The selection of candidate meliponine species for targeted cocoa pollination was based on two main species traits, (1) their natural occurrence in the Brazilian Amazon region and (2) their body size.

To determine the geographic range of occurrence of the meliponine species selected as candidates for targeted cocoa pollination (see 2.1), we consulted the online databases SpeciesLink (https://specieslink.net/), Discover Life (https://www.discoverlife.org/), and GBIF – Global Biodiversity Information Facility (https://www.gbif.org/), as well as scientific literature that provided the locations of data collection (Rasmussen and Gonzalez, 2009; Oliveira, F. F. et al., 2013; Brown and Oliveira, 2014; Misiewicz et al., 2014; Alvarez, 2015; Rasmussen and Delgado, 2019; Giannini et al., 2020; Raudales, 2022). In case georeferenced data were missing, whenever possible, we estimated the geographic position of the collected specimens from the verbatim description of the location indicated in the respective references. Distribution maps were elaborated in QGIS 3.32 (Free Software Foundation, Boston, USA).

To estimate whether and to which extent the stingless bees, selected based on their geographic occurrence and size (see 2.1), may de facto be candidates for targeted cocoa pollination, we analyzed the scientific literature available concerning relevant behavioral traits. Particularly important characteristics in this context are the timing of the species’ foraging activity and their stratum preference. The timing of foraging, especially the foraging onset and peak activity, provide insight into whether the bees may visit cocoa flowers during the period of maximum anther dehiscence and stigma/style receptivity. Moreover, the time of foraging onset and the stratum preference indicate the capacity of a species to forage under reduced light conditions and, thus, the potential to visit cocoa flowers under dense shade as is the case in agroforests. Information on the species’ nesting behavior is relevant to estimate the possibility and facility of management.

The intertegular distance (ITD) of the 52 selected species range from 0.5 mm (Leurotrigona gracilis) to 1.4 mm (Frieseomelitta doederleini). Almost half of the species (23 spp.; 44.2%) have an ITD < 1.0 mm ( Figure 2 ; Table 1 ). Concerning body length (BL), species vary from BL = 2.2 mm (Trigonisca extrema) to 6.0 mm (Scaura latitarsis). Minute species (ITD < 1.0 mm, BL ≈ 2–4 mm) are mostly represented by the genera Leurotrigona (3 spp.), Trigonisca (10 spp.), and Dolichotrigona (7 spp.).

To get an insight into the geographic distribution of the 52 Meliponini selected as candidates for targeted cocoa pollination, we evaluated the number of Latin-American states in which they have been documented. For the Brazilian territory, the highest numbers the selected species occur in the states of Amazonas (37 spp.), Pará (35 spp.), Rondônia (33 spp.), and Acre (22 spp.) ( Figure 1B ). Elevated species richness was also found in Mato Grosso (28 spp.) and Maranhão (15 spp.) ( Figure 1B ), both states with transitions between the Amazon Rainforest and the Cerrado ecoregion (tropical savanna). In neighboring countries, high species numbers are associated with Amazon Rainforest biomes, particularly the Peruvian Amazon jungle plain (Loreto: 21 spp.; Madre de Dios: 18 spp.; San Martin: 16 spp.), the Southwestern Amazon (Beni: 9 spp.) and the adjacent Yungas Forest in Bolivia (La Paz: 10 spp.), as well as the Amazon region in Colombia (7 spp.) ( Figure 1B ).

Stingless bees native to the Brazilian Amazon also occur in regions beyond the center of origin of T. cacao. Importantly, 10 of the classified species have been documented in the state of Bahia ( Figure 1B ), which is currently the second largest cocoa producer of Brazil (109,748 tons of cocoa beans in 2022, 40% of Brazilian production) and the largest in terms of area occupied by cocoa plantations (410,185 ha) (for comparison, Pará: 145,995 tons; 152.837 ha; 53% of Brazilian production) (IBGE, 2023).

Although some of the meliponine species identified in our study show a wide geographic distribution, with documented occurrence in more than 10 Brazilian states (7 spp.; 13.5%), most are restricted to one (7 spp.; 13.5%), two (4 spp.; 7.7%), three (14 spp.; 26.9%), or four (7 spp.; 13.5%) states ( Figure 2 ; Supplementary Material 1 ). The species with the largest distribution is Tetragonisca angustula, ranging from southern Brazil, northern Paraguay and Argentina to Belize and southern Mexico ( Figure 3 ).

The light intensity necessary for foraging decreased exponentially with increasing body size (nonlinear regression, exponential decay: R²adj = 0.971). The smallest species for which we found information on light intensity thresholds was Trigonisca pipioli (ITD = 0.8 mm), who initiates foraging, on average, at 72 lux. Bigger species, Tetragonisca angustula (ITD = 1.0 mm) and Frieseomelitta doederleini (ITD = 1.4 mm) have documented light intensity thresholds of 30 and 10 lux, respectively ( Figure 4A ). In open environments or at canopy level, the illuminance necessary for foraging by these three Meliponini are reached already 30 minutes before sunrise. In the understory, however, light intensities are only 1% to 10% of those at canopy level ( Figure 4B ). In these shady environments, light levels necessary for foraging by T. pipioli are attained only at sunrise, whereas F. doederleini may already be active 30 minutes earlier ( Figure 4B ). Importantly, at the time of full anther dehiscence of coca flowers (3.8 hours after sunrise, hours) or during the period of maximum receptivity of the stigmata (6.8 hours) ( Figure 5B ), light levels in the understory are around 5,000 lux ( Figure 4B ) and, thus, far above activity threshold of small bee species ( Figure 4A ).

Concerning foraging activity, we found information on 21 out of the 52 classified Meliponini in 21 studies ( Table 1 ). For 10 species we obtained data on onset and end of colony foraging activity in addition to the timing of flower visits ( Figure 5A ; Table 1 ; Supplementary Material 2 ). Flower visits occurred between sunrise and almost 10 hours after sunrise (earliest observation = 0.02 hours; latest observation = 9.3 hours; average = 5.3 hours) ( Figure 5A ). Colonies of stingless bee species, for which we found more than a single information on flight activity at the nest entrance, initiated foraging, on average on hour after sunrise, had an activity peak 6 to 7 hours after sunrise, and stopped foraging at sunset (F. doederleini: average foraging onset, FO = 1.6 hours; average foraging peak, FP = 6.5 hours; average foraging end, FE = 10.8 hours, N = 18; F. varia: FO = 1.4 hours, FP = 7.1 hours, FE = 10.9 hours, N = 4; T. angustula: FO = 1.7 hours, FP = 6.9 hours, FE = 11.1 hours, N = 21) ( Figure 5A ).

For the majority of Meliponini classified in our study, we found little information on the variety of plants visited, given that most species had been observed only in studies focusing on a specific pant/crop ( Figure 6A ; Table 1 ; Supplementary Material 2 ). The known food source spectrum broadened considerably in species for which palynological studies were available (F. flavicornis, F. longipes, F. portoi, F. silvestrii, F. trichocerata, F. varia, Nannotrigona minuta, Nogueirapis butteli, Plebeia minima, Scaura latitarsis, S. tenuis, T. angustula) ( Table 1 ). Floral resource richness was significantly associated with the number of studies available (sigmoidal nonlinear regression: R² adj = 0.78; P < 0.0001) ( Figure 6A ), which highlights the need for additional investigations to get a more complete picture of the niche breath a given bee species.

The Meliponini with the highest documented food spectrum were T. angustula (367 spp.), F. varia (197 spp.) F. longipes (98 spp.) ( Table 1 ; Supplementary Material 2 ). In most meliponine species/genera, for which at least 10 plant resources are specified in the literature, trees were the preferred stratum, accounting for between 22.7 to 60.5% of visited plants (average = 39.9 ± 10.4%) ( Figure 6B ). However, putative understory strata combined (herbs, subshrubs, shrubs, and lianas) amounted for between 33.3 and 72.7% of visited plants (average = 49.6 ± 11.4%) ( Figure 6B ).

We found information on the nesting behavior of 38 out of the 52 classified Meliponini in the scientific literature (9 studies). Of these, 21 species build their nests in tree cavities, two nest in the soil, and 15 utilize multiple sites. The nesting habits of 14 species have remained unknown to this date ( Table 1 ).

In the present study, we highlight 52 small stingless bee species native to the Brazilian Amazon region as potential cocoa pollinators. Despite their reduced body size (intertegular distance, ITD = 0.5–1.4 mm; body length = 2.2–6.0 mm), the foraging area of these Meliponini may range between a couple hundred meters to almost one kilometer around their colonies (Grüter and Hayes, 2022). As far as is known, stingless bees collect floral resources during the day (Teixeira and Campos, 2005; Streinzer et al., 2016; Figure 5A ). Although bigger species may initiate foraging at nautical twilight levels, at an illuminance below 2 lux, the eye-size and associated visual physiology of small species restrain their activity to higher light levels (Streinzer et al., 2016; Aguiar et al., 2023). It has been argued that the strongly reduced light regime owing to shading in cocoa plantations, particularly in agroforests, limits the visitation of T. cacao flowers by bees (Young, 1985b). However, foraging of small Meliponini may occur at an illuminance corresponding dawn (around 80 lux; Figure 4B ). Thus, at least from the viewpoint of visual physiology, they should well be capable to visit cocoa flowers after full dehiscence of the anthers and during maximum receptivity of the stigmata even under heavy shade cover (100% dehiscence at 3.8 hours after sunrise: approx. 111,000 lux at canopy level; approx. 5,000 lux in the understory; maximum receptivity at 6.8 hours after sunrise: approx. 151,000 lux at canopy level; approx. 5,000 lux in the understory; Figure 4B ). The bees’ potential to forage under reduced light regimes is corroborated by the finding that, despite a certain preference for trees, between 20 and 60% of the naturally exploited food sources are understory vegetation, including herbs, subshrubs, shrubs, and lianas ( Figure 6B ). Depending on the height above ground, at which these plants bloom, and the trees’ crown area, the light levels may be less than 1% of the illuminance at canopy level (Yoda, 1974; MacDougall and Kellmann, 1992; Matuso et al., 2021). Thus, food collection at understory vegetation, particularly at herbs and subshrubs in forest environments, relies on the capacity to forage under low light conditions. Although some meliponine species prefer sunlit food patches (Biesmeijer et al., 1999), this is an issue of thermal physiology, associated with body coloration rather than size (Pereboom and Biesmeijer, 2003; Ramos et al., 2024). In any case, bees that are capable of foraging in the sun without overheating may take advantage of flowers located on thin branches in the crown, which is common in ancestral forms of T. cacao in the wild (Young et al., 1987). Bees that preferentially forage at shaded patches to avoid overheating, by contrast, may visit flowers on lower branches even under densely shaded conditions as is the case in cocoa agroforests (Young et al., 1987).

Whether or not stingless bees are, de facto, pollinators of cocoa flowers, has remained an open debate. Early studies showed a low efficiency of Tetragonisca angustula as cocoa pollinator under natural conditions (0.8% of visited cocoa flowers that had not been knocked off during the visit produced fruit; control: 0.2% of flowers sheltered from floral visitors produced fruit; Young, 1981) and postulated that foragers of this meliponine species are mere pollen thieves (Young, 1981, 1985). More recently, however, it has been documented that foragers of T. angustula pass through the staminode fence and enter in contact with the stigmatic surfaces of cocoa flowers (Jaramillo et al., 2024), which gives reason to believe that at least some Meliponini may not only visit but also pollinate T. cacao. In contrast to midges, stingless bees have specialized body structures, such as the corbiculae, which allow them to effectively collect and transport large amounts of pollen, therewith facilitating pollination (Michener, 2007; Martins et al., 2014; Engel and Rasmussen, 2020). Given that social bee foragers do not collect floral resources for their own sustenance, but to provide nutrients for the colony, they typically return to previously known food sources, where they visit numerous flowers (Soria, 1975; Young, 1981). The extended duration of these visits for food collection, coupled with the adherence of pollen grains to their bodies when in contact with pistils, make bees key agents in systems depending on cross-pollination as they move between flowers (Westerkamp, 1996).

Although effective pollen deposition on the female parts of cocoa flowers by small stingless bees cannot be ruled out, it has not been observed so far. Howsoever, pollen grains littered by bees while handling the anthers may sediment in the petal hoods, thereby facilitating eventual adhesion to midges or other minute arthropods crawling through the flowers and, thus, pollen transfer to the stigma (Young et al., 1984; Young, 1985a). In addition to promoting self-pollination (thrip-mediated self-pollination of Ranunculus secleratus and Potentilla rivalis: Baker and Cruden, 1991; ant-mediated self-pollination of Blandfordia grandiflora: Ramsey, 1995), the coincidental deposition by crawling flower visitors has been shown to increase the seed set of self-incompatible plants, yet only in association with flower visits by bees (for Conospermum undulatum: Delnevo et al., 2020). Given the high variety and abundance of minute arthropods in cocoa flowers (Bigger, 2012; Toledo-Hernández et al., 2017), joint pollination involving different species or functional groups appears a promising topic for future research.

A major drawback concerning the importance of stingless bees for commercial pollination of T. cacao is their reduced attraction to the flowers, which may be the result of artificial selection and cloning for agricultural purposes (Young and Severson, 1994). Thus, Meliponini must be “convinced” to visit the flowers of cultivars in cocoa plantations and agroforests. A possibility to enhance floral visits by social bees to crop flowers is through targeted pollination (Farina et al., 2023a). Prior to introducing managed colonies into crops, they are fed with sugar syrup scented with the odor of the target flowers, or a synthetic odor mixture that mimics the bouquet (Farina et al., 2020). During repeated exposure to the scented food, bees learn to associate the odor with reward and establish specific olfactory memories (Farina et al., 2005). The olfactory information gained inside the nest biases the foraging decisions of individuals in the field (Arenas et al., 2007, 2008), therewith enhancing the colonies’ visitation rate to target crop flowers (Farina et al., 2020, 2022, 2023b). Importantly for the relatively nectarless cocoa flowers (Young et al., 1984), the success of this protocol for managed pollination has been documented also for crops whose flowers offer little or no nectar reward like pear (Farina et al., 2022). Feeding colonies with scented syrup, presumably, activates mainly nectar foragers (Farina et al., 2023a). However, some foragers may switch to pollen collection, particularly if the “promised” nectar source has a lower productivity than expected by the bees based on their experience in the nest (Arenas and Kohlmaier, 2019).

These studies, all using honey bees (Apis mellifera) as model organism, highlight three conditions that must be met to guarantee the success of targeted crop pollination. (1) The bee species in question needs to be manageable. (2) The bees must show a certain level of olfactory associative learning. (3) The bees must show the capacity to transfer scent information learnt inside the nest to the foraging context. Luckily, stingless bees comply with the first two of these requirements, while the third has remained largely understudied so far. (1) Unlike with A. mellifera, there is no standard hive model for stingless bees. Traditionally, many beekeepers, who manage stingless bees, just leave the colony inside the original tree trunk (Crane, 1999; Quezada-Euán, 2018), which limits the possibilities of manipulation (Contrera et al., 2011). To improve the practicability of meliponiculture, several types of hives have been developed over the past fifty years (Nogueira-Neto, 1970; Crane, 1999; Contrera et al., 2011), varying in in size and format to optimize honey extraction and colony division of a particular species (Oliveira, R. C. et al., 2013; Barbieri et al., 2019). For the Meliponini classified in our study as potential cocoa pollinators, we found references for the nesting behavior of 38 species, 26 of which species build their nests, at least facultatively, in tree cavities ( Table 1 ), which facilitates their transfer to rational hives. Although it is possible to keep soil nesting species in nest-boxes (Imperatriz-Fonseca, 1973, 1978; Ribeiro, 2008), new colonies are usually established in underground cavities (Imperatriz-Fonseca, 1990). Thus, attracting these species to trap-nests, a common practice by beekeepers to obtain new colonies (Oliveira, R. C. et al., 2013), is, at the very least, difficult. For only few of the Meliponini highlighted in our study, successful establishment of colonies under managed conditions has been documented so far (Frieseomelitta varia, F. longipes, F. silvestrii, F. trichocerata, Leurotrigona muelleri, Nannotrigona melanocera, Plebeia minima, Scaura latitarsis, S. longula, S. tenuis, Tetragonisca angustula: Nogueira-Neto, 1970; Mateus et al., 2013; Oliveira, F. F. et al., 2013; Barbieri et al., 2019; Leão et al., 2024). However, species usually thrive well when kept in nest-boxes of adequate size and in appropriate thermal conditions (CMS, MH, VLIF, personal observations). (2) Similarly to honey bees, stingless bees show the capacity to establish long lasting olfactory memories when trained on a specific scent (Aguiar et al., 2023). Particularly interesting for protocols of targeted crop pollination is the possibility to induce these olfactory memories through feeding the colonies with scented sucrose solution inside the nest (Mc Cabe and Farina, 2009, 2010). To this date, unfortunately, only two meliponine species have been investigated concerning the acquisition of olfactory information through in-hive feeding, one of these, however, a species classified by our study as potential cocoa pollinator (T. angustula: Mc Cabe and Farina, 2010). (3) It is most likely that foragers use learnt scent-reward contingencies for their foraging decisions. At least when trained to artificially scented food at outdoor feeders, or when exposed to a particular odor in the laboratory, foragers show a preference for this scent in the field (Mc Cabe and Farina, 2010; Roselino and Hrncir, 2012). Crucial, in any case, for a potential targeted cocoa pollination is the fact that stingless bees, similarly to honey bees, may switch from nectar to pollen foraging in case the nectar source becomes unavailable (shown in Plebeia tobagoensis: Hofstede and Sommeijer, 2006).

An important requirement for the success of any forms of directed pollination by stingless bees is an appropriate habitat quality that permits the maintenance of managed colonies within or near crops and, moreover, allows for gene exchange with natural populations. Small Meliponini, in particular, are highly vulnerable to deforestation (Mayes et al., 2019), since their reduced foraging range limits the access to food when resources become scarce or disconnected (Araújo et al., 2004; Greenleaf et al., 2007; Hrncir, 2022). Thus, at landscape-level, protection of continuous and interconnected forest areas next to crop plantations is essential to ensure the survival of bee populations (Campbell et al., 2018). To increase the biodiversity of small bees, the conservation of natural habitats and the sustainable management of agricultural landscapes are fundamental for providing sufficient resources for natural pollinators (Ulyshen et al., 2023). These resources include suitable nesting habitats (e.g., tree cavities and soil substrates) as well as plants that bees may use as additional pollen and nectar sources (Kremen et al., 2007; Ricketts et al., 2008).

In Brazil, cocoa is mostly cultivated in agroforests (Cuenca and Nazario, 2004), which are a valuable resort for managed stingless bee colonies. Agroforestry is a land management approach that involves the incorporation of trees and shrubs into agricultural and rural landscapes, aiming at enhancing productivity, profitability, diversity, and overall ecosystem sustainability (Kang and Akinnifesi, 2000; Nair, 2007, 2011; Jose, 2009; François et al., 2023). Brazilian agroforests may play an important role in stingless bee conservation due to their high floral and structural diversity (Jose, 2012), which increases the availability of both potential food sources and nesting options (Jha and Vandermeer, 2010; Delgado et al., 2022) and improves the microclimatic conditions for foraging and development (Tscharntke et al., 2011; Betrup et al., 2019). Moreover, agroforests contribute to biodiversity conservation through enhancing landscape connectivity between habitat remnants of natural habitat (Moguel and Toledo, 1999; Betrup et al., 2019). The consequently reduced pressure on biodiversity promotes functional services, including pollination (Toledo-Hernández et al., 2021) and pest control (Andow, 1991; Tscharntke et al., 2011; Ferreira, D. F. et al., 2023), which, eventually, lead to increases in crop yields (Betrup et al., 2019; Sabino et al., 2022).

In Brazil, cocoa cultivation plays a vital role in family farming, providing income and subsistence for smallholder farmers (Cuenca and Nazario, 2004). Enhancing cocoa production through targeted crop pollination, therefore, not only increases yields, but also improves the livelihood of rural communities (Donald, 2004). Although our study highlights the potential of small Meliponini as pollinators of T. cacao, it reveals considerable gaps-of-knowledge. So far, we know only little concerning the nesting behavior and foraging activity of most neotropical species, particularly the smallest ones that, owing to size, might be most suitable for cocoa pollination ( Table 1 ). We know even less about the Meliponini of Africa and Southeast Asia, which currently share more than 80% of the global cocoa production (Suri and Basu, 2022). Intriguingly, a survey of potential pollinators of T. cacao in Ghana reported visits of cocoa flowers by Liotrigona bouyssoui (as L. parvula) (Frimpong et al., 2009), a minute stingless bee species with a body length (BL) of less than 4 mm (Grüter, 2020). Moreover, pollen of T. cacao was detected in the honey of small Indonesian Meliponini, Tetragonula laeviceps (BL = 4.0 mm) and Heterotrogona itama (BL = 6.1 mm) (Anggadhania et al., 2020; Jayadi and Susandarini, 2020; for body size see Purwanto et al., 2022). These findings highlight the global potential of targeted cocoa pollination by stingless bees. Yet, studies are urgent to determine the physiological and behavioral requirements for targeted crop pollination, such as the capacity of olfactory associative learning and the capability to transfer scent information learnt inside the nest to the foraging context. From our list of potential cocoa pollinators, scent learning-studies have only been performed on Tetragonisca angustula (Mc Cabe and Farina, 2010). Thus, important next steps should comprise detailed studies particularly on foraging behavior and olfactory learning to validate the potential of the meliponine species from our list as cocoa pollinators. In sequence, managed colonies of the most promising species should be introduced into cocoa plantations, particularly in agroforests, to quantify the effect of their pollination services for crop yields. Studies investigating the efficiency of such directed pollination by stingless bees, so far, documented a positive effect on the yield of several economically important crops, including acai (Muto et al., 2020), apple (Viana et al., 2014), macadamia (Heard, 1999), and watermelon (Layek et al., 2021). Moreover, significant contributions of pollination by Meliponini to the fruit set in avocado (Can-Alonzo et al., 2005), coconut (Heard, 1999), and coffee (Klein et al., 2003) point to the potential of this social bee group for directed crop pollination. At local level, the incorporation of managed stingless bee colonies into family farms and agroforests may provide an efficient strategy to secure crop productivity, smallholder farmers’ income and biodiversity conservation.