Effects of Drying Methods on the Volatile Compounds of Allium mongolicum Regel

Allium mongolicum Regel (AMR) is a traditional Mongolian food. Various drying methods play an important role in foodstuff flavor. However, the effect of different drying methods on AMR is limited. In this study, freeze drying (FD), vacuum drying (VD), and hot-air drying (HAD) were applied to dry fresh AMR to a moisture content of 8% (wet basis); headspace gas chromatography mass spectrometry was adopted to identify volatile compounds in AMR; and principal component analysis and fingerprint similarity analysis based on the Euclidean distance was used to distinguish the fresh and three dried treatments. In total, 113 peaks were detected and 102 volatile compounds were identified. Drying causes significant changes to the amounts of volatile compounds in AMR, and the drying method plays a key role in determining which volatile compounds appear. Compared to FD, VD and HAD were more appropriate for drying AMR because the volatile compounds after VD and HAD were closer to those of fresh AMR. These findings can provide a scientific basis to help to preserve future seasonal functional food and aid in Mongolian medicine production.


Introduction
Allium mongolicum Regel (AMR), a perennial Liliaceae herb native to Mongolia, Kazakhstan, and China (Inner Mongolia, Gansu, Ningxia, and Xinjiang) [1] and also known as the Mongolia leek, grows in high-altitude desert steppe and desert areas [2]. AMR is resistant to wind erosion, drought, and low temperature and is therefore important in preventing wind erosion as well as in maintaining and improving regional ecological environments.
AMR is a traditional Mongolian medicinal herb [3]. Its leaves are rich in nutrients, possess a unique flavor, and is a natural green uncontaminated food [4] used to make sauces in cold vegetable dishes and stuffing of steamed stuffed buns and dumplings. In addition, AMR is beneficial in managing functional constipation, helping to maintain colon water content and increasing intestinal transit [1]. The aqueous extract of AMR has the potential to be used as a functional food or nutraceutical in the prevention and treatment of obesity and hypertension [3].
AMR is widely distributed on grassland, and it is eaten by both humans and sheep. It is found that sheep meat from AMR-pasture-fed animals tastes better. Zhao et al. [5] found that feeding AMR to sheep could significantly improve the quality of mutton.
Some research yielded AMR extracts to feed sheep [2,6]. The water-insoluble flavonoids from AMR were found to reduce the unpleasant mutton flavor and improve meat quality of sheep. AMR polysaccharide supplementation significantly increased serum total protein, albumin, and globulin concentrations and reduced mortality after LPS challenge [7]. AMR supplementation can also affect nutrient digestibility, CH 4 emission, and the antioxidant capacity of Simmental calves in northwest China [8].
However, fresh AMR is seasonal and perishable because of its high moisture content, making AMR difficult to be enjoyed year round. Overcoming this seasonality is crucial to utilizing AMR. Drying, as an essential processing operation, can reduce moisture to a safe level at which microorganisms and deterioration reactions are inhibited. Freeze drying (FD) has the highest cost, but produces the highest quality food products among all drying methods [9,10]. Hot-air drying (HAD) is low-cost and the most common method that is applied to the storage of agricultural products [11,12]. The main drawbacks of HAD are undesirable physical, structural, chemical, organoleptic, and nutritional changes [13]. Vacuum drying (VD) with low pressure and low drying temperatures can yield a higher quality than atmospheric drying [14].
Flavor is mainly determined by volatile components and is a key factor for consumers to accept dried AMR. Headspace gas chromatography mass spectrometry (HS-GC-MS) is a new and reliable method used to identify volatile compounds in foods, such as fermented rose jams [15], Dezhou braised chicken [16], dry-cured fish [17], yak milk [18], soybeans [19], tea [20], Chinese prickly ash peels [21], canned bamboo shoots [22], and fresh-cut yams [23]. HS-GC-MS was used to analyze the changes of volatile components in fried Tricholoma matsutake Singer under different heating temperatures and times [24], and a total of 40 signals that corresponded to 24 compounds were identified. The effects of spray drying and FD on volatile components of yak milk powder were compared by Feng et al. [18]. Their results indicated that different drying methods affect the flavor of yak milk powder. Volatile components in Acacia honey powders were found to have significant differences under various drying conditions [25]. Changes in the volatile compounds of peppers during the drying process based on HS-GC-MS was studied by Ge et al. [26]. Their results indicated that the change of volatile compounds is significantly determined by the drying temperature.
Once moisture is removed from AMR, it has a potential to be used as a feed supplement and edible spice over a long term. However, dried AMR still has not been widely adopted. Until now, the kinds of drying methods that can be employed to dry AMR and retain flavor to its maximum are still not subject to research. In this study, HS-GC-MS was used to detect volatile components of AMR after processing by HAD, VD, and FD. The effect of these drying methods on the volatile compounds was studied with principal component analysis (PCA) and Euclidean distance analysis. Finally, the knowledge obtained from this study has the potential to be used to improve the industrial processing of AMR traditional feed supplements and edible production.

Materials
Fresh aerial parts of AMR were purchased from a local market in Baotou, Inner Mongolia, China, in August 2021. These parts were washed and the surface water of the sample was removed by centrifugation. Then, samples were cut into segments of 5 cm in length. Portions of each sample were immediately stored at 4 • C in sealed plastic bags as samples. Before the drying experiment began, the AMR in a sealed plastic bag was taken out and placed indoors until its temperature stabilized. The other samples were placed in a polypropylene tray and frozen at −25 • C for at least 8 h as the frozen samples. The initial moisture content of the sample was 91.93% (wet basis).

Drying Experiments
The AMR samples were dried by using different drying methods until the final moisture content was <8% (wet basis). These methods are described below. After drying, samples were vacuum-packed and stored for 10 days at ambient temperature (25 • C ± 5 • C) for subsequent analyses.

FD Process
The frozen samples (500 g) were dried by using a freeze dryer (SCIENTZ-18N, Ningbo Xinzhi Biotechnology Co., Ltd., Ningbo, China). The cold trap temperature was set at −60 • C, and the pressure of the drying chamber was set at 2 Pa during the drying process. Drying was completed after 29 h.

VD Process
The samples (500 g) were dried by using a vacuum dryer (DZF-6050B, Shanghai Yiheng Science Co., Ltd., Shanghai, China). The drying temperature was set at 50 • C and the pressure of the drying chamber was set at 2.0 × 10 4 Pa during the drying process. Drying was completed after 36 h.

HAD Process
The samples (500 g) were dried by using a hot-air dryer (ZXRD-B5110, Shanghai Zhicheng Analytical Instrument Manufacturing Co., Ltd., Shanghai, China). The drying temperature was set at 50 • C. Drying was completed after 46 h.
The column temperature was 60 • C, and the temperature of the MS was 45 • C. Nitrogen of 99.99% purity was used as a carrier gas and drift gas. The carrier gas was programmed to flow as follows: 2 mL/min during 0-2 min, 2-10 mL/min during 2-10 min, 10-100 mL/min during 10-20 min, and 100-150 mL/min during 20-40 min. The drift gas was set at 150 mL/min.

Statistical Analysis
Data from volatile compounds in samples were acquired and processed using Laboratory Analytical Viewer analysis software and Library Search qualitative software (G.A.S., Dortmund, Germany). Laboratory Analytical Viewer was used to view the analytical spectrum, where each dot represents a volatile compound. A reporter plugin was directly used to compare the spectral differences between samples (two-dimensional and three-dimensional views). A gallery plot plugin was used to compare fingerprints and visually and quantitatively compare the differences in volatile components among different samples. A dynamic PCA plugin was used for dynamic PCA and clustering analysis of the samples and to quickly determine the types of unknown samples. GC-IMS Library Search is an application software that was used to qualitatively analyze substances based on the NIST2014 Mass Spectral Library and IMS Library.

Visual Topographic Plot Comparison
The differences in volatile compounds in AMR samples from fresh and different drying methods were analyzed by using HS-GC-IMS. The data are represented by threedimensional topographical visualizations in Figure 1a, where the Y-axis represents the retention time of the gas chromatograph, the X-axis represents the ion migration time for identification, and the Z-axis represents the peak height for quantification. As can be seen from Figure 1a, the volatile compounds and the signal intensity of AMR of fresh and dried samples from different drying methods are significantly different. The volatile compounds in fresh AMR have maximum intensities. Compared with the fresh samples, some volatile compounds were generated and some volatile components disappeared, so the contents of some volatile components decreased and the contents of other volatile components increased in dried samples. This phenomenon was also observed by Guo et al. [24]. They reported that some compounds dramatically decreased and some volatile compounds were formed once processed by HAD. The data are represented by two-dimensional topographical visualizations in Figure 1b. In Figure 1b, the ion migration time and the position of the reactive ion peak were normalized. The fingerprint shows the total volatile components of the AMR samples. Each dot on the fingerprint represents a single volatile compound separated from the total volatile components. The color of the dots represents the signal intensity of the volatile components. The redder the color, the greater the signal intensity and the higher the content of the target volatile compound. As can be clearly seen from Figure 1b, the fingerprint of fresh AMR is obviously different from that of the dried AMR samples, with most of the signals within a retention time of 30-1000 s and with a drift time of 1.0-1.75. The fingerprints of the three dried AMR samples were similar to each other, but their signal intensities were slightly different. This means that each dried AMR sample had a unique specific flavor. The difference comparison model was applied to compare the differences of AMR samples. The topographic plot of fresh AMR was selected as a reference, and the topographic plot of dried samples was deducted from the reference. The results are shown in Figure 2. The white, red, and blue colors in dried samples indicate that the concentration of the volatile compounds is consistent with that of the reference, higher than that of the reference, and lower than that of the reference, respectively. It can be seen that most of the signals in the topographic plot of fresh and dried samples from different areas appear within a retention time of 30-1300 s. After drying, the signal intensities of some compounds decrease, while others disappear completely. In contrast, some signal intensities increase, indicating that the content of some compounds increases after drying. It also can be seen from Figure 2 that the signals for a retention time between 600 and 700 s in FD are intense, but in VD and HAD disappear. The different drying methods lead to variations in the volatile compounds in the dried AMR samples.

Effects of the Different Drying Methods on the Volatile Compounds in the AMR Samples
To further compare the specific volatile compounds in each group of samples, all peaks were selected for fingerprint plot comparison, as shown in Figure 3. Columns represent the detected substance, and rows represent the content of the same volatile compounds in different samples. Individual dots represent a volatile compound, and the color represents the content levels of the volatile compounds (the redder and brighter the color, the higher the content). In the fingerprint plot, unidentified volatile compounds are represented by numbers, and some volatile compounds with monomer and dimer morphologies were detected.
The amounts of methional, N-nitroso-morpholine, propyl 1-propenyl disulfide M, 2,3-diethyl-5-methyl pyrazine, propyl hexanoate, 2-acetyl furan and diallyl disulfide, labelled as E, decreased in dried AMR samples and their signal intensity was similar among different drying methods. Ethyl sulfide and ethyl acetate, labelled as E, suffered the most serious damage under FD and suffered less damage under VD and HAD. The amount of 2-methyl-2-pentenal M, labelled as E, was not decreased in FD, but almost vanished in VD and HAD, that is, 2-methyl-2-pentenal M in fresh AMR samples can only be preserved well by FD. This means that 2-methyl-2-pentenal M can be protected well by low temperature and sublimation.

Compound Identification
After analyzing AMR samples under different drying methods, we tentatively identified a total of 113 typical target signals by comparing the feature retention and drift times with those of the individual standard ion signals. These were confirmed by using the commercial GC-IMS Library Search and are represented by different numbers (1-113). The identified and unidentified compounds are listed in Table 1. The identified compounds include the compound name, CAS number, molecular formula, molecular weight, retention index, retention time, and drift time. The unidentified compounds include the retention index, retention time, and drift time. The tentatively identified volatile compounds in AMR samples included 5 ketones, 9 aldehydes, 12 alcohols, 9 esters, 4 acids, 3 ethers, 5 alkenes, 8 salts, 2 furans, 6 pyrazines, and 10 sulfide compounds.

Cluster Analysis of the Fresh and Dried AMR Samples
PCA is a multivariate statistical analysis technique established using signal intensity to highlight the differences in volatile compounds. The PCA results of the volatile compounds in the fresh and dried AMR samples are presented in Figure 4. The figure shows the distribution map for the first two principal components determined by PCA, which describe 69% and 18%, respectively, of the accumulative variance contribution rate, and a visualization of the data was obtained. These components were thought to show a similarity between the different AMR samples. The PCA results clearly show fresh and different dried AMR samples in a relatively independent space and these would be well distinguished in the distribution map. Dried AMR samples can be well distinguished according to the positive score values of PC1, while fresh samples can be well defined according to the negative scores of PC1, and the difference in different parts from different areas can be separated by combining the scores with the score values of PC2.
From Figure 4, one can see that flavors varied greatly between the fresh and dried AMR samples. Moreover, the flavors of FD AMR samples were vastly different from the flavors of VD and HAD AMR samples. The flavors of VD and HAD AMR samples were relatively similar. The results revealed that the characteristic volatile fingerprints of the fresh and dried AMR samples under different drying methods were successfully established through HS-GC-IMS. The HS-GC-IMS data contained valuable information and can be a useful tool for distinguishing AMR samples. Feng et al. [18] also reported that volatile compounds in yak milk powder have significant differences between FD and spray drying. Figure 5 shows the fingerprint similarity based on Euclidean distance. The Euclidean distance between AMR samples was 38,613-30,496,849. The mean values of FR and FD samples differed by 27,806,152; the mean values of FR and VD samples differed by 24,608,500; and the mean values of FR and HAD samples differed by 21,211,236. When Euclidean distance was larger, the distance between samples was father apart, and the similarity differences in fingerprint spectra became obvious. Thus, compared to VD and HAD, the difference between the FD and fresh AMR samples was more significant. The reason for this is the amount of volatile components that formed after FD, as shown in the C part of Figure 3. Some reactions produced or eliminated volatile components under high-vacuum, low-temperature conditions.

Conclusions
In this study, a total of 113 signal peaks from topographic plots were identified in the fresh and dried AMR samples treated by using different drying methods. The drying process changes the amounts of volatile components found in the fresh AMR. Different drying methods significantly affected the volatile components of AMR. A total of 28 new volatile components formed under the operation of FD, VD, and HAD, of which 14 new volatile components can only be formed in FD. Thus, the differences of volatile components between FD samples and fresh samples were much greater than those between fresh samples and VD and HAD samples. That is, FD is unsuitable for attaining dried AMR samples if the maximum amounts of volatile components in dried samples is required. In this study, the volatile component variation under FD, VD, and HAD was discussed in detail, and the basic experimental data were supplied for different requirements based on the choice of volatile components desired.