Antioxidant and flavor properties of different apple cultivars grown in Korea

Abstract

This study investigated the antioxidant capacity, chemical composition, and sensory quality of various apple cultivars grown in Korea. ‘Hongro’, ‘Picnic’, ‘Eazypple’, ‘Hwangok’ and ‘Fuji’ showed high antioxidant activity, whereas ‘Arisoo’, ‘Arione’, and ‘Ruby-S’ consistently indicated lower capacities. ‘Arisoo’ and ‘Arione’ were characterized by high sensory scores for sweetness, sourness, and fruity, apple, and pineapple flavors, and received high consumer preference ratings, despite having lower antioxidant activities. However, ‘Hongro’ was perceived as less sweet and sour, with dominant cucumber-like off-flavors, resulting in lower overall preference, although it was rich in flavonoids and exhibited strong antioxidant capacity. These results suggest that flavonoid content may contribute to antioxidant activity but is not necessarily associated with favorable flavor characteristics. Overall, this study provides a comprehensive framework for profiling the nutritional and sensory attributes of apples using chemometrics and offers practical insights for selecting cultivars suited to specific applications in both fresh consumption and processing.

Highlights

• •Antioxidant and flavor qualities are governed by distinct chemical networks.
• •Epicatechin, procyanidin B1, and phloridzin drive apple antioxidant capacity.
• •Flavonoid-rich apple cultivars enhance health potential but not perceived flavor quality.
• •Fruity esters and green aldehydes serve as precise chemo-markers of consumer liking.

1. Introduction

Apple (Malus domestica Borkh.) is consumed worldwide in various forms, including fresh fruit, juice, purée, wine, and dried chips because of its attractive sensory and nutritional properties (Dadwal et al., 2023). Epidemiological studies have shown that apple consumption is associated with a reduced risk of certain cancers, cardiovascular disease, asthma, and diabetes (Nezbedova et al., 2021). The health benefits of apples are primarily attributed to their antioxidant compounds, such as polyphenols and flavonoids, which are known to neutralize free radicals that can damage DNA, cell membranes, and other cellular components (Chandimali et al., 2025). The composition of antioxidant compounds in apple fruit is cultivar-dependent, with the most abundant components being dihydrochalcones (phlorizin and phloretin), flavan-3-ols ((+)-catechin and (−)-epicatechin), procyanidins, and anthocyanins (cyanidin-3-galactoside) (Feng et al., 2021; J. Y. Kim et al., 2021). The peel of the apple, in particular, generally contains higher concentrations of phytochemicals and exhibits greater antioxidant activity than the flesh, particularly due to the abundance of anthocyanins in the peel (Kaeswurm et al., 2023; Zhang et al., 2020).

The overall quality of apples is determined not only by their antioxidant properties but also by a complex interplay of taste, flavor, appearance, and nutritional profile. Sensory characteristics are key determinants of consumer preferences and perceptions of apple quality (Drkenda et al., 2021). Among these, flavor is widely regarded as the primary factor in consumer acceptance of apples, often outweighing other factors such as texture and color. Apple flavor quality is primarily driven by the combination of sugars, organic acids, and the emission of volatile compounds (Kim et al., 2023). The perceived sweetness, sourness, and characteristic aroma of an apple can be influenced by the composition of sugars and organic acids, as well as by the presence and complex interactions among various volatile compounds (Kim et al., 2023; Zheng et al., 2025). Each cultivar possesses genetically-determined chemical characteristics that can influence phenotypic traits, such as antioxidant activity and sensory quality (Bouillon et al., 2024; Braga et al., 2021; Mignard et al., 2021). To evaluate the overall quality of apples, it is crucial to investigate chemical profiles associated with both antioxidant activity and sensory quality across various apple cultivars. Although numerous studies have independently examined antioxidant activities and sensory qualities in different apple cultivars, an integrated approach that simultaneously considers these factors is essential for achieving a more comprehensive understanding.

In Korea, apples are one of the most significant cultivated crops, with a production of 394,428 tons across 33,789 ha in 2023 (KOSTAT, 2023). Korean apple cultivars have been selectively bred to meet consumer preferences and to enhance agricultural sustainability. Several notable apple cultivars, including ‘Hongro’, ‘Picnic’, ‘Arisoo’, ‘Eazypple’, ‘Hwangok’, ‘Ruby-S’ and ‘Arione’ have been bred for their unique characteristics such as flavor, color, and resistance to pests and diseases (Ban et al., 2014; Lee et al., 2021). ‘Fuji,’ developed in Japan and renowned for its crisp texture and excellent storage quality, dominates Korean apple production, accounting for approximately 67% of the total cultivation area, and is also one of the most widely cultivated apple varieties worldwide (Li et al., 2019; Win et al., 2024). However, despite the development of these diverse cultivars, a comprehensive comparative analysis of their antioxidant capacities and flavor properties has not yet been conducted for Korean-grown apples.

Therefore, the objectives of this study were: 1) to evaluate the antioxidant capacity and chemical composition of apple fruits; 2) to characterize the flavor attributes of different apple cultivars using trained sensory panels; and 3) to investigate the interactions between the chemical composition associated with antioxidant activity and the flavor quality of apples using chemometrics.

2. Materials and methods

2.1. Chemicals

Trolox, gallic acid, (+)-catechin, ethylenediaminetetraacetic acid (EDTA), 2,2-azino-bis-(3-ethylbenzothiazoline-6-sulphonic acid) (ABTS), potassium ferricyanide, 1,1-diphenyl-2-picrylhydrazyl (DPPH), ferric chloride, ferrous chloride, trichloroacetic acid, 3-(2-pyridyl)-5,6-bis-(4-phenylsulphonic acid)-1,2,4-triazine, Folin–Ciocalteu’s phenol reagent, potassium persulphate, sodium hydroxide, sodium carbonate, sodium nitrite, formic acid, cyclohexanone, hexanal, ethyl acetate, and linalool were purchased from Sigma-Aldrich. (St. Louis, MO, USA). Dimethyl sulfoxide (DMSO) was purchased from Duksan Pure Chemicals (Daegu, South Korea). Solvents for extraction and LC–MS analyses (acetonitrile, methanol, and water) were HPLC or LC–MS grade, obtained from Honeywell (Philadelphia, PA, USA).

2.2. Apple samples

Eight apple cultivars (‘Ruby-S’, ‘Hongro’, ‘Picnic’, ‘Fuji’, ‘Arisoo’, ‘Eazypple’, ‘Hwangok’, and ‘Arione’) were obtained from the Apple Experiment Station, National Institute of Horticultural and Herbal Science, Rural Development Administration of South Korea. The apples were harvested in 2020 at the mature stage from different trees and were carefully selected for their uniformity in size, color, and absence of defects. A total of 20–30 fruits were randomly collected from different trees for each cultivar. After harvest, the fruits from the same batch were transported to the laboratory washed with tap water and sodium bicarbonate to remove surface contaminants. The cleaned samples were then packed in polyethylene zipper bags and stored at 4 °C under 90–95 % relative humidity minimizes quality changes prior to analysis.

For antioxidant activity analysis, ten fruits from each of the eight cultivars were selected, pooled, and used as a composite sample. However, ‘Ruby-S’, which has a smaller fruit size than the other cultivars, twenty fruits were used to obtain a comparable amount of material. Whole fruits were maintained intact; the peel and flesh were separated, and all samples were stored at –80 °C until freeze-drying.

Five apple cultivars (‘Arione’, ‘Arisoo’, ‘Eazypple’, ‘Hongro’, and ‘Picnic’) were selected based on their differences in antioxidant activity and subjected to in-depth analyses to determine physicochemical properties, chemical composition, and flavor profile. These samples were stored at 4 °C until sensory evaluation, which was conducted within two weeks after harvest. Each fruit was divided into three uniform portions to ensure consistency across analyses: one portion was pooled for sensory evaluation, another was used for physicochemical analyses, and the remaining portion was designated for compositional analyses, including sugars, organic acids, and volatile compounds. By dividing each fruit in this way and pooling two fruits per replicate, variation among different parts of the fruit was minimized, thereby ensuring representativeness and reproducibility. For sensory evaluation, the apples were washed, peeled, and cut into uniform cubes (1 × 1.5 × 1.5 cm). Two fruits were randomly selected for each session, and the evaluation was repeated three times. For the chemical composition analysis, the flesh samples were quenched in liquid nitrogen and homogenized with a mortar and pestle. The pretreatment of flesh samples for sensory evaluation followed the procedure described in our previous work (Kim et al., 2023).

2.3. Preparation of ethanol extract

The flesh, peel, and whole samples were freeze-dried in freeze dryer (FDTA-4504, Peron Co. Ltd., Korea) for the antioxidant capacity and chemical profiling. The freeze-dried samples were ground into a fine powder with liquid nitrogen using mortar and pestle. Then, 5 g of freeze-dried apple powders were extracted by 200 mL of 100 % ethanol at room temperature overnight. After extraction, the extract was filtered, and the supernatant was evaporated at 40 °C until constant weight was achieved. The dried extract was further suspended in DMSO and stored at –80 °C until analysis.

2.4. Physicochemical analysis

Apple flesh color was assessed using a calibrated colorimeter (Spectrophotometer; CM-3500d, Minolta Co., Ltd., Osaka, Japan) to determine the chromaticity values L* (lightness), a* (green to red) and b* (blue to yellow). Total soluble solids (TSS, °Brix) were determined using a refractometer (Digital Hand-held Pocket Refractometer; PAL-1, Atago, Tokyo, Japan). Titratable acidity (TA, % malic acid) was evaluated to determine the total acidity of the fruits. Each sample was measured in triplicate.

2.5. Descriptive analysis

The sensory characteristics of five apple fruits were determined using descriptive analysis. Eleven trained panelists (ages = 23 ± 3, male = 3, female = 8) participated in each sensory analysis, and the evaluations were carried out under artificial light and room temperature (22 °C). The panels were familiarized with the reference of sensory attributes during the training sessions. Sensory attributes were assessed by 15-point scale (0 = none, 15 = extremely strong) and evaluated taste (sweetness and sourness) and flavor (apple, pineapple, pear, honey, cucumber, vanilla, fruity, and floral) profiles with reference standards according to previous studies (Table S8) (Aprea et al., 2012; Dixon & Hewett, 2000; Kim et al., 2023). The apple samples with random 3-digit codes were provided with water and crackers to rinse the mouth between samples. This sensory evaluation protocol was approved by the Institutional Review Boards (IRBs 1040191-202006-HR-010-01) of Andong National University, and all experiments were conducted with understanding and written consent of participants, according to the IRB guidelines.

2.6. Determination of antioxidant activity

The total polyphenol and flavonoid contents of apple samples were measured using the method previously described by Kim et al. (2024). Antioxidant activities, including ABTS and DPPH radical scavenging and metal chelating effects, were evaluated as described by Yu et al. (2023).

2.7. Analysis of flavonoid composition

Apple extracts (0.1 g) were subjected to sonication with 1 mL of methanol for 30 min, followed by centrifugation at 12,000 rpm for 30 min. The supernatant was filtered through a 0.2 μm nylon filter, and a 3 μL aliquot was injected into a QTRAP 4500 LC-MS/MS system (Applied Biosystems, Foster City, CA, USA) connected to a Nanospace UPLC system (Shiseido, Tokyo, Japan). Flavonoids were separated using a PREMIER BEH C18 column (1.0 × 150 mm, 2.1 μm particle size; Waters Corp., Milford, MA, USA) with mobile phases A (10 % acetonitrile in water containing 0.1 % formic acid) and B (100 % acetonitrile). The flow rate was set to 0.4 mL/min, and the column temperature was maintained at 40 °C. The gradient elution program was as follows: 2 % B (0–3 min), 2–50 % B (3–25 min), 50–95 % B (25–30.1 min), and 95–2 % B (30.1–45 min). Authentic standards were used for the identification of flavonoids by matching retention times.

2.8. Analysis of anthocyanin composition

Apple extracts (0.5 g) were subjected to sonication with 5 mL of 80 % methanol containing 0.1 % HCl for 15 min, followed by centrifugation at 3,000 rpm for 10 min. This procedure was repeated twice, and the combined supernatants were adjusted to a final volume of 10 mL. The supernatant was then filtered through a 0.2 μm nylon filter, and a 1 μL aliquot was injected into an Acquity™ ultra-performance liquid chromatography-UV (UPLC; Waters Corp., Milford, MA, USA) system for analysis at 515 nm. The anthocyanins were separated using a CORTECS® T3 column (2.1 × 150 mm, 1.6 μm particle size; Waters Corp., Milford, MA, USA) with (A) 0.5 % trifluoroacetic acid in water and (B) 0.5 % trifluoroacetic acid in 50% acetonitrile as the mobile phase. The flow rate was set to 0.3 mL/min, and the column temperature was maintained at 30 °C. The gradient elution program was as follows: 100 % B (0–3 min), 100–24 % B (3–11 min), 24–27 % B (11–20 min), 27–42 % B (20–35 min), 42–45 % B (35–40 min), 45–90 % B (40–50 min), and 90–20 % B (50–60 min). Authentic standards were used for the identification of anthocyanins by matching retention times.

2.9. Analysis of sugar and organic acid compositions

For sugar analysis, freeze-dried apple flesh powders (0.2 g) were weighed and mixed with 10.0 mL of 50 % acetonitrile. The suspensions were vortexed for 5 min and then subjected to ultrasonic-assisted extraction for 30 min at 50 °C. After centrifugation at 3,000 rpm for 15 min, the supernatant was filtered through a 0.45 μm nylon filter, and a 10 μL aliquot was injected into a HPLC-refractive index (RI) detector system (Waters Corp., Milford, MA, USA). Sugars were separated using an Asahipak NH2P-50 column (250 × 4.6 mm, 5 μm particle size; Shodex, Tokyo, Japan) with acetonitrile/water (75:25, v/v) as the mobile phase. The flow rate was 1.0 mL/min, and the column temperature was maintained at 40 °C. Authentic standards were used to identify the sugars by matching retention times.

For organic acid analysis, freeze-dried apple flesh powders (0.2 g) were weighed and mixed with 10 mL of phosphoric acid in water (pH 2). The suspensions were vortexed for 5 min and then extracted using ultrasonic-assisted extraction for 20 min on ice. After centrifugation at 12,000 rpm for 20 min, the supernatant was filtered through a 0.2 μm nylon filter, and a 20 μL aliquot was injected into a HPLC-UV detector system (Osaka Soda, Osaka, Japan) set at 220 nm. Organic acids were separated using a Cadenza CD-C18 column (250 × 4.6 mm, 3 μm particle size; Imtakt Corp., Kyoto, Japan) with isocratic phosphoric acid/water (pH 2) at 0.6 mL/min and 30 °C. Authentic standards were used to identify the organic acids by matching retention times.

2.10. Determination of volatile compounds

Volatile compounds of apple fruits were extracted using headspace solid-phase microextraction (HS-SPME) and analyzed using gas chromatography-mass spectrometry (GC-MS). GC-MS (QP 2020, Shimadzu, Kyoto, Japan) was coupled with autosampler (AOC 6000, Shimadzu, Kyoto, Japan). Apple flesh sample (7 g) was placed into a 20 mL of headspace vial containing 1 mL of distilled water and internal standard (cyclohexanone, 20 μg/mL in methanol). Samples were incubated at 50 °C for 10 min in a thermostatically controlled autosampler followed by the absorption of the divinylbenzene/polydimethylsiloxane (DVB/PDMS) SPME fiber (65 μm film thickness, Supelco, Bellefonte, PA) into the headspace volatiles of each vial for 20 min. After absorption, the volatile compounds were thermally desorbed at 240 °C for 15 min with a splitless injector port of the GC. The GC-MS system was operated in electron impact ionization mode (EI, 70 ionization energy) with a scan range from m/z 30–300. The volatile compounds were separated on a HP-FFAP capillary column (50 m × 0.32 mm i.d., 0.5 μm film thickness; Agilent, Santa Clara, CA, USA) with constant pressure. Helium was used as a carrier gas with a flow rate of 1.0 mL/min. The GC oven temperature program was initially held at 40 °C for 2 min, increased from 40 °C to 64 °C at a rate of 2 °C/min (after a 1 min hold), then increased from 64 °C to 200 °C at a rate of 4 °C/min (after a 10 min hold), and held at the final temperature of 230 °C for 13 min at a rate of 10 °C/min. Ion source and interface temperature were set at 230 °C. Saturated alkanes (C7–C30) were used to determine linear retention index (RI) of each volatile compound. Compound identifications were performed by comparison of mass spectra with NIST 14 library and RI of authentic standards. When authenticated standards were not available, tentative identification was carried out based on experimental RIs from the literature. The relative concentration of each component was determined by internal standard method. The formula content is as follows: component concentration (μg/g) = (A₁ × M₁)/(A₂ × M₂), where A₁ represents the peak area of the detected compound, A₂ is the peak area of the internal standard, M₁ is the mass of the injected internal standard, and M₂ is the sample weight (g).

2.11. Statistical analysis

Statistical analyses were performed using one-way and two-way analyses of variance (ANOVA). One-way ANOVA followed by Tukey’s honestly significant difference (HSD) test was conducted using Statistical Analysis System (SAS 9.4 Institute Inc., Cary, NC, USA). Two-way ANOVA was performed using the Pingouin package (version 0.5.5) in Python (version 3.13.3) to examine the interaction effects between cultivar and part (peel, flesh, and whole fruit). A P-value < 0.05 was considered indicative of a statistically significant difference. Pearson correlation analysis was conducted to examine the relationships between antioxidant activities and bioactive compounds of apple cultivars, using SPSS software (version 20.0; SPSS Inc., Chicago, IL, USA), and the data were visualized using the corrplot package in R (version 4.3.0). Principal component analysis (PCA) and partial least squares regression (PLS regression) were performed using XLSTAT 2019 software (Addinsoft, New York, USA) to characterize the antioxidant activity of apple fruits and to investigate the relationships between their chemical compositions and sensory attributes, respectively.

3. Results and discussion

3.1. Antioxidant activities of different apple cultivars

Polyphenols, flavonoids, and anthocyanins are significant bioactive compounds in fruits and vegetables, highly correlated with antioxidant activity, which contributes to health promotion and the prevention of chronic diseases such as obesity, cancer, and diabetes (de Mello Andrade & Fasolo, 2014). In this study, eight different apple cultivars grown in Korea were evaluated for their antioxidant contents, including total polyphenols, total flavonoids, and total anthocyanins (Fig. 1 and Table S1). Antioxidant activities varied among the eight cultivars. The antioxidant content of all apple cultivars was significantly higher in the peel compared to the flesh and whole fruit (peel + flesh). The TPC of whole apple fruits varied in the range of 287.68 ± 7.64–570.29 ± 11.47 mg CE/100 g DW. The ranking of TPC in whole apple fruits from highest to lowest was as follows: ‘Hwangok’ > ‘Eazypple’ > ‘Hongro’ > ‘Picnic’ > ‘Fuji’ > ‘Arione’ > ‘Arisoo’ > ‘Ruby-S’. In the flesh, TPC ranged from 234.77 ± 6.51 to 531.00 ± 6.26 mg GAE/100 g DW, with ‘Hwangok’ exhibiting the highest polyphenol content and ‘Arisoo’ showing the lowest. In the peel, TPC varied from 473.53 ± 3.67 to 773.37 ± 28.51 mg GAE/100 g DW. The peel of ‘Hwangok’ also had the highest polyphenol content, whereas the lowest was observed in ‘Arisoo’. The TFC of whole apple fruits ranged from 35.36 ± 0.83 to 160.66 ± 6.41 mg CE/100 g DW. The ranking of TFC in whole apple fruits from highest to lowest was as follows: ‘Hongro’ > ‘Eazypple’ > ‘Picnic’ > ‘Fuji’ > ‘Hwangok’ > ‘Arione’ > ‘Ruby-S’ > ‘Arisoo’. Although the TFC ranking did not match the TPC ranking, apple cultivars with higher TPC generally exhibited higher TFC. The TFC in the flesh ranged from 23.08 ± 0.80 mg CE/100 g DW in ‘Hongro’ to 156.41 ± 1.70 CE/100 g DW in ‘Arisso’. In the peel, ‘Hongro’ also exhibited the highest TFC (292.74 ± 12.17 mg CE/100 g DW), while the lowest value was observed in ‘Ruby-S’ (78.25 ± 0.45 mg CE/100 g DW).

Fig. 1.

(A) Representative appearance of eight apple cultivars used in this study. (B) Total phenolic content and (C) total flavonoid content measured in the whole fruit, peel, and flesh of each cultivar (; flesh, ; peel, ; whole). Data are expressed as means ± standard deviation (n = 3).

ABTS and DPPH radical scavenging, reducing power, and chelating effect assays have been widely used to evaluate antioxidant activity, which could reflect various antioxidant mechanisms. The DPPH assay primarily measures antioxidant activity through electron transfer reactions and hydrogen atom abstraction, while the ABTS radical cation is soluble in both aqueous and organic solvents, making it suitable for measuring antioxidant capacity in both hydrophilic and lipophilic media (Marathe et al., 2011). The reducing power and chelating effect assays evaluate the electron-donating capacity of antioxidants and their ability to inhibit metal-catalyzed oxidation, respectively (Yu et al. 2023). The antioxidant activities of eight apple cultivars were evaluated using ABTS and DPPH radical scavenging activities, as well as reducing power and chelating effect assays (Fig. 2 and Table S2). The results indicated that most antioxidant activities, except for chelating effect, were greater in apples with higher TPC and TFC, such as ‘Hongro’, ‘Picnic’, ‘Fuji’, ‘Eazypple’, and ‘Hwangok’. In contrast, apple cultivars with lower TPC and TFC, such as ‘Ruby-S’, ‘Arisoo’, and ‘Arione’, exhibited reduced antioxidant activities. ABTS radical scavenging activity was highest in the peel of ‘Hongro’ (807.31 ± 13.48 mg TEAC/100 g DW), followed by ‘Picnic’ (779.02 ± 36.06 mg TEAC/100 g DW), while ‘Arisoo’ peel showed the lowest value (279.87 ± 6.32 mg TEAC/100 g DW). In the flesh, ‘Hongro’ had the highest activity (458.57 ± 8.31 mg TEAC/100 g DW), while ‘Ruby-S’ flesh exhibited the lowest (93.73 ± 5.57 mg TEAC/100 g DW). The whole fruit samples showed a similar trend, with ‘Hongro’ and ‘Ruby-S’ having the highest and lowest activities, respectively. DPPH radical scavenging activity was also greatest in the peel of ‘Hongro’ (634.44 ± 24.66 mg TEAC/100 g DW) and ‘Picnic’ (557.58 ± 7.64 mg TEAC/100 g DW), whereas ‘Arisoo’ showed the lowest values in both peel and flesh. The flesh of ‘Hongro’ also exhibited the highest activity (380.48 ± 42.65 mg TEAC/100 g DW), followed by ‘Picnic’ (311.16 ± 69.56 mg TEAC/100 g DW). Overall, the peels exhibited significantly higher activity than flesh across all cultivars. Reducing power was highest in the peel of ‘Hongro’ (30.84 ± 0.88 mg TEAC/100 g DW) and ‘Picnic’ (32.91 ± 0.29 mg TEAC/100 g DW). The flesh of ‘Hongro’ also exhibited the highest reducing power (23.69 ± 0.29 mg TEAC/100 g DW), whereas ‘Arisoo’ had the lowest value in both peel (14.26 ± 0.39 mg TEAC/100 g DW) and flesh (6.85 ± 0.35 mg TEAC/100 g DW). The chelating activity varied among cultivars and was not dependent on the other assays. The peel of ‘Picnic’ had the highest chelating ability (383.70 ± 19.40 mg EDTA/100 g DW), followed by ‘Hongro’ flesh (494.74 ± 31.26 mg EDTA/100 g DW). By contrast, ‘Arisoo’ displayed the weakest chelating effect in both peel (414.11 ± 40.07 mg EDTA/100 g DW) and flesh (529.57 ± 46.33 mg EDTA/100 g DW).

Fig. 2.

Antioxidant activities of eight different apple cultivars. (A) ABTS radical scavenging activity, (B) DPPH radical scavenging activity, (C) reducing power, and (D) chelating inhibitory effect (; flesh, ; peel, ; whole). Data are expressed as means ± standard deviation (n = 3).

To further clarify the contribution of cultivar and part (flesh, peel, and whole) to antioxidant activity, a two-way ANOVA was applied (Table S3). Both cultivar and part type had significant effects (P < 0.001) on all antioxidant indices, including TPC, TFC, ABTS, DPPH, reducing power, and chelating effect. The interaction between cultivar and part type was also significant (P < 0.001) for all parameters except DPPH (P = 0.0563), indicating that the influence of cultivar on antioxidant activity varied depending on the tissue. These results suggest that antioxidant potential in apples in governed not only by genetic differences among cultivars but also by anatomical variation between peel and flesh, with tissue type showing a stronger overall effect than cultivar.

Collectively, our findings suggest that apple cultivars such as ‘Hongro’, ‘Picnic’, and ‘Eazypple’, which demonstrate high antioxidant potential owing to elevated TPC and TFC, may offer enhanced antioxidant and health benefits. In contrast, ‘Arisoo’, ‘Arione’ and ‘Ruby-S’, which consistently exhibited lower antioxidant activities than other cultivars, may have limited potential for these health benefits. Our study also clearly demonstrated that apple peels contain significantly higher levels of antioxidant activities compared to flesh and whole fruit. This is consistent with previous studies indicating that the peel is a richer source of polyphenols and flavonoids, likely due to its exposure to environmental factors that stimulate the synthesis of these compounds (Kim et al., 2011; Łata, 2007; Wolfe et al., 2023). These results imply that consuming whole apples, including the peel, could maximize antioxidant intake, while discarding the peel would result in a significant loss of these beneficial compounds.

3.2. Flavonoid and anthocyanin profiles of different apple cultivars

Apple fruits contain significant amounts of flavonoids and anthocyanins, with their composition and prevalence varying depending on the cultivar (Boyer & Liu, 2004). Flavonoids are present in various derivative forms within both the flesh and peel, whereas anthocyanins, the primary pigments responsible for red and purple coloration, are predominantly concentrated in the peel (Serrano et al., 2005; Tsao et al., 2005). In this study, we analyzed the composition of eight major flavonoids (phlorizin, phloretin, quercetin, luteolin, chrysin, eriodictyol, (-)-epicatechin, and procyanidin B1) across the flesh, peel, and whole fruit of apples, as well as four key anthocyanins (cyanidin-3-galactoside, cyanidin-3-arabinoside, cyanidin-3-rhamnoside, and peonidin-3-rutinoside) in the peel.

The flavonoid content of different apple cultivars is presented in Table S4. Our results showed significant variation in flavonoid composition among cultivars and across different parts (flesh, peel, and whole fruit). The peels of most cultivars contained substantially higher flavonoid levels than their flesh. The predominant flavonoid derivatives in apples were phlorizin (47.00 ± 4.06–670.46 ± 100.30 μg/100 g DW), (-)-epicatechin (6.55 ± 1.06–977.52 ± 133.70 μg/100 g DW), procyanidin B1 (0–96.08 ± 11.41 μg/100 g DW), and quercetin (0–8.03 ± 0.49 μg/100 g DW), while phloretin, luteolin, chrysin, and eriodictyol were present in low but detectable amounts. Phlorizin, a dihydrochalcone glycoside, and its aglycone form, phloretin, are known to be abundant in apple peels and contribute significantly to the antioxidant capacity of apples (Jugdé et al., 2008). Our data indicated that apples contained significantly higher levels of phlorizin compared to phloretin. The phlorizin content in whole apple fruits ranged from 76.29 ± 10.26 μg/100 g DW in ‘Arione’ to 235.24 ± 17.97 μg/100 g DW in ‘Eazypple’. The peels of ‘Hongro’ and ‘Eazypple’ exhibited higher levels of phlorizin than those of other cultivars. Quercetin, a dietary antioxidant flavonol, has been reported to possess antiviral, anti-inflammatory, and anticarcinogenic properties (Di Petrillo et al., 2022; Hollman et al., 1997; Lesjak et al., 2018). The peels of ‘Fuji’ (8.03 ± 0.49 μg/100 g DW), ‘Picnic’ (7.36 ± 0.30 μg/100 g DW), and ‘Eazypple’ (7.11 ± 0.26 μg/100 g DW) contained more quercetin compared to other cultivars. Quercetin was absent in the flesh part of apples. Previous studies have demonstrated the antioxidant potential of epicatechin and procyanidins, particularly in inhibiting low-density lipoprotein oxidation in vitro and in vivo (da Silva Porto et al., 2003; Rein et al., 2000; Steffen et al., 2005). In our study, the (-)-epicatechin content of whole apple fruits varied from 19.88 ± 4.25 to 599.79 ± 15.11 μg/100 g DW. The peel of ‘Hongro’ exhibited the highest (-)-epicatechin content (977.52 ± 133.70 μg/100 g DW), followed by that of ‘Picnic’ (857.17 ± 87.45 μg/100 g DW). In contrast, the flesh of ‘Arisoo’ and ‘Ruby-S’ had the lowest (-)-epicatechin contents, at 6.55 ± 1.06 and 8.79 ± 0.98 μg/100 g DW, respectively. The ranking of (-)-epicatechin levels across apple parts from highest to lowest was as follows: ‘Hongro’ > ‘Picnic’ > ‘Fuji’ > ‘Eazypple’ > ‘Arione’ > ‘Arisoo’ > ‘Hwangok’ > ‘Ruby-S’. The highest procyanidin B1 content was also found in the peels of ‘Hongro’ (96.08 ± 11.41 μg/100 g DW) and ‘Picnic’ (57.92 ± 5.62 μg/100 g DW), while the flesh of ‘Ruby-S’ contained the lowest level (0.51 ± 0.81 μg/100 g DW). Procyanidin B1 was absent in the flesh of both ‘Arisoo’ and ‘Hwangok’.

The effects of cultivar, part type (flesh, peel, and whole), and their interaction on the distribution of individual flavonoids were further analyzed using two-way ANOVA (Table S5). Significant differences (P < 0.05) were observed in most compounds for both cultivar and tissue effects. Phlorizin, phloretin, quercetin, epicatechin, and procyanidin B1 exhibited highly significant effects (P < 0.001) for cultivar, part, and their interaction, indicating distinct accumulation patterns across both genetic and anatomical factors. Eriodictyol also showed significant variation among the three factors (P < 0.01), while luteolin and chrysin indicated no significant part or interaction effect, suggesting their distribution was relatively uniform across parts. These findings suggest that most flavonoid compounds in apples are jointly influenced by cultivar and tissue.

In this study, the anthocyanin content in the peels of various apple cultivars was measured (Table S6), with a focus on four key anthocyanins –cyanidin-3-galactoside, cyanidin-3-arabinoside, cyanidin-3-rhamnoside, and peonidin-3-rutinoside– because anthocyanins are primarily concentrated in this part of the fruit. Our results indicated that cyanidin-3-galactoside was the most abundant anthocyanin in apple peels, contributing to the red coloration observed in the peels (Awad et al., 2001; Wolfe et al., 2023). The highest cyanidin-3-galactoside content was found in ‘Arisoo’ (33.60 ± 1.65 mg/100 g DW), followed by ‘Ruby-S’ (32.04 ± 1.86 mg/100 g DW), ‘Picnic’ (31.23 ± 0.70 mg/100 g DW), and ‘Eazypple’ (30.73 ± 1.23 mg/100 g DW). In contrast, ‘Fuji’ (13.07 ± 0.32 mg/100 g DW) and ‘Hwangok’ (17.28 ± 0.33 mg/100 g DW) contained significantly lower amounts of cyanidin-3-galactoside. The lower levels observed in ‘Hwangok’ are likely due to its yellow peel color. Although ‘Fuji’ has a red peel, the relatively low cyanidin-3-galactoside content suggests that other pigments may primarily contribute to its coloration. For cyanidin-3-arabinoside, ‘Picnic’ (2.18 ± 0.16 mg/100 g DW) and ‘Hongro’ (2.55 ± 0.17 mg/100 g DW) showed the highest levels, while ‘Fuji’ and ‘Hwangok’ displayed the lowest concentrations. Elevated levels of cyanidin-3-rhamnoside were observed in ‘Arisoo’ (1.81 ± 0.11 mg/100 g DW), ‘Picnic’ (1.73 ± 0.12 mg/100 g DW), and ‘Eazypple’ (1.16 ± 0.10 mg/100 g DW) compared with other cultivars. In terms of peonidin-3-rutinoside, ‘Picnic’ had the highest content (2.69 ± 1.54 mg/100 g DW), followed by ‘Arisoo’ (2.33 ± 0.10 mg/100 g DW), whereas ‘Fuji’ and ‘Hwangok’ had the lowest levels (0.68 ± 0.05 and 0.83 ± 0.06 mg/100 g DW, respectively). These anthocyanins were not detected in the ‘Arione’ cultivar.

Pearson correlation analysis was performed to evaluate the relationships between antioxidant activity parameters and the physicochemical and sensory characteristics of the apple cultivars (Fig. S1). Strong positive correlations were observed among total polyphenol content (TPC), total flavonoid content (TFC), and antioxidant activity indicators, including DPPH, ABTS, and reducing power (r > 0.88, P < 0.01). In particular, TFC showed very high correlations with ABTS (r = 0.980), DPPH (r = 0.928), and reducing power (r = 0.944), indicating that flavonoid compounds significantly contribute to antioxidant capacity. Antioxidant activities were highly correlated with each other, except for the chelating effect, which exhibited a distinct pattern. This difference may be attributed to the unique mechanism of metal ion chelation, which differs from radical scavenging or electron donation. The antioxidant effects of compounds in apples are therefore likely to involve radical scavenging and reducing mechanisms rather than metal ion chelation. Furthermore, (–)-epicatechin, procyanidin B1, quercetin, and phloridzin showed relatively high correlation coefficients (r > 0.60, P < 0.01) with antioxidant parameters such as TFC and ABTS, suggesting that these flavonoids may play a key role in the antioxidant potential of apple cultivars.

Overall, our findings confirm that apple peels contained higher concentrations of flavonoids compared to the flesh and whole fruit, with considerable variation among cultivars. These differences might be attributed to genetic factors that regulate flavonoid biosynthesis (Feng et al., 2013; Wang et al., 2018), influencing both the total flavonoid content and the potential health-promoting properties of apples. Apple peels, often treated as by-products, have previously been reported to be rich in phytochemicals such as flavonoids and anthocyanins (Asma et al., 2023; Sagar et al., 2018). In this study, these antioxidant compounds were predominantly concentrated in the peel, further underscoring its nutritional relevance. (–)-Epicatechin, procyanidin B1, quercetin, and phloridzin were identified as key contributors to the antioxidant activity of apples. Therefore, processing strategies that enable the efficient extraction and retention of these compounds may support the development of value-added functional apple products. Incorporating peels during processing, rather than discarding them, could enhance the nutritional quality of apple-derived foods. This strategy also aligns with the upcycled food movement by promoting sustainability and providing economic advantages through the transformation of agricultural by-products into functional ingredients (Rakesh & Mahendran, 2024).

3.3. Relationship between antioxidant activities and flavonoid profiles in different apple cultivars

This study conducted a PCA to better understand the relationships among different chemical compositions and antioxidant activities of apple cultivars. The PCA biplots illustrate the distribution of different parts (flesh, peel, and whole) of eight apple cultivars in relation to their flavonoid and anthocyanin compositions, as well as their antioxidant activities (Fig. 3).

Fig. 3.

The relationships between antioxidant activity and chemical composition in eight apple cultivars. (A) Flesh, (B) peel, and (C) whole fruit. Each PLS-DA plot shows associations between samples and antioxidant indicators and key bioactive compounds.

The first two principal components in the flesh portion of the PCA biplot accounted for 59.82% of the total variation (PC1 = 42.34% and PC2 = 17.48%) (Fig. 3A). PC1 is primarily associated with antioxidant activities, including TPC, TFC, DPPH and ABTS radical scavenging activities, and reducing power, along with several flavonoids such as quercetin, (-)-epicatechin, and procyanidin B1. ‘Hongro,’ ‘Picnic,’ and ‘Eazypple’ are positioned towards the positive side of PC1, indicating their high antioxidant content. In contrast, ‘Arisoo’ and ‘Ruby-S’ are located on the negative side of PC1, exhibiting lower polyphenol and flavonoid concentrations, which correlate with weaker antioxidant activities.

For the apple peel, the first two principal components in the PCA biplot explained 67.01% of the total variance (PC1: 46.11%, PC2: 20.9%) (Fig. 3B). PC1 was strongly associated with antioxidant parameters (TPC, TFC, DPPH, ABTS, and reducing power), major flavonoids such as quercetin, (–)-epicatechin, and phloridzin, as well as anthocyanins including cyanidin-3-galactoside, cyanidin-3-rhamnoside, and cyanidin-3-arabinoside. ‘Hongro,’ ‘Fuji,’ and ‘Picnic’ were located on the positive axis of PC1, indicating a high abundance of antioxidant compounds in their peel. Conversely, ‘Arione’ and ‘Ruby-S’ were clustered on the negative side of PC1, reflecting relatively lower antioxidant activity and flavonoid content. These findings suggest that the peel of certain cultivars possesses a more diverse and abundant flavonoid profile, contributing to greater antioxidant potential.

In the PCA biplot of the whole fruit, the first two principal components explained 62.42% of the total variance, with PC1 and PC2 contributing 41.42% and 21.00%, respectively (Fig. 3C). Consistent with the results from the flesh and peel, PC1 exhibited strong positive loadings with TPC, TFC, DPPH and ABTS radical scavenging activities, and reducing power, as well as key flavonoids such as quercetin, (–)-epicatechin, and procyanidin B1. ‘Fuji’ and ‘Hongro’ were located in the upper-right quadrant, reflecting a rich antioxidant profile in the whole fruit. In contrast, ‘Ruby-S’ and ‘Arione’ appeared on the opposite side of PC1, indicating relatively lower levels of polyphenolic compounds and antioxidant activity.

3.4. Physicochemical characteristics and taste/flavor attributes of different apple cultivars

The relationship between antioxidant activity and sensory characteristics was examined through a comparative analysis of five apple cultivars, selected on the basis of their antioxidant potential. Accordingly, the cultivars were classified into high (‘Eazypple’, ‘Hongro’, and ‘Picnic’) and low (‘Arione’ and ‘Arisoo’) antioxidant groups for further comparative analysis. These samples were subjected to comprehensive analyses, including physicochemical properties and taste/flavor characteristics.

The physicochemical characteristics of five apple cultivars are shown in Table S7. The apple cultivars exhibited statistically significant differences in TSS, TA, TSS/TA ratio, and color values (a* and b* value) (P < 0.05). ‘Picnic’ had the highest TSS content (18.09 ± 1.75 °Brix), followed by ‘Eazypple’ (16.76 ± 0.16 °Brix) and ‘Arione’ (16.44 ± 0.54 °Brix), whereas the lowest was observed in ‘Arisoo’ (13.69 ± 1.86 °Brix). In contrast, TA was highest in ‘Arisoo’ (0.43 ± 0.03%) and lowest in ‘Hongro’ (0.13 ± 0.01%). The TSS/TA ratio showed significant variation among the apple cultivars (P < 0.05). ‘Hongro’ had the highest ratio (113.48 ± 8.35), followed by ‘Eazypple’ (63.83 ± 5.71) and ‘Picnic’ (58.61 ± 2.35), while the lowest was observed in ‘Arisoo’ (31.85 ± 6.55).

For color measurements of the edible portion, no significant differences were found in lightness (L*), which ranged from 74.37 ± 6.24 in ‘Hongro’ to 78.20 ± 3.09 in ‘Eazypple’. However, the cultivars exhibited significant differences in their a (redness) and b* (yellowness) values. The a* value ranged from 2.43 ± 4.60 in ‘Eazypple’ to 10.13 ± 1.41 in ‘Hongro’, while the highest b* value was observed in ‘Arione’ (18.14 ± 3.09) and the lowest in ‘Hongro’ (12.69 ± 0.56).

The flavor characteristics of the five apple cultivars were evaluated by trained panels using two taste attributes (e.g., sweetness and sourness) and nine flavor descriptors (e.g., pineapple, apple, pear, honey, vanilla, floral, fruity, green-like, and cucumber) (Fig. 4A). Descriptive profiling revealed significant differences (P < 0.05) in sourness and several flavor notes, including pineapple, apple, green-like, and cucumber. ‘Arione’ exhibited the highest perceived sweetness, while ‘Eazypple’ showed the lowest. For sourness, the cultivars were ranked in the following order: ‘Arisoo’ > ‘Eazypple’ > ‘Picnic’ > ‘Arione’ > ‘Hongro’. ‘Arisoo’ was rated highest in both sweetness and sourness and also recorded high intensities for pineapple, apple, honey, and fruity notes, which contributed to its distinct sensory profile. By contrast, ‘Hongro’ not only displayed relatively low intensities for both taste attributes but also recorded the lowest intensities for pineapple, apple, pear, honey, and green-like notes, while exhibiting the highest intensity for the cucumber note, clearly distinguishing it from the other cultivars. ‘Arione’, ‘Picnic’, and ‘Eazypple’ exhibited moderately high scores in apple, fruity, and green-like notes, while ‘Arione’ had the lowest cucumber intensity. Honey and pear descriptors varied little among cultivars and were perceived at relatively low intensities.

Fig. 4.

Sensory profiles of different five apple cultivars. (A) Radar plot showing the distribution of taste and flavor attributes across cultivars. (B) PCA biplot illustrating the relationships between apple cultivars and sensory attributes.

In the PCA biplot of flavor attributes for the five apple cultivars, the first two principal components explained 60.80% of the total variance, with PC1 and PC2 accounting for 36.45% and 24.35%, respectively (Fig. 4B). ‘Hongro’ was distinctly positioned on the negative side of PC1, influenced by strong cucumber and green-like notes and reduced levels of sweetness, sourness, and fruit-related attributes. In contrast, ‘Arisoo’ and ‘Arione’ were located on the positive side of PC1, characterized by higher intensities of sweetness, apple, and fruity descriptors. Meanwhile, ‘Eazypple’ and ‘Picnic’ clustered near floral and green-like attributes, indicating a shared flavor profile shaped by these aromatic notes. Among the cultivars, ‘Arisoo’ received the highest overall acceptability, followed by ‘Picnic’, ‘Arione’, and ‘Eazypple’, while ‘Hongro’ was rated the lowest (Fig. S1). Although ‘Hongro’ exhibited the highest TSS/TA ratio, which is typically considered an indicator of favorable flavor balance, it received the lowest preference score. This may be attributed to its dominant cucumber flavor and weak perception of sweetness, sourness, and fruity characteristics. These findings align with previous studies reporting that apples with diminished sweetness, acidity, and fruity aroma tend to be less favored by consumer (Kim et al., 2023). Moreover, Moreover, our results suggest that the TSS/TA ratio alone may not adequately reflect the sensory complexity or consumer acceptance of apple cultivars, despite its traditional use as a quality index (Jung et al., 2023; Teerachaichayut & Ho, 2017). For example, although ‘Hongro’ had the highest TSS/TA ratio, it received the lowest overall preference score. In contrast, ‘Arisoo’, which exhibited the lowest TSS/TA value, was perceived as both sweeter and sourer, and received the highest liking score. While ‘Picnic’ had the highest TSS content, it was perceived as only moderately sweet, whereas ‘Arione’ and ‘Arisoo’, with comparatively lower TSS values, were rated as sweeter. These inconsistencies imply that TSS alone is not a reliable predictor of perceived sweetness.

Taken together, these findings highlight that the conventional physicochemical parameters, such as TSS and TA, may have limited value in predicting sensory outcomes and consumer liking. Instead, specific volatile-derived flavor characteristics—particularly fruity, apple-like, and floral notes—appear to be stronger determinants of apple preference. Therefore, future assessments of apple quality should incorporate sensory-driven flavor profiling in conjunction with compositional analysis to more accurately reflect consumer perception and guide cultivar development.

3.5. Sugars, organic acids, and volatile compositions of different apple cultivars

Sugars and organic acids are primarily associated with sweetness and sourness, respectively, while volatile compounds contribute to the overall flavor profile (Sung et al., 2019). The investigation of compositional characteristics is essential for understanding flavor perception based on taste- and flavor-related compounds. In this study, the composition of sugars, organic acids, and volatile compounds in the edible flesh portion of apples was analyzed to identify key contributors to taste and flavor attributes. The total sugar content in the flesh of the five apple cultivars ranged from 767.81 ± 8.37 g/kg in ‘Picnic’ to 790.34 ± 10.58 g/kg in ‘Arione’ (Table 1). Fructose was identified as the predominant sugar in all five apple cultivars, with the highest concentrations observed in ‘Hongro’ (423.17 ± 23.44 g/kg) and ‘Picnic’ (431.10 ± 24.50 g/kg). Sucrose content exhibited considerable variation among cultivars, ranging from a maximum of 267.96 ± 5.13 g/kg in ‘Arisoo’ to a minimum of 146.49 ± 32.96 g/kg in ‘Picnic’. Notably, ‘Eazypple’ contained the highest level of sorbitol (44.21 ± 20.23 g/kg), a sugar alcohol that may influence sweetness perception, with significantly lower levels detected in the other cultivars. Interestingly, these compositional patterns did not entirely align with the sensory evaluation results for sweetness. Although ‘Hongro’ and ‘Picnic’ exhibited high levels of total sugars and fructose, they were perceived as less sweet by the sensory panel. Conversely, ‘Arisoo’ and ‘Arione’ received the highest sweetness scores despite having relatively lower fructose content. This discrepancy suggests that sweetness perception in apples is not determined solely by total sugar concentration or the dominant sugar type.

Table 1.

Sugar and organic acid composition in five apple cultivars.

	Arione	Arisoo	Eazypple	Hongro	Picnic
Sugars (g/kg)
Fructose	349.15 ± 21.66b	414.31 ± 11.17a	307.56 ± 39.67b	423.17 ± 23.44a	431.41 ± 24.50a
Sorbitol	38.63 ± 19.76abc	17.84 ± 4.65bc	44.21 ± 20.73ab	14.45 ± 4.59c	49.61 ± 19.65a
Glucose	176.03 ± 25.80a	88.10 ± 3.11b	158.40 ± 24.98a	153.59 ± 33.26a	140.30 ± 28.84a
Sucrose	226.53 ± 52.42ab	267.96 ± 5.13a	261.25 ± 49.42a	186.34 ± 53.75c	146.49 ± 32.96c
Total sugars	790.34 ± 10.58a	788.20 ± 13.41a	771.43 ± 10.06b	777.55 ± 3.25ab	767.81 ± 8.37b
Organic acids (mg/kg)
Oxalic acid	528.81 ± 51.31ab	645.54 ± 26.52a	431.18 ± 252.09b	691.17 ± 39.53a	722.89 ± 120.56a
Tartaric acid	46.33 ± 80.24ab	125.03 ± 2.75a	50.00 ± 86.60ab	N.D.b	121.95 ± 105.61a
Malic acid	21158.91 ± 4134.40b	26518.06 ± 2818.32a	24215.79 ± 1270.08ab	13232.94 ± 913.77c	23528.97 ± 3149.49ab
Acetic acid	677.13 ± 586.99b	2386.95 ± 514.65a	1143.72 ± 162.34b	1070.36 ± 126.23b	609.55 ± 583.83b
Citric acid	385.26 ± 76.42c	468.35 ± 14.08bc	386.67 ± 52.35c	1119.70 ± 336.84a	641.27 ± 37.75b
Total organic acids	22796.45 ± 4663.90b	30143.93 ± 2790.81a	26227.37 ± 1269.24ab	18141.55 ± 2706.46c	25624.74 ± 2313.46b

The data are expressed as a mean ± standard deviation (n = 3).

^a–eDifferent letters indicate significant statistical differences at P < 0.05 by Tukey’s HSD test. N.D.:not detected.

The composition of organic acids in the flesh of five apple cultivars revealed significant differences among the samples. Malic acid, the main organic acid responsible for sourness in apples, was detected at high levels across all cultivars, with the highest concentrations in ‘Arisoo’ (26,518.06 ± 2,818.32 mg/kg) and ‘Eazypple’ (24,215.79 ± 1,270.08 mg/kg), In contrast, ‘Hongro’ exhibited the lowest malic acid content (13,232.94 ± 913.77 mg/kg). ‘Arisoo’ also exhibited high concentrations of oxalic acid, tartaric acid, and acetic acid, whereas ‘Hongro’ showed comparatively lower levels of most organic acids. ‘Hongro’ had a relatively high citric acid content compared to the other cultivars, despite its lower overall organic acid profile. The concentration of malic acid closely reflected the sensory perception of sourness, reinforcing its role as the primary souring agent in apples. This observation is consistent with the known relative sourness intensities of major organic acids, where malic acid is perceived as the most sour, followed by tartaric acid and then citric acid (Neta et al., 2007).

The volatile composition of five apple cultivars was comprehensively characterized, revealing the presence of 12 esters, 7 alcohols, 5 aldehydes, 1 furan, 1 acid, 1 terpene, and several miscellaneous compounds (Table 2). Esters and alcohols associated with fruity and sweet aromas were the most dominant classes of volatiles identified across all samples. Notably, ‘Arisoo’ and ‘Arione’ exhibited a greater number, and higher concentrations of these fruit- and sweet-related esters compared to the other cultivars. ‘Arisoo’, in particular, showed the highest levels of key esters contributing to apple- and pineapple-like notes, including hexyl acetate (apple), hexyl 2-methylbutyrate (apple), butyl acetate (pear), 2-methylbutyl acetate (fruity), and butyl 2-methylbutyrate (fruity, pineapple), which likely enhanced its sensory preference. In addition, green-like aroma compounds such as hexanol, hexanal, and (E)-2-hexenal were consistently detected across all cultivars, although their concentrations varied considerably, reflecting cultivar-specific compositional differences. These compounds may contribute to the fresh and favorable flavor characteristics observed in certain cultivars. Among the alcohols, 2-methyl-1-butanol—associated with wine- and fusel oil-like notes and commonly linked to off-flavors—was found in relatively high concentrations in ‘Hongro’. On the other hand, the apple cultivars such as ‘Arione’ and ‘Arisoo’, which contained only moderate levels of this off-flavor compound along with high concentrations of desirable fruity esters, received higher overall acceptability scores.

Table 2.

Volatile Compositions in five apple cultivars (μg/g).

Compound	RIA	Aroma descriptorB	Arione	Arisoo	Eazypple	Hongro	Picnic	IdentificationC
Esters
Butyl acetate	1080	Pear	1.38 ± 1.45b	13.18 ± 3.62a	N.D. Db	N.D.b	0.63 ± 1.10b	FI, MS, TI
2-Methylbutyl acetate	1130	Fruity	N.D.b	8.78 ± 2.83a	N.D.b	N.D.b	N.D.b	FI, MS, TI
Amyl acetatens	1183	Banana	N.D.	0.90 ± 1.55	N.D.	N.D.	N.D.	FI, MS, TI
Methyl hexanoatens	1195	Fruity, pineapple	0.11 ± 0.10	0.06 ± 0.10	N.D.	N.D.	0.03 ± 0.05	MS, TI
Butyl butyratens	1224	Fruity, pineapple	0.74 ± 0.55	0.82 ± 0.94	N.D.	0.09 ± 0.16	N.D.	FI, MS, TI
Butyl 2-methylbutyratens	1236	Fruity	1.37 ± 0.95	3.91 ± 2.74	0.35 ± 0.33	4.34 ± 3.13	0.17 ± 0.30	FI, MS, TI
2-Methylbutyl butyratens	1270	Fruity, sweet	0.15 ± 0.13	N.D.	N.D.	N.D.	N.D.	MS, TI
Hexyl acetate	1280	Apple, fruity	2.16 ± 3.74b	25.55 ± 12.38a	N.D.b	N.D.b	N.D.b	FI, MS, TI
2-Methylbutyl 2-methylbutyratens	1285	Apple, fruity	0.56 ± 0.97	N.D.	0.03 ± 0.05	1.84 ± 2.52	N.D.	FI, MS, TI
Butyl hexanoatens	1417	Fruity	0.34 ± 0.58	N.D.	N.D.	N.D.	N.D.	MS, TI
Hexyl butyratens	1418	Apple, fruity	0.15 ± 0.26	N.D.	N.D.	N.D.	N.D.	MS, TI
Hexyl 2-methylbutyrate	1429	Apple, fruity	3.53 ± 1.95b	21.99 ± 11.72a	0.74 ± 0.25b	10.44 ± 8.96ab	1.41 ± 2.02b	FI, MS, TI
Alcohols
Butanol	1156	Fruity	2.37 ± 1.02ab	3.58 ± 1.54a	0.97 ± 0.26ab	2.04 ± 1.17ab	0.09 ± 0.16b	MS, TI
2-Methyl-1-butanol	1213	Wine, fusel-oil like	2.04 ± 0.92b	2.53 ± 1.06b	0.72 ± 0.06b	7.95 ± 2.98a	N.D.b	FI, MS, TI
Hexanol	1360	Green	14.30 ± 4.12b	36.62 ± 9.15a	6.26 ± 1.43b	7.74 ± 2.45b	3.18 ± 0.57b	MS, TI
3-Octanolns	1395	Green	N.D.	N.D.	0.04 ± 0.07	N.D.	N.D.	MS, TI
Heptanolns	1462	Green	N.D.	0.47 ± 0.41	N.D.	N.D.	N.D.	MS, TI
2-Ethyl-1-hexanolns	1491	Floral, sweet	N.D.	N.D.	0.03 ± 0.05	0.10 ± 0.18	N.D.	MS, TI
Octanolns	1561	Green	N.D.	0.14 ± 0.24	N.D.	N.D.	N.D.	FI, MS, TI
Aldehydes
Hexanalns	1098	Green	23.10 ± 1.82	21.18 ± 1.67	24.15 ± 10.28	35.72 ± 12.92	25.95 ± 4.47	FI, MS, TI
(E)-2-Hexenal	1251	Green	8.52 ± 1.58ab	9.12 ± 3.55ab	9.77 ± 0.99ab	5.84 ± 1.13b	12.46 ± 2.88a	MS, TI
Nonanalns	1411	Green	N.D.	N.D.	N.D.	N.D.	0.16 ± 0.28	FI, MS, TI
(E)-2-Octenalns	1464	Green	0.35 ± 0.32	N.D.	0.33 ± 0.07	0.11 ± 0.18	N.D.	MS, TI
Furans
2-Pentylfuranns	1237	Green bean	N.D.	N.D.	0.12 ± 0.21	N.D.	N.D.	FI, MS, TI
Acids
Butyric acidns	1638	Acidic	2.05 ± 3.55	N.D.	N.D.	N.D.	N.D.	MS, TI
Terpenes
(D)-limonenens	1203	Citrus, sweet	0.04 ± 0.06	N.D.	N.D.	N.D.	N.D.	MS, TI
Others
Estragolens	1705	Licorice	0.20 ± 0.35	0.87 ± 0.07	0.88 ± 0.15	0.16 ± 0.27	7.61 ± 7.18	FI, MS, TI
α-Farnesenens	1740	Woody	1.69 ± 0.78	2.89 ± 1.32	3.34 ± 2.59	1.16 ± 0.43	0.14 ± 0.24	MS, TI

The data are expressed as a mean ± standard deviation (n = 3). Different letters (a-b) in the same row indicate statistically significant differences (Tukey’s HSD, p-value < 0.05); ^ARetention indices were determined on HP-FFAP capillary column using n-alkanes (C7−C30) as external references; ^BAroma descriptors from the Cornell University’s Flavornet (http://www.flavornet.org/flavornet.html) and Good Scents Company (http://www.thegoodscentscompany.com/index.html); ^CIdentification with FI (fully identified using authentic standard), MS (mass spectrum consistent with that from the NIST library), TI (tentatively identified based on the NIST library and literature); ^DNot detected; ^nsNot significant.

Collectively, these findings suggest that esters and aldehydes are key contributors to the varietal differences in aroma profiles among apple cultivars. Apple cultivars enriched in fruity esters and green aldehydes, such as ‘Arisoo’ and ‘Picnic’, are more likely to exhibit the complex and desirable sensory attributes commonly associated with consumer-preferred apples, whereas elevated levels of off-flavor alcohols such as 2-methyl-1-butanol may detract from overall flavor preference.

3.6. Relationship between chemical compositions and taste and flavor Attributes in different apple cultivars

In this study, PLS regression was employed to elucidate the relationship between chemical composition (sugar, organic acid, and volatile compounds) and flavor attributes in five apple cultivars (Fig. 5). The regression model demonstrated how non-volatile compounds—including sugars, organic acids, and flavonoids (x variables, n = 20)—and volatile compounds (x variables, n = 28) contributed to the taste and flavor attributes (y variables, n = 11) of the five cultivars. This multivariate approach enabled the visualization of how specific metabolites contribute to perceived flavor characteristics across different cultivars.

Fig. 5.

PLS regression plot of the taste/flavor characteristics and chemical properties of different five apple cultivars. (A) Loading plot based on non-volatile compounds, including sugars, organic acids, and flavonoids. (B) Loading plot based on volatile compounds. The plots illustrate how chemical compositions relate to taste and flavor descriptors among the cultivars.

In the PLS loading plot based on non-volatile compounds (Fig. 5A), ‘Arisoo’ and ‘Arione’ were closely associated with key sensory attributes such as sweetness, sourness, fruity, apple, and pineapple notes. These attributes were positioned in proximity to malic acid, acetic acid, tartaric acid, sucrose, and total acids, suggesting that these organic acids and sugars were major contributors to the perceived taste intensity of ‘Arisoo’ and ‘Arione’. ‘Arione’ also showed a positive association with pear, and honey notes and was positioned near sucrose and acetic acid, indicating a sugar-acid profile that supported its sweet sensory characteristics. However, 'Hongro' was distinctly separated from the other cultivars and was located near glucose, fructose, oxalic acid, citric acid, and several flavonoids including quercetin, phloridzin, and (–)-epicatechin. Although the flavonoids may contribute to antioxidant activity, they were not associated with preferred flavor attributes. Instead, 'Hongro' showed a strong association with cucumber flavor, which may have negatively influenced its overall acceptability. Both ‘Picnic’ and ‘Eazypple’ clustered in the lower half of the PLS loading plot and were associated with sorbitol, luteolin, and chrysin, as well as sensory attributes such as floral, green, and vanilla notes. The non-volatile chemical composition of these two cultivars appears to support a more nuanced flavor profile, in contrast to the dominant sweet and fruity esters observed in ‘Arisoo’. This may contribute to their moderate perception of taste attributes, such as sweetness and sourness, resulting in an overall balanced but less intense flavor character.

The PLS loading plot based on volatile compounds further illustrated distinct flavor profiles among the apple cultivars (Fig. 5B). ‘Arisoo’ and ‘Arione’ were closely associated with numerous fruity esters, including hexyl acetate (apple), butyl acetate (pear), butyl butyrate, amyl acetate (banana), methyl hexanoate (pineapple), and 2-methylbutyl acetate (fruity). These volatiles are well-documented contributors to desirable fruity and sweet aromas and were strongly correlated with the favorable flavor attributes of these cultivars. Notably, ‘Arisoo’ exhibited a volatile profile dominated by ester-type aroma compounds, consistent with its high overall sensory acceptance. On the other hand, ‘Hongro’ was positioned away from favorable sensory attributes and was instead closely associated with 2-methyl-1-butanol and hexanal. These compounds are known to impart off-flavors such as wine-, fusel oil-, and green-like notes. The high concentration of 2-methyl-1-butanol in ‘Hongro' corresponded with its lower overall acceptability, consistent with previous findings that associate this compound with diminished consumer preference. Although ‘Hongro’ was also linked to volatiles typically associated with favorable aromas, such as 2-methylbutyl 2-methylbutyrate (apple) and 2-ethyl-1-hexanol (floral), the impact of these compounds appeared to be relatively minor compared to the dominant off-flavor contributors. Meanwhile, 'Eazypple' and 'Picnic' were moderately associated with floral, green, and vanilla descriptors, along with volatiles such as estragole, nonanal, and (E)-2-octenal. These cultivars exhibited a more balanced flavor profile, lacking in intense fruity esters but enriched in soft aromatic notes, which may explain their intermediate acceptability scores.

4. Conclusions

This study comprehensively evaluated the antioxidant capacity, chemical composition, and sensory attributes of eight apple cultivars grown in Korea. ‘Hongro’, ‘Picnic’, and ‘Eazypple’ exhibited high antioxidant activity, characterized by elevated levels of total polyphenols, flavonoids, and specific antioxidant-related compounds such as (–)-epicatechin, procyanidin B1, quercetin, and phloridzin—particularly concentrated in the peel. In contrast, ‘Arisoo’ and ‘Arione’ received high consumer preference scores due to their favorable sensory profiles, which included high intensities of sweetness, sourness, and fruity/apple-like aroma attributes, despite their relatively low antioxidant capacity. Our results suggest that antioxidant capacity and flavor quality are not necessarily correlated and are governed by distinct compositional drivers. While flavonoid-rich cultivars may offer health-promoting potential, esters and volatile aldehydes were more predictive of consumer-perceived quality and preference. Additionally, conventional physicochemical markers such as TSS and TA were found to be insufficient to fully explain sensory perception, highlighting the importance of integrating compositional and sensory analyses in apple quality evaluation. The key compounds identified—associated with antioxidant activity and flavor perception—may serve as important selection markers in future cultivar development. The present study provides valuable insights for targeted apple breeding and product innovation.

CRediT authorship contribution statement

Keono Kim: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation. Myeongbin Park: Writing – original draft, Methodology, Formal analysis, Data curation. Dagyeong Kwon: Resources, Investigation, Conceptualization. Yu Wang: Writing – review & editing, Writing – original draft, Conceptualization. Jeehye Sung: Writing – review & editing, Writing – original draft, Visualization, Supervision, Project administration, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.