Andrena lathyri is a 10–14 mm long ground-nesting solitary bee, which is distributed over large parts of Europe and Turkey (Gusenleitner and Schwarz 2022). In Central Europe, the species is widespread at lower elevations and regionally often abundant. It belongs to the subgenus Taeniandrena, which comprises about 35 Palearctic species with one species additionally introduced into the Nearctic (Gusenleitner and Schwarz 2002; Wood et al. 2021; Praz et al. 2022; Wood 2022; T. Wood, personal communication). All species of this subgenus, for which the pollen hosts are known, exhibit an exclusive or very strong preference for the pollen of Fabaceae (Westrich 1989; Praz et al. 2022). Together with A. aberrans Eversmann, which exclusively collects pollen on few genera of the tribe Genisteae, such as Chamaecytisus (Westrich 2018), A. lathyri is among the most strongly pollen-specialised representatives of A. (Taeniandrena) exploiting only two closely related genera of the tribe Fabeae, i.e., Lathyrus and Vicia (Westrich and Schmidt 1987; Schaefer et al. 2012). As the proboscis of A. lathyri is too short to reach the nectaries at the base of the Lathyrus and Vicia flowers, nectar cannot be ingested during legitimate flower visits. Instead, A. lathyri gains nectar by primary nectar robbing on its pollen hosts.

Lathyrus pratensis L. and Vicia sepium L. have typical legume flowers with a five-part calyx and corolla (Kugler 1970; Proctor et al. 2003; Fig. 6a–f). The calyx consists of five sepals, which are fused over most of their length ending in five apical lobes (“calyx teeth”). The corolla is composed of five petals, of which the uppermost petal is in dorsal position (“standard”), the median two petals are in lateral position (“wings”) and the lowermost two petals are in ventral position largely concealed by the lateral wings. The lowermost two petals are ventrally and apically fused forming a boat-shaped structure (“keel”) that encloses the single pistil and ten stamina, of which the dorsalmost stamen is free, whereas the filaments of the other nine stamina are fused to a staminal tube surrounding the pistil. The flowers have a secondary pollen presentation in that the pollen is shed at the late bud stage onto a dense brush of fine hairs near the apex of the style (“stylar brush”), from where it is removed by pollen-collecting bees. Nectar is produced at the base of the pistil in a nectary, which is longitudinally crossed by the filament of the uppermost stamen, either slightly above the nectary’s upper rim in L. pratensis or slightly below the upper rim in V. sepium (Fig. 6e, f) The nectar is accessible only from above the staminal tube, where there is a spacious empty flower interior due to the considerably arched base of calyx and standard. To drink nectar from the flowers of L. pratensis, flower visitors have to pass the proboscis through the rather narrow slit on either side of the stamen filament, while the nectar is slightly more easily accessible in V. sepium due to the lower position and the smaller width of the crossing stamen filament.

Field observations and experiments were performed in northern Switzerland near Rekingen (47°33'59"N, 8°18'41"E; site 1), Dagmersellen (47°13'28"N, 7°58'49"E; site 2) and Wädenswil (47°13'06"N, 8°40'45"E; site 3) from May to June 2021 and 2022.

To verify the pollen host specialization of Andrena lathyri to flowers of Lathyrus and Vicia, we microscopically analysed the pollen contained in the hind leg scopa of 30 females collected at 30 different localities in Switzerland (n = 28) and Liechtenstein (n = 2) between 1914 and 2019 using the method described by Westrich and Schmidt (1986). Before removing pollen from the scopa, the amount of pollen was assigned to five classes, ranging from 5/5 (full load) to 1/5 (filled to one-fifth). The pollen grains were stripped from the scopa of one leg with a fine needle, embedded in glycerol gelatin on a slide and identified at a magnification of 400× to family, subfamily or genus level. While pollen of Lathyrus and Vicia can be easily distinguished from that of other Fabaceae taxa in Central Europe by light microscopy based on shape, ornamentation and size of the hydrated pollen grains, there are no reliable characters to separate Lathyrus from Vicia pollen in every case (Beug 2004). Therefore, pollen of these two closely related taxa was recorded as Lathyrus/Vicia type in the pollen samples examined.

To clarify whether Andrena lathyri is an obligatory nectar robber on flowers of Lathyrus and Vicia or whether it exploits numerous other plant taxa for nectar, we analysed the flower-visiting data for A. lathyri contained in the database of the Wildbienen-Kataster Baden-Württemberg. At the time of data retrieval (August 2022), the database comprised 392 records of A. lathyri, which were distributed all over Baden-Württemberg, collected from 1988 to 2021 and provided by M. Haider, M. Klemm, V. Mauss, R. Prosi, A. Schanowski and H.-R. Schwenninger. For 189 out of these 392 records, the plant genus or plant species visited by A. lathyri was known.

To analyse the morphology of the mouthparts of Andrena lathyri, the proboscis was examined under a stereomicroscope at a magnification of 40× and compared with that of nine other western Palearctic Andrena species of the subgenus Taeniandrena, i.e., A. aberrans, A. afzeliella (Kirby), A. gelriae Van der Vecht, A. caesia Warncke, A. intermedia Thomson, A. leucopsis Warncke, A. poupillieri Dours, A. russula Lepeletier, and A. wilkella (Kirby). For measurements, A. wilkella was selected as representative species for the subgenus Taeniandrena as its proboscis was found to be morphologically identical to all A. (Taeniandrena) species other than A. lathyri. For five females and five males each of A. lathyri and A. wilkella originating from different localities in Switzerland, the length of the glossa from the lowermost point of the basiglossal sclerite to its apex and the intertegular width were measured with a micrometer scale to the nearest 0.025. For the same individuals, the angle between the dorsal surface of the labium’s lateral sclerites and the anterior surface of the glossa was determined on close-up photographs taken with a Nikon D750 camera. The morphological terminology of the proboscis follows Michener (2000).

To examine the behaviour of Andrena lathyri during primary nectar robbing, we observed females and males gaining nectar from flowers of Lathyrus pratensis and Vicia sepium with the aid of a threefold magnifying glass at site 1 on 16.6.2021 and at site 2 on 17.5.2022. In total, we observed about 90 nectar-robbing visits.

To understand the movements of the labium during nectar ingestion, two females of Andrena lathyri were observed in the laboratory drinking sugar water from a small bowl.

To investigate whether the long and angled proboscis of Andrena lathyri fits into the flower interior for nectar uptake, three consecutive sections of the labium of ten females and ten males originating from different localities in Switzerland were measured with a micrometer scale to the nearest 0.025 mm under a stereomicroscope at 40× magnification (Fig. 2a): i) length from the sclerotised base of the prementum to the sclerotised base of the labial palpi (“basal section”), ii) dorsal length of the intermediate section extending from above the sclerotised base of the labial palpi to and including the basiglossal sclerite (“intermediate section”), and iii) length of the anterior surface of the glossa from the dorsalmost point of the basiglossal sclerite to the tip of the glossa (“apical section”). The measured lengths were averaged over all 20 individuals, multiplied with ten and plotted onto a 10:1 drawing of an average-sized flower each of Lathyrus pratensis and Vicia sepium to optically evaluate the labium’s fit to the flower interior and its ability to reach the nectary.

To examine the behaviour of Andrena lathyri during pollen collection, we observed females harvesting pollen on flowers of Vicia sepium with the aid of a threefold magnifying glass at site 2 on 17.5.2022 and at site 3 on 24.5.2022. In total, we observed about 30 pollen-collecting visits.

To evaluate the damage caused to flowers by nectar-robbing individuals of Andrena lathyri, we collected 50 robbed flowers each of Lathyrus pratensis and Vicia sepium at site 1 on 4.5., 12.5. and 25.5.2022 and examined both corolla and calyx under a stereomicroscope at a magnification of 20–40×.

To investigate whether the fruits of flowers robbed by Andrena lathyri develop normally, 25 flowers of Vicia sepium that showed the typical sign of having been robbed, i.e., a longitudinal slit in the calyx (see below), were marked with coloured threads at site 3 on 24.5.2022. Two weeks later, the development of the fruits of the marked flowers was assessed.

All 30 female pollen loads of Andrena lathyri from Switzerland and Liechtenstein exclusively consisted of pollen of the Lathyrus/Vicia type (Fig. 1a). This finding supports the species’ narrow pollen specialisation already postulated by Westrich and Schmidt (1987), who found 46 pollen loads from Germany, Austria, Poland and Greece to be composed only of Lathyrus and/or Vicia pollen.

Figure 1. 

Pollen and nectar sources of Andrena lathyri a composition of female pollen loads b flower visits. Dark grey = Lathyrus and Vicia species, light grey = other plant species.

Based on the Wildbienen-Kataster dataset, 69 (87.3%) of 79 flower visits by females of Andrena lathyri and 102 (92.7%) of 110 flower visits by males were recorded on flowers of Lathyrus, such as L. niger (L.) Bernhardi, L. pratensis L. and L. vernus (L.) Bernhardi, and of Vicia, such as V. angustifolia L., V. cracca L., V. sativa L., V. sepium L. and V. villosa Roth (Fig. 1b). Considering that the females mostly collect both pollen and nectar during the same flower visit (see below) and the males exploit the flowers only for nectar, these figures suggest that A. lathyri only exceptionally gains nectar from flowers other than Lathyrus and Vicia, rendering the species an almost obligatory nectar robber.

The comparison of the proboscis of Andrena lathyri with that of nine other Andrena species of the subgenus Taeniandrena revealed that the morphological differences between A. lathyri and its relatives are restricted to the labium, whereas the construction of the maxillae is identical.

The labium of Andrena (Taeniandrena) species consists of five main sclerotised parts (Fig. 2a, b): i) the prementum, which extends till the base of the labial palpi, ii) the two four-segmented labial palpi, which attach laterally to the end of the prementum, iii) a pair of lateral sclerites, which cover the base of the paraglossae from above and form together with the adjacent basiglossal sclerite an intermediate section between the end of the prementum and the anterior surface of the glossa, iv) the two paraglossae, which attach distal to the ventral end of the prementum, run along the lower margin of the two lateral sclerites and are distally bent outwards, and v) the glossa, which originates between the base of the paraglossae, tapers towards its apex and is densely beset with annulate hairs below the basiglossal sclerite.

Figure 2. 

Mouthparts of Andrena lathyri and Andrena wilkella in lateral view a labium of A. lathyri b labium of A. wilkella c maxilla and labium of A. lathyri. pm = prementum, lp = labial palpus, ls = lateral sclerite, bs = basiglossal sclerite, pg = paraglossa, gl = glossa; I = basal section, II = intermediate section, III = apical section.

The labium of Andrena lathyri differs in two main respects from that of all other A. (Taeniandrena) species represented here by A. wilkella (Fig. 2). First, the glossa is on average 1.57 mm long in A. lathyri (range = 1.48–1.68 mm, n = 5 females and 5 males), whereas it is on average 0.50 mm long in A. wilkella (range = 0.45–0.55 mm, n = 5 females and 5 males). By correcting for the difference in body size, which is 10% smaller in A. wilkella based on intertegular width, the glossa of A. lathyri is about 2.9× longer than that of A. wilkella. Second, the angle between the dorsal surface of the lateral sclerites and the anterior surface of the glossa is on average 100.8° in A. lathyri (range = 91–109°, n = 5 females and 5 males), whereas it is on average 140.9° in A. wilkella (range = 130–152°, n = 5 females and 5 males). These differences result in a much longer and distinctly more strongly angled labium in A. lathyri compared to most other Andrena species including the closely related representatives of the subgenus Taeniandrena.

The behaviour of nectar-robbing individuals of Andrena lathyri, which was very uniform, identical in both sexes and invariant between flowers of Lathyrus pratensis and Vicia sepium, can be divided into three phases. In the first phase, the bees crawled headfirst to the side of the flower and inserted the half-extended proboscis and the lower half of the head under the lower margin of the standard (Figs 3a, 6c, d). In the second phase, the bees moved the head between the inner side of the standard and the outer side of the wing towards the base of the flower (Fig. 3b); when the advancing head reached the calyx, the outer mandible was spread out (Fig. 3c), and by moving the head further towards the flower base, the calyx was torn open between the lowest and the second lowest calyx tooth (Fig. 3d); during advancing, the outer mandible was repeatedly moved up and down, which probably facilitated the process of tearing the rather hard calyx tissue by acting as an abutment or support, but not as cutting tool; instead, the calyx tissue tore due to the pressure of the advancing head; sometimes, a soft crackling sound was heard, which probably occurred when the tissue tore; at the end of this phase, the outer mandible rested on the calyx and the inner mandible, the proboscis and one compound eye were completely hidden inside the flower (Fig. 3d). In the third phase, the bees — holding their head obliquely upwards from its ventral position relative to the flower —extended the proboscis inside the flower in longitudinal direction to the head upwards, so that the uppermost point of the angled proboscis was close to the upper roof of the flower base and the lowermost point, i.e., the apex of the glossa, reached the nectary from above to drink nectar (Fig. 4); after nectar uptake, the bees withdrew the proboscis from the calyx slit and left the flower (Fig. 3e).

Figure 3. 

Nectar-robbing behaviour of Andrena lathyri. See text for details.

Figure 4. 

Maximally extended labium of Andrena lathyri plotted true to scale into the flower interior a Lathyrus pratensis b, c Vicia sepium. Black = labium, grey = nectary and calyx slit.

In all observed nectar-robbing visits (n = 90), only one side of the calyx was slit, suggesting that the proboscis can empty the nectary from one side despite the longitudinally crossing stamen filament and that calyces slit on both sides originate from two different visits by Andrena lathyri.

How often individuals of Andrena lathyri gained nectar by secondary nectar robbing, i.e., by using an already existing calyx slit, could not be quantified exactly, because the observer’s view of the flower base became obscured as soon as the bee crawled to the side of the flower to rob nectar. However, secondary nectar robbing occasionally occurred (Fig. 3f).

The observation of nectar-drinking females of Andrena lathyri in the laboratory revealed that the glossa can be moved far forwards and backwards in longitudinal direction to the body due to a ventral joint adjacent to the distal end of the prementum. In addition, the haired part of the glossa itself is movable to all sides, allowing for its precise guidance within the flower. In repose, the glossa is folded back over the dorsal surface of the prementum, which is embedded between the stipites of the maxillae within the proboscidial fossa on the underside of the head. For nectar ingestion, the prementum is moved downwards and forwards followed by the folding out of the glossa. At maximum extension of the proboscis, the sclerotised base of the prementum is situated roughly underneath the labrum resulting in a long maximal reach of the proboscis. By moving the prementum forward at different distances and/or bending the joint at the end of the prementum at varying angles, the mouthparts have a considerable flexibility to adjust to the specific interior of the host flowers, which is expected to slightly differ among the different species of Lathyrus and Vicia that are exploited by A. lathyri (see above).

The basal, intermediate and apical section of the labium of Andrena lathyri measured on average 1.91 mm, 0.30 mm and 1.74 mm, respectively (n = 10 females and males each). The true-to-scale plotting of these three sections, i.e. the labium at its maximum extension, onto flower drawings of Lathyrus pratensis and Vicia sepium revealed - in combination with the observed position of the bee’s head during primary nectar robbing and the fact that the nectaries can be accessed only from above (see above) - that the bee’s mouthparts precisely correspond to the dimensions of the flower interior and the position of the nectary (Fig. 4a–c).

To collect pollen on flowers of Vicia sepium, the females of Andrena lathyri pushed the standard upwards with the front of the head, which created the necessary space for the movements of the fore legs (see below), and simultaneously pressed the wings sideways down with the middle and hind legs, which caused the keel to move downwards and the pollen-bearing stylar brush to emerge from the tip of the keel (Fig. 5a). The bees then harvested the pollen from the stylar brush by rapid strokes of the fore legs, which resulted in pollen masses adhering to both the pilosity of the foretarsi and the long hairs covering the underside of the maxillar stipites (Fig. 5b).

Figure 5. 

Pollen-collecting behaviour of Andrena lathyri. See text for details.

During most flower visits on Lathyrus pratensis and Vicia sepium, the females of Andrena lathyri first collected pollen, before crawling to one side of the flower to rob nectar. Occasionally, however, pollen-only and nectar-only visits also occurred.

Of the 50 corollae each of Lathyrus pratensis and Vicia sepium examined for damage by nectar-robbing individuals of Andrena lathyri, 81 (81%) were intact and 16 (16%) could not be properly assessed as the corolla was partly withered due to the late development stage of the flower; only three (3%) flowers of V. sepium were found to have a 1.8–5.8 mm long longitudinal crack near the lower margin of one of the two lateral wings, most probably caused by A. lathyri during nectar robbing.

Primary nectar robbing by Andrena lathyri on Lathyrus pratensis and Vicia sepium invariably resulted in a longitudinal slit in the calyx (Fig. 6a, b). This slit was located between the lowest and the second lowest calyx tooth in all 100 flowers examined with the exception of one flower of V. sepium, where the slit was located between the second lowest and the third lowest calyx tooth; interestingly, the corolla of the latter flower was one of the very few damaged by nectar robbing (see above), suggesting that this flower might possibly have been robbed by an unexperienced forager. The calyx was slit only on one side in 89 (89%) flowers examined, whereas it was slit on both sides in 11 (11%) flowers, probably resulting from two different flower visits (see above). The length of the calyx slit was significantly shorter in L. pratensis (range = 0.7–2.1 mm, median = 1.3 mm, n = 54) than in V. sepium (range = 2.0–3.9 mm, median = 2.8 mm, n = 57) (Mann-Whitney-U-test, U = 2.5, p < 0.001). In contrast, the distance between the end of the slit and the base of the flower, where the nectary is located in both species, did not significantly differ between L. pratensis (range = 1.4–2.6 mm, median 2.1 mm, n = 54) and V. sepium (range = 1.4–2.9 mm, median = 2.2, n = 57) (Mann-Whitney-U-test, U = 1325.5, p = 0.206).

Figure 6. 

Flowers of Vicia sepium (left) and Lathyrus pratensis (right) a, b longitudinal slit in calyx due to primary nectar robbing by Andrena lathyri c, d flower with calyx removed showing lower side of standard (arrow), under which A. lathyri moves head and proboscis towards flower base e, f flower in top view with calyx, standard and wings removed showing nectary (ne), which is accessible only from above g, h normally developing fruits despite slit in the calyx (arrow) caused by A. lathyri.

In 21 (84%) of the 25 marked flowers of Vicia sepium, the fruits developed normally in spite of having been robbed by Andrena lathyri (Fig. 6g, h), in one flower (4%) the ovary was withered and three flowers (12%) dropped off the plant for unknown reasons, overall suggesting a negligible negative impact of primary nectar robbing on fruit development.