Comparative pollen preferences by africanized honeybees Apis mellifera L . of two colonies in Pará de Minas , Minas Gerais , Brazil

The aim of this study was to investigate the polliniferous floral sources used by Apis mellifera (L.) (africanized) in an apiary situated in Pará de Minas, Minas Gerais state, and evaluate the pollen prefences among the beehives. Two beehives of Langstroth type with frontal pollen trap collectors were used. The harvest was made from September 2007 to March 2008, with three samples of pollen pellets colected per month per beehive. The subsamples of 2 grams each were prepared according to the European standard melissopalynological method. A total of 56 pollen types were observed, identifying 43 genus and 32 families. The families that showed the major richness of pollen types were: Mimosaceae (8), Asteraceae (6), Fabaceae (3), Arecaceae (3), Euphorbiaceae (3), Rubiaceae (3), Caesalpiniaceae (2), Moraceae (2) and Myrtaceae (2). The most frequent pollen types (> 45%) were Mimosa scabrella, Myrcia and Sorocea. The results demonstrated a similarity regarding the preferences of floral sources during the major part of the time. There was a distinct utilization of floral sources among the pollen types of minor frequency. In spite of the strong antropic influence, the region showed a great polliniferous variety, which was an indicative of the potential for monofloral as well as heterofloral pollen production.


INTRODUCTION
Pollen is essential for the development of larvae that, for their survival, depend on available stocks in the combs.The foraging worker bees are adapted to regulate the pollen comb stores according to the intrinsic needs of the colony.The regulatory mechanisms that incite a smaller or bigger interest in searching for pollen in specific plants by the foraging worker bees are intricate and subject to controversy (Sagili and Pankiw 2007).According to Cook et al. (2003), the bigger the number of essential aminoacids a specific plant has, the more visited it is by A. mellifera honeybees.According to Schmidt and Buchmann (1993), honeybees collect pollen in various plant species and, thereby, maintain a good nutricional balance and a high dilution of toxic potential of alkaloids and other poisons.To Gary (1992), the nutrients in some bee pasture are requested by the hive, possibly because of evolutionary influences of the bees' harvest behavior (previous learning), as well as the different levels of competition among the colonies.Roulston et al. (2000) state that the protein variation in pollen grains cannot be directly related to the activity of the pollinizator, as the pollen of the zoophilous species is not richer in proteins than that of the anemophilous species.
In spite of the complexity of factors, two are definitely important: the quantity of pollen stocked in the comb cells and the size of the breed (Dreller et al. 1999).
The results of the melissopalynological research are correlated with the offer of nectar and pollen in cronological terms, showing that, in differents periods during the year, certain flowers can be nectariferous or polliniferous, while in other periods both floral resources are available (Luz et al. 2007a).Apart from this, the pollen types observed in the pollen pellets can vary according to the region where they are offered, a factor which depends on the available surrounding bee pasture in the apiary vegetation.To know, it is one of the most important requisites in choosing a local for installation of the apiaries, because it reflects on the number of beehives and on the production of beekeeping derivates that they can bear.
In researching the botanical origin of apicultural derivates, one obtains a better management of the production according to the required international commercial qualification (Luz et al. 2007b).As melissopalynological analysis is rarely made on pollen pellets that are sold on the brazilian market, generally the floral origin that is written on the label is erroneous in which it indicates a vulgar name of a plant, which pollen is not contained in the product (Barth et al. 2009).
The Brazilian honeybee races are hybrids between the european honeybees and the southeast african honeybee, characterized by a great facility to swarm, high productivity, desease tolerance, good adaptation to colder climates and continuously working in low temperatures, while the european honeybees withdraw during these seasons (Embrapa Meio Norte 2003).The research by Villanueva-G and Roubik (2004) in Mexico on competitive pressure in relation to the pollen harvest among these races showed that the european honeybees, previously adapted to the local, exploited the variety of pollen sources more than the introduced african honeybees; but, in spite of this, the european honeybees utilized significantly only a few plants.The africanized honeybees showed more advantages than the european ones in exploring trees, grasses and sedges, and these were extensively utilized when pollen was available.
Even though the african honeybees were introduced in Brazil only some fifty years ago, what can be seen in the few existing palynological papers on this subject is a great variety of pollen types found in the pollen pellets, wich reveals an intrinsic adaptation between the africanized Apis mellifera and the native polliniferous flora (Barth 2004).
In Minas Gerais, Modro et al. (2007) presented physical-chemical and palynological results of pollen pellets from ten honey beehives, chosen in two apiaries in Viçosa, showing that the nutritional components were correlated with the frequency of specific pollen types, a fact attributed to a more balanced diet.Furthermore, these authors confirmed the occurrence of a selective harvest from different pollen sources possibly due to the intrinsic preference of each colony and to the exploitative competition behavior for the available floral sources.

STUDY AREA
The county of Pará de Minas, Minas Gerais, is located at 19 • 53 S of latitude, 44 • 31 W of longitude, and 970 m high.The climate in Pará de Minas is Cwa, according to Köppen's classification, where June, July and August are the driest months, and November, December and January the rainiest (Pinheiro and Batista 1998).It is located in the area embraced by Seasonal Semideciduous Forest, in the transition between the Atlantic Forest domains and Cerrado (IBGE 1993), with many arboreal species of Leguminosae (lato sensu), Myrtaceae, Lauraceae, Rubiaceae, Annonaceae, Meliaceae, Euphorbiaceae and Flacourtiaceae mainly along the margins of the waterbodies (Meyer et al. 2004).During the driest periods of the year, the region is subject to forest fires, many times caused by the catlle breeders.The apiary selected for the study is close to a natural forest fragment with few introduced fructiferous and ornamental specimens of Coreopsis lanceolata L. ("Yellow daisy"), Vernonia polysphaera Baker ("Assa-peixe"), Taraxacum officinale Weber ex FH Wigg ("Dandelion") and Achyrocline satureioides (Lam.)DC. ("Camomile") (Asteraceae); Begonia sp (Begoniaceae); Ipomea sp. and Ipomea alba L. ("Moonflower") (Convolvulaceae); Chamaecrista sp and Stryphnodendron adstringens (Martius) Coville ("Barbatimão") (Fabaceae); Persea americana Mill.("Avocado") (Lauraceae); Abelmoschus esculentus (L.) Moench ("Okra") (Malvaceae); Miconia albicans (Sw.)Triana ("Canela-de-velha") (Melastoma- The aim of this study was to investigate the influence of the local flora on the pollen harvest by the Apis mellifera L. (africanized) from two beehives in an apiary in Pará de Minas, Minas Gerais, in order to examine the similarity in polliniferous sources preferences.

MATERIALS AND METHODS
Two honeybee colonies of Apis mellifera L. (africanized) were selected for harvest of pollen pellets, installed in beehives of Langstroth type, positioned side by side in the apiary, each one with a nest and ten honeycombs.Each beehive was equipped with a frontal pollen trap (Jean-Prost 1987).
The harvest of pollen pellets was gathered between the 15 th of September and 18 th of November in 2007, and between the 16 th of February and the 29 th of March in 2008, with a seven day interval among each harvest.At the same time, the pollen pellets were collected from the pollen traps in both beehives, summing a total of twenty-nine samples, as no pollen was collected on 01/ 03/2008 in one of them.No harvest was possible during the months of December and January due to rainfalls and invasion of ants that weakened the beehives.
The samples of pollen pellets from beehive A were collected on the following dates: Each pollen sample was manually cleansed and kept in a refrigerator.In sequence, it was homogenized and 2 g (wet weight) of it were macerated and extracted with ethanol.The preparation of the pollen pellets followed the standard European methodology (Maurizio and Louveaux 1965) without the use of acetolysis and with some modifications as suggested by Barth et al. (2009).
The identification of the pollen types was based upon the reference pollen slide collection of the Institute of Botany in São Paulo, as well as on specialized literature data, (Melhem et al. 1984, Roubik and Moreno 1991, Barth 1970a, b, c, d, 1989).
Aproximately 500 pollen grains per sample were counted.The pollen classes and terminology follow Zander (1924) and were implemented later by Louveaux et al. (1978), comprising the dominant pollen class (> 45% of the total pollen sum), the accessory pollen class (15-45% of the total pollen sum) and the important pollen class, subdivided in isolated (3-15% of the total pollen sum) and occasional (< 3% of the total pollen sum).
The Principal Component Analysis (PCA) was performed in order to verify the pollen preference in the beehives A and B, by which the pollen types were grouped per month (Aset and Bset = september, Aout and Bout = october, Anov and Bnov = november, Afev and Bfev = february, Amarc and Bmarc = march).The matrix comprised the absolute value of all taxa found in each sample.The absolute numerical variables were transformed into natural logarithm [log (x+1)] using the FITOPAC program (Shepherd 1996), and thereafter the ordination was done through a covariance matrix using PC-ORD 4.0 (McCune and Mefford 1999).The MINITAB program (2003) was used to compose the percentage similarity dendrogram among the pollen pellet samples.
The illustrations of the pollen grains were digitally obtained using a OLYMPUS BX 50 microscope equipped with a video camera and a PC with the program IMAGE PRO-PLUS 3.1 for Windows.
In both beehives, the heterofloral samples were predominant most of the time.The month of September (samples A1, A2, A3, B1, B2 and B3) showed the greatest pollen richness (28) (Table I).
The most frequent pollen types (> 45%) in beehive A were Mimosa scabrella, Myrcia and Sorocea, while in beehive B were Mimosa scabrella and Myrcia (Table I The variability among the samples of pollen pellets for each month from the two beehives comprised 77,8% on the two first axis in the Principal Component Analysis (PCA) (Fig. 1).Considering the same harvest periods, the correlation among pollen pellets from the two beehives showed a great similarity with respect to the ocurrence of the pollen types.The pollen types Myrcia, Sorocea and Cecropia were the main characteristic components in the PCA in the months of September, October and November 2007, while Mimosa scabrella, Asteraceae, Poaceae, Croton and Tetrapteris were in February and March 2008.
There was a distinct usage of floral sources when the two principle anual harvest periods were compared (September to November 2007, andFebruary to March 2008), given that the pollen pellets formed two main groups with 50,13% of similarity among them (Fig. 2).The samples from September to November showed a similarity of 90,34% between the beehives (Fig. 2).The biggest porcentage similarity in the usage of pollen sources (99,48%) was noticed in the samples from the period between February and March 2008 (samples Afev, Bfev, Amarc and Bmarc), which demonstrates the simultaneous harvest by the honeybees from the two beehives from the flowers of the Asteraceae family and Mimosa scabrella.

DISCUSSION
Researches from Apis mellifera pollen pellets collected in Rio de Janeiro (Barth 1973, 1989, Barth and Luz 1998, Luz and Barth 2001, Luz et al. 2007a) showed that the most common pollen types were Eupatorium, Ricinus communis and Sapindaceae, together with Cecropia, Borreria, Gochnatia, Panicum, Spondias, Triumfetta and Vernonia.Pollen from Eucalyptus and Mimosaceae (identified taxa: Albizia lebbeck, Piptadenia sp., Schrankia sp., Mimosa bimucronata/M.caesalpiniaefolia and Mimosa scabrella/M.pudica) were observed a few times.Various of these pollen types were found in the pollen pellet samples from Pará de Minas, reflecting a characteristic vegetation of the Southeast Brazil.
According to the results obtained in the pollen pellet samples from the apiary in Pará de Minas, during the second half of September 2007, there was a predominance of Cecropia, Myrcia, Ricinus, Sorocea and Trema.On the other hand, a research by Modro et al. (2007) in Viçosa (Minas Gerais), in an apiary of Apis mellifera located in an area of coffee plantation with abandoned pasture, during the same month, there was a predominance of Cecropia, Coffea and Eucalyptus in the pollen pellets.During the first half of October, Myrcia and Sorocea predominated in Pará de Minas, and Myrcia and Coffea in Viçosa.During the second half of October, the pollen from Pará de Minas was monofloral (Myrcia) while in Viçosa it was heterofloral, consisting of Myrcia, Piper, Anacardiaceae and Senecio.The first half of November in Pará de Minas showed a predominance of bifloral pollen pellet, Myrcia and Sorocea, while in Viçosa there was an heterofloral predominance, consisting of Myrcia, Coffea, Anadenanthera and Piper.main supplier of pollen, followed by the representatives of Asteraceae.In Pará de Minas, the Asteraceae was also an important source of pollen, however, this wasn't observed in the harvest of Mabea pollen type during the analyzed period.
The referred study by Modro et al. (2007) in an apiary in an abandoned pasture with coffee plantation in Viçosa also analyzed the pollen preference among five beehives that showed differences in the harvest of pollen sources over the time.This difference was attributed to the previous learning in every colony, because in areas of abandonded pasture, there is no uniformity in the plant species and its blossom times are short, which creates high competition.In Pará de Minas, the main pollen types were very similar and, in the PCA, the samples from the same month were grouped in pairs between the two beehives, even though the beehives showed differences regarding the main floral sources when the samples were analyzed weekly.The main difference regarding the weekly samples was the presence of Trema, exclusively harvested in beehive A on 29/09/2007, which showed a frequency above 15%.There was also a range between the beehives in Pará de Minas when certain pollen types were observed occurring only in the pollen pellets of beehive A or beehive B, mainly Acacia, Apocynaceae, Bauhinia, Bignoniaceae, Chenopodium, Com-melina, Elephantopus, Euphorbiaceae, Euterpe/Syagrus, Fabaceae 2, Ilex, Inga, Jacquemontia, Ludwigia, Mimosa caesalpinaefolia, Mimosa selloi, Mimosa verrucosa, Piptadenia, Polygonum, Struthanthus, Trema and Vigna, most of them harvested in short periods and with low percentage values.According to Villanueva-G and Roubik (2004), the africanized honeybees compete among themselves for food sources, which can cause a decrease of the diet diversity in every colony and an increase of specialization in specific pollen sources, many times used by one of them only as a function of intense competition.In Pará de Minas, this fact may have influenced the honeybees forraging behaviour.
There is a low similarity (50,13%) between the two major harvest periods (September to November 2007, and February to March 2008), which indicates the difference in blossom phases and the phenological plant development in the beehive sourroundings.
We do not know the reason why there had not accumulated any pollen in the pollen trap in beehive B in 01/03/2008 however, there was a big fire close to the local that could have caused this fact.The lack of pollen during this period weakened the beehive.
The great richness of the pollen types in the samples in Pará de Minas shows that Apis mellifera collected pollen in many floral sources, including the plants that less occurred in the area.The harvest in various sources was attributed to a richer and more balanced diet of these bees, according to Modro et al. (2007) in Viçosa.

CONCLUSIONS
The pollen types recognized in the pollen pellets collected by Apis mellifera in Pará de Minas were considered characteristc of the Southeast region in Brazil, mainly from disturbed areas in the forest, proved by the presence of pollen from heliophyte and ruderal plants (Asteraceae, Cecropia, Mimosa scabrella, Ricinus, Sorocea, Trema, among others).
There was no dissimilarity among the main pollen types in the monthly samples from the two beehives.This fact suggests a similar preference of pollen sources, even when differences occurred regarding the main weekly analyzed floral sources, which may be considered a function of competition.
September was the month when the honeybees took most advantage of the floral sources in the vegetation nearby the apiary.The periods that coincided with the biggest harvest from the same pollen source (Myrcia and Mimosa scabrella), October 2007 and March 2008, respectively, were caused by the ample blossoming of these plants in the environment around the apiary.Some pollen types from less common plants occurred in one or the other of the two beehives.This suggests a previous learning process due to intrinsic preferences of each colony or different competition levels in the search for pollen sources.Both possibilities may occur in a vegetation characterized by non-uniformity regarding plant species, wich in turn have short blossom periods.
There was a dissimilarity among the pollen sources when the two main flowering periods of September to November 2007, and February to March 2008, were compared.

ACKNOWLEDGMENTS
Thanks to Erik Bernhard Lidgren for the translation and Dr. Eduardo Custódio Gasparino for assistance with the statistics computer programs.The first author acknowledges the Brazilian "Conselho Nacional de Desenvolvimento Científico e Tecnológico" for the fellowship of "Produtividade em Pesquisa" (CNPq #301220/2009-3).

(
Fig. 1 -PCA biplot for pollen types in pollen pellets samples per month from Pará de Minas, Minas Gerais State, using the absolute value of the variables per sample.