Phylogenetic relationships of the bumblebees Bombus moderatus , B . albocinctus , B . burjaeticus , B . florilegus and B . cryptarum based on mitochondrial DNA markers : a complex of closely related taxa with circumpolar distribution ( Hymenoptera : Apidae : Bombus )

Königinnen von Bombus moderatus aus Alaska/USA und Alberta/Canada und von B. burjaeticus and B. patagiatus aus dem Russischen Transbaikal wurden an verschiedenen Orten im Frühjahr gefangen, um künstliche Kolonien zu züchten. Zusätzlich wurden Männchen von B. florilegus in Hokkaido/Japan gesammelt. Teilsequenzen (Länge 1005 bp) mitochondrialer Cytochrome Oxidase Untereinheit I (COI) wurde sequenziert. Zum Vergleich wurden auch Museumsproben von B. albocinctus und B. burjaeticus sequenziert. Die Divergenz der Sequenzen innerhalb der Taxa beträgt 1 bis 2 Basen-Substitutionen und die Tamura-Nei Genetische Distanz 0.001–0.002. Die Divergenz der Sequenzen zwischen B. moderatus, B. albocinctus und B. burjaeticus beträgt nur 1–5 Basen-Substitutionen und die Tamura-Nei Genetische Distanz 0.001–0.005, während die Divergenz der Sequenzen zwischen B. lucorum, B. magnus, B. patagiatus und B. cryptarum 22-44 Basen Substitutionen beträgt und die Tamura-Nei Genetische Distanz 0.027-0.042. Zusätzlich zu den Clustern für B. lucorum, B. magnus und B. patagiatus zeigt die Topologie des Phylogramms (MrBayes Maximum Likelihood Tree) ein umfangreiches Cluster, in dem B. albocinctus, B. burjaeticus, B. moderatus und B. florilegus vereinigt sind. Da die COI Sequenzen keine Lücken aufweisen, können die einzelnen Nukleotide wie homologe Positionen verwendet werden. Jedes Taxon besitzt 8–20 eigene Substitutionen, die als diagnostische Positionen zur Charakterisierung des Taxons verwendet werden können. Die Analyse der diagnostischen Positionen der Taxa bestätigt die Topologie des Maximum Likelihood Phylogramms. Um die Lücke zwischen den östlichsten bekannten Vorkommen von B. cryptarum im Kaukasus und im Elburz und den Vorkommen von B. burjaeticus/B.albocinctus im Russischen Transbaikal und in Russisch Fernost zu überbrücken, wurden 12 weitere Museumsproben aus Zentralasiatischen Gebirgen und dem Himalaja sequenziert. Durch Analyse der diagnostischen Positionen der teilweise 100 Jahre alten DNA kann gezeigt werden, dass keine dieser Proben zu B. lucorum gehören kann, alle bilden ein Cluster mit den Taxa des cryptarum Komplexes.


Introduction
The taxon Bombus cryptarum (FABRICIUS 1775), initially described from the type locality Hafniae (Copenhagen), re-described as B. lucorum var.pseudocryptarum SKORIKOV (1913) from Poland SKORIKOV (1913) from Poland SKORIKOV and Russia, and as B. lucorum var.lucocryptarum BALL (1914) from Belgium, was established as a species by RASMONT (1981RASMONT ( , 1983)).It took another 25 years until this species, with a European distribution, was accepted as separate from B. lucorum by morphology, male labial gland secretions and DNA sequences (BERTSCH 1997;BERTSCH et al. 2004BERTSCH et al. , 2005;;BERTSCH 2009).Bombus cryptarum is distributed throughout Europe.It is more or less abundant in parts of Great Britain, the Netherlands, Benelux, parts of France, in Northern and Middle Germany, Scandinavia, Finland, Poland and Lithuania.The distribution in Belarus and Russia still has to be investigated, but museum specimens and fieldwork in St. Petersburg and Moscow show that the species is also common there.Museum specimens from Tscheljabinsk/Russia make its distribution throughout Northern Russia very probable, at least till the Urals.The available data also show that B. cryptarum is less abundant in the Southern parts of these countries.The abun dance and distribution in the Alps of France, Switzerland and Austria still has to be fully in vestigated, otherwise the alpine distribution continues through the Balkan Mountains and the highlands of North Eastern Anatolia into the Caucasus and the Talesh and Elburz Mountains in Iran.
By investigating isoenzyme markers, SCHOLL et al. (1990) were able to show that the North American B. moderatus CRESSON, 1863, was closely related to the European B. cryptarum, a view confirmed by DNA sequences (BERTSCH et al. 2010).These sequences also showed that the Russian Far Eastern taxon B. albocintus SMITH, 1854, and the North American B. moderatus belong to the same taxon, and that this Pan-Pacific eratus belong to the same taxon, and that this Pan-Pacific eratus B. albocinctus is not the Far Eastern B. albocinctus is not the Far Eastern B. albocinctus representative of B. lucorum as discussed by VOGT (1911), BISCHOFF (1930), KRÜGER (1951), GER (1951), GER TKALCU (1974), LELEJ & KUPIANSKAYA (2000), DAVYDOVA (2001), andDAVYDOVA &PESENKO (2002) but is closely related, if not conspecific, to the European B. cryptarum.Between the specimens B. cryptarum from the Urals, the Caucasus and the Talesh and Elburz Mountains, and the Russian Far Eastern specimens of B. albocinctus/burjaeticus, a gap of about 5000 km has to be closed.
To investigate the relationships between the European B. cryptarum and the Russian Far Eastern B. albocinctus, we obtained partial sequences from the mitochondrial cytochrome oxidase subunit I (COI) gene of Bombus sensu stricto taxa from the Russian Transbaikal region, the Russian Far East and museum specimens from the Central Asiatic Mountains (Alai, Kirghizian and Dzungarian Alatau) and the Himalayas, and addressed the following issues: (1) whether the Russian Far East Genbank sequences designated as B. cryptarum were identical with the taxon B. albocinctus; (2) whether there were close relationships between B. albocinctus and the Siberian taxon B. patagiatus NYLANDER, 1948, the Transbaikal region NYLANDER, 1948, the Transbaikal region NYLANDER taxon B. burjaeticus KRÜGER, 1951and GER, 1951 and GER the taxon B. florilegus PANFILOV, 1956(= B. terrestris var. japonica B. terrestris var. japonica B. terrestris FRIESE 1909) from the Kuril Islands and Hokkaido; and (3) whether it was possible to close the gap between the Transbaikal region taxa and the B. cryptarum specimens from Anatolia, the Caucasus and Elburz Mountains using museum specimens from the Central Asiatic Mountains.

Bumblebee samples
Table 1 shows the bumblebee samples used for DNA sequencing with their identification codes and the localities of their origins with geographical coordinates.Figure 1 shows a pacific-centred world map with the distributions of B. cryptarum,B. burjaeticus,B. albocintus,and B. moderatus. Figure 6 shows the distribution of B. florilegus between Hokkaido/Japan and the North Kuril Islands.From the museum collections, 18 specimens were also sequenced and Table 2 gives their identification codes, their original designations and the localities of their origins with geographical coordinates.The localities for these museum specimens with their identification numbers are also included in Figure 1.In order to check the identity

Polymerase chain reaction (PCR) and DNA sequencing of mitochondrial COI
Total DNA was extracted from legs using the QIAamp® DNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's specifications for tissue, and eluted in 150 µl of highly purified water (Ampuwa®, Fresenius Kabi, Bad Homburg, Germany).For sequence analysis, overlapping fragments (in all 1027 bp) of mitochondrial COI were amplified using primers specifically designed for Bombus.BO-1-fwd (5' TAGGATCACCAGATATAGC 3') and BO-K-rev (5' GAGCTCAAACAATAAATCC 3') resulted in the amplification of a 609 bp fragment, whereas BO-5-fwd (5' AATGAAAGAGGTAAAAAAGAAAC 3') and BO-A-rev (5' ATGTTGAGGGAAAAATGTTAT 3') resulted in the amplification of a 510 bp fragment.Polymerase Chain Reaction amplifications were performed as described (BERTSCH et al. 2010).A sample of 10 µl from each reaction was checked on a 1 % agarose gel.Polymerase Chain Reaction products were purified using the AMPure® PCR Purification Kit (Agencourt, Beverly, MA, USA).Both strands were sequenced for all specimens.Sequencing reactions were performed using ABI® BigDye Terminator version 3.1 chemistry (Applied Biosystems, Foster City, CA, USA) according to manufacturer's instructions and they were then analysed on an ABI 3100 sequencer (Applied Biosystems).Sequences were aligned using CLUSTALX.No gaps or poorly aligned regions occurred in the alignment, but missing characters were trimmed from the ends of the alignment to produce an equal sequence length of 1005 bp for all individuals (encoding 335 amino acids).Individual alignments were aligned against the complete COI gene sequence of Bombus ignitus between positions 262 and 1267 (GenBank DQ870926;CHA et al. 2007). CHA et al. 2007).CHA

Degraded DNA from museum specimens
The 40-100 year old DNA from the museum specimens was more or less degraded.The integration of these sequences into MrBayes maximum likelihood simulations did not give useful results; depending on the grade of degradation, the specimens clustered in unpredictable ways.Therefore, the original ABI traces of these sequences had to be carefully inspected for degraded positions; whenever double peaks were detected the IUPAC Code for mixed bases (wobbles) was inserted.
Where necessary, the Polymerase Chain Reaction was repeated with slightly adjusted amplification conditions, and by this procedure it was possible to get reliable sequences for the control specimens (M13-1, M13-2 and M14-M16), which were suitable for use in MrBayes simulations.About 97 % (863 out of 891) of the positions in the sequences from the fresh material were invariant, therefore, for MrBayes simulations with the rest of the museum specimens, the variability of the degraded DNA was reduced by deleting all the invariant triplets (of the fresh, not the degraded DNA) and only the triplets with parsimony informative positions were included in the analysis.

Analysis of sequence divergence of mitochondrial COI
The absolute numbers of substitutions were counted based on a pair-wise comparison of the sequences.The analysis of sequence diversity was performed using the maximum composite likelihood method of MEGA 4.0 (TAMURA et al. 2007;KUMAR et al. 2008).Because the distribution KUMAR et al. 2008).Because the distribution KUMAR of nucleotides in the COI of Hymenoptera is known to be heterogeneous, with a strong A + T bias, we selected the Tamura-Nei model of base substitution (TAMURA & NEI 1993) which corrects this bias in its assumption of sequence evolution.The nucleotide frequencies and the parameters necessary for this model were estimated from the sequence data and Tamura-Nei genetic distances were calculated.The best maximum likelihood model was selected using the JMODELTEST (POSADA 2008).The tree topology was inferred by a maximum likelihood tree based on the general time-reversible model (GTR plus gamma), calculated by Bayesian analysis using MrBayes (HUELSENBECK & RONQUIST 2001).Tree topology was also calculated as a Minimum Evolution tree (ME) with bootstrap sampling using MEGA 4.0 (TAMURA et al. 2007).GENEIOUS PRO 4.5 (Biomatters Ltd.) was used to analyse the alignment and detect diagnostic positions and the GREENBUTTON plugin (InterGrid) was used to do the time consuming MrBayes calculations on a supercomputer cluster.The nucleotide changes along the branches

Nucleotide frequencies and substitution parameters
The aligned data matrix of 36 sequences (B.moderatus, B. albocinctus, B. florilegus and B. florilegus and B. florilegus B. burjaeticus) of 891 bp length (encoding 297 amino acids) included 863 (97 %) invariants and 28 variable cus) of 891 bp length (encoding 297 amino acids) included 863 (97 %) invariants and 28 variable cus sites.Of these 28 variable positions, 3 were parsimony uninformative (noise) and 25 parsimony informative (signal).However, differences in this pattern were evident in codon positions, where only 1 informative site was in first position and all of the other 24 informative variant sites were in third position.None of these sites was a replacement position; the singleton replacement position 1007 in the GenBank sequence AF279485 was most probably an amplification error or a miscoding lesion.Table 3 shows the pattern of nucleotide substitutions estimated from the data with the maximum composite likelihood model (MEGA).The nucleotide frequencies were 33.1 (A), 42.1 (T), 12.3 (C) and 12.5 (G), which proves the known strong A + T bias typical for sequences of Hymenoptera.

COI sequence divergence between and within species
Table 4 presents  B. burjaeticus was clearly B. burjaeticus distinct but not a separate species.All these taxa had a Tamura-Nei genetic distance between 0.012 and 0.018 to B. cryptarum, again with the largest genetic distance for the geographically separated B. moderatus, therefore it seemed to be useful to treat all these taxa as members of a closely related complex of taxa, the cryptarum-complex.Bombus patagiatus was well separated from this cryptarum-complex by a Tamura-Nei genetic distance of 0.031 to 0.040 and B. magnus by a Tamura-Nei genetic distance of 0.030 B. magnus by a Tamura-Nei genetic distance of 0.030 B. magnus to 0.037, comparable with B. lucorum, which was separated by a Tamura-Nei genetic distance of 0.036 to 0.040.These Tamura-Nei genetic distances are larger by one order of magnitude and revealed that all three taxa are separate species.

Tree building by maximum likelihood models
The maximum likelihood tree (Fig. 2) generated using the Bayesian Markov Chain Monte Carlo (MCMC) analysis with 1005 bp full-length sequences was based on the general time-reversible model of base substitution (GTR plus gamma

Tree building by diagnostic characters
Because there are no gaps in the alignments of COI sequences, single nucleotide sites can be used as positional homologies (HILLIS 1994).The alignment file (Fig. 3) showed that each taxon was characterised by substitutions which were unique ("private") and could be used as diagnostic characters to define and identify this taxon (FUNK et al. 2001;FUNK et al. 2001;FUNK BERTSCH 2009).In MACCLADE, all of the changes at the nodes and the diagnostic characters at the last branch of the terminal units can be investigated in detail, and a tree with the classical tools for morphological characters can be built (Fig. 4).The conclusions drawn from the genetic distance data and the clustering of the maximum likelihood tree were fully confirmed by analysis of the diagnostic characters.

Degraded DNA of museum specimens
As the alignment of the fresh material showed that 97 % of the positions were invariable, we removed all invariable triplets from the alignment of the museum specimens and generated a maximum likelihood tree (Fig. 5) with the variable triplets using the Bayesian MCMC analysis with the general time-reversible model of base substitution (GTR plus gamma  (GenBank DQ870926;CHA et al. 2007).This procedure implied the assumption that CHA et al. 2007).This procedure implied the assumption that CHA the museum specimens belong to one of the taxa investigated in this report.
The control specimens M-13-1 and M-13-2 of the type series B. burjaeticus and the con-  B. reinigi also joined the cluster γ with the specimens of the cryptarum-complex taxa.The posterior probabilities were high; therefore, the conclusion that these specimens are not connected to B. lucorum but rather to B. cryptarum respectively to the cryptarum-complex taxa is reasonable.

Bombus patagiatus
This species was first described by NYLANDER (1848) and afterwards was neglected for a long time NYLANDER (1848) and afterwards was neglected for a long time NYLANDER until it was established as a good species by TKALCU (1967).The most obvious characteristic is the typical combination of yellow-white colouration (collare, scutellum, tergite), which is not a very useful feature in bumblebee taxonomy of critical taxa.TKALCU (1967) described morphological characteristics for this taxon (form of labrum, punctuation of vertex and ocellar field, ocellar distance, proportions of flagellum segments) and the differences between these characters compared to those of B. lucorum, but nearly all of these characteristics were so similar that terms such as 'a little bit more' and 'a little bit larger' ore 'sometimes a bit larger' were the best that could be said

Is Bombus florilegus a separate species?
Bombus florilegus a separate species?Bombus florilegus The taxon Bombus terrestris var.japonica FRIESE, 1909, described from Yesso (= Hokkaido) and re-described as B. florilegus PANFILOV, 1956PANFILOV, (japonicus , 1956PANFILOV, (japonicus , 1956 ( = name ( = name preoccupied; Dalla Torre 1890) japonicus = name preoccupied; Dalla Torre 1890) japonicus with morphological characters of the sculpture of workers and the male genitalia, is characterised by the completely black abdomen of the females.ITO & SAGAKAMI (1980)  Bombus florilegus is a taxon with a very restricted habitat.Figure 6 shows the distribution from Bombus florilegus is a taxon with a very restricted habitat.Figure 6  Bombus burjaeticus was described by Krüger (from type locality Kulskoje/Burjatia) mainly by characters of colouration.In the VOGT collection (ZMA, Amsterdam) there is a large number of specimens from the type locality Kulskoje (353 queens, collected 19.V.1928 by Klemm, Kulskoje, Uda Valley, Pinus forest), most in very good condition.A thorough investigation of the morpho-Pinus forest), most in very good condition.A thorough investigation of the morpho-Pinus logical characters still remains to be done, and it might well be that different taxa are included in this Kulskoje material, but one characteristic mentioned in Krüger's original description was quite distinctive: many specimens showed a typical greenish tint of the yellow parts of the colouration, and the dorsal part of the episternum was yellow, both characteristics typical for B. cryptarum.DOI: 10.21248/contrib.entomol.60.1.13-32 Two queens from the type series conforming to the description of Krüger were sequenced (M-13-1 and M-13-2).The 80year-old DNA was degraded but none of the 17 diagnostic characters of B. lucorum could be found (Fig. 3 B. turkestanicus B. jacobsoni, B. reinigi).In a group where differentiation using morphological characters is as delicate as with the many taxa resembling and somehow related to B. lucorum, this approach leads to unsatisfactory results.According to WILLIAMS (1991:84), at least 187 names have been published concerning B. lucorum 'in the broadest sense'.Much more fieldwork is necessary but fieldwork alone will not help to clarify the situation.Collecting fresh material in the Transbaikal region and naming these specimens B. burajeticus because of resemblances in colouration with the type material of Krüger is not suf-burajeticus because of resemblances in colouration with the type material of Krüger is not suf-B.burajeticus ficient because the characters available are not conclusive.More detailed information is required and it is necessary to connect designated museum specimens with fresh material collected in the field.DNA and especially COI sequences are obvious tools which are suited to solve at least some of these problems.By carefully adjusting the amplification conditions and manually inspecting the ABI traces at the positions with miscoding lesions, it is possible to get reliable sequences from museum specimens, thus, connecting 80-year-old museum specimens and fresh material is possible.However, the standard sequence length (648 bp) used for DNA bar-coding identification by COI might be a bit too short for such purposes, and we think that in the case of museum specimens, with possibly many degraded positions, it would be safer to use sequences of about double that bar-coding length (about 1200 bp) and to carefully compare the diagnostic positions calibrated against many sequences from fresh specimens from a broad spectrum of geographical provenances.

Conclusions
From the taxa investigated in this report, it should now be possible to use their diagnostic characters to relate unambiguously specimens to taxa, and the next step must be to find morphological characters useful for safe identification.Up until now, this task has given unsatisfactory results because the separation of the variability of characters and the delimitation of characters useful for diagnostic work has been impossible without proper designation of the specimens.The number of misidentifications in museum collections is substantial.Consequently, arguments over identification have quite often been circular and a critical re-evaluation of all the work invested into the morphological separation of taxa somehow related to B. lucorum must be done.Just the invention of a B. lucorum 'in the broadest sense' (WILLIAMS 2010) illustrates the problem, and reminds of times not so far back, when, because of difficulties of morphological separation, B. terrestris and B. terrestris and B. terrestris B. lucorum from the European continent (both with white tails) were only tentatively separated as species (HOFFER 1883;HOFFER 1883;HOFFER SCHMIEDEKNECHT, 1930;KRÜGER 1920) or not at all (FAESTER & FAESTER 1970;FAESTER 1970;FAESTER WARNKE 1981).But fortunately nobody invented a B. terrestris 'in the broadest B. terrestris 'in the broadest B. terrestris sense'.The premature mixing of clearly separable taxa (whatever species definition is used) prohibits scientific progress and insights into biological diversity.Definable unities in nature are the objects of interest; they should be properly named to promote further investigations into their ecology and distribution.Even in museum boxes it is preferable to keep taxa separate.Divide et Impera might also be good strategy to follow in bumblebee taxonomy.
the matrix of genetic distances estimated by the Tamura-Nei model and as p-distances within and between the five taxa investigated: B. cryptarum, B. florilegus, B. albocinctus, B. moderatus and B. burjaeticus.For comparison and discussion the sequences of the European species, B. lucorum (LUC-01-LUC-03), B. magnus (MAG-01-MAG-03) and the Asiatic species B. patagiatus ( B. patagiatus ( B. patagiatus PAT-01-PAT-03) were included in the analysis.The intraspecific genetic variability was low for all taxa (1-2 nucleotides, Tamura-Nei distance 0.001-0.002).By contrast, the interspecific genetic variability between B. lucorum, B. magnus, B. patagiatus and B. patagiatus and B. patagiatus B. cryptarum was approximately one order of magnitude larger (25-44 nucleotides, Tamura-Nei distance 0.027-0.042).The Tamura-Nei genetic distance between B. burjaeticus and B. burjaeticus and B. burjaeticus B. albocinctus is only 0.002, both B. albocinctus is only 0.002, both B. albocinctus taxa are conspecific, and B. moderatus, with a Tamura-Nei genetic distance to B. albocinctus/ B. burjaeticus of 0.004, also belongs to the same taxon; the larger genetic distance manifested the B. burjaeticus of 0.004, also belongs to the same taxon; the larger genetic distance manifested the B. burjaeticus geographic separation of this taxon by the Bering Strait.Bombus florilegus, with a Tamura-Nei genetic distance between 0.010 and 0.014 to B. moderatus/B.albocinctus/B.burjaeticus was clearly

Fig. 2 :
Fig. 2: Tree topology calculated as Maximum-Likelihood tree using Bayesian MCMC analysis with the general time reversal model of base substitutions with gamma distribution.(M-14, M-15, M-16) and Chuchotka (M-17) from the collection of the Zoological Museum of the Academy of Sciences St. Petersburg (ZMAS) were sequenced.Two queens from the type series Kulskoje (M-13-1 & M-13-2) of B. burjaeticus, from the Vogt collection of the Zoölogical B. burjaeticus and the con-B.burjaeticus trol specimens M-14 to M-17 of B. albocinctus joined the cluster B. albocinctus joined the cluster B. albocinctus γ 3 with the fresh material of B. moderatus / B. moderatus / B. moderatus B. albocinctus / B. burjaeticus, which proved that deletion of the invariable sites does not change the overall result.Depending on quality and length (sequences M-02 & M-05 were too short to be included) of the degraded DNA, the specimens M-01, M-03, M-04 from Turkey, the Caucasus and Iran, areas from where B. cryptarum has been identified, joined the B. cryptarum cluster γ 1 or the cryptarum-complex cluster γ.All of the remaining specimens from the Central Asiatic mountains and the Himalayas whether identified as subspecies of B. lucorum (M-06, M-09, M-12), as subspecies of B. magnus (M-07 and M-08) or as separate species B. magnus (M-07 and M-08) or as separate species B. magnus B. reinigi (M-10) B. reinigi (M-10)

Fig. 3 :
Fig. 3: Alignment of all parsimonious informative triplets (with uninformative sites deleted -), and with a pointer for position number (numbered for total COI) and codon position.Diagnostic (= private) positions marked with colour green = Thymine, violet = Cytosine, red = Adenine and yellow = Guanine.

Fig. 4 :
Fig. 4: Observed diagnostic character changes with position numbers mapped onto the Maximum-Likelihood tree.Black box = unambiguous diagnostic character change, grey box = ambiguous diagnostic character change, grey box = ambiguous diagnostic character diagnostic character change, and white box = character change.character change, and white box = character change.character

Fig. 5 :
Fig. 5: Tree topology calculated as Maximum-Likelihood tree using Bayesian MCMC analysis with the general time reversal model of base substitution and gamma distribution for degraded DNA of museum specimens, only parsimony informative triplets included.
listed the morphological differences of females and males of B. florilegus compared to B. florilegus compared to B. florilegus B. albocinctus and B. albocinctus and B. albocinctus B. lucorum, but the variation in most of the characters remains unclear and many of them seem to overlap.
In a short remarkRASMONT et al.  (1986:677)  noted 'the resemblance of the sternite 8 of certain B. burjaeticus males with B. burjaeticus males with B. burjaeticus B. cryptarum' and they alsonoted (1986:678)  that the females of B. albocinctus from Sakhalin B. albocinctus from Sakhalin B. albocinctus are not B. lucorum but are morpho logically closely conform to B. cryp tarum.Specimens of B. albocinctus from Magadan, Kamchatka and the North Kuril Islands deposited in the B. albocinctus from Magadan, Kamchatka and the North Kuril Islands deposited in the B. albocinctus year 2000 in GenBank by Ito & Tanaka have been identified by morphological characters as B. cryptarum (M.Ito in e-mail correspondence).All of these findings and the conclusions corresponded with the results of our sequencing: B. albocinctus and B. albocinctus and B. albocinctus B. burjaeticus belong to B. burjaeticus belong to B. burjaeticus the same taxon, the white colouration of some Kamchatka and Far Eastern specimens is not a specific characteristic and both taxa are genetically very close to the European B. cryptarum.And as the North American taxon B. moderatus also belongs to B. moderatus also belongs to B. moderatus B. albocinctus ( B. albocinctus ( B. albocinctus BERTSCH et al. 2010), we have a taxon with a Pan-Pacific distribution.Unless possible connections between the European B. cryptarum and these Far Eastern taxa are investigated in detail, the best taxonomic conclusions are to treat B. burjaeticus KRÜGER as a junior synonym of GER as a junior synonym of GER B. albocinctus SMITH and B. moderatus as B. albocinctus moderatus CRESSON.

Fig. 6 :
Fig. 6: Distribution of B. florilegus (red) from Cape Nemuro/ B. florilegus (red) from Cape Nemuro/ B. florilegus Hokkaido through the Southern Kuril Islands and of B. albocinctus (blue) from Kamchatka through the Northern Kuril Islands.Arrows = zone of contact.
).For comparison, GenBank data of the European taxa B. lucorum, B. cryptarum and B. magnus were included.This tree confirmed the results from B. magnus were included.This tree confirmed the results from The unpublished sequences from Magadan, Kamchatka and the Northern Kuril Islands, Shumshu and Paramushir, submitted to GenBank in 2000 by Ito & Tanaka, were identified by morphological characteristics as specimens of B. cryptarum by Masao Ito.As so far there are no known specimens of typical B. cryptarum from these areas, it seemed necessary to prove the identity of these GenBank specimens with typical B. albocinctus.Therefore four typical specimens (collare, scutellum and second tergite white-coloured) of B. albocinctus from Kamchatka B. albocinctus from Kamchatka B. albocinctus

albocinctus, a Pan-Pacific species with four taxa
shows the distribution from Bombus florilegus Cape Nemuro at Hokkaido through the South Kuril Islands where B. florilegus is the most abun-B.florilegus is the most abun-B.florilegusdantbumblebee, with a contact zone with B. albocinctus of about 120 km on the small North Kuril Islands Shimushir, Ketoi and Rasshua (ITO & SAGAKAMI 1980; ITO & KURANASHI 2000; LELEJ & KUPIANSKAYA 2000).Compared to the wide distribution of most of the typical KUPIANSKAYA 2000).Compared to the wide distribution of most of the typical KUPIANSKAYA Bombus sensu stricto species, this restricted distribution is typical for island populations, as for instance, B. terrestris sassaricus from Sardinia or B. terrestris sassaricus from Sardinia or B. terrestris sassaricus B. terrestris canariensis from the Canary Islands compared to B. terrestris from the continent.Such isolated island populations are often genetically separated and may have their own diagnostic characters.Bombus florilegus has two unambiguous diagnostic Bombus florilegus has two unambiguous diagnostic Bombus florilegus characters that separate this taxon from B. albocinctus, which is comparable to the three unambiguous diagnostic characters separating B. magnus magnus of the British Isles from B. magnus magnus of the British Isles from B. magnus magnus B. magnus flavoscutellaris of the European continent (BERTSCH 2009) or island populations of B. ignitus from Japan compared to continental populations of B. ignitus from Korea ( B. ignitus from Korea (B.ignitus  TOKORO et al. 2010).Bombus florilegus is a taxon within the Bombus florilegus is a taxon within the Bombus florilegus cryptarum-complex, most probably with the status of a subspecies of B. albocinctus.Until further evidence is available it could be treated as B. albocinctus florilegus PANFILOV.separated only by the white colouration.But the difference in colouration is not always as clear as discussed in the literature; from the 15 queen specimens from Kamchatka (ZMAS, St.Petersburg) many were not clear white (as described by SMITH) but there was a graduation from clear white to light yellow and some were more or less citron yellow, not very different from typical B. lucorum.The same is true for 25 queen specimens (ZMAS, St. Petersburg) from Sakhalin, Magadan and Anadyr.Some of these specimens were clear white but some were citron yellow, at least in part, and all intermediate shades were available.But the Tamura-Nei genetic difference of the COI sequences of B. albocinctus and