Figure S1. Schematic diagram and longitudinal cross-sectional view of the fermentation device (1) Outer shell of the fermentation unit; (2) Insulating layer; (3) Junction of system monitor; (4) Connecting pipe of water tank; (5) Discharge tube; (6) pH meter; (7) Temperature probe; (8) Water inlet line; (9) Heating tube bundle; (10) Blender; (11) Sampling observation port Supplementary Figure S2. Principal coordinate analysis (PCoA) of Bray–Curtis β-diversity of microbial communities in the rumen and feces (A) PCoA of rumen bacterial communities, (B) rumen archaeal communities, (C) fecal bacterial communities, and (D) rumen protozoal communities. Microbial β-diversity was calculated at the operational taxonomic unit (OTU) level based on Bray–Curtis distance metrics. Group differences were assessed using PERMANOVA. Each dot represents one sample, and shaded ellipses indicate 95% confidence intervals. The R² and P-values in each panel are results from PERMANOVA analysis. Supplementary Figure S3. Fecal metabolite profiling between CON and FLF groups (A) OPLS-DA (orthogonal partial least squares–discriminant analysis) score plot comparing CON (red triangles) and FLF (blue circles); each point represents one sample, and separation between groups indicates overall differences in fecal metabolite profiles. Model fit and predictability are summarized by R²X, R²Y, and Q² (higher Q² indicates better predictive ability). (B) Permutation test (200 permutations) for the OPLS-DA model; lower R²Y and Q² values in permuted models compared with the original model indicate low risk of overfitting. (C) Volcano plot of differential metabolites (x-axis: log₂ fold change [FLF/CON]; y-axis: −log₁₀ P). Blue and red dots indicate metabolites up- and downregulated in FLF, respectively; bubble size represents VIP (variable importance in projection) from the OPLS-DA model.