1 Introduction

Stingless bees, a pantropically distributed and species-rich group of eusocial bees, are the main pollinators in tropical ecosystems (Michener 2007; Grüter 2020). Although they are common diurnal visitors to tropical plants, we know little about how nutrients in nectar affect their floral choices (Grüter 2020). Like other bees, they likely show nutrient-sensitive or selective foraging for specific nutrients required by the colonies, which may determine their foraging choices (Roubik 1989; Vaudo et al. 2015; Grüter 2020) and shape their interactions with plants. Understanding these nutritional preferences offers valuable insights into which nutrients are necessary for their survival and how changes in the diversity and composition of food resources, as currently observed in tropical ecosystems as a consequence of human activities, affects their health (Parreño et al. 2022).

Stingless bees need proteins, carbohydrates, lipids, and micro-nutrients to survive and reproduce (Roubik 1989; Michener 2007; Brodschneider and Crailsheim 2010; Grüter 2020). Carbohydrates are the main energy source used by adult bees, e.g., required for flight (Brodschneider and Crailsheim 2010). Proteins and their monomers (amino acids) are primarily needed for the growth and development of bee larvae and thus directly influence colony reproductive fitness (Brodschneider and Crailsheim 2010). Lipids and their monomers (fatty acids) provide energy and are precursors in various biosynthetic pathways and therefore also essential for bee larvae development (Cantrill et al. 1981), as are micro-nutrients (Leonhardt et al. 2024). Here, in particular, sodium is essential for maintaining physiological homeostasis and for counteracting the toxic effects of high doses of potassium which is a characteristic of plant diets (Cohen 2015; Filipiak et al. 2023).

Bees obtain most of these required nutrients from pollen and nectar (Michener 2007). Nectar provides carbohydrates (mono- and disaccharides), and to a smaller extent also amino acids, proteins, minerals (i.e., K+, Na+, Ca++, and Mg++), lipids, and organic acids (Nicolson and Thornburg 2007). However, most of these nutrients are provided by pollen, which is therefore considered the most important food source for larvae development and hence bee reproductive fitness (Leonhardt et al. 2024). In turn, nectar is typically considered the primary energy source of adult bees. Nevertheless, most nutrients are also ubiquitous (in lower quantities) in nectar (Leonhardt et al. 2024), and it is largely unclear how their presence influence the foraging preferences of pollinators. For instance, nectar amino acid composition and/or concentration have been found to affect the foraging choices of honeybees and butterflies (Inouye and Waller 1984; Alm et al. 1990; Rusterholz and Erhardt 2000). However, choice experiments with sucrose solutions differing in amino acid content, type, and concentration testing different bee species showed contrasting responses. For example, honeybees preferred sucrose solutions with specific single amino acids at moderate concentrations (e.g., 2 to 20 mM of proline) or with a mix of amino acids (16 mM), but their preference decreased with higher amino acid levels (Alm et al. 1990; Carter et al. 2006). Moreover, their preference depended on the type of amino acid added (Inouye and Waller 1984; Bertazzini et al. 2010; Hendriksma et al. 2014), with preferences partly driven by their essentiality (EAA: arginine, histidine, isoleucine, leucine, lysine, phenylalanine, threonine, tryptophan, and valine) (de Groot 1952; Hendriksma et al. 2014).

Notably, stingless bees exhibit different amino acid preferences than honeybees. For instance, the Australian stingless bee Tetragonula hockingsi showed no preference for sugar solutions enriched with an amino acid mix (each at a concentration from 0.20 to 0.5 mM) (Gardener et al. 2003). Likewise, eight neotropical stingless bee species, including Trigona fulviventris, Scaptotrigona pectoralis, and Melipona fuliginosa, did not show preferences for sucrose solutions enriched with single amino acids (at concentrations of 35–80 mM); M. fuliginosa even avoided sucrose solutions with specific amino acids, such as arginine (essential for honeybees), serine, glycine, and alanine (all non-essential for honeybees) (Roubik et al. 1995).

Nectar minerals may also affect pollinator preferences, as mineral availability can be scarce, in particular in tropical habitats (Kaspari et al. 2008; Oesker et al. 2010; Dudley et al. 2012). This scarcity likely explains why pollinators occasionally show a preference for water enriched with salts (e.g., Finkelstein et al. 2022). For instance, stingless bees were attracted to salt (NaCl) solutions (Roubik 1996) and preferred sodium in the form of NaCl-soaked baits over water-soaked baits (Afik et al. 2014; Dorian and Bonoan 2016, 2021). Similarly, honeybees preferred water enriched with NaCl (Lau and Nieh 2016; Cairns et al. 2021), but preferences decreased with increasing concentrations (Lau and Nieh 2016). Interestingly, however, NaCl rich diets did not affect the survival of honeybees (de Sousa et al. 2022). In fact, in particular, sodium is often scarce in the bees’ food sources (Filipiak et al. 2017), and they may actively forage for it.

Other, yet still little investigated, nectar components that could play a role in pollinator preferences are lipids. Nectar can contain diglycerides and unmodified free fatty acids, among others (Vogel 1971; Seigler et al. 1978; Buchmann 1987; Ecroyd et al. 1995; Nicolson and Thornburg 2007). Their concentration can even reach levels that change the visual appearance of nectar (from clear to milky) as is the case for nectar in flowers of the genus Jacaranda (average total fatty acid concentration of 0.60 mM) (Kram et al. 2008). Interestingly, Jacaranda nectar is also collected by several bee species, including Trigona (Milet-Pinheiro and Schlindwein 2009). Fatty acids can be consumed from nectar and are used as a metabolic fuel in moths (Levin et al. 2017). However, relatively high concentrations of fatty acids in pollen reduced survival in the bumblebee Bombus terrestris and were thus avoided by foragers (Ruedenauer et al. 2020). The ecological role of fatty acids in nectar is as yet unknown for bees.

To better understand how nectar nutritional composition may affect stingless bee foraging choices and thus their interactions with plants, we investigated nutritional preferences in nectar foragers of the stingless bee Trigona fulviventris Guerin in Costa Rica. In feeding choice experiments in the field, we tested foraging responses to sucrose solution enriched with either a mixture of amino acids, table salt (NaCl), or a mixture of fatty acids, all applied in concentrations as naturally found in nectar (Table I) to determine if enriched sugar solutions were preferred over pure sucrose solutions. The three substance classes were chosen, because they represent important nutrients for bees in general, but their effect on nectar foraging choices has hardly been studied in stingless bees. As salt represents a scarce and limited resource in tropical habitats, it might play an important role in nectar foraging of Trigona fulviventris. Therefore, we expected preferences for sucrose solution enriched with table salt, which has also already been shown for other stingless bee species. Previous studies also found that preferences of other stingless bee species towards sugar solutions with additional amino acids depended on the type and composition of amino acids, indicating that stingless bees might not actively forage for amino acids in nectar. We therefore expected no clear preferences of T. fulviventris for sugar solutions with additional amino acids. We refrain from making predictions for fatty acids due to the lack of previous studies.

Table I Information on the quinine, amino acids, fatty acids, and table salt solutions
Full size table

2 Material and methods

2.1 Study species and location

Trigona fulviventris (Hymenoptera: Apidae) is a medium size (c. 1.5 mm inter-tegulae span (Streinzer et al. 2016)) neotropical bee species found from southern Mexico to Brazil. It can be a rapid food recruiter, and it has relatively large colonies (> 10,000 workers) (Roubik 1989). Trigona fulviventris is a common species at our field site and can easily be trained to sucrose feeders. Several studies have been carried out on this genus over the last years at the same location (e.g., Jarau and Barth 2008; John et al. 2012; Spaethe et al. 2014; Sommerlandt et al. 2014; Streinzer et al. 2016; Eckert et al. 2022).

The experiments were conducted in the botanical garden of the Tropical Research Station La Gamba, district Golfito (8° 42′ 3.78″ N, 83° 12′ 6.14″ W), in the Pacific southwest lowlands of Costa Rica between March and April 2022. For identification, three foragers were collected and killed via freezing (collection and research permit: SINAC-ACOSA-D-PC-085-2022, SINAC-ACOSA-D-PI-R-018-2022, respectively). No bees were collected or killed over the course of the experiment.

2.2 Experimental procedure

2.2.1 Pretraining

First, foragers of the stingless bee T. fulviventris observed to collect nectar from adjacent flowers were trained to a feeder with 1 M sucrose solution. The feeder consisted of one round piece of foam material (7-cm diameter and 1-cm thickness) covered with filter paper, both with a central hole (circa 2-cm diameter) to hold a snap-on cap with sucrose solution (as in Sommerlandt et al. 2014) (Figure 1). The filter paper allowed the bees to mark the feeder (i.e., place pheromones on its surface to communicate with other foragers, Sommerlandt et al. 2014) without directly marking the foam. The foam was cleaned with water and 70% ethanol and reused over the next days and in different experiments. We used different foams on 2 consecutive days to further rule out any pheromone effects.

Figure 1.
figure 1

a Yellow and b blue feeders used in the learning phases and c test phases. Each feeder consisted of one piece of foam material (7-cm diameter and 1-cm thickness) covered with filter paper and a central hole (circa 2-cm diameter) with a snap-on cap. Each feeder was placed on yellow or blue cardboard (11 cm × 12 cm). In the learning phase, the snap-on cap contained the tested solutions (in blue: sucrose solution plus tested substance, in yellow: pure sucrose solution). When conducting the control tests for color preference, we offered only a pure sucrose solution in both the blue and yellow feeder. In the test phase, the feeders (with a snap-on cap containing water) were presented at the same time, so the bees could make a color choice (the bees were not allowed to drink from the water). We tested choices of 20 bees and changed the orientation of the feeders after 10 choices. Every test phase started with a different feeder orientation.

Full size image

The bees were trained to the feeder by offering them a drop of sucrose solution using a pipette. Once the bees had detected the drop of sugar they climbed on the pipette to feed from the sucrose solution. Pipettes with bees were then carefully moved from the flower to the feeder. If bees accepted the feeder, they returned after a short period and over the next couple of days as long as the feeder was presented. Return of foragers was confirmed at the beginning of the experiment by color-marking four bees on their thorax (each bee with a different color) and noting their return to the feeder. We trained approximately 10–20 bees at the beginning. Additional ones were then recruited by their nest mates. Once the bees were trained, we decreased the sucrose concentration to 0.5 M and then to 0.25 M to maintain a constant number of foragers (up to 20–30 individuals) during the experiments. The reason for this is that T. fulviventris is a mass-recruiting species (Slaa et al. 2003; Lichtenberg et al. 2017) and quickly recruits large numbers of foragers to highly attractive food sources, which renders behavioral experiments impossible. It is thus important to constantly adapt the sucrose reward to a level that is (only) medium attractive to the foragers. Note that we cannot make any inferences on the number of colonies tested, as it is nearly impossible to locate wild colonies in a natural rainforest. It is therefore possible that most bees trained in our experiment were from the same colony, as we did not observe any aggression between foragers as typically shown by T. fulviventris towards non-nestmates (Buchwald and Breed 2005).

2.2.2 Test for innate color preference

To establish the experimental setup, we first determined whether the bees had an innate color preference for either blue or yellow. To do so, the feeders were placed on yellow or blue cardboard (11 cm × 12 cm, Figure 1a, b) (hereafter called yellow or blue feeders, respectively). The bees’ innate color preference was tested by offering them both feeders containing the same sucrose solution during the learning and test phases (for a detailed description of the experimental procedure see Section 2.2.3). We found that the bees showed a slight (innate) preference for blue (proportion of blue choices (b) = 0.55 ± 0.11 (sd), n = 21). We therefore chose the blue feeder to always represent the feeder with the enriched solution to reliably disentangle preference and learning effects. We assumed that a preference for or avoidance of the added substances (see Table II) would increase or decrease, respectively, the choice ratio observed for pure sucrose solution, i.e., blue vs. yellow, 55:45 (used as baseline), while the ratio should remain constant if the bees’ were indifferent towards the added substance.

Table II Generalized linear models’ results for the proportion of feeder color choices depending on the tested substance
Full size table

We then tested whether the bees could relate the feeder content to the feeder color. We used a sucrose solution mixed with quinine as negative control, because it is known that the bitter taste of quinine is typically avoided by bees (Chittka et al. 2003). The quinine solution was presented in the blue feeder. Thus, the bees should change their preference for blue to yellow if they were able to relate the feeder content to feeder color. We found that the proportion of blue choices shifted to (b) = 0.43 ± 0.09 (sd) (Table II) during the experiments, demonstrating that the bees were able to associate feeder color and content.

2.2.3 Learning and test phase

After determining the bees’ color preference and establishing a reliable experimental setup, we finally tested how the bees responded to a choice between a solution with sucrose and a solution enriched with each of the test substances (Table I), simulating nectar rich in specific nutrients. All experiments (including the innate color preference test and the negative control) consisted of two phases: a learning phase and a test phase (Figure 2). In the learning phase, we started offering one feeder, e.g., the yellow feeder with the pure sucrose solution (0.25 M or 0.125 M depending on the round). The bees were allowed to forage for 10 min as in Sommerlandt et al. (2014). Afterwards, the yellow feeder was replaced by the blue feeder with sucrose solution (0.25 M or 0.125 M) enriched with the test substance (except for the color preference test where the blue feeder also contained the same pure sucrose solution as the yellow feeder). Bees were again allowed to forage for another 10 min. Then, the blue feeder was removed, and the yellow feeder was presented again. This procedure lasted for 60 min (Figure 2). During the learning phase, each feeder was presented three times, which allowed the bees to learn and associate the feeder color with a specific sucrose solution composition. The next learning phase started with the other feeder (e.g., first blue, then yellow). Thus, each learning phase started with the color not used at the beginning of the previous learning phase (Figure 2).

Figure 2.
figure 2

Graphical scheme of the experimental procedure. The scheme shows the learning and test phases of the experiment. The feeders are represented with a white double circle on a blue or yellow rectangle, depending on the feeder. In the learning phases, the blue feeder always contained the sucrose solution plus the tested substance, and the yellow feeder only sucrose solution. In the test phases, both feeders contained water. During the learning phase, the bees were allowed to drink freely from the feeders, so they learned to relate feeder color with content. The feeder was changed every 10 min. After 60 min of learning phase, a test phase was performed. In this phase, we tested the choices of 20 bees. First, the color decision of 10 bees was noted, then the orientation of the feeders was rotated to prevent side biases, and the decision of additional 10 bees was noted. In this phase, the bees were not allowed to drink from the feeders, and each test phase started with a different orientation. After the choices, a new learning phase started. Each learning phase started with the color not used at the beginning of the previous learning phase.

Full size image

Each learning phase was followed by a test phase. During the test phase, we presented the two feeders (only with water, hereafter called test feeders) simultaneously next to each other (Figures 1c and 2) so that the bees could make a color choice based on their nutritional preference established in the learning phase. We considered each landing of an individual bee on any of the two test feeders a choice. In the test phase, the bees were not allowed to drink from the water but were chased off once they had made a clear color choice. We counted the choices of roughly 20 bees. After 10 bees made a decision, the orientation of the feeders was exchanged to prevent side biases. Each test phase started with a different orientation. After 20 choices, the test feeders were removed, and a new learning phase started (Figure 2). Per day, we performed between four to six learning and test phases, depending on weather conditions. On each day, only one substance was tested with a different substance trained and tested at subsequent days to avoid that learning and memory from previous days affected the bees’ choice behavior. Note that we did not capture the bees to not affect their choices, as in pre-trials, we found that capturing foragers prevented subsequent foragers from approaching the feeders. We ensured however that there were at least 20–30 bees foraging on our feeders during the training bees, and we always saw at least 5–10 different foragers approaching the feeders during the test phase. As we did not capture foragers, we cannot rule out multiple testing of the same foragers.

After all substances had been tested (one substance per day, four to six times per day), a new round started where the bees were trained and tested again for all substances. Before each round, we repeated the test for the bees’ preference for either blue or yellow. Overall, we performed three rounds (so each substance was tested between 14 and 20 times over the course of the entire experiment, Table II). In the learning phases of the first and third rounds, we used a sucrose concentration of 0.25 M, while we had to switch to a 0.125 M sucrose solution in the second round, as too many bees were approaching the feeders, rendering counting and proper training difficult. When using 0.125 M sucrose solution, we adjusted the amount of the tested substances accordingly to keep the same ratios of sucrose to nutrients as in the 0.25 M sucrose solution.

2.2.4 Tested substances

We compared pure sucrose solution with sucrose solution enriched with either quinine, an amino acid mixture at low concentration, an amino acid mixture at high concentration, table salt, or a fatty acid mixture (see Table I for exact concentrations). Note that the fatty acid mixture was only tested in the second and third rounds. The amino acids in the low-concentration mixture (hereafter called low amino acid solution) and in the high-concentration mixture (hereafter called high amino acid solution) corresponded to amino acid compositions and concentrations reported for natural nectar (Table I) (Gardener and Gillman 2001; Petanidou et al. 2006) and consisted of four EAAs and three NAAs (based on honeybee requirements). Moreover, the amino acids used were of the same molarity and of similar composition as used in another study with stingless bees (Gardener et al. 2003) to facilitate the comparison of study results. For the fatty acid mixture, we used concentrations as reported for palmitic and stearic acid in the study of Kram et al. (2008) for Jacaranda nectar. Species within this genus were also found to be visited for nectar by Trigona bees (Milet-Pinheiro and Schlindwein 2009). As we found no reports of concentrations for other fatty acids, we adjusted the concentrations of additional five common fatty acids, also reported in nectar of other plant species (Bernardello et al. 1999; Vesprini et al. 1999; Nicolson and Thornburg 2007), following proportions as reported for pollen (Ruedenauer et al. 2020) (Table I). When mixed with the sucrose solution, the fatty acids did not dissolve completely, resulting in slightly differing individual fatty acid ratios between the 0.25 M and the 0.125 M sucrose solution (Table I). However, the total fatty acid concentration was higher in the 0.25 M sucrose solution (84.1 ng/µl) than in the 0.125 M (44.4 ng/µl) as revealed by analyzing samples of both sucrose solutions via gas chromatography coupled to mass spectrometry and to flame ionization detection (GC-MS, GC-FID, respectively, see SI).

For the salt trials, we used commercial table salt (NaCl), which contained 393 mg of sodium (Na) and between 0.03 and 0.6 mg of potassium (K) and iodine (I) (in the form of potassium iodine (KI)) per gram. As for the other substances, the amount of salt added to the sucrose solution was adjusted to obtain a Na concentration within the natural range of what can be found in nectar (Nicolson and Worswick 1990) (Table I).

2.3 Statistical analysis

Statistical analyses were performed with the software R version 4.1.2 (R Core Team 2022). To account for an (innate) color preference of T. fulviventris, we calculated the average proportion of choices for the blue and yellow feeder in the trials where only sucrose solution was used (preference control). Trigona fulviventris foragers chose the blue feeder more often than the yellow one. We, therefore, considered 55% as the expected value for blue choices if the bees showed neither a preference for nor avoidance of the tested substance (quinine or nutrients), while choice proportions deviating from 55% were considered either a preference (values > 55%) or avoidance by (values < 55%) for the tested substance.

To observe differences in the proportions of choices made to the blue (sucrose solution plus tested substance) and yellow (only sucrose solution) feeders, we implemented generalized linear models (glms, stats R package) with quasibinomial distribution and with the intercept set to 1 (reflecting our null hypothesis, i.e., no difference between blue and yellow) for each tested substance. To add the expected blue proportion, we included an offset term in the models. This offset was based on the calculated logistic quantile function from the average probability of blue choices (55%) (function qlogis, stats R package). The residual diagnosis was done visually using residuals vs. fitted values and quantile-quantile plots (car package, Fox and Weisberg 2019). A significance level (α) of 0.05 was considered for all models.

Round or sucrose concentration was also used as the explanatory variable in different models to test if they affected the observed proportions of choices for the tested substances and, therefore, should be included as random factor in our models (note that we did not test for the effect of these variables on the proportion of blue and yellow choices when only sucrose solution was used). Neither round nor sucrose concentration significantly affected the proportion of observed choices (Table S1, Figure S1), and were therefore not included as random factors in the abovementioned models. P-values and F-values for these models were obtained through an analysis of deviance (function ANOVA, stats R package) (Table S1).

3 Results

We found that T. fulviventris foragers significantly preferred the pure sucrose solution over the sucrose solution enriched with either quinine, the amino acid mixtures (low and high concentrations), or the fatty acid mixture (Figure 3, Table II). The bees, however, did not differentiate between the solution with only sucrose or the one enriched with table salt (Figure 3, Table II).

Figure 3.
figure 3

Proportion of choices to the blue (sucrose solution with tested substance (sucrose + sub.)) and yellow (only sucrose solution (sucrose)) feeder for each tested substance. When conducting the control (gray panel) tests for color preference, we offered only a pure sucrose solution in both the blue and yellow feeder. The black bars represent the confidence interval of 95% obtained with the average proportion of blue and yellow choices for each tested substance. The black diamonds represent the mean value of the data set. Dots indicate the proportions obtained in each test. The horizontal dashed line indicates the proportion of blue choices used as the expected value (0.55, average proportion of blue choices) when only sucrose was used as the tested substance. Triple asterisks “***” indicate a P-value ≤ 0.001, and “ns” stands for no significant results (P > 0.05) based on generalized linear models performed for each substance and using the average proportion of blue choices when only sucrose was tested as the expected value. Number of tests is as follows: only sucrose n = 21, quinine and low amino acids n = 16, high amino acids and fatty acids n = 14, and salt n = 20.

Full size image

4 Discussion

In this study, we explored how nectar chemical composition affects foraging preferences of the stingless bee Trigona fulviventris. As expected, T. fulviventris nectar foragers disfavored the sucrose solution enriched with quinine. Quinine is a plant secondary metabolite, produced by plants for protection against herbivores (Lambers and Oliveira 2019). For humans and likely also bees, it has a bitter taste and is typically avoided (Chittka et al. 2003), though the underlying mechanism is not fully understood (De Brito Sanchez et al. 2005; Bestea et al. 2021). Similarly to bumblebees, honeybees, and kissing bugs (Dyer and Chittka 2004; Rodríguez-Gironés et al. 2013; Avarguès-Weber and Giurfa 2014; Pontes et al. 2014), the bees in our study also quickly learned to relate the quinine sucrose solution to the feeder color, showing that our experimental setup worked. After training, T. fulviventris associated a visual cue with the solution’s content and switched their color preference (from blue to yellow) depending on the solution in the feeder. This capacity to associate color and solution content has also been shown in honeybees (e.g., Bestea et al. 2022). Likewise, an innate preference for blue and the ability to learn to associate feeder and color content has also been reported for other species of stingless bees (e.g., Koethe et al. 2020). This experimental setup allowed us to show that T. fulviventris preferred the pure sucrose solution over the sucrose solution enriched with either high or low concentrations of amino acids, or fatty acids, while the bees made no difference between the pure sucrose solution and the sucrose solution enriched with salt.

Regarding the response to the amino acid solutions, T. fulviventris response was different from responses of the Australian stingless bee T. hockingsi: T. fulviventris preferred a pure sucrose solution over the sucrose solutions enriched with EAAs and NAAs at natural concentrations, while T. hockingsi showed no preferences (Gardener et al. 2003). Prior experiments with T. fulviventris and other stingless bee species similarly showed no difference in foraging choices between pure sucrose solution and sucrose solution enriched with single amino acids (with concentrations from 35 up to 80 mM) (Roubik et al. 1995). However, two Melipona stingless bee species also avoided solutions enriched with specific amino acids (Roubik et al. 1995). In fact, T. fulviventris might have avoided the amino acid mixture in our experiment due to the presence of alanine, which was also avoided by M. fuliginosa (Roubik et al. 1995). Excessive alanine consumption has been linked to decreased survival in some solitary bee species (Felicioli et al. 2018). However, alanine can be found in nectar at comparatively high concentrations (up to 668 mM, Gardener and Gillman 2001), and other stingless bee species and honeybees showed no avoidance behavior towards this amino acid (Inouye and Waller 1984; Roubik et al. 1995; Hendriksma et al. 2014). As we used a mixture with different EAAs and NAAs, we cannot infer specific amino acid effects on T. fulviventris foraging. Our results, together with other studies, indicate, however, that stingless bees might not preferentially forage for amino acid–rich nectar but rather obtain them from pollen and other resources.

We observed a similar effect of sucrose solution enriched with fatty acids as of sucrose solutions enriched with amino acids on T. fulviventris nectar foragers. Although fatty acids are essential for bee development (Vaudo et al. 2015), do occur in nectar, and are consumed by other pollinators (Levin et al. 2017), T. fulviventris preferred pure sucrose solutions over sucrose solutions enriched with fatty acids. As we were the first to test the effect of fatty acids in nectar on bee foraging choices, our results can only be related to studies testing the effect of fatty acids in pollen. These studies showed that pollen enriched with comparatively high amounts of fatty acids also induced avoidance and reduced survival in the bumblebee B. terrestris (Ruedenauer et al. 2020). This suggests that bees strictly regulate fatty acid intake and likely obtain sufficient and appropriate amounts from pollen.

Interestingly, the salt-enriched sucrose solution was the only solution that was equally attractive to T. fulviventris as the pure sucrose solution. This may be explained by the general scarcity of sodium in tropical ecosystems and plant resources (Kaspari et al. 2008; Oesker et al. 2010; Dudley et al. 2012; Welti et al. 2019; Filipiak et al. 2023), leading to a general preference of particularly tropical pollinators for resources enriched with sodium (Kaspari 2020; Demi et al. 2021; de Sousa et al. 2022). For example, when offered raw chicken baits soaked in NaCl, CaCl2, KCl, MgCl2, or deionized water and unsoaked baits, different species of stingless bees visited NaCl baits and unsoaked baits more often than the other baits (Dorian and Bonoan 2021). However, like for other nutrients, preference for mineral solutions also appear concentration, mineral (Afik et al. 2014; Lau and Nieh 2016), and even daytime specific (Cairns et al. 2021). Cairns et al. (2021) examined preference of honeybees for different concentrations and ratios of elements, including sodium, in relation to their preference for sucrose solution at different times of the day. The bees always preferred sodium, except in the morning when they preferred sucrose to the same extent, suggesting that, in the morning, the bees focused on energy intake, while their sodium needs were high all day.

The fact that we did not find such an obvious preference for table salt in T. fulviventris may be explained by the presence of iodine in the table salt used in our study which may have affected the bees’ preference. Unfortunately, to our knowledge, there are no studies on whether bees can perceive specifically the mineral iodine despite its presence in honey (Solayman et al. 2016). Hence, the effect of iodine on T. fulviventris foraging behavior remains inconclusive. Another alternative explanation for our findings is that the bees could not detect or differentiate salt at the offered concentration. We clearly encourage more studies on the effect of this important nutrient group on foraging choices of T. fulviventris and other wild bee species.

Our results indicate that T. fulviventris foragers can perceive (“taste”) other substances in sucrose solution (i.e., nectar) apart from sugar and adjust their foraging behavior accordingly. It remains unclear, however, whether they do have respective taste receptors on their antennae, mouthparts, or tarsae, as was suggested for bumblebees and honeybees (Lim et al. 2019; Ruedenauer et al. 2019, 2021). Alternatively, different substances, like amino acids and fatty acids, rather affect sugar sensitivity (i.e., inhibit receptors responding to sugar) and thus responsiveness, as has been shown for some plant secondary metabolites (De Brito Sanchez et al. 2014; French et al. 2015; Bestea et al. 2021). In fact, a reduced sugar sensitivity might explain why T. fulviventris preferred pure sucrose solutions over sucrose solutions with additional nutrients, independent of their type and concentration.

Clearly, implementing our methodology under tropical field conditions has inherent limitations. One limitation of conditioning bees to feeders to assess nutritional preferences is that we cannot account for the nutritional state of the foragers and colonies, which depends on the current physiological state of the colony as well as on the natural nutrient supply of their environment. During our experiments, the surroundings may have provided all tested nutrients in sufficient amounts, resulting in decreased foraging for the experimental feeders, while the bees’ choices may have been different at times of different nutrient constrain. Moreover, with this setup, we do not know the precise number of tested colonies, but since we saw no aggression among foragers, it is likely that most trained bees originated from the same colony (e.g., Sommerlandt et al. 2014). Thus, our results may reflect the physiological state of the specific colony tested. In addition, we did not capture the bees after they had made a color decision in the test phase so as not to influence their choices. Therefore, the possibility of multiple testing of the same foragers cannot be excluded. We also could not mark individual foragers and do not know whether each tested forager underwent the full training phase before selecting a feeder during the test phase, which may have influenced the observed results. Finally, it is difficult to completely eliminate the influence of other bee individuals on the tested bee and hence exclude social effects on choices during the learning phase. Future studies could consider both environmental nutrient availability and the bees’ physiological state to better understand how current nutritional requirements affect food choices and sensitivity (Seeley 1989; Brodschneider and Crailsheim 2010). They should also test different wild colonies (if possible) and ideally conduct the experiment with individual bees at a time to avoid social effects.

To summarize, our results suggest that the preference of T. fulviventris foragers for nectar depends on its chemical composition. T. fulviventris, and maybe also other stingless bee species, appear to select nectar primarily based on its sugar (Leonhardt 2017) and sodium content, while the amino acids and fatty acids appear to decrease the preference for the sucrose solution impacting bee choices. These results suggest that nectar primarily serves as source for energy and sodium. Notably, the bees’ specific responses and nutrient preferences may still change depending on current environmental nutrient supplies and colony state. In fact, if the environment provides a stable supply of floral resources with pollen and nectar and additional resources (e.g., carrion) that vary little in nutrient concentrations, bees may collect nectar primarily for its sugar content. Conversely, if nutrient supply varied strongly across seasons, the bees might need to adjust their foraging preferences and possibly also their sensitivity to nectar nutrients. Such variation in nutrient supply could even be a major driver of bee–plant interaction patterns.