Stingless bees (Apidae: Meliponini) primarily feed on nectar and pollen from a wide diversity of flowering plants. By doing so they pollinate these flowers thus contributing to biodiversity and ecosystem stability. The pollen they collect provides essential nutrients for brood rearing and colony growth. This study aimed to characterize the floral resources available to stingless bees in a Tucumano-Boliviano Forest, including their pollen protein content and, through construction of an interaction network and preference analysis, understand their foraging behavior. Only 8 out of 25 pollen types sampled within the study site around the meliponaries were collected by the bees. Pollen pots also contained many types of pollen not from to plant species in the study area. Pollen from Capsicum minutiflorum (Solanaceae) was dominant in almost every hive (up to 98.7% of pollen composition). Additionally, protein content of Capsicum minutiflorum pollen (67% w/w) was the highest of all species present at the study site and explained almost 100% of the protein content in the hives of Tetragonisca angustula, and Scaptotrigona depilis, and almost 80% and 75% of the protein content in those of Scaptotrigona polysticta and Melipona rufiventris, respectively. These results suggest that stingless bees preferentially collect pollen with higher protein content.

Bee nutrition, Melipona, plant–pollinator interactions, pollen preferences, Scaptotrigona, Tetragonisca

Bees are considered among the most effective pollinators, and they play a fundamental role in the maintenance of biodiversity and stability of ecosystems. This is especially true in the tropics, where due to the high plant diversity, and consequent low density of the different plant species, plants are more dependent on their pollinators (Absy et al. 2018; Klein et al. 2018; Rodger et al. 2021). Thus, in the tropics nearly 94% of all flowering species are animal pollinated, and most of these animals are bees (Ollerton et al. 2011). In most regions of the neotropics, stingless bees (Apidae: Meliponini) are very abundant and very diverse, with around 400 species of the 550 species described worldwide being found in such habitats (Barbiéri et al. 2022). Stingless bees are eusocial insects that live in perennial colonies with 3,000 to 100,000 individuals (Michener 2007, 2013). In Bolivia, more than 100 stingless bee species have been described (Townsend et al. 2021), and as in many other countries of Latin America, they hold great cultural value for many indigenous and rural populations. (Souza et al. 2013; Reyes-Gonzalez et al. 2014). Stingless bee breeding, formerly called meliponiculture, is primarily practiced for their honey and for other useful resources such as cerumen, propolis, and pollen, which have various religious, medical, nutritional and economic uses (Vit et al. 2013, 2018; Salatino 2019; Adler et al. 2023).

Stingless bees feed on nectar as their primary carbohydrate source and pollen as main protein source, although pollen also provides amino acids, lipids, carbohydrates, sterols, and various micronutrients; thus, pollen constitutes a nutritionally diverse and highly valuable reward (Roulston and Cane 2000; Vossler 2015). Pollen is also important for brood rearing and development, since it is the main ingredient of larval food, and hence its availability directly influences brood rearing and the growth of the colony (Roubik 1982; Abrol 2012). Thus, to maintain a healthy colony, worker bees must be efficient in collecting floral pollen that satisfies its nutritional requirements. In that sense, a preference for high-quality proteins or a high protein content in the diet has been observed in species that are part of the same subfamily (Apinae), to which stingless bees belong. For example, in honey bees (Apis mellifera), depending on the colony’s needs, foragers will search for pollen with higher protein content, although protein-lipid ration and amino acid content plays also a roll in their preference (Ghosh et al 2020). Bumble bees (Bombus) fed artificially with high protein diets, enhance their colony growth (Génissel et al 2002) and in the wild they prefer flower species with higher protein levels that those chosen by honey bees (Leonhardt and Blüthgen 2011). In Stingless bees colonies, higher protein content in the larval food can have a positive effect on body size of worker bees (Quezada-Euán et al. 2011), but if they forage preferentially on pollen with higher protein, has not yet been studied. To avoid competition stingless bees have developed different strategies to recruit their nestmates and tell them where the food sources are located. Depending on the species some have strong recruitment strategies, like most Scaptotrigona species, and some week or no recruitment strategy, like Melipona or Tetragonisca species (Ramalho1989; Michener 2013; Grünter 2020). Although other bee tribes belonging to the “corbiculate” clade, such as Bombini (Vaudo et al. 2016; Ruedenauer et al. 2020) and Apini (Pernal et al. 2001; Cook et al. 2003; Gosh et al. 2020) have distinguishable preferences and foraging behavior towards specific nutrients in pollen to fulfill their nutrient requirement, these preferences have been poorly studied in species from the Meliponini tribe (Grüter 2020).

As meliponiculture in Bolivia continues to grow as a socio-economic activity, particularly for vulnerable populations, it becomes important to conduct in-depth studies of the behavior and requirements of these bees (Adler et al. 2023). A comprehensive understanding of the system will facilitate improved management practices and lead to enhanced productivity and conservation efforts both, of the bees and the ecosystem. This is of particular importance in the Tucumano-Boliviano forest of Chuquisaca, where large-scale meliponiculture is performed by women from different communities. The present study, performed in the protected area of Serranías del Iñao, a region that remains largely understudied, aims to characterize the floral resources available within the foraging range of stingless bees from three meliponaries. Furthermore, we seek to quantify the protein content of various pollen species to enquire about preferential foraging behaviors, thus aiming to provide insights that can lead to sustainable meliponiculture practices and ensure the well-being of both the bees and the communities that rely on them.

A total of 25 plant species categorized into 17 families, were at the flowering stage in the sampling area defined. Most of the observed species were herbs (16 species), and only 4 species of lianas, 4 of trees and one bush showed flowers during the survey period. The dominant plant species at the flowering stage was identified as the liana Heteropterys umbellata (Malphigiaceae), with a relative abundance of 13.64%, followed by the herb Asclepia curassavica (Apocynaceae) (10.91%) (Table 1).

Pollen density, defined as the number of pollen grains per unit volume and calculated using the Neubauer formula, varied significantly among the studied plant species. The highest pollen density was recorded in Piper elongatum, with 1392045 pollen grains per mm³. Ludwigia peruviana followed with less than 1/10 of pollen density (11267), and the flowers with the lowest pollen density were Oxypetalum sp. with just 12 pollen grains per mm³, and Asclepia curassavica with 22 pollen grains per mm³ (Table 1). Unlike volume-based measurements, this approach highlights actual differences in pollen grain abundance, which is particularly relevant for evaluating pollen availability for pollinators.

In terms of cytoplasmic volume, the species with the grates cytoplasmatic volume was Sida anrgentina (Malvaceae), with 1334016 μm³, followed by Sida sp. (Malvaceae) with 782698 μm³. On the other hand, the species with the smallest volume was Piper enlongatum (Piperaceae) with 699 μm3, followed by Oxypetalum sp. (Apocynaceae) with 1921 μm3. No significant correlation was found between pollen quantity and cytoplasmic volume (p = 0.1911, rho = -0.2704), suggesting that pollen grain abundance does not necessarily correspond to pollen grain size.

The protein content from the collected flower species varied from 67% in Capsicum minutiflorum, to 20% in Piper elongatum (Table 1). No significant correlation was found between protein content and pollen quantity (p = 0.495, rho = 0.1463) or pollen volume (p = 0.5248, rho = -0.1365).

In total, in all pot-pollen samples, 23 pollen types were identified: 14 in the pollen types from M. rufiventris, 13 in S. polysticta, 10 in S. depilis and 5 in T. angustula (Table 2).

A very eye-catching result was the dominance of one particular pollen species in almost every hive studied (Fig. 3). Pollen from Capsicum minutiflorum (Solanaceae) was always dominant in every hive, except in hive 1 (H1) from S. polysticta, which showed a more balanced pollen distribution. In this case, Capsicum minutiflorum represented in this hive 17% of the total pollen, while Solanum riparium and Proteaceae sp1. represented 36.3% and 32.5%, respectively.

Figure 3. 

Capsicum minutiflorum A dried specimen B acetolyzed pollen

The average for every bee species, pollen composition between the different species, varied from very different to similar, with a Sorensen index between 26.7% to 69.6% similarity. The Sørensen Similarity Index (SSI) calculated for the percentage of the different pollen types found in the pollen samples from every beehive from the four bee species, demonstrated that between hives of the same species, there was almost always a great similarity (over 50%). The only exception was S. polysticta where some hives showed low similarity indexes (Table 3).

Based on pollen species composition found in the pot pollen for every bee species, the calculated network exhibited an intermediate level of complementary specialization with H’ = 0.516, a high connectance with C = 41.667%, an intermediate level of nestedness with a NODF value of 49.603, and nearly no modularity with Q = 0.163.

Interactions were predominantly driven by the interaction between the four bee species and Capsicum minutiflorum (Fig. 4). However, interaction did not occur equally across the four bee species. In S. depilis and T. angustula, the pollen of C. minutiflorum accounted for 96% and 98.8%, respectively. In contrast, the value of the interaction was 66.6% for M. rufiventris and 71.9% for S. polysticta.

Figure 4. 

Bipartite quantitative interaction matrix based on the pot pollen (green) found in the hives for every bee species (red). Interaction intensity is represented by gray shade.

Regarding the proportion of the protein content in the stored pots that can be attributed to each plant species, Capsicum minutiflorum explained almost 100% of the protein content in T. angustula and S. depilis. However, almost 20% of the protein content in the reserves of S. polysticta could be attributed to Proteaceae sp.1, and almost 25% of the protein content in the reserves of M. rufiventris could be attributed to Plumbago sp., Proteaceae sp.1 and Proteaceae sp.2 pollen grains (Fig. 5).

Figure 5. 

Percentage of protein content collected by each bee species.

According to the width of the taxonomic spectrum of plant species used as pollen sources by bees, they are classified as oligolectic if they forage on plant species belonging to one or several closely related genera, and as polylectic if they obtain pollen from plants from two or more families (Roulston et al. 2000; Murray et al. 2009). However, it has been proposed that even in polylectic species there is some degree of selectivity and bees do not collect pollen from all plant species available (Nicholls and Hempel de Ibarra 2017). Some stingless bees have been considered polyleptic (Oliveira et al. 2009; Rech and Absy 2011; Vossler 2012), while others are more selective in terms of pollen collection (Oliveira et al. 2009; Rech and Absy 2011). For example, Aleixo et al. (2017), and later Vossler (2018) suggests that some known polylectic stingless bee species may collect pollen from only one or a few plant species, even when the floral offer is large; moreover, due to a flexible foraging behavior, false conclusions may be drawn on the category of specialization if data from a single nest is gathered.

The present results show that the species studied are polylectic but nevertheless exhibit pollen selectivity. This conclusion is based on three facts: a) the diversity of pollen types that were collected which belong to more than one family; 9 in M. rufiventris (Asteraceae, Fabaceae, Myrtaceae, Plumbaginaceae, Primulaceae, Proteaceae, Sapindaceae, Solanaceae, and one unidentified family), 10 in S. polysticta (Solanaceae, Proteaceae, Sapindaceae, Myrtaceae, Asteraceae, Primulaceae, Fabaceae, Euphorbiaceae, and two unidentified families), 9 in S. depilis (Solanaceae, Proteaceae, Sapindaceae, Myrtaceae, Amaranthaceae, Asteraceae, Primulaceae, Euphorbiaceae, and Convolvulaceae) and 5 in T. angustula (Solanaceae, Proteaceae, Poaceae, Amarantaceae and one unidentified family, b) only 8 out of 25 pollen types identified around the meliponaries within the study site were collected by the bees studied; 6 in M. rufiventris (Capsicum minutiflorum, Plumbago sp., Sapindaceae sp., Heterosperma sp., Inga adenophylla, I. saltensis), 5 in S. polysticta (C. minutiflorum, Sapindaceae sp., Heterosperma sp., Croton sp., I. saltensis), 5 in S. depilis (C. minutiflorum, Sapindaceae sp., Heterosperma sp., Croton sp., Convolvulaceae sp.), and 1 in T. angustula (C. minutiflorum). And c) the pollen collected by the four bee species belong mostly to a single plant genus; Capsicum.

Bees consume nectar for their own nourishment and typically collect pollen (plus nectar) as food for their offspring, transporting it to the nest via body hairs, scopae or using corbiculae (in the case of eusocial corbiculate bees) (Michener 2007; Nicholls and Hempel de Ibarra 2017). Pollen protein content is crucial for reproduction, growth and longevity of bees, and there is a correlation between performance of bees and the amount of protein consumed (Cane and Buchmann 2000; Cook et al. 2003; Weiner et al. 2010; Quezada-Euán et al. 2011). While earlier studies argued that some bees primarily collect pollen based on floral availability rather than protein content, such as was reported for Apis mellifera (O’Neal and Waller 1984), newer evidence suggests that nutritional quality does influence their foraging decisions, particularly when resources are scarce or colony needs are high. For example, some studies carried out with Bombus and Apis species, which are polylectic and eusocial bees (similarly to the four bee species herein studied), have reported that pollen collected for the provisioning of larvae shows high protein concentration (Robertson et al. 1999; Cook et al. 2003; Ghosh et al. 2020). One study carried out with Apis mellifera reported high protein content in pollen load samples collected directly from hives and demonstrated that pollen types most frequently found in samples have also the highest protein content (Sá-Otero et al. 2009). Nevertheless, more recent studies have expanded on this, showing that protein-lipid ratios also play a critical role in bee foraging preferences and colony health (Vaudo et al. 2020). These findings align with observations reported in this study; Stingless bees showed a preference for high-protein pollen types, such as C. minutiflorum, over other abundant but lower-protein species. Additionally, studies on Stingless bees in various regions, including those by Vossler (2018), have emphasized the strong linkage between pollen preferences and nutritional value. Similar patterns have been observed in Amazonian Melipona species, where Fabaceae and Myrtaceae are frequently dominant in pollen stored inside the meliponaries (Ferreira and Absy 2015). Moreover, studies on Tetragonisca angustula, such as those by Braga et al. (2012) and De Novais et al. (2014), reveal significant variation in pollen use across different ecosystems, further supporting the idea that Stingless bees adapt their foraging strategies to local floral availability and nutritional demands.

Thus, differences in protein content may influence bees’ foraging decisions (Keller et al. 2005), which require a sensorial capacity allowing them to select the type of pollen they collect. Bees sample pollen grains with their mouthparts and antennae during pollen manipulation, a behavior that has been linked to the detection of nutritional quality (Nicholls & Hempel de Ibarra 2017; Ruedenauer et al. 2020). However, preference for pollen is also affected by plant proximity to the hive (Sá-Otero et al. 2009), competition (Ferreira and Absy 2015) and prior foraging experience, factors that may lead to the selection of a familiar pollen type with lower protein concentration (Nicholls & Hempel de Ibarra 2017). Interestingly, Urquizo et al. (2022) demonstrated that Tetragonisca angustula often selects Solanaceae pollen, underscoring the importance of nutrient-rich families in shaping foraging preferences. However, the pronounced reliance on the single genus Capsicum reported in this study contrasts with broader diets, such as those reported for T. angustula by Saravia-Nava et al. (2018). The contrast between those results and the ones reported here may be explained by the influence of both regional and seasonal factors on foraging behavior.

The present results show that bees are indeed able to sample pollen differentially since, in spite of the high abundance and diversity of plant species surrounding the meliponaries they collected few pollen types and showed strong preference for one (i.e., Capsicum minutiflorum) with the highest protein content. Moreover, other species with relatively high protein content and abundance (e.g. Oxypetalum sp., Richardia sp. and Hamelia patens) were collected in much smaller proportions than C. minutiflorum, and three plant species (Asclepia curassavica, Heteropterys umbellate and Serjania sp.) around the hives which had the same or more relative abundance than C. minutiflorum (Asclepia curassavica, Heteropterys umbellate and Serjania sp.) and lower protein content than C. minutiflorum were not found in pollen pots of the hives.

The foraging preferences of Stingless bees presented in this study align with findings obtained in Stingless bees from Brazil, where families like Fabaceae and Myrtaceae frequently dominate the diets (Vossler 2021). However, the dominant use of Capsicum pollen, as reported in this study, is a novel observation because Capsicum has rarely been reported in pollen stores from Brazilian or Amazonian studies. Guimarães et al. (2021) observed that Melipona seminigra shows strong preferences for Fabaceae and Euphorbiaceae in disturbed and natural environments, suggesting that floral availability and nutritional needs drive foraging choices. Similarly, Braga et al. (2012) found that T. angustula relies on diverse pollen sources depending on seasonal floral availability. Nevertheless, none of both studies documented Capsicum dominance. This discrepancy underscores the importance of conducting regional studies to identify how local floral resources influence foraging behavior.

Solitary bees construct a pollen mass inside of each cell where immature develops, as results of several foraging flights (Michener 2007). These pollen masses constitute the food that larvae consume. Thus, longer foraging flights affect female brood productivity and offspring survival; the number of brood cells provisioned per unit time decrease substantially with increasing foraging distance, so energy and time investments resulting from longer foraging distances reduce pollen quantity used for provisioning their offspring (Peterson and Roitberg 2006; Zurbuchen et al. 2010). By the contrary, in Apis mellifera, a eusocial species that builds combs where colonies develop, the amount of pollen is collected in relation to the colony’s need, which depend on the number of larvae and adult bees that constitute the colony (Ghosh et al. 2020). M. rufiventris, S. polysticta, S. depilis and T. angustula are eusocial species that belong to the “corbiculate” group. That is, they have a specialized structure that serves for transport of a large amount of pollen (Michener 1999). We think that these species have similar colony’s needs based on the fact that they exhibit similar pattern of resource use: preference for the same type of pollen in all four species, very high amounts of that pollen in all four species, and high similarity in the composition of pollen between hives of the same species, in three of the four species.

Quantity and size of collected pollen may also influence foraging decisions. Thus, bees may collect higher quantities of nutritionally poor pollen to compensate for its low quality (Wcislo and Cane 1996). This was not the case for M. rufiventris, S. polysticta, S. depilis and T. angustula, according with data herein reported. Thus, only three pollen types categorized as “large” or “very large” (Croton sp., Inga saltensis and Plumbago sp.) were collected by two of the four bee species, and they were collected at very low proportions. Since all species studied are polylectic, they appear to choose the pollen type that offers the highest protein content, independent of its grain size or relative abundance in the home range of bees. Thus, bees do not compensate for low pollen quality by providing their progeny with larger pollen grains and/or higher quantities of nutritionally poor pollen.

Additionally, an interesting observation was the identification of Solanum riparium pollen in pot samples of S. polysticta and S. depilis. Although the amount calculated was very low (6.37 and 0.09, respectively), on the basis of its protein content S. riparium reach the position number 8 out of 24 plant species. In the same line, a recent study reported that pollen of Solanum genus had a higher protein concentration than most of plant genera visited by bees (Pamminger et al. 2019). Taking into account this evidence, we wonder why S. riparium was not more represented in quantity, nor collected by all bee species, because Solanum possess poricidal anthers, it requires pollination from specialized bees that use vibrations to release pollen (i.e., “buz-pollination” bees) and, as far as we know, Scaptotrigona species do not perform buzz pollination. By the contrary, in sample plots of M. rufiventris we did not find S. riparium pollen, which is interestingly because other Melipona species perform buzz-pollination (Nunes-Silva et al. 2010, 2013) and one of them is an efficient pollinator of a Solanum species (Nunes-Silva et al. 2013). Probably the buzz-pollination in Meliponini species is more frequent than is currently known.

Other studies found that crude protein as the protein-to-lipid ratio in pollen are strongly correlated with plant phylogeny and pollinator dependency, as a consequence of selection imposed by pollinator preferences, leading pollen nutrient contents and ratios to co-evolved with the needs of their insect pollinator partners (Ruedenauer et al. 2019). It is also worth mentioning that a specific protein-to-lipid ratio has been observed to affect bumblebees, honeybees and Osmia bees foraging preferences, so it is necessary to deepen into this nutrient proportion in further studies with stingless bees (Vaudo et al. 2016; Vaudo et al. 2020).

The bipartite network of bees and plants showed a high degree of connectance, i.e. the number of observed interactions in relation to the total number of possible interactions. Likely, this is a consequence of the sampling method, as data form pollen pots reflect interactions that are not easy to observe in a plant-centered sampling. This seems to be the case, as we even recorded interactions with plants that were not detected in the sampling area of 160000 m² around each meliponary. Similar high connectance was obtained by Ferreira and Absy (2015), working with pollen data from Melipona species in the Amazonian basin. Low modularity is also expected, given the small size of the network and the inclusion of similar bee species. Nestedness based on overlap and decreasing fills (NODF) has been reported robust to heterogeneous sampling (Rivera-Hutinel et al. 2012), the observed value indicates moderate to low nestedness, suggesting an intermediate level of hierarchical organization. Finally, specialization values may be driven by the overwhelming presence of Capsicum pollen in the pots. Further studies may clarify if this specialization, which seems to be associated to preferences for high quality pollen, remain stable or, more likely, change over the seasons in response to available floral resources.

We would like to acknowledge Reinaldo Lozano for support in the collection and identification of plant species. We are indebted to all meliponiculturists who kindly let us collect the pollen samples, as well as to Fundación PASOS for their logistic support.

This research was funded by the International Science Programme at Uppsala University (ISP), under grant BOL-01. The study was supported by, the Center for International Health of the University Hospital Munich (LMU), and the OH-TARGET project supported by the German Academic Exchange Service (DAAD) with funds from the Federal Ministry for Economic Cooperation and Development (BMZ)–Germany.” and “The APC was funded by the International Science Programme at Uppsala University (ISP), under grant BOL-01 and Universidad San Francisco Xavier ”.