Skip to main content
NCBI home page
Food Chemistry: X logoLink to Food Chemistry: X
. 2026 Sep 10;39:104424. doi: 10.1016/j.fochx.2026.104424

Comparative multi-omics analysis of microbial succession, metabolite profiles, and antioxidant properties in fermented black and yellow highland barley

Mengru Han a, Hanghang Hou a, Honglu Wang a, Miaomiao Zhang a, Xinyu Tang a, Xirong Zhou b, Meijin Liu b, Changshou Ma c, Qinghua Yang a,⁎, Baili Feng a,⁎
PMCID: PMC13586749  PMID: 42761470

Abstract

Colored highland barley is rich in bioactive compounds, but fermentation-associated links among microbes, metabolites, and antioxidant properties remain unclear. Black (BHB) and yellow highland barley (YHB) were comparatively fermented using microbial profiling, non-volatile metabolomics, and antioxidant assays. Both cultivars showed reduced microbial diversity, lactic acid bacterial enrichment, fungal dominance by Saccharomycopsis, increased phenolics and anthocyanins, and decreased flavonoids. BHB retained higher flavonoid and anthocyanin contents and stronger radical-scavenging activities, whereas YHB showed greater increases in total phenolics and antioxidant capacity. Both were dominated by late-stage Pediococcus, but YHB had a denser positive bacterial network, whereas BHB showed greater modularity and enrichment of flavonoid- and phenolic acid-related metabolites. YHB was relatively associated with amino acid- and peptide-related metabolites. Lactiplantibacillus and Levilactobacillus were positively associated with selected phenolic metabolites. These findings identify common and cultivar-specific associations among microbial succession, metabolite variation, and antioxidant characteristics during colored highland barley fermentation.

Keywords: Colored highland barley, Fermentation, Antioxidant capacity, Non-volatile metabolites, Microbial community succession

Highlights

  • •

    Fermentation enhanced antioxidant-related properties in colored highland barley, with cultivar-dependent differences.

  • •

    Black barley retained higher flavonoid and anthocyanin contents and stronger radical-scavenging activity.

  • •

    Black barley favored flavonoid- and phenolic acid-related metabolites, whereas yellow barley favored peptides.

  • •

    Yellow barley formed a denser positive bacterial network, whereas black barley showed greater modularity.

1. Introduction

Colored cereals have received increasing attention because they contain diverse bioactive compounds and may provide nutritional and functional advantages (Gao et al., 2024). Highland barley (Hordeum vulgare var. coeleste L.), an important cereal crop cultivated on the Qinghai–Tibet Plateau, is widely used to produce traditional fermented foods and beverages, including highland barley wine (Xia et al., 2022). Colored highland barley contains phenolic compounds, flavonoids, anthocyanins, and other constituents that contribute to grain pigmentation and antioxidant activity (Ge et al., 2021; Li et al., 2023). Consequently, colored highland barley has potential as a raw material for fermented cereal products with enhanced functional attributes. Previous studies have compared the compositional characteristics and antioxidant-related properties of highland barley cultivars with different grain colors (Ge et al., 2021; Han et al., 2025). In our previous study, black, purple, blue, and yellow highland barley cultivars were systematically evaluated, and black and yellow cultivars showed the greatest differences in compositional characteristics and antioxidant capacity, with black highland barley exhibiting higher antioxidant capacity than yellow highland barley (Han et al., 2025). Such differences in the initial composition and antioxidant properties of the two cultivars may lead to distinct fermentation trajectories, including differences in microbial succession and the transformation of bioactive compounds.

Fermentation is a critical stage in highland barley wine production because it affects microbial growth, substrate conversion, and final product quality (Guo et al., 2025). Colored cereal cultivars differ in the contents and physicochemical characteristics of carbohydrates, proteins, dietary fiber, and phenolic compounds. These differences may alter substrate availability for microorganisms and impose selective pressures on microbial growth during fermentation, thereby influencing microbial succession and metabolic activity (Zhai et al., 2023). For example, phenolic acids and anthocyanins may selectively inhibit certain microorganisms and contribute to shifts in microbial community composition (Zhang, Xu, et al., 2025; Gaur & Gänzle, 2023). In addition, variation in starch, protein, and dietary-fiber contents may affect the availability of fermentable substrates and the activities of microbial enzymes, with subsequent effects on metabolite formation and sensory quality (Bui et al., 2020).

Microbial succession is closely associated with the transformation of cereal components during fermentation. Microorganisms can release extracellular enzymes and convert cereal-derived substrates into a wide range of metabolites, thereby affecting the bioavailability of phenolic compounds and the antioxidant-related properties of fermented products (Chen et al., 2025; Li et al., 2025; Zhang et al., 2024). For instance, fermentation has been reported to promote the release of phenolic compounds from highland barley bran and enhance its antioxidant activity (He et al., 2023). Fermentation may also modify cell-wall polysaccharides and hydrolyze phytic acid, potentially affecting nutrient accessibility and flavor development in barley-based foods (Okarter & Liu, 2010). Although previous studies have demonstrated that fermentation can alter chemical composition and metabolite profiles in highland barley products (Han et al., 2025; He et al., 2023), dynamic and integrated analyses of microbial succession, non-volatile metabolite transformation, and antioxidant-related quality during the fermentation of different colored highland barley cultivars remain limited. In particular, the relationships among cultivar-dependent microbial communities, differential metabolites, and antioxidant properties have not been systematically characterized.

Therefore, this study used a highland barley wine fermentation system to investigate black and yellow highland barley, which showed marked differences in initial chemical composition and antioxidant-related characteristics in our previous study. Under identical fermentation conditions, we examined whether these two cultivars exhibited distinct cultivar-associated patterns in microbial community succession, non-volatile metabolite changes, and antioxidant-related indices. An integrated approach combining microbial community profiling, non-volatile metabolomics, and antioxidant-related quality assessment was applied. Specifically, this study aimed to: (1) compare the changes and differences in antioxidant-related indices, microbial community composition, and non-volatile metabolite profiles of black and yellow highland barley at different fermentation stages; (2) identify core microorganisms and differential metabolites associated with cultivar-specific fermentation characteristics; and (3) examine the statistical associations among microbial taxa, differential metabolites, and antioxidant-related quality attributes. This study contributes to a better understanding of cultivar-dependent quality changes during highland barley fermentation and provides a theoretical reference for developing fermented highland barley products with enhanced functional properties.

2. Materials and methods

2.1. Colored highland barley and sample preparation

Two highland barley cultivars with distinct grain colors were chosen as experimental materials in this study (Fig. 1a), namely Liuleng (black, BHB) and Ganqing10 (yellow, YHB). All test samples were grown in a uniform experimental plot belonging to the Agricultural Science Research Institute of Gannan Tibetan Autonomous Prefecture, which is located at 35°0′ N latitude and 102°54′ E longitude with an altitude of approximately 2960 m, and were harvested in 2023. After collection, the samples were subjected to air-drying under ambient temperature (20 ± 5 °C) prior to subsequent grain fermentation and analysis.

Fig. 1.

Fig. 1

Open in a new tab

Appearance characteristics of black (BHB) and yellow (YHB) highland barley (a) and dynamic changes in antioxidant substance content and antioxidant capacity during fermentation (b). 0, 24, 48, and 72 represent highland barley samples at 0 h, 24 h, 48 h, and 72 h of fermentation, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Highland barley grains (HBGs) were processed according to the traditional brewing procedure for Tibetan highland barley wine (Cao et al., 2012; Chen et al., 2021; Lu et al., 2021). Briefly, HBGs (200 g) were soaked in pure water at 25 °C (room temperature, RT) for 30 min. The soaked HBGs were then heated to boiling on an induction cooker (CH2082; Galanz Electrical Appliances Manufacturing Co., Ltd., Guangdong, China) at 600 W, followed by continuous boiling at 2000 W for 35 min. Upon completion of the boiling step, the HBGs were spread out and cooled to RT. After cooling, HBGs were inoculated with 2% highland barley saccharifying starter, which consisted of Tibetan indigenous Jiuqu and high active dry yeast (Angel Yeast Co., Ltd., Hubei, China) at a mass ratio of 1:1. The Tibetan indigenous jiuqu was obtained from Rumei Town, Mangkang County, Tibet, China. It is a traditional indigenous starter used for brewing Tibetan highland barley wine and is typically prepared from highland barley flour, local plateau herbs, and aged starter. The prepared HBGs were subsequently transferred into vertical cylindrical traditional earthenware jars (13 cm in height and 9 cm in bottom width) and fermented under sealed conditions at 29 °C for 72 h. Samples were collected at 0, 24, 48, and 72 h during fermentation. Three independent fermentation jars were prepared for each sampling time point as biological replicates. Samples were collected separately from the upper, middle, and lower layers of the fermented mash in each jar and then thoroughly mixed to obtain fermented highland barley samples. Each sample was divided into two portions for subsequent analyses. One portion was immediately frozen in liquid nitrogen and stored at −80 °C for microbial community profiling. The other portion was freeze-dried for the determination of antioxidant compound contents and antioxidant activity. In addition, fully fermented HBG samples collected at 72 h were stored at −80 °C for subsequent non-volatile metabolomic analysis.

2.2. Quantification of antioxidant substances and capacity

Total phenolic content was determined according to the methods described by Han et al. (2025) and Feng et al. (2017). Briefly, total phenols were extracted by blending 1 g of sample with 8 mL of 85% methanol solution supplemented with 1% formic acid. A 20 μL portion of the sample extract was combined with Folin-Ciocalteu reagent, and the mixture was vortexed thoroughly for 5 min. Subsequently, 160 μL of 10% (w/v) sodium carbonate solution was added to the reaction system. After incubation at RT for 1 h, the absorbance was measured at 765 nm. Total flavonoid content was determined using a commercial kit (No. G0118F; Suzhou Greus, China). The antioxidant capacity was assessed by measuring DPPH radical scavenging ability kits, ABTS radical scavenging ability kits, hydroxyl radical scavenging ability kits, and superoxide radical scavenging ability kits (Beijing Solarbio Science & Technology Co., Ltd., Nos. BC4750, BC4770, BC1320, and BC1290). The total flavonoid content was quantified with a commercially available assay kit (No. G0118F; Greus Biotechnology, Suzhou, China). Anthocyanin content was determined according to the method described by Zhang, Han, et al. (2025). Briefly, 0.2 g of the test sample was weighed and mixed with 25 mL of ethanol-HCl solution (prepared by mixing 95% ethanol and 1.5 mol/L HCl at a volume ratio of 85:15). The mixture was extracted in a water bath at 80 °C for 30 min. After cooling, the absorbance of the extract at 535 nm (OD₅₃₅) was measured using a UV–visible spectrophotometer, with three replicates per group. The pigment concentration corresponding to OD₅₃₅ = 1 was defined as one anthocyanin unit, and the content was expressed as anthocyanin units per gram of sample (U/g).

Total antioxidant capacity was determined according to the method of Zhang, Han, et al. (2025). Antioxidant activity was evaluated by determining the scavenging capacities against DPPH, ABTS, hydroxyl, and superoxide anion radicals, using corresponding assay kits supplied by Beijing Solarbio Science & Technology Co., Ltd., with product numbers BC4750, BC4770, BC1320, and BC1290 respectively.

2.3. Analysis of microbial community structure

Genomic DNA of the microbial community was extracted from fermented HBGs in accordance with the instructions of the E.Z.N.A.® Soil DNA Kit (Omega Bio-tek, Norcross, GA, USA). The quality of the extracted genomic DNA was detected by 1% agarose gel electrophoresis, and its concentration and purity were determined using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). Using the above-extracted DNA as the template, the bacterial 16S rRNA gene was amplified with the primers 799F (5′-AACMGGATTAGATACCCKG-3′) and 1193R (5′-ACGTCATCCCCACCTTCC-3′). For fungi, the internal transcribed spacer (ITS) region was amplified using the primers ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2R (5′-GCTGCGTTCTTCATCGATGC-3′). The PCR products were recovered by 2% agarose gel electrophoresis, purified using a DNA Gel Extraction Kit (PCR Clean-Up Kit, Yuhua, China), and the concentration of the purified products was detected and quantified with a Qubit 4.0 fluorometer (Thermo Fisher Scientific, USA). Library construction of the purified PCR products was performed using the NEXTFLEX Rapid DNA-Seq Kit, and sequencing was carried out on an Illumina Nextseq 2000 platform by Meiji Biomedical Technology Co., Ltd. (Shanghai, China).

2.4. Detection of non-volatile substances

For the detection of non-volatile substances, samples were first freeze-dried and ground into powder. A 50 mg aliquot was mixed with pre-cooled 70% methanol aqueous solution containing an internal standard, vortexed repeatedly, centrifuged, and filtered; the filtrate was transferred to an injection vial for analysis. UPLC-ESI-MS/MS (UPLC: ExionLC™ AD, https://sciex.com.cn/) was used, with chromatographic separation on an Agilent SB-C18 column via gradient elution. The column effluent was analyzed by ESI-QTRAP-MS/MS under specific conditions: source temperature 550 °C, ion spray voltage 5500 V (positive) and −4500 V (negative), GSI 50 psi, GSII 60 psi, and CUR 25 psi. Qualitative analysis relied on Metware Database secondary spectral data, while quantitative analysis used triple quadrupole mass spectrometry in MRM mode. MultiQuant software was applied for peak integration and concentration calculation.

2.5. Statistical analysis

Data processing and multiple comparison analyses were carried out with SPSS 17.0 software. Each assay was conducted in three independent replicates, and experimental outcomes were presented as mean values accompanied by standard deviations. The figures related to antioxidant properties were plotted with Origin 2025. Microbial community data analysis was carried out on the Majorbio Cloud Platform (https://cloud.majorbio.com). The α-diversity was calculated using mothur software (Schloss et al., 2009), and the intergroup differences in α-diversity were analyzed by the Wilcoxon rank-sum test. PCoA based on the Bray-Curtis distance algorithm was performed to verify the similarity of microbial community structures among samples. LEfSe analysis was used to identify bacterial taxa with significantly different abundances from the phylum to genus levels among different groups. Species were screened according to the criteria of Spearman's correlation coefficient |r| > 0.6 and p < 0.05 for microbial correlation network analysis (Ma et al., 2025). Visualization and topological structure analysis were performed using Gephi 0.10.1, and the stability was comprehensively evaluated via positive and negative cohesion indices (Feng et al., 2025). Functional prediction analysis of the 16S rRNA gene sequencing data was performed using PICRUSt2. Differential metabolites between the two groups were identified with the criteria of VIP > 1 and |log2FC| ≥ 1.0. Metabolites were annotated based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) Compound database (http://www.kegg.jp/kegg/compound/) and mapped to the KEGG Pathway database (http://www.kegg.jp/kegg/pathway.html). Metabolite Set Enrichment Analysis 2 was applied to explore the pathways significantly affected by metabolites, and the statistical significance was evaluated by p-values calculated from the hypergeometric test. The Mantel-test heatmaps and multi-omics correlation corplots were separately analyzed and plotted using R version 4.2.0 and R version 3.5.1 packages.

3. Results

3.1. Dynamic changes in antioxidant properties

The initial compositional characteristics of the unfermented grains are summarized in Table S1. Compared with yellow highland barley (YHB), black highland barley (BHB) contained higher levels of protein, crude dietary fiber, fat, and antioxidant-related compounds. Notably, the total flavonoid, total phenolic, and anthocyanin contents of BHB were approximately two-fold higher than those of YHB, indicating a greater initial abundance of antioxidant compounds in BHB.

Fermentation significantly affected antioxidant-related compounds and antioxidant activities in both cultivars, but the temporal patterns differed between BHB and YHB (p < 0.05, Fig. 1b). Total phenolic content increased overall during fermentation, with a more sustained increase in YHB; consequently, YHB showed higher total phenolic content at the final fermentation stage. In contrast, total flavonoid content declined progressively in both cultivars, although BHB consistently retained a higher level than YHB. Anthocyanins increased during the early and middle fermentation stages, reached their highest levels at 48 h, and then decreased slightly by 72 h. BHB generally maintained higher anthocyanin content, although the difference between cultivars was reduced at 48 h. Antioxidant activities also showed cultivar-dependent trajectories. Total antioxidant capacity increased in both cultivars, with a greater relative increase in YHB; however, BHB maintained higher total antioxidant capacity throughout fermentation. Similarly, BHB generally exhibited stronger DPPH, ABTS, hydroxyl radical, and superoxide radical scavenging activities than YHB. Among these indices, hydroxyl radical scavenging activity increased continuously, particularly during late fermentation, whereas superoxide radical scavenging activity peaked at 24 h and subsequently declined. DPPH scavenging activity decreased during the early and middle stages before partially recovering at 72 h. In contrast, ABTS scavenging activity remained relatively stable in BHB but showed a more evident decline in YHB.

Overall, fermentation promoted the accumulation or release of selected antioxidant-related compounds, particularly phenolics and anthocyanins, while reducing total flavonoid content. Despite the greater late-stage increase in total phenolics and total antioxidant capacity in YHB, BHB retained a stronger overall antioxidant profile, consistent with its higher initial abundance of flavonoids and anthocyanins. These cultivar-specific patterns provided the basis for subsequent analyses of microbial succession and metabolite variation.

3.2. Dynamic characteristics of microbial diversity

Microbial alpha diversity changed substantially during the fermentation of both BHB and YHB (Table S2). Bacterial diversity, as indicated by the Shannon index, decreased continuously in both cultivars and was significantly lower at 72 h than at 0 h, indicating progressive simplification of the bacterial community. In contrast, bacterial richness showed cultivar-specific patterns. Bacterial richness in BHB increased initially and then declined, whereas YHB showed an opposite trend, with richness decreasing at the early stage and increasing later in fermentation. Differences in bacterial richness between the two cultivars were most apparent at the initial and final fermentation stages. Fungal communities showed lower richness and diversity than bacterial communities and underwent more rapid simplification during fermentation. The fungal Chao index generally declined in both cultivars, while the fungal Shannon index approached zero during the middle and late stages. This pattern indicated the progressive dominance of a limited number of fungal taxa. Differences in several alpha-diversity indices were already evident between BHB and YHB at 0 h (p < 0.05), suggesting that the two cultivars harbored distinct initial microbial communities. Nevertheless, fermentation reduced microbial diversity in both systems, with fungal diversity showing the most pronounced decline.

3.3. Composition analysis of microbial community structure

The bacterial and fungal community compositions changed across fermentation stages in both cultivars (Fig. 2a, b). The bacterial communities were primarily composed of Pediococcus, Lactococcus, Sphingobium, Massilia, Enterococcus, Brevundimonas, Microbacterium, Bradyrhizobium, and Zoogloea. During fermentation, Pediococcus became progressively enriched in both cultivars and was dominant at 72 h, indicating its potential importance during late-stage highland barley fermentation. Lactococcus was most abundant at 24 h and subsequently declined, suggesting a greater contribution during the early fermentation stage. Although the overall succession patterns were similar, Pediococcus showed greater enrichment in YHB at 48 h, and Lactococcus was more abundant in YHB at 72 h. The fungal community was almost completely dominated by Saccharomycopsis throughout fermentation. This highly uneven distribution was consistent with the near-zero fungal Shannon index observed at the middle and late stages, indicating that fungal succession was characterized primarily by the persistence and dominance of a single genus.

Fig. 2.

Fig. 2

Open in a new tab

Genus-level community composition of bacteria (a) and fungi (b), and principal coordinate analysis (PCoA) of bacterial (c) and fungal (d) community structures during fermentation of black (BHB) and yellow (YHB) highland barley. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

PCoA further showed that the bacterial communities of BHB and YHB broadly overlapped, and no significant cultivar-level separation was detected (Fig. 2c). Thus, despite differences in the relative abundance of individual genera at specific stages, the overall bacterial community structures were similar between the two fermentation systems. In contrast, fungal communities differed significantly between BHB and YHB (Fig. 2d); however, the low effect size and lack of clear overall separation indicated that this difference was limited.

LEfSe analysis identified stage-specific bacterial biomarkers in each cultivar (Fig. 3). BHB was characterized by the enrichment of Gordonia and Frigoribacterium at 24 h, Enterococcus at 48 h, and Levilactobacillus at 72 h (Fig. 3a, Fig. S1a). In YHB, Lactococcus and Flavobacterium were enriched at 24 h (Fig. 3b, Fig. S1b). Notably, Pediococcus was significantly enriched at 72 h in both cultivars, further supporting its role as a common late-stage taxon. These results suggest that the two cultivars shared an overall bacterial succession pattern but differed in the timing and relative enrichment of specific bacterial taxa.

Fig. 3.

Fig. 3

Open in a new tab

Screening of biomarkers by LEfSe analysis in black (BHB) and yellow (YHB) highland barley at different fermentation stages. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.4. Co-occurrence network analysis

The bacterial co-occurrence networks differed markedly between BHB and YHB fermentation systems (Fig. 4a, b). The YHB network showed more edges, higher average degree, greater network density, and a higher clustering coefficient than the BHB network, indicating a more closely connected and topologically complex bacterial association network. YHB also had a higher proportion of positive correlations, suggesting that co-occurrence patterns were more frequently positive in this system. By contrast, BHB contained a slightly greater proportion of negative correlations and exhibited substantially higher modularity, indicating a more clearly compartmentalized network structure. The two networks had similar average path lengths, suggesting comparable topological connectivity. However, the smaller network diameter of BHB indicated a more compact network, whereas the YHB network showed broader connectivity. At the phylum level, the network compositions were broadly similar, with Proteobacteria predominating in both systems, followed by Firmicutes. Therefore, cultivar-associated differences were mainly reflected in microbial association patterns and genus-level network organization rather than in the dominant bacterial phyla.

Fig. 4.

Fig. 4

Open in a new tab

Co-occurrence network analysis (r > 0.6, p < 0.05) and screening of keystone taxa in fermented colored highland barley. Co-occurrence networks and topological characteristics in fermented black (a, BHB) and yellow (b, YHB) highland barley (n = 12 per group). Identification of keystone taxa in network nodes of fermented black (c) and yellow (d) highland barley based on topological roles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Keystone taxa also differed between the two networks (Fig. 4c, d). The BHB network contained two keystone genera, Bacillus and Zoogloea, whereas the YHB network contained a larger number of keystone taxa, including Lactococcus, Enterococcus, and Aquabacterium. These results indicate that the YHB network involved a broader set of highly connected taxa, consistent with its greater network complexity. It should be noted that co-occurrence-based keystone taxa represent putative ecological roles and do not demonstrate direct microbial interactions or functional effects.

3.5. Comparative analysis of predicted microbial functional profiles

Predicted KEGG functional profiles indicated that the microbial communities in both BHB and YHB fermentation systems were dominated by metabolism-related functions (Fig. 5a). At KEGG level 1, metabolism was the predominant functional category, followed by genetic information processing and environmental information processing. At KEGG level 2, functions related to global and overview maps, carbohydrate metabolism, amino acid metabolism, energy metabolism, membrane transport, and the metabolism of cofactors and vitamins were consistently abundant across samples. The functional heatmap showed that the major predicted functions were broadly similar between cultivars and remained relatively stable throughout fermentation (Fig. 5b). This result is consistent with the overlapping bacterial community structures observed by PCoA and suggests that both fermentation systems retained a common functional foundation related to substrate utilization and energy generation. In contrast, several low-abundance functions, including cell motility, environmental adaptation, aging, and antineoplastic drug resistance, varied among fermentation stages and cultivars. At 72 h, differential analysis identified several predicted KEGG level 2 pathways that differed between BHB and YHB (Fig. 5c). Most of these differential pathways showed higher relative abundance in YHB. However, these functions were generally low in abundance and were derived from prediction based on microbial community composition. Therefore, they should be interpreted as potential functional differences rather than direct evidence of altered metabolic activity.

Fig. 5.

Fig. 5

Open in a new tab

KEGG functional prediction analysis of microbial communities in fermented black (BHB) and yellow (YHB) highland barley. (a) Relative abundance distribution of predicted KEGG level 2 functional pathways in BHB and YHB fermented samples. Different colors represent the corresponding KEGG level 1 functional categories. (b) Heatmap visualization of KEGG level 2 functional pathway abundances in BHB and YHB samples at different fermentation times. The colour gradient from blue to red indicates low to high predicted functional abundance. (c) Differential abundance analysis between BHB and YHB at fermentation 72 h. The bar plot on the left shows the mean relative abundance of each pathway in the two groups, while the dots and error bars on the right indicate the differences in mean proportions between BHB and YHB and their 95% confidence intervals. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.6. Differential analysis of non-volatile metabolites

Non-volatile metabolite profiles showed clear cultivar-dependent differences between BHB and YHB (Fig. 6). Hierarchical clustering and PCA consistently separated BHB from YHB, while the close clustering of biological replicates indicated good within-group reproducibility (Fig. 6a, b). Although most detected metabolites were shared between the two cultivars, BHB contained more cultivar-specific metabolites than YHB (Fig. 6c). These findings suggest that the two cultivars retained a common metabolic background but differed substantially in the abundance and accumulation patterns of individual metabolites.

Fig. 6.

Fig. 6

Open in a new tab

Differential analysis of non-volatile metabolites in fermented black (BHB) and yellow (YHB) highland barley. (a) Cluster heatmap of differential metabolites; (b) Principal component analysis (PCA); (c) Venn diagram; (d) Content distribution of differential metabolites; (e) KEGG pathway enrichment analysis. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The differential metabolites were assigned to 12 major chemical classes, including flavonoids, amino acids and their derivatives, phenolic acids, and alkaloids. In the BHB versus YHB comparison, the metabolites enriched in BHB were predominantly flavonoid- and phenolic acid-related compounds, including derivatives of isoscoparin, tricin, chrysoeriol, caffeoyl compounds, and 1-acetyl-beta-carboline (Fig. 6d). By contrast, several flavonoid derivatives and peptide- or amino acid-related metabolites were less abundant in BHB. KEGG enrichment analysis showed that the differential metabolites were mainly associated with flavonoid-related pathways, including chrysoeriol aglycone biosynthesis, apigenin C-glycoside biosynthesis, luteolin aglycone biosynthesis, anthocyanin biosynthesis, flavone and flavonol biosynthesis, and flavonoid biosynthesis (Fig. 6e). Together, these results indicate that cultivar-dependent metabolic divergence was primarily associated with the differential accumulation of flavonoids, phenolic acids, anthocyanin-related compounds, and related metabolic pathways. This pattern was consistent with the higher flavonoid and anthocyanin contents and stronger antioxidant properties observed in BHB.

3.7. Associations among antioxidant properties, differential metabolites and microbial communities

To evaluate potential associations among antioxidant properties, metabolite variation, and microbial community composition, Mantel tests and Spearman correlation analyses were performed using antioxidant indicators, the top 50 differentially accumulated metabolites (DAMs), and core microbial taxa. The Mantel test was used to evaluate overall associations between data matrices, whereas Spearman correlation analysis was used to identify representative pairwise relationships.

3.7.1. Associations between antioxidant properties and key differential metabolites

Mantel analysis indicated that variation in antioxidant properties was significantly associated with the overall profile of the top 50 DAMs, particularly flavonoids, phenolic acids, and several amino acid-derived metabolites (Fig. S2). Pairwise correlation analysis further showed that antioxidant-related indices were predominantly and positively associated with flavonoid- and phenolic acid-related metabolites (Fig. 7; Supplementary Data 2). Total flavonoid content was strongly associated with several flavonoid glycosides and phenolic derivatives, including chrysoeriol-7-O-rutinoside-5-O-glucoside, 1-O-caffeoyllysine, ellagic acid-4-O-xyloside, quercetin-5,4′-di-O-glucoside, and diosmetin-7-O-glucuronide (r = 0.941–0.943, p < 0.01). Total anthocyanin content was positively associated with gallic acid, 3-hydroxy-5-methoxybenzaldehyde, and 1-acetyl-beta-carboline (r = 0.943, p < 0.01). Among antioxidant activity indices, total antioxidant capacity and radical scavenging activities were also positively associated with selected flavonoids, flavonoid glycosides, phenolic compounds, and leucocyanidin. In contrast, several metabolites, particularly peptide-related compounds, showed negative associations with antioxidant indices. For example, Lys-Thr was negatively associated with total antioxidant capacity, while vitexin-2”-O-glucoside showed negative relationships with ABTS scavenging activity and total antioxidant capacity. These results indicate that the antioxidant properties of fermented highland barley were more closely associated with the accumulation of flavonoid- and phenolic acid-related metabolites than with peptide- or nucleotide-related metabolites. Because these analyses are correlative, they do not establish that individual metabolites directly determine antioxidant activity.

Fig. 7.

Fig. 7

Open in a new tab

Spearman correlation analysis of antioxidant properties and key differential metabolites (top 50 in VIP value) in fermented colored highland barley. *p < 0.05.

3.7.2. Associations between core microbial genera and key differential metabolites

Core microbial taxa showed distinct correlation patterns with the top 50 DAMs (Fig. 8; Supplementary Data 2). Among lactic acid bacteria, Lactiplantibacillus showed positive associations with several flavonoid and phenolic acid-related metabolites, including apigenin-7-O-neohesperidoside, chrysoeriol-7-O-rutinoside-5-O-glucoside, neosakuranin, and tricin-derived metabolites (r = 0.941–1.000, p < 0.01). Levilactobacillus was similarly associated with several antioxidant-related metabolites, including hydroxylated flavones, methyldopa, malonyl isosakuranin, and 6-hydroxydopaquinone (r = 0.880–0.943, p < 0.05). In contrast, several non-lactic bacterial genera were more strongly associated with peptide- and amino acid-derived metabolites. In particular, Aquabacterium, Bradyrhizobium, Lactococcus, and Microbacterium were positively associated with Lys-Thr, whereas Sphingobium was associated with Ile-His. Acidovorax, Brevundimonas, and Massilia were associated with Pro-Val-Thr and related peptide metabolites. In addition, Flavobacterium and Saccharomycopsis showed negative associations with several flavonoid and phenolic acid derivatives, despite positive associations with O-phosphate-L-tyrosine. By comparison, Enterococcus and Pediococcus showed weak or mostly non-significant associations with the top 50 DAMs. Although Pediococcus was a dominant late-stage taxon, its relative abundance was not strongly correlated with the differential metabolites selected for this analysis. On the whole, Lactiplantibacillus and Levilactobacillus were the taxa most closely associated with antioxidant-related flavonoid and phenolic derivatives, whereas several non-lactic genera were associated mainly with peptide- and amino acid-derived metabolites. These associations may contribute to the cultivar-specific metabolic and antioxidant differences observed during highland barley fermentation, but targeted cultivation or inoculation experiments are needed to verify their functional roles.

Fig. 8.

Fig. 8

Open in a new tab

Spearman correlation analysis of core microbial genera and key differential metabolites (top 50 in VIP value) in fermented colored highland barley. *p < 0.05.

4. Discussion

4.1. Dynamic changes in antioxidant capacity during the fermentation of colored highland barley

The antioxidant responses of black highland barley (BHB) and yellow highland barley (YHB) differed during fermentation. YHB showed a more pronounced late-stage increase in total phenolic content, accompanied by an increase in anthocyanin content. Although the initial total phenolic and anthocyanin contents of unfermented BHB were significantly higher than those of YHB (Table S1), the greater relative changes observed in YHB may be associated with differences in the proportions of bound phenolics and anthocyanin-related precursors in the two raw materials. During fermentation, these compounds may have been released from the grain matrix or converted into other extractable phenolic derivatives (He et al., 2023). However, targeted analysis of bound and free phenolic fractions would be needed to verify this interpretation. In contrast, total flavonoid content decreased gradually in both cultivars. This pattern suggests that flavonoids were not uniformly accumulated during fermentation and may have undergone degradation, deglycosylation, or conversion into other phenolic derivatives (Han et al., 2025). Therefore, the reduction in total flavonoid content should not be interpreted as direct evidence of reduced antioxidant potential. The antioxidant capacity of a fermented product may depend on the composition, concentration, redox properties, and interactions of individual metabolites rather than on the abundance of a single compound class alone. The increase in T-AOC, together with the observed changes in total phenolics and anthocyanins, indicated that fermentation was associated with altered antioxidant-related characteristics in both cultivars. In YHB, the greater increase in T-AOC was statistically consistent with the increase in total phenolic content and anthocyanins. Nevertheless, this association does not establish that these compounds directly accounted for the change in T-AOC, because other fermentation-derived metabolites may also affect the assay response. Despite the larger relative increase in T-AOC in YHB, BHB maintained higher DPPH, ABTS, HRSA, and SRSA values throughout fermentation. This result suggests that BHB retained a stronger radical scavenging profile under the conditions evaluated. The difference may be related to its higher initial contents of flavonoids, phenolics, anthocyanins, and other pigment-associated metabolites. Overall, fermentation was associated with distinct antioxidant trajectories in the two colored highland barley cultivars. YHB showed a greater relative response in total phenolics and T-AOC, whereas BHB retained stronger radical scavenging activities. These differences may reflect the combined effects of initial compositional variation and fermentation-associated changes in metabolite profiles (Zhao et al., 2025; Chen et al., 2022). Further targeted metabolite quantification and activity-guided fractionation are required to determine which compounds are primarily responsible for the observed antioxidant differences.

4.2. Microbial succession, co-occurrence patterns, and predicted functions during the fermentation of colored highland barley

To clarify whether the different antioxidant responses of BHB and YHB were associated with cultivar-specific microbial succession, the bacterial and fungal community dynamics during fermentation were further analyzed. The bacterial Shannon index decreased significantly in both BHB and YHB, while the fungal Shannon index approached zero during the middle and late fermentation stages (Table S2). These findings indicate a marked simplification of the microbial communities during fermentation. Such changes are commonly observed in cereal fermentation systems and may be associated with progressive substrate consumption, organic acid accumulation, pH reduction, and microbial competition, which can favor taxa adapted to the changing fermentation environment (Guo et al., 2020; Xu et al., 2018). The opposite Chao index trends in BHB and YHB suggest that bacterial richness changed differently between cultivars. In addition, differences in microbial alpha diversity were already evident at 0 h, indicating that the two raw materials harbored distinct initial microbial communities. These initial differences may be associated with cultivar-dependent variation in grain-associated microbiota, nutrient composition, phenolic composition, or other physicochemical properties (Guo et al., 2020). For example, BHB had a higher initial anthocyanin content than YHB (Table S1), which may have contributed to differences in the chemical environment encountered by microorganisms at the beginning of fermentation. However, the present data cannot determine whether anthocyanins directly shaped microbial colonization or succession. Controlled fermentation experiments using standardized inocula and defined phenolic conditions would be required to test this possibility.

At the genus level, Pediococcus and Lactococcus were dominant bacterial taxa, whereas Saccharomycopsis overwhelmingly dominated the fungal community. The early enrichment of Lactococcus was temporally associated with the initial restructuring of bacterial communities. Previous studies have reported that some Lactococcus strains can participate in acid production during fermentation (Wang et al., 2025; Chen et al., 2023). Similarly, Pediococcus and Lactococcus remained dominant at the late fermentation stage, indicating that lactic acid bacteria continued to shape the acidic and metabolic environment throughout fermentation. The overwhelming dominance of Saccharomycopsis largely accounted for the low fungal diversity, suggesting that fungal succession was strongly constrained by the selective fermentation environment and the competitive advantage of this genus in carbohydrate utilization. Sphingobium and Massilia showed relatively high abundance at the initial fermentation stage in both cultivars. Their early enrichment may be associated with the utilization or transformation of aromatic compounds, including phenolic acid- and lignin-derived substrates. However, their decline during fermentation suggests that these genera were less competitive under the later acidic and nutrient-limited conditions. LEfSe analysis further identified cultivar- and stage-associated biomarkers, including Levilactobacillus in BHB at 72 h and Lactococcus in YHB at 24 h. These biomarkers indicate differential taxonomic enrichment between cultivars and fermentation stages. They should not, however, be interpreted as microorganisms that directly determine cultivar-specific fermentation outcomes, because their occurrence may reflect both initial raw material-associated microbial differences and subsequent fermentation selection.

Because taxonomic composition alone could not fully explain the divergent fermentation characteristics of BHB and YHB, bacterial co-occurrence networks were analyzed to determine whether the two cultivars differed in microbial interaction patterns and potential ecological organization. The YHB network showed greater complexity and a higher proportion of positive correlations, indicating a more tightly connected bacterial community and stronger potential cooperative associations among taxa. Such interaction patterns may be relevant to the stronger fermentation response observed in YHB, because tightly connected taxa are more likely to jointly participate in substrate utilization, metabolite exchange, and the transformation of phenolic- or carbohydrate-related compounds (Hu et al., 2016). In contrast, the higher modularity of the BHB network suggested a more compartmentalized microbial organization. This modular pattern implies that bacterial taxa in BHB may be organized into relatively independent functional subgroups, potentially allowing different taxa to respond separately to substrate availability, acid stress, or phenolic compounds (Faust & Raes, 2012; Ban et al., 2026). The limited differences observed at the phylum level suggest that the varietal effects of BHB and YHB were more likely reflected in genus-level interaction patterns and the ecological roles of keystone taxa. Furthermore, the putative keystone taxa in the bacterial networks differed markedly between BHB and YHB. These highly connected taxa may contribute to network stability and intertaxon associations, and they may influence fermentation-related metabolic processes through their connections with other community members (Banerjee et al., 2018).

To further determine whether the different microbial interaction patterns between BHB and YHB were accompanied by functional differences, microbial functional profiles were predicted. Both fermentation systems shared abundant functions related to carbohydrate metabolism, amino acid metabolism, energy metabolism, and membrane transport. These predicted functions are consistent with the general metabolic requirements of microbial growth and substrate utilization during cereal fermentation. The broadly similar dominant pathways between BHB and YHB suggest a shared predicted functional framework. By contrast, several low-abundance pathways differed between the cultivars, with some showing higher predicted relative abundance in YHB. These differences may indicate subtle variation in functional potential; however, they should be interpreted cautiously because the functional profiles were inferred from taxonomic data rather than measured by metagenomics, metatranscriptomics, enzyme activity assays, or metabolite flux analysis (Chang et al., 2025). It is also important to note that disease-related KEGG categories, including Infectious disease, Cancer, Aging, and Drug resistance, are annotation categories derived from homologous gene assignments. Their presence in predicted functional profiles does not indicate pathogenicity, disease risk, or health-related functions of the fermented samples. In short, BHB and YHB shared a broadly similar predicted core functional profile but differed in microbial succession, co-occurrence patterns, and several low-abundance predicted pathways.

4.3. Differential characteristics of non-volatile metabolites in fermented colored highland barley

To explain why BHB maintained stronger radical scavenging capacity whereas YHB showed a more pronounced fermentation-induced increase in T-AOC, non-volatile metabolite profiles were compared between the two cultivars at the late fermentation stage (72 h). The non-volatile metabolite profiles of BHB and YHB were clearly separated, indicating substantial cultivar-associated differences. Differential metabolites were distributed among several chemical classes, including flavonoids, phenolic acids, amino acids and their derivatives, and alkaloids. This result suggests that both secondary metabolite variation and primary nutrient-related metabolism were associated with the compositional differences between the two fermented products. Among these metabolites, the top 10 upregulated compounds in fermented BHB were mainly flavonoids and phenolic acids with antioxidant relevance (Fig. 6d), which may partly explain the consistently higher radical scavenging activities observed in BHB during fermentation. In contrast, YHB showed greater accumulation of amino acids, small peptides, and related derivatives, indicating that its fermentation process may be more strongly associated with primary nutrient transformation and taste-related metabolite formation. Several compounds enriched in BHB, including chrysoeriol-7-O-(6″-malonyl)glucoside, tricin-7-O-(2″-feruloyl)glucoside-5-O-glucoside, and 1-O-caffeoyllysine, have been reported to possess antioxidant-related or other bioactive properties (Wang et al., 2021). Their relative enrichment may therefore be associated with the antioxidant characteristics of fermented BHB. KEGG enrichment analysis showed that the differential metabolites were mainly associated with flavonoid- and anthocyanin-related pathways, including chrysoeriol aglycone, apigenin C-glycoside, anthocyanin, flavone, and flavonol biosynthesis. This finding supports the view that flavonoid-related metabolic differences were a major feature distinguishing BHB and YHB at the late fermentation stage. The down-regulated metabolites in BHB were mainly primary metabolites, including amino acids and carbohydrates. Their decrease may reflect microbial utilization for growth and energy metabolism, as well as their conversion into organic acids, amino acid derivatives, or other metabolites (Chen et al., 2025; He et al., 2023). The distinct metabolite profiles of BHB and YHB may therefore reflect differences in their initial nutritional composition, phenolic reserves, microbial community composition, and fermentation-associated transformations (Choi et al., 2024; Zhao et al., 2025).

4.4. Correlation-based associations among antioxidant properties, differential metabolites, and core microorganisms

Correlation analyses showed that antioxidant indicators were significantly associated with several flavonoid glycosides, phenolic acids, and related metabolites. For example, total flavonoid content was positively correlated with chrysoeriol, quercetin, and diosmetin derivatives, while T-AOC and radical scavenging indices were associated with selected methoxyflavones, leucocyanidin, flavonoid glycosides, and phenolic derivatives. These findings indicate that variation in antioxidant properties was statistically associated with changes in specific metabolite subsets rather than with all metabolites uniformly (Scarano et al., 2023). The association patterns also differed among antioxidant assays. T-AOC was most strongly correlated with 6,7,8-tetrahydroxy-5-methoxyflavone, whereas ABTS radical scavenging activity was most strongly correlated with abbeokutone. These assay-specific relationships may be related to differences in the reaction mechanisms and chemical sensitivities of the antioxidant methods, as well as variation in the molecular structures and redox properties of the metabolites (Han et al., 2025). Several phenolic acids, ellagic acid derivatives, caffeoyl-containing compounds, and methoxyflavones were positively associated with anthocyanin content and antioxidant indicators. These patterns suggest potential associations between phenolic metabolite variation and antioxidant-related characteristics (Wijaya et al., 2025). In contrast, Lys-Thr and vitexin-2″-O-glucoside showed negative associations with T-AOC or ABTS activity. Such opposing correlations may reflect differences in intrinsic antioxidant properties, microbial utilization, or competition for shared metabolic precursors during fermentation (Leonard et al., 2021). Nevertheless, because correlation analysis cannot establish causality, the functional contribution of these candidate metabolites should be further verified by targeted quantification and activity-guided assays.

Microbial correlation analysis further indicated that Lactiplantibacillus and Levilactobacillus were positively correlated with several flavonoid and phenolic acid derivatives. Specifically, Lactiplantibacillus was positively associated with apigenin-7-O-neohesperidoside and neosakuranin, whereas Levilactobacillus was positively associated with 6,7,8-tetrahydroxy-5-methoxyflavone and malonyl isosakuranin. These results indicate significant positive correlations and suggest potential associations between these lactic acid bacterial genera and antioxidant-related metabolite profiles. Some lactic acid bacteria have been reported to possess β-glucosidase, esterase, and phenolic acid decarboxylase activities, which can participate in phenolic compound transformation under appropriate conditions (Hou et al., 2025; Landete et al., 2021; Leonard et al., 2021; Liu et al., 2026). Several other bacterial genera, including Acidovorax, Aquabacterium, Bradyrhizobium, Lactococcus, Massilia, Microbacterium, and Sphingobium, were positively associated with peptide- and amino acid-derived metabolites, such as Lys-Thr, Ile-His, and Pro-Val-Thr. These correlations suggest that these taxa may be associated with nitrogen-related metabolic variation during fermentation. Although some peptides and amino acid derivatives may exhibit antioxidant activity depending on their sequence and composition (Xia et al., 2012), the peptide-related metabolites identified here showed negative or inconsistent associations with antioxidant indices. Notably, Flavobacterium and Saccharomycopsis displayed association patterns opposite to those of Lactiplantibacillus and Levilactobacillus. They were positively correlated with O-phosphate-L-tyrosine but negatively correlated with several antioxidant-related flavonoid and phenolic acid derivatives, including leucocyanidin and chrysoeriol-7-O-rutinoside-5-O-glucoside. This pattern suggests that these taxa and metabolites changed in opposite directions across the samples. Similarly, the weak or non-significant associations of Enterococcus and Pediococcus with the top 50 DAMs indicate that their relative abundances were not strongly correlated with the selected differential metabolites, although these genera may still be relevant to other aspects of fermentation. In summary, the correlation analyses reveal statistically significant associations among microbial taxa, non-volatile metabolites, and antioxidant properties. They identify Lactiplantibacillus and Levilactobacillus, together with selected flavonoid and phenolic derivatives, as candidate factors associated with antioxidant-related variation.

4.5. Limitations and future perspectives

This study integrated antioxidant-related quality evaluation, microbial community profiling, predicted functional analysis, and non-volatile metabolomics to characterize cultivar-dependent changes during highland barley fermentation. However, several limitations should be acknowledged. First, the microbial, metabolomic, and antioxidant datasets were analyzed primarily through comparative and correlation-based approaches. Therefore, the observed associations do not establish direct causal relationships between individual microorganisms, metabolites, and antioxidant properties. Second, the microbial functional profiles were predicted from taxonomic composition and were not validated using metagenomic, metatranscriptomic, proteomic, or enzyme activity analyses. The inferred KEGG pathways should therefore be regarded as potential functional profiles rather than direct measurements of microbial metabolic activity. Third, physicochemical parameters that may influence microbial succession and phenolic compound behavior, including pH, organic acid concentrations, and free versus bound phenolic fractions, were not comprehensively evaluated.

Future studies should use controlled validation experiments to test the causal hypotheses generated here. Sterile or microbiologically controlled fermentation could be used to distinguish endogenous grain-associated transformations from microbially associated changes. Isolation of candidate strains, followed by mono-culture and defined co-culture fermentation experiments, would help determine whether Lactiplantibacillus, Levilactobacillus, or other taxa are associated with reproducible changes in specific phenolic metabolites (Wen et al., 2026). Inoculation and depletion experiments, together with exogenous addition of purified phenolic compounds or relevant precursors, could further evaluate microbial-metabolite relationships. Where feasible, gene knockout, gene silencing, or heterologous expression of candidate microbial enzymes, such as β-glucosidases and esterases, could provide direct evidence for specific transformation pathways (Yu et al., 2025). Finally, time-resolved targeted metabolomics, metagenomics or metatranscriptomics, and activity-guided fractionation would strengthen the mechanistic interpretation of antioxidant metabolite formation during colored highland barley fermentation (Huang et al., 2024; Ren et al., 2026).

5. Conclusion

This study investigated the associations among microbial succession, metabolite variation, and antioxidant properties during the fermentation of colored highland barley using multi-omics integration analysis. Fermentation showed common effects in both cultivars, including reduced microbial diversity, enrichment of lactic acid bacteria, dominance of Saccharomycopsis in the fungal community, an overall increase in total phenolics and anthocyanins, and a decline in total flavonoids. Nevertheless, BHB retained higher total flavonoid and anthocyanin contents and stronger total antioxidant capacity, DPPH, ABTS, hydroxyl radical, and superoxide radical scavenging activities, whereas YHB showed a greater increase in total phenolic content during fermentation. The bacterial succession patterns were broadly similar, with Pediococcus enriched at the late stage in both cultivars, but cultivar-specific biomarkers and network structures were observed; YHB showed a more densely connected and positively correlated bacterial network, whereas BHB exhibited greater network modularity. BHB was relatively enriched in flavonoid- and phenolic acid-related compounds and related flavonoid biosynthetic pathways, while YHB showed stronger associations with amino acid- and peptide-related metabolites and several low-abundance predicted functional pathways. Correlation analysis indicated that Lactiplantibacillus and Levilactobacillus were positively associated with selected flavonoid and phenolic acid metabolites. However, this study primarily revealed correlations based on multi-omics data, and the causal relationships among specific microorganisms, key metabolites, and antioxidant activity still require further investigation.

CRediT authorship contribution statement

Mengru Han: Writing – original draft, Methodology, Formal analysis, Data curation. Hanghang Hou: Methodology, Formal analysis. Honglu Wang: Conceptualization. Miaomiao Zhang: Methodology, Formal analysis. Xinyu Tang: Methodology. Xirong Zhou: Resources. Meijin Liu: Resources. Changshou Ma: Validation. Qinghua Yang: Writing – review & editing, Validation, Project administration, Formal analysis. Baili Feng: Project administration, Funding acquisition, Formal analysis.

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.

Acknowledgments

This research was supported by the Agriculture Research System of China of MOF and MARA (CARS-06-A26), Zhunge'er Miscellaneous Grains Science and Technology Experimental Station, Key Research and Development Project of Ordos City (YF20250259), and PhD Student Special Program of the Youth Science and Technology Talent Cultivation Project, China Association for Science and Technology.

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.104424.

Contributor Information

Qinghua Yang, Email: qinghuayang@nwsuaf.edu.cn.

Baili Feng, Email: fengbaili@nwsuaf.edu.cn.

Appendix A. Supplementary data

Supplementary material 1

The supplementary tables and figures of this article.

mmc1.docx (1.1MB, docx)
Supplementary material 2

Top 50 differentially accumulated non-volatile metabolites in black and yellow highland barley.

mmc2.xlsx (16KB, xlsx)
Supplementary material 3

Data of correlation analysis among antioxidant properties, differential metabolites and microbial communities.

mmc3.xlsx (61.8KB, xlsx)

Data availability

Data will be made available on request.

References

  1. Ban S., He B., Song Y., An X.L., Xu Y., Wu Q., Shi Y. Biosynthesis gene cluster-containing species drive microbial community stability under thermal stress through enhanced ecological cohesion. iMetaOmics. 2026;3(1) [Google Scholar]
  2. Banerjee S., Schlaeppi K., Van Der Heijden M.G. Keystone taxa as drivers of microbiome structure and functioning. Nature Reviews Microbiology. 2018;16(9):567–576. doi: 10.1038/s41579-018-0024-1. [DOI] [PubMed] [Google Scholar]
  3. Bui A.T., Williams B.A., Hoedt E.C., Morrison M., Mikkelsen D., Gidley M.J. High amylose wheat starch structures display unique fermentability characteristics, microbial community shifts and enzyme degradation profiles. Food & Function. 2020;11(6):5635–5646. doi: 10.1039/d0fo00198h. [DOI] [PubMed] [Google Scholar]
  4. Cao Y., Du M., Kan J., Chen Z. Changes in chemical components during fermentation of highland barley wine with multi-strain starter. Food Science. 2012;33(11):252–256. [Google Scholar]
  5. Chang H., Zhang Q., Soro P.L., Zhang W., Ma L., Feng Z., Gu C. Integrative metagenomic, metabolomic, and transcriptomic analyses unravel flavor development driven by a defined starter cocktail in cocoa fermentation. Food Research International. 2025;225:118042. doi: 10.1016/j.foodres.2025.118042. [DOI] [PubMed] [Google Scholar]
  6. Chen L., Peng Q., Chen Y., Che F., Chen Z., Feng S. Analysis of dominant microorganisms and core enzymes in Qingke baijiu Daqu by high-throughput sequencing and proteomics. Food Research International. 2025;204 doi: 10.1016/j.foodres.2025.115941. [DOI] [PubMed] [Google Scholar]
  7. Chen L., Wang S., Li D., Feng S. Correlations between microbes and metabolites of hulless barley wines fermented with varieties of hulless barley and different starters. LWT. 2021;152 [Google Scholar]
  8. Chen X., Wang Z., Li Y., Liu Q., Yuan C. Survey of the phenolic content and antioxidant properties of wines from five regions of China according to variety and vintage. LWT. 2022;169 [Google Scholar]
  9. Chen Z., Zhang L., Peng M., Zhu S., Wang G. Preharvest application of selenite enhances the quality of Chinese flowering cabbage during storage via regulating the ascorbate-glutathione cycle and phenylpropanoid metabolisms. Food Research International. 2023;163 doi: 10.1016/j.foodres.2022.112229. [DOI] [PubMed] [Google Scholar]
  10. Choi B., Rangel A.T., Kerkhoven E.J., Nygård Y. Engineering of Saccharomyces cerevisiae for enhanced metabolic robustness and L-lactic acid production from lignocellulosic biomass. Metabolic Engineering. 2024;84:23–33. doi: 10.1016/j.ymben.2024.05.003. [DOI] [PubMed] [Google Scholar]
  11. Faust K., Raes J. Microbial interactions: From networks to models. Nature Reviews Microbiology. 2012;10(8):538–550. doi: 10.1038/nrmicro2832. [DOI] [PubMed] [Google Scholar]
  12. Feng C.Y., Wang W.W., Ye J.F., Li S.S., Wu Q., Yin D.D.…Wang L.S. Polyphenol profile and antioxidant activity of the fruit and leaf of Vaccinium glaucoalbum from the Tibetan Himalayas. Food Chemistry. 2017;219:490–495. doi: 10.1016/j.foodchem.2016.09.083. [DOI] [PubMed] [Google Scholar]
  13. Feng Y., Tian L., Zhang F., Zhang Y., Hou H., Wang M.…Gao X. Modulating effects of herbicide-safener co-application on rhizosphere microbiota and soil function in proso millet (Panicum miliaceum L.) Journal of Hazardous Materials. 2025;499:140320. doi: 10.1016/j.jhazmat.2025.140320. [DOI] [PubMed] [Google Scholar]
  14. Gao Y., Ping H., He Z., Liu J., Zhao M., Ma Z. Characterization of the active components and bioaccessibility of phenolics in differently colored foxtail millets. Food Chemistry. 2024;452 doi: 10.1016/j.foodchem.2024.139355. [DOI] [PubMed] [Google Scholar]
  15. Gaur G., Gänzle M.G. Conversion of (poly) phenolic compounds in food fermentations by lactic acid bacteria: Novel insights into metabolic pathways and functional metabolites. Current Research in Food Science. 2023;6 doi: 10.1016/j.crfs.2023.100448. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Ge X., Jing L., Zhao K., Su C., Zhang B., Zhang Q.…Li W. The phenolic compounds profile, quantitative analysis and antioxidant activity of four naked barley grains with different color. Food Chemistry. 2021;335 doi: 10.1016/j.foodchem.2020.127655. [DOI] [PubMed] [Google Scholar]
  17. Guo L., Luo Y., Zhou Y., Bianba C., Guo H., Zhao Y., Fu H. Exploring microbial dynamics associated with flavours production during highland barley wine fermentation. Food Research International. 2020;130 doi: 10.1016/j.foodres.2019.108971. [DOI] [PubMed] [Google Scholar]
  18. Guo T., Zhou X., Zhao S., Yao Y., Dong B., Zhao G. Microbiota interactions as critical determinants of flavor development in fermented foods. Trends in Food Science & Technology. 2025;164 [Google Scholar]
  19. Han M., Zhang X., Kong Y., Cui Z., Qiu Y., Dong X.…Feng B. Analysis of antioxidant capacity and wine quality characteristics of fermented colored highland barley based on metabolomics. Food Chemistry: X. 2025;30:102958. doi: 10.1016/j.fochx.2025.102958. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. He C., Liu X., Zhang H., Mu T., Zhang Y., Ren X.…Wang M. Enhancement of the release of phenolic compounds from white and black Qingke bran by autoclaving and fermentation treatments. Food Bioscience. 2023;53 [Google Scholar]
  21. Hou F., Su T., Chen Y., Dong L., Zhang M., Huang F. Novel insights into bound phenolics: Conversion and release of phenolic compounds in lychee pulp by heat pump drying and lactic acid bacterial fermentation. Food Research International. 2025;208 doi: 10.1016/j.foodres.2025.116201. [DOI] [PubMed] [Google Scholar]
  22. Hu X., Du H., Ren C., Xu Y. Illuminating anaerobic microbial community and cooccurrence patterns across a quality gradient in Chinese liquor fermentation pit muds. Applied and Environmental Microbiology. 2016;82(8):2506–2515. doi: 10.1128/AEM.03409-15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Huang G., Zhang M., Zhang S., Wang J., Zhang R., Dong L.…Deng M. Unveiling biotransformation of free flavonoids into phenolic acids and chromones alongside dynamic migration of bound phenolics in lactobacillus-fermented lychee pulp. Food Chemistry. 2024;457 doi: 10.1016/j.foodchem.2024.140115. [DOI] [PubMed] [Google Scholar]
  24. Landete J.M., Plaza-Vinuesa L., Montenegro C., Santamaría L., Reverón I., De Las Rivas B., Muñoz R. The use of Lactobacillus plantarum esterase genes: A biotechnological strategy to increase the bioavailability of dietary phenolic compounds in lactic acid bacteria. International Journal of Food Sciences and Nutrition. 2021;72(8):1035–1045. doi: 10.1080/09637486.2021.1900078. [DOI] [PubMed] [Google Scholar]
  25. Leonard W., Zhang P., Ying D., Adhikari B., Fang Z. Fermentation transforms the phenolic profiles and bioactivities of plant-based foods. Biotechnology Advances. 2021;49 doi: 10.1016/j.biotechadv.2021.107763. [DOI] [PubMed] [Google Scholar]
  26. Li Z., Chen Y., Zheng J., Su J., Lu Y., Xia Y., Luo H. From fungi to lactic acid bacteria: Dominant microorganisms govern stage-specific esters biosynthesis in pit microecology of strong-flavor baijiu. Food Research International. 2025;222:117797. doi: 10.1016/j.foodres.2025.117797. [DOI] [PubMed] [Google Scholar]
  27. Li Z., Ren Z., Zhao L., Chen L., Yu Y., Wang D., Yang H. Unique roles in health promotion of dietary flavonoids through gut microbiota regulation: Current understanding and future perspectives. Food Chemistry. 2023;399 doi: 10.1016/j.foodchem.2022.133959. [DOI] [PubMed] [Google Scholar]
  28. Liu Y., Zhao M., Zhong S., Wu G., Yang F., Zhou J. Review on mining robust lactic acid bacteria for next-generation silage inoculants via multi-omics. Life. 2026;16(1) doi: 10.3390/life16010108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lu H., Yang F., Jia F., Liu Y., Xue B. The related research and prospect analysis of Tibetan traditional highland barley wine. Light Industry Science and Technology. 2021;37(5):19–21. [Google Scholar]
  30. Ma S., Li Y., Dong Y., Zhou Y., Tu F., Zhang L., Wu C. Stochastic and deterministic processes shape microbial communities and flavor characteristics of strong-flavor daqu during storage. International Journal of Food Microbiology. 2025;429 doi: 10.1016/j.ijfoodmicro.2024.111017. [DOI] [PubMed] [Google Scholar]
  31. Okarter N., Liu R.H. Health benefits of whole grain phytochemicals. Critical Reviews in Food Science and Nutrition. 2010;50(3):193–208. doi: 10.1080/10408390802248734. [DOI] [PubMed] [Google Scholar]
  32. Ren H., Fan J., Guo X., Zhang B., Zhao H., Ma D.…Ding X. Deciphering the characteristics of strong-flavor Daqu at different grades through integrated microbiome and metabolome analysis. Food Chemistry: X. 2026;33:103526. doi: 10.1016/j.fochx.2026.103526. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Scarano A., Laddomada B., Blando F., De Santis S., Verna G., Chieppa M., Santino A. The chelating ability of plant polyphenols can affect iron homeostasis and gut microbiota. Antioxidants. 2023;12(3):630. doi: 10.3390/antiox12030630. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Schloss P.D., Westcott S.L., Ryabin T., Hall J.R., Hartmann M., Hollister E.B.…Weber C.F. Introducing mothur: Open-source, platform-independent, community-supported software for describing and comparing microbial communities. Applied and Environmental Microbiology. 2009;75(23):7537–7541. doi: 10.1128/AEM.01541-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Wang H., Luo Y., Zhang Y., Xiong Y., Bian G., Feng B. Flavor quality during fermentation of the specialty product proso millet (Panicum miliaceum L.) thick wine: Insights from microbiome and metabolomics analyses. Food Research International. 2025;223:117835. doi: 10.1016/j.foodres.2025.117835. [DOI] [PubMed] [Google Scholar]
  36. Wang M., Zhao H., Wen X., Ho C.T., Li S. Citrus flavonoids and the intestinal barrier: Interactions and effects. Comprehensive Reviews in Food Science and Food Safety. 2021;20(1):225–251. doi: 10.1111/1541-4337.12652. [DOI] [PubMed] [Google Scholar]
  37. Wen Z., Wei Y.H., Zhu H.Y., Song L., Han D.Y., Guo L.C.…Han P.J. Integrated microbiome and metabolome analyses reveal spatial heterogeneity of medium-temperature Daqu and its potential impact on simulated strong-flavor baijiu fermentation. Food Chemistry: X. 2026;38:104225. doi: 10.1016/j.fochx.2026.104225. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Wijaya G.Y.A., Vornoli A., Giambastiani L., Digiacomo M., Macchia M., Szymczak B.…Longo V. Solid-state fermented cereals: Increased phenolics and their role in attenuating liver diseases. Nutrients. 2025;17(5) doi: 10.3390/nu17050900. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Xia H., Yan X.Y., Lu L.D., Cao Y.N., Peng L.X., Zou L., Zhao G. Progress on nutritional function and high-value utilization of hulless barley. Science and Technology of Food Industry. 2022;43:403–413. [Google Scholar]
  40. Xia Y., Bamdad F., Gänzle M., Chen L. Fractionation and characterization of antioxidant peptides derived from barley glutelin by enzymatic hydrolysis. Food Chemistry. 2012;134(3):1509–1518. doi: 10.1016/j.foodchem.2012.03.063. [DOI] [PubMed] [Google Scholar]
  41. Xu J., Wu H., Wang Z., Zheng F., Lu X., Li Z., Ren Q. Microbial dynamics and metabolite changes in Chinese Rice wine fermentation from sorghum with different tannin content. Scientific Reports. 2018;8(1):4639. doi: 10.1038/s41598-018-23013-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Yu Q., Li J., Bai J., Huang M., He Y., Zhao Y., Xiao X. Enzymatic mechanism of a β-glucosidase from Lactiplantibacillus plantarum Dy-1 with potential applications in the release of bound phenolics in fermentation barley. Journal of Agricultural and Food Chemistry. 2025;73(7):4164–4173. doi: 10.1021/acs.jafc.4c09865. [DOI] [PubMed] [Google Scholar]
  43. Zhai X.T., Wang H.Q., Han L., Wang F.Y., Zhang Y.H., Li J.…Wang M. Impact of solid-state fermentation by Aspergillus niger 4Q and Cyberlindnera fabianii J2 on antioxidant and flavor properties of whole grain barley. Cereal Chemistry. 2023;100(4):914–926. [Google Scholar]
  44. Zhang X., Han M., Li Y., Liu M., Zhou X., Li Y., Feng B. The differences in nutritional quality and antioxidant capacity of colored highland barley. Journal of the Chinese Cereals and Oils Association. 2025;40(07):80–87. [Google Scholar]
  45. Zhang Y., Kang J., Han B.Z., Chen X. Wheat-origin Bacillus community drives the formation of characteristic metabolic profile in high-temperature Daqu. Lwt. 2024;191 [Google Scholar]
  46. Zhang Y., Xu Z., Gu Z., Cheng L., Hong Y., Li L. Effect of anthocyanins on the in vitro fermentation of high-amylose starch. Carbohydrate Polymers. 2025;353 doi: 10.1016/j.carbpol.2025.123271. [DOI] [PubMed] [Google Scholar]
  47. Zhao Y., Leng F., Fan S., Ding Y., Chen T., Zhou H., Xiao X. Changes of barley bound phenolics and their characteristics during simulated gastrointestinal digestion and colonic fermentation in vitro. Foods. 2025;14(7) doi: 10.3390/foods14071114. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Supplementary material 1

The supplementary tables and figures of this article.

mmc1.docx (1.1MB, docx)
Supplementary material 2

Top 50 differentially accumulated non-volatile metabolites in black and yellow highland barley.

mmc2.xlsx (16KB, xlsx)
Supplementary material 3

Data of correlation analysis among antioxidant properties, differential metabolites and microbial communities.

mmc3.xlsx (61.8KB, xlsx)

Data Availability Statement

Data will be made available on request.

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

RESOURCES