All animal care and experimental protocols were conducted in strict compliance with the Guidelines for the Care and Use of Laboratory Animals issued by the Ministry of Science and Technology of China and ethically approved by the Institutional Animal Care and Use Committee (IACUC) of the Xinjiang Uygur Autonomous Region Academy of Animal Science (Approval No. 320240107; Urumgi, China).

The Rosa rugosa Thunb. pomace (residue from essential oil extraction) was provided by Xinjiang Yutian Guimi Biotechnology Co., Ltd. (Yutian County, Hotan Prefecture, China; GPS coordinates: 36°46′N, 81°50′E). The proximate composition (dry matter basis) of the RP is detailed in Table 1. The active components in rose pomace were analyzed using ultra-performance liquid chromatography coupled with tandem mass spectrometry (UPLC-ESI-MS/MS). After sample pretreatment through solvent extraction and filtration, chromatographic separation was achieved with a reverse-phase C18 column employing gradient elution of acidified aqueous and organic phases. Mass spectrometric detection was performed in both positive and negative ionization modes with multiple reaction monitoring, simultaneous qualification and quantification of diverse bioactive compounds including flavonoids, phenolic acids and terpenoids. As illustrated in Figure 1, flavonoids constituted the predominant bioactive components (relative abundance: 47.4%), followed by terpenoids (9.7%), based on quantification of normalized peak areas.

The trial was conducted from January to March 2024 at the Hu Sheep Feedlot operated by Xinjiang Taihe Agricultural and Animal Husbandry Technology Co., Ltd. (Kashgar, China; 39°46′N, 78°36′E), a facility certified under the GB/T 20014–2018 good agricultural practice standards.

Forty healthy male purebred Hu sheep (Ovis aries), 3–4 months of age, with comparable initial body weight (27.42 ± 1.80 kg; mean ± SD), were selected and randomly allocated into four groups (n = 10/group) after a 14-day quarantine period with ivermectin anthelminthic treatment (0.2 mg/kg BW). The experimental sheep were divided into four enclosures, with 10 sheep in each enclosure, and they were raised separately in their corresponding enclosures. The dietary treatments included:

The corresponding rose pomace in each group was crushed and mixed with other feed ingredients, then extruded into cylindrical pellets with a standardized 8mm diameter using a granulation device (model SZLH508; Jiangsu Liangyou Zhengda Co., Ltd., Suzhou, China) to facilitate lamb consumption.

The 67-day trial comprised a 7-day adaptation period, a gradual transition from the maintenance diet to the experimental diets, and a 60-day feeding trial, with ad libitum feeding of diets formulated according to NRC (2007) nutrient requirements for growing lambs. The composition of the basal diet is given in Table 2 and the analyzed nutritional profiles of the experimental diets are presented in Table 3. During the trial period, fresh water and mineral blocks were provided at all times. Mineral blocks were also replenished monthly for every lamb group.

Prior to trial commencement, all pens were thoroughly cleaned and disinfected using 2% sodium hypochlorite solution. Throughout the experimental period, lambs were group-housed in identical environmental conditions with ad libitum access to feed (offered at 09:00 and 18:00 daily) and fresh water. To ensure consistent nutrient intake, daily feed residues were maintained at 15% of the offered quantity through gravimetric adjustment.

Body weights were recorded before morning feeding on day 1 (initial) and day 61 (final) of the formal trial period. Average daily gain (ADG) was calculated as:

Following the Guidelines for the Ethical Treatment of Experimental Animals, lambs were humanely slaughtered through a standardized procedure. After a 12-h fasting period with free access to water, pre-slaughter live weight was measured. Exsanguination was performed via simultaneous severance of the carotid artery and jugular vein, after which the head, hooves, skin, and internal organs (excluding perirenal fat) were removed. The carcass weight was immediately recorded, and the dressing percentage was calculated as:

Rumen content samples were collected via oral intubation by inserting a rumen fluid collection tube through the mouth into the rumen, followed by syringe aspiration and immediate filtration through four layers of sterile gauze to separate particulate matter. To minimize cross-contamination from oral microbiota, the initial 50 mL of filtrate was discarded. Subsequently, 10 mL of clarified rumen fluid was aliquoted into sterile cryovials and snap-frozen in liquid nitrogen for preservation. Prior to analysis, thawed samples were centrifuged at 15,000 × g for 10 min at 4 °C. The resultant supernatants were analyzed for ammonia nitrogen (NH₃-N) using the phenol-hypochlorite method (Broderick and Kang, 1980) and the absorbance at 630 nm was measured with a UV-1800 spectrophotometer (Shimadzu Corp., Kyoto, Japan). Volatile fatty acid (VFA) concentrations were determined by gas chromatography (Clarke, 1993) (GC-450, Varian Inc., United States), with detection limits of 0.01 mm for acetate, propionate, butyrate, isobutyrate, valerate, and isovalerate.

The determination of amino acids refers to Li J. et al. (2025). Freeze-dried tissue samples (100 mg) were homogenized in 2.0 mL microcentrifuge tubes with 1.2 mL of 10% (w/v) sulfosalicylic acid solution. Following vortex mixing for 3 min, the homogenate was centrifuged at 13,500 × g for 15 min at 4°C. The supernatant was filtered through 0.22 μm polyethersulfone (PES) membranes and transferred to 2.0 mL glass threaded vials with PTFE/silicone septa. Fatty acid concentrations were quantified using UPLC-ESI-MS/MS (Waters Acquity I-Class PLUS; Applied Biosystems QTRAP 6500+) with an HSS-T3 column (1.8 μm, 2.1 × 100 mm). The mobile phase consisted of (A) 0.1% formic acid/5 mM ammonium acetate in water and (B) 0.1% formic acid in acetonitrile, employing a 14-min gradient (2–98% B; 0.35 mL/min, 50°C). ESI parameters were maintained at 550°C and ± 5,500/4,500 V with curtain gas (35 psi). Analytes were detected via optimized MRM transitions, and concentrations were determined by normalizing peak areas to total area using external calibration curves.

The determination of amino acids refers to Li J. et al. (2025). Total fatty acid extraction from frozen meat samples was performed using the direct methylation protocol described by O’Fallon et al. (2007), involving sequential saponification (2 N KOH/methanol, 55°C, 1.5 h) and methylation (14% boron trifluoride/methanol, 100°C, 30 min). Fatty acid methyl esters (FAMEs) were analyzed using a gas chromatograph (GC-450, Varian Inc., United States). Peak identification was based on retention time alignment with certified FAME standards (C4-C24 methyl esters, CRM47885, Sigma-Aldrich, US). Individual fatty acid concentrations were quantified using external calibration curves generated from five-point serial dilutions of the standard mixture.

Bacterial DNA was extracted from rumen fluid using the TGuide S96 Magnetic Soil/Stool DNA kit (Tiangen Biotech, Beijing, China) according to the manufacturer’s instructions. Full-length 16S rRNA gene amplification was performed with the primers 27F (5′-AGRGTTTGATYNTGGCTCAG-3′) and 149 2R (5′-TASGGHTACCTTGTTASGACTT-3′). PCR conditions included an initial denaturation at 95°C for 3 min, followed by 30 cycles of denaturation (95°C, 30 s), annealing (55°C, 30 s), and extension (72°C, 90 s), with a final extension at 72°C for 7 min.

For DNA library preparation, total DNA was amplified using barcoded 16S-specific primers (27F/1492R). PCR products were purified, quantified, normalized, and ligated into SMRT Bell libraries. After passing quality control, libraries were sequenced on the PacBio Sequel II platform. Raw sequencing data (.bam format) were processed through the smrtlink pipeline to generate circular consensus sequencing (CCS) reads. CCS sequences were demultiplexed using lima v1.7.0 to obtain raw-CCS data, followed by primer trimming and length filtering with cutadapt v1.9.1 to generate clean-CCS sequences. Chimeric sequences were removed using UCHIME v4.2, yielding effective-CCS sequences. All downstream analyses were conducted on the Biomarker Biotechnology Cloud Platform.¹

Experimental data were initially processed using Microsoft Excel 2013. Statistical analyses were performed with SPSS 26.0 software. One-way ANOVA followed by Duncan’s multiple range test was applied for comparisons among groups. Results are expressed as mean ± SEM. Statistical significance was defined as P ≤ 0.05, while P ≤ 0.01 was considered highly significant. A P-value between 0.05 and 0.10 was interpreted as indicative of a trend. Correlation analyses were visualized using Origin 2021 software.

As shown in Tables 4, 5, the dry matter intake (DMI) of the RP8 group was significantly higher than that of the RP2 group (P ≤ 0.05). No statistically significant differences (P > 0.05) were observed among groups for other growth performance or slaughter parameters.

As presented in Table 6, increasing dietary rose pomace inclusion elicited linear responses in ruminal concentrations of acetate, propionate, isobutyrate, butyrate, isovalerate, valerate, and TVFA (P = 0.024, P = 0.004, P = 0.006, P = 0.05, P = 0.000, P = 0.003, P = 0.009), while NH₃-N exhibited a quadratic change (P = 0.042). Notably, the RP4 group demonstrated significantly higher acetate levels compared to RP0 (P ≤ 0.05), whereas propionate and TVFA concentrations in RP0 were significantly lower than those in RP2, RP4, and RP8 (P ≤ 0.05). The RP8 group showed marked increases in isobutyrate (P ≤ 0.05) and isovalerate (P ≤ 0.01) relative to RP0, together with elevated valerate levels in both RP4 and RP8 groups (P ≤ 0.05).

As shown in Figure 2, sequences were clustered at a 97% similarity threshold to define operational taxonomic units (OTUs). A total of 2,115 OTUs were identified across all four groups. The OTU counts for RP0, RP2, RP4, and RP8 groups were 1702, 1649, 1684, and 1687, respectively. Among these, 1183 OTUs (55.93% of total) were shared among all groups. Unique OTUs per group numbered 44 (2.08%), 9 (0.43%), 70 (3.31%), and 55 (2.60%) for RP0, RP2, RP4, and RP8, respectively, indicating relatively minor compositional differences in OTU distribution across dietary treatments.

Alpha-diversity analysis revealed no significant differences (P > 0.05) among groups for the ACE, Chao1, Simpson, and Shannon indices, or coverage (Table 7).

Principal coordinates analysis (PCoA) based on Bray-Curtis dissimilarity at the OTU level demonstrated minimal separation among groups, with principal coordinate 1 (PC1) and PC2 explaining 17.55 and 9.36% of total variance, respectively (Figure 3). The close clustering of sample points across dietary treatments indicates that rose pomace supplementation exerted limited structural impacts on the rumen microbial ecosystem in Hu sheep.

As detailed in Table 8, linear responses were observed in the relative abundances of Firmicutes, Bacteroidota, and Patescibacteria with increasing RP inclusion (P = 0.016, P = 0.007, P = 0.014). Notably, the RP8 group exhibited an extremely significant elevation in Bacteroidota abundance compared to RP0 and RP2 (P ≤ 0.01), while showing reduced Patescibacteria levels relative to these groups (P ≤ 0.05). Marginal linear decreases (0.05 ≤ P ≤ 0.10) were detected for Planctomycetota (P = 0.076) and Desulfobacterota (P = 0.086). These shifts suggest selective modulation of rumen microbial ecology by dietary polyphenol-fiber interactions.

As delineated in Table 9, increasing dietary RP inclusion elicited a linear response (P = 0.030) in the relative abundance of the Christensenellaceae_R-7_group, while Prevotella exhibited a marginal linear trend (P = 0.076). Notably, Rikenellaceae_RC9_gut_group demonstrated a quadratic response (P = 0.001), characterized by an initial decrease followed by rebound, with the RP8 group showing extremely significant elevation compared to RP2 and RP4 (P ≤ 0.01), and significant increase over RP0 (P ≤ 0.05).

As illustrated in Figure 4, linear discriminant analysis (LDA) with a threshold score ≥ 3.5 revealed differential microbial signatures across groups at the phylum-to-genus level. The control group (RP0) exhibited three taxa significantly influencing community structure, while RP2 showed no discriminative taxa. The RP4 and RP8 groups demonstrated one and five discriminative taxa, respectively, with the latter displaying the highest number of lineage-specific biomarkers, indicative of dose-dependent modulation of rumen microbiota by RP supplementation.

As shown in Table 10, no significant differences (P > 0.05) were detected in the amino acid composition of longissimus dorsi muscle among the groups. Notably, histidine content exhibited a quadratic trend (P = 0.091) with increasing RP inclusion, suggesting a potential threshold effect of polyphenol-fiber interactions on this essential amino acid.

As shown in Table 11, dietary supplementation with RP did not significantly affect saturated fatty acid (SFA) content in the longissimus dorsi muscle of Hu sheep (P > 0.05). However, notable modulations were observed in specific unsaturated fatty acids (UFAs). Increasing RP inclusion elicited quadratic tendencies in C16:1, C18:1, N-3 PUFAs, C18:2N6, N-6 PUFAs, and PUFAs (P = 0.065, P = 0.087, P = 0.51, P = 0.067, P = 0.081, P = 0.079). Linear tendencies were detected for C22:5N3 and C22:6N3 (P = 0.053 and P = 0.057), while significant linear responses occurred in C24:1, N-3 PUFAs, and C20:2N6 (P = 0.025, P = 0.028, P = 0.017). Notably, the RP4 group exhibited significantly higher N-3 PUFA levels compared to RP0 (P ≤ 0.05), extremely elevated C20:2N6 levels versus the control (P ≤ 0.01), and superior C18:3N3 concentrations relative to RP0, RP2, and RP8 (P ≤ 0.05).

The correlation analysis heatmap constructed using Pearson’s algorithm (Figure 5) revealed complex interrelationships among fatty acids in the longissimus dorsi muscle, rumen microbiota, and rumen fermentation parameters. SFAs and MUFAs exhibited significant negative correlations (P ≤ 0.05) with the A/P, whereas N-3 PUFAs, N-6 PUFAs, and PUFAs showed significant positive correlations (P ≤ 0.05) with increased propionate concentration and relative abundance of Prevotella. Notably, N-6 PUFAs and PUFAs were strongly negatively correlated (P ≤ 0.01) with the Rikenellaceae_RC9_gut_group, while the N-6/N-3 ratio displayed a significant negative association (P ≤ 0.05) with the Rikenellaceae_RC9_gut_group. Acetate positively correlated with Prevotella (P ≤ 0.05), whereas propionate exhibited divergent associations—negatively with Firmicutes (P ≤ 0.05) and positively with Bacteroidota (P ≤ 0.05) and Prevotella (P ≤ 0.01). Valerate demonstrated a highly significant positive correlation with Bacteroidota (P ≤ 0.01) and Prevotella (P ≤ 0.05), but a negative correlation with the Rikenellaceae_RC9_gut_group. TVFAs were positively linked to both Bacteroidota and Prevotella (P ≤ 0.05), and the A/P ratio was significantly associated with increased Patescibacteria abundance. These findings collectively suggest that RP-induced shifts in rumen microbiota (e.g., Prevotella-Bacteroidota synergism and Rikenellaceae_RC9_gut_group suppression) mediate lipid metabolism remodeling in muscle tissue.