Introduction

Allium L. is one of the largest monocotyledonous genera, composed of approximately 900 species (Ekşi et al. 2016). It is widely distributed in the Holarctic region and occurs from the dry subtropics to the boreal zone. The single exception is A. dregeanum, which is native to South Africa (Li et al. 2010). Allium has its primary center of diversity from the Mediterranean Basin to central Asia and Pakistan (Li et al. 2010). Allium species are well known for their steroidal saponins, polyphenolic compounds, thiosulfinates, and other sulfur compounds (Lanzotti et al. 2014; Sobolewska et al. 2016), several of which valued for their pharmacological properties (Sobolewska et al. 2016). Several species are cultivated for their high economic importance, including edible garlic, leek, onion, shallot, bunching onion, and chives. Other species are used as ornamentals, for traditional medicine, and religious connotations (Kamenetsky and Rabinowitch 2006). Most Allium species are perennial herbs characterized by tunicated bulbs, narrow basal leaves, umbellate or head-like inflorescences, and flowers with six free or almost free tepals (Li et al. 2010).

Allium flowers are self-compatible, and although they are protandrous, geitonogamy is possible if within an inflorescence female- and male-stage flowers occur at the same time (Currah and Ockendon 1978; Devi et al. 2015). Thus, both self- and cross-pollination can occur (Shemesh et al. 2008; Devi et al. 2015). The anthers produce sticky pollen grains making insect pollination much more important than wind pollination (Shemesh et al. 2008; Devi et al. 2015). However, since most studies were restricted to species of economic value such as A. cepa (Devi et al. 2015), and rarely on others (e.g., A. oleraceum by Åström and Hæggström 2004), knowledge on the pollination biology of Allium is far from being complete. The few studies available suggest that the species are generalist and pollinated by a wide variety of insects, including bees, flies, butterflies, wasps and moths (Shemesh et al. 2008; Devi et al. 2015).

Allium species display a wide spectrum of floral colors, and Zuraw et al. (2009) suggested that their floral color is important in pollinator attraction. In addition to these visual cues, Allium flowers might also use olfactory (scent) cues to communicate with their pollinators, as it is well known for other animal-pollinated plants (Chittka and Raine 2006; Raguso 2008; Burger et al. 2010). So far, floral volatiles in Allium have only been studied in A. cepa and its hybrids (Soto et al. 2015), and in A. ursinum (Brunke et al. 1993 in Knudsen et al. 2006), whereas most of the species within the genus remain uninvestigated.

In the present study, we identified floral volatiles and flower visitors of five Mediterranean Allium species (A. lehmannii Lojac., A. neapolitanum Cirillo, A. nigrum L., A. obtusiflorum DC. and A. roseum L.), which differed in their flower and inflorescence morphology (Fig. 1), inflorescence size (diameter from 3 to 9 cm), and in flower colors (from white to pink). We also tested the physiological activity of selected floral scent components in antennae of two flower visitors [Apis mellifera (Hymenoptera, Apidae) and Eristalis tenax (Diptera, Syrphidae)] and two congeneric halictid bee species (Lasioglossum calceatum and L. leucozonium). In detail, we (1) collected the inflorescences scent by dynamic headspace, and analyzed the samples by thermal desorption (TD) gas chromatography and mass spectrometry (GC/MS), (2) recorded insect-visiting flowers/inflorescences and thus potentially acting as pollen vectors, and (3) determined which floral compounds were electrophysiologically active for some representative inflorescence-visiting insects and for congeneric species of Allium visitor insects, by using synthetic mixtures of available volatile compounds in gas chromatographic and electroantennographic detections (GC/EAD).

Fig. 1
figure 1

Flower visitors of study species. Lasioglossum transitorium planulum flying toward a flower of Allium lehmannii (a); L. malachurum visiting A. neapolitanum flowers (b); Eristalis tenax on A. nigrum flowers (c); an undetermined bee visiting an A. obtusiflorum flower (d); L. malachurum on a flower of A. roseum (e)

Full size image

Given that flower and inflorescence morphology, size, and colors differ among the species, we predict that their visitors/pollinators differ among the species. As different insect taxa might have different olfactory (scent) preferences and respond to a different set of compounds, we expect differences in scent among the species as shown in other genera (e.g., Narcissus by Dobson et al. 1997; and Mimulus by Byers et al. 2014). Overall, we discuss whether the interspecific variation in floral scents is linked to insects visiting the inflorescences of the different Allium species.

Materials and methods

Plant species

Allium species investigated in this work (Table 1) are Mediterranean spring-flowering species present in the Italian flora.

Table 1 Locality with geographic coordinates of studied Allium species
Full size table

Allium lehmannii (subg. Allium, sect. Codonoprasum, 2n = 16) is a species from Sicily and Calabria (Garbari and Raimondo 1987). It has 4–5 leaves up to 20 cm in length, and it blooms from late spring to early summer (June–July). Tepaloid flowers, grouped in inflorescences of 3.5 cm in diameter, are bell-shaped, whitish-pink with a green–red mid-vein. They are supported by flexuous scapes up to 38 cm tall. This species thrives in calcareous substrates, arid rocky cliffs and hillsides from 200 to 800 m a.s.l. Due to insufficient information available on its distribution, population size, trend and potential threats, A. lehmannii is assessed as Data Deficient in the IUCN red list (Donnini and Branca 2011). In Calabria, it is considered as Data Deficient, whereas in Sicily as Low Risk (Donnini and Branca 2011).

Allium neapolitanum (subg. Amerallium, sect. Molium, 2n = 28) is a species naturally occurring in the Mediterranean Basin and in Northern Africa. It is cultivated for its ornamental value and naturalized all over the world (Pignatti 1982). This species has 2–4 fleshy leaves up to 20–35 cm in length. Its inflorescence (hemispherical umbel) of 8–9 cm in diameter is visible from May to June. It has numerous fragrant and star-shaped white flowers on scapes up to 40 cm tall. In nature, A. neapolitanum grows in meadows, parks, roadsides and vineyards, always in moist shady soils from 0 to 800 m a.s.l.

Allium nigrum (subg. Melanocrommyum, sect. Melanocrommyum, 2n = 16) is a species native to the Eastern Mediterranean region, known for its potential medicinal properties, and widely cultivated as ornamental in Europe for a long time (Fritsch et al. 2010). In Sicily, this species is commonly found in herbaceous beds, vineyards, orchards and olive fields, from 0 to 1000 m a. s. l. Allium nigrum has linear-lanceolate leaves of 20–50 cm in length. The inflorescence, visible from April to May, is a hemispherical umbel of 7–9 cm in diameter, containing numerous star-like flowers (pink or sometimes white) on robust cylindrical scapes of about 80–120 cm, often with bulbils along the margin. Distinctive characters of this species are the ovaries which are blackish-green at anthesis.

Allium obtusiflorum (subg. Polyprason, sect. Scorodon, 2n = 16) is a Sicilian endemic species spreading from the northern coast around Palermo to the Iblean region (Brullo et al. 1994). This is a dwarf species with filiform leaves of 8–13 cm in length. Allium obtusiflorum blooms from late spring to early summer with solitary flexuous scapes, up to 7–18 cm tall. The inflorescence is an ovate-globose umbel containing whitish-pink, 3.5–5 mm long tepaloid flowers, with a dark purplish mid-vein. In its habitat, this species can be found on maritime sands, rocky fields, cliffs and hillsides. This species is classified as Data Deficient in the IUCN red list (Kell 2011).

Allium roseum (subg. Amerallium, sect. Molium, 2n = 16–32) has the broadest natural range, spanning from North Africa to Turkey. In the whole Mediterranean area, A. roseum is known for its nutraceutical properties. It is traditionally used as a cooking ingredient (Zammouri et al. 2009). Moreover, due to its ornamental value, it is also cultivated in many European regions outside its indigenous distribution. In some of these regions, it is naturalized and is considered as a spontaneous weed. Allium roseum is a variable species with 2–4 linear, fleshy leaves, up to 35 cm in length. Its cylindrical scapes of up to 50 cm length appear from late spring to early summer. Each scape supports an inflorescence (hemispherical umbel), up to 7 cm in diameter, containing numerous pink or sometimes white flowers of variable size, with or without bulbils, depending on (wild) variety (Zammouri et al. 2009). For the present study, we used a variety with bulbils and pink flowers. In nature, it grows in well-drained soil from 0 to 700 m a.s.l., and in Sicily, it is typically found on uncultivated fields as well as on roadsides near urban centers.

Floral scent collections

Floral scents of the five Allium species were collected on sunny days (11:00–13:00) in April (A. neapolitanum, A. roseum and A. nigrum) and in June (A. obtusiflorum and A. lehmannii) 2016, from plants growing in their natural habitats as specified in Table 1. For each species, eight different plants in full bloom were sampled by dynamic headspace (Zito et al. 2018). Therefore, one inflorescence per plant (5–38 flowers, depending on species) was enclosed in an oven bag (CUKI® Cofresco S.p.A.; 6 × 7 cm in A. lehmannii; 15 × 10 cm in A. neapolitanum; 18 × 15 cm in A. nigrum; 4 × 6 cm in A. obtusiflorum; 12 × 10 cm in A. roseum) and the emitted scent was trapped in a thermal desorption tube (TD-tube) using a vacuum pump (G12/01 EB, Rietschle Thomas, Germany) with a flow rate of 200 ml/min. TD-tubes were filled with 1.5 mg of Carbotrap B and 1.5 mg of Tenax-TA (both Supelco).

Each inflorescence of quite strongly scented A. neapolitanum, A. nigrum, and A. roseum was bagged for 3 min, and the emitted scent trapped for 3 min. Allium lehmannii and A. obtusiflorum emitted only a very weak scents as detected by our nose, thus, bagging time was increased to 10 min followed by 4 min of trapping time. Scent of leaf and of empty oven bags (one sample per species) was collected as negative controls (blanks) using the methods detailed above. After sampling, the TD-tubes were stored at − 20 °C until their chemical analysis.

Vouchers are deposited at the herbarium of the Department of Biological, Chemical and Pharmaceutical Sciences and Technologies of the University of Palermo.

Chemical analysis

VOCs trapped in the TD-tubes were analyzed using an automatic thermal desorption system (TD-20, Shimadzu, Japan) coupled to a Shimadzu GC/MS QP2010 Ultra equipped with a ZB-5 fused silica column (5% phenyl polysiloxane; length 60 m, i.d. 0.25 mm, film thickness 0.25 μm, Phenomenex). The samples were run with a split ratio of 1:1 and a helium carrier gas flow of 1.5 mL/min. The GC oven temperature started at 40 °C, then increased by 6 °C/min to 250 °C and finally was held for 1 min. The MS interface worked at 250 °C. Mass spectra were taken at 70 eV (EI mode) from m/z 30 to 350. GC/MS data were processed using the GCMSolution package, Version 2.72 (Shimadzu Corporation 2012). Components were tentatively identified by using both, the mass spectral libraries ADAMS (2007), FFNSC 2, W9N11, and ESSENTIAL OILS (available in MassFinder 3), and the Kovats retention indices of the compounds (based on n-alkane series). We only considered compounds which had a calculated Kovats index ± 10 compared to various databases (Adams 2007; El-Sayed 2018; Nist11). Twenty-two of the 31 identified floral scent compounds were confirmed by comparison of mass spectra and retention times with standard components available in the reference collection of the Plant Ecology Lab of the University of Salzburg. For quantitative analysis of VOCs, 100 ng each of ca. 150 components, among them monoterpenes, aliphatic, and aromatic compounds, was injected into the GC–MS system. The mean of the peak areas (total ion current) of these compounds was used to estimate the total amount of scent available in the scent samples (see Marotz-Clausen et al. 2018).

Inflorescence-visiting insects

Insects visiting the inflorescences of the five Allium species were observed and captured during the day of the headspace collections. The sampling of inflorescence-visiting insects was restricted to five different plants per species (35–173 flowers, depending on species and individual selected) in order to minimize potential negative impacts on the local entomofauna and the reproductive success of the plants, some of which included in the IUCN red list (e.g., A. lehmannii and A. obtusiflorum). Observations and captures were performed simultaneously by two researchers staying near the plants for 2 h (from 11:00 to 13:00). Insect visitor observations were limited to the time of scent samplings to allow a direct correlation between scent and pollinator data (see below). In other plant species, it is known that plants change their scent throughout daytime, with, e.g., plants emitting other scent patterns at noon than afternoon, linked with a change in the visitor spectrum (Dötterl et al. 2012). Thus, if on the one hand correlations between scent and visitor data might be biased when recording them at different times, on the other hand, we cannot exclude that the pollinator spectrum recorded in the Allium species may be more complex than observed by our approach. Insects were collected only when they landed on a flower/inflorescence, using an entomological net or small plastic bags (10 × 7 cm). In addition, the insects visiting the inflorescences were photographed with a digital camera (CANON EOS 1000D). The specimens were identified, if possible, to species level. Andrenid bees were identified by using Amiet et al. (2010), apid bees by Amiet (1996), halictid bees by Ebmer (1969, 1970, 1971, 1974, 1987), megachilid bees by Amiet et al. (2001), Anthomyiid flies by Séguy (1923), bombyliid flies by Séguy (1926), calliphorid flies by Akbarzadeh et al. (2015), conopid flies Smith (1969), sarcophagid flies by Povolný and Verves (1997), scatophagid flies by Séguy, (1934), and syrphid flies by Speight and Sarthou (2014). Insects are stored in an entomological box at the Entomological Collection of the Department of Biological, Chemical and Pharmaceutical Sciences and Technologies of the University of Palermo.

Electroantennographic recording (GC-EAD)

For our electrophysiological experiments, we collected insects in the Botanical Garden of the University of Salzburg: Apis mellifera and Eristalis tenax were recorded as flower visitors of Allium (see Results), while Lasioglossum calceatum and L. leucozonium were used for preliminary analyses, as representatives of an inflorescence-visiting genus. This seems reasonable given that even species within families respond similar to floral scents (Jürgens et al. 2014; Braunschmid et al. 2017). Insects were tested on three synthetic mixtures prepared with commercially available compounds (Sigma-Aldrich®; overall 10−4 in acetone; v/v) (hereafter referred to as mix I, mix II and mix III). As some compounds had similar retention times, they were not used in a single mix, but in different mixtures, to better link antennal responses with specific components. Mix I and mix II were tested on all insect species mentioned above; mix III was only tested on A. mellifera as the other insects were not available when these measurements were performed. The three mixtures contained 22 out of the 31 VOCs identified in our scent analyses (see below). Mix I contained 3-methylbutanoic acid, hexanol, benzaldehyde, (Z)-β-ocimene, (E)-linalool oxide furanoid, allo-ocimene, 1,4-dimethoxybenzene, 1,2,4-trimethoxybenzene, 1,3,5-trimethoxybenzene, and indole. Mix II contained (Z)-3-hexenol, heptanol, benzyl alcohol, phenylacetaldehyde, methyl benzoate, 1,2-dimethoxybenzene, methyl salicylate, eugenol, and octyl butanoate. Mix III contained (E)-β-ocimene, 2-phenylethanol, and methyl nicotinate.

For measurements, six individuals of each species were used. An antenna (A. mellifera, L. calceatum/leucozonium) or the head (E. tenax) was cut off from the insect, mounted between two electrodes filled with insect Ringer’s solution (8.0 g/L NaCl, 0.4 g/L KCl, 0.4 g/L CaCl2), and connected to silver wires. Both antennae of a specific individual were used, one for mix I, the other for mix II. The mixture III was tested on other individuals. The reference electrode was placed in contact with the cutting surface of the antenna (bees) or with the caudal side of the head (syrphid flies), while the recording one was brought into contact with the tip of an antenna.

A compound was considered to be EAD-active when it elicited a depolarization response in at least four out of six replicates per each insect taxon tested.

The GC-EAD system consisted of a gas chromatograph (Agilent 7890A, Santa Clara, CA, USA) equipped with a flame ionization detector (FID) and an EAD setup (heated transfer line, 2-channel USB acquisition controller) provided by Syntech (Kirchzarten, Germany). A volume of 1 µl of each mixture was injected (temperature of injector 250 °C) splitless at 40 °C oven temperature, followed by heating the oven at a rate of 10 °C min−1 to 220 °C. The split vent was opened 0.5 min after the injection. A ZB-5 column (5% phenyl polysiloxane; 30 m long, inner diameter 0.32 mm, film thickness, 0.25 µm, Phenomenex) was used for the analyses, and the column flow (carrier gas: hydrogen) was set at 3 mL min−1. The column was split at the end into two deactivated capillaries by a µFlow splitter (Gerstel, Mülheim, Germany) leading to the FID (2 m × 0.15 µm) and EAD (1 m × 0.2 µm) setup. Makeup gas (N2) was introduced in the splitter at 25 mL min−1. The EAD outlet was placed in a cleaned, humidified airflow that was directed over the antennae.

Statistical analysis

To determine pairwise semiquantitative differences in scent (square root transformed percentage of relative amount of compounds/inflorescence/species) among samples, we calculated the Bray–Curtis similarity index in Primer 6.1.6 (Clarke and Gorley 2015). Based on obtained similarity matrix, we performed an analysis of similarity (10,000 permutations) to test for differences in the scents among species. Non-metric multidimensional scaling (NMDS; Primer 6.1.6), based on Bray–Curtis similarity matrix, was used to graphically display similarities in scent among samples/species.

To test for linkage between floral scent and visitor/pollinator taxa, we correlated a Bray–Curtis similarity matrix based on floral scent (square root transformed species means of relative amount per compound were used for calculation) to Bray–Curtis similarity matrices of visitors/pollinators (order, family and morphospecies level) using RELATE in Primer (Spearman’s rank correlation, 10,000 permutations). These analyses were calculated with all insect taxa found and additionally only with insect taxa that occurred more than once in the surveys. Cluster analysis (single linkage; Primer 6.1.6), based on Bray–Curtis similarity matrix, was used to graphically display similarities in visitors/pollinators at morphospecies level.

Results

Floral scent

Of the 36 VOCs detected in inflorescence scent samples of the five Allium species, 31 were identified (Table 2). The number of VOCs ranged from one in A. lehmannii and A. obtusiflorum to eighteen in A. nigrum. We did not find a single compound that occurred in all the species; instead, most of the compounds were species specific (Table 2). The only co-occurrences, with a single compound shared each, were found in A. obtusiflorum and A. roseum (benzyl alcohol), and in A. nigrum and A. roseum (methyl salicylate) (Table 2). The volatiles were from twelve chemical classes and functional groups (Table 2). The total absolute amount of scent trapped from an inflorescence ranged from 1.0 ng/min in A. obtusiflorum to 49.2 ng/min in A. nigrum. When this amount was calculated per flower, it ranged from 0.1 ng/min in A. obtusiflorum to 1.6 ng/min in A. neapolitanum. The scent profiles of the analyzed species were dominated either by one or two aromatic compounds (A. neapolitanum, A. nigrum, A. obtusiflorum and A. roseum) or by a monoterpene (A. lehmannii) (Table 2). The most abundant VOCs (> 20%) were: benzaldehyde in A. neapolitanum, 1,2- and 1,4-dimethoxybenzene in A. nigrum, methyl benzoate and phenylacetaldehyde in A. roseum, benzyl alcohol in A. obtusiflorum, and (E)-linalool oxide furanoid in A. lehmannii. Intraspecific variation in semiquantitative scent composition was very low, whereas interspecific variation was high resulting in strong differences in scent profiles among species (ANOSIM: R4,35 = 0.978, p < 0.001; Fig. 2), which each species differing from the others (p < 0.002).

Table 2 Relative (%, mean ± SE) and absolute amounts of floral volatiles in the five Allium species studied (N = 8 per species)
Full size table
Fig. 2
figure 2

Non-metric multidimensional scaling (NMDS) used to display a the semiquantitative differences in scent profiles among the scent samples collected from five Allium species. The most abundant VOCs found in the different Allium species are also given. Cluster analysis (single linkage) used to display b differences in Allium visitors/pollinators at morphospecies level

Full size image

Inflorescence-visiting insects

We recorded 122 insect specimens on the inflorescences of the five Allium species (Table 3) with bees and flies captured on all these species. Bees were the most abundant visitors in A. neapolitanum, A. obtusiflorum, and A. roseum inflorescences, while flies were the most abundant in A. lehmannii and A. nigrum. Additionally, a few butterflies were recorded on A. nigrum and A. roseum, while a few beetles were recorded on A. lehmannii, A. nigrum and A. roseum. In A. lehmannii, the largest number of insect visitors was Bombyliidae followed by Halictidae, while in A. neapolitanum and A. obtusiflorum, the largest number was Halictidae. In A. roseum, the most abundant insect visitors was Apidae followed by Halictidae. In A. nigrum, Syrphidae and Scatophagidae flies were the most abundant insects captured. Bees, flies, and butterfly visitors were observed nectaring on the flowers, while the pollen grains were passively collected (Fig. 1 and Online Resource 1, 2, 3). At least some bees were also observed to actively collect pollen.

Table 3 Inflorescence-visiting insects of the five Allium species
Full size table

Overall only a few visitors/pollinators were shared among the studied Allium species (Table 3). Co-flowering and syntopic A. lehmannii and A. obtusiflorum (locality Barcarello) shared Lasioglossum transitorium planulum, Halictus smaragdulus, Anthrax sp., and an unidentified morphospecies of Oedemeridae, while A. neapolitanum and A. roseum (locality San Giovanni Gemini) shared an unidentified morphospecies of Anthomyiidae, though they did not have floral scents in common.

Floral scents and inflorescence-visiting insects

Overall floral scent patterns did not correlate with flower-visiting insects (RELATE: order level, Rho = − 0.49, df = 4, p = 0.91; family level, Rho = − 0.07, df = 4, p = 0.56; morphospecies level, Rho = − 0.046, df = 4, p = 0.60). Similar results were obtained when excluding taxa found with one individual only (RELATE: order level, not performed as we did not find orders with only one individual; family level, Rho = 0.026, df = 4, p = 0.50; morphospecies level, Rho = 0.037, df = 4, p = 0.54).

Electrophysiologically active compounds

The antennae of Apis mellifera, Lasioglossum calceatum, L. leucozonium, and Eristalis tenax responded overall to most of the 22 VOCs present in the three synthetic mixtures (Fig. 3, Table 4). Only 3-methylbutanoic acid did not elicit an antennal response in any insect. Apis mellifera was tested on all three mixtures and perceived 21 compounds. Lasioglossum species and E. tenax were only tested on two of these mixtures (mix I and mix II with 19 compounds in total) and responded to 16 and five thereof, respectively. Overall, the compounds in the three mixtures from the different Allium species elicited antennal responses in their specific/congeneric visitors and also in the visitors of other Allium species. Thus, antennal responses of the different insects did not explain differences in visitor spectra among the Allium species. For instance, (E)-linalool oxide furanoid and benzyl alcohol, abundant compounds in A. lehmannii, and A. obtusiflorum, respectively, elicited antennal responses in both Lasioglossum species and in A. mellifera; however, we only recorded Lasiglossum and not A. mellifera as visitors. In contrast, 1,2- and 1,4-dimethoxybenzene, abundant compounds in A. nigrum, and phenylacetaldehyde and methyl benzoate, abundant in A. roseum, elicited also antennal responses in both Lasioglossum species and in A. mellifera; however, we only recorded A. mellifera as visitors. The syrphid fly E. tenax was not captured/observed on A. roseum though it perceived methyl benzoate, a main compound emitted by its inflorescences.

Fig. 3
figure 3

Example of antennal responses (EAD) of Lasioglossum calceatum (mix I) and L. leucozonium (mix II) (red tracks), Apis mellifera (blue tracks), and Eristalis tenax (green tracks) to compounds (FID) of two synthetic mixtures (mix I and mix II). The compounds eliciting antennal responses (EAD-active) are marked with an asterisk (*). Numbers correspond to: (1) 3-methylbutanoic acid; (2) hexanol; (3) benzaldehyde; (4) (Z)-β-ocimene; (5) (E)-linalool oxide furanoid; (6) allo-ocimene; (7) 1,4-dimethoxybenzene; (8) indole; (9) 1,2,4-trimethoxybenzene; (10) 1,3,5-trimethoxybenzene; (11) (Z)-3-hexenol; (12) heptanol; (13) benzyl alcohol; (14) phenylacetaldehyde; (15) methyl benzoate; (16) 1,2-dimethoxybenzene; (17) methyl salicylate; (18) eugenol; (19) octyl butanoate

Full size image
Table 4 Inflorescence scent of the different Allium species used for electrophysiological measurements and their antennal responses (EAD-active) in the insects tested
Full size table

Discussion

The investigated Allium species emitted a low number of compounds, mainly aromatics and monoterpenes, and most of these compounds elicited antennal responses in the flower visitors of these species, mainly honey bees, sweat bees (Lasioglossum spp.), and syrphid flies. Most of the VOCs found in the present study are widespread among floral scents (Knudsen et al. 2006). However, the scent blends of studied species differ from those described from other Allium species. For instance, we found only a few similarities between the Allium species of this study and A. ursinum since only (Z)-3-hexenol, hexanol and benzaldehyde co-occurred (Table 2 and Brunke et al. 1993 in Knudsen et al. 2006). We did not find any similarities between the five Allium species and A. cepa that mainly emits sulfur compounds from their flowers (Sotoet al. 2015). Overall, the floral scent data available for Allium suggest that in this genus scents are species-specific with very limited overlap in scent pattern among species (Fig. 2a). Furthermore, our results show that scent patterns did not correlate with flower-visiting insects (Fig. 2b), similarly to some deceptive Ceropegia flowers (Heiduk et al. 2017).

Generally, a plant pollinated by one order is functionally specialized but may or may not be ecologically specialized (e.g., it may be pollinated by one or many species of that order) (Waser and Ollerton 2006). In the present study, Allium species show to be functionally specialized (mainly bees and flies as visitors/pollinators) but not ecologically specialized (several bee and fly species). Data on the functional specialization, however, need to be taken with care given the limited sampling time. A further sampling will show whether plants are functionally generalized or indeed as specialized as observed by our sampling approach. As different species flowered at different locations and parts of the year, it also would be interesting to test how flowering time and location influenced the observed visitor patterns.

In contrast to the plants, most insects recorded on the inflorescences seem to be generalist flower visitors, which also is true for the bees (Westrich 1989; Scheuchl and Willner 2016). Among the bees, there are several social (e.g., Apis mellifera, H. smaragdulus, L. malachurum) and solitary (A. flavipes, N. facilis) species known to occur in high abundances in southern Europe (Ebmer, unpublished data). Thus, their occurrence as flower visitors of Allium may especially reflect their abundance in Sicily and less their specific preference for Allium. However, differences in their olfactory/visual preferences might still explain the variable visitor spectra of the different Allium species (see also below). The recording of Chelostoma distinctum as flower visitor of Allium was unexpected, given that this bee is specialized on Campanula species and rarely observed on other plants (Westrich 1989). Despite described from Sicily (Dathe et al. 2016; Scheuchl and Willner 2016), we did not record Hylaeus punctulatissimus Smith and Hylaeus bifasciatus (Jurine), both specialized on Allium pollen (Dathe et al. 2016; Scheuchl and Willner 2016), as flower visitors. The third European species specialized on Allium, i.e., Colletes graeffei Alfken, does not occur on Sicily (Ebmer, unpublished data).

Our electroantennographic investigations revealed that both bee and syrphid fly visitors of Allium detect most of the VOCs released by these plants, among them the most abundant ones (Tables 2, 4). The measurements with sweat bees were performed with species differing from the ones visiting the inflorescences. However, given that bees, even independent of family membership, respond similarly to (common) floral scents (present study; Jürgens et al. 2014; Braunschmid et al. 2017), we believe that our results are a good approximation for sensory capabilities of the flower-visiting species. Several of the VOCs identified as being physiologically active in present work were previously known to be electrophysiologically active in insect families, genera or species here investigated. For instance, indole, methyl salicylate, benzaldehyde, phenylacetaldehyde, 1,4-dimethoxybenzene, 1,2,4-trimethoxybenzene, 2-phenylethanol, (E)-β-ocimene, and methyl nicotinate are EAD-active in A. mellifera (Füssel et al. 2007; Salzmann et al. 2007; Kobayashi et al. 2012; Jürgens et al. 2014; Mas et al. 2018); benzyl alcohol, 2-phenylethanol, methyl salicylate, (E)-linalool oxide furanoid, and eugenol in Lasioglossum spp. (Urru et al. 2010; Braunschmid et al. 2017), and 1-heptanol, 2-phenylethanol and (E)-linalool oxide furanoid in syrphid flies (Braunschmid et al. 2017). However, before our study several other compounds were not known to be physiologically active in honey bees, sweat bees and syrphid flies, as highlighted in Table 4. For example, 1,2-dimethoxybenzene, benzyl alcohol, and methyl benzoate were not known for honey bees; 1,2-dimethoxybenzene, 1,4-dimethoxybenzene, phenylacetaldehyde and methyl benzoate were not known for sweat bees, and 1,2-dimethoxybenzene and methyl benzoate were not known for syrphid files (Table 4 for details of all VOCs eliciting antennal responses). Furthermore, some EAD-active compounds identified in Allium inflorescences were previously reported as being behaviorally active in families, genera or species of the insect-visiting Allium flowers. Among these, (Z)-3-hexenol, methyl salicylate, 1,4-dimethoxybenzene, benzyl alcohol, 2-phenylethanol and phenylacetaldehyde are attractants for A. mellifera (Laloi et al. 2000; Dötterl and Vereecken 2010; Jürgens et al. 2014); phenylacetaldehyde is an attractant for Halictidae bees (Dötterl and Vereecken 2010); (Z)-3-hexenol and methyl salicylate are attractants for Syrphidae flies (Zhu and Park 2005; James 2005). Thus, when comparing our results with the literature (see above), we expect that several Allium floral VOCs (e.g., 1,4-dimethoxybenzene, phenylacetaldehyde and (Z)-3-hexenol) play an important role in the attraction of their floral visitors/pollinators. However, behavioral experiments with EAD-active synthetic scent compounds are necessary to prove that these compounds are indeed involved in pollinator attraction in Allium spp.

Generally, floral VOCs are either specific attractants of specialist insect species or common attractants for a large array of generalist ones (Farré-Armengol et al. 2015). By the emission of mostly common VOCs (e.g., benzyl alcohol, benzaldehyde, phenylacetaldehyde, (E)-linalool oxide furanoid, 1,2- and 1,4-dimethoxybenzene), the different Allium species could be able to attract the same or at least related generalist insect visitors to their inflorescences, despite the species-specific scents. This finding is in accordance with previous studies, which reported that plant species with different VOC patterns might attract the same/related generalist pollinators (Filella et al. 2013).

We also found that specific insects, such as honey bees, perceived several compounds (e.g., benzyl alcohol, benzaldehyde, (E)-linalool oxide furanoid and indole) of Allium species, which they did not visit (see Results). This might have to do with the availability of the insect in the habitat of the plant and/or a mismatch between other floral traits, such as morphology and color, and the morphology or preferences of the insects (Burger et al. 2010; Milet-Pinheiro et al. 2012).

Co-flowering and syntopic A. lehmannii and A. obtusiflorum shared some of the visitors/pollinators, for which they might compete. However, even if they share pollinators, they might still be reproductively isolated due to flower constancy (Chittka and Raine 2006) of single pollinator individuals.

Our study investigated the inflorescence scents and insect visitors in Allium, for the first time in their natural habitats, and contributes to the understanding of the pollination biology and scent chemistry in this genus. Furthermore, several compounds have been added to the list of VOCs being EAD-active in bees and syrphid flies. Three of the studied species emitted scent patterns consisting of only 1–2 components, all widespread scents. These components might be generalist visitor attractants involved in the attraction of the 3–9 different insect taxa observed as flower visitors (see also above). Otherwise, such simple scent patterns are rare among plants and especially expected in more specialized systems (Ervik et al. 1999; Wiemer et al. 2009; Maia et al. 2012; Dellinger et al. 2019).