Maximum likelihood inference implies a high, not a low, ancestral

12 Pages·2012·0.4 MB·English

Annalsof Botany109:681–692, 2012
doi:10.1093/aob/mcr302,available onlineatwww.aob.oxfordjournals.org
Maximum likelihood inference implies a high, not a low, ancestral haploid
chromosome number in Araceae, with a critique of the bias introduced by ‘x’
Natalie Cusimano*, Aretuza Sousa and Susanne S. Renner
Department of Biology, University of Munich (LMU), D-80638 Munich, Germany
*For correspondence. E-mail [email protected]
D
Received:6June2011 Returnedforrevision:26July2011 Accepted:7November2011 Publishedelectronically:30December2011 o
w
n
lo
a
†BackgroundandAimsFor84years,botanistshavereliedoncalculatingthehighestcommonfactorforseriesof d
e
d
dhuapcilbolied)crherfoermenocsoemtoesnpuemciebserrseltaotiaornrsivheipastaandsof-rceaqluleedncbiaessiocfnduimffebreern,txn.uTmhibsewrsaisndaocnleadwei.tLhoikuetlichoonosdistmenotde(rlespthroa-t from
treatpolyploidy,chromosomefusionandfissionaseventswithparticularprobabilitiesnowallowreconstruction h
ochfraonmceosstormalechnruommboesromcheannguembinertsheinlaanrgeexmploicniotcforatmfaemwoilryk.AWraecehaaeveanudsetdoatemstodeaerllliinerghayppporothacehsetsoarbeocuotnsbtarusicct ttp://a
o
numbersin the family. b.o
†Methods Using a maximum likelihood approach and chromosome counts for 26% of the 3300 species of xfo
Araceae and representative numbers for each of the other 13 families of Alismatales, polyploidization events rd
jo
and single chromosome changes were inferred on a genus-level phylogenetic tree for 113 of the 117 genera u
ofAraceae. rna
†KeyResultsThepreviouslyinferredbasicnumbersx¼14andx¼7arerejected.Instead,maximumlikelihood ls.o
oCphtriommizoastoiomnerfeuvseioanled(loasns)aisnctheestrparledhoamplioniadntcihnrfoemrroedsoemveennt,uwmhbeerreaosfpnol¼yp1lo6i,diBzaatyieosniaenveinntfseroecnccuerreodflnes¼sf1re8-. at Wrg/
quentlyandmainly towardsthe tips ofthe tree. a
s
†ConclusionsThebiastowardslowbasicnumbers(x)introducedbythealgebraicapproachtoinferringchromo- h
in
some number changes, prevalent among botanists, may have contributed to an unrealistic picture of ancestral g
to
chromosome numbers in many plant clades. The availability of robust quantitative methods for reconstructing n
U
ancestral chromosome numbers on molecular phylogenetic trees (with or without branch length information), n
withconfidencestatistics,makesthecalculationofxanobsoleteapproach,atleastwhenappliedtolargeclades. ive
rs
ity
Keywords: Araceae, Bayesian inference, chromosome evolution, haploid chromosome number, dysploidy, , O
maximumlikelihoodinference, polyploidy. lin
L
ib
ra
INTRODUCTION chromosome numbers, the ‘Index of Plant Chromosome ry - S
numbers’ (http://mobot.mobot.org/W3T/Search/ipcn.html). e
Chromosome numbers in angiosperms vary from n ¼ 2 Given the incomplete knowledge of angiosperm chromo- rials
(Tsvelev and Zhukova, 1974; Singh and Harvey, 1975; some numbers, evolutionary changes in chromosome U
n
Sokolovskaya and Probatova, 1977; Erben, 1996) over n ¼ numbers in most clades can only be estimated. Botanists do it o
250 (Oginuma et al., 2006) and n¼298 (Johnson et al., this by calculating a so-called basic, or monoploid, chromo- n F
1989) ton ¼320 (Uhl, 1978). The range in animals is some number, denoted x, to differentiate it from the haploid eb
ru
similar (Crosland and Crozier, 1986; Imai et al., 2002). Such (usually the gametophytic) number n and the diploid (sporo- a
ry
drastic differences in chromosome number, sometimes even phytic or somatic) number 2n. The concept of x goes back 2
4
withinsmallgroups,raisequestionsabouttheevolutionarydir- to Langlet (1927) who explained it using Aconitum as an , 2
0
ectionandfrequencyoftheimplieddrasticgenomerearrange- example; if different Aconitum species have n ¼ 8, n ¼ 12, 12
ments. Cytogenetic studies have shown that chromosome n ¼ 16 and n ¼ approx. 32, their inferred monoploid
numberscanchangeduetofission,fusionorgenomedoubling number x is 4 (Langlet, 1927: 7). Langlet’s idea took off, at
(Guerra,2008),andthereisampleevidencethatsuchchanges least in botany, where thousands of basic chromosome
can contribute to speciation. It has also been inferred that a numbers have been inferred, even for poorly counted groups.
large fraction of all plant species may have polyploid Thus, for flowering plants, Raven (1975, p. 760) suggested x
genomes (Stebbins, 1971; Goldblatt, 1980; Otto and ¼ 7 as ‘characteristic of all major groups of both dicots and
Whitton, 2000; Ramsey and Schemske, 2002; Cui et al., monocots except Caryophyllidae.’ Similarly, base chromo-
2006; Soltis et al., 2009; Wood et al., 2009; Jiao et al., some numbers ofx ¼ 12or x ¼ 5and6havebeen suggested
2011). Chromosome counts, however, exist only for 60 000 forPoaceae(reviewedinHilu,2004)andx ¼ 7orx ¼ 12for
of the 300 000–352 000 species of flowering plants Triticeae (Heslop-Harrison, 1992; Luo et al., 2009). Many
(Bennett, 1998; http://www.theplantlist.org/browse/A/). Most further examples of divergent base numbers having been cal-
published numbers are listed in an electronic database for culated for a clade could be cited (Soltis et al., 2005; Blo¨ch
#The Author2011.Publishedby OxfordUniversity Presson behalfofthe AnnalsofBotanyCompany.Allrights reserved.
For Permissions,please email:[email protected]
682 Cusimano et al. — Maximum likelihood-inferred chromosome number changes
et al., 2009). Part of the reason why different researchers took into account morphological phylogenetic analyses
arrived at different base numbers (x) has to do with the (Grayum, 1990; Mayo et al., 1997).
unclear definition of x, with some treating it in Langlet’s ori- Molecular phylogenetic work over the past few years has
ginal sense as an algebraically discoverable highest common resulted in aroid relationships at the generic level becoming
factor, others as ‘the lowest detectable haploid number relatively clear (French et al., 1995; Cabrera et al., 2008;
within a group of related taxa’ (Stuessy, 2009: 264; italics Cusimanoetal.,2011).Wehereusethemostrecentphylogen-
ours), and yet others as ‘the haploid number present in the etic analysis of Araceae to inferchromosome evolution in the
initial population of a monophyletic clade’ (Guerra, 2008: family, using the model-based approach of Mayrose et al.
340), i.e. an inferred number, since the ‘initial population’ (2010), in both its ML and Bayesian implementations, the
will not usually have its chromosomes counted. How to latter having the advantage that uncertainty in ancestral state
maketheinferenceisuptotheinvestigator.Zoologists,incon- probabilities is readily quantified. To test the power of their
trast, never became enamoured of the concept of an inferred method, Mayrose et al. (2010) first used simulated data and Do
w
base number x. then four exemplar plant clades (Aristolochia, Carex, n
Criteria for inferring ancestral (perhaps no longer present) Passiflora and Helianthus) with relatively densely sampled loa
d
chromosome numbers from empirical counts could come phylogenetic trees and chromosome counts. Sampling in ed
from phylogenetic analyses, the relative frequencies of differ- these clades ranged from 11 to 100% of the species in the fro
m
ent haploid numbers in various species groups, cytogenetic genera. The Araceae data set analysed here represents an h
work on closely related species or, best, a combination of all entire family that is larger and older by at least an order of ttp
such information. Data from genomics and molecular– magnitude. This poses challenges that we tried to address by ://a
o
b
cytogeneticmethods,suchasfluorescenceinsituhybridization experimentally modifying character codings to take into .o
x
(FISH)-marking chromosomes, are probably the best way to account uncertaintiesinthelarger generaandthe13outgroup fo
search for evidence of past chromosome number changes families. rdjo
u
becausetheycanidentifysynteny,fusionsitesorunusualloca- rn
a
tions of centromeres, in turn providing evidence for duplica- ls
tions, fusions or losses (Bowers et al., 2003; Lysak et al., METHODS .org
2b0e06fe;aPseibrulezziinetlaarlg.,e20c0la9d)e.sSuocrhtmhoestheodwsi,thhofweewvesr,pemcaieysniont Familyand order phylogeny at W/
a
cultivation. ThephylogenetictreeforAraceaeonwhichancestralchromo- sh
In 2010, an approach was developed that moves the infer- some numbers were inferred in this study is based on the six- ing
ence of chromosome number evolution to maximum likeli- plastid marker matrix of Cusimano et al. (2011). Clades are ton U
hood (ML) character state reconstruction (Mayrose et al., named as proposed in that study. We used the ML tree from n
d2e0s1c0r)i.biMngayrthoeseeevtoalul.ti(o2n01o0f) fcohrrmomuloatseodmperonbuambibliesrticacmroosdselas tuhsaintgstBudEyASoTr avn. 1u.l6t.r1am(DetrruicmmBaoynedsiaanndtrReaemnbeawutl,y2o0b0t7a)i.neIdn iversity
phylogenetic tree. Their approach makes use of branch BEAST, we used the GTR+G model with four rate categor- , O
lneunmgtbhesrsasatatphreoxtiypsfoarntdimine aonudtgorofuthpes ftoreqinufeenrctireasnosiftidoinfferraetenst ipeusr,e-abimrtehanYsuulbestmituotdioenl raastethesetimtraeteedpfrrioomr. tThheedaGtaT,Ran+dGa lin Lib
between the different states. Ancestral chromosome numbers model fit the data best, as assessed with Modeltest (Posada rary
have previously sometimes been reconstructed using andCrandall,1998).Theanalysiswas runfor37million gen- - S
maximum parsimony (e.g. Soltis et al., 2005: 178, 298– erations,samplingevery1000thstep.Theburn-infraction,i.e. eria
302). Parsimony, however, assigns all state transitions the thenumberoftreestobediscardedbeforerunsreachedstatio- ls U
sgaemneetiwcebirgahntchanldendgitshre,gwahrdicshintefonrdmsattoiornescuoltntianinaenduinndpehreysltoi-- tnhaeriBtyE,AwSasTapssaecsksaegde)usainndgAthWeTTrYac(eNryvl.a1n.d4e.1reptroagl.r,a2m00(8p)a.rtFoorf nit on
mate of the number of transition events. one set of analyses (below), we included only Araceae. For Fe
b
InthisstudyweusetheapproachofMayroseetal.(2010)to another, we included one exemplar each of the other families ru
a
reconstruct ancestral haploid chromosome numbers in of Alismatales (Stevens, 2001 onwards), using branch lengths ry
Amraaicnelyaet,roapilcaarlgefamanidly,oAldrafcaemaeilyhaovfe ma ohnigohcontuymledboernso.fFcohrroa- gorfo0u.p01uesxecdepbtyfoCruTsiomfiaenldoiaecteaael.(T(2o0fi1e1ld)iaan),dwhhaidchawnaesmthpeiroicuat-l 24, 20
1
mosomes counts, with 862 (26%) of their approx. 3300 branch length. 2
species counted, including at least one species for most of
the 117 genera (Petersen, 1989; Bogner and Petersen, 2007;
Chromosome number coding
Appendix; Supplementary Data Table S1 lists all species
with their n and/or 2n counts and the respective references). Totalnumbers of generaandspecies ofAraceaeweretaken
Two basic chromosome numbers have been suggested for fromthewebsiteCreatingaTaxonomiceScience(CATE;http://
Araceae. Larsen (1969) and Marchant (1973) argued for x ¼ www.cate-araceae.org/) and then updated by the Araceae spe-
7, with higher numbers derived through ancient polyploidiza- cialist Josef Bogner (see Acknowledgements). Of the 117 cur-
tion events or ascending dysploid series. In contrast, Petersen rently recognized genera of Araceae, 29 are monospecific
(1993) hypothesized a base number of x ¼ 14 because 2n ¼ (and hence can be coded unambiguously for chromosome
28isespeciallycommoninthefamily.Theformerhypothesis number), 19 have just two species, 31 have 3–10 species, 25
was put forward without the benefit of a phylogenetic frame- have11–50speciesand13have.50species.Araceaechromo-
work, but Petersen (and also Bogner and Petersen, 2007) some counts were compiled from original literature
Cusimano et al. — Maximum likelihood-inferred chromosome number changes 683
(Supplementary Data Table S1, available online), checking the Zosteraceae n¼6, 9, 10 (numbers from Stevens, 2001
generic assignment of each species against the current classifi- onwards). Those of these families with more than one
cationandforsynonymy.Chromosomenumbersforfourmono- number listed by Stevens were coded as polymorphic in all
typicgenerawerecontributedbyJ.BognerandE.Vosyka(see analyses. The above-described three coding schemes were
Acknowledgements)andarenewlyreportedhere:Filarumman- first run on the phylogenetic tree that included only Araceae
serichense Nicolson (M. Sizemore s.n., voucher in the herbar- and then on the tree that included the 13 outgroups, resulting
ium M), Hestia longifolia (Ridl.) S. Y. Wong & P. C. Boyce in six analyses (labelled A1–A6 in Table 1).
(J. Bogner 3003, M), Philonotion americanum (A. M. E.
Jonker & Jonker) S. Y. Wong & P. C. Boyce (J. Bogner
2911, M) and Pichinia disticha S. Y. Wong & P. C. Boyce Inference of chromosome numberchange
(P. C. Boyce s.n., M; Supplementary Data Table S1). One
genus was coded as unknown (X), namely the monotypic hapFlooridMcLhraonmdoBsoamyeesinanumpbheyrlso,gewneetriceliiendferoennctehseocfharonmceEstvroall Dow
Schottariella, the chromosomes of which have not been program v. 1.2 of Mayrose et al. (2010; http://www.zoology. nlo
counted. The presence of B chromosomes was not coded. a
Overall, our phylogenetic analysis includes 113 of the 117 ubc.ca/prog/ chromEvol.html). This implements eight models ded
accepted genera of Araceae, with 112 of them coded for of chromosome number change (Table 2), two more than fro
describedintheoriginalpaper.Themodelsincludethefollow- m
haploid chromosome number (Appendix). h
Chromosome numbers were coded in three ways to address ingsixparameters:polyploidization(chromosomenumberdu- ttp
the problem of genera with more than one chromosome plicationwithrater,‘demi-duplication’ortriploidizationwith ://a
ratem) and dysploidization (ascending, chromosome gain rate ob
number. First, we coded all reported numbers for each .o
l; descending, chromosome loss rate d) and two linear rate x
genus,regardlessoffrequencyindifferentspecies,butexclud- fo
ing odd numbers (Appendix, column 5; Supplementary Data aplalroawmientgertsh,elm1 taonddedp1e,nfdorotnhethdeyscpulroreidnitzantuiomnberarteosflcharnodmod-, rdjou
Table S1). This resulted in 55 genera coded as polymorphic. somes. Four of the models have a constant rate, whereas the rna
Our second coding scheme (‘reduced polymorphism’ coding) ls
other four include the two linear rate parameters. Both .o
ttoreoaktedinttoheacmcoosutntcotmhemofrneqausentchye aonfcedsiftrfaelresnttatenu(mAbpepresndainxd, modelsetsalsohaveanullmodelthatassumesnopolyploidi- arg/
column 7; Supplementary Data Table S1). For example, zation events. We fitted all models to the data, each with 10 t W
Lemnoideae have many different chromosome numbers, but 000 simulations to compute the expected number of changes ash
of the four transition types along each branch. The in
n¼20 is especially common (Landolt, 1986; Appendix, g
maximum number of chromosomes was set to 10× higher to
Supplementary DataTableS1). For generawithnumbers sug- then the highest number found in the empirical data, and the n U
gesting different ploidy levels, we used the lowest haploid n
minimum number was set to 1. The null hypothesis (no poly- iv
bcheroremdouscoemdetonutmwobesrta(ete.gs.(Achruromm).osPoomlyemnourpmhbisemrss)cpoeurldgetnhuuss ploidy) wastested with likelihood ratio tests using the Akaike ersity
or even a single haploid number, leaving 34 instead of 55 information criterion (AIC). , O
g(‘einneforarmweidth’ cpoodlyinmgo),rpwheictnouomkbinetros.aIcncoautnhtirmdocloedciunlgarscphheymloe- codWinegalsscoheramnea,nbauntaelyxscilsu,duisnigngCtahlelainfboercmauesdepooflymitsorupnhcilsemar- lin Lib
relationships in Araceae (Cusimano et al., 2011). For a final ra
genetic analyses for the genera Philodendron (Gauthier ry
et al., 2008), Biarum and Typhonium (Cusimano et al., sensitivity test, we again used the informed coding scheme - S
but the non-ultrametric ML phylogenetic tree from e
2th0e10e)a,rlayn-dbraasnscihginnegdstpheecsietasteto(cthheroemnotisroemgeennuusm. bTehre) fnouumnbdeirns Cinutshimearenmoaeintianlg. a(n2a0l1y1s)esi.nstead of the ultrametric tree used rials U
tahnudsPientfeerrsreend (w2e0r0e7)c.oTmhpiasrethdirwdiathpptrhooascehilneffetrjruedstbteynBgoegnneerar nit on
F
coded as polymorphic with maximally two states (Appendix, e
column 8; Supplementary Data Table S1, Supplementary RESULTS bru
a
Data Figs S1 and S2). In this third scheme, Lazarum, a Theresultsobtainedinthesixanalyses(A1–A6)aresummar- ry 2
genusof23specieswithafewchromosomecountsandinsuf- ized in Table 1. The three-parameter constant-rate model 4, 2
ficient phylogenetic information (Matthew Barrett, Botanic (Mc2), with the chromosome duplication rate equal to the 01
2
Gardens&ParksAuthority,WestPerth;personalcommunica- demi-duplicationrate,wasthebestexplanationoftheempiric-
tion, 2011) was coded as ‘unknown’ (X) because no ancestral al data in all analyses. All analyses rejected the null model of
haploidnumbercouldbeinferred.Inallcases,changesamong nopolyploidywithhighsignificance(P,0.999).Theinferred
character states (i.e. chromosome numbers) were assigned rates of change, chromosome numbers at nodes (and their
equal probability. probability) and numbers of events were similar regardless
The remaining families of Alismatales were coded as of which of the three schemes for polymorphism coding was
follows: Alismataceae n¼7, 8; Aponogetonaceae n¼12, applied. We therefore show the results obtained from
16, 19; Butomaceae n¼7, 8, 10, 11, 12; Cymodoceaceae Bayesian and ML analyses with the most conservative
n¼7, 8, 10, 14, 15; Hydrocharitaceae n¼notably variable; coding scheme, namely the one including all polymorphisms
Juncaginaceae n¼6, 8, 15; Maundiaceae only Maundia tri- andalloutgroups(Table1,A1;Figs1and2).Forcomparison,
glochinoides, no chromosome count reported; Posidoniaceae theresultsfromanalysisA6,withoutoutgroupsandthephylo-
n¼10; Ruppiaceae n¼8–12, 15; Potamogetonaceae n¼7, genetically informed coding (Appendix, column 8), can be
12, 14–18; Scheuchzeriaceae n¼11; Tofieldiaceae n¼15; found in Supplementary Data Figs S1 and S2.
684 Cusimano et al. — Maximum likelihood-inferred chromosome number changes
Resultsfromthesixanalyses(A1–A6)carriedouttoinferchromosomenumberchangesinAraceaeunderBayesianandmaximumlikelihoodoptimizationT1.ABLE Chromosomeno.atAraceaeChromosomeno.rangesat.PP.rootnodeAraceaerootnodeCodingschemeRateparametersEventsinferredwith05 Bayes:Bayes:nnTree:AllRed.BestBest;2ndbest;SumSumdlrmPPPPnPPnPPAnalysisoutgroupspoly.poly.Inf.modelLogLikAICLossesGainsDupl.Demi.ML ++............A1Mc2–21954454593969–9818414314318;01816;0161616–18058–1809++..........A2Mc2–23644789564063–112201151318;02617;0131617–1905110–20085++...........A3Mc2–24574973582057–1201011913918;02619;0121717–1905210–2009+............866321059318;03817;031717–19086A4–Mc2–196639915041866–+..........A5–Mc2–21324324536056–8720989418;04217;0231717–1909+..........A6–Mc2–22244507581054–94409710518;03719;0341817–19085 Onlythebest-fittingmodelsareshown.Tree(column2)referstowhetheroutgroupswereincludedornot;codingschemereferstohowgenerawithpolymorphichaploidchromosomenumberswerecoded.Allpoly.,allchromosomenumberpolymorphismcoded(scheme1);Red.poly.,reducedpolymorphismcoding(scheme2);Inf.,phylogeneticallyinformedcoding(scheme3).Bestmodel,Mc2¼¼rmdl(constantratemodelwithduplicationrateanddemi-duplicationrate;compareTable2);Logarithmiclikelihood(LogLik)andAICscores;rateparameters(chromosomelossrate,.¼¼rmPP.duplicationrate,demi-duplicationrate);frequencyofthefourpossibleeventtypeswithaposteriorprobability()05;haploidchromosomenumberinferredatthechromosomegainrate,PPPProotnodeunderBayesianoptimizationwiththerespective,andundermaximumlikelihood(ML).Thelastcolumnshowsthechromosomenumberrangeinferredfortherootnode,eachwithits. wiwwi[r1iBcp1cw(etdctifgn(acoAbbnaPaSflnntwI1(oAaDtMPdltZnnnhonyhheoaAron1nfhhvrnllnn0r0eaunreo2feuaePoaaiiiriaaoau¼¼newscfeeeoacfo.errmidsddeITTttttmd0..al4pn¼8yetsitmgenddalnns55ngoohhhuhtbttnnenphsaac¼te.u)ptrhe,o-ahrisateimisgt1cemmeooso11bvatectraeereofir,rnthd-hsilo¼ine8rss(hhe(rdbeee(huuu56dersaFnenrdannidtgotoiiee0AdoomsedAtdcAaiden–oowrs((intettmt::cegeglcaui)hee(dngs.(liaessAcrohngtfshs6p.a22hlhmi1lr34niipetsteooiPtahPsmseeyonenet(csea(.oooltaa)cohct199re.p380)glTweymmGT)eniy(shnbhgashmt;nPuPoaliuiurhrrrleb)ni4r).)sTSeiicrensthedotryie.-laiti,oamtd,,yrtntairkeetrrffi.ab¼eaeagtos)1hm((a¼gfgn¼aeoihubp4irpaOiebminihfePlarAPnetaBPsenbnerbcsod)etanrrmeedtllnrriygaldinoeno.roAiogPenmonroao1Ptlersuy,nura10c0ucenbia(daoeouwuddeiuouorlRalistocn7idnpyfIAslDfi)h.)1n.psiailtyodnehendttumpeea12ne2npmpsh;(nilsysrfehlihtah–clsdnomicsnoi1fl6ntoaitdresoret.,4a9ssirsiwczllasataegCbchtdeaarh1e,cuotgnite)r,)te,,meylntif))eiila¼peitaw)dahpr.nreFce.yhr,aaareaw9n.o:t,0amafrl-v.trodeSFtltogerirhhpnga¼eteysrdessbisidIdne.iiohanokeFTeidAnuiet(gtcoe8am1r.intlotoriiyiepuaAuIodetih)ahsienugeelhofdmcfariaipse.nhhnns54Mssnd,ltp1a.tt¼nhnrgeecalTtttm.uoo4,o,psodeg,oyssepsop0ldbthaf1d.fbgn1rwp¼pps,hciLhimwete1ametdonrhaAotbnslLTrrnec1oe–hus3(ohSnstohohewtveiysfr)ootoinu3ioleer0maharnosl5etemaareectleu2rg1ovmy02eeepaleamifhlmr.tptyoenhn.eeyoob(sneon2yomne0h8hri.m)es1nisdnrrmbbrAatnpmloivdr7iomnn-rc.lacsooBnubnrt,eehyo8dn4aeeeioeedtoeulubs7oncoceedc¼Fb2sttolelheloariud5(aew≤cr–,onunspmShttuyeaaoy(ro¼dhPiira)AvifsAa11iyipe.aonPFpntontprnerge0rstuodnm9eirSbrme0oos(o,ecub1olftgh12ofe,nedimlratadc..PbaprigPispih.dnsrfngiwde2muoase.bwz(e(csvit4,lmnF5huterce5piirfeoTsPiornc1or.he6taaaamtearg2¼aarpautadriisragilmlheetnolwc.1tdebyt,Aongofanoronteeosntues1alh¼apayCii(irvtaroehreisb4ei.iim.ndcommdeousosoAngpoir)vSesoes31mtTcwleoodA,atlh.eosdlen,nitpuaideopam1tibna3)h41teyolem2tage0etalhclhacefh,oan]sUurlps5rythfoh(lrsw4annsbl1s)ir.olooisaoaer1eePasaetsase.5finSareooa(osot)ctrlt74ii.dtbisdosnhrn,,toclvfAana.ieisPf)dfccoodtuumpdnlroy/rolnnpeAetoeorc(enwc¼argeAdalh1hee(nsyotpaei30oms0fypccagdghahrrodPungicenA¼na1poram8ito1aemaswe1)nr.mc.odro)lyghoenielt1rSoifg5F()da3ueesrreea6.)ndsu(nolmeeDnfsP.olhigmafsd4l2e,du0ti)7t9e9(dsiocr)ttryrr(uehntgs;rccaeal3Ts.sP1ho,t9rpi.aao9A8r8irioPSaa–e(goea3pr5tphsatghahbrnsi(ncee.attd)aPm.phhm.r..vrovino9ieisrTttPnah07l5slir,3rt11reddfdg5snepnmnihloeesoueoaeaPnoou)rh(b.en;eat68e¼aope4AtncAidmmn¼f(eemmpoFdi0irmwtbgni(d((aA.i.itenFagn¼nnemo2nlhhtavtovuin92sAnnhAAe.il1aesibietvnnooadb9ihuui≥s)g1neaeeee1nmzu0cirtffem¼rt))g¼ret6ed41((pssoy9npemminee0.6chal,b.wwea¼AAv.n08oota,y)b))srrrrhtlatlt./tttraaonahiSaobbiin136oorr.18t6mmPtttAeerrh,ewaijie312nn81n1onsnchahheeeeroonoieettti1a4i22osrr..71rassP))))y4ddh55d5nonoon4gnd6ddeeeeeeaeeeeessrrrr--ft;:;;t.,,.,,,, at Washington University, Olin Library - Serials Unit on February 24, 2012http://aob.oxfordjournals.org/Downloaded from
Cusimano et al. — Maximum likelihood-inferred chromosome number changes 685
0·5 10 00··76 ZPoostatemraocgeeatoen 9aceae 7
0·8 10 10 10 10 PosidoniaRCceuyampepo i1da0oceceaaec 1e0ae 10
12 10 1·1 JuncaginaMceaauen d8iaceae 10
0·5 8 0·8 11 00··86 ASlcishmeuactahczAeeaprieoa nc8eoagee t1o1naceae 12
9 8 0·7 8 HBuydtormocahcaeraitea c8eae 80·54·90·730·6·90·60·7 2·3 GToymfienldoisatacechayes 1 254
9 17 1·9 15 20·5·6 150·9 OSLyyrsomincpthiluoitmcoan 1r p13u4s 15
1·2 1·5 Spirodela 20
A1raceae 18 Lemnoideae 22 0·5 201·921 0·6 211·30·900··77 WWLALaenoonmtllhdffnffuoiieaarlitl uli22aam00 22 1103
Pothoideae 14 1·7 12 PPeodthicoesl l1a2rum 12 D
12 Pothoidium 12 o
Spirodela1· 3 clade 15 15 105·8 14141154 0·17 SHARSHpltoheleoalotonestdchorcoohishpsplpeahpsemmyairsltmoly hu1nsaam4 et 1i1 o5144n52 14 wnloade
Rhaphidophora clade 15 1303300 MSARcnhoiaanndpsdehtaenipdrdasoru pu3shm0 o3 r03a0 30 d from
True Ar0a·8ceae 15 1L·4asioideae 131313310330 CPALEALaaynpmysscraipytiinpomadrohes ro1spyimur3pleplmnaurhumtm ha3maa 011 31133033 http://aob
n = Podolasia clade0·6 14 14 13 2·403·8011···877000···999 1711133311331 AMADUPDAZGSCangontrrraooyaoamuldalnnallcscoobopatoioporocpilhtianananohccsysesttthupailihis ilmuoaaolue2ac mi1r pst4ada 1od2 7s e s13ni216ias s7 013 1 21174433 .oxfordjournals.o
678910 Aroidea1e4 14 14202102·18 0·270 3212202002·21200 CCZPAFHPANauhgsnueoenreilclrmplactcotuhahaeoeaddodtsddslheomooieotyanhimassrat dyu i2 cnse22rdm1h oen012rinos aa02 s 21021m0806e 20 at Washingrg/
11 1·1 19 1·9 GBoeganruemra 1177 Zantedeschia ton
1123 Spathicarpeae 1717117717 GIMDnioaceranfgfrgeouonnmnbid iaa1icu 71hm7ia 1 177 clade Univ
1111145678 14 0·7 13 1·117017·11877181717 CLSATSCSPaapspyhrryogtcnaaielpaecoattrhhtatnnnooiiaeraocdscunlnattrolmiiatoodgrrhp synr m1e1aapn 7u17a ae11m3 d 1871i 7x81 717 ersity, Olin L
13 Piptospatha 13 ib
1290 113313 BBAuaricdkeoaparuh 1ma3l a1n3dra 13 rary
2212 Philonotion clade 14 1·6 0·6 113313 PSSCPhccsahheyllmioaust dmat1aot8aarditreroualglmcaloo 1n1tt3t3iisu m13 13 - Seria
23 13 Amorphophallus 13 ls
2222246789 14 1414 13131311323·205·1·283113391·019··77 JHUFZSCXSCaiocayalhalesamannpllaoaarpgtauirrhdrcholuoumioianiunmsmssr emipp1puo a71a4am1m 1t1t 3hh1 a31aa1 31 11334 Unit on Febru
30 2·813 ZAommbircoasripneal l1a1 13 ary
45572696 Ambrosina clade 14 11044·15282184271127 CCATAPPPyrreiraosoiopslrlttlptlhieaaeaahotrnpr ouy1undhgmt4moroyyad tn n1o 1o 5e4n24r6 u 722m77 56 Tclyapdheonodorum 24, 2012
84 14 14 ACroiolopcsaiss i1a4 14
14 Remusatia 14
14 14 Steudnera 14
Alocasia 14
Events inferred with PP > 0·5 Arisaema 14
1414 1·30·6Pinellia 14
CChhrroommoossoommee gloasisness : : 8 9·48·1 14141414111·6 TSTLayhapzeuahrriorooumnpmihau otm8un4m u1m 31 312
Duplications : 14·3 1144 1·1 HEmeliicnoiudmic e1r4os 14 Areae
Demi-duplications : 14·3 1144 BAiraurmum 1 413
14 Dracunculus 14
FIG. 1. Chromosome numberevolutionin Araceae inferred under Bayesian optimization, with outgroups included and all polymorphic chromosomestates
coded (analysis A1 in Table 1). Pie charts at nodes and tips represent the probabilities of the inferred chromosome number(s); numbers inside charts have
the highest probability. Numbers at the tips are chromosome numbers inferred with the highest probability, i.e. the inferred ancestral haploid chromosome
numberforeachgenus.Numbersabovebranchesrepresenttheinferredfrequencyofthoseofthefourpossibleevents(gains,losses,duplicationsanddemi-
duplications)thathadaprobability.0.5.Thecolourcodingofchromosomenumbersandeventtypesisexplainedintheinsets.
686 Cusimano et al. — Maximum likelihood-inferred chromosome number changes
Potamogetonaceae
10 Zosteraceae
10 Cymodoceaceae
10 10 10PosidoniaRceuapepiaceae
10 Maundiaceae
12 Juncaginaceae
12 Aponogetonaceae
8 Scheuchzeriaceae
Alismataceae
8 8 8 HBuydtormocahcaeraiteaceae
Tofieldiaceae
Gymnostachys
16 Orontium
8 15 15 SLyysmicphloitcoanrpus
Spirodela
Lemnoideae 21 21 LLaenmdnoaltia
Araceae16 21 21 WWoollffffiiealla
Anthurium
Spirodela clade 15 Poth1o5ideae 14 15 111442112154 SHARSHPPPlpteoohoeleoadttoltonhhestidcchooorcooehissihpdspllpelahipasuemmyairmrsultmolyhumnsaametion Downloaded
15 30 ARnhaadpehniddorpuhmora fro
True Araceae 15 Rhaphidophora clade 33003030 EAMSpmcoiinnpysddretareimpurasmnuusm httpm
Lasioideae 1133 13 CALLaanyssratiipomahsopyrleplurhmmaa ://aob
13 Pycnospatha .o
13 Podolasia x
13 Urospatha fo
Podolasia clade 14 14 13 17131133 AMDDAZGSCanntrraooyaamualnnllccoobpatioorocpihtiannohccsystthupailiisluoaoluecmirpstdaodsesniiass rdjournals.org
14 20 AAggllaaoodnoermuam a/
AroideaeS 1p4at1h3i2c20a109r2p0ea2e0 12171721120770 2200 CCZPFHPANIMDGBnauhnsueooeieacenreiclrgmpaanlctfcothuafnhrrgaeeeaduuoddetosdnslhemmmooorinetbyaanhimaissataadyicnsecrdhoehnrinsoiaaasme Zclaandteedeschia t Washington Univers
14 17171117771717 ATSSGPCSaspphyrooctnaairelacgoattrhhatonnoiiaernocdsunlaittrldmiiatoogrhipsunmeapmuaamdix ity, Olin Lib
Philonotion clade 14 1313 11311318313133 PSBPSBACLahcuircarpiyghhycdktpemieoooastnpstaromatupahactmaanaaroittdrleroahuyrlgnalamnaldeortatis rary - Serials U
14 14 1313 1133 JHSCCAPaymsaaasnepllaloaaguadrrodlpuiinunohmemidourpmahcaolnlutsium nit on Feb
131133 139 UFSCXicalhleaanlaorpturrhhuomoimsssppoaamtthhaaa ruary 2
14 1313 ZZoommiiccaarrppealla 4, 2
14 AArmisbarrousmina 01
14 Peltandra 2
28 Typhonodorum Typhonodorum
28 Arophyton
Ambrosina clade 14 14 27 27 CCPPiraosorltltlieeaatprouhgmyytnoen clade
Ariopsis
1414 14 1144 SRCteoemluodcuanssaeitraiaa
Alocasia
1414 PAirnisealleiama
14 14 TLahzeariroupmhonum
14 Typhonium
14 Sauromatum
14 Eminium Areae
14 Helicodiceros
1144 BAiraurmum
14 Dracunculus
FIG.2. ChromosomenumberevolutioninAraceaeinferredundermaximumlikelihoodoptimization,withoutgroupsincludedandallpolymorphicchromosome
statescoded(analysisA1inTable1).
Cusimano et al. — Maximum likelihood-inferred chromosome number changes 687
TABLE 2. The eight models of chromosome number evolution rearrangements haveoccurred aftergenome doubling,perhaps
implemented in the software of Mayrose et al. (2010), indicating especiallyfollowinghybridization(Hayasakietal.,2000;Lim
the considered parameter estimates (d, l, r, m, d, l), the et al., 2008; Peruzzi et al., 2009; Tu et al., 2009).
1 1
number of parameters included, and, in the case of m, with In general, basic chromosome numbers inferred according
whichcondition to Langlet’s (1927) approach, as the lowest detectable or
somehow calculated haploid number within a group of
Model d l r m d1 l1 No.ofparameters related taxa, will be low, simply because of the way they are
arrived at (see Introduction for Langlet’s original example).
Mc1 + + + – – – 3 For Araceae, the hypothesized ancestral numbers were x¼
Mc2 + + + r¼m – – 3
Mc3 + + + r=m – – 4 14 or x¼7 (Larsen, 1969; Marchant, 1973; Petersen, 1993).
Mc0 + + r¼0 m¼0 – – 2 The present study instead inferred an ancestral haploid
Ml1 + + + – + + 5 numberofn¼16(underML)orn¼18(withBayesianinfer- Do
Ml2 + + + r¼m + + 5 ence) and, moreover, an evolutionary trend from higher to wn
Ml3 + + + r=m + + 6 lo
Ml0 + + r¼0 m¼0 + + 4 ltoowkeerenpuminbemrsi,ndraththeartthnaonnetheofotthheer ewaaryliearrosutunddi.eOsn(eLanreseedns, aded
Mcindicatesmodelswithconstantrates,andMlmodelsthatincludelinear 1969; Marchant, 1973; Petersen, 1993) included Lemnoideae from
rateparameters(d1,l1).Zeroindicatestherespectivenullmodel. irnangAeracoefaec,haromtaxoosonmomeicnudmifbfeerresncfeoutnhdat ignreaetalyrlya-fdfievcetrsgitnhge http
clades (Figs 1 and 2). It is also likely that the high frequency ://a
o
b
of 2n¼28 in the well-counted unisexual Aroideae unduly .o
x
duplication event on the branch leading to the Typhondorum influenced the hypotheses about x being 7 or 14. Finally, the fo
clade (from n ¼ 14 to n ¼ 28). earlierhypotheses weredevelopedwithouttherelativelycom- rdjo
Results of the two additional analyses (inclusion/exclusion plete and solid phylogenetic information that is available urn
of Calla; ultrametric or non-ultrametric trees) did not yield today. als
results substantially different from those obtained in analysis Nevertheless,anyinferencesaboutcharacterevolutionfrom .org
A6 and shown in Supplementary Data Fig. S1. Model Mc2 a taxon sampling of just 112 representatives, however well a/
remained the best-fitting model, and chromosome number coded their states may be, must be regarded with caution. t W
a
reconstructions at nodes and change rates were similar. Every genus with more than one species must have its own, sh
in
perhapscomplex,historyofcytogeneticchange.Itisalsocon- g
to
DISCUSSION ceivablethatdysploidyratesmightchangeindifferentpartsof n U
thetree(e.g.incladesoftaxalivingindifferentenvironments) n
iv
Theresultspresentedhereprovideanexampleofthepowerof and that relatively derived and rapidly radiating clades, ers
ML-based or Bayesian inference of chromosome number perhaps with frequent hybridization, might have different ity
changes. The new approach, which distinguishes (and separ- rates of polyploidization than older, genetically isolated , O
ately infers) chromosome gains, losses, polyploidization and groups. The phylogenetically informed coding scheme (our lin L
demi-ploidization, not only reconstructs numbers at particular scheme three) may be the best way of coding ancestral ib
ra
phylogeneticnodes,butalsoinfersratesofchangethroughout haploid chromosome numbers in larger clades (here genera), ry
thephylogenetictree.Equallyimportantly,BayesianPPsyield an idea that could be tested by cytological work in small - S
e
a statistically well-understood measure of confidence in the genera with solid phylogenetic hypotheses, such as Arum ria
results.Mostpreviousancestralchromosomenumbers,incon- (e.g. Esp´ındola et al., 2010). ls U
t(reaxsat,mphlaevseanbdeecnritiicnaflerdriesdcuswsiiothnoiunt Scooltnifisdeetnacle., 2a0ss0e5s)s.mTehnet somGeivefnustihoensinf(enrereudtrahlilgyhtaenrcmeestdral‘lhoaspselos’idinnumthbeerms,ocdherolsmoo-f nit on
F
experiments we carried out with the different coding Mayrose et al., 2010) must have been common during evolu- e
b
schemes for genera polymorphic for chromosome number tion of Araceae. This hypothesis now needs to be tested. ru
a
revealed surprising robustness of the states inferred at interior Large chromosomes in Araceae, with distally positioned cen- ry 2
nodes, although as expected the inclusion orexclusion of out- tromeres, may be the result of fusion between smaller meta- 4, 2
groups (in our case 13 families) affected the number inferred centric chromosomes (Petersen, 1993). If so, one expects to 01
2
for the basal-most node, albeit only slightly (Table 1). The find interstitial telomeric sites. With probes, using primer
results of the present study further confirm that model-based pairs homologous to the basic plant telomeric repeats, one
chromosome inferenceworks well even with large data matri- can visualize these regions (Ijdo et al., 1991;
ces;thelargestofthefourmatricesanalysedbyMayroseetal. Weiss-Schneeweiss et al., 2004). Such chromosome prepara-
(2010) had 107 terminals, and the present tree had 126. tions are now being carried out in our laboratory on
Chromosome fusion (loss) appears to be the predominant Typhonium species with suspected chromosome fusion (pre-
pattern in the evolution of chromosome number in Araceae; dicted from high or low chromosome numbers in species of
polyploidization events are much less frequent and apparently known phylogenetic relationships).
occurredmainlytowardsthetipsofthetree.However,ancient The results of the present study suggest that quantitative
polyploidization events may be harder to detect than recent methods for inferring ancestral haploid numbers should
ones, because of the genomic restructuring that follows poly- replaceinferencesthatrelyonalgebraicallyfindingthegreatest
ploidization.Onlydetailedstudies,perhapsinvolvingchromo- common factor for a series of numbers or on interpreting the
some painting techniques, will reveal how rapid intergenomic lowest available haploid count as the ancestral condition.
688 Cusimano et al. — Maximum likelihood-inferred chromosome number changes
The new approaches also yield a measure of statistical confi- unravelling its evolutionary history. Botanical Journal of the Linnean
dence or estimates of the rates of polyploidization, fusion or Society163:14–32.
FrenchJC,ChungMG,HurYK.1995.ChloroplastDNAphylogenyofthe
fission, We suggest that the concept ‘x’, which sets botanists
Ariflorae. In: Rudall PJ, Cribb P, Cutler DF, Humphries CJ. eds.
apart from zoologists, be retained only in the context of
Monocotyledons: systematics and evolution, vol. 1. Kew, UK: Royal
small species groups in which the history of polyploidy is BotanicGardens,255–275.
known in detail (Vanzela et al. 2003). GauthierMPL,BarabeD, BruneauA. 2008.Molecular phylogenyof the
genus Philodendron (Araceae): delimitation and infrageneric classifica-
tion.BotanicalJournaloftheLinneanSociety156:13–27.
SUPPLEMENTARY DATA Goldblatt P. 1980. Polyploidy in angiosperms: monocotyledons. In: Lewis
WH. ed. Polyploidy: biological relevance. New York: Plenum Press,
Supplementary data are available online at www.aob.oxford- 219–239.
journals.org and consist of the following. Table S1: chromo- GrayumMH.1990.EvolutionandphylogenyoftheAraceae.Annalsofthe
MissouriBotanicalGarden77:628–697. D
some counts for species of Araceae with references. Figure GuerraM.2008.Chromosomenumbersinplantcytotaxonomy:conceptsand ow
S1: chromosome number evolution in Araceae inferred under implications.CytogeneticandGenomeResearch120:339–350. nlo
Bayesian optimization, with phylogenetically informed HayasakiM,MorikawaT,TarumotoI.2000.Intergenomictranslocationsof ad
coding and outgroups excluded. Figure S2: chromosome polyploidoats(genusAvena)revealedbygenomicinsituhybridization. ed
number evolution in Araceae inferred under maximum likeli- GenesandGeneticSystems75:167–171. fro
Heslop-Harrison JS. 1992. Molecular cytogenetics, cytology and genomic m
hanododouotgprtoimupizsaetixocnl,udwedit.h phylogenetically informed coding Hiluc(KgormWasps.aer2si0)s.o0An4s.u.sPHtrhaeylrlieoadgnietaJnoseut1irc1ns6a:aln9od3f–Bc9ho9rtoa.mnyos5o2m:1al3–ev2o2l.ution in the Poaceae http://ao
IjdoJW,WellsRA,BaldiniA,ReedersST.1991.Improvedtelomeredetec- b.o
ACKNOWLEDGEMENTS tion using a telomere repeat probe (TTAGGG)n generated by PCR. xfo
NucleicAcidsResearch19:4780. rd
WearegratefultoMichael Fayandtwoanonymousreviewers ImaiHT,SattazY,WadayM,TakahatazN.2002.Estimationofthehighest jo
u
for the constructive suggestions, J. Bogner and L. Nauheimer chromosome mumber of eukaryotes based on the minimum interaction rn
a
fcohrectkirienlgesSsluyppdliesmcuesnstianrgy ATarabcleeaSe1evfoorlustyionnonwymithouussoarntdranfosr- JiaotYh,eoWryi.ckJoetutrnNaJl,oAfyTyhaemorpeatilcaaylaBmioSlo,geyta2l1.72:06111–.7A4n.cestralpolyploidyin ls.org
ferred species names, and E. Vosyka for help with chromo- JohnsseoendMplAanTt,sKanedntaonngiAoYsp,eBrmens.nNetattuMreD4,7B3r:a9n7d–h1a0m0.PE.1989.Voanioala at W/
some counts. This work was supported by the Deutsche gerardiihasthehighestknownchromosomenumberinthemonocotyle- a
s
Forschungsgemeinschaft, RE 693/7-1. dons.Genome32:328–333. hin
Landolt E. 1986. Biosystematic investigations in the family of duckweeds g
(Lemnaceae) (vol. 2). The family of Lemnaceae – a monographic ton
study.vol.1.Zu¨rich:Vero¨ffentlichungendesGeobototanischenInstituts U
LITERATURE CITED n
derETH,StiftungRu¨bel,Heft71. iv
e
Bennett MD. 1998. Plant genome values: how much do we know? LangletO.1927.Beitra¨gezurZytologiederRanunculaceen.SvenskBotanisk rs
ProceedingsoftheNationalAcademyofSciences,USA95:2011–2016. Tidskrift21:1–17. ity
Blo¨ch C, Weiss-Schneeweiss H, Schneeweiss GM, et al. 2009. Molecular Larsen K. 1969.Cytologyof vascularplants: III.A studyof aroids. Dansk , O
phylogeneticanalysesofnuclearandplastidDNAsequencessupportdys- BotaniskArkiv27:39–59. lin
ploidandpolyploidchromosomenumberchangesandreticulateevolution LimKY,SoltisDE,SoltisPS,etal.2008.Rapidchromosomeevolutionin Lib
in the diversification of Melampodium (Millerieae, Asteraceae). recently formed polyploids in Tragopogon (Asteraceae). PLoS One 3: ra
BognMerolJe,cuPleatrerPshenyloGg.en2e0t0ic7s.TanhdeEchvroolmutoiosnom53e:n2u2m0b–e2r3s3o.fthearoidgenera. LuoeM33C5,3D.hetatlpa://KdxR.d,oAi.korhgu/1n0o.v1a37E1D/j,ouArnkahlu.pnoonvea.0A00R3,3e5t3.al. 2009.Genome ry - S
e
Aroideana30:82–90. comparisons reveal a dominant mechanism of chromosome number ria
BowersJE,ChapmanBA,RongJ,PatersonAH.2003.Unravellingangio- reduction in grasses and accelerated genome evolution in Triticeae. ls
spermgenomeevolutionbyphylogeneticanalysisofchromosomaldupli- Proceedings of the National Academy of Sciences, USA 106: U
n
cationevents.Nature422:433–438. 15780–15785. it o
CabreraLI,SalazarGA,ChaseMW,MayoSJ,BognerF,DavilaP.2008. LysakMA,BerrA,PecinkaA,SchmidtR,McBreenK,SchubertI.2006. n
Phylogenetic relationships of aroids and duckweeds (Araceae) inferred Mechanisms of chromosome number reduction in Arabidopsisthaliana Fe
b
from coding and noncoding plastid DNA. American Journal of Botany andrelatedBrassicaceaespecies.ProceedingsoftheNationalAcademy ru
95:1153–1165. ofSciences,USA103:5224–5229. ary
CroslandMWJ,CrozierRH.1986.Myrmeciapilosula,anantwithonlyone MarchantCJ.1973.ChromosomevariationinAraceaeIV:fromAcoreaeto 2
4
pairofchromosomes.Science231:1278–1281. Lasieae.KewBulletin28:199–210. , 2
CuiL,WallPK,Leebens-MackJH,etal.2006.Widespreadgenomedupli- Mayo SJ, Bogner J, Boyce PC. 1997. The genera of Araceae. Kew, UK: 01
cationsthroughoutthehistoryoffloweringplants.GenomeResearch16: RoyalBotanicGardens. 2
738–749. MayroseI,BarkerMS,OttoSP.2010.Probabilisticmodelsofchromosome
Cusimano N, Barrett M, Hetterscheid WLA, Renner SS. 2010. A phyl- numberevolutionandtheinferenceofpolyploidy.SystematicBiology59:
ogenyoftheAreae(Araceae)impliesthatTyphonium,Sauromatumand 132–144.
the Australian species of Typhonium are distinct clades. Taxon 59: NylanderJAA,WilgenbuschJC,WarrenDL,SwoffordDL.2008.AWTY
439–447. (arewethereyet?):asystemforgraphicalexplorationofMCMCconver-
Cusimano N, Bogner J, Mayo SJ, et al. 2011. Relationships within the genceinBayesianphylogenetics.Bioinformatics24:581–583.
Araceae:comparisonsofmorphologicalpatternswithmolecularphyloge- Oginuma K, Munzinger J, Tobe H. 2006. Exceedingly high chromosome
nies.AmericanJournalofBotany98:654–668. number in Strasburgeriaceae, a monotypic family endemic to New
DrummondAJ,RambautA.2007.BEAST:Bayesianevolutionaryanalysis Caledonia.PlantSystematicsandEvolution262:97–101.
bysampling trees. BMC Evolutionary Biology 7: 214. http://dx.doi.org/ Otto SP, Whitton J. 2000. Polyploidy incidence and evolution. Annual
10.1186/1471-2148-7-214. ReviewofGenetics34:401–437.
ErbenM.1996.Thesignificanceofhybridizationontheformingofspeciesin PeruzziL,LeitchIJ,CaparelliKF.2009.Chromosomediversityandevolu-
thegenusViola.Bocconea5:113–118. tioninLiliaceae.AnnalsofBotany103:459–475.
Esp´ındola A, Buerki S, Bedalov M, Ku¨pfer P, Alvarez N. 2010. New Petersen G. 1989. Cytologyand systematics of Araceae. Nordic Journal of
insights into the phylogenetics and biogeography of Arum (Araceae): Botany9:119–166.
Cusimano et al. — Maximum likelihood-inferred chromosome number changes 689
PetersenG.1993.ChromosomenumbersofthegeneraofAraceae.Aroideana StuessyTF.2009.Planttaxonomy:thesystematicevaluationofcomparative
16:37–46. data,2ndedn.NewYork:ColumbiaUniversityPress.
Posada D, Crandall KA. 1998. MODELTEST: testing the model of DNA TuY,SunJ,GeX,LiZ.2009.Chromosomeelimination,additionandintro-
substitution.Bioinformatics14:817–818. gression in intertribal partial hybrids between Brassica rapa and Isatis
RamseyJ,SchemskeDW.2002.Neopolyploidyinfloweringplants.Annual indigotica.AnnalsofBotany103:1039–1048.
ReviewofEcologyandSystematics33:589–639. TzvelevNN,ZhukovaPG.1974.Ontheleastmainnumberofchromosomes
RavenPH.1975.Thebasesofangiospermphylogeny:cytology.Annalsofthe infamilyPoaceae.BotanicheskijZhurnal59:265–269(inRussian).
MissouriBotanicalGarden67:724–764. Uhl CH. 1978. Chromosomes of Mexican Sedum II. Section Pachysedum.
SinghBD,HarveyBL.1975.Selectionfordiploidcellsinsuspensioncultures Rhodora80:491–512.
ofHapplopappusgracilis.Nature253:453. VanzelaALL,CuadradoA,GuerraM.2003.Localizationof45SrDNAand
SokolovskayaAP,ProbatovaNS.1977.Ontheleastmainnumberofchromo- telomericsitesinholocentricchromosomesofRhynchosporatenuisLink
somes(2n¼4)inColpodiumversicolor(Stev.)Woronow.Botanicheskij (Cyperaceae).GeneticsandMolecularBiology26:199–201.
Zhurnal62:241–245(inRussian). Weiss-Schneeweiss H, Riha K, Jang CG, Puizina J, Scherthan H,
SoltisDE,SoltisPS,EndressPK,ChaseMW.2005.Phylogenyandevolu- Schweizer D. 2004. Chromosome termini of the monocot plant D
tionofangiosperms.Sunderland,MA:Sinauer. Othocallis siberica are maintained by telomerase, which specifically ow
SoltisDE,AlbertVA,Leebens-MackJ,etal.2009.Polyploidyandangio- synthesises vertebrate-type telomere sequences. The Plant Journal 37: nlo
spermdiversification.AmericanJournalofBotany96:336–348. 484–493. a
d
Stebbins GL. 1971. Chromosomal evolution in higher plants. London: Wood TE, Takebayashi N, Barker MS, Mayrose I, Greenspoon PB, ed
Arnold. RiesebergLH.2009.Thefrequencyofpolyploidspeciationinvascular fro
StevensPF.2001onwards.AngiospermPhylogenyWebsite.Version9,June plants. Proceedings of the National Academy of Sciences, USA 106: m
2008[andmoreorlesscontinuouslyupdatedsince]. 13875–13879. http
://a
o
b
.o
x
fo
rd
jo
u
rn
a
ls
.o
rg
a/
t W
a
s
h
in
g
to
n
U
n
iv
e
rs
ity
, O
lin
L
ib
ra
ry
- S
e
ria
ls
U
n
it o
n
F
e
b
ru
a
ry
2
4
, 2
0
1
2
690 Cusimano et al. — Maximum likelihood-inferred chromosome number changes
APPENDIX
The117generaofAraceaewithnumberofspecies,numberandpercentageofspecieswithchromosomecounts,diploidchromo-
somenumbersandcodedancestralhaploidchromosomenumbersinthethreecodingschemesusedinthisstudy(seeMethods).
X ¼ unknown.
Counteddiploid Reduced
Spp. Spp. chromosome Allpolymorphic polymorphic Informed
Genera number counted % numbers2n¼ n¼ n¼ n¼
1 Aglaodorum 1 1 100 40 20 20 20
2 Aglaonema 23 6 26 14,40,100 7,20,50 7,20,50 20
3 Alloschemone 2 1 50 84 42 42 42 D
4 Alocasia 107 17 16 24,26.28,40,42,56,68,70, 12,13,14,20,21,28, 12,13,14,20,21,28, 14 ow
84 34,35,42 34,35,42 n
lo
5 Ambrosina 1 1 100 22 11 11 11 a
d
6 Amorphophallus 196 47 24 26,28,39 13,14 13,14 13 ed
7 Amydrium 5 2 40 60 30 30 30 fro
8 Anadendrum 11 3 27 60 30 30 30 m
9 Anaphyllopsis 3 1 33 26 13 13 13 h
1101 AAnnachpohmylalunmes 26 23 10500 2460 1230 1230 1230 ttp://a
o
12 Anthurium 903 171 19 14,20,24,26,28,29, 7,13,15,17,18,30 7,13,15,17,18,30 15 b
30+Bs,34,36,40,48,49, .ox
56,60,84,approx.90,approx. ford
124 jo
u
13 Anubias 8 8 100 48 24 24 24 rn
14 Apoballis 12 6 50 26,39,56 13,28 13 13 als
15 Aridarum 10 4 40 24,26 12,13 12,13 12,13 .o
1167 AArriiosapesmisa 1520 971 6550 2208,,2844,,2866,28,32,42,48,52, 1140,,4122,,4133,14,16,21, 1140,12,13,14,16,21, 1144 at Wrg/
56,64,70,72,84,112,140, 24,26,28,32,42,56, 24,26,28,32,42,56,
a
168 70,84 70,84 sh
18 Arisarum 4 2 50 14,28,42,52,56 7,14,21,26,28 7,14,21,26,28 14 ing
19 Arophyton 7 6 86 38,40,54,approx.76 19,20,27 19,20,27 19 to
n
20 Arum 29 26 90 28,29,30,42,56,63,70,84 14,15,21,28,35,42 14 14 U
21 Asterostigma 8 2 25 34 17 17 17 niv
22 Bakoa 2 2 100 26 13 13 13 ers
23 Biarum 21 12 57 16,18,22,24,26,32,36,40, 8,9,11,12,13,16,18, 8,9,11,12,13,16,18, 13 ity
74,approx.96,98,108 20,37,49,54 20,37,49,54 , O
24 Bognera 1 1 100 34 17 17 17 lin
25 Bucephalandra 3 3 100 26 13 13 13 L
26 Caladium 12 6 50 19,22,26,28,30 11,13,14,15 11,13,14,15 13,14 ibra
27 Calla 1 1 100 36,54,60,72 18,27,30,36 18 18 ry
28 Callopsis 1 1 100 36 17 17 17 - S
29 Carlephyton 3 3 100 54,108 27,54 27 27 e
30 Cercestis 10 6 60 approx.36,42 21 21 21 rials
31 Chlorospatha 28 2 7 26 13 13 13 U
3323 CCoollloectaosgiyane 116 51 10301 2464,,2486,,3504,36,38,42,44,46, 1232,,1243,,1257,18,19,21, 1237,14,15,18,19,21, 1247 nit o
n
48,52,58,84,116 22,23,24,26,42,58 22,23,24,26,42,58 F
e
34 Croatiella 1 1 100 34 17 17 17 b
35 Cryptocoryne 60 64 107 20,22,28,30,33,34,36,42, 10,11,14,15,17,18, 10,11,14,15,17,18, 17,18 rua
54,56,66,68,70,72,85,88, 21,27,28,33,34,35, 21,27,28,33,34,35, ry 2
90,102,112,approx.132 36,44,45,51,56 36,44,45,51,56 4
36 Culcasia 24 9 38 approx.40,42 21 21 21 , 20
37 Cyrtosperma 12 4 33 24,26 12,13 12,13 13 12
38 Dieffenbachia 57 14 25 34,36,40,44,68 17,18,20,22,34 17 17
39 Dracontioides 2 1 50 26 13 13 13
40 Dracontium 24 5 21 26 13 13 13
41 Dracunculus 2 2 100 28,32 14,16 14 14
42 Eminium 9 3 33 28 14 14 14
43 Epipremnum 15 3 20 60,70,84 30,35,42 30,35,42 30
44 Filarum 1 1 100 28 14 14 14
45 Furtadoa 2 1 50 40 20 20 20
46 Gearum 1 1 100 34,68 17,34 17 17
47 Gonatopus 5 4 80 34,approx.68 17 17 17
48 Gorgonidium 8 3 38 34 17 17 17
49 Gymnostachys 1 1 100 48 24 24 24
50 Hapaline 8 2 25 26,28 13,14 13,14 13,14
Continued