Bombus handlirschianus Vogt

Williams, Paul H., Huang, Jiaxing, Rasmont, Pierre & An, Jiandong, 2016, Early-diverging bumblebees from across the roof of the world: the high-mountain subgenus Mendacibombus revised from species’ gene coalescents and morphology (Hymenoptera, Apidae), Zootaxa 4204 (1), pp. 1-72 : 52-55

11. Bombus handlirschianus Vogt

( Figs 34 , 50 , 66 )

[ Bombus mendax Gerstaecker ; Morawitz 1881:243, in part, misidentification.]

< Bombus mendax > var. Handlirschianus Vogt 1909:49 , type-locality citation ‘kaukasischen’. Lectotype queen by present designation RMNH examined, ‘ Mamisson Pass’ ( Caucasus , Georgia / Russian border). Note 1.

[ Bombus mendax subsp. caucasicus Skorikov 1910a:78 , 81, published without description. Note 2.]

< Bombus mendax > subsp. shaposhnikovi Skorikov 1910b:329 , type-locality citation (Cyrillic) ‘[Caucasus]’. Lectotype male by present designation ZISP examined, ( Cyrillic ) ‘[ Kubanskoy Oblast ]’ ( Caucasus , Russia). Note 3. Regarded as conspecific with Bombus handlirschianus Vogt by DeMeulemeester et al. (2010), supported here.

Mendacibombus handlirschianus (Vogt) ; Skorikov 1923:149.

Mendacibombus shaposhnikovi ( Skorikov); Skorikov 1923 :149.

Bombias (Mendacibombus) handlirschianus (Vogt) ; Rasmont 1983:10.

Bombias (Mendacibombus) shaposhnikovi (Skorikov) ; Rasmont 1983:10.

Bombus (Mendacibombus) handlirschianus Vogt ; P.H. Williams 1998:100.

Bombus (Mendacibombus) shaposhnikovi Skorikov ; P.H. Williams 1998:100.

Note 1 ( handlirschianus ). Vogt described his var. handlirschianus referring to a part of the population with both a particular area of distribution and a particular colour pattern, showing no evident intention of infrasubspecific status (e.g. without mentioning a higher rank subspecific taxon), so the taxon is deemed to be of subspecific rank ( ICZN 1999: Article 45.6.4).

Vogt’s original description of the taxon handlirschianus cites the type locality as ‘kaukasischen’. The RMNH collection studied by Vogt contains a queen (and other specimens) that agrees with the original description but also carries the labels: (1) white, handwritten ‘Mamisson / Pass ca 2750 / 21 VIII 03; (2) white, printed ‘Collectie / C. et O. Vogt / Acq. 1960’; (3) red, printed ‘ Bombus / mendax Handlirschianus / Vogt 1909 / ZMAN type HYME.0112.4’; (4) red, handwritten ‘ B. MENDAX var. HANDLIRSCHIANUS / VOGT, 1909:43 / LECTOTYPE [female] / P. RASMONT DES. 2008’ (P. Rasmont, unpublished); (5) green, printed ‘ Mendacibombus / MD# 1271 det. PHW’; (6) red, printed ‘LECTOTYPE [female] / Bombus mendax var. / handlirschianus / Vogt, 1909 / det. PH Williams 2012’; (7) white, printed ‘[female] Bombus / ( Mendacibombus ) / handlirschianus / Vogt / det. PH Williams 2012’. This queen, which lacks the right midleg, is regarded as a syntype and is designated here as the lectotype in order to reduce uncertainty in the identity and application of the name.

Note 2 (caucasicus). Skorikov’s attribution of the name caucasicus to a subspecies of B. mendax , supposedly described by Vogt from the Caucasus, may have arisen from confusion with the name B. lapidarius ‘Form (caucasicus RAD.)’ that also appears on Vogt’s (1909) page 49, in a paragraph above his description of B. mendax var. handlirschianus from the Caucasus.

Note 3 ( shaposhnikovi ). Skorikov’s original description of the taxon shaposhnikovi , which does not specify the sex of the specimens examined, cites the type locality as the Caucasus. The ZISP collection studied by Skorikov contains a male that agrees with the original description and carries the labels: (1) white, printed ( Cyrillic ) ‘[gora Loganaky / Kubanskoy obl. / Shaposhnik] VII08 ’; (2) white, handwritten in pencil ‘[illegible] / N 8’; (3) white, printed ‘ k. Skorikova’; (4) red, printed ‘ Lectotypus Bombus / mendax subsp. / shaposhnikovi Skor / design. Podbolotsk. ’ (M. Podbolotskaya, unpublished); (5) green, printed ‘ Mendacibombus / MD# 3522 det. PHW’; (6) red, printed ‘ LECTOTYPE [male] / Bombus mendax ssp. / shaposhnikovi / Skorikov, 1910 / det. PH Williams 2012’; (7) white, printed ‘[male] Bombus / ( Mendacibombus ) / handlirschianus / Vogt / det . PH Williams 2012’. This male, which is complete, is regarded as a syntype and is designated here as the lectotype in order to reduce uncertainty in the identity and application of the name.

A second male collected at Mt Loganaky by Shaposhnikov in 1908 (MD#835, NHM, sent by Skorikov as part of an exchange with the NHM in 1934), closely similar in morphology, is designated here as a paralectotype and is interpreted as conspecific.

Etymology. The species is named after A. Handlirsch, who in 1909 was an entomologist (and later Director) at the Natural History Museum of Vienna.

Taxonomy and variation. The interpretation of this species is based here on DNA and the form of the male genitalia. This disagrees with earlier concepts ( Skorikov, 1923, 1931), diagnosed originally in terms of the hair colour pattern ( Skorikov, 1910b), because the species appears to be much more variable in colour pattern than was originally understood.

All samples Pale hair hue Observed Expected

(n=634) males females females females

(haploid) (diploid) - no dominance - simple dominance

Number yellow 28 136 241 102 white 38 432 327 466

Frequency yellow 0.424 (recessive) Chi-square 1-sample test (df = 1) 79 13.8 Probability that deviation of observation from << 0.0001 <0.001 expectation is due to chance

NE Turkey Pale hair hue Observed Expected

(n= 485) males females females - no females - simple dominance

(haploid) (diploid) dominance

Number yellow 24 104 242.5 121 white 24 381 242.5 364 Frequency yellow 0.500 (recessive) Chi-square 1-sample test (df = 1) 158 3.18 Probability that deviation of observation from << 0.0001> 0.07 not significant expectation is due to chance

Vogt (1909) described females (e.g. MD#1271) of the taxon handlirschianus s. str. briefly as having the pale bands white and Skorikov (1910b) described in more detail females (e.g. MD#3522) of the taxon shaposhnikovi as having the pale bands yellow, the corbicula framed with rusty hairs, and T3 entirely black. Our COI tree shows that specimens of these taxa can be indistinguishable in the available sequences ( Fig. 13 : the white-banded taxon handlirschianus s. str. MD#427 and the yellow-banded taxon shaposhnikovi MD#3859) and are interpreted as conspecific, as parts of the species B. handlirschianus s. l.. Males with both colour patterns share the same diagnostic form of the male genitalia, supporting the white/yellow dimorphism in the pale bands of the females suggested for Turkish specimens from four genes by Cameron et al. (2007), and adding to the evidence of De Meulemeester et al. (2010). Both colour patterns of this species occur in Turkey, the Caucasus, and as far east as the Alborz mountains of Iran ( Pittioni 1937b). Throughout this region, other species also show a similar yellow/ white dimorphism ( Rasmont et al. 2005; P.H. Williams, Byvaltsev, et al. 2012). The white-banded female colour pattern (taxon handlirschianus s. str.) is a particularly characteristic and frequent convergent colour pattern in this region (P.H. Williams 2007).

Because a relatively large sample is available, we can compare the observed yellow/white dimorphism for this species with a simple genetic model in which we assume that the two pale hue states are each determined by one of two alleles at a single gene locus ( P.H. Williams 1991). We assume that there is a single homogeneous population with random interbreeding, unselective predation or other loss, and that this population has been sampled at random. Because males can be assumed usually to be haploid ( Plowright & Pallett 1979), we can estimate the frequency of each allele within the population from the frequency of each colour state among the males. If the observed frequency of one colour state were much higher among the haploid males than among the diploid females, then this colour state might be coded for by a recessive allele. We could then estimate the expected frequency of the colour states among the females from the observed frequency of the alleles among the males using the Hardy-Weinberg equilibrium model. When this expected frequency is tested within the entire sample in our database, the observed frequency of colour states among the females is confirmed to differ from the frequency among the males, although it also differs from the frequency predicted among the females by the model of simple dominance at one locus, so we must reject the model ( Table 8). However, we know that sampling intensity was not uniform across the entire range, and the frequency of alleles or the selectivity of sampling may also have varied geographically. Therefore we repeated the test, but restricting the sample used to the region of north-eastern Turkey around Erzurum, a small fraction of the total extent of occurrence, but reducing the sample size by only about one quarter (from 634 to 485 specimens). When the test was repeated for the geographically restricted subsample, the observed frequency of colour states among the females still differs from the frequency among the males, but the frequency predicted among the females no longer differs significantly from the model of simple dominance at one locus ( Table 9). Confirmation of this model would require breeding experiments to test the pattern of inheritance (cf. Owen & Plowright 1980). Nonetheless, this result is consistent with both the yellow-banded and the whitebanded bees being parts of a single species, B. handlirschianus s. l., supporting the synonymisation of the taxon shaposhnikovi as part of this species.

Diagnostic description. Wings nearly clear. Female hair colour pattern: generally black, but with pale hair (yellow or grey-white) in a small patch ventral to the base of the antenna, usually absent from the vertex of the head, in a transverse band anteriorly on the thoracic dorsum and extending laterally and ventrally to from half to two-thirds the way down the side of the thorax, in a transverse band posteriorly on the thoracic dorsum (scutellum; so that the thoracic dorsum between the wing bases has the hair entirely black), on T1‒2, although T2 has a few black hairs intermixed along the posterior margin, T3 with orange hair only as a very few hairs posteriorly, T4‒6 always orange except T6 medially with a few short unobtrusive black hairs. Hindleg tibia with the corbicular fringes with a mixture of orange and black hairs. Female morphology: labrum with the basal depression narrow, the transverse ridge broad and low, in the median third subsiding only slightly and uninterrupted, with only a few scattered punctures, lateral tubercles almost without punctures. Male morphology: genitalia ( Fig. 34 ) with the volsella distally rounded (finger-shaped) and curled back dorsally but not anteriorly; volsella at its broadest near the midpoint of its length, the dorsal surface just distal to this point without a raised curved ridge just inside the inner margin. Penis-valve inner shoulder located at Ĺ 0.5× the length of the penis valve from the distal end to the broadest point of the spatha; penis valve proximal to the outer shoulder <2× as broad as the penis-valve head; penis-valve breadth just proximal to the penis-valve head 0.06× the length of the penis valve distal to the broadest point of the spatha.

Material examined. 128 queens 454 workers 67 males (plus 14 with sex/caste undetermined), from Armenia, Azerbaijan, Georgia, Iran, Iraq, Russia, and Turkey ( Fig. 66 : AM, HH, HO, MS, NHM, OLL, PW, RMNH, UMONS, ZISP, ZSM; specimens in the UMONS database do not have individual MD# labels), with 5 specimens sequenced (interpretable sequences listed in Figs. 11–13 ).

Habitat and distribution. Flower-rich alpine and subalpine grassland, at elevations 472‒(2338)‒ 3982 m a.s.l.. A species of the Alborz, Caucasus, and Turkish mountains. There is no overlap in distribution with any other Mendacibombus species. Regional distribution maps are available for Turkey ( Rasmont & Flagothier 1996; Özbek 1997) and for western Asia ( Rasmont & Iserbyt 2012). Özbek (1997) reported that the species has been in decline in Turkey, although evidence for this is not apparent now (M. Aytekin, pers. com.).

Food plants. Özbek (1997).

Behaviour. De Meulemeester et al. (2010).

Figures

FIGURE 13. Assigning names to species from an estimate of phylogeny for 130 COI sequences of the subgenus Mendacibombus. Based on a BEAST analysis of the 30 unique haplotypes (Fig. 11), with the remaining sequences interpolated back into the tree with 0 branch lengths from their matching samples (where a sequence matches more than one longer sequence within a species in the Collapse results, it is added after the first matching sequence, with the addition of the two short-sequence yellow unbanded specimens from Pakistan, MD # 4058, 4059). The sequence labels follow the format used in Figs 11, 12. Lineages accepted as species from the GMYC models (Fig. 11) are shown as thick branches, the most recent common ancestors of each species are shown with a black spot, and branches within species are shown as thin branches. Asterisks mark samples used as informal proxies for the type specimens of each of the taxon names (Table 1). The proxy sample for the type specimen for the oldest available name (the valid name) for each species (from Fig. 11) is marked in bold. The interpretations of the species with their valid names are shown in grey to the right.

FIGURES 24 ‒ 35. Morphology of the male genitalia for species of the subgenus Mendacibombus from the dorsal aspect, anterior to the left of the image, posterior to the right: 24, B. superbus, China MD # 339; 25, B. waltoni, China MD # 216; 26, B. convexus, China MD # 215; 27, B. makarjini, Kazakhstan MD # 1240; 28, B. marussinus, Pakistan MD # 408; 29, B. avinoviellus, India MD # 404; 30, B. himalayanus, India MD # 426; 31, B. turkestanicus, China MD # 351; 32, B. defector, Kyrgyzstan MD # 367; 33, B. margreiteri, China MD # 294; 34, B. handlirschianus, Russia MD # 835; 35, B. mendax, Spain MD # 689.

FIGURES 36 ‒ 55. Morphology of parts of the male genitalia from the dorsal aspect of the left side, anterior to the left of the image, posterior to the right, white bars indicating morphological characters referred to in the key to males: 36, B. waltoni; 37, B. mendax; 38, B. superbus; 39, B. convexus; 40, B. waltoni; 41, B. convexus; 42, B. defector; 43, B. mendax; 44, B. mendax; 45, B. turkestanicus; 46, B. avinoviellus; 47, B. mendax; 48, B. mendax; 49, B. makarjini; 50, B. handlirschianus; 51, B. himalayanus; 52, B. turkestanicus; 53, B. himalayanus; 54, B. himalayanus; 55, B. marussinus.

FIGURES 56 ‒ 67. Maps of the global distributions of species of the subgenus Mendacibombus from the material examined, with grey spots for specimen records and black spots for the samples with COI sequences (each species is unknown from regions outside the map boundaries shown); grey crosses for records of all species of Mendacibombus combined. Relief map with hill shading, Cartesian orthonormal projection, grey lines show borders to national administration according to UN maps, north at the top of the map. Image created in ArcGIS using World _ Shaded _ Relief basemap © 2014 Esri.

FIGURES 11 ‒ 12. Recognising species from an estimate of phylogeny for 30 unique COI haplotypes of the subgenus Mendacibombus by BEAST, using GMYC models and the PTP procedure. From the GMYC models (11), the single threshold T is shown by the vertical grey bar so that the intersecting lineages are interpreted as indicating 12 separate species (black spots at coalescent nodes; outgroups not shown). Values below the nodes of the maximum clade-credibility tree are Bayesian posterior probabilities showing branch support (groups with values <0.8 are considered unreliable) from the BEAST analysis with two billion MCMC generations and a 1 % burn-in. The x - scale axis is calibrated in substitutions per nucleotide site. From the PTP procedure (12), the Bayesian PTP solution with highest support also indicates 12 species. Values above the nodes are PTP Bayesian support values that all daughter haplotypes are parts of a single species; values below the nodes are Bayesian posterior probabilities showing branch support. The scale bar is calibrated in substitutions per nucleotide site. Each unique haplotype is represented by one of the longest available sample sequences, labelled with: the sequence length in number of base pairs, a taxon name or name for an unpublished colour pattern (light / dark / yellow / white / pale), and then a code that consists of a specimen identifier (MD #) from the project database and (after the hyphen) a sample identifier from BOLD or from GenBank, followed with its geographic origin.