Morphological and molecular phylogenetic context of the angiosperms: contrasting the 'top-down ...

33 Pages·2006·0.63 MB·English

JournalofExperimentalBotany,Vol.57,No.13,pp.3471–3503,2006
MajorThemesinFloweringResearchSpecialIssue
doi:10.1093/jxb/erl128
Morphological and molecular phylogenetic context of the
angiosperms: contrasting the ‘top-down’ and ‘bottom-up’
approaches used to infer the likely characteristics of the
first flowers
Richard M. Bateman1,*, Jason Hilton2 and Paula J. Rudall1 Do
w
n
1 Jodrell Laboratory, Royal Botanic Gardens Kew, Richmond, Surrey TW9 3AB, UK lo
a
d
2 School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, ed
Birmingham B15 2TT, UK fro
m
h
Received4May2006;Accepted13July2006 ttp
s
://a
c
a
d
e
m
Abstract dictatingtheperceivedevolutionarypolarityofcharac- ic
.o
tertransitionsamongtheearly-divergentangiosperms. up
Recentattemptstoaddressthelong-debated‘origin’of These factors still permit a remarkable range of con- .co
theangiospermsdependonaphylogeneticframework m
derived from a matrix of taxa versus characters; most trasting, yet credible, hypotheses regarding the order /jxb
of acquisition of the many phenotypic characters, /a
assumethatempiricalrigourisproportionaltothesize reproductive and vegetative, that distinguish ‘classic’ rtic
of the matrix. Sequence-based genotypic approaches le
angiospermy from ‘classic’ gymnospermy. The flower -a
increase the number of characters (nucleotides and b
remains ill-defined and its mode (or modes) of origin s
indels) in the matrix but are confined to the highly remains hotly disputed; some definitions and hypoth- trac
restricted spectrum of extant species, whereas mor- eses of evolutionary relationships preclude a role for t/57
phology-based approaches increase the number of the flower in delimiting the angiosperms. We advocate /13
phylogenetically informative taxa (including fossils) at /3
maintenance of parallel, reciprocally illuminating pro- 4
7
theexpenseofaccessingonlyarestrictedspectrumof grammes of morphological and molecular phylogeny 1/4
phenotypic characters. The two approaches are cur- reconstruction, respectively supported by homology 77
5
rently delivering strongly contrasting hypotheses of testing through additional taxa (especially fossils) 39
relationship. Most molecular studies indicate that all and evolutionary-developmental genetic studies that by
extantgymnospermsformanaturalgroup,suggesting explore genes potentially responsible for major phe- gu
e
surprisingly early divergence of the lineage that led to notypic transitions. st o
angiosperms, whereas morphology-only phylogenies n
0
indicate that a succession of (mostly extinct) gymno- Key words: Angiosperm, character optimization, congruence, 3 A
sthpiesrmcsonpflreicctedinedcluadlea:ter(ia)ngthioespvearsmt oprhigeinno.tCyapuicsesanodf dfoesvseill,ogpemneentd,upelvicoalutiotionn,agryym-dneovseploeprmm,emntoarlphgoelongeytic,so,ntoflgoewneyr,, pril 2
0
genotypic lacuna, largely reflecting pre-Cenozoic ex- outgroup,phylogeny,pteridophyte,taxonsampling,treerooting. 19
tinctions, that separates early-divergent living angio-
sperms from their closest relatives among the living
gymnosperms; (ii) profound uncertainty regarding Introduction
which (a) extant and (b) extinct angiosperms are
most closely related to gymnosperms; and (iii) pro- ‘An early hope was that the relationships indicated
found uncertainty regarding which (a) extant and (b) among species by DNA were more likely to be correct
extinct gymnosperms are most closely related to thanthosebasedonmorphology;thisnowseemsna¨ıve.’
angiosperms, and thus best serve as ‘outgroups’ (Judd et al., 1999, p. 99)
*Towhomcorrespondenceshouldbeaddressed.E-mail:[email protected]
ªTheAuthor[2006].PublishedbyOxfordUniversityPress[onbehalfoftheSocietyforExperimentalBiology].Allrightsreserved.
ForPermissions,pleasee-mail:[email protected]
3472 Batemanetal.
‘The primary motivating force for preparing this book Table 1. Representative spectrum of definitions of a flower
was the dramatic change in our understanding of
‘...agrowthcomprisingthereproductiveorgansofseedplants’
angiosperm phylogeny during the past 10 years. Many
(ChambersDictionary,1998,p.617)
long-standing[morphological]viewsofdeep-levelrelation- ‘...anaxisonwhichtherestofthefloralorgansareborne’(Fahn,
shipswerealteredsuddenlyandsubstantivelyasadirect 1974,p.425)
‘...[a]specializedreproductiveshoot,consistingofanaxis
result of molecular analyses.’ (Soltis et al., 2005, p. ix)
(receptacle)onwhichareinsertedfourdifferentsortsoforgans
[sepals,petals,stamens,carpels]’(PenguinDictionaryofBiology,
‘Occam’s razor? But that’s for circumcision, surely?’
1971,p.102)
(Tom Sharpe, 1995; Grantchester Grind, London: ‘...anaxis(orreceptacle)bearing,initscompleteform,fourzonesof
Macmillan, p. 82) appendagesthatareconsideredthehomologsofleaves’(Bierhorst,
1971,p.511)
In this review we address one of the most popular ‘...ahighlymodifiedshootbearingspecializedappendages(modified
leaves)’(Juddetal.,1999,p.53)
discussiontopicsinevolutionarybiology:theoriginofthe
‘...anamphisporangiatestrobilusofdeterminategrowthandwithan
flower, and, by implication, the origin of the flowering involucrumofmodifiedbracts’(ArberandParkin,1907) D
o
plants (note that we pay much less attention to the better ‘...thereproductivestructureofAnthophyta,derivedevolutionarily wn
fromaleafyshootinwhichleaveshavebecomemodifiedintopetals, lo
documented subsequent, family-level radiation of the sepalsandintothecarpelsandstamensinwhichthegametesare ad
angiosperms; cf. Friis et al., 2006). We do not attempt to formed’(DictionaryofBiologicalTerms,1995;Harlow:Longman, ed
offer definitive answers on the fundamental nature of the p.206) fro
‘...ashootbesetwithsporophylls:...leavesbearingsporangia’ m
flowerbut,instead,aimtoestablishamorerigorouscontext (Goebel,1905,p.469) h
for future research. Our main objective is to review and, ‘...asectionofashoot,orbranchresemblingashortshoot,which ttp
s
where possible, clarify several of the key issues in this btreaanrssfolermafeodrgacacnosrwdihnigclhy’se(Wrveebfeorrlisnegx,u1a9l8r9ep,rpo.d2u)ctionandwhichare ://ac
broad field. a
‘...aparticulartypeofdeterminatesporogenousshoot...oneofits d
We willingly follow the modern convention of placing mostdefinitiveorgansisthecarpel;...amegasporophyll[thatis] em
our discussion in the context of an explicit phylogenetic mareorupshuoallolgyiceanlclylosdeisdtiwncittihvienbaehcoaullsoewthbeasoavlupleosrti(oi.ne.dmeseiggansapteodraansgitah)e ic.ou
framework, but within this framework we have chosen to ovary’(FosterandGifford,1971,pp.593–594) p.c
exercise certain prejudices. Unlike most recent contri- ‘...acompressed,bisexualreproductiveaxiscomposedofstamens, o
m
butions to this debate, we do not focus on a single phylo- carpelsandasterileperianth’(BaumandHileman,2006) /jx
‘...abisexualstructurewithmoreorlessrecognizablecarpels;that b
gimenpelitcicatiohnyspoothfesais,vaprrieeftyerrionfgminatsrtiexa-dbastoedcpohnysildoegrentihees i(sa,dlaaxmiailn)asrusrtfrauccet,uarensdlawrigtehlymeonrecloosrinlegssthreecoovgunliezsabolnetshtaemireunpsp’er /artic
derivedfromhighlycontrastingkindsofdata.Inassessing (Frohlich,2006) le-a
‘...adeterminateaxisterminatinginmegasporangiathatare b
the relative merits of these hypotheses, we emphasize surroundedbymicrosporangiaandarecollectivelysubtendedbyat stra
conceptual rigour over statistical robustness. We are leastonesterilelaminarorgan’(presentstudy) c
t/5
especially concerned with identifying the optimal roles 7
/1
fordifferentcategoriesofdatathatpertaineitherdirectlyor 3
/3
indirectly to phylogeny reconstruction. We therefore dis- racted from this aggregate of definitions: form, function, 4
7
1
cuss both molecular and morphological data sources, homology, and taxonomy. /4
7
including fossils, are discussed. This issue is explored in Themajorityofdefinitionsidentify(sexual)reproduction 7
5
3
the context of two contrasting perspectives on land-plant as the primary function of the flower; an uncontentious 9
b
phylogeny: the top-down perspective looks backward statement, but one that does not in itself separate a flower y
g
through evolutionary time from the present, whereas the from the reproductive structure(s) of any other land plant. u
e
s
bottom-upperspectivelooksforward throughevolutionary Two definitions include explicit references to taxonom- t o
time from the deep past. ically (and phylogenetically) delimited groups of organ- n 0
3
isms: the clade of groups that bear flower-like structures A
p
(luAsnivtheocplhaydteeso)fagnrdo,upmsotrheatcobnetarrovseeersdisall(yS,pethrmeamtooprheytiensc)-. ril 20
What is a flower? 1
The definition from Chambers Dictionary encompasses 9
Giventhevasttractsofpublishedtextdevotedtotheorigin gymnosperms as well as angiosperms. It therefore begs
of the flower (and of the flowering plants), this is one immediate rejection by all knowledgeable botanists, who
question that could reasonably be assumed to have been routinely equate the structure ‘flower’ with the (mono-
unequivocally answered. Indeed, most glossaries included phyletic) group ‘angiosperms’, and who consequently
in textbooks and reviews omit the term ‘flower’, pres- attempt to define gymnosperms in part by their absence
umably on the assumption that the term is universally of flowers.
understood and the underlying concept is familiar to even With regard to form, several definitions describe the
the most lackadaisical student of botany. However, comp- flower as a composite structure, referring explicitly to the
arisonof the definitions assembledin Table 1reveals little presence of four differentiable categories of (usually
unanimity. Four ostensibly distinct elements can be ext- physically discrete) organs that are organized in a specific
Phylogeneticcontextoftheflower 3473
and reliable linear sequence along the axis towards its various degrees of completeness, cannot adequately be
apex:sepals,petals,stamens,andcarpels.Afewdefinitions definedinwords’.This,andmanyothersimilarstatements
emphasized the function of stamens and carpels for in the literature, refer primarily to the considerable floral
generating gametes (male and female, respectively), and variationrevealedbysurveysofextantangiosperms.Some
Foster and Gifford (1971) and Frohlich (2006) further species lack sepals, petals, or both. For example, in one
identifiedthecarpelasbeingthemostdiagnosticfeatureof clade(SaururaceaeplusPiperaceae)ofthemagnoliidorder
a flower. Foster and Gifford also noted that definitions of Piperales,flowersareentirelyperianth-less,afeaturethatis
afloweromittingreferencetothesestructures,suchasthat closely correlated (at least, in this group) with possession
of Goebel (1905), implicitly encompass not only all of indeterminate inflorescences (Remizowa et al., 2005).
angiosperms but also all gymnosperms and even some of Manyotherspeciesshowmorphologicalgradationbetween
the more derived, reproductively complex, pteridophyte sepals, petals, and/or stamens. Yet others contravene the
groups (thereby rendering less bizarre the aforementioned requirement for bisexuality by bearing functional stamens
Chambers Dictionary definition: Table 1). and carpels on separate flowers, either on the same D
o
w
Crucial to the majority of definitions is an almost individual (monoecy) or on separate individuals (dioecy); n
lo
universally accepted interpretation of the flower as a det- organs of the opposite gender have either been rendered a
d
erminate reproductive shoot with a defined number of sterile or completely suppressed. For example, the much- ed
floral organs, typically (but not universally) arranged in at discussed ‘primitive’ extant angiosperm Amborella (Fig. fro
m
least four distinct whorls: sepals, petals, stamens, and 1A, B) is commonly (though not reliably) dioecious, h
carpels. Specifically, the flower is homologized with an bearing either exclusively male flowers (lacking even ttp
s
axis bearing sporophylls (i.e. evolutionarily modified sterile carpels) or female flowers that occasionally bear ://a
c
sporangium-bearing leaves or, perhaps more conserva- staminodes, sometimes with a developmental transition a
d
tively, leaf-like structures). Implicit in these definitions is between stamens and carpels (Buzgo et al., 2004). Many em
the dynamic concept of a transition from an earlier, other early-divergent extant angiosperms even lack carpel ic.o
u
profoundly different (and thus recognizable) ancestral closurebytissuefusion(EndressandIgersheim,2000a,b; p
.c
form of reproductive organ. The identity of the ancestral Endress, 2001a) or double fertilization (Williams and o
m
group is almost universally accepted, namely the gymno- Friedman,2002).Thecolumellarinfratectuminthemicro- /jx
b
sperms; an origin among the (by definition) seed-less sporewallofangiosperms,functionallylinkedtochemical /a
pteridophytes requires a set of morphological transitions recognition systems on the stigmatic surface, is also un- rtic
le
that seemingly is too radical for any modern botanist to reliably present in basally divergent extant angiosperms -a
b
seriously contemplate. (Sampson, 2000; Doyle, 2006; Frohlich, 2006). stra
Thus, we can draw on several categories of information Thus,severalsupposedlydefinitivefeaturesoftheflower c
t/5
in order to define a flower. And, having defined a flower, are frequently absent from angiosperms, especially early- 7
/1
surely we have by default defined a flowering plant—an divergent angiosperms: these include hermaphroditism, 3
/3
angiosperm.Whythenhavethemoreinformedamongthe fully closed carpels, and a distinctly whorledarrangement. 4
7
1
land-plantmorphologistsappendedfirmcautionarynotesto Admittedly, the presence on the ovule of a second (outer) /4
7
their preferred definitions? For example, according to integument, and the remarkably conservative structure of 7
5
3
Foster and Gifford (1971, p. 593), ‘As angiosperms are the stamen (two thecae joined by a connective, each con- 9
b
commonly designated the flowering plants, it might be sistingoftwoembeddedmicrosporangia),aremoreconsis- y
g
assumed that there is rather general agreement about the tent features of the angiosperm flower, though they too u
e
s
scientificconceptofaflower.Unfortunately,thisisnotthe have potential homologues in some gymnosperm groups t o
case, and the literature on floral organography, ontogeny, (Doyle, 2006). n 0
3
and structure displays widely divergent viewpoints of the Furthermore, the flower-subtending bract, normally A
p
finutnedrpamreetanttiaoln noafturiets ocfomthpeonflenotweorrgaasnsw(eslelpaalss, opnetathlse, rbeegacrodnesdidaesredexptraar-tfloofr,alt,hecaflnoswtreorn(gel.yg.inRfleumenizcoew, aoretevael.n, ril 20
1
stamens, and carpels). One of the basic difficulties lies in 2005; Buzgo et al., 2006). Bringing a palaeobotanical 9
ourcomplete[sic]ignoranceoftheevolutionaryhistoryof perspective to bear, Stewart and Rothwell (1993, p. 440)
the flower ...; it becomes largely a matter of conjecture coined the phrase ‘accessory reproductive structures’ to
whether it is justifiable to draw comparisons between encompass sterile organs associated with reproductive
modern angiospermous flowers and the spore-producing functions—not only perianth members (tepals, or petals
structures of other tracheophytes [vascular land-plants]. If plussepals)butalsoanother(lesswell-researched)leaf-like
suchcomparisonsareattempted,itisquitepossibletoreach organ,theflower-subtendingbract.Bracts(andbracteoles)
eitheraverybroadoraveryrestrictedconceptordefinition share with perianthmembersthe characteristics of exhibit-
of a flower.’ ingmanyleaf-likefeaturesbutbeingpositionallyfixedwith
Bierhorst (1971, p. 511) preceded his definition by respect to either a flower or an inflorescence. In some
stating that a flower is ‘a structure that, because of its angiosperms, the bract has become intimately integrated
3474 Batemanetal.
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
s
://a
c
a
d
e
m
ic
.o
u
p
.c
o
m
/jx
b
/a
rtic
le
-a
b
s
tra
c
t/5
7
/1
3
/3
4
Fig. 1. (A)Femaleand(B)male flowersoftheputative basallydivergentextant angiosperm Amborellatrichopoda.(C)Male (above)andfemale 71
(below)reproductivestructuresofthemuch-discussedJurassicfossilgymnospermCaytonia(Caytoniales)and(D)hermaphroditereproductivestructure /4
7
of the Jurassic fossil gymnosperm Williamsoniella (Bennettitales); bothare candidate sister-groups to theangiosperms. (C and Dreproduced, with 7
5
permission,fromfigs1.8and1.11,respectively,ofSoltisetal.,2005,andreprintedfromCrane,1985.)Scalebar=100lm. 3
9
b
y
g
u
e
s
into the flower; for example, Amborella (Fig. 1A, B) and ‘anthostrobilus’ for the modern angiosperm flower, and t o
several other early-divergent angiosperms (e.g. Austro- ‘pro-anthostrobilus’ for the type of cone manifested not n 0
3
baileya, Trimenia) exhibit a morphological continuum onlybytheMesozoicbennettiteWilliamsoniella(Fig.1D) A
p
betetawl.e,e2n00b4r)a.ctIsn oatnhderspe(eri.agn.tAhra(cEenader,eCsso,rn2a0c0e1abe,; SBauurzugro- banugtioalsspoerbmys athahtytphoetyhetteicrmaledgr‘oHupemoifanegxiotisnpcetrmstaeem’.-gTrohuepy ril 20
1
aceae) the inflorescence bracts are conspicuously petaloid noted that the bennettite cone possessed a series of sterile 9
and hence perform at least one of the functions (pollinator leaf-likeorgansthattheyinterpretedasan‘undifferentiated
attraction)thataremoretypicallyperformedbypetals.For primitive perianth’. They also insightfully defined a major
example, in two genera (Houttuynia and Anemopsis) of subsetof(derived)angiospermsaspossessing‘aeuanthos-
the perianth-less family Saururaceae, the inflorescence trobilus, of which the distinctive features are the presence
bracts form a pseudocorolla, and the entire inflorescence ofthespecialtypeofmicrosporophylltermedastamen,and
has a flower-like appearance (Tucker, 1981). of closed carpels’ (Arber and Parkin, 1907, p. 75). Thus,
Moresignificantly,manyofthegymnospermousgroups, if flowers are at least partly defined not by possession of
bothextantandextinct,whicharebydefinitionconsidered sepals and/or petals but by possession of the broader
to lack a differentiated perianth, unequivocally possess category of leaf-like accessory reproductive structures, the
bract-likeorgans.ArberandParkin(1907)coinedtheterms flowerisnolongerseenasaunique(anddefining)attribute
Phylogeneticcontextoftheflower 3475
of the angiosperms, and when using the term ‘flowering (carpels) (Arber and Parkin, 1907; Arber, 1937). By
plants’ certain gymnospermous groups, especially extinct contrast, the more diverse pseudanthial theories all per-
taxa such as Bennettitales, should strictly be included (cf. ceive the flower as having condensed from a multiaxial
Crane, 1988). This conclusion is implicit in our own structure (Wettstein, 1907; Melville, 1960; Eames, 1961;
(inevitablyimperfect)definitionofaflower:adeterminate Meeuse, 1975, 1987, Stuessy, 2004). Thus, the difference
axis bearing megasporangia that are surrounded by betweenthetwoconflictinghypothesesrelatesmoretothe
microsporangia and are collectively subtended by at least nature of individual organs than to the flower as a whole.
one sterile laminar organ. In formulating this definition, Several authors have postulated secondary derivation of
the orientation of our discussion has, in practice, switched flower-like structures from inflorescences (i.e. a secondary
from top-down to bottom-up. We note that, although we pseudanthial origin in certain phylogenetic groups), based
do not necessarily exclude multiple origins of flower-like on ontogenetic evidence. These studies relate to both
structures(seebelow),ourdefinitionhasconvergedonthat gymnosperms (Gnetales: Mundry and Stu¨tzel, 2004) and
of Arber and Parkin (1907, p. 75), who characterized the angiosperms (e.g. alismatid monocots: many authors, D
o
w
anthostrobilusas‘aspecialformofamphisporangiatecone, reviewed by Sokoloff et al., 2006). Multiple origins of n
lo
distinguishedbythepeculiarjuxtapositionofthemega-and flower-like structures, both within angiosperms and other a
d
microsporophylls, and by possessing a well-marked peri- seed plants, are implicit in these pseudanthial hypotheses. ed
anth’. We also note that Arber and Parkin primarily (and Studies of flowers prompted by the various insights fro
m
almost uniquely) employed a bottom-up approach in what outlinedabove were, to varying degrees, comparative,and h
becameabenchmarkstudyinfloralevolutionthatremained in some cases hypothesis-testing. However, they often (i) ttp
s
influential throughout the ensuing century. focused on limited suites of morphological characters ://a
c
consideredaprioritobeofparticularimportance,(ii)failed a
d
to determine whether the characters of interest were ever em
Which are the (other) benchmark studies in specifiedbyasinglegenome(i.e.wereobservedinasingle ic.ou
individual), and/or (iii) unjustifiably equated extreme div- p
floral evolution? .c
ersityofformwithincreasedlikelihoodofmultipleorigins o
m
Certain conceptual thresholds can readily be identified in of the feature in question. More importantly, they lacked /jx
b
thestudyofflowersandfloweringplants.Inhisbenchmark aunifyingconceptualframework.Thiscrucialprerequisite /a
classification of flowering plants, Linnaeus (Linne´, 1735) has been provided, first by phylogeny reconstruction and rtic
le
emphasizedthesignificanceofthenumberandarrangement then by evolutionary-developmental genetics, epitomized -a
b
ofstamensandcarpels.Thepossibleequivalenceofsepals, by the ABC model of floral whorl control (e.g. Coen and stra
petals, and stamen filaments (though not explicitly the Meyerowitz, 1991; Irish and Kramer, 1998; Kramer and c
t/5
carpels) to modified leaves was greatly elaborated by Irish, 1999; Lawton-Rauh et al., 2000; Cronk et al., 2002; 7
/1
Goethe (1790) in an essentialist essay partly stimulated Theissenetal.,2002). 3
/3
by studying plant teratologies. However, Goethe’s ideas 4
7
1
about simple equivalence, summarized by the famous /4
7
phrase ‘all is leaf’, had no phylogenetic implications 7
Which are the benchmark studies in 53
(Lo¨nnig, 1994). As noted by adherents of Zimmerman’s 9
reconstructing seed-plant phylogeny? b
telome theory (Zimmerman, 1930, 1938), such as Wilson y
g
(1937), spore-bearing organs (i.e. sporangia) evolved The now ubiquitous tree motif routinely used to repre- u
e
s
before leaves in early land plants. By contrast, Darwin’s sent the supposed sequence of divergence of evolutionary t o
(1859) developmentof a credibleevolutionary mechanism lineages was famously employed by Darwin (1859) and n 0
3
to explain (albeit in a uniformly gradualistic manner) such elaborated by authors such as Haeckel (1894). However, A
p
rhaadrinceaslsinmgoropfhoMloegnidcealliatnragnesinteiotincss twoasaddforellsoswtehde bcoynttrhoel o(1n9ly66w)itahntdheBeraurnlydicnlad(1is9t7ic2w) owrkerseofweentopmroovliodgeidstswHitehnnthige ril 20
1
ofsuchtransitionsbyauthorssuchasDeVries(1906)and rigorous conceptual framework needed to generate such 9
later Wardlaw (1965). trees with a large degree of objectivity from matrices of
These new data sources informed some increasingly coded taxa, each explicitly scored for competing states of
sharp exchanges between proponents of the two main sets a broad suite of characters. Central to this approach is the
of theories competing to explain the origin of the flower. conceptofthecongruencetestofhomology(e.g.Patterson,
The morecommon euanthial theory that waspioneered by 1988). The preferred phylogeny for the scored taxa is the
Goethe(andisimplicitinmostofthedefinitionsofaflower one requiring fewest transitions between character-states.
collated in Table 1) postulates derivation from an un- Wecanthenobserveintheresultingtree(s)whichcharacter-
branched,uniaxialstructure,andhenceinterpretstheflower states delimit which taxonomic groups—in other words,
as a condensed sporophyll-bearing single axis with prox- whichpriorstatementsofhomologybetween species have
imalmicrosporophylls(stamens)anddistalmegasporophylls been upheld in the most-parsimonious tree(s). The
3476 Batemanetal.
particulate nature of the numerically scored character stickfor comparative studies that focus simultaneously on
states is preserved in the resulting trees, thereby simpli- changes in key developmental genes and the phenotypic
fying various kinds of post hoc character analysis. charactersthatthey‘control’—agrowingdisciplinetermed
The cladistic approach was first used to compare the evolutionary-developmental genetics (e.g. Cronk et al.,
majorgroupsofseed-plantsbyHillandCrane(1982).With 2002). This approach is especially attractive, as it repre-
theassistanceofrapidlyimprovingcomputerhardwareand sents an alternative, and potentially more informative, test
software, it spawned an increasing number of morpholog- of a priori homology statements that also explicitly links
ical phylogenies based on parsimony analyses during the genotype with the resulting phenotype, and thence ulti-
1980s and early 1990s. If we consider the development matelywithspecificbiologicalfunction(s).Thus,themajor
of taxonomically broad morphological matrices for seed- phenotypic transitions first explored morphologically by
plants, four main lineages can be recognized, nucleating Goethe (1790), functionally by Darwin (e.g. 1859), and
around PR Crane (e.g. Hill and Crane, 1982; Crane, 1985, genetically by De Vries (1906) can at last begin to be
1988), JA Doyle (e.g. Doyle and Donoghue, 1986, 1987, examined for fundamental causation (e.g. Frohlich and D
o
w
1992; Doyle, 1996, 1998a, b, 2006; Hilton and Bateman, Parker, 2000; Bateman and DiMichele, 2002; Vergara- n
lo
2006), GW Rothwell (e.g. Rothwell and Serbet, 1994; Silva, 2003; Baum and Hileman, 2006; Hintz et al., 2006; a
d
Rothwell and Nixon, 2006), and DW Stevenson (e.g. Theissen, 2006). ed
Loconte and Stevenson, 1990, 1991; Nixon et al., 1994). Despite their conflicting topologies (Fig. 2), these fro
m
Thesestudiesplacedasclosestrelativesofangiosperms,in different approaches share some common elements with h
various paraphyletic combinations, several groups (most respecttoseed-plantrelationships.Angiospermmonophyly ttp
s
of them wholly extinct) that possess reproductive organs is routinely inferred. Nonetheless, a few brave authors ://a
c
showing some flower-like properties, notably Gnetales, generally viewed as mavericks, notably Meeuse (1976, a
d
Bennettitales (Fig. 1D), Pentoxylon, Caytonia (Fig. 1C), 1987), have penned morphologically based (though not em
andglossopterids.Sucharrangementssoonbecameknown matrix-based)argumentsforpolyphyly.Cladisticanalyses, ic.o
u
collectively as the anthophyte hypothesis, implying that both morphological and molecular, consistently place the p
.c
the flower evolved only once and, hence, that all flowers clade containing all extant angiosperm lineages (the o
m
are fundamentally homologous. ‘crown-group’ angiosperms: Fig. 3) on a long branch /jx
b
By1990,these morphologicalanalyses werecompeting with respect to the remaining seed plants. ‘Missing links’ /a
with(andsoonlargelysupersededby)phylogeneticstudies (i.e. stem-group) angiosperms have been postulated from rtic
le
employing as characters the sequences of nucleotides (and the fossil record but have not hitherto obtained universal -a
b
insertion–deletion mutations) in specific regions of the acceptance (e.g. Archaefructus, discussed below). Further- stra
three plant genomes: nuclear-chromosomal, plastid, and more,relationshipsamongextantgymnosperms,andhence c
t/5
mitochondrial.Asthenumberofreadilysequencedregions homologies among their reproductive structures, remain 7
/1
(and hence the number of usable characters) increased contentious. These issues can only be partly circumvented 3
/3
exponentially, driven by advances in sequencing technol- by adopting a top-down approach but are fundamental to 4
7
1
ogy, maximum parsimony was supplemented with more any bottom-up approach. /4
7
mathematically complex methods of generating trees, 7
5
3
notably maximum likelihood and, latterly, Bayesian ap- 9
b
proaches (e.g. Page and Holmes, 1998). The current spe- y
g
cies richness and ecological pre-eminence of angiosperms Which kinds of phylogeny are of greatest value? u
e
s
have caused them to be preferentially sampled for phylo- Why do tensions remain within the phylogenetic com- t o
genetic studies. The result is a framework of relationships munity regarding molecular versus morphological char- n 03
that is generally viewed as well-sampled and increasingly A
acters and extant versus extinct taxa? Over the past p
(etthoaul.g,h20b0y4,no20m0e5a),nsanudniuvnerdsearlplyin)sashirgehliearblcela(ses.igfi.caStoioltniss two decades, molecular phylogenies have in practice ril 20
largely superseded morphological phylogenies. This shift 1
thatarebasedon‘natural’monophyleticgroups(e.g.APGII: 9
of emphasis has had profound consequences, yet it is
Angiosperm Phylogeny Group, 2003).
now rarely discussed. Arguments most commonly adv-
At present, increasing numbers of species have been
anced against morphological rather than molecular phylo-
sequenced for the complete plastid and/or mitochondrial
genetic analyses are:
genomes (e.g. Goremykin et al., 2005), and a few for the
entire nuclear genome (e.g. rice, Arabidopsis; Arabidopsis (i) the limited number of characters available (and the
GenomeInitiative,2000;Bennetzen2002;Yuetal.,2002; existence of an asymptote of the number of phyloge-
Yamadaetal.,2003;InternationalRiceGenomeSequenc- netically informative character states available as taxa
ing Project, 2005). These sequence-based data not only are successively added to a matrix: Bateman, 1992);
allow character-rich (if presently species-poor) phylogeny (ii) the high cost in time expended per unit character
reconstructions of extant species but also provide a yard- coded(bothintermsofactualcharacterscoringandthe
Phylogeneticcontextoftheflower 3477
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
s
://a
c
a
d
e
m
Fig. 2. Crudeconsensusofphylogeneticrelationshipsrecognizedbymorphologicalanalysesusingextantandextincttaxa(A)andsequence-based ic
analysesusingonlyextanttaxa(B),illustratingthedominanceofparaphylyintheformerandmonophylyinthelatter. .o
u
p
.c
o
m
‘informalapprenticeship’thatmustfirstbeundertaken deletion events (in other words, in assigning some
/jx
in order to describe and code characters correctly); states of some taxa to the correct character); b
/a
(iii) an inevitable degree of subjectivity involved in (iv) the availability of the same restricted set of a maxi- rtic
making a priori homology assessments (in other mumoffour(orarguablyfive:A,C,G,T,andabsent) le
-a
words, in delimiting characters before each is in turn states for each position/character artificially masks bs
divided into alternative character states); much of the actual homoplasy (e.g. when a particular tra
c
(iv) the supposed comparatively high level of homoplasy A mutates to a T but later reverts to an A); t/5
7
(this is caused by conflicting character states, which (v) the potential in expressed regions of the genome for /1
3
are considered to mainly reflect similar responses to particular categories of nucleotide to behave differ- /3
4
7
similar pressures of directional or disruptive selection ently (e.g. contrasting mutational rates in different 1
/4
in lineages that are in fact only distantly related: compartments and regions of the genome, first and 7
7
5
Scotland et al., 2003), and; second versus third bases within codons, and radical 3
9
(v) the potential developmental correlation of apparently contrasts in the GC:AT ratio). b
y
unrelated characters (e.g. pleiotropic expression of g
u
Additional phenomena that can negatively affect both e
a single critical mutation). s
morphological and molecular matrices include: t o
n
0
Atleastsomeobserversperceivecorrespondingconstraints (i) strong heterogeneity of rates of change within the 3 A
on molecular phylogenies: sbtruadnychgersoutpo, feanlcseoluyragcoinaglesthcee;mthoisst rnaopwidlythcohroanuggihnlgy pril 20
1
(i) the limited number, and sporadic phylogenetic distri- researched, but still problematic, phenomenon is 9
bution, of taxa available, due to the inability to allow termed long-branch attraction (e.g. Sanderson et al.,
molecularly recalcitrant extinct taxa to participate in 2000; Felsenstein, 2004);
the tree-building procedure (a central issue of this (ii) migration of genes (and the phenotypic characters
paper); that they underpin) between lineages through hybrid-
(ii) the fact that routinely sequenced regions of the three ization, lateral gene transfer, and organelle capture,
genomes do not participate directly in the phenotypic and;
transitions that signal a macroevolutionary event (iii) the necessity of rooting the tree, typically through
(Bateman,1999); outgroup choice, in order to polarize characters and
(iii) an inevitable degree of subjectivity in aligning thereby study character evolution (another central
nucleotides in matrices that are rich in insertion– issue of this paper: see below).
3478 Batemanetal.
andthenmapped,therebyeffectivelygeneratingtautologous
interpretationsofevolution(Bateman,1999).
How significant are optimization and outgroups
choice?
The concept of character mapping leads naturally into
discussionofanotherareaofphylogeneticsthatispivotalto
thequestionsaddressedbythispaper,namelyoptimization.
If we wish to know what the first flower looked like, but
have not found it in the fossil record (such a discovery is
highly improbable, given the undoubted patchiness of the D
o
w
fossil record of land-plants), we need to reconstruct that n
lo
flower conceptually. This is achieved using combinations a
d
ofcharactersfoundinspecieswhosemorphologyhasbeen ed
carefully described and whose phylogenetic relationships fro
m
have been rigorously inferred (this phrase is not oxymor- h
onic; it is important to remember that even the most ttp
s
Fig. 3. Hypotheticalphylogenyillustratingtherelativisticnatureofthe rigorously reconstructed phylogeny remains wholly infer- ://ac
ential).Providedwehaveaccesstoamatrixthatthoroughly a
concepts of crown group and stem group, and the significance of the d
subtendingnodesinanyattempttoreconstructthenatureofhypothetical describesthemorphologyofallorgansofthespeciesunder em
abnacseedstoarpsp.rToahcehedsiatgoradmeliamlsiotincgonmtroanstosphthyelentiocdger-obuaspesd. Danadshaepdobmraonrpchheys- scrutiny, then we can reconstruct the morphology of the ic.ou
subtendextincttaxa,cross-strikesonthebranchimmediatelybelowthean- hypothetical ancestors that lie on each node of the clado- p.c
giospermcrown-groupnodeindicateindividualcharacter-statetransitions. gram. Optimization is most readily achieved by a simple o
m
logical protocol that works downward through the tree /jx
b
In addition, the main source of operator bias influencing from the terminal taxa toward the outgroup node (Fig. 3) /a
the resulting topology is a priori homology assessment (e.g. Maddison and Maddison, 2001), though other, more rtic
le
for morphological data, whereas for molecular data it is complex, models can be applied, sometimes with advanta- -a
b
selectivity among (i) available characters and (ii) tree- geousresults(OakleyandCunningham,2000;Polly,2001; stra
buildingalgorithms. Moreover,bothcategories ofanalysis Webster and Purvis, 2002; Crisp and Cook, 2005). c
t/5
are prone to culling of ‘troublesome’ taxa from matrices, Amost-parsimonioustree,derivedfromaparticulardata 7
/1
and revised outgroup choices made in search of more matrix, yields similar information about the relationships 3
/3
‘intuitively acceptable’ topologies. of the coded taxa and the relative lengths of the various 4
7
1
Many phylogeneticists advocate a compromise approach branches within the tree, irrespective of whether it is /4
7
to the relative treatment of morphological and molecular unrooted or rooted. However, an unrooted tree lacks 7
5
characters. In one frequently used protocol, morphological polarity; it cannot be read in terms of transitions from 39
b
charactersarecombinedwiththemolecularcharactersprior ancestral character states (plesiomorphies) to derived y
g
totreebuilding,butonlyaftertheinitialmorphologicalmatrix character states (apomorphies). This in turn means that it u
e
s
hasbeenreducedtoits bareessentials by cullingcharacters is not possible to optimize character states in order to t o
thatevokesuspicion,mostcommonlyduetothedifficultyof reconstructthehypotheticalancestorsoccupyingthenodes n 0
3
dividing a complex of continuous character into discrete ofthetree.Inearlier(andmoreexperimental)phasesinthe A
p
sStcaotetlsa.nHdo(w20e0v6e)r,caonrvecinecnitndgelytarielefduteasnathlyissiaspbpyroWacohr,tdleeymaonnd- ecvoonlcuetpiotunalloyfdmisetitnhcotdaspporfoapchhyelsogtoenroetoictinagnwaleyrseise,xspelovreerdal, ril 20
1
strating that the culled characters contain the same average but most proved impractical and all appeared at least 9
strength of phylogenetic signal as the supposedly superior partially tautologous.
characters that survive the cull. The alternative, and more Over the last two decades a single method, termed
commonly used, approach is generally termed ‘mapping’. outgroupcomparison(e.g.NixonandCarpenter,1993),has
Here,themorphologicalcharactersareplacedalongthetips become ubiquitous, to the point where it is routinely
of the phylogeny after it has been constructed. Mapping adopted without serious thought by almost all practising
prevents morphological characters from contributing in any phylogeneticists. It requires the analyst to select a priori
waytothetree-buildingexercise.Mappingcanalsodissuade a set of comparable taxa (preferably species) that are of
researchersfromseeking(andthenoccasionallydiscovering) interest and are suspected of being an inclusive, mono-
newphylogeneticallyvaluablecharacters.Alltoooften,well- phyleticgroup(inpractice,thestrongestguidetoperceived
known phenotypic characters of a priori interest are scored monophyly is a previous, and taxonomically broader,
Phylogeneticcontextoftheflower 3479
phylogenetic analysis; thus, the outgroup method is offerthemostobviousmeansofmaximizingthenumberof
vulnerable to accusations of logical tautology). Then, one characters per taxon.
or more additional taxa, thought to lie phylogenetically Given the continuing uncertainties regarding relation-
outside the chosen study group, and hence operationally ships among the major groups of land plants (discussed
termed outgroups, are chosen to simultaneously root the below), and their potential influence on optimizations of
tree and polarize the scored characters. If multiple out- nodes,severalbranch-pointsdistantfromthetargetnode(s),
groups are chosen they constitute a test (albeit flawed) of we sought phylogenies that encompassed all such major
themonophylyoftheingroup,sinceanoutgrouptaxonthat groups. In the case of morphological phylogenies, no
is placed phylogenetically within the ingroup contradicts existing study convincingly stretched from the bryophytes
ingroup monophyly. Clearly, the larger the numbers of to the more derived angiosperms (eudicots sensu APGII:
ingroup and outgroup taxa, the stronger is this test. AngiospermPhylogenyGroup,2003).Wethereforetookthe
Outgroup choice leaves the analyst with a further controversial step of grafting a selected most-parsimonious
quandary. In most cases, the ingroup is chosen because it tree from our recent analysis of 48 lignophytes (angio- D
o
w
is a morphologically and/or molecularly distinct aggregate sperms plus gymnosperms, plus progymnospermous pter- n
lo
ofspecies,andsoislikelytodifferconsiderablyfromeven idophytes as operational outgroups: Hilton and Bateman, a
d
itsphylogeneticallyclosestoutgroups(hence,intheresult- 2006, fig. 10) onto the product of a recent parsimony ed
ingtrees,amonophyleticingroupislikelytobesubtended analysisof52codedtaxathatfocusedonpteridophytesbut fro
m
by a relatively long branch). These differences make used the fossil non-vascular polysporangiophyte Aglao- h
characterdelimitationmoredifficultforbothmorphological phytonasoutgroupandjustfivetaxa(oneprogymnosperm ttp
s
andmolecularanalysts.Forthemorphologist,theoutgroup plus three primitive pteridospermous gymnosperms and ://a
c
is likely to lack structures found in the ingroup, possess the extant Pinus) as representative ‘placeholders’ for the a
d
structuresnotfoundintheingroup,orthetwogroupsmay monophyleticlignophytes(RothwellandNixon,2006,fig. em
possess structures that are broadly similar yet sufficiently 3). The resulting composite phylogeny contains 95 coded ic.o
u
different that their homology cannot be adequately asses- taxa:47whollyextinctand48containingatleastoneextant p
.c
sed. For the molecular researcher, sequence alignment species (Fig. 4). o
m
becomes more challenging for many genic regions, and Selecting among the many broad-brush molecular trees /jx
b
base saturation becomes an ever-increasing risk. These in order to produce Fig. 5 proved even more problematic. /a
problems of strong ingroup–outgroup divergence can, in The controversial analysis by Goremykin et al. (2003, rtic
le
theory, be averted by identifying one or more members of 2004;seealsoSoltisetal.,2004;Martinetal.,2005)stood -a
b
theingroupasoperationaloutgroups,buttheanalystisthen outasbeingespeciallycharacter-rich,sinceitwasbasedon stra
imposing a strong subjective overprint on the topology of nearly completely sequenced plastid genomes. The trade- c
t/5
the resulting tree(s). Certainly, the polarity of characters, offisthattheanalysislackeddatafromtheothertwoplant 7
/1
and thus the optimizations that allow reconstruction of the genomes (nucleus and mitochondrion), and was restricted 3
/3
properties of the hypothetical ancestor of the ingroup, to <20 coded taxa; moreover, the taxa were selected as 4
7
1
would be strongly influenced by such a decision. much for their economic importance as their likely /4
7
But before we contemplate implementing of an optim- phylogenetic significance. This left the trees open to 7
5
3
ization procedure, we must first select our preferred accusations of distortion by long-branch attraction (Soltis 9
b
topology from among the very broad spectrum of land- et al., 2004; Stefanovic et al., 2004; Leebens-Mack et al., y
g
plant phylogenies generated since 1982. 2005),thoughamorerecentstudysuggeststhatlikelihood u
e
s
model mis-specification is a more probable source of t o
systematic error in the trees (Goremykin and Hellwig, n 0
3
2006). The matrix of Soltis et al. (2002) encompassed A
What is the best way to choose among the p
plethora of phylogenies? adisstriimbuiltaerdnaucmrobsesrthoef ecxotdaendt ltaanxda-,plaalnbteipthymloogreenye,veannlyd ril 20
1
Oneofourprimaryobjectiveshereistoexploretherelative sequenced eight genic regions distributed among the 9
merits of maximizing sampling of taxa versus that of plastid (rbcL, atpB, psaA, psbB), mitochondrion (mtSSU,
characters per taxon in taxonomically broad (i.e. at least cox1, atpA), and nucleus (18S rDNA).
Class level) phylogenetic analyses. Morphological studies Attheotherendofthespectrumofmolecularcharacters
that include the best-understood fossils (i.e. conceptual per coded taxon are analyses based on a single region but
whole plants reconstructed from their component parts: with substantial taxon sampling. A good example is the
Chaloner,1986;Bateman,1992)offerthebestopportunity 2538-taxon rbcL study of land plants by Ka¨llersjo¨ et al.
to maximize sampling of taxa that provide strongly (1998, 1999), subsequently reduced to an illuminating
contrasting combinations of character states. By contrast, series of analyses of various combinations of 80 taxa
recentmoleculartrees,someofwhichcompareentireplas- sampled evenly across the phylogenyof extantland plants
tid genomes (e.g. Goremykin et al., 2003, 2004, 2005), by Rydin and Ka¨llersjo¨ (2002). Matrices of intermediate
3480 Batemanetal.
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
s
://a
c
a
d
e
m
ic
.o
u
p
.c
o
m
/jx
b
/a
rtic
le
-a
b
s
tra
c
t/5
7
/1
3
/3
4
7
1
/4
7
7
5
3
9
b
y
g
u
e
s
t o
n
0
3
A
p
ril 2
0
1
9
Fig. 4. Compositemorphologicalphylogenyof95taxa(47fossil,inboldface)obtainedbygrafting(atarrow)aseed-planttreeofHiltonandBateman
(2006,fig.10)ontoapteridophytetreeofRothwellandNixon(2006,fig.3a).