Epidendrum Flammeus Complete INTEGRATIVE

8/13/2019 Epidendrum Flammeus Complete INTEGRATIVE

1/19

Pessoa & al. Different tools to delimitEpidendrumspeTAXON61 (4) August 2012: 721734

INTRODUCTION

Species delimitation is an old issue with an extensie theo-

retical and methodological body of literature (de Queiro, 1998;Sites & Marshall, 2004 and references therein). This subject

has recently attracted renewed attention, which is emphasied

by the wide interest in barcoding approaches (Lahaye & al.,

2008; Hollingsworth & al.,2009; Pires & Marinoni, 2010). Be-

cause species are the main units of ecological and eolutionary

studies, the identification of boundaries among closely related

species, and therefore the inference of the number of extant

species, is an essential target of current systematic studies (Du-

minil & al.,2006; Petit & Excoffier, 2009). To achiee this goal,

the cross-alidation of species identification using different

approaches is desirable (but see Padial & al., 2010), and dif-

ferent tools are aailable to depict complex morphological and

genetic patterns of ariation, such as morphometry (Pinheiro

called integratie taxonomy (Dayrat, 2005; Padial & al., 20

Schlick-Steiner & al.,2010).

The perspectie that species can be delimited based

multiple lines of eidence has been boosted by the increing eidence that reproductie barriers are permeable to g

flow (porous genomes, Wu, 2001), that species lineages m

dierge despite hybridiation and introgression and that a sin

species can originate polyphyletically by parallel eolut

(reiewed in Hausdorf, 2011). Indeed, the general lineage c

cept of species (de Queiro, 1998) can be used as a pro

conceptual framework related to the use of different type

data to identify species. According to this concept, differ

patterns of diergence for different types of data and deli

tation criteria may arise at different times in the proces

speciation (de Queiro, 1998). For example, different spe

criteria such as intrinsic reproductie isolation (Mayr, 194

reciprocal monophyly (de Queiro & Donoghue, 1988)

Integrating different tools to disentangle species complexes:A case study in Epidendrum(Orchidaceae)

Edlley Max Pessoa,Marccus Alves,Anderson Alves-Arajo,Clarisse Palma-Silva& Fbio Pinheiro

1 Universidade Federal de Pernambuco, Centro de Cincias Biolgicas, Departamento de Botnica, 50670-420, Recife, PE, Brazil

2 Instituto de Botnica, 04301-902, So Paulo, SP, Brazil

Author for correspondence:Fbio Pinheiro, [email protected]

Abstract Species delimitation remains a central problem in systematic, taxonomic and evolutionary studies. However, the

precise delimitation of species depends on the criteria used to identify lineages and the specific species concept that is applied.

Recently, multidisciplinary studies using different data sources have significantly improved the delimitation of species within

complex taxonomic groups, leading to an integrative taxonomy. To investigate the species delimitation within the Atlantic

clade ofEpidendrum(subg.Amphyglottium), four different species criteria were examined (phenetic differentiation, mono-

phyly, diagnosability, absence of genetic intermediates). Morphometrics, plastid DNA sequences and nuclear microsatellite

markers were used to explore the agreement between patterns recovered and species criteria tested. The conflicts amongspecies criteria are discussed in light of pollination ecology, patterns of gene f low, reproductive isolation mechanisms and

selective pressures currently acting in deceptive orchid species. Four currently recognized species from the Atlantic clade

could be delimitated, including one newly described species,Epidendrum f lammeus. Three out of f ive species satisfied the

monophyly criterion, and few diagnostic flower characters were found among species. In contrast, nuclear microsatellite

data correctly assigned individuals to their respective species, even in the presence of weak reproductive isolation and ex-

tensive hybridization events reported in the literature. One important implication of this finding is that phylogenetic studies

inEpidendrumspp. should make use of single- or low-copy nuclear loci instead of plastid markers, which may be true for

other plant groups. The results also indicate that the generalized pollination syndrome found among species of the Atlantic

clade, the different levels of gene flow observed between nuclear and plastid markers, and hybridization events are com-

monly observed as the main evolutionary forces within this orchid group. Finally, we discuss evolutionary processes and

taxonomic limits among these species, and we highlight the need to increase the inter-disciplinary approach to investigate

species limits in complex plant groups.

Keywords hybridization; orchid evolution; reproductive barriers; speciation; species complex; species delimitation

Supplementary Material Tables S1S2 and Figures S1S3 (in the Electronic Supplement) and the alignment are available in the

Supplementary Data section of the online version of this article(http://www.ingentaconnect.com/content/iapt/tax).
http://www.ingentaconnect.com/content/iapt/taxhttp://www.ingentaconnect.com/content/iapt/tax


2/19

TAXON61 (4) August 2012: 721Pessoa & al. Different tools to delimitEpidendrumspecies

scenarios has been inestigated, leading to the conclusion that

the application of seeral different criteria should be ealu-

ated considering the releant forces driing speciation process

(reiewed by Padial & al., 2010).

Whereas qualitatie traits are relatiely easy to measure,

the ast majority of phenotypic differences between popula-

tions, species, or higher-order taxa are quantitatie(Lynch

& Walsh, 1998) and will thus require accurate and objectie

measurement and analysis. The use of multiariate methods

to analye quantitatie morphometric data proides the op-

portunity to summarie large numbers of characters in few

discriminant components, allowing the identification of cryptic

borders between closely related species (Borba & al.,2002) or

the mosaic nature of recombinant hybrids (Lexer & al.,2009).

Furthermore, traditional characters used to differentiate spe-

cies, such as morphology, are increasingly supplemented by

approaches based on molecular genetic data. Distinguish-

ing species that hae low leels of genetic diergence, eitherbecause speciation is recent or because species continue to

exchange genes remains a challenging task in eolutionary

biology and molecular taxonomy (Duminil & al., 2006; Lexer

& al., 2009). The inestigation of species limits is greatly fac

tated by the aailability of molecular markers informati

the species leel, from different genomic compartments, m

typically nuclear and cytoplasmic genomes (Duminil &

2006; Moccia & al.,2007; Palma-Sila & al.,2011). Accord

to Petit & Excoffier (2009) and Hausdorf & Hennig (2010),

use of multiple unlinked nuclear markers experiencing h

intraspecific gene flow leels coupled with the use of mod

based assignment methods proides the best results to deli

species. Nuclear microsatellites hae been pointed out as hig

informatie markers for interspecific comparisons (Dum

& al.,2006; Hausdorf & Hennig, 2010; Palma-Sila & al.,20

High cross-transferability among species has been obsered

nuclear microsatellites (Barbar & al., 2007; Pinheiro &

2009b), and the occurrence of homoplasy is oercome by

use of multiple unlinked loci (Duminil & Di Michele, 200

Our study group, Epidendrum L. (Orchidaceae), is

largest (1500 species) and most widespread (Southern UnStates to North Argentina) Neotropical orchid genus (Hgs

& Soto-Arenas, 2005).Epidendrumis renowned for uncerta

ties regarding delimitation within species complexes, beca

Fig. .Geographical distribution

ofEpidendrumspecies of the

Atlantic clade. A,Latin Ameri-

can map; B,detail of Brailian

Atlantic coast.


3/19


many species show impressie morphological ariation. Studies

within the genus are primarily limited to the description of new

species based on qualitatie morphological data, and comple-

mentary taxonomic and eolutionary studies are comparatiely

rare (see reiew by Hgsater & Soto-Arenas, 2005). Blurred

morphological boundaries between many closely related spe-

cies are common within the genus, and detailed data describing

patterns of intraspecific morphological ariation are restricted

to few species (Pinheiro & Barros, 2007a, b). Recently, a highly

informatie set of nuclear (Pinheiro & al., 2008a, b) and plastid

markers (Pinheiro & al.,2009c) was deeloped for Epiden-

drum, which contributed substantially to the inestigation of

eolutionary processes at low taxonomic leels within the genus

(Pinheiro & al., 2010, 2011).

Species of Epidendrum subg. Amphyglottium (Salisb.)Brieger were first recognied as a monophyletic group by Hg-

sater & Soto-Arenas (2005). Recently, Pinheiro & al. (2009a)

expanded the preious phylogenetic study of Hgsater & Soto-Arenas (2005) by analying 14 species. Strongly supported

clades representing distinct biogeographical regions were re-

coered, such as the Andean clade, with species mainly dis-

tributed along the Andean and Guianan mountain ranges, and

the Atlantic clade, which includes species distributed along

the Brailian coast, in the Cerrado and Caatinga egetation

domains (Fig. 1). The Atlantic clade comprises four species

(Epidendrum cinnabarinumSalm. ex Lindl., E. denticula-

tumBarb. Rodr., E. fulgensBrongn. and E. puniceoluteum

Pinheiro & Barros) (Fig. S1 in the Electronic Supplement).

Extensie ariation of morphological traits, allopatric s. sym-

patric/parapatric occurrence and wide geographic distribution

offer interesting possibilities for inestigating species limits,

reproductie isolation and phylogeography within this group.

Hybridiation and late generation introgression betweenE. ful-

gensandE. puniceoluteumwere identified using nuclear and

plastid microsatellites (Pinheiro & al.,2010), which suggests

an important role of interspecif ic gene flow in generating mor-

phological and chromosome number diersification. Moreoer,

pure parental indiiduals of both species and hybrid indiiduals

were identified with high confidence using Bayesian assign-

ment analysis, in agreement with preious obserations based

on morphological characters (Pinheiro & Barros, 2006). Eenwithin the geographical range of E. fulgens it was possible

to identify important phylogeographic breaks and historical

demographic eents (gene flow, bottleneck) that shaped the

current genetic patterns obsered in natural populations (Pin-

heiro & al.,2011).

To explore species limits and speciation hypotheses among

closely related species, the general lineage concept of species

was used as the conceptual framework to test the utility of dif-

ferent species criteria, in the context of integratie taxonomy.

The following four criteria for species delimitation, based on

morphological and DNA polymorphism, were tested: (1) phe-

netic differentiation (Sokal & Croello, 1970); (2) reciprocal

monophyly (de Queiro & Donoghue, 1988); (3) diagnosability,

of a new species was also tested in our study, described be

asE. flammeus. Qualitatie morphological characters, morp

metric data and multiple nuclear and plastid markers cle

indicate species limits within the Atlantic clade, clarify

eolutionary and speciation processes within this group.

cross-alidation of species identification using different to

following the general lineage species concept and an integ

tie taxonomic approach, proed to be useful for depict

complex eolutionary relationships that are commonly fo

among many plant groups.

MATERIALS AND METHODS

Group of interest. All species of the Atlantic clade w

analyed in this study.Epidendrum flammeusoccurs within

geographical domain of the Atlantic clade, and shares m

morphological characters withE. fulgensandE. denticulatAll of the specimens ofE.flammeuswere collected in one

nitic rock outcrop within the Borborema mountain range

northeastern Brail. Samples ofE. flammeuswere found du

recent fieldwork conducted to make floristic inentories of r

outcrops in the Serra da Borborema massif, within Caatin

Preliminary qualitatie morphological comparisons with o

specimens of subg.Amphyglottiumindicated that these col

tions might represent a new species. Additional herbarium m

rial was also found from close geographic localities, identi

asE. fulgens(see commentaries in species description). For

reason, samples ofE. flammeuswere included in this study

its status as a new species was inestigated. Population or

and sample sies are described in Table 1.Morphometric analysis. To quantify continuous

terns of morphological ariation among species of the Atla

clade (E. cinnabarinum,E. denticulatum,E. fulgens,E. pu

ceoluteum) and the new species described, ariation inflower traits (TableS1 in the Electronic Supplement) was m

ured in 420 indiiduals, using a digital ruler and follow

the procedure described in Pinheiro & Barros (2007a). Dis

minant function analysis (DA) was used to assess multi

ate morphological differentiation among populations. Wi

Lambda, jackknifed classification, which assigns unclassispecimens to groups, and F-to remoe statistics, which pro

an indication of the relatie importance of each ariable,

also reported. Mean and standard deiation alues for e

character and each species were calculated, and are depic

in Box-plot graphics. Data were analysed and tested for m

tiariate normality using SYSTAT .11.0. (SYSTAT Softw

Richmond, California, U.S.A.).

Molecular phylogenetic study. A recent molec

phylogenetic study of subg. Amphyglottium(Pinheiro &

2009a) was used as the basic framework to test the phylogen

position ofE. flammeus. The precise phylogenetic relationsh

of the new species were then determined by sequencing th

plastid regions, trnT-trnL, trnL-trnFand rpl32-trnLfor all s


4/19


described E. flammeus. Epidendrum campestreLindl. and

E. nocturnumJacq. were used as outgroups. DNA extraction,

amplification reactions and plastid DNA sequencing were car-

ried out following Pinheiro & al. (2009c). All of the sequences

used in this study were submitted to GenBank (Appendix). A

total of 20 terminal operational taxonomic units (OTUs) were

selected for this study, and fie of them were represented by

more than one accession each (Epidendrum cinnabarinum,

E. fulgens,E. puniceoluteum,E. denticulatum,E. f lammeus).

Each sequence accession of the trnL-trnF, rpl32-trnLand

trnL-trnTregions was aligned with Clustal W option in the

BioEdit Sequence Alignment Editor (Hall, 1999). The resulting

automated alignment was manually edited in BioEdit .5.0.6

and then was exported as a Nexus file for maximum pa

mony (MP) and maximum likelihood (ML) analyses. The

analysis was performed following Pinheiro & al. (2009a), us

the criterion of Fitch (1971), excluding uninformatie char

ters, and with ACCTRAN optimiation. Ten thousand ad

tion sequence replicates were performed by stepwise addit

and holding 10 trees per replicate, and TBR branch swapp

on best trees. The MP analysis was performed with PAU

.4.0b10 (Swofford, 2002). The ML analysis was conduc

using RAxML .7.0.4 (Stamatakis, 2006) and RAxML-G

.1.1 (Silestro & Michalak, 2011). The GTR+ substitut

model, which allows rate ariation among sites, was de

mined using jModeltest .0.1.1 (Posada, 2008) under the Aka

Table .Epidendrumspecies and populations analyed in this study. Sample sies for the morphometric analysis, nuclear microsat-ellite markers (SSR) assays for assignment tests and plastid intergenic spacer sequences (CP) are indicated.

Species Population

Sample Sie

Morphometry SSR CP

Epi

dendrums

ubg.

Amphyglottium

Atlanticclade

E. cinnabarinumSalm. ex Lindl. Pirambu 25 1

Lagoa do Abaet 13 1

E. denticulatum Barb. Rodr. Oliena 9 20

Alcobaa 11

Poos de Caldas 5

Marambaia 10 20 1

Botucatu 13

So Paulo 22

Itapea 9 1

E. flammeusE. Pessoa & M. Ales Pedra do Cachorro 51 53 4

E. fulgens Brongn. Parati 22 1

Bertioga 26 20

Canania 30

Itaja 26

Florianpolis 33

Imbituba 32 20 1

Porto Alegre 17

E. puniceoluteum Pinheiro & Barros Imbituba 13 20 1

Canania 26

Pontal do Sul 27 20 1

E. calanthumRchb. f. & Wars. Serra Pacaraima 1

E. macrocarpum Rich. Recife 1

E. myrmecophorum Barb. Rodr. Araruama 1

E. radicansPa. ex Lindl. Oaxaca 1

E. secundum Jacq. Serra do Rio do Rastro 1

E. xanthinumLindl. Santa Brbara 1

group

E. campestreLindl. Santana do Riacho 1


5/19


information criterion (AIC). To find the optimal likelihood tree,

we ran 100 independent tree searches on the combined matrix.

The support for indiidual branches was ealuated using non-

parametric bootstrapping (Felsenstein, 1985) with 1000 thor-

ough bootstrap replicates. The categories of bootstrap support

(BS) considered were unsupported ( 0.01 for all comparisons). The ordination diagram of

DA indicates a clear separation among the species, consider

the first three canonical axes (Fig. 2, Wilks Lambda = 0.00

P< 0.0001) which represent 97.5% of the total sample ariat

The fie most discriminant ariables in the model are len

of the callus of the lip, column length, length and width of

central lobe of the lip, and lip length. The jackknifed clafication matrix produced by DA shows an aerage of 98%

correct classification of indiiduals into the preiously assig

species, ranging from 100% correct classification inE. fulg

andE. cinnabarinumto 94% inE. f lammeus(Table S2 in

Electronic Supplement). The pattern of ariation of the ab

characters is represented by box-plots (Fig. S2 in the Electro

Supplement). High intraspecific ariability is obsered for

most all of the characters, which oerlap among most spec

OnlyE. cinnabarinumshows clear discontinuities, mainl

the length of the dorsal and lateral sepals, the petal and

callus of the lip (Fig. S2).

The trnL-trnFgene sequences were 1040 bp long, rp

trnL832 bp, and trnL-trnT683 bp, excluding flanking regio

insertions and ambiguously aligned fragments. The combi

molecular matrix consisted of 2557 aligned bp from the

terminal OTUs with nucleotide frequencies of: A = 0.3855,

0.1258, G = 0.1462, and T = 0.3425, with a transition/trans

sion ratio = 2 (kappa = 4.9217825).

The topologies of MP and ML trees recoered in this st

were identical, and for this reason only results from the

analysis are shown (Fig. 3). The main difference obsered

comparison to the study of Pinheiro & al. (2009a), was

position ofE. radicansnested within the Andean clade (FigSubgenusAmphyglottiumis supported as a monophyletic gr

(BS 100; Fig. 3). Strong support (BS 85) was obsered

both the Atlantic clade and subsect. Tuberculata. In the And

cladeE. radicansis sister to the other species,E. calanthum

E. macrocarpum. Within the Atlantic clade,E. cinnabarinu

sister to the other species with BS = 98. BothE. denticula

and E. flammeusare monophyletic, with bootstrap alue

90% and 92%, respectiely. In contrast to the results obtai

by Pinheiro & al. (2009a) using AFLP,E. fulgensandE. p

ceoluteumwere not monophyletic (Fig. 3).

Bayesian assignment results obtained by STRUCTU

and STRUCTURAMAcorrectly assigned all indiidual

preiously recognied species. The statisticL(K) proposed


6/19


Canonical Axis 1 (59.0%)

lacinonaC

)%1.8(3si

xA

lacinonaC

)%4.03(2si

xA

A

B

Fig. .Discriminant scatter

plots showing morphological

similarities amongE. cinnabar-

inum(),E. denticulatum(),

E. f lammeus(),E. fulgens()

andE. puniceoluteum().

A,first and second axes repre-sent 89.4% of the ariation of

the dataset; B,first and third

axes represent 67.1% of the total

ariation among samples.


7/19


probabilities calculated by STRUCTURAMA also showed

K=4 as the most probable number of genetic clusters (Table 2).

Assignment probabilities were higher than 0.9 for almost all

indiiduals, indicating a clear clustering and no indication of

admixture for almost all specimens (Fig. 4). Only indiiduals

ofE. puniceoluteumshowed signs of admixture withE. fulgens.

The eleen nuclear microsatellite loci analyed reealed

a total of 109 alleles among species of the Atlantic clade (ex-

cluding E. cinnabarinum). Of those, PAA identified 49 pri-

ate alleles distributed amongE. denticulatum,E. f lammeus,

E. fulgensand E. puniceoluteum. The number of diagnostic

alleles ranged from 16 inE. denticulatumandE. fulgensto 7 in

E. f lammeus(Table 3). Diagnostic characters were also foundfor both quantitatie and qualitatie morphological traits. Four

exclusie morphological characters were found inE. cinnabari-

num, two inE. denticulatumand one inE. fulgens. Diagnostic

morphological characters were not found inE. f lammeusand

E. puniceoluteum(Table 3).

DISCUSSION

When combining different data types to test hypotheses

about lineage diersification, authors go beyond the naming of

species by reconstructing the eolutionary processes inoled

in their origin (Schlick-Steiner & al., 2010). The characteria-

E. secundum

E. xanthinum

E. cinnabarinum

E. fulgens

E. puniceoluteum

E. denticulatum

E. calanthum

E. radicans

E. myrmecophorum

E. campestre

E. nocturnum

E. macrocarpum

E. cinnabarinum

E. fulgens

E. puniceoluteum

E. denticulatum

E. flammeus

E. flammeus

E. flammeus

E. flammeus

outgroup

100

100

64

92

100

98

100

90

0.0030 substitution/site

I II

III

IV

100

Fig. .Maximum-likelihood

phylogram ofEpidendrum

subg.Amphyglottium, includ

E. campestre andE. nocturn

as outgroups, based on com-

bined trnT-trnL, trnL-trnFa

rpl32-trnL data. The phylogshown is based on the best-

scoring maximum-likelihoo

tree. Clades recognied in th

study are indicated as follow

(I)E. subg.Amphyglottium,

(II)E. subsect. Tuberculata,

(III) Andean clade,(Iv) Atl

tic clade. Maximum likeliho

bootstrap support alues abo

50% are indicated on branch

Table .Estimates of cluster number (K) from STRUCTURE andSTRUCTURAMA analyses using nine microsatellite loci for threedatasets, includingE. flammeusandE. denticulatum,E. flammeus

E. fulgens, andE. flammeusandE. puniceoluteum. STRUCTURE hoc statisticsL(K) and Kand STRUCTURAMA posterior probabdistributionsE(K) are gien.

No. of

popula-tions (K)

STRUCTURE

ad hoc statistics

STRUCTURAposterior probab

distributions

L(K) K E(K)

1 4620.73 0.00

2 3781.25 8.40 0.00

3 3423.44 0.77 0.01

4 3103.32 32.72 0.43

5 3041.53 8.24 0.37

6 3067.23 0.41 0.14

7 3068.70 0.25 0.03

8 3091.25 1.06 0.01

9 3230.99 0.74 0.00

10 3125.18

TheL(K) alue in bold points at the beginning of similar estimatesobtained by log Pr(X|K) and indicates the correct number of group


8/19


cryptic species hae been described, and the number of species

has decreased through the demonstration of conspecificity of

nominal taxa (Schlick-Steiner & al., 2010). Through its inter-

disciplinary approach, in which phylogeography, systemat-

ics, population genetics, ecology and morphology proide the

necessary framework for comparatie analyses of biodiersity,

integratie taxonomy should hae a deep impact across all

leels of biological organiation (Dayrat, 2005; Hendry & al.,

2010). Gien that the Neotropics harbour an impressie number

of species, and a still unknown number of species remain to

be described, the integration of seeral disciplines should be

especially helpful in this region (Dini-Filho & al., 2008; Pires

& Marinoni, 2010; Antonelli & Sanmartn, 2011).

Integratie taxonomy has included new concepts and

methods in order to delimit species, but a lack of consensus

about how data from different sources should be integrated is

still eident (Padial & al., 2010). Congruence among differ-

ent sources of data strengthens species delimitation in mostcases (Saolainen & al., 2006; Moccia & al., 2007; Reees

& Richards, 2011). This approach, named integration by con-

gruence (Padial & al., 2010), tends to promote taxonomic stabil-

ity, since most taxonomists will share similar opinions on the

alidity of a species supported by seeral datasets. Howeer,

recent radiations may often be oerlooked under this po

of iew (Padial & al., 2010; Barrett & Freudenstein, 20

On the other hand, the heterogeneous nature of eolution

forces often precludes full character congruence in specie

complete reproductie isolation, and such conflicts could in

cate ongoing speciation (Borba & al., 2002; Pillon & al., 20

Palma-Sila & al., 2011; Duminil & Di Michele, 2009). Beca

different species concepts often refer to different stages dur

speciation, full congruence among criteria will be achie

only in a late stage of differentiation (de Queiro, 1998)

such situations, species recognition could be based on a sin

dataset, following the framework of integration by cumulat

(Padial & al., 2010). A major adantage of this approach is

species which originated from recent radiations could be ea

delimited (Padial & al., 2010). Howeer, an oerestimation

species number is expected when only a single line of eide

is considered (Padial & al., 2010).

The contrasting approaches of integration by congrueand integration by cumulation need to be coordinated tak

into account the eolutionary forces associated leading

speciation in the taxonomic group of interest (Hendry &

2010; Padial & al., 2010). Since the concordance among d

types is not strictly necessary for justifying the existence

0%

10%

20%

30%

40%

50%

60%70%

80%

90%

100%

E. flammeus E. fulgens E. denticulatum E. puniceoluteum

Bertioga Imbituba OlivenaImbituba Marambaia Pontal do Sul Imbituba

Fig. .Maximum posterior assignment probabilities for 173 specimens ofE. f lammeus(Pedra do Cachorro population),E. fulgens(Bertioga

Imbituba populations),E. denticulatum(Oliena and Marambaia populations) andE. puniceoluteum(Imbituba and Pontal do Sul populatio

analyed with STRUCTURE withK= 4. Each ertical bar represents an indiidual. The proportion of color in each bar represents an indiiduassignment probability to different species. vertical dotted lines delimit different populations within species. See Table 1 for population det

Table .Number of fixed (priate) alleles and presence (1) and absence (0) of diagnostic morphological characters amongE. cinnabarinum,E. denticulatum,E. flammeus,E. fulgensandE. puniceoluteumdetected by PAA.

Character E. cinnabarinum E. denticulatum E. flammeus E. fulgens E. puniceoluteu

Fixed microsatellite alleles 16 7 16 10

Lip with brown dots 0 0 0 1 0

Flowers pink 0 1 0 0 0

Lip callus white 0 1 0 0 0

Dorsal sepal >18 mm long 1 0 0 0 0

Mid-lobe of lip entire 1 0 0 0 0


9/19


distinct species (de Queiro, 1998; Padial & al., 2010; Barrett

& Freudenstein, 2011), we adopted the cumulatie integration

of multiple lines of eidence as a criterion to combine the dif-

ferent data sources inestigated in this study. Furthermore, spe-

cies delimitation will be discussed in the light of eolutionary

trends (Padial & al., 2010), based mainly on specif ic ecological

and genetic characteristics obsered in subg.Amphyglottium.

The phylogeny recoered in this study shows a topology

similar to the tree published by Pinheiro & al. (2009a). Within

the Atlantic clade, support for the monophyly criterion (de

Queiro & Donoghue, 1988) is obsered only for E. cinna-

barinum,E. denticulatumandE. flammeus(Table 4).Epiden-

drum fulgensandE. puniceoluteumare not monophyletic in this

clade. Thus, although monophyly is a proper criterion for spe-

cies delimitation under the general lineage concept, monophy-

letic groups might not be detected in the face of gene flow and

incomplete lineage sorting, factors that are common at lower

taxonomic leels (Knowles & Carstens, 2007).Epidendrum ful-gensoccurs in sympatry with three other species (E. denticula-

tum,E. puniceoluteum,E. secundum), and hybridiation eents

are known to occur among them (Pinheiro & al., in prep.). In

fact, hybridiation and introgression between E. fulgensand

E. puniceoluteumin six sympatric populations were detected

by nuclear and plastid microsatellites (Pinheiro & al., 2010).

Indeed, reproductie isolation mechanisms are not complete

among the species of the Atlantic clade (Pansarin & Amaral,

2008; Pinheiro & al., 2010; Pinheiro, unpub.), and support for

the reproductie isolation criterion (Mayr, 1942) was obsered

only forE. cinnabarinum. Moreoer, shared haplotypes were

detected by Pinheiro & al. (2010) in some indiiduals, suggest-

ing late generation backcrossing, which can mislead results

from phylogenetic trees. Howeer, the monophyly obsered for

E. cinnabarinum,E. denticulatum, andE. f lammeussuggests

the absence of haplotype sharing caused by late-generation

introgression, indicating the existence of strong postygotic

barriers, as F1 hybrids can be easily produced in crossing ex-

periments (Pinheiro, unpub.). Further studies inestigating the

genetic structure of other hybrid ones in the range of species of

the Atlantic clade will help to clarify the strength of reproduc-

tie barriers and species cohesion within this group.

Although plastid markers did not support a pattern ofmonophyly for all species, nuclear microsatellite markers cor-

rectly assigned all indiiduals to the known species (Fig. S3;

Fig. 4), and genetic intermediates were rarely found (E. fulgens

E. puniceoluteum,Fig. 4), strongly supporting the genetic

intermediates criteria (Table 4) (Mallet, 1995). The different

leels of resolution recoered by plastid and nuclear markers

could result from differential rates of gene flow between the

two genomes. According to Petit & Excoffier (2009), genomic

compartments with high leels of intraspecific gene flow are

predicted to experience less introgression (interspecif ic gene

flow), because genetic drift reduces the probability of an in-

trogressed allele to increase in frequency by chance. In plants

pollen moement is often the predominant form of gene f low

Recent empirical data show that seeds hae poorer disp

ibility than pollen, which could explain the higher leel

introgression obsered for maternally inherited plastid D

(Du & al., 2009; Palma-Sila & al., 2011). In fact, Pinheiro &

(2010, 2011) found that gene flow ia pollen inEpidendrum

more than ten-fold higher than that ia seeds. Consequeneen in the presence of hybridiation and introgression (P

heiro & al., 2010), nuclear markers are less introgressed beca

they experience high leels of gene flow. These results stron

agree with patterns that hae been obsered in many ot

angiosperms (reiewed in Petit & al., 2005; Duminil &

2007), and suggest that nuclear markers coupled with mult

cus assignment methods (e.g., Duminil & al., 2006; Hausd

& Hennig, 2010) represent the best option for species delim

tion. Howeer, additional analyses including additional pop

tions may be required to demonstrate that species show d

genetic structuring among populations, which might resu

an oerestimation of the number of true species (Pritch

& al., 2010). One important implication of this finding is

phylogenetic studies in Epidendrumspp. should make us

single- or low-copy nuclear loci instead of plastid mark

which might also be true for other plant groups.

Few morphological diagnostic characters supporting b

the phenetic (Sokal & Croello, 1970) and diagnosability (C

craft, 1983) criteria were found among species of the Atla

clade (Table 3). Four of seen morphological characters w

restricted to E. cinnabarinum(Table 3; Fig. S2), which h

larger flowers and are isited by hummingbirds (Pinhe

unpub.), in contrast to the other species, which hae smaflowers and are mainly pollinated by butterflies (Alme

& Figueiredo, 2003; Pansarin & Amaral, 2008; Fuhro &

2010).Epidendrum denticulatumshowed two diagnostic m

phological characters,E. fulgensone, and no exclusie flo

trait could be found for eitherE. flammeusorE. puniceolute

The morphological delimitation ofE. puniceoluteum(Pinh

& Barros, 2006) andE. flammeus(this study) was possible o

when considering the combination of multiple morphome

characters (Fig. 2; Table S1). Usually,Epidendrumspecie

subsect.Amphyglottiumlack strong premating barriers beca

many species share an extensie number of pollinator spec

(Byerychudek, 1981; Almeida & Figueiredo, 2003; Pans

& Amaral, 2008; Hgsater & Soto-Arenas, 2005). The oe

Table .Conformance ofEpidendrumspecies of the Atlantic cladewith four criteria for species delimitation.

Species Phenetic

Mono-

phyly

Diagnos-

ability

Genoty

clust

E. cinnabarinum Yes Yes Yes Yes

E. denticulatum Yes Yes Yes YesE. flammeus Yes Yes Yes Yes

E. fulgens Yes No Yes Yes

E. puniceoluteum Yes No Yes Yes


10/19


the Atlantic clade. In general, plants with generalist pollination

systems might be less likely to experience strong directional

selection on floral traits (Waser, 1998), and diagnostic flower

characters could be less eident in such species (Pellegrino

& al., 2005).

The relaxed selection related to pollinator behaior could

be responsible for high leels of intraspecific morphological

ariation (Juillet & Scopece, 2010). In food-deceptie or-

chids, outcrossing preails because pollinators aoid plants

in the same patch, promoting gene flow by pollen oer long

distances and reducing the chances of geitonogamous pollina-

tion (Coolino & Widmer, 2005). This behaior is likely to

result in low stabiliing selection for floral traits in deceptie

species, and thus floral polymorphisms arising from muta-

tions can accumulate and increase intraspecific ariation in

natural populations (Juillet & Scopece, 2010). According to

Pinheiro & al. (2010), habitat selection contributes to species

cohesion as a potential post-mating, postygotic barrier thatlimits gene f low. Future studies should include long-term field

experiments, including reciprocal transplants and reproductie

manipulations, to proide direct tests of this hypothesis.

According to the results obtained in this study, species of

the Atlantic clade satisfy most of the species criteria examined

(Table 4), and thus merit recognition as distinct species. All

of the species satisfy at least three of the four species criteria

inestigated: (1) phenetic differentiation based on morphologi-

cal discontinuities obsered on morphometric results (Sokal

& Croello, 1970); (2) monophyly (de Queiro & Donoghue,

1988); (3) diagnosability (Cracraft, 1983), which was found

in both molecular and morphological traits; and (4) geno -

typic clusters recognied by a deficit of genetic intermediates

(Mallet, 1995), as shown by nuclear microsatellite data. The

multidisciplinary approach adopted here to inestigate spe-

cies limits in this orchid group allowed for the delimitation of

closely related species of the Atlantic clade and the recognition

of a new species,E. f lammeus. The new species is described

and illustrated below. An identification key for species from

the Atlantic clade is also proided. Species of subg.Amphy-

glottiumwere used as models to understand diersification

patterns inEpidendrum. The huge number of species in this

genus is an intrinsic impediment for multidisciplinary stud-ies, but different studies focusing on specific and smallerEpi-

dendrumgroups should be a suitable approach to understand

evolutionary patterns within this genus. Moreover, future

studies should consider the use of multiple lines of eidence to

inestigate species limits in different plant groups, in order to

depict eolutionary trends and speciation processes according

to an integratie taxonomic approach.

DESCRIPTION OF THE NEW SPECIES

Epidendrum flammeusE. Pessoa & M. Ales, sp. nov. Type:

Brail. Pernambuco, Municpio de So Caetano, RPPN

Epidendrum flammeus is morphologically related

E. denticulatum, butdiffersby haing yellow, orange or

flowers, these with a cured lip.

Description. Rupicolous, caespitose herb. Stems 10

73.0 cm long, 0.51.0 cm wide, simple, cane-like, terete. Lea

3.09.0 cm long, 0.83.0 cm wide, distichous, distribu

throughout the stems, coriaceous, lanceolate to elliptic, a

rounded, slightly emarginate, margin entire. Inflorescence

minal, racemose; peduncule 14.077.0 cm long, coered by ac

amplexcaulous sheaths; rachis of inflorescence 3.04.0 cm lo

Floral bracts 0.30.9 cm long, 0.10.2 cm wide, much sho

than oary, triangular, lanceolate, apex acute. Flowers 5

simultaneous, nonresupinate, yellow, orange or red. O

pedicellate, 1.22.9 cm long, 0.200.25 cm wide; dorsal se

0.71.2 cm long, 0.320.55 cm wide, oblong-oboate, apex ac

margin entire; lateral sepals 0.71.4 cm long, 0.390.60 cm w

oboate, falcate, apex acute, margin entire; petals 0.81.3

long, 0.300.59 cm wide, oboate to elliptic, apex acute, marentire thoughout or apical half erose. Lip 0.300.67 cm lo

0.91.6 cm wide, united with column, 3-lobed, base of the d

with a pair of ooid callus, and a longitudinal keel 0.250.48

long, 0.140.30 cm wide; lateral lobes 0.350.70 cm long, 0.

0.75 cm wide, semiorbicular, margin serrate to fimbriate, m

lobe 0.220.38 cm long, 0.300.68 cm wide, flabellate, a

bilobed, margin serrate to fimbriate. Column 0.51.1 cm lo

0.300.35 cm wide, apex of entral face with a pair of proj

tions on both sides, directed to lip base. Anther apical, o

Pollinia 4, ooid, laterally compressed. Cuniculus 0.70.9

long, penetrating about half the oary. Capsule 5.05.2 cm lo

1.31.9 cm wide, globose to ooid. Figure 5.

Paratypes. BRAzIL. Alagoas, Quebrangulo, REBIO

dra Talhada, 091517 S 362536 W, 25 Out 2011,B.Amo

1149(UFP); Paraba, So Joo do Tigre, APA das Onas, 6

2005,Dantas & al. 1442(JPB); Pernambuco, Bonito, Rese

municipal, 20 Jul 1995, M. Alves & al. 34545 (UFP!); i

Brejo da Madre de Deus, Faenda Bituri, 19 Apr 1959,D.

drade-Lima 59-3354 (IPA); ibid. So Caetano, RPPN Pedra

Cachorro, 081406 S 361125.5 W, 14 Jul 2007, P. Gom

684 (UFP!, MO!); ibid. So Caetano, RPPN Pedra do Cacho

0814 21.8 S 361120.3 W, 29 Aug 2010,D. Cavalcanti &

289(UFP!, PEUFR!).Etymology. The new species is named on behalf of

color of its flowers, which are mainly red, orange and yell

similar to a flame.

Distribution and ecologyEpidendrum flammeusis kno

from the states of Pernambuco, Paraba and Alagoas in the no

ern part of northeastern Brail (Fig. 1). The species occur

granitic rock outcrops (up to 1000 m high) on the Borbore

plateau, within the Caatinga biome. It grows in association w

other rupicolous plants such as bromeliads and orchid in isla

of egetation. Flowers hae been obsered during the entire y

with a clear flowering peak from December to March.

Many inentories from northeastern Brail (Pereira, 19

Gomes-Ferreira, 1990; Felix & Caralho, 2002) list specim


11/19



12/19


Almeida, A.M. & Figueiredo, R.A. 2003. Ants isit nectaries of

dendrum denticulatum (Orchidaceae) in a Brailian rainforEffects on herbiory and pollination.Brazil. J. Biol. 63: 551

Antonelli, A. & Sanmartn, I. 2011. Why are there so many plant cies in the Neotropics? Taxon60: 403414.

Barbar, T., Palma-Silva, C., Paggi, G.M., Bered, F., Fay, M.FLexer, C. 2007. Cross-species transfer of nuclear microsate

markers: Potential and limitations.Molec. Ecol. 16: 3759376Barrett, C.F. & Freudenstein, J.V. 2011. An integratie approac

delimiting species in a rare but widespread mycoheterotrop

orchid.Molec. Ecol. 20: 27712786.Barros, F. de, Vinhos, F., Rodrigues, V.T., Fraga, C.N. & Pes

E.M. 2011. Orchidaceae. In: Lista de Espcies da Flora do BrJardim Botnico do Rio de Janeiro. http://floradobrasil.jbrj.

.br/2011/FB000179, accessed 20 Apr. 2012).Bierzychudek, P. 1981.Asclepias, Lantana andEpidendrum: A fl

mimicry complex?Biotropica 13: 5458.

Borba, E.L., Shepherd, G.J., Van den Berg, C. & Semir, J. 2Floral and egetatie morphometrics of fiePleurothallis (Ordaceae) species: Correlations with taxonomy, phylogeny, genariability, and pollination systems.Ann. Bot. 90: 219230

Buzzato, C.R., Davies, K.L., Singer, R.B., Santos, R.P. & V

den Berg, C. 2012. A comparatie surey of floral characin Capanemia (Orchidaceae: Oncidiinae).Ann. Bot.109: 1

144.Cozzolino, S. & Widmer, A. 2005. Orchid diersity: An eolution

consequence of deception? Trends Ecol. Evol. 20: 487494.

Cracraft, J. 1983. Species concept and speciation analysis. C

Ornithol. 1: 159187.Davis, J.I. & Nixon, K.C. 1992. Populations, genetic ariation,

the delimitation of phylogenetic species. Syst. Biol. 41: 4214Dayrat, B. 2005.Towards integratie taxonomy.Biol. J. Linn. S

85:407415.De Queiroz, K. 1998. The general lineage concept of species, spe

criteria, and the process of speciation. Pp. 5775 in: Howard,

& Berlocher, S.H. (eds.),Endless forms: Species and speciatOxford: Oxford Uniersity Press.

De Queiroz, K. & Donoghue, M. 1988. Phylogenetic systematics

the species problem. Cladistics4: 317338.Diniz-Filho, J.A.F., Telles, M.P.C., Bonatto, S.L., Eizirik, E., Frei

T.R.O., Marco, P., Santos, F.R., Sole-Cava, A. & Soares, T2008. Mapping the eolutionary twilight one: Molecular mark

populations and geography.J. Biogeogr.35: 753763.Du, F., Petit, R. & Liu, J. 2009. More introgression with less g

flow: Chloroplast ersus mitochondrial DNA in the Picea aspe

Epidendrum fulgensandE. puniceoluteumgrow in south-

ern and southeastern Brail, in coastal sand areas locally called

restingas (Barros & al., 2011).Epidendrum cinnabarinumis

known from restingas and the granitic rock outcrops found in

the Caatinga ecosystem in northeastern Brail, andE. denticu-

latumoccurs in southern, southeastern, and southern parts of

northeastern Brail, also growing on sandy soils along the coast

and in saanna egetation (Cerrado) (Barros & al., 2011, Fig. 1).

Morphological aff inities. The new species described

here has preiously been wrongly identified as E. fulgens,

because of similarities in flower color which ranges from red

to orange or yellow in the latter species (Fig. S1). The mor-

phology of the sepals, petals and lip are also similar in these

two species, as well as in all species ofEpidendrumplaced in

the Atlantic clade.

Epidendrum fulgensandE. cinnabarinumshare a similar

column morphology, but both species lack the pair of lateral

projections on the entral face of the apex directed to the lipsbase seen in other species of the Atlantic clade (E. denticula-

tum,E. flammeus,E. punniceoluteum). AlthoughE. flammeus

andE. cinnabarinum are sympatric (Fig. 1), their flowers differ

in sie and morphology. InE. puniceoluteum, the flowers are

red-purple and larger than inE. denticulatumandE. flammeus,

which hae flowers of similar sie.Epidendrum denticulatum

has pink flowers, whereas in E. flammeus they are yellow,

orange or red. In addition, E. flammeusshows a cured lip,

which is erect inE. denticulatum.

Two morphotypes can be recognied in the populations

ofE. flammeus(Fig. 5). In one of them, plants hae yellow

and slightly larger flowers, and the lip is more incured. The

second morphotype has red and smaller flowers and the lip is

only slightly incured. Specimens with intermediate charac-

teristics such as orange flowers, and with ariation in flower

sie and lip curature are common within populations. Clear

discontinuities between morphotypes were not obsered in

the morphometric analysis (Fig. 2). Fine-scale genetic studies

within populations could clarify the role of microhabitat pref-

erences between morphotypes, and a pollination surey could

clarify the role of color selection and the maintenance of the

flower ariability in this species.

Key to Epidendrum flammeusand related species

of the Atlantic clade of subg.Amphyglottium

1. Dorsal sepal >4 cm long; lateral lobes of the lip fimbriate

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. cinnabarinum

1. Dorsal sepal


13/19


14/19


& Barros, F. 2009a. Phylogenetic relationships and infragenericclassification ofEpidendrumsubgenusAmphiglottium(Laeliinae,Orchidaceae).Pl. Syst. Evol. 283: 165177.

Pinheiro, F., Palma-Silva, C., Barros, F. & Cozzolino, S. 2009b.

Cross-amplification and character iation of microsatellite loci forthe Neotropical orchid genus Epidendrum. Genet. Molec. Biol.32: 337339.

Pinheiro, F., Palma-Silva, C., Barros, F., Flix, L.P., Lexer, C., Coz-zolino, S. & Fay, M.F. 2009c. Chloroplast microsatellite markersfor the Neotropical orchid genus Epidendrum, and cross-ampli-fication in other Laeliinae species (Orchidaceae). Conservation

Genet. Resources1: 505511.Pinheiro, F., Santos, M.O., Barros, F., Meyer, D., Salatino, A.,

Souza, A.P. & Cozzolino, S. 2008a. Isolation and characteriation

of microsatellite loci in the Brailian orchidEpidendrum fulgens.Conservation Genet. 9: 16611663.

Pinheiro, F., Santos, M.O., Palma-Silva, C., Barros, F., Meyer, D.,Salatino, A., Souza, A.P. & Cozzolino, S. 2008b. Isolation and

characteriation of microsatellite loci inEpidendrum puniceolu-teum, an endemic orchid from the Atlantic Rainforest.Molec. Ecol.

Resources8: 11141116.Pires, A.C. & Marinoni, L. 2010. DNA barcoding and traditional

taxonomy unified through Integratie Taxonomy: A iew that chal-

lenges the debate questioning both methodologies.Biota Neotrop.10: 339346.

Posada, D. 2008. jModelTest: Phylogenetic model aeraging. Molec.

Biol. Evol. 25: 12531256.Pritchard, J.K., Wen, X. & Falush, D. 2010. Documentation for

STRUCTUREsoftware, ersion 2.3. http://pritch.bsd.uchicago

.edu/structure.html.Reeves, P.A. & Richards, C.M. 2011. Species delimitat ion under the

general lineage concept: An empirical example using wild North

American hops (Cannabaceae:Humulus lupulus). Syst. Biol.60:4559.

Savolainen, V., Anstett, M.C., Lexer, C., Hutton, I., Clarkson, J

Norup, M.V., Powell, M.P., Springate, D., Salamin, N. & BaW.J. 2006. Sympatric speciation in palms on an oceanic isl

Nature441:210213.

Schlick-Steiner, B.C., Steiner, F.M., Seifert, B., Stauffer, C., Chtian, E. & Crozier, R.H. 2010. Integratie taxonomy: A msource approach to exploring biodiersity.Annual Rev. Entom

55: 421438.Shaw, J., Lickey, E.B., Beck, J.T., Farmer, S.B., Liu, W., Mil

J., Siripun, K.C., Winder, C.T., Schilling, E.E. & Small, R2005. The tortoise and the hare II: Relatie utility of 21 non

ing chloroplast DNA sequences for phylogenetic analysis.AmJ. Bot.92: 142166.

Shaw, J., Lickey, E.B., Schil ling, E.E. & Small, R.L. 2007. Compson of whole chloroplast genome sequences to choose noncod

regions for phylogenetic studies in angiosperms: The tor toisethe hare III.Amer. J. Bot.94: 275288.

Silvestro, D. & Michalak, I. 2011. raxmlGUI: A graphical front-end

RAxML. Organisms Diversity Evol. online, doi:10.1007/s13011-0056-0.

Sites, J.W. & Marshall, J.C. 2004. Operational criteria for delimispecies.Annual Rev. Ecol. Evol. Syst.35: 199227.

Sokal, R.R. & Crovello, T.J. 1970. The biological species concA critical ealuation.Amer. Naturalist104:127153.

Stamatakis, A. 2006. RAxML-vI-HPC: Maximum likelihood-b

phylogenetic analyses with thousands of taxa and mixed modBioinformatics22: 26882690.

Swofford, D.L. 2002. PAUP*: Phylogenetic analysis using parsim(*and other methods), ersion 4. Sunderland, Massachus

Sinauer.Waser, N.M. 1998. Pollination, angiosperm speciation, and the na

of species boundaries. Oikos81:198201.Wu, C.I. 2001. The genic iew of the process of speciation. J. E

Biol. 14: 851865.

Appendix.Specimens used in morphological and molecular analyses with oucher information, countr y, locality, and GenBank accession numbers forrpl32-trnL, trnT-trnLand trnL-trnF, respectiely. Newly generated sequences are indicated by an asterisk.

Epidendrum cinnabarinumSalm. ex Lindl., Brail, Sergipe, Pirambu,F. Pinheiro 609(SP), JQ645971*, JQ645991*, JQ646011*;Epidendrum cinnabarinBrail, Bahia, Lagoa do Abaet,F. Pinheiro 608(SP), JQ645970*, JQ645990*, JQ646010*;Epidendrum dent iculatumBarb. Rodr., Brail, Bahia, OlieF. Pinheiro 626(SP);Epidendrum denticulatum, Brail, Bahia, Alcobaa, F. Pinheiro 627(SP);Epidendrum denticulatum, Brail, Minas Gerais, Pde Caldas,F. Pinheiro 628(SP);Epidendrum denticulatum, Brail, Rio de Janeiro, Marambaia,F. Pinheiro 610(SP), JQ645972*, JQ645992*, JQ6460Epidendrum denticulatum, Brail, So Paulo, Botucatu, F. Pinheiro 629(SP); Epidendrum denticulatum, Brail, So Paulo, So Paulo, F. Pinheiro(SP);Epidendrum denticulatum, Brail, So Paulo, Itapea, F. Pinheiro 611(SP), JQ645973*, JQ645993*, JQ646013*;Epidendrum f lammeusE. PessoM. Ales, Brail, Pernambuco, Pedra do Cachorro,F. Pinheiro 612, JQ645974*, JQ645994*, JQ646014*;Epidendrum flammeus, Brail, Pernambuco, PedrCachorro,F. Pinheiro 613, JQ645975*, JQ645995*, JQ646015*;Epidendrum flammeus, Brail, Pernambuco, Pedra do Cachorro,F. Pinheiro 614, JQ6459

JQ645996*, JQ646016*;Epidendrum flammeus, Brail, Pernambuco, Pedra do Cachorro,F. Pinheiro 615, JQ645977*, JQ645997*, JQ646017*;EpidendfulgensBrongn., Brail, Rio de Janeiro, Parati,F. Pinheiro 631(SP), JQ645978*, JQ645998*, JQ646018*;Epidendrum fulgens,Brail, So Paulo, BertiF. Pinheiro 616(SP);Epidendrum fulgens,Brail, So Paulo, Canania,F. Pinheiro 632(SP);Epidendrum fulgens,Brail, Santa Catarina , Itaja,F. Pinh633(SP);Epidendrum fulgens,Brail, Santa Catarina, Florianpolis,F. Pinheiro 634(SP);Epidendrum fulgens,Brail, Santa Catarina, Imbituba,F. Pinh617(SP), JQ645979*, JQ645999*, JQ646019*;Epidendrum fulgens,Brail, Rio Grande do Sul, Porto Alegre,F. Pinheiro 635(SP);Epidendrum puniceolutPinheiro & Barros, Brail, So Paulo, Ilha Comprida, F. Pinheiro 622(SP), JQ645984*, JQ646004*, JQ646024*;Epidendrum puniceoluteum,Brail,Paulo, Canania,F. Pinheiro 636(SP);Epidendrum puniceoluteum,Brail, Paran, Pontal do Sul,F. Pinheiro 621(SP), JQ645983*, JQ646003*, JQ6460Epidendrum calanthumRchb.f. & Wars., Brail, Roraima, Serra Pacaraima,F. Pinheiro 606(SP), JQ645968*, JQ645988*, JQ646008*;Epidendrum mrocarpumRich., Brail, Pernambuco, Recife,F. Pinheiro 618(SP), JQ645980*, JQ646000*, JQ646020*;Epidendrum myrmecophorumBarb. Rodr., BrRio de Janeiro,Araruama,F. Pinheiro 619(SP), JQ645981*, JQ646001*, JQ646021*;Epidendrum radicansPa. ex Lindl., Mexico, Oaxaca,F. Pinheiro(SP), JQ645985*, JQ646005*, JQ646025*;Epidendrum secundum Jacq., Brail, Santa Catarina, Serra do Rio do Rastro, F. Pinheiro 624 (SP), JQ6459JQ646006*, JQ646026*; Epidendrum xanthinumLindl., Brail, Minas Gerais, Santa Brbara, F. Pinheiro 625 (SP), JQ645987*, JQ646007*, JQ6460OUTGROUP:Epidendrum campestre Lindl., Brail, Minas Gerais, Santana do Riacho,F. Pinheiro 607(SP), JQ645969*, JQ645989*, JQ646009*;Epidrum nocturnumJacq., Brail, So Paulo, Canania,F. Pinheiro 620(SP), JQ645982*, JQ646002*, JQ646022*.
http://pritch.bsd.uchicago.edu/structure.htmlhttp://pritch.bsd.uchicago.edu/structure.htmlhttp://dx.doi.org/10.1007/s13127-http://dx.doi.org/10.1007/s13127-http://dx.doi.org/10.1007/s13127-http://dx.doi.org/10.1007/s13127-http://pritch.bsd.uchicago.edu/structure.htmlhttp://pritch.bsd.uchicago.edu/structure.html


15/19

1

Electronic Supplement (1 pp.) to: Pessoa & al. Different tools to delimitEpidendrumspeciesTAXON61 (4) August 2012Vol. 61 August 2012

International Journal of Taxonomy, Phylogeny and Evolution

Electronic Supplement to

Integrating different tools to disentangle speciescomplexes: A case study in Epidendrum (Orchidaceae)

Edlley Max Pessoa, Marccus Alves, Anderson Alves-Arajo, Clarisse Palma-Silva

& Fbio Pinheiro

Taxon:


16/19

S1

Electronic Supplement (4 pp.) to: Pessoa & al. Different tools to delimitEpidendrumspeciesTAXON61 (4) August 2012

Table S.Morphological characters used in morphometric analyses, means and standard deviations ofEpidendrum cinnabarinum(Ec),E. denticu-

latum(Ed),E. flammeus(Efl),E. fulgens(Ef) andE. puniceoluteum (Ep) and results of discriminant analysis, including F-to-remove and canonicaldiscriminant values for the first three axes (CA 1, CA 2 and CA 3).

Characters Ec Ed Efl Ef Ep

F-to-

remove CA 1 CA 2 CA 3

Pedicel length 31.56 4.80 21.75 4.24 19.27 3.60 19.56 2.92 25.50 3.34 15.12 0.070 0.099 0.052

Dorsal sepal length 20.15 1.45 10.54 1.08 10.18 1.36 13.39 0.98 14.48 1.25 0.74 0.015 0.107 0.183Dorsal sepal width 5.94 0.91 3.85 0.44 4.26 0.52 5.34 0.50 5.72 0.53 3.06 0.062 0.283 0.443

Lateral sepal length 21.56 2.28 11.11 1.07 10.53 1.43 14.06 1.04 15.13 1.29 2.73 0.004 0.199 0.067

Lateral sepal width 5.72 0.62 4.13 0.52 4.70 0.57 5.80 0.46 6.24 0.65 15.21 0.459 0.897 1.032

Petal length 21.24 2.43 10.95 1.02 10.29 1.29 13.31 0.91 14.82 1.28 7.07 0.194 0.403 0.269

Petal width 5.23 1.02 3.63 0.64 4.27 0.72 5.44 0.81 5.82 0.75 4.19 0.082 0.13 0.325

Lip length 8.29 1.16 6.59 1.00 5.25 0.77 6.05 0.59 6.58 0.65 20.96 0.315 0.208 0.617

Lip width 15.41 1.79 12.14 1.62 12.09 1.58 14.56 1.38 17.09 1.51 18.72 0.226 0.013 0.443

Column length 14.56 1.11 6.82 0.91 7.98 1.36 12.21 0.89 10.38 0.87 58.59 0.493 1.000 0.013

Lateral lobe of lip length 7.06 0.36 6.31 0.88 5.15 0.83 6.25 0.67 7.33 0.67 19.26 0.838 0.423 0.279

Lateral lobe of lip width 7.42 0.65 6.05 0.98 5.77 0.86 8.27 0.98 7.80 1.16 20.09 0.241 0.276 0.359

Central lobe of lip length 5.54 0.50 4.28 0.86 3.11 0.39 3.38 0.49 5.15 0.61 21.6 0.201 0.679 0.629Central lobe of lip width 3.70 0.34 6.46 1.00 4.85 0.92 4.69 0.78 7.20 1.22 23.82 0.482 0.093 0.300

Callus of lip length 7.81 0.35 2.71 0.41 3.51 0.54 4.43 0.51 4.04 0.58 77.87 1.404 0.441 0.453

Callus of lip width 1.58 0.19 2.00 0.40 2.35 0.44 2.48 0.39 3.14 0.56 15.46 0.545 0.596 1.001

Table S.Results of jackknifed classification matrix from discriminant analysis performed on 420 individuals from Epidendrum cinnabarinum(38),E. denticulatum(79),E. f lammeus(51),E. fulgens(186) andE. puniceoluteum(66), indicating the percentage of correct classification of individualsin each species. Wilks Lambda = 0.011,P= 0.0000.

Species E. cinnabarinum E. denticulatum E. f lammeus E. fulgens E. puniceoluteum % correct

E. cinnabarinum 38 0 0 0 0 100

E. denticulatum 0 76 3 0 0 96

E. f lammeus 0 0 48 3 0 94

E. fulgens 0 0 0 186 0 100

E. puniceoluteum 0 1 0 0 65 98

Total 38 77 51 189 65 98


17/19

S2


Fig. S.Morphological and color variation within and amongEpidendrumspecies from the Atlantic clade.


18/19

S3


Fig. S.Box-plots for the 16 quantitative characters measured onEpidendrum fulgens(Ef),E. puniceoluteum(Ep),E. f lammeus(Efl),E. den-

ticulatum(Ed) andE. cinnabarinum(Ec.). Rectangles define 25 and 75 percentiles; horizontal lines show median; whiskers are from 10 to 90

percent ile; asterisks indicate extreme values.

Ef Ep Efl Ed Ec Ef E p E fl Ed Ec Ef Ep Efl Ed Ec

Ef E p Efl Ed Ec Ef Ep Efl Ed Ec Ef Ep Efl Ed Ec Ef E p Efl Ed Ec

Ef Ep Efl Ed Ec Ef Ep Efl Ed Ec Ef Ep Efl Ed Ec Ef E p E fl Ed Ec

Ef Ep Efl Ed EcEf Ep Efl Ed EcEf Ep Efl Ed Ec Ef Ep Efl Ed Ec

Ef Ep Efl Ed Ec10

20

30

40

50

PedicelLength[mm]

0

10

20

30

Dorsa

lSepalLength[mm]

2

3

4

5

6

7

8

Dors

alSepalWidth[mm]

0

10

20

30

Latera

lSepalLength[mm]

2

3

4

5

6

7

8

LateralSepalWidth[mm]

0

10

20

30

PetalLength[mm]

2

3

4

5

6

7

8

9

PetalWidth[mm]

3

4

5

6

7

8

9

10

11

LipLength[mm]

0

5

10

15

20

25

LipWidth[mm]

5

10

15

20

ColumnLength[mm]

2

3

4

5

6

7

8

9

10

LateralLobeofLipLength[mm]

2

3

4

5

6

7

8

9

10

11

LateralLobeofLipWidth[mm]

2

3

4

5

6

7

CentralLobeofLipLength[mm

]

2

3

4

5

6

7

8

9

10

11

CentralLobeofLipWidth[mm]

1

2

3

4

5

6

7

8

9

CallusofLipLength[mm]

1

2

3

4

5

CallusofLipWidt[mm]h


19/19


Fig. S.Graphical methods allowing detection of the true number of groupsK, based on ad hoc statistics provided by STRUCTURE. A, Mean log

probabil ity of dataL(K) (SD) as a function ofK, where the values plateaus atK= 4; B, the magnitude of Kas a function ofK, where the modal

value of this distribution indicates the trueKor the uppermost level of structure, here four clusters.

K

L(K)

K

K

A

B

Documents

Epidendrum Flammeus Complete INTEGRATIVE