The objective of the present work was to investigate the volatile profile of multifloral honeys collected from different regions in Cyprus, and whether these specific volatiles could serve as true markers of geographical origin. An effort to highlight differences in the volatile amounts of Cypriot honeys according to altitude of harvesting was also carried out. Aqueous solutions of honeys were subjected to headspace solid phase microextraction coupled to gas chromatography/mass spectrometry analysis (HS-SPME/GC-MS). Results showed that 25 volatile compounds of different classes were identified in all samples by 100% frequency rate. Application of multivariate analysis of variance (MANOVA) showed that 3 volatiles [1-(2-furanyl-ethanone), cis-linalool oxide, para-cymenene] recorded significant variations (p < 0.05) according to geographical origin. Discriminant function analysis (DFA) classified honeys according to geographical origin providing a total classification rate of 85% using the original and 70% the cross validation method based on the aforementioned volatiles. It is possible then for an analyst to characterize the geographical origin of honey, even in cases when multifloral honeys are subjected to research/analysis. Regarding the effect of altitude on Cypriot honey aroma, results showed that the lower the altitude of harvesting the richer the aroma of honeys.
1.
Introduction
The determination of geographical and botanical origin of natural based foods is of great concern to the authorities and organizations that handle food distribution and set specific guidelines and rules for products' identity. Of course, a similar concern on food authentication is exhaustively shown in the recent literature in which numerous articles that cover this topic have been published. A great number of the aforementioned articles deal with the aroma profile of foodstuffs [1,2,3,4,5,6,7,8].
The determination of geographical and botanical origin of natural products has flourished the past decades, as analytical methodologies and chemometric techniques have shown a great evolution. Honey, the sweet product of honeybees (Apis mellifera) has earned great attention in terms of its authentication and adulteration control [2]. Hence, the authentication of a natural based product may be defined as the uniqueness in its properties and chemical composition. Analytical methodologies such as solid phase microextraction in combination with gas chromatography-mass spectrometry (SPME/GC-MS), high performance liquid chromatography (HPLC) or high performance liquid chromatography in combination with tandem mass spectrometry (HPLC/MS-MS), nuclear magnetic resonance (13C NMR/1H NMR), etc., may provide actual data on honey composition in terms of aroma [1,2,3,4,5,6,7,8], minerals [9,10], saccharides [11], phenolic compounds or any other fingerprint/chemical marker related to honey authenticity [12].
Honey may be classified into two main categories: blossom or nectar honeys and honeydew honeys. There are substantial differences among these honey types including pollen contribution, physicochemical composition and sensory properties [14].
At the same time, the plant microflora that dominates a region may be termed as 'characteristic of the region'. However, there is a possibility of a honey type to be of mixed flowers since honeybees cannot be restrictedly controlled and kept only in a specific place while collecting nectar from a wider area. If so, we can easily understand the differences in the pollen, physicochemical and sensory properties of such honey types compared to monofloral ones. In addition, parameters such as sunlight, rainfall, humidity, altitude, bee species, etc. may affect honey composition and plant contribution in the final product [10].
Regarding the preference among consumers, monofloral honeys are normally preferred by consumers since these have a well defined composition and characteristic sensory properties. On the other hand, multifloral ones seem to have a lower price in the market and lesser acceptability.
A fast screening of the literature points out that there are lesser studies on the characterization and geographical differentiation of multifloral or less common honey types [15,16,17]. As a matter of fact, studies on the authentication of Cypriot honeys using volatile compounds analysis are scarce.
Therefore, the objective of the present work was to differentiate multifloral Cypriot honeys according to geographical origin using volatile compounds data in combination with chemometrics. The impact of altitude during honey harvesting was also monitored.
2.
Materials and methods
2.1. Honey samples
Twenty multifloral honey samples were collected directly from beekeepers during two consecutive harvesting seasons (2011–2013) from different regions in Cyprus. Respective number (N) of honey samples that were collected from Limassol was N = 5, from Larnaca N = 10, and from Nicosia N = 5. The contribution of plant taxa and harvesting year of honey samples are given in Table 1.
2.2. Headspace solid phase microextraction coupled to gas chromatography/mass spectrometry analysis (HS-SPME/GC-MS)
2.2.1. HS-SPME
Volatile compounds of multifloral honeys were extracted using a divinyl benzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) fiber 50/30 μm (Supelco, Bellefonte, PA, USA). The fiber was firstly conditioned for 1h at 270 ℃ in the injection port of the GC unit. The optimum SPME analysis conditions were as follows [7,15]: 2 g of honey was placed in a 15 mL screw-cap vial equipped with PTFE/silicone septa, with 2 mL of distilled water, and 0.20 g NaCl (Merck, Darmstadt, Germany). Finally, 20 μL of internal standard (benzophenone, 100 μg/mL, Sigma-Aldrich), were placed in the aqueous honey solution obtained. The vials were then sealed and maintained at 45 ℃ in a water bath under continuous stirring at a speed regulation of 600 rpm during the headspace extraction. A magnetic stirrer (cross shaped PTFE-coated magnetic stirrer, diameter 10 mm, Semadeni, Ostermundigen–Bern, Switzerland) was placed inside the vials to assist in the extraction of volatiles. The fiber was removed from the vial at the end of extraction, and was inserted into the injection port of GC for 15 min. Honey samples were prepared daily prior to HS-SPME/GC-MS analysis and the fiber was cleaned exhaustively prior analysis, using the clean program method (22.5 min). Each sample was run in duplicate and the results were averaged [16].
2.2.2. GC-MS analysis
The GC-MS analysis was performed on an Agilent 7890A gas chromatograph coupled to an Agilent 5975 mass detector with an electron ionization source. The chromatographic separation of volatile compounds was accomplished using a DB-5MS (cross linked 5% PH ME siloxane) capillary column (60 m × 320 μm i.d., ×1 μm film thickness). The GC oven temperature program followed the sequence:i) held at 40 ℃ for 3 min, ii) then was increased to 260 ℃ at 8 ℃/min (6 min hold). The injector and MS-transfer line were maintained at 250 ℃ and 270 ℃, respectively. The mass spectra were recorded from m/z 50 to 550 at ionization energy of 70 eV. Finally, helium was used as the carrier gas (purity 99.999%), at a flow rate of 1.5 mL/min.
2.3. Identification and semi-quantification of volatile compounds
The identification of volatile compounds was achieved using the Wiley 7, NIST 2005 mass spectral library. In addition, a mixture of n-alkanes (C8–C20) dissolved in n-hexane was employed in order to calculate the linear retention time indices (Kovats indices). The mixture was supplied by Supelco (Bellefonte, PA, USA). The calculation was carried out for volatiles eluting between n-octane and n-eicosane. Volatile compounds having ≥90% similarity with Wiley library were tentatively identified using the GC-MS spectra. Data were then expressed as concentration (Canalyte,
μg/kg) according to the eq. 1:
where Eanalyte and Ebenzophenone are the peak areas of the isolated volatile compounds and internal standard; whereas CIS is the final concentration of internal standard inside the aqueous honey solution, assuming a response factor equal to 1 for all the volatile compounds [16]. Benzophenone (m/z = 182) proved to be an ideal internal standard since it did not cause any co-elution problems. It should also be marked that this compound did not comprise a 'natural' honey volatile compound. The reproducibility, that is the closeness of the agreement between the results of consecutive measurements of the same measurand carried out with same methodology described in work, of the metabolite semi-quantification with respect to sample preparation and analysis was ≥90%.
2.4. Statistical analysis
Semi-quantitative data of volatile compounds were subjected to multivariate analysis of variance (MANOVA) in order to investigate which are the significant ones (p < 0.05) that could differentiate Cypriot multifloral honeys according to geographical origin. MANOVA may be termed as a generalized form of univariate analysis of variance (ANOVA) although, unlike univariate ANOVA, it uses the covariance between outcome variables in testing the statistical significance of the average differences [18]. Volatile compounds were taken as the dependent variable, while geographical origin was taken as the independent variable. Wilk's Lambda and Pillai's Trace indices were computed to determine/or not a significant effect of volatile compounds on geographical origin of multifloral honey samples. Discriminant function analysis (DFA) was then applied only to the significant volatile compounds to explore the effectiveness, in terms of prediction rates, of differentiating multifloral honey samples according to geographical origin. The original and cross validation methods for the prediction ability were both considered. More specifically, in the original method the correct prediction rate arises from the contribution of all cases (significant parameters) in the discriminant functions. In cross validation, each case is classified by the functions derived from all cases other than that case [7].
3.
Results and discussion
3.1. Volatile compounds of Cypriot multifloral honeys
In Table 2 are listed the 25 volatile compounds that were tentatively identified and semi-quantified. The volatiles that dominated the aroma of multifloral Cypriot honeys were acids, alcohols, aldehydes, esters, hydrocarbons, ketones, and terpenoids. A typical gas chromatogram of multifloral honey sample from Limassol is given in Figure 1.
At this point it should be stressed that in Table 2 some values for the amount of certain components comprise of standard deviations higher than the average. It is owed to the sample variability regarding the specific volatile compound and not the analytical method variability. More specifically, a wide range in the amounts of specific volatile compounds was observed during the analysis of samples (Table 2). Since honey is a natural product, hence, variations in the volatile composition among different samples of the same geographical origin may also occur. This is also confirmed further by Table 1 in which the contribution of plant species among honey samples of the same geographical origin was different.
Numerous of these compounds have been previously identified in blossom and honeydew honeys harvested in different parts of the world [3,4,5,6,15,16,17].
Acetic acid has been reported previously aiding to the volatile fraction of Spanish and Greek thyme honeys, respectively [7,8]. On the other hand, Spanish lavender honey proved to have a higher amount (μg/kg) of acetic acid compared to that of Spanish [8].
Octanoic, nonanoic, decanoic, and dodecanoic acid ethyl esters showed diverse amounts (μg/kg) with respect to multifloral honey geographical origin. Spanik et al. [5] reported the presence of numerous acid methyl esters in rape, sunflower, acacia, lime, raspberry, and phacelia honeys from Slovakia. Acetic acid ethyl ester was also identified in Spanish lavender and thyme honeys [8].
The same trend, in terms of diversity, showed linear aldehydes such as octanal, nonanal, and decanal. Such aldehydes are typical honey aroma compounds since these have been identified in numerous honey types [3,4,5,6,7,8,9]. For instance, the amounts of octanal and decanal were higher in Larnaca region.
Phenolic aldehydes such as benzaldehyde and phenylacetaldehyde were identified in higher amounts in multifloral honeys from Limassol region. Phenolic volatiles may be characterized as a major class of volatile compounds included in the typical volatile compounds of honey. Variations in their amounts have been reported in relation to botanical and geographical origin of honey [3,7,16].
The ketones, 1-hydroxy-2-propanone and 1-(2-furanyl)-ethanone have been reported previously to enhance the aroma of monofloral thyme honey from Greece and Spain [7,8]. 1-hydroxy-2-Propanone recorded higher amounts (μg/kg) in multifloral honeys from Limassol region. On the other hand, 1-(2-furanyl)-ethanone recorded higher amounts (μg/kg) in Nicosia region.
Linear hydrocarbons such as octane and nonane are typical volatile markers of honey. Octane recorded higher amounts (μg/kg) in multifloral honeys from Nicosia, whereas nonane recorded higher amounts in samples from Limassol. In a previous study involving Italian Citrus spp. honeys decane was the dominant hydrocarbon that was identified. [4].
Terpenoids are among the major class of volatiles found in numerous natural based products, and enhance the aroma of honey by a distinctive note. Volatiles such as alpha-pinene, ortho-cymene and para-cymenene, cis-linalool oxide, delta-3-carene and 1, 8-cineole dominated the volatile fraction of multifloral honeys from Cyprus. Terpenoids have been also identified in Italian, Slovak, Greek Spanish and Irish floral honeys [4,5,6,8,19].
Furan or pyran derivatives are quite common volatiles that are found in honey. These compounds may originate from the thermal processing of honey during the headspace extraction [16].
3.2. Differentiation of multifloral Cypriot honeys according to geographical origin based on specific volatile compounds' semi-quantitative data
Data of volatile compounds (semi-quantitative results) originating from all the analyzed multifloral honey samples were subjected to MANOVA analysis to determine which volatile compounds could effectively contribute to geographical differentiation (p < 0.05). Dependent variables included the 25 volatile compounds while geographical origin was taken as the independent variable. Wilks' Lambda and Pillai's Trace indices showed that there was a multi-significant effect of volatiles on honey geographical origin. Three of the 25 volatile compounds (Table 2) were found to be significant (p < 0.05) for the geographical discrimination of multifloral Cypriot honeys (Wilks' Lambda = 0.327, F = 3.738, df = 6,
p = 0.007; Pillai's Trace = 0.733, F = 3.083, df = 6,
p = 0.017). Afterwards, these 3 volatiles were subjected to DFA.
Discriminant function analysis results showed that two discriminant functions were formed. However, only one was significant and could be used for the differentiation of multifloral honeys according to geographical origin: Wilks' Lambda = 0.327, X2 = 17.864, df = 6,
p = 0.0007 for the first function and Wilks' Lambda = 0.906, X2 = 1.581, df = 2, p = 0.454, for the second. The first discriminant function recorded the higher eigenvalue (1.767) and canonical correlation of 0.799, accounting for 94.4% of total variance. The second discriminant function recorded a much lower eigenvalue (0.104) and canonical correlation of 0.307, accounting for 5.6%.
In Figure 2 it is shown that the geographical regions are satisfactorily separated. The overall correct classification rate was 85% using the original and 70% using the cross validation method, considered a satisfactory rate for this method of "modeling" (Table 4). The unstandardized canonical discriminant functions evaluated at group means (functions at group centroids) are also shown in Figure 2 and were (−1.183, −0.427) for Limassol; (−0.436, 0.278) for Larnaca; and (2.056, −0.128) for Nicosia. The best classification results (100%) were obtained for multifloral honey samples from Limassol, followed by those of Nicosia (80%) and Larnaca (50%). Discriminant function coefficients are given in Table 3. In that sense the canonical discriminant functions based on the 3 volatiles may be written as follows:
The structures of the marker compounds are given in Figure 3. Classification result showed that geographical differentiation of Cypriot multifloral honeys based on specific volatile compounds was in general satisfactory. We should not forget that the discrimination of multifloral honeys according to geographical origin may comprise a more difficult approach compared to research carried out on monofloral honeys. In addition, botanical origin differentiation of honey is normally a simpler procedure and may reach classification rates of ca. 100%
[6].
Stanimirova et al. [3] showed that 26 volatile compounds such as benzaldehyde, octanal, nonanal, decanal, para-cymene, nonanoic acid ethyl ester, decanoic acid ethyl ester, etc., were the marker volatiles for the classification of Corsican and non Corsican honeys by application of DFA. The classification rate was 86.4%.
Karabagias et al. [7] classified Thymus capitatus L. honeys according to geographical origin based on volatile compounds such as formic acid, acetic acid, 1-hydroxy-2-propanone, octane, decanal, etc. and reported a classification rate of 64.3% using DFA and the cross-validation method.
Escriche et al. [8] classified Spanish lavender and thyme honeys according to botanical origin, based on volatile compounds such as hexanal, 1-hexanol, hotrienol, acetic acid, etc. achieved a correct prediction rate of 85.7% using discriminant function analysis.
On the other hand, environmental factors should be also taken into account. Indeed, the altitude of honey harvesting was also an important parameter that showed fluctuations with respect to the total volatile content or individual volatile content (semi-quantitative data) of Cypriot multifloral honeys. The non significant differences (p < 0.05), hence, may be probably due to the limited honey samples grouped into the three zones or the multifloral nature of honey samples analyzed.
It is very important then for a researcher that works in the field of honey authentication to test the classification models by using in the analysis, among other varieties, mixed floral honey types. In addition, when data regarding environmental factors are available, this should be also considered in the final report. In that sense, more rigorous honey authentication approaches may be planned. The present study addresses also this questioning.
3.3. The impact of altitude on total volatile semi-quantitative data of Cypriot multifloral honeys
In order to test whether the altitude could affect volatile compounds' data (semi-quantitative data) honey samples originating, hence, from different regions were classified into three groups: Group A: honey samples harvested at 0–300 m (N = 8); Group B: honey samples harvested at 300–600 m (N = 6); and Group C: honey samples harvested at 600–1000 m (N = 6). Before going any further it is important to stress that this procedure may imply a useful strategy for the evaluation of the impact of altitude on the volatile profile of honey, since grouped honey samples originated from the 3 different regions in Cyprus. Thus, present data may be well generalized in a larger population of samples. For instance, this study could be extended, i.e. to the entire Mediterranean area to a larger number/population of samples, from more wide-spread sampling in order the obtained results to be more accurate.
Therefore, the collected data may be characterized as preliminary in nature. However, results showed that the lower the altitude, the richer the aroma of honeys (Figure 4). This finding shows that the contribution of aromatic plants is favored in regions of lower altitude, compared to those with higher one. This is extremely important information for beekeepers or any other authorities that are involved in honey production; that is the production of a richer in aroma honey. Therefore, the altitude of honey harvesting may comprise an additional criterion for the quality of honey. As we mentioned above, future research using a larger number of honey samples harvested at different altitudes will enhance further present results.
4.
Conclusions
The volatile profile of Cypriot multifloral honeys was investigated for the first time. Results showed that volatile compounds of Cypriot honeys were affected by geographical origin, involving at the same time the plant contribution in honey. The application of HS-SPME/GC-MS proved to be an assistive tool for the differentiation of Cypriot honeys according to geographical origin using DFA. Three marker compounds were found: 1-(2-furanyl-ethanone), cis-linalool oxide and para-cymenene. Eventhough, the classification rate obtained using the cross-validation method was lower compared to previous works [3,6], the present study contributes to the state of the art by showing that geographical origin differentiation of multifloral honeys is a more difficult procedure than that of monofloral honeys [3]. In addition, some promising preliminary data regarding the effect of altitude on honey aroma were also collected.
Acknowledgments
The present study was prepared for publication during the Post Doctoral studies of Dr. Ioannis K. Karabagias. Therefore, Prof. M.G. Kontominas is acknowledged for the access he provided to the laboratory of Food Chemistry at Chemistry Department of University of Ioannina, 45110, Greece. We would like also to thank beekeepers from Cyprus for the donation of honey samples.
Conflicts of Interest
The authors declare no conflicts of interest.