All experiments were approved by and conducted in accordance with the animal care guidelines of the Animal Care Committee of the Institute of Subtropical Agriculture, Chinese Academy of Sciences.

Three types of fermentation substrates were used for in vitro incubation: (1) a basic diet containing ammonium chloride (A), (2) a basic diet containing biuret (B), and (3) a basic diet containing glutamine (G). Maize stover of “Kexiangtian 1” (locally breed) was selected as forage. The diets were dried at 65°C, ground, and stored in air-tight bags. Ammonium chloride, biuret, and glutamine were purchased from the China National Medicines Co. Ltd. (Beijing, China). The composition and nutrient levels of basic experimental diets are presented in Table 1 .

Three ruminally fistulated Xiangdong black goats (a local breed in Hunan Province, China) with an average body weight of 25.0 ± 2.5 kg (mean ± SD) were used in the experiment. The goats were fed a basic diet comprising a mixture of corn–soybean meal concentrate and maize stover (50:50). The diet was provided twice a day as two meals at 08:00 and 17:00 h, and water was provided ad libitum.

In all feed samples, DM (method 945.15), CP (method 954.01), NDF, and ADF (method 973.18) contents were analyzed following published methodologies (Williams et al., 1990). The pH of the fermentation tank was automatically detected by the pH electrode installed in the tank and entered into the computer.

The individual VFA concentrations in the supernatant were measured using gas chromatography (HP5890, Agilent 5890; Agilent, Santa Clara, CA) according to the method described by Chen et al. (2017). The NH₃-N concentration was determined on the basis of the ultraviolet absorption value recorded at 700 nm using a spectrophotometer (UV-2300, Shimadzu, Japan), according to the method described by Wang et al. (2016). Microbial crude protein (MCP) production was estimated according to the purine method described by Hu (2005).

Protozoa counts were obtained according to the method described by Wu (2017). Using staining microscopy, 5 ml of rumen fluid and 10 ml of methyl green formalin solution were collected, mixed well, and stained for 24 h. A disposable pipette was used to remove the stained solution. Then, the suspension was swiftly dropped onto a blood cell counting plate until it covered the entire counting chamber. The liquid-filled counting plate was gently enclosed with a glass coverslip. Finally, cell count was determined under an optical microscope.

In the selected plate count area of 1 mm², five squares each at the four corners and in the middle, totaling 25 squares, were analyzed. The average count was calculated as follows:

where X is the total number of protozoa of five squares, D is the fold-dilution of rumen fluid, and 1,000 is the coefficient for the conversion of 1 mm³ square volume to 1 ml.

Microbial DNA was extracted using a bacterial DNA kit (Omega Bio-Tek Inc., Norcross, GA, USA), according to the manufacturer’s protocol. The extracted DNA samples were detected by 1% agarose gel electrophoresis and spectrophotometry (Nanodrop 2000, America) (260/280 nm density ratio) for quality check. The V3–4 hypervariable region of the bacterial 16S rRNA gene was amplified using primers 338F (ACTCCTACGGGAGGCAGCAG) and 806R (GGACTACHVGGGTWTCTAAT) (Mumbi et al., 2015). For each sample, a 10-digit barcode sequence was added at the 5′-end of the forward and reverse primers (Allwegene Technology Inc., Beijing). Polymerase chain reaction (PCR) (LightCycler 480II, Germany) was performed on a MasterCycler Gradient (Eppendorf, Germany) using a 25-μl reaction system containing 12.5 μl of KAPA 2G Robust Hot Start Ready Mix, 1 µl of forward primer (5 µM), 1 µl of reverse primer (5 µM), 5 µl of DNA (total template quantity = 30 ng), and 5.5 µl of H₂O. The reaction conditions were as follows: 95°C for 5 min, followed by 28 cycles at 95°C for 45 s, 55°C for 50 s, and 72°C for 45 s and final extension at 72°C for 10 min. The PCR products were purified using the QIAquick Gel Extraction Kit (QIAGEN, Germany), quantified using real-time PCR, and sequenced on the MiSeq platform at Allwegene Technology Inc., Beijing.

Raw data were removed from consideration if they were shorter than 200 bp, had a low-quality score (≤20), contained ambiguous bases, or did not exactly match the primer sequences and barcode tags. Qualified reads were separated using sample-specific barcode sequences and trimmed using the Illumina Analysis Pipeline version 2.6. The dataset was then analyzed using QIIME. The sequences were clustered into operational taxonomic units (OTUs) at 97% similarity (Edgar, 2013). The Ribosomal Database Project (RDP) tool was used to classify all sequences into different taxonomic groups (Cole et al., 2009).

The evolutionary distances between microbial communities in each sample were calculated using the TAYC coefficient and represented as an unweighted pair group method with arithmetic mean (UPGMA) clustering tree describing the dissimilarity (1 − similarity) between multiple samples (Jiang et al., 2013).

The statistical model included dietary treatments (n = 3) as the fixed effects, treatment replicates (n = 6) as a repeated measure, and period (n = 3) and fermentation tank (n = 6) as the random effects. Data were subjected to single factor analysis in SPSS 19.1.0. Significance was declared at P < 0.05, and 0.05 ≤ P ≤ 0.10 indicated a significant trend.

DMD, CPD, NDFD, and ADFD were not affected (P > 0.05) by NPN sources ( Table 2 ). However, DMD and ADFD were numerically higher (P > 0.05) but CPD and NDFD were lower (P > 0.05) in group G than those in the other groups.

DM and NDF are important indicators of dietary utilization during ruminal fermentation. Increasing NPN level in in vitro cultures of rumen inocula enhanced DMD (Muthia et al., 2021). Further, NPN source affected DMD; specifically, biuret-straw and urea-straw significantly improved DMD, because starch, as a carbohydrate, readily provides energy for rumen fermentation (Muthia et al., 2021). However, the present experiment showed that DMD, CPD, ADFD, and NDFD did not differ among the treatments. In supplemented diets, nitrogen and energy are available simultaneously and in proportions required by rumen microbes, thus promoting diet digestion (Muthia et al., 2021). Although increased dietary energy levels significantly enhance apparent digestibility, the dietary levels of the three treatments were equal in terms of nitrogen and energy supply in the present study. This may explain the lack of significant differences in apparent digestibility among the experimental groups.

pH reflects the overall state of rumen fermentation; however, it is easily affected by various factors, and a relatively stable pH value is considered beneficial to animal health and feed utilization (Bao et al., 2011). In the present experiment, the pH of approximately 5.5–6.0 maintained the normal state of the fermentation tank. In a study by Wang et al. (2010), treatments including ammonium chloride in different molecular forms of nitrogen showed the highest pH values. Similarly, in the present experiment, ammonium chloride showed the highest pH value. Ammonium chloride is a highly water-soluble ionic compound. Ammonium radicals and chloride ions are ionized in water; thus, ammonia gas accumulation elevates pH. In vitro results showed that the NH₃-N concentration in group G was the highest among the three treatments. Protozoa exhibit a stronger deamination activity than bacteria. Protozoa are the net ammonia-producing microorganisms but cannot use ammonia to synthesize bacterial proteins (Huntington and Joan, 2021). Therefore, in group G, which showed a higher protozoa count, NH₃-N concentration in the rumen was higher than that in groups A and B. In addition, glutamine and ammonium chloride could be directly utilized by microorganisms, and ammonium ions can dissociate, increasing the NH₃-N concentration (Geisseler et al., 2011). These results are also consistent with the lowest NH₃-N concentration in group B of the present experiment. Low ruminal NH₃-N concentrations (<2.94 mM) can limit microbial growth and ruminal fermentation (Wang et al., 2018), thereby reducing the MCP content. These results were consistent with the lowest MCP in group B. NH₃-N acts as the raw material for MCP synthesis and can directly reflect the production of microbial proteins. Simultaneously, in the present experiment, SP3-e08 and Prevotellaceae_UCG-001, which degrade carbohydrates, were more abundant in group A than in group G. As the synthesis efficiency of MCP is affected by many factors, carbohydrates in the feed can be fermented by microorganisms, providing energy and affecting the efficiency of microbial protein synthesis (Feng et al., 1993; Xie et al., 2019).

VFAs are the final products of rumen fermentation and provide energy to the body. In the present study, total VFAs did not differ among the treatments, due perhaps to the equal nitrogen and energy levels and similar DM degradability of the three treatments. Specifically, group G showed the highest propionate and butyrate levels, which may result from the highest NH₃-N concentration in this group. As such, increased ruminal NH₃-N concentrations improve energy production from straw (Lee et al., 1985). Furthermore, the molar ratio of acetate to propionate is an important measure of microbial fermentation. In the present experiment, the lowest acetate-to-propionate ratio was recorded in group G, probably because the propionate content in this group was significantly higher than that in groups A and B. In a study on lactating dairy cows, a linear decrease in isobutyrate with increase ammonium chloride concentration was observed, although it did not significantly affect other VFA. However, these results are inconsistent with our findings of the highest valerate and isovalerate proportion in group A. In dairy cows that are fed a diet based on maize stover as the only source of roughage, rumen valerate and isovalerate concentrations increase significantly (Weng, 2013). Increased pH, enhanced affinity of rumen microorganisms to fiber, and augmented ATP production affect VFA generation (Strobel and Russell, 1986). Furthermore, the type of roughage may affect valerate and isovalerate production (Yue et al., 2009; Ponce et al., 2016). In the present experiment, maize stover was added as roughage to the experimental diets. Nevertheless, the acetate-to-propionate ratio was the highest in group A, which tended toward acetate fermentation, changed the rumen fermentation mode, and increased the fiber content during fermentation. These results are consistent with the highest relative abundance of Ruminococcus_1, a major fiber-degrading bacteria in the rumen. Thus, greater maize stover fermentation in the rumen increased valerate and isovalerate proportion in group A compared with those in the other groups. Therefore, roughage utilization affects VFA generation, regardless of the concentrate-to-roughage ratio.

Ruminal microbiome is a complex and important natural fermentation system, which has long been a focus of ruminant research (Abecia et al., 2014). Rumen microorganisms are always in dynamic equilibrium. Dietary factors are the key influences, and the corresponding rumen microorganisms also affect host metabolism and nutrition (Sommer and Ckhed, 2013). Fibers and carbohydrates in the diet are degraded and fermented by ruminal microorganisms to synthesize MCPs and provide nutrition to the host animals. Short-chain fatty acids produced during anaerobic fermentation can meet 70% of the daily energy requirements of animals (Kelsea et al., 2015; Shabat et al., 2016). In addition, 50% of the ruminant protein demand is derived from ruminal microorganisms (Yeoman and White, 2015). Since the structural composition of the colony could not be deduced based on alpha diversity indices, we further analyzed the differences in bacterial abundances among the three groups. Jami and Mizrahi (2012) characterized the core microbiome of bovine rumen and identified Bacteroidetes, Firmicutes, and Spirochaetae as the top three phyla, consistent with the findings of the present study. The dominant microflora was not affected by diet, although the addition of different nitrogen sources affected the relative abundances of some rare rumen bacteria. Specifically, the relative abundances of Cyanobacteria, Planctomycetes, Elusimicrobia, and Armatimonadetes were the highest in group G and significantly higher than those in the B group. This may be because the NH₃-N concentration in group B was very low, which likely limited the growth of microorganisms and reduced their activity. Surprisingly, the abundance of SR1_Absconditabacteria was highest in group B, with the lowest microbial protein and NH₃-N concentration concentrations. Sun et al. (2019) reported that 45% low-concentration diets increased the relative abundance of SR1_Absconditabacteria on at the phylum level compared with 55% high-concentration diets. SR1_Absconditabacteria may not rely primarily on nitrogen sources for nutrition in the rumen; therefore, its activity was not restricted even at low NH₃-N and protein levels. At the genus level, the relative abundances of Rikenellaceae_RC9_gut_group and Prevotella_1 were the highest in all groups, indicating that these genera play pivotal roles in rumen NPN metabolism. Prevotellaceae_UCG-001 is one of the major protein-degrading bacteria in the rumen (Avgustin et al., 1997). In the present study, the relative abundances of Prevotellaceae_UCG-001 and SP3-e08 were the lowest in group B, which is consistent with the lower ruminal MCP content in this group. However, the relative abundance of Rikenellaceae_RC9_gut_group was the highest in group B and the lowest in group G. Biuret is a urea derivative, and the mechanism through which rumen microorganisms convert urea to bacterial proteins differs from that through which they convert other NPN sources (Balasubramanian et al., 2013). Biuret cannot be directly degraded by microorganisms, but it can be catalyzed by ureases secreted by microorganisms (Strope et al., 2011). Moreover, the solubility of biuret is low, its degradation rate of is slow, and it remains in the rumen for a longer time, which may be conducive to the growth of Rikenellaceae_RC9_gut_group and alter the activity of other microorganisms (Ali, 2007). Butyrivibrio_2 is a fiber-degrading bacterium, and its relative abundance was the highest in group B. Based on the results of apparent indicators of nutrient disappearance rates, NDFD was the highest in group B. This may explain the higher microbial activity of Butyrivibrio_2 in group B. Griffith et al. (2016) reported that ammoniated barley straw feed is more suitable for fiber degradation by rumen microbes. Moreover, a higher relative abundance of Ruminococcus_1 was observed in group A, possibly because ammonium chloride was more easily used by these microbes, which promoted their proliferation. Finally, the relative abundance of Unidentified microbes was 25%, whose phyla and functions were unclear. Further studies on these microbes are warranted to better understand the metabolic mechanisms of NPN in the rumen.

In conclusion, the relative abundances of the major fiber-degrading bacteria Ruminococcus_1 and protein-degrading bacteria SP3-e08 and Prevotellaceae_UCG-001 were lower in group B than in the other groups. Group G showed significantly increased the concentrations of propionate, NH₃-N, and microbial proteins, which decreased the acetate-to-propionate ratio and appeared to shift the fermentation pathways from acetate to propionate. Thus, glutamine is more suitable than biuret as a dietary source of non-protein nitrogen.