1 Introduction

Every inch of land in the Amazon region presents a fascinating amount of biodiversity. If an observer is interested in plants, he will find up to 300 species per hectare in some areas of the Amazon (ter Steege et al. 2000). As for insects, it is possible to identify more than 480 species in 0.16 square kilometers (Wilkie et al. 2010). If interactions attract the eyes of the observer, one can find up to 35 species of bees pollinating the flowers of a single liana species in the Amazon region (Rech et al. 2011). Since the first trips of naturalists into the Amazon, its enormous biodiversity has raised questions about processes behind the observed patterns. Different lines of research have offered hypotheses to explain how Amazon biodiversity has evolved to reach the patterns that are currently observed. By looking at the interactions between bees and flowers, one may be led to ponder some tentative explanations for the observed diversity. In this chapter, we intend to review the studies previously conducted in the Brazilian Amazon that address the interactions between bees and flowers. We aim to give the reader a broad idea about what is presently known about bees and flowers in the Brazilian Amazon, which is primarily recorded using palynological tools, i.e., pollen grains for taxon identification.

1.1 Origin and Evolution of Plant-Bee Interactions

The earliest bees possibly appeared in the xeric interior of the paleo-continent Gondwana, which was presumably also the area of origin for flowering plants (Raven and Axelrod 1974). The morphological and behavioral diversity found in bees may be one of the evolutionary drivers toward the simultaneous expansion of angiosperms and bees during the mid-Cretaceous (Grimaldi 1999; Danforth and Poinar 2011).

Recently, Cardinal and Danforth (2013), supported by molecular and morphological evidence, have suggested that the origin of bees occurred approximately 123 Mya (113–132 Mya). This hypothesis is contemporary to an incremental expansion and abundance within the Eudicot group, a clade mostly dependent upon bees for reproduction. According to Ollerton et al. (2011), globally, approximately 85% of the angiosperms are animal pollinated, and the local percentage varies from 78% to 94% depending on latitude, with the tropics being a location where animals are more important as pollinators at the community level (Rech et al. 2016).

Stingless bees (Meliponini) are species of social bees that produce honey. Nests are often inside hollow trees and constructed of wax that is secreted from dorsal metasomal glands, which bees combine with resin or propolis collected from plants (Engel and Michener 2013). Due to their permanent nests with large populations, most Meliponini are foragers. In the Amazon region, these bees are the main visitors of numerous plant species and, although they seem to pollinate a large number of them, there is a need for quantifying the importance of such apparent pollination (Roubik 1989). Considering that most of these plants are bisexual or obligate outcrossers, they usually need an animal pollinator to carry pollen from one flower to another (Bawa et al. 1985; Bawa 1990; Roubik 1989; Ollerton et al. 2011; Rech et al. 2016). Thus, the bees are essential for the maintenance of plants in the same way that plants are essential for bee survival, as pollen is the principal food of bee larvae. Social bees present a pantropical distribution (Indo-Australia, the Neotropics, and Africa-Madagascar) composed of continental disjunctions and showing a complex history of vicariance of great antiquity (Camargo 2013; Martins et al. 2014).

Bee diversification was likely rapid, with the earliest roots and stem members of the principal families appearing during and after the mid-Cretaceous and radiating through the latter part of the period (Engel 2001, 2004; Ohl and Engel 2007).

The earliest evidence of stingless bees comes from the latest stage of the Cretaceous (Engel and Michener 2013). The Cretaceous period lasted for approximately 80 million years. According to Michener and Grimaldi (1988), Cretotrigona prisca (Michener and Grimaldi) comprises the only Mesozoic record of the Meliponini, and it is the only definitive apid from the great middle age of Earth. This species showed a significant superficial similarity to the modern species of Trigona s. str. (Michener and Grimaldi 1988), while a detailed examination has suggested that its phylogenetic affinities are more closely related to some Old World lineages (Engel 2000). Hence, the Meliponini are not only morphologically and biologically diverse but also ancient (Michener 2013).

According to Engel and Michener (2013), most of the ideas about stingless bee evolution have been grounded in our knowledge of the hundreds of living stingless bee species. At each moment during the evolution of Meliponini, there must have been numerous, and later hundreds, of species. The two authors mentioned above have emphasized that the nearly 550 extant species are a mere fraction of the total historical diversity of stingless bees (Rasmussen and Gonzalez 2013).

Looking for living species of bees and many other tropical organisms, many naturalists have visited the Amazon region since the early days after the European invasion of the Americas. Many species were described from this region by Adolfo Duke and other traveling naturalists (Hemming 2015). Nevertheless, it was only during the twentieth century when organized collection trips that focused on the study of bees were organized by Drs. João Maria Franco Camargo and Warwick Estevam Kerr. These two researchers and a group of collaborators traveled through the main rivers of the Amazon basin, collecting samples at several locations along the rivers not only of bees (mostly Meliponini) but also samples of their nest materials. These materials were very helpful for the further identification of the pollen that comprises the diet of most bees (Absy et al. 1984; Rech and Absy 2011a, b). It is in the context of these trips that these first authors became engaged in the discussion of the trophic ecology of bees in the Amazon region.

2 The Use of Pollen Analysis in the Study of Bees in the Amazon Rainforest

A number of studies have evaluated the pollen collected by stingless bees in the central Amazon region. Those studies have involved the use of classical protocols for the collection of pollen foraged by bees, such as from pollen pots (Absy et al. 1984; Rech and Absy 2011a, b), from nectar and honey samples (Absy et al. 1980), or more commonly, from the pollen loads of worker bees (Absy and Kerr 1977; Marques-Souza et al. 1995, 1996, 2002; Marques-Souza 1996; Oliveira et al. 2009; Ferreira and Absy 2015). Recently, Ferreira and Absy (2013) established a new collection protocol for the postemergence residue that avoids the acetolysis processes without losing relevant information about the texture and ornamentation of the pollen grain (see Fig. 3.1).

Fig. 3.1
figure 1

Methodologies used in previous studies on pollen grains collected by stingless bees in the Central Amazon: (a) corbicular pollen in the nest entrance, (b) honey samples in storage pots, (c) nectar regurgitated by workers in the nest entrance, (d) postemergence residue, and (e) pollen stored inside the nest (Photo MG Ferreira, and AQ Lima)

Full size image

As noted in the introduction, the Amazon region encompasses a huge diversity of plants, and a large portion of them are used as food sources by bees (Absy et al. 1984; Rech and Absy 2011a, b). This diversity brings a great challenge to researchers interested in studying the diets of bees using pollen analysis. Therefore, having a large pollen library may not be sufficient to obtain satisfactory pollen identification. It is also necessary to have good samples and well-prepared slides to observe enough detail in the pollen grains to achieve good results in the identification of pollen types. This is the main reason that we support the use of the Erdtman’s acetolysis method to prepare pollen samples from the Amazon. Following this technique, the Palynology Laboratory of the National Institute of Amazon Research (INPA) has been conducting research on pollen and bees since 1977 (Absy and Kerr 1977). These studies are divided primarily into two different groups. The first group of papers addressed large nest samples collected during taxonomic expeditions (Absy et al. 1984; Rech and Absy 2011a, b). The second group addressed annual studies, considering a single or few species of bees and the possible pollen sources used throughout the year (Absy et al. 1980; Marques-Souza et al. 1995, 1996, 2002, 2007; Marques-Souza 1996, 2010; Oliveira et al. 2009; Ferreira and Absy 2015).

3 Diversity of Plants, Stingless Bees, and Their Interactions in Central Amazon

Studies of pollen identification have revealed the increasing importance of botanical diversity with regard to bees. However, it is an indirect source of evidence; therefore pollen is a proxy for the potentially mutualistic pollination interaction and should not be used to deduce the interaction consequences. Also, considering the huge available diversity of plants and the similarities within some groups, there is a need for nomenclature and technique standardization with the aim of avoiding misinterpretation of pollen features. Therefore, not only within entomopalynology but also in the field of palynology, there is growing support to use the terminology “type” when referring to a taxon or a morphologically conspicuous group (Joosten and de Klerk 2002; de Klerk and Joosten 2007).

From a study of pollen collected by Meliponini in the Central Amazon region (Table 3.1), it was possible to recognize a wide variety of pollen types collected by 48 species of stingless bees and their interactions with the main botanical families that supply trophic resources (see Fig. 3.2). Using a large-scale approach, a pioneering study in the Amazon was carried out by Absy et al. (1984). For this study, pollen material was obtained from the nests of 24 stingless species distributed over the Baixo Tapajos, Trombetas, Medio Amazonas, and Baixo Uatumã rivers. From all analyzed nests, the authors found 122 pollen types, with a Myrtaceae type visited by 14 species of stingless bees, followed by Attalea maripa (Arecaceae) and Tapirira guianensis (Anacardiaceae), both visited by 13 species. Moreover, these authors emphasize that 18 bee species are generalists which collect the pollen of ten or more botanical species.

Table 3.1 Stingless bees referenced in pollen studies conducted in the Amazon between 1977 and 2015
Full size table
Fig. 3.2
figure 2

Bipartite graph representing the network interactions between stingless bees studied within the main botanical families identified and represented by numbers of pollen types

Full size image

Following a similar approach, Rech and Absy (2011a) studied 10 different species of bees. Inside the nests of those bees, the authors found 78 pollen types, encompassing 70 different genera and 42 plant families. The same authors, in another study considering different sets of 14 bee species (from the genera Partamona, Scaura, and Trigona), found 78 pollen types from 36 plant families stored in the bee nests. After quantification of pollen abundance inside the pollen pots, Rech and Absy (2011b) defined 37 plants as attractive to bees (representation >10%) and 16 pollen types as a result of temporary specialization of the bees (representation >90% of a pollen pot).

The data from Absy et al. (1984) and Rech and Absy (2011a, b) were obtained from standard methods and covered a large terrestrial area. Using the species list from these studies, Sfair and Rech (unpublished data) pooled together a large Amazonian network and checked for “modularity” and “nestedness.” Considering the 74 stingless bees and 334 plant species (when a taxon was identified only to the genus level in different places, each was considered a different species), the “global connectance” was 0.03 and the network was considered nested (N total = 3.55, NODF (Er) = 1.83, p (Er) = 0.00; NODF (Ce) = 2.37, p (Ce) = 0.00), encompassing nine modules (Modularity = 0.56, Mrand = 0.466, sigma Mrand = 0.004). From the nine modules, two modules had only two species each, and the other seven were nested, suggesting a fractal structure for the entire network.

This combined structure, defined as nested compartments, was previously described for plant-herbivore interactions (Lewinsohn et al. 2006) and, as far as we know, is first shown here for mutualistic interaction. This structure probably emerges from the differences in abundance and attractiveness of the different pollen sources (Lewinsohn et al. 2006; Rech and Absy 2011a, b). Corroborating abundance as a possible driver of the nested compartments, as proposed for plant-herbivores, pollen from mass flowering plants was already found to represent >90% of the annual income of a given colony of stingless bee (Hrncir and Maia-Silva 2013). Hence, theoretical models suggest that nested networks tend to reduce competition and allow a greater number of species to coexist (Bastolla et al. 2009). In the same way, modularity is supposed to improve community stability (Fortuna et al. 2010). It may also be hypothesized that the general structure of plant-Meliponini interactions reflects its position halfway in the continuum of mutualism/antagonism, since bees sometimes are pollinators and sometimes mere herbivores—an open avenue for future research.

Contrasting with a large spatial scale, temporal studies usually covering 1 year have focused mainly on economically important stingless bee species. These studies have provided important data on the annual distribution of trophic resources for bees and their behavioral responses to fluctuations in food availability. In the Amazon, these types of studies began in 1977 with a collaboration between Drs. Absy and Kerr (1977). These authors found Inga edulis (Fabaceae/Mimosoideae), Bixa orellana (Bixaceae), and Miconia types (Melastomataceae) were three very important pollen sources for bees. Absy et al. (1980) found another 60 pollen types used by two species of bees. Later studies have confirmed the importance of Myrtaceae, Fabaceae, Melastomataceae, and Arecaceae as pollen sources for Meliponini. Hence, Marques-Souza et al. (1995) found Miconia type (Melastomataceae), Myrcia type, Myrcia amazonica (Myrtaceae), and Leucaena type (Fabaceae/ Caesalpinioideae) were important food sources for two species of bees. Moreover, Marques-Souza (1996) found 30 pollen types (22 genera, 19 families) used by Melipona compressipes, with Cassia type Fabaceae (Caesalpinioideae), Miconia type (Melastomataceae), and Solanum type (Solanaceae) being the most important pollen sources. Studying the bee Trigona williana, Marques-Souza et al. (1996) found Cocos nucifera (Arecaceae), Attalea sp. (Arecaceae), Cassia type (Fabaceae/Caesalpinioideae), Carica papaya (Caricaceae), Bellucia grossularioides (Melastomataceae), Artocarpus altilis (Moraceae), and Stachytarpheta cayennensis (Verbenaceae) being used by these stingless bees.

More recent studies that included new species of bees such as Scaptotrigona fulvicutis have noted some well-known important pollen source plant families, such as Fabaceae (Mimosoideae) and Myrtaceae while including others that have also been reported in previous studies, such as Sapindaceae, which was not as important for this particular stingless bee species (Marques-Souza et al. 2007). Moreover, the latter authors also find another 97 pollen types collected by S. fulvicutis, representing 73 genera and 36 plant families. The most important pollen types collected by these bees were Stryphnodendron guianense (Fabaceae/Mimosoideae) and Schefflera morototoni (Araliaceae).

Using a temporal approach and also considering different stingless bee species (Melipona seminigra merrillae, Melipona fulva, Trigona fulviventris, and Cephalotrigona femorata), Oliveira et al. (2009) recorded 90 pollen types from 31 plant families and 67 different genera. In this study, the most frequent pollen types were Miconia myriantha (Melastomataceae), Leucaena leucocephala (Fabaceae/Mimosoideae), Tapirira guianensis (Anacardiaceae), Eugenia stipitata (Myrtaceae), Protium heptaphyllum (Burseraceae), and Vismia guianensis (Hypericaceae). When considering different synchronopatric bee species, it is important to take into account the potential for food competition among the studied species. Ferreira and Absy (2015) have analyzed the trophic overlap between two common bee species that are managed for honey in the Central Amazon region. In their work, the authors show variation in the trophic overlap over time and by resource seasonality. Again, the most important plant families in which the proportion of shared pollen types raised the trophic overlap between the two bee species were Fabaceae, Melastomataceae, Myrtaceae, and Anacardiaceae.

4 Amazonian Bee Diet, Biology, and Suggested Interactions Potentially Leading to Pollination

The ecological relevance of social bees as pollinators comes from their dependence upon large amounts of pollen and nectar as food sources for constant brood production throughout the year (Simpson and Neff 1981; Roubik 1989; Corbet et al. 1991; Free 1993; Ramalho et al. 2007). By visiting flowers to collect trophic resources, these bees establish a complex network of interactions that may go from mutualism, when pollination really occurs, to antagonism, when resources are collected without any fruit set (Junker and Blüthgen 2010; Santamaría and Rodríguez-Gironés 2015). Therefore, although pollen evidence is a valid indication that an interaction had occurred, it is no guarantee of pollination, and more studies in this area are needed. This is especially true in the Amazon, where the canopy is high above the eyes of observers; therefore recording flower visitation with observation is not always an easy task. In this context, a powerful proxy such as pollen on the body of pollinators or in their nest is a welcome alternative to know the plants visited by each bee species. Once plants are known further studies will reveal whether interaction is mutualistic (resulting in pollination) or only trophic (resulting only in resource consumption).

Not only in the study of interactions that occur among tall native trees but also in the forest-agriculture landscape, pollen evidence of interactions may be of great value. From the 38 most common native plant species commercialized in the open market in Manaus-Amazonas (Rabelo 2012), 23 were recorded in pollen studies as collected by bees. Among the most important plant families found in the markets and in the pollen analysis, we may emphasize the following: Arecaceae (Bactris gasipaes Kunth, Astrocaryum aculeatum Mayer, Mauritia flexuosa L.f. and Euterpe ssp.), Anacardiaceae (Spondias mombin L.), and Myrtaceae (Eugenia stipitata McVaugh and Myrciaria dubia McVaugh).

Although evidence of bee visitation has been noted for many important plants in the Amazon, few have confirmed its importance in the process of pollination. However, many species of stingless bees seem to be promising for pollination in agroforestry systems, especially those of the genus Melipona. Hence, Roubik (1979) suggests that 84% of plants visited by meliponine bees are potentially benefiting from its pollination services. This may be associated with the bees’ ability to extract pollen from plants with poricidal and non-poricidal anthers (Buchmann 1983; Proença 1992), especially several species of Myrtaceae, Fabaceae, Melastomataceae, and Solanaceae families (Roubik 1989; Endress 1996). The release of pollen is attained by the bees through the vibration of their thorax using the flight muscles. The whole process is called pollination by vibration or “buzz pollination” (Buchmann and Hurley 1978).

Furthermore, in other regions of Brazil, there is a growing body of evidence corroborating the great value of stingless bees as potential pollinators of economic importance (Malagodi-Braga and Kleinert 2004; Cruz et al. 2004; Del Sarto et al. 2005). The main interest in stingless bees as pollinators derives from the simple requirements to permit their management. Moreover, different species of stingless bees may be managed within the same area, adding conservation value to this activity. From the observation of different species interactions, it is possible to understand complex and significant ecological interactions (Rech et al. 2013), highlighting meliponiculture as a low impact and enjoyable practice.

Although social bees are true generalists that are able to temporarily specialize on profitable food sources, as noted above they are not always good pollinators. Rech and Absy (2011a, b) found that a large proportion of the species used as pollen sources by bees are not necessarily pollinated by them. Several authors note that social bees are very efficient at collecting pollen, sometimes destroying floral parts and behaving as thieves or robbers, and this may reduce pollen transfer to conspecific stigmas (Renner 1983). Considering this possibility, the authors recommended the association of different sources of evidence when the objective is to look at the mutualistic nature of this interaction. Otherwise, it is possible to confuse a potentially mutualistic network with another one that is purely trophic. Both cases are very interesting from an ecological point of view, but the different interpretations may lead to very different conclusions.

Comparing data from observations with pollen evidence, Rech and D’Apolito (unpublished data) studied plant-bee interaction networks in a Campina area (open scrubland surrounded by Amazonian forest) near Manaus-Amazonas. In total, the authors found 19 bee species interacting with nine plants. When data from pollen were considered together with data from observations, the network was considered nested (see Fig. 3.3). When considering only one source of evidence (observation or pollen), the network structure was not nested, and the interaction numbers were halved. The same study found an intense movement of the bees between open areas and the surrounding forest. The plant Pradosia schomburgkiana and the bee Trigona fulviventris were the most connected points in the network (see Fig. 3.3). Although bees were observed visiting many flowers, when only pollen was taken into account, there was more pollen from the forest on the bodies of the bees than from the plants in the open area. This is probably because bees had preferred larger amounts of pollen offered by trees instead of scattered portions of flowers from shrubs or herbs. Considering the bee perspective, many plants that are actually important as food sources will not be found if only pollen evidence is considered, especially when the plants are visited exclusively for nectar (Villanueva-Gutiérrez and Roubik 2016). In contrast, from the plant perspective, some of those bees are working as antagonists and are just acquiring resources without the outcome of pollination.

Fig. 3.3
figure 3

Different bee-plant interaction networks from areas of Campina (open scrubland areas inside the Amazon forest) produced using evidence from pollen and observation (a) only pollen (b) and only observation (c)

Full size image

5 How to Improve Meliponiculture for Sustainable Development in the Amazon

Knowledge about the plants used as trophic or nesting resources by bees is a basis for the conservation and maintenance of promising species to produce honey in the Amazon region. The accumulated data on plants used by bees in the Amazon show that regardless of their huge generalist potential, stingless bees tend to rely upon a few continuous sources of pollen throughout the year (Roubik and Moreno 1990, 2000, 2013 and in this volume). These plants become factors that determine the maintenance of honey production over time. Studies on Melipona (Michmelia) seminigra merrillae and Melipona (Melikerria) interrupta (see Fig. 3.4), which are the two main species used in the beekeeping industry within the Central Amazon region, reveal a predominance of pollen from Fabaceae, Melastomataceae, Myrtaceae, and Anacardiaceae in the bee diet from várzea (floodplains) and riverside areas (see Fig. 3.5). Moreover, ecological indexes show a large degree of seasonality and a high overlap in the pollen resources used, with a high proportion of a few shared species or pollen types (Ferreira and Absy 2015).

Fig. 3.4
figure 4

Stingless bees that are more frequently managed in beekeeping in the Central Amazon: (a) Melipona interrupta and (b) Melipona seminigra merrillae (Photo MG Ferreira)

Full size image
Fig. 3.5
figure 5

Most of the representative pollen types shared by Melipona seminigra merrillae and Melipona interrupta. Anacardiaceae – Tapirira guianensis (a, b); Melastomataceae – Miconia type (c, d) and Bellucia type (e, f); Fabaceae/Faboideae – Swartzia type (g, h); Fabaceae/Mimosoideae, Mimosa pudica (i, j), Mimosa sensitiva (k, l), and Mimosa guilandinae (m, n); and Myrtaceae – Plinia cauliflora (o, p) (Adapted from Ferreira and Absy 2015)

Full size image

Starting with the many studies developed in the Amazon (Absy and Kerr 1977; Absy et al. 1980; Marques-Souza 1996; Oliveira et al. 2009; Ferreira and Absy 2013, 2015), the two main stingless species used for honey in the region have spread beyond the Central Amazon region. The main species managed in the Central Amazon region, M. seminigra merrillae and M. interrupta, have differing reasons for their popularity (Absy et al. 2013). For the first species, characteristic high honey production and easy adaptation to managed conditions are foremost. The second species, M. interrupta, is becoming popular due to the different types and amazing tastes of the honey it produces within the várzea riverside areas. The understanding of the plants mainly used by these managed species has shown the potential to justify forest preservation and increase fruit production due to better pollination, promoted by meliponine bees (Roubik 1995).

The largest human population of the Amazon is located in the várzea riverside region, and it is where most of the honey from stingless bees is produced. However, the várzea vegetation is less diverse than is that of the adjacent terra firme forest. In this region, both flora and fauna are adapted to the seasonally flooded conditions of the river (Kalliola et al. 1993; Peixoto et al. 2009). According to Moure and Kerr (1950), stingless bees maintain a very close relationship with floodplain areas in the Amazon, where most of the species of the genus Melipona are present, displaying a highly concentrated species diversity. Moreover, Dr. Warwick Estevam Kerr (personal communication) suggests that várzea areas promote the occurrence of tree hollows which provide the natural nest sites for many species of stingless bees.

Some species that are often used by bees to produce honey (both species M. seminigra merrillae and M. interrupta – Ferreira 2014) and feed larvae (Ferreira and Absy 2013), such as Triplaris weigeltiana, are endemic in the várzea riverside region and deserve special attention when sustainable strategies such as beekeeping are considered as an economic alternative for local communities (Wittmann et al. 2013).

6 Conclusions

Palynology has contributed to our knowledge about plant-bee interactions. The evidence produced using pollen grains has helped to shed light on the ecology of this potentially mutualistic interaction and also suggests ways to address questions regarding animal behavior and adaptive processes between plants and pollinators. Such knowledge can and must be applied in conservation and wildlife management programs, especially given the economic importance of bees as pollinators and the widespread perception of their decline in recent years.

Although most of the published studies concerning the plant-bee interaction networks are based on focal observations of flower visitation, recent studies also incorporate evidence from pollen analyses. As we have shown here, depending on the ecological context, this innovation may represent a significant difference in methods used to assess evidence for a particular interaction. Moreover, the rich diversity of plants already described in the Amazon region and the limitation to consistently observing all interactions in detail reinforce the need for different sources of evidence. In this context, palynology, especially as supported by Erdtman’s method, emerges as a powerful tool to access evidence that is more closely related to what animals and plants are actually doing.

Finally, it was with the help of palynological tools that the trophic requirements of many bee species were elucidated, and this knowledge has promoted and improved meliponiculture in the Amazon region. This promising partnership between two areas of knowledge must be emphasized once meliponiculture becomes a sustainable economic activity, promoting not only bee multiplication and preservation but also the improvement of fruit production and the development of agro-ecological activities. It may thus come to encompass social, economic, and ecological dimensions. Meliponiculture is therefore likely to promote a real interchange of popular and scientific knowledge and is one of the ways to maintain forests and humans in this region of the planet, as it was for centuries before colonization.