Analysis of the Differences in Volatile Organic Compounds in Different Rice Varieties Based on GC-IMS Technology Combined with Multivariate Statistical Modelling

In order to investigate the flavour characteristics of aromatic, glutinous, and nonaromatic rice, gas chromatography–ion mobility spectrometry (GC-IMS) was used to analyse the differences in volatile organic compounds (VOCs) amongst different rice varieties. The results showed that 103 signal peaks were detected in these rice varieties, and 91 volatile flavour substances were identified. Amongst them, 28 aldehydes (28.89~31.17%), 24 alcohols (34.85~40.52%), 14 ketones (12.26~14.74%), 12 esters (2.30~4.15%), 5 acids (7.80~10.85%), 3 furans (0.30~0.68%), 3 terpenes (0.34~0.64%), and 2 species of ethers (0.80~1.78%) were detected. SIMCA14.1 was used to perform principal component analysis (PCA) and orthogonal partial least squares discriminant analysis, and some potential character markers (VIP > 1) were further screened out of the 91 flavour substances identified based on the variable important projections, including ethanol, 1-hexanol, hexanal, heptanal, nonanal, (E)-2-heptenal, octanal, trans-2-octenal, pentanal, acetone, 6-methyl-5-hepten-2-one, ethyl acetate, propyl acetate, acetic acid, and dimethyl sulphide. Based on the established fingerprint information, combined with principal component analysis and orthogonal partial least squares discriminant analysis, different rice varieties were also effectively classified, and the results of this study provide data references for the improvement in aromatic rice varieties.


Introduction
Rice (Oryza sativa L.) is amongst the most important foods for humans and is irreplaceable.With rapid social and economic development and the continuous improvement in people's lives, the quality [1] and edibility [2] of rice have received increasing attention in China and abroad.Aroma is an important quality indicator of rice and has a direct impact on consumer desires to buy rice.Aromatic rice contains unique volatile aroma substances and is rich in nutrients, such as amino acids, and trace elements and is favoured by consumers and the market [3,4].The study of rice aroma traits and their aroma components has long been an important direction of research on rice-quality research.To date, hundreds of volatile substances have been detected in rice [5], including aldehydes, ketones, esters, alcohols, heterocycles, and alkenes [6].The aroma of rice is generated via the comprehensive action of various aromatic volatile components.The differences in the types and proportions of volatile substances lead to differences in rice aroma [7,8].Therefore, the in-depth study of volatile components in different varieties of aromatic rice is of great significance for the improvement in aromatic rice varieties and the development of high-end, high-quality rice.To visually compare the differences in the volatile flavour substances of diff rice varieties, difference comparison spectra of the volatile substances of six differen varieties were obtained by subtracting the DG163 spectrum, as shown in Figure 2 dark-blue area indicates a low concentration of the substance, and the dark-red are dicates a high concentration of the substance.The results show that the aromatic g nous rice varieties DG2030 and DG2029 were the most abundant in volatile substa followed by the aromatic rice varieties DG163, DG1839 and DG1946, and the non matic rice variety DG1938 had the lowest concentration.Different rice volatiles better separated by the GC-IMS features, so there were relative differences in GC-IMS feature spectra of different rice varieties, suggesting that the content of vo flavouring substances varied from one rice to another (red dashed box area in Figu This may be related to the differences in factors, such as raw material varieties and ferent nutrient chemical compositions [19,20].To visually compare the differences in the volatile flavour substances of differ rice varieties, difference comparison spectra of the volatile substances of six different varieties were obtained by subtracting the DG163 spectrum, as shown in Figure 2. T dark-blue area indicates a low concentration of the substance, and the dark-red area dicates a high concentration of the substance.The results show that the aromatic gl nous rice varieties DG2030 and DG2029 were the most abundant in volatile substan followed by the aromatic rice varieties DG163, DG1839 and DG1946, and the nona matic rice variety DG1938 had the lowest concentration.Different rice volatiles w better separated by the GC-IMS features, so there were relative differences in GC-IMS feature spectra of different rice varieties, suggesting that the content of vola flavouring substances varied from one rice to another (red dashed box area in Figure This may be related to the differences in factors, such as raw material varieties and ferent nutrient chemical compositions [19,20].

Qualitative Analysis of GC-IMS Spectra of Different Rice Varieties
The retention time and migration time of volatile flavour compounds from differ varieties of rice were compared, and the n-ketone C4~C9 calibration solution was u as an external standard reference to calculate the retention indices of the volatile co pounds.The NIST built-in Library Search in GC-IMS was used to calculate the retent To visually compare the differences in the volatile flavour substances of different rice varieties, difference comparison spectra of the volatile substances of six different rice varieties were obtained by subtracting the DG163 spectrum, as shown in Figure 2. The darkblue area indicates a low concentration of the substance, and the dark-red area indicates a high concentration of the substance.The results show that the aromatic glutinous rice varieties DG2030 and DG2029 were the most abundant in volatile substances, followed by the aromatic rice varieties DG163, DG1839 and DG1946, and the nonaromatic rice variety DG1938 had the lowest concentration.Different rice volatiles were better separated by the GC-IMS features, so there were relative differences in the GC-IMS feature spectra of different rice varieties, suggesting that the content of volatile flavouring substances varied from one rice to another (red dashed box area in Figure 2).This may be related to the differences in factors, such as raw material varieties and different nutrient chemical compositions [19,20].

Qualitative Analysis of GC-IMS Spectra of Different Rice Varieties
The retention time and migration time of volatile flavour compounds from different varieties of rice were compared, and the n-ketone C4~C9 calibration solution was used as an external standard reference to calculate the retention indices of the volatile compounds.The NIST built-in Library Search in GC-IMS was used to calculate the retention index.
The database was matched with the IMS migration time database to qualitatively analyse the volatile substances [21] and perform quantitative analysis based on the signal peak intensity [22].Figure 3 shows the qualitative analysis chromatogram obtained for the variety DG163.The qualitative and quantitative analyses of the volatile compounds from the six rice varieties are shown in Table 1.A total of 91 volatile substances were identified amongst 103 signal peaks, including 28 aldehydes, 24 alcohols, 24 ketones, 14 ketones, 12 esters, 5 acids, 3 furans, 3 terpenes, 2 ethers, 1 other, and 11 unknown components.The volatile components are mainly compounds such as aldehydes, alcohols, esters, and ketones [6,23].Aldehydes include (E)-2-octenal, nonanal, octanal, heptanal, hexanal, and valeraldehyde, which are mainly derived from lipid oxidation [24] and generally have a pleasant, fruity, floral, and delicate aroma (for example, (E)-2-octenal generates a green aroma, nonanal generates a fatty aroma, and octanal generates a citrus aroma; these are the main contributors to rice aroma with a low threshold [25]).Alcohols include 1-octen-3-ol, 1-hexanol, 1-pentanol, etc., which generally originate from the oxidation of fat and have an aromatic, vegetal, rancid, and earthy odour and are an important component of rice aroma [26].Ketones include 2-heptanone, acetone, 1-penten-3-one, etc., which are produced by the oxidation of unsaturated fatty acids, the Maillard reaction, etc., and have pleasant odours, such as fresh, creamy, and fruity aromas, and the aroma threshold is relatively low [27].Esters include propyl acetate, ethyl acetate, and ethyl formate, which have fruity, sweet, and wine-like aromas and endow rice with a light fruity aroma.Esters generally do not make a significant contribution to the development of the overall aroma but rather play a role in accentuating the aroma [28].Furans include 2-pentylfuran and 2-ethylfuran, which exhibit beany, fruity, and earthy aromas, amongst others [29].Acids include acetic acid and butyric acid, which are generally produced via the hydrolysis of triglycerides and phospholipids or the oxidation of alcohols and aldehydes.Acetic acid is a low-level saturated acid with a strong pungent odour [30], which is undesirable.Other compounds, such as alkanes, alkenes, terpenes, and ethers, are common in all varieties and are generally considered not to contribute to rice aroma.
pounds from the six rice varieties are shown in Table 1.A total of 91 volatile were identified amongst 103 signal peaks, including 28 aldehydes, 24 alcoh tones, 14 ketones, 12 esters, 5 acids, 3 furans, 3 terpenes, 2 ethers, 1 other, a known components.The volatile components are mainly compounds such as alcohols, esters, and ketones [6,23].Aldehydes include (E)-2-octenal, nonan heptanal, hexanal, and valeraldehyde, which are mainly derived from lipid [24] and generally have a pleasant, fruity, floral, and delicate aroma (for (E)-2-octenal generates a green aroma, nonanal generates a fatty aroma, a generates a citrus aroma; these are the main contributors to rice aroma w threshold [25]).Alcohols include 1-octen-3-ol, 1-hexanol, 1-pentanol, etc., wh ally originate from the oxidation of fat and have an aromatic, vegetal, rancid, odour and are an important component of rice aroma [26].Ketones include 2-h acetone, 1-penten-3-one, etc., which are produced by the oxidation of unsatu acids, the Maillard reaction, etc., and have pleasant odours, such as fresh, cr fruity aromas, and the aroma threshold is relatively low [27].Esters include p tate, ethyl acetate, and ethyl formate, which have fruity, sweet, and wine-li and endow rice with a light fruity aroma.Esters generally do not make a contribution to the development of the overall aroma but rather play a role i ating the aroma [28].Furans include 2-pentylfuran and 2-ethylfuran, whi beany, fruity, and earthy aromas, amongst others [29].Acids include acetic ac tyric acid, which are generally produced via the hydrolysis of triglycerides pholipids or the oxidation of alcohols and aldehydes.Acetic acid is a low-leve acid with a strong pungent odour [30], which is undesirable.Other compoun alkanes, alkenes, terpenes, and ethers, are common in all varieties and are gen sidered not to contribute to rice aroma.

Comparison of GC-IMS Fingerprints of Different Rice Varieties
All peaks were selected to compare the fingerprints and differences in the flavour substances of the different rice varieties (Figure 4).Each row in the figure represents all signal peaks selected in one rice sample, and each column represents the signal peaks of the same volatile organic compound in different rice samples.Due to the higher concentration of some volatile compounds, they will exhibit different forms, such as monomers (Ms) and dimers (Ds), and corresponding migration peaks will appear.The data of all identified flavour substances were selected.The fingerprints were generated via the built-in plug-in of the instrument.GC-IMS fingerprints showed that the contents of major volatile components amongst the six improved rice lines were significantly different, with obvious characteristic peak areas.Relatively high contents of acetic acid, nonanal, and cyclohexanone are in the aromatic rice DG163 (region A).The aromatic rice DG1839 contains relatively high levels of 2-acetyl-1-pyrroline, 6-methyl-5-hepten-2-one, decanal, 1-penten-3-one, and dimethyl disulphide (region B).Relatively high levels of heptanal, isopropanol, 2-pentanone, β-pinene, ethyl isovalerate, and ethyl acetate are in the nonaromatic rice DG1938 (region C).The contents of acrolein, 3-methylbutanal, isobutanol, benzaldehyde, p-xylene, 2-heptanone, 1-butanol dimer, tert-butanol, and dimethyl sulphide in the aromatic rice DG1946 are relatively high (region D).In contrast, most of the volatile components are high in the aromatic glutinous rice varieties DG2029 and DG2030, with relatively high contents of 2-butylfuran, laurin, butyl butyrate, and butenoic acid ethyl ester in DG2029 (region E), and the contents of isobutyric acid, butyric acid, 1-penten-3-ol, 2-pentylfuran, 2,5-dimethylfuran, (E)-2hexenal, (E)-2-butene aldehyde, 2-methylpropanol-M, 2-methylpropanol-D are relatively high (region F).From a macroscopic point of view, the contents of the main volatile compounds are different amongst aromatic rice, aromatic glutinous rice, and nonaromatic rice.The relative contents of volatile components of the two aromatic glutinous rice varieties are all higher, followed by the three aromatic rice varieties, and the nonaromatic rice possesses the lowest content.To clearly present the differences in rice flavour substances between different varieties, the relative contents of flavour substances in the varieties were obtained by normalizing the peak volumes of each flavour substance on the fingerprints (Figure 5).The flavour substances of the different rice varieties contained alcohols, aldehydes, ketones, acids, and esters, in which the relative contents of alcohols, aldehydes and ketones were higher.The relative contents of aldehydes in DG1839, DG163, DG1946, DG1938, DG2030, and DG2029 were 30.38%,28.89%, 29.65%, 28.94%, 31.17%, and 31.04%,respectively; those of alcohols were 34.97%, 35.18%, 35.95%, 40.52%, 35.17%, and 34.85%, respectively; those of ketones were 14.23%, 14.17%, 13.28%, 12.26%, 14.74%, and 14.67%, respectively; those of acids were 10.52%, 10.85%, 9.62%, 10.52%, 8.08%, and 7.8%, respectively; those of esters were 2.83%, 2.82%, 3.55%, 2.3%, 3.87%, and 4.15%, respectively; those of ethers were 1.32%, 1.78%, 1.64%, 0.8%, 1.32%, and 1.23%, respectively; those of furans were 0.4%, 0.44%, 0.42%, 0.3%, 0.62%, and 0.68%, respectively; and those of terpenes were 0.35%, 0.34%, 0.64%, 0.46%, 0.45%, and 0.55%, respectively.Lin et al. [31] used the SPME/GC-MS technique to determine the volatile components of rice grains, which were mainly alcohols, aldehydes, ketones, esters, hydrocarbons, organic acids, and heterocyclic compounds, with hydrocarbons being the most abundant, followed by aldehydes and ketones.In contrast, studies by Sun [32], Bian [33], and Zhu et al. [34] used the GC-IMS technique to detect volatiles from different rice varieties and detected the highest number of aldehyde species followed by alcohols and ketones, which is consistent with the results of this study.The reason for these differences may be the variation in the materials and the detection techniques.Some researchers compared the results of GC-IMS and GC-MS analyses and found differences.The volatile components determined using the GC-IMS method were more informative in terms of characteristic peaks.It was hypothesised that the main reason for this was due to the different pretreatment methods.When volatile compounds are determined via the GC-MS method, water vapour distillation of the samples is required, whereas no pretreatment is needed for the samples determined using the HS-GC-IMS method.The direct determination of samples after crushing maximises the retention of volatile components in the samples and shows some advantages in the identification of characteristic components [35][36][37].
0.4%, 0.44%, 0.42%, 0.3%, 0.62%, and 0.68%, respectively; and those of terp 0.35%, 0.34%, 0.64%, 0.46%, 0.45%, and 0.55%, respectively.Lin et al. [31] SPME/GC-MS technique to determine the volatile components of rice grains, w mainly alcohols, aldehydes, ketones, esters, hydrocarbons, organic acids, and clic compounds, with hydrocarbons being the most abundant, followed by and ketones.In contrast, studies by Sun [32], Bian [33], and Zhu et al. [34] GC-IMS technique to detect volatiles from different rice varieties and detected est number of aldehyde species followed by alcohols and ketones, which is with the results of this study.The reason for these differences may be the varia materials and the detection techniques.Some researchers compared the GC-IMS and GC-MS analyses and found differences.The volatile compone mined using the GC-IMS method were more informative in terms of character It was hypothesised that the main reason for this was due to the different pre methods.When volatile compounds are determined via the GC-MS method, pour distillation of the samples is required, whereas no pretreatment is need samples determined using the HS-GC-IMS method.The direct determination o after crushing maximises the retention of volatile components in the samples a some advantages in the identification of characteristic components [35][36][37].

Principal Component and Orthogonal Least Squares Discriminant Analyses of the Components of Different Rice Varieties
Principal component analysis was performed on the relative contents substances in different rice varieties (Figure 6, left).The contribution rate of 25.9%, the contribution rate of PC2 was 46.3%, and the cumulative contributio 72.2%.As a result of PCA, the aromatic, nonaromatic, and aromatic glutinous ties were clustered; three aromatic rice varieties (DG163, DG1839, and DG1 clustered; two aromatic glutinous rice varieties (DG2029 and DG2030) were and a nonaromatic rice variety (DG1938) was clustered alone.

Principal Component and Orthogonal Least Squares Discriminant Analyses of the Volatile Components of Different Rice Varieties
Principal component analysis was performed on the relative contents of volatile substances in different rice varieties (Figure 6, left).The contribution rate of PC1 was 25.9%, the contribution rate of PC2 was 46.3%, and the cumulative contribution rate was 72.2%.As a result of PCA, the aromatic, nonaromatic, and aromatic glutinous rice varieties were clustered; three aromatic rice varieties (DG163, DG1839, and DG1946) were clustered; two aromatic glutinous rice varieties (DG2029 and DG2030) were clustered; and a nonaromatic rice variety (DG1938) was clustered alone.OPLS-DA was performed on the volatile components of the different rice samples (Figure 6, right).OPLS-DA could effectively differentiate aromatic rice, nonaromatic rice, and aromatic glutinous rice.Three aromatic rice varieties (DG163, DG1839, and DG164) were distributed in the fourth quadrant; two aromatic glutinous rice varieties (DG2029 and DG2030) were distributed in the second quadrant.A nonaromatic variety DG1938 was distributed in the first quadrant.The model is usually evaluated using the independent variable fit index R 2 X (cum), the dependent variable fit index R 2 Y (cum), and the model prediction index Q 2 (cum).R 2 X and R 2 Y denote the explanatory rate of the constructed model for the X and Y matrices, respectively, and Q 2 denotes the predictive power of the model.Q 2 and R 2 close to 1 imply a better model fit and higher than 0.4 indicates that the model is acceptable, while a Q 2 larger than 0.5 indicates that the model has good predictive power [38,39].The model R 2 X = 0.95, R 2 Y = 0.994, and Q 2 = 0.99, which are all close to 1, indicate that the model fits well, has good predictive ability, and is reliable for analysing signature compounds [40].In order to avoid the overfitting phenomenon, the OPLS-DA model was cross-validated (200 permutation fits).The regression slopes of R 2 and Q 2 were both >0, and the intercepts were <0.5 and −0.5, respectively, indicating that the model did not overfit (left in Figure 7), was stable, and had good predictive ability.Variable importance projection values greater than 1 (VIP value > 1) generally indicate key flavour substances.Amongst the 6 different rice varieties, there were 30 volatile substances with VIP values > 1 (including dimers of some substances), including acetic acid, ethanol, acetone,6, hexanal, heptanal, 1-hexanol, nonanal, 1-pentanol, heptanal, dimethyl sulphide, acrolein, propyl acetate, 1-butanol, (E)-2-heptenal, methacrolein, valeraldehyde, 3-methylbutyraldehyde, acetal, acetoin, octenal, trans-2-octenal, 2-methyl-1-propanol, 6-methyl-5-hepten-2-one, and ethyl acetate.In this study, the contents of each key flavour compound were significantly different in aromatic (DG163, DG1839, and DG1946), nonaromatic (DG1938), and aromatic glutinous rice (DG2029, DG2030) varieties (Table 1).OPLS-DA was performed on the volatile components of the different rice samples (Figure 6, right).OPLS-DA could effectively differentiate aromatic rice, nonaromatic rice, and aromatic glutinous rice.Three aromatic rice varieties (DG163, DG1839, and DG164) were distributed in the fourth quadrant; two aromatic glutinous rice varieties (DG2029 and DG2030) were distributed in the second quadrant.A nonaromatic variety DG1938 was distributed in the first quadrant.The model is usually evaluated using the independent variable fit index R 2 X (cum), the dependent variable fit index R 2 Y (cum), and the model prediction index Q 2 (cum).R 2 X and R 2 Y denote the explanatory rate of the constructed model for the X and Y matrices, respectively, and Q 2 denotes the predictive power of the model.Q 2 and R 2 close to 1 imply a better model fit and higher than 0.4 indicates that the model is acceptable, while a Q 2 larger than 0.5 indicates that the model has good predictive power [38,39].The model R 2 X = 0.95, R 2 Y = 0.994, and Q 2 = 0.99, which are all close to 1, indicate that the model fits well, has good predictive ability, and is reliable for analysing signature compounds [40].In order to avoid the overfitting phenomenon, the OPLS-DA model was cross-validated (200 permutation fits).The regression slopes of R 2 and Q 2 were both >0, and the intercepts were <0.5 and −0.5, respectively, indicating that the model did not overfit (left in Figure 7), was stable, and had good predictive ability.OPLS-DA was performed on the volatile components of the different rice samples (Figure 6, right).OPLS-DA could effectively differentiate aromatic rice, nonaromatic rice, and aromatic glutinous rice.Three aromatic rice varieties (DG163, DG1839, and DG164) were distributed in the fourth quadrant; two aromatic glutinous rice varieties (DG2029 and DG2030) were distributed in the second quadrant.A nonaromatic variety DG1938 was distributed in the first quadrant.The model is usually evaluated using the independent variable fit index R 2 X (cum), the dependent variable fit index R 2 Y (cum), and the model prediction index Q 2 (cum).R 2 X and R 2 Y denote the explanatory rate of the constructed model for the X and Y matrices, respectively, and Q 2 denotes the predictive power of the model.Q 2 and R 2 close to 1 imply a better model fit and higher than 0.4 indicates that the model is acceptable, while a Q 2 larger than 0.5 indicates that the model has good predictive power [38,39].The model R 2 X = 0.95, R 2 Y = 0.994, and Q 2 = 0.99, which are all close to 1, indicate that the model fits well, has good predictive ability, and is reliable for analysing signature compounds [40].In order to avoid the overfitting phenomenon, the OPLS-DA model was cross-validated (200 permutation fits).The regression slopes of R 2 and Q 2 were both >0, and the intercepts were <0.5 and −0.5, respectively, indicating that the model did not overfit (left in Figure 7), was stable, and had good predictive ability.Variable importance projection values greater than 1 (VIP value > 1) generally indicate key flavour substances.Amongst the 6 different rice varieties, there were 30 volatile substances with VIP values > 1 (including dimers of some substances), including acetic acid, ethanol, acetone,6, hexanal, heptanal, 1-hexanol, nonanal, 1-pentanol, heptanal, dimethyl sulphide, acrolein, propyl acetate, 1-butanol, (E)-2-heptenal, methacrolein, valeraldehyde, 3-methylbutyraldehyde, acetal, acetoin, octenal, trans-2-octenal, 2-methyl-1-propanol, 6methyl-5-hepten-2-one, and ethyl acetate.In this study, the contents of each key flavour compound were significantly different in aromatic (DG163, DG1839, and DG1946), nonaromatic (DG1938), and aromatic glutinous rice (DG2029, DG2030) varieties (Table 1).
Amongst the key flavour substances of aldehydes, the content of octanal (M + D) reached significantly different levels for aromatic glutinous rice, aromatic rice, and nonaromatic rice.The studies of Yang and Sarika et al. also showed that the octanal content of aromatic rice was significantly higher than that of nonaromatic rice [17,41], which is consistent with the results of this study; the hexanal content (M + D) in aromatic glutinous rice was significantly higher than that in aromatic and nonaromatic rice; hexanal is a lipid oxidation marker during rice ageing [42]; and it has been shown that amylose can react with hexanal to form a V-type crystal complex [43].However, aromatic glutinous rice has a very low content of amylose, which may be due to the fact that glutinous rice produces fewer crystal complexes, resulting in higher levels of detected hexanal.In terms of alcohols, the content of 1-hexanol (M + D) in aromatic glutinous rice was significantly higher than that in nonaromatic and aromatic rice.An analysis by Peng et al. [44] showed that nonaromatic rice had a higher content of 1-hexanol than glutinous rice, which is inconsistent with the present study and may be due to the difference in materials, as well as detection techniques.Amongst ketones, the contents of acetone and 6-methyl-5-hepten-2-one in aromatic rice and aromatic glutinous rice were significantly higher than those in nonaromatic rice.In addition, the characteristic rice aroma compound 2-acetyl-1-pyrroline was detected, and the 2-AP content in aromatic rice was significantly higher than in nonaromatic rice and aromatic glutinous rice.Most studies believe that the aromatic substance 2-AP is caused via the mutation of the gene encoding betaine dehydrogenase Badh2, the loss of betaine dehydrogenase activity, and the production of nonfunctional BADH2 protein, which results in the accumulation of 2-AP [45,46].It has also been reported that the 2-AP content of varieties with high amylose content is considered to be lower than that of varieties with low amylose content (glutinous rice) because 2-AP interacts with amylose to form a V-type complex [43], which may be a reason why the 2-AP content of the nonaromatic rice DG1938 and the aromatic glutinous rice DG2029 and DG2030 did not have any significant difference in the present study, but the appearance of the 2-AP feature profile of the nonaromatic rice DG1938 was much darker in fingerprints.
In the present study, a small amount of 2-AP was also detected in nonaromatic rice DG1938, which showed that both aromatic and nonaromatic rice produce the characteristic aromatic compound 2-AP, with the only difference being the concentration between them.However, since some nonaromatic rice contains higher levels of linoleic acid, linolenic acid, and lipid oxidase, more secondary oxidation products are formed, so this could be an important reason why both aromatic and nonaromatic rice, despite both containing 2-AP, smell less aromatic than the flavoured rice [8].Therefore, the PCA and OPLS-DA analyses based on GC-IMS data all showed significant differences in key flavour substances amongst aromatic rice, nonaromatic rice and aromatic glutinous rice.The results also showed that it was feasible to apply GC-IMS to detect and identify substances that produce flavours in rice.

Experimental Materials
As shown in Table 2, six rice varieties were used in the test, in which Diangu 163, Diangu 1839, and Diangu 1946 were aromatic rice; Diangu 1938 was nonaromatic rice; and Diangu 2029 and Diangu 2030 were aromatic glutinous rice with aroma traits and waxy properties.Further molecular identification of the aroma gene badh2 locus showed that (Figure 8), five lines with the aroma trait appeared with two bands of about 580 bp and 255 bp in size, which were consistent with the band pattern of DT502, and were recessive pure genotypes badh2/badh2, which were aroma rice; one line was without the aroma trait, which appeared with two bands of about 580 bp and 355 bp in size, which were inconsistent with the band pattern of DT502, and were dominant pure genotypes badh2/badh2, which were nonaromatic rice.All materials were provided by the Rice Research Institute of Yunnan Agricultural University.

Flavour-Gene-Specific Primer Analysis
Referring to Bradbury [45], specific primers were used to analyse the badh2 locus of the scent gene.The primer sequences were synthesised by Shanghai Sangong Bioengineering Technology Service Co Ltd. (Shanghai, China) (Table 3).The PCR reaction system was 15 μL, including 7.5 μL mix, 1 μL DNA, 4.5 μlddH2O, and 0.5 μL of each of the four primers.The reaction procedure was as follows: denaturation at 94 °C for 5 min, denaturation at 94 °C for 30 s, annealing at 54 °C for 30 s, and extension at 72 °C for 1 min, with a total of 29 cycles and extension at 72 °C for 10 min.

Instruments and Equipment
The FlavourSpec ® sensitive analyser (G.A.S, Rehden, Germany) and a quality control system (GC-IMS system equipped with a CTC automatic headspace sampler, Laboratory Analytical Viewer (LAV) analysis software, and Library Search qualitative software) were used.

Sample Preparation
The method of Song [47] was used to process samples.Different varieties of rice were collected, dried in the air, browned, and refined; 2.0 g of refined rice was weighed, incubated at 60 °C for 15 min, and placed in a 20 mL headspace flask; and each sample was assayed in parallel three times.

Flavour-Gene-Specific Primer Analysis
Referring to Bradbury [45], specific primers were used to analyse the badh2 locus of the scent gene.The primer sequences were synthesised by Shanghai Sangong Bioengineering Technology Service Co Ltd. (Shanghai, China) (Table 3).The PCR reaction system was 15 µL, including 7.5 µL mix, 1 µL DNA, 4.5 µlddH 2 O, and 0.5 µL of each of the four primers.The reaction procedure was as follows: denaturation at 94 • C for 5 min, denaturation at 94 • C for 30 s, annealing at 54 • C for 30 s, and extension at 72 • C for 1 min, with a total of 29 cycles and extension at 72 • C for 10 min.

Instruments and Equipment
The FlavourSpec ® sensitive analyser (G.A.S, Rehden, Germany) and a quality control system (GC-IMS system equipped with a CTC automatic headspace sampler, Laboratory Analytical Viewer (LAV) analysis software, and Library Search qualitative software) were used.

Sample Preparation
The method of Song [47] was used to process samples.Different varieties of rice were collected, dried in the air, browned, and refined; 2.0 g of refined rice was weighed, incubated at 60 • C for 15 min, and placed in a 20 mL headspace flask; and each sample was assayed in parallel three times.

Figure 2 .
Figure 2. GC-IMS two-dimensional spectra of volatile substances in different rice varieties (d ence comparison map).

Figure 2 .
Figure 2. GC-IMS two-dimensional spectra of volatile substances in different rice varieties (dif ence comparison map).

Figure 2 .
Figure 2. GC-IMS two-dimensional spectra of volatile substances in different rice varieties (difference comparison map).

Figure 3 .
Figure 3. Qualitative analysis spectrum of volatile substances in rice variety DG163 via

Figure 3 .
Figure 3. Qualitative analysis spectrum of volatile substances in rice variety DG163 via GC-IMS.

Figure 4 .
Figure 4. Fingerprints of volatile components in different rice varieties.Note: rows represent all signal peaks selected in the samples, and columns represent signal peaks of volatile components in different samples.

Figure 5 .
Figure 5. Relative contents of volatile substances in different rice varieties.

Figure 5 .
Figure 5. Relative contents of volatile substances in different rice varieties.

Figure 6 .
Figure 6.PCA diagram (left) and OPLS-DA model (right) of volatile substances of different rice varieties.

Figure 7 .
Figure 7. OPLS-DA model validation and VIP values of different rice varieties (Compounds highlighted in red are potential signature markers for VIP > 1).

Figure 6 .
Figure 6.PCA diagram (left) and OPLS-DA model (right) of volatile substances of different rice varieties.

Figure 6 .
Figure 6.PCA diagram (left) and OPLS-DA model (right) of volatile substances of different rice varieties.

Figure 7 .
Figure 7. OPLS-DA model validation and VIP values of different rice varieties (Compounds highlighted in red are potential signature markers for VIP > 1).

Table 1 .
Qualitative results of volatile substances in six different varieties of rice.
Note: D-dimer; M-monomer; different lowercase letters in the same row indicate significant differences (p < 0.05).The numbers in the compound represent unknown compounds.

Table 2 .
Aroma genotypes of aromatic rice Diantun502 and its improved lines.which were nonaromatic rice.All materials were provided by the Rice Research Institute of Yunnan Agricultural University.

Table 2 .
Aroma genotypes of aromatic rice Diantun502 and its improved lines.

Table 3 .
Special primers used for aromatic gene badh2 locus amplification in rice.

Table 3 .
Special primers used for aromatic gene badh2 locus amplification in rice.