The floral development of Neurada procumbens L. (Neuradaceae)

(1)

The floral

_development

of Neurada

_procumbens

L.

(Neuradaceae)

L.P. Ronse+Decraene and E.F. Smets

Laboratory ofPlantSystematics, Botanical Institute, KatholiekeUniversiteit Leuven, Kardinaal Mercierlaan92, B-300IHeverlee,

Belgium

SUMMARY

The floral_ontogenyof Neurada

_procumbens

L. has been

_investigated

to

_clarify

the

_position

of the

_family

Neuradaceae andtoelucidatea

number of

morphological problems,

such as the nature of the

epicalyx

and the

_{10-carpellate gynoecium.}

_{Morphological}

and

ontogenetic

evidence _suggests_aclose

_affinity

with the_Rosaceae,

contrary to _reports ofrbcL sequencedata. There is astrong

gynoecial

resemblance with

_Maloideae,

but other characteristics

suggest thatthe affinities ofNeuradaceaelieatthe base ofthe

Rosaceae.

Key-words:

floral

ontogeny,

epicalyx, gynoecium,

Neurada

procumbens,

Neuradaceae,

Rosaceae.

INTRODUCTION

The Neuradaceae consists of the threesmall _genera: Grielum L.

_(five

to six

_species),

Neurada L.

_(one

_species)

and

_Neuradopsis

Bremek. &

_Obermey.

_(three

_species)

(Melchior

1964;

Hutchinson

_{1964; Cronquist}

_1981;

_Mabberley

_1987).

Most authors agreethat the

family

is

closely

related tothe Rosaceaein which ithas sometimes been

placed

asa

subfamily

ortribe

_(e.g.

Focke

_1894;

Murbeck

_1916,

_1941;Lawrence

_1951;

Hutchinson

_1964,

_{1973; Rowley}

_1978;

Thorne

_1983).

The Neuradaceae differ from the Rosaceae

_by

a few

_conspicuous

_characters,

such as their

_unique

_{pollen morphology,}

unusual

_zygomorphic

_gynoecium,

_habit,

seed

_morphology

and

_embryology

(see

Murbeck 1916; Melchior1964;Corner

_{1976; Cronquist}

_1981).

However, these

differ-ences are

usually

consideredas

merely phyletic

advances

_compared

_{with the gross of the} Rosaceae

_{(cf. Cronquist}

_1981).

Willis

(1966)

takesan

exceptional position

in

pointing

to similarities with the Malvaceae in the leaf

_{shape, vertically}

inserted

_carpels

and similar colour

_changes

ofthecorollaupon

drying.

The Neuradaceae have also been

_compared

with the

_{Chrysobalanaceae}

orthe Geraniaceae

(see

Murbeck

_{1916). Recently, Morgan}

etal.

_{(1994) suggested}

_{that the rbcL sequence data of Neurada}are notcongruentwith

aclose

relationship

with the

Rosaceae;

instead,

Neurada is the sister groupof

Gossypium

(Malvaceae)

on their rbcL tree,in line with the

opinion

ofWillis

(1966).

The

_only

detailed

_{morphological study}

ofthe_{group has}been carriedout

by

Murbeck

(1916)

who _gavea detailedaccount of the _{genera Neurada and} Grielum. Neurada is

highly

unusual in its habit and flower

_morphology.

It is _a low

_spreading

annual herb

growing

inan area

ranging

from the Southern Mediterraneancoasts to Indian deserts.

The flowers and fruitsare

flattened, strongly epigynous saucer-shaped

mounds covered with

_spines

and athick indumentum. The

petals

and

_{hardening styles}

(2)

440 L. P.RONSE DECRAENE AND E. F. SMETS

_gynoecium

differs in essential details from the current state known in the

Rosaceae and shows _a number of

_{specific morphological}

characteristics. At

_maturity

one side of the

gynoecium

is

normally developed,

while theother aborts. This leadsto

a

zygomorphic gynoecium

while the flower remains

_regular

in its other

organs. Thereare

10

_carpels

thatare

generally

believedto have arisen

_by

dedoublementof_an

_original

set of five. This enabled earlier authors to link Neurada with Rosaceae

_having

five

antesepalous carpels,

aswas done

by

Murbeck

₍₁₉₁₆₎

on the basis of the orientation of

the ovules.

_However,

this

assumption

has neverbeen checked

ontogenetically.

Uncertainty

also

_reigns

on the number of ovules per

carpel,

assome authorsaccept

a

single pendulous

ovuleper locule

(e.g.

Focke

_1894;

Willis

_{1966; Cronquist}

₁₉₈₁₎

or one

to two ovules

(e.g.

Murbeck 1916;Melchior

_{1964; Goldberg}

_1986).

For

Murbeck,

two ovules_are

_{originally superposed}

within each locule and the basalonebecomes aborted.

The _presence of floral

_spines

in Neurada _represents _an

_interesting

character for

comparison

with the Rosaceae where similarstructures arefound insome genera

(e.g.

Agrimonia).

Murbeck

₍₁₉₁₆₎

believedthe

spines

ofNeuradatobe

_secondary

_emergences andnot an

epicalyx.

Hutchinson

(1964)

and Willis

₍₁₉₆₆₎

refer to an

epicalyx

of five

bracteoles. Grielumbears_no

_spines

but has short knobs and

Neuradopsis

has

_spines

but not the five

_bracteoles,

which Hutchinson

(1964)

takes as a

diagnostic

character to

distinguish

Neurada from the other genera. Little is known about the

morphology

of the

nectary. Murbeck

₍₁₉₁₆₎

mentionsaweak intrastaminal

‘Ringwulst’

covered with hairs

in Neurada. In _{Grielum these emergences may be scale-like}

_{(‘Schuppen’).}

The above-mentioned

morphological problems,

aswellasthe fact that evidence from

molecular data does not fit with the

_{morphological}

evidence

_justifies

a renewed

investigation

of Neurada. More data can

provide

astrongerbasis for

_discussing

the

relationships

of thegroupin relationto the Rosaceae.

MATERIALS AND METHODS

Flowering

material of Neurada

_procumbens

L. was collected

_by

the first author

during

afield

_trip

onJerba

_(Tunisia).

Flower buds of

_{Agrimonia eupatoria}

L.

_(Rosaceae)

_were

gathered

on Mont Panisel near Mons

_(Belgium).

Reference material

_(pickled:

Ronse Decraene 310 L

t

and 198 L 0

)

and aherbarium

_{specimen (Ronse}

Decraene

₉₆₈₎

are

kept

at the Botanical Institute of the Katholieke Universiteit Leuven

_(LV).

For methods of

preparation

werefertoearlier_reports

(Ronse

Decraene

_1990;

Ronse Decraene & Smets

1991).

Observations were made with a

Philips

501.B

_(Meise)

and a Jeol JSM.6400

scanning

electron

microscope (Leuven).

RESULTS

Flowers arise

_{sequentially alongside}

a

continuously

growing apical

meristem

_{(Figs la,b,}

2A). Inception

startswith the

_unequal

division of_an

_{elliptical primordium}

which arises betweentwo leaflike

_{prophylls (A1}

and

_Bl).

One_partof this division

_gives

rise toafirst flower

_primordium,

while the other

_produces

a new

pair

of

_{prophylls (A2}

and

_B2)

and

repeatstheprocess.

By unequal division,

alanceolate bract-likestructure

_(p)

is detached from each flower

_{primordium (Fig. 2B).}

The

_growth

of this

_appendage

is limited and it

(3)

prophylls

grow

unequally;

prophyll

B1 becomes

displaced

along

the

horizontally

growing

stem, while A1 remainsnext to the first flower that it

_originally

enclosed. B1 encloses the

growing point

of the inflorescence that

_produces

thetwo next

_prophyll

A2 and B2. _{This process is}

_repeated

witha

displacement

of B2 to a

higher

level and the

insertionof B1 next tothe second flower and

A2,

and _{so on.}As_a

_result,

each node bears

a

flower,

asmall

appendage (p)

andtwoleaves of

_unequal

size

_{(A1-B0, A2-B1, A3-B2,}

etc.). Prophyll

B is

_{always larger}

than

_prophyll

A and has been shifted froma lower

level; prophyll

A standsonthe side of the flower

primordium

withan

angle

of 90° tothe bract-like

_appendage

_p

_subtending

the flower

_(Fig.

_la).

B is

always

situatedat the underside in relationto the mainstemand floweron

opposite

sides _{of p, while A is} inserted in an upper lateral

position.

The inflorescence is monochasial and could be

termedacincinnus with aterminal flower

placed

ateach node.

Prophylls

also differin the fact that Acovers a small lateral bud

(s)

that

only develops

afew leaves in later

stages

(Figs

la,b, 2A),

whileB has_no small lateral bud but encloses the main

growing

(4)

442 L. P. RONSE DECRAENE AND E. F, SMETS

(5)

inflorescence apex. Each

subsequent

flower is orientedat an

angle

of 90° in relationto

an older

flower,

concomitant with the

_position

of B.

Calyx inception

is successive

along

a2/5 sequence

(Figs

la, 2A-C).

Sepals

arise as

hemispherical

primordia

on the

periphery

ofa flattened floralapex. The first

sepal

is oriented towards

_prophyll

B

(B3

of

_Fig.

_la,b;

B1 of

_{Fig. 2A)}

and arises

_{simultaneously}

with

_sepals

number 2

_(oriented

towards the inflorescence

_apex)

and 3

_(oriented

towards

prophyll

A; A4on

Fig. la,b,

A2on

Fig. 2A).

Insome casesafourth and fifth

sepal

follow

immediately

orthe fifth

lags

in time. Thefifth

sepal

is situated

against

p

(Figs. la,b, 2A).

Sepal primordia rapidly

become almost

_equal

in size_as

_{they progressively}

_coverthe floral apex.

Sepal shape

is

triangular

with rounded

apices;

sepals

touch each other

laterally

without

_fusing

and

_finally

coverthe bud

_completely

inavalvate aestivation

_(Fig.

2

_C,D).

At that timeunicellular hairsare initiated

basipetally

on each

sepal

in the order of the

calyx

inception (Fig. 2C,D),

finally covering

the whole

_sepal

in athick indumentum.

Sepals

are

persistent

but

_they

donot_{grow much from}this stageon.Thebasal part ofthe flower increases

_dramatically

in size and becomes _a broad

_platform

_on which the

_sepal

lobes_are inserted.

_Removing

the

_sepals

atthis _stagereveals _aflattened

_pentagonal

apex

with a

girdle

of

primordia

on the

periphery

(Fig. 2E).

Petalsand

antesepalous

stamen

primordia

arise

_{simultaneously}

and donot differ in size and

_{morphology (no}

_stageswere

seen with

only

petals

and without

stamens).

Antepetalous

stamen

_primordia

follow

rapidly

assmall

hemispheres (Fig. 2F). They

donot

always

arise

_{simultaneously,}

as can

be seen

by

the

_slight

differences in size

_{(Fig. 3A).}

As aresult of

this,

three

_alternating

whorls canbe

readily

observed

_surrounding

aflattened floral apex.

Antepetalous

stamen

primordia

and

_{petal primordia}

remain of

_equal

size for a

long time,

as

petal growth

is

veryslow. Size differences between thetwo stamenwhorls are

expressed

very

early

with the

_antesepalous

stamens

_{remaining largest (Fig.}

_{3B,D,E); peripheral growth}

lifts the

antesepalous

stamens on a rim above the

_antepetalous

stamens

_(Figs

_{3D, 4D, 5A,B).}

Filaments appear

simultaneously

with anther differentiationatthe time the centralarea

becomes

_{invaginated by peripheral growth (Fig. 3D).}

Thestamensare

slightly

bentover

the floral

apex with the

antesepalous

stamens

overlapping

the lower

antepetalous

stamens in a manner characteristic for Rosaceae. Four

pollen

sacs

develop

on each

anther;

the dorsalsare

larger

and

_diverging

from the

ventrals,

whichconverge towards

each other

_{(Figs 3E, 4D, 5A,B).}

Anthers arebasifixed atanthesis and filaments havean

inflated base. Petal

growth lags considerably

behind that of the stamens.

Only

before

anthesis do the small

_{ligulate primordia}

increase in size and_overtop thestamens

(Fig.

5B). They

cover the floral bud in acontorted aestivation. Petals

drop

off

rapidly.

As

early

asstamen and

petal inception, globular primordia

become_apparentoutside the saucerlike floral

primordium just

below the

_sepal

insertion. These

_primordia

initiate the

_epicalyx

andare situated in

_antepetalous

_{position (Figs 2E, 3D);}

eachone is

_rapidly

followed

_by

two

_{adjacent primordia opposite}

the

_{sepals (Fig.}

_3B,E,F).

A third whorl

(A)View of inflorescence apexshowingtwoflowersand agrowing point (most prophylls removed).Asterisks_representthe youngestprophyllsrelated tothe inflorescence_apex.(B)View of inflorescence _apex and anolder bud with the inceptionof the sepals; oneprophyll removed. Note the separationof the bract-likeappendagefrom the _youngflowerbud (arrow). (C)Older_stageof_{young flower} with initiation of trichomesonsepals1and 2. (D)Olderflower_bud;thesepalsenclosethe bud inavalvate

aestivation. (E)Initiation of theepicalyx, antesepalousstamensandpetals; sepalsremoved.(F)Detail ofpetal and stamenprimordiaatthe earliest_inceptionof the_antepetalousstamens(arrows).Bars=100_pm,_{except (B)} and (F)=50pm.Abbreviations: AI,inflorescenceapex;AC, antesepalousstamenprimordium; F,flowerbud; E,epicalyx primordium;K,petal primordium;P,bract-like_appendage;PA, PB,prophyllsA andB; S,lateral shoot. Numbers indicate orderofinceptionofsepals.

Neurada procumbens.

(6)

444 L. P. RONSE DECRAENE AND E. F. SMETS

©1995 Royal BotanicalSocietyof TheNetherlands,Ada Bot. Neerl. 44,439-451 Neuradaprocumbens (sepalsremoved in allcases). (A)Lateralviewshowinginitiation ofpetalsand androecium. (B)Lateralviewof thesameshowingthebasipetal inceptionof the_{epicalyx primordia.}Arrows pointtotheinceptionof the third whorl. (C) Apical view; antesepalousstamensremoved. Formation ofslight

depressionsonthe flatapex. (D) Slightlyolder bud showingthebeginningof thecurvature of thestamensand anther initiation. (E) Lateral view of older bud with epicalyx members. (F) Lateral view similar to (C) with epicalyx development,Bars=100_pm, except (C)=50_pm. Abbreviations: AC, antesepalous stamen; AK,antepetalousstamen;K,petal.Numbers indicateorder of_inceptionof_epicalyx members.

(7)

arises with two

_antepetalous

_{primordia just}

belowthe

first, together

with another one

betweentwo

_{adjacent pairs}

in

_{antesepalous position (Fig. 3B).}

A fourth whorl arises in the

spaces formed

by

the second and third whorl. Murbeck

₍₁₉₁₆₎

confounded the third and fourth whorlsonhis floral

_{diagram (p.7; Fig. 2A).}

He mentioned

_only

a

single primordium

opposite

the first

_whorl;

wesaw twoof these. More

_primordia

_{may appear}

_basipetally

in alternation with

_previous

ones. It remained difficultto

identify

theirexact numbers and

positions

dueto the abundant indumentum. Primordia grow into

spines,

whichare

largest

ontop of the flower.At

_maturity

_{the upper}

_spines

have

_strongly

inflatedbases.

When the

_periphery

of the flower starts to be _{lifted up,} ten

depressions

become

apparenton the flattened centralareainalternation with thestamen

_{primordia (Fig. 3C).}

The

depressions

become

_{progressively}

slit-like

_by

_{the appearance of}

_{arc-shaped margins}

overarching

each

_depression

_{(Fig. 4A,B);}

_adaxially

_no

_margin

is formed. These _arcs appear

laterally joined

intoasinusoidal

_{girdle taking}

_upthe spacebetween the insertion of the stamenwhorls

_{(Fig. 4A,B),}

but

_they

soon become detached

_by

their

_pronounced

growth

into

U-shaped primordia (Fig. 4C).

Theslits extend in size towards the central

flattenedareaof the

_flower,

while

_they

become elevated inanalmost vertical

_{position by}

peripheral growth.

At that timethe

_carpels

_appearasflattened

_wings

orcurtainsas

they

are

fully

ascidiate

_{(Fig. 4D,E).}

The lower_part of the

_gynoecium

_appears

_congenitally

fused and is

completely

embedded in the

receptacular

tissue

(Fig. 4D,E).

The ascidiate

structure, which _was

_originally

_seen _as _anindividual

_carpel,

_now_curves with the

apical

part

pointing

towards the stamens and is _{lifted up}

_by

the formation of a stalk witha

longitudinally running

groove

(Figs 4F, 5A-C).

The result is a

fairly long style

witha

broad

_stigmatic

area

showing

two lobes and aslit turned towards the stamens

(Figs

5C,D, 6A).

The upperpartof the

_{style, just}

below the

_{stigmatic papillae,}

bears a

large

number ofstomata.Ovule

_primordia

arise

_just

above the

_congenitally

fusedareabetween

the curtain-like

_margins.

_They

are connected with the

carpel margins

on

opposite

sides andarise

_singly

orasan

unequal pair

withineachlocule

_{(Fig. 4C,D).}

One ovuleissmaller and

_readily

aborts. Insome cases asmall

cavity

is seenbelow the

remaining

ovule

_(Fig.

6A,

arrow), corroborating

Murbeck’s observations. The

_remaining

ovule _curves out-wards and becomes enclosed

_by

two

_{integuments (Figs 4F, 5A). During}

ovule maturation theareaof the flower between the ovule insertion and the

_sepals

extends

_horizontally

in

considerable

_proportions.

Ovules tend to be

_pressed

into sinuous bodies within the limited space

(Figs 5C,D, 6A). They

are

_{strongly anatropous}

and canbe detached from

their massive funiculus.

Only

at a very late stage of

development

doesone sideof the

gynoecium

stop

growing, giving

it_a

_zygomorphic

appearance.

The area between the stamens and

gynoecium

becomes

_densely

covered with unicellular hairs

_during

the

_development

of the

_{style (Fig. 5B-D).}

Close

_inspection

of thestamen bases after

_partial

removal of the hairs reveals somesunkenstomata which

may indicate the presence of nectariferous tissue.

However,

at

maturity

thereare no

obvious traces of a nectary. To our

knowledge,

no information is available about

pollination

of the flower. After

anthesis, petals

and stamens

drop,

but the

_sepals

and erect

_styles

remain_on _top of the

_spiny

fruit.

For

comparison,

a few floral buds of

Agrimonia

eupatoria

(Rosaceae)

were also

observed with

_{special emphasis}

on the

_{epicalyx development (Fig. 6B-E).}

Flowers are

essentially diplostemonous

withanoften

_{incomplete antepetalous}

whorl

_{(Fig. 6E).}

The

epicalyx

consists of

_long

bristles that arise ina

centrifugal

fashion similarto Neurada.

(8)

L. P. RONSE DECRAENE AND E. F. SMETS

446

Neurada resembles the Rosaceae in several_aspects of its floral

_development,

such as the

formation ofa

hypanthium

withstamensinsertedat two

levels,

the curvedstamenswith theouter inanupper

position,

the

early dropping petals

with retarded

growth

and small (A)Partial view of the central area ofthe flower showing gynoecial slits. (B)Apicalviewofaslightly olderbud;stamensremoved. (C)Older

stageshowingtheascidiate form of the carpelprimordia around thereceptacularresidue. Notethe ovuleprimordiumonthe flankofacarpel (arrow). (D)Longisection throughflowerbudatabout thesamestageas(C),showing positionofstamensand ovules (arrows). (E) Detail of curtain-like carpellaryfolds. (F)Section throughonecarpel showingoneovuleand

developing style. Bars=100 _pm,except (A)=50_pm. Abbreviations: AC,antesepalousstamen;AK, antepeta-lous stamen;K, petal; R, receptacularresidue.

(9)

insertion

_{base; sepals}

arise in_analmost identical_{manner as} small

_persistent

lobes _on _a

broad

platform;

the

_spines

_are reminiscent of similar structures;aninferior ovarywith

persistent styles

and few axile ovules is also characteristic.

Murbeck

₍₁₉₁₆₎

described the

_{complex sympodial}

inflorescence of Neurada and also

provided

a

drawing.

Our observations _{agree with} his

_description

in that the flowers stand inaterminal

position

between two

unequal

prophylls (‘Vorblatter’). Also,

the

larger prophyll

(Fig.

1:

_B)

encloses the bud that continues the inflorescences and ends with thenext

flower,

while the smaller

(Fig.

1:

_A)

bears_ashort stemin its axil. Each

larger prophyll

is believed to have been shifted to a

higher

level

(next

to a younger

flower),

afact thatwecould also observe. Murbeck

interpreted

the small

appendage (p)

asasmall

ligular stipule belonging

to

prophyll

A. He believed the other

stipule

of the

pair

to be lost duetothe horizontal

_growth

of thestem.

_However,

the

_position

ofp is

opposite

each flower and is innoway linked with the smaller

prophyll (Figs

la,

2A,B).

Moreover,

it arises

independently

of the

_prophylls

andis

_{initially larger.}

Asindicated

_by

Murbeck,

the horizontal

_growth

_{form may well be}

responsible

for the

_displacement

of flowers and bracts and the unusual construction of the inflorescence.

(A)Longisectionwithdevelopingstyleandtwoovules(arrows). (B)Partial view of flower_priortoanthesis. Notethe indumentum between stamensandcarpels and _youngstyles; anthers removed. (C) Longisection throughflowershowingtheerectstyleswith extrorsestigmasandtwoovules (black arrows).Notethe_grooverunning throughthe middle of thestyle(white arrow), (D)Lateral viewof section of the central_partof the flowershowingtheovules and_partsof the locules. All bars= 100_pm.Abbreviations: AC,antesepalousstamen;AK, antepetalousstamen;K., petal; R,central residue of the _apex.

Neuradaprocumbens.

(10)

_epicalyx

is controversial in the Rosaceae and in the Neuradaceae alike. Hutchinson

₍₁₉₆₄₎

_probably

confused the _upper

_spines

of Neurada

(which

alternate with the

_{calyx lobes)}

with

_bracteoles,

as each upper

spine

is connected with

more

appendages

arising

basipetally.

Kania

₍₁₉₇₃₎

_interpreted

the

_epicalyx

of the Rosaceae as emergences, because he could not find _any

_ontogenetic

evidence for a

stipular

nature

_(the

classicalview, cf. Eichler

_1878).

This

_{interpretation}

_wasrefuted

_by

Trimbacher

₍₁₉₈₉₎

who

_presented

a

morphological

sequence

starting

from the

_simple

appendages

of

_Rhodotypos

tothe

complex

epicalyx

of

Agrimonia.

In

Rhodotypos

the

epicalyx primordia

arise as

marginal appendages

of theouter

sepals

and recall

_stipules.

All othertaxaof the Rosaceaeaswellas Neurada have their

epicalyx primordia arising

Longisectonof gynoeciumwith _young ovule and styles.Note the small locular_spacebelowthe ovule (arrow). (B)-(E)

Neurada procumbens.

Fig.6. (A).

(B) Early stageof the initiation of the epicalyx. (C) Lateral viewof older budshowingthe centrifugal inceptionofepicalyx primordia.(D) Idem, nearlymature_{stage. (E)}Lateral view of_youngflowerbud _showingpetals, antesepalousstamensand two

antepetalousstamens (arrows); sepalsremoved. All bars= 100pm. Abbreviations: C, sepal; P, prophyll; K,petal;AC,antesepalousstamen; G, carpelprimordium.

(11)

independently

and outside the

_sepal

whorl.

_They

arise

simultaneously

except in Comarum

_{palustre L.,}

where the

_{epicalyx primordia}

continue the

_spiral

_{sequence of the}

sepals, indicating

their

_stipular

nature. _{Trimbacher concluded that the whorled} arrange-ment is induced

_by

the isolation and

_displacement

of the

_stipules.

Neurada

_closely

resembles

Agrimonia

in the

inception

of its

_epicalyx

(Fig. 6B-D).

There is asimilar

centrifugal

increase of

_{appendages recalling}

the

_secondary

increase ofstamens.Also the floral

_development

of

_Agrimonia

bears similarities with Neurada

_{(Fig. 6E).}

The androecium of Neurada is a

_typical

illustration of

diplostemony

and is

charac-terized

_by

the absence of interactions

_(in

time and

_space)

between

_petals

and stamens

(such

as

stamen-petal

complexes) (see

also Ronse Decraene & Smets

_1995).

Neurada shares a

diplostemonous

androecium with taxaof the Rosoideae

_{(e.g. Stephanandra,}

Agrimonia:

Fig.

6E)

and the

_Quillajeae

_(e.g.

_Quillaja)

of the Rosaceae. Endress &

Stumpf (1991)

also observed the stamens of Neurada

procumbens.

Theanther

shape

with

_larger

dorsal

_pollen

sacs agrees with the Rosaceae. Their observations also agree with ours except for _{their report of}

_hairy

filaments. A thick indumentum is found at the base of the

_filaments,

but it is inserted on the

_hypanthium,

not on the filament

(Fig. 5B,C).

Murbeck

_{(1916) placed}

_a link with Maloideae where_an inferior

_gynoecium

of five

carpels

is characteristic.

_Indeed,

the

_gynoecial

_development

of Amelanchier and other Maloideae resembles Neurada in several

_{points (compare}

with Steeveset al.

1991).

1. The

_{gynoecial primorida}

contribute littleto the

_development

_{of the ovary;}

_instead,

they develop

almost

_exclusively

into the

_style

and

_stigmatic

areabut

they

also contribute to the initiation of the ovules thatappear

laterally

on the

carpellary

flanks

_{(Fig. 4C,D).}

The fact that _a

_wholly

ascidiate

carpel develops

asa

stigmatic

structureis notcommon

in the dicots. The entire

_{carpel develops}

asastalk which is the continuationof the septa.

Carpellary

tissue

hardly

participates

in the

_development

of the

ovary. Processes of

development

occur

deep

in the

_receptacular

tissue and the limits between

_carpellary

and

_receptacular

tissue are not discernable below the insertion of the

_{style (Figs}

4E,

5C,D, 6A).

2 The

_deepening

of the

_hypanthium

induces the

_{carpel primordia}

toextend

vertically

inside the continuous

_cylinder

of meristematictissue

(Figs 4D,E, 5A).

3 The basal

_region

_{of the floral cup between ovule insertion and}

_periphery

_expands

horizontally during

ontogeny and forms the roof of the ovary

(Figs 4F, 5C,D, 6A).

4 The

_gynoecial

_development

of Rosaceae with an inferior ovary and that of Neurada

is not

strictly epigynous (in comparison

with the

_development

_in,

e.g. Asteraceaeor

Dipsacaceae).

Indeed,

the

_epigynous

appearance is enhanced

by

strong

hypanthial

growth

lifting

perianth

and the androecium above the free

_gynoecial

_parts. The

receptacle

remains flattened

_{(Figs 4C,D, 5A-C)}

or forms a domeon which ascidiate

primordia

arise

(e.g.

Rosoideae: van Heel

1981,

1983). Carpels

do not fuse as such with the

_{receptacular tissue; only}

their basal

parts become connected with

_receptacular

tissue.

The difference between the

_{five-carpellate}

condition ofmost Maloideae and the 10

carpels

of Neurada is

_important.

Ten

_carpels

_develop

in

_Neurada,

with a

position

intermediate between thetwo stamenwhorls. It would be

_tempting

toacceptan

original

dedoublementof five

carpels

as is done

by

most authors.

_However,

there is

_absolutely

(12)

_by

the division of a

complex

primordium,

or

they

may arise

_{independently}

and be at the same time connected

_(see

Ronse Decraene & Smets

_1993).

In Grielum there are 5-10

_carpels

_(Focke

_1894).

The

_suggestion

of a

partition

of the locules

_by

false

septaseems more

plausible

than the idea of

dedouble-ment,ascertaingeneraof Maloideae show falseseptawhich

partially

divide the locules

(e.g.

Amelanchier, Malacomeles,

Peraphyllum

(Murbeck

1916;

Steeves et al.

1991;

Rohrer et al.

_1994).

In Neurada this

_invagination

is believed to be

_complete

and

accompanies

theshift of the

_placental

_area to a central

position (Murbeck

1916).

In

this case one would expect that

_only

one ovule would remain within a locule

(as

they

tendtobe

paired

in

_Maloideae).

As

_{suggested by}

Murbeck

_(1916),

the orientation of the ovules in Neurada indicatesan

original

arrangementin

_{pairs. Indeed,}

the ovules

are insertedon one lateralflank ofa

carpel (Fig. 4C,D)

and not in a

strictly

median

position.

On the other

_hand,

_{the presence of}asupernumerary sterile ovule tendsto

_disagree

with this

_assumption

as the second ovule is

_laterally

insertedon the other

_carpel

flank and not

strictly

below the fertile ovule. There is

_occasionally

ahollow space below the fertile

_ovule,

where the lost ovule should have beennestled

(Fig.

6A,

compare with Murbeck

_1916).

Moreover, there isno difference between the real and false

_septa.

Most

Maloideae have two collateral ovulesper

carpel;

in

Cretaegus

and

_{Mespilus they}

_are

superposed

and

_only

one

develops

into a seedas the fruitmatures. Neurada shows the

independent

inception

of 10

_carpels

without sufficient evidence for a

pairing

or the

building

of false_septa.

_However,

the similar_arrangementof all locules in Neurada does not exclude that

_possibility

perse.

On the basis of the

_{morphological evidence,}

_we

may

confidently

suggestthatNeurada

belongs

to the

_vicinity

of the Rosaceae. However,

Morgan

etal.

(1994)

indicate that rbcL data do not _support a close

_relationship

between Neurada and the Rosaceae.

Instead,

a group

composed

of

_Rhamnaceae,

Moraceae and Ulmaceae should be

consideredas the sister groupstothe Rosaceae.Theseresults are

incongruent

with the

morphological

and

_ontogenetic

information that has been assembled

_during

morethan

acentury. As with all characters used _{in taxonomy,} _one mustbe careful in

_deciding

about the

_impact

of different

_approaches.

This seems tobe the case where molecular

dataare in conflict with

morphological

evidence. More characters from othersources

would be also

_helpful

for

_clarifying

this

_{incongruence.}

_Zhang

_(1992),

for

example,

mentions the lack of informationon the woodanatomyof the Neuradaceae.

A

_relationship

with Malvaceae

_(as

_suggested

_by

Willis

₁₉₆₆₎

_on the

superficial

resemblance of the flower is difficultto _support

_against

the wealth of evidence of a

rosoid

_affinity.

Affinities with

_specific

taxaof the Rosaceae are difficult to determine.

Apart

from_strong resemblances in the

_{gynoecial morphology,}

a

relationship

with the Maloideae tendsto be excluded

by

the basic chromosome number of Neurada

(x=7),

which doesnotresemble Maloideae

(x= 17)

but

corresponds

to the basic number of the Rosaceae_as found in the Rosoideae

_(Morgan

etal. 1994).

Also,

certain

_{embryological}

characters tendto be different from the

_Maloideae,

suchasabsence ofanobturator and

endosperm

(Murbeck

1916).

In certain characters Neurada resembles taxa of the Rosoideae

_(e.g.

chromosome

_{number, diplostemony, epicalyx)}

while

_{other, especially}

gynoecial

characters

_point

to the Maloideae. In the latter _case _we

(13)

We thank the director of the National Botanic Garden of

_Belgium

for

_permission

touse

the SEM facilities. We are indebted to Prof Dr C.

_{Vandenberghen}

for

_leading

a

botanical

_trip

toTunisia in 1988. This

_study

was

supported by

research

_{grants (project}

No. 2.0038.91:

_scanning

electron

_microscopy

and

project

No. G.0143.95:

general

research

_project)

from the National Fund for Scientific Research of

_{Belgium (NFWO).}

The

leading

author is a

postdoctoral

researcher of the NFWO.

REFERENCES

Comer, E.J.H,(1976): The Seeds of Dicotyledons. Cambridge University Press,Cambridge, Cronquist,A.(1981):An IntegratedSystemof

Clas-sification ofFloweringPlants. Columbia University Press,NewYork.

Eichler, W.A. (1878): Bliithendiagramme II.

Engelmann,Leipzig.

Endress,P.K. &Stumpf, S. (1991):Thediversity of

stamen structures in ‘lower’ Rosidae (Rosales, Fabales, Proteales, Sapindales).Hot. J. Linn. Soc. 107; 217-293.

Focke, W.O. (1984); Rosaceae. In; _Engler, A. & Prantl,K, (eds): Dienatiirlichen _{Pflanzenfamilien} HI, 3. 1-61. Engelmann, Leipzig.

Goldberg,A. (1986): Classification, evolution and phylogeny of the families of Dicotyledons. Smithson. Contr.Bot. 58:1-314.

Heel, W.A. van (1981): A SEM-investigation on

thedevelopmentof free carpels. Blumea27: 499-522.

Heel,W.A.van(1983):The ascidiformearly devel-opment of free carpels, a SEM-investigation. Blumea 28: 231-270.

Hutchinson, J. (1964): The Genera of Flowering Plants. Vol. 1. Clarendon Press,Oxford. Hutchinson, J. (1973): The Families _ofFlowering

Plants

,

3rd edn. Clarendon Press,Oxford. Kama, W. (1973); Entwicklungsgeschichtliche

Un-tersuchungenanRosaceenbliiten. Bot. Jahrb.Syst.

93: 175-246.

Lawrence, G.H.M. (1951); Taxonomyof Vascular Plants. MacMillan,New York.

Mabberley.D.J.(1987):The Plant-Book. APortable Dictionary of the Higher Plants. Cambridge

University Press,Cambridge.

Melchior,H. (1964): Engler’sSyllabusder Pflanzen-familien,12th edn. Bomtraeger,Berlin.

Morgan, D.R., Soltis, D.E. & Robertson, K.R. (1994):Systematic and evolutionary implications of rbcL sequence variation in Rosaceae. Am. J. Bot. 81: 890-903.

Murbeck,S. (1916):Uber dieorganisation,Biologie und Verwandtschaftlichen Beziehungen der Neuradoideen. Lund Univ. Arrskr.212: 1-29. Murbeck, S. (1941): Untersuchungen fiber das

Androeceum der Rosaceen. Lund Univ. Arrskr. 37, 7:1-56.

Rohrer, J.R., Robertson,K.R. &Phipps,J.B.(1994): Floral morphologyof Maloideae (Rosaceae)and itssystematicrelevance. Am. J.Bot. 81: 574-581. Ronse Decraene,L.P.(1990); Morphologicalstudies in Tamaricales. I. Floral_ontogenyand_anatomyof Reaumuria vermiculata L.Beitr. Biol. Pflanz.65:

181-203.

Ronse Decraene,L.P. &Smets,E.F.(1991): Andro-ecium and floral nectaries ofHarungana

madagas-cariensis(Clusiaceae).PI.Syst. Evol.178: 179-194. Ronse Decraene,L.P. &Smets,E.F.(1993):

Dedou-blement revisited: towards a renewed interpre-tation of theandroeciumof the Magnoliophytina.

Bot. J.Linn. Soc. 113: 103-124.

Ronse Decraene, L.P. & Smets, E.F. (1995): The distribution and systematic relevance of the androecial character oligomery.Bot. J. Linn.

Soc. 118.

Rowley,G.D.(1978):Rosaceae. InHeywood,V.H.

(ed.) Flowering Plants _of the World. 141-144. Oxford_University Press,Oxford.

Sleeves, T.A., Sleeves, M.W. & Randall Olson, A. (1991): Flower developmentin Amelanchier alni-folia(Maloideae).Can. J.Bot. 69: 844-857. Thorne,R.F.(1983):Proposednewrealignmentsin

theAngiosperms.Nord.J. Bot. 3: 85-117. Trimbacher, C.(1989):DerAussenkelch der

Rosa-ceen.In:Weber, A., Vitek,E. &Kiehn,M.(eds)9. Symposium Morphologic, Anatomicund Systema-tik, Zusammenfassungender Vortrdge.66. Wien. Willis, J.C. (1966): A Dictionary oftheFlowering

Plants andFerns, 8th edn. Cambridge Unversity

Press, Cambridge.