Integration of DNA taxonomy and geometric morphometry for species delimitation as a tool for solving taxonomic problems in the genus Gnopharmia Staudinger, 1892 (Lepidoptera, Geometridae, Enominae)

Abstract

The genus Gnopharmia belongs to the tribe Macariini (Ennominae) and includes so far
thirteen traditionally recognised species which, however, are difficult to distinguish by
their external morphology. Departing from a taxonomic revision of this genus we
analysed the morphological variation of the sternite A-8 using geometric morphometrics
(Eigenshape PCA analyses with subsequent AMOVA analyses of informative
characters). Subsequently we analysed a 658 bp fragment of cytochrome oxidase 1
(CO1) to assess the congruence between traditionally defined morphospecies with
mtDNA clusters and with morphometric patterns of variation.
Although some morphological characters in this genus are extreme variable and overlap
significantly between the species, those discrete morphological characters used so far
for species delineation with classical taxonomy (i.e. genitalia characters) support seven
clearly separated groups. ML tree topology with all morphospecies being monophyletic
and genetic distances of mtDNA data (CO1) are consistent with these groupings.
At the other hand, geometric morphometric analysis of Sternite-A8 separates the genus
into two principal groups, and the same pattern is also found with a parsimony network
analysis of CO1 data.
Investigating the correlation of variation of discrete morphological characters and range
patterns with mtDNA clusters and morphometric patterns we infer about the
“significance” of the observed patterns of variations and interpret them in terms of
taxonomy (i.e. to infer the species delineation).

Full text

Integration of cytochrome c oxidase I barcodes
and geometric morphometrics to delimit species
in the genus Gnopharmia (Lepidoptera:
Geometridae, Ennominae)
HOSSEIN RAJAEI Sh
1
*, JAN-FREDERIC STRUWE
1
, MICHAEL J. RAUPACH
2
,
DIRK AHRENS
1
and J. WOLFGANG WÄGELE
1
1
Zoologisches Forschungsmuseum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany
2
Deutsches Zentrum für Marine Biodiversitätsforschung, Senckenberg am Meer, Südstrand 44,
26382 Wilhelmshaven, Germany
Received 9 January 2013; revised 2 May 2013; accepted for publication 9 May 2013
We tested the hypothesis of species taxonomy in the genus Gnopharmia (Macariini, Ennominae) that was recently
established in a review based on discrete morphological characters. For this objective we integrated both
DNA-based and morphometric approaches in order to infer species boundaries. A 658-bp fragment of the
mitochondrial cytochrome c oxidase subunit 1 (CO1) (DNA barcode) was analysed from populations of five species
distributed throughout the Middle East to assess their consistency with traditionally defined morphospecies.
Signals in the morphological variation of the aedeagus of all relevant populations were evaluated using geometric
landmarks. Consistent groupings compatible with the current taxonomic classification were found with both
approaches. The results strongly support the distinction of seven closely related species.
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83.
doi: 10.1111/zoj.12053
ADDITIONAL KEYWORDS: DNA barcode – geometric morphometry – integrative taxonomy – Middle
East – mitochondrial DNA.
INTRODUCTION
Taxonomy has experienced a renaissance in the
past decade in the form of a new integrative, multidis-
ciplinary approach for species delimitation. Today,
taxonomy can be faster and more reliable than previ-
ously (Padial et al., 2010). Amongst the different
methods that have been used and that have demon-
strated benefits (Schlick-Steiner et al., 2010), DNA
taxonomy and barcoding have emerged recently as
rapid and effective methods for species identification,
delineation, and accelerated biodiversity assessment
(e.g. Hajibabaei et al., 2006; Pons et al., 2006; Ahrens,
Monaghan & Vogler, 2007; Stoeckle & Hebert, 2008;
Valentini, Pompanon & Taberlet, 2008; Smith &
Fisher, 2009; Zhou et al., 2009; Hebert, deWaard &
Landry, 2010). Although DNA taxonomy and barcoding
follow different assumptions and approaches (Vogler &
Monaghan, 2007), their effectiveness has been shown
in many studies on both vertebrate and invertebrate
taxa (e.g. Hajibabaei et al., 2006; Clare et al., 2007;
Monaghan et al., 2009; Rivera & Currie, 2009;
Robinson et al., 2009; Raupach et al., 2010; da Silva
et al., 2011; Hausmann, Haszprunar & Hebert, 2011;
Lakra et al., 2011; Astrin et al., 2012). Data from DNA
barcoding can lead to and contribute sound, substan-
tiated results in taxonomy when combined with other
types of data (e.g. distinction of morphospecies).
Modern morphometric analyses have also become
a favoured tool for quantitative and multidi-
mensional comparison of morphological characters
(Schlick-Steiner et al., 2006; Bichain et al., 2007;
Hahn et al., 2011; Toth & Varga, 2011; von Reumont
*Corresponding author. E-mail: eagle426@uni-bonn.de
bs_bs_banner
Zoological Journal of the Linnean Society, 2013, 169, 70–83. With 5 figures
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–8370

et al., 2012). Landmark-based geometric morphom-
etry (Rohlf, 1993; Rohlf & Marcus, 1993; Bookstein,
1997) allows the evaluation of phenotypic differentia-
tions and has a high discriminatory power at species
and even subspecies level (Adams & Funk, 1997;
Alibert, Moureau & Dommergues J-L, 2001; Villegas,
Feliciangeli & Dujardin, 2002; Nolte & Sheets, 2005).
The common species-specific diversity of insect geni-
talia (Eberhard, 1985; Shapiro & Porter, 1989) makes
this trait fit for morphological analyses of species
divergence. Multivariate analyses of these morpho-
logical data allow closely related species to be distin-
guished (Wakeham-Dawson et al., 2004; Toth &
Varga, 2010, 2011).
In this context, DNA barcoding and geometric mor-
phometry were used to identify and classify species of
the genus Gnopharmia Staudinger, 1892 (Geometri-
dae, Ennominae). This taxon has a very complicated
taxonomy, and the validity of species and subspecies
within this genus has been challenged by several
authors (e.g. Lederer, 1870; Staudinger, 1892;
Staudinger & Rebel, 1901; Prout, 1912–1916; Wehrli,
1938, 1939, 1939–1954; Ebert, 1965; Wiltshire, 1967,
1970). However, none of these studies led to satisfac-
tory conclusions because the external morphological
characters of adult specimens (e.g. wing pattern) are
subject to strong infraspecific variability and plastic-
ity, and discrete characters of genitalia are limited.
In a recent taxonomic revision, a large data set
(> 2000 specimens and around 900 genitalia prepa-
rations including type series of all taxa) was studied
with scanning electron microscopy (SEM) and light
microscopy (Rajaei Sh, Stüning & Trusch, 2012).
Based on this morphological study, ten out of 19 taxa
of the genus were synonymized and an updated
checklist of the genus, currently comprising nine
valid taxa in seven species, was provided as follows:
(1) Gnopharmia cocandaria cocandaria (Erschoff,
1874) and Gnopharmia cocandaria afghanistana
Wiltshire, 1967; (2) Gnopharmia colchidaria col-
chidaria (Lederer, 1870), Gnopharmia colchidaria
sinesefida Wehrli, 1941, and Gnopharmia colchidaria
objectaria Staudinger, 1892; (3) Gnopharmia erema
Wehrli, 1939; (4) Gnopharmia irakensis Wehrli, 1938;
(5) Gnopharmia kasrunensis Wehrli, 1939; (6) Gnop-
harmia rubraria Staudinger, 1892; and (7) Gnop-
harmia sarobiana Ebert, 1965 (see Rajaei Sh et al.,
2012 for more details on new taxonomic changes).
However, some taxonomic questions still remained
unresolved. For example, the populations of G. ru-
braria from west and central Turkey are highly vari-
able in wing colour and genitalia in comparison with
specimens from east Turkey, Syria, and Israel. A
second question concerns three morphs (or so-called
subspecies; Rajaei Sh et al., 2012) of G. colchidaria
that show clear differences in their genitalia.
Although Rajaei Sh et al. (2012) considered the vari-
ation in populations of G. rubraria to be intraspecific
clines within the species and defined the three
morphs of G. colchidaria as subspecies, further tests
are in order. We tested the delimitation of these taxa
by using molecular and morphometric methods.
The aim of this study was to delimit the species
of the genus Gnopharmia based on (1) DNA bar-
codes, employing distance clustering and parsimony
network analyses, and (2) assessment of aedeagus
characters based on eight landmarks analysed with
principal component analyses (PCAs) and canonical
variates analysis (CVA). All results are compared
with earlier taxonomic classifications (Rajaei Sh
et al., 2012).
MATERIAL AND METHODS
SAMPLES
We sampled five of the seven species of Gnopharmia
for DNA sequencing, whereas six species were part of
a morphometric analysis (Table 1). For two species,
G. erema (reported only from Iraq) and G. sarobiana
(distributed only in eastern Afghanistan), no suitable
material was available for DNA extraction. Speci-
mens were either freshly collected in Iran or loaned
from different collections: Staatliches Museum für
Table 1. Number of analysed specimens employed in the different analyses
Group
Distance threshold
cluster analysis
Parsimony network
analysis
Geometric
morphometrics
Traditional taxonomy
(Rajaei Sh et al., 2012)
Gnopharmia irakensis 29 29 35 415
Gnopharmia kasrunensis 53 53 50 180
Gnopharmia rubraria 21 21 39 249
Gnopharmia colchidaria 108 108 124 1010
Gnopharmia cocandaria 9 9 30 45
Gnopharmia sarobiana – – 20 47
Gnopharmia erema – – – 8 (type series only)
INTEGRATIVE TAXONOMY OF GNOPHARMIA 71
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

Naturkunde, Karlsruhe (SMNK); Zoologisches Forsc-
hungsinstitut und Museum Alexander Koenig, Bonn
(ZFMK); Zoologische Staatssammlung München
(ZSM); private collections: Dirk Stadie, Lutherstadt
Eisleben (PCDS); Dr Jörg Gelbrecht, Koenig Wuster-
hausen (PCJG); Dr Jörg-Uwe Meinke, Kippenheim
(PCJM); Manfred Sommerer, Munich (PCMS) [for
detailed sampling data see Figure 1 and Table S1
(Supporting Information Appendix)].
All analysed Gnopharmia specimens (adults) were
photographed in dorsal and ventral views. Specimens
were identified based on the genital morphology
and compared with type material (except specimens
that were sequenced as part of other projects; see
Table S1). One leg of each individual was used for
DNA extraction. All voucher specimens are preserved
in their original collections for future reference (see
Table S1).
DNA EXTRACTION, AMPLIFICATION, AND SEQUENCING
For each specimen total genomic DNA was extracted
from a leg following a standardized high-throughput
protocol (Ivanova, deWaard & Hebert, 2006). The
barcode region, consisting of approximately 650 bp
of the mitochondrial cytochrome c oxidase subunit
1 (CO1), was amplified with two alternative sets
of primer pairs: LepF1/LepR1 and MLepF1/MLepR1
(Hajibabaei et al., 2006). PCR was performed in
12.5 ml volume containing 2 mlH
2
O, 1.25 ml10¥ PCR
buffer, 6.25 ml 10% trehalose, 0.0625 ml deoxyribonu-
cleotide triphosphates (10 mM), 0.625 ml MgCl
2
(50 mM), 0.06 ml Platinum Taq polymerase (Invitro-
gen), 0.125 ml of each primer (10 mM), and 2 ml DNA
template. The PCR included an initial denaturation
at 94 °C (1 min), followed by five cycles of 94 °C (dena-
turation, 40 s), 45 °C (annealing, 40 s), and 72 °C
(extension, 1 min), an additional 35 cycles at 94 °C
(40 s), 51 °C (40 s), and 72 °C (1 min), and a final
extension step of 72 °C (5 min). PCR products were
visualized on 96-well precast 2% agarose gels. An ABI
3730xl DNA Analyser (Applied Biosystems) was used
for bidirectional sequencing with BigDye (v. 3.1),
using the same primers as used in PCR. We per-
formed BLAST searches in GenBank to check for
contamination (Altschul et al., 1990). DNA extraction,
PCR, and sequencing were carried out at the Cana-
dian Centre for DNA Barcoding (CCDB) of the Uni-
versity of Guelph. All trace files, images, global
positioning system coordinates, specimen data,
and taxonomic data are accessible via the Barcode
of Life Data System website (BOLD, http://www
.barcodinglife.org) under the project: ‘Gnopharmia of
the Palaearctic [GEAGN]’ (Table S1). Sequences are
also available in GenBank (for accession numbers see
Table S1).
Figure 1. Sampling sites; the numbers indicate localities with at least one sequenced specimen. The circles and ovals
show the geographical distribution of species based on a taxonomic revision (Rajaei Sh et al., 2012). The distribution of
Gnopharmia irakensis is not shown, as it occurs sympatrically with other species in most areas.
72 H. R. Sh ET AL.
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

PHYLOGENETIC ANALYSIS AND SPECIES DELIMITATION
All sequences were edited using BioEdit (v. 7.1.3)
(Hall, 1999) and aligned using MAFFT (v. 6) (Katoh
et al., 2002). Two sequences (Neognopharmia steve-
naria and Isturgia sublimbata) were used as out-
groups (see Table S1 for accession numbers). A
substitution model test was performed with MODEL-
TEST as implemented in MEGA5 (Tamura et al.,
2011). Maximum likelihood and neighbour-joining
trees with 1000 pseudoreplicates (Felsenstein, 1985)
were calculated with MEGA5. Pairwise distances
were calculated within and between the morphospe-
cies using a Kimura two-parameter model. The
barcode gap was calculated following Puillandre et al.
(2012) (Fig. 2).
Statistical parsimony networks were determined
using TCS v. 1.3 (Clement, Posada & Crandall, 2000).
Parsimony network analyses partition the sequence
data into groups of closely related haplotypes con-
nected by branches with a > 95% probability of
being nonhomoplastic. Based on the TCS analysis,
identical haplotypes were removed from the data set
for further analysis. All haplotype networks were
mapped on an ultrametric tree of haplotypes (Fig. 3).
We used the general mixed Yule coalescent model
(GMYC) of Pons et al. (2006) to infer species bounda-
ries by delimiting mtDNA clusters based on the tran-
sition from slow to faster apparent branching rates on
the gene tree expected at the species boundary. The
method reconstructs a threshold ‘age’ such that nodes
before the threshold are considered to be speciation
and nodes subsequent to the threshold reflect coales-
cence occurring within each species.
Distance-based clustering was carried out using
the Species Identifier module within TaxonDNA
(v. 1.6.2) (Meier et al., 2006). This approach clusters
the sequences to the putative species based on their
p-distances under a preselected threshold (Meier
et al., 2006; Hendrich et al., 2010). Species Identifier
groups all distances from a clique (where all individu-
als are connected to each other by distance values
below a cut-off threshold) into a single cluster, as it
does with quasicliques (where some individuals are
connected indirectly, i.e. some distances in the cluster
infringe the threshold). For the calibration of the
cut-off threshold, we determined the percentage of
threshold for the best match between the resulting
clusters and the morphospecies represented in the
data set. Finally, the resulting species were mapped
on the ultrametric tree based on their species cluster
(Fig. 3).
MORPHOMETRIC ANALYSIS
Landmark-based geometric morphometry (Bookstein,
1997; Zelditch et al., 2004; Klingenberg, 2009) was
applied to analyse abdominal sternite VIII, forewing
venation, and aedeagus shape. However, because
neither the shape of abdominal sternite VIII nor of
the wing venation allowed differentiation between
the majority of the morphospecies, analyses of these
two traits are not reported here. In contrast to these
two traits, the landmark-based analysis of aedeagus
shape (Fig. 4) was found to be much more informative
(see also Rajaei Sh et al., 2012). For this data set,
eight homologous landmarks (LMs) (Fig. 4, LMs
3–10) per individual were digitized using the TPS
software package (Rohlf & Marcus, 1993; Rohlf,
2004a, b). LMs 1 and 2 were used for the scaling
process. As an additional character, the fore- and
hindwing was tested separately using 11 LMs each.
Subsequent analyses of the landmark data were
conducted using the IMP software package (Sheets,
2002). The generalized least square Procrustes
superimposition (Rohlf, 1990; Rohlf & Slice, 1990)
integrated into the program CoordGen6f (Sheets,
2002) was used to remove nonshape variation as
effects of size and position by scaling all specimens
to unit size, translating them to a common location,
and rotating them to line up their corresponding
landmarks as closely as possible. By this procedure
the Procrustes superimposition minimizes differences
between landmark configurations without altering
shape.
Shape diversity of samples was analysed with
PCA (Chatfield & Collins, 1980; Morrison, 1990) and
CVA (Mardia, Kent & Bibby, 1979; Campbell &
Atchley, 1981). PCA variables were used to examine
variation amongst individuals within a sample and to
Figure 2. Barcode gap between intra- and interspecific
pairwise distances based on Kimura two-parameter pair-
wise distance analysis.
INTEGRATIVE TAXONOMY OF GNOPHARMIA 73
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

74 H. R. Sh ET AL.
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

eventually find distinct outliers, which might point to
errors in species identification or to misplaced land-
marks. A CVA analysis was performed directly on the
data. The resulting scatter plots show the group’s
consistency and illustrate the relative position of
groups to each other in relation to the overall vari-
ance. The specimens were grouped according to the
results of the CO1 analyses to check for consistency.
The program used to conduct the analyses was
CVAGen6o (IMP package: Sheets, 2002).
An assignment test as implemented in the
CVAGen6o program was used to test for stability
and differentiation of the groups at individual level.
The degree of differentiation between the tested
groups was measured by the Mahalanobis distance
(Mahalanobis, 1936) of each specimen from the mean
value of the CVA scores for each group, and assigning
each specimen to its closest group. The assignment
test determines the probability that the Mahalanobis
distance between an individual and the mean of the
group is larger than expected under a null model of
random variation around the mean of each group. The
approach used follows Cornuet et al. (1999). In this
method, the distribution of distances produced by a
Monte Carlo simulation is used to determine whether
or not the observed distance of a given specimen is
consistent with the null model of random variation
around the mean of the group to which the specimen
is assigned.
We performed jack-knife assignment tests (1000
trials) to analyse the robustness of the data set with
30, 50, 70, and 90% of the samples randomly sorted
out at a time (Quenouille, 1956; Tukey, 1958). Circu-
larity in the assignment process was avoided by not
using the assigned specimens during the axes calcu-
lation. Afterwards, the fit of the abstracted samples of
this new compound data set was tested. Percentage
values for assignment success are specified as signifi-
cant or insignificant and as correct or incorrect
assigned trials. Goodall’s F test (Fisher, 1925) inte-
grated in the TwoGroup6h program (Sheets, 2002)
was used to perform pairwise comparisons of several
populations to test the significance of the difference
between samples using a Procrustes superimposition.
To ensure that the observed differences did not arise
by chance from sampling a single population, speci-
mens of the groups were pooled to draw bootstrap
sets of the original sample size. From these pairs
the F-ratio was computed. The bootstrap sets with an
F-value at least as large as the original set were
counted and divided by the number of iterations to
calculate the probability (P-value) of obtaining the
original samples under the null hypothesis (Snedecor
& Cochran, 1967).
RESULTS
DNA SEQUENCING
Although many of the individuals examined were
pinned museum specimens (usually not older than ten
years), complete DNA barcodes were amplified suc-
cessfully in most cases. Sample size per species ranged
from nine (G. cocandaria) to 105 (G. colchidaria) speci-
mens (see Table S1 for more information). Complete
DNA barcodes were obtained for 218 specimens from
five Gnopharmia and two outgroup species. The com-
plete alignment has 107 variable sites with 93
parsimony-informative positions. Base composition
is AT biased (T: 39.9%; C: 14.8%; A: 30.7%; G: 14.6%),
as is typical for insects (e.g. Wei et al., 2010). A
substitution model test suggested a general time-
reversible + G model (with Bayesian score: 7674.198)
as the best model for the maximum likelihood analysis.
Pairwise distance analysis revealed a significant
difference between intra- and interspecific distances
Figure 3. Ultrametric maximum likelihood (ML) tree of haplotypes [same topology as neighbour-joining (NJ) tree].
Numbers indicate bootstrap support (1000 replicates): upper numbers for ML analysis, lower numbers for NJ analysis.
Green shows the congruence and red shows incongruence between molecular analyses and morphospecies. The arrow
indicates the single sequence of Gnopharmia colchidaria colchidaria and black bars indicate those of Gnopharmia
colchidaria objectaria.
䉳
Figure 4. Position of selected landmarks for geometric
morphometry on the aedeagus.
INTEGRATIVE TAXONOMY OF GNOPHARMIA 75
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

(Table 2) and a barcode gap with a distance of 2.4%
(Fig. 2). Maximum intraspecific distances ranged
from 0.29% (in G. irakensis) to 0.9% (in G. rubraria),
whereas minimum interspecific distances had values
from 2.7% (between G. rubraria-G. kasrunensis)to
5.86% (between G. irakensis-G. colchidaria objec-
taria) (Table 2).
Maximum likelihood and neighbour-joining topolo-
gies were highly similar, with all five sequenced
morphospecies resulting as monophyla (bootstrap
values > 90%; except for G. kasrunensis; Fig. 3).
Minimum distances between morphospecies were
higher than 2.5%.
CO1 data revealed two distinct intraspecific genetic
(or haplotype) lineages within clades in the following
taxa: G. irakensis [maximum intraspecific distances
(D): 0.2%], G. rubraria (D: 0.9%), G. kasrunensis (D:
0.5%), and G. cocandaria (D: 0.5%). In the case of
G. colchidaria colchidaria, G. colchidaria objectaria,
and G. colchidaria sinesefida, none of the infrasub-
specific taxa (subspecies) was monophyletic. Popula-
tions of G. rubraria from west, central, and south
Turkey had a distance of 1.6% from the populations
from south-east Turkey, Israel, and Syria.
SPECIES DELIMITATION ANALYSIS
The parsimony network analysis subdivided the 72
unique haplotypes into five different networks, per-
fectly matching with the morphospecies (Fig. 3). The
highest match between DNA clusters under distance-
based clustering and morphospecies was found at a
threshold of 1.25%. Applying this threshold in Species
Identifier, haplotypes were grouped into six different
clusters (Fig. 3) that were widely consistent with the
morphospecies. Only G. rubraria s.l. was split into
two groups.
The GMYC analysis with a single threshold
resulted in extreme over-splitting, with a total yield of
43 GMYC units and a very narrow confidence interval
of zero entities. The likelihood ratio between GMYC
and the null model was 293.61 [with significance
in the likelihood ratio test P = 0.0001; LnLH(G-
MYC) = 662.2; LnLH(0) = 515.4; LH = likelihood].
MORPHOMETRIC ANALYSIS OF THE AEDEAGUS
As previously stated, analyses of the wings were not
informative and are not presented here. With regards
to the aedeagus, individuals clustered according
to the morphospecies. A remarkable separation of
G. irakensis from the rest of the species was found on
the first PC axis. In an additional analysis this group
was taken out because of the strong difference in
its aedeagus. This had no influence on the results
of the PCAs and CVAs. Therefore, further results
Table 2. Intra- and interspecific Kimura two-parameter distances of all analysed Gnopharmia taxa (in %). Gnopharmia. rubraria s.l. refers to all populations
of G. rubraria, G. rubraria (W) refers to populations from west, central, and south Turkey, and G. rubraria (SE) refers to the populations from south-eastern
Turkey, Syria, Israel, and Palestine
Maximum
distances
(intraspecific)
Minimum distances
(interspecific)
Gnopharmia
rubraria
(W)
G.
rubraria
(SE)
G.
rubraria
s.l.
Gnopharmia
colchidaria
sinesefida
Gnopharmia
kasrunensis
Gnopharmia
irakensis s.l.
Gnopharmia
colchidaria
objectaria
Gnopharmia
cocandaria
0.21 G. rubraria (W)
0.04 G. rubraria (SE) 1.6
0.9 G. rubraria s.l.
0.35 G. colchidaria sinesefida 4.2 4.0 4.1
0.55 G. kasrunensis 3.0 2.7 3.0 4.0
0.29 G. irakensis s.l. 5.3 5.1 5.2 5.6 4.7
0.85 G. colchidaria objectaria 4.7 4.6 4.7 1.1 4.3 5.9
0.50 G. cocandaria 5.0 4.6 4.8 4.0 4.1 5.4 4.6
Only one
individual!
G. colchidaria colchidaria 4.2 4.1 4.1 0.2 4.0 5.7 1.0 4.1
76 H. R. Sh ET AL.
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

are presented with G. irakensis included. Another
distinct separation was G. sarobiana on the second
PC axis (see Fig. 5A). For all other species we
found various degrees of differentiation from each
other on PC axes 3–6, explaining 83.73% of the total
variation.
To get a clearer picture of the segregation of
species, the investigation of group differences was
continued with a CVA, resulting in seven significant
axes (Table S2). The first CV axis confirmed the
distinct positioning of G. irakensis, whereas the
second separates G. rubraria s.l. from the rest
(Fig. 5B). The third CV axis (Fig. 5C) places G. cocan-
daria and G. sarobiana on opposite edges with no
overlap between the groups. Although G. colchidaria
sinesefida is always placed closely to G. colchidaria
objectaria, both groups show only minor overlapping
regions. Both groups are separated from G. cocan-
daria. The fourth CV axis shows a substantial sepa-
ration of G. colchidaria colchidaria from the rest of
the groups (Fig. 5D). This group shows overlapping
regions with G. colchidaria objectaria and G. col-
chidaria sinesefida on all other axes. Gnopharmia
kasrunensis shares the distinction from G. irakensis
on the first, and G. rubraria (SE) and G. rubraria (W)
on the second CV axis (Fig. 5B). The latter two groups
overlap along all axes and could not be separated
according to the CV axes. Gnopharmia sarobiana is
separated as a distinct group on the third and fourth
axes (Fig. 5D), and on others has a minor overlap
with G. colchidaria s.l. The three remaining signifi-
cant CV axes showed continuous overlap in the
scatter plots.
The distinctiveness of shape differences between
the observed groups of Gnopharmia was expressed
in the individual-based assignment test on the basis
of the canonical variates scores. In total, 94.93%
of the specimens were reassigned to their original
group (Table S3). The highest assignment success
(100%) was reached by three groups: G. colchidaria,
G. irakensis, and G. sarobiana. The lowest percentage
of correct assignments was found in G. cocandaria
(86.2%). In this species, 13.8% were assigned to
G. colchidaria objectaria, which itself had 1.9% of
incorrect back assignments to G. cocandaria. Recip-
rocal wrong assignments were also present in G. kas-
runensis, G. colchidaria objectaria, G. colchidaria
sinesefida, and in the G. rubraria (SE) and G. ru-
braria (W) groups.
A jack-knife test of group assignments revealed a
very consistent signal in the specimens’ grouping
even when large partitions of the whole data set were
removed. The discriminatory power of the CV axes
did not break down until 90% of specimens were
excluded. At this point, a total of 70% of correct
assignments was reached (Table S4).
In a pairwise comparison, the F-test was used to
check whether the shape of the aedeagus differs sig-
nificantly. Bootstrap pseudoreplicates were performed
with 2500 iterations to minimize the chance that a
larger population was arbitrarily divided into two
groups.
The chance that the tested pairs were actually one
group was given as P = 0 for all pairs except one. The
pair G. rubraria
(SE) and G. rubraria (W) had a
P-value of P = 4.4487
-10
. The greatest distance was
found between G. irakensis and G. rubraria (W) with a
partial Procrustes distance (PPD) = 0.2731 and the
highest F-value = 225.12. The lowest distance was
again found between G. rubraria (SE) and G. rubraria
(W), with a PPD = 0.0576 and the lowest F-value =
6.18. The mean values were PPD
mean
= 0.1611 and
F
mean
= 75.05.
DISCUSSION
Complex taxonomic hypotheses are most reliably
tested by an integrative taxonomy approach. Here we
combine evidence from mitochondrial CO1 sequences,
geometric morphometry, and the results of previous
traditional morphological study (Rajaei Sh et al.,
2012) to confirm the taxonomic status of the species
in Gnopharmia. The results of these three different
approaches are highly congruent.
GNOPHARMIA KASRUNENSIS
The G. kasrunensis clade is monophyletic and
includes two separate intraspecific lineages. One
lineage is represented by three haplotypes from
the same locality (Bashagerd, south Iran; RD1159,
RD1160, and RD1165) and the second one is com-
posed of all other haplotypes of this species. The low
minimum distance (1.67%) and lack of clear morpho-
logical differences between these two lineages are
evidence for their status as one species. All analysed
specimens also nested in the same cluster and
network (Fig. 3). Interestingly, the Bashagerd popu-
lation appears to be isolated and recently differenti-
ated, most likely after the last glacial maximum’s
bottleneck events (Rajaei et al., 2013).
The separation of G. kasrunensis from all other
taxa on the CV axes can be seen in Figure 5B and C.
Additionally, the clade is further supported by a good
performance in the assignment test. Despite minor
morphometric overlaps with G. colchidaria sinesefida
and G. kasrunensis, these sympatric species are
easily delimited based on many discrete characters of
their genitalia (Rajaei Sh et al., 2012). Gnopharmia
kasrunensis occurs in most parts of the Zagros Moun-
tains, which are the central and western mountains
of Iran.
INTEGRATIVE TAXONOMY OF GNOPHARMIA
77
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

78 H. R. Sh ET AL.
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

GNOPHARMIA RUBRARIA
We found two distinct lineages in this well-supported
clade (bootstrap support > 0.96). Whereas the
maximum parsimony network analysis united all
haplotypes of this species into one single network,
distance-based clustering analysis separated them
into two distinct clusters. However, taking the thresh-
old of the barcode gap (2.4%) into account, the
minimum distance between these two clusters (1.7%)
seems insufficient to separate them into two different
species.
The morphometric data present a similar picture.
In the assignment test (Table S3), the two groups
found by distance-based clustering appear distinct
from each other with a few incorrect assignments.
Taking these groups as one cluster, we get a highly
distinctive complex with 100% correct assignments.
This complex can also be identified on the CV axes,
where both groups tend to appear as one group with
high shape similarity and a distinction of both taxa
cannot be confirmed.
It should be kept in mind that there is no geo-
graphical barrier or separation between these two
groups: in several localities in south-eastern Turkey
specimens with both wing patterns can be observed
(see Rajaei Sh et al., 2012) and even gradual transi-
tions can be found. In this context, the colour pattern
shows a transition from orange in Israel, Palestine,
and Syria, to a mix of orange-grey in eastern Turkey,
and grey in Central and western Turkey. Conse-
quently we do not define new subspecies here; instead
we hypothesize the existence of a cline in colour
pattern. According to this hypothesis, G. rubraria is
distributed from Palestine and Israel to south-eastern
Turkey, including its central and western parts.
GNOPHARMIA COCANDARIA
This clade shows a high minimum distance (> 4%) from
all other species. Morphological characters resemble
those of G. kasrunensis with some subtle differences,
but the minimum distance between the two species
(4.6%) confirms their separation at species level. Both
species are also geographically separated from each
other (G. cocandaria in Uzbekistan, Tajikistan, Kyr-
gyzstan, and northern Afghanistan and G. kasrunensis
in central and western parts of the Zagros Mountains
in Iran). The clade G. cocandaria contains two line-
ages. The minimum distance between the two
intraspecific lineages is 0.9%, indicating intraspecific
variation. Beside the lack of morphological distinctive-
ness (also including genital characters), these two
lineages are not geographically separated (Rajaei Sh
et al., 2012). Consequently it is unlikely that these
two lineages represent subspecies. Gnopharmia
cocandaria afghanistana (Wiltshire, 1967) has been
described based only on a male holotype from Afghani-
stan. Rajaei Sh et al. (2012) retained the validity of this
subspecies as an open question; because of the lack of
adequate material and DNA data it could not be
considered in this analysis.
The results of the morphometric analyses show a
group that has a strong tendency to separate from the
rest. However, a continuous overlap can be observed
with G. colchidaria objectaria (see Fig. 5). This is
quite interesting because although G. cocandaria s.l.
represents a morphologically and geographically (see
Fig. 3) close group to G. colchidaria objectaria, they
are still separated by a genetic distance of at least
4%. This fact supports our hypothesis about their
separation.
GNOPHARMIA COLCHIDARIA SENSU LATO
Three subspecies have been defined by morphology so
far for this species (G. colchidaria colchidaria, G. col-
chidaria objectaria, and G. colchidaria sinesefida),
with clearly separated distribution ranges confirmed
(Rajaei Sh et al., 2012).
Haplotype analysis showed that the only complete
CO1 sequence of G. colchidaria colchidaria belongs to
haplotype 2 (the most widespread haplotype in this
analysis, Fig. 3), and nests in the same genetic clade
as G. colchidaria sinesefida and G. colchidaria objec-
taria. The minimum distance between G. colchidaria
colchidaria and G. colchidaria sinesefida is extremely
Figure 5. A, variation within the groups of Gnopharmia. Gnopharmia irakensis differs greatly from the other groups on
the first principal component (PC) axis, which accounts for over 40% of the total variance. B, confirmation of the very
distant position of Gnopharmia irakensis by the canonical variates analysis (CVA). Separation of Gnopharmia rubraria
s.l. from the rest of the groups is on the second CV axis. No overlap of Gnopharmia sarobiana, Gnopharmia colchidaria
colchidaria, Gnopharmia colchidaria objectaria, and Gnopharmia sarobiana with Gnopharmia kasrunensis on the first
axis. C, clear differentiation between Gnopharmia cocandaria and G. sarobiana on the third CV axis. No overlap between
G. colchidaria objectaria and G. sarobiana and between Gnopharmia colchidaria sinesefida and G. cocandaria. Distinction
between G. kasrunensis and G. cocandaria. D, substantial separation of G. colchidaria colchidaria from the rest on the
fourth CV axis. E, no additional information on the fifth CV axis. F, no overlap between G. rubraria (W) and G. colchidaria
colchidaria on the sixth CV axis.
䉳
INTEGRATIVE TAXONOMY OF GNOPHARMIA 79
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

low (0.22%; Table 2). A recent divergence could be
responsible for low divergence of the CO1 sequences
between these two subspecies, despite the fact that
multiple differences in discrete characters of the geni-
talia (see Rajaei Sh et al., 2012) separate them. Con-
sequently we do not reject the subspecies status of
G. colchidaria sinesefida.
Our morphometric analysis shows a clear separa-
tion of G. colchidaria colchidaria on the fourth
CV axes from the other subspecies (Fig. 5D) and
100% correct assignments in the assignment test
(Table S3). These results match with the results of
previous work based on the morphology of genitalia
(Rajaei Sh et al., 2012) and show a high correlation
with Caucasian populations of G. colchidaria col-
chidaria, which represent geographically separated
populations. Based on all these findings, we regard
the other taxa (G. colchidaria objectaria and G. col-
chidaria sinesefida) as subspecifically but not specifi-
cally distinct from the nominate colchidaria, and
hence continue to treat the nominate form as a dis-
tinct subspecies. An alternative hypothesis of the
taxonomic status of the nominate subspecies is that
this taxon may be a well-established species. In this
situation, the lack of resolution at the molecular level
may indicate introgressive hybridization (e.g. Sota
et al., 2001; Cianchi et al., 2003; Pauls et al., 2009).
More molecular data are definitely needed to test this
hypothesis.
In the ML tree, the haplotypes of the morphologi-
cally defined subspecies (G. colchidaria objectaria and
G. colchidaria sinesefida) are not separated. Although
one lineage of G. colchidaria sinesefida is distinct
(RA2157, SMNK339, SMNK378), the minimum dis-
tance between this lineage and all other analysed
specimens of G. cocandaria s.l. is low (1.6%). Further-
more, there are no significant morphological differ-
ences between this lineage and other G. colchidaria
sinesefida specimens. Cluster as well as haplotype
analysis indicated the existence of one single cluster
that includes all specimens.
Gnopharmia colchidaria objectaria and G. col-
chidaria sinesefida can be separated from each other
quite easily based on morphology (Rajaei Sh et al.,
2012). Furthermore, they are geographically sepa-
rated: G. colchidaria objectaria occurs from south
Turkmenistan in north-east Iran to western parts of
Afghanistan and the north of Pakistan, whereas
G. colchidaria sinesefida is distributed in the Zagros
Mountains in the western and southern parts of Iran.
Separation of these two taxa at species level is not
very likely given their low minimum distances (1.1%).
As revealed by our morphometric analyses, both
groups (G. colchidaria objectaria and G. colchidaria
sinesefida) form a complex of two sets of specimens
that share similarities in geometric shape without
losing their own group consistency. Gnopharmia col-
chidaria colchidaria is distinct from both other sub-
species with 100% correct assignments.
GNOPHARMIA IRAKENSIS
The Gnopharmia irakensis clade is monophyletic,
supported by high bootstrap values in the CO1 tree
(0.92). This clade is composed of two major lineages at
a minimum pairwise distance of 0.9%, one repre-
sented by the three haplotypes (SBB01, SBB02
SBB03) from Pakistan and the other by the
specimens from Iran. Nevertheless, morphological
characters show no evidence of this divergence. Fur-
thermore, cluster and haplotype analysis defined
them in a single cluster and network (Fig. 3). Mor-
phometric analyses dramatically corroborate the
species’ distinction. The first CV axis shows a striking
distance from all other species: mean distances
and F-values were the greatest of all tested groups
and finally we get 100% of correct assignments. All
analyses presented G. irakensis as a well-established
and distinct species. The species is distributed sym-
patrically with other congenerics (Rajaei Sh et al.,
2012) extensively throughout the Middle East.
GNOPHARMIA SAROBIANA
As we had no sequence data that was restricted to
Afghanistan for this species (only old material > 15
years was available), we were only able to include it
in the morphometric analyses. In the PCAs (Fig. 5A)
and CVAs (Fig. 5B–D), this taxon was well separated
from other taxa. All other tests supported this result
(e.g. assignment success: 100%, Table S3). Conse-
quently, we accept the results of traditional morpho-
logical studies here but still await DNA data to verify
the phylogenetic placement.
CONCLUSIONS
As part of our comprehensive morphometric, morpho-
logical, and molecular analyses, we have been able to
clarify some important aspects of the phylogeny and
taxonomic status of species in the genus Gnopharmia.
Our analyses confirmed the validity of the following
six species: G. kasrunensis Wehrli, 1939, G. rubraria
Staudinger, 1892, G. cocandaria (Erschoff, 1874),
G. colchidaria (Lederer, 1870) G. sarobiana Ebert,
1965, and G. irakensis Wehrli, 1938. However, there
are still some open questions, e.g. what is the extent
of genetic differentiation between G. sarobiana or
G. erema as well as nominate subspecies of G. col-
chidaria and the other taxa. Furthermore, additional
specimens of G. cocandaria afghanistana need to be
studied in future to define its validity as a separate
80 H. R. Sh ET AL.
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

subspecies. Finally, a comparative study of nuclear
DNA using a multiple gene approach would provide
valuable evidence to resolve the problematic tax-
onomy within this genus.
ACKNOWLEDGEMENTS
We are especially grateful to Paul Hebert (Biodiver-
sity Institute of Ontario, Guelph, Canada) and Axel
Hausmann for providing the main part of the genetic
data from the BOLD database. The DNA studies were
supported by Genome Canada, the Ontario Ministry
of Research, and the Innovation and Natural Science
and Engineering Research Council of Canada
(NSERC) in the framework of the International
Barcode of Life (iBOL) program. We also thank our
colleagues at ZFMK for support with the DNA studies
of this project: Claudia Etzbauer, Oliver Niehuis,
Patrick Kück, Karen Meusemann, and Carola Greve.
We thank Dieter Stüning (Bonn) and Robert Trusch
(Karlsruhe) for their valuable comments. Addition-
ally, we thank Jörg-Uwe Meineke for his generous
help during different steps of this project, and Dirk
Stadie, Jorg Gelbrecht, Bernd Müller, Axel Haus-
mann, Manfred Sommerer, Gergely Petranyi, and
Jörg-Uwe Meineke, who sent us fresh material
of Gnopharmia for DNA extraction. Thanks to Joy
M. Layton (Bonn) for linguistic correction of the
last version of this paper. This study is part of the
PhD project of the first author. H. R. was funded
by the DAAD (Deutscher Akademischer Austausch
Dienst).
REFERENCES
Adams DC, Funk DJ. 1997. Morphometric inferences on
sibling species and sexual dimorphism in Neochlamisus
bebbiabae leaf beetles: multivariate applications of the thin
plate spline. Systematic Biology 46: 180–194.
Ahrens D, Monaghan MT, Vogler AP. 2007. DNA-based
taxonomy for associating adults and larvae in multi-species
assemblages of chafers (Coleoptera: Scarabaeidae). Molecu-
lar Phylogenetics and Evolution 44: 436–449.
Alibert P, Moureau B, Dommergues J-L DB. 2001. Differ-
entiation at a microgeographical scale within two species of
ground beetle, Carabus auronitens and C. nemoralis (Cole-
optera, Carabidae): a geometrical morphometric approach.
Zoologica Scripta 30: 299–311.
Altschul SF, Gish W, Myers WMEW, Lipman DJ. 1990.
Basic local alignment search tool. Journal of Molecular
Biology 215: 403–410.
Astrin JJ, Stüben PE, Misof B, Wägele JW, Gimnich F,
Raupach MJ, Ahrens D. 2012. Exploring diversity
in cryptorhynchine weevils (Coleoptera) using distance-,
character- and tree-based species delineation. Molecular
Phylogenetics and Evolution 63: 1–14.
Bichain JM, Boisselier-Dubayle MC, Bouchet P, Samadi
S. 2007. Species delimitation in the genus Bythinella (Mol-
lusca: Caenogastropoda: Rissooidea): a first attempt com-
bining molecular and morphometrical data. Malacologia 49:
293–311.
Bookstein FL. 1997. Morphometric tools for landmark data:
geometry and biology. Cambridge: Cambridge University
Press.
Campbell NA, Atchley WR. 1981. The geometry of canonical
variates analysis. Systematic Zoology 30: 268–280.
Chatfield C, Collins AJ. 1980. Introduction to multivariate
analysis. London & New York: Chapman & Hall.
Cianchi R, Ungano A, Marini M, Bullini L. 2003. Differ-
ential patterns of hybridization and introgression between
the swallowtails Papillo machaon and P. hospiton from
Sardinia and Corsica islands (Lepidoptera, Papilionidae).
Molecular Ecology 12: 1461–1471.
Clare EL, Lim BK, Engstrom MD, Eger JL, Hebert PDN.
2007. DNA barcoding of neotropical bats: species identifica-
tion and discovery within Guyana. Molecular Ecology Notes
7: 184–190.
Clement M, Posada D, Crandall K. 2000. TCS: a computer
program to estimate gene genealogies. Molecular Ecology 9:
1657–1660.
Cornuet JM, Piry S, Luikart G, Estoup A, Sollignac M.
1999. New methods employing multilocus genotypes to
select or exclude populations as origins of individuals.
Genetics 153: 1989–2000.
Eberhard WG. 1985. Sexual selection and animal genitalia.
Cambridge: Harvard University Press.
Ebert G. 1965. Afghanische Geometriden Lep.) I. Stuttgarter
Beiträge zur Naturkunde. 142: 1–32.
Felsenstein J. 1985. Confidence limits on phylogenies: an
approach using the bootstrap. Evolution 39: 783–791.
Fisher RA. 1925. Statistical Methods for Research Workers.
Edinburgh: Oliver and Boyd.
Hahn C, Bakke TA, Bachmann L, Weiss S, Harris PD.
2011. Morphometric and molecular characterization of
Gyrodatylus teuchis Lautraite, Blanc, Thiery, Daniel &
Vigneulle 1999 (Monogenea: Gyrodactylidae) from an Aus-
trian brown trout population. Parasitology International 60:
480–487.
Hajibabaei M, Janzen DH, Burns JM, Hallwachs W,
Hebert PDN. 2006. DNA barcodes distinguish species of
tropical Lepidoptera. Proceedings of the National Academy
of Sciences, USA 103: 968–971.
Hall TA. 1999. BioEdit: a user-friendly biological sequence
alignment editor and analysis program for Windows 95/98/
NT. Nucleic Acids Symposium Series 41: 95–98.
Hausmann A, Haszprunar G, Hebert PDN. 2011.
DNA barcoding the geometrid fauna of Bavaria Lepidop-
tera): successes, surprises, and questions. PLoS ONE 6:
e17134.
Hebert PDN, deWaard JR, Landry JF. 2010. DNA bar-
codes for 1/1000 of the animal kingdom. Biology Letters 6:
359–362.
Hendrich L, Pons J, Ribera I, Balke M. 2010. Mitochon-
drial Cox1 sequence data reliably uncover patterns of
INTEGRATIVE TAXONOMY OF GNOPHARMIA 81
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

insect diversity but suffer from high lineage-idiosyncratic
error rates. PLoS ONE 5: e14448. doi:10.1371/
journal.pone.0014448.
Ivanova NV, deWaard JR, Hebert PDN. 2006. An inex-
pensive, automation-friendly protocol for recovering high-
quality DNA. Molecular Ecology Notes 6: 998–1002.
Katoh K, Misawa K, Kuma K, Miyata T. 2002. MAFFT: a
novel method for rapid multiple sequence alignment based
on fast Fourier transform. Nucleic Acids Research 30: 3059–
3066.
Klingenberg CP. 2009. Morphometric integration and modu-
larity in configurations of landmarks: tools for evaluation
a priori hypothesis. Evolution and Development 11: 405–421.
Lakra WS, Verma MS, Goswami M, Lal KK, Mohindra V,
Punia P, Gopalakrishan A, Singh KV, Ward RD,
Hebert PDN. 2011. DNA barcoding Indian marine fishes.
Molecular Ecology 11: 60–71.
Lederer J. 1870. Contributions a La Faune des Lépidoptères
de la Transcaucasie. Annales de la Société Entomologique de
Belgique 13: 39–48, pl. 2, fig. 1.
Mahalanobis PC. 1936. On the generalised distance in sta-
tistics. Proceedings of the National Institute of Sciences of
India 2: 49–55.
Mardia KV, Kent JT, Bibby JM. 1979. Multivariate analy-
sis. London: Academic Press.
Meier R, Shiyang K, Vaidya G, Peter KLNG. 2006. DNA
barcoding and taxonomy in Diptera: a tale of high intraspe-
cific variability and low identification success. Systematic
Biology 55: 715–728.
Monaghan MT, Wild R, Elliot M, Fujisaea T, Balkei M,
Inwardi DJG, Lees DC, Ranaivosolo R, Eggleton P,
Barraclough TG, Vogler AP. 2009. Accelerated species
inventory on Madagascar using coalescent-based models of
species delineation. Systematic Biology 58: 298–311.
Morrison DF. 1990. Multivariate statistical methods, 3rd
edn. New York: McGraw-Hill.
Nolte AW, Sheets HD. 2005. Shape based assignment
tests suggest transgressive phenotypes in natural sculpin
hybrids. Frontiers in Zoology 2: 11.
Padial JM, Miralles A, De la Riva I, Vences M. 2010. The
integrative future of taxonomy. Frontiers in Zoology 7: 16.
Pauls SU, Theissinger K, Ujvarosi L, Balint M, Hasse P.
2009. Patterns of population structure in two closely
related, partially sympatric caddisflies in Eastern Europe:
historic introgression, limited dispersal, and cryptic diver-
sity. Journal of the North American Benthological Society
28: 517–536.
Pons J, Barraclough TG, Jesus Gomez-zurita TG,
Cardoso A, Duran DP, Hazell S, Kamoun S, Sumlin
WD, Vogler AP. 2006. Sequence-based species delimitation
for the DNA taxonomy of undescribed insects. Systematic
Biology 55: 595–609.
Prout LB. 1912–1916. Die spannerartigen Nachtfalter. In:
Seitz A, ed. Die Gross-Schmetterlinge der Erde 4. Stuttgart:
Alfred Kernen Verlag, i–v, 1–479, pls. 1–25.
Puillandre N, Lambert A, Brouillet S, Achaz G. 2012.
ABGD, Automatic Barcode Gap Discovery for primary
species delimitation. Molecular Ecology 21: 1864–1877.
Quenouille MH. 1956. Notes on bias in estimation.
Biometrika 43: 353–360.
Rajaei Sh H, Rödder D, Weigand AM, Dambach J,
Raupach M, Wägele JW. 2013. Quaternary refugia in
southwestern Iran: Insights from two sympatric moth
species and independent lines of evidence. Organisms Diver-
sity & Evolution, doi: 10.1007/s13127-013-0126-6.
Rajaei Sh H, Stüning D, Trusch R. 2012. Taxonomic
revision and zoogeographical patterns of the species of
Gnopharmia Staudinger, 1892 (Geometridae, Ennominae).
Zootaxa 3360: 1–52.
Raupach MJ, Astrin J, Hannig K, Peters M, Stoeckle
MY, Wägele JW. 2010. Molecular species identifications of
Central European ground beetles (Coleoptera: Carabidae)
using nuclear rDNA expansion segments and DNA bar-
codes. Frontiers in Zoology 7: 1–15.
von Reumont BM, Struwe JF, Schwarzer J, Misof B.
2012. Phylogeography of the burnet moth Zygaena trans-
alpina complex: molecular and morphometric differentia-
tion suggests glacial refugia in Southern France, Western
France and micro-refugia within the Alps. Journal of Zoo-
logical Systematics and Evolutionary Research 50: 38–50.
Rivera J, Currie DC. 2009. Identification of Nearctic black
flies using DNA barcodes (Diptera: Simuliidae). Molecular
Ecology Resources 9(Suppl. 1): 224–236.
Robinson EA, Blagoev GA, Hebert PDN, Adamowicz SJ.
2009. Prospects for using DNA barcoding to identify spiders
in species-rich genera. ZooKeys 16: 27–46.
Rohlf FJ. 1990. Rotational fit Procrustes methods. In Rahlf
FJ, Bookstein FL, (eds.) Proceedings of the Michigan
morphometrics workshop, University of Michigan Museum
of Zoology Special Publication 2. Ann Arbor: University of
Michigan Museum of Zoology, 227–236.
Rohlf FJ. 1993. Relative warp analysis and examples of
its application to mosquito wings. In: Marcus LF, Bello I,
García-Valdescasa A, eds. Contributions to morphometrics.
Madrid: Museo Nacional de Ciencias Naturales, 131–159.
Rohlf FJ. 2004a. TpsDig. Stony Brook, NY: Department of
Ecology and Evolution, State University of New York.
Rohlf FJ. 2004b. TpsUtil. Stony Brook, NY: Department of
Ecology and Evolution, State University of New York.
Rohlf FJ, Marcus LF. 1993. A revolution in morphometrics.
Trends in Ecology and Evolution 8: 129–132.
Rohlf FJ, Slice D. 1990. Extensions of the Procrustes
method for the optimal superimposition of landmarks. Sys-
tematic Zoology 39: 40–59.
Schlick-Steiner BC, Steiner FM, Moder K, Seifert B,
Sanetra M, Dyreson E, Stauffer C, Christian E. 2006.
A multidisciplinary approach reveals cryptic diversity in
western Palearctic Tetramorium ants (Hymenoptera: Formi-
cidae). Molecular Phylogenetics and Evolution 40: 259273.
Schlick-Steiner BC, Steiner F, Seifert M, Stauffer B,
Erhard C, Crozier RH. 2010. Integrative taxonomy: a
multisource approach to exploring biodiversity.
Annual
Review of Entomology 55: 421–438.
Shapiro AM, Porter AH. 1989. The lock-and-key hypothesis
– evolutionary and biosystematic interpretation of insect
genitalia. Annual Review of Entomology 34: 231–245.
82 H. R. Sh ET AL.
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83

Sheets D. 2002. Integrated morphometrics package IMP.
Buffalo, NY: Department of Physics, Canisius College. Avail-
able at: http://www2.canisius.edu/~sheets/morphsoft.html
da Silva JM, Creer S, dos Santos A, Costa AC, Cunha
MR, Costa FO, Carvalho GR. 2011. Systematic and evo-
lutionary insight derived from mtDNA CO1 barcode diver-
sity in the Decapoda Crustacea: Malacostraca). PLoS ONE
6: e19449.
Smith MA, Fisher BL. 2009. Invasions, DNA barcodes, and
rapid biodiversity assessment using ants of Mauritius.
Frontiers in Zoology 6: 31.
Snedecor GW, Cochran WG. 1967. Statistical methods, 6th
edn. Ames: Iowa State University Press.
Sota T, Ishikawa R, Ujiie M, Kusumoto F, Vogler AP.
2001. Extensive trans-species mitochondrial polymorphisms
in the carabid beetles Carabus subgenus Ohomopterus
cause by repeated introgressive hybridization. Molecular
Ecology 10: 2833–2847.
Staudinger O. 1892. Neue Arten u Varietäten von paläark-
tischen Geometriden. Deutsche entomologische Zeitschrift
Iris 5: 181–188.
Staudinger O, Rebel H. 1901. Catalog der Lepidopteren
des Palaearctischen Faunengebietes, 3rd edn. Berlin: R.
Friedländer & Sohn.
Stoeckle MY, Hebert PDN. 2008. Barcode of life. Scientific
American 299: 82–88.
Tamura K, Peterson D, Peterson N, Stecher G, Nei M,
Kumar S. 2011. MEGA5: molecular evolutionary genetics
analysis using maximum likelihood, evolutionary distance,
and maximum parsimony methods. Molecular Biology and
Evolution 28: 2731–2739.
Toth JP, Varga Z. 2010. Morphometric study on the genitalia
of sibling species Melitaea phoebe and M. telona (Lepidop-
tera: Nymphalidae). Acta Zoologica Acadmiae Scientiarum
Hungaricae 56: 273–282.
Toth JP, Varga Z. 2011. Inter- and intraspecific variation in
the genitalia of the ‘Melitaea phoebe group’ (Lepidoptera,
Nymphalidae). Zoologischer Anzeiger 20: 258–268.
Tukey J. 1958. Bias and confidence in not quite large
samples. Annals of Mathematical Statistics 29: 614.
Valentini A, Pompanon F, Taberlet P. 2008. DNA
barcoding for ecologists. Trends in Ecology and Evolution
24: 110–117.
Villegas J, Feliciangeli MD, Dujardin JP. 2002.
Wing shape divergence between Rhodnius prolixus
from Cojedes (Venezuela) and Rhodnius robustus from
Mérida (Venezuela). Infection, Genetics and Evolution 2:
121–128.
Vogler AP, Monaghan MT. 2007. Recent advances in DNA
taxonomy. Journal of Zoological Systematics and Evolution-
ary Research 45: 1–10.
Wakeham-Dawson A, Jaksic P, Holloway JD, Dennis
RLH. 2004. Multivariate analysis of male genital structures
in the Hipparchia semele-muelleri-delattini complex (Nym-
phalidae, Satyrinae) from the Balkans: how many taxa?
Nota lepidopterologica 27: 103–124.
Wehrli E. 1938. Nachtrag zu meiner Arbeit in Nr. 31 d. J.
Seite 354 [Neue Gattungen, Untergattungen, Arten und
Rassen von Geometriden Lep.)]. Entomologische Rundschau
55: 433–434.
Wehrli E. 1939. Einige neue Arten und Rassen aus dem
südwestlichen Iran und aus dem Irak, sowie ein neues
Subgenus aus dem letzteren. Lepid. Geometr.). Mitteilungen
der Münchner Entomologischen Gesellschaft 29: 70–71.
Wehrli E. 1939-1954. 6. Subfamilie: Geometrinae. In: Seitz A,
ed. Die Gross-Schmetterlinge der Erde 4, Die Spanner des
palaearktischen Faunengebietes (Suppl.). Stuttgart: Alfred
Kernen Verlag, 254–722, pls 19–53.
Wei SJ, Shi M, Chen XX, Sharkey MJ, van Achterberg C,
Ye GY, He JH. 2010. New views on strand asymmetry in
insect mitochondrial genomes. PLoS ONE 5: e12708.
Wiltshire EP. 1967. Middle East Lepidoptera XX, a third
contribution to the Fauna of Afghanistan. Beiträge zur
naturkundlichen Forschung in Südwestdeutschland 263:
137–169.
Wiltshire EP. 1970. Österreichische entomologische
Expeditionen nach Persien und Afghanistan. Annalen des
Naturhistorischen Museums in Wien 74: 387–398.
Zelditch ML, Swiderski DL, Sheets HD, Fink WL. 2004.
Geometric morphometrics for biologists. New York: Elsevier
Academic Press.
Zhou X, Adamowicz SJ, Jacobus LM, Dewalt RE,
Hebert PDN. 2009. Towards a comprehensive barcode
library for arctic life – Ephemeroptera, Plecoptera, and
Trichoptera of Churchill, Manitoba, Canada. Frontiers in
Zoology 6: 30.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher’s web-site:
Table S1. List of sequenced specimens, with identification, sampling sites, collecting data, accession numbers,
and process ID in the BOLD database.
Table S2. Results from canonical variates analysis (CVA)/MANOVA.
Table S3. The canonical variates analysis (CVA) assignment test shows a high number of correct assignments.
Table S4. Results from the jack-knife assignment test: the discriminatory power of canonical variates analysis
axes is high.
INTEGRATIVE TAXONOMY OF GNOPHARMIA
83
© 2013 The Linnean Society of London, Zoological Journal of the Linnean Society, 2013, 169, 70–83