Flower morphology and nectar quality traits in faba bean affect attraction to bees

Legumes provide nutritious floral resources for bees thus are commonly considered for use in schemes to enhance pollinators in agroecosystems. In particular, legume landraces are important elements in practices towards sustainable agriculture. Differences in flower morphology and nectar may affect the attraction of pollinators. The aim of the present study was to investigate possible differences in pollinator attraction between a Vicia faba L. landrace and a commercial variety and to indicate differences in floral characteristics and nectar quality traits which can be associated with bee visits. To assess this, floral morphological trait measurements and chemical analysis using high‐resolution mass spectrometry‐based non‐targeted screening of the flower nectar were carried out and bee visits on faba bean flowers were recorded. Floral traits that contributed to the separation of the two accessions were ‘intensity of streaks’ and ‘extent of anthocyanin colouration’ on standard petal which were more pronounced in the landrace. All nectar traits examined contributed to the separation of the two V. faba accessions. Chemical substances such as amino acids and sugars which are important for bee nutrition are differentially increased in the landrace compared with the commercial variety. The landrace attracted more bee genera (Apis, Anthophora, Eucera) and bee visits compared with the commercial variety (only Eucera) supporting that the floral and nectar differences between the accessions of the same plant species are important to bees. The well‐fitting interrelation of the landrace flowers and bee visits can have possible implications in agricultural practices such as crop breeding for pollinator friendly varieties through selecting floral traits, improvement of foraging resources for pollinators in agroecosystems and subsequently food production. Exploitation of the V. faba landrace in agroecological schemes to support conservation of flora and fauna genetic resources in situ should be considered also for the benefit of future food stability and security.

Legumes are commonly considered when strategies to enhance pollinator biodiversity and ecosystem services in agroecosystems are designed (Barda et al., 2023;Brandmeier et al., 2023;Cole et al., 2022), in particular in seed mixtures for flower strips in orchards (Cole et al., 2022) as several plant species have been proven to attract insect pollinators by providing nutritious floral resources (Kleijn & Raemakers, 2008).Flowers of faba bean, Vicia faba L.
Furthermore, inclusion of faba bean in agroecological schemes is supported by its high adaptability to climate change (Merga et al., 2019) and many ecological benefits (Etemadi et al., 2019), including the legume contribution to soil nitrogen fixation (Kebede, 2021;Stagnari et al., 2017).
Floral nectar is one of the major rewards for pollinators and it is probably subjected to selection processes that lead to nectar variations across closely related plants pollinated by various animals (Baker & Baker, 1982;Pyke & Waser, 1981).Also, nectar rather than pollen appeared to be the main driver of flora choice by insects (Fowler et al., 2016).Moreover, floral traits influence the functional ecology of plant -pollinator interactions (Galetto & Bernardello, 2004;Ollerton et al., 2007;Randall et al., 2009;Reverté et al., 2016;Stang et al., 2009;Suso et al., 2016;Wilson et al., 2004); knowledge on these interactions can be utilized for the mutual benefit of pollinators and crops.
The chemicals in nectar that contribute to bees and other pollinators attraction vary but some common compounds include sugars, amino acids, flavonoids, terpenes, and alkaloids (Bertazzini et al., 2010;Nicolson, 2022;Nicolson & Thornburg, 2007;Roy et al., 2017;Seo et al., 2019).These molecules help bees to identify the flower as a potential food source and are essential for bee survival.Main sugars in nectar include glucose, fructose, and sucrose, with a plant species being consistent in composition (Nicolson & Thornburg, 2007).Other sugars are present in trace amounts, such as the monosaccharides mannose, arabinose, xylose, disaccharides, such as maltose, melibiose or the oligosaccharides raffinose, melezitose, and stachyose (Baker & Baker, 1982, 1983).The monosaccharide sorbitol is a frequent component of Mediterranean floral nectars (Petanidou, 2005).
All 20 amino acids found in proteins have been identified in various plant nectars and the essential ones may be an important nitrogen source for nectarivorous pollinators (Nicolson & Thornburg, 2007).A few of non-protein amino acids, including β-alanine, ornithine, homoserine, and γ-aminobutyric acid can be found in nectar (Baker & Baker, 1977;Kaczorowski et al., 2005), but whether they have any role in attraction of pollinators is not clear.Amino acids concentration in nectar also contribute to nectar taste (Birch & Kemp, 1989) and have the potential to modify insect behaviour by stimulating insect chemosensory receptors (Gardener & Gillman, 2002;Hansen et al., 1998;Shiraishi & Kuwabara, 1970).For instance, Petanidou et al. (2006) reported that phenylalanine, which is known to be a phagostimulant for bees (Inouye & Waller, 1984), was the most abundant amino acid in the nectars of 73 mainly bee-pollinated Mediterranean plant species.Moreover, Hendriksma et al. (2014) concluded that honeybees prefer essential over non-essential amino acids and that they were attracted by phenylalanine and deterred by glycine.
Other chemical compounds in nectar which are associated with insect interactions include phenolics, terpenoids, alkaloids, and lipids.Phenolic substances are quite widespread in nectars, consisting relatively common scent products of flowers (Baker & Baker, 1982;Deachathai et al., 2006;Knudsen et al., 1993) which might serve as attractants for pollinators (Pichersky & Gershenzon, 2002).Their accumulation may render the nectar toxic and therefore repellent to some visitors (Frankie et al., 1982;Hagler & Buchmann, 1993).Moreover, phenolics have fluorescent properties which might serve as a guide for pollinators to visit the flower, although some authors do not share this view (Kevan, 1976;Thorp et al., 1975).Especially flavonoids in nectar have been shown to attract bees and other pollinators (Dudek et al., 2016;Winkel-Shirley, 2002).
Terpenoids identified in flower nectar produce a pleasant scent attractive to pollinators (Andersson, 2003;Plepys et al., 2002;Raguso, 2004;Tholl et al., 2004).Terpenoids which serve as attractants to visiting pollinators are also produced by cells dispersed throughout the flower (Bergström et al., 1995) but they are cues of little importance if there is not an associated nectar reward (Bitterman et al., 1983).
Alkaloids have been detected in the nectar of many plants (Adler & Wink, 2001;Baker & Baker, 1975;Hazslinsky, 1956) and were generally thought to have a detrimental effect on pollinator visitation (Adler, 2000;Baker & Baker, 1982).However, studies have shown that bees are attracted by low concentrations of caffeine and nicotine (Singaravelan et al., 2005;Thomson et al., 2015).
Lipids have been reported in numerous plant nectars providing pollinators with a rich energy source (Baker & Baker, 1975).
Overall, investigations on the relation of flower morphology and nectar composition with pollinators are encouraged towards a more holistic approach of plant-pollinator interplay and possible implications in agricultural practices (Diekötter et al., 2010;FAO, n.d.;Hevia et al., 2021;Kaczorowski et al., 2005;Mottershead & Underwood, 2020;Sabatino et al., 2021;Waser & Ollerton, 2006), such as increased yields of V. faba due to bee pollination (Beyer et al., 2020b).In legumes, knowledge on plant-pollinator interactions is important for crop breeding and production as well as the growing urgency for halting pollinator decline through the implementation of conservation measures (Suso et al., 2016).The importance of maintaining the diversity and utility of V. faba genetic resources and the need for further knowledge on the interactions of faba bean with insect pollinators is highlighted (Duc et al., 2010;Mamedova & Vishnyakova, 2020).
Furthermore, on-farm conservation of V. faba genetic resources can promote participatory breeding and the production of pre-breeding pools adapted to local agronomic practices, for example, to preferable floral traits by pollinators which will potentially maximize crop production (Duc et al., 2010;Terzopoulos et al., 2003).In parallel, insect pollinators can benefit through the nutritious Fabaceae floral resources which can be essential for bee reproduction and larval development (Cole et al., 2022;Kleijn & Raemakers, 2008).Despite the major role of nectar and the importance of flower traits on the attraction of bees, that has captured the interest of a plethora of studies in various plants, in very few studies, an attempt has been made to investigate the possible role of V. faba flower and nectar traits in a detailed approach (Bailes et al., 2018;Bailes & Glover, 2018;Pierre et al., 1996), especially in field scale.
The aim of the present study was to investigate possible differences in pollinator attraction between a faba bean landrace ('Episkopis') of the region of Tegea, Peloponnese, southern Greece, and a faba bean commercial variety ('Histal'), of similar biological cycle, which are most commonly used in this area and indicate differences in their floral characteristics and nectar quality traits which are possibly associated with bee visits on their flowers.Due to genetic erosion 'Episkopis' is the only faba bean landrace cultivated in the area.Hence, we chose to study a landrace that still can be found on cultivation on farm and has been adapted to edaphoclimatic and ecosystem conditions of the region in parallel with the most famous commercial variety cultivated in the studied area.Efforts to exploit the only faba bean landrace of Tegea plateau named 'Episkopis,' as a food source for insect pollinators has been made in Arcadia, Greece by Barda et al. (2023), who established patches of annual flowering mixtures inside orchards of the protected designation of origin apple cultivar 'Pilafa'.Barda et al. (2023) have recorded several genera of pollinating bees on the faba bean landrace, some of which also visited the apple blossoms.Due to these findings, the authors would like to further investigate bee visits on this landrace, testing the hypothesis that it may attract higher bee visits than the commercial variety.If this hypothesis is true, then the utilization of the landrace in agroecological schemes in the studied area will be supported resulting in further benefits such as the conservation of plant genetic resources in situ.To assess this possible associations, floral morphological trait measurements, chemical analysis using high-resolution mass spectrometry (HRMS)-based non-targeted screening (NTS) of the flower nectar and bee visits recordings during flowering of the faba bean accessions were carried out.

| Experimental design
The study was carried out in a field located in Tegea plateau, Arcadia, Peloponnese, Greece (37.442424 N, 22.403983E)  The soil profile was the same in the two areas of the field (Figure 1).

| Plant material and establishment of faba bean
The faba bean landrace was 'Episkopis' from the village Episkopi Arcadias (37.464416Ν, 22.418946Ε) in the region of Tegea plateau in the prefecture of Arcadia in the Peloponnese, Greece and the commercial variety was 'Histal'.Both varieties have similar biological cycle, are utilized for human food and animal feed and are cultivated in Tegea plateau.Nevertheless, the landrace 'Episkopis' is the most widely cultivated faba bean in the area.Prior to sowing, the germination rate of Episkopis and Histal seeds was tested in the laboratory and was 100%.In field conditions, the germination rate was >98%.
A rotary tiller was used for field preparation prior to sowing of the faba beans.For the establishment of the accessions, broadcast seeding was performed by hand in autumn (November 2021) at a sowing density of 8 seeds/m 2 .Seeds were covered with shallow raking.No irrigation was required since the soil moisture was adequate for seed germination and seedling establishment.Also, no plant protection product or fertilizer was applied.Weeds around the experimental field were destroyed mechanically using a rotary (Brassicaceae), Veronica persica Poir.(Plantaginaceae).These annual weeds were present in the experimental plots of both accessions.In addition, adjacent to the experimental field were a cherry orchard and an apple orchard which provided flower resources for pollinators in early to mid-April (cherry blossom) and mid to late April (apple blossom).The orchards were in equal distance apart from the accessions.

| Bee visits
Attraction of the faba bean accessions to pollinators was assessed by recording the bee (Hymenoptera: Anthophila) visits on faba bean flowers by visual observation, for 1 min per plot (4 m 2 ) at four dates: 5, 13, 19, 21 April 2022, which provided a total of 96 observations.The observations were made by walking around the four sides of each square plot, standing equal time on each side, to be able to inspect visits on the bottom facing flowers of faba bean.Bee visits on faba bean flowers were recorded in each plot for 1 min.All counts (bee visits on faba bean flower/plot/1 min) were made between 10:00 and 14:00 h at temperatures in the range of 14-20°C and no more than half the sky was covered with clouds.Wind velocity was recorded during the measurements and did not exceed 1 m/s.Plant cover (% of plot area) was recorded.Faba bean flower cover (% with open flowers of plot area) for each accession was also recorded in each observation day.Bees that could not be identified in the field, were collected by sweeping net sampling, stored in the freezer (−18°C) and identified under a stereomicroscope (OLYMPUS SZ61, UK).

| Measurements of flower traits
Flowering stage traits of the two accessions were recorded based on the 'Faba bean Descriptors' (IBPGR, 1985) and the 'UPOV tests on distinctness' (2019).Thus, the quantitative traits recorded were: days to flowering (DF), width of standard petal (WS), number of flowers per inflorescence as the mean number of flowers per raceme from two intermediate nodes (NFIN), flower length (FL), and the flowering period (FP).Additionally, the following qualitative traits were recorded: ground colour of standard petal (GCSP), intensity of streaks on standard petal (ISSP), extent of anthocyanin colouration on standard petal (EACSP), wing petal colour (WPC), and colour of melanin spot of the wing (MSW).The flower traits examined were recorded on 40 'Episkopis' and 40 'Histal randomly selected open flowers of stage four to five, as described by Osborne et al. (1997), from the plots of each accession following a zig-zag route in the plantation area of each accession.The phenotypic classes or units for each of the above-mentioned traits are shown in Table 1.were removed from both 'Episkopis' and 'Histal' plants at the same time and date, to exclude potential variations in nectar production with time of day.The flower samples were collected randomly from the plots of each accession following a zig-zag route in the plantation area of each accession.After collection, the flowers were transferred to the laboratory, where the petals and anthers of each flower were removed.The reproductive complex of the flowers was sealed in sets of 4-5 flowers in 0.5 mL microcentrifuge tubes bearing five micro holes at their base (7 replications).These tubes were then placed within 1.5 mL tubes which were centrifuged at 13,000 rpm for 4 minutes, as described by Bailes et al. (2018) and adjusted to meet the aims of the present study.Nectar from 29 'Episkopis' flowers and 29 'Histal' flowers was extracted, in total.

| Ultra-High-Performance Liquid Chromatography (UHPLC) -coupled to orbitrap High-Resolution Mass Spectrometry (HRMS) analysis of flower nectar
UHPLC was performed using a Dionex Ultimate 3000 UHPLC system (Thermo Scientific) for the analysis of the nectar samples.For the separation of the analytes, a Hypersil Gold UPLC C18 (2.1 × 150 mm, 1.9 μm) reversed phased column (Thermo Scientific) was used.The analysis was performed on a Q-Exactive Orbitrap mass spectrometer using a negative and positive heating electrospray ionization source (Thermo Scientific).
The mobile phase consisted of solvents A: aqueous 0.1% (v/v) formic acid and B: MeOHobtained from Merck (Darmstadt, Germany).Ultra-pure water was produced from SG Milipore apparatus.Nylon filters (0.22 μm), used for filtering nectar extracts prior to injection in the UHPLC-HRMS system, were obtained from Macherey-Nagel (Dueren, Germany).Naringenin (>99% purity) was obtained from Extrasynthese (Genay, France).All reagents and chemicals were of analytical grade.A gradient elution methodology from 0 to 30 min has been employed as follows: 0 to 21 min: 5% B, 21-24 min: 95% B, 24-30 min: 5% B. The flow rate was 0.220 mL/min and the injection volume 5 μL.The column temperature was kept at 35°C while the sample tray temperature was set at 10°C.The optimized conditions for analysis were set as follows: capillary temperature, 320°C; spray voltage, 2.7 kV (for negative) and 3.6 kV (for positive); S-lense Rf level, 50 V (for negative) and 55 V (for positive); sheath gas flow, 40 arb.units; aux gas flow, 8 arb.units; aux.gas heater temperature, 200°C.
Analysis was performed using the Fourier transform mass spectrometry mode of the LTQ orbitrap (FTMS) in the full scan ion mode, applying a resolution of 35,000, while acquisition of the mass spectra was performed in every case using the centroid mode.The mass range for full MS was set at 120-1300 m/z.The data-dependent acquisition capability was used at 17,500 resolutions, allowing for MS/ MS fragmentation of the six most intense ions of every peak apply- For in silico fragmentation, the MetFrag tool (Wolf et al., 2010) was used, applying 5 ppm search tolerance and 0.001 mass deviation to match generated fragments against MS/MS peaks.For MetFrag, the candidate structures were retrieved from the Kegg (http:// www.genome.jp/ kegg/ compo und/ ) database.For the selection of the compounds, a final score >0.8 was applied.
TA B L E 1 Flower traits studied and their phenotypic classes/units.

| Statistical analysis
To determine the effect of the V. faba accessions ('Episkopis', 'Histal') on bee visits in each observation date, the means of the bee visits/ plot/min were separated using t-test with two independent samples (α = 0.05).The statistical analyses were performed using the statistical package JMP (v7, SAS Institute Inc., 1989-2023).
In order to detect possible differences of the flower traits between the two accessions, t-test was conducted to compare each of the following quantitative flower traits: NFIN, WS, FL, DS, FP.
In addition, PCAmix (Principal Component Analysis of a mixture of numerical and categorical variables) was performed to examine the contribution of each of the flower traits and nectar chemical substances, in the classification of the two accessions as described in

| Pollinator visits on faba bean flowers
The bee pollinators recorded were Eucera nigrescens Pérez on flowers of both accessions; Apis mellifera L. and Anthophora plumipes Pallas only on flowers of the landrace 'Episkopis'.The mean number of bee visits/4 m 2 /1 min on 'Episkopis' flowers ranged from 0.33 to 3.08 bee visits/plot/1 min, whereas the respective range of means on 'Histal' was from 0.00 to 0.66 bee visits/plot/1 min (Figure 2).
Statistically, significantly more bee visits were recorded on the faba bean landrace flowers compared with the bee visits on the commercial variety in three out of the four observation dates ( 13

| Flower morphological and agronomical traits
The flower morphological traits recorded during the study for each accession are presented in Table 2. Regarding flowering agronomical traits, the landrace 'Episkopis' reached the flowering stage 2 days earlier than the commercial variety 'Histal'.Thus, the two accessions had a similar period from sowing until 50% Concerning the qualitative morphological flower traits, the ground colour of the standard petal was white and the wing petal was spotted with a dark brown melanin spot in both accessions.The mode number of the trait intensity of streaks on the standard petal was 'strong' in 'Episkopis', while 'moderate' in 'Histal'.Moreover, the trait extent of anthocyanin colouration on the standard petal was 'large' in 'Episkopis' and 'medium' in 'Histal' (Figure 4).

| Nectar chemical analysis
HRMS-based NTS analysis has revealed various compounds in the nectar of the two accessions (for UHPLC-HRMS chromatograms in both ionization modes see Figure S1).In addition, four com-  S1 and representative Figures: Figure 5, Figure 6).Table 3 summarizes the putatively annotated compounds that were found to be differentially increased (using the obtained peak area for relative quantification) in the landrace 'Episkopis' compared with the commercial variety 'Histal' (Table 3).Among these compounds, which are present in both accessions, are important compounds for bee attraction, such as amino acids, flavonoids, and sugars.

| Contribution of flower traits and nectar chemical substances in the description of the two accessions
The first two axes of PCAmix explained 85.72% of total variation.
Importantly, the two accessions were clearly separated along axis F1 (Figure 7a).All examined traits contributed to the formation of F1 axis, and as a result, to the separation of V. faba accessions.The numerical traits with lower contribution to axis F1 were NFIN (number of flowers per inflorescence), WS (width of standard petal), FL (flower length), while the categorical traits with lower contribution were ISSP (intensity of streaks on standard petal) and EACSP (extent of anthocyanin colouration on standard petal) (Figure 7b, c).All numerical variables were positively correlated to landrace, except for trait DF (days to flowering), which is positively correlated to the commercial variety (Figure 7b).It should be mentioned that the categorical traits 'ground color of standard petal' (GCSP), 'wing petal colour' (WPC), 'colour of melanin spot of the wing' (MSW) were excluded from the PCAmix analysis as they were identical in both accessions.

| DISCUSS ION
In the present study, a faba bean landrace ('Episkopis') had higher bee visits compared with a commercial variety ('Histal').Τhe faba bean landrace which attracted more bee visits and diverse taxa had longer and wider flowers compared with the commercial variety.Differences in pollinator visitation between varieties of the same plant species have been examined in few studies (Mallinger & Prasifka, 2017) and attributed to floral morphology differences between cultivars (e.g., blueberry cultivars in Courcelles et al., 2013).Flower size is strongly associated with bee visits as it also affects the nectar and pollen accessibility (Elle & Carney, 2003;Herrera, 1989;Stang et al., 2009;Suso et al., 2008).Several studies report that even very small differences in floral morphology can influence pollinator preference and consistency (Green & Bohart, 1975;Gross, 1992;Jones, 1978) and the reduction in floral size, petal width, flower length, and number of flowers can result in decreasing visitation rates by wild bees (Kuppler & Kotowska, 2021), supporting our findings.Moreover, the flowers of the landrace have more dense pigmentation than those of the commercial variety.In fact, the ground colour of the standard petal in both accessions is white but the intensity and extension of colouration (streaks on standard petal) is stronger in the landrace.This may also affect the number of bee visits on the two accessions as flower pigmentation is an important morphological cue for bees to visit a flower (Barda et al., 2021;Davies et al., 2012;Miller et al., 2011).Differences in flower nectar quality between the two accessions in our study align with previously reported intraspecific variability of floral nectar composition (Bertazzini & Forlani, 2016).The flavonoids detected in nectar of both accessions were kaempferitrin, kaempferol-3-O-β-glucopyranosyl-7-O-α-rhamnopyranoside, kaempferol-3-galactoside-6″-rhamnoside-3″'-rhamnoside.Kaempferol derivatives have been detected as the main flavonoids of rosemary and acacia nectar (Gil et al., 1995;Truchado et al., 2008).Naringenin, which was exceptionally higher in the landrace's nectar, plays a fundamental role in the biosynthesis of anthocyanins (Grotewold, 2006).
Also, wogonin which was present only in the landrace flower nectar, has been previously detected in acacia flower nectar (Wang et al., 2017).The apigenin flavonoid derivative 5-hydroxy-2-(4-hydroxyphenyl)-4-oxo-3-(beta-D-xylopyranosyloxy)-4H-chromen-7-yl 6-deoxy-alpha-L-mannopyranoside which was only detected in the commercial variety, has been previously detected in nectar and linked with flower pigments (Mishio et al., 2015).Coloured flower nectar has been associated with pollination by vertebrates (birds and bats) but not with bees (Miller et al., 2011).Nevertheless, we cannot value a possible effect of colour from these nectar compounds on bee attraction as in both accessions, the nectar is not exposed due to flower morphology.Gong et al. (2020) demonstrated that honey bees showed a preference for nectars containing polyphenols and that these compounds were capable of increasing memory retention and affecting sensitivity to bee-alarm odours.In particular, kaempferol increases resilience against insecticides, enhances pathogen tolerance, supports the immune system of honey bees (Bernklau et al., 2019) and benefits bee colony health and the performance of workers (Niño et al., 2022), rewards that might render the nectar of 'Episkopis' more attractive to bees.
Pollinator flower preferences are greatly connected with nectar amino acids and sugars (Tiedge & Lohaus, 2017).In our case, the amino acids in the nectar of both accessions included asparagine, phenylalanine, 2-phenylglycine and N-acetyl-L-tyrosine.In addition, the non-protein amino acid derivative, L-DOPA, was expressed only in the landrace while the L-DOPA 3′-glucoside, which is a tyrosine derivative, was only detected in the flower nectar of the commercial variety.L-DOPA is an essential precursor in the biosynthesis of melanin (Birkett et al., 2023;Soares et al., 2014).To our knowledge, this is the first time that L-DOPA is detected in flower nectar.The chemically related to L-DOPA, octopamine and tyramine which act as insect neurotransmitters, have been reported as flower nectar constituents (Muth et al., 2022).However, in the current study, despite all precautions being taken, the contamination of the nectar by pollen or damaged cell contents of plant tissue cannot be completely excluded.Phenylalanine which was found to be much more increased in the landrace compared to the variety has been reported to be a significant phagostimulant for honey bees (Hendriksma et al., 2014;Inouye & Waller, 1984), and may be associated with the bee visits on the landrace flowers.Asparagine which slightly differed between the accessions has been reported to behave as a repellent for pollinators, whereas tyrosine had no effect on chemosensory cells of two fly species (Petanidou et al., 2006; Seo TA B L E 3 Putative annotation of compounds that were found to be differentially increased, using the obtained peak area for relative quantification, in the landrace 'Episkopis' compared to the commercial variety 'Histal'.et al., 2019;Shiraishi & Kuwabara, 1970).The flower nectar of both accessions contained the nucleosides uridine and adenosine, but differentially increased in 'Episkopis' nectar.Adenosine has been previously found in the nectar of melon flowers (Fernandes et al., 2023) and its receptors have been linked with the impact of caffeine in the nervous system of insects (Mustard, 2020;Wright et al., 2013).
Regarding sugars, nectar secretion of faba bean has been described as sucrose-dominant (Pierre et al., 1996).α-Lactose was detected in both accessions but was differentially increased in the landrace's flower nectar.Lactose can be detected by bees (Allen Sylvester, 1979), has been previously found in flower nectar, for example, in Rhinanthus minor (Kwak et al., 1985) and serves as attractant and feed supplement for bees (Burden & George, 1978).
Thus, higher expression of lactose in the nectar of the landrace may have played an important role in the higher bee visits recorded in the same accession.To strengthen our argument, when Hendriksma et al. (2014) exposed bee foragers to a range of 15-24% sucrose solutions, they recorded preferences of honey bees for higher sucrose concentrations.
In both accessions, the nectar contained the alkaloid trigonelline but this was increased in the landrace.Trigonelline has been previously found in honey samples from Europe and South America (Donarski et al., 2008;Schievano et al., 2015) but, to the best of our knowledge, this is the second study in which trigonelline has been identified in nectar, after the study of Lande et al. (2019).Lande et al. (2019) suggested that honey bees and bumble bees follow cues from within the nest to a nectar source and experience strong associative learning due to alkaloids in the nectar.Further investigation on trigonelline's effect to foraging pollinators on faba bean nectar would be interesting to be conducted.Mildly toxic or narcotic levels of chemical substances in the nectars of some plant species make pollinators to move slowly on the flowers, lengthening the time spent on them and the chances of successful pollination (Ehlers & Olesen, 1997).
Other differences in the nectar chemical compounds of the two accessions include the protocatechuic aldehyde and the iridoid glycosyltransferase 7-Deoxyloganetin.Phenolic aldehydes such as protocatechuic aldehyde which was found in the landrace nectar, contribute to plant protection against stress and oxidative damage and have been detected in honey samples as they transferred via nectar to honey (Ciulu et al., 2016;Hájek, 2023), but have not been linked with bee visits on flowers.The iridoid glycosyltransferase 7-Deoxyloganetin was detected only in the commercial variety nectar but it has been detected on other flowers before (Suárez-Baron et al., 2019).Richardson et al. (2016) reported that high compared with low nectar iridoid glycoside concentrations induced longer flower foraging of bees infected with protozoan parasites.
Τhe recorded differences in bee visits on the landrace may be important for the best exploitation of faba bean in agro-ecological schemes for the conservation of pollinators and the support of crop pollination, for example, the landrace 'Episkopis' attracted A. plumipes.As A. plumipes is active from early spring, it is an important pollinator for plants that bloom in cool spring weather such as blueberries and apples (Batra, 1994).Since, the examined region (Tegea plateau) has low temperatures during spring, V. faba can be utilized to attract Anthophora species to the benefit of early blooming crops, mainly in orchards (e.g., the Protected Designation of Origin apple landrace 'Delicious Pilafa Tripoleos').Moreover, Anthophora spp.are efficient pollinators for blueberries (Stubbs & Drummond, 1999), other legumes such as Lupinus sp.(Wainwright, 1978) and orchards.On the other hand, Eucera nigrescens was recorded on both accessions.
during spring 2022.Two accessions of V. faba plants, a landrace and a commercial variety, were established to compare their attraction to pollinators in flowering under the design of two independent samples with 12 replications [hereafter called plots (2 × 2 m = 4 m 2 each)] for each accession (treatment).Each plot consisted of 32 faba bean plants.The plots within the same accession were 60 cm apart.The plots of each accession were established in two separate areas (70 m 2 /each) of the field which were 11 m apart to avoid inter-plot competition, for example, risk of having individuals/insects visiting both landrace and commercial variety flowers during a single pollinator observation.

F
I G U R E 1 Layout of the sown faba bean plots ['Episkopis' ('EP'), 'Histal' ('HI')] where bee visits on faba bean flowers were recorded; Tegea plateau, Peloponnese, Greece.[Colour figure can be viewed at wileyonlinelibrary.com] 2.5 | Collection of nectar for chemical analysis HRMS-based NTS (non-targeted screening) of the floral nectar of the two accessions was conducted.To extract the floral nectar, open flowers of stage four to five, as described by Osborne et al. (1997), ing a 5 s dynamic exclusion.The stepped normalized collision energy was set at 20, 40, and 60.The Xcalibur version 4 was used for data acquisition.The raw data were exported to Compound Discoverer 3.3 (Thermo Fisher Scientific Inc.) for peak detection, deconvolution, deisotoping, and composition prediction procedures.For the putative annotation of the compounds, the mzCloud (https:// www.mzclo ud.org) and the NIST/EPA/NIH Mass Spectral Library 2020 (National Institute of Standards and Technology.NIST/EPA/NIH Mass Spectral Library.Wiley.com.2020, https:// scien cesol utions.wiley.com/ solut ions/ techn ique/ gc-ms/ nist-epa-nih-mass-spect ral-libra ry-2020-2/ ) databases were used applying m/z tolerance of 5 ppm, taking into consideration the isotope distribution similarity and MS/MS fragmentation pattern.
of the plants flowered(150 and 152 days, respectively).Similarly, the period from the first day of flowering to the end of flowering differed slightly between the accessions, 'Episkopis': 45 days, 'Histal': 40 days.The flower length (FL) and the width of standard petal (WS) were statistically larger in the landrace compared with Histal (p < 0.001 in both traits), whereas the number of flowers per inflorescence (NFIN), the days to flowering (DF), and the flowering period (FP) did not differ statistically.In particular, the mean length of the 'Episkopis' flowers was 3.26 ± 0.02 cm, while in 'Histal' 3.09 ± 0.05 cm.The mean width of standard petal was 2 ± 0.02 cm in 'Episkopis' and 1.77 ± 0.04 cm in 'Histal'.In Figure3, the Box and Whisker Plots show the range of values in each trait for each accession, and the area of values in which 50% of the examined flowers were in, for each trait.F I G U R E 2 Mean total number of bee visits (per plot/1′ ± s.e.m.) and per bee species (A.mellifera, A. plumipes, E. nigrescens) recorded on flowers of the V. faba landrace 'Episkopis' and the commercial variety 'Histal' for 1 min/4 m 2 at four observation dates.Different letters on the top of the columns indicate significant differences between the accessions in each observation date.
Furthermore, A. mellifera and A. plumipes visits were recorded only on the landrace flowers, whereas E. nigrescens visits were recorded on both accessions.All the recorded species are generalist bees which are important to maintain the cohesion and stability of plantbee networks(Memmott et al., 2004).In the same locality, Barda et al. (2023) recorded A. mellifera, Eucera spp., A. plumipes, Bombus spp.and X. violaceae on the faba bean landrace, but Bombus and Xylocopa genera were not recorded in the present study.Also, in other studies, records of bees on V. faba flowers include species of the genera Andrena, Megachile, Chalicodoma and Colletes whichwere not recorded in our study as well as genera that we also report, that is, Apis, Anthophora, and Eucera have been observed on V. faba before(Aouar-Sadli et al., 2008;Kambrekar et al., 2019;Shebl & Farag, 2015).

F
Box and Whisker Plots of the flower traits (a) width of standard petal (WS) and (b) flower length (FL) for the faba bean landrace 'Episkopis' and the commercial variety 'Histal'.[Colour figure can be viewed at wileyonlinelibrary.com]Flowers of V. faba where the difference of the floral traits 'intensity of streaks on standard petal' and 'extent of anthocyanin colouration on standard petal' between the accessions is shown.Left: flower of 'Histal' (commercial variety), right: flower of 'Episkopis' (landrace).The flower pictures are not indicative of flower size and respective traits (FL, WS).[Colour figure can be viewed at wileyonlinelibrary.com]F I G U R E 5 L-DOPA annotation based on MS/MS fragmentation pattern using mzCloud database and applying m/z tolerance of 5 ppm.The top panel is the MS/MS pattern from the sample 'Episkopis' whereas the bottom panel is the MS/MS pattern of L-Dopa from the mzCloud database.[Colour figure can be viewed at wileyonlinelibrary.com]F I G R E 6 Naringenin annotation based on MS/MS fragmentation pattern using mzCloud database and applying m/z tolerance of 5 ppm.The top panel is the MS/MS pattern from the sample 'Episkopis' whereas the bottom panel is the MS/MS pattern of Naringenin from the mzCloud database (naringenin's MS/MS pattern was also verified with the one obtained from authentic analytical standard solution injection).[Colour figure can be viewed at wileyonlinelibrary.com] Apis mellifera visited only 'Episkopis' flowers.Whether this could indicate a preference over 'Histal' by this widespread pollinator in Europe and the most utilized for crop pollination of various crops would be worth to investigate further.F I G U R E 7 Graphical outputs of PCAmix for the flower and nectar traits of two Vicia faba accessions, the landrace ('Episkopis') and the commercial variety ('Histal'): (a) Component map with factor scores of the observations; (b) Component map with factor scores of numerical variables; (c) Plot of the squared loadings of all variables (numerical and categorical).[Colour figure can be viewed at wileyonlinelibrary.com]