A Comprehensive Investigation of the Mineral Composition of Brazilian Bee Pollen : Geographic and Seasonal Variations and Contribution to Human Diet

A composição de minerais foi investigada em 154 amostras de pólen apícola, provenientes de diferentes regiões do Brasil. A técnica de ICP OES foi utilizada para determinar Ca, Cu, Fe, K, Mg, Mn, Na, P, Se e Zn, após mineralização em sistema de microondas. Utilizando parâmetros instrumentais otimizados conseguiu-se boa exatidão e precisão na determinação simultânea dos minerais. Em geral, as amostras dos estados do Nordeste apresentaram teores de minerais mais elevados e produção constante durante o ano. Os minerais Mn, Se, Cu, Zn, Fe são os que poderiam ter maior contribuição na dieta, podendo atingir 70, 37, 27, 17 e 17%, respectivamente, da ingestão dietética diária recomendada no Brasil. A análise de componentes principais mostrou que Ca, Cu, Fe, K, Mg, Mn, Na, P e Zn podem ser usados na classificação do pólen nacional em função da origem geográfica. Os teores dos minerais variaram amplamente ao longo do ano.


Introduction
Bee pollen, often referred to as a natural supplement, is one of the emerging bee products valued by its high content of nutrients and other health-promoting constituents.Brazilian honey has become an important product for domestic consumption and export, the latter having increased by 12% of its value in 2010.The major importing country is the United States, followed by Germany, the United Kingdom, Canada, Austria, France and Spain. 1 Official data for the pollen is not available, but it is known that the current production is not sufficient to meet the national demand.Beekeeping has gradually increased in Brazil favored by plant diversification, vast land area and favorable climatic conditions (tropical climate). 24][5][6][7] Brazilian studies on the macronutrient and micronutrient composition of bee pollen, however, are still limited. 8,9he chemical composition of bee pollen is known to vary with the methods of extraction and storage, the floral origin (plant species), including age and nutritional status of the plant, and environmental conditions. 10According to O'Rourke and Buchmann, 11 pollen composition may vary in different locations, between seasons and years.Funari et al. 8 found variations in mineral composition of Brazilian bee pollen collected in one apiary from Botucatu during a period of four months.Knowledge of the nutritional value and quality of bee pollen produced in different regions may contribute to the choice of the place to install apiaries. 12,13ome mineral elements have already been quantified in bee pollen samples from several countries: potassium (K), phosphorus (P), magnesium (Mg), calcium (Ca), sodium (Na), sulfur (S), iron (Fe), copper (Cu), manganese (Mn), zinc (Zn), chromium (Cr), nickel (Ni) and selenium (Se). 6,8,14- 19The number of samples analyzed, however, is limited, varying from one 19 to fifty 16 per study.Two of them are Brazilian studies, involving 24 samples collected in only one sampling site from the state of São Paulo 8 and 36 samples from Southern Brazil. 9 The main analytical techniques used to determine mineral elements in bee pollen samples are atomic absorption spectrometry (AAS), 7,17,18,20 electrothermal atomization atomic absorption spectrometry (ET-AAS), 21 inductively coupled plasma optical emission spectrometry (ICP OES), 6,16 and X-ray fluorescence spectrometry (XRF). 15aldi-Coronel et al. 17 determined K, Ca, Mg, Na, Fe, Zn, Mn and Cu by AAS in bee pollen ash obtained by calcination of samples at 550 °C.Szczesna found in decreasing order the minerals K > Mg > Na > Ca > Fe > Mn > Zn > Cu in bee pollen samples collected in Poland, South Korea and China, after sample mineralization in acid medium using an open microwave-assisted system and quantification by AAS. 18Somerville and Nicol 16 used ICP OES to quantify K, P, S, Ca, Mg, Na, Fe, Zn, Mn and Cu in bee pollen from New South Wales after sample digestion in closed microwave-assisted system.Serra-Bonvehí and Escolà-Jordà 6 determined Fe, Ca, Zn, K, Na, Cu, Mg, P and Mn in Spanish bee pollen, also using ICP OES after calcination.Comparing dry and wet digestion, Kump et al. 15 obtained the best results by ashing the sample in the presence of diluted H 2 SO 4 (1:4).These authors used XRF to determine K, Ca, Mn, Fe, Ni, Cu, Zn, Rb and Pb in bee pollen samples from Slovenia.
In a previous study, the occurrence of inorganic contaminants in bee pollen from Southeastern Brazil was reported. 22The purpose of the present paper was to analyze bee pollen samples from Brazilian producing regions for the mineral elements Ca, Cu, Fe, K, Mg, Mn, Na, P, Se and Zn.To reach this goal, a method using sample digestion in closed microwave-assisted system and quantification by ICP OES was optimized.While food safety was the concern of the previous paper, the current work deals with the nutritional value, geographic and seasonal variations of this increasingly important food product.

Experimental
Sampling A total of 154 Brazilian dehydrated bee pollen samples ready for commercialization (dry, clean and packaged products) were directly purchased from apiculture producers of Southern, Southeastern, Central, Northeastern and Midwestern regions (Table 1) during a year.Samples (weighing 200 to 300 g) were transported to Campinas, SP, mostly in plastic bags, some in glass containers, simulating commercialization practices.

Sample preparation
In the laboratory, the samples were packed under vacuum in polyamide and polyethylene bags, to avoid absorption of moisture and oxygen, and stored in a freezer at -16 °C until analyses were carried out (maximum of 3 months).On the day of analysis, samples were quartered in a stainless steel quartering equipment and ground in a refrigerated mill with tungsten helix (M20, IKA Labortechnik, Staufen, Baden, Ge).The samples were then sieved using a 30-mesh (600 µm) sieve to standardize pollen particles size.

Analytical determination
Digestion was performed in a closed microwave acid digestion unit with 41 digestion bottles (Microwave Digestion System Start D, Milestone, Sorisole, Italy), using nitric acid 65% (m/m) (Merck, Farmstadt, Germany) and hydrogen peroxide 30% (v/v) (Merck, Darmstad, Germany) as oxidizing agents.A sample of 0.65-0.70g was weighed in the digestion bottle and 10 mL of concentrated nitric acid were added, followed by 3 mL of 30% hydrogen peroxide.Digestion was performed in two phases, with 30 min of cooling after each phase.Power was maintained at 1000 W and external temperature of the reaction vessel was at 110 °C.For phase 1 three stages of 20, 20 and 30 min were undertaken, the internal temperature of the reaction vessel being kept at 140, 160 and 180 °C, respectively.Phase 2 had two stages of 20 and 30 min with the internal temperature of the reaction vessel at 180 and 200 °C, respectively.Once digestion was completed, the bottle's content was transferred to a 25 mL volumetric flask, the volume being completed with a 5% HCl solution (v/v). 22uantification of mineral elements was performed by ICP OES, using a Varian Vista MPX model (Mulgrave Victoria, Australia), with axial vision, equipped with a radio frequency (RF) source of 40 MHz, a Charge Coupled Device (CCD-type) simultaneous multi-element solid state detector, a peristaltic pump, a spraying chamber and a sea spray nebulizer.The system was controlled by ICP Expert software and 99.996% liquid argon was used as plasma gas (Air Liquid, SP, Brazil).Operating conditions of the ICP OES equipment were: forward power, 1000 W; nebulizer Ar flow rate, 0.9 L min -1 ; auxiliary argon flow, 1.5 L min -1 ; cooling Ar flow rate, 15 L min -1 ; background correction, 2 points; integration and reading time, 10 s; replicate number, 3 and axial torch configuration.The analytical line wavelengths were: Ca, 317.933 nm; Cu, 324.754 nm; Fe, 259.940 nm; K, 766.491 nm; Mg, 280.270 nm; Mn, 257.610 nm; Na, 589.592 nm; P, 213.618 nm; Se, 196.026 nm; Zn, 206.200 nm.
Stock solutions at 10,000 mg L -1 for Ca, K, Mg, Na (Titrisol -Merck) and P (Qhemis High Purity) and at 1,000 mg L -1 for Cu, Fe, Mn, Zn (Merck) and Se (Qhemis High Purity) were used for preparing the standard solutions in 5% HCl v/v.The concentration ranges of the standard solutions were: 0.005 to 0.5 mg L -1 of Cu and Se, 0.05 to 2.5 mg L -1 of Fe, Mn and Zn, 0.5 to 15.0 mg L -1 of Na, 5.0 to 75.0 mg L -1 of Ca, P and Mg and 25.0 to 150.0 mg L -1 of K.

Assurance of quality in analytical procedure
Method accuracy was evaluated by recovery tests conducted at two levels, covering the concentration ranges in the samples.The limit of detection (LOD) and limit of quantification (LOQ) values were calculated as suggested by Mermet and Poussel: 23 LOD = (3 RSD BEC)/100 and LOQ = 5 LOD, where RSD is the relative standard deviation and BEC is the background equivalent concentration, determined experimentally (n = 8).Method precision was evaluated by the coefficient of variation of 8 repetitions.
Because of the unavailability of a certified reference material for bee pollen, a certified reference material, NIST 1566a (oyster tissue), was used only to validate analytical control of the laboratory under existing working conditions.This material has certified values for Ca, Cu, Fe, K, Mg, Mn, Na, Se and Zn.Results were assessed by the relative error.

Statistical analysis
In order to verify if the means obtained for mineral elements of dehydrated bee pollen samples from different states could be considered statistically different at a significance level of p < 0.05, Tukey's multiple comparison test was applied (employing the approach method with different repetition numbers per state) using the software Statistica for Windows 5.5 (StatSoft Inc., Tulsa, USA).

Hierarchical cluster analysis (HCA) and principal component analysis (PCA)
Using mean values, a data matrix (154 × 10) was built, in which samples were put in lines and the concentration of minerals were put in columns (variables).Data were preprocessed and then submitted to HCA and PCA, using Statistica program version 5.0.The HCA statistical analysis was performed with the purpose of verifying similarities between samples from different states, clustering samples by similarity from the Euclidean distance calculated.HCA results were presented as a bi-dimensional graph called dendogram.PCA analysis was used as a chemometric tool to verify which mineral elements were important for the separation of the groups formed.

Estimation of microelement daily intake
Contribution of mineral elements in bee pollen to dietary intake was estimated according to the recommended daily intake (RDI) values established by Brazilian regulation for adults, 24 based on the recommended nutrient intakes (RNI) for Ca, Fe, Mg, Zn and Se 25 and the dietary reference intakes (DRI) values for Cu, Mn 26 and P. 27 Potassium DRI level was adopted from the Institute of Medicine, 28 since the Brazilian regulation does not establish a recommendation for this mineral.To estimate the microelement daily intake, the portion size for bee pollen recommended by Lengler was used, which is 25 g (approximately three tablespoons) for adults. 29

Method performance
Sample digestion using a microwave-assisted closed system allowed quantification of the mineral elements Ca, Cu, Fe, K, Mg, Mn, Na, P, Se and Zn in a single sample preparation.The residual carbon content was 0.15 g kg -1 , indicating low level of organic matter after sample digestion.The time spent to digest the samples was approximately 180 min.
The method exhibited good performance, confirmed by r values close to 1, LOD and LOQ with appropriate sensitivity to determine the minerals and satisfactory CV at the detection levels achieved (Table 2).The working standards solutions for the calibration and reagent blanks used for the estimation of the linearity ranges and limits of detection and quantification were subjected to the digestion procedure.Recovery results in polen matrix were within the range of 70 to 106%, mostly close to 100%, indicating adequacy of the analytical method.
The results obtained in the laboratory analytical control evaluation, using the oyster tissue NIST 1566a certified reference material, under the same preparation conditions for bee pollen samples, showed adequacy of the method.For all elements, relative error was below 8% (Table 2).

Composition of mineral elements in Brazilian bee pollen sample
Mean contents, standard deviations and concentration ranges obtained for the minerals in the 154 dehydrated bee pollen samples from 11 states and the Federal District are shown in Table 3.In general, wide within-state and between-state concentration ranges were obtained, especially for Se, Na and Cu, which had standard deviations higher than the means.
A perusal of the between-state variation shows higher Ca levels in PI, CE, SE, BA and MT samples, while those of SE, CE and PI had higher amounts of P.There was no significant variation of Cu and Mn among the states.Samples from MG and DF had higher levels of Fe, than those of the other states, but the differences did not reach significant levels.Samples from SE, DF, PI, CE, BA and MT had greater amounts of K and Mg, but for the latter, PR also had higher  Pronounced variations are observed in the levels of all minerals during one year of bee pollen production.Although this variation presents a challenge to the appraisal of geographic effects, one can still observe that Brazilian pollen production is constant in Northeastern states (BA and SE), and the mineral levels tend to be higher, especially for Zn, Ca, K, Mg and P. In fact, apiculture has grown in this area of the country, which is considered to be prominent for beekeeping activity.The Northeastern region is one of the few in the world with the potential to produce organic bee products in large quantity, due to the great diversity of plant species and microclimates, along with vast unexplored land areas, free of conventional farming activity.
In SP and SC states bee pollen production is higher during the Spring season, which occurred from September  22).In addition, SP samples showed a tendency for higher Fe levels during August, September, and October and higher Mn levels during February and March.
The dendogram obtained by HCA is shown in Figure 2a.Cluster formed among bee pollen samples from the Southern and Southeastern states of SC, RS, PR and SP indicates that these samples are similar in mineral composition.Samples from DF and MT exhibit low similarity.Samples from MG and ES show similarity and samples from the Northeastern region (SE, PI, CE and BA) make up another group with similar composition.
The grouping of the samples through their similarities can be better explained by the PCA (Figure 2b) that indicates which mineral elements are responsible for the clusters.The samples of BA and SE can be characterized by the Na, Ca and Mg contents.Those from PI and CE can be mainly defined by Zn, K, Mn and P. Cu influenced the samples from the Federal District while those of MG, RS, SP and MT are characterized by the Fe level.
Estimated micronutrient contribution of pollen to the diet Contribution of the mineral elements in bee pollen to recommended daily intakes (RDI) established by Brazilian regulation per portion is shown in Table 4. 30 In terms of the mean values, high contribution could come from Mn (69.6% of RDI) and Se (37.5% of RDI).According to Brazilian regulation, 31 a contribution of more than 30% classifies the food as a rich source and above 15% as a source.In addition, Brazilian bee pollen may be a source of Cu (26.7% of RDI), P (17.2% of RDI) and Zn (16.7% of RDI).It is notice worthy that Cu content did not vary

Figure 1 .
Figure 1.Variation of minerals levels of bee pollen during one year of production from the same sites of the following states: Santa Catarina (SC), São Paulo (SP), Sergipe (SE) and Bahia (BA).

Table 1 .
Localization of bee pollen sample collection regions

Table 2 .
Performance characteristics of the method for determining bee pollen mineral elements r, coefficient of correlation; LOD, limit of detection; LOQ, limit of quantification; CV, coefficient of variation; *n = 3.

Table 3 .
Mineral contents of 154 dehydrated bee pollen samples collected from 11 Brazilian states and the Federal District 23 to December 12, decreasing during Summer (December 22 to March 19), Fall (March 20 to June 20), with low, if any production during the Winter season (June 21 to September

Table 4 .
Percentage contribution of the mineral elements to Brazilian recommended daily intakes (RDI) for adults in relation to the mean levels found in the 154 Brazilian bee pollen samples 28Brazilian Ministry of Health.24Considering a 25 g portion.29*DietaryRerence Intakes (DRI), according to Institute of Medicine,28not established by Brazilian regulation.