Nordic Journal of Botany 25: 129
136, 2007

doi: 10.1111/j.2008.0107-055X.00010.x,
Journal compilation # Nordic Journal of Botany 2007. No claim to US government works
Subject Editor: Stefan Andersson. Accepted 2 October 2007

Morphometric analysis of Epidendrum secundum (Orchidaceae)

in southeastern Brazil

F. Pinheiro and F. de Barros

F. Pinheiro and F. de Barros (fdebarros@terra.com.br), Inst. de Botânica, Caixa Postal 3005, BR
01061-970 São Paulo, Brasil.

One of the largest genera of Orchidaceae with about 1125 species, Epidendrum L. presents several taxonomic problems,
mainly due to the great variability of some of its species. The present study aims at evaluating the influence of different
environments on the morphology of individuals from Epidendrum secundum, using morphometry to compare plants
under cultivation and plants collected directly from the field. Eighty specimens maintained in cultivation at the Institute
of Botany of São Paulo (Brazil) and 146 individuals originating from three natural populations, totaling 226 specimens,
were analyzed with multivariate methods (PCA, CVA, DA). The fact that plants growing in rocky outcrops and plants
from the Atlantic rainforest showed the largest morphological differentiation among themselves, indicates that
phenotypes are strongly influenced by the habitat. Because plants collected from both habitats, maintained under
cultivation for at least 20 years maintained their differentiation, such differentiation may have a genetic component.
Because individuals collected in such environments present the largest morphological dimensions, new habitats generated
by human activity, such as highway margins, seem to be quite favorable for these plants.

With about 1125 species (Chase et al. 2003), Epidendrum and Garay 1961, Dunsterville 1979, Dressler 1989), based
L. is the largest neotropical orchid genus. It presents many only on morphological characters. We follow the traditional
taxonomical problems, resulting mostly from its size, taxonomical concept of species, mainly based on morpho-
combined with the great variability of some of its species logical discontinuities, because no other criteria to delimit
(Dressler 1967, 1984, Brieger 1976
1977, Hágsater 1984, species can be used in this group. Different kinds of
Withner and Harding 2004). biological information, employed by other species delimita-
Epidendrum secundum is one of the most variable species tion criteria, like reproductive biology and population
of the genus. This species belongs to the section Amphyglot- genetics (Borba et al. 2002, Moccia et al. 2007), are not
tidae, formerly recognized by some authors (Lindley 1852
available for this species.
1859, Pabst and Dungs 1975, Brieger 1976
1977, Hágsater Apparently, no morphological discontinuities are present
1984) as a group of highly polymorphic species, and is widely in this species, but the authors that studied the group did
distributed in South America. The complexity of this not employ objective methods or sample strategies that
group is reflected in the large number of names associated focused on identifying these discontinuities among speci-
to it, sometimes treated as synonyms of E. secundum
mens from between and within populations. Studies on the
Jacq., sometimes as distinct species (E. elongatum Jacq.,
influence of habitat type on the morphological variation of
E. crassifolium Lindl., E. ellipticum Grah., E. ansiferum Rchb.
the species are also absent. Epidendrum secundum can be
f., E. versicolor Hoehne & Schltr., E. xanthinum Lindl.,
among others). found in numerous habitat types, like the Andean Cordil-
Epidendrum secundum is essentially characterized by lera, the central highlands from Brazil, the coastal Atlantic
being a caespitose plant, without pseudobulbs, with an rainforest, dry inselbergs from Caatinga vegetation and the
elongated stalk bearing coriaceous and distichous leaves, Venezuelan Tepuis. Even in populations located very close
and by inflorescences in densely flowered simple corymbs, to each other, a few kilometers apart, the morphological
with a variable number of small flowers (ca 2.0 cm in variation both among populations and within populations
diameter), commonly lilac, with a 3-lobed lip with lacerate
is very high (Fig. 1).
dentate margin, and a complex callus in its center. Numerical methods are important tools in studies of
This work follows the assertion by Hágsater (1993) who taxonomy (Reinhammar 1995, Marhold 1996, Palestina
considered Epidendrum secundum the oldest name for this and Sosa 2002, Goldman et al. 2004), mainly to evaluate
group, and the delimitation of Vasquez and Ibisch (2004), morphological variation. Their advantage is the objectivity
that joined all the morphological and color forms into a in the way they deal with the data, especially when many
single polymorphic species, like other authors (Dunsterville variables are involved (Manly 1994).

129
Fig. 1. Lips of flowers from E. secundum plants showing the morphological variation. (A) individuals from Santana do Riacho
(population 3), (B) individuals from Cananéia (population 18), (C) individuals from São Bernardo do Campo (population 13). See Table
1 and Fig. 2 for provenance details.

Within Orchidaceae, multivariate methods have pre- analysis. We also aimed to evaluate the influence of
viously been applied as taxonomic tools on various groups different habitats on plant morphology, by comparing
and in various ways, in order to: (a) delimit taxa (Dufréne plants under cultivation and plants collected directly from
et al. 1991, Tyteca and Dufréne 1994, Reinhammar 1995, the field.
Tyteca 1995, Tyteca and Gerbaud 1998, Carlini-Garcia
et al. 2002), (b) help recognize the hybrid origin of some
taxa (Kallunki 1976, Du Puy et al. 1985, Catling and Material and methods
Catling 1997, Knyasev et al. 2000), (c) evaluate the
morphological variability within populations (Shaw 1998, A total of 226 living plants from 18 southeastern Brazilian
Cardim et al. 2001, Pedersen 2004), or (d) combine all populations were studied (Table 1, Fig. 2). One hundred
these different approaches (Tyteca and Gathoye 1993). and forty six individuals were sampled directly from the
Multivariate methods frequently make it possible to field in populations 11, 13 and 14 (Table 1, Fig. 2). Eighty
recognize important morphological discontinuities and to specimens, collected in the populations listed in Table 1,
enhance the selection of diagnostic characters (Tyteca and have been cultivated for at least 20 years at the orchid
Dufréne 1994, Reinhammar 1995, Palestina and Sosa collection of the Botanical Inst. of São Paulo (Brazil).
2002). Nevertheless, in some instances, the use of morpho- Populations 6
13 and 15
18 grow surrounded by small
metrics reveal gradients without clear morphological dis- trees and shrubs, at high elevations in the Atlantic rainforest
continuities, which can be explained as clinal variation mountains (Fig. 2). Population 13 grows in a disturbed
(Kephart et al. 1999, Fritsch and Lucas 2000, Sapir et al. habitat; a roadside cut. Populations 1
5 and 14 grow on
2002). rocky outcrops in both the central Brazilian plateau and the
This work aimed at analyzing the morphological Atlantic rainforest (Fig. 2).
variability within and among populations of individuals Twenty continuous morphological characters were mea-
of the E. secundum complex by means of multivariate sured (Table 2), of which three characters were vegetative

130
Table 1. The studied populations of E. secundum, together with the were employed to investigate possible distortions produced
number of individuals (n) used in morphometric analysis and by a specific method (Everitt 1978); principal component
vouchers housed at SP. * denotes plants from Atibaia, São Bernardo
do Campo and Salesópolis collected directly from the field in analysis (PCA) was used to summarize variation when a
natural populations. The remaining populations were collected in priori knowledge of a population to which individuals
natural populations. but maintained under cultivation for at least 20 belonged was not considered, and canonical variate analysis
years at the living collection from the Institute of Botany of São (CVA) was applied to ordinate pre-defined populations in a
Paulo. Brazil. MGMinas Gerais; RJ Rio de Janeiro; SPSão
Paulo states. way that accounted for variance and covariance among
characters within and among populations. In the PCA, the
Population n Voucher number of informative axes was determined by comparing
1. Pedra Azul
MG 1 Brólio and Silva s.n. (SP365921) eigenvalues with the random expectation in a broken-stick
2. Diamantina
MG 2 Brólio and Silva s.n. (SP365931) distribution (Frontier 1976). The CVA was based on the
3. Santana do 4 Barros s.n. (SP365928) following groups: the three populations collected directly
Riacho
MG from the field (Atibaia, São Bernardo do Campo and
4. Caeté
MG 1 Bicalho s.n. (SP365896)
5. Mariana
MG 2 Brólio et al. s.n. (SP365904) Salesópolis) and all plants under cultivation (data pooled
6. Miradouro
MG 4 Bicalho s.n. (SP365903) over populations). Discriminant analysis (DA) was per-
7. Itutinga
MG 2 Brólio and Silva s.n. (SP365920) formed with the same grouping variables as were used in the
8. Baependi
MG 3 Bicalho s.n. (SP365937) CVA. Wilks’ lambda and jack-knife classifications, which
9. Parati
RJ 14 Targa s.n. (SP365907)
10. São José dos 6 Ribeiro s.n. (SP365897) assign unclassified specimens to groups, and provide F-to-
Campos
SP remove statistics, which give an indication of the relative
11. Salesópolis
SP 50* Pinheiro and Inês 214 importance of the variables used in the model, were also
12. Bertioga
SP 7 Bicalho s.n. (SP365912) calculated.
13. São Bernardo do 32* Pinheiro and Inês 226
Campo
SP
14. Atibaia
SP 64* Pinheiro and Inês 202
15. Anhembi
SP 5 Bicalho and Kuhlmann s.n. Results
(SP365932)
16. Mira-Estrela
SP 5 Kuhlmann s.n. (SP365895)
17. Apiaı́
SP 11 Silva and Brólio s.n. (SP365902) The results of the descriptive analyses show that the mean
18. Cananéia
SP 13 Barros s.n. (SP365910) values of characters from cultivated plants and from
Total 226 individuals collected in a disturbed habitat at São Bernardo
do Campo are larger than the values observed in the Atibaia
and Salesópolis populations (Table 3). Although the
cultivation conditions of the plants at the orchid collection
and 17 characters were floral (Fig. 3). Measures were taken of the Botanical Institute are homogenous, the standard
at the point of the greatest dimension for each of the deviations observed for these plants are high, revealing a
characters. Vouchers are deposited at the herbarium SP great variability of all characters, as is also the case for the
(Table 1). Salesópolis and São Bernardo do Campo populations
The data were analysed with Systat (Wilkinson 2000) to (Table 3). The smallest values were observed in the Atibaia
provide the mean, standard deviation and 5 and 95 population (Table 3).
percentiles of each character. Multivariate analysis were In the PCA, only the first two axes were considered as
carried out using the programs Systat (Wilkinson 2000) and informative. Individuals under cultivation and those
Fitopac (Shepherd 1994). Two types of ordination analysis collected in São Bernardo do Campo and Salesópolis
were mixed in the analysis, and occupied different regions
of the scatter-plot (Fig. 4). Only the specimens from
Atibaia showed a tendency to form a separate cluster.
Many individuals collected in rocky outcrops and main-
tained under cultivation (populations 1
5) were grouped
together with plants from Atibaia. Floral characters
showed a high correlation with the first axis (Fig. 4,
Table 2), and the five most important characters to define
this axis were PT
L, LA
W, LA
L, CA
L and LS
L.
Vegetative characters and IN
L were also correlated, but
in a different direction, and did not define any relevant
pattern in the scatter-plot (Fig. 4).
The first canonical axis in the CVA separates the Atibaia
population from plants under cultivation which, together
with the Salesópolis and São Bernardo do Campo popula-
tions, form a mixed group (Fig. 5), the same pattern as that
observed in the PCA (Fig. 4). The five most important
characters for the first axis were CO
L, LL
L, CL
W, LS

W and LA
W (Table 2). Individuals from São Bernardo do
Fig. 2. Map with the localities of the populations studied in Campo showed a tendency to form a separate cluster along
southeastern Brazil. See Table 1 for locality codes. the second canonical axis (Fig. 5), and the five most

131
Table 2. Morphological characters used in morphometric analyses of E. secundum and results of the PCA (PC1 and PC2). the CVA (CN1 and
CN2) and discriminant analysis (F-to-remove). PC1 and PC2: correlations between the original variables and principal components one and
two. respectively. CN1 and CN2: correlations between the original variables and canonical discriminant axes one and two. respectively. F-to-
remove: relative importance of the variables used in the model to discriminate groups in discriminant analysis (Table 4).

Characters Abbreviation PC1 PC2 CN1 CN2 F-to-remove

1. Leaf length LE
L 0.506 0.650 0.334 0.716 15.72

2. Leaf width LE
W 0.253 0.661 0.424 0.138 6.23
3. Stem length ST
L 0.365 0.642 0.037 0.042 1.29
4. Inflorescence length IN
L 0.521 0.595 0.303 0.177 4.06
5. Pedicel length PE
L 0.827 0.012 0.231 0.752 14.27
6. Dorsal sepal length DS
L 0.890 0.175 0.043 0.226 0.73
7. Dorsal sepal width DS
W 0.813 0.022 0.143 0.361 3.22
8. Lateral sepal length LS
L 0.855 0.158 0.120 0.253 0.77
9. Lateral sepal width LS
W 0.843 0.026 0.201 0.099 0.86
10. Petal length PT
L 0.904 0.158 0.130 0.152 0.59
11. Petal width PT
W 0.802 0.015 0.096 0.245 9.28
12. Lip length LA
L 0.884 0.191 0.108 0.096 0.34
13. Lip width LA
W 0.896 0.077 0.480 0.026 3.71
14. Column length CO
L 0.766 0.127 0.458 0.393 9.08
15. Lateral lobe of lip length LL
L 0.762 0.010 0.425 0.178 4.49
16. Lateral lobe of lip width LL
W 0.721 0.040 0.099 0.359 3.69
17. Central lobe of lip length CL
L 0.808 0.102 0.103 0.086 1.58
18. Central lobe of lip width CL
W 0.816 0.110 0.577 0.101 11.04
19. Callus of lip length CA
L 0.869 0.136 0.178 0.225 2.30
20. Callus of lip width CA
W 0.722 0.176 0.132 0.232 2.40

important characters on this axis were DS
W, CO
L, LL
may be involved in this morphological differentiation from
W, PT
W and CA
L. the other populations, because the Atibaia population in
The Atibaia population showed the highest percentage of addition to populations 1
5 are rupiculous and more
correct classification in the jack-knife procedure (Table 4). exposed to the sun, while individuals from the Atlantic
The five most important characters in the classification rainforest populations grow on humus-rich soil, in a more
function, according to the F-to-remove values, were LE
L, shaded environment surrounded by small trees and shrubs
PE
L, CL
W, PT
W and CO
L (Table 2). (Fig. 4, 5, Table 4). This pattern may be related to a general
size difference in floral characters, where the smallest values
are those of the Atibaia population (Table 3), rather than a
Discussion difference in shape: all the floral characters showed a strong
correlation with only one axis, both in the PCA and in the
The PCA, the CVA, and the DA generated very similar CVA.
results (Fig. 4, 5, Table 4). The Atibaia population, which In the PCA, the CVA and in the jack-knife classification
grows on rocky outcrops, is separated from the other matrix (Table 2, 4), the floral characters were the most
populations in all the analyses. Therefore, the habitat type important to define the observed patterns, except for
character LE
L, which was the most important in the
jack-knife classification matrix. LA
W, CO
L, and CL
W
appeared among the five most important characters in at
least two of the three analyses performed (Table 2, 4). These
characters are localized on the lip (Fig. 3), a structure
traditionally used to delimit the species of the E. secundum
complex (Pabst and Dungs 1975, Brieger 1976
1977,
Sastre 1990a, 1990b), and are mainly qualitative, reflecting
shape of the lip callus and the type of lip margins
(denticulate, crenate, fimbriate, erose, etc.). Because these
qualitative characters showed some relation with the groups
obtained, displaying a high variation between populations
and within a single population (Fig. 1), they must be used
with caution as diagnostic characters in the taxonomy of the
group.
Several authors (Dunsterville 1979, Dressler 1989,
Hágsater and Arenas 2005) have already noted the ability
of individuals of E. secundum to colonize new kinds of
Fig. 3. Outline of flower segments and quantitative floral environments. Their commonly ruderal habit can have
characters surveyed in this study (see Table 2 for character codes). several implications for the group’s evolution, as hybridiza-
Scale bar 4 mm. tion (Carson and Templeton 1984, Rieseberg 1997) and

132
Table 3. Summary of descriptive statistics for quantitative morphological characters measured on 226 individuals from E. secundum. (5 percentile-) mean9standard deviation (95 percentile). CH
polyploidy (Levin 2001) often occur in such habitats. A
significant morphological variation was observed by Shaw

(279.27 ) 317.039151.18 (354.80)

(246.74 ) 267.81984.36 ( 288.88)
(59.44) 62.88913.74 ( 66.31)
(1998) in individuals from different species of Dacthylorhiza

(21,74) 23,0395,16 ( 24,32)

(15.38) 15.9892.43 ( 16.59)

(2.96) 3.0590.37 ( 3.14)

(2.90) 3.0090.42 ( 3.11)
(3.17) 3.3090.53 ( 3.43)
(2.66) 2.7790.43 ( 2.87)
(4.37) 4.5490.68 ( 4.71)
(2.78) 2.8790.38 ( 2.97)
(4.04) 4.1790.53 ( 4.30)
(7.65) 7.8790.87 ( 8.08)
(4.59) 4.6890.38 ( 4.77)
(2.30) 2.3990.36 ( 2.48)
(8.08) 8.2690.75 ( 8.45)
(3.20) 3.3190.41 ( 3.41)
(8.47) 8.6490.71 ( 8.82)
(3.75) 3.8490.37 ( 3.93)
(8.22) 8.4090.73 ( 8.58)
(Orchidaceae) in industrial waste sites in England, with a
Atibaia population high occurrence of hybridization. The differentiation of the
individuals collected in the population of São Bernardo do
Campo, in a highway margin (Fig. 5, Table 4), provides a
circumstantial evidence of the importance of disturbed
environments in the differentiation of populations of
E. secundum. Highways can act as corridors that geographi-
cally connect isolated species, allowing hybridization be-
tween E. secundum and allied species (Pansarin pers. comm.).
Most characters of plants under cultivation are larger
than those observed in specimens collected directly from
the field (Table 3), as observed in species of Cerastium L.
(Caryophyllaceae) by Brysting and Elven (2000), and in
(325.84 ) 405.009278.55 (484.16)
(354.04 ) 408.809192.67 (463.56)

species of Pleurothallis R. Br. (Orchidaceae) by Borba et al.

(78.32) 83.64918.72 ( 88.96)

(10.33) 10.6090.96 ( 10.87)

(10.15) 10.4190.93 ( 10.68)

(22.13) 24.1096.94 ( 26.07)

(21.38) 22.2993.19 ( 23.19)

(9.97) 10.2390.90 ( 10.48)

(2002). This could be the result of suitable growth

(3.28) 3.4290.50 ( 3.56)

(3.30) 3.4390.46 ( 3.56)
(5.65) 5.8990.87 ( 6.14)
(3.86) 4.0190.51 ( 4.15)
(5.15) 5.3390.62 ( 5.50)
(9.21) 9.5391.11 ( 9.84)
(5.48) 5.6490.54 ( 5.79)
(3.10) 3.2590.53 ( 3.40)
(3.90) 4.1490.83 ( 4.38)
(3.72) 3.8690.47 ( 3.99)

(4.29) 4.4290.45 ( 4.55)

(3.56) 3.7390.60 ( 3.90)

Salesópolis population

conditions resulting from constant watering and fertiliza-

tion. However, the morphological variation observed in
the specimens under cultivation was as high as that
observed in the specimens collected directly from the field
(Table 3, Fig. 4, 5), indicating that the morphological
variation has a genetic component in addition to an
environmental one. The comparison of cultivated plants
with individuals collected directly from the field was
important in the study of Brysting and Elven (2000), since
they were able to evaluate which characters that were more
influenced by the environment, and which had a strong
genotypic component. In some studies, all analyzed
São Bernardo do Campo population

individuals were previously cultivated (Loos 1993, Borba

(530.00 ) 598.139188.94 (666.25)
(492.67 ) 553.139167.67 (613.58)
(100.70 ) 109.47924.31 ( 118.23)

et al. 2002, Goldman et al. 2004) to suppress the

(9.33) 9.7591.19 ( 10.18)
(9.78) 10.2391.25 ( 10.68)

(9.75) 10.1991.24 ( 10.64)

(23.55) 25.5395.49 ( 27.51)

(19.42) 20.6693.44 ( 21.90)

(9.35) 9.8191.30 ( 10.28)

(5.96) 6.3291.00 ( 6.69)

(3.71) 3.9390.61 ( 4.15)
(3.32) 3.5190.51 ( 3.69)
(5.76) 5.9690.55 ( 6.15)
(3.06) 3.2590.51 ( 3.43)
(3.87) 4.1190.67 ( 4.35)
(3.32) 3.5490.62 ( 3.76)
(3.82) 3.9890.44 ( 4.14)

(4.30) 4.4790.48 ( 4.65)

(3.39) 3.6090.57 ( 3.80)

(4.87) 5.1790.83 ( 5.47)

environmental influence on the characters. In this study,

all measures taken from cultivated plants showed a
variation similar to that of individuals collected directly
from the field (Table 3), therefore, it was not possible to
evaluate which characters that are more influenced by the
environment. The morphological similarity between speci-
mens from the Atibaia population and the individuals
originating from populations 1
5 (and maintained under
cultivation) can also be related to a genetic component.
Molecular data used together with morphometric data
have proved to be important tools to delimit species and/
or lineages in plants displaying high morphological
variation (Kjaer et al. 2004, Pedersen 2004, Bernardos
(428.99 ) 465.449163.80 (501.89)
(393.16 ) 435.639190.84 (478.09)

et al. 2005).
(77.34) 80.60914.66 (83.86)

(9.91) 10.1991.28 ( 10.48)

(10.25) 10.5091.16 ( 10.76)
(22.33) 23.3994.75 ( 24.44)

(23.06) 23.8893.66 ( 24.69)

(9.99) 10.2491.09 ( 10.48)

(10.42) 10.7191.29 ( 10.99)

Because this study was a first attempt to understand the

Plants under cultivation

(4.01) 4.1490.56 ( 4.26)

(3.88) 4.0390.69 ( 4.19)
(5.46) 5.5990.62 ( 5.73)
(3.04) 3.1790.58 ( 3.30)
(4.41) 4.6190.88 ( 4.80)
(3.50) 3.6390.58 ( 3.76)
(6.59) 6.8391.04 ( 7.06)
(3.97) 4.0790.43 ( 4.16)

(4.53) 4.6490.50 ( 4.76)

(3.63) 3.7590.58 ( 3.88)

(5.36) 5.5190.70 ( 5.67)

morphological relationships among populations of E.

secundum, more questions than answers were raised. To
date, it is not possible to judge if the morphological
variation observed is sufficient to circumscribe more than
one species in this sample, as the variation between
populations may be due to phenotypic plasticity related to
environmental conditions. The taxonomical doubts on how
characters (see Table 2).

many species and/or lineages that do exist, due to the

influence of habitat type on the morphological variation
between populations, could be tested if more populations
from rocky outcrops and the Atlantic rainforest are
sampled, and if other methods like molecular markers
CA
W
DS
W

CL
W
LA
W
PT
W

CO
L

LL
W
LS
W
LE
W

(e.g. isozymes or microsatellites) are employed, to reveal

CA
L
DS
L

CL
L
LA
L
PT
L
IN
L
ST
L

PE
L

LL
L
LS
L
LE
L
CH

patterns of genetic variability.

133
Fig. 4. PCA of 226 specimens from E. secundum based on 20 characters (Table 2). Specimens from populations 1
5 are indicated by
arrows. Principal compontent (PC) 1 and 2 explain 58.1% and 9.2% of the total variation respectively.

Fig. 5. CVA of 226 specimens from E. secundum based on 20 characters (Table 2) and the groups ‘Cultivated plants’. ‘Atibaia’.
‘Salesópolis’. and ‘São Bernardo do Campo’ populations. Individuals from populations 1
5 are indicated by arrows. Axes 1 and 2 explain
63.5% and 24.3% of the total variation. respectively. The ellipses are centered on the sample means and comprise 70% of the sample
from each group.

Table 4. Results of jack-knife classification analysis with plants under cultivation and natural populations from São Bernardo do Campo,
Atibaia and Salesópolis as groups. Wilks’ lambda0.1147. p0.00001.

Grouping variables Cultivated plants São Bernardo do Campo Atibaia Salesópolis Percentage correct

Cultivated plants 54 6 8 12 68
São Bernardo do Campo 3 23 2 4 72
Atibaia 0 0 63 1 98
Salesópolis 11 7 4 28 56
Total 68 36 77 45 74

134
Acknowledgements
Thanks to the staff of the ‘‘State Orchidarium Frontier, S. 1976. Study of the decrease of eigenvalues in principal
Section’’ of the Inst. of Botany, São Paulo, Brazil, for their component analysis: comparison with the broken stick model.
technical support. Thanks also to Dr George J. Shepherd for his
J. Exp. Mar. Biol. Ecol. 25: 67
75.
valuable comments on the manuscript. This work was supported Goldman, D. H. et al. 2004. Morphometric circumscription of
by a grant from ‘‘Fundação de Amparo à Pesquisa do Estado de species and infraspecific taxa in Calopogon R. Br. (Orchida-
São Paulo’’ (Fapesp no. 00/07814
3) to the first author. The ceae).
Plant Syst. Evol. 247: 37
60.
second author acknowledges the ‘‘Conselho Nacional de Desen- Hágsater, E. 1984. Towards an understanding of the genus
volvimento Cientı́fico e Tecnológico’’ (CNPq grant no. 303962/ Epidendrum.
In: K. W. Tan (ed.), Proc. 11th World Orchid
2004
6) for the grant received. Conf., Miami, pp. 195
201.
Hágsater, E. 1993. Epidendrum anceps or Epidendrum secundum?

Orquidea, Mex. 13: 153
158.
References
Hágsater E. and Arenas M. A. S. 2005. Epidendrum L. In:

Pridgeon, A. M. et al. (eds), Genera Orchidacearum. Vol. 4.
Oxford Univ. Press, pp. 236
251.
Bernardos, S. et al. 2005. The section Pseudophrys (Ophrys, Kallunki, J. 1976. Population studies in Goodyera (Orchidaceae)
Orchidaceae) in the Iberian Peninsula: a morphological and with emphasis in the hybrid origin of G. tesselata.
Brittonia
molecular analysis.
Bot. J. Linn. Soc. 148: 359
375. 28: 53
75.
Borba, E. L. et al. 2002. Floral and vegetative morphometrics of Kephart, S. et al. 1999. Varietal relationships in Silene douglasii
five Pleurothallis (Orchidaceae) species: correlation with (Caryophyllaceae): morphological variability at the population
taxonomy, phylogeny, genetic variability and pollination level.
Syst. Bot. 24: 529
544.
systems.
Ann. Bot. 90: 219
230. Kjaer, A. et al. 2004. Investigation of genetic and morphological
Brieger, F. G. 1976
1977. Gattungsreihe Epidendra.
In: Brieger, variation in the Sago Palm (Metroxylon sagu; Arecaceae) in
F. G. et al. (eds), Schlechter die Orchideen. Vol. 3. Paul Parey,
Papua New Guinea.
Ann. Bot. 94: 109
117.
Berlin, pp. 509
549.
Knyasev, M. S. et al. 2000. Interspecific hybridization in northern
Brysting, A. K. and Elven, R. 2000. The Cerastium alpinum
C.
Eurasian Cypripedium: morphometric and genetic evidence of
articum complex (Caryophyllaceae): numerical analyses of
hybrid origin of C. ventricosum.
Lindleyana 15: 10
20.
morphological variation and a taxonomic revision of C.
Levin, D. A. 2001. The recurrent origin of plant races and species.
articum Lange s.l.
Taxon 49: 189
216.

Syst. Bot. 26: 197
204.
Cardim, D. C. et al. 2001. Variabilidade intra-especı́fica em cinco
Lindley, J. 1852
1859. Epidendrum.
In: Lindley, J. (ed.), Folia
populações de Oncidium varicosum Lindl. (Orchidaceae

Orchidacea. J. Matthews, London, pp. 1
97.
Oncidiinae) em Minas Gerais.
Rev. Bras. Bot. Suppl. 24:
553
560. Loos, B. P. 1993. Morphological variation in Lolium (Poaceae) as
Carlini-Garcia, L. A. et al. 2002. A morphometric analysis of floral a measure of species relationships.
Plant Syst. Evol. 188: 87

characters in Miltonia spectabilis and Miltonia spectabilis var. 99.
moreliana (Maxillarieae: Oncidiinae).
Lindleyana 17: 122
Manly, B. F. J. 1994. Multivariate statistical methods, a primer.
129.
Chapman and Hall.
Carson, H. L. and Templeton, A. R. 1984. Genetic revolutions in Marhold, K. 1996. Multivariate morphometric study of the
relation to speciation phenomena: the founding of new Cardamine pratensis group (Cruciferae) in the Carpathian and
populations.
Annu. Rev. Ecol. Syst. 15: 97
131. Pannonian area.
Plant Syst. Evol. 200: 141
159.
Catling, P. M. and Catling, V. R. 1997. Morphological Moccia, M. D. et al. 2007. The strength of reproductive isolation
discrimination of Platanthera huronensis in the Canadian rocky in two hybridizing food-deceptive orchid species.
Mol. Ecol.
mountains.
Lindleyana 12: 72
78. 16: 2855
2866.
Chase, M. W. et al. 2003. DNA data and Orchidaceae systematics: Pabst, G. F. J and Dungs, F. 1975. Orchidaceae Brasilienses. Vol.
a new phylogenetic classification.
In: K. W. Dixon et al. 1.
Kurt Schmersow, Hildesheim.
(eds), Orchid conservation. Natural History Publications, Palestina, R. A. and Sosa, V. 2002. Morphological variation in
Sabah, pp. 69
89. populations of Bletia purpurea (Orchidaceae) and description
Dressler, R. L. 1967. The genera Amblostoma, Lanium and of the new species B. riparia.
Brittonia 54: 99
111.
Stenoglossum.
Brittonia 19: 237
243. Pedersen, H. E. 2004. Dactylorhiza majalis s.l. (Orchidaceae) in
Dressler, R. L. 1984. La delimitacion de géneros en el complejo acid habitats: variation patterns, taxonomy, and evolution.
Epidendrum.
Orquidea, Mex. 9: 277
298.
Nord. J. Bot. 22: 641
658.
Dressler, R. L. 1989. Will the real Epidendrum ibaguense please Reinhammar, L. 1995. Evidence for two distinctive species of
stand up?
Bull. Am. Orch. Soc. 58: 796
800. Pseudorchis (Orchidaceae) in Scandinavia.
Nord. J. Bot. 15:
Dufréne, M. et al. 1991. Biostatistical studies on western 469
481.
European Dactylorhiza (Orchidaceae
the D. maculata group. Rieseberg, L. H. 1997. Hybrid origins of plant species.
Annu.

Plant Syst. Evol. 175: 55
72. Rev. Ecol. Evol. Syst. 28: 359
389.
Dunsterville, G. C. K. 1979. Orchids of Venezuela
Epidendrum Sapir, Y. et al. 2002. Morphological variation of the Oncocyclus
elongatum.
Bull. Am. Orch. Soc. 48: 447
454. irises (Iris: Iridaceae) in the southern Levant.
Bot. J. Linn.
Dunsterville, G. C. K. and Garay, L. A. 1961. Venezuelan orchids Soc. 139: 369
382.
illustrated. Vol. 2.
Andre Deutsch, London. Sastre, C. 1990a. Epidendrum bambusiformes de Guadeloupe
Du Puy, D. J. et al. 1985. A numerical taxonomic analysis of et de Martinique 1.
L’ Orchidophile 93: 149
158.
Cymbidium section Iridorchis (Orchidaceae).
Kew Bull. 40: Sastre, C. 1990b. Epidendrum bambusiformes de Guadeloupe
421
434. et de Martinique 2.
L’ Orchidophile 94: 197
203.
Everitt, B. S. 1978. Graphical techniques for multivariate data. Shaw, P. J. A. 1998. Morphometric analyses of mixed Dactylorhiza

North-Holland. colonies (Orchidaceae) on industrial waste sites in England.
Fritsch, P. W. and Lucas, S. D. 2000. Clinal variation in the
Bot. J. Linn. Soc. 128: 385
401.
Halesia carolina complex (Styracaceae).
Syst. Bot. 25: 197
Shepherd, G. J. 1994. FITOPAC 1: Manual de usuário.
Depto
210. de Botânica, Univ. Estadual de Campinas, Brazil.

135
Tyteca, D. 1995. Multivariate analyses of western European Tyteca, D. and Gerbaud, O. 1998. Nouvelles observations sur
allogamous populations of Epipactis helleborine (L.) Crantz Dactylorhiza lapponica (Laest. ex Hartman) Soó en France.
L.
s.l., with special emphasis on Epipactis tremolsii Pau in Orchidophile 131: 60
65.
southeastern France.
Ber. Arbeitskrs. Heim. Orchideen. Vasquez, R. and Ibisch, P. L. 2004. Orquı́deas de Bolı́via. Vol. 2:
12: 4
49. Subtribus Laeliinae, Polystachinae, Sobraliinae.
Editorial
Tyteca, D. and Dufréne, M. 1994. Biostatistical studies of western FAN, Santa Cruz de la Sierra.
European allogamous populations of the Epipactis helleborine Wilkinson, L. 2000. Systat Statistics ver. 10.0 for Windows.
(L.) Crantz species group (Orchidaceae).
Syst. Bot. 19:
SPSS, Chicago.
424
442. Withner, C. L. and Harding, P. A. 2004. The Cattleyas and their
Tyteca, D. and Gathoye, J.-L. 1993. On the morphological relatives. Vol. 7: the debatable Epidendrums.
Timber Press,
variability of Dactylorhiza praetermissa (Druce) Soó (Orchida- Portland.
ceae).
Belg. J. Bot. 126: 81
99.

136