Bee Diversity and Solanum didymum ( Solanaceae ) Flower – Visitor Network in an Atlantic Forest Fragment in Southern Brazil

Brazil’s Atlantic Forest biome is currently undergoing forest loss due to repeated episodes of devastation. In this biome, bees perform the most frequent pollination system. Over the last decade, network analysis has been extensively applied to the study of plant–pollinator interactions, as it provides a consistent view of the structure of plant–pollinator interactions. The aim of this study was to use palynological studies to obtain an understanding of the relationship between floral visitor bees and the pioneer plant S. didymum in a fragment of the Atlantic Forest, and also learn about the other plants that interact to form this network. Five hundred bees were collected from 32 species distributed into five families: Andrenidae, Apidae, Colletidae, Megachilidae, and Halictidae. The interaction network consisted of 21 bee species and 35 pollen types. The Solanum-type bee species with the highest number of interactions were Anthrenoides sp. 1, Augochlora sp. 2, and Augochloropsis notophos, representing 71.78% of their interactions. Augochloropsis notophos and Augochlora sp. 2 were the only common species in the flowers of S. didymum. Given the results of our study, we conclude that Solanum is an important source of pollen grains for several native bee species, mainly for the solitary species that are more diverse in the south of Brazil. Moreover, our results indicate that bees from the families Halictidae (A. notophos, Augochlora) and Andrenidae (Anthrenoides) are the pollinators of S. didymum.


Introduction
In the Atlantic Forest, the most frequent pollination system is accomplished by bees, which mainly look for pollen and nectar in the various plant species, which are used as nutritional sources [1].This is a mutualistic relationship, in which the plants provide food for pollinators, and in return receive the benefits of pollen transfer [2].
Several groups of bees have morphological, physiological, and behavioral adaptations that enable them to collect floral resources efficiently.Among these adaptations, there is a specialized behavior for pollen collection by the vibration of flowers with poricidal anthers [3], such as the Solanum [4].This feature requires pollinators to show vibratile behavior (buzz-pollinating) [3] to mechanically remove a larger amount of pollen grains, and a more efficient pollination of their flowers [5].Its flowers do not offer nectar as a reward, and are therefore dependent on buzz-pollinating bees.Only some bees are capable of vibrating the flowers (buzzing) in order to collect their pollen [6].
The genus Solanum is one of the largest among the angiosperms, comprising approximately 1500 species and maintaining a wide diversity in South America [7].As most Solanaceae, these plants have an important role as colonizers of open and disturbed sites such as pastures, clearings, forest edges, and roadsides, thus contributing to the regeneration process of altered areas [8].
Solanum didymum Dunal is one of the Solanum species that occurs in the understory of the mixed Araucaria Forest (Araucaria Forest), an Atlantic Forest plant physiognomy [9] that is one of the main plant formations in southern Brazil [10].Araucaria Forest communities are characterized by the presence of woody species that are phytogeographically related to Austral-Antartic and Andean floras [11,12].The most physiognomically important tree species is Araucaria angustifolia.Some other typical species found in these forests are Podocarpus lambertii, Drimys brasiliensis, Dicksonia sellowiana, and several species of Myrtaceae, Melastomataceae, Lauraceae, and Solanaceae.
The only information in the literature about Solanum didymum is that it is a selective hygrophytic and heliophytic bush [13].Therefore, by using palynological studies, the aim was to determine the interaction network between native bees and S. didymum flowers in a fragment of a mixed Ombrophilous Forest (Araucaria Forest), and also their indirect interaction with the other plants in the same forest.
The network theory has substantially helped understandings of the community structure of mutualisms by revealing some invariant properties of these systems [14], and has been widely used to study plant-pollinator interactions [15], as it provides a consistent view of the structure of plant-pollinator interactions [16].Besides this analysis, this study aimed to obtain an understanding of the relationship between floral visitors' bees and the pioneer plant S. didymum in a fragment of the Atlantic Forest, using palynological studies, and with this tool also to know the other plants that interact to form this network.
According to the Köppen climate classification, the climate is the Cfb type, moist with no dry season, and with an average temperature in the warmer months below 22 • C and severe frosts [18].

Sample Design
As S. didymum is a small herbaceous plant that occurs in groups, we studied one of these groups in the understory of the Araucaria Forest fragment of the Araucarias Municipal Park in Guarapuava (PR), southern Brazil.The bees were captured three times per week from September to November 2013, which corresponds to the period of flowering of S. didymum in this park.The collection time was established after observing the floral receptivity, which occurred from 9am to 4pm.The bees were collected directly from the flowers using transparent plastic containers.

Pollen Collection
Pollen grain material was retrieved from the floral visiting bees' bodies using 70% alcohol.From this material, the slides were prepared for observation using an optical microscope from the Instituto de Botânica (São Paulo state) based on the method proposed by Erdtman [19].
Bees from the same species collected on the same day and time were grouped.The pollen grains were simultaneously removed from their bodies to prepare one slide to represent the occurrence of that species on that day and at that time.Plants in a flowering state, or presenting flower buds that were found in the proximity, were collected to prepare the slides, which helped identify pollen types found on the bees' bodies.
The pollen types found were analyzed qualitatively and quantitatively, and the first 1000 grains per slide were counted.Afterwards, the percentage occurrence of each pollen type was observed, according to the Vergeron classifications [20].
The identified and acetolyzed pollen grains prepared for light microscopy (LM) were digitally photographed using an OLYMPUS BX 50 photomicroscope coupled to a video camera, and a computer using the CellSens software for Windows from the Botanical Institute of São Paulo.

Assemblage of Floral Bee Visitors of S. didymum
The frequency of occurrence (FO) and species dominance were calculated for each bee species obtained.Frequency of occurrence is the percentage of the number of collections with a given species, and was calculated as FO = (F/N) × 100 [21], where "F" is the number of collections with the species, and "N" is the total number of collections performed.The bee species were classified as primary (FO > 50%), secondary (FO = 25-50%) or accidental (FO < 25%).
The species dominance of bees (D) was calculated as D = (d/n) × 100 (Palma, 1975), where "d" is the abundance of a specific species, and "n" is the total abundance.The species were classified as dominant (D > 5%), accessory (D = 2.5-5%), or accidental (D < 2.5%).According to Palma [22], the FO and D indices, when used together, group and determine the species as common, intermediary, or rare.

Interaction Network
The interaction network was built between the bees found in S. didymum flowers, the pollen types found in PMA and in these bees' bodies, and the pollen types of plants that occur in this park.
The network size was calculated through the formula M = ap, where M is the maximum number of possible interactions, a is the number of bees, and p is the number of pollen types.
Connectance (C = E/(ap)) measures the ratio between the number of observed interactions (E) and the number of possible interactions, given that p is the number of plants (pollen types), and a is the number of bee species in the network.To transform these values into percentages, the results were multiplied by 100 [23].The connectance is a qualitative measure of the network specialization, and also represents the density of interactions in a network.Thus, a highly specialized community presents a low value of connectance [23,24].
In qualitative interaction networks (binary matrix), the average degree ( The average degree ҟ of interaction between plant specie degree distribution was heterogeneous, following the truncated po with bees (ҟ = 7.2).
The bipartite graph (Figure 4) and the NODF = 64.5 index ind with most of the interactions concentrated in a few species.
) was determined, which is the average number of observed connections for species of plants or bees [25,26].From the adjacency matrix, which was built with data regarding the presence and absence of plant and bee species, Eulerian graphs were prepared using R software version 3.0.1.The evaluation of the network's degree of nesting, based on the adjacency matrix, was performed using the metrical base NODF (Nestedness Metric Based on Overlap and Decreasing Fill) [27] for more consistent statistical properties.The analyses were produced using ANINHADO software version 3.0 [28].

Results
Five hundred bees from 32 species distributed in five families were collected from its flowers: Andrenidae, Apidae, Colletidae, Halictidae, and Megachilidae.The highest species richness was for the families Halictidae, Andrenidae, and Apidae (Table 1).
The families with the highest number of individuals were Halictidae and Andrenidae.Most of the species were classified as rare.Only Augochloropsis notophos Vachal and Augochlora sp. 2 were common species in the flowers of S. didymum.Anthrenoides sp. 1, Anthrenoides sp. 2, Ceratalictus sp., Augochloropsis sp.A and Augochlora sp. 1 were classified as intermediate species (Table 1).
Out of the 500 bees collected, only 306 participated in the pollen analysis, because they contained a pollen load in their bodies.Two hundred and forty nine slides comprised the network analysis, which recorded 64,442 pollen grains distributed in 35 pollen types of 26 families (Figures 1-3).
The interaction network consisted of 21 bee species (a) and 35 pollen types (p), which enabled 735 interactions (M), and 147 interactions observed between the plant species (p) and the bee species (a) (c = 0.20 (20%)).
Sixteen bee species (66.67%) had interactions with five or less pollen types, and seven (33.33%) bee species had seven or more interactions.Among the most abundant floral bee visitor species of S. didymum that comprise the network, Anthrenoides sp. 1 (Andrenidae/intermediate) interacted with 23 pollen types found, followed by Augochlora sp. 2 (Halictidae/common) and Augochloropsis notophos (Halictidae/common), which interacted with 22 and 20 pollen types, respectively.From the species considered intermediate, Augochloropsis sp.A (Halictidae) interacted with 16 pollen types, Anthrenoides sp. 2 (Andrenidae) with seven types, and Augochlora sp. 1 (Halictidae) with five types.For the rare species, the number of interactions was below five, except for Augochloropsis cfr sparrilis Vachal (Halictidae), which interacted with seven pollen types, and Halictilus sp.(Halictidae), which interacted with five types.There was no interaction of any bee species with all of the pollen types found.
The average degree (ҟ) of interaction between plant species was 4.26, and the network of degree distribution was heterogeneous, following the truncated power law, and the same happened with bees (ҟ = 7.2).
The bipartite graph (Figure 4) and the NODF = 64.5 index indicated a nested network pattern, with most of the interactions concentrated in a few species.Concerning the plants, out of the 35 pollen types found, only three interacted with 10 or more bee species, and Solanum was the pollen type that had the highest number of interactions with the different bee species, and therefore a more general type.Nine pollen types (T.Ilex, T. Butia, T. Stenachaerum, T. Jaquemontia, T. Stylosanthes, T. Senna, T. Struthanthus, Pytolacaceae, and T. Monocotyledon) had interactions with only one bee species, and therefore was more specialized.
The average degree ( Concerning the plants, out of the 35 pollen types found, only three interacted with 10 or more bee species, and Solanum was the pollen type that had the highest number of interactions with the different bee species, and therefore a more general type.Nine pollen types (T.Ilex, T. Butia, T. Stenachaerum, T. Jaquemontia, T. Stylosanthes, T. Senna, T. Struthanthus, Pytolacaceae, and T. Monocotyledon) had interactions with only one bee species, and therefore was more specialized.
The average degree ҟ of interaction between plant species was 4.26, and the network of degree distribution was heterogeneous, following the truncated power law, and the same happened with bees (ҟ = 7.2).
The bipartite graph (Figure 4) and the NODF = 64.5 index indicated a nested network pattern, with most of the interactions concentrated in a few species.
) of interaction between plant species was 4.26, and the network of degree distribution was heterogeneous, following the truncated power law, and the same happened with bees ( cerning the plants, out of the 35 pollen types found, only three interacted with 10 or more ies, and Solanum was the pollen type that had the highest number of interactions with the bee species, and therefore a more general type.Nine pollen types (T.Ilex, T. Butia, T. erum, T. Jaquemontia, T. Stylosanthes, T. Senna, T. Struthanthus, Pytolacaceae, and T. tyledon) had interactions with only one bee species, and therefore was more specialized.
average degree ҟ of interaction between plant species was 4.26, and the network of istribution was heterogeneous, following the truncated power law, and the same happened s (ҟ = 7.2).bipartite graph (Figure 4) and the NODF = 64.5 index indicated a nested network pattern, st of the interactions concentrated in a few species.

= 7.2).
The bipartite graph (Figure 4) and the NODF = 64.5 index indicated a nested network pattern, with most of the interactions concentrated in a few species.

Discussion
The bees of the family Halictidae were the richest and most abundant in S. didymum flowers, and corresponded to the common species in their flowers (A.notophos and Augochlora sp. 2).From the five species classified as intermediate, three belong to this family (Ceratalictus sp., Augochloropsis sp.A, and Augochlora sp.1), and two belong to the family Andrenidae (Anthrenoides sp. 1 and Anthrenoides sp. 2).The network analysis results also show a strong interaction between these bees and the pollen type Solanum.In several studies on bee communities in southern Brazil, the family Halictidae has shown to be the most diverse, or similar in diversity to that of the family Apidae [29][30][31].
Gonçalves and Melo [30] conducted a standardized survey of bees visiting several species of blooming plants in an area covered by natural grasslands in the Vila Velha State Park, State of Paraná, southern Brazil, from October 2002 to October 2003.Augochlorini and Halictini presented 32 and 26 species, respectively.According to these authors, the numbers are mainly due to the genera Augochlora, Augochloropsis, and Dialictus, which have a high number of species and dense populations.
In our study, Anthrenoides sp. 1 was the most generalist species, interacting with 23 pollen types.This result is interesting, because it suggests that these bees are polylectic, whereas Freitas and Sazima [32], in high-altitude areas in the southeast of the Brazil, found different results.They observed that the main pollinators of Viola species were female bees belonging to the genus Anthrenoides, which search mainly for pollen [32].According to them, these bees seem to be oligolectic, and obtain large amounts of pollen from Viola by vibrating the flowers or by moving the hook repeatedly back and forth.From the 124 plant species surveyed as part of a study of pollination biology at the community level in the high-altitude grasslands of Serra da Bocaina in southeast Brazil, [33], these bees were observed only on flowers of V. cerasifolia A.St.-Hil.Anthers of Viola, which have dehisce longitudinally, as a function as a single poricidal anther, due to the intimate contact of anthers and the arrangement of the connective projections [32].
Other species that were generalists in our study were Augochlora sp. 2 (22 interactions) and A. notophos (20 interactions), which also interacted strongly with S. didymium.Species of Augochloropisis were in the flowers of Solanum paniculatum L. in a study carried out in the northeast of Brazil, but not those of the Augochlora genus that interacted with several families of plants, least with the Solanaceae species [34].
On the island of Santa Catarina, also in southern Brazil, the broad spectrum of pollen sources characterized Augochlora (A.) esox (Vachal) as a polylectic forager.The pollen from the female's scopae and from the brood cells included 18 distinct types [35].Pollen from the families Asteraceae and Melastomataceae were the most frequent, and those from Solanaceae were not registered.For Moure and Hurd [36], the polylectic habit is a common trait in halictid bees, and may help to explain the wide geographical distribution of this species.
If the bees of the Halictidae family are actually polylectic, it seems that there are exceptions, because Dalmazzo and Vossler [37] used pollen analyses in the nests of Augochlora amphitrite and bee flower visitation, and found that the Ludwigia type dominated pollen counts.Based on these results, they concluded that Ludwigia is the most important pollen source in the diet of this bee.
Ludwigia was also one of the pollen types present in the body of the bees collected in the flowers of S. dydimum, but in a small quantity.Their flowers produce pollen grains in large tetrads, with the presence of viscin threads [38], and only specialist bees with long rigid hairs-with few branched in their scopa-and rapid body and leg movements can be the potential pollinators of these plants [39,40], which explains the presence of oligolectic bees in the flowers of Ludwigia in the south and southeast of Brazil [41][42][43][44][45][46].
The relationship of Ludwigia with oligolectic bees was verified in southern Brazil by Buschini et al. [47] when they analyzed the pollen in the Megachile (Moureapis) sp nests.The pollen of Ludwigia peruviana (L.) H. Hara (36%) and Ludwigia sericea (Cambessides) H. Hara (63.6%) accounted for 99.6% of the total amount of pollen collected, and they concluded that this is an oligolectic bee species, and that these plants are their important pollen source.However, when Gonçalves and Buschini [48] collected bees in the flowers of these plants and analyzed the pollen in their bodies, they saw that these plants are generalist in their interactions with bee species, and that the same occurs for most of the bees in relation to these plants.Augochlora amphitrite was among the generalist bee species in their network, interacting with many types of pollen, and not just with Ludwigia type, as observed by Dalmazzo and Vossler [37].Only the Megachilidae species, which largely depend on the Ludwigia pollen grains, have oligolectic habits.
Another oligolectic bee species that occurs in Guarapuava and has only been recorded from Brazil [49] is Centris tarsata Smith [50,51].This species uses exclusively Solanum pollen to feed offspring.An analysis of the pollen spectrum based on samples from its nests indicated the presence of 20 pollen types, but the family Solanaceae was visited most often (71%).Solanum americanum Mill.(28.6%) and Sol.variabile Mart.(42.4%) were the primary pollen sources for C. tarsata in Guarapuava [52].No pollen of S. dydimum was found, because this bee nested only in grassland areas [50,51], and, as previously commented, S. dydimum occurs only in the forest areas in this park.
Another network study carried out in the Guarapuava region, mapping the interactions between Eugenia uniflora L. flowers and floral visitor bees, found that although the most abundant species was Apis mellifera (Linnaeus), the subfamily with the highest number of species was Halictidae (23 species).Andrenidae was represented by just one species (Anthrenoides paolae (Urban)) [53,54].Similar to S. dydimum, E. uniflora is abundant in the understory of the Araucaria Forest, but their interaction networks were quite different, and no pollen type of Solanum was present in the networks of E. uniflora.
Besides the Solanum pollen type, Platymiscium (Leguminosae), Vernonia (Asteraceae), Baccharis (Asteraceae), and Erythroxylom (Erythroxylaceae) were the most abundant pollen types in the network.Platymiscium was the second most frequent pollen type, but, although it has poricidal anthers, in most of the studies on melittophilus plants, bees are seldom collected on the flowers of these plants, and those of the family Asteraceae are one of the most visited [30,31,34,[55][56][57][58].
In the collection of bees visiting several species of blooming plants in the Park of Vila Velha, 150 kms from Guarapuava, Asteraceae was the richest family (43 species), the one that received the most bee visits (55%), and that housed the largest number of species of bees (80 species) [30].These authors did not record any bees on the flowers of Platymiscium and Erythroxylum.The high number of bee visits to plants of this family may be due to the abundance and richness of this family in the tropics, the accessibility of the resources in their flowers for collection [59], its widespread entomophilic syndrome, the dominance of herbs in secondary fields [60], or presenting massive blooms.
Flowers of Erythroxylum campestre St. Hil., E. suberosum St. Hil., and E. tortuosum Mart were indistinctly visited by 14 wasp species, 14 bee species, and two diptera in the in the central-western region of Brazil.Brachygastra, Polistes, Polybia, and Pepsis wasps were considered effective pollinators of these plants, based on their efficiency in contacting the stigmas, while Trigona spinipes (Fabr.)and Apis mellifera were considered occasional pollinators.Augochloropsis aphrodite was frequent only on the flowers of E. campestre, and Augochloropsis cleopatra (Schrottky) on the flowers of E. suberosum and E. tortuosum These plants are distylic; that is, they have flowers with long styles (pin) as well as short ones (thrum).Their flowers are similar, being small, sweet-scented, creamish, and diurnal, as well as nectar producers (sugar concentration about 20.2%) [61].In the northeast of Brazil, Erythroxylon catingae Plowman was one of the plants most frequently visited by bees, and A. mellifera was the species of bees most common in its flowers [34].
Given the results of our study, we conclude that Solanum is an important source of pollen grains for several native bee species, mainly for the solitary species that are more diverse in the south of Brazil.Although the behavior of the bees on the flowers of S. didymum was not followed, our results indicate that the bees of the families Halictidae (A. notophos, Augochlora) and Andrenidae (Anthrenoides) are the pollinators of this plant.

Diversity 2018, 10 , 3 Figure 3 .
Figure 3. Pollen types from floral visiting bees' bodies of Solanum dy Park in Guarapuava, Paraná state, Brazil.(a-f) Angiosperms, Zanthoxylum.(b) Sapindaceae, Allophylus.(c) Sapindaceae, Matayb Solanaceae, Solanum didymun (esxicatte).(f) Turneraceae, Turnera.(ba Concerning the plants, out of the 35 pollen types found, only bee species, and Solanum was the pollen type that had the highest different bee species, and therefore a more general type.Nine Stenachaerum, T. Jaquemontia, T. Stylosanthes, T. Senna, T. Stru Monocotyledon) had interactions with only one bee species, and thThe average degree ҟ of interaction between plant specie degree distribution was heterogeneous, following the truncated po with bees (ҟ = 7.2).The bipartite graph (Figure4) and the NODF = 64.5 index ind with most of the interactions concentrated in a few species.

Table 1 .
Number of bee species on flowers of Solanum didymum (N) and their respective dominance indices (FO: species frequency of occurrence; D: species dominance).