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

Diversity and functional analysis of rumen and fecal microbial communities associated with dietary changes in crossbreed dairy cattle

  • Felix M. Kibegwa ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft, Writing – review & editing

    Affiliation Department of Animal Production, Faculty of Veterinary Medicine, University of Nairobi, Nairobi, Kenya

  • Rawlynce C. Bett,

    Roles Conceptualization, Formal analysis, Methodology, Project administration, Supervision, Writing – review & editing

    Affiliation Department of Animal Production, Faculty of Veterinary Medicine, University of Nairobi, Nairobi, Kenya

  • Charles K. Gachuiri,

    Roles Conceptualization, Investigation, Methodology, Project administration, Supervision, Writing – review & editing

    Affiliation Department of Animal Production, Faculty of Veterinary Medicine, University of Nairobi, Nairobi, Kenya

  • Eunice Machuka,

    Roles Formal analysis, Methodology, Visualization, Writing – review & editing

    Affiliation Biosciences Eastern and Central Africa—International Livestock Research Institute (BecA-ILRI) Hub, Nairobi, Kenya

  • Francesca Stomeo,

    Roles Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Writing – review & editing

    Current address: Max Planck Institute for Plant Breeding Research, Cologne, Germany

    Affiliation Biosciences Eastern and Central Africa—International Livestock Research Institute (BecA-ILRI) Hub, Nairobi, Kenya

  • Fidalis D. Mujibi

    Roles Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Software, Supervision, Writing – review & editing

    Affiliation Genomics Division, USOMI Limited, Nairobi, Kenya


The objective of this study was to investigate the effect of varying roughage and concentrate proportions, in diet of crossbreed dairy cattle, on the composition and associated functional genes of rumen and fecal microbiota. We also explored fecal samples as a proxy for rumen liquor samples. Six crossbred dairy cattle were reared on three diets with an increasing concentrate and reducing roughage amount in three consecutive 10-day periods. After each period, individual rumen liquor and fecal samples were collected and analyzed through shotgun metagenomic sequencing. Average relative abundance of identified Operational Taxonomic Units (OTU) and microbial functional roles from all animals were compared between diets and sample types (fecal and rumen liquor). Results indicated that dietary modifications significantly affected several rumen and fecal microbial OTUs. In the rumen, an increase in dietary concentrate resulted in an upsurge in the abundance of Proteobacteria, while reducing the proportions of Bacteroidetes and Firmicutes. Conversely, changes in microbial composition in fecal samples were not consistent with dietary modification patterns. Microbial functional pathway classification identified that carbohydrate metabolism and protein metabolism pathways dominated microbial roles. Assessment of dietary effects on the predicted functional roles of these microbiota revealed that a high amount of dietary concentrate resulted in an increase in central carbohydrate metabolism and a corresponding reduction in protein synthesis. Moreover, we identified several microbial stress-related responses linked to dietary changes. Bacteroides and Clostridium genera were the principal hosts of these microbial functions. Therefore, the roughage to concentrate proportion has more influence on the microbial composition and microbial functional genes in rumen samples than fecal samples. As such, we did not establish a significant relationship between the rumen and fecal metagenome profiles, and the rumen and fecal microbiota from one animal did not correlate more than those from different animals.


Demand for animal source foods is rapidly steadily rising: for example, milk demand in low-income countries is expected to increase by 136% by 2030 compared to 2000 [1]. This increase in demand has mostly been ascribed to population expansion, increased urbanization, and income growth [2]. To meet this huge demand, developing countries must significantly increase livestock production [2, 3], through simultaneous interventions in animal feeds, genetics, and health [4]. In East Africa, many interventions to improve milk production have had minimal gains because farmers depend mainly on rain-fed forage, pastures/cultivated fodder and crop by-products. Additionally, the high cost of conventional feed resources, such as cereals and grain legumes, the food-feed competition between man and livestock, and the absence of suitable technology to optimize the use of these feeds in animal husbandry prohibit their wide-scale use [5]. On the other hand, despite continuous improvement of the cattle genotypes through cross-breeding and upgrading, the proportion of improvement in milk production attributable to this genetic improvement remains relatively low [6]. There is therefore a need for a paradigm shift that incorporates use of new technologies or customization of existing technologies to improve milk production while using the available feed resources and cattle genotypes. Previous studies have reported that animal’s diet (quantity and quality) is closely related to animal production and rumen microbial composition [7, 8]. This is because the rumen microbiota symbiotically degrades forages into nutrients, such as volatile fatty acids and microbial proteins, which are utilized by the host animal [9, 10]. However, few studies have been conducted to evaluate the microbial composition in crossbred cattle reared by dairy farmers in the East African tropics and subjected to constant fluctuations in diet composition.

To facilitate rumen microbial composition evaluation, the standard sample collection method is rumen cannulation of cattle [11]. However, only a small number of ruminally cannulated cattle are accessible to any investigator in a single study, thus restricting the number of cattle that can be used to assess the ruminal microbiome, if only rumen cannulated cows are used. A more conventionally accepted method of collecting rumen contents from non-cannulated cattle is using stomach tubing. A study by Paz et al. [11] showed that samples collected from the same rumen cannulated animals and stomach tubing did not result in a significant difference in the composition of ruminal microbiome. However, this approach cannot be used to repeatedly collect rumen samples from the same cattle within a short period because the cattle are stressed during the sampling [12]. Additionally, the sampling technique requires the use of expensive equipment and well-trained personnel. It is therefore important to explore other potential non-invasive alternative samples/sampling techniques that can act as proxies for assessing the composition of the rumen microbial community. Previously, buccal swab, regurgitated digesta collected from the cow’s mouth (bolus), salivary content and feces have been tried as potential non-invasive sampling techniques for rumen microbial analysis [1315]. However, results from these studies have been inconclusive, leaving this question open for further investigation.

For many years, the structure of rumen microbial community has been evaluated by the culturing of samples using selective media [16, 17]. However, this method is time-consuming and heavily biased by the selected isolation media and methods employed, thus providing an incomplete depiction of the microbial community structure [8, 18]. The advent of next-generation sequencing (NGS) has overcome the limitations of microbial culturing through direct genetic analysis of genomes within an environmental sample, in a field called metagenomics [19]. Metagenomics can be characterized as targeted or unselective (shotgun) metagenomics, based on their directed or random sequencing strategies respectively [20]. Targeted metagenomics focuses on the isolation, extraction and sequencing of amplicons corresponding to entire (or specific portions of) phylogenetic marker genes (e.g., 16S rRNA) or specific genomic regions such as the Internal Transcribed Spacer (ITS) regions (for fungal species) [21]. On the other hand, shotgun metagenomics allows for universal microbe detection regardless of the type (viruses, bacteria, fungi, and parasites) [22]. It provides more detailed information about the microbial taxonomic composition, microbial interaction, and recovery of whole genome sequences. It further provides unlimited access to functional gene composition information derived from microbial communities inhabiting ecosystems [19]. However, shotgun metagenomics is costlier, requires significantly more efforts in sequencing, data processing, and analysis compared with targeted metagenomics [23]. This restricts the number of samples sequenced in most studies [2426]. Accordingly, we hypothesized that given the advancements in shotgun metagenomics, it would be possible to detect even the slightest changes in microbial composition and microbial functional genes across different sections of the cattle gastrointestinal (GI) tract. Therefore, the objective of this study was to determine the effect of varying roughage and concentrate proportions, in diet of crossbreed dairy cattle, on the composition and associated functional genes of rumen and fecal microbiota. Additionally, rumen liquor and fecal metagenome profiles were contrasted to determine if the fecal metagenome was predictive of rumen metagenome.

Materials and methods

Animals and diets

This study was approved and performed in accordance with the guidelines of University of Nairobi’s Faculty of Veterinary Medicine Animal care and use committee (ACUC). Experienced animal health professionals handled animals to minimize discomfort and injury.

Experimental animals were chosen from the dairy herd of the Faculty of Veterinary Medicine Farm at the University of Nairobi (latitude 1° 14’33.4"S, 36° 42’36.3"E). The experimental design was discussed in our previous study [27]. This research used six crossbred lactating cows (300 ± 50 kg body weight; 174 ± 15 days in milk) in first-parity. The animals were crossbreeds resulting from two successive crossing of indigenous Zebu female cattle with pure breed Friesian bulls. Before recruitment into the experiment, the animals were reared in different herds, and grazed on mixed perennial grasses until being transferred to individual cages for the experiment. Given the animal’s diverse dietary background and the need for adaptation to confinement, selected animals were put on a 10-day acclimatization period. During the acclimatization period, all the animals were fed ad libitum mixed diet of Pennisetum clandestinum (Kikuyu grass) and Chloris gayana (Rhodes grass) hay, supplemented with dairy meal (a commercial concentrate), at 0.01/kg body weight. The feeds were divided into two and offered during morning and evening milking. The animals were then fed on three distinct diets in three successive 10-day periods after this 10-day acclimatization period. The experimental diets contained an increased amount of concentrate and a reduced amount of roughage. The roughage ratios to concentrate in the diets were 90:10 (Diet 1), 75:25 (Diet 2), and 60:40 (Diet 3). The three diets were developed using the NRC-Nutrient Requirements of Dairy Cattle Software v 1.9 [28] to meet the energy demands of dairy cows producing 20 kg of milk/day with 4.0% milk fat and 3.5% true protein. For the roughage, the dietary components were Chloris gayana (Rhodes grass), hay, Pennisetum purpureum (Napier grass), Pennisetum clandestinum (Kikuyu grass), and dry Zea mays (maize) Stover. These were mixed with dairy meal and urea in different proportions to make total mixed rations of the three diets, as shown in Table A in S1 Text, and offered to the animals at 8 am and 6 pm. The chemical composition of the dietary ingredients was evaluated according to the Association of Official Analytic Chemist (AOAC) methods [29], while dietary fiber was determined according to a previously described method [30] (Table B in S1 Text). The cattle were housed in individual stalls and had free access to fresh water and mineral lick supplements throughout the 30 days of the experiment.

Sample collection and DNA extraction

A total of 36 (18 fecal and 18 rumen liquor) samples were collected in this study. The samples were obtained from individual animals on the final day of each experimental diet, i.e., day 10 (for diet D1), day 20 (for diet D2), and day 30 (for diet D3). Roughly, 200 g of fecal samples were obtained by rectal grab, and a sub-sample was placed into a sterile 50 mL falcon tube. Rumen liquor samples were collected by a suction pump and a flexible stomach tube as previously described [31]. Approximately 250 mL of total (i.e., solid and liquid) rumen contents were collected, 50 mL of the rumen liquor was placed in sterile 50 mL falcon tubes after discarding the first 200 mL of sample to reduce saliva contamination. The samples were transported to the Biosciences eastern and central Africa—International Livestock Research Institute, (BecA ILRI) Hub laboratory in Nairobi, and stored at -20°C awaiting further analysis. The samples were kept in a cooler box with dry ice immediately after collection and during transportation to the laboratory.

Before DNA extraction, samples were thawed at room temperature (~22°C) and thoroughly mixed with vortexing at maximum speed for 30 seconds. Total DNA was extracted from a representative subsample of fecal and rumen liquor samples using a QIAamp® DNA Stool Mini Kit (Qiagen, Valencia, CA, USA), following the manufacturer’s guidelines with a few changes. The modifications were made to increase the DNA yield while reducing the amount of RNA obtained. The modifications included (i) doubling the recommended amount of the sample and (ii) adding 2 μl of RNAse A to the sample and proteinase K mixture. After elution, the DNA was visualized with UV light on 1% agarose gel electrophoresis. The DNA quantity and quality were assessed by using the NanoDrop spectrophotometer (ThermoScientific, USA) and Qubit 2.0 Fluorometer (Thermo Fisher Scientific, USA), following the manufacturer’s recommendations.

Preparation of DNA library and Illumina MiSeq sequencing

The Nextera DNA Preparation Kit and the Nextera Index Kit (Illumina, San Diego, CA, USA) were used to prepare the DNA Libraries following the manufacturer’s instructions with a few modifications. About 50 ng of total genomic DNA per sample underwent simultaneous tagmentation and the addition of adapter sequences at 55°C for 10 min. The resulting tagmented mixture was purified using the Zymo DNA Kit (Zymo Research Corporation, Irvine, CA, USA). The Zymo IIC spin columns were used, and all the centrifugation steps were performed at 10,000 ×g. This contrasted with the manufacturers’ recommendation of using a Zymo-Spin I-96 Plate centrifuged at 1300×g. A limited-cycle (5 cycles) polymerase chain reaction (PCR) was then conducted to amplify the insert DNA using a unique combination of barcode primers. This PCR reaction also added index sequences on both ends of the DNA. Finally, PCR products were cleaned up, and short library fragments, including indexes, were removed using AMPure XP beads (A63881, Beckman Coulter, Brea, CA, USA). The final library concentration was measured using the Qubit dsDNA HS Assay Kit (Thermo Fischer Scientific, USA) and the average library size was estimated using the Bioanalyzer TapeStation 2200 (Agilent Technologies, Santa Clara, USA). The libraries were diluted to 4 nM stocks, pooled in equimolar ratios, spiked with 1% PhiX, and paired-end (200 cycles) sequenced on the MiSeq® (Illumina, USA) platform at the BecA-ILRI Hub, Nairobi, Kenya.

Analysis and processing of sequence reads

The quality of all paired-end raw fastq sequencing reads were evaluated using FastQC (version 0.11.5) ( Before further sequence analysis, filters were used to remove reads with poor quality from all the samples. SolexaQA v3.1.5 [32] was used to calculate sequence quality statistics and perform quality filtering of the raw reads. Reads were then trimmed at a threshold of Q20 using DynamicTrim in SolexaQA++. Filtered reads were re-assessed for quality using FastQC, and any samples that still had a poor quality were further processed by truncating at any site having an average quality score < 20 using the FASTX-trimmer in the FASTX-toolkit v0.0.14 ( The cleaned fastq sequences were uploaded onto the publicly available Meta Genome Rapid Annotation using the Subsystem Technology server (MG RAST, v3.3) [33]. The taxonomic domain groups were allotted using MG RAST against the M5NR database. The data were also analyzed using the SEED Subsystems platform in MG RAST to identify putative protein-encoding sequences. This functional classification in SEED hierarchical classification had a percentage identity cut-off of 60%, an expected value (E) cutoff of 10−5, and a minimum alignment cut-off of 15 bp [34].

Statistical analysis

All the operational taxonomic unit (OTU) counts were normalized by relative abundance and log transformation, [log(x + 1)], before quantitative characterization using the Paleontological STatistics software package (PAST v3.13), [35]. Alpha diversity analysis was conducted in PAST v3.13 to evaluate the taxonomic richness and diversity using Chao1 minimal richness index [36], inverse Simpson diversity index [37], and Shannon diversity index [38]. Statistical differences per diet within each sample type were determined by a non-parametric Friedman test, using Genstat version 14 [39], and correction for multiple testing was performed using the Bonferroni adjustment. To group diets within sample types according to their characteristics, cluster analysis of the metagenomes was performed on all taxa using principal coordinate analysis (PCOA). The effect of diet and sample type were then calculated using a two-way PERMANOVA on 9,999 permutations (p < 0.01). For comparisons of microbial taxa between sample types and within diets for each sample type, a general linear model was fitted with diet and samples as the main effect and their interaction using Genstat software. Linear discriminant analysis Effect Size (LEfSe) [40], was used in the specific identification of OTUs that differed within diets in each sample type and diets and between the two sample types (fecal and rumen liquor). Relative abundances of all fecal and rumen liquor samples from all animals across all diets were compared to determine a core microbe. Any taxa found to have a relative abundance of ≥ 0.01% across all samples were then defined to be part of the core microbe. The core OTUs relative abundances were used as an input for LEfSe. The analysis conditions were set as follows: i) alpha value for the factorial Kruskal-Wallis test among classes at < 0.05; ii) alpha value for the pairwise Wilcoxon test among subclasses at < 0.05; iii) the threshold on the logarithmic LDA score for discriminative features at < 2.0; and iv) multi-class analysis was set as all-against-all. Additionally, two-group analysis was performed, applying a Fisher’s exact t-test with a 95% confidence interval to assess the differences in the abundance of microbes between fecal and rumen liquor samples. Pearson correlation analysis was performed with Genstat (version 14) to evaluate the relationship between fecal and rumen liquor samples within and between animals. Next, a t-test was performed on the correlation values to weigh if the correlation between the rumen and fecal profiles of a cow was more significant than the correlation between samples from different cows. These correlations were visualized as a heatmap using the R library corrplot. To estimate the differences in functional pathways, SEED Level 1 and 2 functional classification, one-way ANOVA to assess dietary effects, and a t-test to evaluate sample type effects were conducted using Genstat.


Sequence assessment

Metagenomic sequencing of the samples resulted in a total of 9.6 and 11.9 million raw reads for fecal and rumen liquor samples, respectively. After quality control, the total number of reads was reduced by 27.63% for fecal samples and 26.39% for rumen liquor samples. Upon MG-RAST annotation, less than 1% of reads per diet for both fecal and rumen liquor sample types were classified as rRNA based on hits against 16S rRNA gene sequence databases. About 73.56–81.22% and 74.54–78.32% of the reads from fecal and rumen liquor samples, respectively, were classified into various functional subsystems (Table C in S1 Text). The shotgun metagenomic approach applied in this study allowed us to infer the actual proportions of all domains of life and thus an overall identification of the rumen microbial community. The Bacteria domain was dominant irrespective of sample type and diet, consisting over 98% of all identified sequences. The domains of Archaea, Eukaryota, and Virus comprised remaining, less than 2%, of all the sequences found in both fecal and rumen metagenomes (Fig 1). Moreover, roughly 1% of the reads could not be categorized into any recognized OTU. The classification of individual OTUs identified 55 phyla, 131 classes, 253 orders, 443 families, 901 genera and 2357 species in four domains. Given the large number of classified groups and to facilitate the interpretation of results, our investigation focused on the most abundant taxonomic groups within each domain.

Fig 1. Phylogenetic classification of the rumen microbiome as revealed by metagenomic analysis.

a. Pie chart for microbial classification at the domain level. Averages are from all fecal and rumen liquor samples. b. Stacked bar plot of relative abundance of the top 10 bacterial phyla in the three diets for fecal and rumen liquor samples. c. Stacked bar plot of relative abundance of the dominant phyla in Archaeal domain. d. Stacked bar plot of relative abundance of one dominant phylum in viral domain, and e. Stacked bar plot of relative abundance of the dominant phyla in the Eukaryota domain.

Diversity indices

The alpha diversity indices for describing richness, diversity and evenness within each diet in fecal and rumen liquor sample types are shown in Fig 2A and Table D in S1 Text. ANOVA revealed no significant differences, in all diversity and evenness indices, between diets within fecal sample type. On the contrary within rumen liquor samples, richness estimator (Chao1) and diversity indices (Simpson and Shannon) showed significant differences between diets. When a general linear model was fitted with sample type and diet as the main effects and their interaction, all the indices showed significant differences between rumen liquor and fecal sample types whereas diet and sample type by diet interaction did not show differences among groups. Moreover, Bray-Curtis dissimilarity index was estimated from OTU counts to measure diversity across samples (beta diversity). The first two dimensions from the Bray-Curtis dissimilarity matrix (Fig 2B) revealed a significant different distance between fecal and rumen liquor sample types. PERMANOVA analysis indicated that sample type (p = 0.001), diet (p = 0.003) and their interaction (p = 0.001) contributed significantly to the differences. PCOA undertaken on diets within each sample type indicated significant differences between diets (S1 Fig). These differences were more pronounced in rumen liquor samples compared to fecal samples.

Fig 2. Alpha and beta diversity analysis.

a. Box-and-whiskers plot for estimated alpha diversity indices in the fecal and rumen liquor microbiota. NS = Not Significant, * = Significant at p = 0.05, ** = Significant at p = 0.01, *** = Significant at p = 0.001. b. principal coordinate analysis (PCOA) of fecal and rumen liquor microbial communities. The axis of principal component 1 (PC1) described 51.8% and principal component 2 (PC2) described 25.64% of total data variability. PCOA was performed using PAST v3.13 [35]. S = sample type, D = diet and SXD = sample type X diet interaction.

Effect of diet on the diversity and relative abundance of fecal and rumen microbiota

When the relative abundance of each taxon was compared across all samples, diet had a significant effect on several phyla, genera and species (Fig 3). Phylum level analysis within fecal samples revealed that Firmicutes (p = 0.01), Proteobacteria (p = 0.01), Actinobacteria (p = 0.02), Cyanobacteria (p = 0.01), Fibrobacteres (p = 0.01), Streptophyta (p = 0.03), and unclassified derivatives of Viruses (p = 0.01) were significantly different among the diets (Table E in S1 Text). Of these, Firmicutes, Actinobacteria, Fibrobacteria, Streptophyta, and unclassified derivatives of Viruses had the highest abundance in Diet 1, while Proteobacteria had the highest abundance in Diet 3, compared with the other diets. The rest of the phyla had the highest abundance with Diet 2. Within rumen liquor samples, all phyla, except Chordata (p = 0.18), significantly differed with changes in diet. Diet 1 was associated with higher proportions of all the phyla in the Archaea domain, Firmicutes (p = 0.01), Bacteroidetes (p < 0.01), Actinobacteria (p < 0.01), Spirochaetes (p = 0.04), Fusobacteria (p < 0.01), Lentisphaerae (p = 0.03), Cyanobacteria (p = 0.01), Fibrobacteres (p < 0.01), unclassified derivatives from Eukaryota (p < 0.01), Ascomycota (p = 0.01), and Streptophyta (p = 0.03), whereas Diet 3 had the lowest proportions in the same phyla. Conversely, Proteobacteria (p < 0.01) and unclassified derivatives of Viruses (p = 0.01) had the highest and lowest abundance in Diet 3 and Diet 1, respectively. Notably, Firmicutes and Bacteroidetes showed a consistent decrease in abundance, whereas Proteobacteria consistently increased when animals were transitioned from a high-fiber diet (Diet 1) to a high-concentrate diet (Diet 3) (Table E in S1 Text). A comparison of the core OTUs by linear discriminant analysis effect size (LEfSe) identified 19 in and 26 features to be significantly discriminative, with an absolute LDA score > 4.0, between diets in fecal and rumen liquor samples respectively. Of these features, Proteobacteria was differentially higher in Diet 1, Bacteroidetes in diet 2 and Firmicutes in diet 3 in fecal samples (Fig 4A). Within rumen liquor samples, Bacteroidetes and Firmicutes were higher in Diet 3 while Proteobacteria was enriched in Diet 1 (Fig 4B).

Fig 3. Comparison of relative abundance of most abundant taxa within diets in each sample type and between fecal and rumen liquor sample types.

a. heatmap hierarchical clustering of most abundant phyla. In the heatmap, F = fecal, RL = rumen liquor, S = sample type, D = diet and SXD = sample type X diet interaction. Phyla color correspond to domain, blue = Bacteria, red = Archaea, green = Eucharyota and brown = Virus. NS = Not Significant, * = Significant at p = 0.05, ** = Significant at p = 0.01, *** = Significant at p = 0.001. b. scatter plot of 20 most abundant genera, and c. bar chart of 15 most abundant species. In fig b and c, a log transformed scale of the relative abundances is used for beater visualization of low abundance taxa.

Fig 4. Cladograms and histograms illustrating significantly different microbial taxa in LEfSe analysis.

a. Diets within fecal sample type, b. Diets within rumen liquor sample type, c. Fecal and rumen liquor sample types. The cladogram illustrates the presence of microbial phyla and genera that are significantly different based on linear discriminant analysis (LDA) scores in the histogram. The LDA score at log 10 > 4 is set as threshold and the length of each bin, i.e., LDA score represents the extent to which the microbial taxa differ among the groups.

At the genus level, the most abundant genera, Bacteroides, Prevotella, Clostridium, Pseudomonas, Acinetobacter, and Ruminococcus, were from bacteria domain (Fig 3B). Further analysis on genera within the dominant phyla revealed that diet influenced several OTUs irrespective of the sample type (Table E in S1 Text). Of these, the abundance of six genera, Methanococcus, Methanosarcina, Clostridium, Eubacterium, Ethanoligenens, and Ruminococcus, was significantly different with diet changes in fecal samples. All these genera, except Methanosarcina, had the highest abundance with Diet 1 and lowest abundance with Diet 3. In rumen liquor samples, all the featured genera showed a similar abundance distribution at the genus level as that of the phyla that they hailed from, with different diets.

Species level composition identified 201 species having a relative abundance of ≥ 0.01% in all the samples (S1 Table). A statistical comparison of these species revealed that dietary changes had a significant effect on 53 species in fecal samples. However, these variations did not display specific trends as the concentrate content in diet was increased from Diet 1 to Diet 3. The top five species in fecal samples were Alistipes putredinis, Bacteroides capillosus, Bacteroides fragilis, Bacteroides vulgatus and Ruminococcus albus, with abundances ranging from 2.18 to 2.71%, 2.18 to 2.50%, 1.78 to 2.15%, 1.72 to 1.98 and 1.68 to 2.02% respectively (Fig 3C). In rumen liquor samples, dietary changes affected 154 species. The most abundant microbial species was Prevotella ruminicola, with relative abundances ranging from 9.72% to 17.77%. Other abundant species were Acinetobacter baumannii (0.11% to 5.78%), Pseudomonas putida (0.09% to 3.12%), Prevotella melaninogenica (1.2% to 2.61%) and Bacteroides vulgatus (1.17% to 2.31%). The abundance of Prevotella ruminicola, Prevotella melaninogenica and Bacteroides vulgatus decreased as the diet transitioned from the high-forage diet to the high-concentrate diet. Conversely, Acinetobacter baumannii and Pseudomonas putida increased in abundance as concentrate content in diet increased from Diet 1 to Diet 3 (Fig 3C and S1 Table). Despite their low abundance, other species identified in the rumen samples were Fibrobacter succinogenes, Ruminococcus flavefaciens, and Ruminococcus albus.

Variation in microbial taxa in fecal and rumen liquor sample types

Given that obtaining rumen liquor samples is more tedious and invasive than collecting fecal samples, we investigated the theory that the fecal metagenome of a cow could be used as a predictor of its rumen metagenome. For further analysis, we compared the relative abundance of the most common phyla (relative abundance > 0.1%) and genera (relative abundance > 0.05%) between fecal and rumen liquor samples (Table E in S1 Text). The relative abundance values were the mean of all the treatments and animals within each sample type. Relative abundance revealed that Firmicutes (p < 0.01), Actinobacteria (p < 0.01), Euryarchaeota (p = 0.03), and Fusobacteria (p = 0.01) phyla had significant differences between fecal and rumen liquor samples, while five genera showed significant differences between fecal and rumen liquor sample types (Fig 3A, Table E in S1 Text). Furthermore, a comparison of the core OTUs by linear discriminant analysis effect size (LEfSe) identified 847 features to be significantly discriminative between fecal and rumen liquor samples. Of these significantly discriminative features, 30 had an absolute LDA score > 4.0. One phylum, Firmicutes, was differentially abundant for fecal samples, while Bacteroidetes and Proteobacteria were much more enriched in rumen liquor samples. The genera, Prevotella and Clostridium, were the highest sources of variation between the communities, with an absolute LDA score factor of roughly 4.8 (Fig 4C). To test if the fecal microbial profile of a cow could predict the rumen liquor metagenome profile, the correlation between each fecal and rumen liquor metagenome profile was determined. These correlations were then evaluated to determine if they were greater for samples from the same animal than between animals. The correlations between fecal samples and rumen liquor samples in the same animal ranged from 0.4 to 0.66. Further, we observed no significant differences between correlations from the same animal and those from between animal samples (t-test; p = 0.914) (Fig 5).

Fig 5. Correlation heatmap representing correlations.

Correlation heatmap between fecal and rumen liquor microbial abundance in all the animals. F = fecal, RL = rumen liquor, D = Diet and numbers 1 to 6 represent the individual animals. Fecal -rumen liquor pairs from the same animal are highlighted by a black square. Overall comparison between fecal and rumen liquor sample types was not significant p = 0.91.

Functional capacity of ruminal and fecal microbiota

Normalization was carried out before the metabolic potential among the samples was undertaken. Normalization was done to account for differences in community structure, library size and to compare functional categories with low abundance [41]. The normalization was performed together for all pathways at Level 1 and within each specific pathway at level 2 classification. Table 1 illustrates the 28 most abundant predicted functional pathways found at level 1 classification. The most abundant level 1 pathway was carbohydrate metabolism ranging from 14.08% to 14.19% and 12% to 14.62% in fecal and rumen liquor samples, respectively, followed by clustering-based subsystems (functional coupling evidence but unknown function; fecal: 13.97% to 14.28%, rumen liquor: 11.99% to 13.21%), and then protein metabolism (12.5% to 13.37% in fecal and 8.67% to 11.67% in rumen liquor samples). Other dominant pathways were amino acids and derivatives (8.78–10.13%) and miscellaneous (5.52–6.14%), suggesting the dominant role of these functional pathways in all the samples. ANOVA revealed that dietary changes caused significant variation in only two pathways changes within fecal samples, whereas changes in diet resulted in significant variations in all the functional pathways, except miscellaneous (p = 0.152), in rumen liquor samples. All pathways were significantly varied between fecal and rumen sample types except DNA metabolism (p = 0.057) and photosynthesis (p = 0.279) (Table 1).

Table 1. Relative abundance (%) of level 1 subsystems classification of microbial functional pathways.

A total of 168 level 2 pathways were identified in the rumen and feces samples. Further detailed, level 2 analysis was undertaken on carbohydrates and protein metabolism pathways as they were the most abundant functional pathways at level 1 classification. The carbohydrate metabolism included eleven level 2 pathways. Amino sugars, central carbohydrate metabolism, fermentation, one-carbon metabolism, and organic acids significantly changed in fecal samples when the diet was varied. Conversely, only one pathway, CO2 fixation (p = 0.295), did not vary across diets in rumen liquor samples. Central carbohydrate metabolism, fermentation, organic acids, and sugar alcohols pathways increased with an increase in concentrate, whereas monosaccharides, disaccharides, oligosaccharides, and polysaccharides decreased with a greater proportion of dietary concentrate (S2 Table). Five different pathways, protein biosynthesis, protein degradation, protein folding, protein processing and modification, and selenoproteins were identified at level 2 classification of protein metabolism pathway. The maximum sequences were attributed to protein biosynthesis, followed by protein degradation. Protein degradation (p = 0.01) and selenoproteins (p = 0.001) were the only categories that varied in fecal samples. The variation in protein degradation did not follow any specific dietary pattern, whereas selenoproteins increased with increased concentration in the diet. Within the rumen liquor samples, the abundance of all the categories varied across the diets. With an increase in concentrate level, all the categories except protein biosynthesis increased (S2 Table).

MG-RAST-BLAT integration revealed that carbohydrate metabolism pathways were spread across 41, 333 and 28 genera from Archaea, bacteria, and Eukaryota domains, respectively. Protein metabolism pathways were generally scattered across 43 Archaea, 341 Bacteria, and 38 Eukaryota microbial genera. An assessment of the microbes responsible for the most abundant level 2 pathways (protein biosynthesis and central carbohydrate metabolism) revealed that within these two pathways, and irrespective of the diet, the prominent genera were Bacteroides and Clostridium in fecal samples. However, in rumen liquor samples, Diet 1 was dominated by Bacteroides and Clostridium; Diet 2 was dominated by Bacteroides and Pseudomonas; and Diet 3 was dominated by Acinetobacter, Bacteroides, and Pseudomonas (Fig 6).

Fig 6. Stacked bar charts of relative abundance (%) of predominant microbiota associated with various functional pathways.

a. protein biosynthesis, b. central carbohydrate metabolism, c. sulfur metabolism, d. oxidative stress, e. cell wall and capsule and f. dormancy and sporulation pathways.

Despite their low normalized abundance at level 1 classification, a detailed assessment was carried out on sulfur metabolism, stress response, cell wall and capsule, and dormancy and sporulation functional pathways. This assessment was because previous studies have shown the importance of sulphur in rumen microbial synthesis. Moreover, the rumen microbiome goes through an array of dietary stresses. To adapt to these constraints, rumen microbes have developed several stress responses like the capacity to enter dormant states (spores) or a cell wall/capsule development. For the fecal samples, only stress response (p = 0.031) and sulfur metabolism (p = 0.017) were significantly affected by dietary changes; however, all the pathways were significantly influenced by dietary variations for rumen liquor samples. Changes in pathways did not follow any variation in the diet for fecal samples, while in the rumen liquor samples, all the pathways except cell wall and capsule decreased with an increase in concentrate proportion in the diet.

In the sulfur metabolism pathway, the relative abundance of the level 2 functional category of inorganic sulfur assimilation reduced with increased concentration in the diet. However, organic sulfur assimilation increased with increased concentrations in the diet (S2 Table). The OTUs involved in sulfur metabolism belonged to several genera, as shown in (Fig 6). The stress response pathway contained seven categories at level 2 classification. The most abundant pathways in the rumen and feces were oxidative stress, osmotic stress, and heat shock, whereas acid stress, periplasmic stress, cold shock, and desiccation stress had the least abundance. The diet had significant effects in all the pathways within fecal samples except desiccation stress (p = 0.035) and periplasmic stress (p = 0.063). Acid stress (p = 0.798) and desiccation stress (p = 0.155) were the only two pathways that were not significantly affected by diet in rumen liquor samples.

The pathways that were significantly affected by dietary changes did not follow any dietary trend in fecal samples. Conversely, in rumen liquor samples, osmotic stress, oxidative stress, and cold shock increased with increased concentration. This contrasted with heat shock, which decreased with an increase in concentrate proportion (S2 Table). A total of 31 phyla and 372 genera spread across Archaea, Bacteria, and Eukaryota domains were affiliated with the stress response pathways. The results further showed that for oxidative stress response, Bacteroides and Clostridium were the most prominent genera, irrespective of the diet, in the fecal samples. In contrast, Bacteroides dominated Diet 1, Bacteroides and Pseudomonas were the highest genera with Diet 2, and genera Acidovorax, Acinetobacter, and Pseudomonas were the highest dominant with Diet 3 (Fig 6).

Four pathways were identified in level 2 classification of the cell wall and capsule in all the samples. These pathways were capsular and extracellular polysaccharides, the cell wall of mycobacteria, gram-negative cell wall components, and gram-positive cell wall components. The gram-negative cell wall components increased with an increasing amount of dietary concentrate, whereas the capsular and extracellular polysaccharides reduced. Genus-wise affiliation of the most abundant genera associated with the cell wall and capsule pathways were like those observed for the sulfur metabolism category, as shown in (Fig 6). Level 2 pathways were associated with spore DNA protection and unknown (null) function within the dormancy and sporulation pathway. Changes in diet did not exert significant changes in these pathways for both fecal and rumen liquor samples (S2 Table). Clostridium was the dominant genus involved in dormancy and sporulation for all fecal samples, while several diverse genera were associated with different diets in rumen liquor samples (Fig 6).


Inadequate feed resources are one of the greatest challenges facing dairy farmers in the East African tropics. This problem is aggravated by the high cost of commercially formulated feeds. For these reasons, dairy farmers subject their animals to constant changes in the quality and quantity of feeds [42]. These feed resources, which consist mainly of indigestible plant cell wall components, are degraded by rumen microbes into volatile fatty acids and microbial proteins that are critical for the hosts’ survival and production [43]. Among other factors, diet is a major driver of shifts in cattle GI tract microbial communities [8, 44]; however, information on the shifts in rumen GI tract microbiota of crossbred cattle reared by dairy farmers in the East African tropics is limited. Additionally, given the difficulty in collecting rumen liquor samples from cattle, it would be expedient if there was a relationship between the microbial structure of rumen fluid and other samples that are easy to obtain, like feces. In this study, we characterized shifts in ruminal and fecal microbiota and associated microbial functional roles occurring due to increasing dietary concentrate proportion. Additionally, we assessed the suitability of using fecal metagenome as a proxy for rumen metagenome in crossbred dairy cattle.

Microbial diversity and taxonomic assessment

Our results indicated that dietary changes influenced the community composition and abundance of several rumen microbial taxa, while very few OTUs were varied in the feces. Similar to previous studies [45, 46], in rumen liquor samples, the microbial richness decreased with an increase in concentrate proportion in the diet. This observation could have been due to extensive fermentation activities and a low rumen pH due to increased easily fermentable nutrients from the higher dietary concentrate proportion [47]. Thus, this environment was perhaps less favorable for some members of rumen microbiota, and as such, microbial richness declined. PCOA-based clustering analysis showed differences in treatments within rumen liquor samples with Diet 1 being significantly separated from Diet 2 and Diet 3. We hypothesize that this observation may have resulted from the relatively higher abundance of OTUs, especially from Bacteroidetes phylum, recovered when the animals were fed on a diet with high roughage proportion (Diet 1).

As reported in previous studies of cattle [18, 48], the most abundant phyla were Bacteroidetes, Firmicutes, and Proteobacteria, irrespective of sample type and diet. Similar to findings reported by Fernando et al. [49], changes from a high-roughage diet (Diet 1) to a high-concentrate diet (Diet 3) caused a consistent rise in the abundance of Proteobacteria and a decrease in the abundance of Bacteroidetes and Firmicutes in rumen liquor samples. Previous studies have demonstrated that members of the Bacteroidetes phylum can utilize starch, xylan, pectin, galactomannan, and arabinogalactan [50], because they have higher mean polysaccharide lyases (PLs) and glycoside hydrolases (GHs) genes per genome, as well as signal peptide-containing PLs and GHs, compared to members of any other bacterial phyla in the GI tract [51]. As such, members of the Bacteroidetes phylum are among the primary degraders of the many complex polysaccharides in the plant cell wall resulting in production of butyrate, a significant player in energy metabolism in the rumen [52]. The Firmicutes, on the other hand, can utilize carbohydrates such as xylan, cellulose, hemicellulose, and galactomannan as energy sources [53, 54]. This explained their higher abundance when animals were fed on Diet 1 and subsequently decreased as the concentration increased. Conversely, increased Proteobacteria abundance with an increase in concentrate proportion in the diet suggests a greater need for bacteria to digest the newly available fermentable carbohydrates [45, 49]. Moreover, when animals are fed higher amounts of starchy feed, the rumen pH decreases which in turn reduces the competitiveness of members of Bacteroidetes phyla, leading to their decrease. This allows opportunistic phyla, such as Proteobacteria, to proliferate faster per unit of time, resulting in an increase in the proportions of Proteobacteria [55]. Moreover, Acinetobacter was identified among the most abundant phyla in this microbial communities. This phylum is involved in enhancing the intestinal digestive activities and improving feed utilization [56]. The most abundant genera were Prevotella, Bacteroides, and Clostridium. While the abundance of these genera was relatively stable in fecal samples, their abundance decreased with an increase in concentrate proportion in rumen liquor samples. The predominance of these genera can be explained in part because i) Prevotella genus is comprised of genetically and metabolically diverse members [57] that are numerically high in animals fed on both high-grain and high-forage diets [58]. Prevotella species can use starches, simple sugars, and other non-cellulosic polysaccharides as energy sources [58]. Furthermore, Prevotella species, including members with hemicellulolytic and proteolytic activity [59], are involved in hemicellulose and pectin digestion [60] and protein or peptide metabolism [61] in the rumen. ii) The genus Bacteroides has been shown to extensively contribute to carbohydrate, small molecule, and organic acid metabolisms and plays a significant role in other Bacteria-linked metabolic processes [62]. iii) Clostridia contains cellulolytic strains that are mainly commensals in the GI tract. Clostridia members make up a substantial part (10–40%) of the total bacteria in the gut microbiota [63]. As such, Clostridia likely plays a crucial role in gut homeostasis by interacting with the other resident microbe populations and providing specific and essential functions [64].

Despite their lower abundance, other important microbial genera identified in this study include Ruminococcus, Bifidobacterium, Methanobrevibacter. Ruminococcus genus, one of the important fibrolytic microorganisms in Firmicutes phylum, displayed a decrease with adaptation to a grain diet. Fernando et al. [49] and Tajima et al. [65] have reported a gradual decrease in abundance of this genus. Ruminococcus degrades hemicellulose, pectin, and cellulose present in the plant cell wall by producing several types of cellulolytic and hemicellulolytic enzymes [66]. The abundance of lactic acid-producing genus Bifidobacterium reduced as the animals were shifted to a more starch-based diet. This could be because of reduction in the substrates; amylopectin, maltotriose, and maltodextrin, that the genus digests to produce lactic acid [67]. The genus is also a beneficial enteric bacterium that provides protection against enteric infection due to synergistic adhesion effect [68]. In Archaea, Methanobrevibacter was the main genus, confirming what previously reported in studies on rumen liquor from cattle [69]. Methanobrevibacter is an obligate anaerobe distinctly differentiated from other organisms, that produces methane as a major catabolic product. The variation in community composition observed in this study may be due to the rumen environment and the diet [70].

At the species level, the most abundant species, Prevotella ruminicola, decreased in abundance as animals were fed on diets with a high concentrate diet affirms their high polysaccharide degradation potential [71]. Similarly, other dominant fibrolytic bacterial species (Fibrobacter succinogenes, Ruminococcus flavefaciens, and Ruminococcus albus) exhibited a decreasing trend as the fiber contents decreased in the diet. Previous studies have reported that ruminal fibrolytic bacterial populations were positively related to dietary fiber contents and negatively related to dietary starch contents [65]. Therefore, these results suggest that rumen bacterial composition was mainly affected by the proportions of structured and unstructured carbohydrates in the different diets.

Dietary variations significantly affected very few OTUs in fecal samples at the phylum and genus level. This was further supported by a distinct clustering trend in all fecal microbiota samples in PCOA. This was in congruence with a previous study that characterized rumen and fecal microbiome in bloated and non-bloated cattle [72]. The few OTUs that were significantly affected by diet did not follow any pattern corresponding with the diets as previously shown by [8, 44, 49]. The stability of the fecal metagenome with changes in diet observed in this study could be attributed to the stability of the hindgut, especially in its pH, thus promoting greater substrate availability and consequently microbial stability [73]. Most species identified were normal commensals in the feces and were not affected by dietary changes. For example, Alistipes putredinis has been identified in cattle feces [74]. More research is required on these species roles in digestion.

It is challenging to obtain rumen liquor samples; hence, it would be advantageous if the microbe profiles of the rumen and other samples, such as fecal samples, which can be easily collected, are considerably overlapped. Given the depth and precision of NGS technology in assessment of microbial populations, we hypothesized that any relationship, however small, between rumen and fecal microbiota would be identified. Moreover, diet composition can affect fecal bacterial concentration with highest values associated with high energy diets due to a higher microbial growth in the rumen [75]. Similarly, increasing the maize silage proportion in the diet reduced the ruminal richness and diversity of the bacterial community but increased the number of total bacteria, which in turn should be reflected in the fecal bacteria [76]. However, we could not find proof of a significant link between rumen liquor and fecal profiles. This finding was in line with previous studies using 18 Angus calves [77], seven cows [9] and 22 Holstein dairy cows [78]. The dissimilarity in fecal and rumen liquor samples was further shown by diversity indices, PCOA and LEfSe analysis. The differences between fecal and rumen liquor metagenomes may be largely related to the role of the two environments; the rumen microbiota may be intensely selected to remain functional as the host relies on them for digestion, while the fecal metagenome may be less restricted [9]. Additionally, luminal pH, differences in gut motility, host secretions, and nutrients are other factors that affect the microbial structure in various sections of the gut [79]. We also found no indication to support our hypothesis that the fecal microbial profile of a cow correlated more closely with that of its rumen liquor than that of any other cow. Although this observation agreed with an earlier study [9], more advanced analysis and additional sample types may be needed to explore this theory.

Functional metagenomic classification

In the present study, fecal and rumen liquor sample types showed a significant variation in functional classification between the various diets. The largest percentage of gene functions were linked to protein and carbohydrate metabolism which are vital for microbial survival and proliferation [80]. The high abundance of carbohydrate and protein metabolism functional pathways was consistent with the findings of previous studies on cattle [81], humans [82], and mice [83]. The decreased carbohydrate pathway may be because when ruminants are fed fiber deficient rations, ruminal pH declines and microbial ecology is altered. Specifically, fibrolytic bacterial numbers are very sensitive to the pH change [66]. Therefore, a decrease in the number of these species may have had a corresponding decrease in this metabolic pathway. This was further supported by Bacteroides being the main genus responsible for this pathway.

Level 2 classification of protein metabolism pathway revealed that most ontologies were associated with protein biosynthesis. This high abundance of protein biosynthesis pathway could be explained in part by 19 different tRNA aminoacylation categories for different amino acids identified on further (level 3) analysis of this category. When the effect of diet was tested on level 2 pathways in protein metabolism, differences in metabolic potential were observed, indicating a selective pressure exerted on the microbes with metabolic capabilities. For example, the selenoproteins metabolism pathway increased with an increased amount of concentrate in the diet. Selenoproteins are involved in combating oxidative stress [84]. As seen in this study, an increase in concentrate concentration caused an increase in oxidative stress [85]. Consequently, selenoproteins pathway increased to counter this increase in oxidative stress. In carbohydrate metabolism, the increasing levels of central carbohydrate metabolism and fermentation pathways with the expected higher content of easily available carbohydrates in a high-concentrate, low-roughage diet.

Our results reported an increase in sulphur metabolism as the animals were transitioned from a high-fiber to a high-concentrate diet. We theorize that the high amount of concentrate led to a reduction of members of rumen microbiota due to a less favorable environment because of lactate accumulation. To cater for this, sulphur metabolism was increased to improve microbial growth and consequently utilization of lactate by the rumen bacteria by contributing to the synthesis of amino acids, especially methionine and cysteine [86]. This theory was supported by an apparent increase, although not significant, of the level 2 Lysine, threonine, methionine, and cysteine pathway within the Amino Acids and Derivatives metabolism pathway.

Microbes survive in the rumen under different stresses, which may be natural or associated with the feed. Some of the feed-associated stresses, such as anti-nutritional factors, act as natural antimicrobial agents by limiting the growth of some microbes [34]. To adapt to these constraints, rumen microbes, especially bacteria, can enter dormant states (spores) or develop a cell wall/capsule [87]. Spores serve to protect the bacterium from harmful environmental conditions by reducing it into a desiccated, cryptobiotic and highly defensive state which conveys resistance to many environmental assaults that would otherwise harm and kill the vegetative form of the bacterium [34]. We theorize that the increase in the relative abundance of stress response and gram-negative cell wall component pathways with an increase in the amount of concentrate diet was a counter mechanism by the gut microbiota to avoid cellular lysis and death because of lowered pH due to dietary changes. Further, the identification of spore DNA protection suggests the potential for long-term dormancy of cells through DNA protection.


This study compared the diversity and functional roles of fecal and rumen liquor microbial communities in crossbred cows under different diets. Our findings indicate that dietary modifications had a significant effect on several rumen and fecal microbial OTUs. In the rumen, an increase in dietary concentrate amount resulted in an upsurge in the abundance of Proteobacteria, while reducing the abundance of Bacteroidetes and Firmicutes. Conversely, the changes in microbial composition in fecal samples were not consistent with the dietary modification patterns. Thus, there was no significant relationship between the rumen and fecal metagenome. Moreover, fecal microbiota from one animal did not correlate more than that from different animals. Functional classification identified that microbial genes were dominated by those associated with carbohydrate metabolism and protein metabolism. The assessment of dietary effects on microbial functions revealed that an increased roughage level in the diet boosted protein synthesis while decreasing central carbohydrate metabolism. This study identified that Bacteroides, Clostridium, and Pseudomonas genera were the principal hosts of these microbial functions. Our current study suggests a connection between the feed of the host dairy cattle and their resident rumen microbiota. The study also reveals potential candidate taxa that may prove useful for future inoculation studies given their association with either roughage or concentrate based diets. However, given that little is known about the possible causes of the negative and positive correlations noted among the microbial and functional pathways analyzed here, additional work is needed to evaluate potential roles of enteric microbes responsible for feed digestion in the rumen, as well as other potential sample types that can serve as proxies for rumen microbial composition in crossbred dairy cattle.

Supporting information

S1 Text. Supporting information, containing tables on diet formulation, chemical analysis, metagenomic sequence sample details, estimators of diversity, and microbial phyla and genus relative.

Table A. Roughage-based diet supplemented with dairy meal concentrate. Table B. Chemical composition of the dietary components. Table C. Analysis of Variance (ANOVA) for the 36 metagenomic sequence sample details. Table D. Estimators of diversity within each diet for fecal and ruminal fluid samples (Mean ± Standard Error). Table E. Mean relative abundance (%) of microbial phyla and genus by sample type and diet.


S1 Fig. Principal coordinate analysis (PCOA) of dietary effects on microbial communities within fecal and rumen liquor sample types.

a. Fecal microbial communities PCOA, and b. Rumen liquor microbial communities.


S1 Table. Comparisons of relative abundance of dominant species within diets in fecal and rumen liquor sample types and between the two sample types.


S2 Table. Normalized mean abundances for selected SEED level 2 functional categories of diets within fecal and rumen liquor samples.



We acknowledge all staff in the University of Nairobi Veterinary farm for outstanding support during the experiment and sample collection.


  1. 1. FAO. Mapping supply and demand for animal-source foods to 2030. In: Robinson TP, Pozzi F, editors. Animal Production and Health Working Paper: Rome. 2011; 2:1–54.
  2. 2. Thornton PK. Livestock production: recent trends, future prospects. Philos Trans R Soc Lond B Biol Sci 2010 Sep 27;365(1554):2853–67. pmid:20713389
  3. 3. Delgado C, Rosegrant M, Steinfeld H, Ehui S, Courbois C. Livestock to 2020: The Next Food Revolution. Outlook Agric 2001 Mar 1;30(1):27–9.
  4. 4. Marshall K, Gibson JP, Mwai O, Mwacharo JM, Haile A, Getachew T, et al. Livestock Genomics for Developing Countries–African Examples in Practice. Front Genet 2019 Apr 24;10:297 pmid:31105735
  5. 5. Onwuka CFI, Isah OA, Oni AO, Aderinboye RY. Ruminant animal nutrition. University os Agriculture, Abeokuta, ANN503. Abeokuta, Ogun State, Nigeria: College of Animal Science and Livestock Production, Federal University of Agriculture (FUNAAB); 2011.
  6. 6. Kiplagat S, Limo KM, Kosgey SI. Genetic Improvement of Livestock for Milk Production. In: Narongsak C, editor. Milk Production—Advanced Genetic Traits, Cellular Mechanism, Animal Management and Health. Rijeka: InTech Publishers; Rijeka 2012 Sep 26;77–96.
  7. 7. Morgavi DP, Kelly WJ, Janssen PH, Attwood GT. Rumen microbial (meta)genomics and its application to ruminant production. Animal 2013 Mar;7(s1):184–201. pmid:23031271
  8. 8. 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 Nov 1;78(2):256–65. pmid:21671962
  9. 9. Ross EM, Moate PJ, Bath CR, Davidson SE, Sawbridge TI, Guthridge KM, et al. High throughput whole rumen metagenome profiling using untargeted massively parallel sequencing. BMC Genet 2012 Dec;13(1):1–4. pmid:22747657
  10. 10. Lawhead JB, Baker M. Introduction to Veterinary Science. 3rd ed. Boston: Cengage Learning; 2015.
  11. 11. Paz HA, Anderson CL, Muller MJ, Kononoff PJ, Fernando SC. Rumen Bacterial Community Composition in Holstein and Jersey Cows Is Different under Same Dietary Condition and Is Not Affected by Sampling Method. Front Microbiol 2016 Aug 3;7:1206. pmid:27536291
  12. 12. Kim M, Park T, Yu Z. Metagenomic investigation of gastrointestinal microbiome in cattle. Asian Australas J Anim Sci 2017 Nov;30(11):1515. pmid:28830126
  13. 13. Indugu N, Hennessy M, Kaplan-Shabtai VS, de Assis Lage CF, Räisänen SE, Melgar A, et al. Comparing noninvasive sampling techniques with standard cannula sampling method for ruminal microbial analysis. JDS Communications 2021 Nov 1;2(6):329–33. pmid:36337103
  14. 14. Tapio I, Shingfield KJ, McKain N, Bonin A, Fischer D, Bayat AR, et al. Oral Samples as Non-Invasive Proxies for Assessing the Composition of the Rumen Microbial Community. PLoS One 2016 Mar 17;11(3):e0151220. pmid:26986467
  15. 15. Amin N, Schwarzkopf S, Kinoshita A, Tröscher-Mußotter J, Dänicke S, Camarinha-Silva A, et al. Evolution of rumen and oral microbiota in calves is influenced by age and time of weaning. Anim Microbiome 2021 Dec;3(1):1–5.
  16. 16. McAllister TA, Cheng KJ, Okine EK, Mathison GW. Dietary, environmental and microbiological aspects of methane production in ruminants. Can J Anim Sci 1996 Jun 1;76(2):231–43.
  17. 17. Tajima K, Arai S, Ogata K, Nagamine T, Matsui H, Nakamura M, et al. Rumen Bacterial Community Transition During Adaptation to High-grain Diet. Anaerobe 2000 Oct;6(5):273–84.
  18. 18. Jami E, Mizrahi I. Composition and similarity of bovine rumen microbiota across individual animals. PLoS One 2012 Mar 14;7(3):e33306. pmid:22432013
  19. 19. Oulas A, Pavloudi C, Polymenakou P, Pavlopoulos GA, Papanikolaou N, Kotoulas G, et al. Metagenomics: Tools and insights for analyzing next-generation sequencing data derived from biodiversity studies. Bioinform Biol Insights 2015 Jan;9:75–88. pmid:25983555
  20. 20. Suenaga H. Targeted metagenomics: a high-resolution metagenomics approach for specific gene clusters in complex microbial communities. Environ Microbiol 2012 Jan;14(1):13–22. pmid:21366818
  21. 21. Santamaria M, Fosso B, Consiglio A, de caro G, Grillo G, Licciulli F, et al. Reference databases for taxonomic assignment in metagenomics. Brief Bioinform 2012 Nov 1;13(6):682–95. pmid:22786784
  22. 22. Sleator RD, Shortall C, Hill C. Metagenomics. Lett Appl Microbiol 2008 Nov;47(5):361–6. pmid:19146522
  23. 23. Snipen L, Angell IL, Rognes T, Rudi K. Reduced metagenome sequencing for strain-resolution taxonomic profiles. Microbiome 2021 Dec;9(1):1–9.
  24. 24. Kibegwa FM, Bett RC, Gachuiri CK, Stomeo F, Mujibi FD. A Comparison of Two DNA Metagenomic Bioinformatic Pipelines While Evaluating the Microbial Diversity in Feces of Tanzanian Small Holder Dairy Cattle. Biomed Res Int 2020 Apr 23;2020:1–12. pmid:32382536
  25. 25. Fan P, Bian B, Teng L, Nelson CD, Driver J, Elzo MA, et al. Host genetic effects upon the early gut microbiota in a bovine model with graduated spectrum of genetic variation. ISME Journal 2020 Jan 1;14(1):302–17. pmid:31624342
  26. 26. McCann JC, Wickersham TA, Loor JJ. High-throughput methods redefine the rumen microbiome and its relationship with nutrition and metabolism. Bioinform Biol Insights 2014 Jan;8:109–25 pmid:24940050
  27. 27. Habimana V, Cheruiyot RB, Amimo JO, Kibegwa FM, Githae D, Jung’a JO. Metagenomic analysis of enteric bacterial pathogens affecting the performance of dairy cows in smallholder productions systems. Afr J Microbiol Res 2018 May 7;12(17):387–98.
  28. 28. NRC. Nutrient requirements of dairy cattle. 7th ed, National Research Council. National Academy Press; Washington D.C. 2001;1–401.
  29. 29. AOAC. Official Methods of Analysis of AOAC International. 16th ed. Association of Official Analysis Chemists International. Association of Official Analytical Chemists; 2005.
  30. 30. Van Soest PJ, Robertson JB, Lewis BA. Methods for Dietary Fiber, Neutral Detergent Fiber, and Nonstarch Polysaccharides in Relation to Animal Nutrition J Dairy Sci 1991 Oct 1;74(10):3583–97.
  31. 31. Muizelaar W, Bani P, Kuhla B, Tapio I, Yáñez-Ruiz D, van Gastelen S. Rumen fluid sampling via oral stomach tubing method. In Methods in cattle physiology and behaviour research 2020. Publisso.
  32. 32. Cox MP, Peterson DA, Biggs PJ. SolexaQA: At-a-glance quality assessment of Illumina second-generation sequencing data. BMC Bioinformatics. 2010 Dec 27;11(1):1–6. pmid:20875133
  33. 33. Glass EM, Wilkening J, Wilke A, Antonopoulos D, Meyer F. Using the Metagenomics RAST Server (MG-RAST) for Analyzing Shotgun Metagenomes. Cold Spring Harb Protoc. 2010 Jan 1;2010(1):pdb-rot5368. pmid:20150127
  34. 34. Singh KM, Reddy B, Patel AK, Panchasara H, Parmar N, Patel AB, et al. Metagenomic analysis of buffalo rumen microbiome: Effect of roughage diet on Dormancy and Sporulation genes. Meta Gene 2014 Dec 1;2:252–68. pmid:25606408
  35. 35. Hammer Ø, Harper DAT, Ryan PD. PAST: paleontological statistics software package for education and data analysis, ver. 1.89. Palaeontol Electron 2001 Jun 22;4(1)1–31.
  36. 36. Chao A, Shen TJ. Nonparametric estimation of Sannon’s index of diversity when there are unseen species in sample. Environ Ecol Stat 2003 Dec;10(4):429–43.
  37. 37. Simpson EH. Measurement of Diversity. Nature 1949 Apr 30;163(4148):688.
  38. 38. Shannon CE. The Mathematical Theory of Communication. The Bell system technical journal 1948 Jul;27(3):379–423.
  39. 39. Payne RW. Genstat Release 14 Reference Manual. Part 3: Procedure library PL21. VSN International: Oxford 2011
  40. 40. Segata N, Izard J, Waldron L, Gevers D, Miropolsky L, Garrett WS, et al. Metagenomic biomarker discovery and explanation. Genome Biol 2011 Jun;12(6):1–8. pmid:21702898
  41. 41. Shi Y, Tyson GW, Eppley JM, Delong EF. Integrated metatranscriptomic and metagenomic analyses of stratified microbial assemblages in the open ocean. ISME Journal 2011 Jun 9;5(6):999–1013. pmid:21151004
  42. 42. Whitaker DA, Smith EJ, da Rosa GO, Kelly JM. Some effects of nutrition and management on the fertility of dairy cattle. Veterinary Record 1993 Jul 1;133(3):61–4. pmid:8212484
  43. 43. Khiaosa-ard R, Zebeli Q. Cattle’s variation in rumen ecology and metabolism and its contributions to feed efficiency. Livest Sci 2014 Apr 1;162:66–75.
  44. 44. Pitta DW, Pinchak WE, Dowd SE, Osterstock J, Gontcharova V, Youn E, et al. Rumen Bacterial Diversity Dynamics Associated with Changing from Bermudagrass Hay to Grazed Winter Wheat Diets. Microb Ecol Apr 29;59(3):511–22. pmid:20037795
  45. 45. 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. Ren X, editor. PLoS One 2013 Dec 31;8(12):e83424 pmid:24391765
  46. 46. Plaizier JC, Li S, Danscher AM, Derakshani H, Andersen PH, Khafipour E. Changes in microbiota in rumen digesta and feces due to a grain-based subacute ruminal acidosis (SARA) challenge. Microb Ecol 2017 Aug;74(2):485–95. pmid:28175972
  47. 47. Faniyi TO, Adegbeye MJ, Elghandour MMMY, Pilego AB, Salem AZM, Olaniyi TA, et al. Role of diverse fermentative factors towards microbial community shift in ruminants. J Appl Microbiol 2019 Jul 1;127(1):2–11. pmid:30694580
  48. 48. Brulc JM, Antonopoulos DA, Berg Miller ME, Wilson MK, Yannarell AC, Dinsdale EA, et al. Gene-centric metagenomics of the fiber-adherent bovine rumen microbiome reveals forage specific glycoside hydrolases. Proc Natl Acad Sci 2009 Feb 10;106(6):1948–53. pmid:19181843
  49. 49. Fernando SC, Purvis HT, Najar FZ, Sukharnikov LO, Krehbiel CR, Nagaraja TG, et al. Rumen microbial population dynamics during adaptation to a high-grain diet. Appl Environ Microbiol 2010 Nov 15;76(22):7482–90. pmid:20851965
  50. 50. Martens EC, Lowe EC, Chiang H, Pudlo NA, Wu M, McNulty NP, et al. Recognition and degradation of plant cell wall polysaccharides by two human gut symbionts. PLoS Biol 2011 Dec 20;9(12):e1001221. pmid:22205877
  51. 51. El Kaoutari A, Armougom F, Leroy Q, Vialettes B, Million M, Raoult D, et al. Development and validation of a microarray for the investigation of the CAZymes encoded by the human gut microbiome. PLoS One 2013 Dec 31;8(12):e84033. pmid:24391873
  52. 52. Henderson G, Cox F, Ganesh S, Jonker A, Young W, Janssen PH. Rumen microbial community composition varies with diet and host, but a core microbiome is found across a wide geographical range. Sci Rep. 2015;5(April):14567. pmid:26449758
  53. 53. Morrison M, Miron J. Adhesion to cellulose by Ruminococcus albus: a combination of cellulosomes and Pil-proteins? FEMS Microbiol Lett 2000 Apr 1;185(2):109–15. pmid:10754233
  54. 54. Dassa B, Borovok I, Ruimy-Israeli V, Lamed R, Flint HJ, Duncan SH, et al. Rumen cellulosomics: divergent fiber-degrading strategies revealed by comparative genome-wide analysis of six ruminococcal strains. PLoS One 2014 Jul 3;9(7):e99221. pmid:24992679
  55. 55. Khafipour E, Li S, Plaizier JC, Krause DO. Rumen microbiome composition determined using two nutritional models of subacute ruminal acidosis. Appl Environ Microbiol 2009 Nov 15;75(22):7115–24. pmid:19783747
  56. 56. Mao SY, Zhang RY, Wang DS, Zhu WY. Impact of subacute ruminal acidosis (SARA) adaptation on rumen microbiota in dairy cattle using pyrosequencing. Anaerobe 2013 Dec;24:12–9. pmid:23994204
  57. 57. Stevenson DM, Weimer PJ. Dominance of Prevotella and low abundance of classical ruminal bacterial species in the bovine rumen revealed by relative quantification real-time PCR. Appl Microbiol Biotechnol 2007 May;75(1):165–74. pmid:17235560
  58. 58. Purushe J, Fouts DE, Morrison M, White BA, Mackie RI, Coutinho PM, et al. Comparative Genome Analysis of Prevotella ruminicola and Prevotella bryantii: Insights into Their Environmental Niche. Microb Ecol 2010;60(4):721–9. pmid:20585943
  59. 59. Matsui H, Ogata K, Tajima K, Nakamura M, Nagamine T, Aminov RI, et al. Phenotypic characterization of polysaccharidases produced by four Prevotella type strains. Curr Microbiol 2000 Jul;41(1):45–9. pmid:10919398
  60. 60. Matsui H, Ushida K, Miyazaki K, Kojima Y. Use of ratio of digested xylan to digested cellulose (X/C) as an index of fiber digestion in plant cell-wall material by ruminal microorganisms. Anim Feed Sci Technol 1998 Apr 1;71(3–4):207–15.
  61. 61. Wallace R. The proteolytic systems of ruminal microorganisms. Annales de Zootechnie. 1996; 45(Suppl. 1):301–8.
  62. 62. Wirth R, Kádár G, Kakuk B, Maróti G, Bagi Z, Szilágyi Á, et al. The planktonic core microbiome and core functions in the cattle rumen by next generation sequencing. Front Microbiol. 2018 Sep 24;9:2285. pmid:30319585
  63. 63. Manson JM, Rauch M, Gilmore MS. The commensal microbiology of the gastrointestinal tract. GI microbiota and regulation of the immune system. 2008;15–28. pmid:18841700
  64. 64. Lopetuso LR, Scaldaferri F, Petito V, Gasbarrini A. Commensal Clostridia: leading players in the maintenance of gut homeostasis. Gut Pathog 2013 Dec;5(1):1–8.
  65. 65. Tajima K, Aminov RI, Nagamine T, Matsui H, Nakamura M, Benno Y. Diet-Dependent Shifts in the Bacterial Population of the Rumen Revealed with Real-Time PCR. Appl Environ Microbiol 2001 Jun 1;67(6):2766–74. pmid:11375193
  66. 66. Zhang L, Chung J, Jiang Q, Sun R, Zhang J, Zhong Y, et al. Characteristics of rumen microorganisms involved in anaerobic degradation of cellulose at various pH values. Rsc Adv 2017 Aug 16; 7(64):40303–10.
  67. 67. Bi Y, Zeng S, Zhang R, Diao Q, Tu Y. Effects of dietary energy levels on rumen bacterial community composition in Holstein heifers under the same forage to concentrate ratio condition. BMC Microbiol 2018 Dec;18(1):1–1.
  68. 68. Collado MC, Meriluoto J, Salminen S. Measurement of aggregation properties between probiotics and pathogens: in vitro evaluation of different methods. J Microbiol Methods 2007 Oct 27;71(1):71–4. pmid:17719109
  69. 69. Sirohi SK, Chaudhary PP, Singh N, Singh D, Puniya AK. The 16S rRNA and mcrA gene based comparative diversity of methanogens in cattle fed on high fibre based diet. Gene 2013 Jul 10;523(2):161–6. pmid:23603353
  70. 70. Buccioni A, Decandia M, Minieri S, Molle G, Cabiddu A. Lipid metabolism in the rumen: New insights on lipolysis and biohydrogenation with an emphasis on the role of endogenous plant factors. Anim Feed Sci Technol 2012 Jun 1;174(1–2):1–25.
  71. 71. Avguštin G, Wallace RJ, Flint HJ. Phenotypic diversity among ruminal isolates of Prevotella ruminicola: proposal of Prevotella brevis sp. nov., Prevotella bryantii sp. nov., and Prevotella albensis sp. nov. and redefinition of Prevotella ruminicola. Int J Syst Bacteriol 1997 Jan 1;47(2):284–8. pmid:9103611
  72. 72. Azad E, Derakhshani H, Forster RJ, Gruninger RJ, Acharya S, McAllister TA, et al. Characterization of the rumen and fecal microbiome in bloated and non-bloated cattle grazing alfalfa pastures and subjected to bloat prevention strategies. Sci Rep 2019 Dec 12;9(1):4272. pmid:30862851
  73. 73. Khafipour E, Li S, Plaizier JC, Dowd SE, Krause DO. Microbiome analysis of the rumen, cecum, and feces of dairy cows with subacute ruminal acidosis. J Anim Sci 2011;89:489.
  74. 74. Rautio M, Eerola E, Väisänen-Tunkelrott ML, Molitoris D, Lawson P, Collins MD, et al. Reclassification of Bacteroides putredinis (Weinberg et al., 1937) in a new genus Alistipes gen. nov., as Alistipes putredinis comb. nov., and description of Alistipes finegoldii sp. nov., from human sources. Systematic and applied microbiology. 2003 Jan 1;26(2):182–8.
  75. 75. van Vliet PCJ, Reijs JW, Bloem J, Dijkstra J, de Goede RGM. Effects of cow diet on the microbial community and organic matter and nitrogen content of feces. J Dairy Sci 2007 Nov 1;90(11):5146–58. pmid:17954755
  76. 76. Lettat A, Hassanat F, Benchaar C. Corn silage in dairy cow diets to reduce ruminal methanogenesis: Effects on the rumen metabolically active microbial communities. J Dairy Sci 2013 Aug 1;96(8):5237–48 pmid:23769352
  77. 77. Lourenco JM, Kieran TJ, Seidel DS, Glenn TC, Silveira MF da, Callaway TR, et al. Comparison of the ruminal and fecal microbiotas in beef calves supplemented or not with concentrate. Duarte M de S, editor. PLoS One. 2020 Apr 13;15(4):e0231533. pmid:32282837
  78. 78. Mu Y, Lin X, Wang Z, Hou Q, Wang Y, Hu Z. High-production dairy cattle exhibit different rumen and fecal bacterial community and rumen metabolite profile than low-production cattle. MicrobiologyOpen 2019 Apr;8(4):e00673. pmid:30277648
  79. 79. Salonen A, de Vos WM. Impact of Diet on Human Intestinal Microbiota and Health. Annu Rev Food Sci Technol 2014 Jan 2;5(1):239–62. pmid:24387608
  80. 80. Thomas M, Webb M, Ghimire S, Blair A, Olson K, Fenske GJ, et al. Metagenomic characterization of the effect of feed additives on the gut microbiome and antibiotic resistome of feedlot cattle. Sci Rep 2017 Sep 25;7(1):1–13.
  81. 81. Mao S, Zhang M, Liu J, Zhu W. Characterising the bacterial microbiota across the gastrointestinal tracts of dairy cattle: membership and potential function. Sci Rep 2015 Nov 3;5(1):1–4. pmid:26527325
  82. 82. Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, et al. Enterotypes of the human gut microbiome. Nature 2011 May;473(7346):174–80. pmid:21508958
  83. 83. Lu K, Abo RP, Schlieper KA, Graffam ME, Levine S, Wishnok JS, et al. Arsenic exposure perturbs the gut microbiome and its metabolic profile in mice: an integrated metagenomics and metabolomics analysis. Environ Health Perspect. 2014 Mar;122(3):284–91. pmid:24413286
  84. 84. Lu J, Holmgren A. Selenoproteins. J Biol Chem 2009 Jan 9;284(2):723–7. pmid:18757362
  85. 85. Ma Y, Zhang Y, Zhang H, Wang H. Thiamine Alleviates High-Concentrate-Diet-Induced Oxidative Stress, Apoptosis, and Protects the Rumen Epithelial Barrier Function in Goats. Front Vet Sci 2021 May 20;8:663698. pmid:34095275
  86. 86. da Silva CJ, Leonel F de P, Pereira JC, Costa MG, Moreira LM, de Oliveira TS, et al. Sulfur sources in protein supplements for ruminants. Revista Brasileira de Zootecnia. 2014;43:537–43.
  87. 87. Romero-Pérez GA, Ominski KH, McAllister TA, Krause DO. Effect of environmental factors and influence of rumen and hindgut biogeography on bacterial communities in steers. Appl Environ Microbiol. 2011 Jan 1;77(1):258–68. pmid:21075877