Specialized Structures in the Leaf Epidermis of Basal Angiosperms

17 Pages·2006·1.46 MB·English

AmericanJournalofBotany93(5):665–681.2006.
S
PECIALIZED STRUCTURES IN THE LEAF EPIDERMIS OF BASAL
: , , 1
ANGIOSPERMS MORPHOLOGY DISTRIBUTION AND HOMOLOGY
KEVIN J. CARPENTER2
CanadianInstituteforAdvancedResearch,BotanyDepartment,UniversityofBritishColumbia,3529-6270University
Boulevard,Vancouver,BritishColumbiaV6T1Z4Canada
The morphology of specialized structures in the leaf epidermis of 32 species of basal (ANITA: Amborella, Nymphaeales,
Illiciales, Trimeniaceae, and Austrobaileyaceae) angiosperms, representing all seven families and 11 of 14 genera, was
investigatedusinglightandscanningelectronmicroscopy.Distribution,density,andsizeofstructureswerealsomeasured,and
characterevolutionwasanalyzed.HydropotesareasynapomorphyofNymphaealesandetherealoilcellsareasynapomorphyof
Austrobaileyales,butuniseriatenonglandulartrichomesappeartohavearisenindependentlyseveraltimes.Specializedstructures
are frequently characterized by adjacent epidermal cells that have striking similarities in their form and arrangement (i.e.,
architecture)tosubsidiarycellsofcertaintypesofstomatalcomplexes.Additionally,formsintermediatetooilcellsandstomata,
totrichomesandstomata,andtohydropotesandoilcellsarepresentinsometaxa.Thus,allofthesespecializedstructuresand
theiradjacentepidermalcellsformcomplexesthatmaybehomologouswith,andevolutionarilyderivedfromstomatalcomplexes,
andthespecializedstructure,orportionthereof,maybehomologoustothestomaorguardmothercell.Improvedknowledgeof
the morphology and evolution of these structures in the earliest branching extant angiosperm lineages has a bearing on many
diverseareasofbotany.
Key words: Amborellaceae, Austrobaileyales, evolution, hydropotes, leaf epidermal anatomy, Nymphaeales, oil cells,
trichomes.
Since1999and2000,whenseverallarge-scalephylogenetic completedacomparativesurveyofstomatalarchitectureacross
analyses(e.g.,MathewsandDonoghue,1999;Qiuetal.,1999, all ANITA-grade families and Chloranthaceae (Carpenter,
2000; Graham and Olmstead, 2000) placed Amborella 2005). However, the morphology, distribution, and evolution
trichopoda Baill., Nymphaeales, and Austrobaileyales at the of other specialized structures in the leaf epidermis of these
base of the extant angiosperm phylogenetic tree, the renewed plants (e.g., trichomes, ethereal oil cells, and hydropotes—
interestinthesegroupsresultedinnumerousstudiesofvarious specialized trichome-like structures in Nymphaeales) have
aspects of their biology (e.g., Endress and Igersheim, 2000; been little examined or discussed.
Carlquist and Schneider, 2001, 2002; Bernhardt et al., 2003; Becauseplantscommunicatewiththeirexternalenvironment
Feild et al., 2004; Carpenter, 2005). One aspect of ANITA and protect and maintain essential internal physiological and
(acronym of the families and orders within the first three biochemical processes through such specialized epidermal
clades: Amborella, Nymphaeales, Illiciales, Trimeniaceae, and structures, information on their morphology and evolution has
Austrobaileyaceae) angiosperms that has attracted relatively bearing on a wide variety of issues. Aside from their proven
little attention is leaf epidermal anatomy. Carlquist (2001) and value in the systematics and taxonomy of extant and fossil
Baranova (2004) presented brief treatments of Austrobaileya plants (e.g., Stace, 1965; Upchurch, 1984; Baranova, 1992a),
scandens C. T. White, and Amborella trichopoda was specializedepidermalstructuresrepresentadaptationstoawide
summarized by Carlquist and Schneider (2001). I recently range of ecologies (cf. hydropotes and stomatodes of aquatic
plants as in Kaul [1976] and Wilkinson [1979], and trichomes
1Manuscriptreceived4August2005;revisionaccepted20February2006. inxerophyticplantsasinEhleringerandClark[1987])andare
The author thanks G. Vermeij for guidance during the course of this ofpracticalinterestinagriculturebecauseoftheirinfluenceon
projectandforcriticalreviewofthemanuscript;D.PotterandJ.Jernstedt the uptake of pesticides and fertilizers and their role in host–
for the same, as well as for the use of laboratory and photomicroscope parasite interactions (e.g., Harr et al., 1991; Harr and
facilities, respectively; G. Upchurch for instruction on specimen Guggenheim, 1995). Recent workers interested in the
preparation;andD.Canington,B.Ertter,H.Gang,R.Harris,J.Henrich, molecular basis of plant development have been attracted to
D. Lorence, C. Prychid, M. Romanova, P. Romanov, P. Rudall, R.
the leaf epidermis and its specialized structures such as
Saunders,andthefollowinginstitutionsformaterialandotherassistance:
trichomes, as a system offering many advantages and
ConservatoryofFlowers(SanFrancisco,California,USA);RoyalBotanic
interesting questions for study (e.g., Ramsay and Glover,
Gardens:theJodrellLaboratoryandMicromorphologygroup(Kew,UK);
National Tropical Botanical Garden (McBryde, Kalaheo, Hawaii, USA); 2005).
SouthChinaBotanicGardenandIBSC(Guangzhou,China);Universityof The potential importance of specialized leaf epidermal
California: Berkeley Botanic Gardens, UC, DAV, Davis Botanical structures in ANITA angiosperms in particular has been
Conservatory, and Santa Cruz Arboretum; and HKU. Research was suggested by the few studies on such structures in these taxa
supported by Jastro Shields Graduate Research Grants, the Davis
prior to the formation of the ANITA hypothesis in 1999 and
BotanicalSociety,theCenterforBiosystematics,andaUCDDissertation
2000. The presence of ethereal oil cells in leaves and other
Year Fellowship. This represents a portion of a doctoral dissertation
organsofAustrobaileyawasconsideredbyBaileyandSwamy
carriedoutinthePlantBiologyGraduateGroup,UniversityofCalifornia,
Davis. (1949)tobeamajorlineofevidenceagainstarelationshipwith
2 Author for correspondence (e-mail: [email protected], Dilleniaceae.Bailey andSwamy(1948) considered thelack of
[email protected]) ethereal oil cells in Amborella to be an important character
665
666 AMERICAN JOURNAL OF BOTANY [Vol. 93
separating it from Monimiaceae. The presence of ethereal oil removed under a dissecting microscope, leaving only the epidermes to be
cellsintheleafepidermisinparticular,ledBaranova(1992a)to mounted.SpecimenswereexaminedandphotographedwithanOlympusBH-2
argue against a relationship between Austrobaileya and lightmicroscope(Tokyo,Japan)andMicrofiredigitalcamera(Tokyo,Japan).
Leaf samples of Amborellaceae and Austrobaileyales were prepared for
Annononaceae, or Myristicaceae, in favor of a relationship
scanningelectronmicroscopybysamplingpiecesasdescribed.Sampleswere
with Schisandraceae (as supported by Qiu et al., 1999, 2000).
placedin1500-lLplasticreactiontubes,immersedin100%chloroform,and
The increasing diversity of leaf epidermal secretory cells, hair placedinawaterbathsonicatorfor10min.toremoveepicuticularwaxesand
bases, and other leaf characters seen through the Early otherdebris.Sampleswerethenrinsedseveraltimesindeionizedwater,andset
Cretaceous,ledUpchurch(1984)toconcludethatangiosperms todryovernightonpapertowelscoveredbyinvertedbeakers.Leafsamplesof
were undergoing a major adaptive radiation at that time. Yet Victoria amazonica were immersed in 10% Tween overnight, rinsed in
deionizedwater,sonicatedinchloroformasbefore,rinsedindeionizedwater,
despite these intriguing findings, these structures have been
dehydratedinanethanolseries,andcriticalpointdriedinaTousimisSamdri-
mentioned sporadically, and even less frequently illustrated in
780Acriticalpointdryer(Rockville,Maryland,UnitedStates).Sampleswere
the literature on ANITA taxa. Some authors who mention
then mounted onto aluminum stubs with Ted Pella double stick adhesive
ethereal oil cells in the leaf mesophyll of a given species or (Redding,California,UnitedStates),sputtercoatedwithgoldwithaPelcoAuto
familyfailtomentiontheiroccurrenceintheleafepidermis,as Sputter Coater SC-7 (Redding, California, United States), and examined and
in Metcalfe’s (1987) treatment of Austrobaileya, Philipson’s photographedat10kVinaPhilipsXL30TMPscanningelectronmicroscope
(1993)treatmentofTrimeniaceae,andKeng’s(1993)treatment (Eindhoven,Netherlands).
of Schisandraceae. With the current strongly supported and
widelyacceptedrootingoftheextantangiospermtree,Ibelieve Character coding—For each sample (i.e., leaf), 20 of each type of
specializedstructurepresentoneachsurface(abaxialandadaxial)werecoded
thatacomparativestudyofthesestructuresfromalargesample
forcharactersenumeratedintheResults(Characters)section.Whilesomeof
representing all ANITA families will yield data of taxonomic
thesecharactershavebeenpreviouslymentionedintheliterature(e.g.,cuticular
and systematic value for studies of living and fossil striationson etherealoil cells and adjacent epidermalcells; Upchurch,1984;
angiosperms, as well as hypotheses pertaining to many Baranova 1992a), many have not, and I noted additional characters for each
questions about their taxonomic distribution, evolution, and typeofspecializedstructureaftercarefulexaminationofallthetaxaincluded
homology. here. Densities of structures on abaxial and adaxial surfaces were also
calculated.Aminorityofspecimenshaveaverylowfrequencyofspecialized
Inthisstudy,Iexaminedthemorphologyofspecializedleaf
structures,and for these,fewer than 20 structures wereexamined.Structures
epidermal structures in mature leaves of 32 species of ANITA
were measured in Adobe Photoshop Elements 2.0. Distances in pixels were
angiosperms, representing all seven families and 11 of the 14
obtainedandconvertedtomicrometers.
genera. Structures inmostofthesespecieswereexamined and
illustratedforthefirsttime.Iusedtheresultstoaddressseveral Analysisofcharacterevolution—Characterevolutionwasexaminedusing
important questions including (1) What characters can be MacClade 3.08 software (Maddison and Maddison, 1999) by mapping
observed in the various structures, and how do these vary characters onto the basal portion of the most parsimonious tree obtained by
across the taxa? (2) What is the taxonomic distribution of the DoyleandEndress(2000),whichincludesChloranthaceaeasthenextlineage
structures?Arepreviousreportsconcerningthelackofoilcells to diverge above the ANITA grade. Within this framework, topologies of
NymphaealesandChloranthaceaewereexpandedaccordingtotheanalysesof
in Amborella trichopoda (e.g., Bailey and Nast, 1948)
Les et al. (1999) and Qiu et al. (1999, 2000) respectively. Characters were
confirmed?(3)Whatistheancestralstateofangiospermswith
assumed to be unordered, and the most parsimonious reconstruction/node
regard to presence of structures and their character states? (4) option was used. I coded Chloranthaceae as lacking oil cells in the leaf
HowhavethesestructuresevolvedwithintheANITAtaxa?(5) epidermis(seeDiscussion,Characterevolution).
What does the evidence suggest about the homologies of the For analysis of the evolution of trichomes, the species in the genera of
structures?Areleaf epidermal oilcellshomologouswith those Schisandraceae (i.e., Kadsura and Schisandra) that I examined are glabrous,
in the leaf mesophyll and elsewhere in the plant? Is there butbothgeneraareknowntohavespecieswithpubescence(Saunders,1998,
2000).IincludedthesepubescentspeciesintheanalysisalongwiththeonesI
evidence that some or all of these specialized epidermal
examined and ordered relationships according to the topologies obtained by
structures might be homologous? (6) How do extant basal
Saunders (1998, 2000). Chloranthaceae are coded as questionable. Metcalfe
angiospermscomparetoearlyCretaceousangiospermfossilsin (1987)mentionedthatmembersofthefamilyareglabrous,butEklundetal.
termsofthesestructures?Whatdoesthisimplyaboutancestral (2004) noted hairs in scattered species of the family. In Trimeniaceae, the
character states and the taxonomic affinities of these fossils? specimen of Trimenia weinmanniaefolia that I examined had trichomes, but
Metcalfe(1987)mentionedthatsomeotherspecieslackthese;henceIplacedT.
weinmanniaefoliaassistertoacladeofglabrousTrimeniaceaetodepictthis
MATERIALS AND METHODS
visually,althoughthishadthesameeffectascodingthefamilyasquestionable.
Taxonomicsamplingandspecimenpreparation—Tomaximizephyloge-
netic diversity sampled within ANITA taxa, sampling was guided by recent RESULTS
phylogenetic analyses of basal clades and other pertinent literature including
Haoetal.(2000),Lesetal.(1999),Saunders(1998,2000),andSmith(1947). Specializedstructureswereobservedintheleafepidermisof
Materialwasobtainedfromherbaria,aswellasfromplantsgrowinginbotanic all seven ANITA grade families and in 11 of the 14 genera
gardensandthewild.CollectionandvoucherdataaregivenintheAppendix. examined here (Table 1). Uniseriate, nonglandular trichomes
Leafclearingsweremadebyexcisingseveralsamplepiecesofroughly0.25
and/or their associated abscission scars and foot cells were
cm2 from near the midblade. A single mature leaf was sampled for all taxa
found in Amborellaceae (Figs. 5–12) and Trimeniaceae (Figs.
examinedexceptforAmborellatrichopodaandAustrobaileyascandens,which
wereeachrepresentedbytwoleavesfromeachoftwoindividuals(i.e.,four 36–39).Etherealoilcellswereobservedinthefourfamiliesof
leaves for each species). Samples were then immersed in two consecutive Austrobaileyales (Austrobaileyaceae, Trimeniaceae, Schisan-
treatments of 5% KOH (12–24 h per treatment), rinsed in deionized water, draceae,andIlliciaceae;Figs.33–35,40–52).Hydropoteswere
treated for a few minutes in glacial acetic acid, and cleared in bleach (6% observed in Nymphaeaceae (Figs. 17–32). These consist of
sodium hypochlorite). Following clearing, samples were dehydrated in an
a unicellular or multicellular uniseriate hairlike portion that is
ethanolseries,stainedin1%safraninO(in100%ethanol)foraminimumof3
abscisedatmaturity,leavingabaseofthreeorfourspecialized
d, and mounted onto microscope slides in Bioquip’s Euparal (Rancho
Dominguez,California,UnitedStates).Wherepossible,mesophylltissuewas epidermalcellssetinsideoneanother(Fig.32).Mucilagehairs
May 2006] CARPENTER—LEAF EPIDERMAL STRUCTURES IN BASAL ANGIOSPERMS 667
TABLE1. Distributionofspecializedleafepidermalstructuresintaxaofbasalangiosperms.States:0¼absent;1¼present;?¼uncertain.
Taxonandspecimen Trichomecomplexes Hydropotecomplexes Oilcellcomplexes Cuticularstriations
AmborellatrichopodaK.J.Carpenter11,Leaf1 0 0 0 0
AmborellatrichopodaK.J.Carpenter11,Leaf2 0 0 0 0
AmborellatrichopodaK.J.Carpenter27,Leaf1 1 0 0 0
AmborellatrichopodaK.J.Carpenter27,Leaf2 1 0 0 0
Braseniaschreberi 0 1 0 0
Nupharadvena 0 1 0 0
Nupharluteum 0 1 0 0
Nupharpolysepalum 0 1 0 0
Nymphaeacaerulea 0 1 0 0
Nymphaeaflava 0 1 0 0
Nymphaeanouchali 0 1 0 0
Euryaleferox 0 1 0 0
Victoriaamazonica 0 1 0 0
Victoriacruziana 0 1 0 0
AustrobaileyascandensK.J.Carpenter12,Leaf1 0 0 1 1
AustrobaileyascandensK.J.Carpenter12,Leaf2 0 0 1 1
AustrobaileyascandensK.J.Carpenter42,Leaf1 0 0 1 1
AustrobaileyascandensK.J.Carpenter42,Leaf2 0 0 1 1
Trimeniaweinmanniaefolia 1 0 1 1
Schisandrachinensis 0 0 1 1
Schisandragrandiflora 0 0 1 1
Schisandraincarnata 0 0 1 1
Schisandralongipes 0 0 1 0
Schisandrarubriflora 0 0 1 1
Schisandrasphenanthera 0 0 1 0
Kadsuraborneensis 0 0 1 1
Kadsuracoccinea 0 0 1 0
Kadsuraheteroclita 0 0 1 1
Kadsuraoblongifolia 0 0 1 1
Kadsurascandens 0 0 1 1
Illiciumangustisepalum 0 0 1 0
Illiciumdunnianum 0 0 1 1
Illiciumfloridanum 0 0 1 1
Illiciumhenryi 0 0 1 0
Illiciumlanceolatum 0 0 1 0
Illiciumparviflorum 0 0 1 1
Illiciumsimonsii 0 0 1 1
Illiciumverum 0 0 1 1
Chloranthaceaea ? 0 0 1
aSeeMaterialsandMethods,AnalysisofcharacterevolutionandDiscussion,CharacterevolutionforexplanationsofcodingforChloranthaceae.
(consideredhereasatypeofhydropote—seeDiscussion)were types do not entail such a distinction, and some of these terms
found in Cabombaceae (Fig. 13–15). areusedhere.Examplesincludethestephanocytictype,which
comprises a rosette of subsidiary cells (Baranova, 1987), the
Characters—The four specialized leaf epidermal structures similaractinocytictype,whichhasarosetteofsubsidiarycells
examinedhereare similar inconstruction and six of thecoded marked by radial elongation (Metcalfe and Chalk, 1950;
charactersarecommontoallofthem.Eachtypeofspecialized Wilkinson, 1979), and the anomocytic type, which lacks
structure also has additional characters that are specific to it. subsidiary cells altogether (Metcalfe and Chalk, 1950).
Some of the observed character states are mentioned later in Stephanocytic, actinocytic, and other related forms (i.e.,
this section and in the Discussion, but a complete listing is comprising rosettes of subsidiary cells) that I previously
given in Table 2. defined (Carpenter, 2005) are included in the stephanocytic
category.Thosecomplexeslackingsubsidiariesareincludedin
Characters common to all specialized epidermal struc- theanomocyticcategory.Complexeswithoneormorestrongly
tures—(1) Complex type. The epidermal cells bordering the specializedcells(i.e.,anatomicallyspecializedcells,whichare
specialized structure frequently take the form of recognized notably different from adjacent nonspecialized cells in size,
stomatalcomplextypes,soIusedasubsetofstomatalcomplex wall contour, or some other feature), even those that may
terminology to describe these patterns. Because a distinction resemble a paracytic or other such complex, are consigned to
between stomatal poles (i.e., the two areas where the guard the irregular category. The complex types associated with
cells contact) and the lateral face of the stoma (sides of the specialized structures are compared to those of stomatal
guardcells)underliesthedefinitionsofmanystomatalcomplex complexes within each taxon in Table 3, but it is important
types(e.g.,paracytic,diacytic,laterocytic),itisnotpossibleto to note that stomatal complexes with strongly specialized
apply such terms to epidermal cells bordering specialized subsidiarycellssuchaslaterocyticandparacyticareconsidered
structures, because these have no definable polar or lateral here as irregular to facilitate comparison to complexes
regions. However, the definitions of some stomatal complex associated with specialized leaf epidermal structures where
668 AMERICAN JOURNAL OF BOTANY [Vol. 93
measured at its widest point, which also corresponds to the
diameteroftheupperportionoftheoilcellthatemergesatthe
leafsurface.(9)Cell shape. Theshape oftheoil cell(oinFig.
1)wasrecorded.(10)Celldiameter.Thediameteroftheoilcell
was measured at its widest point, which lies below the
epidermal surface (see Fig. 43).
Forhydropotes,characters7and8areequivalenttothethose
for oil cells; hydropotes likewise have a ‘‘base’’ formed by the
outline of the epidermal cells (b in Fig. 4) that surrounds
a ‘‘bowl-shaped’’ cell (see Results, Nymphaeales), which is
levelwiththeepidermis.Characters9and10aretheshapeand
diameter,respectively,ofthe‘‘bowl-shaped’’cell(BcinFig.4).
For the mucilage hairs of Cabombaceae, characters 7 and 8
are shape and diameter, respectively, of the foot cell (f in Fig.
3)—the specialized cell that is level with the epidermis and
uponwhichthedisk-shapedcells(seecharacters9and10)rest.
Figs. 1–4. Illustrations of specialized leaf epidermal structures and
Characters9and10areshapeanddiameter,respectively,ofthe
their characters in surface view. 1. Ethereal oil cell complex typical of
disk-shaped cell (d in Fig. 3)—the cell to which the hair was
thosein Austrobaileyales comprisingoilcell(o),depicted withadashed
attached.
line to indicate that the majority of the cell lies below the epidermis, its
nucleus(n),base(b)formedbyanticlinalcontactcellwalls,andcuticular For the nonglandular hairs of Amborellaceae and Trimenia-
striations (s). A radial wall (r) and tangential wall (t) are indicated. 2. ceae, characters 7 and 8 are shape and diameter, respectively,
Trichome complex typical of Amborellaceae and Trimeniaceae showing ofthefootcell,i.e.,thecelltowhichthehairisattached—fin
abscission scar(a), foot cell towhich thetrichome was attached (f), and Fig. 2; Characters 9 and 10, shape and diameter, respectively,
astronglyspecializedcontactcell(Sc).3.Mucilagehaircomplextypical of the trichome abscission scar (a in Fig. 2), refer to the place
ofCabombaceaewithtwodisk-shapedcells(d)towhichthemucilagehair wherethetrichomeabscisesandaremoreorlessequivalentto
is attached, and a foot cell (f), level with the epidermis, upon which the
the cross-sectional shape and diameter of the hair itself at or
disk-shaped cells rest. 4. Hydropote complex typical of Nymphaeaceae
near its base.
withbase(b)formedbyanticlinalcontactcellwalls,thelens-shapedcell
(L),andthebowl-shapedcell(Bc).Insurfaceview,theBcoftenappears
asadarkringsurroundingtheL.Asubepidermalfootcell(f)liesbeneath Results by taxon—Amborellaceae—Small (less than 150
theBcandL. lminlength),uniseriate,nonglandulartrichomescomposedof
one to three cells coated by a thick cuticle were observed on
both leaves sampled of specimen K.J. Carpenter 27 (Figs. 5–
terms such as paracytic and laterocytic cannot be applied. (2) 12). Both leaves had a sparse pubescence on the adaxial
Complexdiameter.Thediameterofthecomplexwasmeasured surface, while only one had a few trichomes on the abaxial
atitswidestpoint.(3)Numberofcontactcells.Acontactcellis surface. A minority of trichomes were fully intact but most
defined by Upchurch (1984) as any epidermal cell, whether were broken at various places, usually somewhere near the
specializedornot,thatcontactsthestoma.Iusedthistermhere base (Fig. 8). The epidermal cells surrounding the foot cell
similarly to denote any epidermal cell in contact with the often conform to stephanocytic (Figs. 7, 10) or irregular
specialized structure (or one of the cells—see Figs. 1–4). (4) patterns (Figs. 6, 11), both of which are more common than
Numberofstronglyspecializedcontactcells.Contactcellswith anomocytic.Bothleavesoftheotherindividual,specimenK.J.
strong anatomical specialization per complex were counted Carpenter 11, were entirely glabrous. Ethereal oil cells were
(Sc, in Fig. 2). Complexes with one or more strongly not observed in the leaf epidermis of either specimen.
specialized cells are usually consigned to the irregular type
(Character 1), but a minority of stephanocytic types also may Nymphaeales—Floating leaves of 10 species representing
have strongly specialized cells (e.g., Fig. 10). (5) Presence or five of the eight genera of both families in this order were
absence of cuticular striations. Many taxa have striations investigated. All had a dense covering of hydropotes on the
radiatingfromthespecializedcellandextendingoveradjacent abaxial surface (e.g., Figs. 25, 29). The hairlike portion of the
contact cells (e.g., Figs. 33, 34, 42, 45, 47). (6) Presence or hydropote was lacking in the majority of specimens; most had
absenceofanucleus.Somespecializedcellshaveanucleusor only the basal portions of these structures. In specimens with
its remnant (n in Fig. 1; Figs. 41, 45, 48, 52). hairlikeportionsremaining,thenumberofcellsrangedfrom3–
5inEuryaleferoxandVictoriaamazonica(Fig.25)toadozen
Characters specific to particular specialized epidermal or more in the other genera. In leaf transverse sections, the
structures—Because the first six characters are common to hydropote bases of V. amazonica have an outermost cell
all specialized leaf epidermal structures, I continued the shaped like a biconvex lens, set inside a larger bowl-shaped
numbering of characters at seven for the additional characters cell (Fig. 32), which completely encircles it in surface view
specific to each type of structure. This was done for (Fig. 29). This in turn is supported by a rectangular sub-
convenience and does not imply that characters numbered epidermal ‘‘foot’’ cell (terminology from Lu¨ttge and Krapf,
seven and higher are necessarily homologous across these 1969 and Wilkinson, 1979). The lens-shaped cell and bowl-
different structures. shapedcelloftenstainmoredarklythanadjacentcells,andthe
Specific characters for ethereal oil cells include (7) Base former frequently appears to contain small structures, possibly
shape. The base is formed by anticlinal walls of abutting vesicles or crystals.
epidermalcellsthatsurroundtheupperportionoftheoilcell(b Victoria amazonica and E. ferox had hydropote-like
in Fig. 1). (8) Base diameter. The diameter of the base was structures regularly distributed over their adaxial surfaces
May 2006] CARPENTER—LEAF EPIDERMAL STRUCTURES IN BASAL ANGIOSPERMS 669
Figs.5–12. Light(LM)andscanningelectronmicrographs(SEM)ofuniseriate,nonglandularleaftrichomesandassociatedabscissionscars,footcells,
andcontactcellsinAmborellatrichopoda,K.J.Carpenter27.LMareofclearedleaves.5.Abaxial,tricellular,uniseriatetrichome,Leaf1.6.Adaxial
trichome complex with irregular architecture, Leaf 2. Asteriks denote two strongly specialized subsidiary cells. 7. Adaxial trichome complex with
actinocyticarchitecture,Leaf1.8.SEMofadaxialtrichomefootcellwithtrichomeabscissionscar,Leaf3.9.SEMofadaxialtrichomeandportionofits
basecell.Leaf3.10.Trichomecomplexwithactinocyticarchitecturewithonestronglyspecializedsubsidiary(asterik)formedbytangentialdivision,Leaf
2. 11. Adaxial trichome complex of irregular architecture, Leaf 2. Asteriks denote two strongly specialized subsidiary cells. 12. Adaxial, unicellular
trichomewithitsfootcellandthecontactcells(Leaf2).Bars¼20lm.
(Figs. 24, 26–28, 30, 31), while V. cruziana had only a few complexes in Austrobaileyales and are filled with numerous,
such structures on the adaxial surface. In Victoria, these small prismatic crystals (Fig. 16).
structures differ from abaxial hydropotes in that, in surface
view, they show a highly conspicuous subepidermal complex Austrobaileyales—Representativesofallfamiliesandgenera
ofcloselyappressedcellsthatsometimesresemblesatetragonal of this order were examined, and all were observed to have
pollen tetrad (Fig. 28). Other ones have three (Fig. 31) or two ethereal oil cells in varying densities on the abaxial leaf
tightly appressed cells, while some have only one. Addition- surface; however, the majority lack these on the adaxial
ally, many of these structures appear to have a nucleus in the epidermis. In Austrobaileyaceae, oil cells were found exclu-
lens-shaped cell (Fig. 30). The adaxial structures in Euryale sively on the abaxial epidermis in the four leaf samples (Figs.
were not observed to have nuclei. Nymphaea flava had rare 33–35). Trichomes are lacking. Trimenia weinmannieafolia
structuresontheadaxialsurface that appear tobeintermediate Seem., (Trimeniaceae) had trichome foot cells, abscission
between stomata and oil cells; in these, a large, spheroidal scars,andoilcellcomplexesonbothleafsurfaces.Theoilcell
saclike objectisattached atitsbasetoapoorlydevelopedpair complexes are typical for this order, but the trichome bases
ofguardcells(Fig.19).Unlikeleafepidermaletherealoilcells, differ from those in other ANITA families in that their
however, these saclike structures are positioned entirely above associated epidermal cells have a very high proportion of
the leaf surface. irregulararchitecture.Inthese,contactcellsoftenappearmuch
The mucilage hairs of Brasenia schreberi (Cabombaceae) smaller than other epidermal cells (Fig. 37). The trichome
are structurally more similar to hydropotes of Nymphaeaceae abscission scars (Fig. 39), however, are similar to those in
thantothetrichomesofAmborellaceaeorTrimeniaceae.They Amborellaceae. The two genera of Schisandraceae differed
consist of a small unicellular or bicellular hair attached to two somewhatindistributionofoilcellcomplexes.Onlyoneofthe
disk-shaped cells stacked one atop the other (Figs. 14, 15). six species of Schisandra (S. chinensis) investigated had
Thesetwodisk-shapedcellsinturnrest uponalargerfootcell adaxialoilcellcomplexes,butthesewerefoundinfouroffive
that is level with the other epidermal cells (Fig. 13). The species of Kadsura (Fig. 41). Some species of Kadsura have
arrangement of these three cells as seen in leaf transverse very high densities, and K. oblongifolia Merrill is one of two
sectionissimilartothethree-cellarrangementofhydropotesin ANITA-grade taxa with a higher density of oil cells on the
Nymphaeaceae (cf. Figs. 14 and 32). Also, B. schreberi has adaxial surface than the abaxial surface. The species of
two spherical, thin-walled, oil cell-like structures above the Schisandraceae examined here were all glabrous. The eight
adaxial epidermis. These have bases typical of oil cell speciesofIlliciaceaeexamined werehighlyvariableindensity
670 AMERICAN JOURNAL OF BOTANY [Vol. 93
Figs.13–32. LM(clearedleaves)andSEMofleafhydropotes,mucilagehairs,andunidentifiedoilcell-likestructuresinNymphaeales.13.Abaxial
stephanocyticmucilagehaircomplexinBraseniascherberi(Cabombaceae).Diskcells(d)andfootcell(f)areindicated.14.AbaxialmucilagehairofB.
schreberiintransversesection,twodiskcells(d)atopafootcell.15.Abaxialbicellularmucilagehair,B.schreberi.16.Adaxialoil-cell-likestructure,B.
schreberi.Notethecircularoutlineofthecellandnumeroussmallprismaticcrystalswithin;itsbaseissimilartothoseinetherealoilcellcomplexesin
May 2006] CARPENTER—LEAF EPIDERMAL STRUCTURES IN BASAL ANGIOSPERMS 671
of abaxial oil cell complexes. Only Illicium parviflorum had Austrobaileyales are absent in the nymphaealean taxa I
them on the adaxial surface (Fig. 50), where they occurred in examined, there are intriguing structures in a few species that
greater density than on the abaxial surface. Illiciaceae have do not conform to anything previously reported. In V.
amuchhigherfrequencyofundulatebases(withaconcavityin amazonica, Wilkinson (1979, p. 165) noted ‘‘a few’’ hydro-
thewallsoftheepidermalcellsformingthebase;Fig.49),and potesontheadaxialsurface,butinthesampleIexamined,the
typically, larger oil cells than any other ANITA grade family. hydropote-like structures on the adaxial surface appear as
All investigated species of Illiciaceae are glabrous. In general, a regular feature; they are more or less evenly spaced and are
Illiciaceae have larger oil cells and bases than Schisandraceae, present in all four slides I prepared. These differ from abaxial
while those in both families generally exceed those in hydropotes in their unusual subepidermal complexes (see
Trimeniaceae. Austrobaileyaceae have the smallest oil cells Results, Results by taxon) and in the presence of nuclei in the
and bases in the order. lens-shapedcellofsome,acharacterfoundinsomeoilcellsof
Austrobaileyales.
Character evolution—Analysis of character evolution in- Perhaps even more intriguing are the rare adaxial structures
dicatedthatleafepidermaletherealoilcellsareasynapomorphy of Nymphaea flava that appear intermediate between stomata
of Austrobaileyales (Fig. 60) and that hydropotes are and oil cells in having a large saclike structure attached to
a synapomorphy of Nymphaeales (Fig. 61). Trichomes are a pair of poorly developed guard cells (Fig. 19). Brasenia
reconstructed as absent in the common ancestor of Nymphae- schreberi likewise has rare structures on the adaxial surface
laes and in the common ancestor of Austrobaileyales, but are that bear some similarity to oil cell complexes in Austro-
equivocalinthecommonancestorofallangiospermsandafew baileyales (Fig. 16). Goleniewska-Furmanowa (1970, p. 24)
other internal nodes (Fig. 62). The common ancestor of all mentioned that the mucilage hairs of this species may vary
angiosperms is thus reconstructed as lacking hydropotes and widely andinclude‘‘low glandularhairs with a bicellularstalk
ethereal oil cells, although the presence of trichomes is andunicellularhead’’.Sheillustratedarangeofhairs(depicted
unresolved. The evolution of most of the characters coded on the abaxial surface), including one with a globose,
for these structures was not traced due to extensive variability unicellular head supported by two flattened, disk-shaped cells
within genera and for other reasons explained in the sitting one atop the other. Despite this similarity in cell shape,
Discussion, Character evolution. the structures I observed differ in occurring on the adaxial
surface and in containing numerous tiny prismatic crystals (as
DISCUSSION
are common in Nymphaeaceae), which Goleniewska-Furma-
Morphology and distribution—Evidence presented here nowa (1970) commented were absent in Cabombaceae. I was
adds to that from a growing list of studies in which oil cells not able to observe these structures in leaf transverse sections
werefoundtobelackinginAmborellaceae(BaileyandSwamy, andthuswasnotable toconfirm thepresenceofthetwodisk-
1948; Money, et al., 1950; Carlquist and Schneider, 2001), shaped base cells, but in surface view, the base appears quite
countertotheearlierclaimoftheirpresencebyPerkins(1898). similar to the polygonal to elliptical bases seen in oil cell
The variability in distribution and density of trichomes in the complexes of Austrobaileyales. Even supposing that these
four samples examined here also accords with Bailey and structuresarerareadaxial,mucilagehairs,Ifindtheirsimilarity
Swamy’s (1948) observations that the presence and density of to oil cell complexes in Austrobaileyales to be nevertheless
hairs and other leaf characters (e.g., size, form) often vary notable and intriguing.
greatly, even among leaves of the same plant. The hairs InAustrobaileyales,morphologyanddistributionofethereal
observed here are similar to the ones noted by Bailey and oil cells largely conform to previously published descriptions
Swamy (1948, p. 248), except that the ‘‘protuberant, multicel- and illustrations (e.g., Bailey and Nast, 1948; Bailey and
lular pedestal[s]’’they noted in some were not found here, nor Swamy, 1948; Upchurch, 1984; Baranova, 1992a).
were other specialized forms such as the ‘‘tufted, fan-shaped,
stellate,orpeltateforms’’notedbyMoneyetal.(1950,p.374). Character evolution—Although the analysis of character
Oil cells have not been noted in Nymphaeales in any of evolution resolved the common ancestor of all extant
a number of works that have addressed leaf anatomy (e.g., angiosperms as lacking leaf epidermal ethereal oil cells and
Metcalfe and Chalk, 1950; Goleniewska-Furmanowa, 1970; hydropotes,therearetwopointsthatmustbeconsideredbefore
Kaul, 1976; Williamson et al., 1989; Schneider and adoptingthisreconstruction.First,sinceAmborellatrichopoda
Williamson, 1993; Williamson and Schneider, 1993; Les et islikelytheproductofoneoftheoldestlineage-splittingevents
al., 1999). While oil cell complexes of the type found in in extant angiosperms, is the sole representative of its family,

Austrobaileyales. 17. Abaxial hydropote complex with stephanocytic architecture, Nuphar advena. 18. Abaxial hydropote complex with irregular
architecturewithonestronglyspecializedsubsidiary,N.polysepalum.19.Adaxialoil-cell-likestructures,Nymphaeaflava.Notetheellipticoutlineofthe
cell(arrows),whichisattachedtoapairofguardcells.20.Abaxialactinocytichydropotecomplex,N.nouchali.21.AbaxialhydropotecomplexesinN.
flava;upperis actinocytic, lowerhasirregular architecture (onestrongly specialized subsidiary). 22.Twoabaxial actinocytichydropote complexes, N.
caerulea. 23. Abaxial actinocytic hydropote complex, Euryale ferox. 24. Adaxial hydropote complex with irregular architecture and one strongly
specialized subsidiary, E. ferox. Note the surrounding stomata. 25. SEM of abaxial surface of Victoria amazonica showing many hydropotes without
hairlike portions (arrows), and a few with these still attached. 26–28. Different focal planes of one adaxial hydropote complex (weakly actinocytic
architecture)ofV.amazonica.26.Topoflens-shapedandbowl-shapedcells.27.Middleofthehydropotewithcruciformpatternofsubepidermalcomplex
visible.28. Lowerplaneshowing 4-celled subepidermal cellcomplex.29.SEM ofabaxial surfaceofV.amazonica withdetail ofthebasal portionof
hydropotecomplexes.30.AdaxialhydropoteofV.amazonicashowingnucleus.31.A3-celledsubepidermalcellcomplexofanadaxialhydropote,V.
amazonica.32.Abaxialhydropote(transversesection),V.amazonica.L¼lens-shapedcell,B¼bowl-shapedcell,F¼footcell.Bars¼20lm.
672 AMERICAN JOURNAL OF BOTANY [Vol. 93
TABLE 2. Observed character states and means, standard deviations, or ranges (in parentheses) for characters pertaining to specialized leaf epidermal
structures in ANITA angiosperms. Standard deviations for density and characters 5 and 6 are not given when only one sample was examined.
Characters7–10varyaccordingtostructureandareexplainedintheResults,CharactersandinFigs.1–4.Dashesindicatethatstructuresareabsent.
Questionmarksindicatethattheidentityofthestructureisuncertain.Abbreviations:A¼anomocytic;Ab¼Abaxial;Ad¼Adaxial;C¼circular;CCs
¼contactcells;E¼elliptical;Ep¼epidermis;I¼irregular;O¼ovate;P¼polygonal;S¼stephanocytic;Scs¼stronglyspecializedcontactcells;T¼
triangular;U¼undulate.Adashbetweenabbreviationsindicatesarangeofmorphologiesgradingbetweenthetwostates.Characterstatesindicated
byabbreviationsarelistedinorderofdecreasingfrequency.Rareindicatesthatfiveorfewerofthestructurewereobservedthroughoutthesample.
1and2
Complextype 7and8: 9and10:
Structure Density anddiameter 3and4:No.CCs/ 6:Nucleus Shapeanddiameter Shapeanddiameter
Genus Ep. (No.sampleswithstructure) (permm2) (lm) Scspercomplex 5:Striae(%)a (%)a (lm)b (lm)c
Amborella Ad Trichomes(2of4) 1.260.5 S,I,A 7.6(6–10) 060 060 P–E,T C–E
152.4621.9 0.9(0–7) 54.263.8 21.262.9
Ab Trichomes(1of4) Rare S,A 7.5(7–8) 0 0 T C
123.161.4 0(0–0) 38.467.5 16.766.5
Brasenia Ad Oilcell-likestructures? Rare A,S 5.5(5–6) 0 0 P–E P–E
(1of1) 64.7617.1 0(0–0) 18.765.8 30.465.9
Ab Mucilagehairs 347.4 A,S,I 6.4(4–9) 0 0 P–E C–E
(1of1) 131.4621.2 0.1(0–1) 39.965.1 19.562.7
Nuphar Ad — — — — — — — —
Ab Hydropotes(3of3) 148.5666.6 A,I,S 4.8(3–8) 060 1.762.9 C–E P–E,C–E
103.869.9 0.3(0–2) 17.960.4 28.662.0
Nymphaea Ad — — — — — — — —
Ab Hydropotes(3of3) 188.3651.7 S,A,I 6.7(5–9) 060 060 P–E,C–E,O P–E,C–E,O,I
108.3623.3 0.4(0–6) 19.162.6 24.363.0
Euryale Ad Hydropotes(1of1) 0.76 S,A,I 7.6(6–11) 0 0 P–E P–E
63.367.5 0.2(0–1) 12.461.5 16.661.2
Ab Hydropotes(1of1) 479.9 S,I,A 7.2(4–10) 0 0 P–E P–E
78.369.4 0.8(0–2) 20.861.8 24.861.4
Victoria Ad Hydropotes(2of2) 2.3d S,I,A 8.7(8–11) 060 40656.6 P–E P–E
104.5619.3 0.2(0–2) 21.460.1 26.962.1
Ab Hydropotes(2of2) 145.8658.1 S,A,I 7.2(5–9) 060 060 P–E,C–E,O P–E,C–E,O,I
129.8616.0 0.1(0–1) 26.063.4 31.963.8
Austrobaileya Ad — — — — — — — —
Ab Oilcells (4of4) 0.460.2 S,I,A 6.5(5–8) 10060 15.0619.1 P–E P–E,I
162.866.8 0.7(0–4) 25.761.5 34.262.2
Trimenia Ad Oilcells(1of1) 1.76 A,I,S 4.9(4–6) 95.0 30.0 P–E P–E,O
190.3628.9 0.5(0–5) 27.963.8 48.664.2
Ad Trichomes(1of1) 1.76 I,S 5.7(4–7) 70.0 0 P–E,O C–E
150.5625.0 3.5(1–6) 55.464.2 19.361.3
Ab Oilcells(1of1) 6.03 I,A,S 5.7(4–7) 5.0 15.0 P–E,U P–E
168.5622.0 0.7(0–4) 30.463.7 42.763.4
Ab Trichomes(1of1) 4.02 I,A 6.6(5–9) 20.0 0 C–E,P,O C–E
170.8638.6 1.6(0–5) 47.264.4 17.762.2
Schisandra Ad Oilcells(1of6) 0.54 I,S 6.0(5–7) 100.0 67.0 P–E,I P–E
191.6626.7 2.0(0–4) 36.364.2 44.163
Ab Oilcells(6of6) 2.861.9 A,I,S 6.3(4–9) 49.2643.8 34.2629.7 P–E,I,U,O P–E,I,O
211.5639.7 0.5(0–4) 37.363.1 44.062.0
Kadsura Ad Oilcells(4of5) 5.965.4 I,S,A 5.5(4–9) 58.3644.0 75.56 P–E,U,I P–E,O,I
158.8618.9 1.1(0–5) 39.2 35.565.0 52.168.3
Ab Oilcells(5of5) 5.963.0 I,A,S 5.5(3–10) 68.0641.5 60.0617.7 P–E,I,U,O P–E,I,O
165.2624.0 0.9(0–6) 37.266.0 49.667.0
Illicium Ad Oilcells(1of8) 6.36 S,A,I 5.7(4–7) 10.0 10.0 U,P–E P–E
134.8612.3 0.4(0–2) 36.564.9 86.065.9
Ab Oilcells(8of8) 5.066.0 A,S,I 6.2(4–9) 45.6648.1 47.5637.0 P–E,U,I P–E,O
163.6640.7 0.5(0–6) 37.963.4 59.1612.1
aPercentageofstructureswithcuticularstriations(striae)orwithnucleus.
b Characters 7 and 8, respectively: for trichomes, shape and diameter of the foot cell; for mucilage hairs, shape and diameter of the foot cell; for
hydropotesandoilcells,shapeanddiameterofthestructure’sbase(seeResultsandFigs.1–4).
cCharacters9and10,respectively:fortrichomes,shapeanddiameteroftheabscissionscar;formucilagehairs,shapeanddiameterofthediskcell;for
oilcells,shapeanddiameteroftheoilcell;forhydropotes,shapeanddiameterofthebowl-shapedcell(seeResultsandFigs.1–4).
dDensityisbasedonlyonVictoriaamazonica;adaxialhydropotesarerareinV.cruziana.
and has a very restricted present-day distribution, it is difficult 2005). Likewise, Nymphaeales have become highly modified
to believe that neither extinction nor phyletic evolution have for an aquatic existence, and a loss of epidermal oil cells may
occurred in Amborellaceae since its origin. Hence, the lack of have occurred alongside, or as a result of, other modifications
leaf epidermal ethereal oil cells may represent a derived in the leaf epidermis. Second, examination of early fossil
condition, just as I inferred that the predominantly paracytic angiospermleavesshowsthattaxawithleafepidermalethereal
stomatal architecture in Amborella is derived (Carpenter, oil cells and/or trichomes appeared early in the history of the
May 2006] CARPENTER—LEAF EPIDERMAL STRUCTURES IN BASAL ANGIOSPERMS 673
TABLE 3. Comparison of the architecture of specialized leaf epidermal angiospermclade,asdiscussedlater(seealsoUpchurch,1984).
complexes with that of stomatal complexes within each taxon. The The occurrence of uniseriate trichomes in scattered species of
mostcommontypeofarchitecturepresentineachstructureisgiven. woody ANITA taxa including Amborella, some Schisandra-
Data for stomatal architecture are taken from Carpenter (2005) and ceae (Saunders, 1998, 2000) and some Trimeniaceae (Results,
refertoabaxialstomatalcomplexes,exceptfortaxaofNymphaeales Results by taxon, and Metcalfe, 1987), as well as in early
(Brasenia,Nuphar,Nymphaea,Euryale,andVictoria)inwhichthey
angiosperm fossil cuticles (Upchurch, 1984) may suggest that
areadaxial.Adashindicatesdataareunavailable.Aslashseparating
the presence of structural genes for such trichomes may be
abbreviations indicates that both types are equally abundant.
ancestral in angiosperms, or at least may have evolved early.
Abbreviations: A ¼ anomocytic; ab ¼ abaxial; ad ¼ adaxial; I ¼
Thesubsequentlossorsuppressionoftrichomesinsomeextant
irregular; S¼stephanocytic. For purposes of comparison, stomatal
types with strong specialization of subsidiary cells are considered ANITA taxa could possibly be explained through the MYB-
‘‘irregular’’ here so as to be compatible with criteria by which bHLH-WD40 protein complex of transcription factors, which
complexes of specialized stuctures were judged. The most common isthoughttooperateinallangiospermsandknowntohavethe
typeofsuch‘‘irregular’’stomatalcomplexesisgiveninparentheses. abilitytoquicklyevolvetochangeexpressionofepidermalcell
types (Larkin et al., 2003; Ramsay and Glover, 2005). As
Taxon(Specimen) Structure:Architecture Stomatalarchitecture
Ramsay and Glover (2005) point out, regulatory genes and
Amborellatrichopoda Trichomes(ad):I I(paracytic) their DNA-binding sequences are known to be able to evolve
(K.J.Carpenter27) Trichomes(ab):A/S more quickly than the structural genes whose expression they
Braseniaschreberi Hydropotes(ab):A S mediate. The differing complement of trichomes in closely
Nupharadvena Hydropotes(ab):A S related species (e.g., as in Schisandraceae) may provide an
Nupharluteum Hydropotes(ab):A A
intriguing system for further investigation of the role of this
Nupharpolysepalum Hydropotes(ab):A S
protein complex in angiosperms as a whole.
Nymphaeacaerulea Hydropotes(ab):S S
Nymphaeaflava Hydropotes(ab):S S Itisalsonotablethatinthisanalysis(i.e.,asbasedonDoyle
Nymphaeanouchali Hydropotes(ab):A/S S andEndress’[2000]topology),leafepidermaletherealoilcells
Euryaleferox Hydropotes(ab):S S are a synapomorphy of Austrobaileyales, rather than Austro-
Victoriaamazonica Hydropotes(ad):S S baileyales þ Other Angiosperms (Fig. 60). Although Chlor-
Hydropotes(ab):S
anthaceae(thenextlineage tobranchaboveANITAaccording
Victoriacruziana Hydropotes(ad):S S
to Doyle and Endress, 2000) are mentioned as having oil cells
Hydropotes(ab):S
Austrobaileyascandens Oilcells(ab):S I(laterocytic) in the leaf mesophyll and other tissues (e.g., Swamy, 1953;
(K.J.Carpenter12,Leaf1) Metcalfe,1987;Eklundetal.,2004),noneofthesestudies,nor
Austrobaileyascandens Oilcells(ab):S I(laterocytic) arecentstudyoftheleafepidermisinthisfamily(Kong,2001),
(K.J.Carpenter12,Leaf2) mention or illustrate oil cells in the leaf epidermis, so I coded
Austrobaileyascandens Oilcells(ab):S I(laterocytic)
Chloranthaceae as lacking these. Additionally, I have not
(K.J.Carpenter42,Leaf1)
observed leaf epidermal oil cells in the family either (personal
Austrobaileyascandens Oilcells(ab):S I(laterocytic)
(K.J.Carpenter42,Leaf2) observation). While oil cells are commonly held to be
Trimeniaweinmanniaefolia Oilcells(ad):A a symplesiomorphy of magnoliid plants (e.g., West, 1969), it
Oilcells(ab):I I(paracytic) appearsthatatleastthoseintheleafepidermishaveoriginated
Trichomes(ad):I (orwerelost)independentlyanumberoftimes.However,ifthe
Trichomes(ab):I
Doyle and Endress (2000) placement of Chloranthaceae is
Schisandrachinensis Oilcells(ad):I I(laterocytic)
rejected, as in the APG II (2003) topology, then it is possible
Oilcells(ab):A
Schisandragrandiflora Oilcells(ab):I I(laterocytic) thatthis interpretation may change. However,withtheAPGII
Schisandraincarnata Oilcells(ab):A I(laterocytic) (2003)topology,untilgreaterresolutionisachieved,especially
Schisandralongipes Oilcells(ab):A — among Chloranthaceae, monocots, and ‘‘magnoliids’’ (Piper-
Schisandrarubriflora Oilcells(ab):A I(laterocytic) ales, Canellales, Laurales, Magnoliales), which currently form
Schisandrasphenanthera Oilcells(ab):A I(laterocytic)
a trichotomy sister to Ceratophyllum þ Eudicots, it is not
Kadsuraborneensis Oilcells(ad):I I(paracytic)
possible to draw any firm conclusions. One possible scenario
Oilcells(ab):I
Kadsuracoccinea Oilcells(ab):A I(laterocytic) thatcouldsupporthomologyofleafepidermaletherealoilcells
Kadsuraheteroclita Oilcells(ad):S I(laterocytic) in Austrobaileyales and magnoliids (as circumscribed by APG
Oilcells(ab):A II,2003)wouldbeifmagnoliidsaresistertoChloranthaceaeþ
Kadsuraoblongifolia Oilcells(ad):I — Monocots,allofwhicharesistertoCeratophyllumþEudicots.
Oilcells(ab):I
Then, it could be hypothesized that leaf epidermal oil cells
Kadsurascandens Oilcells(ad):I I(laterocytic)
Oilcells(ab):A/S arose in the common ancestor of Austrobaileyales þ Other
Illiciumangustisepalum Oilcells(ab):S I(paracytic) Angiosperms, were retained in magnoliids, and lost in
Illiciumdunnianum Oilcells(ab):S I(paracytic) Chloranthaceae, monocots (although Acorus does have oil
Illiciumfloridanum Oilcells(ab):A I(paracytic) cells in the mesophyll (See Doyle and Endress, 2000) and
Illiciumhenryi Oilcells(ab):A I(paracytic) CeratophyllumþEudicots.
Illiciumlanceolatum Oilcells(ab):A I(paracytic)
Ofthecharacterspertainingtothespecializedleafepidermal
Illiciumparviflorum Oilcells(ad):S I(paracytic)
structures I coded, those showing systematic value include
Oilcells(ab):A
Illiciumsimonsii Oilcells(ab):A I(paracytic) adaxialhydropotes,asynapomorphyoftheVictoriaþEuryale
Illiciumverum Oilcells(ab):I I(paracytic) clade (see Les et al., 1999) and perhaps cuticular striations,
which may represent a synapomorphy of Austrobaileyales þ
Other Angiosperms (although some Austrobaileyales lack
these). A high frequency of undulate ethereal oil cell bases
(exceeding all other types of bases) seems a potentially useful
674 AMERICAN JOURNAL OF BOTANY [Vol. 93
Figs.33–52. LM(clearedleaves)andSEMofleafepidermaloilcellandtrichomecomplexesinAustrobaileyales.33.SEMofabaxialetherealoilcell
complex, Austrobaileya scandens, K.J. Carpenter 12, Leaf 3. Note the raised outer portion of the oil cell with deep, radiating striations. 34. Abaxial
etherealoilcellcomplexwithirregulararchitecture(threestronglyspecializedsubsidiaries),A.scandensK.J.Carpenter12,Leaf1.35.Abaxialactinocytic
etherealoilcellcomplex.Notethesmoothcontourofthearcdescribedbythetangentialwallsofthesubsidiarycells.A.scandens,K.J.Carpenter12,