Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Effect of different forage-to-concentrate ratios on ruminal bacterial structure and real-time methane production in sheep

  • Runhang Li,

    Roles Writing – original draft

    Affiliation College of Animal Science and Technology, Jilin Agricultural University, Changchun, PR China

  • Zhanwei Teng,

    Roles Methodology

    Affiliation College of Animal Science and Technology, Jilin Agricultural University, Changchun, PR China

  • Chaoli Lang,

    Roles Formal analysis

    Affiliation College of Animal Science and Technology, Jilin Agricultural University, Changchun, PR China

  • Haizhu Zhou,

    Roles Validation, Visualization

    Affiliation College of Animal Science and Technology, Jilin Agricultural University, Changchun, PR China

  • Weiguang Zhong,

    Roles Investigation

    Affiliation Jilin Academy of Agricultural Sciences, Changchun, PR China

  • Zhibin Ban,

    Roles Methodology, Software

    Affiliation Jilin Academy of Agricultural Sciences, Changchun, PR China

  • Xiaogang Yan,

    Roles Investigation, Visualization

    Affiliation Jilin Academy of Agricultural Sciences, Changchun, PR China

  • Huaming Yang,

    Roles Methodology, Supervision

    Affiliation Jilin Academy of Agricultural Sciences, Changchun, PR China

  • Mohammed Hamdy Farouk,

    Roles Conceptualization, Resources

    Affiliation Animal Production Department, Faculty of Agriculture, Al-Azhar University, Nasr City, Cairo, Egypt

  • Yujie Lou

    Roles Conceptualization

    Affiliation College of Animal Science and Technology, Jilin Agricultural University, Changchun, PR China


Emission from ruminants has become one of the largest sources of anthropogenic methane emission in China. The structure of the rumen flora has a significant effect on methane production. To establish a more accurate prediction model for methane production, the rumen flora should be one of the most important parameters. The objective of the present study was to investigate the relationship among changes in rumen flora, nutrient levels, and methane production in sheep fed with the diets of different forage-to-concentration ratios, as well as to screen for significantly different dominant genera. Nine rumen-cannulated hybrid sheep were separated into three groups and fed three diets with forage-to-concentration ratios of 50:50, 70:30, and 90:10. Three proportions of the diets were fed according to a 3 × 3 incomplete Latin square, design during three periods of 15d each. The ruminal fluid was collected for real-time polymerase chain reaction (real-time PCR), high-throughput sequencing and in vitro rumen fermentation in a new real-time fermentation system wit. Twenty-two genera were screened, the abundance of which varied linearly with forage-to-concentration ratios and methane production. In addition, during the 12-hour in vitro fermentation, the appearance of peak concentration was delayed by 26-27min with the different structure of rumen bacteria. The fiber-degrading bacteria were positively correlated with this phenomenon, but starch-degrading and protein-degrading bacteria were negative correlated. These results would facilitate macro-control of rumen microorganisms and better management of diets for improved nutrition in ruminants. In addition, our findings would help in screening bacterial genera that are highly correlated with methane production.


Of the total methane emission in China, the emission from ruminants was estimated to be approximately 17%, turning them into one of the largest anthropogenic sources of methane emission [1]. The emission of methane associated with agriculture is expected to see a significant increase. Therefore, new strategies were needed for reducing the emission and improving livestock productivity, which had been extensively studied and reviewed [2].

Rumen is the main site of methane production [3], which provided a habitat for a variety of microbes, including numerous species of bacteria, archaea, viruses, protozoa and fungi [4]. In the anaerobic environment of the rumen, several organic compounds present could eventually be decomposed into methane by a number of microorganisms [5]. The composition of ruminal microbiome was affected by different factors, such as age, breed, general well-being of the animal, its location as well as administration of feed and antibiotics [68]. Furthermore, feedstuffs were the main factors regulating the composition and functional patterns of ruminal microbiome [911]. Among the nutritional indices of diets, protein and energy levels were the major factors affecting the fermentation of ruminal microbiome [12]. Fibers, including hemicellulose and cellulose, were the main source of energy [13], which could be degraded into methane by the microbes present in the rumen. Leng and Nolan [14] pointed out that 80% of the nitrogen available to ruminal bacteria came from ammonia and 20% was derived from amino acids or oligopeptides. Therefore, the low content of ammonia promotes microorganisms to degrade other nitrogen sources in a diet with high forage-to-concentration ratios (F:C), which delays fermentation. Grovum and Leek [15] found that non-structural carbohydrates were degraded much faster than structural ones. Easily degradable carbohydrates provide energy and carbon sources for faster microbial fermentation and increase fermentation rate. Methane production can be affected by the above-mentioned factors.

In the current models established for the same rumen microflora to predict methane production, nutritional indicators had been used as parameters [1621]. A large number of calibration parameters are required for the models to adapt to plentiful situations, thereby limiting the scope of these models. Some mechanistic models considering the role of rumen microbes [2223] had been established by the extrapolation of mathematical formula used by computers. Because of high operation cost, it is difficult to apply these models to actual production systems. Therefore, the application scope of these models will be greatly expanded if some important microorganisms can be related with the models using nutrient indicators as parameters.

Previous studies have indicated that archaea are the main microorganisms producing methane in the rumen [24]. However, other recent studies involving high-throughput sequencing have shown that change in methane production is irrelevant to archaea flora, but highly correlated with bacterial flora [25]. The main function of bacteria is to break down the nutrients in the feed into simple compounds and additional products used by animals, including hydrogen, carbon dioxide and volatile fatty acids which are raw materials for methane synthesis [3]. To establish better models for methane prediction with wide range of application, characteristic microorganisms should be screened from rumen bacterial communities to serve as effective parameters.

Simple devices for in vitro fermentation have been used to establish the prediction models [21]. However, in such cases, methane production could only be detected either at specific time points or at the final time point, and therefore, did not reveal the overall fermentation status well. Sun et al. [26] used a new real-time in vitro fermentation system to determine the methane production time course when they studied the effect of cysteamine hydrochloride and nitrate supplementation on methane production and productivity in cattle. This system makes it possible to determine a more subtle fermentation state. Thus, in exploring the relationship between rumen microbial structure and methane production, this system may provide more detailed reference data.

We hypothesized that the real-time methane production of sheep would be highly correlated with the abundance of bacteria in the rumen. The objective of this study was to investigate the relationship among different structures of bacterial flora in the rumen, dietary levels, and methane production, using the in vitro fermentation system. The genera of bacteria that showed high correlation with methane production were screened in order to serve as the reference for accurate prediction of methane production.

Material and methods

Ethics statement

All research involving animals was conducted according to Guide for the Care and Use of Laboratory Animals which was approved by the ethics committee of Jilin Agricultural University, P. R. China. The ethics committee of Jilin Agricultural University, P. R. China approved this study, and the approved permit number for this study is “JLAC20171104”.

Animals and diets

A total of nine rumen-cannulated (cannulated at one year of age) hybrid sheep (Chinese merino fine wool sheep × Dorper sheep) were selected, which was 2 years old and whose average weight was 87.83 ± 8.11kg. Randomly assigned to three groups, these sheep were separately fed at random. Jilin Agricultural University, Changchun, China prepared Guide for the Care and Use of Laboratory Animals which provided guidance for all animal-related procedures.

Chosen as the forage, Leymus chinensis was mixed with the concentrate in three proportions of F:C including 50:50 (L), 70:30 (M) and 90:10 (H). The composition and nutrition levels of the three diets based on the NRC [27] are shown in Table 1. The three diets were fed according to the 3 × 3 incomplete Latin square design over 45d in three periods of 15d each, including 14d of pre-feeding and the 15th day for sampling. Three distinct flora structures were established under different treatments.

Sampling and DNA extraction

Ruminal fluid (400mL) was collected by pump and pre-warmed thermos before feeding (07:00) and saturated with CO2. Filtrated through a double-layered gauze, the collected fluid was used to measure the pH value to confirm the health of rumen. The fluid with the pH value between 5.5 and 7.5 from three sheep of one group was mixed. All the samples with 10mL were respectively stored in sterile centrifuge tubes (without any treatment) with 2mL at -80°C for DNA extraction. Another 300mL of ruminal fluid from each group was warmed to 39°C for in vitro rumen fermentation right after sampling.

Microbial genomic DNA was extracted from all ruminal fluid samples with 220mg using the methods of Murray and Thompson [28] and Zhou et al. [29]. Agarose gel electrophoresis was applied to confirm the successful extraction of DNA [30]. The qualified DNA continued to be tested for real-time polymerase chain reaction (real-time PCR) and high-throughput sequencing. A total of 9 samples for the three diets were collected and each sample was tested twice in order to have six replicates for each diet.

Real-time PCR for total bacteria, methanogens, protozoa and anaerobic fungi

Real-time PCR were tested on Applied Biosystems StepOne Real-time PCR System based on the methods of Denman and McSweeney [31]. The designed primers were shown in the Table 2. Below is the reaction system (25μL): SYBR Premix Ex Taq (RR420A, Takara) 12.5μL, forward primer 0.5μL, reverse primer 0.5μL, DNA template 2.0μL, sterile distilled water 9.5μL. Below is the reaction conditions: 95°C 2min; 95°C 5s, 60°C 30s, 40 cycles; 95°C 15s; 60°C 1min; 95°C 15s. The relative abundance of rumen microorganisms was expressed as a percentage or a thousandth of 16S rDNA relative to the total rumen bacteria. The formula is as follows: Target = 2 -(Ct target −Ct total bacteria) × 100; The Ct value is defined as the number of amplification cycles that are passed when the fluorescence signal of the amplification product reaches a set threshold during real-time PCR amplification.

PCR amplification of 16S rDNA, amplicon sequence and processing of sequence data

Based on previous comparisons [3436], 16S rDNA had V4 hyper variable regions which performed PCR amplification for microbial genomic DNA extracted from ruminal fluid samples and were adopted in the rest of the study. PCR primers which flanked bacterial 16S rDNA’s V4 hypervariable region were designed. The forward primer with a barcode was 338F 5’-ACTCCTACGGGAGGCAGCAG-3’ while the reverse primer referred to 806R 5’-GGACTACHVGGGTWTCTAAT-3’ based on the approach of Fan et al. [37]. Below is the PCR reaction system (TransGen AP221-02, 20μL): 5×FastPfu Buffer, 2.5 mM dNTPs, Forward Primer (5μM), Reverse Primer (5μM), FastPfu Polymerase, BSA and Template DNA and ddH2O with 4μL, 2μL, 0.8μL, 0.8μL, 0.4μL, 0.2μL and 10ng respectively were added up to 20μL totally. Below are PCR conditions: One pre-denaturation cycle, 27 denaturation cycles, annealing, elongation and one post-elongation cycle at 95°C, 95°C, 55°C, 72°C and 72°C for 3min, 30s, 30s, 45s and 10min respectively. Separated on 1% garose gels, products of PCR amplicon were obtained by the extraction of gels. Sequencing only adopted PCR products which were void of contaminant bands and primer dimers by means of synthesis. Illumina MiSeq PE300 proposed the paired-end approach which was taken for the sequencing of barcoded V4 amplicons. The filtering of effective reads was based on the methods of [3841]. Sequences with lower mean phred score (20bp), equivocal bases and primer mismatching or sequence lengths below 50bp were removed. The assembly of only the sequences which had an overlap above 10bp and no mismatch was completed in accordance with their overlap sequence. Reads which were unable to be assembled were abandoned. Barcoded and sequencing primers were removed from the sequence which was assembled.

Taxonomic classification and statistical analysis

A web-based program called Usearch (version 7.0, was applied to carry out taxon-dependent analysis. 16S rRNA gene sequences were used for phylogenetically consistent bacterial taxonomy according to the method of a Bayesian classifier, Ribosomal Database Project Classifier [41]. The Silva Database (Silva_128_16s, was compared to calculate the operational taxonomic units (OTUs) for all samples to show the abundance of bacterial species with 97% of identity cutoff, whereas the species for which the sum of OTUs of all the samples was more than 20 reads were retained. The richness of OTUs for each sample was produced at the level of genus. The length of all the valid bacterial sequences with primers was 440bp on the average. The calculation of abundance at the level of genus was transformed according to log2 and normalized as the method of Niu et al. [42]. Inter-group OTUs were compared by the generation of a Venn diagram. The bacterial community indices adopted contained Chao and Shannon’s coverage. The diversity of bacteria was presented by the quantity of OTUs.

In vitro rumen fermentation

The substrate was made with the feed of group M in the feeding experiment by drying and grinding through a 0.45mm sieve. Collected from different dietary treatments during the feeding experiment, the ruminal fluid was filtrated by four-layer cheesecloth and mixed with pre-heated artificial saliva [43] at a ratio of 2:1 (buffer: ruminal fluid, v:v). The ruminal fluid (150mL) which was buffered was dispensed into pre-warmed 200-mL incubation flasks. Two grams of each substrate was blended with the buffered ruminal fluid in each incubation flask which was incubated at the temperature of 39°C for 12h in water. The production ratio of methane was measured by real-time in vitro fermentation system (produced by Jilin Academy of Agricultural Sciences, code Qtfxy-6), which was tested for the effluent gas discharged from each incubation flask. The nitrogen (purity 99.99%) was passed into the incubation flask from the bottom at the speed of 200mL/min. Methane was carried by nitrogen into an AGM10 sensor (Sensors Europe GmbH, Erkrath, FRG) and the concentration of methane was measured and recorded every 6min [26]. The fermentation was terminated by placing flasks on ice. After opening the incubation flask, pH was measured using a PHS-3C pH meter (Shanghai INESA Scientific Instrument Co., Ltd., China), and 2mL of incubation medium was collected for NH3-N [44]. Another 1 mL of incubation medium was analyzed for volatile fatty acids (VFAs), including acetic acid (AA), propionic acid (PA), and butyric acid (BA) using gas chromatography (Agilent Technologies 7890A GC System, USA) and the method of Castro-Montoya et al. [45]. The left fluid was dried in a forced-air oven at 60°C for 72h and placed in sealed containers in order to analyze the in vitro dry matter digestibility (IVDMD) [46].

Experimental feeds and chemical analyses

Collected in plastic bags, the samples for diets were reserved at -20°C. After the feeding experiment, the samples were warmed at 65°C to a fixed weight. Thereafter, a 0.45mm sieve and a high-speed universal pulverizer were used to grind them for analysis. The filter bag technique of ANKOM A200 (AOAC 973.18) was adopted to analyze neutral detergent fiber (NDF), acid detergent lignin (ADL) as well as acid detergent fiber (ADF). A Kjeltec 8400 analyzer unit (Foss, Sweden) was applied to measure the content of crude protein (CP) on the basis of the Kjeldahl method (AOAC 984.13). In addition, a Soxhlet apparatus was used to measure the content of ether extract (EE) based on Soxhlet extraction method (AOAC 920.85). Methods of Horwitz et al. [46] were the foundation of all chemical analyses.

Data analysis

Sequences of good quality were deeply studied through its uploading to QIIME [39]. A comparison was made between valid bacterial sequences and sequences present in the Silva Database which classified the abundance calculation of each taxon with the optimal choice of classification [36]. QIIME filed the sequence length. Mothur was used for the generation of abundance and diversity indexes. After the implementation of a pseudo-single relevancy algorithm, there was 97% of OTUs identity cutoff [4748]. For all the parameters, data was analyzed by the R-Studio software (version 7.2). Methane production was up to the approach of Sun et al. [26]. A one-way analysis of variance (ANOVA) was carried out late in each bioassay to compare selected taxonomic groups (abundant phyla or genera), bacterial community indices observed OTUs or methane production indices. Duncan’s test was adopted to perform the mean comparison at the significance level of P < 0.05. Redundancy analysis (RDA) was conducted to assess the association between the nutrients in the feed and the bacterial abundance in the rumen. The relationship among bacterial abundance, methane production, peak concentration (Cmax) and the time to peak concentration (Tmax) was assessed by means of Pearson’s correlations. All the data was presented as means ± S.E. (standard error).


Relative quantification of total bacteria, methanogens, protozoa and anaerobic fungi

Firstly, the results of real-time PCR (Table 3) showed that the numbers of methanogens and protozoa were increased with the decreasing F:C but not significantly. Conversely, the numbers of total bacteria and anaerobic fungi were decreased with the decreasing F:C. And the difference of total bacteria in three groups was significant (P = 0.017), while the difference of anaerobic fungi was not. These results exhibited that the change of F:C has extremely effect only on the number of total bacteria but not on the other kinds of microorganism.

Table 3. Relative quantification of total bacteria, methanogens, protozoa and anaerobic fungi with different forage-to-concentrate ratios.

Analysis of DNA sequence data

After quality was controlled preliminarily, 517,492 paired-end 440-bp reads were obtained in total. Each sample got 28,750 sequences averagely. Reads had an overall length of 2.28 gigabases (GB), and each sample had a mean read length of 0.13GB with 191,537, 171,125, and 154,794 raw reads in L, M and H groups respectively (Table 4). Based on 97% species similarity, 132,987, 104,640 and 92,194 OTUs were separately obtained from the samples in L, M and H groups (Table 4). Among all the samples, 708 OTUs were identified, of which 542 existing in all the groups were known as key OTUs (Fig 1A). Key OTUs took up about 76.6% of all OTUs, whereas 6, 5 and 11 OTUs were individually identified in groups L, M and H respectively. Good’s coverage was 99.4%, 99.3% as well as 99.2% for L, M and H groups separately, indicating the capture of dominant phylotypes by this study. The three groups were similar in diversity (Fig 2A). The richness (P<0.01) of the rumen microbiota was related to F:C (Fig 2B).

Table 4. Raw reads and OTUs with different forage-to-concentrate ratios.

Fig 1. Venn diagrams of OTUs and genera with different forage-to-concentrate ratios.

OTUs, operational taxonomic units; L, forage-to-concentrate ratio, 50:50; M, forage-to-concentrate ratio, 70:30; H, forage-to-concentrate ratio, 90:10. The number of observed OTUs sharing ≥ 97% nucleotide sequence identity is shown. (1A) Venn diagram showing the common and unique OTUs among the three groups. (1B) Venn diagram showing the common and unique genera among the three groups.

Fig 2. Chaos and Shannon indexes of OTUs with different forage-to-concentrate ratios.

OTUs, operational taxonomic units; L, forage-to-concentrate ratio, 50:50; M, forage-to-concentrate ratio, 70:30; H, forage-to-concentrate ratio, 90:10. (2A) Bacterial diversity as determined from the Shannon index. (2B) Bacterial richness as reflected in the Chao index. **P < 0.01.

Bacterial community structure at the levels of phylum and genus

According to the results in Fig 3, DNA sequences were distributed in different phyla. The three groups shared 14 phyla, namely Bacteroidetes, Actinobacteria, Cyanobacteria, Chloroflexi, Elusimicrobia, Synergistetes, Fibrobacteres, Firmicutes, Lentisphaerae, Verrucomicrobia, Proteobacteria, Saccharibacteria, Spirochaetes and Tenericutes. As the main components of the 14 phyla (P<0.01) in spite of the diet, Bacteroidetes and Firmicutes occupied over 90% of all sequences. The three groups showed differences in the bacterial richness of different phyla. The remarkable differences of bacterial richness in five out of 14 phyla were discovered in the three groups (Table 5). As the dominant phylum in group L, Bacteroidetes (P<0.01) accounted for about 66.14% of the sequences. Groups M and H assigned a lower percentage (60.05% and 56.80%) of the sequences to Bacteroidetes. Ranking the second as a phylum in all the groups, Firmicutes (P<0.01) comprised roughly of 24.13%, 27.45% and 32.01% sequences in the L, M and H groups respectively. The proportion of Firmicutes increased with the increase of the ratio. Moreover, the richness of Proteobacteria, Spirochaetes and Synergistetes changed with F:C (Table 5). With the increase of the ratio, the proportion of Proteobacteria and Spirochaetes (Table 5) decreased (P<0.01) and the proportion of Synergistetes (P < 0.01) (Table 5) increased.

Fig 3. Phyla distribution of rumen florawith different forage-to-concentrate ratios.

L, forage-to-concentrate ratio, 50:50; M, forage-to-concentrate ratio, 70:30; H, forage-to-concentrate ratio, 90:10.

Table 5. Relative abundance of five distinct phyla (%) and Pearson’s correlations with different forage-to-concentrate ratios.

At the level of genus, the identification of 150 genera in all the samples was conducted despite F:C (Fig 3). L, M and H groups had 141, 146 and 141 genera respectively and shared 130 genera, whereas Coriobacteriaceae and Gemella were special for group L (Fig 1B). 15 richest genera, comprising over 78.96% of all sequences, included Prevotella, Ruminococcus, Lachnospira, Rikenella, Succiniclasticum, Fibrobacter, Christensenella, Saccharofermentans, Eubacterium, Papillibacter, Quinella, Phocaeicola, Verllonella, Moryella and Fretibacterium. The bacterial richness of 22 genera varied with the diet ratio. The abundance of 14 bacteria increased, whereas that of eight bacteria decreased (Table 6). Significantly different among all the groups, Lachnospira, Fibrobacter and Clostridium were not linearly related to the diet ratio. Among the linearly changed genera, Prevotella was the predominant genus, accounting for 50.79%, 43.06% and 34.09% of the total sequences in L, M and H groups respectively.

Table 6. Relative abundance of 25 distinct genera (%) and Pearson’s correlations with different forage-to-concentrate ratios.

Nutrition index in rumen and its correlation with the rumen microbiota

In terms of the RDA, our dataset changed, which was principally interpreted by the increasing F:C (Fig 4). It suggested that 100% change in bacteria was explained by all the nutrition indices whose order of contribution was CP > ADF > NDF > Starch > EE > ADL (Table 7). The two sorting axes accounted for 95.48% of the changes based on this model with the first sorting axis explaining a change of 66.37% and 29.11% for the second sorting axis. The rumen microbiota in group L was concentrated in the regions with high CP, starch content and low NDF and ADF contents, whereas the rumen microbiota in group H was concentrated in the regions with high NDF and ADF contents and low CP and starch contents. The rumen microbiota in group M was concentrated in the regions with intermediate nutrient levels. According to the RDA analysis, the relevance of CP accounted for 0.72 of the microbiota (P<0.01) as the main nutrient factor affecting the structure of microbiota. Insignificant, the relevance of EE and ADL was the lowest (R2 = 0.41, P>0.05; R2 = 0.36, P>0.05) and they were not significant. Under the different levels of F:C, the different kinds of bacteria bacterial community were established, which could reflect the state of the microflora in the rumen fluid at the beginning of in vitro fermentation.

Fig 4. Redundancy analysis of nutrition index and the rumen microbiota with different forage-to-concentrate ratios.

L, forage-to-concentrate ratio, 50:50; M, forage-to-concentrate ratio, 70:30; H, forage-to-concentrate ratio, 90:10. CP, crude protein; ADF, acid detergent fiber; NDF, neutral detergent fiber. Two sorting axes accounted for 95.48% of the changes with the first sorting axis explaining a change of 66.37% and 29.11% for the second sorting axis.

Table 7. Relevance of nutrition indices in the redundancy analysis a.

In vitro rumen fermentation characteristics, real-time methane production and its correlation with the rumen microbiota

After 12 h fermentation, the concentrations of pH, AA and A/P were decreased greatly with the decreasing F:C. Simultaneously, the concentrations of PA, BA, NH3-N, and IVDMD were increased with the decreasing F:C. The greatly growth of IVDMD had led to the massive production of VFAs (Table 8). The Cmax (P<0.01, Table 9, Fig 5) and total production (P<0.05, Table 9) of methane decreased with the increase in F:C, whereas Tmax (P<0.01, Table 9, Fig 5) of methane increased with the increase of F:C. At the level of phylum, Bacteroidetes, Proteobacteria as well as Spirochaetae showed a positive correlation with the Cmax and total production, and a negative correlation with Tmax (Table 10). Firmicutes and Saccharibacteria were positively correlated with Tmax, but negatively correlated with Cmax and total production (Table 10). At the level of genus, Prevotella, Quinella, Verllonella, Ruminobacter, Oribacterium, Succinivibrio, Syntrophococcus and Olsenella was positively correlated with Cmax and total production in bacterial abundance, but negatively correlated with Tmax (Table 11). Ruminococcus, Rikenella, Succiniclasticum, Eubacterium, Papillibacter, Pseudobutyrivibrio, Butyrivibrio, Candidatus, Anaerotruncus and Ruminiclostridium was positively correlated with Tmax in bacterial abundance, but negatively correlated with Cmax and total production (Table 11).

Table 8. Effect of different forage-to-concentrate ratios on in vitro fermentation.

Table 9. Cmax, Tmax and total production of methane in vitro and Pearson’s correlations with different forage-to-concentrate ratios.

Table 10. Pearson’s correlations between five distinct phyla and in vitro Cmax, Tmax and total production a of methane with different forage-to-concentrate ratios.

Table 11. Pearson’s correlations between 25 distinct genera and in vitro Cmax, Tmax and total production a of methane with different forage-to-concentrate ratios.

Fig 5. Methane production curve in vitro with different forage-to-concentrate ratios.

L, forage-to-concentrate ratio, 50:50; M, forage-to-concentrate ratio, 70:30; H, forage-to-concentrate ratio, 90:10. SEM, standard error of mean; Total gas, fermentation gas and carrier gas.


The rumen bacterial community structure was changed significantly with the different levels of F:C, but the numbers of methanogens, protozoa and anaerobic fungi were not changed obviously. The in vitro fermentation has been changed significantly with the change of F:C. And a close relationship was found between the real-time methane production and bacterial community structure. It seems to provide a new understanding of rumen fermentation model and methane production adjustment.

Firstly, the results of real-time PCR showed that the number of methanogens, protozoa and anaerobic fungi, under the change of F:C, was changed significantly except bacterial. Archaea was considered the producer of methane. But with the change of F:C in this study, the quantity of archaea was stable. It showed no significant linear relationship between the structure of archaea and methane production, which is similar to the research of Lengowski et al. [25]. The relationship between archaea and methane production has been discussed in many studies. Moreover, the main raw materials for methane synthesis, such as hydrogen, carbon dioxide and VFAs, were produced by bacteria [49]. Methane synthesis was a passive behavior of archaea to maintain rumen pressure and pH balance in the case of too high ratios of bacterial synthesis [5051]. Therefore, methane production was more probably related to the concentration of synthetic raw materials in the rumen and the bacteria producing these materials.

The second goal of this experiment was to explore changes in the genus level of the rumen flora with different F:C. With the increase of the ratio, the proportion of different genera showed significant differences, revealing the effectiveness of experimental gradient design. In this study, the proportion of Prevotella showed a linearly increasing trend with the increase of protein levels in diets, which was consistent with the results of Xu and Gordon [52]. As a genus, Prevotella has many functions, mainly including promoting protein degradation and assisting other strains in enhancing the utilization of fiber materials in ruminants [53]. Ruminococcus, a cellulolytic bacterium [54], increased with the increasing fiber. Succinivibrio, Ruminobacter, amylophilus and Selenomonas were starch-degrading bacteria that could produce acetic acid and succinic acid during starch degradation [55]. Succinic acid was eventually transformed to PA [56] to provide energy for microbial protein synthesis in the rumen. With the decrease of starch content in the diets, the proportion of these three genera decreased significantly in this experiment, indicating the change of carbohydrate fermentation substrate from a non-structural carbohydrate to a structural carbohydrate. Butyrivibrio and Pseudobutyrivibrio were carbohydrate-degrading bacteria producing butyric acid [57]. In this experiment, the proportion of Butyrivibrio and Pseudobutyrivibrio decreased linearly with the increase of starch, but increased linearly with the increase of NDF and ADF in the diets, which showed that Butyrivibrio and Pseudobutyrivibrio were more likely to produce energy by using structural carbohydrates. The proportion of Eubacterium with the function of degrading structural carbohydrates was similar to that of Butyrivibrio [58].

In this study, the third aim was to gain a preliminary understanding of the relationship between nutrition levels and the diversity and richness of rumen microbiota in sheep under various F:C. In this experiment, CP was the most important nutrient factor contributing to the change in bacterial diversity. Bodine and Purvis [59] found that the effect of supplementation of non-structural carbohydrate is largely determined by the level of protein in the diet. Adding protein to the diet can improve the balance of energy and nitrogen and increase digestibility. CP could provide nitrogen resource for the self-replication and enzyme synthesis of bacteria [60]. ADF, NDF and starch were important nutrient factors providing carbon resource for self-replication and energy. However, starch showed a negative correlation with bacterial diversity compared to ADF and NDF because of its easier decomposition as a non-structural carbohydrate. According to the studies of Kononoff and Heinrichs [61] and Drackley et al. [62], the rumen fermentation was mainly in the AA-mode when NDF and ADF contents in the diet were high and mainly in the PA-mode when starch content in the diet was high. The Chao index of OTUs increased with the increase in F:C, showing more strains of bacteria were required by the degradation of NDF and ADF to cooperate than those required by the degradation of CP and starch. These were the changes of microbiota in the rumen. EE and ADL in the diets had no significant effects on the changes in the rumen microbiota. Jenkins [63] found that only about 8% of fat in the rumen was degraded. There might be two reasons: The designed levels of EE and ADL content in the diets were too close to result in the similarity of the microbial community or these nutrients were not main energy resources for bacterial activity in the rumen so that bacteria were not sensitive to the low levels of EE [64]. Based on the above results, three kinds of rumen bacterial community were proved to be successfully established.

In this study, the final goal was to preliminarily understand the relationship between methane production and the rich and diversity of rumen microbiota in sheep under various F:C. In the anaerobic environment of the rumen, a variety of organic compounds could eventually be transformed to methane through decomposed by a number of microorganisms [5]. Leng and Nolan [14] showed that 80% of the nitrogen available to ruminal bacteria came from ammonia and 20% from amino acids or oligopeptides. With the increase of F:C, Cmax of methane was delayed. For the diet with higher CP and starch contents, methane production could reach Cmax more quickly, showing a significant correlation with the rumen microbiota. With lower CP content in diet, bacteria required more time for protein decomposition to provide materials for their reproduction and methane synthesis, which indicated that methane synthesis needed to go through a “start-up” phase before the normal fermentation mode. Previous studies only found that Cmax of methane occurred at about 2 h after food intake [6566]. The missing of the delay phenomenon could be attributed to the insufficient frequency of detection. On the other hand, both Cmax and total production of methane with lower CP and starch contents in the diet during fermentation were less than those in the diets with lower NDF and ADF contents. It was probably because related bacteria like Prevotella and Butyrivibrio could decompose nitrogen compounds to provide sufficient raw materials for the reproduction and synthesis of methanogenic archaea [67]. As indicated by the correlation analysis, fiber-degrading bacteria were positively correlated with Tmax of methane, but negatively correlated with Cmax and total production of methane. Compared with fiber-degrading bacteria, starch-degrading and protein-degrading bacteria showed an opposite correlation. More readily available nitrogen and degradable carbohydrates could be preferentially used by microorganisms [68], providing more effective support for methane synthesis. Furthermore, Cmax and Tmax of methane could be effective parameters for predicting the type of rumen fermentation, which however remained to be confirmed by further research.

Based on nutritional parameters, a new model of methane prediction with a wider range of applications is being developed in accordance with the results of the present study. The genera of bacterial, as the parameters for the prediction models, had been narrowed down. There were significant correlations between specific bacterial at the starting of fermentation and real-time methane production in vitro. However, the dynamic changes of bacterial at the moment such as Tmax during the fermentation need to be explored in the following study. Additionally, the fermentation in vivo was more complex. For instance, nitrogen cycling in ruminants might provide bacteria with initial nitrogenous material [69]. Thus, further studies are required to confirm the occurrence of this delay phenomenon in vivo and illustrate the process.


With the change in F:C, bacterial community structure and methane production in the rumen showed significant changes. Crude protein was the most important nutrient factor that contributed to the change in bacterial diversity. Among the 150 genera identified in the rumen, the abundance of 22 varied linearly with F:C. These genera would be further screened to serve as effective parameters for the methane prediction model. In addition, during the 12 h in vitro fermentation, as F:C increased, the Cmax and total production of methane decreased significantly, and Tmax was delayed by 26–27 min. The fiber-degrading bacteria were positively correlated with this phenomenon, but starch-degrading and protein-degrading bacteria were negatively correlated with it.

Supporting information


We are grateful to the technical staff of the Animal Nutrition and Feed Science Laboratory at Jilin Agricultural University (Changchun, China) for their help in this work. The in vitro fermentation system was provided by Animal Nutrition Institute of Jilin Academy of Agricultural Sciences (Changchun, China). The work on high-throughput sequencing was outsourced to Shanghai Majorbio Technology Co., Ltd.


  1. 1. NCCC. National coordination committee on climate change. Second national communication on climate change of the people’s republic of china. Beijing: China planning press; 2017.
  2. 2. Bennetzen EH, Smith P, Porter JR. Decoupling of greenhouse gas emissions from global agricultural production: 1970–2050. Global Change Biol. 2016; 22: 763–781. pmid:26451699
  3. 3. Wolfe RS. Unusual coenzymes of methanogenesis. Annu. Rev. Biochem. 1990; 10(10): 396–399.
  4. 4. Hobson PN, Stewart CS. The rumen microbial ecosystem. London: Springer Science & Business Media; 2012.
  5. 5. Wolin MJ. The Rumen Fermentation: A model for microbial interactions in anaerobic ecosystems. Adv. Microb. Ecol. 1979; 3: 49–77.
  6. 6. Vlková E, Trojanová I, Rada V. Distribution of Bifidobacteria in the gastrointestinal tract of calves. Folia Microbiol. 2006; 51(4): 325–328.
  7. 7. Lee HJ, Jung JY, Oh YK, Lee SS, Madsen EL, Jeon CO. Comparative survey of rumen microbial communities and metabolites across one caprine and three bovine groups, using bar-coded pyrosequencing and ¹H nuclear magnetic resonance spectroscopy. Appl. Environ. Microb. 2012; 78: 5983–5993. pmid:22706048
  8. 8. Jiao J, Li X, Beauchemin KA, Tan Z, Tang S, Zhou C. Rumen development process in goats as affected by supplemental feeding v. grazing: age-related anatomic development, functional achievement and microbial colonisation. Br. J. Nutr. 2015; 113(6): 888–900. pmid:25716279
  9. 9. de Menezes AB, Lewis E, O’Donovan M, O’Neill BF, Clipson N, Doyle EM. Microbiome analysis of dairy cows fed pasture or total mixed ration diets. FEMS Microbiol. Ecol. 2011, 78, (2), 256–65. pmid:21671962
  10. 10. Carberry CA, Kenny DA, Han S, McCabe MS, Waters SM. The effect of phenotypic residual feed intake (RFI) and dietary forage content on the rumen microbial community of beef cattle. Appl. Environ. Microbiol. 2012; 78(14): 4949–4958. pmid:22562991
  11. 11. Petri RM, Schwaiger T, Penner GB, Beauchemin KA, Forster RJ, McKinnon JJ, et al. Characterization of the core rumen microbiome in cattle during transition from forage to concentrate as well as during and after an acidotic challenge. PLoS One 2014; 8(12): e83424. pmid:24391765
  12. 12. Clark JH. Lactational responses to postruminal administration of proteins and amino acids. J. Dairy Sci. 1975; 58(8): 1178–1197. pmid:1099124
  13. 13. Cunha IS, Barreto CC, Costa OY, Bomfim MA, Castro AP, Kruger RH, et al. Bacteria and archaea community structure in the rumen microbiome of goats (Capra hircus) from the semiarid region of Brazil. Anaerobe 2011; 17(3): 118–124. pmid:21575735
  14. 14. Leng RA, Nolan JV. Nitrogen metabolism in the rumen. J. Dairy Sci. 1984; 67: 1072–1089. pmid:6376562
  15. 15. Grovum WL, Leek BF. Parotid secretion and associated efferent activity inhibited by pentagastrin in sheep. Peptides 1988; 9(3): 519–526. pmid:3420011
  16. 16. Moe PW, Tyrrell HF. Methane Production in Dairy Cows. J. Dairy Sci. 1979; 62(10): 1583–1586.
  17. 17. Holter JB, Young AJ. Methane prediction in dry and lactating Holstein cows. J. Dairy Sci. 1992; 75(8): 2165–2175. pmid:1401368
  18. 18. Mills JA, Kebreab E, Yates CM, Crompton LA, Cammell SB, Dhanoa MS, et al. Alternative approaches to predicting methane emissions from dairy cows. J. Anim. Sci. 2003; 81(12): 3141–3150. pmid:14677870
  19. 19. Ellis JL, Kebreab E, Odongo NE, Mcbride BW, Okine EK, France J. Prediction of methane production dairy and beef cattle. J. Dairy Sci. 2007; 90: 3456–3466. pmid:17582129
  20. 20. Jentsch W, Schweigel M, Weissbach F, Scholze H, Pitroff W, Derno M. Methane production in cattle calculated by the nutrient composition of the diet. Arch. Anim. Nutr. 2007; 61(1): 10–19. pmid:17361944
  21. 21. Dong RL, Zhao GY. Relationship between the methane production and the CNCPS carbohydrate fractions of rations with various concentrate / roughage ratios evaluated using in vitro incubation technique. Asian Austral. J. Anim. 2013; 26(12): 1708–1716. pmid:25049761
  22. 22. Dijkstra J, Neal HD, Beever DE, France J. Simulation of nutrient digestion, absorption and outflow in the rumen: model description. J. Nutr. 1992; 122(11): 2239–2256. pmid:1331382
  23. 23. Mills JA, Dijkstra J, Bannink A, Cammell SB, Kebreab E, France J. A mechanistic model of whole-tract digestion and methanogenesis in the lactating dairy cow: Model development, evaluation, and application. J. Anim. Sci. 2001; 79: 1584–1597. pmid:11424698
  24. 24. Lange M, Ahring BK. A comprehensive study into the molecular methodology and molecular biology of methanogenic archaea. FEMS Microbiol. Rev. 2001; 25(5): 553–571. pmid:11742691
  25. 25. Lengowski MB, Witzig M, Möhring J, Seyfang GM, Rodehutscord M. Effects of corn silage and grass silage in ruminant rations on diurnal changes of microbial populations in the rumen of dairy cows. Anaerobe 2016; 42: 6–16. pmid:27451293
  26. 26. Sun YK, Yan XG, Ban ZB, Yang HM, Hegarty RS, Zhao YM. The effect of cysteamine hydrochloride and nitrate supplementation on in-vitro and in-vivo methane production and productivity of cattle. Anim. Feed Sci. Tech. 2017; 232: 49–56.
  27. 27. NRC. Nutrient requirements of small ruminants. Washington, DC: National Academies Press; 2007.
  28. 28. Murray MG, Thompson WF. Rapid isolation of high molecular weight plant DNA. Nucleic Acids Res. 1980; 8(19): 4321–4326. pmid:7433111
  29. 29. Zhou JZ, Bruns MA, Tiedje JM. 1996. DNA recovery from soils of diverse composition. Appl. Environ. Microb. 1996; 62(2): 316–322.
  30. 30. Rathgeber C, Yurkova N, Stackebrandt E, Schumann P, Humphrey E, Beatty JT, et al. Porphyrobacter meromictius sp. nov., an appendaged bacterium, that produces Bacteriochlorophyll a. Current Microbiol. 2007; 55: 356–361. Available from: pmid:17882507
  31. 31. Denman SE, McSweeney CS. Development of a real-time PCR assay for monitoring fungal and cellulolytic bacterial populations within the rumen. FEMS Microbiol. Ecol. 2006; 58(3): 572–582. pmid:17117998
  32. 32. Denman SE, Tomkins NW, McSweeney CS. Quantitation and diversity analysis of ruminal methanogenic populations in response to the antimethanogenic compound bromochloromethane. FEMS Microbiol. Ecol. 2007; 62(3): 313–322. pmid:17949432
  33. 33. Sylvester JT, Karnati SKR, Yu Z, Morrison M, Firkins JL. Development of an assay to quantify rumen ciliate protozoal biomass in cows using real-time PCR. J. Nutr. 2004; 134(12): 3378–3384. pmid:15570040
  34. 34. Meng H, Zhang Y, Zhao LL, Zhao WJ, He C, Honaker CF, et al. Body weight selection affects quantitative genetic correlated responses in gut microbiota. PloS One 2014; 9: e89862. pmid:24608294
  35. 35. Youssef N, Sheik CS, Krumholz LR, Najar FZ, Roe BA, Elshahed MS. Comparison of species richness estimates obtained using nearly complete fragments and simulated pyrosequencing-generated fragments in 16S rRNA gene-based environmental surveys. Appl. Environ. Microb. 2009; 75: 5227–5236. pmid:19561178
  36. 36. Lan YM, Wang Q, Cole JR, Rosen GL. Using the RDP classifier to predict taxonomic novelty and reduce the search space for finding novel organisms. PloS One 2012; 7: e32491. pmid:22403664
  37. 37. Fan WG, Tang YR, Qu Y, Cao FB, Huo GC. Infant formula supplemented with low protein and high carbohydrate alters the intestinal microbiota in neonatal SD rats. BMC Microbiol. 2014; 14: 279. pmid:25403909
  38. 38. Magoä T, Salzberg SL. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics 2011; 27: 2957–2963. pmid:21903629
  39. 39. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME allows analysis of high-throughput community. Nat. Methods 2010; 7(5): 335–336. pmid:20383131
  40. 40. Bokulich NA, Subramanian S, Faith JJ, Gevers D, Gordon JI, Knight R, et al. Quality-filtering vastly improves diversity estimates from Illumina amplicon sequencing. Nat. Methods 2013; 10: 57–59. pmid:23202435
  41. 41. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 2010; 26: 2460–2461. pmid:20709691
  42. 42. Niu Q, Li P, Hao S, Zhang Y, Kim SW, Li H, et al. Dynamic distribution of the gut microbiota and the relationship with apparent crude fiber digestibility and growth stages in pigs. Sci. Rep.-UK 2015; 5: 9938. pmid:25898122
  43. 43. Menke K, Steingass H. Estimation of the energetic feed value obtained from chemical analysis and in vitro gas production using rumen fluid. Anim. Res. Dev. 1988; 28(1): 7–55.
  44. 44. Preston TR. Tropical animal feeding: a manual for research workers. In: Animal production and health. Rome: Food and Agriculture Organization; 1998. pp. 126.
  45. 45. Castro-Montoya J, Campeneere SD, Ranst GV, Fievez V. Interactions between methane mitigation additives and basal substrates on in vitro methane and VFA production. Anim. Feed Sci. Technol. 2012; 176(1–4): 47–60.
  46. 46. Horwitz W. Official methods of analysis of AOAC international. Maryland: AOAC International Publishing; 2006.
  47. 47. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl. Environ. Microb. 2009; 75: 7537–7541. pmid:19801464
  48. 48. Kim HB, Borewicz K, White BA, Singer RS, Sreevatsan S, Tu ZJ. Microbial shifts in the swine distal gut in response to the treatment with antimicrobial growth promoter, tylosin. P. Natl. Acad. Sci. 2012; 109: 15485–15490. pmid:22955886
  49. 49. Akin DE, Benner R. Degradation of polysaccharides and lignin by ruminal bacterial and fungi. Appl. Environ. Microb. 1988; 54(5): 1117–1125.
  50. 50. Thauer RK, Kaster A, Seedof H, Buckel W, Hedderich R. Methanogenic archaea: ecologically relevant differences in energy conservation. Nat. Rev. Microbiol. 2008; 6(8): 579–591. pmid:18587410
  51. 51. Sakai S, Imachi H, Sekiguchi Y, Tseng L, Ohashi A, Harada H, et al. Cultivation of methanogens under low-hydrogen conditions by using the coculture method. Appl. Environ. Microb. 2009; 75(14): 4892–4896. pmid:19465530
  52. 52. Xu J, Gordon JI. Honor thy symbionts. Proc. Natl. Acad. Sci. 2003; 100: 10452–10459. pmid:12923294
  53. 53. Petri RM, Schwaiger T, Penner GB, Beauchemin KA, Forster RJ, Mckinnon JJ, et al. Changes in the rumen epimural bacterial diversity of beef cattle as affected by eiet and induced ruminal acidosis. Appl. Environ. Microb. 2013; 79: 3744–3755. pmid:23584771
  54. 54. Grinberg IR, Yin GH, Borovok I, Miller MEB, Yeoman GJ, Dassa B, et al. Functional phylotyping approach for assessing intraspecific diversity of ruminococcus albus within the rumen microbiome. FEMS Microbiol. Lett. 2014; 362(3): 1. pmid:25673657
  55. 55. Sun B, Wang X, Bernstein S, Huffman MA, Xia DP, Gu Z, et al. Marked variation between winter and spring gut microbiota in free-ranging Tibetan Macaques (Macaca thibetana). Sci. Rep. 2016; 6: 26035. pmid:27180722
  56. 56. Li XZ, Park BK, Shin JS, Choi SH, Smith SB, Yan CG. Effects of dietary linseed oil and propionate precursors on ruminal microbial community, composition, and diversity in yanbian yellow cattle. PloS One 2015; 10: e0126473. pmid:26024491
  57. 57. Gill JW, King KW. Nutritional characteristics of a Butyrivibrio. J. Bacteriol. 1958; 75: 666–673. Available from: pmid:13549370
  58. 58. Taguchi H, Koike S, Kobayashi Y, Cann IKO, Karita S. Partial characterization of structure and function of a xylanase gene from the rumen hemicellulolytic bacterium Eubacterium ruminantium. Anim. Sci. J. 2015; 75: 325–332.
  59. 59. Bodine TN, Purvis HT. Effects of supplemental energy and/or degradable intake protein on performance, grazing behavior, intake, digestibility, and fecal and blood indices by beef steers grazed on dormant native tallgrass prairie. J. Anim. Sci. 2003; 81(1): 304–317. pmid:12597402
  60. 60. Russell JB. 1983. Fermentation of peptides by Bacteroides Ruminicola B14. Appl. Environ. Microb. 45, 1566–1574. Available from:
  61. 61. Kononoff PJ, Heinrichs AJ. The effect of corn silage particle size and cottonseed hulls on cows in early lactation. J. Dairy Sci. 2003; 86: 2438–2451. pmid:12906062
  62. 62. Drackley JK, Beaulieu AD, Elliott JP. Responses of milk fat composition to dietary fat or nonstructural carbohydrates in Holstein and Jersey cows. J. Dairy Sci. 2001; 84: 1231–1237. pmid:11384050
  63. 63. Jenkins TC. Lipid metabolism in the rumen. Prog. Lipid Res. 1993; (76): 3851–3863.
  64. 64. Knapp DM, Grummer RR, Dentine MR. The response of lactating dairy cows to increasing levels of whole roasted soybeans. J. Dairy Sci. 1991; 74: 2563–2572. pmid:1918534
  65. 65. Kinsman R, Sauer FD, Jackson HA, Wolynetz MS. Methane and carbon dioxide emissions from dairy cows in full lactation monitored over a six-month period. J. Dairy Sci. 1995; 78: 2760–2766. pmid:8675759
  66. 66. Van Haarlem RP, Desjardins RL, Gao Z, Flesch TK, Li X. 2008. Methane and ammonia emissions from a beef feedlot in western Canada for a twelve-day period in the fall. Can. J. Anim. Sci. 88, 641–649.
  67. 67. Zhou M, Mcallister TA, Guan LL. Molecular identification of rumen methanogens: Technologies, advances and prospects. Anim. Feed Sci. Tech. 2011; 166–167: 76–86.
  68. 68. Mould FL, Mann RO. Associative effects of mixed feeds. I. Effects of type and level of supplementation and the influence of the rumen fluid pH on cellulolysis in vivo and dry matter digestion of various roughages. Anim. Feed Sci. Tech. 1983; 10: 15–30.
  69. 69. Russell JB, O’Connor JD, Fox DG, Van Soest PJ, Sniffen CJ. A net carbohydrate and protein system for evaluating cattle diets: I. Ruminal fermentation. J. Anim. Sci. 1992; 70: 3551–3561. pmid:1459918