The experiment was conducted at the Larima National Cattle Farm, located in Xinlong County, Ganzi Prefecture, Sichuan Province, China (30.23–31.32°N, 99.37–100.54°E) at an altitude of 3,600 meters. The study period extended from February to April 2024, during which the average daily temperature was approximately −10°C. All animal use protocols and experimental procedures were approved by the Animal Care and Use Committee of the Laboratory Animal Center of Sichuan Agricultural University (Approval No. SCAUAC202402-4).

Twenty-seven healthy male Maiwa yaks (48 months of age), with an average body weight of 249.10 ± 31.93 kg, were selected for this study. One week prior to the experiment, cattle pens were treated with insecticidal powder for deworming purposes. The yaks were randomly divided into three treatment groups using a randomized block design, with each treatment consisting of three replicates, and three yaks assigned to each replicate. Each pen had an area of approximately 20 m² (5 m in length and 3.8 m in width). The yaks were fed pelleted total mixed ratios formulated with three different NEg levels: A low energy group (LE: 4.06 MJ/kg), a medium energy group (ME: 4.46 MJ/kg), and a high energy group (HE: 4.87 MJ/kg), while maintaining a constant crude protein (CP) content of 13.0% across all diets. The diets were provided in two equal portions daily, at 09:00 and 17:00. The three diets, ranging from the lowest to the highest energy levels, provided 1.0, 1.1, and 1.2 times the NEg requirements for growing cattle as per the Agricultural Industry Standard of the PRC (2004). The NEg level in the LE group was set based on a reference body weight of 200 kg and an ADG of 0.8 kg. The composition and nutritional content of the experimental diets are presented in Table 1. The experiment employed a one-way ANOVA (Analysis of Variance) experimental design, and included a 7-day adaptation period followed by a 60-day measurement period (days 0–60).

All yaks were weighed using a weighbridge (Shanghai Yaohua Weighing System Co. in Shanghai, China) prior to feeding on days 0 and 60 of the experiment period. Feed samples (200 g) were collected on days 1, 30, and 60, and stored at –20° for subsequent analysis. Throughout the experimental period (days 1–60), the daily feed intake and feed refusals for each replicate within each treatment group were recorded.

From days 55 to 60, fecal samples (200 g per day) were collected from each group using plastic pans placed beneath the animals. The total fecal output was weighed, and a subsample (100 g) was placed in self-sealing bags and stored at –20°C for later analysis of nutrient digestibility. Additionally, fecal samples (5 g) were collected directly from the rectum of each yak and stored at –80°C in sterile centrifuge tubes for subsequent microbiota analysis.

On day 60, approximately 150 mL of rumen fluid was collected from each yak using an oral stomach tube (Kelibo Equipment Co., Ltd., Wuhan, China) at three time points: 0, 2, and 6 h post-morning feeding. To minimize saliva contamination, the initial 50 mL of rumen fluid was discarded. The pH of the rumen fluid was measured immediately using a pH meter (pH 828-10, Zhongxing Weiye Instrument Co., Ltd., China). Subsequently, the rumen fluid was filtered through four layers of cheesecloth and divided into four aliquots in 50 mL centrifuge tubes, which were subsequently stored at –80°C for further analysis. For the analysis of volatile fatty acids (VFAs), 10 mL of rumen fluid was mixed with 25% metaphosphate acid in a 4:1 volume ratio. For ammonia nitrogen (NH₃-N) determination, 5 mL of rumen fluid was mixed with methanol at a 1:1 volume ratio. The remaining rumen fluid was stored in two 10 mL centrifuge tubes at –80°C for the analysis of Microbial Protein (MCP) and rumen microbiota.

On day 60, prior to the morning feeding, blood samples (20 mL) were collected from the jugular vein of each yak using evacuated tubes without anticoagulant (Honghu Taining Medical Equipment Co., Ltd., Honghu, China). The samples were kept on ice for 1 h and then centrifuged at 3,000 g for 15 min to separate the serum. The serum samples were stored at –80°C for analyses of serum biochemical parameters, energy metabolism-related hormones, and untargeted metabolomics.

The ADG in the ME and HE groups was significantly higher than that observed in the LE group, while the FCR in the LE group was significantly higher than in the ME and HE groups (P < 0.05). However, no significant differences in ADG or FCR were detected between the ME and HE groups (Table 2). With increasing dietary energy levels, the apparent digestibility of DM, OM, CP, NDF, EE, Ca, and P significantly increased (P < 0.05). The apparent digestibility of ADF and CF in the ME and HE groups was also significantly higher than that in the LE group (P < 0.05) (Table 3).

The concentration of MCP increased with rising dietary energy levels, while the concentration of NH₃-N decreased. Specifically, the MCP concentration in the HE group was significantly higher, and the NH₃-N concentrations was significantly lower than those in the LE group (P < 0.05) (Table 4). The concentrations of GLU, TG, and ALP in the ME and HE groups were significantly higher than those in the LE group (P < 0.05). Conversely, the NEFA concentration in the LE group was significantly higher than in the ME and HE groups (P < 0.05) (Table 5). Among serum hormones, the GH concentration increased with higher dietary energy levels. Additionally, the concentration of INS, IGF-1, LEP, and T4 in the ME and HE groups were significantly higher than those in the LE group (P < 0.05) (Table 6). No significant differences were observed among the groups for the remaining measured indicators.

Following 16S rDNA sequencing of rumen and fecal samples, a total of 9,503 and 5,647 valid ASVs were obtained for the rumen and feces, respectively. Specifically, the number of ASVs in the rumen samples form the LE, ME, and HE groups was 4,853, 4,741, and 5,155, respectively (Figure 1A), while the number of ASVs in the fecal samples from the LE, ME, and HE groups was 3,753, 3,039, and 2,611, respectively (Figure 2A). The rarefaction curves of both rumen and fecal samples plateaued with increasing sequencing depth, indicating that the sequencing depth was sufficient for capturing the majority of microbial diversity (Figures 1B, 2B). In the rumen samples, both the Simpson and Shannon diversity indices increased with rising dietary energy levels, with values in the LE group significantly lower than those in the HE group (P < 0.05) (Figures 1C,D; Table 7). Conversely, in the fecal samples, the Shannon, Chao1, and observed features indices decreased as the dietary energy level increased, with these indices significantly lower in the HE group compared to the LE group (P < 0.05) (Figures 2C–E; Table 8). PCoA demonstrated clear separations among the LE, ME, and HE groups in both rumen and fecal microbiota compositions (Figures 1E, 2F). At the phylum level, Firmicutes and Bacteroidetes were the dominant phyla in the rumen microbiota across all groups. Notably, the relative abundance of Proteobacteria was significantly higher in the ME group compared to the LE and HE groups (P < 0.05) (Figures 3A,B). At the genus level, Prevotella was the predominant genus across all groups, with a higher relative abundance observed in the ME group. In addition, the relative abundance of Prevotellaceae UCG-001 was higher in the LE group compared to the ME and HE groups (Figures 3C,D). In the fecal microbiota, Firmicutes, and Bacteroidetes remained the dominant phyla across groups. The relative abundance of Spirochaetota was significantly higher in the HE group than in the LE and ME groups, while the relative abundance of Desulfobacterota was significantly lower in the HE group compared to the LE and ME groups (P < 0.05) (Figures 4A,B). At the genus level, the relative abundance of Treponema increased significantly with rising dietary energy levels (P < 0.05). Additionally, the relative abundance of Monoglobus was significantly higher in the LE group than in the ME and HE groups, whereas the relative abundance of UCG-005 was significantly lower in the LE group compared to the ME and HE groups (P < 0.05) (Figures 4C,D).

The Pearson correlation coefficients among all datasets exceeded 0.99 (Figure 5A), indicating strong stability and high data quality throughout the detection process. PCA score plots demonstrated good clustering within groups and clear separation between groups across the three dietary treatments, confirming the validity of the experimental data (Figure 5B). The PLS-DA plot showed that the samples were distributed within the Hoteling T² ellipse range, and permutation tests between groups (LE vs. ME, R²Y = 0.95, Q²Y = 0.35; LE vs. HE, R²Y = 0.92, Q²Y = 0.41; ME vs. HE, R²Y = 0.84, Q²Y = 0.21) further confirmed the stability, reliability and validity of the model (Figures 5C–E).

Metabolic differences between groups were screened using criteria of P < 0.05 and VIP > 1. A total of 48 differential metabolites were identified between the LE and ME groups, including phosphatidylcholine 36:3 (PC 36:3), phosphatidylcholine 40:5 (PC 40:5), phosphatidylcholine 38:4 (PC 38:4), indole-3-acrylic acid, and pyridoxamine 5-phosphate, among others (Figure 6A). In the comparison between the LE and HE groups, 58 differential metabolites were detected, with significant metabolites including phosphatidylcholine 34:1 (PC 34:1), phosphatidylcholine 38:6 (PC 38:6), phosphatidylcholine 40:5 (PC 40:5), pyridoxine, and pyridoxamine (Figure 6B). Additionally, 35 differential metabolites were identified between the ME and HE groups, with key metabolites including phosphatidylcholine 38:6 (PC 38:6), DL-indole-3-lactic acid, pyridoxamine, pyridoxine, and adenine (Figure 6C).

Enrichment analysis of the differential metabolites using the KEGG database, based on an impact coefficient (Impact > 0.1) and a P < 0.05, revealed that these serum differential metabolites across the three groups were primarily enriched in pathways related to amino acid metabolism, vitamin metabolism, nucleotide metabolism, lipid metabolism, and other secondary metabolite pathways (Figures 7A–C).

Spearman correlation analysis was conducted to examine the associations between rumen microbiota and serum metabolites, as well as between rumen and fecal microbiota, with the results visualized through correlation heatmaps. Initially, the top 30 differential serum metabolites and the top 10 rumen bacterial genera across the three dietary groups were selected for analysis. The results revealed that the genera Quinella, Selenomonas, Methanobrevibacter, and Christensenellacece R-7 group were positively correlated with multiple serum differential metabolites, showing positive associations with six, five, five, and one serum metabolites, respectively. In contrast, Candidatus Saccharimonas exhibited negative correlations with six serum differential metabolites (Figure 8). Overall, the serum differential metabolites correlated with Quinella, Selenomonas, and Methanobrevibacter were primarily associated with lipid metabolism pathways. Further analysis was performed on the top 10 dominant genera in the rumen and feces across the three groups to investigate the interrelationships between rumen and fecal microbiota. The results demonstrated specific correlations between microbial genera in the rumen and those in the feces. Notably, Prevotellaceae UCG-001 in the rumen was negatively correlated with Clostridium sensu stricto-1 in the feces. while Succiniclasticum in the rumen showed a positive correlation with Prevotellaceae UCG-003 in the feces. Additionally, Methanobrevibacter in the rumen exhibited a negative correlation with UCG-005 in the feces (Figure 9).

Genetics, nutrition, and feeding management are key factors influencing yak production. Compared with traditional grazing system, house-feeding and supplementary feeding have been shown to significantly improve yak growth performance (Hu et al., 2021; Dai et al., 2022). Among nutritional strategies, modulating dietary energy and protein levels remains a common and effective approach to enhance growth performance (Sun et al., 2015a). Research on the energy and protein requirements for yaks has gradually increased, with the Indian Council of Agricultural Research (Indian Council of Agricultural Research, 2013) providing guidelines for the nutrient requirements of Yak and Mithun. Additionally, the Qinghai Yak Engineering Technology Research Center has developed a metabolic energy model for yaks at different physiological stages (Hu, 1997; Hao et al., 2020). Studies have also investigated the dietary energy requirements of various yak breeds, including Tianzhubai yak (Liu, 2022), Datong yak (Yao et al., 2022), Maiwa yak (Wang et al., 2017), and Niangya yak (Jiang et al., 2024). However, due to differences in breeds and regional conditions, a unified nutritional standard for yaks has yet to be established. In a study by Yao et al. (2022), three dietary NEg levels (5.51 MJ/kg with 16.72% CP, 6.22 MJ/kg with 16.96% CP, 6.94 MJ/kg with 17.27% CP) were evaluated, and it was found that the ADG increased by 29.69% in the 6.22 MJ/kg group and by 21.88% in the 6.94 MJ/kg group compared with the 5.51 MJ/kg group. Similarly, Ahmad et al. (2020) reported that yaks fed diets with NEg levels of 5.5, 6.2, and 6.9 MJ/kg at a concentrated-to-forage ratio of 30:70 exhibited ADG of 0.63, 0.88, and 0.86 kg/d. respectively. These findings are consistent with the results of the present study, which demonstrated that the ADG in the ME and HE groups was significantly higher than that in the LE group. Similarly, Yang et al. (2013) observed that yak calves fed diets with NEg levels of 6.50 and 7.40 MJ/kg had significantly higher ADG compared with those in the 5.60 MJ/kg group, with the FCR being significantly higher in the 5.60 MJ/kg group than in the other groups.

The digestibility of dietary nutrients plays a crucial role in the growth and development of ruminants (Zhang et al., 2022). Previous studies have shown that increasing dietary energy levels enhances the apparent digestibility of DM, OM, CP, and EE, a finding consistent with research conducted in steers (Navarrete et al., 2017) and cows (Su et al., 2017). The energy-to-protein ratio is an important indicator for evaluating the balance between dietary energy and protein, which directly influences the potential for growth in animals (Medhi et al., 2018; Liu H. et al., 2022). In the present study, the CP level in the diets was kept consistent across groups, and it was observed that increasing dietary energy levels improved the energy-to-protein ratio, thereby supporting enhanced yak growth. Bao et al. (2020) investigated the effects of three net energy (NE) levels (7.64 MJ/kg with 9.82% CP, 8.95 MJ/kg with 10.57% CP, 10.18 MJ/kg with 10.94% CP) in TMR diets for house-fed fattening yaks, reporting that the apparent digestibility of CP, ADF, NDF, and P improved with higher nutritional levels. Similarly, in the present study, higher dietary energy levels were associated with improved apparent digestibility of P, CP, OM, and NDF. However, no significant differences were observed in the apparent digestibility of ADF and CF between the ME and HE groups. Zhu et al. (2023) evaluated metabolizable energy levels of 7.20, 7.89, and 8.58 MJ/kg with 9.0% CP, and found that NDF digestibility was significantly higher in the 7.89 and 8.58 MJ/kg groups compared with the 7.20 MJ/kg group, whereas ADF digestibility did not differ across the groups. This finding suggest that within a certain range of dietary energy levels, further increases in energy intake may not lead to additional improvements in the digestibility of certain nutrients.

The rumen pH value is a critical indicator reflecting the stability of the rumen environment, with a normal range for ruminants typically between 6.2 and 7.2 (Van Soest et al., 1991). In yaks, rumen pH generally ranges from 6.4 to 7.7, which is slightly higher than that observed in other beef cattle (Liu et al., 2019; Pang et al., 2022; Liu X. et al., 2023). In the present study, the pH value ranged from 7.18 to 7.27, consistent with these previous findings. VFAs provide more than 70% of the energy required by ruminants (Zanton and Heinrichs, 2009; Achmadi et al., 2012). The production of acetate is primarily associated with the proportion of roughage in the diet, while the concentration of propionate is linked to the proportion of dietary concentrate (Dórea et al., 2020). In the present study, no significant differences were observed in the concentration of acetic acid and propionic acid among the three groups. However, with increasing dietary energy levels, the concentration of propionic acid increased, while the concentration of acetic acid decreased, which aligns with the findings of Jing et al. (2020) and Chen et al. (2024). This trend may be attributed to the influence of dietary energy levels on the composition and abundance of rumen microbial populations (Li et al., 2019). The optimal concentration of NH₃-N in the rumen has been reported to range between 6 and 30 mg/dL (Preston et al., 1965). In the current study, the NH₃-N concentration fell within this optimal range, indicating that microbial growth and activity were not adversely affected. Nitrogen-containing compounds are degraded into ammonia nitrogen in the rumen, which is then utilized by microbes as a nitrogen source for the synthesis of microbial protein (Pisulewski et al., 1981; Zhou et al., 2022). As dietary energy levels increased in this study, the concentration of NH₃-N decreased, while the content of MCP significantly increased. These results suggest that increasing dietary energy levels enhance the availability of nitrogen for rumen microbes, thereby promoting the conversion of NH₃-N into MCP and facilitating an improved energy-nitrogen balance (Zhou et al., 2019). Similar trends in NH₃-N concentrations have also been reported by Wang et al. (2017) in their study on yaks fed with diets containing NEg levels of 4.17, 4.48, and 4.79 MJ/kg and a CP level of 14.81% under house-feeding condition.

Blood physiological parameters, including biochemical indicators and endocrine hormone levels, serve as quantitative indicators of the internal physiological state of an organism and play a critical role in evaluating animal health. Glucose, a primary energy substrate, typically ranges from 3.9 to 5.3 mmol/L in yaks (Liu et al., 2015; Wang, 2015). Previous studies have demonstrated that serum glucose levels in yaks increase with a higher energy-to-nitrogen ratio (Liu H. et al., 2022). In the present study, glucose concentrations in the ME and HE groups were significantly higher than those in the LE group, consistent with the observed trend in ADG. TG contribute to energy regulation by serving as a reservoir for hydrolyzable non-esterified fatty acids (De Koster and Opsomer, 2013). The TG levels was significantly higher in the ME and HE groups compared to the LE group, whereas NEFA levels were correspondingly lower in these groups. ALP, a marker enzyme associated with osteoblast differentiation and bone metabolism (Sheng et al., 2022), exhibited a trend similar to that of TG and glucose. These findings suggest that increasing dietary energy levels can enhance the conversion of NEFA to TG in the blood, thereby improving glucose and lipid metabolism, supporting body weight gain, and promoting energy balance and fat deposition in yaks. The concentration of β-HB are indicative of an animal’s energy status, with elevated concentrations signaling increased ketone body production (Bonfatti et al., 2019). β-HB concentration exceeding 0.8 mmol/L are typically considered indicative of ketosis. In this study, β-HB levels ranged from 0.13 to 0.14 mmol/L, well below the ketosis threshold, suggesting a favorable energy status in all groups (Pichler et al., 2014). ALT and AST are biomarkers for liver function, with elevated levels commonly associated with hepatic damage (Sun et al., 2015b; Liu H. et al., 2022). In the present experiment, ALT and AST levels remained within the normal physiological range, and no significant differences were observed among the treatment groups, indicating that the dietary energy levels employed did not adversely affect liver function (Qiao, 2013). Indicators such as UN, TP, GLB, and ALB reflect protein metabolism and amino acid balance (Qin et al., 2022). No significant differences were observed in these parameters among the dietary groups, likely due to the consistent crude protein content across all experimental diets. The finding is consistent with results reported by Zhu et al. (2023), who found no significant changes in protein metabolism indices in yaks fed diets with metabolizable energy levels of 7.20, 7.89, and 8.58 MJ/kg and identical CP levels.

Hormones play a critical role in regulating energy metabolism, growth, and development in animals. Numerous studies have demonstrated that GH is positively correlated with the ADG of ruminants (Sarkar et al., 2008; Tomczak et al., 2019). It has been reported that dietary energy level influences the growth performance of yaks through the actions of both GH and IGF-1 (Miyake et al., 2016). Specifically, IGF-1 contributes to the enhancement of ADG with increasing dietary energy levels and does not rely solely on GH to mediate growth performance (Yang et al., 2019). In the present study, IGF-1 levels in the ME and HE groups were significantly higher than those in the LE group, while no significant difference in ADG was observed between the ME and HE groups. IGF-1 is also positively correlated with INS, jointly promoting the production performance in beef cattle (Ciccioli et al., 2003; Dance et al., 2017). In this study, both IGF-1 and INS concentrations in the ME and HE groups were significantly higher than those in the LE group, indicating that increasing dietary energy levels appropriately can enhance glucose metabolism in yaks, thereby promoting growth, development, and production performance. This finding is consistent with results reported in Tibetan sheep (Zhou et al., 2019). LEP, secreted by adipocytes, plays a role in inhibiting excessive fat deposition, while the expression of the leptin gene in cattle has been shown to promote weight gain (Nkrumah et al., 2005). In the current experiment, leptin levels in the ME and HE groups were significantly higher than those in the LE group. The increase in dietary energy level likely led to elevated fat deposition in yaks, thereby enhancing leptin secretion. Additionally, Thyroid-stimulating hormone (TSH) promotes the release of leptin, which in turn stimulates the production of T4 (Duntas and Biondi, 2013). This mechanism may explain the significantly higher serum T4 concentrations observed in the ME and HE groups compared to the LE group in this study.

The rumen microbiota plays a critical role in maintaining the health and normal digestive function of ruminants, contributing to the harvesting of 85–90% of the gross energy available from the diet (Ley et al., 2008). In the present study, the Shannon and Simpson indices revealed that microbial diversity was lowest in the LE group, indicating that excessively low dietary energy levels reduce microbial richness (Yang et al., 2021). Similarly, a study examining the NEg levels of 5.5, 6.2, and 6.9 MJ/kg in yak rumen microbiota reported that the Shannon and Simpson indices were higher at 6.2 and 6.9 MJ/kg compared to 5.5 MJ/kg, a trend that consistent with the findings of this study (Ahmad et al., 2020). PCoA analysis revealed that dietary energy level exerted a substantial influence on the composition of rumen microbiota (Islam et al., 2021). Firmicutes and Bacteroidota were identified as the dominant phyla in the rumen microbiota of yaks, consistent with previous studies (Pang, 2023; Zhang et al., 2023). Firmicutes are primarily involved in the degradation of fibrous materials, whereas Bacteroidetes mainly degrade non-fibrous materials (Evans et al., 2011; Reigstad and Kashyap, 2013). The present findings indicated that with increasing dietary energy levels, the relative abundance of Bacteroidetes increased, while that of Firmicutes decreased, although no significant differences were observed among the groups. This contrasts with some previous research results, which reported significant variations in these phyla with dietary energy level adjustments (Zhou et al., 2017; Liu et al., 2019). Proteobacteria play an important role in biofilm formation, fermentation, and the digestion of soluble carbohydrates (Pitta et al., 2016). Previous research in dairy cows has shown that the abundance of Proteobacteria increases with high-concentrate diets (Kibegwa et al., 2023). In the current study, the abundance of Proteobacteria was significantly highest in the ME group. It is hypothesized that the rumen environment developed under the ME diet represents a threshold for a linear increase in Proteobacteria abundance, thereby creating favorable conditions for efficient rumen fermentation. At the genus level, the relative abundance of Prevotella was highest in the ME group, consistent with findings in yak by Ma et al. (2023). Prevotella is essential for protein and starch metabolism and is capable of utilizing pectin and cellulose, contributing to improved feed utilization efficiency (Bekele et al., 2010). These findings suggest that the energy level of the ME diet in this study may be optimal for enhancing dietary utilization efficiency in yaks. Prevotellaceae UCG-001 has been reported to promote the resynthesis of branched-chain fatty acids through the elongation of valeric acid and propionic acid (Bainbridge et al., 2018). In this study, the lowest concentration of propionic acid in the rumen was found in the LE group, which exhibited the highest abundance of Prevotellaceae UCG-001. This may indicate that Prevotellaceae UCG-001 utilized propionic acid for the synthesis of other branched-chain fatty acids under low-energy diets, thereby influencing the rumen volatile fatty acid profile in yaks.

Fermentation in the hindgut contributes approximately 10–15% of the gross energy available to adult cattle, underscoring the critical role of hindgut bacteria in energy extraction and growth in ruminants (Fan et al., 2019). In dairy cows, high-concentrate diets have been reported to remodel the composition and function of hindgut microbiota (Zhang et al., 2021). In the present study, the alpha diversity indices, including observed features, Chao1, and Shannon, were significantly higher in the LE group compared to the HE group. Beta diversity analyses demonstrated clear differences in microbial community distribution among the three groups, indicating that dietary energy levels exert a substantial influence on the composition of fecal microbiota. These findings suggest that excessive dietary energy may inhibit the proliferation of fecal microbiota. Firmicutes and Bacteroidota accounted for over 90% of the fecal microbiota, consistent with observations in beef cattle (Zhang et al., 2023) and Holstein calves (Gao, 2018). The phylum Spirochaetota has been associated with the energy metabolism of ruminants, with its abundance increasing alongside elevated dietary energy levels (Bi et al., 2018; Pal et al., 2021). In this study, the abundance of Spirochaetota significantly increases with higher dietary energy levels, indicating that raising the dietary energy may enhance energy metabolism in the hindgut. Desulfobacterota contributes to creating an anaerobic environment favorable for gastrointestinal fermentation (Cai, 2021). The lowest abundance of Desulfobacterota was observed in the HE group, indicating that excessively high dietary energy levels may negatively affect hindgut fermentation conditions. Additionally, the abundance of Treponema, a genus within the phylum Spirochaetota, increased significantly with higher dietary energy levels. Treponema plays a beneficial role in nutrient metabolism, consistent with findings in yaks reported by Shen et al. (2024). The relative abundance of UCG-005 was highest in the ME group in this study. A previous study in goats indicated that the abundance of UCG-005 increased with higher NDF levels (Zhang and Wang, 2018). Monoglobus, a genus within Firmicutes known for its role in degrading pectin and fiber substances (Kim et al., 2019), exhibited significantly lower abundance in the HE groups compared to the LE group. This aligns with the current findings, suggesting that while appropriate increases in dietary energy support the growth of fiber-degrading bacteria, excessive energy levels may exert inhibitory effects. These results collectively indicate that the selection of dietary energy levels should be approached from a comprehensive perspective, considering both energy metabolism and hindgut microbial balance to align with specific feeding objectives in yak production systems.

Metabolomics data provide valuable insights into the physiological and metabolic status of animals (Vinayavekhin et al., 2010). Phosphatidylcholine (PC) acts as a storage molecule for choline, with its metabolites participating in the glycerophospholipid metabolism pathway (Mohamad et al., 2015). Choline supplementation has been shown to aid in preventing fatty liver in early lactating dairy cows (Lu et al., 2023; Lima et al., 2024). In the present study, the levels of phosphatidylcholine substances (PC 40:5, 34:1, 38:6) were significantly upregulated in the HE group compared to LE and ME groups. This suggests that a high-energy diet may influence lipid metabolism by altering lipid levels and their metabolites, thereby increasing the availability of substrates for energy metabolism in the liver and peripheral tissues (Bäckhed and Crawford, 2010). Pyridoxine, pyridoxamine, and pyridoxamine 5-phosphate present the active forms of vitamin B6, which are essential for efficient protein synthesis and critical for growth, lactation, and overall health in animals (Magan et al., 2020; Wu et al., 2022). In this study, the content of active vitamin B6 was highest in the ME group, indicating that this dietary energy level may be more suitable for supporting the growth and development of yaks. DL-indole-3-lactic acid and indole-3-acrylic acid are key metabolites of the tryptophan metabolic pathway. As a functional essential amino acid, tryptophan plays a significant role in influencing growth performance, feed intake, lactation, and the body’s antioxidant and immune functions (Li et al., 2016). The results demonstrated that the content of indole-3-acrylic in the ME group was higher than that in the LE group, while the level of DL-indole-3-lactic acid in the ME group exceeded that observed in the HE group. This finding indicates that appropriate increases in dietary energy levels can support improved growth performance and feed intake in yaks. Adenine is integral to energy metabolism and is involved in the synthesis of DNA and RNA within cells. Supplementation with nucleotides in the diets of piglets has been reported to enhance growth performance and immunity (Diehl et al., 2022). In the current study, adenine was found to be highly expressed in the ME group, suggesting that a medium energy diet may facilitate the promotion of growth and enhance cellular energy metabolism in yaks.

Quinella exhibited a positive correlation with several differential serum metabolites, including X11-deoxyprostaglandin F1α, X16-heptadecyne-1,2,4-triol, estriol, and prostaglandin A3. These metabolites are associated with lipid metabolism pathway according to the KEGG metabolic classification.⁴ The highest relative abundance of Quinella was observed in the HE group, indicating that an increase in dietary energy level can increase Quinella abundance, thereby potentially influencing serum lipid metabolism in yaks. Selenomonas was also found to be positively correlated with PC 36:5, X16-heptadecyne-1,2,4-triol, and prostaglandin A3. As Selenomonas ferment glucose to produce acetic acid and propionic acid (Zhao et al., 2022), and PC 36:5 serves as a choline storage molecule involved in the glycerophospholipid metabolism pathway (Athamena et al., 2011), these findings suggest that Selenomonas may contribute to the regulation of both lipid and glucose metabolism in rumen environment.

Additionally, Succiniclasticum in the rumen was positively correlated with Prevotellaceae UCG-003 in feces. Succiniclasticum utilizes succinic acid to produce propionic acid (Liu X. et al., 2022), whereas Prevotellaceae UCG-003 plays a role in the degradation of protein and starch. A negative correlation was observed between Methanobrevibacter in the rumen and UCG-005 in feces. Although both taxa are involved in fiber utilization, they produce different metabolic substrates (He et al., 2015; Zhang and Wang, 2018). The correlations between rumen and fecal microbiota indicate a functional and compositional linkage between these microbial communities, with distinct yet complementary roles in the digestive process of yaks. Notably, a significant relationship was identified between the relative abundances of Prevotellaceae UCG-001 and Clostridium sensu stricto-1, underscoring the need for further investigation into their functional roles within the yak gastrointestinal ecosystem.