Characterization of aroma-active compounds in Dongli by quantitative descriptive analysis, gas chromatography-triple quadrupole tandem mass spectrometry, and gas chromatography-olfactometry

Abstract

Dongli, or frozen pear, is a traditional Chinese snack with a unique flavor. This study identified the aroma-active volatile compounds (VOCs) in Dongli using quantitative descriptive analysis (QDA), gas chromatography-triple quadrupole tandem mass spectrometry (GC-MS/MS), and gas chromatography-olfactometry (GC-O). QDA indicated that Dongli of all cultivars presented increased sweet and wine aromas. A total of 21 VOCs were identified by GC-MS/MS. Bidirectional orthogonal partial least square (O2PLS) analysis, GC-O analysis, detection frequency analysis (DFA), and relative odor activity values (ROAV) showed that: estragole and anethole contributing “anise, green” aromas were the key aromatic VOCs of fresh pears, while ethyl butanoate, butyl acetate, heptyl acetate, benzaldehyde, and geranyl acetone contributing “sweet, fruity, green” aromas were the key aromatic VOCs of Dongli. The results revealed that the repeated freezing treatment promoted a unique aroma in pears. This study would contribute to developing new pear products.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s13197-022-05463-8.

Introduction

Pears, which are popular with consumers for their unique fragrance, subtle aroma, and pleasant sweetness (Li et al. 2014; Ma et al. 2012), are harvested in the middle of September with a wide range of cultivation in China, especially in the northeastern part of the country (Zhou et al. 2015). Since the Liao dynasty (A.D. 907–A.D. 1125), pears have been frozen in winter due to their limited shelf life and very large yield (Wang et al. 2015), so that people were able to consume pears conveniently in winter.

“Dongli”, or frozen pears were processed by repeated freezing and thawing, rather than simple constant temperature freezing. Currently, Dongli is popular with its peculiar and rich taste as a special local product in the northeastern areas of China despite its black skin after freezing. However, Dongli sold in the market at present are mostly produced by the growers through the traditional processing method, which is only to leave the surplus pears outside in winter in northern China, relying on the natural winter temperature. The quality and flavor of these Dongli products vary greatly by temperature, environment, cultivar, and other factors, which greatly limits its industrial production and promotion. Furthermore, although the price of Dongli has not changed much, the storage period is significantly extended, which could help to solve the problem of overcapacity of local pears and to increase the income of growers. In-depth research on the aroma of Dongli will provide an important reference for the quality evaluation and cultivar selection for the industrial production of Dongli.

Aroma is an important aspect that customers consider when they are choosing fruit (Nuzzi et al. 2008). The aroma of fruits is composed of a mixture of more than 1000 volatile compounds, including esters, aldehydes, terpenes, alcohols, ketones, and some sulfur compounds (Chervin et al. 2000; Lara et al. 2003; Takeoka et al. 1992). Zlatić et al. (2016) found that the change in ‘Bartlett’ pear flavor in storage conditions, and shelf-life was attributed to the increase in ester and aldehyde content. Although Dongli has strong potential economic value, it has not been widely commercialized. However, there are rare reports about the aroma of Dongli. Thus, it is important to investigate the volatile compounds of Dongli before and after freezing. This has been a hot topic for exploring the key aromatic compounds of processed fruits. Sensory characteristics are also an important factor affecting the preferences of consumers for fruit. Quantitative descriptive analysis (QDA) is a useful method to evaluate the sensory characteristics of food (Li et al. 2019). Generally, the contents and the odor threshold of the volatiles of fruits were closely related (Dai et al. 2019).

In this work, the correlation of aroma characteristics of fresh pears and Dongli that was explored by QDA and principal component analysis (PCA). The volatiles of 4 cultivars of pears (fresh and frozen) were identified by headspace solid-phase microextraction with gas chromatography-triple quadrupole tandem mass spectrometry (HS-SPME-GC-MS/MS). Multivariate analyses, including correlation analysis (CA) and partial least square discriminate analysis (PLS-DA), were used to compare the key volatile compounds (VOCs) of fresh pears and Dongli. Bidirectional orthogonal partial least square (O2PLS) was applied to analyze the correlation between aroma characteristics and the key VOCs of fresh pears and Dongli. Gas chromatography-olfactometry (GC-O), detection frequency analysis (DFA), and relative odor activity values (ROAV) were applied to identify the aroma activity of VOCs. This study aims to identify the common and unique aroma-active VOCs from the 4 cultivars of fresh pears and Dongli, which will provide an important reference for the quality evaluation and cultivar selection for the industrial production of Dongli. This study would contribute to developing new pear products.

Materials and methods

Materials and chemicals

AP (Pyrus ussuriensis Maxim., Anli pear), BP (Pyrus bretschneideri Rehd., Baili pear), JP (Pyrus ussuriensis Maxim., Jianba pear), and NP (Pyrus ussuriensis Maxim., Nanguo pear) were obtained in October 2017 from the same local orchard in Anshan, Liaoning Province, China. Fruits were harvested at the maturity of 90%, then were placed in foam boxes, and transported to the laboratory on the same day. Six well-preserved fruits of each cultivar were selected for experiments.

The internal standard as 3-nonanone was purchased from Aladdin Reagent Co., Ltd (Shanghai, China). C7-C40 saturated n-alkane mixture (1000 µg/mL, catalog 49,452-U, Sigma Aldrich Co., St. Louis, MO, USA).

Sample preparation

Fresh pears were prepared by storing them for 7 days at room temperature (25 ± 2 °C) after harvest. Dongli were prepared by storing the pears at − 18 °C for 7 days, thawing at room temperature for 4 h, and freezing at − 18 °C for 2 days (Dongli was processed using an improved traditional method based on numbers of preliminary experiments). The traditional processing method of Dongli is to place the pears outside at around − 20 °C in Northeast China for 7 days in winter and the appearances of pears change with the outside temperature. As shown in Fig. S1, Dongli was thawed at room temperature just before the SPME operation. The pears were peeled and diced as one sample. Three pears from each treatment were analyzed by GC-MS/MS from extraction to determination.

Aroma characteristics by QDA

The comparison of aroma characteristics of fresh pears and Dongli that was performed by QDA (He and Chung 2018). The evaluation team consisted of 15 trained sensory evaluators who evaluated the pears with the juice of the samples. The six attributes that best expressed the aroma characteristics of pears were listed as sweet, sour, fruit, floral, grass, and wine (Zhu and Xiao 2019). The aroma intensities were assigned scores of 0–5, with higher scores indicating a stronger odor for that particular aroma attribute. After removing the highest and lowest scores, the average scores were applied for radar and PCA analysis in Fig. 1.

Fig. 1.

A–D Radar charts of the aroma characteristics of AP versus FAP, BP versus FBP, JP versus FJP and NP versus FNP by QDA. E Score scatter plot and F biplot chart of PC1 versus PC2 scores based on QDA data by PCA

Identification of VOCs using HS-SPME-GC-MS/MS

An aliquot of 2.0 g of pear sample with 20 uL 3-nonanone internal standard was placed in a 20 mL head-space bottle for SPME at 50 °C for 1 h in a water bath (Chen et al. 2020). The SPME fiber (Supelco, Bellefonte, PA, USA) was equipped with a 75 μm CAR/PDMS (carboxen-polydimethylsiloxane) fiber.

The VOCs of the pear samples were analyzed using a TSQ 8000 Evo GC-MS/MS (Thermo Fisher Scientific, Palo Alto, CA, USA) equipped with a non-polar TG-5MS column (30 m × 0.25 mm i.d., 0.25 μm film thickness; Code 26,098–1640, Thermo Fisher Scientific). The samples were injected in splitless mode. The SPME fiber was parsed out at 250 °C for 5 min and then purged at a flow rate of 2 mL/min with helium. The heating program started at 40 °C, heated to 80 °C at a rate of 3 °C/min, maintained at 80 °C for 1 min, heated to 150 °C at a rate of 5 °C/min, and then heated to 280 °C at a rate of 10 °C/min. MS/MS was performed with an EI source and an ionization voltage of 70 eV. The matched ion fragments were compared with standard ion fragments in NIST 2008.

Identification and relative quantification of VOCs

The retention indices (RIs) of VOCs were determined with 1 μL of C7-C40 saturated n-alkane mixture as described previously (Vandendool and Kratz 1963). The VOCs with a similarity index greater than 800 were identified. The internal standard (3-nonanone) was used to detect the stability of the instrument and the extraction effectiveness of VOCs. Qualitative identification was further performed by matching the RIs with the NIST RI library. Relative quantification of the VOCs was determined based on peak area percentage as the following equation (Yang et al. 2015):

$$\text{C}=\frac{{\text{S}}_{\text{peak}}}{{\text{S}}_{\text{total peak}}}\times 100\%$$

where C represents the relative content of the compound; S_peak represents the peak area of the compound; S_{total peak} represents the total peak area of all the compounds.

Gas chromatography-olfactometry (GC-O) analysis and detection frequency analysis (DFA)

The GC-O analysis was performed on an Agilent 7890 gas chromatography (Agilent Technologies, Inc., Santa Clara, CA, U.S.A.) coupled with a Gerstel olfactory detection port (ODP3, Gerstel, Inc., Linthicum, MD, U.S.A.). The SPME extract was separated on a TG-5MS column (30 m × 0.25 mm i.d., 0.25 μm film thickness; Code 26,098–1640, Thermo Fisher Scientific) with the same GC parameters as previously described in “Identification of VOCs using HS-SPME-GC-MS/MS” . Nine trained (ISO 8586-2:2008) panelists (5 females and 4 males, aged from 20 to 40) were selected for aroma-active compounds characterization. Twice repeated detection frequency analysis (DFA) was applied by the method of (Zhang et al. 2019). The retention time and odor description were recorded. The VOCs in which detection frequency (DF) ≥ 7 (reported by at least 7 panelists) could be considered as potential aroma-active compounds.

Relative odor activity values (ROAV)

ROAV was applied to assess the potential aroma-activity of the VOCs (Zhuang et al. 2016). The calculation formula was as follows:

$${\text{ROAV}}_{\text{i}}=\frac{{\text{C}}_{\text{i}}\%}{{\text{C}}_{\text{s}}\%}\times \frac{{\text{T}}_{\text{s}}}{{\text{T}}_{\text{i}}}\times 100$$

where C_s% and T_s represent the highest relative contents in the sample and the odor threshold in water, respectively; C_i% is the VOC relative contents; T_i is the VOC odor threshold in water. The VOCs in which ROAV_i ≥ 1 are considered as the potential aroma contributors for samples.

Statistical analysis

All measurements were performed three times (The samples were extracted 3 times for one determination using SPME-GC-MS/MS). Before the multivariate analysis (PCA, CA, and PLS-DA), data normalization was performed using the normalization sum. Then, log transformation (base 10) was performed using the method of Dieterle et al. (2006). Furthermore, auto-scaling was applied to compare the compounds based on correlations, which was mean-centered and divided by the standard deviation of each variable. Multivariate analyses were performed with PCA, CA, and PLS-DA using MetaboAnalyst 4.0 (Chong et al. 2019). CA was used for the cluster heatmap. PCA and CA were applied to visualize the multivariate data based on VOCs, and PLS-DA was applied to find the key contributor of the aroma of fresh pears and Dongli. Bidirectional orthogonal partial least square (O2PLS) was applied to analyze the correlation between aroma characteristics and the aroma-active compounds of fresh pears and Dongli using SIMCA 15.1 (Umetrics, Umea, Sweden).

Results and discussion

Assessment of the aroma characteristics of Dongli

Radar chart of QDA

AP, BP, JP, and NP are the main cultivars of Dongli produced in Northeast China. As shown in Fig. S1, the skin color of these 4 cultivars changed from different variations of yellow to brown or even black after freezing. Dongli was processed through repeated freezing and thawing, not just freezing. The change of temperature was also an important factor in causing browning, and freezing alone cannot make the pear skin brown so severely. The browning of the skin of Dongli was mainly related to enzymatic browning and non-enzymatic browning (Du et al. 2012). The polyphenol oxidase in the pulp can promote the reaction of oxygen and reduce substances to cause browning (López-Nicolás et al. 2007). The repeated freezing and thawing process changed the original cell structure of the pear so that more oxygen was in contact with the pulp, which intensified the Maillard reaction (Damasceno et al. 2008).

To assess the aroma characteristics of Dongli, the responses of different aroma attributes for the pears were displayed by radar charts in Fig. 2A–D. All fresh pears showed different aroma characteristics. AP smelt sour with some grass and fruit aromas. The sweet aroma of BP was more prominent. JP had softer and more balanced aroma attributes. NP was full of floral aroma, which was stronger than the other aromas. However, Dongli of all cultivars presented increased sweet and wine aromas, and no changes were observed in the sour aroma. The changes in characteristics may indicate that freezing treatment can promote the formation of a unique flavor of Dongli.

Fig. 2.

Clustering heatmap of VOCs from the pear samples

PCA model of QDA data

Multivariate statistical analysis methods were useful to exclude objective factors of the data. The QDA data of 4 cultivars of fresh pears and Dongli were classified by PCA (Fig. 1E, F). The first two principal components had a 78.3% cumulative variance in the PCA model, which can explain the integral aroma characteristics properly. The first principal component (PC1) and the second principal component (PC2) provided 46.9% and 31.4% of the variances, respectively. The score plot (Fig. 1E) indicated that all pear samples were clustered into two clusters: fresh pears and Dongli. It was no table that the difference in aroma characteristics between pear cultivars was less obvious than that between the fresh pears and Dongli by observing intuitively. The result was similar to the characterization of QDA data in that the freezing process had a strong correlation with the formation of pear aroma. A biplot chart (Fig. 1F) was used to merge the score and loading plots, the location of which directly illuminated the correlation between aroma attributes and samples. From the biplot chart, it can be seen that the aroma attributes of Dongli were closely related to sweet and wine aromas. Moreover, the aroma attributes of fresh pears were closely related to floral and grass aromas. The results indicated that freezing treatment may produce special flavor substances in pears.

Effects of freezing treatment on the VOCs

A total of 21 VOCs were identified in these samples using GC-MS/MS. As shown in Table 1, the 21 VOCs were divided into 7 categories based on functional groups, including 10 esters, 2 aldehydes, 4 hydrocarbons, 1 alcohol, 2 ethers, 1 ketone, and 1 other compound. Overall, esters and aldehydes were the major VOCs in all pear samples. Consistently, Qin et al. (2012) also found that esters and aldehydes were equally dominant in 33 Chinese pear cultivars in Xingcheng, Liaoning province. Additionally, more numbers of esters were identified in Occidental pear (Pyrus communis L.) than in AP, BP, JP, and NP (Chen et al. 2018). Although a smaller number of esters were identified in these 4 Chinese pear cultivars in this study, hexyl acetate or ethyl hexanoate has a higher relative content, especially in NP. Esters are the key aroma compounds in pears (Wang et al. 2019). Therefore, changes in esters’ content may lead to changes in the overall aroma of pears. Compared to fresh pears stored at room temperature, all 4 cultivars of Dongli had elevated ester contents. A previous study showed that oxygen content and temperature variation were two important factors affecting the variation of pear aroma (Zlatić et al. 2016). The levels of butyl acetate and hexyl acetate in pears increased, which was stored under an ultra-low oxygen atmosphere and lower storage temperature. Zhou et al. (2015) found that low-temperature conditioning can increase the expression and activity of alcohol acyltransferase (AAT), lipoxygenase (LOX), alcohol dehydrogenase (ADH), and accumulated polyunsaturated fatty acids in NP, which prevented the loss of aroma-related esters. However, in this study, ethyl hexanoate disappeared from FNP, and more ethyl butanoate, butyl acetate, etc. were subsequently produced. The repeated freeze-thaw treatment could change the breakdown pattern of fatty acid in Dongli and produced a wider variety of esters (Zhou et al. 2015), which may lead to the unique aroma of Dongli.

Table 1.

Relative percentage (%) of VOCs in 4 cultivars of fresh pears and Dongli using SPME–GC–MS/MS (n = 3)

No	RIa	Volatile compound	IDb	CAS	Fresh pear (%)		Dongli (%)
					AP	BP	JP	NP	FAP	FBP	FJP	FNP
Esters					14.5	53.1	66.0	88.9	39.0	61.4	67.0	99.1
1	609	Ethyl acetate	MS,RI,O	141-78-6	–d	–	19.9 ± e 1.0	0.9 ± 0.5	8.7 ± 1.0	40.8 ± 3.6	35.9 ± 1.5	4.8 ± 0.6
2	783	Ethyl butanoate	MS,RI,O	105-54-4	–	–	–	20.6 ± 1.7	–	–	–	38.4 ± 1.5
3	804	Butyl acetate	MS,RI,O	123-86-4	–	–	–	–	16.6 ± 1.9	19.9 ± 1.1	–	52.8 ± 2.5
4	941	Methyl hexanoate	MS,RI,O	106-70-7	–	–	–	0.8 ± 0.0	–	–	–	–
5	961	Ethyl 2-methyl-2-butenoate	MS,RI	55,514–48-2	–	–	2.1 ± 1.5	–	–	–	1.2 ± 0.2	–
6	1047	Ethyl hexanoate	MS,RI,O	123-66-0	–	–	–	64.5 ± 2.2	–	–	–	–
7	1057	Hexyl acetate	MS,RI,O	142-92-7	14.5 ± 0.2	53.1 ± 2.5	43.4 ± 2.1	–	8.8 ± 0.7	–	27.5 ± 1.3	–
8	1097	Ethyl 2-hexenoate	MS,RI,O	1552-67-6	–	–	0.6 ± 0.1	–	3.1 ± 0.1	–	1.0 ± 0.4	1.0 ± 0.0
9	1172	Heptyl acetate	MS,RI,O	112-06-1	–	–	–	0.5 ± 0.0	1.8 ± 0.0	0.7 ± 0.1	1.4 ± 0.1	–
10	1225	Ethyl octanoate	MS,RI,O	106-32-1	–	–	–	1.6 ± 0.2	–	–	–	2.1 ± 0.1
Aldehydes					22.4	4.4	24.9	0.0	29.7	10.3	21.9	0.0
11	780	Hexanal	MS,RI,O	66-25-1	22.4 ± 1.2	4.4 ± 0.5	24.9 ± 1.5	–	–	–	21.9 ± 1.0	–
12	987	Benzaldehyde	MS,RI,O	100-52-7	–	–	–	–	29.7 ± 1.2	10.3 ± 0.4	–	–
Hydrocarbons					8.0	7.0	5.0	5.1	30.7	18.5	1.3	0.9
13	1102	Undecane	MS,RI	1120-21-4	8.0 ± 0.3	7.0 ± 0.4	5.0 ± 0.8	5.1 ± 0.3	–	–	–	–
14	1475	Cedrene	MS,RI,O	11,028–42-5	–	–	–	–	–	1.2 ± 0.1	–	–
15	1527	Curcumene	MS,RI	644-30-4	–	–	–	–	1.8 ± 0.0	1.4 ± 0.1	–	–
16	1546	β-Bisabolene	MS,RI,O	495-61-4	–	–	–	–	28.9 ± 1.3	15.9 ± 1.5	1.3 ± 0.1	0.9 ± 0.1
Alcohol					0.0	0.0	0.0	2.7	0.0	9.8	0.0	0.0
17	875	(S)-3,4-Dimethylpentanol	MS,RI	NDc	–	–	–	2.7 ± 0.5	–	9.8 ± 0.4	–	–
Ethers					7.5	6.1	4.1	3.3	0.0	0.0	0.0	0.0
18	1270	Estragole	MS,RI,O	140-67-0	1.7 ± 0.3	1.3 ± 0.1	1.0 ± 0.0	0.8 ± 0.0	–	–	–	–
19	1315	Anethole	MS,RI,O	104-46-1	5.8 ± 0.6	4.8 ± 0.2	3.1 ± 0.1	2.6 ± 0.2	–	–	–	–
Ketone					0.0	0.0	0.0	0.0	0.6	0.0	0.0	0.0
20	1505	Geranyl acetone	MS,RI,O	3796-70-1	–	–	–	––	0.6 ± 0.0	–	–	–
Other					47.6	29.4	0.0	0.0	0.0	0.0	9.8	0.0
21	883	Hexyl hydroperoxide	MS	4312-76-9	47.6 ± 1.9	29.4 ± 1.6	–	–	–	–	9.8 ± 0.5	–

^aRI, retention index

^bMS, mass spectrum; RI, retention index; O, Odor descriptions of VOCs were obtained by matching with references

^cND, not found

^d‘–’, not detected

^eValues were described as the mean ± SD

As shown in Fig. 2, a heatmap was generated based on the 21 VOCs using clustering analysis. The obvious color difference in Fig. 2 illustrated that the VOCs in fresh and Dongli were significantly different. Thus, the freezing process will change the VOCs composition of pears. It was interesting that estragole (No. 18; all compound numbers refer to compounds listed in Table 1) and anethole (19) were ethers found in all fresh pear cultivars but none in Dongli. Estragole and anethole are the only two ethers identified that present grass or floral aromas (Atkinson 2016; Chen et al. 2020; Yauk et al. 2015).

CA, PCA, and PLS-DA of VOCs

CA and PCA of the VOCs in fresh pears and Dongli

The correlation between the samples based on their VOCs was analyzed by CA. As shown in Fig. 3A, it was indicated that the VOCs in fresh pears and Dongli were significantly different. The red parts of Fig. 3A showed a highly significant correlation between the groups. There was a positive correlation between 4 Dongli and a positive correlation between 4 fresh pears. It was interesting that this result was similar to the previous assessment of aroma characteristics by PCA (Fig. 3E). Besides, JP and FJP, NP and FNP as a single cultivar also presented correlations. Thus, more changes were produced (no correlation was found) in the VOCs of AP and BP after freezing treatment. The PCA score plot of (Fig. 3B) with the 21 VOCs revealed two clusters corresponding to fresh pears and Dongli under component 1 (41.0%) and component 2 (22.7%). The biplot chart (Fig. 3C) illustrated the contribution of the 21 VOCs to the classification of the samples. Undecane, estragole, and anethole were the important contributors to AP and BP. Ethyl 2-methyl-2-butenoate was an important contributor to FJP. (S)-3,4-Dimethylpentanol contributed greatly to FNP, and cedrene and butyl acetate contributed greatly to FBP. There were more contributors identified for FAP.

Fig. 3.

A Correlation heatmap between the samples based on their VOCs by CA. B Score scatter plot and C biplot chart of PC1 versus PC2 scores based on the relative contents of VOCs by PCA D Differences in the relative contents of VOC categories in fresh pears and Dongli

Furthermore, Fig. 3D depicted the differences in the categories of VOCs in fresh pears and Dongli. From AP to FAP, the freezing treatment produced more esters and hydrocarbons and lost other VOCs. Moreover, more ethers and others were detected in BP, and more alcohols were produced in FBP. According to the VOCs in JP, the VOC categories of FJP changed a little. Compared to NP, the esters increased obviously in FNP.

PLS-DA of the VOCs in fresh pears and Dongli of a single cultivar

PLS-DA can screen characteristic volatiles between samples (Huang et al. 2019). To compare the VOCs in fresh pears and Dongli of a single pear cultivar, PLS-DA was applied to analyze the VOC data (Fig. 4A). The total variance (R²_cum) of the PLS-DA models from Fig. 4A–D were 0.998, 0.100, 0.934, and 0.975, respectively. All the R²_cum were greater than 0.900, which has sufficient explanation (Fig. S3). Additionally, the PLS-DA models presented acceptance via cross-validation (CV) as shown in Fig. S3. Q² is an estimate of the predictive ability of the model, and is calculated via CV. In each CV, the predicted data are compared with the original data, and the sum of squared errors is calculated. The prediction error is then summed over all samples (Predicted Residual Sum of Squares or PRESS). For convenience, the PRESS is divided by the initial sum of squares and subtracted from 1 to resemble the scale of the R². Good predictions will have low PRESS or high Q². In these PLS-DA models, all the accuracy, R², and Q² were greater than 0.8. Therefore, the PLS-DA models were of good statistical quality with good acceptance. The important VOCs (VIP > 1) were selected by PLS-DA. As shown in Fig. 4A, hexyl hydroperoxide (21), undecane (13), hexanal (11), and anethole (19) were the characteristic VOCs of AP, and benzaldehyde (12), β-bisabolene (16), heptyl acetate (9), curcumene (15), ethyl 2-hexenoate (8), geranyl acetone (20), ethyl acetate (1) and butyl acetate (3) were the characteristic VOCs of FAP. Furthermore, hexyl acetate (7), hexyl hydroperoxide (21), anethole (19), and undecane (13) were the characteristic VOCs of BP, and ethyl acetate (1), benzaldehyde (12), butyl acetate (3), β-bisabolene (16), (S)-3,4-dimethylpentanol (17) and cedrene (14) increased obviously in FBP. The characteristic VOCs of AP were similar to those of BP, and the characteristic VOCs of FAP were similar to those of FBP. Furthermore, anethole (19), estragole (18), undecane (13) and hexyl acetate (7) were the characteristic VOCs of JP, and hexyl hydroperoxide (21), heptyl acetate (9), β-bisabolene (16) and ethyl acetate (1) increased obviously in FJP. Moreover, ethyl hexanoate (6), undecane (13), anethole (19), methyl hexanoate (4), estragole (18), heptyl acetate (9), and (S)-3,4-dimethylpentanol (17) were the characteristic VOCs of NP, and butyl acetate (3), ethyl 2-hexenoate (8), β-bisabolene (16) and ethyl butanoate (2) were the characteristic VOCs of FNP. Compared with the other cultivars, NP had more characteristic VOCs. However, the number of characteristic VOCs of FNP decreased after freezing treatment.

Fig. 4.

A PLS-DA of the VOCs (compound numbers refer to the compounds listed in Table 1). B Circle plot combining the results of the above PLS-DA models. The key VOCs (VIP > 1) were screened by PLS-DA. Each line represented an important VOC of the sample, and different colors represented different samples. The VOCs with more lines contained only in fresh pears or Dongli could be the key VOCs

As shown in Fig. 4B, the correlation between these important VOCs and samples was notable. The circle plot combined the results of the above PLS-DA data, in which each line represents an important VOC of the sample, and different colors represented different samples. In Fig. 4B, the VOCs with more lines contained only in fresh pears or Dongli could be the key VOCs. As result, methyl hexanoate (4), ethyl hexanoate (6), hexyl acetate (7), hexanal (11), undecane (13), estragole (18), and anethole (19) were the key VOCs of fresh pears. Ethyl acetate (1), ethyl butanoate (2), butyl acetate (3), ethyl 2-hexenoate (8), benzaldehyde (12), cedrene (14), curcumene (15), β-bisabolene (16), and geranyl acetone (20) were the key VOCs of Dongli.

Correlation between the VOCs and aroma characteristics

O2PLS was applied to analyze the correlation between the VOCs and aroma characteristics, and the correlation between the VOCs and samples. VOCs content and the aroma characteristics were used as X variables and Y variables in the O2PLS model, respectively. O2PLS model could reflect the difference between the two variable matrices as a whole to the sample group, and the two-variable matrices did not interfere with each other (Zheng et al. 2017). Generally, R² and Q² showed variance and predictability. The O2PLS model in which R² = 0.997 and Q² = 0.990 had excellent variance and predictability. The VIP values reflected the influence of VOCs on the formation of aroma characteristics. Compounds in which VIP_(pred) > 1 presented the key VOCs of Dongli as potential aroma contributors (Fig. 5A). Network model of the correlation between the VOCs and samples as shown in Fig. 5B and network model of the correlation between the VOCs and aroma characteristics as shown in Fig. 5C were performed by O2PLS modeling. Two variables connected with a red line represented a positive correlation, and two variables connected with a blue line represented a negative correlation. As shown in Fig. 5B, the aroma of fresh pears and Dongli was highly correlated with 13 and 14 VOCs, respectively. In particular, the aroma of Dongli was highly correlated with some esters and hydrocarbons. Besides, as shown in Fig. 5C, the aroma characteristics of fresh pears and Dongli were related to their VOCs.

Fig. 5.

A Changes in VIP_(pred) values of O2PLS model by relative contents of VOCs and QDA data. The VOCs in which VIP_(pred) > 1 were considered as the potential key VOCs of the pear samples. B Network model of the correlation between the VOCs and samples by O2PLS modeling. C Network model of the correlation between the VOCs and aroma characteristics by O2PLS modeling

Identification of the key aromatic VOCs of Dongli

The odor descriptions of VOCs were identified by GC-O. Besides, human perception of the aroma of food was closely related to the odor threshold of these aroma-active VOCs. Although odor activity values (OAV) were used to identify whether VOCs had aroma-activity or not, it needs too many standards and wastes time and money. Zhuang et al. (2016) found that ROAV is a simple and effective method to identify aromatic compounds compared to OAV. DFA and ROAV helped to compare the aroma-active compounds. There were 16 aromatic VOCs identified by DFA (Table S2) and 13 aromatic VOCs identified by ROAV (Table S3). Considering the potential contributors to the aroma of Dongli, compounds were defined as the key aromatic VOCs by the following conditions: (1) VIP_(pred) > 1; (2) DF ≥ 7; (3) identified as the key VOCs by PLS-DA; (4) had odor descriptions by GC-O. The result revealed that 7 compounds were identified as the key aromatic VOCs of fresh pears and Dongli as shown in Fig. S2-A, including estragole, anethole, ethyl butanoate, butyl acetate, heptyl acetate, benzaldehyde, and geranyl acetone. Furthermore, O2PLS analysis revealed that estragole and anethole were the key aromatic VOCs of fresh pears (Fig. S2-B). While, ethyl butanoate, butyl acetate, heptyl acetate, benzaldehyde, and geranyl acetone were the key aromatic VOCs of Dongli (Fig. S2-B). Fresh pears had more estragole and anethole contributing to “anise, green” aromas. Dongli had more ethyl butanoate, butyl acetate, heptyl acetate, benzaldehyde, and geranyl acetone contributing to “sweet, fruity, green” aromas. It revealed that the aroma of Dongli was characterized by higher sweet and wine aromas that came from the increase of these five VOCs. Thus, the freezing treatment produced a unique aroma in pears.

Estragole and anethole are isomeric ethers, which could be found in fresh fruits and vegetables, such as apple (Yauk et al. 2015), date palm fruit (Khalil et al. 2017), and Chinese yam (Chen et al. 2020). Estragole and anethole are important aromatic compounds, which can change aroma characteristics (Atkinson 2016). Moreover, ethyl butanoate, butyl acetate, and heptyl acetate are the common aromatic esters in fruits, such as apple (Fuhrmann and Grosch 2002) and banana (Selli et al. 2012). Besides, benzaldehyde and geranyl acetone are also important VOCs in tomato (Hayase et al. 1984) and green tea (Zhu et al. 2018), respectively. It is especially worth noting that only Dongli contained β-bisabolene, which is not an aromatic VOC. It found that β-bisabolene is a major compound in plant essential oils (Crockett et al. 2008; Phi et al. 2006).

Conclusion

This study was the first time to identify the aroma-active volatile compounds (VOCs) in Dongli using quantitative descriptive analysis (QDA), gas chromatography-triple quadrupole tandem mass spectrometry (GC-MS/MS), and gas chromatography-olfactometry (GC-O). The result showed that the effects of freezing treatment on the formation of aroma characteristics in pears were obvious. QDA and PCA showed that Dongli express higher sweet and wine aromas. A total of 21 VOCs were identified in pear samples by SPME-GC-MS/MS. The VOCs in fresh pears and Dongli showed significant differences. There were 13 aromatic VOCs identified by ROAVs, and 16 aromatic VOCs identified by DFA. O2PLS analysis, GC-O analysis, DFA, and ROAV showed that estragole and anethole contributing to “anise, green” aromas were the key aromatic VOCs of fresh pears, while ethyl butanoate, butyl acetate, heptyl acetate, benzaldehyde, and geranyl acetone contributing to “sweet, fruity, green” aromas were the key aromatic VOCs of Dongli. The repeated freezing treatment produced a unique aroma in pears. This study would contribute to developing new pear products.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

JNC, HTH and CJL: conceived and supervised the whole study; QG, XWW and JWZ: conducted of the experiment; JNC and HTH: conceived the study, performed the data analyses and contributed to the drafting of the manuscript; MT and YLX: gave experimental advice, contributed to the drafting of the manuscript, and edited and revised the final manuscript. All the authors read and approved the final manuscript.

Funding

This work was supported by the Liaoning Revitalization Talents Program (XLYC1807270) and the Open Research Fund of the Key Laboratory of Storage of Agricultural Products, Ministry of Agriculture and Rural Affairs (kf2019001).

Data availability

The data that support the findings of this study are available on request from the corresponding author.

Code availability

Not applicable.

Declarations

Conflict of interest

The authors have no conflict of interest to disclose.

Ethical approval

Ethics approval was not required for this research.