Jurnal Inggris

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

LWT - Food Science and Technology 124 (2020) 109203

Contents lists available at ScienceDirect

LWT - Food Science and Technology


journal homepage: www.elsevier.com/locate/lwt

Characterization of volatile and sensory profiles of apple juices to trace fruit T


origins and investigation of the relationship between the aroma properties
and volatile constituents
Jing Guo∗, Tianli Yue, Yahong Yuan, Nan Sun, Peng Liu
College of Food Science and Engineering, Northwest A&F University, No. 28, Xinong Road, Yangling, Shaanxi, China

A R T I C LE I N FO A B S T R A C T

Keywords: To classify apple juices based on their volatile composition and aroma properties, and to determine the corre-
HS-SPME-GC/MS lation between the volatile compounds and odor attributes, the aroma profiles of 31 apple juice samples from
Sensory analysis four varieties and four PDO regions (China) were determined. Thirty-nine volatile compounds were identified by
Multivariate statistical techniques headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS). Twenty-
Partial least squares regression (PLSR)
three odor attributes belonging to five categories (fruity, vegetable, spices, floral and other) were detected via
Malus domestica borkh
descriptive sensory analysis. Principal component analysis (PCA) and stepwise linear discriminant analysis
(SLDA) were separately used on the volatile concentrations and geometric means of the aroma descriptors to
Chemical compounds studied in this article:
Hexyl acetate (PubChem CID: 8908) establish classification models. PCA and SLDA allowed the discrimination of the juice samples from different
Butyl acetate (PubChem CID: 31272) apple varieties according to the GC and sensory datasets. For the samples from different geographical origins,
Ethyl butanoate (PubChem CID: 7762) SLDA provided a 93.5% prediction accuracy based on both datasets. Both the volatile composition and aroma
2-Methylbutyl acetate (PubChem CID: 12209) characteristics could serve as effective indices to determine the variety and geographic origin. Moreover, partial
1-Butanol (PubChem CID: 263) least squares regression (PLSR) yielded satisfactory models to predict seven aroma descriptors in apple juices
1-Hexanol (PubChem CID: 8103) (melon, pineapple, banana, apple, strawberry, cut green and hay) and clarified the effects of volatile components
Hexanal PubChem CID: 6184) on the formation of these aroma sensations.
(E)-2-hexenal (PubChem CID: 5281168)

1. Introduction and sensory characteristics. The effect of a single volatile constituent on


the juice aroma has been generally evaluated by the odor activity value
Due to its nutritional characteristics and unique flavor properties, (OAV). The OAV is the ratio between the concentration of the volatile
apple (Malus domestica Borkh) juice is a widely consumed fruit juice component and its perception threshold (Komthong, Katoh, Igura, &
worldwide. The aroma is a crucial standard when estimating the quality Shimoda, 2006). Moreover, gas chromatography-olfactometry (GC-O)
and characteristics of apple juice. Apple juice contains dozens of vola- has been employed to recognize the most effective volatile components
tile components belonging to different groups, such as esters, ketones, that contribute to the aromas of apple products (Antón, Suárez, García,
fatty acids, alcohols, aldehydes and terpenoids (Braga et al., 2013; & Picinelli, 2014; Mehinagic, Royer, Symoneaux, Jourjon, & Prost,
Nikfardjam & Maier, 2011). Some of them have been identified as 2006). Nevertheless, the perceived aroma of apple juice is the result of a
playing a major role in the formation of apple juice aroma, including mixture of volatile components. There might be synergistic effects
(E)-2-hexenol, (E)-2-hexenal, ethyl butanoate, butyl acetate, hexanal, 1- among volatile compounds, including suppression or masking interac-
hexanol and 1-butanol (Elss, Preston, Appel, Heckel, & Schreier, 2006; tions when their concentrations are above the threshold, or super-
Nikfardjam et al., 2011). Several factors, including the growth en- imposing effects when their concentrations are below the threshold.
vironment, apple variety, maturity level and processing procedure have Then the sensory contribution of an individual volatile component
been considered to affect the aroma of apple juice (Braga et al., 2013; identified by the OAV or GC-O has little correlation with its aroma
Dixon & Hewett, 2000; Elss et al., 2006). contribution in such a complicated system as apple juice. Therefore, it is
To fully understand the aroma properties of apple juice, it is vitally of great significance to study the sensory characteristics of apple juice
important to acquire information concerning its volatile constituents and to explore the relationship between the sensory characteristics and


Corresponding author.
E-mail addresses: fguojing0203@nwsuaf.edu.cn (J. Guo), yuetl305@nwsuaf.edu.cn (T. Yue), yuan324@msn.com (Y. Yuan), S15806100601@163.com (N. Sun),
18292755106@163.com (P. Liu).

https://doi.org/10.1016/j.lwt.2020.109203
Received 1 September 2019; Received in revised form 13 February 2020; Accepted 23 February 2020
Available online 26 February 2020
0023-6438/ © 2020 Elsevier Ltd. All rights reserved.
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

volatile profiles. However, there are no comprehensive data regarding aroma. During enzymatic browning, the volatile release of acetate es-
the sensory characteristics of apple juice, and the relationships between ters increases, while (E)-2-hexenal decreases (Komthong et al., 2006).
aroma descriptors and volatile compounds are not known. Each time, 20 fruits were washed with distilled water to remove surface
China is the world's largest producer of apples, accounting for more dirt, and then were cored and squeezed with a JP351 Midea juicer
than half of the total global production in recent years. A large pro- (Shanghai Midea Group Co., Ltd, China). A portion of 100 mL of fresh
portion of apples in China are processed into juice and sold in the in- juice was filtered by a 0.15 mm cloth filter and was used for the de-
ternational market. Due to the favorable soil and climate environment, termination of the titratable acidity and soluble solids content
apples grown in certain areas in China have been awarded the title of (Table 1). To measure the titratable acidity, 10 mL of fresh juice was
‘‘Protected Designation of Origin” (PDO) products. The apple juices titrated with NaOH (0.1 mol/L) to a pH of 8.1 (phenolphthalein was
from these PDO districts gain additional value and are more likely to be used as an indicator). The soluble solids content was measured by a
adulterated and counterfeited. Therefore, finding reliable marker PAL-1 hand-held Atago digital refractometer (Tokyo, Japan). A portion
compounds or indices to determine the geographical origin and apple of 200 mL of fresh apple juice was promptly poured into a beaker. To
variety of the juices is highly desirable for pricing and guaranteeing inhibit enzymatic browning, 4 g of NaCl and 0.2 g of ascorbic acid were
quality. added (Komthong, Igura, & Shimoda, 2007). Excessive foam and pulp
Multivariate statistical methods have been regarded as an effective were removed by a 0.15 mm cloth filter. The whole process of juice
tool to recognize food adulteration. Based on certain marker com- preparation was operated at low temperature (5 °C). Finally, the tri-
pounds or indices, such as volatile compounds, polyphenols and 1H plicates were pooled and stored at −18 °C for approximately one month
NMR spectra, some authors have employed chemometric methods to for GC-MS and sensory analysis. Prior to the analysis, the juice sample
classify apple juices by the apple variety and/or geographical origin was defrosted overnight in a refrigerator at 4 °C.
(Belton et al., 1998; Gan, Soukoulis, & Fisk, 2014; Guo, Yue, & Yuan,
2012; Guo, Yue, Yuan, & Wang, 2013; Reid, Woodcock, O'Donnell, 2.3. SPME/GC-MS analysis
Kelly, & Downey, 2005). To our knowledge, no study has been carried
out to discriminate apple juices using the sensory profile as character- Headspace solid-phase microextraction (HS-SPME) in combination
istic indices. Moreover, as mentioned above, the relationships between with gas chromatography-mass spectrometry (GC-MS) was employed to
the aroma characteristics and volatile compounds in apple juice are not analyze the volatileconstituents of apple juice samples (Jing et al.,
known. 2012). The juice sample (8 mL) was placed in a 15 mL headspace bottle.
Therefore, the detection of volatile compounds and sensory analysis NaCl (2.8 g) and internal standard (20 μL, 41 mg/L 2-octanol ethanol
were carried out in apple juice samples from PDO regions in China. The solution) were added to the headspace bottle. NaCl was added to pro-
aims of the present study were as follows: (1) discriminate the apple mote the release of volatile compounds due to the salting out effect
juices by variety and geographic origin and (2) identify the correlation (Jonker & Muijs, 2010). The juice sample was equilibrated for 20 min at
between the volatile compounds and aroma characteristics using partial 50 °C in a water bath at 800 rpm. Then, a 50/30 μm divinylbenzene/
least squares regression (PLSR). Apples used in this study were four of carboxen/polydimethylsiloxane coated fiber (DBV/CAR/PDMS, Su-
the most widely planted varieties from four areas within the Chinese pelco, Bellefonte, PA, USA) was exposed to the headspace over the juice
PDO districts. sample for 45 min at 50 °C. The fiber was desorbed at 250 °C in the
injection port of the GC instrument for 5 min. Analyses were carried out
2. Materials and methods in triplicate.
The aroma components were analyzed using a Thermo Finnigan
2.1. Reagents and standards Trace GC that was interfaced with a Trace DSQ mass spectrometer
(Thermo Elecron, San Jose, Calif., USA). Samples were run on a
Ascorbic acid, NaCl and NaOH were purchased from the Yangling 30 m × 0.25 mm × 0.25 μm HP-INNOWax capillary column. Helium,
Chemical Reagents Company (Xi'an, China) and were of analytical the carrier gas, was circulated at 0.8 mL/min at a constant flow rate.
grade. The pure reference standards were from Sigma-Aldrich Ltd. The GC oven temperature program was as follows: 40 °C for 2.5 min,
(Beijing, China). The standards were methyl butanoate, ethyl acetate, 5 °C/min ramp to 220 °C, 10 °C/min ramp to 240 °C and holding for
propyl acetate, isobutyl acetate, ethyl-2-methyl butanoate, propyl 1.5 min. The transfer line temperature was 280 °C. The mass was car-
propionate, 2-methylpropyl butanoate, 2-methylbutyl acetate, hexyl ried out in an electron-impact ionization mode, the ionizing voltage was
acetate, butyl acetate, (E)-2-hexenyl acetate, butyl butanoate, hexyl 70 eV, and the source temperature was 230 °C. The mass was ranged
butanoate, ethyl hexanoate, 2-methyl-1-propanol, 1-butanol, 1-hex- from 35 to 400 (m/z).
anol, 2-octanol, 6-methyl-5-hepten-2-ol, (Z)-2-hexen-1-ol, 2-ethyl-1- The volatile constituents were identified by comparing their mass
hexanol, hexanal, (E)-2-hexenal, benzaldehyde, α-farnesene, linalool, spectra with those from standards and with the data stored in the NIST
6-methyl-5-hepten-2-one and p-allylanisole. 11.L library, as well as by comparing the linear retention index (LRI).
The internal standard method was employed to quantify the identified
2.2. Apple juice samples compounds. The concentrations of the constituents were calculated by
interpolating the relative areas vs. the internal standard (2-octanol)
A total of 31 apple juice samples of four varieties (Starkrimson, area in the calibration graphs of the standards. GC peak areas were
Gala, Jonagold and Fuji) from four PDO regions of China (Luochuan calculated according to total ion current chromatograms. Volatile
and Xunyi counties in Shaanxi province, Xixia city in Shandong pro- constituents with no pure reference utilized the calibration graphs of a
vince and Lingbao city in Henan province) were considered in this compound having a similar chemical structure. The linear ranges, de-
study (Table 1). These apple varieties are the most widely planted termination coefficients (R2), limits of determination (LODs) and limits
varieties in China. The apples were hand-picked at commercial ma- of quantification (LOQs) of the method are listed in Supplementary
turity in 2017. The starch-iodine indices of the apples that were rated Table 1.
according to the Cornell generic starch scale 1–8 (Blanpied & Silsby,
1992) ranged from 4 to 6 (Alberti et al., 2016) (Table 1). For each juice 2.4. Sensory analysis
sample, a total of 60 apple fruits were used for juicing. To reduce the
exposure time of the juice sample in the air to decrease enzymatic The sensory evaluation was executed according to the QDA meth-
oxidation, 60 apple fruits were divided into three parts and juiced se- odology (Lawless & Heymann, 1998) by a trained panel consisting of
parately. Enzymatic browning induces major changes in the apple juice eleven experienced judges (5 males and 6 females). Two training

2
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

Table 1
Apples used in the study and the basic compositional parameters of the apple juices.
Sample Variety Geographical origin Starch index ± SD (n = 6) Soluble solids (%) ± SD Titratable aciditya ± SD (n = 3)
(n = 3)

S1 Starkrimson Luochuan 5.6 ± 0.4 12.3 ± 0.6 4.7 ± 0.3


S2 Starkrimson Luochuan 5.7 ± 0.3 11.9 ± 0.5 5.2 ± 0.6
S3 Starkrimson Xunyi 5.0 ± 0.5 13.1 ± 0.8 4.6 ± 0.3
S4 Starkrimson Xunyi 5.3 ± 0.2 12.7 ± 0.6 5.3 ± 0.4
S5 Starkrimson Xixia 5.7 ± 0.6 12.5 ± 0.6 4.9 ± 0.1
S6 Starkrimson Lingbao 5.4 ± 0.2 12.0 ± 0.2 5.9 ± 0.5
G1 Gala Luochuan 4.6 ± 0.4 14.6 ± 0.4 6.0 ± 0.3
G2 Gala Luochuan 4.6 ± 0.5 12.3 ± 0.3 5.4 ± 0.3
G3 Gala Luochuan 4.4 ± 0.3 12.1 ± 0.2 4.6 ± 0.6
G4 Gala Luochuan 4.7 ± 0.7 14.3 ± 0.2 5.6 ± 0.6
G5 Gala Xunyi 4.4 ± 0.4 14.8 ± 0.3 5.5 ± 0.5
G6 Gala Xixia 4.3 ± 0.2 14.3 ± 0.2 5.0 ± 0.4
G7 Gala Lingbao 4.8 ± 0.5 13.8 ± 0.6 4.7 ± 0.3
G8 Gala Lingbao 4.9 ± 0.7 14 ± 1 4.9 ± 0.5
J1 Jonagold Luochuan 4.4 ± 0.3 13.5 ± 0.6 5.4 ± 0.4
J2 Jonagold Xunyi 4.7 ± 0.4 12.8 ± 0.8 4.8 ± 0.3
J3 Jonagold Xunyi 4.8 ± 0.6 12.6 ± 0.5 5.7 ± 0.4
J4 Jonagold Xunyi 4.5 ± 0.3 12.1 ± 0.7 6.3 ± 0.2
J5 Jonagold Xunyi 5.5 ± 0.6 13 ± 1 5.9 ± 0.3
J6 Jonagold Xixia 4.8 ± 0.3 12.0 ± 0.7 5.8 ± 0.5
J7 Jonagold Lingbao 4.9 ± 0.6 12.30 ± 0.03 5.7 ± 0.4
J8 Jonagold Lingbao 4.5 ± 0.2 13.8 ± 0.3 6.4 ± 0.5
F1 Fuji Luochuan 5.6 ± 0.2 15.2 ± 0.3 4.3 ± 0.5
F2 Fuji Luochuan 5.7 ± 0.8 13.8 ± 0.9 5.4 ± 0.2
F3 Fuji Xunyi 5.7 ± 0.1 14.5 ± 0.2 4.7 ± 0.1
F4 Fuji Xunyi 5.6 ± 0.3 14.6 ± 0.9 5.3 ± 0.5
F5 Fuji Xunyi 5.8 ± 0.4 14.9 ± 0.7 4.9 ± 0.6
F6 Fuji Xunyi 5.9 ± 0.1 14.79 ± 0.04 4.5 ± 0.5
F7 Fuji Xixia 5.4 ± 0.2 14.1 ± 0.8 5.0 ± 0.4
F8 Fuji Lingbao 5.6 ± 0.5 15.1 ± 0.8 4.0 ± 0.4
F9 Fuji Lingbao 5.5 ± 0.2 14.7 ± 0.9 5.0 ± 0.6

a
malic acid in g/L juice.

sessions were performed by the panel, where the descriptive terms the selection criterion. The significance of the change in λ is determined
defining the aroma of juices in this research were developed by the by an F value. Before selecting a new variable to enter the classification
consensus method. During group discussions, each descriptive term was model, the stepwise process scans to figure out if all of the previously
extensively discussed by the group members to reach a consensus on its selected variables remain significant. If any one of them is no longer
meaning. significant, it will be removed from the classification model. When
A juice sample of 30 mL was presented and assessed at 12 °C in there are no other variables that can meet the standard for entering or
glasses according to the International Organization for Standardization when the variable to be entered next is the one discarded just now, the
ISO 3591 (1997). The panelists evaluated the apple juice, recorded each variable selection process stops. The classification performance of SLDA
aroma descriptor perceived and rated its intensity using a linear scale, was evaluated by leave-one-out cross-validation. ANOVA, PCA and
in which 0 denoted that the odor descriptor could not be detected and SLDA were carried out using SPSS 22.0 software (SPSS Inc., Chicago, IL,
1–5 denoted that the intensity of the odor descriptor was very weak, USA). The PLSR method was applied to reveal the link between the GC
weak, moderate, strong and very strong, respectively. For each apple and sensory data sets and to recognize the crucial volatiles that had a
juice, the relative intensity (I%), relative frequency (F%) and geometric profound impact on the aroma perception. PLSR analysis was executed
mean (GM%) of every aroma descriptor were calculated. The GM% is using 11.0 SIMCA-P software (Umetrics, Umeå, Sweden). The leave-
the square root of the product of F% and I% (Tao, Liu, & Li, 2009). one-out method was performed to verify the regression model. The
cumulative variance explained by extracted components (R2y(cum)) and
2.5. Statistical processing the cumulative cross-validation coefficient (Q2(cum)) were used to assess
the performance of the model. The reliability of the PLSR model was
Averaged volatile concentrations of the three replicate analyses for evaluated by CV-ANOVA.
each sample were used to carry out the subsequent statistical analysis.
To estimate the significant differences between volatile contents and 3. Results and discussion
the GM% data of aroma descriptors among the juices, a two factor-
variety and geographical origin-analysis of variance (ANOVA) was 3.1. Volatile profile-based classification
carried out. Differences at P < 0.05 (Duncan's test) were considered to
be statistically significant. Then, the volatile concentrations and GM% 3.1.1. Volatile composition of the apple juices
data were standardized by the z-score method prior to principal com- GC-MS analysis of the apple juices identified 39 volatile constituents
ponent analysis (PCA), stepwise linear discriminant analysis (SLDA) that belonged to six compound categories: 22 esters, 8 alcohols, 4 al-
and partial least squares regression (PLSR). PCA and SLDA were sepa- dehydes, 3 terpenoids, 1 ketone and 1 phenol (Table 2, Supplementary
rately performed on the volatile concentrations and GM% data to Tables 2–5). The volatile composition of the apple juices was in ac-
highlight the sample grouping and to establish classification models by cordance with the published data on apple juices and apple samples
the fruit variety and geographic origin, respectively. In SLDA, the most from China (Gong et al., 2019; Liu, Deng, Bi, Wu, & Zhang, 2019; Niu
significant variables for sample classification are selected by a stepwise et al., 2019; Wu, Wang, Yue, & Yuan, 2017).
process. When assessing the effect of each new variable, the Wilks' λ is Esters are the most dominate group that contributes mainly to the

3
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

Table 2
Volatile compounds identified in apple juices (μg/L) and significant differences of variety (V), geographical origin (G) factors and interaction V × G according to two-
way analysis of variance (ANOVA).
ID Compound LRIa ICb Starkrimson (n = 6) Gala (n = 8) Jonagold (n = 8) Fuji (n = 9) Significancec

Mean SD Mean SD Mean SD Mean SD V G V×G

Ester
1 Ethyl acetated 898 S, LRI, MS 124a 42 0.4c 0.3 106b 43 0.5c 0.1 *** ns ns
2 Ethyl propionatee 962 LRI, MS 21a 9 0.6c 0.2 7b 2 0.7c 0.2 *** ns ns
3 Propyl acetated 983 S, LRI, MS 289a 149 106b 13 17c 4 12c 3 *** ns ns
4 Methyl butanoated 996 S, LRI, MS 44a 22 2c 3 34 ab 37 12bc 6 ** ns ns
5 Isobutyl acetated 1025 S, LRI, MS 15a 9 1.8c 0.6 10b 3 1c 3 *** ns ns
6 Ethyl butanoatee 1052 LRI, MS 3114a 175 26c 41 5c 14 179b 168 *** ns *
7 Propyl propionated 1058 S, LRI, MS 54a 33 20b 4 0.4c 0.2 16b 4 *** ns *
8 Ethyl-2-methyl butanoated 1064 S, LRI, MS 10a 4 5b 3 3bc 1 2c 1 *** ns ns
9 Butyl acetated 1088 S, LRI, MS 2728 1355 2690 1658 4163 2537 4514 2856 ns *** **
12 2-Methylbutyl acetated 1136 S, LRI, MS 2929a 494 453b 88 187c 41 490b 125 *** ns ns
13 Propyl-2-methyl butanoatee 1152 LRI, MS 0.7b 0.5 2b 4 4 ab 4 6.1a 0.6 ** ns ns
14 Butyl propionatee 1156 LRI, MS 0.8c 0.2 1.6c 0.8 20b 8 27a 2 *** ns ns
16 2-Methylpropyl butanoated 1178 S, LRI, MS 46 21 35 88 11 7 7 3 ns ns ns
17 Pentyl acetatee 1185 LRI, MS 193 141 102 133 135 194 135 233 ns *** ***
18 Butyl butanoated 1234 S, LRI, MS 697c 246 2016a 738 244d 227 1294b 80 *** ns ns
20 Ethyl hexanoated 1248 S, LRI, MS 325a 149 19c 30 112b 15 176b 14 *** ns ns
21 Hexyl acetated 1287 S, LRI, MS 6265a 3204 2103b 658 1981b 196 1701b 319 *** ns ns
22 Propyl Hexanoatee 1336 LRI, MS 1758a 1021 3b 1 7b 6 41b 24 *** ns *
23 (E)-2-hexenyl acetated 1348 S, LRI, MS 212a 115 53b 10 2b 1 3b 6 *** ns ns
27 Hexyl butanoated 1430 S, LRI, MS 172a 63 NDd 0 132b 15 42c 12 *** ns ns
28 2-Methylhexyl butanoatee 1441 LRI, MS 124a 68 17c 15 78b 9 54b 5 *** ns ns
37 Hexyl hexanoatee 1608 LRI, MS 18a 16 15 ab 3 10.0b 0.7 0c 1 *** ns ns
Total 19139 7671 7266 8714

Alcohol
11 2-Methyl-1-propanold 1108 S, LRI, MS 99b 110 28c 34 160b 42 220a 50 *** ns ns
15 1-Butanold 1162 S, LRI, MS 4806a 2502 2260b 396 1917b 884 2288b 231 ** ns ns
25 1-Hexanold 1368 S, LRI, MS 900b 332 2643a 331 690bc 99 476c 36 *** ns ns
26 (Z)-2-hexen-1-old 1426 S, LRI, MS 144a 130 1b 4 44b 5 10b 3 *** ns ns
30 1-Heptanole 1466 LRI, MS 145 ab 154 210a 44 114b 20 109b 22 * ns ns
31 6-Methyl-5-hepten-2-old 1475 S, LRI, MS 70b 26 330a 53 32c 5 NDf 0 *** ns ns
32 2-Ethyl-1-hexanold 1500 S, LRI, MS 11c 10 45a 4 27b 4 5.6d 0.7 *** ns **
36 1-Octanole 1568 LRI, MS 21b 5 26a 5 22b 3 5c 2 *** ns ns
Total 6233 5543 3006 3114

Aldehyde
10 Hexanald 1098 S, LRI, MS 1711b 557 1948b 262 1907b 353 3668a 1067 *** ns ns
19 (E)-2-Hexenald 1238 S, LRI, MS 6155a 2489 1232c 449 3020b 371 2391bc 130 *** ns ns
33 Decanale 1509 LRI, MS 95 76 111 110 105 51 82 43 ns *** **
34 Benzaldehyded 1535 S, LRI, MS 22 43 16 15 11 6 4 3 ns ns ns
Total 7983 3307 5042 6146

Terpenoid
29 (Z)-linalool oxidee 1453 LRI, MS 4b 5 62a 14 1b 2 5b 2 *** ns ns
35 Linaloold 1549 S, LRI, MS 8a 5 3bc 4 4b 2 ND 0 ** ns ns
39 α-Farnesened 1766 S, LRI, MS 9b 8 19a 12 4b 2 9b 4 ** * **
Total 22 84 9 14

Ketone
24 6-Methyl-5-hepten-2-oned 1354 S, LRI, MS 26 11 24 9 47 58 31 10 ns ** ***

Phenol
38 p-Allylanisoled 1680 S, LRI, MS 80b 21 17c 6 121a 17 ND 0 *** ns ns

Notations: values with different roman letters (a-d) in the same row are significantly different among varieties.
c
*, ** and *** indicate significance at P < 0.05, P < 0.01 and P < 0.001, respectively, ns indicates no significant difference according to the Duncan test.
a
LRI, Linear retention index determined in a HP-INNOWax column.
b
IC, the identification criteria. Comparison with injected pure compound (S), comparison of LRI with literature data (LRI), mass spectra matching with the
standard NIST library (MS).
d
Quantified by calibration curve.
e
Quantified by utilizing the calibration curve of the compound having similar chemical structure.
f
ND, not detected.

sweet and fruity odors of apple juice (Komthong et al., 2006). The total significant contribution to the sweet, fruity and floral notes, while butyl
esters of all samples ranged from 3643 to 32143 μg/L. Hexyl acetate acetate makes a contribution to the ethereal-fruity notes (Komthong
and butyl acetate were the most plentiful esters, ranging from 1357 to et al., 2006). Ethyl butanoate and 2-methylbutyl acetate were also
11443 μg/L and 714–9619 μg/L in juice samples, respectively, which found in relatively high concentrations, reaching 5856 μg/L and
were consistent with the results of previous reports (Gan et al., 2014; 5191 μg/L, respectively, in the Starkrimson juice from Luochuan.
Komthong et al., 2007; Wu et al., 2017). Hexyl acetate makes a Alcohols are the degradation products of unsaturated fatty acids.

4
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

Fig. 1. Principal component analysis (PCA)


plot of volatile profile. (a) PCA score plot. St
( ), Starkrimson; Ga ( ), Gala; Jo ( ),
Jonagold; Fu ( ), Fuji. (b) PCA loading plot
of volatile compounds. Numbers identify
the volatile compounds as reported in
Table 2. The same symbol indicates the
volatile compounds come from the same
group. , ester; , alcohol; , aldehyde; ,
terpenoid; , ketone; , phenol.

Table 3 (Komthong et al., 2006).


Eigenvalues, explained variances, canonical correlations in the discriminant
functions for volatile and sensory profiles based classification. 3.1.2. ANOVA of the volatile compounds
Functiona Eigenvalue Explained variance (%) Canonical correlation To ascertain the significant differences in the volatile concentrations
among different varieties and geographical origins, the data were
DF 1-1 62.236 50.9 0.992
evaluated by two-factor ANOVA (Table 2). Thirty-three volatile com-
DF 1-2 43.821 35.8 0.989
DF 1-3 16.292 13.3 0.971 pounds varied significantly among varieties (19 esters, 8 alcohols, 2
DF 2-1 22.011 69.1 0.978 aldehydes, 3 terpenoids and 1 phenol). The most outstanding samples
DF 2-2 9.555 30.0 0.951 were Starkrimson juices. The concentrations of 20 volatile compounds,
DF 2-3 0.305 1.0 0.483 including 16 esters, in Starkrimson juices were significantly higher than
DF 3-1 28.168 52.9 0.983
DF 3-2 19.556 36.7 0.975
those in other juices. The total esters in Starkrimson juice were more
DF 3-3 5.531 10.4 0.920 than twice of those in the other varieties. The genes of the related en-
DF 4-1 31.876 97.10 0.985 zymes attending the synthetic process strictly control the biosynthesis
DF 4-2 0.887 2.70 0.686 of volatile compounds. Therefore, the volatile compounds of apple juice
DF 4-3 0.065 0.20 0.248
from a certain variety express its genetic characteristics to a significant
a
DF 1-1 ~ DF 1–3: the discriminant functions to classify the juices by apple
extent. There are only five volatile compounds that were significantly
varieties based on the volatile profiles; DF 2–1 ~ DF 2–3: the discriminant different among different geographical origins. Although compared
functions to classify the juices by geographic origins based on the volatile with the variety, the geographical origin had a relatively small impact
profiles; DF 3–1 ~ DF 3-3: the discriminant functions to classify the juices by on the volatile profile. The contents of volatile compounds varied
apple varieties based on the sensory profiles; DF 4–1 ~ DF 4–3: the discriminant greatly for juices from the same variety, demonstrating that the geo-
functions to classify the juices by geographic origins based on the sensory graphical origin also had an important effect on the volatiles. Finally,
profiles. there were only nine volatile compounds that presented a significant
interaction between the variety and geographical origin. The interac-
The total alcohols of all samples ranged from 995 to 10332 μg/L. tion between the variety and geographical origin was not significant for
Compared with aldehydes and esters, alcohols usually have a higher most of the volatile compounds, which meant that compared with the
odor threshold; therefore, they are regarded as the second largest variety, the interaction between the variety and geographical origin
contributors to the aroma of apples (Echeverria, Graell, Lopez, & Lara, also had a smaller impact on the volatile compounds in apple juice.
2004). 1-Butanol and 1-hexanol were the most dominant alcohols
identified, which has been previously recorded (Gan et al., 2014; 3.1.3. PCA of the volatile compounds
Nikfardjam et al., 2011). Due to its sweet aroma characteristic, 1-bu- PCA was applied to the concentrations of the volatile compounds to
tanol has been considered to contribute to the positive aroma features discriminate among the juice samples according to either their variety
of apple juice (Nikfardjam et al., 2011). 1-Hexanol with an unpleasant or geographical origin. The first two principal components accounted
and earthy odor has been found to serve as a negative contributor to the for 62.58% of the total variance. The first principal component (PC1)
aroma of the apple model mixture, which suppresses the apple-like odor and the second principal component (PC2) accounted for 41.21% and
(Bult et al., 2002). 21.37% of the total variance, respectively. The scores scatter plot of the
The total aldehydes of all samples were from 2615 to 12477 μg/L. samples is given in Fig. 1a, and the corresponding loadings establishing
For individual aldehydes, hexanal (977–4522 μg/L) and (E)-2-hexenal the relative importance of the variables are shown in Fig. 1b.
(343–10022 μg/L) were the most predominant constituents. These two All of the juices could be well separated according to apple varieties.
C6 aldehydes smell green grassy and have been identified to make a Starkrimson juices were mainly connected to PC1. These juices were
major contribution to the green aroma of apple juice (Zheng, Kim, Kim, characterized by remarkably high contents of esters, especially ethyl
Leem, & Lee, 2004). The C6 aldehydes, C6 alcohols and their derivatives propionate, propyl acetate, ethyl butanoate, propyl propionate, 2-me-
originate from the metabolism of fatty acids by lipoxygenases. The high thylbutyl acetate, hexyl acetate, propyl hexanoate and (E)-2-hexenyl
concentrations of these compounds in the juices might be because the acetate. Starkrimson is known to have an especially high ester emission
apples in this experiment were not fully mature (Onsekizoglu, Bahceci, compared to other varieties (Dixon & Hewett, 2000). In addition,
& Acar, 2010). Moreover, the enzymatic formation attributed to cell Starkrimson juices had very high contents of (E)-2-hexenal and (Z)-2-
disruption also resulted in high concentrations of these compounds hexen-1-ol. The Gala juices were positively related to PC2. These juices

5
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

Fig. 2. Stepwise linear discriminant analysis (SLDA) plot showing grouping of the juice samples: (a) volatile compounds based classification according to apple
variety (b) volatile compounds based classification according to geographical origin (c) aroma attributes based classification according to apple variety (d) aroma
attributes based classification according to geographical origin. St ( ), Starkrimson; Ga ( ), Gala; Jo ( ), Jonagold; Fu ( ), Fuji; Lu ( ), Luochuan; Xu ( ), Xunyi; Xi
( ), Xixia; Li ( ), Lingbao.

with lower concentration of esters had higher concentrations of alco- recognition ability and the prediction ability tested by the leave-one-
hols, such as 1-hexanol, 2-ethyl-1-hexanol and 6-methyl-5-hepten-2-ol. out strategy were both 100%.
In contrast, the Jonagold and Fuji juices were negatively associated Second, the geographical origins were set as categorical variables.
with PC2, and the most outstanding compounds in them were butyl Six volatile compounds were selected as input variables in LDA classi-
propionate and (E)-3-hexen-1-ol. fication: butyl acetate, pentyl acetate, 6-methyl-5-hepten-2-one, 1-
However, the juice samples from different geographical origins heptanol, 6-methyl-5-hepten-2-ol and decanal (F = 2 to enter, F = 1 to
could not be separated on the scores scatter plot. There were obvious remove). Three functions were established (DF 2–1 - DF 2–3, Table 3).
overlaps among the juice samples from different regions (Figure not These discriminant functions resulted in 96.8% total recognition ability
shown). and 93.5% prediction ability. The apple juices from Luochuan, Xixia
and Lingbao were all correctly grouped. However, two samples from
3.1.4. SLDA of the volatile compounds Xunyi, which composed 18.2% of the total from Xunyi, were mistaken
The concentrations of the volatile compounds were plugged into for those from Luochuan in the cross-validation section. This result
SLDA to classify the samples by apple varieties and geographic origins, might be due to the very close geographical positions of these two re-
respectively. First, four varieties were defined as categorical variables. gions. The orchards in Xunyi and Luochuan are located within a 100 km
Five independent variables (ethyl acetate, 2-methyl-1-propanol, 6-me- radius, and the weather conditions they experience are similar. The
thyl-5-hepten-2-ol, linalool and p-allylanisole) were screened out and results acquired from this series of data are illustrated Fig. 2b. The
viewed as input variables in LDA classification (F = 3.84 to enter, apple juice samples showed a sufficient division in the scatter diagram.
F = 2.71 to remove). The discriminant analysis for this series of data
generated three functions (DF 1-1, DF 1–2 and DF 1–3; Table 3). The
graphic representation of SLDA is demonstrated in Fig. 2a. The

6
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

Table 4
Geometric mean (%) of sensory descriptors of apple juices, and significant differences of variety (V), geographical origin (G) factors and interaction V × G according
to two-way analysis of variance (ANOVA).
ID Flavour descriptorsa Descriptor category Starkrimson (n = 6) Gala (n = 8) Jonagold (n = 8) Fuji (n = 9) Significanceb

Mean SD Mean SD Mean SD Mean SD V G V×G

1 Citrus Fruity 8a 3 1.9c 0.5 7a 4 4.6b 0.3 *** ns *


2 Pear Fruity 11a 2 8b 2 4c 1 7b 3 *** ns ns
3 Melon Fruity 56a 16 22b 9 19b 12 17b 4 *** ns ns
4 Apricot Fruity 8c 2 9a 2 4c 2 7b 1 *** ns ns
5 Pineapple Fruity 31a 11 13b 9 30a 12 39a 9 *** ns ns
6 Banana Fruity 40a 20 14b 5 17b 8 18b 7 ** * ns
7 Apple Fruity 53a 11 9c 4 18b 7 23b 4 *** ns ns
8 Strawberry Fruity 46a 19 16b 3 9b 4 14b 4 *** ns *
9 Almond Fruity 4 1 4 2 3 1 4 1 ns ns ns
10 Kiwifruit Fruity 0c 0 8a 3 4.0b 0.9 3.6b 0.9 *** ns ***
11 Lemon Fruity 4c 1 9a 1 4c 1 7b 1 *** ns ns
12 Cherry Fruity 9a 2 5b 1 3c 2 8a 1 *** ns ns
13 Cut green Vegetable 22 7 21 3 16 4 20 5 ns ns ns
14 Hay Vegetable 38 23 36 5 23 5 27 11 ns ns ns
15 Pumpkin Vegetable 7a 1 1b 1 7.2a 0.8 8a 3 *** ns ns
16 Anise Spices 2 1 2.1 0.9 1 1 2.0 0.7 ns ns ns
17 Cloves Spices 0b 0 0b 0 5a 1 0b 0 *** ns ns
18 Vanilla Spices 3 2 3 2 2 1 3 2 ns *** ns
19 Acacia Floral 7a 2 0b 0 0b 0 0b 0 *** ns **
20 Camomille Floral 3 1 4 1 4 1 4 1 ns *** ns
21 Geranium Floral 4 1 3 2 2 2 3.8 0.8 ns ns ns
22 Honey Other 9 2 7 1 8 1 7.1 0.9 ns ns ns
23 Coffee Other 0b 0 0b 0 2.4a 0.8 2.4a 0.8 *** ns ns

b
*, ** and *** indicate significance at P < 0.05, P < 0.01 and P < 0.001, respectively, ns indicates no significant difference according to the Duncan test.
a
Descriptors with GM > 10% are highlighted in bold letters.

Fig. 3. Principal component analysis (PCA)


plot of sensory profile. (a) PCA score plot. St
( ), Starkrimson; Ga ( ), Gala; Jo ( ),
Jonagold; Fu ( ), Fuji. (b) PCA loading plot
of aroma attributes. Numbers identify the
sensory descriptors as reported in Table 4.
The same symbol indicates the sensory de-
scriptors come from the same group. ,
fruity; , vegetable; , spices; , floral; ,
other.

3.2. Sensory profile-based classification interaction between the variety and geographical origin had relatively
small impacts on the aroma profile.
Sensory analysis showed that the apple juices were characterized
with 23 odor descriptors, which belonged to the following five cate-
3.2.2. PCA of the odor descriptors
gories: 12 fruity descriptors, 3 vegetable descriptors, 3 spice de-
The PCA of sensory data was employed to highlight the sample
scriptors, 3 floral descriptors and the other two descriptors. The geo-
grouping. PC1 and PC2 accounted for 30.83% and 21.92% of the total
metric means of the 23 odor descriptors are demonstrated in Table 4.
variance, respectively. The juice samples were basically clustered ac-
cording to apple variety (Fig. 3a).
3.2.1. ANOVA of the odor descriptors Starkrimson juices associated with PC1 were characterized by fruity
The two-factor ANOVA was executed on the geometric means (GM (melon, banana, apple, strawberry) and acacia odors (Fig. 3a and b).
%) of the odor descriptors. Fifteen out of twenty-three odor descriptors The Gala juices were positively related to PC2 and were characterized
varied significantly among the juices from different varieties. by apricot and lemon odors. For Jonagold juices (negatively associated
Meanwhile, only three descriptors varied significantly among geo- with PC2), the fruity odors were not highlighted except for citrus, and
graphical origins, and four descriptors varied significantly for the in- the prominent odors were pumpkin, cloves and coffee. The Gala juices
teraction between the variety and geographical origin. This result was were more equilibrated with regards to the aspect of odor descriptors.
consistent with the ANOVA outcome of GC data analysis, displaying Due to the evident overlaps on the scores scatter plot (Figure not
that compared with the variety, the geographical origin and the shown) the juice from different geographic origins could not be

7
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

Table 5
Odor activity values of volatile compounds reaching a concentration above the odor threshold (OAV > 0.5) in at least one variety.
Compound Odor descriptor Odor thresholda (μg/L) Starkrimson Gala Jonagold Fuji

Ester
Ethyl propionate Fruity 10 2.10 0.06 0.67 0.07
Methyl butanoate Fruity, cheese 60 0.73 0.03 0.56 0.19
Ethyl butanoate Fruity, apple, strawberry 1 3113.99 25.84 4.86 179.00
Ethyl-2-methyl butanoate Fruity, strawberry 0.006 1646.67 843.33 456.67 366.67
Butyl acetate Sweets, fruity 66 41.33 40.76 63.08 68.39
2-Methylbutyl acetate Overall aroma, characteristic apple, banana like 5 585.87 90.55 37.31 97.97
Propyl-2-methyl butanoate Fruity, apple 11 0.06 0.18 0.34 0.55
Pentyl acetate Fruity, banana 50 3.86 2.05 2.69 2.70
Butyl butanoate Rotten apple 100 6.97 20.16 2.44 12.94
Ethyl hexanoate Fruity, green apple 1 324.56 18.94 111.87 175.88
Hexyl acetate Sweet fruity, floral 115 54.48 18.29 17.22 14.79
Hexyl butanoate Green 250 0.69 0 0.53 0.17
2-Methylhexyl butanoate Fruity 22 5.63 0.79 3.56 2.45
Alcohol
1-Butanol Overall flavour, sweet aroma 500 9.61 4.52 3.83 4.58
1-Hexanol Fresh, green, earthy 150 6.00 17.62 4.60 3.17
(Z)-2-hexen-1-ol Green apple 70 2.06 0.02 0.62 0.15
Aldehyde
Hexanal Green 5 342.28 389.67 381.39 733.59
(E)-2-Hexenal Green 17 362.06 72.49 177.66 140.66
Ketone
6-methyl-5-hepten-2-one Citrus, strawberry 50 0.52 0.48 0.94 0.61

a
Odor descriptors and odor threshold were reported in literatures (Dixon et al., 2000; Komthong et al., 2006; Mehinagic et al., 2006).

separated by PCA. compounds. In this study, the apple maturity was assessed by starch-
iodine index. The starch-iodine test is one of the most widely used
methods to evaluate the maturity level of apples (Braga et al., 2013).
3.2.3. SLDA of the odor descriptors
Nevertheless, it should be noticed that as a harvest indicator, the re-
SLDA was performed to assess the capability of the aroma profile to
liability of the starch-iodine index might be affected by pre-harvest
classify the juice samples based on apple varieties and geographical
factors. A light cropping tree might have fruit stained darker because of
origins, respectively. At first, for variety-based classification, the geo-
the higher concentration of starch, but this does not necessarily mean a
metric means of sensory descriptors were defined as independent
postponed maturation (Blanpied & Silsby, 1992). In addition, the
variables in SLDA, and apple varieties were defined as categorical
binding ability of the starch and iodine molecules depends on the starch
variables. Six sensory descriptors that had the highest discrimination
composition. Compared with amylopectin, amylose has a much higher
power were selected (F = 3.84 to enter, F = 2.71 to remove). They
binding ability and contributes more to staining (Doerflinger, Miller,
were apple, hay, pumpkin, cloves, acacia and coffee. Three functions DF
Nock, & Watkins, 2015). A certain apple variety presents its specific
3–1, DF 3–2, and DF 3-3 were established by using the discriminant
changes in starch composition during the maturation process, and thus,
analysis (Table 3). A 100% recognition ability and 100% prediction
in future research the relationship between the starch composition and
ability were acquired. The discriminant analysis of the geometric means
starch-iodine index in each apple variety should be considered, and
of the sensory descriptors allowed an unambiguous classification of the
strictly standardized starch regression for each variety should be per-
juices from four apple varieties (Fig. 2c).
formed. Another crucial factor for aroma forming is year-to-year dif-
Then, for geographical origin-based classification, the independent
ferences due to climate variability. Although the pre- and postharvest
variables were geometric means of the sensory descriptors, and the
treatments are same, apples may display qualitatively and quantita-
categorical variables were the geographical origins. The descriptors of
tively different aromas in consecutive seasons (Echeverria, Fuentes,
melon, hay, vanilla and camomille were retained (F = 2 to enter, F = 1
Graell, Lara, & Lopez, 2004). Therefore, before the model of this re-
to remove) and were considered to have the greatest capacity for geo-
search can be applied in the apple industry, a more extensive set of
graphical origin based discrimination. The canonical discriminant
experiments with a larger number of juice samples from different apple
analysis generated three functions (DF 4–1 ~ DF 4–3, Table 3). The
varieties, geographic origins and harvest years is required.
recognition ability was 96.8%, and prediction ability was 93.5%. The
prediction abilities for juices from Luochuan, Xunyi, Xixia and Lingbao
were 88.9%, 90.9%, 100% and 100%, respectively, which could be 3.3. Correlation between the volatile compounds and aroma descriptors
considered satisfactory. The graphic representation of SLDA is demon-
strated in Fig. 2d. The juice samples showed a relatively sufficient di- The correlation between volatile compounds and aroma descriptors
vision into different origins. was explored by PLSR analysis. PLSR was applied between volatile
Combined with multivariate analysis, both the GC-MS data and constitutes (OAV > 0.5) and odor descriptors (GM > 10%, pear,
aroma characteristics could serve as effective indices to determine the melon, pineapple, banana, apple, strawberry, cut green and hay).
variety and geographic origin of apple juice. Meanwhile, it is note- Although there are hundreds of volatile compounds in apples and their
worthy that the growth environment, apple variety, maturity level, products, only a few of them are crucial for the aroma. Therefore, OAV
processing procedure and storage conditions affect the volatile con- was employed to evaluate the effect of every volatile constitute. Wine
stituents and aroma characteristics of the juice at the same time. Among sensory reconstitution research has demonstrated that if a volatile
these, fruit maturity has been considered to be a main factor. As a compound has an OAV > 0.5 in a model wine system, its aroma char-
climacteric fruit, apple accumulates more esters during maturation. The acteristics is remarkably similar to that in the wine. However, if a vo-
concentrations of esters achieve the highest level at the climacteric latile compound has an OAV < 0.5, the addition of it will not enhance
peak (Ortiz, Graell, & Lara, 2011). Therefore, apples harvested at im- the aroma characteristic of the wine (Ferreira, Ortin, Escudero, Lopez,
mature or overripe stages produce low concentrations of volatile & Cacho, 2002; Tao & Zhang, 2010). Therefore, volatile constitutes with

8
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

an OAV > 0.5 are potential contributors to the global aroma of the

2-methylbutyl acetate (1.10)


apple juice and were employed to establish aroma prediction models.

Methyl butanoate (1.08)

Ethyl propionate (1.35)


Ethyl propionate (1.38)

Ethyl propionate (1.27)


Hexyl butanoate (1.02)
(Z)-2-hexen-1-ol (1.29)
The cumulative variance and the cumulative cross-validation variance explained by the partial least squares regression (PLSR) model and the most significant volatile compounds contributing to the model.

3.4. OAVs of the volatile compounds

Nineteen out of 39 quantified volatile compounds had an


OAV > 0.5 in at least one variety, thus contributing to the juice aroma
VIP 5

(Table 5). Therefore, they were used to build sensory prediction


models. The highest OAVs were exhibited by ethyl butanoate (fruity,
apple and strawberry) and ethyl-2-methyl butanoate (fruity and
Ethyl-2-methyl butanoate (1.38)

Ethyl-2-methyl butanoate (1.30)

strawberry). Significant differences were observed for these two com-


2-Methylbutyl acetate (1.36)

pounds among varieties. Starkrimson was the variety that had the
(Z)-2-hexen-1-ol (1.11)
Ethyl hexanoate (1.22)
Ethyl butanoate (1.29)

highest values for these two compounds (OAV = 3113.99 and


1646.67). Other esters with higher OAVs were 2-methylbutyl acetate
Hexanol (1.04)

and ethyl hexanoate. Among the alcohols, 1-hexanol, 1-butanol and (Z)-
2-hexen-1-ol had OAVs higher than 0.5. Starkrimson was the variety
that had the highest OAV values for 1-butanol and (Z)-2-hexen-1-ol,
VIP 4

while Gala had the highest OAV value for 1-hexanol. Two aldehydes
with green grassy characteristics (hexanal and (E)-2-hexenal) presented
OAV values far in excess of 0.5 in all juice samples, indicating that these
2-Methylbutyl acetate (1.30)

two compounds were the general contributors to the green aroma


characteristic of the apple juices. Hexanal and (E)-2-hexenal showed
Butyl butanoate (1.32)
Ethyl butanoate (1.38)
Hexyl acetate (1.39)

Hexyl acetate (1.39)

maximum OAV values in the Fuji and Starkrimson varieties, respec-


1-butanol (1.25)

tively.
Hexanal (1.47)

3.5. PLSR analysis


VIP 3

Data were standardized by the z-score method prior to PLSR ana-


lysis. The reliability of the PLSR models was evaluated by CV-ANOVA.
VIP (n), the name of the most significant volatile compounds contributing to the model and its weight in parenthesis.
Butyl butanoate (1.37)
Ethyl butanoate (1.43)
Ethyl butanoat (1.41)

Seven PLSR models (p < 0.05) except the pear attribute were con-
(E)-2-Hexenal (1.40)
Hexyl acetate (1.30)

sidered reliable and were reported. The number of components esti-


Hexanol (1.63)

Hexanol (1.49)

mated, R2y(cum), Q2(cum) and the first 5 most significant variable im-
portance for the projection (VIP) values are displayed in Table 6. The
VIP value was used to show the importance of a variable to the model.
VIP 2

For instance, the best PLSR model of the “banana odor” was discussed.
The model, which explained 93.4% of the observed variance and 84.5%
of the cross-validation variance, was developed using 2 principal
Propyl-2-methyl butanoate (1.40)

components. Ethyl butanoate, hexyl acetate, pentyl acetate, 2-methyl-


2-Methylbutyl acetate (1.42)

2-Methylbutyl acetate (1.46)

butyl acetate and (Z)-2-hexen-1-ol were the major contributors to the


banana aroma. Although each compound had its own odor separately
Ethyl hexanoate (1.40)
Pentyl acetate (1.38)

(Table 5), when blended, they generated a complex character similar to


the banana aroma attribute. It's worth noticing that the major con-
Hexanal (1.95)
Hexanal (1.56)

tributors for the banana aroma, hexyl acetate, ethyl butanoate and 2-
Q2(cum), the cumulative cross-validation variance explained by the model.
No. of components, the number of the components used by the model.

methylbutyl acetate, were also the major contributors for the melon,
VIP 1d

apple and strawberry aroma models. It was more interesting that the
first 5 most significant VIPs for melon and strawberry odor models were
the same, although they had different contribution weights. These
Q2(cum)c

R2y(cum), the cumulative variance explained by the model.


0.768
0.784
0.845
0.694
0.885
0.577
0.775

confirmed that a complicated interaction among volatiles generates


odor sensations, and that the different sensations are produced when
the volatiles mixed with different ratios (Le Berre et al., 2008).
R2y(cum)b

0.804
0.874
0.934
0.729
0.920
0.784
0.929

4. Conclusions

A comprehensive study on the volatile contents and aroma profiles


No. of componenta

of the apple juices from the PDO areas in China was applied for the first
time. The discrimination of apple juices was carried out based on GC-
MS and quantitative descriptive sensory analysis. PCA and SLDA al-
lowed the discrimination of the juice samples from different apple
varieties according to both GC and sensory datasets. For the samples
1
3
2
1
2
3
4

from different geographical origins, SLDA provided a 93.5% prediction


Flavour descriptors

accuracy according to both datasets. Combined with multivariate ana-


lysis, both the volatile composition and aroma characteristics could be
effective indices to determine the variety and geographic origin of apple
Strawberry
Pineapple

Cut green

juice. This promising result encourages the supplementation of this


Banana
Table 6

Melon

Apple

Hay

analytical approach to identify adulterate or counterfeit apple juices.


d
b
a

Meanwhile, it is noteworthy that the growth environment, apple

9
J. Guo, et al. LWT - Food Science and Technology 124 (2020) 109203

variety, maturity level, processing procedure and storage conditions Doerflinger, F. C., Miller, W. B., Nock, J. F., & Watkins, C. B. (2015). Relationships be-
affect the volatile constituents and aroma characteristics of the juice at tween starch pattern indices and starch concentrations in four apple cultivars.
Postharvest Biology and Technology, 110, 86–95.
the same time. Therefore, before the model of this research can be Echeverria, G., Fuentes, T., Graell, J., Lara, I., & Lopez, M. L. (2004). Aroma volatile
applied in the apple industry, further study is required. A more ex- compounds of 'Fuji' apples in relation to harvest date and cold storage technology - a
tensive set of experiments with a larger number of juice samples from comparison of two seasons. Postharvest Biology and Technology, 32, 29–44.
Echeverria, G., Graell, J., Lopez, M. L., & Lara, I. (2004). Volatile production, quality and
different apple varieties, geographic origins and harvest years is needed aroma-related enzyme activities during maturation of 'Fuji' apples. Postharvest Biology
to optimize the model. In addition, the model should be assessed by and Technology, 31, 217–227.
external validation with a broader data set to ensure its robustness. Elss, S., Preston, C., Appel, M., Heckel, F., & Schreier, P. (2006). Influence of technolo-
gical processing on apple aroma analysed by high resolution gas chromatography-
Furthermore, PLSR models for the aroma attributes have been es- mass spectrometry and on-line gas chromatography-combustion/pyrolysis-isotope
tablished. A clear relationship between specific volatile constitutes and ratio mass spectrometry. Food Chemistry, 98, 269–276.
aroma attributes has been found. The compounds that had significant Ferreira, V., Ortin, N., Escudero, A., Lopez, R., & Cacho, J. (2002). Chemical character-
ization of the aroma of Grenache rose wines: Aroma extract dilution analysis,
contributions to the odor characteristics in apple juices have been
quantitative determination, and sensory reconstitution studies. Journal of Agricultural
identified. It can be confirmed that it is feasible to predict the intensities and Food Chemistry, 50, 4048–4054.
if sensory characteristics from volatile compounds. Gan, H.-H., Soukoulis, C., & Fisk, I. (2014). Atmospheric pressure chemical ionisation
mass spectrometry analysis linked with chemometrics for food classification - a case
study: Geographical provenance and cultivar classification of monovarietal clarified
CRediT authorship contribution statement apple juices. Food Chemistry, 146, 149–156.
Gong, D., Bi, Y., Jiang, H., Xue, S., Wang, Z., Li, Y., ... Prusky, D. (2019). A comparison of
Jing Guo: Methodology, Conceptualization, Writing - review & postharvest physiology, quality and volatile compounds of ‘Fuji’and ‘Delicious’ apples
inoculated with Penicillium expansum. Postharvest Biology and Technology, 150,
editing, Project administration, Funding acquisition. Tianli Yue: 95–104.
Methodology, Conceptualization. Yahong Yuan: Software, Validation, Guo, J., Yue, T., & Yuan, Y. (2012). Feature selection and recognition from nonspecific
Investigation. Nan Sun: Writing - original draft. Peng Liu: volatile profiles for discrimination of apple juices according to variety and geo-
graphical origin. Journal of Food Science, 77, C1090–C1096.
Methodology, Software. Guo, J., Yue, T., Yuan, Y., & Wang, Y. (2013). Chemometric classification of apple juices
according to variety and geographical origin based on polyphenolic profiles. Journal
Declaration of competing interest of Agricultural and Food Chemistry, 61, 6949–6963.
ISO 3591 (1977). Sensory analysis. Apparatus wine tasting glass.
Jonker, M. T. O., & Muijs, B. (2010). Using solid phase micro extraction to determine
The authors declare that they have no known competing financial salting-out (Setschenow) constants for hydrophobic organic chemicals. Chemosphere,
interests or personal relationships that could have appeared to influ- 80, 223–227.
Komthong, P., Igura, N., & Shimoda, M. (2007). Effect of ascorbic acid on the odours of
ence the work reported in this paper.
cloudy apple juice. Food Chemistry, 100, 1342–1349.
Komthong, P., Katoh, T., Igura, N., & Shimoda, M. (2006). Changes in the odours of apple
Acknowledgments juice during enzymatic browning. Food Quality and Preference, 17, 497–504.
Lawless, H. T., & Heymann, H. (1998). Descriptive analysis. Sensory evaluation of food.
Principles and practices. New York: Chapman & Hall.
This study was supported by the National Natural Science Le Berre, E., Beno, N., Ishii, A., Chabanet, C., Etievant, P., & Thomas-Danguin, T. (2008).
Foundation of China (31601436) and the Fundamental Research Funds Just noticeable differences in component concentrations modify the odor quality of a
for the Central Universities (2452017145). blending mixture. Chemical Senses, 33, 389–395.
Liu, X., Deng, J., Bi, J., Wu, X., & Zhang, B. (2019). Cultivar classification of cloudy apple
juices from substandard fruits in China based on aroma profile analyzed by HS-
Appendix A. Supplementary data SPME/GC-MS. Lwt-Food Science and Technology, 102, 304–309.
Mehinagic, E., Royer, G., Symoneaux, R., Jourjon, F., & Prost, C. (2006). Characterization
of odor-active volatiles in apples: Influence of cultivars and maturity stage. Journal of
Supplementary data to this article can be found online at https:// Agricultural and Food Chemistry, 54, 2678–2687.
doi.org/10.1016/j.lwt.2020.109203. Nikfardjam, M. P., & Maier, D. (2011). Development of a headspace trap HRGC/MS
method for the assessment of the relevance of certain aroma compounds on the
sensorial characteristics of commercial apple juice. Food Chemistry, 126, 1926–1933.
References
Niu, Y., Wang, R., Xiao, Z., Zhu, J., Sun, X., & Wang, P. (2019). Characterization of ester
odorants of apple juice by gas chromatography-olfactometry, quantitative measure-
Alberti, A., Machado dos Santos, T. P., Ferreira Zielinski, A. A., Eleuterio dos Santos, C. ments, odour threshold, aroma intensity and electronic nose. Food Research
M., Braga, C. M., Demiate, I. M., et al. (2016). Impact on chemical profile in apple International, 120, 92–101.
juice and cider made from unripe, ripe and senescent dessert varieties. Lwt-Food Onsekizoglu, P., Bahceci, K. S., & Acar, M. J. (2010). Clarification and the concentration
Science and Technology, 65, 436–443. of apple juice using membrane processes: A comparative quality assessment. Journal
Antón, M. J., Suárez, V. B., García, H. A., & Picinelli, L. A. (2014). Aromatic profile of of Membrane Science, 352, 160–165.
ciders by chemical quantitative, gas chromatography-olfactometry, and sensory Ortiz, A., Graell, J., & Lara, I. (2011). Volatile ester-synthesising capacity throughout on-
analysis. Journal of Food Science, 79, 92–99. tree maturation of 'Golden Reinders' apples. Scientia Horticulturae, 131, 6–14.
Belton, P. S., Colquhoun, I. J., Kemsley, E. K., Delgadillo, I., Roma, P., Dennis, M. J., et al. Reid, L. M., Woodcock, T., O'Donnell, C. P., Kelly, J. D., & Downey, G. (2005).
(1998). Application of chemometrics to the 1H NMR spectra of apple juices: Differentiation of apple juice samples on the basis of heat treatment and variety using
Discrimination between apple varieties. Food Chemistry, 61, 207–213. chemometric analysis of MIR and NIR data. Food Research International, 38,
Blanpied, G. D., & Silsby, K. J. (1992). Predicting harvest date windows for apples, 221. 1109–1115.
Cornell Cooperative Extension. Tao, Y., Liu, Y.q., & Li, H. (2009). Sensory characters of cabernet sauvignon dry red wine
Braga, C. M., Ferreira Zielinski, A. A., da Silva, K. M., Fernandes de Souza, F. K., Moura from changli county (China). Food Chemistry, 114, 565–569.
Pietrowski, G.d. A., Couto, M., et al. (2013). Classification of juices and fermented Tao, Y., & Zhang, L. (2010). Intensity prediction of typical aroma characters of cabernet
beverages made from unripe, ripe and senescent apples based on the aromatic profile sauvignon wine in Changli County (China). Lwt-Food Science and Technology, 43,
using chemometrics. Food Chemistry, 141, 967–974. 1550–1556.
Bult, J. H. F., Schifferstein, H. N. J., Roozen, J. P., Boronat, E. D., Voragen, A. G. J., & Wu, H., Wang, J., Yue, T., & Yuan, Y. (2017). Variety-based discrimination of apple juices
Kroeze, J. H. A. (2002). Sensory evaluation of character impact components in an by an electronic nose and gas chromatography-mass spectrometry. International
apple model mixture. Chemical Senses, 27, 485–494. Journal of Food Science and Technology, 52, 2324–2333.
Dixon, J., & Hewett, E. W. (2000). Factors affecting apple aroma/flavour volatile con- Zheng, C. H., Kim, T. H., Kim, K. H., Leem, Y. H., & Lee, H. J. (2004). Characterization of
centration: A review. New Zealand Journal of Crop and Horticultural Science, 28, potent aroma compounds in Chrysanthemum coronarium L. (Garland) using aroma
155–173. extract dilution analysis. Flavour and Fragrance Journal, 19, 401–405.

10

You might also like