Botanical and geographical origin of Turkish honeys by selected‐ion flow‐tube mass spectrometry and chemometrics

Abstract BACKGROUND Honey has a very important commercial value for producers as a natural product. Honey aroma is formed from the contributions of several volatile compounds, which are influenced by nectar composition, botanical origins, and location. Selected‐ion flow‐tube mass spectrometry (SIFT‐MS) is a technique that quantifies volatile organic compounds simply and rapidly, even in low concentrations. In this study, the headspace concentration of eight monofloral (chestnut, rhododendron, lavender, sage, carob, heather, citrus, and pine) and three multiflower Turkish honeys were analyzed using SIFT‐MS. Soft independent modeling of class analogy (SIMCA) was used to differentiate honey samples based on their volatiles. RESULTS This study focused on 78 volatile compounds, which were selected from previous studies of selected honeys. Very clear distinctions were observed between all honeys. Interclass distances greater than 8 indicate that honeys were significantly different. Methanol and ethanol were abundant in the honeys. Chestnut honey collected from the Yalova region had the highest total concentration of volatiles followed by heather honey and chestnut honey collected from the Düzce region. CONCLUSION Honeys with different botanical and geographical origins showed differences in their volatile profile based on chemometric analysis. Of the honey samples, methanol, ethanol, acetoin, ethyl acetate, and isobutanoic acid had the highest discriminating power. Methanol and ethanol, and then acetic acid, were the volatiles with the highest concentrations in most honeys. © 2020 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.


INTRODUCTION
Honey is one of the oldest foods in existence, and is consumed not only for its effects on health but also for its taste, nutritional value, and unique flavor. Honey contains mainly water and sugar. 1 Vitamins, minerals, enzymes, free amino acids, and plentiful volatile compounds are present as secondary constituents. 1 Different honeys have different minor compound compositions due to their botanical and geographical origin, harvesting season, and processing conditions. The variation in this composition can be used to identify botanical and geographical origins as well as their quality. 2 Honey has a very important commercial value for producers as a natural product. 3 World honey production was 1.786.996 t in 2016 and Turkey ranks second after China with 105.532 t. 4 The most widely found honey types in Turkey are wildflower, pine, chestnut, thyme, linden, citrus, cotton, and sunflower honey. 5 Volatile compounds present in honey are characteristic markers of botanical origin. 6,7 Various volatile compounds and representative chemical groups are present at high levels in different honeys. 7 Some of the main marker compounds are 3-hydroxy-5-methyl-2-hexanone, methyl anthranilate, and sinensal isomers in citrus honey, nerolidol oxide, coumarin, hotrienol, hexanal, and hexanol in lavender honey, and 2-cyclopenten-1,4-dione, 2-aminoacetophenone, 2-hydroxyacetophenone, guaiacol, propyl anisol, p-anisaldehyde, and p-cresol in heather honey. 8 Chestnut honeys are noticeable for their high concentrations of acetophenone, 1-phenylethanol, and 2-aminoacetophenone, 7 while lilac aldehyde and 2-aminoacetophenone are indicators for rhododendron. 9 According to Tananaki et al., 10 octanal, 3-carene, camphene, octane, nonanal, decanal, ⊍-pinene, ⊎-pinene, toluene and 1.2.3-trimethylindene are marker compounds for pine honey. Characteristic volatiles of sage honey were tetrahydro- 2,2,5,5-tetramethylfuran, lilac aldehyde, 2-methylbenzene, heptanoic acid, and benzeneacetic acid. 11 The evaluation of the botanical and geographical origin of honey is very complicated. The fingerprint of specific honey samples can be determined by measuring organoleptic properties, melissopalynological characteristics, and physicochemical characteristics. 12 Alternative and faster methods are being considered for characterization of non-volatile and volatile markers of unifloral honeys. 6 Unifloral honey aroma is mainly formed by a nectar of the specific flower. Selected-ion flow-tube mass spectrometry (SIFT-MS) is a fast and sensitive analytical technique for real-time analysis of trace gases by using the chemical ionization of the target gases. 13 The principal objective of this study was to determine if chestnut, rhododendron, lavender, sage, carob, heather, citrus, pine, and wildflower honeys can be distinguished based on their volatile organic compounds using a SIFT-MS technique combined with multivariate statistical analysis.

Honey samples and their botanical origin
Twelve honey samples were selected from different parts of Turkey (Table 1). Local beekeepers collected the honey from bees that were kept in hives near fields containing predominantly chestnut, rhododendron, lavender, sage, carob, heather, citrus, pine, or mixed wildflower ( Table 1). The concentration of the volatiles was measured by employing the predetermined reaction rate constant for the volatile with a selected precursor ion and accounting for the dilution of the sample gas into the carrier gas (helium) in the flow tube. 14 Trace volatile analyte compounds were introduced in the reactor at an optimized sample inlet flow rate of 0.35 Torr·L/s (26 cm 3 min −1 ).

Measurement of volatile concentrations
The range of the mass-to-charge ratio was set to 10-250 m/z, with a total SIM scan time of 120 s. The concentration of measured volatile compounds, which was calculated through known kinetic parameters, is listed in Table 2. Concentrations were measured in μg L −1 in the headspace above the honey sample. During the analysis, some compounds produce the same mass for a given precursor ion, so the interfering compounds have to be reported as a mixture. In this study, several mixtures were identified at different charge-to-mass ratios, such as 2-methyl-2-butanol and butanoic acid at 71 m/z, acetone and isoamyl alcohol at 88 m/z, acetic acid and 2-cyclopenten-1,4-dione at 90 m/z, dimethyl disulfide and phenol at 94 m/z, acetic acid and p-cresol at 108 m/ z, acetoin and ethyl acetate at 118 m/z, phenylacetaldehyde and isopropyl benzene at 120 m/z, 2-phenylethanol and santene at 122 m/z, dimethyl trisulfide and hydroxymethylfurfural at 126 m/z, ⊍-pinene, ⊎-pinene, 2-hydroxyacetophenone and 4-methoxybenzaldehyde at 136 m/z, octanoic acid and nonanol at 144 m/z, and hotrienol and p-menth-1-en-9-al at 152 m/z.

Statistical analysis
The concentrations of volatile compounds were analyzed in triplicate. One-way analysis of variance (ANOVA) using Tukey's procedure with a 95% confidence interval was performed to determine statistical differences among samples; significance was defined as P ≤ 0.05 using SPSS (version 25, SPSS Inc., Chicago, IL, USA). Multivariate statistical analysis was conducted using SIMCA with Pirouette software for Windows Comprehensive Chemometrics Modeling, version 4.0 (Infometrix Inc., Bothell, WA, USA) to identify distributions of volatiles in honey samples.

RESULTS AND DISCUSSION
Volatile composition of honeys Many compounds have been detected in honey using different techniques. This study focused on 78 volatile compounds, which were selected from previous studies of selected honeys. Twelve honey samples of known botanical and geographical origins were analyzed (Table 3). Methanol and ethanol, as in other types of honey and food, were abundant in the analyzed samples. Even though these alcohols were commonly found in natural products due to the metabolism of yeasts, 15 or reduction of aldehydes, 16 they can be the most effective discriminators based on either their high volatility or discriminating power (Fig. 1). These compounds have been found to discriminate among different type of honeys, such as thyme and lavender honey. 17 Acetic acid was the third highest concentrated volatile followed by methanol and ethanol in the honeys, except for chestnut honeys from Yalova region and lavender (Table 3). Acetic acid is formed through degradation of alcohols and produce acidic aroma in honey. 18 Menthol was the second highest compound for chestnut honey from Yalova and phenol was third followed by 1-octen-3-ol. The acetone concentration was the third highest in lavender honey. Menthol is a mint essential oil, which is allowed to be used in formulations against mites and ticks. 19 Acetone was the fourth highest volatile of chestnut_Düzce, pine, wildflower_Ardahan and wildflower_Kırşehir, while 2-cyclopenten-1,4-dione was the fourth leading volatile compound for rhododendron, heather and wildflower_Sivas. Alpha-pinene was the fourth higest concentrated compound measured in sage honey, while ethyl acetate was measured in carob and dimethyl sulfide in citrus as fourth highest compound. Acetone is responsible for a pungent or fruity odor and ethyl acetate gives a fruity aroma in honey. 20 In the presence of ethanol, ethyl acetate is formed through esterification of acetic acid via microorganisms. 21 Alpha-pinene was one of the compounds detected in the honey profile that comes directly from the flower. 22 Ethyl acetate, which is the ester formed from ethanol and acetic acid, was one of the most abundant compounds in carob honey after ethanol, methanol, and acetic acid. Some studies have focused on the volatile compounds found in carob; 23,24 however, only one published article focused on the volatile characteristics of carob honey, which is mainly characterized by nonanal and octanal. 25 Heather, citrus, wildflower honey from Sivas, and wildflower honey from Kırşehir also had high amounts of ethyl acetate.
Dimethyl sulfide was one of the compounds with the highest concentration detected in citrus after ethanol, methanol, and acetic acid. It was also found in both raw and heat-treated citrus honey from Spain. 26 The concentration of dimethyl sulfide was relatively high in rare unifloral honeys in Spain such as Persea americana (38.5%), Spartocytisus supranubius (25.2%), Quercus ilex (7.4-337%), Satureja montana (22.8%), and Agave honey (19.4%). 27 Hotrienol is one of the marker volatiles for lavender honey; 8,28 however, in our study it was of a lower concentration compared to other compounds. Sage honey had a significantly higher amount of hotrienol than other honeys (Table 3); however, previous studies did not report it in sage honey. 11,29 The geographic areas of the lavender and sage honey were in the same province, which may lead the bees harvesting from both area and caused similarities in volatile composition. Hotrienol comes from the flower, during ripening of the honey in the hive and is thermally generated during pasteurization. 30 Several alcohols were identified in lavender honey, and ethanol, methanol, isoamyl alcohol, lemonol, 2-butanol, hotrienol and 1,3-butanediol were the highest concentrations. Radovic et al. 31 determined that ethanol, 2-methyl-1-propanol, 3-methyl-1-butanol, 3-methyl-3-buten-1-ol, hotrienol, and furfuryl alcohol were the main alcohols present in lavender honey collected from France and Portugal.

Effect of botanical and geographical origins
Multivariate statistical analyses allow the determination of botanical and geographical discrimination between honey samples. Interclass distances (ICDs) greater than 3 indicate that samples were significantly different. 32 Better separation of honeys is achieved with higher interclass distances between two honeys. All of the ICD values of measured honeys were greater than 3, or, in this case, greater than 8 (Fig. 1), which indicates that these honey samples can be discriminated based on their volatile composition. Ethanol and methanol showed the highest discriminating power (Table 4). Because ethanol and methanol had the highest concentration of volatiles, they may cause a decrease in discriminating power, by repressing the influence of other volatiles on the volatile profile. Multivariate statistical analysis was therefore also applied to the data set without ethanol and methanol (Fig. 2). After exclusion of these two compounds, menthol, dimethyl disulfide, phenol and dimethyl sulfide showed the highest discriminating power (Table 5). Langford et al. 33 also identified dimethyl disulfide as the compound with the highest discriminating power in monofloral New Zealand honeys. When comparing different botanical sources from the same province, chestnut and rhododendron (Düzce), lavender and sage (Burdur), and carob, heather, and citrus (Antalya) showed clear differentiation (Fig. 1). The concentrations of 2-butanol, 2-methyl-2-butanol, 3-methylbutanal, acetoin, acetone, butanoic acid, ethanol, ethyl acetate, furfuryl alcohol, isoamyl alcohol, isobutanoic acid, methanol, and propyl anisol were significantly different in chestnut and rhododendron (Düzce) ( Table 3). Hexanal, hotrienol, and lilac aldehyde concentration were different in lavender and sage honey. Furfuryl alcohol is one of the characteristic compounds for chestnut honeys. 34 Castro-Vázquez et al. 8 differentiated citrus and heather honey based on their volatile composition. Similar to our study, (Z)-3-hexen-1-ol, acetic acid, 2-cyclopentene-1-4-dione and butanoic acid were found to be higher in heather honey compared to citrus. These compounds were discriminating volatiles for heather and citrus. Even though it is difficult to determine the botanical source of honey accurately by many techniques, 6 it is clearly seen that SIFT-MS with chemometrics was effective. Agila and Barringer 18 identified differences in the volatiles of honeys from different botanical sources (blueberry, clover, cranberry, and wildflowers) collected from the state of Indiana, USA. Langford et al. 33 also applied SIFT-MS technology to distinguish New Zealand monofloral honeys.
When the same flower source was compared with different locations, such as chestnut honeys from Yalova and Düzce, or wildflower honeys from three different provinces, varied volatile compositions were detected (Fig. 1). The composition of honey not only depends on the nectar-providing plant species but also depends on other factors such as environmental factors, bee species, harvesting season and technology, processing, and storage. 35 Chestnut honey collected from Yalova and Düzce regions had no statistical similarities in volatile compound concentration, except for p-menth-1-en-9-al (Table 3). Chestnut honey from Yalova had a higher concentration of all compounds than chestnut honey from Düzce. The reason for this significant difference between the volatile levels was probably the geographical location. Castro-Vázquez et al. 36 reported clear differentiation of chestnut honeys from different geographical origins according to their volatile composition, using multivariate statistical analysis.
The concentrations of 2,3-butandione, 2-butanol, 2-cyclopenten-1,4-dione, acetic acid, acetoin, acetone, butanoic acid, ethyl acetate, furfural, isoamyl alcohol, isobutanoic acid, isopropyl benzene, maltol, methanol and phenylacetaldehyde were different in the three wildflower honeys from different locations. The aroma composition of wildflower honeys can be dissimilar from each other because of the variation and differences of flowers contingent upon the location.
Karabagias et al. 37 investigated the geographical characterization of citrus honeys in Mediterranean countries. While ethyl acetate was determined as a key discriminating compound in citrus honeys collected from Morocco, it was not detected in honeys collected from Egypt, Greece, and Spain. Ethyl acetate was found only in Moroccan citrus honey, although ethyl octanoate and ethyl nonanoate were reported to be in higher concentration in Greek citrus honeys and ethyl nonanoate was high in Egyptian citrus honeys. Ethyl acetate may therefore be one of the compounds that can be used to geographically discriminate between Mediterranean citrus honeys.

CONCLUSION
SIFT-MS is a fast and simple method to enhance the difference between Turkish honeys based on their volatile composition. The application of SIFT-MS technique with the aid of chemometrics for floral and geographical origin determination of honeys can be very useful. The data analysis takes place in twodimensional matrices with a chemometric approach, which allows for a better separation of the samples.
Honeys with different botanical and geographical origin showed differences in their volatile profile based on their interclass distances. Between the honey samples, methanol, ethanol, acetoin, ethyl acetate, and isobutanoic acid had the highest discriminating power and also methanol and ethanol, and then acetic acid, were the volatiles at the highest concentration in most honeys. In general, chestnut from the Yalova region had the highest total concentration of volatiles followed by heather and chestnut from the Düzce region, and wildflower from the Ardahan region had the lowest total concentration. The volatile composition of each honey type was affected by several factors. Future studies with a broader variety of honeys or geographical origins with different harvesting seasons may be required for a better understanding of the honey fingerprint.