ArticlePDF Available

Establishing criteria for higher-level classification using molecular data: The systematics of Polyommatus blue butterflies (Lepidoptera, Lycaenidae)

Authors:

Abstract and Figures

Most taxonomists agree on the need to adapt current classifications to recognize monophyletic units. However, delineations between higher taxonomic units can be based on the relative ages of different lineages and⁄or the level of morphological differentiation. In this paper, we address these issues in considering the species-rich Polyommatus section, a group of butterflies whose taxonomy has been highly controversial. We propose a taxonomy-friendly, flexible temporal scheme for higher-level classification. Using molecular data from nine markers (6666 bp) for 104 representatives of the Polyommatus section, representing all but two of the 81 described genera ⁄ subgenera and five outgroups, we obtained a complete and well resolved phylogeny for this clade. We use this to revise the systematics of the Polyommatus blues, and to define criteria that best accommodate the described genera within a phylogenetic framework. First, we normalize the concept of section (Polyommatus) and propose the use of subtribe (Polyommatina) instead. To preserve taxonomic stability and traditionally recognized taxa, we designate an age interval (4–5 Myr) instead of a fixed minimum age to define genera. The application of these criteria results in the retention of 31 genera of the 81 formally described generic names, and necessitates the description of one new genus (Rueckbeilia gen. nov.). We note that while classifications should be based on phylogenetic data, applying a rigid universal scheme is rarely feasible. Ideally, taxon age limits should be applied according to the particularities and pre-existing taxonomy of each group. We demonstrate that the concept of a morphological gap may be misleading at the genus level and can produce polyphyletic genera, and we propose that recognition of the existence of cryptic genera may be useful in taxonomy.
Content may be subject to copyright.
Establishing criteria for higher-level classification using molecular
data: the systematics of Polyommatus blue butterflies (Lepidoptera,
Lycaenidae)
Gerard Talavera
a,b
, Vladimir A. Lukhtanov
c,d
, Naomi E. Pierce
e
and Roger Vila
a,
*
a
Institut de Biologia Evolutiva (CSIC-UPF), Passeig Marı´tim de la Barceloneta, 37, 08003 Barcelona, Spain;
b
Departament de Gene
`tica
i Microbiologia, Universitat Auto
`noma de Barcelona, 08193 Bellaterra (Barcelona), Spain;
c
Department of Karyosystematics, Zoological Institute of
Russian Academy of Science, Universitetskaya nab. 1, 199034 St Petersburg, Russia;
d
Department of Entomology, St Petersburg State University,
Universitetskaya nab. 7 9, 199034 St Petersburg, Russia;
e
Department of Organismic and Evolutionary Biology and Museum of Comparative Zoology,
Harvard University, 26 Oxford Street, Cambridge, MA 02138, USA
Accepted 11 June 2012
Abstract
Most taxonomists agree on the need to adapt current classifications to recognize monophyletic units. However, delineations
between higher taxonomic units can be based on the relative ages of different lineages and or the level of morphological
differentiation. In this paper, we address these issues in considering the species-rich Polyommatus section, a group of butterflies
whose taxonomy has been highly controversial. We propose a taxonomy-friendly, flexible temporal scheme for higher-level
classification. Using molecular data from nine markers (6666 bp) for 104 representatives of the Polyommatus section, representing
all but two of the 81 described genera subgenera and five outgroups, we obtained a complete and well resolved phylogeny for this
clade. We use this to revise the systematics of the Polyommatus blues, and to define criteria that best accommodate the described
genera within a phylogenetic framework. First, we normalize the concept of section (Polyommatus) and propose the use of subtribe
(Polyommatina) instead. To preserve taxonomic stability and traditionally recognized taxa, we designate an age interval (4–5 Myr)
instead of a fixed minimum age to define genera. The application of these criteria results in the retention of 31 genera of the 81
formally described generic names, and necessitates the description of one new genus (Rueckbeilia gen. nov.). We note that while
classifications should be based on phylogenetic data, applying a rigid universal scheme is rarely feasible. Ideally, taxon age limits
should be applied according to the particularities and pre-existing taxonomy of each group. We demonstrate that the concept of a
morphological gap may be misleading at the genus level and can produce polyphyletic genera, and we propose that recognition of
the existence of cryptic genera may be useful in taxonomy.
The Willi Hennig Society 2012
Despite current progress in morphological and molec-
ular studies of ‘‘Blue’’ butterflies, subfamily Polyom-
matinae (Forster, 1936, 1938; Stempffer, 1937,
Stempffer, 1967; Nabokov, 1945; Eliot, 1973; Als et al.,
2004; Zhdanko, 2004; Stekolnikov and Kuznetzov,
2005; Wiemers et al., 2009; Stekolnikov, 2010), their
higher-level systematics remain controversial. Eliot
(1973) divided this subfamily into four tribes: Lycae-
nesthini, Candalidini, Niphandini and Polyommatini
(Table 1). Among these tribes, the Polyommatini is the
most diverse and arguably one of the most systemati-
cally difficult groups of butterflies, as stated by Eliot
himself: ‘‘I have to admit complete failure in my efforts
to subdivide it into natural groups, simply organizing it
into 30 sections’’ (Eliot, 1973). His division of Poly-
ommatini into sections has nevertheless been widely
accepted by the scientific community (Hirowatari, 1992;
Mattoni and Fiedler, 1993; Ba
´lint and Johnson, 1994,
1995, 1997; Io, 1998; Pratt et al., 2006; Robbins and
Duarte, 2006). Some entomologists prefer considering
*Corresponding author.
E-mail address: roger.vila@ibe.upf-csic.es
The Willi Hennig Society 2012
Cladistics 29 (2013) 166–192
Cladistics
10.1111/j.1096-0031.2012.00421.x
Table 1
Polyommatinae classification according to Eliot (1973)
Tribe Section Genera
Lycaenesthini Lycaenesthes Moore, 1866; Anthene Doubleday, 1847;
Cupidesthes Aurivillius,
1895; Neurypexina Bethune-Baker, 1910;
Neurellipes Bethune-Baker 1910;
Monile Ungemach, 1932; Triclema Karsch, 1893
Candalidini Candalides Hu
¨bner, 1819; Erina Swainson, 1833
(= Holochila C. Felder, 1862);
Cyprotides Tite, 1963; Microscena Tite, 1963;
Adaluma Tindale, 1922; Nesolycaena
Waterhouse & Lyell, 1905; Zetona Waterhouse, 1938;
Holochila sensu auctt. nec C. Felder
Niphandini Niphanda Moore, 1875
Polyommatini
Cupidopsis Cupidopsis Karsch, 1895
Una Una de Nice
´ville, 1890; Orthomiella de Nice
´ville, 1890
Petrelaea Petrelaea Toxopeus, 1929; Pseudonacaduba
Stempffer, 1943
Nacaduba Nacaduba Moore, 1881; Prosotas H. H. Druce, 1891;
Ionolyce Toxopeus, 1929;
Catopyrops Toxopeus, 1929; Erysichton Fruhstorfer,
1916; Paraduba Bethune-Baker,
1906; Neolucia Waterhouse & Turner, 1905; Hypojamides
Riley, 1929
Theclinesthes Theclinesthes Ro
¨ber, 1891;
Thaumaina Bethune-Baker, 1908;
Utica Hewitson, 1865, invalid, praeocc.
Upolampes Upolampes Bethune-Baker, 1908; Caleta Fruhstorfer,
1922; Pycnophallium Toxopeus, 1929;
Discolampa Toxopeus, 1929 (= Ethion Shirozu &
Saigusa, 1962); Pistoria Hemming, 1964
(= Mambara Bethune-Baker, 1908, praeocc.)
Danis Danis Fabricius, 1807 (= Thysonotis Hu
¨bner, 1819;
Hadothera Billberg, 1820; Damis Boisduval,
1832); Psychonotis Toxopeus, 1930; Epimastidia
H. H. Druce, 1891
Jamides Jamides Hu
¨bner, 1819; Pepliphorus Hu
¨bner, 1819
(= Peplodyta Toxopeus, 1929)
Catochrysops Catochrysops Boisduval, 1832; Rysops Eliot, 1973
Lampides Lampides Hu
¨bner, 1819 (= Cosmolyce Toxopeus, 1927;
Lampidella Hemming, 1933)
Callictita Callictita Bethune-Baker, 1908
Uranothauma Uranothauma Butler, 1895
Phlyaria Phylaria Karsch, 1895
Cacyreus Cacyreus Butler, 1898 (= Hyreus Hu
¨bner, 1819, praeocc.);
Harpendyreus Heron, 1909
Leptotes Leptotes Scudder, 1876; Syntarucoides
Kaye, 1904; Cyclyrius Butler, 1897; Syntarucus Butler,
1900 (= Langia Tutt, 1906, praeocc.)
Castalius Castalius Hu
¨bner, 1819; Tarucus Moore, 1881
Zintha Zintha Eliot, 1973
Zizeeria Zizeeria Chapman, 1910; Zizina Chapman, 1910;
Pseudozizeeria Beuret, 1955
Famegana Famegana Eliot, 1973
Actizera Actizera Chapman, 1910
Zizula Zizula Chapman, 1910
Brephidium Brephidium Scudder, 1876; Oraidium Bethune-Baker, 1914
Everes Everes Hu
¨bner, 1819 (= Ununcula van Eecke, 1915);
Cupido Schrank, 1801 (= Zizera
Moore, 1881); Tiora Evans, 1912; Bothrinia
Chapman, 1909 (= Bothria Chapman, 1908,
praeocc.); Tongeia Tutt, 1908; Shijimia Matsumura,
1919; Talicada Moore, 1881; Binghamia Tutt, 1908
167G. Talavera et al. / Cladistics 29 (2013) 166–192
these sections, including the Polyommatus section, as
tribes (Higgins, 1975; Zhdanko, 1983). Thus the Poly-
ommatus section sensu Eliot, 1973 is equivalent to
Polyommatini sensu Higgins, 1975.
The Polyommatus section is the most species-rich
group within the blue butterflies, including about 460
species. It is generally cosmopolitan, but with most
genera and species restricted to the Palearctic, Neotrop-
ical and Nearctic regions. Of a total of ca. 340–350
Palearctic species, ca. 130 belong to the monophyletic
Agrodiaetus. About 20 species occur in North America
(Opler and Warren, 2004) and at least 91 in the
Neotropics (Lamas, 2004). Explosive chromosome evo-
lution has evolved independently in at least three
separate lineages, Agrodaietus,Lysandra and Plebicula
(Kandul et al., 2004). Some lineages (e.g. Polyommatus
Table 1
(Contiuned)
Tribe Section Genera
Pithecops Pithecops Horsfield, 1828; Eupsychellus Ro
¨ber, 1891
Azanus Azanus Moore, 1881
Eicochrysops Eicochrysops Bethune-Baker, 1924
Lycaenopsis Lycaenopsis C. & R. Felder, 1865; Neopithecops Distant, 1884; Parapithecops Moore, 1884;
Megisba Moore, 1881; Pathalia
Moore, 1884; Arletta Hemming, 1935 (= Moorea
Toxopeus, 1927, praeocc.); Celastrina
Tutt, 1906; Notarthrinus Chapman, 1908; Acytolepis
Toxopeus, 1927; Oreolyce Toxopeus, 1927;
Monodontides Toxopeus, 1927; Akasinula Toxopeus, 1928; Ptox Toxopeus,
1928; Udara Toxopeus, 1928; Rhinelephas Toxopeus, 1928;
Uranobothria Toxopeus, 1928; Parelodina Bethune-Baker,
1904; Vaga Zimmerman, 1958;
Papua Ro
¨ber, 1892, invalid, praeocc.; Cyanirioides
Matsumura, 1919, invalid, praeocc.
Glaucopsyche Glaucopsyche Scudder, 1872; Phaedrotes Scudder, 1876;
Scolitantides Hu
¨bner, 1819;
Apelles Hemming, 1931; Philotes Scudder, 1876; Turanana Bethune-Baker, 1916
(= Turania Bethune-Baker, 1914, praeocc.);
Palaeophilotes Forster, 1938;
Praephilotes Forster, 1938; Pseudophilotes Beuret, 1955; Shijimiaeoides Beuret, 1955;
Sinia Forster, 1949; Iolana Bethune-Baker, 1914; Maculinea van Eecke, 1915; Caerulea
Forster, 1938; Phengaris Doherty, 1881
Euchrysops Euchrysops Butler, 1900; Lepidochrysops Hedicke, 1923
(= Neochrysops Bethune-Baker,
1923, praeocc.); Thermoniphas Karsch, 1895; Oboronia
Karsch, 1893; Athysanota Karsch, 1895
Polyommatus Polyommatus Latreille, 1804; Plebejus Kluk, 1802;
Lycaeides Hu
¨bner, 1819; Cyaniris
Dalman, 1816; Nomiades Hu
¨bner, 1819; Aricia
R. L., 1817 (= Gynomorphia Verity, 1929);
Pseudoaricia Beuret, 1959; Kretania Beuret, 1959;
Ultraaricia Beuret, 1959; Agriades Hu
¨bner,
1819; Vacciniina Tutt, 1909; Albulina Tutt, 1909;
Bryna Evans, 1912; Meleageria Sagarra, 1925;
Agrodiaetus Hu
¨bner, 1822 (= Hirsutina Tutt, 1909);
Lysandra Hemming, 1933 (= Uranops
Hemming, 1929, praeocc.); Plebicula Higgins, 1969;
Eumedonia Forster, 1938; Plebulina
Nabokov, 1944; Icaricia Nabokov, 1944; Chilades
Moore, 1881; Edales Swinhoe, 1910;
Luthrodes H. H. Druce, 1895; Freyeria Courvoisier,
1920; Hemiargus Hu
¨bner, 1818;
Itylos Draudt, 1921; Pseudochrysops Nabokov,
1945; Cyclargus Nabokov, 1945;
Echinargus Nabokov, 1945; Pseudolucia Nabokov,
1945; Paralycaeides Nabokov, 1945;
Nabokovia Hemming, 1960 (= Pseudothecla Nabokov,
1945; praeocc.); Parachilades
Nabokov, 1945
168 G. Talavera et al. / Cladistics 29 (2013) 166–192
s.s. and Agrodiaetus) have extremely high rates of
diversification, resulting in numerous species in these
lineages despite their young age (Kandul et al., 2004,
2007). In fact, Agrodiaetus displays one of the highest
known diversification rates in the animal kingdom
(Coyne and Orr, 2004). Homoploid hybrid speciation
(considered to be rare in animals) has been hypothesized
in the genus Plebejus (Gompert et al., 2006). The group
displays an interesting pattern of wing colour evolution,
including multiple independent cases of discoloration, a
change in colour from blue to brown (Ba
´lint and
Johnson, 1997) and rapid colour changes that may
reflect reinforcement (Lukhtanov et al., 2005) or eco-
logical adaptation (Biro et al., 2003). Studies of the
biology of these butterflies have focused on evolutionary
processes (Krauss et al., 2004; Lukhtanov et al., 2005;
Gompert et al., 2006; Kuhne and Schmitt, 2010; Lu-
khtanov, 2010), ecology (Vandewoestijne et al., 2008;
Rusterholz and Erhardt, 2000), biogeography (Mensi
et al., 1988; Schmitt et al., 2003; Schmitt, 2007; Vila
et al., 2011), conservation (Brereton et al., 2008; Vila
et al., 2010), cytogenetics (White, 1973; Lukhtanov and
Dantchenko, 2002; Kandul et al., 2007; Vershinina and
Lukhtanov, 2010), ecological physiology (Goverde
et al., 2008), physiology and genetics of colour vision
(Sison-Mangus et al., 2008), climate change (Carroll
et al., 2009) and symbiosis (Pierce et al., 2002; Trager
and Daniels, 2009).
A robust phylogenetic framework is fundamental for
the advancement of these fields of research. Several
modifications have been suggested to the tentative
classification proposed by Eliot in 1973 (Ba
´lint and
Johnson, 1997; Zhdanko, 2004; Stekolnikov, 2010), but
no comprehensive revision has been published so far.
The systematics of the section are especially prob-
lematic at the genus level. As many as 81 genera have
been described within the section, but their morpho-
logical delineations are generally unclear and a wide
array of taxonomic combinations are currently in use.
Two extreme approaches exist: lumpers and splitters.
The lumpers include the maximum number of species
in one or a few genera. Examples include the mono-
graphs by Scott (1986) and by Gorbunov (2001), where
nearly all the Holarctic species of the Polyommatus
section are lumped into a single large genus (Plebejus).
Splitters recognize numerous genera, with a genus
described for every small species group. This approach
has been a common practice for the Polyommatus
section since the work of Forster (1938). The main
consequence of the taxonomy of both lumpers and
splitters is the same in one respect: they generate
unstructured and uninformative classifications that do
not reflect evolutionary relationships between the
members of the section.
For example, some researchers divided the Holarctic
species into three genera: Chilades,Plebejus and Poly-
ommatus (Zhdanko, 1983; Hesselbarth et al., 1995),
whereas others opted for four: Chilades,Plebejus,Aricia
and Polyommatus (Kudrna, 2002). This created confu-
sion as taxa of the Aricia,Eumedonia,Albulina,Agriades
and Vacciniina species groups are sometimes included
within the genus Plebejus (Hesselbarth et al., 1995) and
sometimes within the genus Polyommatus (Zhdanko,
1983).
In the past 10 years, several molecular phylogenies
have been published that focused on particular genera
within the Polyommatus section (e.g. Agrodia-
etus—Wiemers, 2003; Kandul et al., 2004, 2007; Vila
et al., 2010; Polyommatus—Wiemers et al., 2010), or
on more general issues such as biogeography and
evolution (Schmitt et al., 2003; Krauss et al., 2004;
Kuhne and Schmitt, 2010; Vila et al., 2011) and DNA
barcoding (Wiemers and Fiedler, 2007; Lukhtanov
et al., 2009). These studies were based on the analysis
of limited numbers of molecular markers and most
did not contain a representative collection of all the
taxa of the Polyommatus section. Nevertheless,
together these studies showed that most genera are
young and closely related, explaining the controversial
systematics of the group.
A recent seven-marker phylogeny was the first
detailed hypothesis published for relationships in the
Polyommatus section (Vila et al., 2011) with special
attention to New World taxa. This study revealed that
all the Neotropical genera—Pseudolucia,Nabokovia,
Eldoradina,Itylos,Paralycaeides,Hemiargus,Echinar-
gus,Cyclargus and Pseudochrysops—together formed a
well supported monophyletic clade that is sister to the
Old World and Nearctic taxa. The analyses showed that
all Neotropical taxa belong to the Polyommatus section,
thus the hypothesis that the Neotropical group is
polyphyletic and that several taxa belong to other
sections (Ba
´lint and Johnson, 1994, 1995, 1997) was not
supported. Vila et al. (2011) also determined that the
Everes section is sister to the Polyommatus section.
However, this study did not include a complete sampling
for the Old World taxa.
We address here the analysis of phylogenetic rela-
tionships among worldwide taxa of the Polyommatus
section. We use a combination of three mitochondrial
genes and six nuclear markers to infer phylogenetic
relationships between representatives of nearly all gen-
era, subgenera and distinct species groups described
within the section. We discuss principles of taxonomic
classification above the species level (subgenus, genus,
section and subtribe) and propose explicit criteria for
defining genera in this group. We review the importance
of molecular versus morphological data in evaluating
our systematic hypothesis, and propose that the recog-
nition of ‘‘cryptic genera’’ may be a useful concept in
taxonomy. Finally, we rearrange the classification of the
Polyommatus section and propose a new list of genera.
169G. Talavera et al. / Cladistics 29 (2013) 166–192
Materials and methods
Taxon sampling
We used 104 representatives of the Polyommatus
section, including at least one representative of each
described genus subgenus for all but two genera that we
were unable to collect (Xinjiangia Huang & Murayama,
1988 and Grumiana Zhdanko, 2004). Four representa-
tives for the Everes and one for the Leptotes sections
were used as outgroups. All specimens used in this study
are listed in Table 2. The samples (bodies in ethanol and
wings in glassine envelopes) are stored in the DNA and
Tissues Collection of the Museum of Comparative
Zoology (Harvard University, Cambridge, MA, USA).
DNA extraction and sequencing
Genomic DNA was extracted from a leg or from a
piece of the abdomen of each specimen using the
DNeasy
TM
Tissue Kit (Qiagen Inc., Valencia, CA,
USA) and following the manufacturerÕs protocols.
Fragments from three mitochondrial genes—cytochrome
oxidase I (COI)+leu-tRNA +cytochrome oxidase II
(COII); and from six nuclear markers—elongation
factor-1 alpha (EF-1a), 28S ribosome unit (28S), histone
H3 (H3), wingless (Wg), carbamoyl-phosphate synthetase
2aspartate transcarbamylase dihydroorotase (CAD) and
internal transcribed spacer 2 (ITS2) were amplified by
polymerase chain reaction and sequenced as described in
Vila et al. (2011). The primers employed are shown in
Table S1 (Appendix S1). The sequences obtained were
submitted to GenBank under accession numbers
JX093196–JX093497 (Table S2, Appendix S1).
Alignment
A molecular matrix was generated for each indepen-
dent marker. All sequences were edited and aligned,
together with those obtained in Vila et al. (2011), using
Geneious 4.8.3 (Biomatters Ltd., 2009). ITS2 sequences
were aligned according to secondary structure using the
ITS2 Database Server (Koetschan et al., 2010), as
described in Schultz and Wolf (2009). The HMM-
Annotator tool (Keller et al., 2009) was used to delimit
and crop the ITS2 margins (E-value < 0.001, metazoan
HMMs), preserving the proximal stems (25 nucleotides
of 5.8S and 28S rDNA). The secondary structure of
ITS2 was predicted by custom homology modelling
using the template structure of Neolysandra coelestina
(MW99013) inferred by Wiemers et al. (2010), and at
least 75% helix transfer was used (ITS2PAM50 matrix;
gap costs: gap open 15, gap extension 2). For the
outgroup taxa in Everes and Leptotes sections, the more
closely related taxa Chilades trochylus MW99425 and
Tarucus theophrastus MW02025 were used, respectively,
as references for secondary structure prediction. For the
few cases with incomplete proximal stem (3¢end), the
short missing sequence was completed using the equiv-
alent fragment from the template. These additions were
necessary to obtain a correct alignment, and were
removed for the posterior phylogenetic analysis. Se-
quences and secondary structures were aligned synchro-
nously with 4SALE 1.5 (Seibel et al., 2006, 2008) using
an ITS2-specific 12 ·12 scoring matrix.
Regions of the matrix lacking more than 50% of data,
as well as ambiguously aligned regions, were removed
using Gblocks ver. 0.96 under a relaxed criterion with
the following parameters: )b2 = (50% + 1 of the
sequences) )b3 = 3 )b4 = 5 )b5 = all (Castresana,
2000; Talavera and Castresana, 2007). This step was not
applied to the ITS2 alignment. The final combined
alignment consisted of 6666 bp: 2172 bp of COI +
leu-tRNA +COII, 1171 bp of EF-1a, 745 bp of CAD,
811 bp of 28S, 370 bp of Wg, 1069 bp of ITS2,and
328 bp of H3 (see Data S1).
Phylogenetic inference and dating
Maximum parsimony (MP), maximum likelihood
(ML) and Bayesian inference (BI) were employed to
estimate evolutionary relationships within Polyomma-
tina. For MP analysis, the nine markers were concat-
enated in a single matrix and used as an input for the
software PAUP ver. 4.0b10 (Swofford, 2000). Heuristic
searches were performed with TBR branch swapping
and 10 000 random taxon addition replicates, saving no
more than 10 equally parsimonious trees per replicate.
To estimate branch support on the recovered topology,
nonparametric bootstrap values (Felsenstein, 1985)
were assessed with PAUP ver. 4.0b10. One hundred
bootstrap pseudoreplicates were obtained under a
heuristic search with TBR branch swapping with 1000
random taxon addition replicates, saving no more than
10 equally parsimonious trees per replicate. Model-
based approaches were conducted with BEAST ver.
1.6.0 (Drummond and Rambaut, 2007) for BI, and
GARLI-PART ver. 0.97 (Zwickl, 2006) for ML. The
data were partitioned by six markers, considering
COI +leu-tRNA +COII a single evolutionary unit
in the mitochondrial genome. jModeltest ver. 0.118
(Posada, 2008) was executed to select the best-fitting
models for DNA substitution for each marker data set
according to the Akaike information criterion (AIC).
As a result, the HKY model was used for H3, the TN
model for CAD, and a GTR model for the rest of the
markers, in all cases with a gamma distribution (+G)
and a proportion of invariants (+I) to account for
heterogeneity in evolutionary rates among sites. The
gamma distribution was estimated automatically from
the data using six rate categories. Branch support was
assessed by 100 bootstrap replicates for ML, and the
170 G. Talavera et al. / Cladistics 29 (2013) 166–192
Table 2
Samples used in this study: taxon name, specimen label, sample accession number at MCZ and sample collection locality used in the analysis
Subtribe Genus Species & ssp. Sample code Locality
Polyommatina Agriades glandon VL-05-Z994 Russia, Altai, Sailugem Range
Polyommatina Agriades optilete optilete VL-01-B424 Russia, St. Petersburg, Tamengont
Polyommatina Agriades optilete yukona JB-05-I879 Canada, Yukon, Dempster Hwy km 359
Polyommatina Agriades orbitulus AD-03-B064 Russia, Altai, Aktash
Polyommatina Agriades pheretiades NK-00-P690 Kazakhstan, Dzhambul reg., Kirgizski range
Polyommatina Agriades podarce AS-92-Z130 USA, California, Leek Spring
Polyommatina Agriades pyrenaicus dardanus AD-00-P259 Armenia, Gnishyk, Aiodzor Mts.
Polyommatina Alpherakya sarta VL-02-X098 China, Xinjiang, Kuqa
Polyommatina Aricia agestis NK-00-P712 Kazakhstan, Kayandy
Polyommatina Aricia artaxerxes AD-02-W127 Russia, Primorski Krai Khanka Lake
Polyommatina Aricia chinensis VL-05-Z997 Russia, Buryatia, Sosnovka, 900 m
Polyommatina Aricia crassipuncta AD-00-P528 Armenia, Transcaucasus, Alibek Mt.
Polyommatina Aricia nicias AD-03-B041 Russia, Altai, Aktash env.
Polyommatina Aricia vandarbani VL-03-F745 Azerbaijan, Lerik, Talysh, 900–1000 m
Polyommatina Chilades lajus DL-99-T242 Thailand, Prachuap Khiri Khan Province,
Ampuh Thap Sakae
Polyommatina Cyaniris semiargus belis AD-00-P369 Armenia, Zangezur mts., Akhtchi
Polyommatina Cyaniris semiargus semiargus AD-00-P206 Russia, Low Volga, Volgograd reg., Kamytshinky
Polyommatina Cyclargus ammon JE-01-C283 USA, Florida, Big Pine Key
Polyommatina Echinargus isola AS-92-Z185 USA, California, Alpine, Carson River
Polyommatina Eldoradina cyanea RV-05-M735 Peru, Lima, Oyo
´n
Polyommatina Eumedonia eumedon AD-03-B062 Russia, Altai, Aktash
Polyommatina Eumedonia persephatta minshelkensis NK-00-P743 Kazakhstan, Shymkent reg., Karatau Mts.
Polyommatina Freyeria putli RE-02-A007 Australia, Queensland, Trinity Beach
Polyommatina Freyeria trochylus VL-01-L462 Turkey, Artvin, Kilic¸ kaya
Polyommatina Glabroculus cyane VL-02-X159 Kazakhstan, Karaganda region, Aktchatau
Polyommatina Glabroculus elvira NK-00-P793 Kazakhstan, Baltakul vlg.
Polyommatina Hemiargus hanno bogotanus SR-03-K069 Colombia, Caldas, Chinchina
Polyommatina Hemiargus hanno ceraunus MH-01-I001 Puerto Rico, Culebra Island, Flamenco Beach
Polyommatina Hemiargus hanno gyas AS-92-Z255 USA, California, Los Angeles, Pyramid Lake
Polyommatina Hemiargus hanno gyas DL-02-P801 USA, Arizona, Chiricahua Mts.
Polyommatina Hemiargus huntingtoni RE-01-H234 Costa Rica, P.N. Santa Rosa, Guanacaste
Polyommatina Hemiargus martha RV-04-I212 Peru, Hua
´nuco
Polyommatina Hemiargus ramon MFB-00-N223 Chile, Arica, Molino
Polyommatina Icaricia acmon AS-92-Z184 USA, California, Alpine, Carson River
Polyommatina Icaricia icarioides AS-92-Z065 USA, California, Nevada, Donner Pass
Polyommatina Icaricia saepiolus AS-92-Z069 USA, California, Nevada, Donner Pass
Polyommatina Icaricia shasta AS-92-Z465 USA, California, Nevada, Castle Peak
Polyommatina Itylos huascarana RV-04-I403 Peru, Ancash, Pitec
Polyommatina Itylos koa RV-03-V327 Peru, Junı
´n, Huasahuasi
Polyommatina Itylos mashenka MFB-00-N166 Peru, Junı
´n
Polyommatina Itylos sigal MFB-00-N220 Chile, Socoroma
Polyommatina Itylos tintarrona RV-03-V182 Peru, Arequipa, Can
˜o
´n del Colca
Polyommatina Itylos titicaca MFB-00-N206 Chile, P.N. Lanca, Las Cuevas
Polyommatina Kindermannia morgiana VL-02-X393 Iran, Kerman, Kuh-e-Lalizar Mts.
Polyommatina Kretania alcedo VL-01-L319 Turkey, Erzurum Prov., Ko
¨pru
¨ko
¨y
Polyommatina Kretania eurypilus VL-01-L152 Turkey, Gu
¨mu
¨shane Prov., 35 km
SW Gu
¨mu
¨shane, Dilekyolu
Polyommatina Kretania eurypilus zamotajlovi SH-02-H006 Russia, Krasnodar Region, Abrau
Polyommatina Kretania pylaon AD-00-P066 Russia, Volgograd, Kamyshinsky
Polyommatina Kretania zephyrinus AD-00-P121 Armenia, Transcaucasus, Sevan, Shorzha
Polyommatina Luthrodes cleotas CJM-07-J018 PNG, New Ireland Prov, Simberi Is.
Polyommatina Luthrodes galba HU-08-D004 Cyprus, Ayios Nikolaos
Polyommatina Luthrodes pandava MWT-93-A009 Malaysia, Kepong
Polyommatina Lysandra bellargus AD-00-P129 Armenia, Transcaucasus, Amberd Valley, Aragatz Mt.
Polyommatina Lysandra coridon borussia AD-00-P192 Russia, Tula region, Tatinki, 120 m
Polyommatina Lysandra punctifera NK-02-A027 Morocco, High Atlas, Col-Tagh pass
Polyommatina Maurus vogelii RVcoll09-X164 Morocco, Khenifra, S. Timahdite, Col du Zad
Polyommatina Nabokovia cuzquenha RV-03-V234 Peru, Cuzco, Pisac
Polyommatina Nabokovia faga MFB-00-N217 Chile, Socoroma
Polyommatina Neolysandra coelestina alticola AD-00-P092 Armenia, Gegadyr, Gegamsky Mts.
Polyommatina Neolysandra diana AD-00-P081 Armenia, Gegadyr, Gegamsky Mts., 1800m
171G. Talavera et al. / Cladistics 29 (2013) 166–192
software SumTrees in the DendroPy phylogenetic
Python library (Sukumaran and Holder, 2010) was
used to generate a majority-rule bootstrap consensus
tree.
BI with BEAST ver. 1.6.0 was used to estimate
divergence times. Normally distributed tmrca priors
including maximum and minimum ages from Vila et al.
(2011) within the 95% HPD distribution were estab-
lished on four well supported nodes, shown in Fig. 1.
The resulting 95% HPD ranged from 1.5 to 3.3 Myr for
node 1; from 5.5 to 13.1 Myr for node 2; from 8.4 to
16.8 Myr for node 3; and from 2.5 to 11.3 Myr for
node 4. The uncorrelated relaxed clock (Drummond
et al., 2006) and a constant population size under a
coalescent model were established as priors. The rest of
the settings and priors were set by default. Two
Table 2
Samples used in this study: taxon name, specimen label, sample accession number at MCZ and sample collection locality used in the analysis
Subtribe Genus Species & ssp. Sample code Locality
Polyommatina Pamiria chrysopis VL-05-Z998 Tajikistan, East Pamir, Sarykolski Range, Dunkeldyk Lake
Polyommatina Paralycaeides inconspicua RV-03-V188 Peru, Arequipa, Can
˜o
´n del Colca
Polyommatina Paralycaeides vapa RV-03-V198 Peru, Puno, Chucuito
Polyommatina Patricius lucifer VL-05-Z995 Russia, Altai, Chikhacheva Range, Sailugem Mt; 2300–2400 m
Polyommatina Plebejidea loewii AD-00-P266 Armenia, Gnishyk, Aiodzor Mts.
Polyommatina Plebejus anna AS-92-Z072 USA, California, Nevada, Donner Pass
Polyommatina Plebejus argus NK-00-P135 Ukraine, Krim, Ai-Petri Mt.
Polyommatina Plebejus argyrognomon AD-00-P560 Russia, Tula, Tatinki
Polyommatina Plebejus idas armoricanella NK-00-P165 Russia, St. Petersburg, Luga
Polyommatina Plebejus idas ferniensis NGK-02-C411 Canada, British Columbia, Castlegar
Polyommatina Plebejus melissa AS-92-Z005 USA, California, Nevada, Verdi
Polyommatina Plebulina emigdionis CCN-05-I856 USA, California, Kern, W. Onyx
Polyommatina Polyommatus amandus NK-00-P596 Kazakhstan, Altai, Oktyabrsk
Polyommatina Polyommatus amandus AD-00-P053 Russia, Volgograd region, Kamyshinsky
Polyommatina Polyommatus amandus MAT-99-Q840 Spain, Pyrenees, Uru´ s
Polyommatina Polyommatus amandus amurensis AD-02-W109 Russia, Primorski Krai, S. Ussuri, Khanka Lake, Poganichnoye
Polyommatina Polyommatus cornelia VL-01-L135 Turkey, Gu
¨mu
¨shane Prov., 35 km SW Gu
¨mu
¨shane, Dilekyolu
Polyommatina Polyommatus damocles krymaeus NK-00-P103 Ukraine, Crimea, Kurortnoe
Polyommatina Polyommatus damon damon MAT-99-Q841 Spain, Pyrenees, Uru´ s
Polyommatina Polyommatus daphnis NK-00-P108 Ukraine, Crimea, Kurortnoe
Polyommatina Polyommatus dorylas armena AD-00-P312 Armenia, Gnishyk, Aiodzor Mts.
Polyommatina Polyommatus erotides AD-03-B040 Kazakhstan, Tarbagatai Mts., Petrovskoe env.
Polyommatina Polyommatus erschoffii AD-02-L274 Tajikistan, East Pamir, Sarykolski Range, Dunkeldyk Lake
Polyommatina Polyommatus escheri MAT-99-Q838 Spain, Pyrenees, Uru´ s
Polyommatina Polyommatus glaucias AD-02-M278 Iran, Gorgan Prov., Shahkuh
Polyommatina Polyommatus hunza VL-05-Z996 Tajikistan, East Pamir, Sarykolski Range, Dunkeldyk Lake
Polyommatina Polyommatus icarus NK-00-P562 Kazakhstan, Altai, Oktyabrsk
Polyommatina Polyommatus marcida AD-02-W258 Iran, Mazandaran, Geduk Pass and Veresk
Polyommatina Polyommatus myrrha cinyraea AD-00-P389 Armenia, Zangezur Mts., Akhtchi
Polyommatina Polyommatus nivescens MAT-99-Q904 Spain, Lleida, Ru´ bies
Polyommatina Polyommatus ripartii budashkini NK-00-P859 Ukraine, Crimea, Karabi yaila
Polyommatina Polyommatus stempfferi VL-02-X324 Iran, Esfahan, Khansar
Polyommatina Polyommatus surakovi surakovi AD-00-P006 Armenia, Aiodzor mts., Gnishyk
Polyommatina Polyommatus thersites MAT-99-Q947 France, Languedoc region, Mende
Polyommatina Polyommatus thersites AD-00-P019 Armenia, Aiodzor Mts., Gnishyk, 1800 m
Polyommatina Polyommatus venus NK-00-P810 Kazakhstan, Karzhantau vlg.
Polyommatina Pseudochrysops bornoi MAC-04-Z114 Dominican Republic, Punta Cana
Polyommatina Pseudolucia asafi RV-03-V020 Chile, Ce
´spedes, Illapel
Polyommatina Pseudolucia charlotte BD-02-B813 Chile, Temuco
Polyommatina Pseudolucia chilensis MFB-00-N227 Chile, Farellones
Polyommatina Pseudolucia sibylla RV-03-V112 Chile, Coquimbo,
´o La Laguna
Polyommatina Pseudolucia vera BD-02-B812 Chile, Temuco, Volca
´n Villarica
Polyommatina Rimisia miris NK-00-P575 Kazakhstan, Altai, Oktyabrsk
Polyommatina Rueckbeilia fergana NK-00-P777 Kazakhstan, Shymkent Reg., Karatau Mts., Turpan Pass
Cupidina Cupido comyntas AS-92-Z312 USA, California, Davis
Cupidina Cupido minimus AD-00-P540 Russia, Tula, Tatinki
Cupidina Talicada nyseus JXM-99-T709 India, Karala, Trivandrum
Cupidina Tongeia fischeri NK-00-P594 Kazakhstan, Altai, Oktyabrsk
Leptotina Leptotes trigemmatus RV-03-V095 Chile, Coquimbo, Alcohuas
172 G. Talavera et al. / Cladistics 29 (2013) 166–192
AD00P092.Neolysandra coelestina
.MAT99Q904 Polyommatus nivescens
RV03V095.Leptotes trigemmatus
BD02B813.Pseudolucia charlotte
MH01I001.Hemiargus hanno
ZVL05 996.Polyommatus hunza
.
AS92Z465 Icaricia shasta
CCN05I856.Plebulina emigdionis
VL05Z994.Agriades glandon
NK00P596.Polyom matus amandus
AD00P206.Cyaniris semiargus
MFB00N220.Itylos sigal
AD00P540.Cupido minimus
VL01L462.Freyeria trochylus
.RV03V327 Itylos koa
VL02X159.Glabroculus cyane
NK00P575.Rimisia miris
.AS92Z005 Plebejus melissa
.AD00P121Kretania zephyrinus
AD00P192.Lysandra coridon
.AD00P389Polyommatus myrrha
2AS9 Z065.Icaricia icarioides
.AS92Z312 Cupido comyntas
AD00P019.Polyom matus thersites
RE02A007.Freyeria putli
.NGK02C411 Plebejus idas
.RV04I212 Hemiargus martha
.VL01B424.Agriades optilete
AS92Z255.Hemiargus hanno
BD02B812.Pseudolucia vera
RV03V020.Pseudolucia asafi
MFB00N227.Pseudolucia chilensis
VL05Z995.Patricius lucifer
NK00P562.Polyommatus icarus
NK00P859.Polyommatus ripartii
.AD00P066 Kretania pylaon
9TJXM9 709.Talicada nyseus
.RE01H234 Hemiargus huntingtoni
MC04Z114.Pseudochrysops bornoi
T9MA 9 Q840.Poly ommatus amandus
NK00P594.Tongeia fischeri
VL02X098.Alpherakya sarta
MFB00N223.Hemiargus ramon
.AD00P006 Polyommatus surakovi
AD02W258.Polyommatus marcida
NK00P135.Plebejus argus
NK00P810.Polyommatus venus
.AS92Z072 Plebejus anna
SR03K069.Hemiargus hanno
VL05Z997 Aricia chinensis
.AD02L274 Polyommatus erschoffii
.AD00P560 Plebejus argy rognomon
RV03V188.Paralycaeides inconspicua
A9Q .MT9841Polyommatus damon
AS92Z130.Agriades podarce
DL99T242.Chilades lajus
NK00P108.Polyommatus daphnis
AD00P266.Plebejidea loewii
RVcoll.09X164.Maurus vogelii
CJM07J018.Luthrodes cleotas
MAT99Q947.Polyommatus thersites
AD02W127.Aricia artaxerxes
.NK00P165 Plebejus idas
F
M B00N166.Itylos mashenka
.
NK00P793 Glabroculus elvira
AD03B040.Polyomm atus erotides
NK02A027.Lysandra punctifera
.AD02M278Polyommatus gla ucias
.JB05I879.Agriades optilete
2VL0 X324.Polyommatus stempfferi
AD00P369.Cyaniris semiargus
AD00P053.Polyomm atus amandus
AD00P081.Neolysandra diana
.VL01L135 Polyommatus cornelia
.AD03B064.Agriades orbitulus
NK00P777.Rueckbeilia fergana
.gal
HU08D004 Luthrodes ba
2AS9 Z184.Icaricia acmon
MFB00N217.Nabokovia faga
AD00P312.Polyommatus dorylas
MFB00N206.Itylos titicaca
AD00P129.Lysandra bellargus
AS92Z185.Echinargus isola
.SH02H006 Kretania eurypilus
3.
VL0 F745 Aricia vandarvani
NK00P712.Aricia agestis
5VL0 Z998.Pamiria chrysopis
JE01C283.Cyclargus ammon
RV04I403.Itylos huascarana
AD03B062.Eumedonia eumedon
RV03V234.Nabokovia cuzquenha
DL02P801.Hemiargus hanno
.NK00P103 Pol yommatus damocles
VL02X393.Afarsia morgiana
RV05M735.Eldoradina cyanea
RV03V112.Pseudolucia sibylla
.
AD00P528 Aricia crassipuncta
VL01L319.Kretania alcedo
AD02W109.Polyommatus amandus
AD00P259.Agriades pyrenaicus
NK00P690.Agriades pheretiades
.
AD03B041.Aricia nicias
KP .N00743Eumedonia persephatta
S2A9Z069.Icaricia saepiolus
VL01L152.Kretania eurypilus
MAT99Q838.Polyommatus escheri
T.
MW 93A009Luthrodes pandava
.
RV03V182 Itylos tintarrona
RV03V198.Paralycaeides vapa
0.05.010.015.020.025.030.0
Pseudochrysops Nabokov, 1945
Cyclargus Nabokov, 1945
Echinargus Nabokov, 1945
Hemiargus Hübner, 1818
Eldoradina Balletto, 1993
(=Polytheclus Bálint & Johnson, 1993)
Nabokovia Hemming, 1960
(=Pseudothecla Nabokov, 1945)
Pseudolucia Nabokov, 1945
(=Pallidula Balletto, 1993)
Facula Balletto, 1993
Cherchiella Balletto, 1993
Paralycaeides Nabokov, 1945
Boliviella Balletto, 1993
Itylos Draudt, 1921
(=Ithylos Forster, 1955)
Parachilades Nabokov, 1945
Ityloides Balletto, 1993
Madeleinea Bálint, 1993
(=Nivalis Balletto, 1993)
Chilades Moore, [1881]
Luthrodes Druce, 1895
Edales Swinhoe, [1910]
Lachides Nekrutenko, 1984
Freyeria Courvoisier, 1920
Plebulina Nabokov, [1945]
Icaricia Nabokov, [1945]
Pamiria Zhdanko, 1994
Rueckbeilia gen. nov.
Patricius Bálint, [1992]
Themisia Zhdanko, 2002
Plebejus Kluk, 1780
(=Rusticus Hübner, [1806])
(=Lycoena Nicholl, 1901)
Lycaeides Hübner, [1919]
Kretania Beuret, 1959
Plebejidea Koçak, 1983
Maurus Bálint, [1992]
Eumedonia Forster, 1938
Cyaniris Dalman, 1816
(=Nomiades Hübner, [1819])
Glaucolinea Wang & Rehn, 1999
Rimisia Zhdanko, 1994
Alpherakya Zhdanko, 1994
Lysandra Hemming, 1933
(=Uranops Hemming, 1929)
(=Argus Scopoli, 1763)
Neolysandra Koçak, 1977
Afarsia Korb & Bolshakov, 2011
(=Farsia Zhdanko,1992)
Aricia Reichenbach, 1817
(=Gynomorphia Verity, 1929)
Umpria Zhdanko, 1994
Pseudoaricia Beuret, 1959
Ultraaricia Beuret, 1959
Polyommatus Latreille, 1804
Actisia Koçak & Kemal, 2001
Admetusia Koçak & Seven, 1998
Agrodiaetus Hübner, 1822
(=Hirsutina Tutt, [1909])
Antidolus Koçak & Kemal, 2001
Bryna Evans, 1912
Damaia Koçak & Kemal, 2001
Meleageria De Sagarra, 1925
Musa Koçak & Kemal, 2001
Paragrodiaetus Rose & Schurian, 1977
Peileia Koçak & Kemal, 2001
Phyllisia Koçak & Kemal, 2001
Plebicula Higgins, 1969
Sublysandra Koçak, 1977
Thersitesia Koçak & Seven, 1998
Transcaspius Koçak & Kemal, 2001
Xerxesia Koçak & Kemal, 2001
Elviria Zhdanko, 1994
Glabroculus Lvovsky, 1993
Agriades Hübner, [1819]
(= Latiorina Tutt, [1909])
Albulina Tutt, 1909
Himalaya Koçak & Seven, 1998
VacciniinaTutt, 1909
Xinjiangia Huang & Murayama, 1988
Plebejides Sauter, 1868
Cupidina Subtribe
Polyommatina Subtribe
Leptotina Subtribe
Dagmara Koçak & Kemal, 2001
0 20406080100
Time (Myr)
Number of Lineages
Number of splits
25 20 15 10 5 0
Genera minimum age
Subtribes minimum age
Juldus Koçak & Kemal, 2001
Agriades
podarce
Polyommatus
icarus
Aricia
nicias
Lysandra
coridon
Eumedonia
eumedon
Cyaniris
semiargus
Plebejus
melissa
Icaricia
acmon
Plebulina
emigdionis
Kretania
zephyrinus
Pseudolucia
chilensis
Nabokovia
faga
Eldoradina
cyaena
Itylos
titicaca
Pseudochrysops
bornoi
Hemiargus
hanno
Paralycaeides
vapa
Afarsia
iris
Alpherakya
sartoides
Patricius
lucifuga
Pamiria
chrysopis
Echinargus
isola
Cyclargus
ammon
Mestore Koçak & Kemal, 2007
3
4
2
1
Myr
Lineages Through Time Plot
Fig. 1. Bayesian chronogram for the newly proposed subtribe Polyommatina based on nine genes: COI,leu-tRNA,COII,EF-1a,Wg,ITS2,CAD,
28S and H3 (6666 bp). Thick lines indicate supported relationships (posterior probabilities 0.95); node bars show estimated divergence times
uncertainty. Nearly all the extant genera are included in the phylogeny; representatives from the subtribes Cupidina and Leptotina were used as
outgroups. Valid genus names are presented in bold. Subjective synonyms (that may yet be shown to represent valid subgenera with additional
research) are shown after the valid names. Objective synonyms are indicated by ‘‘=’’. Normally distributed tmrca from inferred divergence times in
Vila et al. (2011) were used as priors on the nodes 1–4. The phylogeny revealed unexpected relationships with respect to traditional classification. We
rearranged the systematics of the group and proposed a new list of genera according to the following criteria: (i) taxa older than 5 Myr are considered
genera; (ii) for taxa between 4 and 5 Myr we are conservative in the sense that we consider a clade to be a genus only if it has already been described,
and do not consider it a genus if it has not; and (iii) taxa younger than 4 Myr are considered subgeneric. The 4–5-Myr time interval is highlighted in
red. Applying these criteria resulted in the retention of 31 of the 81 genera formally described in the group, and necessitated the addition of one new
genus. Minimum age thresholds used to define genera and subtribes are indicated in the lineage through time plot. The upper side and underside of
representative adult specimens of the Polyommatina are shown on the right.
173G. Talavera et al. / Cladistics 29 (2013) 166–192
independent chains were run for 50 million generations
each, sampling values every 1000 steps. A conservative
burn-in of 500 000 generations was applied for each run
after checking Markov chain Monte Carlo (MCMC)
convergence through graphically monitoring likelihood
values in Tracer ver. 1.5 (Rambaut and Drummond,
2007). Independent runs were combined in LogCom-
biner ver. 1.6.0 implemented in the software package
BEAST and all parameters were analysed using the
program Tracer to determine whether they had also
reached stationarity. Tree topologies were assessed using
TreeAnnotator ver. 1.6.0 in the BEAST package to
generate a maximum clade credibility tree of all sampled
trees with median node heights. Finally, FigTree ver. 1.2.2
(Rambaut, 2009) was used to visualize the consensus tree
along with node ages, age deviations and node posterior
probabilities.
Ancestral states reconstruction
Character evolution was reconstructed by estimating
probabilities for ancestral character states with MES-
QUITE ver. 2.6 (Maddison and Maddison, 2007). Both
MP and ML approaches were applied to the Bayesian
tree for two discrete (absence or presence) morpholog-
ical characters traditionally used to define the genus
Vacciniina: (i) metallic marginal spots on the hind wing
underside; and (ii) inner apical part of the valvae in the
male genitalia with sclerotized ventral fold. A reduced
phylogenetic tree excluding the basal Neotropical clade
and outgroup was used.
Results and discussion
Higher-level systematics
The taxonomic system employed by Eliot (1973)
grouped the genera in the rather unconventional cate-
gory ‘‘section’’. This system is still widely used, and it
coexists with several arrangements that use the more
formal categories ‘‘tribe’’ and ‘‘subtribe’’. Since this
study represents the first comprehensive revision of the
group since Eliot, our goal is to normalize the system-
atics above the level of the genus. Our phylogeny
(Fig. 1) shows that the Polyommatus section is mono-
phyletic (see also Vila et al., 2011). We propose to use
the term ‘‘Polyommatina subtribe’’ to replace EliotÕs
‘‘Polyommatus section’’, and generally use the designa-
tion ‘‘subtribe’’ instead of ‘‘section’’ throughout. Thus
Cupidina would be the sister to the Polyommatina, and
Leptotina the sister to both. We estimate the ages of
divergence for these subtribes to range between 22.8 and
25.7 Myr. In the lineages through time plot (Fig. 1), a
relatively long period without diversification events,
from 22.8 to 13.6 Ma, is observed. We have designated
this period as a gap defining subtribes, and therefore
consider subtribes to be those lineages older than
15 Myr. The three sections previously recognized by
Eliot (1973) for the studied group fall within this
definition of subtribe, as do most of the rest of sections
in Polyommatini (Vila et al., 2011). In order to evaluate
the four tribes within the subfamily Polyommatinae (e.g.
Candalidini, Lycaenesthini, Niphandini and Polyomma-
tini), an adequate threshold will need to be set for the
tribal level using a more thorough phylogenetic analysis
of the Lycaenidae that includes these taxa.
Genus concept
Since our aim is to establish a phylogenetically based
classification system for the Polyommatina, criteria for
delineating genera are important to establish. This is
especially true given the wide array of taxonomic
classifications that have been proposed for this group
at the genus level, including drastic approaches that split
the group into numerous nearly monotypic (consisting
of a single species) genera (Forster, 1938; Zhdanko,
2004), or lumped all species into only a few genera
(Zhdanko, 1983; Scott, 1986; Hesselbarth et al., 1995;
Gorbunov, 2001; Kudrna, 2002).
Monophyly. One important criterion defining a genus is
that it should be monophyletic. The majority of taxon-
omists currently believe that monophyly, in the narrow
sense used by Hennig (Hennig, 1950, 1966; Envall, 2008;
Ho
¨randl and Stuessy, 2010) (= holophyly sensu Ash-
lock, 1971) is mandatory, at least for taxonomic
categories above the species level (genus, family, etc.)
(Schwenk, 1994; Groves, 2004). Paraphyletic taxa are
incompatible with the principles of phylogenetic sys-
tematics (Schmidt-Lebuhn, 2011) and have relatively
few defenders (Brummitt, 2003; Ho
¨randl and Stuessy,
2010). Using paraphyletic groups in higher-level taxon-
omy poses serious problems as it can result in taxa that
are neither mutually exclusive nor wholly inclusive of
one another (Nelson et al., 2003). This gives rise to
uncertainties and discrepancies in classifications. Thus
avoiding paraphyletic groups and focusing on mono-
phyletic entities sensu Hennig is the preferable option in
practical terms. It is important to note, too, that the
concept of monophyly applies to whole organisms.
Trees inferred from single markers sometimes display
paraphyletic relationships that reflect the evolutionary
histories of individual genes rather than the species
being studied. It is thus advisable to base taxonomic
conclusions on multilocus analyses using the principle of
character congruence as advocated by Kluge (1989) and
Brower et al. (1996).
Still, the monophyly criterion alone is not enough to
construct a taxonomic system. Nearly every phylogeny
is a complicated structure consisting of numerous nested
174 G. Talavera et al. / Cladistics 29 (2013) 166–192
monophyletic lineages. The number of these nested
clades is often much greater than the number of
traditional taxonomic ranks. Therefore additional crite-
ria need to be used to select which monophyletic lineages
should be considered genera and which not, and similar
criteria should be established for other ranks.
The morphological gap and the concept of cryptic
genera. One criterion that can be used in defining a
genus is the existence of a discontinuity in the distribu-
tion of morphological characters between one mono-
phyletic group and another. The morphological gap
(= morphological hiatus) seen between genera should
be significantly larger than the gaps seen between species
of the same genus. This criterion is widely used, but it is
not ideal. First, it may be difficult to decide when a
morphological gap is sufficient to separate genera (and it
may be difficult to measure morphological gaps in the
first place). Second, and most importantly, using this
criterion can result in artificial taxonomic systems due to
homoplasy. For example, the genus Vacciniina in its
traditional conception includes three morphologically
similar species: V. optilete,V. alcedo and V. fergana
(Tuzov et al., 2000) (Fig. 2). However, our study dem-
onstrates that these species represent three different
evolutionary lineages that are not closely related (Fig. 1).
In fact, we describe the new genus Rueckbeilia for the
traditional species V. fergana, and include the species
V. alcedo in the genus Kretania and the species V. opti-
lete in the genus Agriades. Thus Vacciniina sensu
auctorum represents three cryptic genera, i.e. three
species clusters that cannot be separated from one
another based on their morphological characters and,
at the same time, cannot be lumped into a single genus as
their combination would be polyphyletic. As a conse-
quence, we suggest that the recognition of cryptic genera
(Vilnet et al., 2007; Lucky and Sarnat, 2008) may be
useful, in the same manner that the recognition of cryptic
species is now widely used (Descimon and Mallet, 2009).
Cryptic genera are the consequence of unrecognized
parallelisms in evolution of some morphological char-
acters or of the long preservation of plesiomorphic
states that are mistakenly considered synapomorphies;
or of both processes acting simultaneously in different
characters. For example, the species V. optilete seems to
have independently evolved a wing pattern similar to
those of V. alcedo and V. fergana (Fig. 2), whereas the
‘‘Polyommatus-like’’ structures of the male genitalia of
these lineages (Stekolnikov, 2010) (Fig. 3) probably
represent an ancestral condition that has been preserved
for at least 6 million years (Fig. 4).
Age of lineage as a universal and unbiased criterion?
Hennig (1966) proposed to synchronize taxonomic ranks
universally according to geological ages. This would
have the effect of making groups comparable and ranks
definable. Since geological time is universal, the age of
evolutionary lineages, generally estimated by the dating
of nodes in phylogenetic trees, seems to be the only truly
unbiased criterion by which taxonomic classifications
above the level of biological species can be erected
(Hennig, 1966). Avise and Johns (1999) devised a specific
temporal-banding scheme to fit conventional Linnaean
ranks. They proposed considering as genera those
lineages that originated in the Pliocene (ca. 2–5 Ma); as
subgenera the lineages above the level of species that
(a)
(b)
(c)
(d)
(e)
(f)
Fig. 2. Taxa representing three cryptic genera. (a,b) Rueckbeilia fergana (= ‘‘Vacciniina’’ fergana); (c,d) Kretania alcedo (= ‘‘Vacciniina’’ alcedo);
(e,f) Agriades optilete (= ‘‘Vacciniina’’ optilete). These taxa were all considered species of the same genus (Vacciniina), although in fact they form
three distinct genera according to the criteria described in this study. Despite their genetic differences, this artificial assemblage is strikingly
convergent with respect to wing colour and pattern. They share the violet-blue colour of the wing upper side in males, and the similar wing underside
with blue metallic scales that seems to have evolved independently at least twice.
175G. Talavera et al. / Cladistics 29 (2013) 166–192
originated in the Pleistocene (0–2 Ma); and as tribes the
lineages that originated in the Miocene (5–24 Ma).
However, this proposal has two main problems: it is
not directly applicable to fossil organisms (Griffiths,
1973), and it would necessitate a major, even radical,
rearrangement for current taxonomy. Acknowledging
these difficulties, Avise and Mitchell (2007) launched the
‘‘timeclip proposal’’, which consists of labelling classic
Linnaean taxa with timeclips that indicate their geolog-
ical ages of origin. This could provide relevant addi-
tional information that could be updated easily without
the need to alter taxonomy. Although the timeclip
proposal is interesting, it still relies on a taxonomic
system, and does not invalidate the need to establish
true relationships within and between taxa and to decide
how to determine taxonomic ranks.
We agree with the concept of relative ages, but we think
this should be modified in at least two respects. First, the
age thresholds must take into account the systematics
and relative ages of different groups of organisms.
Second, once the taxonomic ranks are established,
diagnostic morphological characters should be explained
or explored. Moreover, the rank of subtribe, which is
especially useful in insect systematics, should be incor-
porated in the proposal of Avise and Johns (1999).
Stability and preservation of traditionally recognized
taxa. The stability and preservation of traditionally
recognized taxa must be taken into account in estab-
lishing classification guidelines (Godfray and Knapp,
2004). Indeed, stability is a concept that is positively
valued by the International Commissions of Nomencla-
ture, and that can, in some instances, take precedence
over other principles. Applying a universal system of
thresholds would result in taxonomic upheaval, mostly
because at present there is deep discrepancy in the
average age of the taxa accepted for different groups of
organisms (Avise and Liu, 2011). In mammals, for
example, many recognized genera are relatively young
(3–5 Myr) (Castresana, 2001; Rowe et al., 2008; Abram-
son et al., 2009) with an estimated mean of 9.6 Myr
(0.1–40) (Avise and Liu, 2011), whereas other groups
may be relatively old, such as Decapoda, with an
estimated mean of 60.2 Myr (16.8–135) (Avise and Liu,
2011) or Diptera (Drosophilidae, Chironomidae) with
estimated means ranging from 30–40 Myr to more than
100 Myr (Avise and Johns, 1999; Cranston et al., 2010).
Strong temporal banding heterogeneity among different
organismal assemblages also occurs at higher taxonomic
levels such as families and orders (Avise and Liu, 2011;
Hedges and Kumar, 2009). Consequently, a universal
system would require such a complete reorganization of
the systematics of most groups of organisms that the
overall effect would be deleterious to communication
and understanding of taxonomic relationships.
Even if the most extreme cases, such as the relative
ages of genera in Diptera or Decapoda, were to be
modified to create a more balanced general classification,
we propose that a temporal scheme should adapt to some
degree to the particularities and pre-existing taxonomy
of each group. Differences in the age thresholds might be
necessarily pronounced in distantly related groups of
taxa whose rates of diversification are likely to differ
depending on intrinsic biological differences such as
generation time and or population size (e.g. Li, 1997),
differences in the efficiency of DNA repair mechanisms
(Britten, 1986), or differences in metabolic rate (Martin
and Palumbi, 1993). The increased rate of nucleotide
changes at several loci, including such usual phylogenetic
markers as COI and CytB genes, can be affected in some
phylogenetic lineages by positive selection due to their
role in adaptation to specialized metabolic requirements
(da Fonseca et al., 2008).
In the case of Polyommatina, the following thresholds
provide a balanced classification that corresponds well
with current evidence about relationships between
groups: genera can be recognized as those lineages that
originated in the late Miocene (older than 5 Myr), and
subtribes those that originated in the early Miocene or
(a)
(b)
(c)
(d)
Fig. 3. Valva in the male genitalia. Inner part of valva with mem-
branous ventral fold indicated by arrow (a–c) and without membra-
nous ventral fold (d). (a) Rueckbeilia fergana; (b) Kretania alcedo; (c)
Agriades glandon; (d) Plebejus idas. After Stekolnikov (2010) with
modifications.
176 G. Talavera et al. / Cladistics 29 (2013) 166–192
late Oligocene (older than 15 Myr). In the lineages
through time plot, an increase in diversification can be
seen starting at 5–4 Ma (Fig. 1), so we set the minimum
age for genera at this point to avoid excessive splitting.
This approach (plotting the number of lineages or
branching events over time) is useful to illuminate
diversification patterns in the group under study.
Substantial changes in the rate of diversification mark
key moments in the evolution of a group as a whole, and
these are logical points to be used as age thresholds
delimiting taxonomic ranks.
In our case, age thresholds were also selected so as to
minimally affect the existing nomenclature and avoid the
need for descriptions of new genera. A generic threshold
of 3–4 Myr requires the creation of two new genera (the
splitting of Icaricia and description of the new Rueckbe-
ilia), while a 5–6 Myr threshold would have entailed a
wide-scale synonymization (50 subjective synomyms)
with excessive loss of phylogenetic information, and
would have still required the description of Rueckbeilia.
Wider thresholds would have also involved losing
substantial input from the molecular data.
Accounting for uncertainty in age estimates. Addition-
ally, any system of classification should recognize the
uncertainty inherent in estimating evolutionary age given
intrinsic errors associated with the methods of inference,
especially when no paleontological material is available to
calibrate a molecular clock. Absolute age is likely to vary
depending on the analysis, and new information helping
to calibrate the molecular clock, or additional method-
ological improvements, might affect age estimates. In
contrast, relative ages among lineages are less affected by
these factors because they do not depend on external
information for tree calibration. The greater uncertainty
in absolute age estimates compared with those based on
relative ages is another reason to apply a temporal scheme
specific to the group being studied, which could be
adapted eventually to a different molecular substitution
rate without major implications for the taxonomy of the
group. A universal temporal scheme would suffer from
taxonomic instability caused by uncertainty in absolute
age, which is necessary when comparing taxa that are not
closely related to each other. Divergent lineages some-
times display disparate molecular substitution rates,
whereas closely related taxa tend to be more uniform in
this regard (Martin and Palumbi, 1993; Li, 1997). The
subtribe Polyommatina, a clade that evolved ca. 22.8 Ma,
contains many taxa to be compared, but these are
sufficiently evolutionarily and ecologically similar that
they do not exhibit excessive variability in substitution
rates among lineages.
In order to reduce the effect of the uncertainty in age
estimates, and to avoid taxonomic instability because of
(a) (b)
Fig. 4. Ancestral state reconstructions for two morphological characters traditionally defining the polyphyletic genus Vacciniina (taxa in bold).
(a) Metallic marginal spots on the hind wing underside present (black circle) or absent (white circle). (b) Inner part of valvae in the male genitalia with
membranous ventral fold (black circle) or without membranous ventral fold (white circle). Maximum parsimony (upper circles) and Maximum
likelihood (lower grey circles) inferences are represented at nodes. Figures of genitalia are given after Stekolnikov (2010).
177G. Talavera et al. / Cladistics 29 (2013) 166–192
small differences obtained using different phylogenetic
analyses and or novel calibration points, we propose
using a time interval to set the limits of genus age, rather
than a single date (e.g. 4.0–5.0 Myr for genus minimum
age in our case). Thus lineages with a mean age within
these intervals can be dealt with using this relatively
conservative approach, as described below.
Importance of morphological diagnostic characters. Once
the classification of a group is produced using the
previously discussed criteria, the next critical step is to
explore and explain the diagnostic morphological
characters that define the proposed taxa. The exercise
of integrating the molecular-based classification into a
morphological framework has multiple benefits. It does
not create a discontinuity with the previous morphol-
ogy-based classifications; it avoids wasting the mor-
phological data painstakingly gathered; and it allows
for the reinterpretation of earlier work. It also facili-
tates the placement of extinct taxa and those that have
not yet been sequenced, and overcomes the major
drawback of a system based purely on molecular data.
Genera within the Polyommatina. In practice, we apply
these criteria in the following manner.
1. We define as genus any lineage older than 5.0 Myr.
2. Between 4.0 and 5.0 Ma we are conservative, in the
sense that we consider a clade to be a genus only if it has
already been described, and do not consider it a genus if
it has not.
3. We lump into another genus any lineage younger
than 4.0 Myr.
Applying this taxonomy-friendly, flexible temporal
scheme to the phylogeny and dating produced the
division of the subtribe Polyommatina into 32 genera
(Table 3). From this classification scheme, one new
genus needs to be described and 39 names can be
regarded as subjective synonyms or valid subgenera. The
further designation of these 39 taxa as either subgenera
or subjective synonyms requires additional data for all
the species that each one represents, which is beyond of
the scope of this paper.
For the 32 established genera, monophyly was statis-
tically supported for the three phylogenetic methods
used, with the sole exception of Kretania, where the
phylogenetic position of K. alcedo was not resolved in
the MP analysis (Table 3).
Composition and phylogenetic relationships of genera
in the Polyommatina: before and after this study.
The nine-marker phylogeny revealed that the subtribe
Polyommatina includes two major clades: the Neotrop-
ical clade (the genera Pseudolucia,Nabokovia,Eldoradi-
na,Itylos,Paralycaeides,Hemiargus,Echinargus,
Cyclargus and Pseudochrysops) and the non-Neotropical
clade (the remaining genera).
The Neotropical clade. Relationships within the Neo-
tropical clade have already been discussed in detail in a
previous publication (Vila et al., 2011). Briefly, the
Neotropical taxa are divided into four well supported
clades. Two of these, probably sister clades, are formed
by Andean, typically high-altitude taxa that occur south
of Central Colombia. These are Eldoradina Balletto,
1993, Nabokovia Hemming, 1960 and Pseudolucia
Nabokov, 1945 on one hand; and Itylos (= Madeleinea
Ba
´lint, 1993) and Paralycaeides Nabokov, 1945 on the
other. The other two clades are formed by lowland taxa,
including all the Caribbean representatives and species
occurring north of Central Colombia, plus a few with
more southern distributions. One clade is formed by
Cyclargus Nabokov, 1945; Echinargus Nabokov, 1945
and Hemiargus Hu
¨bner, 1818; the other by Pseudo-
chrysops Nabokov, 1945. The position of Pseudochrys-
ops with respect to the other three clades is unresolved,
probably due to its early divergence and very long
branch.
The non-Neotropical clade. The non-Neotropical clade
of the subtribe Polyommatina is strongly asymmetrical,
with multiple nested lineages that are discussed below.
ChiladesLuthrodes clade.Within the non-Neotropical
clade, Chilades Moore, [1881] (TS: Papilio lajus Stoll,
[1780]) and Luthrodes Druce, 1895 (TS: Polyommatus
cleotas Gue
´rin-Me
´neville, [1831]) form a clade that is
sister to the rest. The age of divergence between these
two groups is 6.0 Myr, thus we consider them good
genera despite the fact that most recent studies (Bridges,
1988) have lumped them together. Representatives of
both Chilades and Luthrodes have a similar, most likely
plesiomorphic, pattern on the wing underside with the
presence of all the basic elements typical of the non-
Neotropical Polyommatina. However, the male genitalia
in Luthrodes are very distinct—clearly different from
those in other genera—in the shape of the valvae, which
are broad and trapeziform, and in the presence of a
dorsal process at the distal end of the valvae that is
markedly elongated and directed downwards (Bethune-
Baker, 1913; Zhdanko, 1983, 2004; Stekolnikov and
Kuznetzov, 2005). Within Luthrodes, the taxa Edales
Swinhoe, [1910] (TS: Lycaena pandava Horsfield, [1829])
and Lachides Nekrutenko, 1987 (TS: Lycaena galba
Lederer, 1855) are aged less than 4.0 Myr, and conse-
quently should be considered subjective synonyms or
subgenera of Luthrodes. However, this question may be
better assessed after a study including additional species.
Bethune-Baker (1913) studied the male genitalia of
Chilades lajus and showed that, unlike Luthrodes, the
valvae are elongated and have a short dorsal process. In
fact, the genital morphology of Chilades is more similar
to that of Freyeria than to Luthrodes.
178 G. Talavera et al. / Cladistics 29 (2013) 166–192
Freyeria clade. Freyeria Courvoisier, 1920 (TS: Lycaena
trochylus Freyer, 1845) is frequently treated by modern
authors as a subgenus of Chilades (Ba
´lint and Johnson,
1997; Tolman and Lewington, 1997). Valvae in the male
genitalia of Freyeria are elongated and have a short
dorsal process (Zhdanko, 2004), and are generally
similar to those of Chilades. However, molecular data
demonstrate that Freyeria is not closely related to
Chilades and represents a distinct clade that cannot
possibly be subsumed within Chilades as it would result
in a paraphyletic assemblage.
Our analysis includes one specimen of Freyeria from
Turkey (F. trochylus) and one from Australia (F. putli
(Kollar, [1844])). The taxon F. putli has until recently
been considered a subspecies of F. trochylus (Common
and Waterhouse, 1981; Parsons, 1999), but now most
authors treat it as a good species (Ba
´lint and Johnson,
1997; Braby, 2000). In our analysis, F. trochylus and
F. putli appear as sister taxa, and we estimate that they
diverged ca. 3.6 Ma. This is a surprisingly old diver-
gence, and supports the recognition of F. putli as a
distinct species.
Table 3
Posterior probabilities and bootstrap values for monophyly in BI ML MP inferences, ages (mean and stdev), number of species, and larval food
plant families for the 32 genera within the subtribe Polyommatina
Genus Monophyly stability values Genus age (Myr) Number of species Food plant
Polyommatus 100 100 100 4.3 [3.0–5.6] 183 Fabaceae
Neolysandra 100 100 100 4.3 [3.0–5.6] 6 Fabaceae
Lysandra 100 100 100 4.9 [3.4–6.4] 15 Fabaceae
Aricia 100 100 99 5.3 [3.7–6.9] 15 Geraniaceae
Glabroculus 100 100 100 5.1 [3.6–6.7] 2 Limoniaceae
Alpherakya Single specimen 5.1 [3.6–6.7] 5 Crassulaceae
Agriades 100 73 81 4.2 [2.9–5.8] 19 Primulaceae, Saxifragaceae, Ericaceae
Fabaceae
Rimisia Single specimen 4.2 [2.9–5.8] 1 Fabaceae
Cyaniris 100 100 100 4.4 [3.0–5.7] 2 Fabaceae
Eumedonia 100 100 100 4.0 [2.7–5.4] 3 Geraniaceae
Plebejidea Single specimen 4.0 [2.7–5.4] 2 Fabaceae
Maurus Single specimen 4.4 [3.1–5.9] 1 Geraniaceae
Kretania 99 77 4.6 [3.1–6.1] 17 Fabaceae
Afarsia Single specimen 4.6 [3.1–6.1] 9 Fabaceae
Plebejus 100 99 98 4.0 [2.7–5.5] 40 Fabaceae, Elaeagnaceae
Empetraceae
Ericaceae
Pamiria Single specimen 4.0 [2.7–5.5] 7 Unknown
Patricius Single specimen 4.4 [2.9–5.9] 7 Unknown
Rueckbeilia Single specimen 6.9 [4.9–9.0] 2 Fabaceae
Icaricia 96 66 99 5.5 [3.8–7.4] 7 Polygonaceae
Fabaceae
Plebulina Single specimen 5.5 [3.8–7.4] 1 Chenopodiaceae
Freyeria 100 100 100 9.5 [6.8–12.2] 3 Boraginaceae, Phyllanthaceae, Fabaceae
Luthrodes 100 100 100 6.0 [3.9–8.3] 9 Fabaceae
Cycadaceae
Chilades Single specimen 6.0 [3.9–8.3] ca. 12 Rutaceae
Tiliaceae
Pseudolucia 100 100 98 8.1 [5.6–10.7] 46 Fabaceae
Polygonaceae
Portulacaceae
Cuscutaceae
Nabokovia 100 100 100 5.0 [3.2–6.9] 3 Fabaceae
Eldoradina Single specimen 5.0 [3.2–6.9] 2 Unknown
Itylos 99 74 76 4.6 [3.1–6.3] 24 Fabaceae
Paralycaeides 100 100 100 4.6 [3.1–6.3] 3 Fabaceae
Hemiargus 100 100 100 6.1 [4.2–8.1] ca. 5 Fabaceae
Cucurbitaceae
Oxalidaceae
Echinargus Single specimen 6.1 [4.2–8.1] 1 Fabaceae
Cyclargus Single specimen 7.0 [4.9–9.3] 7 Asteraceae
Fabaceae
Malpighiaceae
Sapindaceae
Pseudochrysops Single specimen 11.4 [8.2–14.7] 1 Unknown
179G. Talavera et al. / Cladistics 29 (2013) 166–192
IcariciaPlebulina clade.In the original descriptions of
the genera Icaricia Nabokov, [1945] (TS: Lycaena
icarioides Boisduval, 1852) and Plebulina Nabokov,
[1945] (TS: Lycaena emigdionis Grinnell, 1905), the
author clearly indicated morphological characters that
distinguished these genera from all other lycenids. In
particular, Nabokov noted that Plebulina remarkably
amalgamates the form of aedeagus similarly to Plebejus,
with uncus, subunci, and valvae similar in shape to those
found in Albulina. On the other hand, Icaricia remark-
ably combines a wing pattern similar to that of Plebejus
with a shape of aedeagus similar to that found in Aricia
(Nabokov, 1945). Since their description, however, the
genera Icaricia and Plebulina generally have been treated
as junior subjective synonyms, or as subgenera of either
Aricia Reichenbach, 1817 or Plebejus Kluk, 1780 (Scott,
1986; Ba
´lint and Johnson, 1997; Gorbunov, 2001;
Brock and Kaufman, 2003; Opler and Warren, 2004).
In all our analyses, the taxa within Icaricia and
Plebulina, as well as the taxon Lycaena saepiolus
Boisduval, 1852, form an exclusively Nearctic clade that
is sister to all the rest of the Holarctic taxa. Such a
topology in the phylogeny is unexpected given modern
taxonomic treatments of these groups, and implies that
Icarica and Plebulina cannot possibly be included within
Plebejus or Aricia. This strongly supported result
confirms that of Vila et al. (2011), who showed that
this clade is the result of a relatively old colonization of
the New World that occurred ca. 9.3 Ma. The age of
divergence of Icaricia (including the taxon I. saepiolus)
from the Plebulina lineage is 5.5 Myr. As a consequence,
we maintain the monotypic Plebulina as a separate
genus, a decision reinforced by the fact that P. emigdi-
onis Grinnel, 1905 feeds on a different host-plant family
(Chenopodiaceae) from the Icaricia taxa (Fabaceae and
Polygonaceae), and by certain peculiarities of its larval
morphology (Ballmer and Pratt, 1988). Interestingly,
divergence ages within the Icaricia lineage are fairly old,
reaching 4.8 Myr for the I. acmon–I. shasta versus
I. icarioides–I. saepiolus split, which still falls within
the 4.0–5.0 Myr genus timeframe. Since no separate
genus name has ever been proposed for the I. acmon–
I. shasta clade, we conservatively retain Icaricia as a
single unit.
The genus Rueckbeilia (= ‘‘Vacciniina’’ fergana
clade).The next well supported lineage found in our
analysis is represented by a single species traditionally
known as Vacciniina fergana. This species is recovered
as sister to the rest of the Holarctic taxa, except for the
IcariciaPlebulina clade. This result is unexpected (but
see Kandul et al., 2004; Lukhtanov et al., 2009) as the
external morphology of V. fergana is extremely similar
to V. optilete, the type-species of Vacciniina. This
position of V. fergana in the phylogeny is strongly
supported in all the analyses and thus cannot be
considered an artifact. The deep divergence of the
V. fergana lineage (6.9 Ma) indicates that it should be
treated as a distinct genus, which we describe in the
Appendix 1 under the name Rueckbeilia gen. nov.
Interestingly, the isolated systematic position of V. ferg-
ana was not apparent in a detailed morphological study
of this species (Stekolnikov, 2010). In fact, V. fergana
exhibits a combination of primitive male genitalic
characters that are found in some other taxa (Stekolni-
kov, 2010), and wing patterns that may represent a
plesiomorphic condition in Rueckbeilia,Glabroculus and
Afarsia +Kretania, but independently evolved in Agri-
ades optilete (Fig. 4). A more detailed description of
Rueckbeilia is given in the Appendix 1.
Patricius + (Pamiria + Plebejus) clade (lineage of Ple-
bejus sensu lato).The grouping Patricius +(Pami-
ria +Plebejus) is recovered as a well supported clade in
our phylogeny. This result is not trivial, as Patricius and
Pamiria have usually been regarded as closely related to
Albulina (Ba
´lint and Johnson, 1997). However, the close
relationship of Patricius (TS: Lycaena lucifera Stau-
dinger, 1867), Pamiria (TS: Lycaena chrysopis Grum-
Grshimailo, 1888) and Plebejus (TS: Papilio argus
Linnaeus, 1758) had already been recognized by
Zhdanko (2004), who noted that these genera shared
similar Plebejus-like male genitalia. Within this clade,
the genus Patricius is sister to the rest (divergence age
4.4 Myr), while Pamiria and Plebejus diverged 4.0 Ma.
In all our analyses, the studied species of Lycaeides
Hu
¨bner [1819] (TS: Papilio argyrognomon Bergstrasser
[1779], also includes idas,melissa and anna) form a clade
that is sister to Plebejus argus, but its recent age
(3.1 Myr) recommends the inclusion of Lycaeides within
Plebejus. Noticeably, Nearctic Lycaeides representatives
appear as polyphyletic, with unexpected, yet strongly
supported, sister relationships between Old and New
World taxa. This result is similar to that obtained
independently by other researchers (Nice et al., 2005;
Gompert et al., 2008; Vila et al., 2011) and deserves
further analysis. A number of authors consider Agri-
ades,Alpherakya,Vacciniina,Plebejides and Plebejidea
as synonyms or subgenera of Plebejus (Ba
´lint and
Johnson, 1997; Gorbunov, 2001), but our results show
that these taxa are more closely related to Aricia and
Polyommatus than they are to Plebejus. Thus the
prevalent use of Plebejus as a supergenus is not possible
according to the recovered topology.
Polyommatus sensu lato clade.The rest of the Poly-
ommatina taxa form a large clade consisting of 14
genera from Alpherakya to Polyommatus (Fig. 1). It
corresponds to Polyommatus sensu Zhdanko, 1983 (but
not to Polyommatus sensu Zhdanko, 2004) and can be
defined by characters of male genitalia similar to those
of Polyommatus sensu stricto (Zhdanko, 1983, 2004).
180 G. Talavera et al. / Cladistics 29 (2013) 166–192
However, these genitalic characters may not constitute a
true synapomorphy. Stekolnikov (2010) demonstrated a
degree of heterogeneity in the male genitalia of this
group, and a similar type of genitalia was found in
Rueckbeilia fergana, which is not closely related. While
the Polyommatus sensu lato clade is strongly supported,
it is formed by several subclades that are older than
4 Myr. The evolutionary relationships among these
supported subclades are in some cases unresolved, and
we will discuss each in the following paragraphs.
The genus Alpherakya. Alpherakya Zhdanko, 1994 (TS:
Lycaena sarta Alpheraky, 1881) is recovered as sister to
Glabroculus, although this relationship is not well
supported. It should also be noted that the wing
patterns and food plants of these two taxa are different
(Table 3). The morphology of this genus is characterized
by a unique combination of traits that make its
identification unmistakable. Alpherakya differs from all
genera of the Polyommatus sensu lato clade, except for
Lysandra, in having chequered wing fringes. It differs
from Lysandra in having hairs on the eyes that are
scarce and short, whereas in Lysandra they are long and
dense. In male genitalia, the structure of the valvae is
also diagnostic: valvae are comparatively short and
broad, with a robust sclerotized inner fold, with a spade-
shaped dorsal element in the apex and sclerotized
ventral elements. Alpherakya can be separated from
the taxa in the Patricius +(Pamiria +Plebejus) clade
by the wide uncus (Zhdanko, 2004) and by the structure
of valvae (Fig. 4). The larval food plants of the
Alpherakya species are also peculiar: they feed on
Crassulaceae (Zhdanko, 1997), whereas most species
and genera of the subtribe Polyommatina are associated
with Leguminosae or Geraniaceae. Ba
´lint and Johnson
(1997) considered Alpherakya as part of the genus
Plebejus. However, our analysis, like the morphological
analysis by Zhdanko (2004), does not support this
hypothesis and demonstrates that these two genera are
phylogenetically distant.
Glabroculus clade.Zhdanko (2004) synonymized Glab-
roculus Lvovsky, 1993 (TS: Lycaena cyane Eversmann,
1837) with Plebejidea, and considered Elviria (TS:
Lycaena elvira Eversmann, 1854) to be a subgenus of
Plebejidea.Ba
´lint and Johnson (1997) considered Glab-
roculus (= cyane-group) as part of the genus Polyomm-
atus sensu lato. Our data support none of these
hypotheses. We show that neither Plebejidea nor Poly-
ommatus is closely related to Glabroculus. Instead,
Glabroculus appears as a sister to Alpherakya, although
with low statistical support.
Morphologically, Glabroculus differs from Polyomm-
atus by hairs on the eyes that are scarce and short (in
Polyommatus they are long and dense) and by the
presence of metallic marginal spots on the underside of
hind wings. Glabroculus differs from the phylogeneti-
cally most closely related genus, Alpherakya, in having
unchequered wing fringes. Moreover, the food plants of
Glabroculus and Alpherakya are different (Table 3). The
taxon E. elvira (the type-species of the nominal genus
Elviria) was recovered as a sister to G. cyane, and the
time of their divergence was estimated as ca. 2.0 Mya.
Therefore Elviria can be considered a synonym of
Glabroculus.
Aricia clade.The taxa representing Aricia (TS: Papilio
agestis Denis & Schiffermu
¨ller, 1775) and aratxerxes),
Umpria (TS: Lycaena chinensis Murrey, 1874),Pseudo-
aricia (TS: Polyommatus nicias Meigen, 1829) and
Ultraaricia (TS: Lycaena anteros Freyer, 1839; includes
the studied species crassipuncta and vandarbani) form a
strongly supported clade. Since the divergences among
them are younger than 4 Myr, the three latter taxa are
subsumed within Aricia. The position of Aricia within
the Polyommatini has been a subject of much discus-
sion. Ba
´lint and Johnson (1997) considered Aricia as
closely related to the Neotropical taxon Madeleinea.
Zhdanko (2004) also considered Aricia as one of the
most basal within the Polyommatus section. In contrast,
Stekolnikov (2010) found it to represent a young lineage
closely related to Polyommatus. Our molecular data
support the latter hypothesis, although the position of
Aricia within the Polyommatus sensu lato clade is
unresolved. Indeed, we recover Aricia as sister to
Alpherakya +Glabroculus, but with low support.
Morphologically the genus is quite distinct. In the
male genitalia, the aedeagus is lanceolate, with caulis
developed, and entirely sclerotized, which is not
observed in other taxa of the subtribe (Zhdanko,
2004). Among external characters, the naked eyes and
absence of metallic spots on the underside of hindwings
are characteristic, although they are not unique within
the subtribe.
The genus Afarsia. (TS: Cupido hyrcana Lederer,
1869—an invalid name; the valid synonym is Cupido
morgiana Kirby, 1871). The taxon C. morgiana was
recognized as a distinctive entity by Zhdanko (1992,
2004) and Ba
´lint and Johnson (1997), but its relation-
ships with other taxa have never been properly docu-
mented. Ba
´lint and Johnson (1997) placed it in the same
group as Patricius,Pamiria,Plebejidea,Vacciniina and
Albulina. In our reconstruction, it is recovered as sister
to Kretania, but the support for this relationship is low.
Its rather deep divergence (4.6 Myr) suggests that it
should be treated as an independent genus. The genus
name Farsia Zhdanko, 1992, for which C. morgiana is
the type species, was preoccupied and the new name
Afarsia Korb and Bolshakov, 2011 (= Farsia Zhdanko,
1992; nec Farsia Amsel, 1961) has recently been
proposed as replacement (Korb and Bolshakov, 2011).
181G. Talavera et al. / Cladistics 29 (2013) 166–192
The morphology of the male genitalia of the genus
Afarsia is similar to Kretania sensu lato (see below), but
these two taxa are distinct in wing pattern: in Afarsia a
discal spot on the fore wing upper side is always present
and usually strongly enlarged, and one of the marginal
metallic spots of the hind wing underside is enlarged.
These characters of the wing pattern are also found in
the genus Albulina (that was the reason why some
authors placed Afarsia within or close to Albulina—see
above). However, male genitalia in Afarsia are consid-
erably different from those in Albulina, both in the
structure of uncus, which is basally narrow with long
slender arms, and in the shape of the valvae, which have
a characteristically concave dorsal margin (Zhdanko,
2004).
Kretania clade.In all our analyses, the taxa within
Plebejides (TS: Lycaena pylaon Fischer von Waldheim,
1832 and P. zephyrinus)andKretania sensu stricto (TS:
Lycaena psylorita Freyer, 1845, includes the studied
species K. eurypilus and K. zamotajlovi), as well as the
species V. alcedo, form a distinct, statistically well
supported clade in ML and BI analyses that originated
4.6 Mya and should be considered as a genus. Within this
genus, the species V. alcedo appears as sister to the rest,
although the position of this taxon is unresolved in the
MP analysis. The statistical support for the subclade
Kretania s.s. + Plebejides is very high (100 100 100)
and the time of divergence of this subclade is quite recent
(ca. 1.9 Mya). The close relationship of Kretania s.s. and
Plebejides was first suggested by Wiemers (2003) based
on the molecular analysis of COI barcodes and nuclear
ITS2. Interestingly, the close relationship between V. al-
cedo,Kretania s.s. and Plebejides has never been recog-
nized by morphologists, who usually consider them as
members of different, not closely related groups: Plebej-
ides as a member of the Plebejus lineage (Zhdanko, 1983;
Ba
´lint and Johnson, 1997), Kretania as a member of the
Polyommatus lineage (Ba
´lint and Johnson, 1997), and
the taxon V. alcedo as a species of Vacciniina (Ba
´lint and
Johnson, 1997). Nevertheless, these butterflies are fairly
similar phenotypically. In fact, species of Kretania s.s.
differ from Plebejides and V. alcedo largely in discol-
oured (brown) upper wings in males, but this is a labile
character that has low value in genus-level taxonomy, as
it seems to have evolved independently numerous times
in the evolution of the Polyommatina (Ba
´lint and
Johnson, 1997; Lukhtanov et al., 2005). As a result, we
propose the following new combinations: Kretania
alcedo comb. nov.,Kretania pylaon comb. nov.,Kretania
zephyrinus comb. nov.
The structure of the valvae in Kretania sensu lato
(including Plebejides and the taxon K. alcedo) is typical
of the genera Polyommatus or Aricia (Stekolnikov, 2010)
(but not typical of the genus Plebejus as suggested by
Zhdanko, 2004), the uncus is narrow (Zhdanko, 2004)
and the wing pattern is extremely similar to that found
in Plebejus. The combination of these morphological
characters makes the genus Kretania sensu lato quite
distinct.
The genus Maurus. The north African endemic species
Lycaena vogelii Oberthu
¨r, 1920 has been included either
within Plebejus or in the monotypic genus Maurus
Ba
´lint, [1992]. Our analysis recovers it as sister to the
PlebejideaEumedonia clade with low support, but its
age (4.4 Myr) is sufficient to maintain the genus Maurus.
The morphology of the genitalia of M. vogelii has been
described as close to that of Plebejus (Zhdanko, 2004).
The external morphology of the genus is distinctive and
can be recognized by the combination of chequered wing
fringes and strongly enlarged discal spot on the fore
wing upper side.
PlebejideaEumedonia clade.The genus Plebejidea (TS:
Lycaena loewii Zeller, 1847) is usually considered to be
close to Glabroculus (Tuzov et al., 2000; Zhdanko,
2004), Polyommatus (Ba
´lint, 1991), or Albulina (Ba
´lint
and Johnson, 1997). Our data support none of these
taxonomic hypotheses. Instead, in our reconstruction,
Plebejidea appears as sister to Eumedonia with high
statistical support. This result is unexpected, as repre-
sentatives of Plebejidea and Eumedonia clearly differ in
wing pattern and coloration and also in ecology: the
species of Eumedonia inhabit humid biotopes and their
larval food plants are species of Geraniaceae, whereas
the species of Plebejidea inhabit very dry semi-desert
biotopes and their larval food plants are xerophilous
species of Astragalus (Fabaceae). The morphology of
the male genitalia in Plebejidea is similar to that of
Glabroculus (Zhdanko, 2004), but differs by a noticeable
basal sclerotization of the subcostal groove of the valvae
(Stekolnikov, 2010).
The genus Eumedonia. (TS: Papilio eumedon Esper,
[1780]) has been considered to be close to Aricia (Ba
´lint
and Johnson, 1997; Tuzov et al., 2000) in part because
they share the same larval food plants (Geraniaceae).
However, our results do not support this close relation-
ship, and differences in the structure of the uncus in the
male genitalia (Zhdanko, 2004) also suggest that these
genera are not closely related. In fact, the genus
Eumedonia is morphologically quite distinct. It shares
a similar form of the valvae in male genitalia with
Plebejidea, the phylogenetically most closely related
genus, as well as with the more distant Polyommatus,
Lysandra,Neolysandra,Aricia,Glabroculus and Al-
pherakya, but differs from them in the narrow uncus
and hairless eyes. The aedeagus in Eumedonia is com-
paratively slender and more pointed, resembling that in
Agriades (Zhdanko, 2004), yet the wing patterns are
very different between Eumedonia and Agriades.
182 G. Talavera et al. / Cladistics 29 (2013) 166–192
The genus Cyaniris. The genus Cyaniris (TS: Zephyrus
argianus Dalman, 1816, now regarded as a synonym of
Papilio semiargus Rottemburg, 1775) is often considered
to be close to Polyommatus s.s. (Hesselbarth et al., 1995;
Ba
´lint and Johnson, 1997), but this relationship was
questioned on the basis of morphological (Zhdanko,
2004) and molecular analyses (Wiemers et al., 2010).
Indeed, our data indicate that Cyaniris is not closely
related to Polyommatus s.s. Instead, it forms a clade
together with Rimisia and Agriades sensu lato, although
the support for this relationship is not high. The age of
divergence of the Cyaniris lineage (4.4 Myr) is sufficient
to maintain it as an independent genus.
Cyaniris differs from Polyommatus,Lysandra,Neoly-
sandra,Aricia,Glabroculus,Alpherakya and Plebejidea
in having a narrow, nearly pointed uncus. It differs from
other taxa that also have narrow uncus in the presence
of hairs densely covering the eyes and by having a longer
aedeagus (Zhdanko, 2004). Additionally, representatives
of the genus have no marginal and submarginal pattern
on the wing underside. The combination of these
characters is characteristic for the genus Cyaniris.
The genus Rimisia. The monotypic Central Asian genus
Rimisia (TS: Lycaena miris Staudinger, 1881) has been
considered to be close to Glabroculus (Ba
´lint and
Johnson, 1997; Tuzov et al., 2000), with which it shares
a similar pattern on the underside of the wings. This
hypothesis is not supported by our data, since Rimisia is
recovered as sister to Agriades with a divergence of more
than 4 Myr. The genus Rimisia displays an unusual
combination of morphological characters: valvae in the
male genitalia similar to those of the species Polyomm-
atus icarus, short and S-shaped aedeagus, naked eyes
and peculiar female genitalia with small papillae anales
(Zhdanko, 2004). Rimisia miris is considered to have no
metallic marginal spots on the hind wings (Zhdanko,
2004), but our analysis of the morphology revealed that
the species is variable with respect to this character and
some specimens bear metallic scales on the marginal
spots.
Agriades clade.According to our results, the genus
Agriades (TS: Papilio glandon Prunner, 1798) originated
4.2 Mya and includes three monophyletic lineages that
may be considered as subgenera: Albulina (orbitulus)
(originated 3.6 Mya), Vacciniina s.s. (optilete) and
Agriades s.s. (glandon,pheretiades,podarce and pyrenai-
cus) (the latter two split 3.2 Mya). These three taxa are
often considered to be distinct genera (e.g. Higgins,
1975), and they indeed differ in their wing patterns
(Fig. 5) and larval food plants (Table 3). The close
relationship between Albulina and Vacciniina was rec-
ognized by Ba
´lint and Johnson (1997). Our analysis
strongly supports the grouping of Agriades s.s., Albulina
and Vacciniina s.s. Within this group, Agriades s.s. and
Vacciniina s.s. are sister taxa and Albulina is sister to the
rest. As our study resulted in the fusion of the taxa
Agriades s.s., Albulina and Vacciniina s.s in one genus,
the following new combinations result: Agriades optilete
comb. nov, Agriades orbitulus comb. nov.
Lysandra + (Neolysandra +Polyommatus) clade.This
clade is recovered with a high support in our analysis,
and it is estimated to have diverged ca. 5.7 Mya. Within
this clade, three genera—Lysandra,Neolysandra and
Polyommatus—are recognized in accordance with the
criteria discussed above.
Lysandra clade.The genus Lysandra (TS: Papilio coridon
Poda, 1761) is monophyletic and sister to the clade
Neolysandra +Polyommatus with good support. The
most characteristic morphological feature of the genus is
the clearly chequered wing fringes. This character is not
exclusive within the subtribe Polyommatina, and it is
found in the distantly related genera Alpherakya,Maurus
and Grumiana, as well as in some genera of the Neotrop-
ical clade. The hypothesis that Lysandra is a synonym of
Meleageria (which includes the species daphnis and
marcida) (Hesselbarth et al., 1995) is not supported by
our phylogeny (see also Wiemers et al., 2010).
Neolysandra clade.In our analysis, the genus Neolysan-
dra (TS: Lycaena diana Miller, 1912) emerges as a well
supported lineage that is a sister to Polyommatus.
Morphologically Neolysandra differs from other genera
by the markedly wide and elliptical uncus. Moreover, it
differs from the most similar genera Lysandra and
Polyommatus in having short and scarce hairs covering
the eyes and in displaying a reduced marginal and
submarginal pattern on the wing underside (Zhdanko,
2004). In the molecular reconstruction made by
Wiemers et al. (2010), Neolysandra was recovered as a
polyphyletic taxon. Several reasons might explain this:
the taxon sampling (the type species N. diana was not
included), lack of resolution (the phylogeny was based
on two relatively short sequences), and incomplete
outgroup sampling (only the phylogenetically distant
taxa Cyaniris semiargus and Freyeria trochilus were used
to root the tree). What we consider Neolysandra
(including the taxa diana and coelestina) corresponds
to WiemersÕNeolysandra group I.
Polyommatus clade.In our analysis, the genus Poly-
ommatus (TS: Papilio icarus Rottemburg, 1775)
emerged as a distinct lineage about 4.3 Mya. It is
composed of taxa sometimes included in the gen-
era subgenera Actisia Koc¸ ak & Kemal, 2001 (TS:
Lycaena actis Herrich-Scha
¨ffer, 1851—a junior syno-
nym, the valid synonym is Lycaena atys (Gerhard,
1851); Admetusia Koc¸ ak & Seven, 1998 (TS: Papilio
admetus Esper, 1783); Agrodiaetus Hu
¨bner, 1822
183G. Talavera et al. / Cladistics 29 (2013) 166–192
(= Hirsutina Tutt, [1909]) (TS: Papilio damon Denis &
Schiffermu
¨ller, 1775); Antidolus Koc¸ ak & Kemal, 2001
(TS: Papilio dolus var. antidolus Rebel, 1901); Bryna
Evans, 1912 (TS: Lycaena stoliczkana Felder & Felder,
1865); Damaia Koc¸ ak & Kemal, 2001 (TS: Lycaena
dama Staudinger, 1892); Meleageria De Sagarra, 1925
(TS: Papilio daphnis Esper, 1778); Musa Koc¸ ak &
Kemal, 2001 (TS: Polyommatus musa Koc¸ ak & Hos-
seinpour, 1996); Paragrodiaetus Rose & Schurian, 1977
(TS: Lycaena glaucias Lederer, 1870); Peileia Koc¸ ak &
Kemal, 2001 (TS: Polyommatus peilei Bethune-Baker,
1921); Phyllisia Koc¸ ak & Kemal, 2001 (TS: Papilio
damon var. phyllis Christoph, 1877); Plebicula Higgins,
1969 (TS: Papilio argester Bergtra
¨sser, 1779); Poly-
ommatus Latreille, 1804 (TS: Papilio icarus
Rottemburg, 1775); Sublysandra Koc¸ ak, 1977 (TS:
Lycaena candalus Herrich-Scha
¨ffer, 1851); Thersitesia
Koc¸ ak & Seven, 1998 (TS: Lycaena thersites Cantener,
1834); Transcaspius Koc¸ ak & Kemal, 2001 (TS: Lyca-
ena kindermanni var. transcaspica Heyne, 1895); and
Xerxesia Koc¸ ak & Kemal, 2001 (TS: Lycaena damone
var. xerxes Staudinger, 1899). Several of these taxa are
recovered as monophyletic, but no subclade is older
than 4 Myr. Thus, according to our criteria, they
should not be treated as genera. The composition and
relationships obtained are notably similar to those
obtained by Zhdanko (2004) based on a morphological
analysis (e.g. Lysandra and Neolysandra are separate
genera), but differ in some details (e.g. in the position
of Agrodiaetus). Wiemers et al. (2010) specifically
addressed relationships in this genus based on molec-
ular data from two genetic markers and a different set
of outgroup taxa. Deeper relationships are frequently
not supported in their study and do not always match
(a)
(b)
(g)
(h)
(c)
(d)
(i)
(j)
(e)
(f)
(k)
(l)
Fig. 5. Representative taxa of the genus Agriades. Similarly to other species-rich genera in the subtribe Polyommatina, despite their monophyly and
genetic similarities, the genus Agriades is morphologically quite diverse with respect to both wing upper side and underside colours and patterns. (a,b)
Agriades orbitulus; (c,d) Agriades glandon; (e,f) Agriades pheretiades; (g,h) Agriades pyrenaicus; (i,j) Agriades podarce; (k,l) Agriades optilete.
184 G. Talavera et al. / Cladistics 29 (2013) 166–192
those obtained here. The most characteristic morpho-
logical features of the genus are the marked downward
expansion of the ventral margin of the uncus and the
presence of all the basic elements of the wing pattern
(Zhdanko, 2004). Polyommatus differs from Lysandra
in having white or grey (not chequered) fringes. It
differs from Neolysandra in the presence of long hairs
densely covering the eyes.
One of the subclades in our analysis is formed by the
taxa traditionally included in Agrodiaetus (P. damocles,
P. ripartii,P. surakovi and P. damon)andParagrodia-
etus (P. glaucias and P. erschoffii), thus our results
confirm previous results showing that Agrodiaetus is a
monophyletic entity that includes Paragrodiaetus (Kan-
dul et al., 2004, 2007; Wiemers et al., 2010). Morpho-
logically, the subgenus Agrodiaetus differs from other
genera and subgenera of the subtribe Polyommatina in
two autapomorphic characters of the male genitalia:
distal extremity of aedeagus pronouncedly swollen
(Zhdanko, 1983) and uncus markedly constricted dor-
soventrally (Zhdanko, 2004). Our data also strongly
support that the taxon P. stempfferi is sister to the
Agrodiaetus clade, and that P. escheri is sister to the
P. stempfferi +Agrodiaetus clade. The taxa P. myrrha
and P. cornelia, representative of the taxon Sublysandra,
form another subgroup of Polyommatus that is recov-
ered with low support and with unresolved position.
Sublysandra is usually considered to be a subgenus of
Polyommatus (Ba
´lint and Johnson, 1997; Zhdanko,
2004; Wiemers et al., 2010) and is morphologically
similar to Polyommatus s.s. The subclade representing
Meleageria (P. daphnis and P. marcida) is recovered
with good support as sister to the species P. amandus.
The close relationship between P. amandus and P. daph-
nis +P. marcida is surprising and has not been pro-
posed previously.
The last supported subclade is formed by Polyomm-
atus s.s. + (Plebicula +Thersitesia). The sister rela-
tionship of the taxa representing Plebicula (P. dorylas
and P. nivescesns)andThersitesia(P. thersites) was first
recovered by Wiemers et al. (2010). Polyommatus s.s.
was recovered as monophyletic with high support.
Within this clade, the Central Asian species P. hunza
and P.venus (which sometimes have been placed
together in the genus Bryna) form a clade that is sister
to the rest (erotides and icarus). This Central Asian
subclade was also recovered by Wiemers et al. (2010).
Conclusion
A multilocus molecular phylogeny has clarified rela-
tionships within the Polyommatina, and molecular age
estimates have helped to establish criteria specific for the
higher-level taxonomy of this group. Each of the
resulting clades that we designate to be a genus displays
a distinguishing combination of morphological charac-
ters, but most of these characters are not unique to a
single genus. The high evolutionary lability of many
morphological characters traditionally used to infer
relationships in this lineage of butterflies (metallic spots
in the hind wing underside, blue versus brown male wing
colour, shape of the valvae, membranous ventral fold in
the inner part of valvae, marked discal spot on the fore
wing upper side, number of segments in the antennal
club, pilosity in the eyes, presence of small tails in the
hind wing, etc.) is apparent, and explains why the
taxonomy of the Polyommatina has been so controver-
sial. Based on our phylogenetic results and the criteria
outlined above, we propose the following systematic
arrangement for the subtribe Polyommatina (in paren-
theses we list objective and subjective synonyms for the
generic names, objective synonyms are indicated by the
sign ‘‘=’’; in brackets we provide a tentative list of
species for each genus in alphabetical order; likely
synonym