The DPPH and ABTS free radical scavenging abilities and the FRAP total antioxidant capacity of ZJPEO at different concentrations (0.5–10 mg/mL) were determined with reference to the method of Flieger and Flieger (30); Priyanthi and Sivakanesan (31) with slight modifications. The DPPH and ABTS free radical scavenging abilities were expressed as the half-inhibitory concentration (IC₅₀) and the FRAP total antioxidant capacity was expressed as mmol/g. The standard curve was y = 0.7309 x + 0.0366 (R² = 0.9991).

HS-GC-IMS is an emerging analytical technique that enables rapid, highly sensitive detection of VOCs, offering strong separation capability and intuitive visualization of aroma characteristics (43). In the three-dimensional HS-GC-IMS spectrum obtained in this study (Figure 2A), the x-axis, y-axis, and z-axis represent drift time, retention time and signal intensity, respectively. The overall spectral patterns of SDEO and SFEO appeared similar, suggesting comparable VOC compositions; however, variations in peak intensities indicate differences in their relative abundance. Because the three-dimensional spectrum is comparatively coarse, the data were further processed into a two-dimensional top-view plot to facilitate more detailed comparison (Figure 2B). In this plot, the background is blue and a red vertical line at an abscissa value of 1.0 corresponds to the reaction ion peak (RIP). Each signal point denotes a specific VOC, with its color intensity reflecting concentration: white indicates low abundance, red represents high abundance, and darker shades correspond to progressively higher levels. Most VOCs in both samples eluted within a retention time range of 100–1,200 s and a drift time of 1.0–2.0 ms, with those eluting between 100 and 600 s showing higher density, while VOCs between 600 and 1,200 s were more dispersed. This variation is likely due to differences in compound polarity, causing polar and non-polar constituents to exhibit distinct retention times when interacting with the non-polar stationary phase of the chromatographic column (44).

To visualize the differences in VOCs between extraction methods, a two-dimensional GC-IMS difference spectrum was generated (Figure 2C). Here, SDEO was set as the reference and the spectra of other samples were subtracted from it. In the resulting plot, the blue background indicates signals similar to the reference, red denotes higher signal intensity and blue signifies lower intensity relative to the reference. Color depth corresponds to the magnitude of difference. Regions marked blue (a) indicate lower abundance than SDEO, whereas red regions (b) indicate higher abundance. For more detailed comparison, these data can be further transformed into a fingerprint profile.

The results of the VOCs identification and analysis for ZJPEO are presented in Table 1. A total of 69 signal peaks were detected by HS-GC-IMS. By comparing drift times and retention indices against standards, 63 VOCs were identified; the remaining peaks could not be identified due to database limitations, underscoring the need for an expanded HS-GC-IMS database for more comprehensive VOC profiling (24). Notably, some compounds produced multiple signal peaks (e.g., monomers, dimers, trimers), which is attributed to variations in VOCs concentration (45). As shown in Supplementary Figure S2A, the VOCs of SDEO and SFEO were broadly similar, encompassing 22 aldehydes (34.92%), 12 esters (19.05%), 11 terpenoids (17.46%), 7 alcohols (11.11%), 7 other compounds (11.11%), and 4 ketones (6.35%), with a greater variety of compounds detected compared to the FTIR results. As shown in Table 1, the major VOCs of SDEO were ethyl butanoate, 2-methyl methylbutyrate, (E)-2-octenal, mesityl oxide-M and 3-carene-M. In contrast, the major VOCs of SFEO were ethanol, 2-propylpyrazine, octanal-M, butanal-D, and ocimene. The ethanol content in SFEO (4.49%) was significantly higher than that in SDEO (1.11%). According to literature reports, after SFE extraction, the pipeline system and valves of the SFE equipment must be cleaned using solvents (e.g., ethanol, water, or mixtures thereof), which may lead to partial retention of ethanol in the final product (46). This may explain the higher ethanol content observed in SFEO. In the study by Song et al. (47), ethanol levels as high as 9.17%−11.89% were detected by GC-IMS. In contrast, although the ethanol content in SFEO in the present study is higher than that in SDEO, it remains within the commonly reported range. Moreover, high ethanol concentrations can generate strong background signals in E-nose detection, thereby reducing its ability to discriminate odors. It has also been reported that the presence of other components can attenuate the influence of ethanol on the E-nose response (48). Therefore, the overall odor discrimination capability of the E-nose results from the combined effects of multiple volatile constituents. Aldehydes were abundantly extracted in both SDEO (31%) and SFEO (36.83%), with SFEO showing a comparatively higher proportion of aldehyde compounds. Ketone content was significantly lower in SFEO compared with SDEO (P < 0.05; Supplementary Figure S2B). No significant differences were observed among other compound classes, likely because the high temperature during SD may cause degradation of thermolabile compounds (16). In addition, 13 monomeric compounds exhibited significantly different abundances between extraction methods (Supplementary Figure S2C), confirming that extraction parameters influence EOs composition and may consequently alter biological activities. For example: linalool-M was more abundant in SDEO (1.24 times that of SFEO), suggesting greater potential for anticancer and anxiolytic applications (49). In contrast, citronellal-D was more prevalent in SFEO (1.61 times that of SDEO), indicating stronger antimicrobial and repellent functionality, making SFEO more suitable for natural antimicrobial/repellent product development (49, 50). Overall, the extraction method can be strategically selected to modulate ZJPEO composition and optimize its functional properties.

To directly compare the VOC profiles of SDEO and SFEO, fingerprint plots were constructed using the Gallery Plot plugin (Figure 2D). In these fingerprint spectra, each row corresponds to a sample (including three biological replicates) and each column to a specific VOC, with darker colors representing higher relative abundance. The substance in the red box represents alcohols. Among them, linalool-D, 1-hexanol, and 1-octen-3-ol were more abundant in SDEO, whereas ethanol was more higher in SFEO. The substance in the orange box represents aldehydes. Among them, butanal-D, (E)-2-hexenal, (E)-2-methyl-2-pentenal, furfural, isovaleraldehyde, and (Z)-4-heptenal were present at higher levels in SDEO. The marked increase in (E)-2-hexenal may stem from polyunsaturated fatty acid degradation (51). The substance in the yellow box represents ketones, all ketones were substantially higher in SDEO, suggesting that SD may enhance their release and extraction. The substance in the green box represents esters. Propyl acetate, butyl acetate M/D, Methyl 2-methylbutyrate, and methyl acetate were enriched in SDEO, while (Z)-3-hexenyl formate and ethyl formate M/D were more abundant in SFEO. The substance in the purple box represents terpenoids. The two extraction methods showed no significant difference in terpenoid extraction efficiency, with minimal variation in content.

PCA revealed clear clustering between samples (Figure 2E), demonstrating that the two extraction methods generated distinct VOCs profiles. Using a VIP threshold of > 1, five key differential VOCs were identified: nonanal-D, octanal-M, mesityl oxide-M, (E)-2-hexenal, and ethyl formate-M (Figure 2F). These VOCs may serve as chemical markers for distinguishing ZJPEO obtained via different extraction methods. These findings provide a theoretical basis for establishing a scientific quality control and traceability system, thereby supporting standardized production, process optimization and the targeted development of high-value ZJPEO-based products.

HS-SPME-GC-MS was employed to comprehensively identify and characterize the VOCs in ZJPEO. As shown in Figure 3A, a total of 1,250 VOCs were detected, with the total ion chromatogram presented in Supplementary Figure S3. These compounds were categorized into 15 chemical classes (Supplementary Table S2), including 329 terpenoids (26.32%), 203 esters (16.24%), 126 ketones (10.08%), 108 heterocyclic compounds (8.64%), 100 alcohols (8%), 74 aldehydes (5.92%), 69 hydrocarbons (5.52%), 54 phenols (4.32%), 53 acids (4.24%), 38 amines (3.04%), 36 ethers (2.88%), 31 aromatic hydrocarbons (2.48%), 16 nitrogen-containing compounds (1.28%), 7 sulfur-containing compounds (0.56%), and 6 halogenated hydrocarbons (0.48%). HS-SPME-GC-MS analysis showed that terpenoids, esters, and ketones were the most diverse classes of compounds, comprising 658 species (52.64%). In contrast, HS-GC-IMS results indicated that aldehydes, esters, and terpenoids were the major contributors to the volatile aroma profile, accounting for 45 species (71.43%), with aldehydes alone representing 34.92%. The observed discrepancies may be attributed to the differing sensitivities of HS-GC-IMS and HS-SPME-GC-MS techniques in detecting VOCs. HS-GC-IMS exhibits higher sensitivity toward low-molecular-weight and low-boiling-point VOCs, whereas it shows limited response to high-boiling-point compounds (52, 53). When detecting high-boiling-point VOCs via GC-IMS, their low volatility often leads to adsorption within the chromatographic column or transfer lines, thereby compromising detection accuracy (54). In contrast, GC-MS demonstrates superior sensitivity toward high-molecular-weight and high-boiling-point VOCs. Furthermore, studies indicate that GC-MS can identify specific alkanes that remain undetected by GC-IMS due to their low proton affinity, which hampers efficient ionization in IMS (55). Methodologically, the two techniques also differ: GC-MS typically requires complex sample pretreatment procedures, while GC-IMS enables direct headspace sampling, simplifying preprocessing and allowing intuitive visualization of VOC profiles between samples through color-contour imaging (56). Consequently, reliance on a single analytical technique is insufficient for comprehensive VOC characterization. This underscores the necessity of employing complementary analytical approaches to achieve a more complete and accurate representation of the VOC composition in samples (57). Also upgrading and improving existing instrumental methods can enhance their sensitivity and accuracy in chemical composition detection. For example, a photoionization detector (PID) enhanced with a thin zeolite 5A layer can improve the selectivity for fatty acid markers (e.g., palmitic acid) in complex oil matrices, thus providing a fast and sensitive solution for the screening of key markers in complex matrices (58). In addition to this, the complementary sensor-based strategy of this study provides a proven solution to such problems. PCA demonstrated clear separation between SDEO and SFEO samples (Figure 3B), indicating substantial differences in their VOC profiles. This conclusion was reinforced by hierarchical clustering analysis, which highlighted marked variations in VOCs abundance between the two extraction methods (Figure 3C).

To further investigate compositional differences between SDEO and SFEO, an OPLS-DA model was constructed. The score plot exhibited distinct separation between the two groups (Figure 3D), indicating significant metabolic divergence. A permutation test (n = 200) was conducted to validate the OPLS-DA model. With R²X (0.76), R²Y (0.999), and Q² (0.895) all exceeding 0.5, the model demonstrated good reliability and fitting performance, making it suitable for subsequent differential VOCs screening analysis (Figures 3E, F). Using the filtering criteria of FC ≥ 2 or FC ≤ 0.5 and VIP > 1, a total of 334 differential VOCs were identified between SFEO and SDEO (176 upregulated, 158 downregulated) (Figure 3G). Terpenoids constituted the most abundant class among these differential VOCs, with 119 compounds (69 upregulated, 50 downregulated), followed by esters (45; 24 upregulated, 21 downregulated), ketones (34; 14 upregulated, 20 downregulated), and alcohols (26; 10 upregulated, 16 downregulated) (Figure 3H). Notable upregulated terpenoids included kessane, dihydrocarvyl acetate, isopulegol acetate, panaxene, and (-)-spathulenol, while major upregulated esters comprised methyl methanthranilate, (Z)-hex-3-enyl (E)-2-methylbut-2-enoate and 2-isopropenyl-5-methylcyclohexyl acetate. These results indicate that terpenoids and esters are the primary differential VOCs classes distinguishing SFEO from SDEO (Supplementary Table S3). To visualize this, a heatmap of the top 50 esters and terpenoids VOCs by VIP value (Figure 4) confirmed significant differences in their metabolic levels between the two samples, suggesting that these compounds likely contribute to the distinct aroma and bioactivity of ZJPEO.

The KEGG database is a major public pathway repository used to investigate metabolite accumulation within biological networks (59). KEGG pathway annotation was subsequently performed to explore the biological pathways associated with differential VOCs accumulation. Most differential VOCs were associated with sesquiterpenoid and triterpenoid biosynthesis, as well as monoterpenoid biosynthesis (Figure 5A). This was supported by KEGG enrichment analysis, which showed significant enrichment of these VOCs specifically in sesquiterpenoids and triterpenoids biosynthesis pathways (Figure 5B). To elucidate the metabolic routes involved, a terpenoid biosynthesis map was constructed (Figure 5C). CEOs are rich in monoterpenes and sesquiterpenes, which accumulate in specialized oil glands of the peel and in the oil bodies of juice sacs (3). Within the terpenoid biosynthesis pathway, SFEO exhibited higher levels of key differential VOCs. These compounds originate from the terpenoid precursors farnesyl pyrophosphate (FPP) and geranyl pyrophosphate (GPP), supplied by the MVA and MEP pathways. These intermediates are converted by terpene synthases into metabolites such as D-carvone, (-)-carvone, germacrene B, β-selinene, caryophyllene, longifolene, and δ-cadinene, many of which possess anti-inflammatory, antibacterial, antioxidant, analgesic, and anticonvulsant activities (60–62).

Zhoupigan is highly appreciated for its distinctive flavor profile and nutritional properties. VOCs are key determinants of flavor quality and provide objective insight into sensory attributes. To assess the sensory contribution of differential VOCs, all 334 differential VOCs were annotated using established odor databases (https://www.odour.org.uk; https://www.flavornet.org; https://www.femaflavor.org/flavor-library). The 10 most frequently annotated flavor attributes (rOAV ≥ 1) were selected to construct a flavor wheel (Figure 5D). Compared with SDEO, SFEO exhibited a more diverse and intense flavor profile, showing stronger associations with sweet, green, woody, floral, herbal, fruity, waxy, spicy, citrus and balsamic notes. These findings suggest that the distinctive flavor of SFEO results from the combined contribution of its uniquely enriched differential VOCs.

CEOs are rich in natural antioxidants capable of mitigating oxidative stress and related disorders, making them promising alternatives to synthetic antioxidants in the food industry (63). In this study, the antioxidant activities of SFEO and SDEO were evaluated using DPPH, ABTS, and FRAP assays. Both types of ZJPEO displayed concentration-dependent scavenging activities against DPPH and ABTS radicals (Figures 6A, B). The total antioxidant capacity measured by the FRAP assay indicated that SFEO (FRAP = 0.1741 mmol/g) was significantly higher than that of SDEO (FRAP = 0.1135 mmol/g, P<0.05; Figure 6C). For DPPH radical scavenging activity, the IC50 of SFEO was 2.31 mg/mL, while that of SDEO was 3.54 mg/mL, and the IC50 of the positive control (vitamin C) was 5.45 μg/mL; SFEO exhibited stronger DPPH radical scavenging activity, which was significantly superior to that of SDEO (P<0.05), though both were lower than vitamin C. Similarly, the IC50 of SFEO for ABTS radical scavenging was 3.71 mg/mL, that of SDEO was 4.14 mg/mL, and that of the positive control (vitamin C) was 52.16 μg/mL; the ABTS radical scavenging activity of SFEO was also significantly higher than that of SDEO (P<0.05; Figure 6D). Overall, SFEO consistently demonstrated superior antioxidant activity compared to SDEO. We speculate that prolonged exposure to elevated temperatures during steam distillation may induce the oxidation or degradation of thermolabile constituents, subsequently reducing the antioxidant capacity of SDEO (64). The antioxidant properties of CEOs are known to depend on both the abundance and composition of bioactive constituents (9), with oxygenated monoterpenes contributing substantially to enhanced antioxidant potential (65). The present findings confirm that ZJPEO possesses notable antioxidant activity and represents a promising natural antioxidant source. Moreover, the extraction method significantly influences antioxidant performance, with SFEO exhibiting stronger activity than SDEO. These results highlight SFEO's potential as a natural food preservative or functional skincare ingredient, supporting its further industrial development.

CEOs exhibit broad-spectrum antibacterial activity in vitro, although the extent of inhibition varies greatly depending on citrus species and extraction techniques (66, 67). To date, no studies have reported on the antibacterial properties of ZJPEO. In this study, antibacterial activity was assessed using the Oxford cup method and micro-broth dilution assays. As shown in Table 2 and Figure 6E, both SDEO and SFEO inhibited the growth of Staphylococcus aureus (S. aureus) and Escherichia coli (E. coli). The inhibitory effect was significantly stronger against Gram-positive S. aureus than Gram-negative E. coli (P<0.05). This difference may be attributed to structural variations in the bacterial cell wall: Gram-negative bacteria possess a hydrophilic lipopolysaccharide (LPS) outer membrane that impedes the penetration of hydrophobic EOs components, whereas Gram-positive bacteria lack this barrier and are therefore more susceptible to EO-induced membrane disruption, cytoplasmic leakage, and metabolic inhibition (68). The extraction method also significantly affected antibacterial efficacy. Based on inhibition zone diameters, antibacterial potency followed the order SDEO < SFEO. MIC and MBC values (Table 3) further confirmed the stronger activity of SFEO. For S. aureus, SFEO exhibited MIC and MBC values of 0.125 μL/mL and 2 μL/mL, respectively, whereas SDEO exhibited 4 μL/mL and 16 μL/mL. Against E. coli, SFEO showed MIC and MBC values of 4 μL/mL and 32 μL/mL, compared with 16 μL/mL and 64 μL/mL for SDEO. It is worth noting that, as previously mentioned, the EOs may contain trace residual ethanol. According to literature reports, ethanol concentrations above 45% can affect bacterial growth and survival (69). However, the maximum concentration of EOs used in our experiments was 512 μL/mL, which corresponds to an ethanol content of approximately 2%, well below the minimum inhibitory concentration reported in the literature. Therefore, the antibacterial contribution from ethanol in the EOs of this study is indeed very limited. Collectively, SFEO demonstrated significantly stronger antibacterial activity against both bacterial strains than SDEO (P<0.05). This suggests that SFE more effectively extracts or preserves antimicrobial constituents compared to SD. From an industrial perspective, this finding holds significant practical value. Although SFE technology involves relatively high equipment investment, its green production advantages, including solvent recycling and residue-free extraction, contribute to higher resource efficiency in large-scale production systems (70, 71). The notable antibacterial activity of SFEO, coupled with the technical feasibility of large-scale continuous production and the trend toward equipment standardization, collectively enhances its potential for industrialization in applications such as personal care products and pharmaceuticals with antibacterial functions. Of course, in view of the complexity of the EOs composition, further clarification of its specific antimicrobial components and mechanism of action is still needed to support its industrial development and standardized production.

To explore the associations between bioactivities and differential VOCs, a Pearson correlation analysis was performed. The analysis included antioxidant activities (IC50 values used for ABTS and DPPH, FRAP expressed as total antioxidant capacity), antibacterial activities (MIC-JP for S. aureus and MIC-DC for E. coli), and the top 50 terpenoids and esters (ranked by VIP values). As shown in Figure 7, 24 terpenoid components, including kessane, aristolochene, and longifolene, exhibited significant negative correlations with DPPH and ABTS, but significant positive correlations with FRAP total antioxidant capacity. Additionally, 22 esters, such as 3,6-nonadien-1-yl acetate, cis-methyl 3-phenyl-2-propenoate, and isobornyl propionate, showed significant negative correlations with DPPH. Twenty-three esters also displayed negative correlations with ABTS and positive correlations with FRAP. These discrepancies may arise from the structural complexity of VOCs and the distinct reaction mechanisms underlying the DPPH, ABTS, and FRAP assays (72). Regarding antibacterial activity, terpenoids such as thymol, linalool, and linalyl acetate, as well as esters including furfuryl heptanoate, cis-3-hexenyl salicylate and 1-octen-3-yl acetate, showed significant positive correlations with MIC-JP, suggesting strong contributions to the antibacterial activity against S. aureus (73, 74). These results agree with previous studies demonstrating the antimicrobial efficacy of thymol and linalool. Conversely, some VOCs exhibited negative correlations with antibacterial activity against E. coli. For instance, terpenoids such as aristolochene, γ-elemene and α-trans-bergamotene, and esters such as (Z)-hex-3-enyl (E)-2-methylbut-2-enoate, 2-isopropenyl-5-methylcyclohexyl acetate and ethyl mandelate showed significant negative correlations with MIC-DC. This study is the first to systematically elucidate the correlation between VOC composition in ZJPEO extracted by different methods and its bioactivities. These findings provide important insights for uncovering the mechanisms of active constituents and identifying novel bioactive compounds, offering valuable guidance for the development of new pharmaceuticals and functional cosmetics.