Does the DNA barcoding gap exist? – a case study in blue butterflies (Lepidoptera: Lycaenidae)

BioMed CentralFrontiers in Zoology

ss

Open AcceResearchDoes the DNA barcoding gap exist ndash a case study in blue butterflies (Lepidoptera Lycaenidae)Martin Wiemers and Konrad Fiedler

Address Department of Population Ecology Faculty of Life Sciences University of Vienna Althanstrasse 14 1090 Vienna Austria

Email Martin Wiemers - martinwiemersunivieacat Konrad Fiedler - konradfiedlerunivieacat

Corresponding author

AbstractBackground DNA barcoding ie the use of a 648 bp section of the mitochondrial genecytochrome c oxidase I has recently been promoted as useful for the rapid identification anddiscovery of species Its success is dependent either on the strength of the claim that interspecificvariation exceeds intraspecific variation by one order of magnitude thus establishing a barcodinggap or on the reciprocal monophyly of species

Results We present an analysis of intra- and interspecific variation in the butterfly familyLycaenidae which includes a well-sampled clade (genus Agrodiaetus) with a peculiar characteristicmost of its members are karyologically differentiated from each other which facilitates therecognition of species as reproductively isolated units even in allopatric populations The analysisshows that there is an 18 overlap in the range of intra- and interspecific COI sequence divergencedue to low interspecific divergence between many closely related species In a Neighbour-Joiningtree profile approach which does not depend on a barcoding gap but on comprehensive samplingof taxa and the reciprocal monophyly of species at least 16 of specimens with conspecificsequences in the profile were misidentified This is due to paraphyly or polyphyly of conspecificDNA sequences probably caused by incomplete lineage sorting

Conclusion Our results indicate that the barcoding gap is an artifact of insufficient samplingacross taxa Although DNA barcodes can help to identify and distinguish species we advocate usingthem in combination with other data since otherwise there would be a high probability thatsequences are misidentified Although high differences in DNA sequences can help to identifycryptic species a high percentage of well-differentiated species has similar or even identical COIsequences and would be overlooked in an isolated DNA barcoding approach

BackgroundMolecular tools have provided a plethora of new opportu-nities to study questions in evolutionary biology (eg spe-ciation processes) and in phylogenetic systematics Onlyrecently however have claims been made that thesequencing of a small (648 bp) fragment at the 5 end ofthe gene cytochrome c oxidase subunit 1 (COI or cox1)

from the mitochondrial genome would be sufficient inmost Metazoa to identify them to the species level [12]This approach called DNA barcoding has gainedmomentum and the Consortium for the Bar Code of Life(CBOL) founded in September 2004 intends to create aglobal biodiversity barcode database in order to facilitateautomated species identifications Right from the start

Published 7 March 2007

Frontiers in Zoology 2007 48 doi1011861742-9994-4-8

Received 1 December 2006Accepted 7 March 2007

This article is available from httpwwwfrontiersinzoologycomcontent418

copy 2007 Wiemers and Fiedler licensee BioMed Central Ltd This is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby20) which permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

of 16(page number not for citation purposes)

Frontiers in Zoology 2007 48 httpwwwfrontiersinzoologycomcontent418

however this approach received opposition especiallyfrom the taxonomists community [3-8] Some argumentsin this debate are political in nature others have a scien-tific basis Concerning the latter one of the most essentialarguments focuses on the so-called barcoding gapAdvocates of barcoding claim that interspecific geneticvariation exceeds intraspecific variation to such an extentthat a clear gap exists which enables the assignment ofunidentified individuals to their species with a negligibleerror rate [1910] The errors are attributed to a smallnumber of incipient species pairs with incomplete lineagesorting (eg [11]) As a consequence establishing thedegree of sequence divergence between two samplesabove a given threshold (proposed to be at least 10 timesgreater than within species [10]) would indicate specificdistinctness whereas divergence below such a thresholdwould indicate taxonomic identity among the samplesFurthermore the existence of a barcoding gap would evenenable the identification of previously undescribed spe-cies ([11-13] but see [14]) Possible errors of thisapproach include false positives and false negatives Falsepositives occur if populations within one species aregenetically quite distinct eg in distant populations withlimited gene flow or in allopatric populations with inter-rupted gene flow In the latter case it must be noted thatdepending on the amount of morphological differentia-tion and the species concept to be applied such popula-tions may also qualify as cryptic species in the view ofsome scientists False negatives in contrast occur whenlittle or no sequence variation in the barcoding fragmentis found between different biospecies (= reproductivelyisolated population groups sensu Mayr [15]) Hence falsenegatives are more critical for the barcoding approachbecause the existence of such cases would reveal exampleswhere the barcoding approach is less powerful than theuse of other and more holistic approaches to delimit spe-cies boundaries

Initial studies on birds [10] and arthropods [916]appeared to corroborate the existence of a distinct barcod-ing gap but two recent studies on gastropods [17] andflies [18] challenge its existence The reasons for these dis-crepancies are not entirely clear Although levels of COIsequence divergence differ between higher taxa (eg anexceptionally low mean COI sequence divergence of only10 was found in congeneric species pairs of Cnidariacompared to 96ndash157 in other animal phyla [2]) Mol-lusca (with 111 mean sequence divergence betweenspecies) and Diptera (93) are not peculiar in thisrespect Meyer amp Paulay [17] assume that insufficient sam-pling on both the interspecific and intraspecific level cre-ate the artifact of a barcode gap Proponents of barcodingmight argue however that the main reason for this over-lap is the poor taxonomy of these groups eg cryptic spe-cies may have been overlooked which are differentiated

genetically but very similar or even identical in morphol-ogy

If the barcode gap does not exist then the thresholdapproach in barcoding becomes inapplicable Althoughmore sophisticated techniques (eg using coalescence the-ory and statistical population genetic methods [19-21])can sometimes help to delimit species with overlappinggenetic divergences these approaches require additionalassumptions (eg about the choice of population geneticmodels or clustering algorithms) and are only feasible inwell-sampled clades

Barcoding holds promise nonetheless especially in theidentification of arthropods the most species-rich animalphylum in terrestrial ecosystems Identification of arthro-pods is often extremely time-consuming and generallyrequires taxonomic specialists for any given group More-over the fraction of undescribed species is particularlyhigh as opposed to vertebrates Hence there is substantialdemand for improved (and rapid) identification tools byscientists who seek identification of large arthropod sam-ples from complex faunas Therefore arthropods deserveto be considered the yard-stick for the usefulness of bar-coding approaches among Metazoa and it is not surpris-ing that several recent studies have tried to apply DNAbarcoding in arthropods [911-1316181922-27] Diver-sity is concentrated in tropical ecosystems but measuringintra- and interspecific sequence divergence in tropicalinsects is hampered by the fragmentary knowledge ofmost taxa In contrast insects of temperate zones andmost notably the butterflies of the Holarctic region arewell known taxonomically compared to other insects Thespecies-rich Palaearctic genus (or subgenus) Agrodiaetusprovides an excellent example to test the existence of thebarcode gap in arthropods This genus is exceptionalbecause of its extraordinary interspecific variation in chro-mosome numbers which have been investigated for mostof its ca 120 species ([28-30] and references therein) As aresult several cryptic species which hardly or not at all dif-fer in phenotype have been discovered (eg [31-39])Available evidence suggests that apart from a few excep-tions (eg due to supernumerary chromosomes) differ-ences in chromosome numbers between butterfly speciesare linked to infertility in interspecific hybrids [40] This isdue to problems in the pairing of homologous chromo-somes during meiosis Since major differences in chromo-some numbers are indicative of clear species boundariesthey are helpful also to infer species-level differentiationfor allopatric populations Agrodiaetus butterflies thereforeare an ideal case for testing the validity of the barcodingapproach If valid then it must be possible to safely recog-nize all species that can be distinguished by phenotypekaryotype or both character sets with reference tosequence divergences alone On the contrary failure of

Page 2 of 16(page number not for citation purposes)


DNA barcodes to differentiate between species that aredistinguished by clear independent evidence wouldundermine the superiority of the barcoding approachwhich has especially been attributed to taxa with diffi-cult classical taxonomy such as Agrodiaetus

ResultsIntraspecific divergenceThe average divergence in 1189 intraspecific comparisonsis 102 (SE = 113) 95 of intraspecific comparisonshave divergences of 0ndash32 The few values higher than32 are conspicuous and probably due to misidentifica-tions (Lampides boeticus Neozephyrus japonicus Arhopalaatosia Agrodiaetus kendevani see below) unrecognizedcryptic species (Agrodiaetus altivagans [41] Agrodiaetusdemavendi [30]) hybridization events (Meleageria marcida[3042]) or any of those (Agrodiaetus mithridates Agrodiae-tus merhaba)

The evidence for the possible misidentifications is the fol-lowing

bull Lampides boeticus is the most widespread species ofLycaenidae and a well-known migrant which occursthroughout the Old World tropics and subtropics fromAfrica and Eurasia to Australia and Hawaii Apart from asingle unpublished sequence (AB192475) all other COIGenBank sequences of this species (from Morocco Spainand Turkey) are identical with each other or only differ ina single nucleotide (= 015 divergence) They are alsonearly identical to two specimens of Lampides boeticus inthe CBOL database (BOLD) [43] from Tanzania andanother sequence of this species from Papua New Guinea(Wiemers unpubl data) The GenBank sequenceAB192475 (of unknown origin but possibly from Japan)however differs strongly (82ndash87) from all other Lam-pides boeticus sequences and therefore we assume this torepresent a distinct species Its identity however remains amystery because it is not particularly close to any otherGenBank sequence and a request for a check of thevoucher specimen has remained unanswered for morethan a year

bull The questionable unpublished sequence of Neozephyrusquercus (AB192476) is identical to a sequence of Favoniusorientalis and therefore probably represents this latter spe-cies which is very similar in phenotype but well differen-tiated genetically (48 divergence)

bull A similar situation constitutes the questionable unpub-lished sequence of Arhopala atosia (AY236002) which isvery similar (04) to a sequence of Arhopala epimuta

bull Agrodiaetus kendevani is a local endemic of the ElbursMts in Iran The two sequences of this species in the NCBI

database which exhibit a divergence of 54 have beenpublished in two different papers by the same work group[2944] While one of them is identical to a sequence ofAgrodiaetus pseudoxerxes the other one is nearly identicalto Agrodiaetus elbursicus (02 divergence) These lattertwo species however belong to separate species groups[30] and thus conspecificity of the two sequences of Akendevani is very improbable as there is no evidence ofhybridization between members of different speciesgroups in Agrodiaetus [30]

Higher intraspecific divergence values are also foundbetween North African and Eurasian populations of Poly-ommatus amandus (38) and Polyommatus icarus (57ndash68) In the former species the North African populationis also well differentiated in phenotype (ssp abdelaziz)while in the latter species phenotypic differences havenever been noted Cases with substantial but lowergenetic divergence between North African and Europeanpopulations which do not correspond to differentiationin phenotype also occur in the butterflies Iphiclides (poda-lirius) feisthamelii (21 [30]) and Pararge aegeria (19[45]) In all cases these allopatric populations may actu-ally represent distinct species although we do not cur-rently have additional evidence in support of thishypothesis

Although some of the other higher divergence values gt2are possibly due to cryptic species (eg in Agrodiaetusdemavendi) or hybridization between closely related spe-cies (eg in the species pair Lysandra corydonius and L oss-mar as evidenced by the comparative analysis of thenuclear rDNA internal transcribed spacer region ITS-2[30]) most of those values represent cases in which thereis hardly any doubt regarding the conspecificity of sam-ples The highest such value is 29 between distant pop-ulations of the widespread Agrodiaetus damon (from Spainand Russia) Outside the genus Agrodiaetus high values arealso found between North African and Iranian popula-tions of Lycaena alciphron (27) Spanish and Anatolianpopulations of Polyommatus dorylas (23) and evenbetween Polish and Slovakian populations of Maculineanausithous (23) Table 1 lists mean intraspecific diver-gences in those species that are represented by more thanone individual in the data set

Interspecific divergenceThe average divergence in 236348 interspecific compari-sons is 938 (SE = 365) ranging from 00 to 232(between Baliochila minima and Agrodiaetus poseidon) Ofthese 57562 are congeneric comparisons with an averagedivergence of 507 (SE = 173) ranging from 00(between 23 Agrodiaetus as well as 3 Maculinea speciespairs) to 124 (between Arhopala abseus and Arhopalaace) 94 of those comparisons are within Agrodiaetus



Table 1 Intraspecific nucleotide divergences

Species No of individuals Mean percent divergence Standard error () Range () Monophyly corrected

Acrodipsas aurata 3 02 01 02 ndash 03 Mono MonoAcrodipsas brisbanensis 8 10 05 02 ndash 16 Mono MonoAcrodipsas cuprea 6 05 03 02 ndash 09 Mono MonoAcrodipsas hirtipes 2 10 --- Mono MonoAcrodipsas mortoni 2 02 --- Mono MonoAgrodiaetus admetus 4 17 07 05 ndash 25 Poly PolyAgrodiaetus ainsae 4 03 02 0 ndash 06 PolyAgrodiaetus alcestis 6 08 04 0 ndash 15 Poly PolyAgrodiaetus altivagans 9 18 15 0 ndash 55 PolyAgrodiaetus antidolus 4 03 03 0 ndash 07 Poly PolyAgrodiaetus arasbarani 2 10 --- Poly PolyAgrodiaetus baytopi 4 19 12 05 ndash 31 Poly PolyAgrodiaetus birunii 10 02 02 0 ndash 07 Para ParaAgrodiaetus caeruleus 3 05 05 0 ndash 1 Mono MonoAgrodiaetus carmon 4 13 06 06 ndash 2 Poly PolyAgrodiaetus cyaneus 6 02 02 0 ndash 07 Poly PolyAgrodiaetus damocles 4 11 08 01 ndash 18 Poly PolyAgrodiaetus damon 5 16 08 0 ndash 29 Mono MonoAgrodiaetus damone 3 06 00 06 ndash 06 Para ParaAgrodiaetus dantchenkoi 6 00 00 0 ndash 0 Poly PolyAgrodiaetus darius 3 00 00 0 ndash 0 Mono MonoAgrodiaetus demavendi 17 21 13 0 ndash 36 Poly PolyAgrodiaetus elbursicus 9 05 08 0 ndash 21 Poly PolyAgrodiaetus erschoffii 3 02 02 0 ndash 03 Mono MonoAgrodiaetus fabressei 3 01 01 0 ndash 02 Poly PolyAgrodiaetus femininoides 2 18 --- Poly PolyAgrodiaetus firdussii 9 05 04 0 ndash 13 Poly MonoAgrodiaetus fulgens 2 02 --- Poly PolyAgrodiaetus glaucias 2 02 --- Mono MonoAgrodiaetus gorbunovi 5 01 01 0 ndash 02 ParaAgrodiaetus haigi 3 00 00 0 ndash 0 PolyAgrodiaetus hamadanensis 4 04 03 0 ndash 07 Mono MonoAgrodiaetus hopfferi 3 15 13 02 ndash 28 Para ParaAgrodiaetus huberti 7 05 04 0 ndash 13 PolyAgrodiaetus humedasae 2 02 --- Mono MonoAgrodiaetus iphidamon 4 00 00 0 ndash 0 Mono MonoAgrodiaetus iphigenia 8 07 06 0 ndash 2 Mono MonoAgrodiaetus iphigenides 3 16 08 07 ndash 22 Poly PolyAgrodiaetus kanduli 2 27 --- PolyAgrodiaetus kendevani 2 54 --- PolyAgrodiaetus khorasanensis 2 05 --- MonoAgrodiaetus klausschuriani 3 00 00 0 ndash 0 Mono MonoAgrodiaetus kurdistanicus 3 00 00 0 ndash 0 Poly PolyAgrodiaetus lorestanus 2 00 --- MonoAgrodiaetus lycius 2 08 --- Mono MonoAgrodiaetus menalcas 5 06 04 0 ndash 13 Mono MonoAgrodiaetus merhaba 3 24 12 11 ndash 35 Poly PolyAgrodiaetus mithridates 2 46 --- Poly PolyAgrodiaetus mofidii 2 10 --- Poly PolyAgrodiaetus nephohiptamenos

2 00 --- Mono

Agrodiaetus ninae 5 07 03 02 ndash 13 Poly PolyAgrodiaetus paulae 2 00 --- Para ParaAgrodiaetus phyllides 4 07 02 04 ndash 09 Poly PolyAgrodiaetus phyllis 4 17 07 05 ndash 25 Para MonoAgrodiaetus pierceae 3 03 02 02 ndash 05 Mono MonoAgrodiaetus poseidon 5 05 04 0 ndash 1 Poly MonoAgrodiaetus posthumus 3 01 01 0 ndash 02 Mono MonoAgrodiaetus pseudactis 2 10 --- PolyAgrodiaetus pseudoxerxes 2 18 --- Poly PolyAgrodiaetus putnami 3 00 00 0 ndash 0 PolyAgrodiaetus ripartii 17 14 08 0 ndash 33 Poly PolyAgrodiaetus rjabovi 2 11 --- Mono MonoAgrodiaetus rovshani 4 02 02 0 ndash 04 Mono MonoAgrodiaetus sekercioglu 2 05 --- PolyAgrodiaetus shahrami 2 03 --- Poly PolyAgrodiaetus sigberti 2 09 --- PolyAgrodiaetus surakovi 2 02 --- Para Poly



Agrodiaetus tankeri 3 17 07 1 ndash 23 Poly PolyAgrodiaetus tenhageni 2 00 --- Mono MonoAgrodiaetus turcicolus 5 08 04 0 ndash 13 PolyAgrodiaetus turcicus 4 08 03 05 ndash 11 Mono MonoAgrodiaetus valiabadi 2 00 --- Mono MonoAgrodiaetus vanensis 5 05 03 0 ndash 08 MonoAgrodiaetus wagneri 2 01 --- ParaAgrodiaetus zapvadi 4 00 01 0 ndash 01 PolyAgrodiaetus zarathustra 2 00 --- Mono MonoArhopala achelous 6 13 09 02 ndash 31 Poly PolyArhopala antimuta 2 11 --- Mono MonoArhopala atosia 3 29 17 1 ndash 43 Poly PolyArhopala barami 2 03 --- Mono MonoArhopala democritus 2 10 --- Mono MonoArhopala epimuta 6 02 03 0 ndash 08 Para ParaArhopala labuana 2 05 --- Mono MonoArhopala major 2 03 --- Mono MonoArhopala moolaiana 2 23 --- Mono MonoAricia agestis 6 07 10 0 ndash 24 Poly PolyAricia artaxerxes 3 18 05 12 ndash 21 PolyCelastrina argiolus 2 13 --- Mono MonoChrysoritis nigricans 2 11 --- Mono MonoChrysoritis pyroeis 2 00 --- Mono MonoCyaniris semiargus 4 10 04 03 ndash 15 Mono MonoFavonius cognatus 3 03 02 01 ndash 05 Poly PolyFavonius jezoensis 2 06 --- Mono MonoFavonius korshunovi 2 04 --- Mono MonoFavonius orientalis 3 07 05 01 ndash 1 Poly MonoFavonius saphirinus 8 11 07 0 ndash 2 Mono MonoFavonius taxila 3 01 01 0 ndash 01 Mono MonoFavonius ultramarinus 5 10 04 04 ndash 14 Poly PolyFavonius yuasai 3 05 04 01 ndash 08 Mono MonoFlos anniella 2 26 --- Mono MonoJalmenus evagoras 12 06 03 02 ndash 11 Mono MonoLampides boeticus 4 43 45 02 ndash 87 Poly MonoLucia limbaria 2 17 --- Mono MonoLycaeides melissa 5 07 05 0 ndash 12 Poly PolyLycaena alciphron 2 27 --- Mono MonoLysandra albicans 3 09 01 08 ndash 09 ParaLysandra bellargus 6 02 03 0 ndash 08 Mono MonoLysandra coridon 5 16 05 07 ndash 21 Poly PolyLysandra corydonius 4 17 12 0 ndash 27 Poly PolyLysandra ossmar 2 21 --- PolyMaculinea alcon 7 00 01 0 ndash 02 Poly MonoMaculinea arion 10 02 02 0 ndash 06 Para ParaMaculinea arionides 4 05 04 0 ndash 09 Poly PolyMaculinea nausithous 3 22 03 19 ndash 24 Mono MonoMaculinea rebeli 3 01 02 0 ndash 03 PolyMaculinea teleius 5 09 05 02 ndash 16 Mono MonoMeleageria daphnis 4 21 04 15 ndash 26 Poly MonoMeleageria marcida 2 44 --- PolyNeolysandra fatima 2 00 --- Mono MonoNeozephyrus japonicus 2 48 --- PolyPlebejus argus 5 10 08 0 ndash 19 Mono MonoPolyommatus amandus 3 26 20 03 ndash 38 Para ParaPolyommatus cornelia 3 11 05 06 ndash 15 Para ParaPolyommatus dorylas 4 16 04 12 ndash 23 Mono MonoPolyommatus eroides 2 14 --- Poly PolyPolyommatus escheri 2 20 --- Mono MonoPolyommatus icarus 8 22 23 0 ndash 68 Poly PolyPolyommatus menelaos 2 00 --- Mono MonoPolyommatus myrrhinus 3 01 01 0 ndash 01 Mono MonoPolyommatus thersites 5 09 06 0 ndash 16 Mono MonoPseudophilotes vicrama 2 00 --- Mono MonoQuercusia quercus 2 06 --- Mono MonoVacciniina alcedo 2 00 --- Mono Mono

Mean and range of intraspecific nucleotide divergences for 133 Lycaenidae species using Kimuras two parameter model The column Monophyly states if conspecific sequences form a monophylum (Mono) a paraphylum (Para) or a polyphylum (Poly) and the subsequent column gives the corrected status (if presumable errors are excluded and critical taxa are lumped together)

Table 1 Intraspecific nucleotide divergences (Continued)



Only congeneric comparisons were included in subse-quent analyses in order to make comparisons feasibleacross taxonomic levels Table 2 lists mean interspecificdivergences in genera of which at least two species are rep-resented in the data set Sequence divergence in 95 ofinterspecific (congeneric) comparisons is above 19and 876 of such comparisons reveal distances above3

The barcode gapAs apparent in Figure 1 (and Figure 2 for comparisonswithin Agrodiaetus only) no gap exists between intraspe-cific and interspecific divergences Since some (014)interspecific divergences are as low as 0 no safe thresh-old can be set to strictly avoid false negatives Althoughspecies pairs with such low divergences include somewhose taxonomic status as distinct species is debatablethey also include many pairs which are well differentiatedin phenotype have a very different karyotype (in Agrodia-etus) and occur sympatrically without any evidence forinterbreeding Examples include Agrodiaetus peilei ndash Amorgani (00) Agrodiaetus fabressei ndash A ainsae (02)Agrodiaetus peilei ndash A karindus (02) Polyommatus myr-rhinus ndash P cornelia (04) or Agrodiaetus poseidon ndash Ahopfferi (06)

The minimum cumulative error based on false positivesplus false negatives is 18 at a threshold level of 28

(Figure 3) Minimum errors are very similar for Agrodiae-tus (186 at 30 threshold not shown) and otherLycaenidae (186 at 20 threshold not shown) butmuch lower in Arhopala (53 at 34 threshold Figure4)

For safe identification minimum distances between spe-cies (Figure 5) are critical and not average distances InAgrodiaetus all but two species (= 983) have close rela-tives with interspecific distances below 3 In the othergenera combined only 74 of taxa are affected but thislower rate is probably due to undersampling and wouldrise if more sequences of more closely related speciesbecome available for the analysis

Identification with NJ tree profileThe approach of species identification with a Neighbour-Joining (NJ) tree profile as proposed by [9] does not nec-essarily depend on the barcoding gap but on the coales-cence of conspecific populations and the monophyly ofspecies (details see Data analysis)

The success rate in the identification of our Lycaenidaedata set with this method was 58 Five out of 158 misi-dentifications or ambiguous identifications (32) can beattributed to incorrectly identified specimens (Lampidesboeticus Neozephyrus japonicus Agrodiaetus kendevani seeabove) Further 90 cases (57) were among closely

Table 2 Interspecific nucleotide divergences

Genus No of species Mean percent divergence Standard error () Range ()

Acrodipsas 9 31 10 05 ndash 57Agriades 2 47 ---Agrodiaetus 117 51 17 0 ndash 101Arhopala 30 68 17 04 ndash 124Aricia 7 34 19 02 ndash 75Chrysoritis 19 70 25 08 ndash 109Euphilotes 2 103 ---Favonius 9 40 09 01 ndash 54Glaucopsyche 2 13 ---Lycaeides 3 17 09 05 ndash 30Lycaena 9 45 11 12 ndash 68Lysandra 9 22 07 07 ndash 40Maculinea 7 28 14 0 ndash 60Meleageria 2 26 16 01 ndash 44Neolysandra 5 46 16 1 ndash 63Phengaris 3 38 21 13 ndash 51Plebejus 5 56 16 24 ndash 74Polyommatus 12 59 25 01 ndash 105Pseudophilotes 4 27 16 06 ndash 45Satyrium 3 45 05 4 ndash 49Trimenia 2 61 ---Turanana 2 48 ---Vacciniina 3 72 03 68 ndash 75

Mean and range of interspecific nucleotide divergences for species in 22 Lycaenidae genera using Kimuras two parameter model



related sister species whose taxonomic status is in dispute(Table 3) If these cases are not taken into account (iecounted as successful identifications an unrealistic bestcase scenario for barcoding success) the success ratewould rise to 84 In Agrodiaetus the success rate wouldremain lower (79) while in the remaining genera itwould reach 91 But even with these corrections 61cases of misidentifications (16) remain 46 of these inAgrodiaetus (affected taxa in Table 4) The complete Neigh-bour-joining tree (available for download as additionalfile 1 NJ-tree) shows the reason for this failure Only 46of conspecific sequences form a monophyletic group onthis tree while the others are either paraphyletic (10) oreven polyphyletic (44) In Agrodiaetus only 34 of spe-cies are monophyletic (Table 1) while the others are par-aphyletic (11) or polyphyletic (55) If incorrectlyidentified specimens are excluded and critical taxa (Table3) are lumped together still only 59 of species aremonophyletic (43 in Agrodiaetus) while 7 are para-phyletic and 34 polyphyletic (49 in Agrodiaetus)

ConclusionWe found an upper limit for intraspecific sequence diver-gences in a wide range of species of the diverse butterflyfamily Lycaenidae but no lower limit for interspecific

divergences and thus no barcoding gap This result is espe-cially well documented in the comprehensively sampledgenus Agrodiaetus (114 of ca 130 recognized speciessequenced) while the smaller overlap in Arhopala can beattributed to the lower percentage of species sampled (33of more than 200 species) The choice of species by [46]was to maximize coverage of divergent clades while mini-mizing the total number of species which is a commonand sensible approach for phylogenetic studies butundermines the power of such sequence data as criticaltests for the barcoding approach The general level ofsequence divergence is not exceptionally low in Lycaeni-dae compared to other Lepidoptera The mean congenericinterspecific sequence divergence of 51 in Lycaenidae(51 in Agrodiaetus and 50 in the other genera) wasonly slightly lower than the mean value of 66 found by[2] in various families of Lepidoptera

We thus confirm the results of Meyer amp Paulay [17] andMeier et al [18] Our results also agree with those from arecent study in the Neotropical butterfly subfamily Ithom-iinae (Nymphalidae) [47] which records highly variablelevels of divergence in mtDNA (COI ampCOII) between taxaof the same rank Our results however fail to agree withthose of Barrett amp Hebert [9] on arachnids In that study

Frequency distribution of intraspecific and interspecific (congeneric) genetic divergence in LycaenidaeFigure 1Frequency distribution of intraspecific and interspecific (congeneric) genetic divergence in Lycaenidae Total number of comparisons 1189 intraspecific and 57562 interspecific pairs across 315 Lycaenidae species Divergences were cal-culated using Kimuras two parameter (K2P) model

0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance

IntraspecificInterspecific



the mean percent sequence divergence between conge-neric species was 164 (SE = 013) and thus three timeshigher than in our study while the divergence among con-specific individuals was only slightly higher with 14 (SE= 016) The contradiction between our study and theirscan be explained by the very incomplete and sparse taxonsampling in their data set amounting to just 1 of the spe-cies contained within the families We conclude that thereported existence of a barcode gap in arachnids appearsto be an artifact based on insufficient sampling acrosstaxa

Despite these difficulties species identification of uniden-tified samples with the help of barcodes is entirely possi-ble The NJ tree profile approach which does not rely on abarcode gap enabled the correct assignment of manysequences and other methods (eg applying populationgenetic approaches) might further increase the successrate However 17 of test sequences could still not beidentified correctly even in some sympatric species pairswhich clearly differ in phenotype and chromosomenumber (eg Agrodiaetus ainsae [n = 108ndash110]fabressei [n= 90] Agrodiaetus hopfferi [n = 15]poseidon [n = 19ndash22])The main reason for this failure is that a large proportionof species are not reciprocally monophyletic eg due to

incomplete lineage sorting which is in accordance with aprevious study [48] Moreover the success with thismethod is again completely dependent on comprehensivesampling If the correct species is not included in the pro-file the assignment must by necessity be incorrect andmisleading Because of the non-existence of a barcodinggap this error will often be impossible to detect This lim-its possible applications of the barcoding approach Forexample cryptic species can only be detected with thehelp of a barcoding approach at high genetic divergencefrom all phenotypically similar species An example isAgrodiaetus paulae which was discovered in this way [41]In contrast and on the one hand the sympatric speciespairs Agrodiaetus ainsae-fabressei A hopfferi-poseidon and Amorgani-peilei would have gone unnoticed by barcodingapproaches even though their strong phenotypical andkaryological differentiation (n = 108 vs n = 90 n = 15 vsn = 19ndash22 and n = 27 vs n = 39 respectively) clearly indi-cates their specific distinctness On the other handsequence divergence in what is currently believed to rep-resent one species does not per se prove the specific dis-tinctness of the entities in question In Polyommatus icarusor P amandus for example the high divergences betweenNorth African and Eurasiatic samples is a strong hint forthe presence of unrecognized cryptic species but this

Frequency distribution of intraspecific and interspecific (congeneric) genetic divergences in AgrodiaetusFigure 2Frequency distribution of intraspecific and interspecific (congeneric) genetic divergences in Agrodiaetus Total number of comparisons 737 intraspecific and 54209 interspecific pairs across 114 Agrodiaetus species Divergences were calcu-lated using Kimuras two parameter (K2P) model

0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance

intraspecificinterspecific



needs to be rigorously tested with sequence data fromsamples that cover the geographic range more compre-hensively Also in practical application the problem ofmisidentified specimens and sequences in GenBankremains a real threat to the accuracy of barcode-basedidentifications An example is the GenBank sequenceAB192475 of Lampides boeticus which is also used in theCBOL database (see above) This underscores the impor-tance of voucher specimens and documentation of local-ity data an issue raised by barcoding supporters butunfortunately still much neglected by GenBank Anothercase of misidentification (GenBank sequence AF170864of Plebejus acmon which was originally submitted as Euphi-lotes bernardino) [30] has already been corrected with thehelp of the voucher specimen

In conclusion the barcoding approach can be very help-ful eg in identifying early stages of insects or when onlyfragments of individuals are available for analysis How-ever correct identification requires that all eligible speciescan be included in the profile and that sufficient informa-tion is available on the amount of intraspecific geneticvariation and genetic distance to closely related species

The barcoding procedure is not very well suited for iden-tifying species boundaries but it may help to give mini-

mum estimates of species numbers in very diverse andinadequately known taxonomic groups at single localitiesOur case study on Agrodiaetus shows that a substantialnumber of species would have gone unnoticed by the bar-coding approach as false negatives Thus especially inclades where many species have evolved rapidly as a resultof massive radiations with minimum sequence diver-gence the barcoding approach holds little promise ofmeeting the challenge of rapid and reliable identificationof large samples Yet it is exactly these situations whichpose the most problematic tasks in the morphologicalidentification of insects

Although molecular data can be helpful in discoveringnew species a large genetic divergence is not sufficientproof since it must be corroborated by other data Further-more most closely related species which are difficult toidentify with traditional means are also similar geneti-cally and would go unnoticed by an isolated barcodingapproach Mathematical simulations have shown thatpopulations have to be isolated for more than 4 milliongenerations (ie 4 million years in the mostly univoltineAgrodiaetus species) for two thresholds proposed by thebarcoding initiative (reciprocal monophyly and a geneticdivergence between species which is 10 times greater thanwithin species) to achieve error rates less than 10 [49]

Cumulative error based on false positives plus false negatives for each threshold value in 315 Lycaenidae species including only congeneric comparisonsFigure 3Cumulative error based on false positives plus false negatives for each threshold value in 315 Lycaenidae spe-cies including only congeneric comparisons The optimum threshold value is 28 where error is minimized at 180

0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value

false negativesfalse positives



This might help to explain why the barcoding approachappears to be more successful in the Oriental genus Arho-pala which is thought to represent a phylogenetically olderlineage of Lycaenidae estimated to be about 7ndash11 Millionyears old [50] while the origin of the Palaearctic genusAgrodiaetus is dated at only 25ndash38 Million years [44]

Our data show that the lack of a barcoding gap and recip-rocal monophyly in Lycaenidae is not confined to thegenus Agrodiaetus with its extraordinary interspecific vari-ation in chromosome numbers but also to other generaof Lycaenidae with stable chromosome numbers Itshould also be noted that in Agrodiaetus there is neitherevidence for exceptional rapid radiation as in cichlids ofthe East African lakes [51] nor for unusual (ie sympatric)speciation patterns caused by karyotype evolutionRather karyotype diversification seems to have been amere by-product of the usual mode of allopatric specia-tion [293044]

MethodsData sourcesA total of 694 barcode sequences were used for our analy-sis We used a 690 bp fragment at the 5 end of cyto-chrome c oxidase subunit I (COI) of 309 Lycaenidaesequences from a molecular phylogenetic study by Wiem-

ers [30] Most sequences belong to Agrodiaetus (198) theothers (111) mostly to closely related Polyommatinae Allsequences have been deposited in GenBank [52](AY556844-AY556867 AY556869-AY556963AY556965-AY557155) with LinkOuts provided to imagesof the voucher specimens deposited with MorphBank[53] These sequences were supplemented by 385 furthersequences of Lycaenidae deposited in GenBank as ofMarch 2006 (Table 5) They include sequences from fur-ther studies on Agrodiaetus [2944] the Palaearctic genusMaculinea [54] Nearctic Lycaeides melissa [55] the Orien-tal genus Arhopala [4650] the Australian genera Acrodip-sas [56] and Jalmenus [57] and the South AfricanChrysoritis [58] as well as a few sequences which have onlybeen used as outgroups in non-Lycaenidae studies (eg[5960]) Sequence length in the 5 region as defined byCBOL ranged between 240 bp and the maximum of 987bp (18 COI sequences from a study on Japonica only con-tained a 3end fragment and therefore were not included)Of these 89 are at least 648 bp long as recommendedby CBOL and 98 at least 500 bp long which is deemedsufficient for barcode sequences [13] However sequenceoverlap for sequences from different studies was some-times lower because of slightly different sequence loca-tions within the barcode region (Figure 6) It should benoted that these inconsistencies in barcode comparisons

Cumulative error based on false positives plus false negatives for each threshold value in 30 Arhopala speciesFigure 4Cumulative error based on false positives plus false negatives for each threshold value in 30 Arhopala species The optimum threshold value is 34 where error is minimized at 53

0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value




are a common situation in barcode sequences due to dif-ferences in primer use (eg [2])

Laboratory protocolsDNA was extracted from thorax tissue recently collectedand preserved in 100 ethanol using Qiagenreg DNeasy Tis-sue Kit according to the manufacturers protocol formouse tail tissue In a few cases only dried material wasavailable and either thorax or legs were used for DNAextraction

Amplification of DNA was conducted using the polymer-ase chain reaction (PCR) The reaction mixture (for a totalreaction volume of 25 microl) included 1 microl DNA 168 microl

ddH20 25 microl 10 times PCR II buffer 32 microl 25 mM MgCl205 microl 2 mM dNTP-Mix 025 microl Taq Polymerase and 0375microl 20 pm of each primer The two primers used were

Primer 1 k698 TY-J-1460 TAC AAT TTA TCG CCT AAACTT CAG CC [61]

Primer 2 Nancy C1-N-2192 (CCC) GGT AAA ATT AAAATA TAA ACT TC [61]

PCR was conducted on thermal cyclers from Biometrareg

(models Uno II or T-Gradient) or ABI Biosystemsreg (modelGeneAmpreg PCR-System 2700) using the following pro-files

Frequency distribution of minimum interspecific (congeneric) genetic distances across 263 Lycaenidae speciesFigure 5Frequency distribution of minimum interspecific (congeneric) genetic distances across 263 Lycaenidae species

Minimum interspecific distances

0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus




Table 3 Sister species or species complexes with disputable species borders

Agrodiaetus altivagansdamoclesectabanensisgorbunovikandulimaraschiwagneri [30 41]Agrodiaetus artvinensisbilginifirdussiipseudactissigberti [30]Agrodiaetus aserbeidschanushubertininaeturcicoluszapvadi [30]Agrodiaetus baytopiiphicarmon [30]Agrodiaetus carmonschuriani [30]Agrodiaetus cyaneuskermansisparacyaneusAgrodiaetus demavendilorestanus [30]Agrodiaetus khorasanensisnephohiptamenosripartii [30]Agrodiaetus phyllisvanensis [30]Agrodiaetus poseidonputnami [30]Agrodiaetus sekercioglusurakovi [30]Aricia agestisartaxerxes [30]Lysandra albicanscaelestissimuscoridongennargenti [30]Lysandra caucasicuscorydoniusossmar [30]Maculinea alconrebeli [30 54]Meleageria daphnismarcida [30 42 54]Polyommatus andronicusicarus [30]Polyommatus eroseroides [30]

List of disputable species complexes due to eg incomplete speciation and gene flow or in Agrodiaetus very similar phenotype and only slight differences in karyotype The taxonomically oldest name is marked in bold

Table 4 Taxa misidentified with the NJ tree profile approach

Agrodiaetus admetus (78ndash80)demavendi (asymp67)nephohiptamenos (asymp90)Agrodiaetus ainsae (108ndash110)fabressei (90)Agrodiaetus alcestis (19ndash21)dantchenkoi (40ndash42)eriwanensis (28ndash32)interjectus (29ndash32)Agrodiaetus altivagans (18ndash22)ciscaucasicus (16)Agrodiaetus antidolus (42)femininoides (27)kurdistanicus (62)Agrodiaetus arasbarani (25)elbursicus (16)lukhtanovi (22)paulae (17)zarathustra (asymp22)Agrodiaetus baytopi (27ndash28)tankeri (20ndash21)Agrodiaetus birunii (10ndash11)brandti (19)Agrodiaetus carmon (81ndash82)surakovi (50)Agrodiaetus ciscaucasicus (16)mofidii (35)Agrodiaetus cyaneus (19)pseudoxerxes (15ndash16)Agrodiaetus damone (66ndash68)iphigenides (67)juldusus (67)karatavicus (67)phyllides (67)Agrodiaetus elbursicus (17)turcicolus (20)Agrodiaetus hopfferi (15)poseidon (19ndash22)Agrodiaetus lorestanus (68)ripartii (90)Arhopala achelousmutaFavonius cognatusultramarinusMaculinea arionarionidesPolyommatus amandus abdelaziz Meleageria daphnisPolyommatus corneliamyrrhinus

List of taxa which were misidentified with the NJ tree profile approach (excluding possible errors and critical taxa listed in Tab3) Misidentified test taxa (in bold font) and their identifications are placed jointly in a single line Haploid chromosome numbers of Agrodiaetus species (taken from [29 30 44]) are given in parenthesis


Initial 4 minutes denaturation at 94degC and 35 cycles of 30seconds denaturation at 94degC 30 seconds annealing at55degC and 1 minute extension at 72degC

PCR products were purified using purification kits fromPromegareg or Sigmareg and checked with agarose gel electro-phoresis before and after purification

Cycle sequencing was carried out on Biometrareg T-Gradientor ABI Biosystemsreg GeneAmpreg PCR-System 2700 thermalcyclers using sequencing kits of MWG Biotechreg (for Li-correg

automated sequencer) or ABI Biosystemsreg (for ABIreg 377automated sequencer) according to the manufacturersprotocols and with the following cycling times initial 2minutes denaturation at 95degC and 35 cycles of 15 secondsdenaturation at 95degC 15 seconds annealing at 49degC and15 seconds extension at 70degC Primers used were thesame as for the PCR reactions for the ABI (primer 1 wasused for forward and primer 2 for independent reversesequencing) but for Li-cor truncated and labelled primerswere used with 3 bases cut off at the 5 end and labelledwith IRD-800 For ABI sequencing the products werecleaned using an ethanol precipitation protocol Electro-phoresis of sequencing reaction products was carried outon Li-correg or ABIreg 377 automated sequencers using themanufacturers protocols

Data analysisSequences were aligned with BioEdit 7041 [62] andpruned to a maximum of 987 bp the section proposed byCBOL for barcoding Pairwise sequence divergences werecalculated separately for intraspecific as well as for inter-specific but intrageneric comparisons with Mega 31 [63]using Kimuras two parameter (K2P) distance model This

is not necessarily the best model to analyze the data (see[64]) but it was chosen to facilitate comparisons withother barcode studies of Hebert and co-workers [19-1216] who have been using this model Distance tableswere processed to calculate divergence means (incl stand-ard errors and ranges) within and between species

The taxonomy was taken from GenBank in most cases buttwo minor spelling inconsistencies were corrected In fourcases where a taxon within Agrodiaetus was treated as aspecies taxon by one author but only as a subspecies byanother we matched them by treating those taxa as dis-tinct species The generic subdivision of Lycaenidae is verymuch in flux Some genera are only treated as subgeneraby some authors and many genera (like Polyommatus orPlebejus) are probably paraphyletic or polyphyletic how-ever we undertook no revision of the GenBank taxonomysince it appeared consistent enough for our analysis Theremaining inconsistencies only affect few taxa in our anal-ysis and include the treatment of Sublysandra (distinctgenus or subgenus of Polyommatus) Eumedonia (distinctgenus or subgenus of Aricia) Otnjukovia (synonym toTuranana) Maculinea (synonym to Phengaris) and Callipsy-che (synonym to Satyrium) (A complete list of sequenceswith corresponding taxa names and voucher numbers isfound in the additional file 1 NJ tree)

A Lycaenidae species profile was created according to [9]Of the 694 barcode sequences we excluded 9 short Arho-pala sequences with a barcode length of only 240 bp (Tocheck the position of those sequences a separate analysiswas run containing only the Arhopala sequences) Of theremaining 685 sequences we randomly selected 1sequence from each of the 308 Lycaenidae species for

Table 5 Material

GenBank accession no Number of sequences Reference Taxa in focus

AY556844 ndash AY556867AY556869 ndash AY556963AY556965 ndash AY557155

309 [30 41] Agrodiaetus



AY235861 ndash AY235903AY235955 ndash AY236006

52 [46 50] Arhopala

AY675402 ndash AY675448 47 [54] MaculineaDQ234691 ndash DQ234695 5 [55] LycaeidesAY091712 ndash AY091741 30 [56] AcrodipsasDQ249942 ndash DQ249953 12 [57] JalmenusAF279217 ndash AF279244 28 [58] ChrysoritisAF170864 1 [59] PapilionidaeAY350456 ndash AY350459 4 [60] LepidopteraDQ018938 ndash DQ018948 11 [67] Papilionoidea amp HesperioideaAB195510 ndash AB195545 36 Odagiri et al (unpubl) FavoniusAB192475 ndash AB192476 2 Tanikawa et al (unpubl) Hesperiidae

List of GenBank accession numbers used for analysis including references and taxa which were the focus of these studies



inclusion into a COI species profile We chose a sequenceof Apodemia mormo (GenBank accession numberAF170863) from the family Riodinidae as outgroupbecause this family appears to represent the sister group toLycaenidae [65-67] The other 377 sequences which hadnot been included in the profile were used as testsequences They were singly added to the test profile inrepeated Neighbour-joining analyses and their classifica-tion success was recorded A test was recorded as success-ful if the test sequence grouped most closely with theconspecific profile sequence and not with another speciesResults of three GenBank sequences which were not iden-tified to species level (all belonging to the genus Agrodiae-tus) were not counted After the classification test anotherNJ analysis was run including all sequences in order tounderstand possible failures in classification The main

reason for using the Neighbour-joining as a tree-buildingmethod is its computational efficiency Although thismethod is well suited for grouping closely relatedsequences it should be noted that other methods (such asMaximum Parsimony Maximum Likelihood or Bayesianinference of phylogeny) are usually superior in construct-ing phylogenetic trees

Competing interestsThe author(s) declare that they have no competing inter-ests

Authors contributionsMW carried out the molecular genetic studies sequencealignment statistical analysis and drafted the manuscriptKF participated in the design of the study and the statisti-

Sequence overlap for pairwise barcode comparisonsFigure 6Sequence overlap for pairwise barcode comparisons Length of sequence overlap in 246229 cross-comparisons of 694 aligned sequences

Sequence overlap for pairwise barcodecomparisons

0

5

10

15

20

25

0 200 400 600 800 1000

bp



cal analysis and helped to draft the manuscript Allauthors read and approved the final manuscript

Additional material

AcknowledgementsMost of the sequencing work was carried out by the first author at the molecular lab of the Alexander Koenig Research Institute and Museum of Zoology in Bonn We thank the late Clas Naumann for supervision and assistance in many ways Bernhard Misof for supervision of the molecular work Esther Meyer Ruth Rottscheidt Meike Thomas Manuela Brenk and Claudia Huber for assistance in DNA sequencing Axel Hille Claudia Etz-bauer Rainer Sonnenberg Anja Schunke and Oliver Niehuis for general assistance in the lab Karen Meusemann Jurate De Prins and Vladimir Lukhtanov for karyological preparations Wolfgang Eckweiler Klaus G Schurian Alexandre Dantchenko John Coutsis Joseacute Munguira and Otakar Kudrna for specimen samples Sabine Fischer for assistance with computer-ized analyses James Mallet and an anonymous reviewer for corrections and helpful comments to the first draft of the manuscript This study was sup-ported by the Deutsche Forschungsgemeinschaft (DFG grant Na 9014)

References1 Hebert PD Cywinska A Ball SL deWaard JR Biological identifica-

tions through DNA barcodes Proc Biol Sci 2003270(1512)313-321

2 Hebert PD Ratnasingham S deWaard JR Barcoding animal lifecytochrome c oxidase subunit 1 divergences among closelyrelated species Proc Biol Sci 2003 270 Suppl 1S96-9[httpwwwjournalsroyalsocacukopenurlaspgenre=articleampid=doi101098rsbl20030025]

3 Ebach MC Holdrege C DNA barcoding is no substitute for tax-onomy Nature 2005 434(7034)697

4 Moritz C Cicero C DNA barcoding promise and pitfalls PLoSBiol 2004 2(10)e354

5 Smith VS DNA barcoding perspectives from a Partnershipsfor Enhancing Expertise in Taxonomy (PEET) debate SystBiol 2005 54(5)841-844

6 Sperling FA DNA Barcoding Deus ex Machina Newsl Biol SurvCanada (Terr Arthropods) 2003 22(2)50-53

7 Will KW Mishler BD Wheeler QD The perils of DNA barcod-ing and the need for integrative taxonomy Syst Biol 200554(5)844-851

8 Will KW Rubinoff D Myth of the molecule DNA barcodes forspecies cannot replace morphology for identification andclassification Cladistics 2004 2047-55

9 Barrett RDH Hebert PD Identifying spiders through DNA bar-codes Can J Zool 2005 83481-491

10 Hebert PD Stoeckle MY Zemlak TS Francis CM Identification ofBirds through DNA Barcodes PLoS Biol 2004 2(10)e312

11 Hebert PD Penton EH Burns JM Janzen DH Hallwachs W Tenspecies in one DNA barcoding reveals cryptic species in theneotropical skipper butterfly Astraptes fulgerator Proc NatlAcad Sci U S A 2004 101(41)14812-14817

12 Smith MA Fisher BL Hebert PD DNA barcoding for effectivebiodiversity assessment of a hyperdiverse arthropod groupthe ants of Madagascar Philos Trans R Soc Lond B Biol Sci 2005360(1462)1825-1834

13 Smith MA Woodley NE Janzen DH Hallwachs W Hebert PD DNAbarcodes reveal cryptic host-specificity within the presumedpolyphagous members of a genus of parasitoid flies (DipteraTachinidae) Proc Natl Acad Sci U S A 2006 103(10)3657-3662

14 Brower AVZ Problems with DNA barcodes for species delim-itation ten species of Astraptes fulgerator reassessed (Lepi-doptera Hesperiidae) Syst Biodiv 2006 4(2)127-132

15 Mayr E Principles of systematic zoology New York McGraw-Hill 1969

16 Hajibabaei M Janzen DH Burns JM Hallwachs W Hebert PD DNAbarcodes distinguish species of tropical Lepidoptera Proc NatlAcad Sci U S A 2006 103(4)968-971

17 Meyer CP Paulay G DNA barcoding error rates based oncomprehensive sampling PLoS Biol 2005 3(12)e422

18 Meier R Shiyang K Vaidya G Ng PKL DNA barcoding and tax-onomy in Diptera a tale of high intraspecific variability andlow identification success Syst Biol 2006 55(5)715-728

19 Pons J Barraclough TG Gomez-Zurita J Cardoso A Duran DPHazell S Kamoun S Sumlin WD Vogler A Sequence-based spe-cies delimitation for the DNA taxonomy of undescribedinsects Syst Biol 2006 55(4)595-609

20 Matz MV Nielsen R A likelihood ratio test for species mem-bership based on DNA sequence data Philos Trans R Soc Lond BBiol Sci 2005 360(1462)1969-1974

21 Nielsen R Matz M Statistical approaches for DNA barcodingSyst Biol 2006 55(1)162-169

22 Hogg ID Hebert PDN Biological identification of springtails(Collembola Hexapoda) from the Canadian Arctic usingmitochondrial DNA barcodes Can J Zool 2004 82749-754

23 Janzen DH Hajibabaei M Burns JM Hallwachs W Remigio E HebertPD Wedding biodiversity inventory of a large and complexLepidoptera fauna with DNA barcoding Philos Trans R Soc LondB Biol Sci 2005 360(1462)1835-1845

24 Monaghan MT Balke M Gregory TR Vogler AP DNA-based spe-cies delineation in tropical beetles using mitochondrial andnuclear markers Philos Trans R Soc Lond B Biol Sci 2005360(1462)1925-1933

25 Monaghan MT Balke M Pons J Vogler AP Beyond barcodes com-plex DNA taxonomy of a South Pacific Island radiation ProcBiol Sci 2006 273(1588)887-893

26 Paquin P Hedin M The power and perils of lsquomolecular taxon-omyrsquo a case study of eyeless and endangered Cicurina (Ara-neae Dictynidae) from Texas caves Mol Ecol 2004133239-3255

27 Scheffer SJ Lewis ML Joshi RC DNA barcoding applied to inva-sive leafminers (Diptera Agromyzidae) in the PhilippinesAnn Entomol Soc Am 2006 99(2)204-210

28 Lesse H Speacuteciation et variation chromosomiques chez lesLeacutepidoptegraveres Rhopalocegraveres Annls Sci nat Zool (seacuter 12) 1960 2(1-14)1-223

29 Lukhtanov VA Kandul NP Plotkin JB Dantchenko AV Haig D PierceNE Reinforcement of pre-zygotic isolation and karyotypeevolution in Agrodiaetus butterflies Nature 2005436(7049)385-389

30 Wiemers M Chromosome differentiation and the radiation ofthe butterfly subgenus Agrodiaetus (Lepidoptera Lycaeni-dae Polyommatus) ndash a molecular phylogenetic approach phDthesis 20031-198 [httphssulbuni-bonndediss_onlinemath_nat_fak2003wiemers_martin] Bonn University of Bonn

31 Lesse H Description de deux nouvelles expegraveces drsquoAgrodiaetus(Lep Lycaenidae) seacutepareacutees agrave la suite de la deacutecouverte deleurs formules chromosomiques Lambillionea 1957 57(910)65-71

32 Lesse H Note sur deux espegraveces drsquoAgrodiaetus (Lep Lycaeni-dae) regravecemment seacutepareacutees drsquoapregraves leurs formules chromo-somiques Lambillionea 1959 59(1-2)5-10

33 Lesse H Les nombres de chromosomes dans la classificationdu groupe drsquoAgrodiaetus ripartii FREYER (LepidopteraLycaenidae) Revue fr Ent 1960 27(3)240-263

34 Lesse H Agrodiaetus iphigenia HS et son espegravece jumelle Atankeri n sp seacutepareacutees drsquoapregraves sa formule chromosomique(Lepid Lycaenidae) Bull Soc ent Mulhouse 1960 196075-78

35 Lesse H Variation chromosomique chez Agrodiaetus dolusHB (Lep Lycaenidae) Alexanor 1962 2283-286

36 Lukhtanov VA Dantchenko A Descriptions of new taxa of thegenus Agrodiaetus Huumlbner [1822] based on karyotype inves-

Additional file 1Neighbour-joining tree (Distance model Kimura-2-Parameter) of profile and test taxa includes a list of GenBank sequences with taxa names and corresponding voucher codesClick here for file[httpwwwbiomedcentralcomcontentsupplementary1742-9994-4-8-S1xls]



Publish with BioMed Central and every scientist can read your work free of charge

BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime

Sir Paul Nurse Cancer Research UK

Your research papers will be

available free of charge to the entire biomedical community

peer reviewed and published immediately upon acceptance

cited in PubMed and archived on PubMed Central

yours mdash you keep the copyright

Submit your manuscript herehttpwwwbiomedcentralcominfopublishing_advasp

BioMedcentral

tigation (Lepidoptera Lycaenidae) Atalanta 2002 33(12)81-107 col pl I

37 Lukhtanov VA Dantchenko AV Principles of the highly orderedarrangement of metaphase I bivalents in spermatocytes ofAgrodiaetus (Insecta Lepidoptera) Chromosome Research 200210(1)5-20

38 Lukhtanov VA Wiemers M Meusemann K Description of a newspecies of the brown Agrodiaetus complex from South-EastTurkey Nota lepid 2003 26(1265-71 [httpwwwsoceurlepcomdownloadspdf_nota_lnota_26_065_071pdf]

39 Olivier A Puplesiene J van der Poorten D De Prins W Wiemers MRevision of some taxa of the Polyommatus (Agrodiaetus) tran-scaspicus group with description of a new species from Cen-tral Anatolia (Lepidoptera Lycaenidae) Phegea 199927(1)1-24

40 Lorkovic Z The butterfly chromosomes and their applicationin systematics and phylogeny In Butterflies of Europe Volume 2Introduction to Lepidopterology Edited by Kudrna O Wiesbaden Aula1990332-396

41 Wiemers M De Prins J Polyommatus (Agrodiaetus) paulae spnov (Lepidoptera Lycaenidae) from Northwest Iran dis-covered by means of molecular karyological and morpho-logical methods Entomol Z 2004 114(4)155-162

42 Schurian KG Zur Biologie Oumlkologie und Taxonomie von Poly-ommatus (Meleageria) daphnis brandti (Pfeiffer 1938) undPolyommatus (Meleageria) daphnis marcida (Lederer 1870)aus Nordiran (Lepidoptera Lycaenidae) Entomol Z 2006116(5)219-225

43 Barcode of Life Data Systems (BOLD) [httpwwwboldsystemsorg]

44 Kandul NP Lukhtanov VA Dantchenko AV Coleman JW Sekerci-oglu CH Haig D Pierce NE Phylogeny of Agrodiaetus Huumlbner1822 (Lepidoptera Lycaenidae) inferred from mtDNAsequences of COI and COII and nuclear sequences of EF1-alpha karyotype diversification and species radiation SystBiol 2004 53(2)278-298

45 Weingartner E Wahlberg N Nylin S Speciation in Pararge(Satyrinae Nymphalidae) butterflies ndash North Africa is thesource of ancestral populations of all Pararge species Syst Ent2006 31(4)621-632

46 Megens HJ van Nes WJ van Moorsel CHM Pierce NE Molecularphylogeny of the Oriental butterfly genus Arhopala (Lycaeni-dae Theclinae) inferred from mitochondrial and nucleargenes Syst Entomol 2003 29115-131

47 Whinnett A Zimmermann M Willmott KR Herrera N Mallarino RSimpson F Joron M Lamas G Mallet J Strikingly variable diver-gence times inferred across an Amazonian butterfly suturezone Proceedings of the Royal Society B 2005 2722525-2533

48 Funk DJ Omland KE Species-level paraphyly and polyphylyFrequency causes and consequences with insights fromanimal mitochondrial DNA Annu Rev Ecol Evol Syst 200334397-423

49 Hickerson MJ Meyer CP Moritz C DNA barcoding will often failto discover new animal species over broad parameter spaceSyst Biol 2006 55(5)729-739

50 Megens HJ van Moorsel CH Piel WH Pierce NE de Jong R Tempoof speciation in a butterfly genus from the Southeast Asiantropics inferred from mitochondrial and nuclear DNAsequence data Mol Phylogenet Evol 2004 31(3)1181-1196

51 Sturmbauer C Meyer A Genetic divergence speciation andmorphological stasis in a lineage of African cichlid fishesNature 1992 358578-581

52 National Center for Biotechnology Information [httpwwwncbinlmnihgov]

53 MorphBank [httpwwwmorphbanknet]54 Als TD Vila R Kandul NP Nash DR Yen SH Hsu YF Mignault AA

Boomsma JJ Pierce NE The evolution of alternative parasiticlife histories in large blue butterflies Nature 2004432(7015)386-390

55 Gompert Z Nice CC Fordyce JA Forister ML Shapiro AM Identi-fying units for conservation using molecular systematics thecautionary tale of the Karner blue butterfly Mol Ecol 200615(7)1759-1768

56 Eastwood R Hughes JM Molecular phylogeny and evolutionarybiology of Acrodipsas (Lepidoptera Lycaenidae) Mol Phylo-genet Evol 2003 27(1)93-102

57 Eastwood R Pierce NE Kitching RL Hughes JM Do ants enhancediversification in Lycaenid butterflies Phylogeographic evi-dence from a model myrmecophile Jalmenus evagoras Evolu-tion 2006 60(2)315-327

58 Rand DB Heath A Suderman T Pierce NE Phylogeny and life his-tory evolution of the genus Chrysoritis within the Aphnaeini(Lepidoptera Lycaenidae) inferred from mitochondrialcytochrome oxidase I sequences Mol Phylogenet Evol 200017(1)85-96

59 Caterino MS Reed RD Kuo MM Sperling FA A partitioned likeli-hood analysis of swallowtail butterfly phylogeny (Lepidop-teraPapilionidae) Syst Biol 2001 50(1)106-127

60 Vila M Bjorklund M The utility of the neglected mitochondrialcontrol region for evolutionary studies in Lepidoptera(Insecta) J Mol Evol 2004 58(3)280-290

61 Caterino MS Sperling FA Papilio phylogeny based on mitochon-drial cytochrome oxidase I and II genes Mol Phylogenet Evol1999 11(1)122-137

62 Hall TA BioEdit a user-friendly biological sequence align-ment editor and analysis program for Windows 9598NTNucl Acids Symp Ser 1999 4195-98

63 Kumar S Tamura K Nei M MEGA3 Integrated software forMolecular Evolutionary Genetics Analysis and sequencealignment Briefings in Bioinformatics 2004 5150-163

64 Nei M Kumar S Molecular Evolution and PhylogeneticsOxford Oxford Univ Press 2000

65 Campbell DL Brower AV Pierce NE Molecular evolution of thewingless gene and its implications for the phylogenetic place-ment of the butterfly family Riodinidae (Lepidoptera Papil-ionoidea) Mol Biol Evol 2000 17(5)684-696

66 Eliot JN The higher classification of the Lycaenidae (Lepidop-tera) a tentative arrangement Bulletin of the British Museum(Natural History) Entomology 1973 28(6)371-505

67 Wahlberg N Braby MF Brower AV de Jong R Lee MM Nylin SPierce NE Sperling FA Vila R Warren AD Zakharov E Synergisticeffects of combining morphological and molecular data inresolving the phylogeny of butterflies and skippers Proc BiolSci 2005 272(1572)1577-1586

68 Lukhtanov VA Vila R Rearrangement of the Agrodiaetus dolusspecies group (Lepidoptera Lycaenidae) using a new cyto-logical approach and molecular data Insect Syst Evol 200637325-334


Abstract

Background

Results

Conclusion

Background

Results

Intraspecific divergence

Interspecific divergence

The barcode gap

Identification with NJ tree profile

Conclusion

Methods

Data sources

Laboratory protocols

Data analysis

Competing interests

Authors contributions

Additional material

Acknowledgements

References


however this approach received opposition especiallyfrom the taxonomists community [3-8] Some argumentsin this debate are political in nature others have a scien-tific basis Concerning the latter one of the most essentialarguments focuses on the so-called barcoding gapAdvocates of barcoding claim that interspecific geneticvariation exceeds intraspecific variation to such an extentthat a clear gap exists which enables the assignment ofunidentified individuals to their species with a negligibleerror rate [1910] The errors are attributed to a smallnumber of incipient species pairs with incomplete lineagesorting (eg [11]) As a consequence establishing thedegree of sequence divergence between two samplesabove a given threshold (proposed to be at least 10 timesgreater than within species [10]) would indicate specificdistinctness whereas divergence below such a thresholdwould indicate taxonomic identity among the samplesFurthermore the existence of a barcoding gap would evenenable the identification of previously undescribed spe-cies ([11-13] but see [14]) Possible errors of thisapproach include false positives and false negatives Falsepositives occur if populations within one species aregenetically quite distinct eg in distant populations withlimited gene flow or in allopatric populations with inter-rupted gene flow In the latter case it must be noted thatdepending on the amount of morphological differentia-tion and the species concept to be applied such popula-tions may also qualify as cryptic species in the view ofsome scientists False negatives in contrast occur whenlittle or no sequence variation in the barcoding fragmentis found between different biospecies (= reproductivelyisolated population groups sensu Mayr [15]) Hence falsenegatives are more critical for the barcoding approachbecause the existence of such cases would reveal exampleswhere the barcoding approach is less powerful than theuse of other and more holistic approaches to delimit spe-cies boundaries

Initial studies on birds [10] and arthropods [916]appeared to corroborate the existence of a distinct barcod-ing gap but two recent studies on gastropods [17] andflies [18] challenge its existence The reasons for these dis-crepancies are not entirely clear Although levels of COIsequence divergence differ between higher taxa (eg anexceptionally low mean COI sequence divergence of only10 was found in congeneric species pairs of Cnidariacompared to 96ndash157 in other animal phyla [2]) Mol-lusca (with 111 mean sequence divergence betweenspecies) and Diptera (93) are not peculiar in thisrespect Meyer amp Paulay [17] assume that insufficient sam-pling on both the interspecific and intraspecific level cre-ate the artifact of a barcode gap Proponents of barcodingmight argue however that the main reason for this over-lap is the poor taxonomy of these groups eg cryptic spe-cies may have been overlooked which are differentiated

genetically but very similar or even identical in morphol-ogy

If the barcode gap does not exist then the thresholdapproach in barcoding becomes inapplicable Althoughmore sophisticated techniques (eg using coalescence the-ory and statistical population genetic methods [19-21])can sometimes help to delimit species with overlappinggenetic divergences these approaches require additionalassumptions (eg about the choice of population geneticmodels or clustering algorithms) and are only feasible inwell-sampled clades

Barcoding holds promise nonetheless especially in theidentification of arthropods the most species-rich animalphylum in terrestrial ecosystems Identification of arthro-pods is often extremely time-consuming and generallyrequires taxonomic specialists for any given group More-over the fraction of undescribed species is particularlyhigh as opposed to vertebrates Hence there is substantialdemand for improved (and rapid) identification tools byscientists who seek identification of large arthropod sam-ples from complex faunas Therefore arthropods deserveto be considered the yard-stick for the usefulness of bar-coding approaches among Metazoa and it is not surpris-ing that several recent studies have tried to apply DNAbarcoding in arthropods [911-1316181922-27] Diver-sity is concentrated in tropical ecosystems but measuringintra- and interspecific sequence divergence in tropicalinsects is hampered by the fragmentary knowledge ofmost taxa In contrast insects of temperate zones andmost notably the butterflies of the Holarctic region arewell known taxonomically compared to other insects Thespecies-rich Palaearctic genus (or subgenus) Agrodiaetusprovides an excellent example to test the existence of thebarcode gap in arthropods This genus is exceptionalbecause of its extraordinary interspecific variation in chro-mosome numbers which have been investigated for mostof its ca 120 species ([28-30] and references therein) As aresult several cryptic species which hardly or not at all dif-fer in phenotype have been discovered (eg [31-39])Available evidence suggests that apart from a few excep-tions (eg due to supernumerary chromosomes) differ-ences in chromosome numbers between butterfly speciesare linked to infertility in interspecific hybrids [40] This isdue to problems in the pairing of homologous chromo-somes during meiosis Since major differences in chromo-some numbers are indicative of clear species boundariesthey are helpful also to infer species-level differentiationfor allopatric populations Agrodiaetus butterflies thereforeare an ideal case for testing the validity of the barcodingapproach If valid then it must be possible to safely recog-nize all species that can be distinguished by phenotypekaryotype or both character sets with reference tosequence divergences alone On the contrary failure of



















2 00 --- Mono



























0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance








0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance










0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value









0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value














0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus



















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References


















2 00 --- Mono



























0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance








0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance










0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value









0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value














0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus



















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References

























0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance








0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance










0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value









0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value














0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus



















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References






0

5

10

15

20

250

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

K2P Distance










0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value









0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value














0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus



















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References








0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value









0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value














0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus



















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References







0

10

20

30

40

50

60

70

80

90

1000

00

40

81

21

62

02

42

83

23

64

04

44

85

25

66

06

46

87

27

68

08

48

89

29

610

0

threshold value














0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus



















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References












0

10

20

30

40

50

60

70

80

90

000

100

200

300

400

500

600

700

800

901 01

1

K2P Distance

Nu

mb

er o

fsp

ecie

s

Others

Arhopala

Agrodiaetus



















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References


















Table 5 Material







52 [46 50] Arhopala











0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References








0

5

10

15

20

25

0 200 400 600 800 1000

bp




Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References



Additional material



















































BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References











BioMedcentral



































Abstract

Background

Results

Conclusion

Background

Results



The barcode gap


Conclusion

Methods

Data sources


Data analysis

Competing interests


Additional material

Acknowledgements

References

Does the DNA barcoding gap exist? – a case study in blue butterflies (Lepidoptera: Lycaenidae)

Documents

Transcript of Does the DNA barcoding gap exist? – a case study in blue butterflies (Lepidoptera: Lycaenidae)