phylogenetic relationships and chromosome number evolution

138

Systematic Botany (2006) 31(1) pp 138ndash150q Copyright 2006 by the American Society of Plant Taxonomists

Phylogenetic Relationships and Chromosome NumberEvolution in Passiflora

A KATIE HANSEN1 LAWRENCE E GILBERT1 BERYL B SIMPSON1 STEPHEN R DOWNIE2ARMANDO C CERVI3 and ROBERT K JANSEN1

1Section of Integrative Biology The University of Texas Austin Texas 787122Department of Plant Biology University of Illinois at Urbana-Champaign Urbana Illinois 61801

3Departamento de Botanica Universidad Federal do Parana Curitiba Parana Brasil 81531990

Communicating Editor Matt Lavin

ABSTRACT The phylogenetic relationships and chromosomal evolution of the diverse tropical genus Passiflora (Passiflor-aceae) are explored using data from two chloroplast markers the rpoC1 intron and the trnLtrnT spacer region A surveyof the presence or absence of the rpoC1 intron in 136 species representing 17 of Killiprsquos (1938) 22 subgenera of Passiflora andfour other genera in the Passifloraceae revealed intron losses in 46 taxa A minimum of two losses were confirmed by aparametric bootstrap approach on sequence data from the trnLtrnT chloroplast non-coding region for 61 taxa The resultsof phylogenetic analyses of the trnLtrnT sequence data support the reduction of Killiprsquos 22 subgenera to four as proposedin a new classification system by Feuillet and MacDougal (2004) The monophyly of the lsquon56rsquo and lsquon59rsquo chromosomal andmorphological groups is strongly supported In addition these data indicate that Passiflora biflora or closely related speciesis the likely continental sister to the red-flowered Caribbean taxa while P auriculata is weakly supported as the New Worldsister to the Old World Passifloras Finally character optimization of chromosome numbers on the phylogenetic tree supportsx512 as the base chromosome number for Passiflora

Passiflora L (Passifloraceae) is a genus of more than530 species of climbing herbs trees and woody lianasThey are a conspicuous part of the Neotropical floraand their distribution extends from southern Argenti-na northward throughout Central America and Mex-ico into the southern United States In addition 20 spe-cies are restricted to the Old World in the tropical andsub-tropical regions of southeast Asia Australia andNew Zealand The wide distribution of this species-rich genus affords a variety of life-history strategiesfrom the weedy colonizers of secondary vegetation tothe large canopy lianas of primary forest

Passiflora is widely cultivated today for its ornamen-tal flowers and edible fruit The presence of a coronaand an androgynophore gives the flowers a strikingmorphology that has long been a favorite of horticul-turists and hobbyists alike This unique morphologycaught the attention of the conquistadors who took itas a symbol of the crucifixion of Christ and conse-quently a sign that the New World would be convertedto Christianity (Uribe 1955 Kugler 2004) This reli-gious symbolism gave the plant their common namelsquolsquoPassion Flowerrsquorsquo referring to the passion of Christ

Species in Passiflora are typically tendril-bearingvines with a non-pedunculate inflorescence and one ortwo sessile pentamerous flowers Although most spe-cies are herbaceous vines or woody lianas membersof Passiflora subgenus Astrophea (DC) Mast tend to beshrubs or true trees The size of the flowers and thedegree of complexity in the corona vary widelythroughout the genus The innermost row of the co-rona the operculum interacts with a membrane (li-men) at the base of the androgynophore to form a lipor cup over the nectary preventing access by ineffec-

tive pollinators These three characters corona oper-culum and limen have historically been heavily reliedupon as taxonomic characters for delimiting relation-ships within Passiflora

The most recent monograph for Passiflora (Killip1938) divided the genus into 22 subgenera It is prob-lematic for taxonomists for several reasons First over120 new species have been described since Secondonly the New World species were included Thirdmany of the ranks below subgenus were invalidly pub-lished Although new insights have been gained intoPassiflora relationships at the subgeneric level (De Meloet al 2001 Muschner et al 2003 Yockteng and Nadot2004) there is still insufficient resolution and supportfor monophyletic groups below the this level More in-formation is still needed to ultimately address evolu-tionary questions at the species level

The two largest lineages in the genus correspondto Killiprsquos subgenera lsquolsquoPlectostemmarsquorsquo and lsquolsquoGranadillarsquorsquo(Decaloba and Passiflora respectively following ICBNrules) Species of Decaloba (220 spp) are mostly her-baceous vines with small flowers and fruit They occurthroughout the entire distribution of the genus includ-ing the Old World Decaloba species have an ancestralchromosome number of n56 (Snow and MacDougal1993) and as such are informally referred to as thelsquolsquon56 grouprsquorsquo In addition to Decaloba this group in-cludes Killiprsquos subgenera Astephia Psilanthus Pseudo-murucuja Murucuja Chloropathanthus and Apodogyne(Feuillet and MacDougal 2004) Conversely species insubgenus Passiflora (220 spp) are woody vines withshowy flowers and edible fruit The ancestral chro-mosome number of n59 has led to the designation ofsubgenus Passiflora along with subgenera Adenosepala

2006] 139HANSEN ET AL SYSTEMATICS OF PASSIFLORA

TABLE 1 Data analyses and tree statistics for the four parsimony analyses of the trnLtrnT spacer region The number of excludedcharacters refer to the number of base pairs comprising gap regions that were removed in analysis B or the number of positions removeddue to homoplasy concerns in analysis D C I values were calculated in PAUP with uninformative characters excluded

AnalysisOptimality

criterion

ofcharactersexcluded

Gapinterleavedcharacters Gap treatment

Parsimonyinformativecharacters

Numberof trees

Treelength C I R I H I

ABCD

ParsimonyParsimonyParsimonyParsimony

NoneNone27964

None7575

None

Missing dataMissing dataRemovedMissing data

881209979

200000200000200000200000

265384302229

0800074507190795

0913088308800916

0200025502810205

Tacsonia Manicata Calopathanthus Dysosmia Dysosmioi-des Distephana Rathea Tacsonioides and Tacsoniopsis asthe lsquolsquon59 grouprsquorsquo (Feuillet and MacDougal 2004)

This dichotomy between the lsquolsquon56rsquorsquo and lsquolsquon59rsquorsquogroups has been well known among Passiflora research-ers for decades (L Escobar unpubl P M Joslashrgensen JMacDougal C Feuillet pers comm Presting 1965)However the first molecular evidence to support theevolutionary integrity of these groups was publishedby Downie et al in 1996 In this study Downie sur-veyed a wide diversity of angiosperms for the presenceor absence of the rpoC1 intron to assess its phyloge-netic utility Although the marker was homoplastic forhigher order relationships preliminary evidence sug-gested that it might be useful in studies at the subfa-milial level particularly in Passiflora In their study of10 species of Passiflora all taxa with a chromosomenumber of n56 lacked the rpoC1 intron while thosethat had n59 retained it The sample size was toosmall in the Downie et al study to make any conclu-sive phylogenetic inferences for Passiflora as a wholebut indicated that a more intensive sampling strategymight provide additional support for the delineationof these two groups As part of our work exploring thephylogenetic relationships in Passiflora we expandedthe intron survey and combined this character with achloroplast DNA sequence phylogeny The combinedresults are presented here along with discussions ofPassiflora relationships and taxonomy chromosomenumber evolution and the relative utility of these mak-ers for resolving evolutionary history

MATERIALS AND METHODS

Chloroplast Phylogeny Sampling included 61 species 57 fromgenus Passiflora representing 16 of Killiprsquos subgenera and one spe-cies each from genera Adenia Forssk Dilkea Mast Tetrapathea DCand Tetrastylis Killip as outgroups (see Appendix) All species in-cluded in the phylogenetic analyses were also examined in theintron survey Total DNA was isolated from single individuals us-ing either fresh leaf material silica dried field collections or her-barium specimens (Appendix 1) Protocols for DNA isolation am-plification and sequencing followed those described in Barber etal (2002) Both strands were sequenced for all taxa

Sequences were edited using Sequencher v 31 (Gene CodesCorp) Clustal X (Thomson et al 1994) was used to perform mul-tiple alignments using pairwise comparisons to obtain an initialalignment In order to identify potential repeat motifs we usedthe Megalign program in DNAstar (DNAstar Madison WI) This

was followed by manual adjustment using two general guidelines1) the number of gaps needed to align sequences was minimized(Golenberg 1993) and 2) in areas of tandem repeats sequenceswere aligned to maximize percent similarity and minimize thenumber of substitutions between sequences (Aldrich 1988)

Parsimony as implemented in PAUP v 40b10 (Swofford 2002)was used for analyses A-D (Table 1) We used the heuristic searchmethod with 100 random addition replications with step-wise ad-dition of taxa tree bi-section reconnection and MulTrees on Inorder to increase the probability that all islands of most-parsi-monious trees would be sampled we ran the analysis on threedifferent computers to find the length of the shortest tree Thiswas then set as the upper limit for tree length for the final analysesand only 2000 trees equal to this value were saved for each rep-licate All subsequent analyses were run on lsquolsquophyloclusterrsquorsquo a NPA-CI Rocks (httprocksclusterorg) cluster comprised of 24 AMD18001 processors Levels of homoplasy were estimated by calcu-lating the CI HI and RI in PAUP Support for monophyleticgroups was evaluated by bootstrap analyses using 100 replicateseach with a lsquoMaxTreesrsquo limit of 2000 per replicate These methodswere utilized for all parsimony analyses

We evaluated the effects of gap coding by doing four differentanalyses using the parsimony methods described above (Table 1)The first analysis (A) included all sequence characters and the gapregions were treated as missing data Gaps were then coded asbinary characters utilizing the simple gap coding method of Sim-mons and Ochoterena (2000) and subsequently appended to thesequence data Since indel events are frequent occurrences in chlo-roplast non-coding regions (Kelchner 2000) removing the result-ing gap regions from the analysis reduced the number of infor-mative characters by 18 Due to the loss of such a significantamount of data in an already information-poor region we ana-lyzed the data with gap regions included and treated as missingdata (analysis B) and with gap regions removed (analysis C) Fi-nally for analysis D the region corresponding to base pairs 425ndash489 was removed This region appears to be a lsquolsquohot-spotrsquorsquo for indelformation and was removed from this analysis to estimate theimpact of such regions on levels of homoplasy in our data Analy-ses A B C and D correspond to TreeBASE (study accessionS1330)

We used Bayesian analyses to evaluate statistically potentialmonophyletic groups in Passiflora The GTR1G model of sequenceevolution was chosen by ModelTest v 306 (Posada and Crandall1998) as the best fit for these data MrBayes v 30 (Huelsenbeckand Ronquist 2003) was used for all analyses Uniform prior dis-tributions were used for 1) alpha shape parameter for the gammadistribution (00 100) 2) instantaneous rate matrix (00 1000)and 3) branch lengths (00 100) Base frequencies were estimatedand the starting trees were randomly generated We used a Me-tropolis-coupled MCMC analysis where four chains were incre-mentally heated and run simultaneously Two separate analyseswere done running for 1ndash3 X106 respectively and every 100th treewas sampled Trees that were generated prior to the plateau ofparameter values were deleted The remaining trees were openedin PAUP vs 410b (Swofford 2002) and used to generate a 50Majority Rule tree Each analysis was subsequently repeated toverify the results

Intron Survey Representative taxa from each subgenus sec-

140 [Volume 31SYSTEMATIC BOTANY

FIG 1 Top Diagrammatic representation of rpoC1 intronArrows indicate the relative positions of the two priming sitesShaded boxes represent coding regions of the rpoC1 genewhile the intron disrupting this gene is depicted as white Afragment size of 1100 bp illustrates the retention of the rpoC1intron while a fragment of 400 bp illustrates an intron lossBottom Gel photograph of rpoC1 intron results

tion and series were sampled when available In total 136 speciesfrom 19 of the Killiprsquos 22 subgenera within Passiflora were sur-veyed for the presence of the rpoC1 intron In addition four othergenera in the Passifloraceae were included as outgroups (Appen-dix 1)

DNA was amplified by polymerase chain reaction (PCR) usingthe forward 59 rpoC1 exon and reverse 39 rpoC1 exon primers de-scribed in Downie et al (1996) PCR amplification methods wereidentical to those described for the trnLtrnT sequencing studyPositive and negative intron controls (Passiflora caerulea and P sub-erosa respectively) that were confirmed by sequencing (Downie etal 1996) were included in each set of amplification reactions aswell as a negative cocktail control (no template) Methods for de-termining the presence or absence of the rpoC1 intron followedthose described in Downie et al (1996) Faint bands were notscored Due to the significant difference in size of fragments withor without the intron gel interpretation was unambiguous (Fig1) These results are listed in Appendix 1

A parametric bootstrap approach (SOWH test Swofford et al1996 Goldman et al 2000) was utilized to determine whether sin-gle or multiple intron loss hypotheses were equally likely givenour data The data was pruned to only include the clade in whichintron losses had been detected This data set was then analyzedusing the Parsimony optimality criterion as implemented in PAUPv 40b10 with 100 random addition replicates Multrees on andTBR A second search where the intron loss was constrained tobe monophyletic followed using the same methods as the uncon-strained search The tree with the best likelihood score from thissecond analysis was used to simulate 100 data sets using Seq-Gen(Rambaut and Grassly 1997) Each data set was then analyzed inPAUP with and without the constraint to generate a distributionof tree-length differences An unconstrained vs constrained tree

length difference greater than 95 of the simulated data setswould allow us to reject the null hypothesis of a single intron loss

RESULTS

Chloroplast DNA Sequence Phylogeny There wassubstantial length variation among taxa in the trnLtrnT spacer consistent with other analyses of this re-gion (Kelchner 2000) Unaligned sequences varied inlength between 454 bp in Passiflora deidamioides and577 bp in Adenia mannii while the aligned length is 631bp Gap coding resulted in 75 characters appended tothe sequence data for analyses B and C

The nucleotide sequence diversity of the trnLtrnTspacer is conserved among taxa as illustrated by thelow number of parsimony informative variable char-acters (Table 1) However the region is clearly hetero-geneous in both base composition (AT rich with 42A 125 C 167 G and 288 T) and the propensityfor indels to occur This is particularly evident withinthe large poly AndashT tract corresponding to bp 425ndash489in which numerous indel events of varying lengths ap-pear to have occurred Determining sequence homol-ogy within this and similar areas was problematicThis region comprising 64 aligned bp was thereforeexcluded from analysis D For all data sets the numberof transversions estimated in MacClade were nearlytwice the number of transitions similar to values re-ported for other chloroplast non-coding regions (Mor-ton 1995)

The tree statistics for all parsimony analyses aresummarized in Table 1 When gaps were treated asmissing data (analysis A) all tree statistics were slight-ly better than analyses when gaps were coded and ap-pended to the sequence data (Analyses B and C) How-ever bootstrap support for most monophyletic groupswas significantly higher when gaps were coded andthe gap regions treated as missing data (Analysis BTable 1) This is likely due to the greater number ofparsimony informative characters available in thisanalysis When the poly A-T regions were removed(Analysis D) there was no improvement in eithernumber of trees found or tree statistics In fact boththe CI and the HI were slightly lower than whenthat region was included in Analyses A The drop inthe percentage of parsimony informative charactersfrom 144 (Analysis A) to 137 (Analysis D) prob-ably accounts for this slight difference There were nodifferences in topology between any of the analyses forstrongly supported groups Since the tree statistics arenot significantly different for any of the analyses ourresults do not support the exclusion of poly A-T re-gions as we anticipated

ModelTest selected the GTR1G model of evolutionfor use in our Bayesian analyses There were no majordifferences in topology between analyses run for onemillion and three million generations These results


FIG 2 Strict consensus of 200000 most parsimonius treesusing the trnLtrnT chloroplast spacer region All trees were265 steps in length Bootstrap values above 50 are shown inbold No gap data were included in this analysis (Analysis ATable 1) The classification used in this figure is that of Feuilletand MacDougal (2004) All specific epithets are from the ge-nus Passiflora unless otherwise noted

FIG 3 50 majority rule tree from a Bayesian analysis ofthe trnLtrnT run for three million generations Numbersabove the branches are posterior probabilities Bars depict thesubgenera of Passiflora sensu Feuillet and MacDougal (2004)Highlighted taxa are those that possess the Andean hum-mingbird pollination syndrome and optimization of this char-acter onto the tree is illustrated by a black box Killiprsquos (1938)subgenera are in bold and abbreviated after each species nameas follows Apodogyne (apo) Astrophea (astr) Calopathanthus(cal) Decaloba (dec) Deidamioides (dei) Distephana (dis) Dys-osmia (dys) Dysosmioides (dysd) Manicata (man) Murucuja(mur) Passiflora (pas) Polyanthea (pol) Pseudomurucuja (pse)Rathea (rat) Tacsonia (tac) Tacsonioides (tacd)

were confirmed by running each analysis in duplicateThe likelihood values reached a plateau at 224182594for one million generations and 224164027 for threemillion generations The numbers of burn-in trees de-leted were 60700 trees for one million generations and118700 trees for three million generations In both cas-es the number of generations utilized clearly was inexcess of that needed for stabilization of all values Re-cent work has suggested that the posterior probabili-ties of Bayesian analyses can often be misleading in-dicating much higher levels of support for a particularclade when compared to bootstrap values (Suzuki etal 2002 Wilcox et al 2003) Although the levels of sup-port in the Bayesian analyses were substantially higherthan when Parsimony methods were used the topol-ogies of all of the trees are nearly identical (Figs 2 3)A single most parsimonious tree is depicted as a phy-logram (fig 4)

There were no major differences in topology be-tween any of the analyses for groups that had 70 orhigher bootstrap support or a posterior probability of070 The only detectable variation between any of thedifferent treatments is on branches that have little sup-port in both parsimony and Bayesian analyses and col-lapse on the parsimony consensus trees (Figs 2 4)Tree statistics for all parsimony analyses are summa-

rized in Table 1 The lsquolsquon59rsquorsquo and lsquolsquon56rsquorsquo groups eachformed monophyletic groups in all trees regardless ofhow the data was treated

Intron Distribution Interpretation of intron datawas straightforward Visualization of the PCR prod-ucts on the gels revealed a fragment of 1100 bp whenthe rpoC1 intron was present and a fragment of 400 bpwhen absent (Fig 1) The presence or absence of theintron is listed by species in the Appendix The intronwas absent in subgenera Astephia ChloropathanthusMurucuja Pseudomurucuja Psilanthus and in all sec-tions of Decaloba with the exception of supersectionsHahniopathanthus Disemma and Pterosperma sensuFeuillet and MacDougal (2004) (Appendix 1) The in-tron is present in all species surveyed from subgeneraApodogyne Adenosepala Tryphostemmatoides Deidamioi-des Tacsonia Manicata Distephana Calopathanthus Pas-siflora Dysosmia Dysosmioides Polyanthea and AstropheaTaxa from four additional genera of Passifloraceae Dil-kea Tetrapathea Adenia and Tetrastylis also retained theintron Decaloba is the only subgenus in which there


FIG 4 Phylogram of 1 of 200000 trees from analysis A of trnLtrnT sequence data illustrating the distribution of chro-mosome numbers in the genus Not all taxa in each group have been counted See Appendix for list of taxa with knownchromosome numbers Any known exceptions are highlighted Branch lengths represent character state changes and the numberof changes is listed above each branch


TABLE 2 Summary of subgeneric classification of Passiflora according to Feuillet and MacDougal (2004) The number of species ineach of the four subgenera is in parentheses Killiprsquos (1938) subgenera are listed in parentheses under the subgeneric names usedby Feuillet and MacDougal (2004) All ranks are subgeneric except for Tetrastylis and Tetrapathea which are generic and denoted withan

Subgenus Defining characters Distribution

Astrophea (DC) Mast (57) (Astrophea) Tendrils present or absentSimple unlobed leaves margins entireGlands at base of petiole or margins of leaf bladeHypanthium campanulate or tubularOperculum entire membranous or tubularLianas trees or shrubs

Lowland tropical SouthAmerica

Decaloba (DC) Rchb (220) (ApodogyneAstephia Chloropathanthus DecalobaMurucuja Pseudomurucuja Psilanthus)

Tendrils presentPalmate venationGlands are laminarHypanthium flatPlicate operculumShort fleshy limenHerbaceous vines

Throughout Central andSouth America andSouthern US 20 speciesin South East Asia Aus-tralia and New Zealand

Deidamioides (Harms) Killip (13) (Deida-mioides Polyanthea Tetrastylis Tetrapa-thea Tryphostemmatoides)

Peduncles terminate in tendrilLeaves entire or lobedGlands on petioleHypanthium campanulate or flatOperculum plicateLianas or herbaceous vines

Throughout Central andSouth America and Mex-ico One species in NewZealand

Passiflora L (240) (Adenosepala Calopa-thanthus Dysosmia Dysosmioides Diste-phana Manicata Passiflora Rathea Tac-sonia Tacsonioides Tacsoniopsis)

Tendrils presentLeaves entire or lobedGlands on petioles stipules and margins of leavesHypanthium tubular or campanulateOperculum tubular or filamentousMembranous limenHerbaceous vines or woody lianas

Throughout Central andSouth America andSouthern US

was any variation in intron presence or absence de-tected with our sampling strategy

Distribution of Intron on cpDNA SequencePhylogeny The loss of the rpoC1 intron was opti-mized onto the phylogeny constructed from cpDNAsequence data using the trnLtrnT spacer region(Fig 2) The loss occurred on the branch leading toPassiflora auriculata in a weakly supported clade withOld World species P herbertiana and P aurantiawhich retain the intron Passiflora holosericea whoserelationship is unresolved also has a loss Anotherintron loss occurs on the strongly supported branchincluding seven species traditionally considered topart of the lsquolsquon56rsquorsquo group Thus the distribution ofthe intron loss on the cpDNA sequence phylogenyindicates that this change has occurred a minimumof two times in the evolution of Passiflora

Multiple losses were confirmed by performing aparametric bootstrap analysis (Swofford et al 1996Goldman et al 2000) Due to computational limita-tions the data set was pruned to include only subge-nus Decaloba and its sister group (Passiflora deidamioidesTetrastylis ovalis and P cirrhiflora Fig 2) The uncon-strained parsimony analysis of this group found onlyten trees and all intron losses occurred within oneclade (BS5100) The results of the parametric boot-strap analysis allow us to reject the null hypothesis of

a single loss with a p value 005 We can say withconfidence that there have been a minimum of twolosses of the rpoC1 intron in Passiflora

DISCUSSION

Chloroplast Sequence Phylogeny Non-coding re-gions of cpDNA were originally thought to be a po-tentially rich source of variation for phylogeneticanalyses (Curtis and Clegg 1984 Taberlet et al 1991Clegg et al 1994) Recent studies have shown that thisis not always the case and in fact these regions areoften surprisingly conserved (Bailey and Doyle 1999Potter and Luby 2000 Raubeson and Jansen 2004) Thisis clearly the case for the trnL-trnT spacer region inPassiflora Although indels are prevalent in this regionthe difficulty in determining their homology limitstheir phylogenetic utility The overall tree topology ishighly congruent with both morphological data (Table2) and chromosome number trends (Fig 4) This ismost evident in the strongly supported lsquolsquon59rsquorsquo andlsquolsquon56rsquorsquo clades (Fig 4) There is however little supportfor the monophyly of most of Killiprsquos (1938) subgeneraand sections as has long been suspected by manyworkers Although many of the terminal nodes are un-resolved in our phylogeny particularly in subgenusPassiflora species relationships within Decaloba are suf-


ficiently resolved to enable us to evaluate the utility ofthe rpoC1 intron for identifying monophyletic groupswithin the genus Additionally several strongly sup-ported clades provide valuable insight into questionsof historical interest in Passiflora including pollinationsyndromes and chromosome number evolution andallow us to evaluate a newly proposed classificationsystem for the genus

INTRON DISTRIBUTION IN PASSIFLORA Chloroplaststructural rearrangements have been shown to bepowerful characters for elucidating phylogenetic rela-tionships (Jansen and Palmer 1987 Bruneau et al 1990Downie and Palmer 1992 Raubeson and Jansen 19922004 Soltis and Soltis 1998) However certain types ofstructural changes are more likely to show homoplasythan others For example inversions in the chloroplastgenome have been found to be extremely reliable char-acters with low levels of homoplasy (Raubeson andJansen 2004) while intron and gene losses are morelikely to occur in parallel (Downie et al 1991 1996)In Passiflora the distribution of the rpoC1 intron cor-relates strongly with chromosomal and morphologicaldata to support subgenera Astephia ChloropathanthusMurucuja Pseudomurucuja Psilanthus and Decaloba as amonophyletic group Indeed all New World specieswith a known chromosome number of n56 lack theintron Only P holosericea and P lobata with haploidchromosome numbers of n57 have a loss and a chro-mosome number other than n56 We did not detectany losses outside of the Decaloba group sensu Feuilletand MacDougal (2004) This is congruent with the pre-vious study (Downie et al 1996) Interpretation of theresults within subgenus Decaloba is more complicatedThe intron is present in only seven of the 54 speciessampled from this subgenus Two of these species Paurantia and P herbertiana are Old World in distribu-tion Passiflora microstipula and P lancetillensis belong tothe newly described supersection Pterosperma (Mac-Dougal and Hansen 2003) The remaining three Pmembranaceae P hahnii and P guatemalensis are mem-bers of supersection Hahniopathanthus within subgenusDecaloba

In the absence of a phylogenetic tree for the genusmultiple losses would have been an unlikely conclu-sion with our sampling strategy However by map-ping the intron loss distribution onto the phylogenetictree as shown in Fig 2 it suggests that multiple losseshave in fact occurred This hypothesis is further con-firmed by the parametric bootstrap analysis that al-lowed us to reject a single loss hypothesis at the 005level and to conclude that the rpoC1 intron loss is nota reliable predictor of relationships within Passiflora be-low the subgeneric level

lsquolsquon56 GROUPrsquorsquo The lsquolsquon56 grouprsquorsquo correspondingto subgenera Decaloba Astephia Murucuja Pseudomu-rucuja Apodogyne and Psilanthus in Killip (1938) forms

a monophyletic group with 100 bootstrap support inthe parsimony analysis and a posterior probability of100 in the Bayesian analysis (Figs 2 3) This group ofspecies is less showy than the lsquolsquogranadillasrsquorsquo of sub-genus Passiflora with much smaller flowers and a typ-ically herbaceous habit The leaves are palmatelyveined and usually with laminar nectaries The floraltubehypanthium is flat while the operculum is plicateand the limen when present is short and fleshy Thepollen is distinct from the lsquolsquon59 grouprsquorsquo in that it hassix pairs of colpi that anastamose toward the poleswhile the lsquolsquon59 grouprsquorsquo has only three pairs of colpi(Amela et al 2002 Spirlet 1965 Presting 1965)

The Old World species of Passiflora represented hereby Passiflora aurantia and P herbertiana both native toAustralia come out nested within subgenus Decalobawith very strong bootstrap support of 87ndash92 and aposterior probability of 10 (Figs 2 3) Although theirNew World sister group has long been debated theyhave historically been given their own section withinDecaloba (DeWilde 1972 Feuillet and MacDougal 2004)Morphologically they possess several characters typicalof Decaloba including plicate opercula laminar nectar-ies and a chromosome number of n56 (Snow andMacDougal 1993) Our Bayesian phylogeny indicatesthat P auriculata or close relative is the likely NewWorld sister to this group (Fig 3) However an ex-panded sampling strategy including southeast AsianPassiflora will be needed to verify this placement Aphylogenetic study already underway (S Krosnickpers comm) including all the Old World taxa is ex-pected to clarify the biogeographic relationships be-tween New and Old World Passiflora

The most surprising outcome in this clade was theplacement of Passiflora microstipula and P lancetillensiswithin subgenus Decaloba They belong to a recentlydiscovered group that occurs in Mexico and CentralAmerica (Gilbert and MacDougal 2001) and has aunique morphology These large lianas bear flowersfrom the tendril an uncommon but not unknowncharacter state in Passiflora The seeds are unusuallylarge and bear conspicuous wings on the edges acharacter unique in Passiflora This combination of char-acters indicated that this group represented a very bas-al lineage within the genus and was placed withinsubgenus Deidamioides by Feuillet and MacDougal(1999) However our data place this group in a basalposition within subgenus Decaloba with strong sup-port Morphological evidence such as variegated juve-niles cernuous new tip growth and a plicate opercu-lum also support this placement (J MacDougal percomm) Passiflora microstipula and P lancetillensis alongwith three Central American species are placed in thenew supersection Pterosperma of subgenus Decaloba(MacDougal and Hansen 2003 Feuillet and Mac-Dougal 2004)


Passiflora microstipula deserves some added discus-sion as it comes out in both subgenera Passiflora andDecaloba in the chloroplast tree We can find no biolog-ical evidence for its placement within the showy gran-adillas of subgenus Passiflora Although P microstipuladoes have a haploid chromosome number of n59 thekaryotype is much more similar to members of sub-genus Decaloba than those of subgenus Passiflora (Snowand MacDougal 1993) In order to rule out contami-nation of DNA or PCR products DNAs from two dif-ferent accessions of P microstipula one from Veracruzand one from Oaxaca were re-isolated from living ma-terial These PCR products were subsequently clonedand sequenced All clones of the Oaxaca populationcame out within subgenus Decaloba while clones fromthe Veracruz population were split between the Passi-flora and Decaloba subgenera The strong statistical sup-port for the sequence similarities between P nitida(subgenus Passiflora) and P microstipula is surprisingbut experimentally repeatable Passiflora nitida is occa-sionally cultivated in Central and South America forits fruit but it is not known to occur in Mexico Field-work in Veracruz and herbarium studies conducted byJohn MacDougal (pers comm) found no evidence thatP nitida ever co-occurred with P microstipula In addi-tion several crosses between both populations of P mi-crostipula and various species of subgenus Passiflorawere carried out in the greenhouses at The Universityof Texas over a span of several years without any ev-idence of fruit set or other indication of compatibility(L Gilbert unpubl) However our work on plastid in-heritance in Passiflora has clearly demonstrated thatheteroplasmy does occur in this genus (K Hansen etal in prep) So although we can find no historical ev-idence at this time to support a chloroplast capturescenario between these taxa we must consider the ideathat the relationship between these two chloroplasttypes is the result of a past biological phenomenon Wedo not feel that this evidence should have any bearingon P microstipularsquos taxonomic placement due to themultitude of morphological and molecular evidencesupporting its strong relationship with other speciesin subgenus Decaloba (Fig 2 Muchner et al 2003 Yock-teng and Nadot 2004)

Most of the Caribbean members of Passiflora havetraditionally been placed in their own subgenera dueto their highly modified flowers suggestive of variouspollinator syndromes The taxa that were placed in Kil-liprsquos subgenera Murucuja and Pseudomurucuja are pol-linated primarily by hummingbirds while bat polli-nation has been documented in P penduliflora of sub-genus Astephia (Kay 2001) Feuillet and MacDougal(1999 and 2004) recognized that these flowers wereprobably highly modified representatives of subgenusDecaloba Our data clearly support this placement and

identify the P biflora lineage as the likely continentalsister group

lsquolsquon59 GROUPrsquorsquo The large flowered granadillasform a monophyletic group with very strong bootstrap(99ndash100) and posterior probability (10) supportThese correspond to Killiprsquos subgenera Passiflora Tac-sonia Manicata Calopathanthus Dysosmia Dysosmioidesand Distephana and are all placed in subgenus Passiflorain the new classification by Feuillet and MacDougal(2004) In addition to a chromosome number of n59(Snow and MacDougal 1993) this group is held to-gether morphologically by having a tubular or cam-panulate hypanthium tubular or filamentous opercu-lum a membranous limen foliaceous bracts and high-ly reticulate pollen with three pairs of colpi (Presting1965 Spirlet 1965 Amela 2002) The showy flowersand edible fruit of this group have led to their culti-vation and introduction throughout the Old and NewWorld This group appears to have undergone a rela-tively rapid radiation as many of the relationshipswithin subgenus Passiflora are unresolved in our dataas well as in two other recent Passiflora phylogenies(Muchner et al 2003 Yockteng and Nadot 2004)

One particularly interesting revelation is that speciesthat exhibit the Andean hummingbird pollination syn-drome do not form a clade (shaded taxa in Fig 3)Endemic to the high elevations of the Andes in SouthAmerica taxa in this group have long tubular redpink or orange flowers and are visited by humming-birds Called lsquolsquotacsosrsquorsquo these plants produce a sweetbanana-shaped fruit that is cultivated throughout theAndes and is prevalent in local markets This is a dif-ficult group taxonomically and they have historicallybeen split into three closely allied subgenera (TacsoniaManicata and Rathea) based primarily on coronalshape degree of bractal fusion and the point of petalinsertion in the floral tube (Escobar 1988) Rather thanforming a monophyletic assemblage as we expectedthe tacsos are split into two very strongly supportedgroups (shaded taxa in Fig 3) In the Bayesian phy-logeny P antioquiensis P ampullacea P parritae P cum-balensis P manicata and P andina form a monophyleticgroup with a posterior probability of 100 Passiflora tri-foliata along with the bat-pollinated P trisecta (Ulmerand MacDougal 2004) is nested within a group of trop-ical lowland bee-pollinated species with a posteriorprobability of 099 That the tacsos do not form amonophyletic group in our data suggests that the An-dean hummingbird pollination syndrome had at leasttwo independent origins Only Masters in his 1871monograph indicated that this syndrome might haveoriginated more than once in this group

Chromosome Number Evolution Chromosomenumber is a very useful predictor of relationships inPassiflora There are 83 published chromosome num-bers which account for roughly 17 of the genus (re-


viewed by Snow and MacDougal [1993] and De Meloet al [2001]) The two largest subgenera Passiflora andDecaloba have ancestral haploid numbers of n59 andn56 respectively Each of these groups is monophy-letic with 90 or greater bootstrap support (Figs 24) Conversely subgenera Astrophea and Deidamioideshave haploid numbers of n512 Polyploidy in the ge-nus is infrequent with only six reports to date

Several numbers have been proposed as the basenumber for Passiflora most commonly x53 x56 x59or x512 (Storey 1950 Raven 1975 Morawetz 1986Snow and MacDougal 1993 De Melo et al 2001 DeMelo and Guerra 2003) Insightful discussions of thistopic have been published in recent years (Snow andMacDougal 1993 De Melo et al 2001) In the Passi-floraceae chromosome numbers of n512 (Tetrapathea inHair and Beuzenberg 1959 and Adenia in De Rocher etal 1990) and probable aneuploid derivatives in n511(Deidamia Noronha ex Thouarsand Crossostemma Plan-chon ex Hook in Gadella 1969 1970) are the onlycounts available for other genera in the family Al-though chromosome numbers are known for only fiveof the 17 genera in the Passifloraceae these numberswould suggest a base chromosome number of x512for the genus Passiflora and the Passifloraceae HoweverDe Melo and Guerra (2003) examined the number of5S and 45S sites revealed by chromosomal staining inthe major karyotype groups within the genus andfound compelling but inconclusive evidence that sup-ports a base chromosome number of x56 for Passiflora

Using a different approach we examined the basenumber for the genus by optimizing chromosomenumbers on the most parsimonious trees usingMacClade v 40 (Madison and Madison 2000) An un-published count of n511 or n512 for P guatemalensis(J MacDougal pers comm) part of a basal clade insubgenus Decaloba is critical for understanding chro-mosome evolution in Passiflora and was included in ouranalysis (Fig 2) An ancestral base chromosome num-ber of x512 required five steps while x56 and x59each required six steps

A difference of a single step is too slight to designatea base chromosome number of x512 for Passiflora withconfidence especially when the convincing evidencepresented in favor of x56 (De Melo and Guerra 2003)is considered However since n56 is not found outsideof subgenus Decaloba in either the family or genus itseems likely that n56 had a single origin in the ances-tor to this monophyletic group In our opinion thisevidence combined with the fact that n512 and n511are the only known numbers in the Passifloraceae sup-ports x512 as the base chromosome number for Pas-siflora Clearly more information is needed to under-stand chromosome number evolution in Passiflora Ofparticular importance would be more chromosomecounts for other genera in the Passifloraceae and for

the poorly known subgenera Deidamioides and Astro-phea within the genus Passiflora

Taxonomic Implications Killiprsquos monograph of theNew World Passifloraceae has been the classificationsystem most widely used since its publication in 1938However the rapidity with which new species havebeen discovered in recent years and the omission ofthe Old World species has limited its utility A newinnovative classification system by Feuillet andMacDougal (2004) has built on the solid foundationestablished by Killip by including all known species ofPassiflora and correcting many nomenclatural errorsBy reducing Killiprsquos 22 subgenera to four this treat-ment more accurately represents the evolutionary re-lationships between species (Figs 2 3) Without excep-tion Feuillet and MacDougal have successfully iden-tified monophyletic species assemblages at this levelSome of the supersections and series do not formmonophyletic groups in the cpDNA phylogeny but thelow resolution in our phylogeny precludes a detailedevaluation of their system at this level

A brief comparison of Feuillet and MacDougalrsquosclassification system with that of Killiprsquos is shown inTable 2 Our data unequivocally support their speciesgroupings at the subgeneric level This is illustrated inFig 2 where subgenera Passiflora Astrophea Decalobaand Deidamioides are all monophyletic with strong sta-tistical support

The relationship of Tetrapathea a monotypic NewZealand genus with dioecious plants and four-merousflowers to the rest of the subgenera is unresolved inour data With the exception of Killip (1938) most Pas-siflora systematists beginning with Masters (1870) haveconsidered Tetrapathea to be a member of the genusPassiflora However there has been little consensus asto which subgenera it is most closely related to orwhether it deserves its own monotypic rank Based onmorphological characters Feuillet and MacDougal(2004) included Tetrapathea within subgenus Deidamioi-des while Green (1972) gave Tetrapathea its own sub-genus Most recently Yockteng and Nadot (2004) us-ing the nuclear encoded chloroplast gutamate synthasegene (ncpGS) found a highly supported relationshipbetween Tetrapathea and subgenus Decaloba Interest-ingly our Bayesian analyses show a strong relationshipbetween Tetrapathea and Dilkea with a posterior prob-ability of 090 Dilkea (South America 9 spp) was de-scribed by Masters (1870) and excluded from Passiflorabased on its lack of an androgynophore and four-mer-ous flowers Although the relationship between Tetra-pathea and Dilkea was not evident in parsimony analy-ses the common character of four-merous flowers sug-gests that it may not be an artifact of our data A morevariable marker may be able to clarify Tetrapathearsquosplacement within Passiflora and its relationship to Dil-kea


ACKNOWLEDGEMENTS The authors thank John Macdougal andChristian Feuillet for sharing their knowledge of Passiflora rela-tionships and for providing an advance copy of their classificationsystem We are very grateful to Elsa Zardini Teresa Floretin-PenaCarmen Cristobol Marcio and Rosana Cardoso Volker BittrichMaria do Carmo Amaral Ricardo Secco Katya Romoleroux Ren-ato Valencia Mauricio Lopez Heidi Meudt Juan Jose CastillioSeth Hansen and Mark Price for help in obtaining permits trans-lating expertise logistics and access to facilities during field workin Paraguay Argentina Brazil Ecuador and Guatemala TomWendt and the Plant Resources Center at UT Austin secured loansfrom BM MA K F COL NY MO CAY QCA and US which werecritical for this work Derek Zwickl and Leah Larkin kindly pro-vided assistance with data analyses Funding was generously pro-vided by a NSF Doctoral Dissertation Improvement Grant (DEB9701092) to A Katie Hansen the American Society of Plant Tax-onomists Sigma Xi The University of Texas Botany MemorialFund The Garden Club of America The Institute for Latin Amer-ican Studies the Lawrence Award through the Hunt Botanical In-stitute and a scholarship from Ron Boender at Butterfly World

LITERATURE CITED

ADAMS S P I J LEITCH M D BENNETT M W CHASE and AR LEITCH 2000 Ribosomal DNA evolution and phylogenyin Aloe (Asphodelaceae) American Journal of Botany 87 1578ndash1583

ALDRICH J B W CHERNEY E MERLIN and L CHRISTOPHERSON1988 The role of insertionsdeletions in the evolution of theintergenic region between psbA and trnH in the chloropastgenome Current Genetics 14 137ndash146

AMELA-GARCIA M B GALATI and A ANTON 2002 Microspo-rogenesis microgametogenesis and pollen morphology ofPassiflora spp (Passifloraceae) Botanical Journal of the LinneanSociety 139 338ndash394

BAILEY C D and J J DOYLE 1999 Potential phylogenetic utilityof the low-copy nuclear gene pistillata in dicotyledonousplants Comparison to nrDNA ITS and trnL intron in Sphaer-ocardamum and other Brassicaceae Molecular Phylogenetics andEvolution 13 20ndash30

BALDWIN J 1949 Polyploidy in Passiflora lutea Rhodora 51 29BARBER J C J F ORTEGA A SANTOS-GUERRA K TURNER and

R JANSEN 2002 Origin of Macaronesian Sideritis L (Lamioi-deae Lamiaceae) inferred from nuclear and chloroplast se-quence datasets Molecular Phylogenetics and Evolution 23 293ndash306

BEAL P 1969 Chromosome numbers of the exotic Passiflora speciesin Australia Queensland Journal of Agricultural and Animal Sci-ences 26 407ndash421

mdashmdashmdash 1971 Chromosome numbers in some recently introducedspecies of Passiflora in Australia Queensland Journal of Agri-cultural and Animal Sciences 28 179ndash180

BERRY P 1987 Chromosome number reports XCV Taxon 36 493BOWDEN W 1940 The chromosome complement and its evolu-

tionary relationship to cold resistance in the higher plantsChronicles of Botany 6 123ndash125

mdashmdashmdash 1945 A list of chromosome numbers in higher plants IIMenispermaceae to Verbenaceae American Journal of Botany32 191ndash201

BROWER A V Z 1997 The evolution of ecologically importantcharacters in Heliconius butterflies (Lepidoptera Nymphali-dae) a cladistic review Zoological Journal of the Linnean Society119 457ndash472

BRUNEAU A J J DOYLE and J D PALMER 1990 A chloroplastDNA inversion as a subtribal character in the Phaseoleae (Le-guminosae) Systematic Botany 15 378ndash386

CLEGG M T B S GAUT G H LEARN JR and B R MORTON

1994 Rates and patterns of chloroplast DNA evolution Pro-ceedings of the National Academy of Sciences USA 91 6795ndash6801

CURTIS S and M T CLEGG 1984 Molecular evolution of chloro-plast DNA sequences Molecular Biology and Evolution 1 291ndash301

DE MELO N F and M GUERRA 2003 Variability of the 5S and45S rDNA sites in Passiflora L species with distinct base chro-mosome numbers Annals of Botany 92 309ndash316

mdashmdashmdash A C CERVI and M GUERRA 2001 Karyology and cyto-taxonomy of the genus Passiflora L (Passifloraceae) Plant Sys-tematics and Evolution 226 69ndash84

DE ROCHER E K HARKINS D GALBRAITH and H BOHNERT1990 Developmentally related systemic endopolyploidy insucculents with small genomes Science 250 99ndash101

DE WILDE J 1972 The indigenous Old World Passifloras Blumea20 227ndash250

DELIDOW B J J LYNCH J PELUSO and B A WHITE 1993 Poly-merase chain reaction in PCR protocols methods in molecularbiology eds B A White Totowa Humana Press Inc

DIERS L 1961 Der Anteil an Polypliden in den Vegetationsgurtelnder Westkordillere Perus Zeitschrift fur Botanik 49 443

DOWNIE S R and J D PALMER 1992 Use of chloroplast DNArearrangements in reconstructing plant phylogeny Pages 14ndash35 in Molecular systematics of plants eds P S Soltis D ESoltis and J J Doyle New York Chapman and Hall

mdashmdashmdash E LLANAS and D S KATZ-DOWNIE 1996 Multiple inde-pendent losses of the rpoC1 intron in angiosperm chloroplastDNArsquos Systematic Botany 21 135ndash151

DOYLE J J and J L DOYLE 1987 A rapid DNA isolation proce-dure for small quantities of fresh leaf material PhytochemistryBulletin 19 11ndash15

ESCOBAR L K 1986 New species and varieties of Passiflora Pas-sifloraceae from the Andes of South America Systematic Bot-any 11 88ndash97

mdashmdashmdash 1988 Passifloraceae Pages 1ndash138 in Flora de Colombia Bo-gota Colciencias

FABIJAN D J PACKER and K DENFORD 1987 The taxonomy ofthe Viola nuttali complex Canadian Journal of Botany 65 2562ndash2580

FEUILLET C and J C MACDOUGAL 1999 Infrageneric classifica-tion of Passiflora in XVI International Botanical Congress StLouis Mo

mdashmdashmdash 2003 [2004] A new infrageneric classification of PassifloraL (Passifloraceae) Passiflora 13 (2) 34ndash35 37ndash38

GADELLA T 1969 Chromosome numbers of some angiospermscollected in Cameroon and the Ivory Coast Proceedings of theRoyal Netherlands Academy of Science C 72 306ndash310

mdashmdashmdash 1970 Chromosome numbers of some Angiospermae col-lected in Cameroon and the Ivory Coast Acta Botanica Neer-landica 19 431ndash435

GILBERT L E 1975 Ecological consequences of a coevolved mutualismbetween butterflies and plants Austin University of Texas Press

mdashmdashmdash and J M MACDOUGAL 2000 Passiflora microstipula a newspecies of Passifloraceae from southeast Mexico Lundellia 31ndash5

GOLDMAN N J P ANDERSON and A G RODRIGO 2000 Likeli-hood-based tests of topologies in phylogenetics SystematicBiology 49 652ndash70

GOLENBERG E M M T CLEGG M L DURBIN J DOEBLEY andD P MA 1993 Evolution of a noncoding region of the chlo-roplast genome Molecular Phylogenetics and Evolution 2 52ndash64

GREEN P S 1972 Passiflora in Australia and the Pacific Kew Bul-letin 26 539ndash558

GUERRA M and S DOS 1986 Citogenetica de angiospermas col-etadas en Pernambuco I Revista Brasileria de Genetica 9 21ndash40

HAIR J and E BEUZENBERG 1959 Contributions to a chromosome


atlas of the New Zealand Flora 2 New Zealand Journal of Sci-ence 1 148ndash156

HARMS H 1925 Passifloraceae Pages 470ndash507 in Pflazenfamilieneds A Engler and K Prantl Leipzig W Engelmann

HARVEY M 1966 IOPB chromosome number reports VII Taxon15 155ndash163

HEISER C 1963 Numeracion cromosomica de plantas ecuatori-anas Ciencia y Naturaleza 6 2ndash6

HEITZ E 1926 Der Nachweis der Chromosomen Vergleichendestudien uber ihre Zahl Grosse und Form Pflanzenreich IZeitschrift fur Botanik 18 625ndash681

HILLIS D M 1995 Approaches for assessing phylogenetic accu-racy Systematic Biology 44 3ndash3

HUELSENBECK J P and F RONQUIST 2003 MrBayes Bayesian in-ference of phylogeny httpmorphbankebcuusemrbayesinfophp

JANAKI AMMAL E 1945 Chromosome atlas of cultivated plants edsC D Darlington and E K Janaki Ammal London GeorgeAllen and Unwin Ltd

JANSEN R K and J D PALMER 1987 A chloroplast DNA inversionmarks and ancient evolutionary split in the sunflower family(Asteraceae) Proceedings of the National Academy of SciencesUSA 84 5818ndash5822

KAY E 2000 Observations on the pollination of Passiflora penduli-flora Biotropica 33 709ndash713

KELCHNER S A 2000 The evolution of non-coding chloroplastDNA and its application in plant systematics Annals of theMissouri Botanical Garden 87 482ndash498

KILLIP E P 1938 The American species of Passifloraceae Publi-cations of the Field Museum of Natural History Botanical Series19 1ndash613

KUGLER E E and L A KING 2004 A brief history of the pas-sionflower Pages 15ndash26 in Passiflora passionflowers of the worldby T Ulmer and J MacDougal Portland Timber Press

LA COUR L 1952 Chromosome counts of species and varieties ofgarden plants Annual Report-John Innes Horticultural Institute41 23

LLOYD R 1963 Tetraploid Passiflora incarnata in North CarolinaRhodora 65 79ndash80

LOOCKERMAN D and R K JANSEN 1996 The use of herbariummaterial for DNA studies Pages 205ndash220 in Sampling the greenworld innovative concepts of collection preservation and storage ofplant diversity eds T Stuessy and S Sohmer New York Co-lumbia University Press

MACDOUGAL J C 1994 Revision of subgenus Decaloba sectionPseudodysosmia (Passifloraceae) Systematic Botany Monographs41 1ndash146

mdashmdashmdash and A K HANSEN 2003 A new section of Passiflora sub-genus Decaloba (Passifloraceae) from Central America withtwo new species Novon 13 459ndash466

MADDISON W and D MADDISON 2000 MacClade version 40Sunderland Sinauer Associates Inc

MASTERS M T 1872 Passifloraceae Pages 536ndash540 in Flora Brasi-liensis ed C F P de Martius Leipzig Monachii

MORAWETZ W 1986 Remarks on karyological differentiation pat-terns in tropical woody plants Plant Systematics and Evolution152 49ndash100

MORTON B R and M T CLEGG 1995 Neighboring base com-position is strongly correlated with base substitution bias ina region of the chloroplast genome Journal of Molecular Evo-lution 41 597ndash603

MUSCHNER V A LORENZ A C CERVI S BONATTO T SOUZA-CHIES F SALZANO and L FREITAS 2003 A first molecularanalysis of Passiflora (Passifloraceae) American Journal of Bot-any 90 1229ndash1238

POSADA D and K CRANDALL 1998 ModelTest testing the modelof DNA substitution Bioinformatics 14 817ndash818

POTTER D J J LUBY and R E HARRISON 2000 Phylogenetic

relationships among species of Fragaria (Rosaceae) inferredfrom non-coding nuclear and chloroplast DNA sequencesSystematic Botany 25 337ndash348

PRESTING D 1965 Zur Morphologieder Pollenkorner der Passi-floraceen Pollen et Spores 7 193ndash247

RAMBAUT A and N GRASSLY 1997 Seq-Gen an application forthe monte carlo simulation of DNA sequence evolution alongphylogenetic trees Computational Applied Biosciences 13 235ndash238

RAUBESON L A and R K JANSEN 1992 A rare chloroplast-DNAstructural mutation is shared by all conifers Biochemical Sys-tematics and Ecology 20 17ndash24

mdashmdashmdash and mdashmdashmdash 2004 Chloroplast genomes of plants Pages45ndash68 in Diversity and evolution of plantsmdashgenotypic and phe-notypic variation in higher plants ed R Henry London CABIPublishing

RAVEN P 1975 The bases of angiosperm phylogeny cytology An-nals of the Missouri Botanical Garden 62 724ndash764

SAMBROOK J E F FRITSCH and T MANIATIS 1989 Molecular clon-ing a laboratory manual ed 2 New York Cold Springs HarborLaboratory Press

SIMMONS M P and H OCHOTERENA 2000 Gaps as characters insequence-based phylogenetic analyses Systematic Biology 49369ndash381

SMALL R L J A RYBURN R C CRONN T SEELANAN and J FWENDEL 1998 The tortoise and the hare choosing betweennoncoding plastome and nuclear Adh sequences for phylog-eny reconstruction in a recently diverged plant group Amer-ican Journal of Botany 85 1301ndash1315

SNOW N and J M MACDOUGAL 1993 New chromosome reportsin Passiflora (Passifloraceae) Systematic Botany 18 261ndash273

SOLIS-NEFFA V and A FERNANDEZ 1993 Estudios cromosomicosen especies de Turnera (Turneraceae) Bonplandia 7 101ndash118

SPIRLET M L 1965 Utilization taxonomique des grains de pollende Passifloraceaees Pollen et Spores 7 249ndash301

STOREY W 1950 Chromosome numbers of some species of Passi-flora occurring in Hawaii Pacific Science 4 37ndash42

SUZUKI Y G V GLAZKO and M NEI 2002 Overcredibility ofmolecular phylogenies obtained by Bayesian phylogeneticsProceedings of the National Academy of Sciences USA 99 16138ndash16143

SWOFFORD D L 2002 PAUP phylogenetic analysis using parsi-mony (and other methods) version 40b10 Sunderland Sin-auer Associates

mdashmdashmdash G OLSEN P WADDEL and D M HILLIS 1996 Phyloge-netic inference Pages 407ndash515 in Molecular systematics eds DM Hillis C Moritz and B K Mable Sunderland SinauerAssociates

TABERLET P L GIELLY G PAUTOU and J BOUVET 1991 Universalprimers for amplification of three non-coding regions of chlo-roplast DNA Plant Molecular Biology 17 1105ndash1110

THOMPSON J D D G HIGGINS and T J GIBSON 1994 CLUS-TALW improving the sensitivity of progressive multiple se-quence alignment through sequence weighting positions-specific gap penalties and weight matrix choice Nucleic AcidsResearch 22 4673

ULMER T and J MACDOUGAL 2004 Passiflora passion flowers of theworld Portland Timber Press

URIBE-URIBE L 1955 Passifloraceas y Begoniaceas de la Real Ex-pedicion Botanica de Nuevo Reino de Granada Ediciones Cul-tura Hispanica 26 1ndash98

VAN HAM R C H J H T HART T H M MES and J M SAND-BRINK 1994 Molecular evolution of noncoding regions of thechloroplast genome in the Crassulaceae and related speciesCurrent Genetics 25 558ndash566

WILCOX T P D J ZWICKL T A HEATH and D M HILLIS 2002Phylogenetic relationships of the dwarf boas and a compar-


ison of Bayesian and bootstrap measures of phylogeneticsupport Molecular Phylogenetics and Evolution 25 361ndash371

YOCKTENG R S and S NADOT 2004 Phylogenetic relationshipsamong Passiflora species based on the Glutamine Synthasenuclear gene expressed in the chloroplast (ncpGS) MolecularPhylogenetics and Evolution 31 379ndash396

APPENDIX 1Species in alphabetical order sampled for rpoC1 intron survey

and chloroplast trnLtrnT spacer sequence phylogeny Subgenericplacement (Killip 1938 followed by Feuillet and MacDougal 2004)is provided after species names (lsquolsquoNArsquorsquo if not treated) Presence ofabsence of rpoC1 intron is denoted by a (1) or (-) respectivelyChromosome numbers if known and references are listed Plantmaterial was field collected by A L Escobar L Gilbert K Han-sen J MacDougal or Jose Panero or cultivated at the University ofTexas by L E Gilbert or at Butterfly World by Ron Boender Allvouchers are held at HUA DUKE TEX or UPR Sequences aredeposited in GenBank under accession numbers DQ096733ndashDQ096793

Adenia mannii Engl NA NA 1 Africa L Escobar 92ndash38 (TEX)DQ96755 n512 (de Rocher et al 1990)

Tetrapathea tetranda Cheeseman NA NA 1 Brazil L Escobar 71(TEX) DQ96754 n512 (Hair amp Beuzenberg 1959)

Tetrastylis ovalis Vell Ex M Roem) Killip NA Deidamioides 1Brazil K Hansen 152 (TEX) DQ96752

Passiflora actinia Hook Passiflora Passiflora 1 Brazil K Hansen36 (TEX) DQ96792 n59 (De Melo et al 2001) P acuminata DCPassiflora Passiflora 1 French Guiana L Escobar sn (UPR)DQ96740 P adenopoda DC Decaloba Decaloba - J MacDougal 2012(MO) NA P adulterina L Tacsonia Passiflora 1 Colombia L Es-cobar sn (HUA) NA P affinis Englem Decaloba Decaloba - TexasK Hansen 251 (TEX) DQ96764 P alata Curtis Passiflora Passiflora1 Brazil K Hansen 248 (TEX) DQ096733 n59 Guerra 1999 Pallantophylla Mast Decaloba Decaloba - Guatemala J MacDougal638 (DUKE) NA P alnifolia Kunth Decaloba Decaloba - ColombiaK Hansen 141 (TEX) NA P ambigua Hemsl Passiflora Passiflora1 Costa Rica L Escobar 88-22 (TEX) NA P amethystina JC Mi-kan Passiflora Passiflora 1 Brazil K Hansen 40 (TEX) NA n59(De Melo et al 2001) P amoena LK Escobar NA Astrophea 1French Guiana L Escobar 9869 (UPR) NA P ampullacea (Mast)Harms Tacsonia Passiflora 1 Colombia L Escobar 88-53 (TEX)DQ96776 P andina Killip Rathea Passiflora 1 Ecuador K Hansen50 (TEX) DQ96751 P anfracta Mast amp Andre Decaloba Decaloba- Ecuador K Hansen sn (TEX) NA P antioquiensis H KarstTacsonia Passiflora 1 L Escobar 88ndash51 (TEX) DQ96779 n59(Snow amp MacDougal 1993) P apetala Killip Decaloba Decaloba -Costa Rica K Hansen 147 (TEX) NA P arbaelezii L Uribe NADeidamioides Cultivated UT L Escobar 92ndash8 (TEX) NA P arboreaSpreng Astrophea Astrophea 1 Ecuador L Escobar 8581 (TEX)NA P aurantia G Forester NA Decaloba 1 Australia K Hansen201 (TEX) DQ96784 n56 (Beal 1971) P auriculata Kunth DecalobaDecaloba - Costa Rica James P Folsom 9078 (TEX) DQ96765

P biflora Lam Decaloba Decaloba - Costa Rica K Hansen sn(TEX) DQ96787 n56 (Beal 1971) P boenderii JM MacDougal NADecaloba - Costa Rica K Hansen 210 (TEX) NA P buchtienii KillipDistephana Passiflora 1 L Escobar 92ndash60 (TEX) NA

P caerulea L Passiflora Passiflora 1 Paraguay K Hansen 16(TEX) NA n59 (Heitz 1926) P campanulata Mast DysosmioidesPassiflora 1 Brazil K Hansen 46 (TEX) DQ96775 P candida(Poepp amp Endl) Mast Astrophea Astrophea 1 French Guiana LEscobar 9901 (TEX) DQ96786 P candollei Triana amp Planch NADecaloba - Colombia K Hansen 186 (TEX) DQ96770 n56 (Snowand MacDougal 1993) P capsularis L Decaloba Decaloba - L Es-cobar sn (TEX) NA n56 (Beal 1971) P ciliata Aiton DysosmiaPassiflora 1 L Escobar 92ndash71 (TEX) DQ96790 P cincinnata MastPassiflora Passifora 1 Argentina K Hansen 175 (TEX) NA P cir-rhiflora Juss Polyanthea Deidamioides 1 French Guiana K Hansen

222 (TEX) DQ96756 n59 (Beal 1971) P citrifolia (Juss) Mast As-trophea Astrophea 1 L Escobar 9921 (TEX) NA P coactilis (Mast)Killip Tacsonia Passiflora 1 EcuadorJose Panero (TEX) NA P coc-cinea Aubl Distephana Passiflora 1 Brazil K Hansen 145 (TEX)DQ96777 P complanata JM MacDougal inedit NA Decaloba -Mexico K Hansen 171 (TEX) NA n59 (Beal 1971) P conzattiannaKillip Decaloba Decaloba - Mexico K Hansen 143 (TEX) NA Pcoriaceae Juss Decaloba Decaloba - Belize K Hansen 246 (TEX)NA n56 (Snow and MacDougal 1993) P costaricensis Killip De-caloba Decaloba - Costa Rica K Hansen 146 (TEX) NA n56 (Beal1971) P crenata Feuillet amp Cremers NA Passiflora 1 French Gui-ana K Hansen 216 (TEX) DQ96737 n56 (Snow and MacDougal1993) P crispolanata L Uribe Tacsonia Passiflora 1 L Escobar 92ndash58 (TEX) NA P cumbalensis (H Karst) Harms Tacsonia Passiflora1 Ecuador L Escobar 88ndash46 (HUA) DQ96773

P deidamioides Harms Deidamioides Deidamioides 1 Brazil KHansen 152 (TEX) DQ96757 P discophora P Joslashrg amp Lawesson NADeidamioides 1 Ecuador K Hansen sn (TEX) NA

P edulis Sims Passiflora Passiflora 1 Colombia L Escobar 88ndash14(TEX) DQ96743 P eichleriana Mast Passiflora Passiflora 1 BrazilCultivated UT NA n59 (Janaki Ammal 1945) P escobariana JMMacDougal NA Decaloba - Colombia L Escobar sn (TEX) NAP ernestii Harms Adenosepala Passiflora 1 GH NA n56 (Snowand MacDougal 1993) P exura Feuillet NA Passiflora 1 FrenchGuiana L Escobar sn (UPR) NA

P foetida L Dysosmia Passiflora 1 Mexico L Escobar 88ndash16(TEX) NA P ferruginea Mast Decaloba Decaloba - French GuianaL Escobar 92ndash15 (TEX) NA n59 Guerra 1986 n510 Storey 1950n511 (Harvey 1966)

P garkei Mast Passiflora Passiflora 1 French Guiana K Hansen177 (TEX) DQ96763 P gilbertiana JM MacDougal NA Decaloba- L Escobar sn (TEX) NA P glandulosa Cav Distephana Passiflora1 L Escobar 9881 (TEX) DQ96791 n56 (Snow and MacDougal1993) P guatemalensis S Watson Decaloba Decaloba 1 GuatemalaK Hansen 201 (TEX) DQ96762 n59 (De Melo et al 2001)

P hahnii (E Fourn) Mast Decaloba Decaloba 1 Cultivated UTNA n511n512 John Macdougal P herbertiana Ker Gawl NADecaloba 1 Australia L Escobar 88ndash48 (TEX) DQ96783 P holo-sericea L Decaloba Decaloba - Guatemala K Hansen 182 (TEX)DQ96759 n56 (Beal 1969)

P incarnata L Passiflora Passiflora 1 Texas L Escobar 9264(TEX) NA n57 (Snow and MacDougal 1993)

P jamesonii (Mast) Bailey Tacsonia Passiflora 1 Ecuador K Han-sen 60 (TEX) NA n59 (Heitz 1926)

P karwinskii Mast Decaloba Decaloba - Mexico K Hansen 227(TEX) NA P kawensis Feuillet NA Astrophea 1 French GuianaL Escobar 9910 (TEX) DQ96769 n56 (MacDougal 1983)

P lancearia Mast Decaloba Decaloba - Costa Rica K Hansen 172(TEX) NA P lancetillensis JM MacDougal amp Meerman NA De-caloba 1 L Gilbert sn (TEX) DQ96760 P laurifolia L PassifloraPassiflora 1 French Guiana L Escobar 88ndash46 (TEX) DQ096734 Pligularis Juss Passiflora Passiflora 1 Ecuador K Hansen 81(TEX) DQ96736 n59 (Storey 1950) P lindeniana Planch Ex Trianaamp Planch Astrophea Astrophea 1 Venezuela L Gilbert sn (TEX)NA n59 (Storey 1950) P lobata (Killip) Hutch ex JM MacDougalTetrastylis (genus) Decaloba - Costa Rica K Hansen 258 (TEX)NA n512 (Berry 1987) P lutea L Decaloba Decaloba - L Escobarsn (TEX) NA n57 (MacDougal 1983)

P macrophylla Spruce ex Mast Astrophea Astrophea 1 EcuadorL Gilbert sn (TEX) NA 2n524 (Baldwin 1949) 2n584 (Bowden1940) n56 (De Melo et al 2001) P magnifica LK Escobar NAPassiflora 1 Colombia K Hansen 154 (TEX) DQ96735 P malifor-mis L Passiflora Passiflora 1 Colombia K Hansen 156 (TEX)DQ967 n59 (Snow and MacDougal 1993) P manicata (Juss) PersTacsonia Passiflora 1 Ecuador K Hansen 51 TEX) DQ96778 n59(Storey 1950) P mathewsii (Mast) Killip Tacsonia Passiflora 1 Ec-uador J MacDougal (TEX) NA n59 (Storey 1950) P mayarum JMMacDougal NA Passiflora 1 Guatemala K Hansen 117 (TEX)


NA P membranacea Benth Decaloba Decaloba 1 Mexico L Escobar88ndash50 (TEX) NA P menispermifolia Kunth Passiflora Passiflora 1Costa Rica L Escobar 88ndash13 (TEX) DQ96785 P microstipula LEGilbert amp JM MacDougal NA Decaloba 1 Vera Cruz MexicoL Gilbert sn (TEX) DQ96782 DQ96789 P microstipula LE Gilbertamp JM MacDougal NA Decaloba 1 Oaxaca Mexico L Gilbertsn (TEX) DQ96758 n59 (Snow and MacDougal 1993) P miersiiMast Passiflora Passiflora 1 Brazil K Hansen 153 (TEX)DQ96766 n59 (Snow and MacDougal 1993) P misera Kunth De-caloba Decaloba - Ecuador L Escobar 88ndash26 (TEX) NA P mixta Lf Tacsonia Passiflora 1 Ecuador L Escobar 88ndash45 (TEX) NA n56(De Melo et al 2001) P mucronata Lam Passiflora Passiflora 1Cultivated UT NA n59 (La Cour 1952) P multiflora L ApodogyneDecaloba 1 Florida K Hansen 189 (TEX) DQ96793 n59 (De Meloet al 2001) P murucuja L Murucuja Decaloba - Puerto Rico KHansen 219 (TEX) DQ96788

P nelsonii Mast amp Rose Passiflora Passiflora 1 Costa Rica LEscobar 88ndash24 (TEX) NA P nephrodes Mast Passiflora Passiflora 1Brazil K Hansen 211 (TEX) NA P nitida Kunth Passiflora Passi-flora 1 French Guiana K Hansen 217 (TEX) DQ96739

P oblongata Sw Pseudomurucuja Decaloba - Jamaica L Escobarsn (UPR) DQ96772 n59 (De Melo et al 2001) P oerstedii MastPassiflora Passiflora 1 Costa Rica L Gilbert sn (TEX) NA P affOerstedii Mast Passiflora Passiflora 1 Ecuador K Hansen 242(TEX) NA P organensis Gardner Decaloba Decaloba - Brazil KHansen 195 (TEX) NA

P pallens Poepp ex Mast Passiflora Passiflora 1 Florida L Es-cobar sn (TEX) NA P palmeri Rose Dysosmia Passiflora 1 Cul-tivated UT NA P parritae (Mast) LH Bailey Tacsonia Passiflora1 Colombia L Escobar sn (HUA) DQ96780 P penduliflora Berteroex DC Astephia Decaloba - Jamaica K Hansen 241 (TEX) NA Pperfoliata L Pseudomurucuja Decaloba - L Escobar 88ndash36 (TEX)DQ96761 n56 (Beal 1971) P pitteri Mast Astrophea Astrophea 1Costa Rica K Hansen 155 (TEX) DQ96768 n56 (Beal 1971) Pplatyloba Killip Passiflora Passiflora 1 Cultivated UT DQ96738 Pporphyretica Mast Pseudogranadilla Decaloba - L Escobar sn(HUA) NA

P quadrangularis L Passiflora Passiflora 1 Costa Rica K Hansen183 (TEX) NA n56 (Snow and MacDougal 1993) P quindiensisKillip Tacsonia Passiflora 1 Colombia L Escobar sn (HUA) NAn59 (Janaki Ammal 1945) P quinquangularis J MacDougal Deca-loba Decaloba - Guatemala K Hansen 146 (TEX) NA

P racemosa Brot Calopathanthus Passiflora 1 Brazil K Hansen157 (TEX) DQ96746 n56 (Snow and MacDougal 1993) P reflexi-

flora Cav Tacsonioides Passiflora 1 Ecuador K Hansen 72 (TEX)DQ96750 n59 (Bowden 1945) P resticulata Mast amp Andre Passi-flora Passiflora 1 Cultivated UT NA P retipetala Mast PassifloraPassiflora 1 Trinidad L Escobar sn (HUA) NA n59 (Escobar1986) P rovirosae Killip Decaloba Decaloba - Mexico L Gilbert 8056(TEX) NA P rubra L Decaloba Decaloba - Ecuador K Hansen 245(TEX) NA n56 (Snow and MacDougal 1993)

P sanguinolenta Mast amp Linden Decaloba Decaloba - Ecuador LEscobar sn (HUA) NA n56 (Snow and MacDougal 1993) P see-mannii Griseb Passiflora Passiflora 1 Colombia L Escobar 92ndash83(TEX) DQ96744 n56 (Snow and MacDougal 1993) P serratifoliaL Passiflora Passiflora 1 Belize K Hansen 159 (TEX) NA n59(Storey 1950) P serratodigitata L Passiflora Passiflora 1 TrinidadK Hansen 158 (TEX) DQ96742 P serrulata Jacq Passiflora Passiflo-ra 1 Colombia L Escobar sn (HUA) NA P sexflora Juss Deca-loba Decaloba - L Escobar sn (HUA) NA P speciosa GardnerDistephana Passiflora 1 Brazil L Escobar sn (TEX) DQ96747 Psuberosa L Decaloba Decaloba - Mexico K Hansen 142 (TEX) NAP sublanceolata (Killip) MacDougal Dysosmia Dysosmia CultivatedUT DQ96781 P subpeltata Ortega Passiflora Passiflora 1 Culti-vated UT NA n56 Diers 1961 n512 (Storey 1950)

P tacsonioides Griseb Pseudomurucuja Decaloba - Jamaica KHansen 223 (TEX) NA n59 (Storey 1950) P talamancensis KillipDecaloba Decaloba - Cultivated UT NA P tiliifolia L PassifloraPassiflora 1 Cultivated UT NA P trialata Feuillet and JMMacDougal NA Passiflora 1 French Guiana K Hansen 230(TEX) DQ96741 P tricuspis Mast Decaloba Decaloba - Bolivia KHansen 208 (TEX) NA P trifasciata Lem Decaloba Decaloba - Ec-uador L Escobar sn (HUA) NA n56 (De Melo et al 2001) Ptrifoliata Cav Tacsonia Passiflora 1 Colombia L Escobar sn(HUA) DQ96745 P tripartita (Juss) Poir Tacsonia Passiflora 1Peru L Escobar 88ndash44 (TEX) NA P trisecta Mast Tacsonia Passi-flora 1 Colombia L Escobar sn (HUA) DQ96749 n59 (Heiser1963) P tuberosa Jacq Decaloba Decaloba - Trinidad L Escobar 88ndash6 (TEX) NA P tulae Urban Decaloba Decaloba - Puerto Rico KHansen 167 (TEX) NA

P vespertilio L Decaloba Decaloba - Brazil K Hansen 192 (TEX)DQ96771 P villosa Vell Dysosmioides Passiflora 1 Brazil K Han-sen 38 (TEX) DQ96767 P viridiflora Cav Decaloba Decaloba - Mex-ico L Escobar sn (HUA) NA P vitifolia Kunth Distephana Passi-flora 1 Brazil K Hansen 174 (TEX) DQ96748

P xiikzodz JM MacDougal NA Decaloba - Guatemala L Es-cobar sn (TEX) NA n59 (Storey 1950)

P yucatanensis Killip Decaloba Decaloba - Mexico K Hansen 150(TEX) NA


TABLE 1 Data analyses and tree statistics for the four parsimony analyses of the trnLtrnT spacer region The number of excludedcharacters refer to the number of base pairs comprising gap regions that were removed in analysis B or the number of positions removeddue to homoplasy concerns in analysis D C I values were calculated in PAUP with uninformative characters excluded

AnalysisOptimality

criterion

ofcharactersexcluded

Gapinterleavedcharacters Gap treatment

Parsimonyinformativecharacters

Numberof trees

Treelength C I R I H I

ABCD

ParsimonyParsimonyParsimonyParsimony

NoneNone27964

None7575

None

Missing dataMissing dataRemovedMissing data

881209979

200000200000200000200000

265384302229

0800074507190795

0913088308800916

0200025502810205

Tacsonia Manicata Calopathanthus Dysosmia Dysosmioi-des Distephana Rathea Tacsonioides and Tacsoniopsis asthe lsquolsquon59 grouprsquorsquo (Feuillet and MacDougal 2004)

This dichotomy between the lsquolsquon56rsquorsquo and lsquolsquon59rsquorsquogroups has been well known among Passiflora research-ers for decades (L Escobar unpubl P M Joslashrgensen JMacDougal C Feuillet pers comm Presting 1965)However the first molecular evidence to support theevolutionary integrity of these groups was publishedby Downie et al in 1996 In this study Downie sur-veyed a wide diversity of angiosperms for the presenceor absence of the rpoC1 intron to assess its phyloge-netic utility Although the marker was homoplastic forhigher order relationships preliminary evidence sug-gested that it might be useful in studies at the subfa-milial level particularly in Passiflora In their study of10 species of Passiflora all taxa with a chromosomenumber of n56 lacked the rpoC1 intron while thosethat had n59 retained it The sample size was toosmall in the Downie et al study to make any conclu-sive phylogenetic inferences for Passiflora as a wholebut indicated that a more intensive sampling strategymight provide additional support for the delineationof these two groups As part of our work exploring thephylogenetic relationships in Passiflora we expandedthe intron survey and combined this character with achloroplast DNA sequence phylogeny The combinedresults are presented here along with discussions ofPassiflora relationships and taxonomy chromosomenumber evolution and the relative utility of these mak-ers for resolving evolutionary history

MATERIALS AND METHODS

Chloroplast Phylogeny Sampling included 61 species 57 fromgenus Passiflora representing 16 of Killiprsquos subgenera and one spe-cies each from genera Adenia Forssk Dilkea Mast Tetrapathea DCand Tetrastylis Killip as outgroups (see Appendix) All species in-cluded in the phylogenetic analyses were also examined in theintron survey Total DNA was isolated from single individuals us-ing either fresh leaf material silica dried field collections or her-barium specimens (Appendix 1) Protocols for DNA isolation am-plification and sequencing followed those described in Barber etal (2002) Both strands were sequenced for all taxa

Sequences were edited using Sequencher v 31 (Gene CodesCorp) Clustal X (Thomson et al 1994) was used to perform mul-tiple alignments using pairwise comparisons to obtain an initialalignment In order to identify potential repeat motifs we usedthe Megalign program in DNAstar (DNAstar Madison WI) This

was followed by manual adjustment using two general guidelines1) the number of gaps needed to align sequences was minimized(Golenberg 1993) and 2) in areas of tandem repeats sequenceswere aligned to maximize percent similarity and minimize thenumber of substitutions between sequences (Aldrich 1988)

Parsimony as implemented in PAUP v 40b10 (Swofford 2002)was used for analyses A-D (Table 1) We used the heuristic searchmethod with 100 random addition replications with step-wise ad-dition of taxa tree bi-section reconnection and MulTrees on Inorder to increase the probability that all islands of most-parsi-monious trees would be sampled we ran the analysis on threedifferent computers to find the length of the shortest tree Thiswas then set as the upper limit for tree length for the final analysesand only 2000 trees equal to this value were saved for each rep-licate All subsequent analyses were run on lsquolsquophyloclusterrsquorsquo a NPA-CI Rocks (httprocksclusterorg) cluster comprised of 24 AMD18001 processors Levels of homoplasy were estimated by calcu-lating the CI HI and RI in PAUP Support for monophyleticgroups was evaluated by bootstrap analyses using 100 replicateseach with a lsquoMaxTreesrsquo limit of 2000 per replicate These methodswere utilized for all parsimony analyses

We evaluated the effects of gap coding by doing four differentanalyses using the parsimony methods described above (Table 1)The first analysis (A) included all sequence characters and the gapregions were treated as missing data Gaps were then coded asbinary characters utilizing the simple gap coding method of Sim-mons and Ochoterena (2000) and subsequently appended to thesequence data Since indel events are frequent occurrences in chlo-roplast non-coding regions (Kelchner 2000) removing the result-ing gap regions from the analysis reduced the number of infor-mative characters by 18 Due to the loss of such a significantamount of data in an already information-poor region we ana-lyzed the data with gap regions included and treated as missingdata (analysis B) and with gap regions removed (analysis C) Fi-nally for analysis D the region corresponding to base pairs 425ndash489 was removed This region appears to be a lsquolsquohot-spotrsquorsquo for indelformation and was removed from this analysis to estimate theimpact of such regions on levels of homoplasy in our data Analy-ses A B C and D correspond to TreeBASE (study accessionS1330)

We used Bayesian analyses to evaluate statistically potentialmonophyletic groups in Passiflora The GTR1G model of sequenceevolution was chosen by ModelTest v 306 (Posada and Crandall1998) as the best fit for these data MrBayes v 30 (Huelsenbeckand Ronquist 2003) was used for all analyses Uniform prior dis-tributions were used for 1) alpha shape parameter for the gammadistribution (00 100) 2) instantaneous rate matrix (00 1000)and 3) branch lengths (00 100) Base frequencies were estimatedand the starting trees were randomly generated We used a Me-tropolis-coupled MCMC analysis where four chains were incre-mentally heated and run simultaneously Two separate analyseswere done running for 1ndash3 X106 respectively and every 100th treewas sampled Trees that were generated prior to the plateau ofparameter values were deleted The remaining trees were openedin PAUP vs 410b (Swofford 2002) and used to generate a 50Majority Rule tree Each analysis was subsequently repeated toverify the results

Intron Survey Representative taxa from each subgenus sec-







RESULTS
































DISCUSSION




























LITERATURE CITED

































































































































RESULTS
































DISCUSSION




























LITERATURE CITED






















































































































































DISCUSSION




























LITERATURE CITED





















































































































































LITERATURE CITED



























































































































phylogenetic relationships and chromosome number evolution

Documents