The study included 270 specimens (Suppl. material 1: Appendix 1) which were either newly collected or provided by the Natural History Museum of Vienna (NHM), the Upper Austrian State Museum and private collections. Pin-mounted specimens were obtained from collections. Fresh bees were collected from July to September 2017 in Vienna, Lower Austria (Retz and Ollersdorf), Upper Austria (Linz) and Burgenland around Lake Neusiedl, as well as in Poland (around the city of Sierakow). Collecting was carried out with an insect net and the bees were either euthanised in 96% ethanol to preserve DNA for DNA-barcoding or in a vial with ethyl acetate vapour.

To find additional, previously-unknown morphological characters for species distinction, both females and males were examined by light microscopy and compared between species. Since a description of the males of C. pannonicus has not yet been published, special attention was paid to this species and a detailed description is given in this work. Much attention was given to the proboscis of the females and to sterna 6–8 and genitalia of the males. Therefore, these body parts were manually dissected. Morphological descriptions were chiefly based on the terminology of Michener (2007) and Boudinot (2013). However, in order to get a better overview of the male’s structures of sterna 6–8 and genitalia, specific terms were additionally introduced.

For illustration of the species-specific differences, photography, as well as scanning electron microscopy, was used:

Stacked digital images of the different parts of sterna 6–8 and genitalia were acquired with a Leica DFC490 camera attached to a Leica Z16 APO zoom microscope, using Leica Application Suite 4.10.0 software. Afterwards, the digital images were stacked with ZerenaStacker 64-bit and processed with Adobe Photoshop 7.0. After illustration of the entire genital capsule, further dissection became necessary to see all important structures: a median cut between the two valvulae was performed, followed by removing the valvulae from the gonostyli. These parts were also illustrated by photography.

To illustrate different structures of proboscis and head of females, a scanning electron microscope (Philips XL 30 ESEM) was used. After dissection, the samples were washed for dehydration three times in 100% ethanol and three times in 100% acetone for 15 minutes each. For drying, the critical point dryer (LEICA EM MED020) was used for around 1.56 minutes with the settings: velocity – medium; delude time – 120 seconds; exchange steps – 5; cycles – 18; heating process – slow and speed – slow. Afterwards, the dried proboscides were glued to a copper foil with conductive silver and mounted on a carbon-taped stub. The dried heads were glued directly to the stub. For gold coating, a sputter coater (LEICA EM CPD300) was used for around 120 seconds. Images were taken with the scanning electron microscope and the programme Scandium 5.1 was used to add the scale bar.

In total, 103 females were used to obtain morphometric data for analysis (Suppl. material 2: Appendix 2): 10 specimens of C. halophilus, 11 specimens of C. pannonicus, 14 specimens of C. brevigena, 20 specimens of C. collaris, 22 specimens of C. hederae and 26 specimens of C. succinctus. The following distances, measured on the head and between the tegulae, were investigated (Figs 1, 2):

head length (HL) – maximum head length, measured in exact frontal view along the mid-line, from vertex to distal margin of clypeus;

head width (HW) – maximum head width, measured in exact frontal view from one outer edge of the compound eye to the other;

eye length (EL) – length of the eye, measured in latero-frontal view from the most dorsal point to the most ventral point of the left compound eye;

upper interocular distance (UID) – shortest distance between the dorsal margins of the compound eyes, measured in dorsal view;

lower interocular distance (LID) – shortest distance between the ventral margins of the compound eyes, measured in frontal view;

median interocular distance (MID) – longest distance between the inner margins of the compound eyes, measured in frontal view;

clypeus length (CL) – maximum clypeus length, measured in frontal view from the anterior to the posterior margin;

cheek length (CHL) – minimum length of the gena, measured in latero-frontal view from lower eye margin to mandible;

thorax width (TW) – maximum distance between the two mesal edges of the tegulae, measured in dorsal view.

Figure 1. 

Frontal Measuring distances for a female of the Colletes succinctus group. HL – head length, HW – head width, EL – eye length, UID – upper interocular distance, LID – lower interocular distance, MID – middle interocular distance, CL – clypeus length, CHL – cheek length.

Figure 2. 

Dorsal measuring distance for a female of the Colletes succinctus group. TW – thorax width.

Due to their very similar morphological characters C. brevigena and C. pannonicus were examined in more detail. As already done in the species description of Hölzler and Mazzucco (2011), the relationship of head width (HW) and thorax width (TW) for C. brevigena and C. pannonicus was calculated in form of an index:

Head-thorax index = TW / HW × 100

Measurements were conducted at different magnifications (24–76.8×) using a calibrated LEICA MZ6 binocular microscope with an ocular micrometre and later converted to millimetres. The statistics programme PAST3 (Hammer et al. 2001) was used to conduct principal component analysis (PCA) and discriminant analysis (LDA). The logarithmic morphometric values were used for all analyses. The PCA was based on a correlation matrix and 95% confidence intervals of variances were calculated using 1,000 Bootstrap re-samplings. For LDA, the specimens were assigned to six hypothetical groups, based on their morphological characters (C. succinctus, C. collaris, C. brevigena, C. halophilus, C. hederae and C. pannonicus) and the Jackknife method was used for re-sampling. Due to different morphological characters, the examined species C. brevigena was divided into two groups (Austrian and Mediterranean specimens) and analysed separately by LDA.

For pollen analyses, the pollen loads of 32 fresh and 41 dried female specimens (n = 73) were examined (Suppl. material 3: Appendix 3). The filling ratio of the pollen loads was determined between the grades one and five (see Müller and Kuhlmann 2008): “Five” defined fully-loaded hind legs and propodeum, so that no hair was visible and “one” meant that only one fifth of the collecting structures was loaded. Filling ratios in between (2–4) were estimated according to the researcher’s personal assessment. Furthermore, these filling ratios were used to calculate the correlation between the different filling loads and the amount of pollen types. Therefore, the Pearson correlation coefficient was calculated using Past3 (Hammer et al. 2001). The pollen was removed with a fine needle and placed on a glass object slide. After prevention of clotting accumulation by adding a drop of 85% ethanol, the pollen grains were fixed and dyed with a mixture of Kaisers Glycerine-Gelatine (ROTH) and alkaline Fuchsine. Afterwards, up to 300 pollen grains per slide were counted by using a NIKON ECLIPSE E800 light microscope.

For pollen determination, literature (Beug 2004, Hesse et al. 2009), as well as the databases paldat, ponet and pollen.tstebler, were used. Illustrations were made with the programme NIS-Elements D (4.51.01) and edited with Adobe Photoshop 7.0. Furthermore, the percentage for each pollen type was calculated and all types below 5% were identified as impurities.

The legs of 46 specimens were used for DNA barcoding (Suppl. material 4: Appendix 4). In preparation for DNA extraction, the ethanol, in which some of the animals were stored, had to be washed out with PBS (phosphate buffered saline). Afterwards, the samples were shock-frozen in a mixture of dry ice and ethanol and crushed. The following DNA extraction was performed using Qiagen’s “DNeasy Blood and Tissue Kit” and the protocol “Purification of Total DNA from Animal Tissues (Spin-Column Protocol)”.

After extraction, the samples were analysed with a nanodrop (Nanodrop 200/2000c Spectrophotometer) for DNA quantification and only samples containing sufficient DNA were used for further analysis. For Polymerase Chain Reaction (PCR) 25 μl MM Biozym Red HS Taq Master Mix, 21 μl molecular grade water (Sigma Aldrich) as well as 1.5 μl of each primer and the respective DNA sample (1–2 μl) were mixed. The PCR machine “TProfessional Thermocycler” (Biometra) was used to amplify the gene using the primer-pair LCO1490 and HCO2198 (Folmer et al. 1994) in the following settings: the starting denaturation cycle at 95 °C for 3 minutes was followed by 33 cycles for 30 seconds each, 33 alignment cycles at 47 °C for 30 seconds each, 33 elongation cycles at 72 °C for 45 seconds each and one final cycle at 72 °C for 10 minutes.

For the subsequent gel electrophoresis, a mixture of 1× TBE buffer and agarose gel (1% w/v) was used. In total, 3 μl of the samples, together with 1.5 μl Orange DNA Loading Dye (Thermo Science) and the Sizemarker GeneRuler 1kb – DNA Ladder (Thermo Science), were loaded on to the gel.

The purification was carried out with Qiagen’s “QIAquick PCR Purification Kit” and the DNA sequencing was executed by Microsynth AG.

Unfortunately, only 21 samples were sequenced successfully: 20 specimens of the European C. succinctus group and one specimen of C. creticus, which was used as an outgroup (Suppl. material 4: Appendix 4). The remaining samples could not be used for further analyses due to either contamination or unsuccessful DNA extraction. Furthermore, some sequences could not be used because of infestation with Wolbachia (a genus of gram-negative bacteria common in sexual organs of arthropods) and very poor electropherograms which could not be evaluated. In general, electropherograms were not ideal, which is probably due to an artefact band (200 bp). Therefore, base-calling in all electropherograms had to be checked carefully.

The obtained electropherograms were proof-read, aligned and cut to the same length by removing the primer sequences with Bioedit 7.2.6. and FinchTV 1.4.0 (Geospiza, Inc). POPArt (Bandelt et al. 1999) was used to create a median-joining network from the sequences obtained, as well as from reference data (47 sequences) from the Barcode of Life Database (BOLD) (Ratnasingham and Hebert 2007) (Suppl. material 5: Appendix 5). In addition, genetic distances were calculated and a neighbour-joining tree was generated with MEGA 7.0.26 (Kumar et al. 2016) before it was illustrated with FigTree 1.4.3.

In addition to already-known morphological features, new characters were discovered to better distinguish between the Austrian species of the Colletes succinctus group (Table 2, Fig. 3).

Figure 3. 

Important morphological characters for species differentiation of the Austrian species of the Colletes succinctus group: (A–C) Female C. succinctus: A. Fovea facialis; B. Terga 1 and 2; C. Proboscis; (D–F) Female: C. collaris: D. Fovea facialis; E. Terga 1 and 2; F. Cuticle of frons; (G–I) Female C. brevigena: G. Fovea facialis; H. Terga 1 and 2; I. Cuticle of mesopleura; (J–L) Female C. hederae: J. Fovea facialis; K. Terga 1 and 2; L. Proboscis; (M–O) Female C. pannonicus: M. Fovea facialis; N. Terga 1 and 2; O. Clypeus. ant – antenna, ce – complex eye, fov – fovea facialis, oc – lateral ocellus, mxp – maxillary palpus, T1 – tergum 1, T2 – tergum 2.

In addition to external morphological characters, the male specimens can also be distinguished by shape and pubescence of sterna 6–8 and the morphology of the genitalia. It was possible to find differences in their shape and pubescence.

General description, Colletes succinctus group:

Sternum 6: large, lacking lateral tubercles; the convex hind margin medially protruded and with variably-developed blunt corners at each side. Before hind margin with two deep lateral grooves (lgr). An oval-shaped translucent area (ota) of variable size in the middle.

Sternum 7: base (bs) curved, narrow, connected with distal structures via a bridge (br) with a narrow sclerotised medial stalk. Distal part strongly modified: a central, diamond-shaped, distally bifid medial elevation (mel) leads to strongly sclerotised basal shoulders (sh), which bear the paired wings (wg). Except for C. collaris (see description of C. collaris), each wing consisting of a sclerotised, densely pilose medial processes (mp) and a weakly sclerotised, flexible lateral part with a hairy basal arm (ba) and a distal, almost bold membrane (mbr).

Sternum 8: sub-rhomboidal. Anterior spiculum (spi) strongly elongated. Lateral processes (lpr) bifid, bearing muscle attachments. Distal process (dpr) curved ventrally, with dense tuft of long setae on dorsal margin.

Genital capsule: stout, most parts, including gonobase, heavily sclerotised. Gonopod (gpo) smooth and hairless, dorsolaterally with oblique depression; ventrally fused with gonostylus, dorsally separated from it by a deep fissure. Gonostylus (gst) curved ventrally; mesoventrally with a ridge bearing a row of setae (rs), distally with a hairy, medially curved gonostylus membrane (gme) of approximately triangular shape. Volsella (vol) strongly developed; basivolsella (bvo) short; digitus (dig) and cuspis (cus) both plate-shaped, their opposing surfaces with numerous stout, short teeth. Penisvalva (val) with slender base; distal part curved ventrally, with several modifications: a heavily-sclerotised mesodorsal ridge (mdr), a laterodorsal membrane (ldm), a basolateral groove (blg) often surrounded by stout spines, and a lateral area (lar) bearing numerous, often spine-like setae.

Compared to other species groups, the distal part of sternum 7 is smaller and distinctly shorter in the C. succinctus group (Noskiewicz 1936). The specific structures, described above, cannot be clearly homologised with the structures of other species groups.

Specific characters, C. collaris :

Sternum 6: Stout, lateral edges extended to posterior, appears long and broad. Little hair on the disc, blunt corners weakly developed. Lateral grooves (lgr) small and spherical (Fig. 4A). – Sternum 7: Short, transversely oval paired wings (wg) with densely hairy basal arm (bs), without membrane and medial process. Strongly developed shoulders (sh) (Fig. 4B). – Sternum 8: Strongly sclerotised median process of the anterior spiculum (spi) reaching the posterior apex, with short bifurcation in its centre. Distal process (dpr) sitting on small outwardly curved shoulders (Fig. 4C, D). – Genital capsule: Prominent volsella (vol) with slender digitus (Fig. 5B, D).

Figure 4. 

Male terminal structures of Colletes collaris, specimen no. 16 (LM): A. Ventral view of sternum 6; B. Ventral view of sternum 7; C. Ventral view of sternum 8; D. Lateral view of sternum 8. ba – basal arm, br – bridge, bs – base, dpr – distal process, lgr – lateral grooves, lpr – lateral process, mbr – membrane, mel – median elevation, mp – median process, ota – oval-shaped translucent area, sh – shoulder, spi – spiculum, wg – wing.

Figure 5. 

Male genitalia of Colletes collaris, specimen no. 16 (LM): A. Dorsal view of genital capsule; B. Ventral view of genital capsule; C. Lateral view of genital capsule; D. Mediolateral view of gonopod and gonostylus with volsella; E. Lateral view of penis valva. blg – basolateral groove, bvo – basivolsella, us – cuspis, dig – digitus, gme – gonostylus membrane, gpo – gonopod, gst – gonostylus, lar – lateral area, ldm – laterodorsal membrane, mdr – mesodorsal ridge, rs – row of setae, val – penis valva, vol – volsella.

Specific characters, C. succinctus :

Sternum 6: Prominent inwardly inclined lateral grooves (lgr), long and oval shaped. Weakly pronounced blunt corners on convex hind margin and oval-shaped translucent area (ota), 1.5–2 times larger than lateral grooves (lgr) (Fig. 6A). – Sternum 7: Wide in appearance, with strong shoulders (sh). Basal arm (bs) of the wing (wg) with setae, especially in its proximal half. Horizontally directed distal margin of membrane (mbr) with long setae (Fig. 6B). – Sternum 8: Distal process (dpr) basal with wide outwardly curved shoulders. Sclerotised anterior spiculum (spi) with strong central bifurcation (Fig. 6C, D). – Genital capsule: No species-specific characters were detected (Fig. 7).

Figure 6. 

Male terminal structures of Colletes succinctus, specimen no. 243 (LM): A. Ventral view of sternum 6; B. Ventral view of sternum 7; C. Ventral view of sternum 8; D. Lateral view of sternum 8. ba – basal arm, br – bridge, bs – base, dpr – distal process, lgr – lateral grooves, lpr – lateral process, mbr – membrane, mel – median elevation, mp – median process, ota – oval-shaped translucent area, sh – shoulder, spi – spiculum, wg – wing.

Figure 7. 

Male genitalia of Colletes succinctus, specimen no. 243 (LM): A. Dorsal view of genital capsule; B. Ventral view of genital capsule; C. Lateral view of genital capsule; D. Mediolateral view of gonopod and gonostylus with volsella; E. Lateral view of penis valva. blg – basolateral groove, bvo – basivolsella, cus – cuspis, dig – digitus, gme – gonostylus membrane, gpo – gonopod, gst – gonostylus, lar – lateral area, ldm – laterodorsal membrane, mdr – mesodorsal ridge, rs – row of setae, val – penis valva, vol – volsella.

Specific characters, C. halophilus :

Sternum 6: Prominent oval-shaped translucent area with hairless centre (ota) (Fig. 8A). – Sternum 7: Hind margins of membrane (mbr) of distal processes of sternum 7 (wg) almost straight (Fig. 8B). – Sternum 8: Distal process (dpr) basal with outwardly curved shoulders. Weakly pronounced bifurcation of median spiculum (spi) (Fig. 8C, D). – Genital capsule: In lateral view, membrane of gonostylus (gme) of genital arched dorsally, with dorsobasal knob and densely hairy (Fig. 9D).

Figure 8. 

Male terminal structures of Colletes halophilus, specimen no. 180 (LM): A. Ventral view of sternum 6; B. Ventral view of sternum 7; C. Ventral view of sternum 8; D. Lateral view of sternum 8. ba – basal arm, br – bridge, bs – base, dpr – distal process, lgr – lateral grooves, lpr – lateral process, mbr – membrane, mel – median elevation, mp – median process, ota – oval-shaped translucent area, sh – shoulder, spi – spiculum, wg – wing.

Figure 9. 

Male genitalia of Colletes halophilus, specimen no. 180 (LM): A. Dorsal view of genital capsule; B. Ventral view of genital capsule; C. Lateral view of genital capsule; D. Mediolateral view of gonopod and gonostylus with volsella; E. Lateral view of penis valva. blg – basolateral groove, bvo – basivolsella, cus – cuspis, dig – digitus, gme – gonostylus membrane, gpo – gonopod, gst – gonostylus, lar – lateral area, ldm – laterodorsal membrane, mdr – mesodorsal ridge, rs – row of setae, val – penis valva, vol – volsella.

Specific characters, C. hederae :

Sternum 7: Basal arm (bs) narrow at base and widening distally, densely hairy. Membrane (mbr) convexly curved, proximal part with dense setae, short setae at distal edge. Medial bifid elevation (mel) strongly pronounced (Fig. 10B). – Sterna 6, 8 and genital capsule: no species-specific characters were detected (Figs 10A, C, D, 11).

For species-specific characters of the male terminal structures and genitalia of C. brevigena, see the next chapter.

Figure 10. 

Male terminal structures of Colletes hederae, specimen no. 197 (LM): A. Ventral view of sternum 6; B. Ventral view of sternum 7; C. Ventral view of sternum 8; D. Lateral view of sternum 8. ba – basal arm, br – bridge, bs – base, dpr – distal process, lgr – lateral grooves, lpr – lateral process, mbr – membrane, mel – median elevation, mp – median process, ota – oval-shaped translucent area, sh – shoulder, spi – spiculum, wg – wing.

Noskiewicz (1936) described C. succinctus ssp. brevigena, based on specimens from a large distribution area: Macedonia, Dalmatia, Cyprus, Crete, North Persia and the Caucasus; however, he questioned the synsubspecificity of a male from Austria near Neusiedlersee (note that this is the type area of C. pannonicus). Noskiewicz (1936) did not designate a holotype; a lectotype was not selected by subsequent authors. The depositories of the syntype series were not published by Noskiewicz (1936), but later it was reported that no types of C. brevigena are represented in the Noskiewicz collection in the Museum of Natural History, University of Wroclaw, Poland (Wanat et al. 2014; and Marek Wanat, pers. comm. to Herbert Zettel).

The Natural History Museum Vienna keeps eleven male specimens, all identically labelled “Colletes ♂ succinctus L. ssp. brevigena Nosk. det. Noskiewicz.”, which were studied by Noskiewicz in the course of preparing his monograph and, therefore, are putative syntypes. For the reason of taxonomic stability, we select a male (no. 270) from Crete (Greece) as the lectotype (see Figs 12–14). Arguments for selecting this specimen were (1) its almost perfect condition and (2) avoidance of a type locality in the southern Caucasus which would make future molecular research on this taxon more difficult.

Figure 11. 

Male genitalia of Colletes hederae, specimen no. 197 (LM): A. Dorsal view of genital capsule; B. Ventral view of genital capsule; C. Lateral view of genital capsule; D. Mediolateral view of gonopod and gonostylus with volsella; E. Lateral view of penis valva. blg – basolateral groove, bvo – basivolsella, cus – cuspis, dig – digitus, gme – gonostylus membrane, gpo – gonopod, gst – gonostylus, lar – lateral area, ldm – laterodorsal membrane, mdr – mesodorsal ridge, rs – row of setae, val – penis valva, vol – volsella.

Figure 12. 

Male of Colletes brevigena, lectotype: A. Lateral view; B. Dorsal view; C. Frontal view; D. Terga 1 and 2. T1 – tergum 1, T2 – tergum 2.

Lectotype (♂, present designation, Natural History Museum Vienna): “Oertzen Creta 1884.” [= Crete Island, Greece; printed], “Colletes ♂ succinctus L. ssp. brevigena Nosk. det. Noskiewicz.” [mostly handwritten], “270” [handwritten], “Colletes brevigena Noskiewicz, 1936 det. K. Zenz 2018“ [printed], “Lectotype Colletes succinctus brevigena Noskiewicz, 1936 des. Katharina Zenz et al. 2020” [printed on red paper].

Paralectotypes deposited in the Natural History Museum Vienna: 1 ♂ (legs partly broken) labelled as the lectotype; 1 ♂ (legs and antennae partly broken) labelled “Pola Schlett.” [= Pula, today in Croatia, leg. Schletterer; printed]; 8 ♂♂ (in various conditions) labelled “Transkauk. Helenendorf 1886.” [= Goygol in Azerbaijan; printed “6” on some labels handwritten]; all specimens with Noskiewicz’s and Zenz’s identification label as the lectotype and a type label “Paralectotype Colletes succinctus brevigena Noskiewicz, 1936 labelled by K. Zenz 2020” [printed on red paper].

Specific characters of terminal structures and genitalia of the male C. brevigena :

Sternum 7: Very pronounced shoulders (sh) with broad basal arms (ba), densely hairy. Concavely curved distal margin of membrane (mbr) bearing few hairs (Fig. 13B). – Sternum 8: Spiculum (sp) with a broad, sclerotised bifurcation (Fig. 13C, D). – Sternum 6 and genital capsule: no species-specific characters were detected (Figs 13A, 14).

For species-specific characters of the male terminal structures and genitalia of C. pannonicus, see the next chapter.

Figure 13. 

Male terminal structures of Colletes brevigena, specimen no. 270 (LM): A. Ventral view of sternum 6; B. Ventral view of sternum 7; C. Ventral view of sternum 8; D. Lateral view of sternum 8. ba – basal arm, br – bridge, bs – base, dpr – distal process, lgr – lateral grooves, lpr – lateral process, mbr – membrane, mel – median elevation, mp – median process, ota – oval-shaped translucent area, sh – shoulder, spi – spiculum, wg – wing.

Figure 14. 

Male genitalia of Colletes brevigena, specimen no. 270 (LM): A. Dorsal view of genital capsule; B. Ventral view of genital capsule; C. Lateral view of genital capsule; D. Mediolateral view of gonopod and gonostylus with volsella; E. Lateral view of penis valva. lg – basolateral groove, bvo – basivolsella, cus – cuspis, dig – digitus, gme – gonostylus membrane, gpo – gonopod, gst – gonostylus, lar – lateral area, ldm – laterodorsal membrane, mdr – mesodorsal ridge, rs – row of setae, val – penis valva, vol – volsella.

Colletes pannonicus is a recently described species that has been solely found in the Seewinkel near Lake Neusiedl (Hölzler and Mazzucco 2011). The “preliminary description” was based on a female specimen (holotype) and since the males of C. pannonicus have not yet been described, this is done on the following, based on dry specimens collected sympatrically with C. pannonicus females in Seewinkel on Tripolium pannonicum (sea aster).

Examined material. Austria, Burgenland: 1 ♂, Podersdorf, 4.9.1991, leg. M. Madl, coll. H. Zettel (spec.no. 174); 1 ♂, Illmitz, Hölle; 4.9.2006, leg. & coll. H. Wiesbauer (spec.no. 166), 1 ♂, Podersdorf, 9.9.2012, leg. & coll. H. Wiesbauer (spec.no. 175); 1 ♂, Podersdorf; 29. 8.2014, leg. & coll. H. Wiesbauer (spec.no. 176); 2 ♂♂, Illmitz, Hölle, 29–30.8.2014, leg. & coll. H. Wiesbauer (spec.no. 167–168); 1 ♂, Neusiedl am See, Kalvarienberg, 47°56'33.69"N, 16°51'39.23"E, 8. 9.2016, leg. & coll. L.W. Gunczy (spec.no. 171).

Description. Face with long yellow-whitish hair. Galea reticulated between sensilla, segments of maxillary palpi long and narrow. Mesonotum coarsely and densely punctured, in some specimens with a shiny centre of varying size, with long orange-brownish hair. Mesopleura densely punctured, (sporadically) punctures merge and form wrinkles, with yellow-whitish hair. Hair on propodeum yellowish-white to yellow-orange coloured. Stripes of setae on terga yellow-whitish. Punctures on tergum 1 large, but smaller than on mesonotum. Tergum 1 deeply and coarsely punctured, on disc dense, to the sides with interspaces of the size of 0.5–1 puncture diameter; cuticle shiny; basal declivity with long yellow hair. Tergum 2 densely punctured, sporadically with intervals of the size of one puncture diameter; cuticle shiny (Fig. 15D).

Figure 15. 

Male of Colletes pannonicus: A. Lateral view; B. Dorsal view; C. Frontal view; D. Terga 1 and 2. T1 – tergum 1, T2 – tergum 2.

Specific characters of terminal structures and genitalia of the male C. pannonicus :

Sternum 6: Lateral grooves (lgr) oval and large (Fig. 16A). – Sternum 7: Slender wings (wg), with setae on their proximal half. Hairless membrane (mbr), sigmoid-curved hind margin bearing short hairs (Fig. 16B). – Sternum 8: Distal process (dpr) basal with outwardly curved shoulders, with sclerotised, short and discontinuous median bifurcation (Fig. 16C). – Genital capsule: no species-specific characters were detected (Fig. 17).

Figure 16. 

Male terminal structures of Colletes pannonicus, specimen no. 171 (LM): A. Ventral view of sternum 6; B. Ventral view of sternum 7; C. Ventral view of sternum 8; D. Lateral view of sternum 8. ba – basal arm, br – bridge, bs – base, dpr – distal process, lgr – lateral grooves, lpr – lateral process, mbr – membrane, mel – median elevation, mp – median process, ota – oval-shaped translucent area, sh – shoulder, spi – spiculum, wg – wing.

Figure 17. 

Male genitalia of Colletes pannonicus, specimen no. 171 (LM): A. Dorsal view of genital capsule; B. Ventral view of genital capsule; C. Lateral view of genital capsule; D. Mediolateral view of gonopod and gonostylus with volsella; E. Lateral view of penis valva. blg – basolateral groove, bvo – basivolsella, cus – cuspis, dig – digitus, gme – gonostylus membrane, gpo – gonopod, gst – gonostylus, lar – lateral area, ldm – laterodorsal membrane, mdr – mesodorsal ridge, rs – row of setae, val – penis valva, vol – volsella.

Using a range of external morphological characters (Table 1), as well as the identified differences in male genitalia, it was possible to establish identification keys for both females and males. All investigated specimens were subjected to a detailed examination and determined according to the following identification keys for females and males:

All studied specimens were subjected to a control of their species affiliation. The specimens were re-identified with the help of the new key presented herein. A total of 21 (19 females and two males) of the 270 specimens were assigned to another species of the C. succinctus group than previously. Three males were excluded from the Colletes succinctus group due to the absence of lateral pits on sternum 6 (Suppl. material 1: Appendix 1). In an additional 29 specimens, it was not possible to assign them to a species with certainty, due to their character combination (Suppl. material 6: Appendix 6). These aberrant specimens showed either a mix of species-specific characters of C. succinctus and C. brevigena or of C. succinctus and C. hederae. One specimen even showed characters of all three species. One specific male specimen from Velden in Carinthia (no. 124) could be assigned to the C. succinctus group due to its deep lateral pits on sternum 6. However, it differs from all other Austrian species by peculiar characters: the mesopleura are so densely punctured that the punctures merge into each other (as is only known from C. halophilus) and tergum 1 shows a very sparse puncturation with distances on disc about twice as long as a diameter of puncture. Tergum 2 is only superficially punctured.

Variation within species

Identified females of all species showed a pronounced intraspecific variation (Table 3). Especially, the head width (HW) seems to be very variable in all species. The intraspecific variation is most pronounced in C. succinctus. Here, the females differ very strongly in all measuring distances with the exception of the cheek length (CHL). Colletes collaris and C. halophilus, on the other hand, show a pronounced intraspecific variability in their head length (HL) and width (HW) and C. brevigena varies both in head width (HW) and thorax width (TW). Colletes hederae differ mostly in head width (HW) and upper interocular distance (UID). Colletes pannonicus, on the other hand, exhibits the lowest interspecific polymorphism, although the specimens differed to a large extent in their head width (HW).

On average, the females of C. hederae are the largest specimens (Table 3). They have by far the largest head, the longest clypeus, the longest lower interocular distance and the longest cheeks. In addition, they have the longest eyes, closely followed by C. brevigena. In turn, the largest average upper interocular distance is shown in females of C. halophilus, as well as the longest middle interocular distance, followed by C. hederae. The specimens of C. brevigena show the widest thorax on average, closely followed by C. halophilus and C. hederae.

Overall, the females of the species C. succinctus have the smallest mean head size (Table 2). They show the lowest values for head width, clypeus length, upper ocular distance, lower ocular distance and middle ocular distance. In terms of head length, they present the smallest mean value together with C. pannonicus.

Principal components analysis of all species (PCA)

All measurements were analysed using a principal component analysis (PCA). Based on 1,000 bootstrap re-samplings, the first principal component (PC1) explains just under 71% of the total sample variance, the second explains over 11% and the third explains about 7% of the variance.

The loadings of the nine measuring distances show that PC1 (principal component 1) correlates with all variables except for variable CHL (cheek length) (Table 4). Therefore, it is primarily a measure for body size. Principal component 2 dependents strongly on CHL. For example, when plotting PC1 and PC2, specimens of C. hederae are defined by high loadings on both axes, which means the majority have large bodies and long cheeks (Fig. 23). PC3 is strongly positively influenced by TW and negatively influenced by HL (Table 4). Thus, specimens with a wider thorax tend to have a shorter head.

Fig. 23 shows principal components 1 and 2, based on the measured values of the investigated species. Overall, the morphometric data of all species overlap with each other to some extent. By comparing only two species, it is possible to distinguish between some of them. The specimens of C. hederae and C. succinctus overlap only minimally. The same can be seen when comparing the data of C. pannonicus with C. hederae or C. collaris. However, measurements of all other species overlap greatly.

Figure 23. 

Scatter plot of PC 1 and PC 2, based on nine morphometric values of six European members of the Colletes succinctus group.

Linear discriminant analysis of all species (LDA)

Based on their morphological characters, the species could not be separated efficiently. Even with a Jackknife re-sampling, only 54.13% of all measured specimens were classified as their previously-assigned species (hypothetical group). Thus, around half of all specimens were assigned to different species by the LDA (Suppl. material 7: Appendix 7). A detailed listing of assignments for each measured specimen, by the author as well as by the LDA, with or without Jackknife re-sampling, is given in Suppl. material 8: Appendix 8.

Linear discriminant analysis of C. brevigena

For a more detailed analysis of the species C. brevigena, all specimens (n = 18) were divided into two hypothetical groups: thirteen specimens from the Mediterranean region were grouped as “C. brevigena MED” and the remaining five females from Austria were grouped as “C. brevigena A”. Based on the morphometric data, the discriminant analysis calculated that the specimens belonged to their previously-assigned hypothetical species in 100% of cases. Only after a Jackknife re-sampling, the affiliation of five Mediterranean C. brevigena and two Austrian C. brevigena was reversed, resulting in different assignments in about 40% (Table 5) of the specimens.

The taxonomically-challenging species C. brevigena and C. pannonicus

Head-thorax index: By comparing the head-thorax index of C. brevigena and C. pannonicus, the species cannot be differentiated. In relation to the thorax width, the examined females of C. brevigena show broader heads than the specimens of C. pannonicus; however, the two species overlap in their minimum to maximum head-thorax index range to a great extent (C. brevigena 71.7 to 87.5 vs. C. pannonicus 70.7 to 82.2).

Principal component analyses (PCA): Principal component 1 explains around 70% of the total sample variance and is defined by high positive loadings of all variables, except for variable CHL (cheek length) and TW (thorax width). Principal component 2, however, explains only 13% of the variance and shows high positive loadings of CHL and high negative loadings of TW. Therefore, PC1 is interpreted as a measure for body size, whereas PC2 is mainly a measure for sizes of CHL and TW (Table 6).

The examined specimens of C. pannonicus and C. brevigena overlap in their measurements to a great extent. On average, the specimens of C. brevigena are larger, but like in the PCA of all investigated species, the intraspecific variation is larger than the interspecific differences (Fig. 24).

Figure 24. 

Scatter plot of PC 1 and PC 2, based on nine morphometric values of two European members of the Colletes succinctus group: (pink) C. pannonicus, (dark green) C. brevigena.

Pollen determination

Based on their morphological characters, the pollen grains found on the studied specimens were assigned to Asteraceae (liguliflorae and tubuliflorae), Araliaceae, Ericaceae, Resedaceae and Rutaceae (Fig. 25). In addition, tricolporate/reticulate pollen grains were also found in the pollen load of one specimen, but it was not possible to determine them more precisely. In some cases, it was possible to determine the pollen grains to genus level (Citrus sp., Reseda sp.) or even species level (Tripolium pannonicum, Calluna vulgaris and Hedera helix). The pollen of the family Asteraceae has a spiny (echinate) surface structure. They are either entirely spiked (tubuliflorae) or have spineless windows on their surface (liguliflorae). Ivy (Hedera helix) belongs to the Araliaceae family. Its pollen grains show a reticulate surface and are tricolporate: it has three colpi as well as three pores from which the pollen tubes emerge. Heather (Calluna vulgaris) belongs to the family Ericaceae and is arranged as tetradae: four spheroidal-shaped pollen grains are associated with each other. They possess pores and a scabrate to verrucate structure. The pollen of Citrus sp. (Rutaceae) is tetracolporate and with a reticulate surface.

Figure 25. 

Selection of pollen grains collected by the females of the Austrian species of the Colletes succinctus group, scale bar 10 µm: A. Asteraceae tubuliflorae; B. Asteraceae tubuliflorae (Tripolium pannonicum); C. Asteraceae liguliflorae; D. tricolporate/reticulate; E. Resedaceae (Reseda sp.); F. Ericaceae (Calluna vulgaris); G. Araliaceae (Hedera helix); H. Rutaceae (Citrus sp.).

Relationship between filling ratio and number of pollen types

Females with smaller pollen packages and therefore a lower filling ratio, collected fewer different pollen types than females with larger pollen loads and a high filling ratio (r = 0.343, p = 0.003).

Pollen preferences of the Austrian species of the C. succinctus group

Colletes succinctus: Pollen analysis indicated that C. succinctus is polylectic. More than half (57%) of their average pollen load consisted of Reseda sp., closely followed by 40% Ericaceae pollen, which could be assigned to Calluna vulgaris. Only 1% of Asteraceae liguliflorae and 2% of Asteraceae tubuliflorae could be identified in the package and the rest was interpreted as contamination (Asteraceae, Ericaceae and undetermined). Fourteen of the 21 investigated C. succinctus females were collected in Retz, nine of them on Calluna vulgaris (Ericaceae) and five on Reseda sp. (Resedaceae): only seven of the nine specimens collected on C. vulgaris showed pure pollen packages consisting of that specific pollen. One female (no. 88) also preferred C. vulgaris (86.3%), but additionally collected a small proportion of Asteraceae liguliflorae (13.7%). Only in one specimen (no. 87), which was caught on C. vulgaris, no pollen of the same plant could be found. Its pollen package consisted of 100% Reseda sp. The five captured specimens on Reseda sp. possessed over 90% pollen from this plant in their collecting devices. One specimen (no. 92) additionally collected C. vulgaris in smaller quantities (8.3%) (Suppl. material 3: Appendix 3). Two C. succinctus females captured in Ollersdorf possessed pollen packages with over 99% Reseda sp. All specimens from Oberweiden (no. 160) and Bisamberg (n = 4) collected preferably on Reseda sp. Of those, only one female (no. 98) additionally collected 28.7% of its pollen package on Asteraceae tubuliflorae (Suppl. material 3: Appendix 3).

Colletes collaris: In summary, the 13 investigated females of C. collaris show a polylectic pollen-collecting behaviour with a strong preference for Asteraceae (66%). This is closely followed by Reseda sp. (31%). Only 2% are due to pollen of the type tricolporate/reticulate and 1% remained indeterminable. Flower consistency of specimens was rarely observed. A single female (no. 1), captured at Bisamberg, had a 100% pure pollen package of Reseda sp. and a further specimen (no. 3) also from Bisamberg collected 100% of Asteraceae of the type tubuliflorae. Resedaceae and Asteraceae tubuliflorae are also very popular with the remaining females from Bisamberg (nos. 4–7). One of them (no. 4) collected approximately equal parts of Reseda sp. (36.7%), Asteraceae tubuliflorae (33.3%) and tricolporate/reticulate pollen (30%). Another one (no. 6) preferred Reseda sp. (46%), closely followed by Asteraceae tubuliflorae (32%) and Asteraceae liguliflorae (16%) and a further 6% of the pollen load is attributed to contamination (Suppl. material 3: Appendix 3). A female (no. 9) from Langenlois preferred Asteraceae tubuliflorae (78.3%) to Reseda sp. (21.7%) as did the specimen from Engabrunner Haide (no. 10). The latter collected 76.7% of Asteraceae tubuliflorae and 23.3% of Reseda sp. (Suppl. material 3: Appendix 3). A female from Ollersdorf (no. 11) again preferred Asteraceae liguliflorae (56.7%) to Asteraceae tubuliflorae (43.3%). Another female from Ollersdorf (no. 195), caught on Reseda sp., collected mainly on this plant (77.3%), followed by Asteraceae tubuliflorae (21.3%). It shows a small amount of 1.3% that is presumably contamination-related (Suppl. material 3: Appendix 3). The three studied females from Lake Neusiedl (nos. 191–193) preferred Asteraceae tubuliflorae, which can be assigned to Tripolium pannonicum. Only no. 192 additionally collected 18.3% of its pollen load on Asteraceae liguliflorae (Suppl. material 3: Appendix 3).

Colletes hederae: Colletes hederae is the most represented species in this pollen analysis (n = 27). It shows a polylectic pollen-collecting behaviour with a strong preference for Hedera helix of the Araliaceae family (79.5%). In addition, 20.3% of the collected pollen comes from Citrus sp. of the family Rutaceae. There is only one specimen that shows contamination (0.2%) by Citrus sp. All females from Stammersdorf (no. 198), Hainburg (nos. 29–30), Linz (no. 221) and different parts of Vienna (nos. 221, 225 and 228–230) collected pure pollen packages of H. helix flowers. The specimens captured at Donaupark on H. helix (nos. 211–212 and 219) in turn possess not only pollen of H. helix, but also pollen of Citrus sp., some in smaller and some in larger quantities. One female (no. 217) collected 50% H. helix and 50% Citrus sp. (Suppl. material 3: Appendix 3).

Colletes brevigena: The only specimen of C. brevigena (no. 190) represented in this pollen analysis was caught at the same time as a specimen of C. collaris (no. 195) in Ollersdorf on the flowers of Reseda sp. This specimen collected Reseda pollen in large quantities (~ 98%) and only a small proportion of the pollen load is due to contamination (2.3%).

Colletes pannonicus: All specimens of C. pannonicus (n = 5) were captured near Lake Neusiedl and show an oligolectic behaviour, collecting pollen on Asteraceae tubuliflorae (99.6%), whereas only 0.4% of the load is due to contamination. Their individual pollen packages contain 98–100% pollen of this Asteraceae type, which can be assigned to Tripolium pannonicum and the contamination rate is ~ 2% (Suppl. material 3: Appendix 3).

Specimens without assignment to a species: Due to unclear morphological characters, seven specimens, which were used for pollen analyses, could not be clearly assigned to a specific species. Thus, they are marked with “???” in Suppl. material 3: Appendix 3. Most of these females were captured on the Bisamberg (n = 5) and chiefly collected Reseda sp. pollen. Two of them (nos. 62 and 64) additionally collected a small amount of pollen from Asteraceae tubuliflorae (13–19%). The other three (nos. 61, 63 and 65) had pollen loads with over 90% Reseda sp. and some amounts of contamination (0.3–2.0%). A female from Spitzerberg near Hainburg (no. 66) also preferred Reseda sp. (99.3%) and a single female (no. 148), for which both the date and location of capture are unknown, had a pure pollen package with Calluna vulgaris pollen (100%).

Phylogeny

The obtained sequences were aligned for comparison (Suppl. material 9: Appendix 9). In the CO1 sequences of the specimens of C. collaris, 27 single base-pair differences were found, which clearly separate them from the other species. These, in turn, show little to no fixed substitutions in their alignment and therefore cannot be separate from each other.

The 21 obtained sequences, as well as 47 reference sequences received from the DNA-Barcode of Life Database (BOLD) (Suppl. material 5: Appendix 5), were used for further analysis by the neighbour-joining tree (NJT) method (Fig. 26). In the retrieved phylogenetic tree, the species C. collaris clusters in two clades strongly supported with Bootstrap values of 99% each and thus appears paraphyletic (as later found out, by a misidentification): clade 1 contains all the specimens from the present study (n = 4), as well as data from BOLD (n = 16), whereas clade 2 consists of two BOLD sequences (KC469653 and KC469654). The other four species (C. brevigena, C. succinctus, C. hederae and C. pannonicus) collapse into one large clade, which is supported by a moderate Bootstrap value of 88% (Fig. 26). They appear as the sister group to one specimen of C. brevigena (no. DQ085546), with a Bootstrap value of 99%.

Figure 26. 

Neighbour-joining tree of the sequences of the species of the Colletes succinctus group obtained in this study (n = 21, marked in red) and reference data (n = 47) from BOLD with outgroup and Bootstrap values (1,000 re-samples). The specimens KC469653 and KC469654 in BOLD (marked in green), originally assigned to C. collaris, were later identified as a different species, C. luzhouensis Kuhlmann, 2007, that does not belong to the C. succinctus group. The scale bar represents 0.01 substitutions per site.

For a better illustration of the relationships amongst these clustering species, a median joining network was created (Fig. 27). Most of the closely-related specimens share one common haplotype. As in the NJT, the two C. collaris clades are highly distinct. Clade 2 of C. collaris (KC469653 and KC469654), later identified as C. luzhouensis Kuhlmann, 2007, is separated from the main haplotype (C. succinctus, C. brevigena, C. hederae and C. pannonicus) by fifty substitution steps. One specimen of C. succinctus (no. BCZSMHYM02017) and one specimen of C. brevigena (no. DQ085546) differ from the group by three substitution steps and two hypothetical haplotypes that could not be found in the sample.

Figure 27. 

Median-joining network of the analysed species of the Colletes succinctus group (n = 20) and reference data (n = 47) from BOLD (Barcode of Life Database). (tick marks) substitution steps, (black dots) hypothetical haplotype that cannot be found.

Genetic distances

The interspecific distances between the investigated species are lowest between C. brevigena and C. hederae and highest between C. collaris and all the other species with the exception of the outgroup (Colletes creticus) (Table 7). The intraspecific distance, in turn, is slightly higher than the interspecific distance in C. brevigena, C. succinctus, C. pannonicus and C. hederae. It is the lowest amongst the C. collaris examined in this study (Table 8). Some of the investigated specimens of C. brevigena, C. succinctus and C. hederae have identical sequences and show no genetic distance (Table 7). A pairwise genetic distance matrix can be found in Suppl. material 10: Appendix 10.

Including sequence data from BOLD shows no differences. In addition, in this combined dataset, C. collaris is the most differentiated species (Tables 7 and 8). However, two sequences from BOLD, labelled as C. collaris, were highly differentiated resulting in a higher intraspecific distance than the interspecific distances to the other ingroup species and even higher than distances between ingroup and outgroup (Table 7).

New features were found for species differentiation in the Austrian C. succinctus group. Nevertheless, there is a pronounced intraspecific variation (between populations of different collecting sites) in all species. This also concerns previously-described diagnostic characters for females (Noskiewicz 1936; Schmidt and Westrich 1993; Burger 2010; Hölzler and Mazzucco 2011), like the wrinkles on the clypeus, the puncturation on the mesonotum and the presence/absence of the reticulate micro-sculpture on the galea. Due to these sometimes very confusing character combinations, it was not possible to clearly assign all specimens of this study to a species. A few female specimens showed combined traits of the species C. succinctus and C. hederae or C. succinctus and C. brevigena, one even a mixture of characters of all three species. This can best be explained by the hypothesis that, in these young species, not all diagnostic characters are yet fixed in all populations.

By combining the new features with the already-known characters from literature (Noskiewicz 1936; Verhoeff 1944; Schmidt and Westrich 1993; Hölzler and Mazzucco 2011; Amiet et al. 2014), it was possible to establish identification keys for females from Austria and males from Central Europe. With these identification keys, the species C. succinctus, C. collaris and C. hederae can be reliably distinguished in both sexes. Colletes halophilus and C. pannonicus live in very similar habitats, but were never compared to each other. The initial hypothesis that they could be synonyms was not corroborated: the dorsobasal knob on the membrane of the gonostylus distinguishes C. halophilus from all other Central European species. Colletes brevigena and C. pannonicus remain to be difficult to determine as they are very similar in appearance. Only slight differences of the dorsal shape of the female’s fovea facialis, as well as the male’s genitalia, provide indications for the two-species hypothesis. The cuticle of the fovea facialis bears many secretory cells and is more strongly developed in females than in males (Schuberth and Schönitzer 1993). The differently shaped dorsal margins of the fovea are often used as a character for species identification in other genera, especially Andrena (Schmid-Egger and Scheuchl 1997).

The determination of females by use of the new identification key is not supported by the discriminant analysis (LDA). Based on their morphometric data, the LDA assigned only about half of all specimens to the same species as previously determined by the authors. For this method, specimens of C. collaris were used as reference species. Colletes collaris possesses several morphological characters (pilosity of propodeum, narrow band of setae on terga etc., compare Amiet et al. 2014) which do not show intraspecific polymorphism. As C. collaris was observed to have been mis-assigned by the discriminant analysis, the character sets, which were used for the LDA, are not suitable to differentiate the species.

In general, the morphometric analysis of the selected head characters and thorax width alone did not prove to be informative enough to distinguish females of the examined species of the C. succinctus group, as, for example, suggested for C. pannonicus by Hölzler and Mazzucco (2011). No morphometric studies have yet been published for the studied species, but other species of the genus Colletes have already been investigated: because of their similar appearance, especially in their puncturation on terga, C. inexpectatus Noskiewicz, 1936 and C. daviesanus Smith, 1846 were regarded as synonyms (Warncke 1978). Prǐdal (1999) was able to verify that both represent independent species, amongst other characters, by the measurements of the male’s hind tarsus. Therefore, the present results call for subsequent examinations. Maybe it is possible to gain more information about species differentiation by measuring legs or antennae. Additionally, males should be added to morphometric studies.

In this study, the species of the Colletes succinctus group occurring sympatrically in Austria were both polylectic and oligolectic: the investigated females of C. succinctus, C. collaris and C. hederae showed a polylectic pollen-collecting behaviour, C. pannonicus appeared to be oligolectic, but examination was based on a single population. In addition, a correlation between the filling ratio of the pollen packages and the number of different pollen types could be determined: the larger the pollen load, the more different pollen types could be found.

The present study showed a preference of the “heather bee” C. succinctus for Reseda sp. (hitherto unknown as a pollen source), closely followed by Calluna vulgaris and Asteraceae, whereas Müller and Kuhlmann (2008) found that the species collected pollen on Ericaceae, Araliaceae, Asteraceae and Apiaceae. Interestingly, some specimens of the C. succinctus, collected in Retz, had the expected pollen of Calluna vulgaris, but some others used the pollen of Reseda sp. thriving in close vicinity, although there was no obvious shortage of Calluna flowers. That the females had either pure Calluna vulgaris or Reseda sp. pollen loads can be explained by flower consistency (cf. Waser 1986).

The single analysed female of C. brevigena had a pollen load of pure Reseda sp.; this species is described as polylectic by Müller and Kuhlmann (2008), collecting pollen on a variety of different families. These authors also described C. hederae as polylectic, whereas Bischoff et al. (2005) described the species as oligolectic due to the fact that examined nest cells and pollen loads only showed pollen of Hedera helix. In this study, most of the females collected pure pollen loads from Hedera helix flowers. Only females from Donaupark in Vienna showed polylectic behaviour by adding pollen of Citrus sp. to their pollen packages. This plant genus is recognised for the first time as a pollen source for C. hederae. The other females of C. hederae were captured in Linz and several sites in Vienna, directly on H. helix. During the rather late flight period of C. hederae (September–November; Kuhlmann et al. 2009), there is no particularly large selection of flowering plants available. Ivy flowers offer an easily accessible and nutrient-rich pollen resource. Nonetheless, it was observed that, if other nutritious pollen sources are available, C. hederae disregards Hedera sp. and starts collecting pollen from the other sources. Therefore, the polylectic pollen-collecting behaviour of C. hederae is sometimes referred to as pseudo-oligolectic; this means that they are sometimes only oligolectic by lack of choice (Teppner and Brosch 2015).

Little is known about the pollen-collecting behaviour of C. pannonicus. Field observations led to the assumption that C. pannonicus is strictly oligolectic on Tripolium pannonicum (sea aster) (Hölzler and Mazzucco 2011). The first pollen analysis, conducted within this study, confirmed that C. pannonicus from the same population is oligolectic on Asteraceae. However, as all investigated females were collected on T. pannonicum, the results can either be explained by strict oligolectic behaviour or by flower consistency.

Although the examined females of C. collaris showed a strong preference for Asteraceae, which is in accordance with findings by Müller and Kuhlmann (2008), they also collected pollen of Reseda sp. Asteraceae are omnipresent, provide large amounts of pollen and nectar and flower, depending on the species, from spring to autumn, but they have a low protein and amino acid content (Somerville and Nicol 2006) and a possibly toxic pollen kit (Williams 2003) and the extraction of important nutrients from the pollen plasma is difficult (Peng et al. 1985). Therefore, expensive physiological adaptations of the bees to this pollen type are necessary: for coping with a low protein and amino acid content, for detoxification and for an easier extraction of important nutrients from the pollen plasma (Müller and Kuhlmann 2008). Asteraceae are the perfect nutrient supplier for adapted bees, being available almost without competition. In Colletes, Asteraceae are preferentially collected by oligolectic species, but are largely avoided by the majority of polylectic species; exceptions are foremost found of the Colletes succinctus group: C. succinctus, C. brevigena and C. hederae (Müller and Kuhlmann 2008).

Only specimens of C. collaris can be clearly separated from the other species. By analysing only sequences of the material of this study, C. collaris forms a monophylum and the sister group to the other analysed species, which collapse into one large clade and show little to no genetic distance to each other. However, after adding sequences from BOLD, C. collaris separates into two clades and forms a paraphylum. This can be explained by checking the two C. collaris sequences that appear secluded in the tree: an additional BOLD blast showed that both specimens belong to Colletes luzhouensis Kuhlmann, 2007, a species native to China, which explains not only the high intraspecies genetic distance of C. collaris, but also the high genetic distances to the other species. Thus, C. collaris forms a monophylum and is the sister group to a clade comprising the remaining studied species of the C. succinctus group. A previous study of Kuhlmann et al. (2009), albeit including only one specimen of C. collaris, analysed the gene fragments CO1 and 28S and achieved the same result: C. collaris represented the sister group to C. intricans, C. succinctus, C. hederae, C. brevigena and C. halophilus.

In this study, C. pannonicus was examined for the first time by using DNA barcoding. However, it cannot be distinguished from the other investigated species, except C. collaris. Since the species collapse into one large clade/haplogroup (except C. collaris), it is not possible to assign specimens to species using the CO1 sequence, which is in accordance with previous studies (Kuhlmann et al. 2007, 2009). For future investigations of the group, it would be recommended to investigate other genes.

Colletes brevigena and C. pannonicus proved to be the most challenging species. Due to their similar appearance, it is difficult to distinguish them strictly by morphology. There is a small difference in the female’s fovea facialis that was not mentioned in the original description of the holotype of C. pannonicus (Hölzler & Mazzucco, 2011), whereas the previously-stated morphometric differences could not be approved in larger material. The previously undescribed male of C. pannonicus shows only discrete differences in its genitalia. Furthermore, some Austrian specimens of C. brevigena showed different morphological character states than specimens from the Mediterranean. Therefore, both species were investigated more closely.

All examined specimens of C. pannonicus were found solely near to the type locality in the Seewinkel where they can be observed flying around the flowers of Tripolium pannonicum (sea aster). That all known specimens were caught around Lake Neusiedl is surprising, as there are no geographical barriers which would prevent a wider distribution. Specialisation to a distinct habitat, salt meadows, seems the most likely and hitherto accepted explanation that C. pannonicus could not be found elsewhere in Austria. The proposed (Hölzler and Mazzucco 2011) and here confirmed apparent dependence on T. pannonicum could be explained by the fact that hardly any other Asteraceae are blooming at these sites in late autumn. Tripolium pannonicum is a widespread plant found in Europe and in temperate regions of Asia (Euro+Med 2006) and it can be expected that C. pannonicus has a wider distribution than presently known. The seeming endemism is either caused by under-collecting in other suitable areas or, possibly, by confusion with similar species, mainly C. brevigena.

Noskiewicz (1936) described C. succinctus ssp. brevigena, based on specimens from a large distribution area spanning from the Balkans, to Crete, Persia and the Caucasus, but many specimens are untraceable. The selection of a lectotype was necessary to define this problematic species. The type locality is Crete (Greece).

A clear morphological distinction could be detected between Austrian specimens (from Bisamberg and Ollersdorf, Lower Austria) of C. brevigena and specimens from the Mediterranean region. The Austrian females are unusually large and show a puncturation on terga, mesonotum and mesopleura that is similar to C. succinctus and/or C. hederae. Only some specimens from Spitzerberg (Lower Austria) are more similar to Mediterranean C. brevigena. A linear discriminant analysis (LDA) of morphometric data resulted in clear separation of Austrian and Mediterranean specimens (100%). Subsequently, in a Jackknife re-sampling, only 60% of all specimens could be assigned to their original group (Austrian or Mediterranean). This lower value may be explained by the small number of Austrian specimens (n = 5) in comparison to the higher number of Mediterranean specimens (n = 13). To determine whether Austrian and Mediterranean specimens differ in their measurements or are more similar than assumed in this study, a larger number of samples would be needed for measurement. Unfortunately, due to the rarity of this species in Austria, this was not possible during this study.

Zettel et al. (2006) also listed questionable females of C. brevigena in Ollersdorf. These specimens showed a puncturation on their terga that is similar to the puncturation of C. hederae, but as typical for females of C. brevigena, they have longitudinal wrinkles on their clypeus. Therefore, the authors classified them with reservation as C. brevigena (Zettel et al. 2006).

Concluding the findings for C. brevigena, this study raises doubts about the close relationship between specimens/populations of C. brevigena occurring in Austria and specimens from the Mediterranean region. In addition to the morphological differences, also the phenology differs: in the Mediterranean, C. brevigena is bivoltine (Kuhlmann 2003; Standfuss 2009): the first generation flies in May and the second generation later in the year (according to collection data from September to November). In Austria, however, there is only one generation active from August to September (Zettel et al. 2006). It can be assumed that C. brevigena is a Mediterranean species that post-glacially migrated northwards, where it must have adapted to different environmental conditions, most importantly to a shorter warm season. In this case, it would be expected that – like in some other bee species (e.g. Andrena pontica Warncke, 1972; Scheuchl and Willner 2016) – the second generation is omitted because of the longer-lasting development. In the peculiar case of C. brevigena, however, the spring generation is omitted. As most specimens of the genus Colletes in Central Europe spend their diapause (hibernation) as a pre-pupa (in the last larval stage) (Westrich 1989), this delayed development into an imago should be genetically fixed and, therefore, must be a trait that was passed on by its ancestor. A similar case is also known from the species complex of Andrena argentata Smith, 1844. This bivoltine species shows a trans-Palaearctic distribution. In England and Sweden, it either has no spring generation or the summer generation is richer in specimens. Additionally, in this case, it is assumed that the two generations do not belong to the same species (Scheuchl and Willner 2016).

Thus, it would be quite possible that the Austrian specimens in question are of a different species that is very similar to C. brevigena in the Mediterranean. However, it is also possible that C. brevigena shows a highly pronounced geographic variation. The studied populations in Austria and the Mediterranean region are geographically far apart and differ to such an extent that they can be regarded as conspecific only with difficulty. It would be advisable to study C. brevigena populations from intermediate areas, for example, from Hungary or the northern Balkans. Thereby, it may be possible to find transitional morphs that could corroborate the conspecificity of the two morphologically different groups. Both possibilities would merit further investigation.

Colletes brevigena and C. pannonicus share a very similar morphological character set. Genetic data (mitochondrial gene CO1) are not useful to differentiate between them. It is difficult to distinguish the two species by the discriminating characters described by Hölzler and Mazzucco (2011), as morphometric data greatly overlap. For example, Hölzler and Mazzucco (2011) stated that the female of C. pannonicus possesses a wider head in relation to the thorax width, when compared to C. brevigena. This was not confirmed by this study. The values of both species greatly overlap, with C. brevigena even showing mean values for a slightly wider head than C. pannonicus. These findings can support the assumption from the previous chapter, that C. pannonicus can easily be misidentified as C. brevigena. In the salt meadows of Seewinkel, however, the two species do not occur sympatrically, but C. pannonicus shares the habitat with C. collaris. Colletes brevigena, on the other hand, is only found in steppe biotopes (Zettel et al. 2006). Due to the few subtle differences, it is quite possible that specimens of C. pannonicus were misidentified as C. brevigena in other areas of south-eastern Europe. Additionally, the oligolectic collecting behaviour on Tripolium pannonicum in Seewinkel could simply be based on the lack of alternative resources. Therefore, it is important to include pollen preferences, as well as habitat preferences, to differentiate between both species.

For 21 specimens, it was not possible to assign them to a species, based on their morphological characters alone. They showed mixed characters of several species (C. succinctus, C. brevigena and C. hederae), especially, regarding the puncturation on terga, mesopleura and mesonotum, as well as structures on clypeus and galea.

The intraspecific variation of the species of the C. succinctus group has always been an issue for taxonomists (Noskiewicz 1936; Verhoeff 1944; Schmidt and Westrich 1993; Hölzler and Mazzucco 2011). Some specimens show mixed characters and cannot be clearly identified. Therefore, the ecology (pollen preferences of females and phenology) has been included in species differentiation (Kuhlmann et al. 2007). For example, female no. 148 shows C. succinctus-typical terga, mesopleura and clypeus, but does not have a shiny centre on the mesonotum. Since this was already mentioned by Noskiewicz (1936) as a typical C. succinctus characteristic, further aspects have to be investigated. When examining the pollen package of the specimen, it consists purely of Calluna vulgaris. Thus, the probability is very high that this specimen is a specimen of C. succinctus which, however, did not develop all characters typical for this taxon. In addition, the females (nos. 61–66), which were collected at Bisamberg (Vienna), show combined characters of C. succinctus, C. brevigena and C. hederae. These specimens were collected mid-September on Reseda sp. (pollen loads studied). Although Reseda sp. is only known as a pollen source for C. brevigena and C. collaris in literature (Müller and Kuhlmann 2008) and for C. succinctus in the present study, it is not impossible that C. hederae also collects pollen on this flower. Colletes hederae is a polylectic species, which shows preferences for Hedera helix, but also collects pollen on other plants, should their preferred plant not yet be in bloom (Müller and Kuhlmann 2008; Westrich 2008; Teppner and Brosch 2015). Moreover, flower visits of C. hederae on Reseda sp. have been observed after this study was carried out (H. Zettel, unpubl.).