The polyphenol extract used in this study was a commercial sugarcane-derived product (Polygain™; The Product Makers Australia, Keysborough, Australia). Its chemical composition was not analyzed in the present study; therefore, compositional information was obtained from previously published characterizations of the same extract (Deseo et al., 2020; Daneshmand et al., 2021). Untargeted GC–MS profiling has revealed more than 100 metabolites, including amino acids, organic acids, sugars, sugar alcohols, and phenolic compounds, with trans-4-hydroxycinnamic acid, vanillic acid, pyroglutamate, and mono-/disaccharides (e.g., fructose, sorbose, glucose) consistently reported as the most abundant constituents (Daneshmand et al., 2021). Published analyses also report high antioxidant capacity based on ORAC assays and detectable levels of minerals and carbohydrates in the extract (Deseo et al., 2020). Because these data originate from previous studies and the product is plant-derived, the exact concentrations of individual metabolites may vary between production batches.

The daily dose (50 mL/cow per day) was selected based on proportional scaling from ruminant studies using the same extract. Prathap et al. (2024) administered 3.2 mL/day to 40-kg sheep; when adjusted by body weight, an equivalent dose for a 650-kg cow approximates 50 mL/day. This scaling ensured biological comparability with existing ruminant research and was consistent with doses previously reported to influence ruminal fermentation and antioxidant responses.

The cows were fed a total mixed ration (TMR) formulated by NASEM (2021), and composed of sorghum silage, brewer’s grains, a commercial concentrate (Agroindústria Três Irmãos, Mococa, SP, Brazil), and Tifton 85 hay (Table 1). The TMR was offered twice daily (07:00 and 14:00 h) during milking. Feed was provided ad libitum, with daily intake determined by weighing the feed offered and the orts collected the following morning.

Representative samples of each ingredient and the TMR were collected daily and pooled weekly for chemical analysis. Dry matter (DM; method 930.15), ashes (method 942.05), ether extract (EE; method 920.39), and crude protein (CP; N × 6.25; method 984.13) were determined according to the Association of Official Analytical Chemists (2012). Starch was determined using the commercial kit Total Starch Assay Kit AA/AMG – Megazyme (method 996.11). Ash-free neutral detergent fiber (NDF), acid detergent fiber (ADF), and lignin were determined according to the procedures of Van Soest et al. (1991). Non-fiber carbohydrates (NFC) were estimated using the equation proposed by Mertens (1997): NFC (%DM) = 100 – (CP + EE + NDF + Ashes), where CP, EE, NDF, and ash are expressed as a percentage of DM.

Blood samples were collected from all cows on days 0, 21, and 42 of the experimental periods, 2 h after the morning feeding, by jugular venipuncture. Samples were obtained using vacuum tubes containing sodium fluoride with antiglycolytic agents (Vacuette^®, Campinas, Brazil) and tubes without anticoagulant (Vacuette^®, Campinas, Brazil). Immediately after collection, the tubes were packed in insulated boxes with ice and transported to the laboratory. Samples were centrifuged at 2,000 × g for 15 min at 4 °C, and the resulting plasma and serum were transferred into plastic microtubes and stored at −20°C until analysis. The concentrations of glucose (ref. 1331), total protein (ref. 99.1), urea (ref. 104.1), and creatinine (ref. 96.1) were determined using an automated biochemical analyzer (SBA-200, CELM, Barueri, Brazil) and commercial enzymatic kits (Labtest Diagnóstica S.A., Lagoa Santa, Brazil). An inter and intra assay specific control was used for each analyte according to each commercial kit. Differences between assays were accepted if were lower than 3%.

Methane emissions were quantified between weeks 4 and 5 using the sulfur hexafluoride (SF₆) tracer technique originally described by Johnson and Johnson (1995) adapted for Brazilian conditions (Primavesi et al., 2004). Permeation tubes were pre-conditioned for 6 weeks at 39°C in a forced-air oven and weighed weekly to determine individual SF₆ permeation rates (mg·d^-¹). After conditioning, tubes were orally dosed into the reticulorumen one week before sampling. All cows were adapted to the halter–capillary–canister system for 7 days to ensure normal feeding and rumination behavior before measurements.

Gas collection was carried out for 6 consecutive 24-h periods. A stainless-steel capillary tube attached to the halter continuously sampled air from the breathing zone and delivered it into an evacuated stainless-steel canister. The capillary dimensions (internal diameter and length) were adjusted to achieve a filling curve that resulted in a final absolute pressure of approximately 51 kPa (0.5 atm) in the canister after each 24-h collection period. Analyzers and calibration gases were tested daily to ensure linearity and stable flow.

At the end of each sampling period, canisters were pressurized to approximately 122 kPa (1.2 atm) using high-purity nitrogen to ensure adequate sample volume and analytical precision. Methane and SF₆ concentrations were determined by gas chromatography equipped with a flame ionization detector for CH₄ and an electron capture detector for SF₆, following analytical procedures described by Johnson et al. (2007). Quality control involved blank canisters, duplicate injections, and standard curves prepared with certified mixtures.

Daily methane emissions (g·cow^-¹·d^-¹) were calculated using the individual SF₆ release rate and the CH₄:SF₆ molar concentration ratio obtained from each canister, assuming steady release during the 24-h period. The final emission value for each cow was the mean of the 6 collection days.

Rumen fluid samples were collected on days 0, 21, and 42 of the experiment, 2 h after the morning feeding, using an esophageal probe. The probe consisted of a flexible hose (1.5m in length, 1.27cm internal diameter, 0.3 cm wall thickness) with a rounded open end and no side holes, connected to a vacuum pump (Model TE-0581, Tecnal Ltda., Piracicaba, SP, Brazil). Sampling was performed using disposable gloves. The initial 50 mL of rumen fluid was discarded to minimize saliva carryover. If visual inspection indicated the presence of mucus, excessive froth, or abnormally low particle content, characteristics consistent with salivary contamination, additional fluid was discarded until the sample showed typical rumen appearance. Sample integrity was therefore monitored continuously during collection according to the visual criteria described by Terré et al. (2013). Because rumen sampling by esophageal probe may still result in partial dilution by saliva, even after discarding the first aliquot, our procedure follows previously validated approaches in dairy cows, where salivary dilution does not prevent accurate evaluation of pH, SCFA concentrations, or microbial community structure when standard visual screening is applied (de Assis Lage et al., 2020). The collected rumen fluid was filtered through cotton gauze.

Immediately after collection, one aliquot was used to determine pH with a digital potentiometer (Model TEC-5, Tecnal Ltda., Piracicaba, SP, Brazil). Three additional aliquots were transferred into sterile plastic tubes (in triplicate) and stored at −20 °C until further analysis of short-chain fatty acids (SCFA) and ruminal ammonia, and another was stored at -80°C for microbiome analysis.

The SCFA and ruminal ammonia nitrogen (NH₃-N) were analyzed at the Animal Nutrition and Reproduction Laboratory, Department of Animal Science, USP/ESALQ, according to the method described by de Paula et al. (2017).

Data on ruminal fermentation profile and blood metabolites were analyzed as repeated measures using the MIXED procedure of SAS (version 9.4; SAS Institute Inc., Cary, NC, USA). The model included supplementation with Polyphenol as a fixed effect and block as a random effect. Differences were declared significant at P ≤ 0.05, and trends were considered at P < 0.10. Least squares means were estimated using the LSMEANS command.

To conduct a statistical analysis of the microbiome, we first assessed the homogeneity of variances and the normality of the data using Levene’s and Shapiro-Wilk’s tests. Then, the prokaryotic community structure and its correlation with metabolites were evaluated using Redundancy Analysis (RDA), and differences among treatments were tested with PERMANOVA. The analysis of diversity was performed using PAST 4.01 [25] and compared using Tukey’s method. Differences in microbial abundance at the phylum and ASV levels were compared using the Statistical Analysis of Metagenomic Profiles (STAMP) (Parks and Beiko, 2010; Parks et al., 2014), with p-values calculated using a two-sided Tukey-Kramer test and corrected for multiple comparisons according to Benjamini and Hochberg (1995). Additionally, to validate differential-abundance results, we performed a Random Forest analysis, which provided an independent assessment of the ASVs contributing most strongly to treatment discrimination.

The average dry matter intake was 21.8 ± 0.48 kg/d per cow. Supplementation with polyphenols tended to increase total SCFA concentration in the rumen compared with the Control (66.44 vs. 58.44 mM; P = 0.08), although no treatment effects were detected for the molar proportions of individual SCFA (Table 2). A tendency for a greater propionate proportion was observed in cows fed polyphenols (18.89 vs. 18.13 mol/100 mol; P = 0.08), whereas acetate, butyrate, valerate, the acetate-to-propionate ratio, branched-chain fatty acids, and ammonia-N were not significantly affected by the supplementation (Table 2). Week influenced acetate proportion (P = 0.05; Supplementary Figure S1A), isobutyrate proportion (P = 0.01; Supplementary Figure S1B), and isovalerate (P = 0.07; Supplementary Figure S1C). Ruminal pH was not affected by supplementation, but varied across weeks (P < 0.01; Supplementary Figure S1D). Methane emissions were not influenced by polyphenol supplementation. Regarding blood parameters, polyphenol supplementation significantly increased plasma glucose concentration (56.74 vs. 53.18 mg/dL; P < 0.01). No treatment effects were observed for plasma total protein, urea, or creatinine, although all parameters varied significantly over time (P < 0.01; Table 2; Supplementary Figure S2).

The β-diversity analysis revealed significant effects of both treatment (R2 = 0.061, P = 0.038) and time (R2 = 0.112, P = 0.037), as well as a positive correlation (P < 0.05) between valerate concentrations and total protein with Polyphenol supplementation (Figure 1A). The α-diversity metrics (ASV richness and Shannon index) did not differ significantly between groups (Figure 1B).

A total of nine bacterial phyla were detected. The rumen microbiota was dominated by Bacteroidota and Firmicutes, which together accounted for the majority of sequences in all groups (>80% of all sequences). Other phyla, including Proteobacteria, Spirochaetota, Verrucomicrobiota, Patescibacteria, Cyanobacteria, Fibrobacterota, and Desulfobacterota, were present at lower abundances (< 0.5%; Figure 2A). Polyphenol supplementation reduced the abundance of unclassified bacteria and significantly increased the abundance of Planctomycetota relative to the Control (P < 0.05; Figure 2B). Planctomycetota displayed a strong positive response to the polyphenol amendment, with mean relative abundance rising from 0.75% in the Control to 2.70% under polyphenol addition, an absolute increase of +1.95 percentage points, corresponding to a ~259% relative enrichment. In the control group, Desulfobacterota abundance varied over time (P < 0.05), with the lowest at T1, the highest at T2, and intermediate at T3. Planctomycetota peaked at T2, while Thermoplasmatota, the main archaeal group detected in our dataset, increased from T1 to T2 and remained stable at T3. Campylobacterota reached its highest abundance at T2, differing from T3 but not from T1. In the polyphenol group, Chloroflexi abundance was greater at T2, while Thermoplasmatota increased from T1 to T2 and stabilized at T3. Campylobacterota was also enriched at T2, with no significant difference between T3 and earlier time points (Figure 2B).

A total of 121 ASVs differed between the Control and Polyphenol groups across all time points. Hierarchical clustering grouped them into five major clusters (I–V), with distinct patterns of enrichment: clusters I and II were more abundant in Control, whereas clusters III and V showed ASVs enriched under Polyphenol supplementation (Figure 3A). Figure 3B presents a pairwise comparison between treatments, highlighting the 20 most differentially abundant ASVs, which include members of Prevotella, Saccharofermentans, Lachnospiraceae, Treponema, and Muribaculaceae. In total, 61 ASVs were more abundant in the Control and 60 in the Polyphenol group, indicating that supplementation modulated microbial composition without markedly increasing or reducing the overall number of enriched taxa. This outcome is further supported by the Random Forest analysis, which independently identified Prevotella/Prevotellaceae, Lachnospiraceae, Treponema, and the Rikenellaceae RC9 gut group as the most discriminatory taxa between treatments (Supplementary Figure S3), substantially overlapping with those highlighted in Figure 3B. Together, these convergent results demonstrate that the treatment-associated differences are robust across analytical approaches.

Further, we compared the effect of time within each treatment to determine whether microbial responses followed consistent temporal trajectories under Control and Polyphenol conditions. This analysis allowed us to identify ASVs that shifted specifically at each time point and to characterize how these dynamics differed between treatments. The results revealed marked temporal restructuring of the microbiome, with distinct groups responding at T1, T2, and T3 within each condition. This analysis revealed 16, 25, and 39 ASVs altered at T1, T2, and T3, respectively, in the control group. Shifts occurred mainly within Rikenellaceae, Bacteroidales, Prevotellaceae, Anaeroplasmataceae, Christensenellaceae, and Lachnospiraceae (Figure 4A). In the Polyphenol group, 18, 16, and 26 ASVs were differentially abundant at T1, T2, and T3, including members of Prevotellaceae, Rikenellaceae, Erysipelotrichaceae, Absconditabacteriota, Oscillospirales, Succiniclasticum, and Treponema (Figure 4B). Notably, several of these treatment-associated taxa were also independently recovered as the most important discriminatory features in the Random Forest analysis (Supplementary Figure S4). In the Control group, Prevotella/Prevotellaceae, Rikenellaceae, Bacteroidales, and Selenomonas were highlighted by both approaches, while in the Polyphenol group, Prevotella/Prevotellaceae, Rikenellaceae RC9, Treponema, Lachnospiraceae, Butyrivibrio, and Schwartzia were consistently identified. This strong overlap between methods demonstrates the robustness of the observed shifts in microbial composition over time. Overall, the number of altered ASVs increased across time points in both groups, particularly in the Control group, whereas Polyphenol supplementation appeared to selectively enrich specific microbial taxa (Figure 4).

The tendency for greater total SCFA, and specifically higher propionate, in Polyphenol-supplemented cows suggests a shift in hydrogen flow within the rumen fermentation network. Many plant polyphenols, including tannins and other phenolic compounds, can inhibit protozoa and certain hydrogen-producing bacteria, or selectively inhibit hydrogenotrophic methanogens (Liu et al., 2023; Zhao et al., 2025), thereby redirecting reducing equivalents (H₂) toward alternative electron sinks such as propionate production via the succinate or acrylate pathways (Romero et al., 2023). This mechanism has been proposed to explain increases in the molar proportion of propionate observed after polyphenol or tannin supplementation in ruminants (Teng et al., 2024; Tian et al., 2025).

An increase in propionate is physiologically meaningful in lactating dairy cows because it is the primary gluconeogenic precursor in ruminants, contributing roughly 60–74% of the carbon required for hepatic glucose synthesis (Zhang et al., 2015). Studies have shown that propionate can upregulate the expression of gluconeogenic genes such as PCK1, PCK2, and G6PC, indicating a direct role in regulating glucose synthesis (Wang et al., 2023). Moreover, approximately 90% of glucose in lactating dairy cows is generated via gluconeogenesis, with 50–60% of this glucose derived from ruminal propionate (Kong et al., 2025).

Importantly, the rise in plasma glucose occurred without alterations in plasma total protein or urea concentrations, indicating that nitrogen metabolism remained stable and that the improved glucose status was not accompanied by increased amino acid catabolism. This observation supports the interpretation that the modest increase in propionate supply was sufficient to enhance gluconeogenesis without compromising protein utilization. The observed increase in plasma glucose in polyphenol-supplemented cows is therefore consistent with a slightly greater ruminal propionate supply and subsequent hepatic gluconeogenesis, reinforcing the biological plausibility of the metabolic response even when SCFA changes are minor or trend-level.

It is also important to note that the effects of polyphenols on total SCFA and proportions are dose- and matrix-dependent. Some studies report reduced total SCFA or adverse effects on fiber digestion with high tannin levels, while others report a neutral or positive impact at lower, targeted doses, within a specific dose window (Orzuna-Orzuna et al., 2021; Khejornsart et al., 2024). Our results, showing a trend rather than substantial differences, are compatible with a modest dose that modulates fermentation without grossly impairing fiber digestion.

Several studies have reported reductions in enteric CH₄ following supplementation with polyphenols or tannins, although results remain inconsistent across experiments and animal types (Vasta et al., 2019; Beauchemin et al., 2020). In the present study, supplementation with a polyphenol-rich sugarcane extract did not result in a statistically significant reduction in enteric methane emissions, averaging 350 ± 27 g/cow·d and 17 ± 2 g/kg milk for polyphenol-supplemented cows versus 348 ± 27 g/cow·d and 21 ± 2 g/kg milk for controls. A post-hoc power analysis indicated that, given the observed variability and sample size, the study had limited power (23–39%) to detect a 15–20% reduction in methane emissions, highlighting that the null result should be interpreted cautiously.

Multiple, non–mutually exclusive explanations may account for this outcome. First, the efficacy of polyphenols as antimethanogenic agents depends on the dietary background and the availability of alternative hydrogen sinks (e.g., propionate formation or reductive acetogenesis), which differ between high-forage and high-concentrate diets (Bai et al., 2025; Williams et al., 2025). Second, mitigation responses are often dose-dependent and can require adaptation. For example, Prathap et al. (2024) observed variable reductions in CH₄ yield in lambs supplemented with sugarcane polyphenol extract depending on inclusion level, and Osei-Amponsah et al. (2025) reported decreases in daily CH₄ emissions in dairy calves after adaptation. Our sampling window (weeks 4–5 of supplementation) may therefore have missed a transient response or reflected a dose insufficient to reduce emissions beyond natural variability. Third, methodological and statistical considerations must be acknowledged. Methane measurements using SF₆ tracers are characterized by high between-animal variability, and sample sizes required to detect modest (<20%) reductions are often significant (Moate et al., 2021). Taken together, these considerations suggest that our null result is not inconsistent with the literature but rather underscores the conditional nature of polyphenol-driven methane mitigation in ruminants.

We observed no change in α-diversity (richness, Shannon) but clear treatment and time effects on community structure (β-diversity), as well as specific taxonomic shifts, an increase in Planctomycetota, a decrease in unclassified bacteria, and specific ASV clusters enriched in the polyphenol group. This pattern, characterized by stable within-sample diversity but altered community composition, is consistent with previous reports showing that additives can impose selective pressures without inducing broad reductions in microbial richness (Wang et al., 2012; Mao et al., 2013).

Polyphenols often function as selective antimicrobial agents, reducing specific taxa while allowing more tolerant or functionally redundant populations to expand, resulting in compositional rearrangements detectable through β-diversity metrics (Min et al., 2003; Vasta et al., 2019). The enrichment of Planctomycetota in polyphenol-supplemented cows is noteworthy, although any mechanistic interpretation remains speculative. Members of this phylum have been reported to encode a relatively high density of carbohydrate-active enzymes (CAZymes) compared with genome size (Gharechahi et al., 2022), suggesting that even low-abundance taxa may influence polysaccharide turnover and substrate availability. If present here, such activity could theoretically favor propionate-producing pathways and contribute to the modest increase in circulating glucose observed in the supplemented group. However, this hypothesis cannot be confirmed with 16S rRNA amplicon data alone. Future studies using shotgun metagenomics, metatranscriptomics, or targeted qPCR assays (e.g., mcrA for methanogens, or succinate/propionate pathway genes in Prevotella) will be necessary to evaluate whether Planctomycetota or associated taxa actively modulate carbohydrate metabolism and electron-flow dynamics in response to polyphenol supplementation.

Taken together, the data support a model in which polyphenol supplementation exerts targeted antimicrobial or modulatory effects that reconfigure the rumen microbiome (β-diversity), increasing the relative activity of propionate-producing pathways and modestly raising blood glucose levels via enhanced hepatic gluconeogenesis. The apparent lack of a methane decrease could be due to diet, dose, or power limitations, while microbiome shifts indicate that the supplement had biological activity at the community level. This pattern has been observed in other studies using sugarcane-derived polyphenol products or plant polyphenols more broadly, which often report microbial rearrangements and variable methane outcomes (Prathap et al., 2024; Bai et al., 2025).

The key limitations of our study include the sample size for methane detection and the fact that microbiome sequencing targeted 16S amplicons, resulting in taxonomic resolution limited by the method (functionality was inferred indirectly). Future work should include metagenomic or metatranscriptomic approaches to resolve functional shifts (e.g., genes/pathways for succinate/propionate conversion, methanogenesis genes), longer measurement windows for methane (and potentially complementary techniques), and dose–response trials to define the effective window for fermentation modulation vs. adverse effects on fiber digestion.

From a broader perspective, sugarcane-derived polyphenols represent a promising class of natural feed additives that can improve ruminal efficiency and animal metabolic responses while aligning with the global push for sustainable livestock systems. Their integration into precision feeding strategies could enhance nutrient use efficiency, reduce reliance on synthetic additives, and mitigate the environmental footprint of dairy production. Further work should clarify their long-term effects and optimal inclusion rates to enable effective implementation in commercial systems.