Effect of Drying Methods on Volatile Compounds of Citrus reticulata Ponkan and Chachi Peels as Characterized by GC-MS and GC-IMS

To reflect the volatile differences of dried citrus peel as affected by cultivars and drying methods, the volatile compounds of dried citrus peel of two cultivars (Citrus reticulata “Chachi” and Citrus reticulata “Ponkan”), prepared under three drying methods (sun-drying (SD), hot-air-drying (AD), and freeze-drying (FD)), were analyzed by GC-MS, odor activity values (OAVs), and GC-IMS. GC-MS data indicated that SD was favorable to preserve terpenic alcohols (linalool, α-terpineol and terpinene-4-ol), β-cymene, methyl methanthranilate, and monoterpenes; while AD was favorable to preserve aliphatic aldehydes and sesquiterpenes; and SD was more similar with AD in GC-MS analysis of volatile profile (of higher MW) for both cultivars from the PCA outcome. Furthermore, significant difference in volatile isomeric composition of different samples was also clearly demonstrated through extracted ion chromatogram (EIC) by GC-MS analysis. GC-IMS analysis showed the favorability of FD to preserve ketones, phenols, esters, and aromatic aldehydes; and SD was more similar with FD in GC-IMS analysis of volatile profile (of smaller MW) for both cultivars from the PCA outcome. Moreover, the OAVs indicate that 2-methoxy-4-vinylphenol contributed much to the flavor of dried Ponkan peel, while 2-methoxy-4-vinylphenol, methyl methanthranilate, and methyl anthranilate played an important role in the flavor of dried Chachi peel; and the highest OAVs for monoterpenes were observed at SD for both cultivars. Thus, the combination of GC-MS and GC-IMS analyses with PCA in this paper suggested the superiority of SD to preserve volatiles and characteristic aroma in dried citrus peel, and that SD contributed much to the quality of dried Chachi peel.


Introduction
Citrus, belonging to the Rutaceae family, are well known as one of the world's major fruit crops that are produced in many countries, such as China, USA, Brazil, etc. [1,2]. Due to the flavorful and pro-health properties, citrus fruits are one of the most readily consumed fruits in the world. In addition to large scale consumption as fresh fruits, the citrus fruits are mainly processed to produce juice, which generates a huge amount of citrus waste [3].

Drying of the Fresh Citrus Peels
About 100 g fresh citrus peel pieces were spread out evenly and subject to three different drying methods (sun-drying, hot-air-drying, and freeze-drying), and drying lasted until there was no change on the weight of peel. Finally, the water content (wt/wt) of the dried citrus peel prepared by different drying methods was all at the level of about 9%.
In sun-drying (SD), the fresh citrus peel pieces were left under direct sun daylight at temperatures about 25 • C for 4 days with about 32 h of daylight. In hot-air-drying (AD), the fresh citrus peel pieces were dried in an electric thermostatic drying oven (DHG-9240A; Shanghai Qixin Scientific Instrument Co., Shanghai, China) at 70 • C (the temperature was chosen based on the survey of production of dried Ponkan peel from Fujian province) for about 4 h. In freeze-drying (FD), the fresh citrus peel pieces were first frozen at −80 • C for 8 h, and then were quickly placed into a freeze dryer (Scientz-10N, Ningbo Xinzhi Biotechnology Co., Ningbo, China) and dried under vacuum (−60 • C, 1 Pa) for about 48 h. All dried citrus peel, that is, sun-dried peel from Citrus reticulata "Chachi" (Guangdong) (GSD) and Citrus reticulata "Ponkan" (Fujian) (FSD), hot-air-dried peel from Citrus reticulata "Chachi" (Guangdong) (GAD) and Citrus reticulata "Ponkan" (Fujian) (FAD), and freeze-dried peel from Citrus reticulata "Chachi" (Guangdong) (GFD), and Citrus reticulata "Ponkan" (Fujian) (FFD) were all stored at −20 • C until further analyzed.

HS-SPME-GC-MS Analysis
HS-SPME-GC-MS analysis of dried citrus peel samples was conducted according to the method of Qiu et al. [27] with some modifications. A certain amount of samples (2 g) was finely ground in high-speed universal grinder (FW100, Tianjin Tester Instrument Co., Tianjin, China.) and mixed with 100 µL of phenyl ethyl acetate (20 µL/mL in methanol, an internal standard), then immediately sealed in a 20 mL vial covered with an aluminum seal with a PTFE septum. The sample vial was then equilibrated at 90 • C for 5 min on a heating platform. The extraction was conducted by inserting the preconditioned SPME fiber (DVB/PDMS) (75 µm) into the head space of the vial for 30 min at 90 • C. At the end of extraction, the fiber was desorbed into the injection port of GC for 5 min.
GC-MS analysis was performed on Agilent 8890/7000D (Agilent technologies Inc., Santa Clara, CA, USA). Volatiles were separated using a HP-5MS (30 m × 0.25 mm × 0.25 µm, Agilent technologies Inc., Santa Clara, CA, USA). The carrier gas was helium (>99.999%) at a constant flow rate of 1 mL/min. The oven temperature was initially held at 70 • C for 2 min, then increased to 210 • C at 4 • C/min, and held for 10 min. The injector temperature was maintained at 250 • C with a split ratio of 1:20. The transfer line to the mass spectrometer was maintained at 250 • C and the ion source temperature was 230 • C. The mass detector was operated in the electron impact mode with ionization energy of 70 eV and a scanning rage of 35-500 a.m.u. Compounds were identified by comparing their mass spectra and linear retention index (LRI) [28] with those contained in the NIST mass spectrum database, as well as the available reference standards. The content of the volatiles was approximately semi-quantified by comparing the peak area with that of the internal standard (phenyl ethyl acetate) in the total ion chromatogram. The peak area normalization method was also used to obtain the relative percentage content of each compound in the samples.

HS-GC-IMS Analysis
HS-GC-IMS analysis was completed on a GC-IMS instrument (FlavourSpec ®® , Gesellschaft für Analytische Sensorsysteme, Dortmund, Germany) equipped with an autosampler unit (CTC Analytics AG, Zwingen, Switzerland), which uses a 1 mL air-tight heated syringe to directly sample from the headspace. The GC was equipped with a FS-SE-54-CB-1 capillary column (15 m × 0.53 mm ID, 1 µm). Sample analysis was determined based on the method reported in the literature [29] with some slight modifications. A total of 0.5 g of finely ground samples were transferred into a 20 mL headspace glass sampling vial and incubated at 60 • C for 20 min with rotation speed of 500 r/min. A total of 500 µL of headspace samples was automatically injected into the injector (85 • C, splitless mode) by means of a heated syringe at 85 • C. Nitrogen (99.999%) was used as carrier gas and its flow rate was first set at 2 mL/min for 2 min, then increased to 10 mL/min within 8 min, then increased to 100 mL/min within 10 min, and then increased to 150 mL/min within 5 min, and held for 20 min. The analytes were eluted and separated at 60 • C, then driven into an ionization chamber (quasi-enclosed radioactive material) and ionized by a 3 H ionization source in positive ion mode. The resulting ions were driven to a drift tube (98 mm in length) which operated on a constant temperature (45 • C) and voltage (5 kV). The retention index (RI) of each compound was calculated using n-ketones C4-C9 as external references. Volatile compounds were identified based on RI and drift time compared to GC-IMS library (NIST and IMS databases).

Calculation of Odor Activity Values (OAVs)
To assess the attribution of the volatile compounds to the overall dried citrus peel aroma, the odor activity values (OAVs) were calculated by dividing the concentrations of aroma compounds with their sensory thresholds from the literature [30]. Only the compounds with an OAV greater than 1 contribute individually to the dried citrus peel aroma.

Statistical Analysis
The results given in the tables and figures were means and standard deviations of triplicate experiments. Experimental data were analyzed using Origin 8.0 (Microcal Software, Inc., Northampton, MA, USA). Significant differences between samples were analyzed using analysis of variance (ANOVA) and Duncan's multiple-range test (p < 0.05). The correlation analysis and principal component analysis (PCA) were also conducted using Origin 8.0. All identified volatile compounds were engaged in the principal composition analysis. The gallery plot plug-in was used in GC-IMS analysis for fingerprint comparison, i.e., visual and quantitative comparisons of volatile differences between different samples.

GC-MS Analysis of Volatile Profile of Dried Citrus Peel as Effected by Drying Methods
In present study, by using SPME-GC-MS analysis, a total of 56 volatile compounds were tentatively identified and semi-quantified in dried Citrus reticulata "Chachi"/"Ponkan" peel prepared from different drying methods, as listed in Table 1 (the approximate concentration of each compound in samples) and Table S1 (the relative percentage content of each compound in samples through the peak area normalization method). For all 56 compounds, identification was based on the chromatographic peak LRI, similarity index higher than 80%, and the available reference standards. In Table 1, the volatiles could be classified as alcohols, aromatic hydrocarbons and ethers, phenols, N-containing compounds, aldehydes, monoterpenes, sesquiterpenes, and esters, with the approximate concentration and relative percentage content of each category shown in Figure 1a,b, respectively. It could be found in Figure 1a that, for the total approximate concentration of all detected compounds, there was the order of SD > AD > FD for dried peel of both Chachi and Ponkan. Thus, it could be deduced that sun-drying may partly account for the quality of dried Chachi peel.  Figure 1a,b, respectively. It could be found in Figure 1a that, for the total approximate concentration of all detected compounds, there was the order of SD > AD > FD for dried peel of both Chachi and Ponkan. Thus, it could be deduced that sun-drying may partly account for the quality of dried Chachi peel.

Alcohols
As shown in Table 1, the alcohols identified in dried citrus peel were mainly terpenic alcohols, with 2-methyl-3-buten-2-ol only identified in GFD and β-copaen-4α-ol only identified in GAD/GSD. Moreover, linalool and α-terpineol were the main alcohols in both cultivars, with the highest content of linalool and α-terpineol observed in FSD and GSD, respectively. In addition, for both cultivars, SD and AD showed much higher content of linalool and α-terpineol than FD, demonstrating the unfavourability of FD to preserve alcohols in dried citrus peel. Since FD was also unfavorable to keep other categories of volatile compounds, the percentage of alcohols in volatiles of freeze-drying citrus peel was no less than citrus peel made from SD and AD, as shown in Figure 1b. Thus, the combination of content and relative percentage was necessary to investigate the effect of drying methods on the volatile composition.
Since much isomers were detected in volatiles of dried citrus peel, the extracted ion chromatogram (EIC) of different isomers is shown in Figure 2, to more clearly demonstrate the chemical composition of dried citrus peel as influenced by drying methods for the two cultivars. Figure 2a shows the chromatograms of three terpenic alcohol isomers (C 10 H 18 O: linalool, terpinen-4-ol, α-terpineol) through EIC of m/z 136. It can be seen from Figure 2a that dried Chachi peel showed much higher content of α-terpineol and terpinene-4-ol (both cyclic terpenic alcohol), especially for SD; and smaller content of linalool (acyclic terpenic alcohol). Thus, the ratio of cyclic to acyclic terpenic alcohol might be a potential indictor to discriminate dried Chachi peel and dried Ponkan peel.

Aromatic Hydrocarbons and Ethers
Two aromatic hydrocarbons (β-cymene, p-cymenene) and one aromatic ether (2isopropyl-5-methylanisole) were detected in the two cultivars. For both cultivars, it can be found from Table 1 that p-cymenene were only detected in dried Ponkan peel and SD was the best to preserve β-cymene.

Phenols
In Table 1, four phenols (p-thymol, thymol, carvacrol and 2-methoxy-4-vinylphenol) were detected in two cultivars, and FD citrus peel showed the largest peak area and percentage of phenols as compared with dried citrus peel from other two drying methods, suggesting the favorability of FD to preserve phenols in dried citrus peel. Previous reports showed that phenols were bound volatiles with 2-methoxy-4-vinylphenol of the most abundant bound volatile compound in the citrus fruits [31]. In this experiment, phenols were detected in free fractions, and the main phenol in dried Chachi and Ponkan peel were different, with 2-methoxy-4-vinylphenol as the main phenol in FFD and thymol as the main phenol in GFD. Moreover, for the three phenol isomers (p-thymol, thymol, carvacrol), the EIC of m/z 150 (Figure 2b) could clearly demonstrate the composition of phenol isomers (C 10 H 14 O: MW 150) as influenced by drying methods for the two cultivars: FD was the best to preserve all detected phenols in dried citrus peel; and dried Chachi peel showed more categories and content of phenol isomers (C 10 H 14 O) than dried Ponkan peel.
of volatile compounds, the percentage of alcohols in volatiles of freeze-drying citrus peel was no less than citrus peel made from SD and AD, as shown in Figure 1b. Thus, the combination of content and relative percentage was necessary to investigate the effect of drying methods on the volatile composition.
Since much isomers were detected in volatiles of dried citrus peel, the extracted ion chromatogram (EIC) of different isomers is shown in Figure 2, to more clearly demonstrate the chemical composition of dried citrus peel as influenced by drying methods for the two cultivars. Figure 2a shows the chromatograms of three terpenic alcohol isomers (C10H18O: linalool, terpinen-4-ol, α-terpineol) through EIC of m/z 136. It can be seen from Figure 2a that dried Chachi peel showed much higher content of α-terpineol and terpinene-4-ol (both cyclic terpenic alcohol), especially for SD; and smaller content of linalool (acyclic terpenic alcohol). Thus, the ratio of cyclic to acyclic terpenic alcohol might be a potential indictor to discriminate dried Chachi peel and dried Ponkan peel.

N-Containing Compounds
There were many reports on the presence of methyl methanthranilate in dried Chachi peel. In this experiment (Table 1), despite of different drying methods, methyl methanthranilate was detected both in dried Chachi and Ponkan peel with the former of much higher content (as can be seen in Figure 1). The other two N-containing compounds (methyl anthranilate, 4-acetyl-3,5-dimethyl-2-pyrrolecarboxylic acid, methyl ester) were only detected in dried Chachi peel with low content, suggesting the content and number of N-containing compounds could be used to control the quality of dried Chachi peel, in agreement with the previous report [22]. The same effect of drying methods on the percentage of N-containing compounds was observed for the two cultivars: FD > AD > SD. However, for the content of N-containing compounds, there was no significant effect of drying methods on dried Ponkan peel, while SD and AD were better to preserve N-containing compounds than FD for dried Chachi peel.

Aldehydes
The aldehydes detected in dried citrus peel in this experiment include aromatic aldehydes (furfural, benzaldehyde), cyclic aldehyde (perillal), saturated aliphatic aldehydes (octanal, decanal, undecanal, dodecanal and tetradecanal), and unsaturated aliphatic aldehydes (β-sinensal, α-sinensal, 7-pentadecene-7-carbaldehyde). Aliphatic aldehydes, such as octanal, decanal, dodecanal, and sinensal were reported to show significant aroma activity in citrus fruit oil [32,33]. It can be found from Table 1 that dried Ponkan peel showed higher content of both aromatic and aliphatic aldehydes than dried Chachi peel except for α-sinensal and perillal, and dried Chachi peel contains much higher content of α-sinensal (the most aldehyde detected in dried Chachi peel) than dried Ponkan peel; thus, total aldehydes were observed to be higher in dried Chachi peel. For the peel of both cultivars, FD was the best to preserve aromatic aldehydes while AD was the best to preserve aliphatic aldehydes.

Monoterpenes
In agreement with previous reports [16], monoterpenes (C 10 H 16 : MW 136) were often the main compounds in volatiles of dried citrus peel except for the FD peel, as shown in Figure 1a; and among all monoterpenes, D-limonene showed the highest content, followed by γ-terpinene. β-Myrcene and D-limonene were characterized as the potent odorants in the volatile fraction from Pontianak orange peel and wild citrus Mangshanyegan peel oil [34,35]. Under different drying methods, dried Chachi peel always showed the smaller peak area ratio of D-limonene/γ-terpinene than dried Ponkan peel, in accordance with the preliminary review [36]. Moreover, if dried Ponkan peel was made from SD rather than the often-used SD, the ratio of D-limonene/γ-terpinene could also be used to discriminate dried Chachi and Ponkan peel, thus suggesting again that the ratio of D-limonene/γterpinene could be used to identify dried Chachi peel. In Table 1 and Figure 1, it can be also found that SD was the best way to preserve monoterpenes, followed by AD, and FD was unfavorable to preserve monoterpenes. Moreover, to clearly demonstrate the monoterpene composition as influenced by drying methods in the two cultivars peel, the EIC (m/z 136) of all dried peel are also shown in Figure 2c, which clearly shows that the content and number of monoterpenes was highest in dried Chachi peel, especially for GSD, suggesting that monoterpenes were the characteristic compounds of dried Chachi peel and, again, SD may account for the quality of dried Chachi peel.

Sesquiterpenes
For sesquiterpenes (C 15 H 24 : MW 204), they were also the other important part in volatiles of dried citrus peel, and some sesquiterpenes, such as copaene, β-elemene, caryophyllene and farnesene were common in different citrus germplasms [17,37]. The sesquiterpenes were reported to have efficacy in defending against insects and fungus [38,39]. As shown in Figure 1a and Table 1, FD was also unfavorable to preserve sesquiterpenes; however, AD was the best way to preserve sesquiterpenes, which was unlike monoterpenes. The EIC (m/z 204) of all dried peel are also shown in Figure 2d to clearly demonstrate that the sesquiterpene composition is influenced by drying methods in the two cultivars. It can be seen from Figure 2d that the content and number of sesquiterpene isomers was highest in AD peel for both two cultivars. Moreover, each cultivar peel had its characteristic sesquiterpenes, such as γ-elemene, trans-β-farnesene and germacrene B only in dried Ponkan peel while α-humulene and (Z, E)-α-farnesene only in dried Chachi peel; and dried Chachi peel contains much higher content of α-farnesene (the most sesquiterpene detected in dried Chachi peel) than dried Ponkan peel, resulting in a higher level of total sesquiterpenes in dried Chachi peel.

Esters
In Table 1, four unsaturated fatty acid esters (neryl acetate, methyl palmitoleate, methyl linoleate, and methyl linolenate) and one saturated fatty acid ester (methyl palmitate) were detected in dried peel of the two cultivars. Neryl acetate was a major monoterpene ester in most citrus germplasms [17] and showed decreased content in the oleocellosis peel of orange [40]. For both cultivar peels, it can be found from Table 1 that straight-chain fatty acid methyl esters (methyl palmitoleate, methyl palmitate, methyl linoleate, and methyl linolenate) were only detected in FD peel with saturated fatty acid methyl ester (methyl palmitate) of highest peak area and percentage; while branched monoterpene ester (neryl acetate) was undetected in FD peel. Then, for both cultivars, esters accounted for about 20% for FD peel while it had an almost negligible percentage for AD/SD peel (Table S1 and Figure 1b). Thus, there were few reports on straight-chain fatty acid methyl esters in dried citrus peel since citrus peel was always dried by AD or SD.

Correlation and PCA Analysis
For overall analysis of all detected volatiles by GC-MS in dried citrus peel from different drying methods, correlation and PCA analysis was employed in Table 1 with results shown in Figures 3, 4 and S1, respectively. In Figure 3, the Pearson's correlation coefficient between samples showed that the triplicate samples of the same citrus peel with the same drying method were highly correlated, indicating the good repeatability of the samples within the group. Sample correlation analysis also showed that the differences between FAD and FSD, GAD and GSD were the smallest, followed by the differences between GSD and FAD, GSD and FSD, GFD and FFD, GSD and GFD, and the differences between GFD and FAD, GFD and FSD were the largest. In PCA analysis, the first three principal components covered 88.54% of total variance (principal component-1 = 46.30%, principal component-2 = 30.70%, principal component-3 = 11.54%). Figure 4 shows the first two principal component score/loading plots (covering 77.00% of total variance) while the plots of principal component-1 vs. principal component-3 and principal component-2 vs. principal component-3 are shown in Figure S1. It can be directly observed from Figures 4a and S1a,c that both drying methods and cultivars showed great influences on the volatile composition of dried citrus peel, and SD peel was more similar with AD peel in GC-MS analysis of volatiles. For overall analysis of all detected volatiles by GC-MS in dried citrus peel from different drying methods, correlation and PCA analysis was employed in Table 1 with results shown in Figures 3, 4, and S1, respectively. In Figure 3, the Pearson's correlation coefficient between samples showed that the triplicate samples of the same citrus peel with the same drying method were highly correlated, indicating the good repeatability of the samples within the group. Sample correlation analysis also showed that the differences between FAD and FSD, GAD and GSD were the smallest, followed by the differences between GSD and FAD, GSD and FSD, GFD and FFD, GSD and GFD, and the differences between GFD and FAD, GFD and FSD were the largest. In PCA analysis, the first three principal components covered 88.54% of total variance (principal component-1 = 46.30%, principal component-2 = 30.70%, principal component-3 = 11.54%). Figure 4 shows the first two principal component score/loading plots (covering 77.00% of total variance) while the plots of principal component-1 vs. principal component-3 and principal component-2 vs. principal component-3 are shown in Figure S1. It can be directly observed from Figures 4a and S1a,c that both drying methods and cultivars showed great influences on the volatile composition of dried citrus peel, and SD peel was more similar with AD peel in GC-MS analysis of volatiles.    In Figures 4b and S1b, all three N-containing compounds, four sesquiterpenes ((-)-βcaryophyllene, (Z, E)-α-famesene, α-selinene, α-farnesene), and thymol were of positive higher loadings at principal component-1, while five aldehydes (dodecanal, β-sinensal, furfural, decanal, octanal) and four sesquiterpenes (germacrene B, γ-elemene, trans-βfamesene, δ-elemene) were of negative higher loadings at principal component-1, suggesting these compounds represented most important information of principal component-1. Based on the interpretation of principal component-1, it can be seen from Figures 4a and S1a that N-containing compounds, sesquiterpenes, and aldehydes were important compounds to clearly discriminate dried Chachi and Ponkan peel, with both possessing its individual characteristic sesquiterpenes and aldehydes, and much higher content of Ncontaining compounds in dried Chachi peel.
For principal component-2, as shown in Figures 4b and S1d, two monoterpenes (Dlimonene and β-myrcene) and two alcohols (linalool and elemol) were of positive higher loadings, while four straight-chain fatty acid methyl esters (methyl palmitoleate, methyl palmitate, methyl linoleate and methyl linolenate) and two phenols (p-thymol, carvacrol) were of negative higher loadings, suggesting these compounds represented most important information of principal component-2. Based on this interpretation, it can be seen from Figures 4a and S1c that monoterpenes (including the main compound D-limonene), straight-chain fatty acid methyl esters and phenols were important compounds to clearly discriminate FD peel and AD/SD peel, with FD favorable to preserve phenols and esters while unfavorable to preserve monoterpenes when being compared with AD or SD.

OAVs and Sensory Attributes
The OAV value can be used to measure the contribution of aroma compounds to the overall aroma of the sample. The OAVs of the aroma compounds detected by GC-MS analysis, together with their odor threshold values, are listed in Table 2. In Figures 4b and S1b, all three N-containing compounds, four sesquiterpenes ((-)-βcaryophyllene, (Z, E)-α-famesene, α-selinene, α-farnesene), and thymol were of positive higher loadings at principal component-1, while five aldehydes (dodecanal, β-sinensal, furfural, decanal, octanal) and four sesquiterpenes (germacrene B, γ-elemene, trans-βfamesene, δ-elemene) were of negative higher loadings at principal component-1, suggesting these compounds represented most important information of principal component-1.
Based on the interpretation of principal component-1, it can be seen from Figures 4a and S1a that N-containing compounds, sesquiterpenes, and aldehydes were important compounds to clearly discriminate dried Chachi and Ponkan peel, with both possessing its individual characteristic sesquiterpenes and aldehydes, and much higher content of N-containing compounds in dried Chachi peel.
For principal component-2, as shown in Figures 4b and S1d, two monoterpenes (D-limonene and β-myrcene) and two alcohols (linalool and elemol) were of positive higher loadings, while four straight-chain fatty acid methyl esters (methyl palmitoleate, methyl palmitate, methyl linoleate and methyl linolenate) and two phenols (p-thymol, carvacrol) were of negative higher loadings, suggesting these compounds represented most important information of principal component-2. Based on this interpretation, it can be seen from Figures 4a and S1c that monoterpenes (including the main compound D-limonene), straight-chain fatty acid methyl esters and phenols were important compounds to clearly discriminate FD peel and AD/SD peel, with FD favorable to preserve phenols and esters while unfavorable to preserve monoterpenes when being compared with AD or SD.

OAVs and Sensory Attributes
The OAV value can be used to measure the contribution of aroma compounds to the overall aroma of the sample. The OAVs of the aroma compounds detected by GC-MS analysis, together with their odor threshold values, are listed in Table 2.
Despite of different drying methods, the higher OAVs for 2-methoxy-4-vinylphenol and linalool were observed for dried Ponkan peel while the higher OAVs for methyl anthranilate, methyl methanthranilate, perillal, α-sinensal, γ-terpinene, and terpinolene were observed for dried Chachi peel. The OAVs indicated that these compounds could differentiate the aroma of dried Ponkan and Chachi peel. Furthermore, for both dried Ponkan and Chachi peel, the highest OAVs for monoterpenes were both observed for the SD, indicating monoterpenes may be the characteristic aroma compounds of SD citrus peel. In addition, although D-limonene was the most abundant compound among the volatile components of citrus fruits, its contribution to the aroma was not so much, in accordance with the literature [41].

GC-IMS Analysis of Volatile Profile of Dried Citrus Peel as Effected by Drying Methods
This study also adopted HS-GC-IMS to determine the volatiles in dried citrus peel of two cultivars prepared from different drying methods. The flavor compounds were characterized by comparing their drift time and retention index from the HS-GC-IMS spectrum ( Figure S2) with results shown in Table 3. In Table 3, 60 typical flavor compounds were identified from the topographic maps of dried citrus peel via GC-IMS library searches, and some individual compounds generated several signals due to the existence of different polymerization. Moreover, the identified compounds listed in the gallery plot were combined to show the characteristic flavor fingerprints of different dried Chachi and Ponkan peels in Figure 5, where one row represents one sample, and one column represents the intensity of the compound in different samples. A brighter signal of each compound indicates stronger signal intensity, while a darker signal indicates weaker signal intensity. 1.1901 MW represents molecular mass, RI represents the relative retention index calculated using n-ketones C4-C9 as external standard on FS-SE-54-CB column, Rt represents the retention time in the capillary GC column, Dt represents the drift time in the drift tube. As shown in Figure 5 and Table 3, 14 alcohols, 16 aldehydes, 4 esters, 11 ketones, 4 monoterpenes, 3 acids, and 8 other compounds were detected in volatiles of different dried citrus peel by GC-IMS analysis. By comparing Table 1 and Table 3, it can be found that most detected compounds were different between GC-MS and GC-IMS analyses, suggesting that GC-MS and GC-IMS analyses are complementary in volatile analysis, in agreement with the literature [42].

Aldehydes and Alcohols
Aldehydes were the most widely detected compounds in dried citrus peel, with 16 aldehydes detected in Figure 5b. Among these aldehydes, most aldehydes contained no more than 10 carbons. The aldehydes were reported from degradation of fatty acid [43]. It was shown that aldehydes were important odorants, imparting the flavor of grass, fruit, almond, and sweet [44]. Hexanal and (Z)-3-hexenal were reported to be important As shown in Figure 5 and Table 3, 14 alcohols, 16 aldehydes, 4 esters, 11 ketones, 4 monoterpenes, 3 acids, and 8 other compounds were detected in volatiles of different dried citrus peel by GC-IMS analysis. By comparing Tables 1 and 3, it can be found that most detected compounds were different between GC-MS and GC-IMS analyses, suggesting that GC-MS and GC-IMS analyses are complementary in volatile analysis, in agreement with the literature [42].

Aldehydes and Alcohols
Aldehydes were the most widely detected compounds in dried citrus peel, with 16 aldehydes detected in Figure 5b. Among these aldehydes, most aldehydes contained no more than 10 carbons. The aldehydes were reported from degradation of fatty acid [43]. It was shown that aldehydes were important odorants, imparting the flavor of grass, fruit, almond, and sweet [44]. Hexanal and (Z)-3-hexenal were reported to be important odorants in the headspace above the Pontianak peels [35]. Dried Ponkan peel showed higher concentration of furfural and benzaldehyde than dried Chachi peel, in agreement with GC-MS analysis; while dried Chachi peel showed higher concentration of 3-methylbutanal (malt) and (E)-2-heptenal (soap, fat, almond [44]) than dried Ponkan peel. AD was unfavorable to preserve (E)-2-pentenal (strawberry, fruit, tomato [44]).
The detected alcohols were only smaller than aldehydes with 14 alcohols detected as shown in Figure 5a. Alcohols were also important odorants in citrus fruit: linalool and 1,8-cineole were characterized as the most potent odorant in citrus fruit oil [32,33,35]. Linalool, with the brightest points (implying high concentration) in GC-IMS analysis, was also detected in GC-MS analysis, and higher concentration of linalool was observed in dried Ponkan peel through linalool-polymer point, in agreement with GC-MS analysis. It can be also found from Figure 5a that n-hexanol (resin, flower, green [44]) was characteristic for SD peel; and FD was favorable to preserve 2-methyl-1-butanol (wine, onion [44]).

Monoterpenes
According to the brightness of points, it can be found that monoterpenes in Figure 5c were of the highest content in different dried citrus peels, in accordance with GC-MS data in Table 1. Monoterpenes impart typical aromas of citrus, mint and turpentine, contributing the aromatic notes in orange peel oil [35,45]. However, different from GC-MS analysis, the effect of drying methods on monoterpenes of dried citrus peels was unclear in GC-IMS analysis, probably because of saturated detection due to much higher concentration of monoterpenes in dried citrus peels and great sensitivity of the IMS detector.

Esters, Ketones, Acids, and Others
Esters, usually of fruity or floral smell [44], are the main components contributing to citrus head scent; and ethyl acetate and ethyl butyrate were often considered as quality indexes of citrus aroma [46]. For esters detected in GC-IMS analysis, although different esters were detected as compared with GC-MS analysis, FD was always the best way to preserve esters.
Ketones and acids were undetected in GC-MS analysis, while 11 ketones and 3 acids were detected in GC-IMS analysis (Figure 5d,f). Ketones impart the odor of butter, cream, and ether. It can be found from Figure 5d that AD was unfavorable to preserve most ketones, including 2,3-butanedione, 2-hexanone, acetophenone, cyclohexanone, and 6-methyl-5hepten-2-one, which were higher in dried Chachi peel than in dried Ponkan peel. While hydroxyacetone were higher in AD peel since hydroxyacetone was formed through pyrolysis of carbohydrate at high temperature during the AD process [47]. Among the three detected acids (Figure 5f), 2-methylbutanoic acid (cheese, sweat [44]) was higher in GSD/GFD than other dried peels. Overall, SD could preserve ketones (2,3-butanedione, 2-hexanone, acetophenone, cyclohexanone and 6-methyl-5-hepten-2-one) and 2-methylbutanoic acid as FD, and dried Chachi peel possessed higher concentration of these ketones and acid than dried Ponkan peel. Thus, ketones (2,3-butanedione, 2-hexanone, acetophenone, cyclohexanone and 6-methyl-5-hepten-2-one) and 2-methylbutanoic acid could also be considered as the characteristic compounds of dried Chachi peel and the contribution of SD to the quality of dried Chachi peel might be partly from the preservation of these ketones and acid.
For other detected compounds in GC-IMS analysis, including S-containing compounds, 2-acetylfuran, 2-ethyl-5-methylpyrazine, styrene, and two phenols, it could be found that 2-acetylfuran was a characteristic compound in GFD; while two phenols were characteristic for dried Chachi peel with much higher concentration in GFD/GSD than GAD, suggesting again that FD was favorable to preserve phenols and SD could preserve these two detected phenols in GC-IMS analysis as FD.

Correlation and PCA Analysis
Correlation and PCA analysis were also employed in GC-IMS analysis with results showing that the samples within the group have good repeatability, and the Pearson's correlation coefficient between SD and FD was higher than that between SD and AD (as shown in Figure S3); and for both cultivars, SD citrus peel was more similar with FD citrus peel (as shown in Figure S4), suggesting SD could preserve volatile flavor compounds (with low molecular weight) as FD. Thus, GC-IMS analysis could serve as a tool to discriminate dried citrus peel in the market from SD and AD, and GC-IMS analysis again demonstrated the contribution of SD to the quality of dried Chachi peel.

Conclusions
The volatile compounds of dried Chachi and Ponkan peel prepared from three drying methods were analyzed by GC-MS and GC-IMS with a total of 56 and 60 volatile compounds being detected, respectively. The result of GC-MS indicated that for both cultivars, sun-drying was favorable to preserve terpenic alcohols (linalool, α-terpineol and terpinene-4-ol), β-cymene, methyl methanthranilate and monoterpenes, and hot-airdrying was favorable to preserve aliphatic aldehydes and sesquiterpenes. Moreover, the extracted ion chromatogram of isomers in the GC-MS analysis was useful in illustrating the effect of drying methods on the isomeric composition of volatiles in dried citrus peel. GC-IMS data showed that freeze-drying was the best to preserve esters and phenols; hotair-drying was unfavorable to preserve most ketones; and n-hexanol was characteristic for sun-drying peel.
In addition, the PCA outcomes demonstrated that for dried citrus peel of both cultivars, sun-drying showed the similar volatile profile (of higher MW) with hot-air-drying through GC-MS analysis, and also the similar volatile profile (of smaller MW) with freeze-drying through GC-IMS analysis, suggesting sun-drying was the best drying method to preserve total volatiles in dried Chachi and Ponkan peel, and sun-drying played an important role in the quality of dried Chachi peel.
Furthermore, the odor activity values indicate that 2-methoxy-4-vinylphenol contributed much to the flavor of dried Ponkan peel, while 2-methoxy-4-vinylphenol, methyl methanthranilate, and methyl anthranilate played an important role in the flavor of dried Chachi peel; and the highest odor activity values for monoterpenes were both observed at hot-air-drying for both cultivars, indicating monoterpenes may be the characteristic aroma compounds of hot-air-drying citrus peel.  Figure S4: Plots of principal component scores (a) and loadings (b, c, d) in PCA analysis for GC-IMS data of dried Chachi and Ponkan peels prepared from three drying methods; Table S1: The relative percentage content of each compound in GC-MS total ion chromatogram of different dried citrus peel samples through peak area normalization.