Palynological and physicochemical characteristics of three unifloral honey types from central Argentina

The characteristics of 59 unifloral honeys of Condalia microphylla Cav. (“piquillín”), Centaurea solstitialis L. (“yellow starthistle”) and Prosopis spp., from La Pampa, Argentina, were studied. Pollen features (abundance and frequency of pollen types) and some physicochemical parameters (colour, electrical conductivity, free acidity, glucose content, glucose:water ratio, moisture and pH) were determined. Two different but related sets of calculations were done: the first involved single-factor variance analysis, while the second set involved two multivariate methods, principal component analysis and cluster analysis. Variance and multivariate analysis allowed differentiation of the three honey types according to their physicochemical properties. The variables that best explained this differentiation were pH, electrical conductivity, colour, glucose content and the glucose:water ratio. Pollen analysis showed that the pollen frequency traditionally used (> 45%) for a botanical origin assignment in honey was not valid for the unifloral honeys studied. Therefore, pollen analysis should be combined with the above physicochemical analysis order to obtain a successful classification of these unifloral honeys. Additional key words: botanical origin, Centaurea solstitialis, Condalia microphylla, melissopalynology, multivariate analysis, pollen analysis, Prosopis spp.


Introduction
Honey is a natural substance produced by bees (Apis mellifera L.) from flower nectar, and from honeydew.The composition and properties of honey depend on the botanical origin of the nectar or secretion used.Consequently, its composition is influenced by many factors and is subject to variation.Several studies have attempted to establish suitable parameters for honey from the same botanical source.Honey is characterized by its palynological, chemical and physical properties.
Pollen analysis appears to be the most frequently used method of honey identification (Louveaux and Vergeron, 1964;Louveaux et al., 1978;Anklam, 1998;von der Ohe et al., 2004), although in some honeys it is difficult to establish their exact origin (von der Ohe, 1994;Hermosin et al., 2003).
As honey is a complex natural food, clear characterization of honey samples requires the use of several parameters.To establish the combinations of these parameters closely related to the origin of honey quality control methods and multivariate statistical analysis need to be used.These methods will help to evaluate honey samples in their totality and give more precise classifications (Anklam, 1998;Ruoff et al., 2007).
The objective of this work was to characterize unifloral honeys from three different botanical sources produced in La Pampa Province, Argentina.This was done using data from melissopalynological and physicochemical analysis to attempt the classification of honey samples according to their botanical origin.

Study area
Unifloral honeys from Prosopis spp.and Condalia microphylla are from the Caldén District of Espinal Phytogeographical Province.Honeys from Centaurea solstitialis L. are from Pampean Province (Cabrera, 1994).
The Caldén District -usually called Caldenal -spreads over the central semiarid region of Argentina.In central Argentina Centaurea solstitialis ("yellow starthistle") is a winter annual or biennial adventitious species.In natural areas and on rangelands it forms dense impenetrable stands that displace desirable vegetation.Thus, yellow starthistle is a principal nectar source during the summer in disturbed areas of steppe and caldenal.

Sample collection
Fifty nine (n = 59) typical honey samples, from Apis mellifera, were collected by beekeepers in 2003, 2004 and 2005.They were obtained by centrifugation and stored at room temperature until analyzed.
The honeys were harvested from different areas of La Pampa Province, Argentina (Fig. 1).
Sensory analysis (crystallization type) was considered as a complementary criterion.Some Prosopis spp.and Condalia microphylla honeys were rejected, in the first case because of creamy crystallization and the second because of non-crystallization.Both cases showed atypical characteristics of these honey types.

Pollen analysis
The pollen spectrum of the honey samples was determined using the acetolytic method (Erdtman, 1960) and the method of Louveaux et al. (1978).Honeys from central Argentina show very few honeydew elements; therefore they were not calculated (Tellería, 1996;Andrada, 2001;Fagúndez and Caccavari, 2003).The different pollen types were identified by comparing them with a reference collection, made from plants of the area.The preparations were deposited in the Palynological Collection of the Facultad de Agronomía, Universidad Nacional de La Pampa.The identified pollen was classified, according to frequency, into four classes: dominant (> 45%) = D, secondary (16-45%) = S, important minor (3-15%) = M, trace (<3 %) = T (Louveaux et al., 1978).To determine frequency class, 1000 pollen grains were counted.
The absolute pollen content (APC) of the honey samples (i.e., the number of pollen grains in 10 g of honey) was calculated using tablets of Lycopodium spores (Stockmarr, 1971).Following Louveaux et al. (1978) five groups were considered: Group I (honey low in pollen < 20,000/10 g), Group II (normal honey 20,000-100,000/10 g), Group III (honey rich in pollen 100,000-500,000/10 g), Group IV (honey extremely rich in pollen 500,000-1,000,000/10 g), Group V (pressed honeys >1,000,000/10 g).Quantitative analysis of honey samples with over -or under-represented pollen was conducted according to Moar (1985) who suggested standard honey as Trifolium repens L. honey, which has 45% of dominant pollen and is in Group II.Moar (1985) also explained how to estimate an adjusted absolute pollen content and an adjusted minimal pollen percentage for a honey sample to be classified as unifloral.
The honey samples were analyzed using the following methods: -Colour was determined with a Coleman spectrophotometer by reading the absorbance in aqueous solutions at 635 nm (10 g honey in 20 mL water).
Table 1 shows honey colours and their absorbance and mm Pfund values, obtained using the following algorithm (Bianchi, 1990): mm Pfund = -38.7 + 371.39 x Absorbance.-Electrical conductivity (EC, mS cm -1 ) was determined with a Luftman C400 conductivity meter in 20 (w/v) aqueous honey solution (dry matter basis).-Free acidity: acidic components were neutralized with a standard solution of sodium hydroxide in aqueous honey solution (10 g in 75 mL of double distilled water).-Glucose content was determined by the glucose oxidase -peroxidase method (GOD-POD).Absorbance was measured at 595 nm using in a Metrolab 1700 spectrophotometer.-Glucose:water ratio (G:W) was obtained from water and glucose content percentage (White et al., 1962).-Moisture was determined with an Abbe-type refractometer.The refractive index was correlated using Chataway Charts.-Active acidity (pH) was determined, in aqueous solution, with a Horiba B-213 pH meter.

Statistical analysis
Analysis of variance and multivariate analysis were performed using Statgraphics Plus 3.1 software.Differences among means were determined for significance at P = 0.05 using the least significant difference (LSD) test.Principal components analysis (PCA) and cluster analysis (CA) were used to reduce the dimensions of the 7 x 59 data matrix, to determine relationships between physicochemical properties (variables) and honey samples (objects) through optimal graphical 2-D and to define groups between unifloral honeys.To determine similarities among samples and variables, the Euclidean distance and group average method were used.

Pollen analysis
Table 2 shows the frequency of occurrence of pollen types in the three unifloral honeys.A total of 71 pollen types, from 35 plant families, were identified in the honey samples analysed.
In Prosopis spp.honeys 43 pollen types were identified with 5 to 20 types per sample.Brassicaceae and Schinus spp.were present as secondary pollen.In the minor importance class were Centaurea solstitialis, Vicia spp., Eucalyptus spp., Condalia microphylla and Lycium spp.
In Condalia microphylla honeys 31 pollen types were identified with 5 to 18 types per sample.These honeys were characterized by Vicia spp.and Eucalyptus spp. as secondary pollen and Brassicaceae, Schinus spp., Prosopis spp.and Larrea spp.pollen being of minor importance.
Table 3 shows the adjusted (APC adj , DP adj ) and nonadjusted (APC, DP) values for absolute pollen content and the percent dominant pollen for each honey type.
The absolute pollen content indicated that pollen richness is a characteristic of Condalia microphylla honeys (Group III), whereas Prosopis spp.honeys belong to Group II and Centaurea solstitialis honeys to Group I respectively.The percent dominant pollen adjusted (DP adj ) according to Moar (1985) indicated that the C. solstitialis honeys require 31.5% of their pollen to be considered unifloral.The Condalia microphylla and Prosopis spp.honeys would require the 64.5% and 75% respectively.honey samples.All parameters showed high discrimination power in these honeys.However, moisture content only differentiated Prosopis spp.honeys from the others.In terms of colour, pH, free acidity and EC Condalia microphylla honeys had the highest values while Centaurea solstitialis and Prosopis spp.honeys had the highest glucose content and G:W ratio.

Multivariate analysis (CA and PCA)
Cluster analysis showed that there were two clusters at a linkage distance level of 6 corresponding to the different botanical origins (Fig. 2).From right to left, the first cluster is composed of Condalia microphylla honey samples.The second cluster has two sub-clusters, one composed of Centaurea solstitialis samples and the other of Prosopis spp.samples.
Multivariate CA of variables using the group average method and squared Euclidian distance showed two distinct groups.One group was pH, conductivity, colour and free acidity and the other group was glucose content, G:W ratio and moisture (Fig. 3).
To cluster the three botanical types of honey based on their physicochemical properties, a standardized PCA was applied.Principal component 1 (PC 1 ) and PC 2 accounted for 76.8 (i.e., 53.1 + 23.7) of the total variance.Fig. 4 shows the correlation circle where moisture content is located near the origin of two PCs; this variable does not influence group formation in Fig. 5.The first component was positively correlated to colour, EC and pH and negatively related to glucose content and the G:W ratio.The second component was positively correlated to pH and EC and negatively correlated with free acidity.Condalia microphylla honeys had high, positive, PC 1 scores (reflecting dark colour and high values of EC, free acidity and pH).Centaurea solstitialis and Prosopis spp.honeys had high negative PC 1 scores (reflecting high glucose content and G:W ratios).The last variable is related to fast granulation observed in the Centaurea solstitialis and Prosopis spp.honeys, while Condalia microphylla honeys had low, or no granulation (personal observation).Despite Prosopis spp.and Centaurea solstitialis honeys appearing very close on the biplot, they still formed two different groups; the first on the positive side and the second on the negative side.
Prosopis spp.honeys belong to Group II and would require 64% dominant pollen to be considered a unifloral honey because their dominant pollen is slightly overrepresented in the samples.Thus, the usual pollen frequency of > 45% to assign honey botanical origin is not valid for the unifloral honeys studied in this work.
With regard to physicochemical analysis all the variables analyzed are widely recognized in the evaluation of the botanical origin of honey, Honey colour is closely linked to botanical origin is used for honey classification.Generally, colour is related to sensory properties such as flavour and odour.Several factors can influence honey colour such as floral source, mineral content and storage conditions (Tha-wley, 1969;Salinas et al., 1994).Prosopis spp.honeys were a white water colour (± 7.9 mm Pfund) while Centaurea solstitialis honeys were white (± 22 mm Pfund) and Condalia microphylla honeys were dark and ranged from light amber to dark (> 140 mm Pfund).The honey colour of the C. microphylla and Prosopis spp.honey agreed with the results of Andrada (2001).
Honey EC is closely related to the mineral concentration (total ash), salts, organic acids and protein.The EC varies greatly with honey floral origin because the conductivity and the ash content depend on material collected by the bees during foraging (Terrab et al., 2002;Serrano et al., 2004;Ouchemoukh et al., 2006).The EC results varied widely depending on honey type.The C. microphylla honeys had the highest EC values compared with the Prosopis spp.and Centaurea solstitialis honeys.
Variation in free acidity among different honeys can be attributed to floral origin (El-Sherbiny and Rizk, 1979) or to variation due to the harvest season (Pérez-Arquillué et al., 1994).Free acidity differed widely among the three honey types, it was lowest in the Prosopis spp.honeys while Condalia microphylla honeys had the highest values.Andrada (2001) reported free acidity values in C. microphylla honeys which were lower than in these samples.This could be related to the presence of secondary pollen from Prosopis spp. in those honeys.
The glucose content of any honey type depends largely on nectar source (Anklam, 1998).Honey samples of different botanical origin had a wide range of glucose content.Values in Condalia microphylla honeys were low as found by Tamame and Naab (2003).The granulation rate and the tendency to granulate are directly related to parameters such as the glucose, water and fructose content (White et al., 1962;Manikis and Thrasivoulou, 2001).The average ratio of glucose:water is a criterion for the prediction of granulation tendency; its application to these honeys showed it was a good predictor of granulated and non-granulated honeys as found by Manikis and Thrasivoulou (2001) in Greek honeys.The low G:W ratio in the C. microphylla honey samples (≤1.7) confirms that these are less prone to granulation and would remain liquid for longer periods.The Prosopis spp.and Centaurea solstitialis honeys presented G:W ratio ≥2.1.However, Prosopis spp.samples showed faster crystallization than C. solstitialis honeys.
In non adulterated honeys the moisture content depends on botanical origin, harvest season, processing techniques and storage conditions.The moisture content of the samples indicated a proper degree of maturity in agreement with international requirements (Codex Alimentarius, FAO-OMS, 1990).The relatively high moisture values in Prosopis spp.honeys could be due to the early, spring harvest.Basilio and Nöetinger (2002) reported a similar moisture content in Prosopis spp.honeys from the Chaco Region of Argentina.On the other hand, the low moisture in the Centaurea solstitialis honeys can be related to the low relative humidity of the semiarid climate of the study area as found by Andrada (2001).
Floral honeys are acidic, with a pH of 3.0 to 4.3 (Bogdanov et al., 1999).The pH values in these samples accorded with the acceptable range for floral honeys.Condalia microphylla honeys had significantly higher pH values than the other honeys.
The multivariate analysis offered gave quantitative results for the classification of unifloral honeys in agreement with Ruoff et al. (2007).
The PCA and CA showed that selected chemical parameters (colour, EC, free acidity, glucose content, G:W ratio, moisture and pH) provided enough information to develop a botanical classification for honeys studied.Consequently, determination of the chemical properties of unifloral honeys can be a useful tool to complement melissopalynological studies.Quantitative pollen analysis showed that the usual pollen frequency (> 45%) for a correct botanical origin assignment in honey was not valid for the unifloral honeys studied.
All the analyzed honeys had excellent quality properties according to Argentinean and International standards (Codex Alimentarius, FAO-OMS, 1990;Bogdanov et al., 1997).
The results of this study allowed the classification of three central Argentinean unifloral honeys and justify the use of these parameters with other Argentinean unifloral honeys.

Figure 1 .
Figure 1. A. Geographical location of La Pampa Province, Argentina; B. La Pampa Province subdivided into departments.The study area is gray.

Figure 3 .
Figure 3. Dendrogram of cluster analysis (CA) of physicochemical variables (group average method, squared Euclidean distance).Ordinate shows distance units.

Figure 5 .
Figure 5. Ordination from principal component analysis of 59 honey samples from three botanical origins by seven physicochemical properties.Samples are located in the space of the two first principal components.

Table 1 .
Table4shows the results (mean, standard deviation and range) obtained from physicochemical analysis of the Honey colour expressed in absorbance and mm Pfund values

Table 4 .
The physicochemical parameters of the honey samples.Upper line: mean ± SD.Lower line: range (minimum and maximum).Any mean in a row followed by different letters are significantly different Figure 2. Dendrogram of cluster analysis (CA) of honey samples (group average method, Euclidean distance).Ce, Centaurea solstitialis; P, Prosopis spp.; Co, Condalia microphylla.Ordinate shows distance units.