Volatile organic compounds of Thai honeys produced from several floral sources by different honey bee species

The volatile organic compounds (VOCs) of four monofloral and one multifloral of Thai honeys produced by Apis cerana, Apis dorsata and Apis mellifera were analyzed by headspace solid-phase microextraction (HS-SPME) followed by gas chromatography and mass spectrometry (GC-MS). The floral sources were longan, sunflower, coffee, wild flowers (wild) and lychee. Honey originating from longan had more VOCs than all other floral sources. Sunflower honey had the least numbers of VOCs. cis-Linalool oxide, trans-linalool oxide, ho-trienol, and furan-2,5-dicarbaldehyde were present in all the honeys studied, independent of their floral origin. Interestingly, 2-phenylacetaldehyde was detected in all honey sample except longan honey produced by A. cerana. Thirty-two VOCs were identified as possible floral markers. After validating differences in honey volatiles from different floral sources and honeybee species, the results suggest that differences in quality and quantity of honey volatiles are influenced by both floral source and honeybee species. The group of honey volatiles detected from A. cerana was completely different from those of A. mellifera and A. dorsata. VOCs could therefore be applied as chemical markers of honeys and may reflect preferences of shared floral sources amongst different honeybee species.


Introduction
Honey is a well-known natural product derived from honeybees. Worker honeybees collect nectar from flower blossom and store them in a honey sack before they return to the hive. The nectar is then mixed with enzymes from bees resulting in the breakdown of complex sugars in the nectar to simple sugars such as glucose and fructose [1]. Besides sugar content, honey contains proteins, lipids and vitamins [2]. Honey may have aromatic compounds at very low concentrations in the form of volatile mixtures. Each volatile compound in the mixture may contribute a different aroma, taste and function leading to the uniqueness of honey. Volatile a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 organic compounds (VOCs) contributing to the aroma of honey vary in quality and quantity because of different nectar sources, microbes in honey, transformation of plant compounds by bees, honey processing, and condition of storage [3,4].
Numerous organic compounds have been detected and identified as characteristic compounds of honey [5][6][7]. Three well-known principal categories of honey volatiles are terpenes and their derivatives, norisoprenoids, and benzene derivatives [8]. Terpenes and their derivatives are known as important compounds that provide flavor, odor and biomedical properties in honey [9,10]. Norisoprenoids have very low olfactory perception thresholds which strongly influence honey odor [11]. Aside the impact of norisoprenoids on honey odor, some of them are known to be anticarcinogenic [12]. Norisopreniods are products of Carotenoids [13,14]. Furthermore, benzene derivatives can be used as pollution-monitor volatiles [15], although they have been found to be the main antibacterial volatile in New Zealand honeys [16]. All these categories of compounds have previously been detected in Thai honeys [17] in which terpenes were found to be the most abundant VOCs in both wild and coffee honeys. Although, there have been previous studies on honeys produced by Apis mellifera, information on VOCs from honey derived from other honeybee species are still lacking. In Northern Thailand, three species of honeybees (Apis mellfiera, Apis cerana and Apis dorsata) are known to produce honeys for human consumption. To bridge the information gap on honeys produced by other bee species, this present study aimed at determining the VOCs released by Thai honeys produced from different floral sources and three different honey species. Knowledge from this could inform the characteristics of honeys and subsequently used to distinguish honeys from different floral sources and honey bee species.

Honey samples
Honeys produced from five different floral sources by three species of honey bees were obtained from retailers (processed honey) and beekeepers (unprocessed honey) in different provinces of Thailand (Chiang Mai, Chiang Rai, Lumphun, Nan, and Lopburi). The honey samples and their bee producers were Apis mellifera (longan honey, wild honey, lychee honey and sunflower honey), Apis cerana (longan honey, wild honey and coffee honey) and Apis dorsata (wild honey). All samples were kept in a dark room at 23-25˚C until volatile samples were collected from them by solid-phase microextraction (SPME).

Analysis of volatile organic compounds
Sample preparation. For each of the honey samples from different floral sources, 2.5 g of honey was weighed into a 20 ml screw-top glass vial with PTFE/silicone lid. As internal standard, 5 μl of 1 mg/ml benzophenone (Sigma-Aldrich, Milan, Italy) in methanol (Sigma-Aldrich) was used, gently putting it on the inner wall of the vial.
Solid-phase microextraction of volatile samples. The volatile samples were analyzed by SPME where vials containing the honey samples and the internal standard were conditioned at 45˚C in the agitator of the GC set-up for 15 minutes before SPME extraction. A 2 cm 50/30 μm divinylbenzene/carboxen/polymethylsiloxane (57299-U) SPME fiber was put into the vial to expose it to the headspace of the honey samples. The volatile samples were allowed 40 minutes to be adsorbed onto the SPME fiber at 45˚C before GC injection. At injection into the GC, thermal desorption of volatile organic compounds from the SPME fiber lasted 5 minutes.
Gas chromatography-mass spectrometry. GC-MS analysis was done by using a GC (7890A, Agilent Technologies, Santa Clara, USA) coupled with an MS (5975C Network, Agilent Technologies). The GC had an HP 5MS column (non-polar column, Agilent Technologies), 30 m x 0.25 mm internal diameter (i.d.) and 0.25 μm film thickness. The carrier gas was helium flowing at a rate of 1.2 mL/ minute. The initial oven temperature was 40˚C held for 3 minutes. It was raised to 200˚C at 3˚C/minute and then 16˚C/minute until it reached 240˚C and held at this temperature for 1.2 minutes. Mass spectra were recorded in a scanning range of 24-360 m/z. Enhanced ChemStation (ver. E.02.02.1431, Agilent Technologies) was used for identification by comparing the mass spectra of VOCs with those in the database of NIST 11(Gaithersburg, MD, USA) and Wiley 7N (John Wiley, NY, USA). The linear retention indices (LRI) of the VOCs were calculated using the retention times of n-alkane series from C9 to C20 as reference compounds [18]. The calculated LRI were then compared to those in NIST webbook. Each peak area of characterized compound was compared to peak area of the internal standard (5 μL of 1 mg/mL benzophenone), the amounts of detected VOCs were expressed as relative amount (%) ± standard error.
Statistical analysis. The relative amounts of detected VOCs in the honey samples were analyzed for normality (Shapiro-Wilk test), Non-metric multidimensional scaling (NMDS) and grouping plot illustrations were generated using Paleontological statistics; PAST version 2. The normally distributed data were analyzed by one-way ANOVA (SPSS v.17.0; IBM, Chicago, IL, USA). Mann-Whitney tests (SPSS) were applied to data, which were not normally distributed. Significant levels were set at α = 0.05.

Results
A total of 62 compounds were identified from the headspace volatiles of the honey samples studied. Of the total number of compounds, 48 were present in longan honey, 41 in wild honey, 21 in lychee honey, 23 in coffee honey and 8 in sunflower honey (Table 1). Moreover, the identified VOCs belonged to 7 chemical classes namely alcohols, aldehydes, ketones, esters, acids, hydrocarbons, and terpenes. Terpenes were the most abundant compounds among all the volatiles in all the samples. Three volatiles namely cis-linalool oxide (18.9-1,043.5%), furan-2,5-dicarbaldehyde (1.2-45.9%) and trans-linalool oxide (8.2-29.5%) were present in all honey samples (Fig 1). Some volatiles were absent in particular honeys, e.g. 2-phenylacetaldehyde, isophorone and methyl nonanoate were absent from longan honey from A. cerana while linalool, ho-trienol, benzyl ethanol, isophorone and epoxylinalool isomer II were absent from sunflower honey from A. mellifera. However, significantly high amounts of cis-and trans-linalool oxide were found in lychee (A. mellifera) and longan honeys (A. mellifera and A. cerana) than that found in the other honeys. Interestingly, 2-phenylacetaldehyde described as honey, sweet, and floral odor was detected in all honey sample except longan honey produced by A. cerana. When comparing the numbers of the identified volatile compounds, it was found that longan honey produced by A. mellifera had the highest amount of volatiles (48 VOCs) while, sunflower honey produced by A. mellifera has the least amount of volatiles (8 VOCs). The volatiles of longan honey produced by A. mellifera were double of ones of A. cerana, 48 and 22, respectively. In addition, the quantities of each volatile compound of longan honey produced by A. mellifera were generally more than the ones produced by A. cerana. When considering wild honeys, it was found that the ones produced by A. dorsata (31 VOCs) contained the highest number of volatile, followed by A. mellifera (26 VOCs) and A. cerana (14 VOCs), respectively.

A. cerana (n = 5)
A. mellifera (n = 3) honey from A. mellifera (Table 2). Diethyl hexanedioate and diethyl decanedioate were only found in longan honey from A. cerana. Methyl caproate, benzaldehyde and 5-hydroxymethylfurfural were only found in wild honey from A. dorsata while methyl 10-oxodecanoate, 2,6,6-trimethyl-2-cyclohexen-1-ol and methyl oleate were only found in wild honey from A. mellifera. (Z)-5,6-dimethyl-1,3-cyclohexadiene and were only found in lychee honey. 2-Furanmethanol, butyryl lactone, phenylmethanol, anisaldehyde, anise alcohol and 3,4,5-trimethylphenol were only found in coffee honey from A. cerana (Table 2). Interestingly, 2-hydroxy-3,5,5-trimethyl-2-cyclohexen-1,4-dione was only found in lychee from A. mellifera and longan honey from both A.mellifera and A. cerana. Non-metric multidimensional scaling and plot grouping analysis revealed that differences in floral sources and honeybee species lead to differences in VOC groupings. Among volatiles from honey produced by the same honeybee species, A. mellifera, longan honey was the largest group with numerous volatile compounds (Fig 2a). Volatiles in longan honey slightly overlapped with volatiles from three different groups of floral sources namely, lychee, wild and sunflower (Fig 2a). Likewise, the volatile group of wild honey overlapped with the longan, lychee and sunflower groups. However, volatiles belonging to lychee honey were completely separated from volatiles belonging to sunflower honey (Fig 2a). There were clear differences among volatiles of honey produced by A. cerana from three different floral sources (longan, wild and coffee) (Fig 2b). Volatiles from longan honey produced by A. mellifera and A. cerana were distinctly different (Fig 2c). Moreover, volatiles from wild honey produced by A. cerana was distinctly different from those produced by A. mellifera and A. dorsata (Fig 2d) but wild honey from A. mellifera and A. dorsata slightly overlapped (Fig 2d). These results reflect the differences among VOC groups due to floral sources and honeybee species. Floral sources and honey bee species impacts Thai honey volatiles

Discussion
This study is the first to investigate the volatile organic compounds of Thai honeys from several floral sources and different honeybee species. In Thailand, honey is often marketed as unifloral honey. The floral origin of the honey gives it a unique flavor and fragrance. In fact, it is not only the botanical origin of flora that influence the quality of honey but also some extended factors such as honeybee species and storage condition. This present study focused on the influence of floral source and honeybee species on aroma of honey.
An important finding of this study is that volatiles emitted by honeys are dependent on both the floral source and honeybee species that produced the honey. Honeybees feeding on different floral sources produce honeys with different quality and quantity of volatiles. The highest number of volatiles detected were from longan honey followed by wild honey, lychee honey, coffee honey and sunflower honey, which had the least number of volatiles. This finding supports the evidence that longan honey is the most fragrant honey, while sunflower honey is the least fragrant honey among monofloral honey as highlighted by Thai beekeepers. In addition, this is in agreement with many previous studies that found floral sources to be a major factor for differentiating honey aroma [4,7,20,21]. It is well known that wild honey is multifloral honey derived from a variety of nectar sources. It is therefore reasonable to suggest that VOCs of wild honey mimic a combination of VOCs from various monofloral honeys. This is in agreement with our results, which showed that the group of VOCs from wild honey overlapped with every other group of monofloral VOCs. In particular, plants from same family tended to have similar quality and quantity of volatile compounds. It was therefore not surprising that longan and lychee both of which belong to the family Sapindaceae [19] had groups of volatiles that overlapped as seen in Fig 2a. As expected, the VOCs of sunflower (family: Asteraceae) [22], barely overlapped with other monofloral groups.
So far, relatively few studies have focused on the honeybee species as a factor that causes differences in honey aroma [23,24]. As it is depicted in Fig 2, showing volatile organic compounds plot grouping of honey produced from different floral sources and bee species. The volatile compounds of honey produced by A. mellifera were overlapped (Fig 2a). On the contrary, as shown in Fig 2b, volatile compounds of honey produced by A. cerana were clearly grouped. Additionally, Fig 2c and 2d confirm that bee species affect the volatiles of honey even if they feed on the same floral source even though a slight overlap of VOCs was observed between wild honey from A. mellifera and A. dorsata. This study put the spotlight on the honeybee as an important factor that could influence the aroma of honey. The differences in the volatile profiles of the different monofloral honeys are likely influenced by the different honeybee species. For instance, we found that 2-phenylacetaldehyde was absent only in longan honey produced by A. cerana. Meanwhile longan honey from A. mellifera produced more volatiles than any other honey. The foraging behavior of bees that keeps nectar in their honey stomach for some time during their flight back to hives before transferring them into honey combs contribute to these observed differences. During the period that nectar is in the honey stomach, they may be transformed and that likely influence the VOCs. A recent study (18) in which a unique lactic acid bacteria (LAB) symbiont was discovered gives credence to this assertion. In that study, it was established that symbionts from the honey stomach of honeybees influence the production of metabolites including volatile compounds. Most importantly, the LAB symbionts in the honeybees' honey stomach have also been found to influence antimicrobial properties of tested honeys. Furthermore, the lactic acid bacteria vary in different honeybee species [25]. Therefore, it may be possible that the metabolites produced due to LAB symbionts from different honeybee species would have impact not only on antibacterial properties but could also vary the volatile compounds in honeys. Foraging distance in each honeybee species could also be a factor that contributed to varied volatile compounds in tested honeys. It was found that in wild honey collected from A. dorsata provided more volatiles than ones from A. mellifera and A. cerana. It is known that A. dorsata can forage beyond 5 km [26], A. mellifera may forage up to 3 km [27] while A. cerana around 1-1.5 km [27]. The longer foraging distance may increase the chance to collect nectar from various sources leading to more detected volatile compounds in tested honeys. However, more investigation is required.

Conclusion
There were a total of 63 compounds detected in the headspace of the honey samples. The highest (48) and the least (8) numbers of volatiles were identified in longan honey and sunflower honey, respectively. Each of the Thai honey samples studied had a complex and unique volatile organic compound profile which varied in quality and quantity from other honey samples. The variation in volatile compounds in the honey samples lead to uniqueness of fragrances, tastes and potential biomedical properties. These characteristic depend not only on the nectarproviding plant species, but also on honeybee species. Therefore, volatile compound profiles may be used as chemical markers for tracing the origin of honey. Additionally, NMDS and grouping plot can be used for controlling honey produced by different floral sources and bee species.