Despite their role as key pollinators among insects¹, the biological fundamentals for pollen source selection by honey bees (Apis mellifera) in Mexico are still mostly unknown. Botanical studies with apicultural interests are not particularly abundant if is considered that Mexico is a large and mega diverse country, classified in the first places in apicultural production and exports in the world². Since bees depend entirely on the vegetation for their survival, it is crucial to understand their feeding preferences as well as the specifics about pollen availability throughout the year. Pollen contains the nutrient protein³ needed for the brood and young workers survival and proper development. It also contains lipids, vitamins and minerals⁴ as incidental components.

Foragers collect pollen in a trend and proportion that vary greatly according to the availability of the resources, distance to the source, nutritional value⁵, needs of the hive, i.e. life cycle and physiology of workers, queen and drones and weather conditions⁶. This is of particular interest to beekeepers and researchers because the gathering behaviour of the bees does not seem to have fixed patterns. Each season bees will collect pollen in regards to different variables and even in arbitrary ways, i.e. without regard to its nutritional value or from resources that are not as close or highly available as others⁷. This type of information is fundamental to beekeeping and to assess the potential of any determined area, for the production of pollen and for all efforts related to the conservation of biodiversity⁸, particularly where the populations of bees are declining.

Analysis of the pollen collected provides information about its botanical origin, the preferred plant species and aids in understanding the foraging behaviour of the bees. The objective of this study was to find what pollen types were collected by honey bees during the honey production season.

With this in mind, a sampling project was designed to collect and analyze corbicular pollen to determine the spectrum of the pollen used by A. mellifera. The pollen grains were primarily identified by means of a special pollen reference collection made from plants in bloom in the locality. These plants were identified by botanical specialists from the University of Guadalajara, and the voucher specimens remain at the Botanical Institute Herbarium of the Centro Universitario de Ciencias Biológicas y Agropecuarias (CUCBA). A site was selected within an area of importance for beekeeping, in one apiary in the village of Huejotitan, municipality of Jocotepec, state of Jalisco. The experimental site was located at 20°21’13.45’’N, 103°29’6.97’’W. The elevation at the site is 1,597 m asl. around the apiary, the land cover is dominated by seasonal cultivated crops, pastures and secondary vegetation interspersed with tropical deciduous forest.

Among 23 bee hives in the apiary, three were chosen for their strength to be sampled once a month for four months with modified Ontario pollen traps⁹. From August to November 2012, traps were installed and kept in place for 24 to 48 h and then removed. This period corresponds to the honey preharvest and harvest season. The corbicular pellets were gathered from the trays, cleared of debris, put in plastic containers, tagged and frozen. At the laboratory, 1.5 g of pollen were taken from each of the three samples corresponding to one given month and mixed together to form one single larger sample of the pollen collected from the three hives together. In the end there were four samples from the original 12, one for each month.

Before processing, the pellets were carefully and softly mashed in a mortar. The pollen grains were processed by acetolysis technique to remove the exine; permanent glycerine jelly slides were made for the preservation and analysis; the pollen grains were identified by their size and shape, with an Olympus BH-2® upright microscope equipped with a 100X ocular micrometer to measure each individual species pollen grain, using immersion oil; volume of the individual pollen grain was calculated with the formula: V=4/3πa²b where "V" is volume, "a" is the major axe of the pollen grain and "b" the minor axe¹⁰. Identification was made by comparison with the pollen reference collection of the Institute of Geology, Universidad Nacional Autónoma de México. In order to obtain the relative percentage, all the pollen grains were counted in each slide. A reference collection with the pollen grains of local plants in bloom was prepared as an ancillary means of identification.

Every month for 4 mo a circuit between 3 and 5 km long in the surroundings of the apiary was walked to sample all blooming plants. The pollen grains where obtained by extracting the anthers from the flowers and then processed for acetolysis according to the same technique¹¹mentioned above. The information was used to determine whether a plant was a source of nectar, pollen or both, as well as their migratory status. Information was also taken from the available domestic bee flora publications^11-15. From the pellet samples, 23 different pollen types belonging to 17 plant families were recorded (Table 1) and from these 13 were identified at species level, 5 at genus level and 5 at family level. In August there was no dominant pollen type, however there were three secondary types, Aster sp., Eucalyptus citriodora and Ricinus communis, one important minor, Cyperaceae, and traces of other ones. Thus the four types were significant, with percentages above ten. In September E. citriodora was the dominant type with Poaceae and Psidium guajava as secondary types and traces of others. In this month three types were significant, with percentages above ten. In October no dominant type was obtained but there were again three secondary types, E. citriodora, Hyptis albida and L. leucocephala, all significant, with percentages above ten, and traces of others. In November E. citriodora was considerably dominant over the two secondary types, Asteraceae and Pseudosmodingium sp., but the three were significant, with percentages above ten, and traces of others. E. citriodora was significant in the four samples and Asteraceae and L. leucocephala were found in three. R. communis, Sicyos angulatus, Citrus sp., Pseudosmodingium sp. and Poaceae appeared in two samples each.

Each month the represented families changed, however, there was a consistency in their overall presence and percentages of representation (Table 2).

Asteraceae was present in the four samples, Myrtaceae in three, Anacardiaceae, Fabaceae and Rosaceae in two, and Betulaceae, Cucurbitaceae, Cyperaceae, Euphorbiaceae, Lamiaceae and Oleaceae in one. 78 different species of plants in bloom, belonging to 30 families and 71 genres, were documented during the 11 mo (Table 3) in order to have as many as possible species of pollen grains documented for reference. The five best represented families were Asteraceae with 33.33 %, Fabaceae with 8.97 %, Solanaceae with 6.41 %, Lamiaceae with 5.12 % and Verbenaceae with 3.84 %. These five families represent 29.41 % of the total number of families and 57.67 % of the total number of species. 17 of all the species have been reported to be nectar producers, seven pollen producers, 17 nectar and pollen producers and 37 are not documented in terms of their importance for honey bees; 50 % were forbs, 30.77 % shrubs and 19.23 % trees. Considering all the species, 88.46 % were native and 11.54 % were exotic. Twenty-six (26) species were documented to be visited by honey bees.

The months with more species in bloom were September and November with 16 species each, then October with 13 and finally August with 12. Bees represent the primary pollinators among insects and honey bees are becoming the only ones in areas where intensive crop monoculture is gradually wiping out the wild native insects. One of the reasons is that A. mellifera belongs to one of the few bee genera known to have polylectic habits¹⁶. Yield increases are reported to be up to 96% in cultivated crops pollinated by them¹⁷. In terms of the sources used by the honey bees as revealed in this study, Myrtaceae had the second highest percentage in August and by far the first in September, October and November. This family was prominently represented by E. citriodora, an introduced species. Originally evolved in the Austro-Malaysian region¹⁸, has been introduced in many countries for its value as timber, fuel wood, wood fiber and ornament¹⁹. The floral phenology of Eucalyptus tends to be synchronous among different individuals within one stand but at the same time shows great variation in flowering time and even intermittent flowering periods over the greater part of the year²⁰; honeybee has been documented to be one of the most prevalent visitors its flowers²¹.

In August the Asteraceae family was dominant over Myrtaceae. Asteraceae is the most abundant family in Mexico^22-25 and represents an estimated 10 % of all know plants in the world²⁶ and its center of diversification is located in Mexico, where it is the largest and most representative group, containing from 7 to 32 % of the country's flora and 12.5 % of Jalisco´s²⁷. In September, the second most abundant family was Poaceae, but neither species nor genre were determined. This is also an extensive group, with more than 500 species worldwide, which includes the cereals humans consume and the grasses for cattle feed²⁶. In October, the second most important families, in equal percentages were Fabaceae, represented by L. leucocephala and A. farnesiana, and Lamiaceae, by H. albida. Although the percentages in this study refer to the number of pollen grains and not to their volume, the sizes are still relevant because the ratios change when analyzed in terms of a different variable. E. citriodora pollen grains are small, 25 µm average, and have the shape of a flattened triangular prism. The volume of one these grains averages approximately 3,125 µm³. Contrastingly, the pollen grains of A. farnesiana are 60 µm average, and ellipsoid in shape. Their volume is approximately 78,539.8 µm³. E. citriodora represented 34 % of the sample and A. farnesiana (Fabaceae) 2.4 % in terms of number of grains. Nevertheless, if their total volumes compared, the proportions change radically: 106,250 µm³ for E. citriodora and 196,349.5µm³ for A. farnesiana, almost twice the volume of E. citriodora.

In November Asteraceae was the second most important family; some pollen species, regardless of their frequency, were present in at least two of the pollen load samples indicating their presence for longer than a single month thus representing a long-term food resource during the year. Such is the case with E. citriodora, present in the four samples, and Asteraceae and L. leucocephala (Fabaceae), present in three. Of the 23 pollen types found in the samples, seven have been reported as pollen sources for honey bees in other states A. farnesiana, Fraxinus uhdei, Heliocarpus terebinthinaceus, L. leucocephala, P. guajava, R. communis, and S. angulatus^11-15,27,28.

Of the total number of plant species in bloom observed in the area throughout the year only 34.21% have been reported to be used by the honeybees^22-26,29,30. This might be explained by the selectiveness of honey bees depending on the relative abundance and quality of nectar, pollen and distance to the sources.