The experiment was carried out on the experimental farm of the Inner Mongolia Agricultural University, Hohhot, China. Dietary treatments consisted of 500 g/kg concentrate and 500 g/kg roughage, on a DM basis, composed of a variety of forage sources supplied in various quantities. The employed roughage sources were Chinese wildrye, alfalfa and CK. Whole-crop of CK at 60 days of growth were harvested, they were chopped and dried in the shade for 7 days, and dried CK were stored and protected from light until used. CK were used as sources of CT, and CT were measured using the butanol-HCL-method (Makkar, 2003). Table 1 provides information on the chemical composition and CT concentration in CK. A total of 15 crossbred sheep of the Dumont breed, with an average age of 1.5 years and an initial body weight of 40 ± 3 kg, each group consisted of five sheep. The sheep received one of the following feed regimens according to the treatment groups: (1) 400 g/kg concentrate +450 g/kg Chinese wildrye +150 g/kg alfalfa (CON); (2) 400 g/kg concentrate +150 g/kg Chinese wildrye +250 g/kg alfalfa +200 g/kg CK (2% CKT); (3) 400 g/kg concentrate +40 g/kg Chinese wildrye +160 g/kg alfalfa +400 g/kg CK (4% CKT); The three above dietary of sheep treatments had similar energy and CP levels (Table 2). The sheep were housed in the same barn and reared in individual tie stalls (1.5 × 2.5 m) with individual feed trough, the sheep were able to have direct visual and tactile contact and the animals were checked daily by trained personnel to verify the absence of symptoms of disease. Each sheep were given ad libitum to clean drinking water in a plastic bucket throughout the whole experiment. The experimental diet was given twice daily at 07:00 and 17:00. Table 2 displayed the components and nutritional content of the diet. The study spanned a total of 74 days, with a 14-day period dedicated to adaptation followed by 60 days of conducting treatments.

The feed samples were sent to the Inner Mongolia Key Laboratory of Animal Nutrition and Feed Science, College of Animal Science, Inner Mongolia Agricultural University, for DM and chemical analysis. To determine the DM content, the samples were oven-dried at 65°C until a constant weight was achieved. Upon drying, the samples were through a 1-mm screen. The ground samples were analyzed for ashes after 4 h of combustion in a muffle furnace at 550°C [Association of Official Analytical Chemists, 1990: method 942.05]. The DM, ash and ether extract (EE) of feed and feces samples were analyzed according to Association of Official Analytical Chemists, (1990) using the methods No. 934.01, 920.39, and 924.05, respectively. The OM of the samples was calculated by DM minus ash. The NDF and ADF were analyzed on an Ankom A200i Fiber Analyser (Ankom Technology Co., New York, NY, United States) using the methods of Damiano et al. (2023). CP using a Kjeldahl nitrogen analyzer [Association of Official Analytical Chemists, 1990: method 984.14]; Starch [Association of Official Analytical Chemists, 1990: method 996.11]; Calcium (Ca) [Association of Official Analytical Chemists, 1990: method 927.02]; Phosphorus (P) [Association of Official Analytical Chemists, 1990: method 965.17].

The trial to determine digestibility was carried out one week prior to the conclusion of the experiment. Feces from each sheep were collected daily using a plastic screen and weighted prior to being fed. Before determining chemical composition, the feed and feces samples were oven-dried at 65°C for 48 h. Then, determining DM, CP and ash according to Association of Official Analytical Chemists, (1990). The OM was determined by deducting the ash content from the DM of the samples. The determination of NDF and ADF was conducted using the approach outlined by Damiano et al. (2023). According to Chen et al. (2015), nutrient apparent digestibility = (total nutrients intake - nutrients in feces)/total nutrients intake.

Samples of rumen fluid were obtained at 6 h after morning feeding on the final day of the experiment through utilizing a rumen tube with vacuum pump. To prevent contamination from saliva, the initial 5 mL of rumen fluid were discarded. The pH value of rumen was immediately measured after collection. Subsequently, rumen fluid sample underwent filtration using 4 layers of cheesecloth to analysis volatile fatty acids (VFA), ammonia nitrogen (NH3-N), ruminal methanogens, and metabolites. The filtered sample was divided into 4 parts and frozen (−20°C) for further analysis.

Gas chromatography (GC-2014, Shimadzu International Trading CO. LTD) was used to detect the total concentrations of VFA, following the method outlined in Liu et al. (2019). 15-mL rumen fluid from each sheep was centrifuged (10,000 × g) for 10 min. Following this, 4 mL of the resulting liquid was moved to a centrifuge tube that already contained 1 mL of metaphosphoric acid with a concentration of 25%. After being held for a duration of 30 min and then centrifuged (10,000 × g) for 15 min. For gas chromatographic determination, a sample bottle was filled with a 1.5-mL supernatant using a pipette.

Each sheep’s rumen fluid sample, measuring 10 mL, underwent centrifugation at 10,000 × g for 10 min. Then, 0.5 mL of the resulting supernatant was combined with 4.5 mL of hydrochloric acid (0.2 mol/L) to determine NH₃-N concentrations, following the method outlined in Pellikaan et al. (2011).

The open-circuit respirometry system with 6 metabolism cages (Sable Systems International, NV, United States) was utilized to measure CH₄ production. A total of 15 sheep were split into three groups for determination. After a 48-h adaptation period, there was a subsequent 48-h duration dedicated to measuring methane emissions, The specifics regarding its explanation, functioning, and calibration are given in Salazar et al. (2018). The net CH₄ concentration was determined by subtracting the ambient-air CH₄ concentration from the CH₄ concentration of air released from each chamber. To calculate the overall daily CH₄ emission, the net CH₄ concentration for each chamber was multiplied by each chamber’s daily air volume.

The DNA Kit (Omega Bio-tek, Norcross, GA, United States) was used to extract the total DNA of sample. The NanoDrop2000 (Thermo Fisher, NY, United States) was used to detect the purity and concentration of DNA. A universal primer was used to amplify the 16S rRNA gene’s V3-V4 region 338F (5’-GGTGGTGTMGGATTCACACARTAYGCWACAGC-3′) and 806R (5’-TTCATTGCRTAGTTWGGRTAGTT-3′), with microbial DNA serving as the template. The data were processed and analyzed on the Majorbio cloud platform.¹

The correlation analysis was performed utilizing the tools provided by OmicShare. Spearman’s correlation coefficient was used to calculate the associations between distinct methanogens and metabolites, rumen fermentation parameters, and CH₄ production. Additionally, the correlation coefficient was employed to determine the significant relationships between specific metabolites and CH₄ production. The correlation coefficient (r) ranged from −1 to 1, with positive correlation represented by r > 0 and negative correlation represented by r < 0.

In order to determine the effect of treatments on the nutrient apparent digestibility, the alpha diversity indexes, rumen fermentation parameters, and CH₄ production, in the present study, each sheep was considered as an experiment unit. The normality of the distribution of variables was tested by the Shapiro - Wilk test and all variables were normally distributed. All results of 3 dietary treatment groups were subjected to ANOVA where diets were treated as the fixed effect, period, and animal were considered as the random effect using the GLM procedure of SAS (SAS Inst. Inc., Cary, NC, USA; version 9.0). When significant differences were observed, Tukey’s test was used to adjust for multiple comparisons. The PROC MIXED model was used including random and fixed effects as follows: Y_ij = μ + L_i + T_j + ε_ij, where Y_ij is the dependent variable, μ is the overall mean, L_i is the random effects of sheep (i = 5), T_j is the fixed effect of CKT supplements (j = 0, 2 and 4%), and ε_ij is the error term. Variability in the data was expressed as the standard error means and a probability level of p < 0.05 was considered statistically significant.

The rumen fermentation parameters were listed in Table 3. In comparison to the control group, the 2 and 4% CKT groups exhibited a significant decrease (p < 0.05) in NH₃-N and acetate concentrations. Rumen fluid pH, total VFA (TVFA), propionate, butyrate and valetate concentration and acetate: propionate (A/P) were not altered (p > 0.05) by addition of CKT.

As shown in Table 4, dietary treatments did not show any effect on DM, OM, CP, NDF and ADF intake of animal. Similarly, the addition of 2% CKT did not have impact (p > 0.05) on the apparent digestibility of NDF. However, the addition of 4% CKT significantly increased (p < 0.05) the apparent digestibility of NDF. The addition of CKT at 2 and 4% resulted in a significant reduction (p < 0.05) in the apparent digestibility of CP. Both 2 and 4% CKT had no impact (p > 0.05) on the apparent digestibility of DM, OM, ADF and the intake of DM, OM, CP, NDF, ADF.

Table 5 shows that the inclusion of 2 and 4% CKT resulted in a significant reduction (p < 0.05) in daily CH₄ emission (L CH₄/d). CH₄ production of DMI, OMI, NDFI and metabolic weight (BW^0.75) did not differ significantly (p > 0.05) among the various treatments.

A total of 193,249 effective 16S rRNA sequences were obtained from 15 rumen fluid samples and 97 OTU were obtained by performing OTU clustering on nonrepetitive sequences according to 97% similarity. The analysis of α-diversity indicated a decrease in the Chao index in the 4% CKT group (p < 0.05), and the ACE index for both the 2 and 4% CKT groups showed a significant decrease (p < 0.05) compared to the CON group. Meanwhile, a significant increase(p < 0.05) in the Simpson index was observed in the 4% CKT group. Which demonstrated that adding CKT decreased the diversity and richness of ruminal methanogen communities (Table 6).

The β-diversity analysis was performed to explore differences in the rumen methanogens community between the treatment groups (Figure 1). The PCoA and NMDS plots, using the Bray-Curtis distance matrix, revealed a clear separation between the rumen methanogens in the control group and the 4% CKT group. Different quadrants of the coordinate axis held these points, indicating the intake of 4% CKT had a noticeable impact on both the species and abundance of rumen methanogens.

At the phylum level, norank_d_Bacteria (34.62, 37.50 and 53.48%) and Euryarchaeota (52.12, 47.83 and 18.86%) were the dominant microbe in the control, 2% CKT and 4% CKT groups, respectively. The relative abundance of norank_d_Bacteria was increased in the 4% CKT compared with the CON and 2% CKT groups (p < 0.05). In contrast, Euryarchaeota was decreased in rumen with the 4% CKT addition (p < 0.05). At the genus level, norank_d_Bacteria (34.62, 37.50 and 53.48%), Methanobrevibacter (50.61, 45.76, and 16.60%), unclassified_d_Unclassified (13.26, 14.67, and 27.65%) and Methanosphaera (0.86, 1.38, and 1.63%) were the predominant genera in the control, 2% CKT and 4% CKT groups, respectively. The relative abundance of norank_d_Bacteria and unclassified_d_Unclassified were increased in the 4% CKT compared with the CON and 2% CKT groups (p < 0.05). In contrast, Methanobrevibacter was decreased in rumen with the 4% CKT addition (p < 0.05) (Table 7).

Untargeted metabolomics techniques were used to analyze the rumen metabolites, resulting in the identification of a grand total of 261 metabolites. The PCA analysis, which is a type of unsupervised multivariate statistical analysis, indicated that there was a clear distinction in the ruminal metabolites between the sheep in the CON and 4% CKT groups (Figure 2A). The variations in metabolites were additionally confirmed through the score plots of OPLS-DA (Figure 2B). The stability and reliability of the OPLS-DA model are confirmed by the cumulative values of R²Y (0.786) and Q² (0.530) in the OPLS-DA plot.

In Table 8, it was found that there were 25 distinct metabolites in the sheep’s rumen that differed between the control group and the 4% CKT group, with VIP scores exceeding 1.5 and p-values less than 0.05. Two clusters were formed to distinguish the differential metabolites between the two groups. The relative expression of dibenzofuran, heptanoic acid, 4-hydroxypyridine, N-methy-L-glutamic acid, N-carbobenzoxy-L-leucine, 1-octanol, N-alpha-acetyl-L-lysine, 2,3-butanediol and apigenin were elevated in the rumen of sheep fed 4% CKT compared with the CON group. However, The relative expression of tyramine, cyclohexanecarboxylic acid, 5-hydroxyindole-2-carboxylic acid, 2,6-diaminopimelic acid, asarylaldehyde, 4-O-methylphloracetophenone and 3,7-dihydroxyflavone etc. were higher in the CON group than that of 4% CKT group.

To assess if the identified differential metabolites were crucial for distinguishing between different groups, a ROC curve was generated. As shown in Figure 3, the AUC of tyramine, N-methylglutamic acid, heptanoic acid and 1-octanol were all above 0.85, demonstrating that these specific ruminal metabolites effectively indicate the impact of a 4% CKT diet on rumen metabolism.

The KEGG pathway enrichment analysis revealed that the addition of 4% CKT affected primarily CH₄ metabolism (p < 0.0001), glycine, serine and threonine metabolism (p = 0.001), mminoacyl-tRNA biosynthesis (p = 0.057), glycerolipid metabolism (p = 0.036), glyoxylate and dicarboxylate metabolism (p = 0.001), pentose phosphate pathway (p = 0.039), cysteine and methionine metabolism (p = 0.060), isoquinoline alkaloid biosynthesis (p = 0.083), carbon fixation in photosynthetic organisms (p = 0.025) and sphingolipid metabolism (p = 0.024) in the rumen of sheep (Figure 4).

Correlation analysis between rumen methanogens and CH₄ production showed that the emission of CH₄/kg of DMI was positively associated with unclassified_f_Methanobacteriaceae (r = 0.550, p = 0.034) and unclassified_p_Euryarchaeota (r = 0.584, p = 0.022). CH₄/day was negatively associated with unclassified_f_Methanomassiliicoccaceae (r = −0.386, p = 0.046). CH₄/kg of OMI was positively associated with unclassified_p_Euryarchaeota (r = 0.515, p < 0.05) (Figure 5A).

The correlation between rumen methanogens and differential metabolites associated with methane metabolism is shown in Figure 5B. Methanobrevibacter was negatively associated with fructose- 1,6 -diphosphate (r = −0.4227, p = 0.037). Methanosphaera was positively associated with fructose- 1,6 -diphosphate (r = 0.3675, p = 0.018). Unclassified_c_Thermolplasmata was positively associated with N-methylglutamic acid (r = 0.5095, p = 0.029).

Regarding rumen pH, no differences were found among the sheep, which aligned with the findings of Phesatcha and Wanapat (2017) who reported that average pH values of 6.5 ± 0.2 at CT concentration of <2%. Maintaining a stable pH in the rumen is essential for the proliferation of microorganisms and the fermentation of feed (María et al., 2020). Our results further indicated that CKT did not destroy rumen homeostasis.

To be certain to put in evidence any effect on CH₄ production, various research studies have demonstrated alterations in VFA production and nutrient digestibility when employing quebracho tannins, fumarate, and propionate precursors, further affecting CH₄ synthesis and metabolism. For instance, there was a decrease in acetate concentration and the A/P ratio when 10 and 20 g/kg DM of quebracho tannin extracts were added to sheep dietary (Beauchemin et al., 2007). This subsequently lowers the quantity of hydrogen available for methanogenesis, as mentioned in a research by Patra and Jyotisna (2009). Study carried out utilizing propionate precursors, like acrylate (Newbold et al., 2005), demonstrated a decrease in CH₄ yield while observing an increase in acetate and TVFA production. Amlan et al. (2017) found that the decrease in CH₄ emissions from the digestive system was affected by the digestibility of CP. Similar responses were discovered in this research, where the decrease in intestinal propionate concentration and CH₄ production was impacted by the digestibility of NDF. Molina-Botero et al. (2019) observed a reduction in the apparent digestibility of NDF by including E. cyclocarpum and G. sepium (contain CT), this could be explained by the multiple adverse effects of CT’s on animal fermentative processes. CT negatively affected the adhesion of cellulolytic bacteria to the substrate, reduced the activity of fibrolytic enzymes, and inhibited the growth of the cellulolytic population or form complexes with cellulose. However, in the present study, the digestibility of NDF was increased in 4% CKT group, this could be attributed to the different activity and molecular structure of tannins, which needs to be further studied.

NH₃-N serves as a crucial nitrogen provider for the process of rumen fermentation and the development of microorganisms. Increased availability of NH₃-N in the rumen results in higher nitrogen loss and stunted growth (Kabasa et al., 2004). In this research, the inclusion of 2 and 4% CKT in the diet resulted in the concentration of NH₃-N decreased significantly, suggesting a reduction in ruminal CP degradation. Other research studies have also yielded comparable findings when assessing quebracho CT at a daily dosage of 18 g/kg in cows (Aguerre et al., 2016). The decrease in NH₃-N concentration corresponded to the decrease in CP digestibility in the rumen of sheep that were fed 2 and 4% CKT. The decreased digestibility of CP observed in the 2% CKT and 4% CKT group is associated with the capacity of tannin to attach to protein. Furthermore, it is plausible that a portion of this compound (tannin-proteins) did not separate in the abomasum (Waghorn, 2007). Furthermore, when tannin is released into the duodenum, it has the potential to deactivate enzymes in the intestines or attach itself to proteins (Warren et al., 1998), ultimately leading to a decrease in the digestibility of proteins. Similar results were observed by Tiemann et al. (2008) with black wattle CT, there was a gradual decrease in CP digestibility as the levels of up to 20 g/kg of DM were included in the diets of Jersey steers. Al-Dobaib (2009) reported a negative impact of 15 g kg⁻¹ CT from quebracho on protein degradation in an in vitro experiment with rumen fluid from sheep. Other studies reported that condensed or hydrolysable tannin concentrations below 15 g/kg had no impact on in vivo ruminal protein degradation (Benchaar et al., 2008; Krueger et al., 2010). The inclusion of 2 and 4% CKT resulted in a significant reduction in daily CH₄ emission can be attributed to the reduced digestibility of CP in the rumen, as observed in current research.

Rumen microorganisms degrade carbohydrates to form VFAs, CO₂, H₂, CH₄, adenosine triphosphate (ATP), and a few other compounds. VFAs account for 70–80% of the energy sources of ruminants (Jayanegara et al., 2011). Vasta et al. (2019) provided a summary of the existing literature on the alterations in TVFA levels and the acids acetic, propionic, butyric, and valeric following tannin usage. Five parameters showed varying levels, ranging from increased to unchanged to decreased. The precise impact of tannin on the alteration of the molar ratios of VFAs in the rumen remains uncertain, and several authors have proposed that differences in tannin structure and activity due to different plant sources and extraction conditions may explain this inconsistency (Patra et al., 2017). Hence, it is crucial to examine the impacts of tannin inclusion on a per-instance basis. The absence of disparities in TVFA, propionic, butyric, and valeric acids in the present research might be attributed to the restriction of animal numbers per treatment, which minimized the variability linked to the impacts of diet or the insignificant impact of dietary additives on VFA absorption through the rumen wall (Bodas et al., 2012). These results align with previous studies (Aguerre et al., 2016; Piñeiro-Vázquez et al., 2018). The decrease in acetate levels could be associated with the impact of CT on the breakdown of fiber in the rumen. According to Castro-Montoya et al. (2011), cellulolytic bacteria exhibit greater susceptibility to elevated levels of CT in comparison to other microorganisms. These substances not only create complexes with plant cell wall carbohydrates but also directly hinder the activity of CT in these microorganisms. The CT caused a shift in the usage of carbon chain for the creation of VFAs in the rumen, resulting in a reduction in the concentration of acetic acid. This alteration in rumen fermentation could potentially lead to a decreased amount of free metabolic hydrogen, ultimately resulting in limited availability for CH₄ production (Hassanat and Benchaar, 2013). Which was another reason for CH₄ emission reduction effect of CKT.

In present work, we confirm the strong effect of depression of CKT on methanogensin population. In addition, as above mentioned, the mechanisms in which CKT reduce CH₄ emission have two mechanisms, i.e., (1) through reduction in CP digestibility and increase in NDF digestibility, and (2) through decrease concentration of acetic acid, this change in the rumen fermentation may have led to a decrease in the amount of free metabolic hydrogen available for CH₄ production. Nevertheless, Cammack et al. (2018) reported that microorganisms had a significant impact on the effective functioning of the rumen. Protozoa utilize starch, cellulose, hemicellulose, pectin, and soluble sugars to generate VFA and H₂. The methanogens that are attached to the surface of protozoa employ available H₂ to produce CH₄, thereby reduce the negative pressure in the rumen (Caridad et al., 2012). In the absence of methanogens, the levels of methane can be reduced by approximately 11% (Newbold et al., 2015). Thus, high-throughput sequencing technology was used to detect the population diversity of methanogensin the current study, results showed that the relative abundance of Euryarchaeota, which contained most of the methanogens, was lower in the 4% CKT group than that of CON and 2% CKT groups. In addition, the relative abundance of Methanobrevibacter, as the prominent methanogen, also was lower in the 4% CKT group than that of the CON and 2% CKT groups. In this sense, several studies provide support for our findings through various studies conducted under in vitro conditions using cattle rumen liquor, for example, Tan et al. (2011) observed that the total methanogens, as well as methanogens belonging to the Methanobacteriales, showed a significant decline when CT inclusions exceeded 1 mg/50 mg DM. Meanwhile, Animut et al. (2007a) also noted that goats fed diets containing CT had decreased populations of methanogens in their rumen. Liu et al. (2011) observed the phenomenon in vivo, the addition of 30 g of CT/kg of diet from chestnut decreased the populations of methanogens in sheep. One potential explanation for the antimethanogenic impact is the suppressive influence of tannins on the functions of enzymes essential for the development of methanogens (Patra and Saxena, 2011). In addition, several studies (Animut et al., 2007b; Bhatta et al., 2009) found that CT had antiprotozoal properties. As we know, methanogens attached to the surface of ciliate protozoa and symbiosed with them (Finlay et al., 1994). Hence, a decline in protozoa populations could potentially lead to a decrease in methanogen populations. However, the underlying mechanism behind this process needs further investigation.

The use of metabolomics analysis offers a chance to comprehensively assess a wide range of small molecule metabolites in cells, tissues, and biofluids. While the complete description of the majority of differential rumen metabolites identified in the research is still lacking, the metabolomics findings also indicate that CKT supplementation has a significant impact on methane metabolism in sheep. Tyramine is a monoamine compound derived from the tyrosine, which can be converted to pyruvate under the action of tyrosine aminotransferase (Zhang et al., 2019). The H₂ produced by the catabolism of pyruvate is the raw material for methane synthesis (Min et al., 2020), thus, the decrease in tyramine levels was consistent with the decrease in the daily CH₄ emission. Therefore, the decreased levels of tyramine might imply the potential of CKT to downregulate methane synthesis in sheep. Studies have shown that animals can produce formaldehyde through physiological and biochemical reactions and metabolic reactions, which is an important intermediate metabolite (Ma et al., 2020). Through substitution reaction, methane can be converted into methanol, and formaldehyde is the oxidized substance of methanol. N-methy-L-glutamic acid is analog of L-glutamic acid, which can produce formaldehyde through a series of metabolic pathways (Guido et al., 2018). In the present research, the rice in N-methy-L-glutamic acid levels implies that the transition from methane to formaldehyde may be inhibited. The negative correlation between N-methy-L-glutamic acid and Methanobrevibacter could be accounted for. The above-mentioned compounds are significant candidates for future investigations due to their reaction to CKT supplementation. Further investigation is warranted to explore the metabolic processes and action mechanism of CKT in the sheep rumen.

However, some limitations need to be acknowledged. First of all, it is obvious that supplementing tannin with CK instead of green hay is not appropriate, since tannin is only one of the main active ingredients in CK, further extraction and purification of this compound is needed to carry out further research. In addition, this preliminary model was developed on a single experimental group at one farm (crossbred sheep of the Dumont breed). Therefore, its generalizability remains to be assessed in multi-farm studies and meta-analyses and is yet to be externally validated. Recognizing these limitations is crucial to properly contextualize the study and the research questions that need to be investigated in order to gain a more comprehensive understanding of the impact of CKT on CH₄ emission of sheep.