DNA Barcodes As a Tool in Biodiversity Research: Testing Pre-Existing Taxonomic Hypotheses in Delphic Apollo Butterflies (Lepidoptera, Papilionidae)

Systematics and Biodiversity

ISSN: 1477-2000 (Print) 1478-0933 (Online) Journal homepage: http://www.tandfonline.com/loi/tsab20

DNA barcodes as a tool in biodiversity research: testing pre-existing taxonomic hypotheses in Delphic Apollo butterflies (Lepidoptera, Papilionidae)

V. A. Lukhtanov, A. Sourakov & E. V. Zakharov

To cite this article: V. A. Lukhtanov, A. Sourakov & E. V. Zakharov (2016) DNA barcodes as a tool in biodiversity research: testing pre-existing taxonomic hypotheses in Delphic Apollo butterflies (Lepidoptera, Papilionidae), Systematics and Biodiversity, 14:6, 599-613, DOI: 10.1080/14772000.2016.1203371

To link to this article: http://dx.doi.org/10.1080/14772000.2016.1203371

View supplementary material Published online: 08 Jul 2016.

Submit your article to this journal Article views: 65

View related articles View Crossmark data

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=tsab20

Download by: [University of Guelph Library] Date: 16 January 2017, At: 11:07 Systematics and Biodiversity (2016), 14(6): 599À613

Research Article DNA barcodes as a tool in biodiversity research: testing pre-existing taxonomic hypotheses in Delphic Apollo butterﬂies (Lepidoptera, Papilionidae)

V. A. LUKHTANOV1,2,3, A. SOURAKOV3 & E. V. ZAKHAROV4 1Department of Karyosystematics, Universitetskaya nab. 1, 199034 St. Petersburg, Russia 2Department of Entomology, St. Petersburg State University, Universitetskaya nab. 7/9, 199034 St. Petersburg, Russia 3McGuire Center for Lepidoptera and Biodiversity, Florida Museum of Natural History, University of Florida, Gainesville, Florida 32611, USA 4Biodiversity Institute of Ontario, University of Guelph, Guelph, Ontario, Canada N1G 2W1 (Received 10 June 2015; accepted 26 April 2016)

Numerous studies have demonstrated that DNA barcoding is an effective tool for detecting DNA clusters, which can be viewed as operational taxonomic units (OTUs), useful for biodiversity research. Frequently, the OTUs in these studies remained unnamed, not connected with pre-existing taxonomic hypotheses, and thus did not really contribute to feasible estimation of species number and adjustment of species boundaries. For the majority of organisms, taxonomy is very complicated with numerous, often contradictory interpretations of the same characters, which may result in several competing checklists using different specific and subspecific names to describe the same sets of populations. The highly species-rich genus Parnassius (Lepidoptera: Papilionidae) is but one example, such as several mutually exclusive taxonomic systems have been suggested to describe the phenotypic diversity found among its populations. Here we provide an explicit flow chart describing how the DNA barcodes can be combined with the existing knowledge of morphology- based taxonomy and geography (sympatry versus allopatry) of the studied populations in order to support, reject or modify the pre-existing taxonomic hypotheses. We then apply this flow chart to reorganize the taxa within the Parnassius delphius species group, solving long-standing taxonomic problems. Key words: biodiversity, COI, DNA barcoding, insects, Lepidoptera, Parnassius, taxonomy

Introduction & Wheeler, 2005). With some limitations, modern high- throughput sequencing technologies already allow the anal- Given the poor level of taxonomic knowledge of many ysis of multilocus sequence data, even using DNA groups of organisms, any given species’ boundaries extracted from old museum specimens (Bi, Linderoth, Van- should normally be considered as an evolutionary hypoth- derpool, Nielsen & Moritz, 2013; Guschanski et al., 2013). esis (Hey, Waples, Arnold, Butlin, & Harrison, 2003). In It is also obvious that these technologies will become more the course of taxonomic research, this hypothesis can sensitive and powerful in the near future. However, hun- undergo changes: it can be rejected (the species’ name dreds of thousands of organisms (e.g., most insects) require then would become invalid), supported, or modiﬁed (if alpha taxonomy, i.e., detection of new species, describing, research showed that previously established species’ and naming them. Only after this is accomplished, can a boundaries need to be reconsidered). comprehensive analysis of species boundaries be conducted Ideally, testing any given hypotheses should include (e.g., Huemer & Mutanen, 2015). multiple sources of evidence. In case of taxonomic In groups that are large and are poorly studied, a good research, the evidence can come from morphology, ecol- place to start is massive single-locus sequencing studies, ogy, molecular or karyological studies, etc. (Knowles & such as DNA-barcoding research (Hebert, Cywinska, Carstens, 2007; Rubinoff & Holland, 2005; Talavera, Ball, & deWaard, 2003). DNA barcoding is based on Lukhtanov, Rieppel, Pierce, & Vila, 2013; Will, Mishler, comparative analysis of short, standardized gene regions (Hebert, Penton, Burns, Janzen, & Hallwachs, 2004a). Correspondence to: V.A. Lukhtanov. E-mail: [email protected]

The method has become universally accepted as a useful (1) Two or more sympatric clusters of individuals tool for the identification of specimens that correspond to should be classified as members of different species if a a known species and for detecting DNA clusters. The lat- presence of distinct DNA barcode gap between them cor- ter can be viewed as operational taxonomic units (OTUs) responds to a morphological hiatus; in other words, if useful for consequent biodiversity research (Barrett & there is a concordance between morphological traits (at Hebert, 2005; Hebert, Stoeckle, Zemlak, & Francis, least one trait should be analysed) and DNA barcodes. 2004b; Janzen, Hajibabaei, & Burns, 2005; Kekkonen & Two or more distinct DNA barcode clusters found in Hebert, 2014; Lukhtanov, Shapoval, & Dantchenko, sympathry may represent different species; alternatively, 2014; Lukhtanov, Sourakov, Zakharov, & Hebert, 2009; they may represent variants of intrapopulational polymor- Smith, Wood, Janzen, Hallwachs, & Hebert, 2007; Soura- phism. To distinguish between interspecific and intrapo- kov & Zakharov, 2011; Talavera, Dinca, & Vila, 2013; pulational variations, we suggest using a genotypic cluster Vershinina & Lukhtanov, 2010; Vila, Lukhtanov, Tala- approach (Mallet, 2001, 2006; Mallet & Willmott, 2003), vera, Gil-T, & Pierce, 2010). However, with few excep- in which a bimodal distribution of genetic and morpholog- tions (e.g., Brower, 2006, 2010; Hausmann, Sciarretta, & ical characters of specimens in an area is considered as Parisi, 2016; Huemer & Mutanen, 2015; Sourakov et al. evidence of the co-existence of two species. Because gene 2015; Sourakov & Zakharov, 2011) in these studies the flow between members of different biological species is OTUs remained unnamed and not connected with pre- normally either absent or weak (Coyne & Orr, 2004) even existing taxonomy and nomenclature. Flow charts have in their contact zone, a non-random association (i.e., link- been suggested for DNA barcode-based taxonomic analy- age disequilibrium) of at least two unlinked alleles typical sis of super-diverse and insufficiently studied groups of of each group can be expected. Conversely, when two organisms (Butcher, Smith, Sharkey, & Quicke, 2012; groups of individuals represent variants of intraspecific Puillandre et al., 2012; Riedel, Sagata, Suhardjono, polymorphism, the unrestricted gene flow and respec- T€anzler, & Balke, 2013). The main aim of these flow tively a random combination of these markers can be charts is discovering and describing the biodiversity. expected. While association of several morphological In better studied groups, such as butterflies, the taxo- characters may be due to physical linkage between nomic landscape may be very complicated with numer- markers located on the same chromosome, if mitochon- ous, often contradictory interpretations of the same drial and nuclear gene markers are used (with cytoplasmic characters, which may result in several competing check- and chromosomal inheritance, respectively), it strongly lists using different specific and subspecific names to reinforces any conclusions based on bimodal distribution describe the same sets of populations. In order to result in of characters (Lukhtanov & Shapoval, 2008; Lukhtanov, accurate estimation of species diversity and species Shapoval, & Dantchenko, 2008). Mitochondrial DNA boundaries, the results of DNA barcode analyses should characters (e.g., COI barcodes) in combination with mor- be incorporated in these pre-existing taxonomic hypothe- phological traits represent a good tool for the analysis ses, and to our knowledge nobody has explicitly described because mitochondrial genes do not code any morphologi- how this synthesis can be reached. cal traits and are not physically linked with nuclear genes Here we provide a scheme describing how the DNA that are responsible for formation of morphological char- barcodes can be combined with the previous information acters (Lukhtanov, Dantchenko, Vishnevskaya, & Saifit- on morphology-based taxonomy and geography (sympatry dinova, 2015).1 versus allopatry) of the studied populations. We use this (2) Two or more allopatric clusters of individuals can scheme to rearrange a complex of species related to Par- be classified as different species if a presence of distinct nassius delphius Eversmann, 1843 ( D subgenus Koramius DNA barcode gap between them is correlated with a mor- Moore, 1902). The taxa that this complex is comprised of phological hiatus, and if the COI genetic distance between are found throughout the mountains of Central Asia, and them is deeper than the “standard” DNA-barcode species though there are many described species, subspecies, and threshold (3%). forms, the group’s taxonomy is not well understood: the When taxa are allopatric, the direct application of any taxonomic status of most of the named populations remains species concepts that invokes reproductive isolation as a controversial, and classifications vary throughout the recent criterion would be difficult without conducting time-con- literature (e.g., Churkin, 2009; Kaabak, Tarasov, & Tuzov, suming mating experiments. Therefore, we considered 1997; Kreuzberg, 1985; Weiss, 1992). two or more allopatric taxa as separate species if they are genetically distant. This criterion is based on empirical observation that a 3% and higher level of uncorrected Criteria to be considered when testing p-distance between COI barcodes is practically always taxonomic hypotheses associated with separate species (Hebert et al., 2003; but We suggest a set of the following six criteria for testing see Zakharov, Lobo, Nowak, & Hellma, 2009). Although primary morphology-based taxonomic hypotheses using there is no ’magic number’ describing genetic divergence DNA barcodes. within the DNA-barcode region that would allow DNA barcoding in biodiversity research 601 distinguishing between species, the 3% level can with hypotheses using DNA barcodes (Fig. 1). All nominal confidence be used as a level of divergence, the presence (i.e., formally described and available in the sense of the of which would strongly support a separate-species nomenclature rules) taxa at the ranks of species and sub- hypothesis for any given pair of existing taxa. species as well as their monophyletic combinations should (3) Two or more allopatric clusters of individuals can be considered a mere hypothesis that can and should be be classified as different subspecies of the same species if tested. DNA barcodes should be obtained for these taxa, (a) presence of distinct DNA barcode gap between them pairwise comparisons of all these taxa should be con- is correlated with a morphological hiatus, (b) the clusters ducted and DNA barcode-based trees should be con- belong to the same monophyletic lineage (do not consti- structed. For each pair of the taxa the following questions tute a poly- or paraphyletic assemblage), and (c) the COI should be asked: (1) Is the DNA barcode gap distinct? (2) genetic distance between them is lower than ’standard’ If yes, is there a correlation between DNA barcode gap DNA-barcode species threshold (3%). and morphological hiatus? (3) Are the taxa sympatric or This criterion is based on the idea of polytypic species allopatric? (4) Is there a deep (>3%) barcode distance (Mayr, 1963; Poulton, 1904), i.e., species that consist of between the taxa? The answers to these questions should morphologically and molecularly differentiated groups of result in a set of secondary species hypotheses that should populations. be additionally tested for monophyly in accordance with (4) Non-monophyletic taxa should be avoided. the criterion 1. If the data available (combination of DNA Monophyly is the basic principle of phylogenetics and barcodes, morphology and geography) are not sufficient should also be applied to taxonomy. While the majority of for reaching taxonomic conclusions, the pre-existing taxo- taxonomists currently believe that monophyly, in the nar- nomic hypothesis should be upheld. Following the above row sense used by Hennig (Envall, 2008; Hennig, 1950, procedure, a previously described valid name should be 1966;Horandl€ & Stuessy, 2010)(D holophyly sensu Ash- assigned to the studied populations, or new taxa should be lock, 1971) is a mandatory criterion for any described described in accordance with the rules of taxonomic taxon, many taxa still represent non-monophyletic entities. nomenclature. Avoiding non-monophyletic groups is also more practical (Talavera, Lukhtanov, Pierce, & Vila, 2013). The COI barcodes alone can provide weak evidence for monophyly Materials and methods of taxa since trees inferred from single markers sometimes display relationships that reflect the evolutionary histories DNA barcoding, study sites, and sampling of individual genes rather than of the species being studied. We studied standard COI barcodes (658-bp 5’ segment of Despite this limitation, we argue that the validity of taxa mitochondrial cytochrome oxidase subunit I) obtained represented by monophyletic clusters resulting from the from 76 specimens of the P. delphius species-group and DNA barcode analysis is much better supported than taxa from two specimens of the P. charltonius species group that are para- or polyphyletic. (P. davydovi Churkin, 2006 and P. loxias Pungeler,€ (5) If DNA barcode-based evidence is weak, preserving 1901). The butterflies were collected in mountain regions stability can become a priority, as stability is a concept of Kazakhstan, Kyrgyzstan, Uzbekistan, and Tajikistan that is positively viewed by the International Commis- (Table 1, see online supplemental material, which is avail- sions of Nomenclature. able from the article’s Taylor & Francis Online page at (6) The lack of a barcode gap could mean that the two http://dx.doi.org/10.1080/14772000.2016.1203371). None species are not differentiated in the barcode region but may of the specimens were subjected to any chemical treat- be differentiated at various other genes and behave as sepa- ment before desiccation. DNA was extracted from a single rate species. However, we argue that in a situation where leg removed from each voucher specimen employing a two competing pre-existing hypotheses have been explic- standard DNA barcode glass fibre protocol (Ivanova, itly formulated (one suggesting that two clusters represent deWaard, & Hebert, 2006). All polymerase chain reac- two different species, and the other that two clusters are tions (PCR) and DNA sequencing were carried out fol- conspecific), the absence of a distinct barcode gap would lowing standard DNA barcoding procedures for be a reason to select the latter hypothesis as a provisional Lepidoptera as described previously (deWaard, Ivanova, taxonomic solution until better evidence appears. Hajibabaei, & Hebert, 2008; Hajibabaei et al., 2005). Pho- tographs of specimens used in the analysis and collecting dataareavailableintheBarcodeofLifeDataSystem Scheme describing procedure of testing pre- (BOLD) at http://www.barcodinglife.org/. All sequences are available through GenBank (#KX422310- KX422387). existing species hypotheses using DNA All voucher specimens are deposited in Zoological Institute barcodes of the Russian Academy of Sciences and in the personal All the six criteria described above can be combined into a collection of B. Khramov (St. Petersburg) and are identified single scheme describing the procedure of testing species with the corresponding unique BOLD Process IDs, which 602 V. A. Lukhtanov et al.

Fig. 1. A scheme describing how the DNA barcodes can be combined with taxonomy and geography to test the pre-existing, morphologically based taxonomic hypotheses (see text for details). are automatically generated by BOLD at the time of the ini- original descriptions of the studied taxa is provided in Ohya tial data submission. Species and subspecies were tenta- (1996), Sotshivko and Kaabak (1996)andWeiss(1992). tively identiﬁed using characters and nomenclature We used 39 additional sequences from GenBank in the described in Kaabak et al. (1997). The complete list of the analysis (see supplemental material online, Table S2) DNA barcoding in biodiversity research 603

Table 1. Parnassius delphius species group: specimens included in the present study.

Taxon (traditionally accepted name Taxon (proposed name Range and combination) and combination) BOLD# Country or locality

P. cardinal P. cardinal cardinal LOWAM348 Tajikistan Darvaz P. cardinal P. cardinal cardinal LOWAM349 Tajikistan Petr I Mts. P. cardinal P. cardinal cardinal LOWAM350 Tajikistan Petr I Mts. P. staudingeri hunza P. cardinal hunza LOWAM346 Tajikistan Beik P. davydovi P. davydovi LOWAM226 Kyrgyzstan P. delphius P. delphius LOWAM396 Kazakhstan Bayankol P. delphius P. delphius LOWAM397 Kazakhstan Bayankol P. delphius P. delphius LOWAM221 Kyrgyzstan Baibichetoo P. delphius P. delphius LOWAM222 Kyrgyzstan Baibichetoo P. delphius P. delphius LOWAM223 Kyrgyzstan Baibichetoo P. delphius P. delphius LOWAM215 Kyrgyzstan Chatkal P. delphius P. delphius LOWAM402 Kyrgyzstan Chatkal P. delphius P. delphius LOWAM403 Kyrgyzstan Chatkal P. delphius P. delphius LOWAM404 Kyrgyzstan Chatkal P. delphius P. delphius LOWAM392 Kazakhstan Dzhungar P. delphius P. delphius LOWAM394 Kyrgyzstan Kaindy P. delphius P. delphius LOWAM395 Kyrgyzstan Kaindy P. delphius P. delphius LOWAM398 Kyrgyzstan Kungey P. delphius P. delphius LOWAM399 Kyrgyzstan Kungey P. delphius P. delphius LOWAM224 Kyrgyzstan Naryntoo P. delphius P. delphius LOWAM400 Kyrgyzstan Sary-Beles P. delphius P. delphius LOWAM401 Kyrgyzstan Sary-Beles P. delphius P. delphius LOWAM216 Kyrgyzstan Suusamyr

P. delphius P. delphius LOWAM217 Kyrgyzstan Talassky P. delphius P. delphius LOWAM218 Kyrgyzstan Talassky P. delphius P. delphius LOWAM219 Kyrgyzstan Terskey P. delphius P. delphius LOWAM220 Kyrgyzstan Terskey P. delphius P. delphius LOWAM390 Kazakhstan Zailiysky P. delphius P. delphius LOWAM391 Kazakhstan Zailiysky P. maximinus P. delphius maximinus LOWAM351 Uzbekistan Karzhantau P. maximinus P. delphius maximinus LOWAM352 Uzbekistan Karzhantau P. maximinus P. delphius maximinus LOWAM354 Kyrgyzstan Kuraminsky P. maximinus P. delphius maximinus LOWAM355 Kyrgyzstan Kuraminsky P. maximinus P. delphius maximinus LOWAM353 Kyrgyzstan Pskem P. maximinus P. delphius maximinus LOWAM356 Kazakhstan Talasky P. maximinus P. delphius maximinus LOWAM357 Kazakhstan Talasky P. staudingeri darvasicus P. infernalis darvasicus LOWAM365 Tajikistan Vanch P. staudingeri darvasicus P. infernalis darvasicus LOWAM366 Tajikistan Vanch P. staudingeri darvasicus P. infernalis darvasicus LOWAM367 Tajikistan Vanch P. staudingeri inaccessibilis P. infernalis inaccessibilis LOWAM363 Tajikistan Petr I Mts. P. staudingeri inaccessibilis P. infernalis inaccessibilis LOWAM364 Tajikistan PetrI Mts. P. kiritshenkoi dunkeldykus P. infernalis dunkeldykus LOWAM379 Tajikistan Dunkeldyk P. kiritshenkoi dunkeldykus P. infernalis dunkeldykus LOWAM380 Tajikistan Dunkeldyk P. kiritshenkoi dunkeldykus P. infernalis dunkeldykus LOWAM381 Tajikistan Dunkeldyk P. kiritshenkoi kiritshenkoi P. infernalis kiritshenkoi LOWAM382 Tajikistan Chechekty P. kiritshenkoi kiritshenkoi P. infernalis kiritshenkoi LOWAM383 Tajikistan Minhajir P. kiritshenkoi kiritshenkoi P. infernalis kiritshenkoi LOWAM384 Tajikistan Minhajir P. kiritshenkoi dunkeldykus P. infernalis dunkeldykus LOWAM375 Tajikistan Dunkeldyk P. kiritshenkoi dunkeldykus P. infernalis dunkeldykus LOWAM376 Tajikistan Dunkeldyk P. kiritshenkoi dunkeldykus P. infernalis dunkeldykus LOWAM377 Tajikistan Dunkeldyk

(continued) 604 V. A. Lukhtanov et al.

Table 1. (Continued )

Taxon (traditionally accepted name Taxon (proposed name Range and combination) and combination) BOLD# Country or locality

P. kiritshenkoi dunkeldykus P. infernalis dunkeldykus LOWAM378 Tajikistan Dunkeldyk P. staudingeri jacobsoni P. jacobsoni LOWAM361 Tajikistan Ak-Buura P. staudingeri jacobsoni P. jacobsoni LOWAM362 Tajikistan Ak-Buura P. staudingeri jacobsoni P. jacobsoni LOWAM358 Tajikistan Dzhelandy P. staudingeri jacobsoni P. jacobsoni LOWAM359 Tajikistan Dzhelandy P. staudingeri jacobsoni P. jacobsoni LOWAM360 Tajikistan Dzhelandy P. staudingeri jacobsoni P. jacobsoni LOWAM368 Tajikistan Rushan P. staudingeri jacobsoni P. jacobsoni LOWAM369 Tajikistan Rushan P. staudingeri jacobsoni P. jacobsoni LOWAM370 Tajikistan Rushan P. loxias P. loxias LOWAM227 Kyrgyzstan Kaindy P. patricius patricius P. patricius patricius LOWAM342 Kyrgyzstan Naryntau P. patricius uzyngyrus P. patricius uzyngyrus LOWAM336 Kyrgyzstan Kirgizsky P. patricius uzyngyrus P. patricius uzyngyrus LOWAM333 Kyrgyzstan Suusamyrsky P. patricius uzyngyrus P. patricius uzyngyrus LOWAM334 Kyrgyzstan Suusamyrsky P. patricius priamus P. patricius priamus LOWAM340 Kyrgyzstan Inylchek P. patricius priamus P. patricius priamus LOWAM341 Kyrgyzstan Kaindy P. patricius kardakoffi P. patricius priamus LOWAM338 Kyrgyzstan Kungey P. patricius priamus P. patricius priamus LOWAM339 Kyrgyzstan Terskey P. patricius lukhtanovi P. patricius lukhtanovi LOWAM345 Kazakhstan Bajankol P. patricius luedwigi P. patricius luedwigi LOWAM343 Kyrgyzstan Chatkal P. patricius luedwigi P. patricius luedwigi LOWAM344 Kyrgyzstan Chatkal P. staudingeri difficilis P. staudingeri difficilis LOWAM371 Tajikistan Turkestansky P. staudingeri difficilis P. staudingeri difficilis LOWAM372 Tajikistan Turkestansky

P. staudingeri difficilis P. staudingeri difficilis LOWAM373 Tajikistan Turkestansky P. staudingeri difficilis P. staudingeri difficilis LOWAM374 Tajikistan Turkestansky P. staudingeri hissaricus P. staudingeri hissaricus LOWAM387 Tajikistan Gissar P. staudingeri staudingeri P. staudingeri staudingeri LOWAM388 Tajikistan Gissar P. staudingeri staudingeri P. staudingeri staudingeri LOWAM389 Tajikistan Gissar

(Lukhtanov et al., 2009; Michel, Rebourg, Cosson, & to species and subspecies level (e.g., Fig. 2À13) and, in a Descimon, 2008; Nazari, Zakharov, & Sperling, 2007). few cases, male genitalia structures can also be used to Only GenBank records in which we were confident in delineate the taxa (Kreuzberg, 1985). Therefore, since we their identifications were used. Such confidence was justi- are very familiar with morphology of the genus Parnas- fied if the images of the voucher specimens were available sius in general and the P. delphius-group in particular, we online, or if species identification was unambiguous due did not reanalyse it and assume that all the taxa studied to absence of morphologically similar taxa in the area. are separated by a morphological hiatus. Only GenBank sequences that overlap with the barcode region in at least 630 bp were used. Parnassius apollonius Eversmann, 1847 and the species of the P. charltonius Geography Gray, 1853 group ( D Kailasius Moore, 1902) were Information about sympatric vs. allopatric distribution of selected as out-groups. the analysed species pairs was extracted from published literature (Churkin, 2009; Kaabak et al., 1997; Kreuzberg, 1985; Weiss, 1992). Morphology The wing pattern of Parnassius species and subspecies was a subject of several studies (Kaabak et al., 1997; Sequence analysis Korb, 2010; Kreuzber, 1985). They demonstrated all the Sequences were aligned using BioEdit version 7.1.7 soft- taxa used in our research to have unique wing patterns ware (Hall, 1999) and edited manually. For the purpose of that, though variable, allow for identification of the taxa phylogenetic reconstruction, sequences were collapsed DNA barcoding in biodiversity research 605

Figs. 2À13. Wing pattern diversity within some taxa of the P. delphius complex recognized in our study. Proposed species names are provided. Fig. 2. P. cardinal hunza, Tajikistan, East Pamir, Beik pass, 20 July 1996, 4300 m, A. Sochivko leg. Fig. 3. P. cardinal cardinal, Tajikistan, Darvaz range, Khobu-Rabat, 3200 m, July 1993, Ivanov leg. Fig. 4. P. cardinal ruth, Afghanistan, Hindukush Range, Anjuman, Jamak pass, 7 July 1772, Asisi leg. Fig. 5. P. jacobsoni, Tajikistan, East Pamir, Tokhtomush, 4300 m, 19 August 1991, V. Lukhtanov leg. Fig. 6. P. delphius, Kazakhstan, Terskey Alatoo Range, Bayankol, 3200À3300 m, 5À7 July 2002, V.Lukhtanov leg. Fig. 7. P. delphius maximinus, Uzbekistan, West Tian-Shan, Kamchik pass, 1À10 July 1997. Fig. 8. P. patricius kardakofﬁ, Kungei-Alatoo Range, 3300 m, Oi-Tal River, Berzhnoi leg. Fig. 9. P. infernalis darvasicus, Tajikistan, West Pamir, Vanch Range, Gushkhon pass, 3600 m, 20 July 1992, I. Pljushch leg. Fig. 10. P. infernalis inaccessibilis Shchetkin, 1979, Tajikistan, Peter the Great range, 14 August 1985. Fig. 11. P. infernalis kiritshenkoi, Tajikistan, East Pamir, Mynkhadjir, 4200 m, 23 August 1991, V. Lukhtanov leg. Fig. 12. P. infernalis illustris, Kyrgyzstan, Transalai Range, Aram-Kungei, 4100À4300 m, 11À12 August 1992, V.Lukhtanov leg. Fig. 13. P. staudingeri staudingeri, Tajikistan, Ghissar Mts., Ansob pass, 3400À2500 m, 24À27 July 1994, V. Lukhtanov leg. 606 V. A. Lukhtanov et al. DNA barcoding in biodiversity research 607

into 56 haplotypes using TCS version 1.18 (Clement, Results Posada, & Crandall, 2000). Phylogenetic relationships Major DNA barcode clusters and the were inferred using Bayesian Inference (BI), maximum likelihood (ML) and maximum parsimony (MP) analyses. sister group jModelTest was used to determine optimal substitution Alignment of COI sequences was unambiguous; no inser- models for ML inference (Posada, 2008). tions or deletions occurred in the dataset. Of 658 posi- Bayesian analyses were conducted using MrBayes, ver- tions, 153 were variable and 117 were parsimony- sion 3.2 (Ronquist et al. 2012)(Fig. 14). Datasets were informative. The results of the BI, ML and MP analyses partitioned by codon position. Substitution models used are shown in Fig. 14 that demonstrates the topology for each partition were chosen according to jModelTest recovered by BI and the branch supports recovered by BI, (Posada 2008): nst D 2 and rates D invgamma for the first ML, and MP. position, nst D 2 and rates D gamma for the second posi- The analysis revealed five major groups of the COI tion, and nst D 6 and rates D gamma for the third position barcodes (Fig. 14). The representatives of these groups of COI barcodes. Two runs of 10,000,000 generations are shown in Fig. 2À13. The first group with four chains (one cold and three heated) were per- (ICIICIIICIVCV) includes populations from the south- formed. Chains were sampled every 1000 generations, ern-most part of the distribution area (S. Tajikistan, and burn-in was determined based on inspection of log Afghanistan, Pakistan, N. India, and Nepal). This group likelihood over time plots using TRACER, version 1.4 was strongly supported by BI, but weakly supported by (available from http://beast.bio.ed.ac.uk/Tracer). ML and not supported by MP. Within this cluster P. steno- The ML trees were inferred by using MEGA6 (Tamura semus Honrath, 1890, and P. stoliczkanus Felder & et al. 2013) based on the Tamura 3-parameter model Felder, 1865 formed a separate clade (IIICIVCV) on the (Tamura, 1992) (Fig. S1, see supplemental material BI tree, but not on the MP tree. Parnassius stoliczkanus online). A discrete Gamma distribution was used to model was found to be paraphyletic with respect to evolutionary rate differences among sites (5 categories P. stenosemus. (CG, parameter D 0.2384)). Branch support was assessed The second group (VICVIICVIII), which has moderate using 1000 bootstrap replicates. support (0.84) in BI and low support in ML and MP, MP analysis was performed using a heuristic search as includes taxa that are found in the northern-most distribu-

implemented in MEGA6 (Tamura et al., 2013) (Fig. S2, tion area (Kazakhstan, NW China, Kyrgyzstan, Northern see supplemental material online). A heuristic search was Uzbekistan, and Tajikistan). The third group (IX) consists carried out using the close-neighbour-interchange algo- of taxa inhabiting the western part of the distribution area rithm with search level 3 (Nei & Kumar, 2000) in which (Western Uzbekistan and Western Tajikistan), and is well the initial trees were obtained with the random addition of supported in all the analyses. The fourth group (X) is rep- sequences (100 replicates). We used non-parametric boot- resented by a single species, P. hide, which is endemic to strap values (Felsenstein, 1985) to estimate branch sup- the eastern part of the distribution area (Tibet). port on the recovered tree. The bootstrap consensus tree The fifth and final group is comprised of the outgroup was inferred from 1000 replicates. (P. charltonius species complex). Within this group, The p-distances between haplotypes were calculated in DNA-barcodes were previously unknown for P. davydovi MEGA6 (Tamura et al. 2013) and represented in Table S1 Churkin, 2006. In our study the COI sequences of P. davy- (see supplemental material online). As a distinct barcode dovi showed an uncorrected p-distance of 4.6% from its gap we consider a situation in which COI distinction morphologically most similar species P. loxias. Further- results in significant topological differentiation between more, BI and MP analyses showed that P. davydovi is the two lineages, i.e., in two lineages with high branch closer to P. inopinatus (p-distance D 3.5%) from Afghani- support (at least 0.9 in BI, ML, and MP, or in one of stan, and not P. loxias (which instead forms a cluster with them). P. autocrator Avinov, 1913). All these taxa are allopatric,

Fig. 14. Consensus Bayesian tree of the P. delphius species complex inferred from COI haplotypes. Branches corresponding to parti- tions reproduced in less than 50% of the trees are collapsed. Posterior probability values (>50%) for BI and bootstrap support values (>50%) for ML and MP are shown for each node, with nonmatching clades using different analyses indicated by ‘À’. The scale bar represents 0.004 substitutions/position. Node labels include traditionally accepted names (mostly after: Kaabak et al., 1997) and localities; numbers of studied specimens representing each haplotype in different localities are given in parentheses. Clusters IÀX correspond to the species that were recognized in our analysis. Subclades 1À4 are described in the text in details. Proposed species names are provided in the right part of the figure. 608 V. A. Lukhtanov et al. and thus, using criterion 2, P. davydovi should be consid- species distinct? (2) If yes, is there a correlation between ered as a separate species. DNA barcode gap and morphological hiatus? (3) What is the p-distance between taxa? (4) Are the taxa sympatric or allopatric? The answers to these questions are provided in the Analysis of clade I Table 2. (1) The distinct barcode gap was found between This clade is comprised of the following taxa (pre-existing P. staudingeri hunza and P. cardinal, and between P. primary taxonomic hypotheses): P. staudingeri hunza staudingeri hunza and [P. cardinal C P. staudingeri Grum-Grshimailo, 1888, P. staudingeri chitralica Verity, ruth]. (2) The distinct gap corresponded to a distinct dif- 1911, P. staudingeri affinis Peschke et Eisner, 1934, P. ference in butterfly wing pattern: P. s. hunza (Fig. 2)is staudingeri kohibaba Clench et Shoumatoff, 1956, P. car- the dark-coloured form without red eyespots, while P. dinal and P. staudingeri ruth Kotzsch, 1936. The mono- cardinal (Fig. 3) and P. s. ruth (Fig. 4) have large red eye- phyletic clade [P. cardinal C P. staudingeri ruth] that is spots. (3) p-distance over 3% was found in four of the strongly supported by BI, ML, and MP analyses was cases. (4) All taxa are allopatric. treated as an additional taxonomic hypothesis. In accor- There are no sympatric taxa in the P. staudingeri dance with the scheme (Fig. 1) the following questions hunzaÀP. staudingeri ruth clade, therefore we used crite- are considered: (1) Is the DNA barcode gap between ria 2 and 3 for interpretation of our data. Within the

Table 2. Parnassius delphius species group: DNA-barcode gap between pairs of taxa, morphological hiatus, DNA-barcode distance, and mode of distribution (a D allopatric, s D sympatric).

Taxon pair Cluster distinct gap DNA gap/morphology correlation p-distance distribution hunza-chitralica I no n/a 0.016 a hunza-affinis I no n/a 0.015 a hunza-kohibaba I no n/a 0.021 a hunza-cardinal I yes yes 0.034 a hunza-ruth I no n/a 0.026 a hunza-(cardinalCruth)I yes yes 0.026 a chitralica-affinis I no n/a 0.022 a chitralica-kohibaba I no n/a 0.021 a chitralica-cardinal I no n/a 0.038 a chitralica-ruth I no n/a 0.030 a chitralica-(cardinalCruth) I no n/a 0.030 a affinis-kohibaba I no n/a 0.018 a affinis-cardinal I no n/a 0.028 a affinis-ruth I no n/a 0.023 a affinis-(cardinalCruth) I no n/a 0.023 a kohibaba-cardinal I no n/a 0.027 a kohibaba-ruth I no n/a 0.022 a kohibaba-(cardinalCruth) I no n/a 0.022 a cardinalCruth I no n/a 0.011 a stenosemus-zogilaica III no n/a 0.015 a stenosemus-stoliczkanus III-IV no n/a 0.011 s stenosemus-nobuko III-V yes yes 0.043 a zogilaica-stoliczkanus III-IV no n/a 0.017 a zogilaica-nobuko III-V yes yes 0.046 a stoliczkanus-nobuko IV-V yes yes 0.037 a delphius-maximinus VI no n/a 0.000 a subclade 1- subclade 2 VI-VII yes yes 0.017 s subclade 1 - subclade 3 VI-VII yes yes 0.011 s subclade 1 - subclade 4 VI-VIII yes yes 0.022 s subclade 2 - subclade 3 VI no n/a 0.011 a subclade 2 - subclade 4 VII-VIII yes yes 0.018 a subclade 3 - subclade 4 VII-VIII no n/a 0.018 a DNA barcoding in biodiversity research 609 studied clades, the lineage [P. cardinal C P. staudingeri existing taxonomic hypothesis (Tshikolovets 2005) should ruth] has a high branch support in BI, ML, and MP, and be upheld, and the taxa P. stenosemus and P. stenosemus genetic distances within this lineage are small (<1.1%). zogilaica should be considered conspecific (subspecies of However, in light of the available data, resolving taxon- the same species). omy within the lineage remains difficult, since there is no Between P. stenosemus and P. stoliczkanus there is a no significant topological differentiation between this lineage distinct barcode gap, and the taxa are sympatric. These and the rest of the clade. Interpretation of P. staudingeri taxa have always been considered as different species. In hunza and P. cardinal (or P. staudingeri hunza and accordance with criterion 5, the pre-existing taxonomic [P. cardinalCP. staudingeri ruth]) as two different spe- hypothesis (Weiss 1992) should be upheld, and the taxa cies is also problematic despite distinct barcode gap and P. stenosemus and P. stoliczkanus should be considered as high level of DNA differentiation between them. If we different species. accept that the above species-pairs are indeed formed by Between the allopatric pairs of taxa P. stenosemus and different species, it becomes impossible to interpret the P. stoliczkanus nobuko, P. stoliczkanus nobuko and P. sto- genetic diversification within the rest of the group, which liczkanus, P. stoliczkanus nobuko and P. stenosemus zogi- consists of P. staudingeri chitralica, P. staudingeri affinis laica, there is (a) a distinct barcode gap, which (b) and P. staudingeri kohibaba, as these taxa are not sepa- corresponds with morphology, and (c) the genetic distance rated by a distinct gap from both P. staudingeri hunza and between them is above 3%. In accordance with criterion 2, P. cardinal. these taxa should be considered as non-conspecific ones. It seems that in accordance with criterion 5, the only Between allopatric taxa P. stoliczkanus and P. stenose- reasonable solution in this situation is the preservation of mus zogilaica, there is a no distinct barcode gap. The the pre-existing hypothesis that viewed P. staudingeri as a taxon P. stenosemus zogilaica has been treated as a sub- widely distributed polytypic species, and P. cardinal as a species of P. stenosemus (Tshikolovets 2005), which has monotypic species within a more restricted area. How- been recognized as separate species from P. stoliczkanus ever, this solution conflicts with criterion 4 (avoiding non- (Weiss 1992). Hence we invoke criterion 5 (preserving monophyly) since P. staudingeri sensu lato appears as a stability) and preserve P. stoliczkanus and P. stenosemus clearly polyphyletic taxon consisting of five unlinked line- zogilaica as representatives of separate species. ages on the tree (P. staudingeri hunzaÀP. staudingeri kohibaba, P. staudingeri ruth, P. staudingeri jacobsoni, P. staudingeri darvasicusÀP. staudingeri inaccessibilis Analysis of the clades VI, VII, and VIII and P. staudingeri difficilisÀP. staudingeri hissaricus). This complex involves a large number of populations with Hence the whole P. staudingeri hunzaÀP. staudingeri relatively low level of DNA barcode differentiation. In ruth lineage should be considered a polytypic species as it fact, this group includes four supported subclades that is (1) monophyletic, (2) consists of only allopatric popula- show little topological differentiation: P. delphius del- tions), and (3) the p-distance with the closest (most likely phius/P. maximinus, P. patricius luedwigi Kreuzberg, sister) allopatric lineage (P. staudingeri jacobsoni)is 1989/P. patricius uzungyrus Weiss, 1979, P. patricius higher than 3%. Parnassius cardinal is the oldest avail- lukhtanovi Rose, 1992/P. patricius kardakoffi Bryk & Eis- able name in the hunza-ruth clade, therefore this name ner, 1930 and P. staudingeri darvasicus Avinov, 1916/P. should be used when describing the polytypic species staudingeri inacessibilis Shchetkin, 1977 (Fig. 14). While combining the taxa P. cardinal hunza, P. cardinal chitral- p-distances within this clade are below 3.4%, some of ica, P. cardinal affinis, P. cardinal kohibaba, P. cardinal these taxa are known to exist in sympatry (Table 2). cardinal, and P. cardinal ruth. Within the first P. delphius delphius/P. maximinus subclade, taxon P. maximinus was described as and for a long time was considered a subspecies. It was elevated to a Analysis of the clades II, III, IV, and V separate species by Kreuzberg (1985) who discovered a The monophyletic lineage II (P. staudingeri jacobsoni contact zone between P. delphius and P. maximinus and Avinov, 1913) is topologically distinct from clade I, described the differences in their wing pattern (Figs 6 and which is likely to be its sister taxon. In other words, clades 7). However, later Churkin (2009) found extensive I and II are separated by a distinct barcode gap of 3.4%, hybridization and numerous intermediates in the contact which corresponds to distinct morphological differences. zone between P. maximinus and P. delphius, indicating a Thus, in accordance with criteria 2 and 4, P. jacobsoni unimodal distribution of morphological traits, and con- and P. cardinal should be considered as non-conspecific cluded that the two taxa should be treated as subspecies. taxa. In our study no distinct barcode gap was found between Between P. stenosemus and P. stenosemus zogilaica P. delphius and maximinus, and p-distances were below there is a no distinct barcode gap, and the taxa are allopat- 1.4%. Thus, in accordance with criterion 6, in a situation ric (Table 2). In accordance with criterion 5, the pre- where there are two competing pre-existing hypotheses 610 V. A. Lukhtanov et al.

(one argues that two clusters are two different species, and infernalis is the oldest available name to describe subclade the other argues that two clusters are conspecific), we con- 4, it should be used for this polytypic species comprised of sider the lack of distinct barcode gap as a support for con- subspecies P. i. darvasicus, P. i. inaccessibilis, P. i. dunkel- specificity. Therefore, we select the hypothesis of Churkin dykus Sotshivko & Kaabak, 1996, P. i. kiritshenkoi Avi- (2009) and treat P. delphius delphius and P. delphius max- nov, 1910, P. i. mustagata Rose, 1990, P. i. illustris Grum- iminus as subspecies of the same species. Grshimailo, 1888, and P. i. infernalis Elwes, 1886. Pairwise comparisons of subclades 1À4 resulted in the following conclusions: Analysis of the lineages IX and X (1) Between the pair subclade 1 and subclade 2, there There is no significant topological differentiation within are (a) a distinct barcode gap, (b) a correspondence clade IX, and the taxa within it are allopatric and genetically between barcode gap presence and morphology, similar (p-distance from 0.2% to 0.9%). Thus, the pre-exist- and (c) sympatricity of taxa. Thus, in accordance ing hypothesis of Kaabak et al. (1997) about the conspecific- with criterion 1, subclades 1 and 2 should be classi- ity of P. staudingeri staudingeri, P. s. difficilis Murzin, 1989 fied as members of different species. and P. s. hissaricus Eisner,1968shouldbepreserved. (2) The same can be said about the pairs of subclade 1/ The lineage X is represented by a single specimen in subclade 3, and subclade 1/subclade 4, and these pairs our analysis. Thus, no taxonomic changes can be made in should be classified as members of different species. this lineage represented by species P. hide. (3) In the pair subclade 2Àsubclade 3, there is a no distinct barcode gap, and the taxa are allopatric. In accordance with criteria 5 and 6, the pre-existing New taxonomic arrangement of P. delphius taxonomic hypothesis (Kreuzberg, 1985) should be species group upheld, and subclades 2 and 3 should be considered Combining the newly obtained DNA barcoding data with conspecific. the existing knowledge of morphology-based taxonomy and (4) Between subclade 2 and subclade 4, there are (a) a distribution results in the following taxonomic arrangement: distinct barcode gap, (b) correspondence between barcode gap and morphology, (c) the minimum P. cardinal Grum-Grshimailo, 1887

genetic distance between taxa is lower than 3%, and (d) the taxa are allopatric. While, in accordance P. cardinal hunza Grum-Grshimailo, 1888; P. cardinal with criterion 3, these are the signs of subspecies; if chitralica Verity, 1911; P. cardinal affinis Peschke & Eis- we are to merge these clusters into a single taxon, it ner, 1934; P. cardinal kohibaba Clench & Shoumatoff, will lead to a non-monophyletic assemblage, which 1956; P. cardinal cardinal Grum-Grshimailo, 1887; P. should be avoided (criterion 4). In this situation, in cardinal ruth Kotzsch, 1936 accordance with criterion 5, the pre-existing taxo- P. jacobsoni Avinov, 1913 nomic hypothesis (Kreuzberg, 1985) should take P. stenosemus Honrath, 1890 prevalence: subclade 2 should be treated as a part À of P. patricius; subclade 4 as P. staudingeri. P. stenosemus stenosemus Honrath, 1890; P. stenosemus (5) Between subclade 3 and subclade 4, there is no dis- zogilaica Tytler, 1926 tinct barcode gap, and the taxa are allopatric. In accordance with criterion 5, the pre-existing taxo- P. stoliczkanus Felder & Felder, 1865 nomic hypothesis (Kreuzberg, 1985) should pre- P. nobuko Ohya, 1996 vail: subclade 3 represents P. patricius; subclade 4 P. delphius Eversmann, 1843 À P. staudingeri. P. delphius delphius Eversmann, 1843; P. delphius maxi- Subclade 4 is represented by allopatric, genetically minus Staudinger, 1891 undifferentiated taxa that are considered as subspecies of P. patricius Niepelt, 191 P. staudingeri in the classification of Kreuzberg (1985)and Kaabak et al. (1997). However, such taxonomic treatment P. patricius luedwigi Kreuzberg, 1989; P. patricius uzyngy- contradicts our first criterion 4 (avoiding grouping non- rus Weiss, 1979; P. patricius patricius Niepelt, 1911; P. monophyletic assemblages into a single species): P. stau- patricius lukhtanovi Rose, 1992; P. patricius priamus dingeri sensu lato appears as a clearly polyphyletic taxon Bryk, 1914; P. patricius kardakoffi Bryk & Eisner, 1930 consisting of five unlinked lineages on the tree obtained (P. staudingeri hunzaÀP. staudingeri kohibaba, P. staudin- P. infernalis Elwes, 1886 geri ruth, P. staudingeri jacobsoni, P. staudingeri darvasicusÀP. staudingeri inaccessibilis and P. staudingeri P. infernalis darvasicus Avinov, 1916; P. infernalis inac- difficilisÀP. staudingeri hissaricus)(seeabove).AsP. cessibilis Shchetkin, 1977; P. infernalis dunkeldykus DNA barcoding in biodiversity research 611

Sotshivko & Kaabak, 1996; P. infernalis kiritshenkoi Avi- Kreuzberg is a polyphyletic assemblage consisting of nov, 1910; P. infernalis mustagata Rose, 1990; P. inferna- numerous, sometimes distant, lineages. lis illustris Grum-Grshimailo, 1888; P. infernalis Hence, when we combine the newly obtained DNA bar- infernalis Elwes, 1886 coding data with the data on morphology and geography (sympatry versus allopatry) and use the six criteria out- P. staudingeri A. Bang-Haas, 1882 sensu stricto lined in this paper, we can essentially improve pre-existing hypotheses and solve long-standing taxonomic P. staudingeri difﬁcilis Murzin, 1989; P. staudingeri stau- problems. dingeri A. Bang-Haas, 1882; P. staudingeri hissaricus Eisner, 1968 P. hide Koiwaya, 1987 Notes 1. Although possibility of nuclear mitochondrial pseudogenes (Numts) should be always considered (Schizas, 2012). 2. P. acdestis Grum-Grshimailo, 1891 was not included Discussion in this list, because molecular studies (Michel, Until the 1970À1980s, it was believed that the Parnassius Rebourg, Cosson, & Descimon, 2008) demonstrated 2 delphius group includes four polytypic species : P. del- that it belongs to the P. charltonius Gray, 1853 - P. phius sensu lato, widely distributed from south-eastern imperator Oberthur,€ 1883 species group. Kazakhstan in the north to northern India in the south; P. patricius Niepelt, 1911, sympatric with P. delphius but limited to the Tian-Shan mountains in Central Asia; and Acknowledgements sympatric with each other P. stenosemus Honrath, 1890 We thank Boris Khramov for samples provided for the and P. stoliczkanus Felder & Felder, 1865 found in north- study. ern Pakistan and northern India. In the 1970À1980s, Shchetkin (1979) and Kreuzberg (1985) demonstrated sig- niﬁcant heterogeneity in morphology and ecology of

P. delphius sensu lato as well as the sympatry between P. Disclosure statement delphius inaccessibilis Shchetkin, 1977 and P. delphius No potential conflict of interest was reported by the cardinal Grum-Grshimailo, 1887. Hence, Kreuzberg authors. (1985) divided P. delphius sensu lato into four species: P. delphius sensu stricto, P. cardinal, P. staudingeri A. Bang-Haas, 1882, and P. maximinus Staudinger, 1891. Funding Kreuzberg (1985) treated many taxa, such as P. staudin- The financial support for this study was provided by grant geri hunza, P. staudingeri chitralica, P. staudingeri affi- N 14-14-00541 from the Russian Science Foundation to nis, P. staudingeri kohibaba, P. staudingeri ruth, P. the Zoological Institute of the Russian Academy of staudingeri jacobsoni, P. staudingeri infernalis, P. stau- Sciences. dingeri illustris, P. staudingeri darvasicus, P. staudingeri kiritshenkoi, P. staudingeri staudingeri, P. staudingeri hissaricus and P. staudingeri inaccessibilis as subspecies of P. staudingeri. At the same time, he considered P. del- Supplemental data phius, P. maximinus, P. patricius, P. cardinal, P. stenose- Supplemental data for this article can be accessed at http://dx. mus, and P. stoliczkanus as distinct species. doi.org/10.1080/14772000.2016.1203371. The system of Kreuzberg (outlined in Table 1, see supplemental material online, and in node labels in Fig. 14) has become widely accepted (e.g., Kaabak et al., 1997; References Weiss, 1992). The molecular studies (Michel et al., 2008), Ashlock, P. D. (1971). Monophyly and associated terms. System- while partially supported the Kreuzberg’s hypothesis, atic Zoology, 20,63À69. demonstrated the heterogeneity of P. staudingeri sensu Barrett, R. D. H., & Hebert, P. D. N. (2005). Identifying spiders Kreuzberg, 1985. through DNA barcodes. Canadian Journal of Zoology, 83, Our study demonstrated that Kreuzberg’s views should 481À491. be significantly revised. We confirm the point of view Bi, K., Linderoth, T., Vanderpool, D., Nielsen, R., & Moritz, C. (2013) Unlocking the vault: Next-generation museum popu- expressed by Churkin (2009) that P. maximinus is conspe- lation genomics. Molecular Ecology, 22: 6018À6032. cific with P. delphius. Furthermore, our data from DNA Brower, A. V. Z. (2006). Problems with DNA barcodes for spe- barcoding clearly demonstrate that P. staudingeri sensu cies delimitation: ‘ten species’ of Astraptes fulgerator 612 V. A. Lukhtanov et al.

reassessed (Lepidoptera: Hesperiidae). Systematics and Bio- Hey, J., Waples, R. S., Arnold, M. L., Butlin, R. K., & Harrison, diversity, 4, 127À132. R. G. (2003). Understanding and confronting species uncer- Brower, A. V. Z. (2010). Alleviating the taxonomic impediment tainty in biology and conservation. Trends in Ecology and of DNA barcoding and setting a bad precedent: Names for Evolution, 18, 597À603. ten species of ‘Astraptes fulgerator’ (Lepidoptera: Hesperii- Horandl,€ E., & Stuessy, T. F. (2010). Paraphyletic groups as nat- dae: Eudaminae) with DNA-based diagnoses. Systematics ural units of biological classification. Taxon, 59, and Biodiversity, 8, 485À491. 1641À1653. Butcher, B. A., Smith, M. A., Sharkey, M. J., & Quicke, D. L. J. Huemer, P., Mutanen, M. (2015) Alpha taxonomy of the genus (2012). A turbo-taxonomic study of Thai aleiodes (aleiodes) Kessleria Nowicki, 1864, revisited in light of DNA-barcod- and aleiodes (arcaleiodes) (Hymenoptera: Braconidae: ing (Lepidoptera, Yponomeutidae). ZooKeys, 503:89À133. Rogadinae) based largely on COI barcoded specimens, with doi: 10.3897/zookeys.503.9590 rapid descriptions of 179 new species. Zootaxa, 3457, Ivanova, N. V., deWaard, J. R., & Hebert, P. D. N. (2006). An 1À232. inexpensive, automation-friendly protocol for recovering high Churkin, S. (2009). Notes on Parnassius Latreille, 1804 from quality DNA. Molecular Ecology Resources, 6,998À1002. Tian-Shan and Alai. Part 2: Parnassius staudingeri A. Bang- Janzen, D. H., Hajibabaei, M., & Burns, J. M. (2005). Wedding Haas, 1882 - Parnassius delphius Eversmann, 1843 (Lepi- biodiversity inventory of a large and complex Lepidoptera doptera, Papilionidae). Atalanta, 40, 436À460. fauna with DNA barcoding. Philosophical Transactions of Clement, M., Posada, D., & Crandall, K. (2000). TCS: A com- the Royal Society of London. Series B, Biological Sciences, puter program to estimate gene genealogies. Molecular 1462, 1835À1846. Ecology, 9, 1657À1660. Kaabak, L. V., Tarasov, E. A., & Tuzov, V. K. (1997). Papilioni- Coyne, J. A., & Orr, H. A. (2004). Speciation. Sunderland, dae. In V. K. Tuzov (Ed.), Guide to the butterflies of Russia Massachusetts: Sinauer Associates. and adjacent territories (Vol. 1, pp. 135À151). Sofia-Mos- deWaard, J. R., Ivanova, N. V., Hajibabaei, M., & Hebert, P. D. cow: Pensoft. N. (2008). Assembling DNA barcodes: Analytical protocols. Kekkonen, M., & Hebert, P. D. N. (2014). DNA barcode-based In C. C. Martin (Ed.), Environmental genomics, methods in delineation of putative species: Efficient start for taxonomic molecular biology (Vol. 410, pp. 275À283). Totowa, New workflows. Molecular Ecology Resources, 14, 706À715. Jersey: Humana Press. Knowles, L. L., & Carstens, B. C. (2007). Delimiting species Envall, M. (2008). On the difference between mono-, holo-, and without monophyletic gene trees. Systematic Biology, 56, paraphyletic groups: A consistent distinction of process and 887À895. pattern. Biological Journal of Linnean Society, 94, Koiwaya, S. (1987). A new species of the genus Parnassius from 217À220. Kunlun Mts., China. Gekkan-Mushi Supplement, 201,3À6. Felsenstein, J. (1985). Confidence limits on phylogenies: An Korb S.K. (2010). Evolution of wing pattern, island effect and À approach using the bootstrap. Evolution, 39, 783 791. aspects of systematics of butterfly superspecies Parnassius Guschanski, K., Krause, J., Sawyer, S., Valente, L.M., Bailey, S., staudingeri A. Bang-Haas, 1882 (Lepidoptera, Papilioni- Finstermeier, K., … Savolainen, V. (2013). Next-generation dae). Altai Zoological Journal, 4,15À31 (in Russian). museomics disentangles one of the largest primate radia- Kreuzberg, A. (1985). Papilionids of the delphius, charltonius, tions. Systematic Biology, 62: 539À554. doi: 10.1093/ simo groups of the fauna of USSR. In Investigations of flora sysbio/syt018 and fauna of Central Asia (pp. 25À68). Tashkent, (in Hajibabaei, M., deWaard, J. R., Ivanova, N. V., Ratnasingham, Russian). S., Dooh, R. T., Kirk, S. L., ... Hebert, P. D. N. (2005). Criti- Lukhtanov V. A., Dantchenko A. V., Vishnevskaya M. S., Saifit- cal factors for assembling a high volume of DNA barcodes. dinova A. F. 2015. Detecting cryptic species in sympatry Philosophical Transactions of the Royal Society of London. and allopatry: Analysis of hidden diversity in Polyommatus Series B, Biological Sciences, 360, 1959À1967. (Agrodiaetus) butterflies (Lepidoptera: Lycaenidae). Biolog- Hall, T. A. (1999). BioEdit: A user-friendly biological sequence ical Journal of the Linnean Society, 116, 468À485. alignment editor and analyses program for Windows 95/98/ Lukhtanov, V. A., & Shapoval, N. A. (2008). Detection of cryp- NT. Nucleic Acids Symposium Series, 41,95À98. tic species in sympatry using population analysis of unlinked Hausmann, A., Sciarretta, A., & Parisi, F. (2016). The Geometrinae genetic markers: A study of the Agrodiaetus kendevani spe- of Ethiopia II: Tribus Hemistolini, genus Prasinocyma (Lepi- cies complex (Lepidoptera: Lycaenidae). Doklady Biologi- doptera: Geometridae, Geometrinae). Zootaxa, 4065,1À63. cal Sciences, 423, 432À436. Hebert, P. D. N., Cywinska, A., Ball, S. L., & deWaard, J. R. Lukhtanov, V. A., Shapoval, N. A., & Dantchenko, A. V. (2008). (2003). Biological identifications through DNA barcodes. Agrodiaetus shahkuhensis sp. n. (Lepidoptera, Lycaenidae), Proceedings of the Royal Society B: Biological Sciences, a cryptic species from Iran discovered by using molecular 270, 313À321. and chromosomal markers. Comparative Cytogenetics, 2, Hebert, P. D. N., Penton, E. H., Burns, J. M., Janzen, D. H., & 99À114. Hallwachs, W. (2004a). Ten species in one: DNA barcoding Lukhtanov, V. A., Shapoval, N. A., & Dantchenko, A. V. (2014). reveals cryptic species in the neotropical skipper butterfly Taxonomic position of several enigmatic Polyommatus Astraptes fulgerator. Proceedings of the National Academy (Agrodiaetus) species (Lepidoptera, Lycaenidae) from Cen- of Sciences, USA, 101, 14812À14817. tral and Eastern Iran: Insights from molecular and chromo- Hebert, P. D. N., Stoeckle, M. Y., Zemlak, T. S., & Francis, C. somal data. Comparative Cytogenetics, 8, 313À322. doi: M. (2004b). Identification of birds through DNA barcodes. 10.3897/CompCytogen.v8i4.8939 Public Library of Science Biology, 2, 1657À1663. Lukhtanov, V. A., Sourakov, A., Zakharov, E. V., & Hebert, P. Hennig, W. (1950). Grundzuge€ einer Theorie der phylogeneti- D. N. (2009). DNA barcoding Central Asian butterflies: schen Systematik. Berlin: Deutcher Zentralverlag. Increasing geographical dimension does not significantly Hennig, W. (1966). Phylogenetic systematics. Urbana, Illinois: reduce the success of species identification. Molecular Ecol- University of Illinois Press. ogy Resources, 9, 1302À1310. DNA barcoding in biodiversity research 613

Mallet, J. (2001). Species, concepts of. In S. Levin (Ed.), Ency- Sotshivko, A. V., & Kaabak, L. V. (1996). A new subspecies of clopedia of biodiversity. (Vol. 5, pp. 427¡440). San Diego, Parnassius staudingeri A. Bang-Haas, 1882 from the Sary- CA: Academic Press. kolsky Mountain Ridge in Tadjikistan (Lepidoptera, Papilio- Mallet, J. (2006). Species concepts. In C. W. Fox, & J. B. Wolf nidae). Atalanta, 27, 199À202. (Eds.) Evolutionary genetics: Concepts and case studies, Sourakov, A., Plotkin, D., Kawahara, A.Y., Xiao, L., Hallwachs, Oxford: Oxford University Press. W., & Janzen, D. (2015). On the taxonomy of the erythrina Mallet, J., & Willmott, K. (2003). Taxonomy: Renaissance or moths Agathodes and Terastia (Crambidae: Spilomelinae): Tower of Babel? Trends in Ecology and Evolution, 18, Two different patterns of haplotype divergence and a new 57À59. species of Terastia. Tropical Lepidoptera Research, 25, Mayr, E. (1963). Animal species and evolution. Cambridge, 80À98. Massachusetts: Harvard University Press. Sourakov, A., & Zakharov, E. (2011). “Darwin’s butterflies”? Michel, F., Rebourg, C., Cosson, E., & Descimon, H. (2008). DNA barcoding and the radiation of the endemic Caribbean Molecular phylogeny of Parnassiinae butterflies (Lepidop- butterfly genus Calisto (Lepidoptera, Nymphalidae, Satyri- tera: Papilionidae) based on the sequences of four mitochon- nae). Comparative Cytogenetics, 5, 191À210. drial DNA segments. Annales de la Societ e entomologique Talavera, G., Dinca, V., & Vila, R. (2013). Factors affecting spe- de France, 44,1À36. cies delimitations with the GMYC model: Insights from a Nazari, V., Zakharov, E. V., & Sperling, F. A. (2007). Phylog- butterfly survey. Methods in Ecology and Evolution, 4, eny, historical biogeography, and taxonomic ranking of Par- 1101À1110. nassiinae (Lepidoptera, Papilionidae) based on morphology Talavera, G., Lukhtanov, V. A., Pierce, N. E., & Vila, R. (2013). and seven genes. Molecular Phylogenetics and Evolution, Establishing criteria for higher-level classification using 42, 131À156. molecular data: The systematics of Polyommatus blue Nei, M., & Kumar, S. (2000). Molecular evolution and phyloge- butterflies (Lepidoptera, Lycaenidae). Cladistics, 29, netics. New York: Oxford University Press. 166À192. Ohya, A. (1996). Two new subspecies of Parnassius from Nepal Talavera G., Lukhtanov V.A., Rieppel L., Pierce N.E., & Vila R. and Tibet. Gekkan-Mushi, 309,3À9. (2013). In the shadow of phylogenetic uncertainty: The Posada, D. (2008). jModelTest: Phylogenetic model averaging. recent diversification of Lysandra butterflies through chro- Molecular Biology and Evolution, 25, 1253À1256. mosomal change. Molecular Phylogenetics and Evolution, Poulton, E. B. (1904). What is a species? Proceedings of the 69, 469À478. Entomological Society of London, 1903, lxxviiÀcxvi. Tamura, K. (1992). Estimation of the number of nucleotide sub- Puillandre, N., Modica, M. V., Zhang, Y., Sirovich, L., Boisse- stitutions when there are strong transition-transversion and lier, M.-C., Cruaud, C., …, Samadi, S. (2012). Large-scale G C C-content biases. Molecular Biology and Evolution, 9, species delimitation method for hyperdiverse groups. Molec- 678À687. À ular Ecology, 21, 2671 2691. doi:10.1111/J.1365- Tshikolovets, V. (2005). The butterflies of Ladakh (N.-W. India). 294X.2012.05559.x Kyiv-Brno: Tshikolovets Publications, published by the author. Riedel, A., Sagata, K., Suhardjono, Y. R., T€anzler, R., & Balke, Vershinina, A. O., & Lukhtanov, V. A. (2010). Geographical dis- M. (2013). Integrative taxonomy on the fast track À towards tribution of the cryptic species Agrodiaetus alcestis alcestis, more sustainability in biodiversity research. Frontiers in A. alcestis karacetinae and A. demavendi (Lepidoptera, Zoology, 10, 15. Lycaenidae) revealed by cytogenetic analysis. Comparative Ronquist, F., Teslenko, P., van der Mark, D., Ayres, A., Darling, Cytogenetics, 4,1À11. S.H., Hohna,€ B.,… Huelsenbeck, J.P. (2012). MrBayes 3.2: Vila, R., Lukhtanov, V. A., Talavera, G., Gil-T., F., & Pierce, N. Efficient Bayesian phylogenetic inference and model choice E. (2010). How common are dot-like distribution ranges? across a large model space. Systematic Biology, 61, Taxonomical oversplitting in Western European Agrodiae- 539À542. tus (Lepidoptera, Lycaenidae) revealed by chromosomal and Rubinoff, D., & Holland, B. S. (2005). Between two extremes: molecular markers. Biological Journal of the Linnean Soci- Mitochondrial DNA is neither the panacea nor the nemesis ety, 101, 130À154. of phylogenetic and taxonomic inference. Systematic Biol- Weiss, J. C. (1992). The Parnassiinae of the World. Part 2. Ven- ogy, 54, 952À961. ette, France: Sciences Nat. Schizas, N. V. (2012). Misconceptions regarding nuclear mito- Will, K. W., Mishler, B. D., & Wheeler, Q. D. (2005). The perils chondrial pseudogenes (Numts) may obscure detection of of DNA barcoding and the need for integrative taxonomy. mitochondrial evolutionary novelties. Aquatic Biology, 17, Systematic Biology, 54, 844À851. 91À96. Zakharov, E. V., Lobo, N. F., Nowak, C., & Hellma, J. J. (2009). Shchetkin, Y. Y. (1979). On the specific independence of Par- Introgression as a likely cause of mtDNA paraphyly in two nassius cardinal Gr.-Gr. and a new subspecies of Parnassius allopatric skippers (Lepidoptera: Hesperiidae). Heredity, delphius Ev. (Lepidoptera: Papilionidae). Doklady Akademii 102, 590À599. Nauk Tadzhikskoi SSR, 22,63À66 (in Russian). Smith, M. A., Wood, D. M., Janzen, D. H., Hallwachs, W., & Associate Editor: Andrew Brower Hebert, P. D. N. (2007). DNA barcodes affirm that 16 species of apparently generalist tropical parasitoid flies (Dip- tera, Tachinidae) are not all generalists. Proceedings of the National Academy of Sciences, USA, 104, 4967À4972.