Introduction

Phenolic compounds rank among the most important compounds present in honey. They exhibit many biological effects, including their actions as antioxidant, antibacterial, anti-inflammatory, and anti-allergic agents. Therefore, honey, despite being considered a highly nutritious natural product, is also used for the prevention and treatment of many diseases. In recent decades, different methods have been implemented for the authentication of honey (Karabagias et al. 2014, Dinca et al. 2015, Ciulu et al. 2016, Pascual-Maté et al. 2017), and the analysis of phenolic compounds is well documented as a very promising method for determining the floral and geographical origins of honey. In general, an analytical procedure for the determination of phenolic compounds includes three steps: isolation, separation and identification, and the most common method of separation is high-performance liquid chromatography (HPLC) or capillary electrophoresis (CE) (Pyrzyńska and Biesaga 2009). In addition, statistical methods, such as hierarchical cluster analysis, principal component analysis, or discriminant analysis are increasingly introduced for the study of the usefulness of different analytical technique used to authenticate honey (Kečkeš et al. 2013; Kuś et al. 2014; Kasprzyk et al. 2018).

This study presents the determination of the phenolic profiles of various honey samples by using high-performance thin layer chromatography (HPTLC) as a detection method. Although HPLC still remains the most dominant technique for the separation and identification of phenolic compounds (Badjah Hadj Ahmed et al. 2016, Pascual-Maté et al. 2017), HPTLC seems to be a very promising alternative for the analysis of these compounds in honey (Locher et al. 2017). This is because HPTLC is simpler, more flexible, more accessible, and cheaper than other commonly used methods. Furthermore, HPTLC allows a parallel separation and quantitative determination of many samples at the same time (Ristivojević et al. 2017). HPTLC chromatograms can be used to determine the differences in botanical origins of honey samples by exhibiting a characteristic set of common bands that is typical for each analyzed branch. Thus, the entire chromatogram is treated as a unique, distinctive fingerprint for each selected type of honey. Chromatograms could, thus, complement quality-control tools for honey. An equally important feature of the HPTLC analysis of phenolic compounds is the detection of individual chemical species that are present in certain honey. Separated substances can be detected by determining the color of the substance, the fluorescence of the substance in UV light, or the color reactions of separated substances. Individual classes of phenolic compounds exhibit different spot coloration under UV light with visualizing reagents. Specifically, flavonoids, like quercetin, show orange or yellow spots, while flavones appear dull brown. Anthocyanins are detected as purple-red-orange or sometimes yellow bands in visible light. Light and dark blue spots are characteristic for caffeic, p-coumaric, and ferulic acids. On the other hand, simple phenolics, such as gallic or salicylic acids, do not fluoresce under UV light and quench to give dark absorbing spots (Waterman 1994). Thus, Głowniak et al. 2005 analyzed phenolic compounds in the flowers of Lavatera trimestris L. The fluorescence colors of phenolic acid and flavonoid standards were blue or light blue for caffeic, p-coumaric, ferulic, and chlorogenic acids. Additionally, caffeic, ferulic, and chlorogenic acids, after spraying with a 2% aqueous solution of ferric chloride, produced brown-green red, light-orange, and gray-green spots, respectively. Milojković Opsenica et al. 2016 analyzed phenolic compounds in propolis samples. Propolis of so-called O-type showed orange (typical for flavonoids), light blue, and some green bands. Orange, green-brown, light blue, and green bands were further identified as quercetin, chrysin, caffeic acid, and naringenin, respectively. It is noteworthy that only one paper has been devoted to the analysis of the chemical composition of honey by means of HPTLC. Locher et al. 2017 explored HPTLC fingerprinting for the authentication of honey, such as jarrah (Eucalyptus marginata), marri (Corymbia calophyla), and manuka (Leptospermum scoparium), but they did not study the use of the HPTLC method for the determination of phenolic compounds.

The aim of the present work was fingerprinting of phenolics by HPTLC to determine their characteristic patterns in willow, buckwheat, heather, pine honeydew, and manuka honey.

Materials and Methods

Honey Samples

Seventeen honey samples of five different varieties of honey (willow, heather, buckwheat, pine honeydew, and manuka) were analyzed. The types of pollen present in honeys analyzed are shown in Table 1. Two categories of honey were distinguished: pure honeys (W2-W3, H1-H2, B1, M1-M2) and honeys with pollen content slightly lower than 45% (W1, B2-B4). In the case of pine honeydew honeys, the authentication is not carried out on the basis of the pollen grains, and their classifications is made by using electrical conductivity parameter. According to Polish Quality Standard (PN-88/A-77626 1998) and Bogdanov et al. 2004, the electrical conductivity should be higher than 800 μS/cm, then honey is a honeydew. The all honeydew honey samples had an electrical conductivity in the range of 1160–1580 μS/cm. The honey samples were of different harvesting years (2015–2016) and geographical origins (Poland, New Zealand, and Australia) (Table 1). The products were stored at 4 °C in the dark.

Table 1 Characteristic of studied honey samples
Full size table

Chemicals and Reagents

All chemicals and solvents used for the chromatography column, HPLC and HPTLC analyses were of analytical purity grade. Methanol, ethanol dibasic sodium phosphate heptahydrate (Na2HPO4 × 7H2O), 85% phosphoric acid, chloroform, ethyl acetate, and formic acid were purchased from POCH S.A. (Gliwice, Poland). The standards (chrysin, myricetin, 3,4-dihydroxybenzoic acid, caffeic acid, chlorogenic acid, ellagic acid, ferulic acid, gallic acid, p-coumaric acid, rosmarinic acid, and (±)-abscisic acid) were purchased from Sigma-Aldrich (Poznań, Poland). Amberlite XAD-2, which was used as an adsorption resin for the extraction step, was obtained from Supelco (Bellefonte, PA, USA).

Physicochemical Properties

Selected physicochemical parameters: moisture content, pH, and electrical conductivity have been determined using the Polish Quality Standard (PN-88/A-77626).

Phenolic Extraction (Solid-Phase Extraction, SPE)

The extraction was performed using the method described by Jasicka-Misiak et al. (2017). The honey (25 g) was thoroughly mixed with five parts (125 mL) of acidified water (pH = 2 with HCl) and homogenized in an ultrasonic bath for 60 min. The solutions were then filtered through filter paper to remove any solid particles. Amberlite XAD-2 (Supelco, Bellefonte, PA, USA, pore size 9 nm, particle size 0.3–1.2 mm) was packed into a glass column (2.5 cm × 55 cm) and washed with both methanol and acidified water. Next, the honey extract was transferred to a glass column and kept for 30–45 min. The solid phase was washed with acidified water (0.05 M HCl, 250 mL) and neutral distilled water (300 mL) to remove all sugars and other polar compounds. The polyphenolics were eluted from the sorbent with methanol (500 mL). The methanol extracts were evaporated under reduced pressure at 40 °C. The dried residue was then dissolved in methanol (HPLC-grade).

HPTLC Separation

The 10 μL of the honey extracts and 4 μL of the standard solutions (concentration of 1 mM) were applied as 8 mm bands on HPTLC sheets at rate of 150 nL s−1 using a semiautomated HPTLC application device (Linomat 5, CAMAG, Muttenz, Switzerland). The chromatographic separations were performed on 20 × 10 cm (0.25 thick) TLC plates (HPTLC Sillica gel 60, Merck) using a mixture of chloroform:ethyl acetate:formic acid (5:4:1) as the mobile phase. Visualization was performed in two ways: (1) under long-wave UV light (254 nm, 366 nm) and (2) by spraying with 1% methanolic AlCl3, Ce-P-Mo spraying reagent (10 g of Ce(SO4)2, and 20 g of phosphomolybdic acid in 1 L of 10% H2SO4) (Żyszka et al. 2017) or 1% methanolic 2-aminoethyl diphenylborate (Neu’s reagent), drying, and visualizing under both 254 nm and 366 nm. After spraying, the plates were dried for approximately 3 min in warm air until colored spots appeared. The obtained chromatographic results were analyzed using HPTLC software (visionCATS, CAMAG).

HPLC Separation

An UltiMate 3000 Dionex HPLC system was used to analyze the polyphenolic compounds in the honey samples. The chromatographic separation was achieved using a reserved-phase Gemini 5 μm C18 column (250 × 4.6 mm, particle size 5 μm). The mobile phases were methanol (solvent A) and 0.01 M aqueous phosphate buffer at pH = 2.5 (solvent B) at a flow rate of 1 mL/min in a gradient elution mode. The gradient separation was as follows (A:B ratio): 10–90% t = 0; 40–60% t = 13.5 min; 90–10% t = 39 min; 100–0% t = 42 min; 10–90% t = 55 min. The standards and honey extracts were injected into the system with an auto-sampler, and the phenolic compounds were detected using a photodiode array detector (PDA). Stock solutions of each standard were prepared with HPLC grade methanol at a concentration of 200 μg/ml and were stored at 4 °C while protecting from light.

Method of Validation

Calibration and Linearity

The calibration curves were generated using nine data points: 0.05, 0.1, 0.5, 1, 5, 10, 50, 100, and 200 μg/ml. Standard solutions were prepared in HPLC grade methanol and analyzed in triplicate. Linear calibration plots were obtained over the calibration range of 0.05–200 μg/ml. The correlation coefficient (r2) was found to be > 0.9972 (Table 2) for each regression equation.

Table 2 Correlation coefficients, LOD, LOQ, and precision of standards
Full size table

Precision

The repeatability of the sample application and measurement of the peak area was tested by injecting three replicates of the standard (100 μg/ml) on the same day and over 3 days (three replicates each day). The intra- and inter-day variations were achieved with relative standard deviation (RSD) values lower than 3.3% for all the analyzed compounds (Table 2).

Limit of Detection and Limit of Quantitation

To estimate the limit of detection (LOD) and limit of quantitation (LOQ), a blank sample (methanol) was spiked three times with each standard at low concentration levels, and the signal-to-noise ratio was determined. The LOD was considered as 3:1, and the LOQ as 10:1. The LOD values were between 0.010 and 0.218 μg/ml, while the LOQ values were between 0.029 and 0.662 μg/ml for each standard compound (Table 2).

Statistical Analysis

Statistical analyses were performed using the Statistica 13.1 software (StatSoft). Means were separated by least significant difference test (LSD) and statistical significance was set at α = 0.05. Hierarchical cluster analysis (HCA) and principal component analysis (PCA) were used to compare concentrations of identified compounds of honey samples (phenolic acids, flavonoids, and abscisic acid).

Results and Discussion

Physicochemical Parameters

The physicochemical parameters studied are moisture content, pH, and electrical conductivity. The mean values for each physicochemical parameter are presented in Table 1. The honey moisture content is one of the most important parameters, and its level should not exceed 20% in accordance with international regulations of quality (Codex Alimentarius 2001) and the Polish Quality Standard (PN-88/A-77626 1998). The levels higher than 20% are indicative of the acceleration of fermentation processes during storage (Semkiw et al. 2008, Muruke 2014).

The electrical conductivity values in honey samples are in the range of 0.38–1.58 mS/cm. The highest values (1.16–1.58 mS/cm) obtained for pine honeydew honeys are according to Polish Quality Standard and Codex Alimentarius. The significant difference in electrical conductivity is used to differentiate honeydew honeys from nectar honeys.

Generally, honeys have an acidic character. In the case of honey samples analyzed, the pH ranged between 3.92 and 4.62. These values were similar to those previously reported in other honeys (Fechner et al. 2016, Chuttong et al. 2016). The highest levels of pH were observed in the case of honeydew honeys while willow had the lowest level of pH. Some authors suggest that pH has an effect on the extraction and storage of honey, but not on the botanical origin (Oroian and Sorina 2017).

HPTLC Fingerprints of Honey

Each of the received HPTLC images was evaluated to assess whether they provided useful information for the discrimination of the honey brands. The HPTLC chromatograms showed differences in the chemical composition of the honey samples with different origins. All of the honey extracts were enriched in phenolic compounds, and their patterns were dominated by light blue, blue, navy blue, yellow, and black bands. The HPTLC fingerprint analysis allowed the differentiation of honey samples depending on the fluorescent color of the phenolic compounds and showed that these compounds were present in some honey samples in varying amounts (Table 3).

Table 3 Average Rf values of identified standards separated by HPTLC and their respective colors
Full size table

The fingerprint of certain honeys exhibits a unique profile of bands that correspond to the phenolics and can be considered unique because they visually differ from each other. Fig. 1 shows the typical HPTLC fingerprints of willow honey samples (W1-W3). Major bands at Rf of 0.03 (light blue), 0.05 (light blue), 0.13 (blue), 0.40 (yellow), 0.45 (blue), 0.50 (blue), 0.59 (navy blue), and 0.61 (yellow) under 366 nm irradiation can be seen. Under 254 nm, the most prominent bands are at Rf of 0.03 (light blue), 0.05 (light blue), 0.13 (blue), 0.40 (green), 0.45 (blue), 0.50 (green), 0.52 (black), 0.59 (blue), and 0.61 (green) (Fig. 2). After derivatization with NEU and 1% methanolic AlCl3 spraying reagent, the bands at Rf 0.40 and 0.61 are intensely yellow, and the appearance of new bands at 0.70 (orange) and 0.71 (yellow) can be observed. It is also seen that the phenolic compounds in willow honey change their composition over time from the harvest period. The track profiles for samples of differing harvesting years exhibited the same band profiles, but the older samples (W1 and W2) exhibit bands with a lower intensity (Rf = 0.40). We have identified the following phenolic compounds (based on the similarity of the Rf and the color of the bands): Rf = 0.03 chlorogenic acid, Rf = 0.40 myricetin, Rf = 0.45 caffeic acid, Rf = 0.50 3,4-dihydroxybenzoic acid, Rf = 0.52 abscisic acid, and Rf = 0.59 ferulic acid, and the most intense signals were derived from chlorogenic acid (light blue), myricetin (yellow), and abscisic acid (black).

Fig. 1
figure 1

HPTLC chromatogram at 366 nm, W1-W3 willow honey samples, H1-H2 heather honey samples, M1-M2- manuka honey samples, R references, FA ferulic acid, DHBA 3,4-dihydroksybenzoic acid, M myricetin, ChA chlorogenic acid, Ch chrysin, p-CA p-coumaric acid, CA caffeic acid, EA ellagic acid, RA rosmarinic acid

Full size image
Fig. 2
figure 2

HPTLC chromatogram at 254 nm. W1-W3 willow honey samples, H1-H2 heater honey samples, M1-M2 manuka honey samples, R references, FA ferulic acid, DHBA 3,4-dihydroksybenzoic acid, M myricetin, ChA chlorogenic acid, p-CA p-coumaric acid, CA caffeic acid, EA ellagic acid, RA rosmarinic acid, AA abscisic acid

Full size image

The heather honey samples (H1-H2) exhibit notable bands at Rf 0.03 (light blue), 0.07 (light blue), 0.09 (yellow), 0.13 (yellow), 0.16 (blue), 0.19 (yellow), 0.29 (yellow), 0.34 (yellow), 0.37 (yellow), 0.40 (blue), 0.47 (blue), 0.55 (dark blue), 0.58 (navy blue), and 0.66 (yellow) under 366 nm (Fig. 1). Additionally, visualization with 254 nm results in spots of Rf of 0.03 (light blue), 0.07 (light blue), 0.09 (yellow), 0.13 (yellow), 0.16 (yellow), 0.19 (yellow), 0.29 (yellow), 0.34 (intense yellow), 0.37 (yellow), 0.40 (blue), 0.47 (black), 0.55 (blue), 0.58 (blue), and 0.66 (yellow) (Fig. 2). Based on the color reactions of polyphenols before and after the derivatization, the following components were identified: Rf = 0.03 chlorogenic acid, Rf = 0.34 myricetin, Rf = 0.40 caffeic acid, Rf = 0.47 abscisic acid, Rf = 0.58 p-coumaric acid and Rf = 0.66 chrysin. It is noteworthy that the chromatogram was dominated by the intense yellow and black bands that were identified as myricetin and abscisic acid, respectively. In addition, the heather honey profiles at 366 nm irradiation showed great similarity in the phenolic signals, which indicated that all of the samples belong to the same type of honey (Fig. 5). The strongest signal was observed at Rf = 0.34, which originated from myricetin.

The buckwheat honey samples (B1-B4) exhibited bands of interest at Rf of 0.03 (light blue), 0.05 (dark blue), 0.08 (blue), 0.14 (dark blue), 0.18 (blue), 0.25 (yellow), 0.36 (navy blue), 0.38 (yellow), 0.42 (dark blue), 0.46 (dark blue), 0.54 (dark blue), 0.57 (navy blue), and 0.68 (dark blue) under 366 nm (Fig. 3). UV irradiation at 254 nm, there are bands at Rf of 0.03 (light blue), 0.05 (dark blue), 0.08 (white), 0.10 (dark blue), 0.14 (dark blue), 0.25 (yellow), 0.36 (light blue), 0.38 (yellow), 0.42 (dark blue), 0.46 (dark blue), 0.54 (dark blue), 0.57 (navy blue), and 0.68 (dark blue) (Fig. 4). Two phenolic compounds were identified here: Rf = 0.38 myricetin and Rf = 0.57 p-coumaric acid (Fig. 4).

Fig. 3
figure 3

HPTLC chromatogram at 366 nm. B1-B4 buckwheat honey samples, PH1-PH6 pine honeydew honey samples, R references, p-CA p-coumaric acid, M myricetin, FA ferulic acid, CA caffeic acid

Full size image
Fig. 4
figure 4

HPTLC chromatogram at 254 nm. B1-B4 buckwheat honey samples, PH1-PH6 pine honeydew honey samples, R references, p-CA p-coumaric acid, M myricetin, FA ferulic acid, CA caffeic acid, GA gallic acid

Full size image

Figure 3 shows the various pine honeydew extract chromatograms. Visualization at 366 nm unveils notable bands at Rf values of 0.03 (light blue), 0.05 (blue), 0.11 (dark blue), 0.13 (dark blue), 0.30 (black), 0.34 (orange), 0.38 (yellow), 0.39 (blue), 0.46 (black), 0.51 (navy blue), and 0.66 (dark blue). The derivatized samples (NEU reagent) have a pronounced band at Rf = 0.54 (yellow). UV irradiation with 254 nm, the samples contain additional spots at Rf = 0.08 (light blue), 0.17 (light blue), 0.43 (yellow), 0.51 (dark blue), and 0.54 (yellow) (Fig. 4). HPTLC led to the identification of four phenolic acids, chlorogenic (Rf = 0.03), gallic (Rf = 0.30), caffeic (Rf = 0.39) and ferulic (Rf = 0.51), and one flavonoid, myricetin (Rf = 0.38). As observed in Fig. 5, various pine honeydew honey sample profiles under 366 nm display very similar profiles with key peaks at Rf = 0.03, 0.13, 0.46, and 0.51.

Fig. 5
figure 5

HPTLC honey extract profiles at 366 nm. H1-H2 heater honey samples, PH1-PH6 pine honeydew honey samples, M1-M2 manuka honey samples

Full size image

Figures 1 and 2 illustrate the chromatograms of manuka honey obtained from different suppliers and locations with varying ages (Table 1). Still, the manuka honey extract under 366 nm irradiation exhibits a very similar fingerprint with key peaks at Rf = 0.03 (light blue), 0.06 (light blue), 0.10 (light blue), 0.14 (blue), 0.21 (light blue), 0.24 (white), 0.34 (light blue), 0.38 (yellow), 0.44 (navy blue), 0.47 (dark blue), 0.57 (dark blue), 0.69 (navy blue), 0.78 (dark blue), 0.89 (dark blue), and 0.28 (dark blue), which are characteristic for only the M1 honey samples (Fig. 1). Under 254 nm, there are prominent additional bands at Rf = 0.24 (yellow), 0.47 (dark blue), 0.50 (light blue), and 0.62 (light blue), while bands at Rf = 0.78 and Rf = 0.89 are not visible (Fig. 2). Manuka samples contained rosmarinic acid (Rf = 0.28), ellagic acid (Rf = 0.44),p-coumaric acid (Rf = 0.57), and myricetin (Rf = 0.38). As is illustrated in Fig. 5, despite very similar profiles, differences between the samples M1 and M2 are observed. Under 366 nm irradiation, a loss in the band intensity at 0.10, 0.21, 0.34, and 0.47 and an intense band appearance at 0.14, 0.38, and 0.57 in sample M1 are observed, which is in contrast to the M2 sample. These insignificant differences are most likely the result of the geographical origin of these honey samples.

Identification of Phenolic Compounds by HPLC

Eleven polyphenolics were identified in honey samples by HPTLC, and the obtained results were further confirmed by HPLC analysis. HPLC was used for the quantification of phenolic compounds in all honey samples. Initially, the validation parameters, such as linearity, precision, limit of detection, and quantification of the developed method (HPLC-DAD), were established (Table 2). The concentrations of all of the identified phenolic compounds in the honey samples are given in Table 4. The HPLC analysis of willow honey (W1-W3) confirmed the presence of chlorogenic acid, myricetin, caffeic acid, 3,4-dihydroxybenzoic acid, abscisic acid, and ferulic acid. Their content varied in the range from 5.276 to 7.643 mg per 100 g of honey. Abscisic acid was found in high quantities. In relation to the literature data, abscisic acid was found in Polish (Jerkovic et al. 2014) and Croatian willow honey (Tuberoso et al. 2011) in concentrations of 25.8–35.6 and 31.7–98.2 mg/kg, respectively. A similar dependence between the composition of the willow honey samples and the harvest year was observed by both HPTLC and HPLC. The highest amounts of caffeic acid, ferulic acid, and myricetin were found in the W3 sample collected in 2016 compared with the W1 and W2 samples that were collected in 2015. The level of these compounds gradually decreased with the age of the honey (caffeic acid content decreased to 60%, ferulic acid to 81%, and myricetin to 73%). In the case of other identified phenolic compounds, no significant differences in their concentrations were found.

Table 4 Quantitative determination of phenolic compounds in honey samples by HPLC-DAD
Full size table

The heather honey exhibited the highest level of abscisic acid (8.959–11.772 mg/100 g of honey) and myricetin (0.639–0.861 mg/100 g of honey). Similar content values of these compounds were observed by Jasicka-Misiak et al. 2012 (abscisic acid 8.272–12.061 mg/100 g of honey; myricetin 0.161–1.485 mg/100 g of honey). Additionally, a high level of abscisic acid (2.5–16.66 mg/100 g honey) in heather honey was identified by Ferreres et al. 1996 who considered this compound as a potential marker for this honey.

Myricetin and p-coumaric acid dominated in buckwheat honey. The content of these phenolics ranged from 0.243 to 0.606 and 0.267 to 0.489 mg/100 g of honey, respectively. A similar phenolic profile was found by Jasicka-Misiak et al. 2012 with a myricetin level of 0.082–1.907 mg/100 g of honey and p-coumaric acid of 0.026–4.551 mg/100 g of honey. Moreover, according to Pasini et al. (2013), the quantified levels of p-coumaric acid in buckwheat honey were 0.058–1.297 mg/100 g of honey. In another study, carried out by Karabagias et al. (2014), it was reported that myrycetin and chrysin were determined in thyme honeys. Myrycetin ranged from 0.74 to 244.67 mg/kg and chrysin ranged from 0.01 to 5.60 mg/kg.

The most prominent components of pine honeydew honey were chlorogenic acid (0.179–0.824 mg/100 g), gallic acid (1.315–4.793 mg/100 g), caffeic acid (0.049–0.337 mg/100 g), ferulic acid (0.039–0.620 mg/100 g), and myricetin (0.179–0.842 mg/100 g). Gallic acid was found to be in the largest amounts, whereas caffeic acid, chlorogenic acid, ferulic acid, and myricetin were at lower levels. The level of phenolic compounds in honeydew honey was higher than those obtained in other studies (Oroian and Sorina 2017; Socha et al. 2011), but the values obtained may vary depending on the type of honeydew used by bees for production of certain honey.

The chromatography data confirmed the already described presence (Yao et al. 2003) of rosmarinic, ellagic, p-coumaric acids, and myricetin in manuka honey. The Leptospermum honey, including Australian and New Zealand manuka honey, presented standard phenolic profiles with only slight differences between the varieties. Thus, the quantified levels of p-coumaric, rosmarinic, and ellagic acid in Australian manuka honey were slightly higher than those in New Zealand manuka honey, which additionally exhibited lower values of myricetin.

Multivariate Data Analysis

This study aimed to determine the compounds such as phenolic acids, flavonoids, and abscisic acid which could be very useful when classifying honeys according to their variety. HCA and PCA were used to establish the relationship between the chemical composition of honeys and their floral origin. The results of HCA are presented as dendrograms in Fig. 6 a. All honey samples were classified into three clusters (A, B, and C). The cluster A contains of pine honeydew samples. This group clearly differed from other tested honey. The samples manuka and buckwheat honeys formed cluster B, which showed weak similarity with samples heather and willow honeys, distributed cluster C. Among the samples in the third group, willow honey samples clearly differed from heather honey samples. Moreover, this tendency was repeated in the PCA classification. The components explained 74.95% of the total variance (PC1 39.16% and PC2 35.79%). Figure 6b show the clustering of honey samples into three main groups, such as willow and heather (first), buckwheat and manuka (second), and pine honeydew honeys (third). Principal component analysis clearly differentiates pine honeydew located on the right side from other honey samples. Willow and heather honeys samples are grouped into one cluster. This is probably due to the fact that these samples were characterized by large amounts of abscisic acid. The cluster buckwheat and manuka honey samples partly overlap, but a certain separation can be observed. In both honeys, p-coumaric acid and myricetin were identified, but the manuka honeys were additionally characterized by the occurrence of rosmarinic acid and ellagic acid. Comparing HCA and PCA technique obtained results that exhibited possible relationship to botanical origin of honey samples. Accordingly to the phenolic properties, honeydew honey samples were perfectly classified, while manuka and buckwheat or willow and heather samples partly overlap. Additionally, the similarity of honeys with large percentage of pollen type (W2-W3, B1) and with pollen of less than 45% (W1, B2-B4) was observed, which shows that the slightly lower value of pollen in honeys is not different from honey with dominance of one type pollen.

Fig. 6
figure 6

Hierarchical cluster analysis dendrograms (a) and principal component analysis (b) of the tested honey samples. W1-W3 willow, H1-H2 heather, B1-B4 buckwheat, Ph1-Ph6 pine honeydew, M1-M2 manuka honey

Full size image

Conclusions

The low instrumentation price, short analysis time, wide range of adsorbents and solvents, small sample requirements for analysis, good precision, and accuracy make HPTLC a method of choice for the differentiation of the botanical origin of honey samples. In the present study, the HPTLC fingerprint analysis allowed the identification of the type of the following samples of honey: willow, buckwheat, heather, pine honeydew, and manuka. This showed that on the basis of the similarities and differences in the observed bands, it was possible to determine the botanical origin of certain honey. Additionally, 11 phenolic compounds were identified, and their presence was independently confirmed by an HPLC analysis. HPTLC has been proven to be a good method for the rapid separation of honey components and well suited for the identification of phenolic compounds present in honey samples. It should also be emphasized that the HPTLC analysis of honey might assist in developing HPTLC-supported quality control tools and might improve consumer confidence and support to the product.

This work was supported by The National Science Centre of Poland in the frame of Grant No. 2014/15/B/N29/02182.