The hidden diet – examination of crop content reveals distinct patterns of pollen host use by Central European bees of the genus Hylaeus (Hymenoptera, Colletidae)

Masked or yellow-faced bees of the genus Hylaeus (Colletidae) differ in their mode of pollen transportation from most other bees in that they ingest the pollen directly on the flowers and carry it back to the nest inside the crop located in the anterior half of the metasoma. Due to this hidden mode of pollen transportation, the examination of pollen collected by Hylaeus females requires the dissection of the metasoma. Although this method has never been applied in Europe, the great majority of the Central European Hylaeus species were supposed to be pollen generalists based on observations of flower visits. The microscopical analysis of pollen removed from 30 crops each of 36 Central European Hylaeus species revealed that the proportion of species exhibiting an exclusive or strong preference for pollen from a single plant taxon is much higher than hitherto assumed and that the current assumption of the genus Hylaeus to largely consist of pollen generalists is wrong. Nineteen of the 36 species examined are strictly or largely dependent on a single plant taxon for collecting pollen, such as Apiaceae (n = 11 species), Rosaceae (n = 3), Reseda (Resedaceae) (n = 2), Allium (Amaryllidaceae) (n = 1), Asteraceae (n = 1) and Melilotus (Fabaceae) (n = 1). The 36 Hylaeus species examined collected pollen from the flowers of 31 plant families, of which the Apiaceae and Rosaceae (particularly Potentilla and Rubus ) were by far the most important contributing almost 60% to the pollen host spectrum of the entire genus. The comparison between pollen host spectrum and flower visiting records showed that the pollen generalists use the flowers of the Asteraceae as nectar rather than pollen sources, corroborating earlier findings that the digestion of Asteraceae pollen requires physiological adaptations to cope with its unfavourable or protective properties. In summary, the patterns of pollen host use by bees of the genus Hylaeus do not substantially differ from those of other Palaearctic bee taxa despite the masked bees’ unusual habit to ingest the pollen directly on the flowers and to transport it inside their body back to the nest.


Introduction
Bees are vegetarian wasps, whose larvae usually develop on a mixture of pollen and nectar within the brood cells of the nests built by the mother bees (Westrich 1989;Michener 2007). In most species, nesting females carry pollen collected on flowers back to the nest on the hind legs and/or on the underside of the metasoma (Westrich 1989;Michener 2007). Due to this external mode of pollen transportation, pollen is easily accessible for examination. In fact, extensive pollen analytical work starting with the seminal publications by Chambers (1968); Raw (1974); Schmidt (1986, 1987) and Westrich (1989) and followed by numerous further investigations for example by Müller (1996Müller ( , 2018; Michez et al. (2008); Müller and Kuhlmann (2008); Sedivy et al. (2008Sedivy et al. ( , 2013; Haider et al. (2014); Wood et al. (2016) or Wood and Roberts (2017) led to a fairly good knowledge of the pollen host preferences of large parts of the Central European bee fauna.
Bees of the genus Hylaeus -a cosmopolitan taxon of colletid bees comprising several hundred species worldwide (Ascher and Pickering 2020) -differ in their mode of pollen transportation from most other bees in that they ingest the pollen directly on the flowers and carry it back inside the metasomal crop to the nest, where it is regurgitated into the brood cells (Westrich 1989;Michener 2007). Due to this hidden mode of pollen transportation, pollen is accessible for examination only after dissection of the metasoma, a method that has never been applied for European species of this genus.
Based on field observations, all Central European Hylaeus species are currently assumed to be pollen generalists ("polylectic") except for three species, which are most probably pollen specialists ("oligolectic") on Allium (Amaryllidaceae), Reseda (Reseda) and Asteraceae, respectively (Scheuchl and Willner 2016;Westrich 2018;Wiesbauer 2020). However, reliable identification of Hylaeus in the field down to species level is strongly hampered by the small size and the uniform morphology of most Central European species, casting doubt on the field-based assumption that the vast majority of species is polylectic. Furthermore, as the spectrum of flowers exploited for nectar is often much wider than for pollen, observations of flower visits without careful differentiation between pollen and nectar uptake poorly reflect pollen host preferences and often conceal pollen specializations (Westrich and Schmidt 1987). In fact, the analysis of pollen remains in larval faeces of three North American Hylaeus species revealed that all three species collected pollen almost exclusively on Rosaceae in spite of long lists of flower visitation records comprising taxa belonging to numerous different plant families (Scott 1996).
In the present study, the pollen host preferences of 36 Central European Hylaeus species including four species restricted in their distribution to higher elevations in the Alps were analysed by microscopical analysis of pollen removed from the crops of collected females. Specifically, the following questions were addressed: i) What are the pollen host spectra of the Central European species? ii) Which plant taxa serve as the most important pollen hosts for the genus in Central Europe? iii) Are there differences between the pollen host spectrum of the genus as assessed in the present study and records of flower visits in the field?

Bee species
Masked or yellow-faced bees of the genus Hylaeus Fabricius (Colletidae) are distributed on all continents except for Antarctica (Michener 2007). Currently, about 760 species are known, of which 47 occur in Central Europe belonging to ten subgenera (Dathe et al. 2016;Ascher and Pickering 2020). The Central European represen-tatives of Hylaeus are small, black, nearly hairless bees usually ranging in length from 3.5 mm to 7 mm. Most species are characterised by the presence of white or yellow markings on the face. The proboscis of all species is very short, limiting nectar uptake to flowers with easily reachable nectar, which is either exposed or -if secreted at the base of the flowers -accessible thanks to the small body size of the bees. The Central European species nest in preexisting cavities such as insect borings in dead wood, hollow stems, soil fissures, abandoned above and below ground nests of aculeate Hymenoptera, abandoned Lipara reed galls or between stones; more rarely, they excavate the nests in the pith of plant stems (Westrich 2018). The brood cell walls are constructed with glandular secretions, which solidify after application by the specialised bilobed tongue to a transparent and cellophane-like waterproof membrane (Batra 1980;Almeida 2008).
For the present study, 36 Central European Hylaeus species were selected representing about 80% of Hylaeus species diversity in Switzerland, Germany and Austria (Dathe et al. 2016). The species identification was based on Amiet et al. (2014) and Dathe et al. (2016). In addition, the publications by Doczkal and Schmid-Egger (1992) and Straka and Bogusch (2011) were used for the proper identification of the two very similar species Hylaeus pictipes and H. taeniolatus and the three species of the Hylaeus gibbus group, respectively.

Pollen host spectrum
To assess the pollen host spectra of the 36 Hylaeus species, the crop content of a total of 1027 pinned females from museum and private collections captured between the middle of the 20 th century and 2022 was analysed by light microscopy. For each species, 30 pollen-containing crops were dissected originating from females collected at 30 different localities within the study area, which encompassed Switzerland, Baden-Württemberg (Germany) as well as Vorarlberg and Tirol (Austria). Localities were defined as different if the data on the collection labels differed with respect to collection site and/or collection date. For the four rare species Hylaeus crassanus, H. glacialis, H. incongruus and H. moricei, the targeted number of 30 different crop contents from 30 different localities was not attained and part of the females originated from outside the study area (see Table 1).

Categories of pollen host range
To characterise the degree of host plant association, such as "narrow oligolecty", "broad oligolecty", "polylecty with strong preference" or "polylecty", definitions proposed by Müller and Kuhlmann (2008) were followed. Two contrasting approaches were applied to infer oligolecty for a given species. The first approach averaged pollen host use across all individuals: a species was classified as oligolectic if 95% or more of the pollen grain volume belonged to the same plant family or genus. The second approach relied on the incidence of pure and mixed pollen loads: a species was classified as oligolectic if 90% or more of the females collected pure loads of one plant family or genus. In the present study, the two approaches differed only for one species, i.e. Hylaeus taeniolatus, which was classified as polylectic with strong preference for Apiaceae by the first approach and broadly oligolectic on Apiaceae by the second approach. As all crops except for one contained pollen of Apiaceae and most related species of the subgenus Paraprosopis proved to be Apiaceae specialists, H. taeniolatus was categorised as broadly oligolectic on Apiaceae.
Comparison between pollen host spectrum and flower visiting records To clarify possible differences between pollen and nectar host use in the Central European Hylaeus species, the pollen host spectrum as assessed in the present study was compared with the flower records of females contained in the database of the Wildbienen-Kataster Baden-Württemberg. At the time of data retrieval in September 2021, the database comprised 3175 female flower records from 29 Central European Hylaeus species without differentiation between pollen and/or nectar uptake. These

Pollen host spectrum at bee species level
Among the 36 Central European Hylaeus species, 19 (53%) exhibited an exclusive or strong preference for pollen from a single plant taxon (

Pollen host spectrum at bee genus level
The 36 Central European Hylaeus species collected pollen from the flowers of 31 plant families (Table 1, Fig. 5). However, only a few families were represented in high percentages in the pollen host spectrum of the genus as a whole. When summing the percentages of the plant families found in the host plant spectrum of each species across all species, the Apiaceae contributed 39.6% to the pollen host spectrum, followed by the Rosaceae with 18.7%, the Resedaceae with 7.0%, the Fabaceae with 6.1%, the Asteraceae with 3.7%, the Amaryllidaceae with 3.5% and the Campanulaceae with 3.4% (Figs 1, 5). These seven plant families accounted for more than 80% of the plants that the Central European Hylaeus species exploited for pollen, whereas the other 24 families were all represented by less than 3.0% in the genus' pollen host spectrum.
Pollen of Apiaceae was collected by all Central European Hylaeus species except for Hylaeus difformis and three oligolectic species specialised on Asteraceae or Resedaceae (Table 1, Figs 2-4). Based on field observations and the strongly differing morphology and size of the Apiaceae pollen grains recorded in the crop contents, all species including the oligolectic ones exploited several different genera among the Apiaceae. In contrast, almost 95% of all pollen of Rosaceae originated from the two genera Potentilla and Rubus (Fig. 5), pollen of Resedaceae and Amaryllidaceae exclusively came from the genera Reseda and Allium, respectively, over 99% of all pollen of Fabaceae was from the genus Melilotus, among the Asteraceae solely the two subfamilies Asteroideae and Carduoideae served as hosts and among the Campanulaceae only the two genera Campanula and Jasione were exploited.
About 89% of the pollen collected by the 36 Hylaeus species originated from herbs. Pollen of shrubs, such as Clematis, Frangula, Hedera, Rosa, Rubus, Sambucus and Vitis, was represented by slightly more than 10% with Rubus alone accounting for 9.6%. Pollen of trees, such as Castanea and Tilia, contributed only 0.2% to the host plant spectrum of the genus, while 0.4% of the pollen could not be attributed to one of the three vegetation layers.
About 93% of the pollen collected by the 36 Hylaeus species originated from flowers with easily accessible nectar, which is either exposed or secreted at the base of flowers that can be reached by the short-tongued Hylaeus bees thanks to their small body size. The remaining pollen came from flowers that either do not produce nectar or whose nectaries are not accessible due to their position at the base of narrow flower tubes. Pollen of nectarless flowers, such as Agrimonia, Aruncus, Filipendula, Hypericum, Plantago, Sambucus, Sanguisorba and Rosa, accounted for 2.3% of the flowers exploited for pollen. Pollen of flowers with inaccessible nectar, such as Carduoideae (Asteraceae) and Trifolium, was represented by 2.5% in the host plant spectrum of the genus, while 1.7% of the pollen could not be attributed to one of the three classes of nectar availability.

Comparison between pollen host spectrum and flower visiting records
The high importance of Apiaceae as host plants for the Central European Hylaeus species was also evident from the flower visiting records of 29 species from Baden-Württemberg. Out of 3175 flower visiting females observed, 1258 (39.6%) were recorded on Apiaceae, which is similar to the percentage of Apiaceae pollen in the crop contents of the same 29 species amounting to 42.4%. In striking contrast, with 838 (26.4%) flower visiting records the Asteraceae were the second most important plant family after the Apiaceae, whereas the percentage of Asteraceae pollen in the crop contents was only 4.3% across all 29 species. By excluding the Asteraceae specialist Hylaeus nigritus, this discrepancy   was even more pronounced with the percentage of female flower visits to the Asteraceae being 20.5% and the percentage of Asteraceae pollen in the crops being 1.1%. Although Asteraceae pollen was found in the crops of 22 out of the 26 polylectic Hylaeus species, its proportion was usually very small and ranged from 0.1-4.3% (mean 1.2%); the only exception was H. dilatatus, whose host plant spectrum included 13.7% Asteraceae pollen.

Discussion
The results of the present study show that the proportion of Central European Hylaeus species exhibiting an exclusive or strong preference for pollen from a single plant taxon is much higher than hitherto assumed and that the current assumption of the genus Hylaeus to largely consist of pollen generalists is wrong. Nineteen of the 36  Central European species examined are strictly or largely dependent on a single plant taxon for collecting pollen. For eleven of these species, flowers of the Apiaceae are the exclusive or strongly preferred hosts. The high significance of this plant family is also substantiated by the finding that the Apiaceae serve as pollen hosts for all Central European Hylaeus species with the exception of one polylectic species and three oligolectic species specialised on plant taxa other than the Apiaceae.
Phylogenetic inference is a powerful tool to reconstruct the evolution of pollen host preferences in bees (Müller 1996;Larkin et al. 2008;Sedivy et al. 2008Sedivy et al. , 2013Haider et al. 2014). To date, no phylogeny of the genus Hylaeus including its Central European representatives is available, rendering any hypotheses on the evolution of pollen host use in this group of bees premature. Nevertheless, the results of the present study allow for some preliminary insights. First, species that show an exclusive or strong preference for Apiaceae occur in six out of the ten Central European subgenera; this finding suggests that the preference for Apiaceae might be an ancestral trait in the Palaearctic Hylaeus fauna or, alternatively, has independently evolved several times in the evolutionary history of the genus. Second, most species of the subgenus Paraprosopis are Apiaceae oligoleges, which suggests that the ancestor of the subgenus was specialised on Apiaceae. Third, the two morphologically similar and probably very closely related species Hylaeus pfankuchi and H. rinki (both belong to the subgenus Lamdopsis) have an almost identical pollen host spectrum with roughly 95% of the collected pollen originating from Potentilla, Rosa (both Rosaceae) and Apiaceae. Considering that these two Hylaeus species distinctly differ in their habitat choice with the former mainly occurring in wetlands with reed beds and the latter in forest clearings and along forest edges (Westrich 2018), the largely concordant pollen host choice likely has a genetic basis. The same might apply to two closely related species of the subgenus Dentigera, i.e. H. gredleri and H. kahri, which exhibit a strong preference for Apiaceae, as well as to the three species of the Hylaeus gibbus group, i.e. H. confusus, H. gibbus, H. incongruus, whose pollen host spectra are all dominated by Rosaceae and additionally include Apiaceae, Campanulaceae, Cistaceae, Hypericaceae and partly Fabaceae.
Flowers of 31 plant families serve as pollen hosts for the Central European Hylaeus species. With 33 families, the number of plant taxa exploited for pollen is similar in the western Palaearctic species of the related genus Colletes (Colletidae), and nearly 70% of the plant families used by the Hylaeus bees as pollen sources are also exploited by the Colletes bees (Müller and Kuhlmann 2008). Furthermore, there is no plant family in the pollen host spectrum of the genus Hylaeus, whose pollen is not collected by other short-tongued Central European bees, such as species of Andrena or Lasioglossum (Westrich 2018). Thus, the peculiar habit of Hylaeus bees to ingest the pollen directly on the flowers and to transport it back to the nest inside the crop does not translate into a pollen host spectrum different from other bee taxa.
The finding that 89% of the pollen collected by the Central European Hylaeus species originated from herbs and a further 9.6% from Rubus, which usually grows as a prostrate shrub, suggests that Hylaeus females restrict pollen harvesting mainly to the herbal layer. However, this finding might be biased since the females dissected for the present study were all netted by hand, which possibly resulted in an underrepresentation of specimens harvesting pollen in the shrub or tree layer. In fact, part of the pollen diet of Hylaeus communis in five European cities originated from trees (Casanelles-Abella et al. 2022).
About 93% of the plant taxa used by the Central European Hylaeus species as pollen hosts can also be exploited for nectar due to the easy access to the nectaries. In contrast, approximately 5% of the pollen hosts lack nectar or secrete nectar that is inaccessible to the Hylaeus bees. To compensate for this lack or inaccessibility of nectar, the females must visit other flowers to gain enough nectar for provisioning their brood cells, as is probably exemplified by the Asteraceae specialist Hylaeus nigritus and the pollen generalist H. dilatatus, for which flowers of Carduoideae (Asteraceae) are important pollen hosts. Although neither species is able to reach the nectaries at the base of the long-tubed Carduoideae flowers with their short proboscis, pollen of Carduoideae contributed 77.4% and 11.3% to the host plant spectra of H. nigritus and H. dilatatus, respectively. Interestingly, 25 out of 30 crop contents of H. nigritus contained a mixture of pollen from Carduoideae and Asteroideae, whereas only two contained solely Carduoideae pollen. Similarly, pollen of Carduoideae was recorded in 7 out of 30 crops in H. dilatatus but never constituted the only pollen type. This finding is likely explained by the necessity to combine mere pollen visits to the Carduoideae with visits to the Asteroideae or other plant taxa to obtain nectar.
The comparison between pollen host spectrum and flower visiting records revealed a striking discrepancy in the use of Asteraceae as host plants by the Central European Hylaeus species. After exclusion of the Asteraceae specialist Hylaeus nigritus, the percentage of Asteraceae pollen in the crop contents averaged only 1.1%, whereas more than 20% of all flower visiting females were observed on this plant family. The most likely explanation for this discrepancy is that the flowers of Asteraceae serve as nectar sources, but not or only marginally as pollen sources. This pattern of use of Asteraceae pollen by the Hylaeus bees supports recent findings that the pollen of this plant family possesses unfavourable or protective properties, which render its digestion difficult and necessitate physiological adaptations to successfully utilize it, resulting in a reduced ability to use alternative hosts (Müller and Kuhlmann 2008;Praz et al. 2008;Wood and Roberts 2018;Vanderplanck et al. 2020). This scenario -known as the Asteraceae paradox -neatly applies to the polylectic Hylaeus species, which hardly exploit Asteraceae for pollen, and to Hylaeus nigritus, which is specialised on Asteraceae. However, it does not apply to H. dilatatus, which is the only polylectic Hylaeus species in Central Europe that collects pollen to a considerable degree on Asteraceae. This species might have inherited the ability to successfully utilise Asteraceae pollen from an ancestor specialised on this plant family, as is possibly the case in the Colletes succinctus group, which comprises both Asteraceae oligoleges and polyleges that partly exploit Asteraceae for pollen (Müller and Kuhlmann 2008).
Bee diversity and abundance have considerably declined in large parts of Europe during the last decades (Nieto et al. 2014;Powney et al. 2019). Species of the genus Hylaeus are no exception: thirteen of the 40 Hylaeus species recorded for Switzerland and six of the 31 species occurring in Baden-Württemberg are red-listed (Westrich et al. 2000;Müller and Praz in prep.). The results of the present study enable the targeted improvement of the food supply for these species at risk. Moreover, given the high importance of Apiaceae, Rosaceae, Resedaceae and Fabaceae as pollen hosts, the promotion of summer flowering Apiaceae (particularly Daucus), of Potentilla and Rubus (both Rosaceae), of Reseda (Resedaceae) and of Melilotus (Fabaceae), for example by including them into wildflower seed mixtures for pollinators, benefits a large part of the Central European Hylaeus species.

Conclusions
Although species of the genus Hylaeus differ from most other bees by their unusual habit to ingest the pollen directly on the flowers and to transport it internally back to the nest, their patterns of pollen host use are comparable to those of numerous other Palaearctic bee taxa in that i) the genus comprises species that cover the whole spectrum of host plant associations ranging from narrow oligolecty to broad polylecty, ii) a similar set of pollen hosts is used as in many other short-tongued bees, such as Andrena, Colletes or Lasioglossum, and iii) Asteraceae are hardly exploited for pollen by the polylectic species.