Skip to main content
NCBI home page
Food Chemistry: X logoLink to Food Chemistry: X
. 2026 Sep 6;39:104398. doi: 10.1016/j.fochx.2026.104398

How light spectra reshape the volatile metabolome and aroma profile of celery: natural greenhouse light vs. artificial LED light

Haijie Dou a,1, Yongsheng Liu b,1, Zejin Zhang c,1, Zidan Wen b, Zhixin Li b, Jinxiu Song d, Na Lu e, Yanjie Yang b, Zhengnan Yan b,⁎
PMCID: PMC13579921  PMID: 42751486

Abstract

Light spectral quality strongly influences volatile formation, yet its effects on celery petioles remain unclear. Celery petioles grown under natural greenhouse light (CK), warm-white (3000K), cool-white (6500 K), and red-blue (RB) LEDs were evaluated using HS-SPME-GC–MS-based volatile metabolomics and relative odor activity values. A total of 1789 VOCs were detected and tentatively identified, dominated by terpenoids, esters, ketones, heterocyclic compounds, and alcohols. Among the four light treatments, RB produced the clearest multivariate separation and the largest number of up-accumulated aroma-related volatiles, whereas the two white LEDs generated comparatively similar profiles. Red-blue light responsive volatiles were associated with terpenoid, phenylalanine, amino acid, and central carbon metabolism. The rOAV analysis highlighted (5Z)-octa-1,5-dien-3-one, 2-methoxy-3,5-dimethylpyrazine, limonene, and ocimene isomers as candidate odor-active markers of celery-petiole aroma. These results support red-blue spectral management for flavor-oriented celery production in controlled environments.

Keywords: Apium graveolens L., Volatile metabolomics, Light spectral regulation, Odor-active compounds, Flavor quality

Highlights

  • •

    A 1789-compound celery-petiole volatilome was tentatively identified under four light regimes.

  • •

    Warm- and cool-white LEDs yielded similar volatile profiles.

  • •

    Red-blue light generated the strongest separation in aroma-related volatiles.

  • •

    rOAV revealed candidate odor-active markers of celery-petiole aroma.

  • •

    Volatile responses support aroma-oriented spectral management for celery production.

1. Introduction

Celery (Apium graveolens L.) is a widely consumed leafy vegetable valued for its nutritional attributes and distinctive aroma and flavor, which strongly influence consumer preference and market quality (Yan et al., 2024a). Celery possesses a characteristic flavor profile defined by a complex mixture of volatile organic compounds (VOCs) that contribute herbal, green, citrus, and spicy notes (Gold & Wilson, 1963; Tang et al., 2025). These VOCs are therefore key determinants of sensory quality in both fresh and processed celery products. Flavor quality has become an increasingly important attribute in fresh vegetables as consumers demand produce that combines nutritional value with desirable sensory characteristics (Ji et al., 2025). Consequently, understanding the biochemical basis and environmental regulation of aroma-related VOCs has become a major research focus in horticultural commodities.

The volatile composition of celery has been extensively characterized and is known to comprise multiple chemical classes, including terpenoids, alcohols, aldehydes, ketones, esters, and heterocyclic compounds (Gold & Wilson, 1963; Li et al., 2018; Sun et al., 2023; Yan et al., 2024b). Terpenoid derivatives such as limonene, β-myrcene, and β-ocimene, are key contributors to citrus, herbal, and fresh sensory characteristics, while lipid-derived C6 aldehydes and alcohols, often referred to as green leaf volatiles (GLVs), provide fresh and green sensory notes (Fan et al., 2023; Yan et al., 2024a). In addition, nitrogen- and sulfur-containing compounds and certain ketones contribute roasted, nutty, or earthy nuances. These studies demonstrate that celery aroma is determined by the combined effects of multiple volatile classes rather than a single compound group. Moreover, celery VOC profiles are often tissue-specific and can be strongly influenced by genotype, environmental conditions, and post-harvest handling (Ji et al., 2025; Sun et al., 2023).

Environmental conditions play a crucial role in shaping plant VOC composition by modulating both primary and secondary metabolic pathways (Mao et al., 2019; Turner et al., 2021; Zhou et al., 2022). Among these factors, light is particularly influential because it simultaneously drives photosynthesis and regulates metabolic signaling networks (Dou et al., 2025; Guo et al., 2025; Ji et al., 2025; Li et al., 2023). For example, supplementary lighting duration has been reported to increase overall volatile content in celery petioles, especially monoterpenes such as β-myrcene, β-pinene, limonene, β-ocimene, and γ-terpinene (Yan et al., 2025a), suggesting that lighting regimes significantly influence celery aroma. More recently, light spectral composition (light quality) has emerged as a major regulator of plant secondary metabolism (Li et al., 2023; Yu et al., 2025). A recent review further emphasized that light spectral composition regulates terpenoid biosynthesis, abundance, and diversity, thereby influencing plant flavor and quality traits (Contreras-Avilés et al., 2024). Studies in horticultural crops such as basil, lettuce, tomato, and tea have shown that specific wavelength combinations, especially red-blue LED spectra, can enhance the accumulation of terpenoids, phenylpropanoids, and lipid-derived volatiles associated with sweet, herbal, and floral aromas (Fan et al., 2023; Fu et al., 2015; Kivimäenpää et al., 2022; Meng et al., 2024; Yue et al., 2020). In addition, light-based technologies have also been shown to modulate volatile composition and sensory attributes in processed food systems, such as fermented mulberry juice, where pulsed light treatment altered key aroma compounds and flavor perception (Kwaw et al., 2018; Ren et al., 2021). For example, red and blue light wavelengths upregulate terpene synthase (TPS) genes, key enzymes in terpenoid biosynthesis, thereby enhancing the accumulation of monoterpenes and sesquiterpenes in tea leaves (Fu et al., 2015). Similarly, the phenylpropanoid pathway, derived from phenylalanine via phenylalanine ammonia-lyase (PAL), produces benzenoid and phenylpropanoid volatiles associated with floral and fruity aromas, and its activity is also light-responsive (Sreelakshmi & Sharma, 2008). In celery, preliminary evidence suggests that supplemental lighting duration and intensity can influence volatile abundance, particularly monoterpenes and GLVs (Yan et al., 2025a). It was also reported that a 3R1B light treatment increased the activities of flavonoid-synthesizing enzymes such as flavonoid synthase (FNS), chalcone isomerase (CHI), and anthocyanidin synthase (ANS), along with upregulation of their corresponding genes compared to white or monochromatic light (Tang et al., 2025). Collectively, these findings indicate that light spectral composition can regulate metabolic flux through different branches of secondary metabolism and consequently reshape plant volatile profiles.

In modern controlled environment agriculture (CEA), particularly in greenhouses and plant factories with artificial lighting (PFALs), light-emitting diode (LED) systems are widely adopted due to their high energy efficiency and flexible spectral control (Song et al., 2022). In practical production systems, broad-spectrum white LEDs with different correlated color temperatures (CCT), such as warm-white (around 3000 K) and cool-white (around 6500 K), are commonly used as primary lighting sources. These light sources differ substantially in their spectral distribution, especially in the proportions of blue and red wavelengths, which may influence plant metabolic processes and flavor formation. Previous studies have characterized celery volatile profiles across tissues and cultivars and examined their responses to ultraviolet radiation and supplementary light duration (Yan et al., 2024a; Yan et al., 2025b; Zhang et al., 2025). However, the volatile and odor-active profiles of the edible petiole under commonly used warm-white, cool-white, and red-blue LED spectra remain poorly characterized.

To address these knowledge gaps, the present study conducted a comprehensive volatile metabolomics analysis of celery petioles grown under natural greenhouse light (CK), 3000 K white LED, 6500 K white LED, and a red-blue (RB) LED spectrum. By integrating HS-SPME-GC–MS-based volatile profiling, multivariate analysis, differential-metabolite screening, KEGG pathway mapping, and relative odor activity value (rOAV) evaluation, this work aimed to (i) characterize changes in the celery-petiole volatilome under different light environments; (ii) identify candidate odor-active compounds contributing to aroma differentiation; and (iii) identify metabolic pathways associated with treatment-responsive volatile accumulation. The findings provide celery-specific evidence for light-mediated regulation of petiole aroma and support flavor-oriented spectral management in CEA production.

2. Materials & methods

2.1. Plant materials and growth conditions

The celery (Apium graveolens L.) cultivar ‘Dayehuang’ was used as the experimental material in this study. Seeds were sown in 128-cell plug trays filled with a substrate mixture consisting of peat, vermiculite, and perlite at a volumetric ratio of 3:1:1. Seedlings were cultivated in a controlled-environment growth chamber at Qingdao Agricultural University. When the seedlings reached the five-true-leaf stage, uniform plants were selected and transplanted into 1.1-L plastic pots (13.5 cm in diameter and 11 cm in depth) containing the same substrate mixture (peat: vermiculite: perlite = 3:1:1, v/v/v). The plants were then grown under controlled environmental conditions for subsequent light-quality treatments.

Four light treatments were applied in this experiment, including natural greenhouse sunlight (CK) at Qingdao Agricultural University, and three artificial lighting treatments in a controlled-environment growth chamber, specifically, warm-white LED (3000 K), cool-white LED (6500 K), and combined red-blue LED (RB). All LED light sources were provided by Zhongshan Aier Lighting Technology Co., Ltd. (Zhongshan, China). During the experimental period, the mean daily light integral (DLI) in the greenhouse (CK) was approximately 5 mol·m−2·d−1. For the three LED treatments (3000 K, 6500 K, and RB), canopy photosynthetic photon flux density (PPFD) was maintained at 200 μmol·m−2·s−1 for 12 h·d−1, corresponding to a DLI of 8.6 mol·m−2·d−1. The spectral properties (Table 1) and spectral distribution (Fig. 1) of the LED light sources were measured at the plant canopy level using a spectrometer (PG100N, United Power Research Technology Corporation, Miaoli, China).

Table 1.

Spectral characteristics and photon flux parameters of light emitting diode (LED) treatments.

Parameter Lighting Source
3000 K 6500 K RB
Fraction z (%)
PPFD y (400 to 700 nm)		 100 100 100
Blue light (400 to 499 nm) 12.0 28.8 32.6
Green light (500 to 599 nm) 42.7 48.7 0.0
Red light (600 to 700 nm) 45.3 22.5 67.4
Light quality
R:B ratio x 		3.78 0.78 2.07
Open in a new tab
z

Data are photon flux-based proportions of blue, green, and red lights.

y

PPFD, photosynthetic photon flux density.

x

R:B ratio, red to blue light ratio.

Fig. 1.

Fig. 1

Open in a new tab

Spectral photon flux distribution of the light emitting diode (LED) treatments. Relative spectral photon flux density (SPFD) profiles of warm white (3000 K), cool white (6500 K), and red-blue (RB) LED light sources measured at canopy level. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The air temperature in the greenhouse and growth chamber was maintained at 25 ± 1 °C during the light period and 18 ± 1 °C during the dark period, relative humidity was controlled at 60–70%, and CO2 concentration was not artificially regulated. Plants were irrigated with half-strength Hoagland nutrient solution every two days.

2.2. Sample preparation and extraction

Each light treatment was replicated in three blocks, with 40 celery plants per treatment. After 40 d of light treatment, one composite biological replicate was collected from each block. Each biological replicate consisted of petiole tissues pooled from five randomly selected plants within the corresponding block, resulting in three independent composite biological replicates per treatment (n = 3). No plant was included in more than one replicate. Petioles were collected from fully expanded leaves at a uniform canopy position, immediately frozen in liquid nitrogen, and stored at −80 °C until further analysis.

Sample preparation and volatile extraction were performed by Wuhan Metville Biotechnology Co., Ltd. Each biological replicate was separately ground to a fine powder under liquid nitrogen, and a 500 mg aliquot of powdered tissue was transferred into a headspace vial. Subsequently, 3-hexanone-2,2,4,4-d4 solution was added as an internal standard. Headspace solid phase microextraction (HS-SPME) was used for automatic sample extraction. The extraction procedure was based on previous studies (Dai et al., 2018; Wang et al., 2023a; Wang et al., 2023b).

2.3. Determination of volatile compounds by HS-SPME-GC–MS

Volatile compounds were analyzed by HS-SPME-GC–MS in scheduled selective ion monitoring mode (Mei et al., 2023; Wang et al., 2023a; Wang et al., 2023b). GC–MS analysis was performed on an Agilent 8890-7000D equipped with a capillary column (30 m × 0.25 mm × 0.25 μm, DB-5MS, Agilent). The temperature of the injector was maintained constant (250 °C). The oven temperature was programmed from 40 °C (3.5 min), and the increase method in temperature was based on previous studies (Wang et al., 2023a; Wang et al., 2023b). Helium was used as the carrier gas at 1.2 mL min−1. The ion source, quadrupole, and transfer-line temperatures were set to 230, 150, and 280 °C, respectively. Mass spectra were acquired in electron ionization (EI) at 70 eV over an m/z range of 35–500. Compounds were tentatively identified by matching their retention times and characteristic ions against a curated widely targeted volatile database established using multispecies datasets, published literature, NIST 2020 mass spectra, authentic-standard-derived library records, and retention-index information (Yuan et al., 2022, Yuan et al., 2024). Relative semi-quantification was performed using the quantifier ion and an internal standard.

2.4. Screening of differentially accumulated metabolites

Orthogonal partial least-squares discriminant analysis (OPLS-DA) was performed for pairwise comparisons to characterize treatment discrimination. VIP values were derived from the OPLS-DA models and represented the contribution of individual metabolites to group separation, whereas fold change represented the magnitude of the treatment response. An initial screening of treatment-responsive volatile metabolites was performed using VIP > 1, and |log2 Fold change| ≥ 1 according to Yan et al. (2025b). VIP values were derived from the OPLS-DA models and reflected the contribution of each metabolite to treatment discrimination, whereas fold change represented the magnitude of the treatment response. For the more stringent candidate DAM set used in the volcano plots and subsequent analyses, a p < 0.05 from pairwise univariate comparisons was additionally required.

2.5. Calculation of odor activity values (rOAV)

To prioritize volatile compounds potentially contributing to celery-petiole aroma, relative odor activity values (rOAVs) were calculated using semi-quantitative concentrations estimated from GC–MS responses normalized to the deuterated internal standard. Odor thresholds were obtained from Volatile Compounds in Food (VCF) online, 2026 (https://www.vcf-online.nl/) and published literature. Compounds with rOAV >1 were considered candidate odor-active compounds contributing to celery-petiole aroma.

2.6. Statistical analysis

Principal component analysis (PCA), hierarchical cluster analysis (HCA), orthogonal partial least-squares discriminant analysis (OPLS-DA), Venn diagram, volcano plot, K-means clustering, flavor-wheel visualization, and Sankey diagrams were performed using the Metware Cloud platform (https://cloud.metware.cn). The annotated volatile metabolites were matched against the KEGG Compound database (https://www.kegg.jp/kegg/compound/), and mapped to metabolic pathways in the KEGG Pathway database (https://www.kegg.jp/kegg/pathway.html) for pathway enrichment analysis. Data are presented as the mean ± SD of three independent biological replicates (n = 3).

3. Results

3.1. Volatile metabolite profiling

The volatile metabolite profile of celery petioles displayed pronounced chemical diversity. Terpenoids and esters represented the dominant classes, together accounting for a major proportion of the total volatiles, followed by ketones, heterocyclic compounds, and alcohols. In addition, various oxygen-containing volatiles, including aldehydes, acids, and phenolic compounds, were detected, along with minor contributions from amines, hydrocarbons, and sulfur-containing compounds, reflecting the complexity of the celery petiole volatilome (Fig. 2a). Clear separation among the four treatments was observed based on PCA of the overall metabolite dataset. The first two principal components explained 54.87% (PC1) and 27.91% (PC2) of the total variance, respectively, indicating that a substantial proportion of the metabolic variability could be captured by these two axes (Fig. 2b). Samples from each group clustered closely. The 3000 K and 6500 K treatments showed partially overlapping distributions, whereas the RB treatment was distinctly separated from the other three groups along the principal components, indicating that samples under red-blue light had a distinct volatile metabolic profile. Consistent clustering patterns were observed in the HCA, supporting the separation trends identified by PCA and indicating that light spectral quality systematically influenced the petiole volatilome (Fig. 2c). Pairwise comparisons identified substantial differences in DAMs among treatments (Fig. 2d). The initial VIP- and fold-change-based screening identified 772, 729, and 585 candidate DAMs in RB vs. 3000 K, RB vs. CK, and RB vs. 6500 K, respectively, of which the majority were upregulated (91.5%, 72.0%, and 89.2%, respectively). In contrast, both 3000 K and 6500 K treatments mainly induced downregulated DAMs compared with CK, accounting for 85.5% of 408 DAMs and 59.5% of 439 DAMs, respectively. The smallest number of DAMs was observed between 3000 K and 6500 K (193 in total), among which 165 were upregulated under 6500 K.

Fig. 2.

Fig. 2

Open in a new tab

Global profiling of volatile metabolites in celery petioles under different light spectral treatments. (a) Classification and relative composition of tentatively identified volatile metabolites. (b) Principal component analysis (PCA) score plot based on the tentatively identified volatile metabolites. (c) Hierarchical clustering analysis (HCA) of metabolite abundance across treatments. (d) Number of differentially accumulated metabolites (DAMs) identified in pairwise comparisons.

3.2. Screening of differentially accumulated metabolites

To further screen and characterize DAMs between treatments, supervised multivariate and univariate statistical analyses were performed. Clear discrimination between each pair of treatments was observed in the OPLS-DA score plot, with samples from each group clustering tightly, indicating good reproducibility within groups and pronounced metabolic differences between treatments (Fig. 3a-3f). The robustness of the OPLS-DA model was further validated by permutation testing (n = 200). The original model exhibited substantially higher R2Y (R2Y = 1) and Q2 (Q2 varies from 0.959 to 0.996) values than those obtained from randomly permuted models (p < 0.005), demonstrating that the observed group separation was statistically reliable and not due to overfitting (Fig. 3g-3l). Based on the combined criteria of VIP > 1, fold change ≥2 or ≤ 0.5, and p < 0.05, a total of 725, 762, and 573 candidate DAMs were identified in RB vs. CK, RB vs. 3000 K, and RB vs. 6500 K, respectively, predominantly characterized by up-accumulated metabolites. The comparison between 3000 K and 6500 K yielded the fewest DAMs of 189 (Fig. 3m-3r).

Fig. 3.

Fig. 3

Open in a new tab

Multivariate analysis and differential screening of differentially accumulated metabolites (DAMs) under different light spectral treatments. (a-f) Orthogonal partial least squares discriminant analysis (OPLS-DA) score plots for pairwise comparisons, 3000 K vs. CK, 6500 K vs. CK, RB vs. CK, 6500 K vs. 3000 K, RB vs. 3000 K, and RB vs. 6500 K. (g-l) Permutation tests validating OPLS-DA models. (m-r) Volcano plots showing DAMs screened using VIP > 1, FC ≥ 2 or ≤ 0.5, and p < 0.05, 3000 K vs. CK, 6500 K vs. CK, RB vs. CK, 6500 K vs. 3000 K, RB vs. 3000 K, and RB vs. 6500 K.

3.3. Class-level distribution patterns of DAMs

To further characterize how light spectral quality reshaped the volatile metabolite landscape at the chemical class level, DAMs were grouped according to compound categories (Fig. 4). Clear treatment-dependent trends were observed in both the magnitude and direction of changes across volatile classes. Comparisons between 3000 K and CK showed a predominance of down-regulated compounds across most chemical classes, including aldehydes, terpenoids, aromatics, esters, heterocyclic compounds, alcohols, and phenols, indicating an overall suppression of multiple volatile groups under this spectrum. A similar but less uniform pattern was observed in 6500 K vs CK, where both up- and down-regulated compounds were present, suggesting moderate restructuring rather than broad suppression.

Fig. 4.

Fig. 4

Open in a new tab

Distribution of differentially accumulated volatile metabolites across chemical classes under different light spectral comparisons.

RB-related comparisons displayed a strikingly different pattern. The RB vs CK comparison showed substantial increases in terpenoids, esters, heterocyclic compounds, ketones, alcohols, and phenols, with up-regulated compounds far outnumbering down-regulated ones in most categories. Similar trends were observed in RB vs 3000 K and RB vs 6500 K comparisons, where terpenoids, esters, heterocyclic compounds, phenols, and aldehydes were predominantly up-regulated. These results indicate that RB illumination promoted a broad enhancement of multiple aroma-related volatile classes rather than selective changes in isolated compounds.

The comparison between the two white light treatments (6500 K vs 3000 K) exhibited the smallest number of DAMs and more balanced changes across classes, suggesting that differences in CCT alone induced relatively minor reorganization of the volatile metabolome.

3.4. KEGG functional annotation and enrichment analysis of DAMs

To further characterize the dynamic accumulation patterns of DAMs under different light treatments, K-means clustering analysis was conducted. The DAMs were classified into seven distinct subclasses based on their standardized accumulation profiles across CK, 3000 K, 6500 K, and RB treatments (Fig. 5). Among these subclasses, subclass 5 and 7 comprised the largest numbers of DAMs and displayed pronounced accumulation under the RB treatment, whereas other subclasses exhibited gradual or treatment-specific decreases across light conditions. These distinct clustering patterns indicate coordinated yet divergent metabolic responses to variations in light spectral composition.

Fig. 5.

Fig. 5

Open in a new tab

K-means clustering of differentially accumulated metabolites (DAMs) under different light spectral treatments. The y-axis represents standardized metabolite abundance (Z-score), and the x-axis represents individual samples.

To further elucidate the functional relevance of these DAMs, KEGG pathway enrichment analysis was performed (Fig. 6). The DAMs were mainly enriched in pathways related to secondary metabolism and amino acid metabolism, as well as central carbon metabolism. Notably, phenylalanine metabolism, biosynthesis of secondary metabolites, and sesquiterpenoid and triterpenoid biosynthesis were repeatedly enriched with relatively high rich factors and DAM counts, particularly in comparisons involving the RB treatment, highlighting these pathways as major metabolic modules associated with light-induced metabolic differentiation.

Fig. 6.

Fig. 6

Open in a new tab

Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis of differentially accumulated metabolites (DAMs). (a-f) Enriched metabolic pathways identified in pairwise comparisons, 3000 K vs. CK, 6500 K vs. CK, RB vs. CK, 6500 K vs. 3000 K, RB vs. 3000 K, and RB vs. 6500 K.

3.5. Descriptor-based aroma profiling of differential metabolites

The effects of different light spectra on the aroma profile of celery petioles were comprehensively evaluated using flavor wheel visualization, Sankey diagram analysis, and quantitative profiling of aroma-related metabolites. The flavor wheel diagrams revealed treatment-dependent differences in the overall distribution of sensory attributes. Comparisons among CK, 3000 K, and 6500 K treatments showed relatively similar aroma structures, with green, fresh, and fruity notes constituting the major components and only moderate variations in their relative proportions (Fig. 7a, b, e). In contrast, comparisons involving the RB treatment displayed more pronounced redistribution of sensory attributes, characterized by noticeable shifts in the relative contributions of sweet, fruity, woody, and citrus-related notes (Fig. 7c, d, f). These structural changes in the flavor wheel indicate that the inclusion of red-blue light was associated with a marked reorganization of the volatile-derived aroma-descriptor profile.

Fig. 7.

Fig. 7

Open in a new tab

Descriptor-based aroma profiling of differential volatile metabolites under different light spectral treatments. (a-f) Flavor wheels representing aroma profiles derived from differentially accumulated metabolites, 3000 K vs. CK, 6500 K vs. CK, RB vs. CK, 6500 K vs. 3000 K, RB vs. 3000 K, and RB vs. 6500 K, respectively. (g-l) Sankey diagrams illustrating the relationships between volatile compounds and associated aroma attributes, 3000 K vs. CK, 6500 K vs. CK, RB vs. CK, 6500 K vs. 3000 K, RB vs. 3000 K, and RB vs. 6500 K, respectively.

The Sankey diagrams further illustrated the associations between individual volatile compounds and their corresponding sensory descriptors. While many volatile compounds contributed to similar sensory categories under CK and the two white-light treatments, RB-related comparisons showed a denser and more complex network of compound-attribute connections (Fig. 7g-7l). Several candidate volatile compounds were associated with multiple sensory categories under RB treatment, suggesting that the altered aroma profile was supported by a broader redistribution of volatile contributors rather than changes in a few isolated compounds. In contrast, the comparison between 3000 K and 6500 K showed the most similar connection patterns, indicating minimal differentiation in compound-level sensory contributions between these two white-light conditions.

3.6. Candidate odor-active compounds driving aroma differentiation under different light spectra

To further quantify these differences, a heatmap of differentially accumulated metabolites associated with aroma attributes was generated (Fig. 8, Fig. 9, Fig. 10). The heatmap showed that comparisons between CK and the white LED treatments (3000 K and 6500 K) exhibited relatively modest changes in metabolite abundance (Fig. 8), with most compounds displaying comparable accumulation patterns. In contrast, comparisons involving the RB treatment (CK vs RB, 3000 K vs RB, and 6500 K vs RB) displayed more extensive shifts in metabolite levels, as reflected by larger regions of contrasting high and low signal intensities (Fig. 9, Fig. 10). Notably, the comparison between 3000 K and 6500 K showed the most homogeneous pattern (Fig. 9), indicating minimal quantitative divergence in candidate aroma-related metabolites between these two treatments.

Fig. 8.

Fig. 8

Open in a new tab

Heatmaps of candidate odor-active differentially accumulated metabolites in celery petioles for 3000 K vs. CK and 6500 K vs. CK. Aroma-active compounds were screened based on rOAV >1 and ranked according to rOAV values.

Fig. 9.

Fig. 9

Open in a new tab

Heatmaps of candidate odor-active differentially accumulated metabolites in celery petioles for RB vs. CK and 6500 K vs. 3000 K. Aroma-active compounds were screened based on rOAV >1 and ranked according to rOAV values.

Fig. 10.

Fig. 10

Open in a new tab

Heatmaps of candidate odor-active differentially accumulated metabolites in celery petioles for RB vs. 3000 K and RB vs. 6500 K. Aroma-active compounds were screened based on rOAV >1 and ranked according to rOAV values.

To prioritize volatile compounds potentially contributing to the aroma profile, rOAVs were calculated from relative concentrations and reported odor thresholds. A subset of compounds exhibited high rOAV values across treatments, indicating their strong sensory relevance. Among these, (5Z)-octa-1,5-dien-3-one showed the highest rOAV values in all treatments (Table 2), with a pronounced increase under the RB condition, suggesting a major contribution to metallic and green sensory nuances. Similarly, 2-methoxy-3,5-dimethylpyrazine displayed consistently high rOAV values, particularly under CK and RB treatments, indicating its importance in roasted and bread-like aroma attributes. Several terpenoid and furanone compounds, including 3-buten-2-one, 4-(2,6,6-trimethyl-1-cyclohexen-1-yl)- and 2(5H)-furanone, 5-ethyl-3-hydroxy-4-methyl-, also exhibited high rOAV values and showed treatment-dependent variation, suggesting their involvement in floral, sweet, and caramel-like notes.

Table 2.

Representative candidate odor-active compounds in celery petioles under different light treatments, including natural greenhouse light (CK), warm-white LED (3000 K), cool-white LED (6500 K), and red-blue LED (RB). Values are presented as the mean ± SD of three independent biological replicates (n = 3). Corresponding literature sources of odor thresholds are indicated by superscript letters.

Compounds Class Formula CAS Odor Threshold rOAV
CK 3000 K 6500 K RB
(5Z)-Octa-1,5-dien-3-one Ketone C8H12O 65767-22-8 geranium, metallic 3.0 × 10–6 a 5872099 ± 576720 4756883 ± 344994 5952295 ± 433248 14382683 ± 363314
2-Methoxy-3,5-dimethylpyrazine Heterocyclic compound C7H10N2O 92508-08-2 bread, mousy 4.0 × 10–7 a 915089 ± 74421 321222 ± 24110 337739 ± 20010 906370 ± 19125
2(5H)-Furanone, 5-ethyl-3-hydroxy-4-methyl- Ketone C7H10O3 698-10-2 sweet, fruity, caramel, maple, fenugreek, brown, sugar, nutty, chicory, praline, butterscotch 2.0 × 10–6 a 466226 ± 21902 453458 ± 29,986 425179 ± 8219 351049 ± 16,186
3-Mercapto-3-methylbutyl formate Ester C6H12O2S 50746-10-6 sulfury, catty, caramel, onion, roasted coffee, roasted meat, tropical 2.0 × 10–6 a 450460 ± 19174 311066 ± 10315 257091 ± 7041 407355 ± 14,648
2,4-Undecadienal Aldehyde C11H18O 13162-46-4 green, buttery, spicy, baked, fruity, fatty, aldehydic, chicken 1.0 × 10–5 a 277125 ± 16866 433973 ± 2654 430704 ± 6840 894941 ± 45681
3-Octen-2-one Ketone C8H14O 1669-44-9 earthy, spicy, herbal, sweet, mushroom, hay, blueberry 3.0 × 10–5 b 181300 ± 10417 146352 ± 4635 148695 ± 3424 255457 ± 5586
Limonene Terpenoids C10H16 138-86-3 citrus, herbal, terpene, camphor 1.0 × 10–2 c 5709 ± 320 6085 ± 198 8087 ± 231 11631 ± 323
trans-β-Ocimene Terpenoids C10H16 3779-61-1 sweet, herbal 3.4 × 10–2 d 4234 ± 250 3337 ± 124 3355 ± 105 6091 ± 135
Open in a new tab

Note: Representative compounds were selected from Supplementary Table S1 based on their relatively high rOAVs, treatment-dependent variation, coverage of major odor-active chemical classes and aroma descriptors, and relevance to the characteristic aroma of celery petioles. The complete list of 30 candidate odor-active compounds is provided in Supplementary Table S1. rOAV values were calculated using semi-quantitative concentrations obtained by internal-standard normalization and odor thresholds reported in water. The values are relative odor-potency indices and should not be interpreted as absolute OAVs. Odor-threshold sources: a VCF Online (2026), bTian et al. (2019), cXue et al. (2022), dXiao et al. (2022).

Aldehydes and sulfur-containing compounds, such as 2,4-undecadienal and 3-mercapto-3-methylbutyl formate, displayed elevated rOAVs and marked accumulation differences across light treatments, particularly under RB illumination, indicating their potential roles in enhancing fatty, green, and roasted sensory attributes. Notably, several terpenoids, including limonene and ocimene isomers, also showed increased rOAV values under RB conditions, consistent with the enhanced citrus and herbal aroma notes observed in the flavor wheel analysis.

In contrast, comparisons between the two white light treatments (3000 K and 6500 K) showed relatively smaller differences in rOAV distribution, suggesting that changes in CCT alone induced limited shifts in dominant odor-active compounds. Overall, these results indicate that red-blue lighting produced the most pronounced reconfiguration of candidate odor-active volatiles, providing quantitative evidence of changes in their relative odor potency that complements the metabolomic and aroma-descriptor analyses.

4. Discussion

The volatile profile of celery petioles was dominated by terpenoids, esters, ketones, heterocyclic compounds, and alcohols, which together accounted for 67.47% of the detected volatilome (Fig. 2). This compositional pattern agrees with previous descriptions of celery aroma, in which terpenoid derivatives and lipid-derived GLVs provide characteristic herbal and green notes, while nitrogen- and sulfur-containing compounds contribute earthy and roasted attributes (Bate & Rothstein, 1998; Reale et al., 2021; Fan et al., 2023; Yan et al., 2024a). Previous celery studies have mainly compared tissues, cultivars, geographic origins, or individual supplementary light conditions. By comparing the edible petiole under warm-white, cool-white, and red-blue LEDs together with a natural greenhouse light reference, the present study reveals light spectra-responsive changes in both volatile abundance and odor-potency-weighted composition. Across these light environments, red-blue light produced the most distinct volatile profile and the broadest accumulation of aroma-related compounds. Since no authentic standards were analyzed alongside the present sample batch, all compound assignments are regarded as tentative. The 207 compounds corresponding to database records established using authentic-standard-derived parameters are listed in Supplementary Table S2.

Among the three LED treatments, PCA, HCA, and OPLS-DA consistently separated RB from the warm-white and cool-white treatments. RB-related comparisons also contained the largest numbers of DAMs, with up-accumulated compounds predominating in comparison with CK, 3000 K, and 6500 K (Fig. 2, Fig. 3). In comparison, 3000 K and 6500 K showed partially overlapping PCA distributions and the smallest pairwise differences in DAMs and odor-active compounds. These patterns indicate that redistribution of spectral energy into red and blue wavebands produced a stronger response in the celery-petiole volatilome than changing CCT within broad-spectrum white light. Similar increases in terpenoid- and phenylpropanoid-related volatiles under red-blue spectra have been reported in sweet basil (Kivimäenpää et al., 2022). The present results extend this spectral response to celery petioles and show that warm-white and cool-white LEDs generated comparatively similar aroma-related profiles, whereas red-blue light produced a distinct chemical signature.

At the chemical-class level, comparisons involving RB treatment showed broad increases in terpenoids, esters, ketones, heterocyclic compounds, alcohols, and phenols (Fig. 4). Terpenoids such as limonene and β-ocimene are important contributors to the citrus and herbal character of celery and are derived from isoprenoid metabolism (Carvalho & Folta, 2016). Their increased accumulation under RB treatment agrees with reports that red and blue wavelengths regulate terpene formation in tea and basil (Fu et al., 2015; Hammock et al., 2021; Kivimäenpää et al., 2022). Red-blue-responsive ketones and aldehydes, including 3-octen-2-one and 2-nonenal, are associated with mushroom-like, fatty, and green odor descriptors. Methoxypyrazines contribute roasted, earthy, and vegetable-like attributes, whereas esters and furanones contribute fruity, sweet, and caramel-like notes. The concurrent changes in these classes were reflected in the broader distribution of citrus, herbal, fruity, green, roasted, and sweet odor descriptors in RB-related comparisons (Fig. 7, Fig. 8). Therefore, the RB profile was characterized by coordinated changes across several odor dimensions rather than by an increase in a single volatile class.

The rOAV analysis further prioritized compounds according to their relative abundance and odor thresholds. (5Z)-Octa-1,5-dien-3-one showed the highest rOAV across treatments and increased markedly under RB treatment, prioritizing it as a candidate contributor to metallic and green notes. 2-Methoxy-3,5-dimethylpyrazine also showed high rOAV values, particularly under CK and RB treatments, supporting its contribution to bread-like and roasted attributes. Several terpenoids, including limonene and ocimene isomers, as well as furanone and sulfur-containing compounds, combined high rOAV values with treatment-dependent accumulation (Table 2). Overall, terpenoids, esters, and ketones represented the principal odor-active classes associated with treatment differentiation, while heterocyclic and sulfur-containing compounds contributed additional odor attributes despite their lower numerical abundance. This pattern is particularly relevant to celery because its characteristic aroma arises from the combination of fresh and herbal terpenoid notes with highly potent green, earthy, and roasted compounds. Integrating relative abundance with rOAV therefore provides an aroma-oriented prioritization of celery-petiole volatiles and highlights candidate markers that distinguish the RB treatment from the two white LED treatments. Because the odor thresholds used in the rOAV calculation were determined in water, matrix-dependent retention and release may modify the odor contributions of these compounds in celery petioles. Targeted validation in the celery matrix, together with sensory evaluation, would further clarify their contributions to perceived aroma.

KEGG mapping associated red-blue-responsive metabolites with sesquiterpenoid and triterpenoid biosynthesis, phenylalanine metabolism, amino acid metabolism, pyruvate metabolism, the tricarboxylic acid cycle, and 2-oxocarboxylic acid metabolism (Fig. 6). These pathways represent interconnected biochemical routes related to the formation of volatile precursors and end products. Pyruvate and tricarboxylic-acid-cycle intermediates contribute carbon substrates used in acetyl-CoA-dependent terpenoid and fatty-acid metabolism, while phenylalanine metabolism is associated with aromatic volatile formation. Amino acid and 2-oxocarboxylic acid pathways are linked to branched-chain and nitrogen-containing volatiles (Maoz et al., 2022; Wang et al., 2019). The simultaneous enrichment of these pathways is consistent with the coordinated accumulation of terpenoids, ketones, esters, and heterocyclic compounds under RB treatment. Similar associations between central carbon, amino acid, and volatile metabolism have been observed in tea leaves exposed to different light qualities (Li et al., 2022; Yue et al., 2020). In celery petioles, the KEGG results provide a pathway-level context for the observed volatile patterns and identify terpenoid, phenylalanine, amino acid, and central‑carbon-related pathways as major modules associated with the red-blue light response.

The spectral pattern observed in celery shares several features with responses reported in other aromatic crops, particularly the sensitivity of terpenoid-related compounds to red and blue wavelengths (Fu et al., 2015; Kivimäenpää et al., 2022; Yue et al., 2020). In leafy fennel, supplemental narrow wavelength red LED lighting increased aroma-active phenylpropanoids, particularly (E)-anethole and estragole (Liu et al., 2026). In the present study, the celery-petiole response was characterized by changes in terpenoids, ketones, esters, methoxypyrazines, furanones, and sulfur-containing compounds. The prominence of (5Z)-octa-1,5-dien-3-one, 2-methoxy-3,5-dimethylpyrazine, limonene, and ocimene isomers defines an odor-active pattern closely associated with the green, metallic, herbal, citrus, roasted, and sweet attributes of celery. This crop- and organ-specific profile demonstrates the value of evaluating lighting responses directly in the harvested petiole. A limitation of this study is that the greenhouse CK received a lower mean DLI than the LED treatments (∼5 vs. 8.6 mol m−2 d−1) and was maintained in a different cultivation setting, which may have contributed to the observed CK-to-LED differences. Accordingly, since p values were not adjusted for multiple testing, the screened DAMs are presented as candidate treatment-responsive compounds for subsequent validation.

From a production perspective, the comparison among the three LED treatments is directly relevant to spectral selection because PPFD and photoperiod were maintained at common levels. The similar volatile profiles under 3000 K and 6500 K indicate that changing white-light CCT produced comparatively modest adjustments in the dominant odor-active compounds, whereas RB treatment generated a broader aroma-related response. Red-blue light can therefore be considered a targeted spectral option for flavor-oriented celery cultivation. The red-blue-light-responsive compounds prioritized here provide candidate chemical markers for future evaluation of lighting strategies across celery cultivars, production cycles, and controlled-environment systems.

5. Conclusion

In conclusion, the celery-petiole volatilome responded distinctly to LED spectral composition, with red-blue lighting generating a broader and more differentiated odor-active profile than warm- or cool-white LEDs. This response was characterized by coordinated increases in terpenoids, esters, ketones, and heterocyclic compounds and by the prioritization of (5Z)-octa-1,5-dien-3-one, 2-methoxy-3,5-dimethylpyrazine, limonene, and ocimene isomers as candidate odor-active markers of celery-petiole aroma. The association of RB-responsive volatiles with terpenoid, phenylalanine, amino acid, and central carbon metabolism provides a pathway-level framework for interpreting the spectral response of the edible petiole. More broadly, the findings show that combining volatile abundance with rOAV-based prioritization provides a more informative basis for evaluating aroma-related responses than compound counts alone. Red-blue LED lighting therefore represents a promising targeted option for aroma-oriented celery cultivation. These findings establish a celery-specific framework linking light spectra with volatile composition and candidate odor-active compounds, and provide measurable chemical endpoints for sensory validation across cultivars and production cycles.

CRediT authorship contribution statement

Haijie Dou: Writing – review & editing, Writing – original draft, Software, Methodology, Investigation, Data curation, Conceptualization. Yongsheng Liu: Writing – original draft, Methodology, Data curation. Zejin Zhang: Writing – original draft, Supervision, Software, Data curation, Conceptualization. Zidan Wen: Software, Data curation, Conceptualization. Zhixin Li: Supervision, Data curation, Conceptualization. Jinxiu Song: Supervision, Methodology, Conceptualization. Na Lu: Supervision, Methodology, Conceptualization. Yanjie Yang: Supervision, Project administration, Methodology, Funding acquisition, Conceptualization. Zhengnan Yan: Writing – review & editing, Resources, Project administration, Methodology, Funding acquisition, Conceptualization.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.

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.

Acknowledgements

This work was supported by Shandong Provincial Natural Science Foundation (ZR2026MS0396); the Modern Agricultural Industrial Technology System of Shandong Province (SDAIT-05); University-Enterprise Cooperation Project (6602423221); the Postgraduate Innovation Program of Qingdao Agricultural University (QNDC20250164).

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.104398.

Appendix A. Supplementary data

Supplementary material
mmc1.docx (76.7KB, docx)

Data availability

Data will be made available on request.

References

  1. Bate N.J., Rothstein S.J. C6-volatiles derived from the lipoxygenase pathway induce a subset of defense-related genes. The Plant Journal. 1998;16(5):561–569. doi: 10.1046/j.1365-313x.1998.00324.x. [DOI] [PubMed] [Google Scholar]
  2. Carvalho S.D., Folta K.M. Green light control of anthocyanin production in microgreens. Acta Horticulturae. 2016;1134:13–18. [Google Scholar]
  3. Contreras-Avilés W., Heuvelink E., Marcelis L.F., Kappers I.F. Ménage à trois: Light, terpenoids, and quality of plants. Trends in Plant Science. 2024;29(5):572–588. doi: 10.1016/j.tplants.2024.02.007. [DOI] [PubMed] [Google Scholar]
  4. Dai C., Huang X., Lv R., Zhang Z., Sun J., Aheto J.H. Analysis of volatile compounds of Tremella aurantialba fermentation via electronic nose and HS-SPME-GC-MS. Journal of Food Safety. 2018;38(6) [Google Scholar]
  5. Dou, H., Li, Z., Liu, Q., Jiang, P., Song, J., Ji, F., Lu, N., Xu, L., & Yan, Z. (2025). Organ-specific physiological and metabolic differentiation in celery (Apium graveolens L.) to supplemental blue light in controlled environment agriculture. Horticulturae, 11(9), article 1074.
  6. Fan X., Lu N., Xu W., Zhuang Y., Jin J., Mao X., Ren N. Response of flavor substances in tomato fruit to light spectrum and daily light integral. Plants. 2023;12 doi: 10.3390/plants12152832. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Fu X., Chen Y., Mei X., Katsuno T., Kobayashi E., Dong F.…Yang Z. Regulation of formation of volatile compounds of tea (Camellia sinensis) leaves by single light wavelength. Scientific Reports. 2015;5 doi: 10.1038/srep16858. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Gold H.J., Wilson C.W. The volatile flavor substances of celery. Journal of Food Science. 1963;28:484–488. [Google Scholar]
  9. Guo K., Guo Z., Ge S., Wang S., Liang L., Peng W., Liu X., Huang X., Qin C., Luo Z., Ouyang K., Pan T., Jiang C., Li M., Zheng Y., Wang S., Lu W. Effect of different light-dark cycles on the growth and nutritional quality of celery. Agriculture. 2025;15(21) [Google Scholar]
  10. Hammock H.A., Kopsell D.A., Sams C.E. Narrowband blue and red LED supplements impact key flavor volatiles in hydroponically grown basil across growing seasons. Frontiers in Plant Science. 2021;12 doi: 10.3389/fpls.2021.623314. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Ji C., Liu C., Wang R., Ruan M., Yao Z., Wan H., Li Z., Cheng Y., Ye Q. The science behind vegetable aromas: Types, synthesis, and influencing factors of volatile compounds. Plant Signaling & Behavior. 2025;20 doi: 10.1080/15592324.2025.2527958. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Kivimäenpää M., Mofikoya A., Abd El-Raheem A.M., Riikonen J., Julkunen-Tiitto R., Holopainen J.K. Alteration in light spectra causes opposite responses in volatile phenylpropanoids and terpenoids compared with phenolic acids in sweet basil (Ocimum basilicum) leaves. Journal of Agricultural and Food Chemistry. 2022;70:14262–14272. doi: 10.1021/acs.jafc.2c03309. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Kwaw E., Ma Y., Tchabo W., Apaliya M.T., Sackey A.S., Wu M., Xiao L. Effect of pulsed light treatment on the phytochemical, volatile, and sensorial attributes of lactic-acid-fermented mulberry juice. International Journal of Food Properties. 2018;21(1):213–228. [Google Scholar]
  14. Li F., Boateng I.D., Chen S., Yang X.M., Soetanto D.A., Liu W. Pulsed light irradiation improves degradation of ginkgolic acids and retainment of ginkgo flavonoids and terpene trilactones in Ginkgo biloba leaves. Industrial Crops and Products. 2023;204 [Google Scholar]
  15. Li M.Y., Hou X.L., Wang F., Tan G.F., Xu Z.S., Xiong A.S. Advances in the research of celery, an important Apiaceae vegetable crop. Critical Reviews in Biotechnology. 2018;38:172–183. doi: 10.1080/07388551.2017.1312275. [DOI] [PubMed] [Google Scholar]
  16. Li Y., He C., Yu X., Zhou J., Ntezimana B., Yu Z.…Ni D. Study on improving aroma quality of summer-autumn black tea by red-light irradiation during withering. LWT-Food Science and Technology. 2022;154 [Google Scholar]
  17. Liu J., Sumner A.T., Harris A.T., Li S., Bargmann B.O.R., Haak D.C., Yin Y. Narrow-wavelength red LED lighting enhances growth and aroma of hydroponic leafy fennel (Foeniculum vulgare mill.) Food Chemistry: X. 2026;38 doi: 10.1016/j.fochx.2026.104178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Mao H., Hang T., Zhang X., Lu N. Both multi-segment light intensity and extended photoperiod lighting strategies, with the same daily light integral, promoted Lactuca sativa L. growth and photosynthesis. Agronomy. 2019;9(12) [Google Scholar]
  19. Maoz I., Lewinsohn E., Gonda I. Amino acids metabolism as a source for aroma volatiles biosynthesis. Current Opinion in Plant Biology. 2022;67 doi: 10.1016/j.pbi.2022.102221. [DOI] [PubMed] [Google Scholar]
  20. Mei S., Ding J., Chen X. Identification of differential volatile and non-volatile compounds in coffee leaves prepared from different tea processing steps using HS-SPME/GC–MS and HPLC-Orbitrap-MS/MS and investigation of the binding mechanism of key phytochemicals with olfactory and taste receptors using molecular docking. Food Research International. 2023;168 doi: 10.1016/j.foodres.2023.112760. [DOI] [PubMed] [Google Scholar]
  21. Meng L., Song J., Ni D., Kamaruddin M.A. Effects of far-red light on growth, endogenous hormones, antioxidant capacity and quality of lettuce. Food Production, Processing and Nutrition. 2024;6 [Google Scholar]
  22. Reale S., Di Cecco V., Di Donato F., Di Martino L., Manzi A., Di Santo M., D’Archivio A.A. Characterization of the volatile profile of cultivated and wild-type Italian celery (<i>Apium graveolens</i> L.) varieties by HS-SPME/GC-MS. Applied Sciences. 2021;11(13) [Google Scholar]
  23. Ren M., Yu X., Mujumdar A.S., Yagoub A.E.G.A., Chen L., Zhou C. Visualizing the knowledge domain of pulsed light technology in the food field: A scientometrics review. Innovative Food Science & Emerging Technologies. 2021;74 [Google Scholar]
  24. Song J., Fan Y., Li X., Li Y., Mao H., Zuo Z., Zou Z. Effects of daily light integral on tomato (Solanum lycopersicon L.) grafting and quality in a controlled environment. International Journal of Agricultural and Biological Engineering. 2022;15(6):44–50. [Google Scholar]
  25. Sreelakshmi Y., Sharma R. Differential regulation of phenylalanine ammonia lyase activity and protein level by light in tomato seedlings. Plant Physiology and Biochemistry. 2008;46:444–451. doi: 10.1016/j.plaphy.2008.02.001. [DOI] [PubMed] [Google Scholar]
  26. Sun Y., Li M., Li X., Du J., Li W., Lin Y., Tang H. Characterization of volatile organic compounds in five celery (Apium graveolens L.) cultivars with different petiole colors by HS-SPME-GC-MS. International Journal of Molecular Sciences. 2023;24 doi: 10.3390/ijms241713343. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Tang L., Chu Q., Liu K., Lu Y., Cheng S., Pan T., Zhou X., He Z. The impact of light quality on the growth and quality of celery. Horticulturae. 2025;11 [Google Scholar]
  28. Tian H., Xu X., Chen C., Yu H. Flavoromics approach to identifying the key aroma compounds in traditional Chinese milk fan. Journal of Dairy Science. 2019;102(11):9639–9650. doi: 10.3168/jds.2019-16796. [DOI] [PubMed] [Google Scholar]
  29. Turner, L., Lignou, S., Gawthrop, F., & Wagstaff, C. (2021). Investigating the relationship of genotype and climate conditions on the volatile composition and sensory profile of celery (Apium graveolens). Foods, 10, article 1335. [DOI] [PMC free article] [PubMed]
  30. VCF online. Volatile Compounds in Food 16.10. BeWiDo BV, The Netherlands, 2026. Available at: https://www.vcf-online.nl/ (accessed 26 March 2026).
  31. Wang H., Cheng Q., Liu J., Yang H., Liu J., Li M.…Shen H. Mapping of the AgPPur gene for the purple petiole in celery (Apium graveolens L.) Scientia Horticulturae. 2023;317 [Google Scholar]
  32. Wang M., Zhang L., Boo K.H., Park E., Drakakaki G., Zakharov F. PDC1, a pyruvate/α-ketoacid decarboxylase, is involved in acetaldehyde, propanal and pentanal biosynthesis in melon (Cucumis melo L.) fruit. The Plant Journal. 2019;98(1):112–125. doi: 10.1111/tpj.14204. [DOI] [PubMed] [Google Scholar]
  33. Wang T., An J., Chai M., Zhu Z., Jiang Y., Huang X., Han B. Volatile metabolomics reveals the characteristics of the unique flavor substances in oats. Food Chemistry: X. 2023;20 doi: 10.1016/j.fochx.2023.101000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Xiao Z., Cao X., Zhu J., Chen F., Niu Y. Characterization of the key aroma compounds in three world-famous black teas. European Food Research and Technology. 2022;248:2237–2252. [Google Scholar]
  35. Xue J., Liu P., Yin J., Wang W., Zhang J., Wang W., Le T., Ni D., Jiang H. Dynamic changes in volatile compounds of shaken black tea during its manufacture by GC × GC-TOFMS and multivariate data analysis. Foods. 2022;11(9) doi: 10.3390/foods11091228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Yan J., Chen J., Huang Z., He L., Wu L., Yu L., Zhu W. Characterisation of volatile compounds in nine celery varieties and breeding selections using GC-IMS and GC-MS. Food Chemistry: X. 2024;24 doi: 10.1016/j.fochx.2024.101936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Yan J., He L., Huang Z., Wang H., Yu L., Zhu W. Investigating the impact of origins on the quality characteristics of celery seeds based on metabolite analysis through HS-GC-IMS, HS-SPME-GC-MS and UPLC-ESI-MS/MS. Foods. 2024;13 doi: 10.3390/foods13101428. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Yan Z., Li Z., Wang H., Dou H., Song J., Ji F., Lin D. Volatile metabolomics analysis reveals the flavor response of different parts of celery to ultraviolet radiation. Food Chemistry: X. 2025;25 doi: 10.1016/j.fochx.2025.102213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Yan Z., Wang H., Li Z., Song J., Ji F., Yang Y., Lin D. Unraveling the growth and volatile compound profiles of celery under different supplementary light durations via comprehensive evaluation and GC-IMS. BMC Plant Biology. 2025;25 doi: 10.1186/s12870-025-07045-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Yu, P., Wang, C., Tursun, R., Zeng, X., Cai, W., Song, J., & Geng, W. (2025). Synergistic regulation of growth and quality in substrate-grown spinach by LED light quality and intensity in PFALs. Horticulturae, 11(9), article 1065.
  41. Yuan H., Cao G., Hou X., Huang M., Du P., Tan T.…Luo J. Development of a widely targeted volatilomics method for profiling volatilomes in plants. Molecular Plant. 2022;15:189–202. doi: 10.1016/j.molp.2021.09.003. [DOI] [PubMed] [Google Scholar]
  42. Yuan H., Jiangfang Y., Liu Z., Su R., Li Q., Fang C.…Luo J. WTV2.0: A high-coverage plant volatilomics method with a comprehensive selective ion monitoring acquisition mode. Molecular Plant. 2024;17:972–985. doi: 10.1016/j.molp.2024.04.012. [DOI] [PubMed] [Google Scholar]
  43. Yue C., Wang Z., Shi X., Jiang X., Li Y., Yang P. Effects of light quality on aroma metabolites in tea: A review of recent literature. Food Science. 2020;41(5):299–305. [Google Scholar]
  44. Zhang Y., Wang T., Wang X., Dong R., Wang D., Zhao G. Apium graveolens Linn. (celery) seeds: A comprehensive review of phytochemistry profile, pharmacological effects and future trends. Food Reviews International. 2025;42(4):2668–2695. [Google Scholar]
  45. Zhou J., Li P., Wang J. Effects of light intensity and temperature on the photosynthesis characteristics and yield of lettuce. Horticulturae. 2022;8(2) [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary material
mmc1.docx (76.7KB, docx)

Data Availability Statement

Data will be made available on request.

Articles from Food Chemistry: X are provided here courtesy of Elsevier

RESOURCES