, fusion; Moseley, 1961; Hu et al., 2009).

Angiospermy and syncarpy are two homoplasioustraits that have been described as major innovationswithin the main lineages of angiosperms (Endress,1982, 2001; Armbruster, 2002) and our study points tosome of the consequences of early experiments incarpel closure and joining for pollen tube growth.

It has been estimated that 93% of phylogeneticallyderived angiosperm species with multicarpellategynoecia are syncarpous (Endress, 1982) and mostforms of syncarpy cause pollen tubes from separatestigmas or stigmatic lobes to meet and then competewithin a common internal transmitting tract(Endress, 1982; Armbruster, 2002), an internal com-pitum (Carr & Carr, 1961). This is not the case inNymphaea and other core Nymphaeaceae where syn-carpy instead produces an extragynoecial compitum,the fluid-filled stigmatic cup. The extragynoecial com-pitum of core Nymphaeales, and probably also of thesister group Barclaya Wall. (Williamson & Schneider,1994), allows pollen tubes to access any one carpelfrom the stigma but does not produce a shared inter-nal pollen-tube pathway. This is an advantage oversyncarpy in the outgroup Nuphar (sister to the rest ofNymphaeaceae), in which carpel joining also producesa stigmatic platform but not a pool of stigmatic fluid;narrow stigmatic crests on each carpel are completelyseparate from each other. The rest of Nymphaealesare apocarpous, such that spaces between separatecarpels prevent crossing over of pollen tubes betweenstigmas.

The structure of the internal pollen-tube pathwayof N. odorata is heterogeneous. First, pollen germi-nates and pollen tubes grow freely within stigmaticsecretions. The whole upper surface of each carpel iscovered with a dense field of uniseriate, multicellularpapillae and functions as a secretory stigma (Iger-sheim & Endress, 1998). A distinct secretory subder-mal tissue, consisting of several layers of looselypacked cells running parallel to the stigmatic surface,underlies the papillate stigma. Pollen tubes enter thesubstigmatic transmitting tissue growing withinintercellular secretions, turn sharply and then aremechanically directed towards the postgenitally fusedradial slit. Pollen tubes turn abruptly downwards inthis distinctly different portion of the transmittingtract, which consists of interlocked cells with muchnarrower intercellular spaces than seen in the sub-stigmatic transmitting tissue. In both the substig-matic and postgenitally fused portions of thetransmitting tract, pollen tubes never enter adjacentground tissues. As they exit the zone of postgenitalfusion, pollen tubes do not follow inner epidermalsurfaces of the ovary to the funiculus and outerintegument of the ovule, but instead wander freelywithin secretions until they near an ovule, then theyturn sharply and grow into the micropyle. Moseley

Figure 6. Timing of pollen tube growth of Nymphaeaodorata. Data represent the mean (SD) lengths of leadingpollen tubes per carpel (mean N = 10.5 independent handpollinations per time point). Two lower dashed linesrepresent pollen-tube pathway length from proximaland distal locations on the stigma to the first ovule(mean ± 95% confidence level = 2.04 ± 0.15 mm and3.10 ± 0.29 mm, respectively) (N = 47). Upper dashed linesrepresent the corresponding distances to the deepestovule: 5.30 ± 0.34 mm and 6.30 ± 0.46 mm (N = 47). Somepollen tubes are longer than the pollen-tube pathway atlater stages because of wandering in the ovary.



(1961) described pollen tubes of N. odorata and N.tetragona Georgi following epidermal surfaces of thestylar canal into the ovary, but in our material it wasfar more common to see pollen tubes growing entirelyfree from morphological surfaces as soon as theyexited the postgenitally fused tissue (see also Orban& Bouharmont, 1995). Growth along inner (adaxial)carpel epidermal surfaces from stigma to ovule, asoccurs in Winteraceae, has been thought to reflect anancestral angiosperm pattern of pollen tube transmis-sion (references in Frame, 2003). However, otherNymphaeales, and also Amborella Baill. and Aus-trobaileya C.T.White (Austrobaileyaceae), show apattern much more like that of Nymphaea (Williams,2008, 2009; Taylor & Williams, 2009).

Classifications of the angiosperm transmitting tracthave been reviewed recently (Endress, 1994; Erbar,2003), although these apply only as analogies,because of the independent origin of internal trans-mitting tissues in Nymphaeaceae. The simplest clas-sification results from the observation that mosttransmitting tracts support pollen tube growth either(1) along a secretory epidermal surface (stigma, openstylar canal, ovary) or (2) within a subdermal layer(s)(see Erbar, 2003). Nymphaea clearly falls in the lattercategory as pollen tubes grow within secretions in amultilayered subdermal tissue and then a complexzone of postgenital fusion, in which ontogeny subse-quent to interlocking of carpel margins has obscureddermal surfaces. Pollen tubes did not grow alongpapillar cell surfaces, nor did they generally followepidermal surfaces of the ovarian cavity, placenta orovule. Although there are structural differencesbetween the substigmatic and postgenitally fused por-tions of the transmitting tract, both are several celllayers thick. The functional result is that the growthof many pollen tubes can be supported, as would benecessary in Nymphaea with its many ovules. Multi-layered internal transmitting tracts are correlatedwith multi-ovulate ovaries in phylogenetically derivedangiosperms (Endress, 1994).

The carpels of the nearest outgroups to Nym-phaeaceae (Cabombaceae and Hydatellaceae as wellas Amborellaceae, which sometimes appears as sisterto Nymphaeales) differ from those of Nymphaea intwo respects: all have open, secretion-filled carpelsand few (1–3) ovules. In uniovulate Amborella andTrithuria Hook.f. (Hydatellaceae), few pollen tubesenter the open mouth of the stylar canal (Williams,2009; M. L. Taylor & J. H. Williams, unpubl. data).More pollen tubes enter the stylar canals of Braseniaschreberi and Cabomba caroliniana (Cabombaceae),but they typically circumvent the open mouth;instead they grow directly through the ground tissueof the carpel wall to reach the canal (Taylor & Will-iams, 2009). The solid portion of the transmitting

tissue in Cabombaceae is not a zone of fusion, andthere are no large intercellular spaces between cells(Endress, 2005), a quite different tissue structurefrom the internal transmitting tissues of Nymphaea.

The postgenitally fused zone of the Nymphaeacarpel comprises only a small portion of the pollen-tube pathway and is formed by interlocking of uppercarpel margins late in carpel development (for carpelontogeny see Troll, 1933; Endress, 2001; Schneideret al., 2003; Hu et al., 2009). Fused tissues generallyhave unique properties relative to solid groundtissues (Lolle & Pruitt, 1999) and, in the case ofNymphaea, the interlocked cells within the postgeni-tally fused area apparently undergo divisions duringthe fusion process (Moseley, 1961). Thus, pollen tubesgrowing in this area are not necessarily followingsubsumed epidermal surfaces, but grow within aweakly differentiated tissue. It is not clear whetherthere are differences in the composition of intercellu-lar secretions between the substigmatic and postgeni-tally fused portions of the transmitting tissues ofNymphaea, but they represent quite different forms ofsubdermal tissue formation. The transmitting tissueof Cabombaceae, at least in terms of its developmen-tal origin and structure, represents yet a third type ofsolid transmitting tissue in which pollen tubes grow.

It is tempting to conclude that pollen tube growthwithin a solid transmitting tissue, such as throughthe wall of the open carpel in Cabombaceae, origi-nated before growth within the fused tissues of a trueclosed carpel, as inferred for the common ancestor ofNymphaeaceae. For the reverse to be true, one wouldhave to infer that the open carpel of Cabombaceaerepresents a reversal from an ancestrally closed con-dition, a less likely scenario given the distribution ofopen carpels among early-divergent angiosperms(Endress & Doyle, 2009). It is notable that, in Ambo-rella, Hydatellaceae, Nymphaeaceae and Cabom-baceae, pollen tubes in the final portion of the pollen-tube pathway have been shown to grow between cellsof a ‘solid’ but exceptionally thin nucellar tissue(Orban & Bouharmont, 1995; Friedman, 2008; Will-iams, 2008, 2009; Rudall et al., 2009; Taylor & Will-iams, 2009). More detailed studies of both pollen tubegrowth and carpel anatomy in these groups areneeded to reconstruct the evolutionary steps leadingto early forms of carpel closure and the pollen tubetransmitting tract within the water lilies – a lineagethat is as old as that leading to all other extantangiosperms, other than Amborella.

THE PROGAMIC PHASE AND POLLEN TUBE

GROWTH RATE

The progamic phase of N. odorata lasts approximately2.5–3.5 h (at which time pollen tubes first enter upper



and lower ovules, respectively). In N. capensis pollentubes are first seen in the ovary within 1 h and inovules < 6 hap (Orban & Bouharmont, 1995), whereasin other Nymphaeales pollen tubes first reach ovulesfrom < 1 h to 6–8 h (Taylor & Williams, 2009; Will-iams, 2009; M. L. Taylor & J. H. Williams, unpubl.data). These are among the shortest periods of pollentube growth known in angiosperms and are shorterthan those in all other early-divergent woody peren-nial angiosperms (Williams, 2009). Among a broaderset of angiosperms, exceedingly short progamicphases seem to be restricted to aquatic lineages, inAlismatales and Nelumbo nucifera Gaertn. (Ohga,1937; Ackerman, 1993; Wang, Tao & Lu, 2002).

The short progamic phases of water lilies are madepossible by the acceleration of their male gameto-phyte development relative to that of other early-divergent angiosperms. In N. odorata substantialpollen germination occurs within 15 min after handpollination. Similarly short times to germination areknown for other water lilies, but longer times aretypical of most early-divergent woody perennial lin-eages (Williams, 2009). Leading pollen tubes of N.odorata grew at average rates of approximately1 mm/h in a diversity of media: stigmatic fluid, trans-mitting tissues and ovarian secretions. Pollen tubegrowth rates have not been reported for other speciesof Nymphaea; however, for N. capensis a similar rate(1.02 mm/h) can be calculated for the leading pollentube in Orban & Bouharmont (1995: fig. 10; assuming15 min for pollen germination). Average growth ratesof leading pollen tubes in other Nymphaeales rangefrom approximately 0.6–1.0 mm/h (Williams, 2008;Taylor & Williams, 2009). These rates are comparablewith those reported for many monocots and eudicots(e.g. Marshall & Diggle, 2001; Higashiyama & Inat-sugi, 2006; Lee et al., 2008), including aquatics, butare much faster than those of any early-divergentwoody perennial angiosperm (Williams, 2009).

The evolution of rapid pollen germination andpollen tube growth must involve either a shorteningof the time between pollination and fertilization or alengthening of the pollen-tube pathway without thetrade-off of having to delay fertilization, or somecombination of both. The pollen-tube pathway of N.capensis is c. 2–6 mm long and individual pollentubes grow up to c. 10 mm long as a result of wan-dering. These distances are similar to those of otherNymphaeales, except for the minute Hydatellaceae,but are relatively short when considering pollen-tubepathways of the many phylogenetically derivedangiosperms that have evolved such fast pollen tubegrowth rates (Williams, 2008). Short pollen-tubepathway length is a plesiomorphic feature of flower-ing plants that is ubiquitous among early-divergentlineages. In fact, the repeated association of excep-

tionally short progamic phases and rapid pollen tubegrowth rates in three independent aquatic lineages(Nymphaeales, Nelumbonaceae and Alismatales) thateach retained ancestrally short pollen-tube pathwaysis striking. It suggests that the necessity for rapidreproduction during colonization of, and/or long-termpersistence in, an aquatic environment was thedriving force for acceleration of male gametophytedevelopment (Williams, 2008). Slight lengthening ofpollen-tube pathways in these lineages evolved as aconsequence of the origin of fast pollen tube growthrates, rather than the other way around. In support ofthis conclusion, evolutionary extension of the pollen-tube pathway occurred in Brasenia Schreb. withoutcausing acceleration of its pollen tube growth rates(Taylor & Williams, 2009).

ACKNOWLEDGEMENTS

We are grateful to T. Arias and N. Buckley for labo-ratory assistance. Funding to J.H.W. was provided bythe Department of Ecology and Evolutionary Biology(L. R. Hesler fund) and the National Science Foun-dation (DEB 0640792).

REFERENCES

Ackerman JD. 1993. Pollen germination and pollen tubegrowth in the marine angiosperm, Zostera marina L.Aquatic Botany 46: 189–202.

APG III. 2009. An update of the Angiosperm PhylogenyGroup classification for the orders and families of floweringplants: APG III. Botanical Journal of the Linnean Society161: 105–121.

Armbruster WE. 2002. Evolution of syncarpy inangiosperms: theoretical and phylogenetic analyses of theeffects of carpel fusion on offspring quantity and quality.Journal of Evolutionary Biology 15: 657–672.

Borsch T, Hilu KW, Wiersema JH, Löhne C, Barthlott W,Wilde V. 2007. Phylogeny of Nymphaea (Nymphaeaceae):evidence from substitutions and microstructural changes inthe chloroplast trnT-trnF region. International Journal ofPlant Sciences 168: 639–671.

Borsch T, Löhne C, Wiersema J. 2008. Phylogeny andevolutionary patterns in Nymphaeales: integrating genes,genomes and morphology. Taxon 57: 1052–1081.

Brewbaker JL. 1967. Distribution and phylogenetic signifi-cance of binucleate and trinucleate pollen grains inangiosperms. American Journal of Botany 54: 1069–1083.

Capperino ME, Schneider EL. 1985. Floral biology of Nym-phaea mexicana Zucc. (Nymphaeaceae). Aquatic Botany 23:83–93.

Carr SGM, Carr DJ. 1961. The functional significance ofsyncarpy. Phytomorphology 11: 249–256.

Corriveau JL, Coleman AW. 1988. Rapid screening methodto detect potential biparental inheritance of plastid DNA



and results for over 200 angiosperm species. AmericanJournal of Botany 75: 1443–1458.

Endress PK. 1982. Syncarpy and alternative modes of escap-ing disadvantages of apocarpy in primitive angiosperms.Taxon 31: 48–52.

Endress PK. 1994. Diversity and Evolutionary Biology ofTropical Flowers. Cambridge: Cambridge University Press.

Endress PK. 2001. The flowers in extant basal angiospermsand inferences on ancestral flowers. International Journalof Plant Sciences 162: 1111–1140.

Endress PK. 2005. Carpels in Brasenia (Cabombaceae) arecompletely ascidiate despite a long stigmatic crest. Annalsof Botany 96: 209–215.

Endress PK, Doyle JA. 2009. Reconstructing the ancestralangiosperm flower and its initial specializations. AmericanJournal of Botany 96: 22–66.

Endress PK, Igersheim A. 2000. Gynoecium structure andevolution in basal angiosperms. International Journal ofPlant Sciences 161S: S211–S223.

Erbar C. 2003. Pollen tube transmitting tissue: place ofcompetition of male gametophytes. International Journal ofPlant Sciences 164: S265–S277.

Frame D. 2003. The pollen tube pathway in Tasmanniainsipida (Winteraceae): homology of the male gametophyteconduction tissue in angiosperms. Plant Biology 5: 290–296.

Franchi GG, Nepi M, Dafni A, Pacini E. 2002. Partiallyhydrated pollen: taxonomic distribution, ecological and evo-lutionary significance. Plant Systematics and Evolution 234:211–227.

Friedman WE. 2008. Hydatellaceae are water lilies withgymnospermous tendencies. Nature 453: 94–97.

Higashiyama T, Inatsugi R. 2006. Comparative analysis ofbiological models used in pollen tube growth. Plant CellMonographs 3: 265–286.

Hu GW, Lei LG, Liu KM, Long CL. 2009. Floral develop-ment in Nymphaea tetragona (Nymphaeaceae). BotanicalJournal of the Linnean Society 159: 211–221.

Igersheim A, Endress PK. 1998. Gynoecium diversity andsystematics of the paleoherbs. Botanical Journal of theLinnean Society 127: 289–370.

Jansen RK, Cai Z, Raubeson LA, Daniell H, dePamphilisCW, Leebens-Mack J, Muller KF, Guisinger-Bellian M,Haberle RC, Hansen AK, Chumley TW, Lee S-B, PeeryR, McNeal JR, Kuehl JV, Boore JL. 2007. Analysis of 81genes from 64 plastid genomes resolves relationships inangiosperms and identifies genome-scale evolutionary pat-terns. Proceedings of the National Academy of Sciences ofthe United States of America 104: 19369–19374.

Lee CB, Page LE, McClure BA, Holtsford TP. 2008. Post-pollination hybridization barriers in Nicotiana sectionAlatae. Sexual Plant Reproduction 21: 183–195.

Löhne C, Borsch T, Wiersema JH. 2007. Phylogeneticanalysis of Nymphaeales using fast-evolving and non-codingchloroplast markers. Botanical Journal of the LinneanSociety 154: 141–163.

Löhne C, Borsch T, Wiersema JH. 2009. The unusualOndinea, actually just another Australian water-lily of

Nymphaea subg. Anecphya (Nymphaeaceae). Willdenowia39: 55–58.

Lolle SJ, Pruitt RE. 1999. Epidermal cell interactions: acase for local talk. Trends in Plant Science 4: 14–20.

Lora J, Herrero M, Hormaza JI. 2009. The coexistence ofbicellular and tricellular pollen in Annona cherimola(Annonaceae): implications for pollen evolution. AmericanJournal of Botany 96: 802–808.

Marshall DL, Diggle PK. 2001. Mechanisms of differentialpollen donor performance in wild radish, Raphanussativus (Brassicaceae). American Journal of Botany 88:242–257.

Moore MJ, Bell CD, Soltis PS, Soltis DE. 2007. Usingplastid genome-scale data to resolve enigmatic relationshipsamong basal angiosperms. Proceedings of the NationalAcademy of Sciences 104: 19363–19368.

Moseley MF. 1961. Morphological studies of Nymphaeaceae.II. The flower of Nymphaea. Botanical Gazette 122: 233–259.

Ohga I. 1937. On the fertilization of Nelumbo nucifera. Cyto-logia Fujii Jubilee Volume: 1033–1035.

Orban I, Bouharmont J. 1995. Reproductive biology ofNymphaea capensis Thunb. var. zanzibariensis (Casp.)Verdc. (Nymphaeaceae). Botanical Journal of the LinneanSociety 119: 35–43.

Prance GT, Anderson AB. 1976. Studies of the floral biologyof tropical Nymphaeaceae. Acta Amazonica 6: 163–170.

Rudall PJ, Sokoloff DD, Remizowa MV, Conran JG,Davis JI, Macfarlane TD, Stevenson DW. 2007. Mor-phology of Hydatellaceae, an anomalous aquatic familyrecently recognized as an early-divergent angiospermlineage. American Journal of Botany 94: 1073–1092.

Rudall PJ, Eldridge T, Tratt J, Ramsay MM, Tuckett RE,Smith SY, Collinson ME, Remizowa MV, Sokoloff DD.2009. Seed fertilization, development, and germination inHydatellaceae (Nymphaeales): implications for endospermevolution in early angiosperms. American Journal of Botany96: 1581–1593.

Saarela JM, Rai HS, Doyle JA, Endress PK, Mathews S,Marchant AD, Briggs BG, Graham SW. 2007. Hydatel-laceae identified as a new branch near the base of theangiosperm phylogenetic tree. Nature 446: 312–315.

Schnarf K. 1929. Embryologie der Angiospermen. Berlin:Gebrüder Borntraeger.

Schneider EL, Chaney T. 1981. The floral biology of Nym-phaea odorata (Nymphaeaceae). Southwestern Naturalist26: 159–165.

Schneider EL, Tucker SC, Williamson PS. 2003. Floraldevelopment in the Nymphaeales. International Journal ofPlant Sciences 164: S279–S292.

Soltis PS, Brockington SF, Yoo MJ, Piedrahita A,Latvis M, Moore MJ, Chanderbali AS, Soltis DE.2009. Floral variation and floral genetics in basalangiosperms. American Journal of Botany 96: 110–128.

Taylor DW. 2008. Phylogenetic analysis of Cabombaceae andNymphaeaceae based on vegetative and leaf architecturalcharacters. Taxon 57: 1082–1095.



Taylor ML, Williams JH. 2009. Consequences of pollinationsyndrome evolution for post-pollination biology in anancient angiosperm family. International Journal of PlantSciences 170: 584–598.

Troll W. 1933. Beiträge zur Morphologie des Gynaeceums. IV.Über das Gynaeceum der Nymphaeaceen. Planta 21: 447–485.

Van Miegroet F, Dujardin M. 1992. Cytologie et histologiede la reproduction chez le Nymphaea heudelottii. CanadianJournal of Botany 70: 1991–1996.

Wang XF, Tao YB, Lu YT. 2002. Pollen tubes enter neigh-bouring ovules by way of receptacle tissue, resulting inincreased fruit-set in Sagittaria potamogetifolia Merr.Annals of Botany 89: 791–796.

Wiersema JH. 1988. Reproductive biology of Nymphaea(Nymphaeaceae). Annals of the Missouri Botanical Garden75: 795–804.

Wiersema JH, Hellquist CB. 1997. Nymphaeaceae. In:

Committee FoNAE, ed. Flora of North America. New York,NY: Oxford University Press.

Williams JH. 2008. Novelties of the flowering plant pollentube underlie diversification of a key life history stage.Proceedings of the National Academy of Sciences of theUnited States of America 105: 11259–11263.

Williams JH. 2009. Amborella trichopoda (Amborellaceae)and the evolutionary developmental origins of theangiosperm progamic phase. American Journal of Botany96: 144–165.

Williamson PS, Schneider EL. 1994. Floral aspects ofBarclaya (Nymphaeaceae) – pollination, ontogeny andstructure. Plant Systematics and Evolution 159–173.

Zhang Q, Liu Y, Sodmergen. 2003. Examination of thecytoplasmic DNA in male reproductive cells to determinethe potential for cytoplasmic inheritance in 295angiosperm species. Plant and Cell Physiology 44: 941–951.



pollen tube growth and the pollen-tube pathway of nymphaea odorata

Documents