Traceability of Satsuma Mandarin (Citrus unshiu Marc.) Honey through Nectar/Honey-Sac/Honey Pathways of the Headspace, Volatiles, and Semi-Volatiles: Chemical Markers

Headspace solid-phase microextraction (HS-SPME) and ultrasonic solvent extraction (USE), followed by GC-MS/FID, were applied for monitoring the nectar (NE)/honey-sac (HoS)/honey (HO) pathways of the headspace, volatiles, and semi-volatiles. The major NE (4 varieties of Citrus unshiu) headspace compounds were linalool, α-terpineol, 1H-indole, methyl anthranilate, and phenylacetonitrile. Corresponding extracts contained, among others, 1H-indole, methyl anthranilate, 1,3-dihydro-2H-indol-2-one and caffeine. The major HoS headspace compounds were linalool, α-terpineol, 1,8-cineole, 1H-indole, methyl anthranilate, and cis-jasmone. Characteristic compounds from HoS extract were caffeine, 1H-indole, 1,3-dihydro-2H-indol-2-one, methyl anthranilate, and phenylacetonitrile. However, HO headspace composition was significantly different in comparison to NE and HoS with respect to phenylacetaldehyde and linalool derivatives abundance that appeared as the consequence of the hive conditions and the bee enzyme activity. C. unshiu honey traceability is determined by chemical markers: phenylacetaldehyde, phenylacetonitrile, linalool and its derivatives, as well as 1H-indole, 1,3-dihydro-2H-indol-2-one, and caffeine.


Introduction
The most common and economically important varieties of Satsuma mandarin (Citrus unshiu Marc.) in Croatia (Neretva valley, Opuzen area) are Wakiyama, Chahara, Okitsu, Kawano Wase, Owari, Saigon, Kuno, Zorica, Ichumare, and Seto [1]. Around 2.5 million mandarin trees have been planted [1] in the Opuzen area (ca. 2500 ha) providing a good nectar source for unifloral honey production and potential for independent commercialization (not just as Citrus honey without distinction of the species). C. unshiu honey has not yet been characterized in detail.
USE extract of HoS contained higher aliphatic compounds, the major ones were (Z)-octadec-9-en-1-ol (45.3%), octadecan-1-ol (8.6%), and hexadecan-1-ol (11.6%), Table 4. These chemical structures are related to the composition of cuticular waxes and less to pheromones, but have been also found in NE (Table 2). Fatty acids and alcohols were previously identified as the major compounds of the solvent organic extract from the sacs of the bees that collected Mentha spp. nectar, and methyl syringate, terpendiol I and vomifoliol were attributed to the plant origin [27]. Other important  Table 2 were caffeine (11.5%), 1H-indole (1.6%), 1,3-dihydro-2H-indol-2-one (6.4%), and methyl anthranilate (1.4%). USE extracts of HoS and NE were very similar. Only 1-hydroxylinalool appeared (5.4%) in HoS, which can be indication of the beginning of linalool transformations triggered by the enzymes and continued later in the combs.

The Chemical Composition of C. unshiu Honey Headspace, Volatiles and Semi-Volatiles
On returning to the hive, the content of the honey-sac is regurgitated into the honeycomb and ripened into honey. Under the honeycomb oxidative atmosphere sensitive honey organic compounds can undergo oxidation [28]. There are only a few studies in which the organic extractives of the honey-sac have been correlated with those of the corresponding honey. The comparison of the components of the extracts of Linden honey and honey-sac contents showed that nectar and honey-sac contents contain many aldehydes which were found as corresponding acids in the honey, while the aliphatic compounds, isoprenoids and the alkaloids remained unchanged [28]. In another research, the major identified terpene in the honey-sac was 3,7-dimethylocta-1,5-dien-3,7-diol (terpendiol I) and it was found in Mentha spp. honey solvent extracts, but also can transform to hotrienol, the most abundant compound in the honey headspace [27].
Phenylacetaldehyde was dominant compound (34.4%-47.2%; 38.3%-49.1%) of the C. unshiu honey headspace, followed by benzaldehyde (5.8%-9.8%%; 3.3%-6.6%). Among other benzene derivatives, abundant was phenylacetonitrile (2.7%-9.9%; 3.4%-10.2%). Phenylacetaldehyde was strikingly more abundant in comparison with the nectar headspace (HS-NE) and the headspace of the honey-sac (HS-HoS), shown in Table 1, indicating its formation during the honey ripening in the hive, since heat was not applied to the samples. This can be generated from phenylalanine either by enzyme catalysis or by Strecker degradation [29]. A high percentage of phenylacetaldehyde was found in the honey headspace of Asphodelus microcarpus Salz. et Viv. [30]. Phenylacetonitrile was present within percentage ranges similar to those seen in the HS-NE and HS-HoS (Table 1), while benzaldehyde percentages were elevated. In addition to phenylacetonitrile, two aliphatic nitriles were detected in several honey samples with minor percentages (Table 3): ethylisocyanide and 3-methylbutanenitrile. Benzaldehyde was found to be the major volatile from the honey of cambara and willow, but also in lemon and orange honey [17,29]. Phenylacetonitrile was found in the headspace of dandelion and thyme honeys [31,32].
1H-indole and methyl anthranilate were occasionally present, but not in the headspace of all honey samples, and with markedly lower percentages in comparison to HS-NE and HS-HoS.
In comparison with other Citrus honey VOCs, several similarities can be pointed out. Namely, the suitability of methyl anthranilate, originating from the plant, as the chemical marker of Citrus honey types has been already found [34]. Along with the detection of methyl anthranilate, more than 60 different VOCs were also reported [11,12,17] in Citrus honey types. Similar to the present research, benzaldehyde, phenylacetaldehyde, and linalool derivatives (e.g., linalool oxides, lilac aldehyde isomers, or p-menth-1-en-9-al) were found among important Citrus honey headspace compounds. Caffeine was also previously found as a characteristic compound of Citrus honey [35]. However, despite previous studies on Citrus honeys, few particular compounds were present in C. unshiu honey not already mentioned in other Citrus honeys, such as 1H-indole, 1,3-dihydro-2H-indol-2-one, and phenylacetonitrile.
During C. unshiu honey flow, a part of the returning foragers were collected. The bees were frozen in the field by liquid nitrogen and were stored in a deep-freezer until their honey-sac contents were investigated. After thawing, the abdomen of 100 bees was dissected by peeling off the tergit with forceps in order to expose the honey sac. The honey sacs were removed and frozen. After freezing, the entire content of the honey-sacs was pooled and put in a glass vial (5 mL) at 4 • C until the volatiles were isolated.
Twelve C. unshiu honey samples were investigated. The combs from specially prepared colonies, which were formed from 2 kg of packaged bees on wax foundation, were placed in the area of predominantly C. unshiu trees growing in the Neretva valley, Opuzen area, Croatia, but the samples were also collected from local beekeepers. All of the samples were stored in hermetically closed glass bottles at 4 • C until the volatiles were isolated. Melissopalynological analysis was performed by the method recommended by the International Commission for Bee Botany [36]. Microscopical examination was carried out on a Hund h 500 (Wetzlar, Germany) light microscope attached to a digital camera (Motic m 1000, Motic Deutschland GmbH, Wetzlar, Germany) and coupled to an image analysis system (Motic Images Plus software, Motic Deutschland GmbH) for morphometry of pollen grains.

Headspace Solid-Phase Microextraction (HS-SPME)
The headspace extraction was performed using a manual SPME holder using two fibres: divinylbenzene/carboxene/polydimethylsiloxane (DVB/CAR/PDMS) and polydimethylsiloxane/divinylbenzene (PDMS/DVB) obtained from Supelco Co. (Bellefonte, PA, USA). The fibres were conditioned prior to use according to the instructions by Supelco Co. For HS-SPME, the nectars (1 mL) were placed separately in 5 mL glass vials and hermetically sealed with PTFE/silicone septa. The content of honey-sacs was put as described above in 5 mL glass vial and hermetically closed with PTFE/silicone septa. The honey/saturated water solution (5 mL, 1:1 (v/v); saturated with NaCl) of each honey sample was placed in a 15 mL glass vial and hermetically sealed with PTFE/silicone septa.
The vials were maintained in a water bath at 60 • C during equilibration (15 min) and HS-SPME (45 min) and were partially submerged so that the liquid phase of the sample was below the water level. All of the experiments were performed under constant stirring (1000 rpm) with a magnetic stirrer. After sampling, the SPME fibre was withdrawn into the needle, removed from the vial, and inserted into the injector (250 • C) of the GC-FID and GC-MS for 6 min where the extracted volatiles were thermally desorbed directly to the GC column.

Ultrasonic Solvent Extraction (USE)
Ultrasound-assisted solvent extraction (USE) was performed in an ultrasound cleaning bath (Clean 01, MRC Scientific Instruments, London, UK) by the indirect sonication mode at a frequency of 37 kHz at 25 ± 3 • C. Two solvents were separately used for USE: a mixture of pentane/diethyl ether, 1:2 (v/v) and dichloromethane.
The nectars (0.5 mL) were separately dissolved in flasks (5 mL) in 0.5 mL distilled water, MgSO 4 (0.05 g) was added and the sample was vortexed (5 min). Dichloromethane (1 mL) was used for USE of dissolved nectars.
The content of honey-sacs was dissolved in distilled water (0.5 mL) in 5 mL flask, MgSO 4 (0.03 mg) was added, and the sample was vortexed (5 min). USE was performed using dichloromethane (1.5 mL). Forty grams of each C. unshiu honey sample was dissolved in distilled water (22 mL) in a 100-mL flask. Magnesium sulphate (1.5 g) was added and each sample was vortexed (10 min). Both solvents (20 mL) were separately used for USE of the honey samples.
The sonication was maintained for 30 min. After sonication, the organic layer was separated by centrifugation and filtered over anhydrous MgSO 4 . The aqueous layer was returned to the flask and another batch of the same extraction solvent was added and extracted by ultrasound for 30 min. The organic layer was separated in the same way as the previous one and filtered over anhydrous MgSO 4 , and the aqueous layer was sonicated a third time for 30 min with another batch of the extraction solvent. Combined organic extracts were concentrated to 0.2 mL by distillation with a Vigreaux column, and 1 µL was used for GC-FID and GC-MS analyses.

GC-FID and GC-MS Analyses
The GC-FID analyses were carried out with an Agilent Technologies (Palo Alto, CA, USA) gas chromatograph model 7890A equipped with a flame ionization detector (FID) and a HP-5MS capillary column (5% phenyl-methylpolysiloxane, Agilent J and W). The GC conditions were similar to those described previously [24]. In brief, the oven temp. was programmed isothermal at 70 • C for 2 min, increasing from 70-200 • C at 3 • C·min −1 , and held isothermally at 200 • for 15 min; carrier gas, He (1.0 mL·min −1 ).
The GC-MS analyses were performed using an Agilent Technologies (Palo Alto, CA, USA) gas chromatograph model 7820A equipped with a mass selective detector (MSD) model 5977E (Agilent Technologies) and a HP-5MS capillary column, under the same conditions as described for the GC-FID analysis. The MSD (EI mode) was operated at 70 eV, and the mass range was 30-300 amu, as previously reported [24].
The identification of the volatile constituents was based on the comparison of their retention indices (RI), determined relative to the retention times of a homologous series of n-alkanes (C 9 -C 25 ), with those reported in the literature [25] and their mass spectra with authentic compounds available in our laboratories or those listed in Wiley 9 (Wiley, New York, NY, USA) and NIST 14 (D-Gaithersburg) mass spectral libraries [25]. The percentage composition of the samples was computed from the GC peak areas using the normalization method (without correction factors). The average component percentages in the tables were calculated from duplicate GC-FID and GC-MS analyses.

Conclusions
Applied HS-SPME/GC-MS/FID and USE/GC-MS/FID methodologies of monitoring nectar/honey-sac/honey pathways of the headspace, volatiles, and semi-volatiles was successful and complementary for the characterisation of C. unshiu honey.