Ashdin Publishing Metagenomics Vol. 1 (2012), Article ID 235571, 11 pages doi:10.

4303/mg/235571

Research Article

The Bacterial Community Composition of the Bovine Rumen Detected Using Pyrosequencing of 16S rRNA Genes
Sitao Wu,1 Ransom L. Baldwin 6th,2 Weizhong Li,1 Congjun Li,2 Erin E. Connor,2 and Robert W. Li2
1 Center 2 USDA-ARS,

for Research in Biological Systems, University of California, San Diego, CA, USA Bovine Functional Genomics Laboratory, Beltsville, MD, USA Address correspondence to Robert W. Li, robert.li@ars.usda.gov Received 8 April 2012; Revised 23 April 2012; Accepted 24 April 2012

Abstract The rumen bacterial composition of both preruminant dairy calves and cows and beef steers was surveyed using pyrosequencing of the 16S rRNA gene. Sequences were analyzed using taxonomy-dependent and -independent clustering methods. The core rumen microbiome, regardless of the rumen developmental status or breeds, consisted of 8 phyla, 11 classes, 15 families, and 17 genera. Principal component analysis and clustering demonstrated that the bacterial communities in the rumen of pre-ruminant dairy calves, dairy cows, and beef steers were clearly distinguishable. Approximately 66% of phyla and 41% of Operational Taxonomic Units (OTUs) in a typical rumen bacterial community differed in relative abundance between the developing and mature rumen. Greater abundance of Fibrobacteraceae and Ruminococaceae in the rumen of beef steers likely reected the need for enhanced ber-digesting capacity in beef cattle. Our results should facilitate understanding of the structural and functional relationships in the rumen microbial ecosystem. Keywords 16S rRNA; cattle; hypervariable regions; metagenomic; rumen; pyrosequencing 1 Introduction The rumen as a complex microbial ecosystem has played a critical role in sustainable agriculture throughout human civilization. Microorganisms in the rumen, such as bacteria, archaea, protozoa, and fungi perform essential fermentation, including the conversion of plant ber to small molecules, such as volatile fatty acids (VFA) and vitamins, for the production of meat and milk for human consumption, thereby inuencing the hosts nutrition. In addition, rumen microorganisms modulate host immunity and enhance host resistance to invading pathogens [23]. Furthermore, certain bacteria detoxify naturally occurring compounds in the diet that are harmful to the host [13]. Microorganisms in the rumen are highly responsive to changes in diet, host genetics, and physiology, as well as geographical and environmental factors.

A solid understanding of major components of rumen microbial ecosystems and their interactions is a prerequisite to the development of strategies for ruminal manipulation, such as increasing the efciency of ber digestion and reducing ruminal methanogenesis. Approximately 11% of the rumen bacteria appear to be culturable [7]. This lack of methodology for successful culturing and unambiguously quantifying rumen microbes and their interactions has hindered our ability to use mathematical models to accurately predict the rumen ecosystem output and ruminant nutrition. It is important to recognize that the rumen microbiome functions as a tightly integrated system in which all resident, as well as transient species, contribute to its emergent properties. Furthermore, rumen microbial fermentation is a highly orchestrated and yet poorly understood process conducted by the interacting rumen microbiota constituents. While the predominant species in the ecosystem perform all major microbial conversions, it is conceivable that seemingly minor species play critical roles in this process. Therefore, identication and quantication of predominant (and permanent), as well as transient species, in the rumen should facilitate a better understanding of the rumen ecosystem function. Toward this end, numerous efforts have been made to study the genetic diversity of rumen microorganisms and rumen ecosystem dynamics using denaturing gradient gel electrophoresis (DGGE) [6, 27], restriction fragment length polymorphism (RFLP) [8], single strand conformation polymorphism (SSCP) [28], suppressive subtractive hybridization [11], and Sanger sequencing of 16S/18S rRNA genes [35]. However, these procedures are laborintensive and have relatively low resolution (such as DGGE and RFLP). Recently a meta-analysis of all 16S rRNA gene sequences of rumen origin deposited to the RDP database identied 5,271 operational taxonomic units (OTUs), representing 19 phyla [21], providing a glimpse of the dazzlingly global diversity of rumen microbiota. The advent of low-cost next-generation sequencing technologies, such as those based on 454/Roche pyrosequencing and Illumina

Metagenomics

platforms has stimulated the research in rumen microbial diversity and rumen ecosystem dynamics [15, 19, 22, 24]. For example, the rumen microbiota of dairy cows harbors 54.5 6.1 genera (mean sd) and 127.3 4.4 operational taxonomic units (OTUs), respectively [24]. Collectively, 21 prokaryotic phyla have been identied in the rumen of lactating dairy cows. However, the microbial diversity and bacterial population dynamics in the rumen of beef cattle, especially during their critical physiological stages such as adaptation to a high-grain diet in feedlots, have not received sufcient attention. In this study, we examined the ruminal bacterial composition of beef steers using pyrosequencing of 16S rRNA genes. Moreover, we compared the bacterial community composition of the rumen between dairy and beef cattle. The difference in taxonomical proles of the microbiota between the mature rumen and the developing rumen of dairy cattle was investigated as well. 2 Materials and methods 2.1 Animals and diet A total of 15 animals were used in this study, including 8 Angus beef steers ( 1 year old), 4 Holstein cows in mid-lactation ( 3 years), and 3 pre-ruminant Holstein bull calves of 42 days old. All animal procedures were conducted according to the protocols approved by the Beltsville Area Animal Care and Use Committee. Eight Angus beef steers (272.5 17.6 kg initial body weight), tted with ruminal and abomasal infusion catheters, were used in a randomized complete block design. Prior to experimentation steers were dewormed, vaccinated, and acclimated for 30 d. During this time, steers were halter broken and introduced to the tie stall facility. During the acclimation period and throughout the experiment, steers received a pelleted basal diet and were fed to 1.5 maintenance metabolizable energy requirements (160 Kcal metabolizable energy/(kg BW0.75 d)) for growing steers of this frame size. The basal diet consisted of 89.45% timothy grass hay, 5% corn grain, 5% Soybean meal (48% Crude Protein), 0.50% trace mineral-salt and 0.05% vitamin premix on a dry matter basis. Steers received the diet in 12 portions daily at 2-h intervals and were allowed ad libitum access to water. Steers were housed in individual tie stalls (1.1 1.8 m) in a temperature-controlled barn and were exercised for 1 h twice weekly when infusion lines were cleaned. Two post-ruminal treatments were administered per abomasal infusion and included: casein (C; N = 4) or starch (D; N = 4) for 42 d. Treatments were administered on an equal energy basis (167.4 kJ Metabolizable Energy/kg metabolic BW). Abomasal infusion treatments were introduced incrementally (25, 50, 75, 100%; every other day) during the rst 7 d. Steers were weighed weekly and rations offered were adjusted accordingly on a weekly basis. On day 42, rumen contents ( 200 ml) were collected via rumen stula.

Four ruminally-cannulated Holstein cows used in this study were described previously [24]. These multiparous cows were fed ad libitum standard lactation rations as a Total Mixed Ration (TMR; 50% corn silage and 50% concentrate at a dry matter basis). Diets were formulated to provide or exceed NRC dietary recommendations for lactation. Complete dietary chemical composition was determined weekly on composited daily samples of TMR composited weekly. Cows were moved to a tie stall barn for adaptation and acclimation at least ve days prior to the collection of rumen contents. Similarly, the rumen contents of preruminant calves fed milk replacer (Cornerstone 22:20; Purina Mills, Gray Summit, MO, USA) were collected as described [22]. The rumen liquor pH was measured using a standard pH meter. The liquid fraction of rumen contents was passed through a 300-M metal sieve and centrifuged at 16, 000 g for 3 min at 4 C. The supernatant was decanted and the remaining solid materials were snap frozen in liquid nitrogen and stored in 80 C until DNA extraction. 2.2 DNA extraction, amplicon library preparation, and pyrosequencing Metagenomic DNA extraction and sequencing were conducted in essentially the same manner as described previously [25, 43, 26]. Briey, a QIAamp DNA stool kit (Qiagen, Valenica, CA) was used with modications. A six-minute incubation at 95 C was used to replace the 70 C lysis recommended in the standard protocol. A 570 bp region of the 16S rRNA gene (E. coli position 357 to 926) containing hypervariable regions V3V5 of the 16S rRNA gene, was amplied from 40 ng of metagenomic DNA with 8-bp sample-specic barcoded primers using 2.5 units of High Fidelity AccuPrime Taq DNA Polymerase (Invitrogen, Carlsbad, CA) in a 50 L reaction buffer containing 200 nM primers, 200 nM dNTP, 60 mM Tris-SO4, 18 mM (NH4)2SO4, 2.0 mM MgSO4, 1% glycerol, and 100 ng/L bovine serum albumin (New England Biolabs, Ipswich, MA). PCR was performed using the following cycling prole: initial denaturing at 95 C for 2 minutes followed by 25 cycles of 95 C for 30 s, 50 C for 30 s, and 72 C for 120 s. The V3V5 regions were selected because of their high variability [2, 17, 39]. Amplicon purication, quantication and libray preprartion were essentially the same as described previously [25]. The puried amplicon library was further veried and quantied using a BioAnalyzer 2000 (Agilent, Santa Clara, CA, USA) and subjected to 454/Roche pyrosequencing. Unidirectional pyrosequencing of amplicon libraries was performed according to the manufacturers instructions with a modication (App No 001-2009, Roche Applied Science, Indianapolis, IN). This modication, using a specic fusion primer design, accommodates amplication using the GS FLX Titanium emPCR Kits (Lib-L). Pyrosequencing was conducted using

Metagenomics

a GS FLX Titanium System (Roche) following the manufacturers protocol. A total of 15 samples, from the rumen of 8 beef steers in 2 treatment groups, abomasal infusion of casein and starch, 4 dairy cows, and 3 calves were sequenced. The number of raw sequences reads per sample was 37, 004.6 6, 612.4 (mean sd) at 456 bp/read. 2.3 Sequence analysis Raw reads were rst decoded based on 8-bp bar codes; their quality was checked and artifacts were removed as previously described [2]. Sequences were ltered to remove lowquality reads and initially analyzed using BLAST against the GreenGene database (http://greengenes.lbl.gov/). The sequences of non-16S rRNA gene origin, such as host contamination, were removed. Resultant quality sequence reads were then analyzed using RDP Classier [41] from the Ribosomal Database Project (release 10; http://rdp.cme.msu.edu/) at a 80% condence threshold for taxonomic classication. Raw read counts were normalized. A square root transformation was applied to the relative abundance data. Bray-Curtis similarity matrix was then calculated using PRIMER v6 software (Plymouth Marine Laboratory, Plymounth, UK). Principal component analysis (PCA) was performed using the ade4 package in R as described elsewhere [43]. The 16S rRNA gene sequences also were analyzed using taxonomy-independent methods, such as CD-HIT-OTU [44] (http://weizhong-lab.ucsd.edu/cd-hit-otu/), ESPRIT-tree [4], and CROP [14] for taxonomic inference at the OTU level. The CD-HIT-OTU algorithm uses a greedy incremental clustering process to identify OTUs from 16S rRNA gene sequences, which involves 3 major steps: raw read ltering and trimming, selection of error-free reads, and clustering selected representative reads into individual OTUs at a userspecic cutoff (at both 0.03 and 0.05 distance levels). The program avoids over estimation of OTUs, a common problem for many existing programs, and results in a rapid and more accurate estimation of microbial diversity in complex microbial ecosystems. The consensus sequences of clusters identied using CD-HIT-OTU were then annotated using FR-HIT [29] against the GreenGene database. Unadjusted P values were calculated based on a modied t-test. 3 Results 3.1 Taxonomical proles of the bovine rumen microbiota The bacterial community composition in the rumen of pre-ruminant calves and mature dairy and beef cattle was estimated from the 16S rRNA gene sequences generated by Roche/454 pyrosequencing. The mean number of 16S sequences generated per sample was 37, 004.6 6, 612.4 (mean sd; N = 15). Rarefaction analysis was performed at the family, genus, and OTU levels and suggested that the sequencing depth at this scale was adequate (Figure 1).

(a)

(b)

(c)

Figure 1: Rarefaction curves of the rumen microbial community based on 16S rRNA gene sequences generated using Roche/454 pyrosequencing. (a) Family; (b) genus; and (c) operational taxonomic units (OTUs) at a 0.03 distance level.

Metagenomics

Table 1: The number of taxa identied in the bovine rumen using pyrosequencing of the 16S rRNA genes. The ruminal taxonomical proles at the phylum to genus levels were derived from RDP Classier. The operational taxonomic units (OTUs) at the 0.03 distance level (bacterial species) was identied using greedy heuristic clustering-based CD-HIT-OTU. Shared by all samples (may represent the core rumen microbiome). Collectively identied in this study, including both the developing and mature rumen of dairy and beef cattle.
Phylum Developing rumen (core ) mean ( sd) Mature rumen (core) Dairy cows core mean ( sd) Beef steers core mean ( sd) Total 11 15.6 ( 1.2) 21 16 22.3 ( 2.3) 31 22 39.4 ( 4.6) 93 37 76.5 ( 12.7) 219 175 499.6 ( 34.1) 1,079 14 17.3 ( 0.8) 16 22.8 ( 1.1) 26 41.0 ( 4.1) 50 73.8 ( 5.5) 368 537.5 ( 30.5) 10 12.3 ( 1.2) 10 Class 18 20.0 ( 1.4) 14 Family 35 49.0 ( 6.5) 18 Genus 53 97.0 ( 11.9) 30 OTU 59 133.3 ( 10.3) 102

Several quality control procedures were applied to raw reads, including removal of putative artifacts. The RDP classier was able to assign 96% of input 16S sequences to a given phylum at an 80% condence threshold. The RDP classier identied 21 phyla, 31 classes, 93 families, and 219 genera in the study, from both the developing and mature rumen, regardless of breed (Table 1). In the developing rumen of pre-ruminant calves fed milk replacer, the mean numbers of phyla and genera per sample identied were 12.3 1.2 (mean sd) and 97.0 11.9, respectively. In contrast, a greater number of phyla were identied in the mature rumen of both dairy cows and beef steers (P < 0.001; Table 1). Sequences assigned to the domain archaea were rare and detected in very low abundance only in 3 of 4 dairy cows and 2 of 8 beef steers analyzed. The primers used in this study to amplify the V3V5 regions of the 16S rRNA genes were not designed specically for archaea. This was conrmed by the fact that the very low abundance of Euryarchaeota was observed in our dataset. The core rumen microbiome of both the developing and mature rumen consisted of 8 phyla, 11 classes, 15 families, and 17 genera. These 8 phyla were Bacteroidetes, Firmicutes, Proteobacteria, Spirochaetes, Fibrobacteres, Verrucomicrobia, Synergistetes, and Actinobacteria, in descending order of relative abundance. These 8 phyla, accounting for 99.5% of input 16S sequences, are likely representing minimal components of the rumen bacterial community. The mean number of genera in the rumen microbiota under study was 79.9 14.0 ( sd). The 17 genera consisting of the core microbiome represented only 21.3% of all genera present in each sample. These genera, such as Butyrivibrio, Fibrobacter, Oscillibacter, Paraprevotella, Prevotella, Ruminococcus, Succinivibrio, and Treponema, accounted for 67.6% of 16S sequence reads and likely contributed to the basic function of the rumen microbial ecosystem.

Taxonomy-dependent algorithms such as RDP Classier have been widely used to estimate bacterial community composition. However, their inherent limitation was obvious due to the fact that vast majorities of rumen microorganisms are not described and captured in the existing databases such as the Ribosomal Database Project and GreenGene databases. To overcome these limitations, we also surveyed the rumen microbial community composition using three taxonomy-independent clustering programs, CD-HIT-OTU, ESPRIT-tree, and CROP. The numbers of OTUs identied at the 0.03 and 0.05 distance levels (parameters loosely used to dene bacterial species and genera, respectively) by the CD-HIT-OTU method in the developing rumen were 133.3 12.7 and 113.0 9.2, respectively. In the mature rumen of dairy and beef cattle, the numbers of OTUs identied using these two cutoffs were 512.3 39.2 and 343.0 21.7, respectively. The difference in the number of OTUs detected using both cutoff values between the developing and mature rumen was signicant (P < 1.0 109 ). However, no differences were detected in the number of OTUs at both the 0.03 and 0.05 levels between dairy and beef cattle (P > 0.05). Of note, the number of genera (or OTUs at the 0.05 distance level) detected between RDP Classier and taxonomy-independent methods in the rumen of either pre-ruminant calves or dairy and beef cattle was rather different. Taxonomy-independent methods tend to identify a signicantly larger number of genera, especially in the mature rumen (343.0 21.7 vs 75.6 11.4 by RDP Classier), suggesting that the rumen microbiota may harbor many novel microorganisms yet to be captured in the public database. However, caution is warranted in using taxonomyindependent clustering methods in analyzing short sequence reads of the 16S rRNA gene. Different algorithms use different equations to compute the dissimilarity distance

Metagenomics

Figure 2: Clustering analysis based on the familylevel Bray-Curtis similarity generated using the Primer 6 program. The relative abundance data were square-root transformed. The linkage function: complete linkage. A: pre-ruminant dairy calves (developing rumen); B: dairy cows (mature rumen); C: beef steers with abomasal casein infusion (mature rumen); and D: beef steers with abomasal starch infusion (mature rumen). between clusters, as discussed [37]. Liberal applications of pre-dened nominal distance levels, such as 0.03 (or 97% similarity) and 0.05 (or 95% similarity) for taxonomic assignment at bacterial species and genus levels, respectively, often lead to inated numbers of OTUs and subsequently overestimate microbial diversity in a given microbial community. In this study, the ESPRITtree program partitioned the 16S sequences into 28,495 and 14,186 OTUs using the 0.03 and 0.05 distance levels, respectively while CROP identied 5,434 and 1,215 OTUs from the same dataset using the same thresholds (at the 0.03 and 0.05 distance levels, respectively). 3.2 Unique characteristics of the bacteria composition in the mature rumen The bacterial composition between the developing and mature rumen displayed a striking difference. Bray-Curtis similarity matrix analysis and clustering using Primer v6 grouped all mature rumen samples together based on the relative abundances of bacterial composition at family level and the separation between the developing and mature rumen was distinct (Figure 2). PCA also conrmed the clustering results that the three groups of samples, from the developing rumen, dairy cows, and beef steers were clearly different, while the two treatment groups within the beef steers were indistinguishable (Figure 3). Among 15 phyla, 22 classes, and 436 OTUs consisting of a typical rumen bacterial community, 10 phyla (66%),

Figure 3: Differences in the microbial community composition in the rumen of pre-ruminant dairy calves (group A, developing rumen), Dairy cows (group B), Beef steers with abomasal casein infusion (group C), and beef steers with abomasal starch infusion (group D). Principal component analysis was performed using the ade4 package in R based on relative abundance of 20 most abundant families. Of note, while the developing rumen and the mature rumen of dairy cows and beef steers were distinctly separate, the 2 groups of beef steers with abomasal infusion of casein and starch, respectively (C and D) were indistinguishable. 13 classes (61%), and 173 OTUs (41%) were different in their relative abundance (P < 0.05) between the developing and mature rumen. At the family level, Bacteroidaceae was the most abundant family in the developing rumen, accounting for 59.8% of the 16S sequences. However, its relative abundance in the mature rumen was 0.04% and 0.54% in the rumen of dairy cows and beef steers, respectively. The difference in the abundance of Bacteroidaceae between the developing and mature rumen was signicant (P < 0.0001). The family Comamonadaceae displayed a similar pattern; its abundance was higher in the developing rumen but barely detectable in the mature rumen (Table 2). On the other hand, the relative abundance of the family Prevotellaceae was very great in the rumen of dairy cows (69.1%) and high in the rumen of beef steers (32.0%) but was small in the rumen of pre-ruminant calves (6.96%) (P < 0.0001). Table 3 lists selected genera that displayed signicant changes in relative abundance between the developing and mature rumen. Twenty six of the 30 genera had a signicant greater abundance in the developing rumen than in the mature rumen while only 4 genera had a higher abundance

Metagenomics

Table 2: The family-level percentage composition of the core microbiome of the bovine rumen. Group A: pre-ruminant dairy calves (N = 3). Group B: dairy cows (N = 4). Group C: beef steers with abomasal casein infusion (N = 4). Group D: beef steers with abomasal starch infusion (N = 4).
Rumen Developing (D) Family Bacteroidaceae Comamonadaceae Porphyromonadaceae Lachnospiraceae Prevotellaceae Ruminococcaceae Synergistaceae Desulfovibrionaceae Veillonellaceae Erysipelotrichaceae Clostridiales IS XIII Succinivibrionaceae Fibrobacteraceae Spirochaetaceae Coriobacteriaceae Group A 59.76 20.51 5.65 7.88 8.40 3.04 4.59 4.79 6.96 2.40 3.69 0.80 1.20 0.13 0.29 0.38 0.68 0.39 0.36 0.50 0.12 0.08 0.01 0.00 0.01 0.00 0.08 0.07 0.06 0.09 Dairy cows Group B 0.04 0.02 0.01 0.01 2.59 1.19 9.39 1.93 69.10 3.34 5.82 1.47 0.09 0.02 0.01 0.00 7.46 1.52 0.12 0.04 0.26 0.09 0.61 0.32 1.79 0.60 1.20 0.16 0.04 0.01 Mature (M) Beef steers Group C Group D 0.62 0.61 0.46 0.29 0.03 0.01 0.01 0.00 3.00 1.51 1.65 0.55 32.89 8.44 35.80 5.24 35.93 11.81 28.05 3.76 14.94 2.23 19.63 3.79 0.18 0.08 0.18 0.06 0.02 0.02 0.02 0.01 2.58 0.84 3.10 0.46 0.11 0.09 0.13 0.13 0.77 0.39 1.30 0.32 1.28 0.83 0.56 0.60 2.86 0.41 3.72 1.38 3.66 1.71 3.86 0.81 0.11 0.06 0.10 0.05
P values

D vs M 0.0000 0.0143 0.0001 0.0198 0.0072 0.0244 0.0000 0.0143 0.0260 0.1055 0.0542 0.0606 0.0014 0.0110 0.5358

Dairy vs Beef 0.0548 0.0900 0.7342 0.0000 0.0000 0.0002 0.0386 0.2964 0.0000 0.9751 0.0062 0.4735 0.0259 0.0025 0.0282

in the mature rumen, including Fibrobacter ( 570-fold higher in the mature rumen), Prevotella ( 21-fold higher); Succiniclasticum ( 36-fold higher) and Syntrophococcus (almost undetectable in the developing rumen). The 10 most abundant OTUs at the 0.03 distance level in the developing rumen accounted for 82.6% of 16S sequences from the community. All 10 OTUs displayed a signicant difference in their abundance between the developing and mature rumen (P < 0.01). Annotation of the consensus sequences representing these OTUs indicated that the 2 most abundant OTUs were dominant in the community and represented by 55.9% sequences. Not surprisingly, both OTUs were annotated to the genus Bacteroides, the most abundant genus in the developing rumen based on the RDP Classier results. The third most abundant OTU belonged to the genus Prevotella, which comprised 5.35% of the 16S sequences. This number was somewhat higher than that derived from the taxonomydependent RDP Classier (Table 3). Interestingly, the consensus sequences from OTU clusters detected by CDHIT-OTU allowed for species identication. Two species, Comamonas kerstersii and Porphyromonas levii, comprising 5.19 and 5.14% in the community in the developing rumen, respectively, were barely detectable in the mature rumen. Intriguingly, only 4 OTUs were shared by all samples tested (of pre-ruminant calves, dairy cows, and beef steers). Two of the OTUs can be annotated to the family Ruminococcaceae while the remaining two were related to the order Bacteroidales and the genus Paludibacter (in the family Porphyromonadaceae). Together, our results demonstrate that the microbial communities in

the rumen from different developmental stages have unique compositional characteristics. 3.3 Microbiological distinctions in the rumen between dairy and beef cattle Bacterial community structure and composition in the rumen between dairy and beef cattle were readily differentiated (Figure 3). The 5 most abundant phyla in the rumen of dairy cows were Bacteroidetes (68.2%), Firmicutes (23.3%), Proteobacteria (5.1%), Fibrobacteres (1.4%), and Spirochaetes (1.0%). Together, these 5 phyla accounted for 99.1% of 16S sequences analyzed. These phyla were also among the 5 most abundant in the rumen of beef steers, which represented 99.3% of 16S sequences. However, the order was different; the relative abundance of Bacteroidetes and Proteobacteria was signicantly less in the rumen of beef steers than of dairy cows (P < 0.001) while the relative abundance of 3 other phyla was greater, compared to in the rumen of dairy cows. A total of 8 phyla exhibited differences in their abundance between beef steers and dairy cows (P < 0.05). Similarly, 10 classes, 20 families, 42 genera, and 345 OTUs had altered relative abundance between beef and dairy cattle (Figure 4). Table 4 lists the 20 families wherein relative abundance was affected, including 9 families that consist of the core microbiome, such as Fibrobacteriacease, Lachnospiraceae, Prevotellaceae, Ruminococaceae, and Veillonellaceae. The signicantly greater abundance of Fibrobacteriacease and Ruminococaceae in the rumen of beef steers is likely reective of the percentage of hay in the diet and possibly a greater need for enhanced ber-digesting capacity in the rumen of beef cattle.

Metagenomics

Table 3: Genera with a signicant difference in their relative abundance between the developing and mature rumen (P < 0.01). The numbers denote mean sd (percentage composition).
Genus Developing Mature Acinetobacter 0.034 0.031 0.003 0.005 Alistipes 0.206 0.210 0.002 0.006 Bacteroides 70.420 21.360 0.687 0.864 Butyricimonas 2.029 0.982 0.016 0.035 Campylobacter 0.531 0.203 0.020 0.025 Cloacibacillus 1.031 0.725 0.004 0.009 Clostridium XIX 0.006 0.007 0.000 0.000 Eggerthella 0.005 0.006 0.000 0.000 Fibrobacter 0.010 0.004 5.703 3.010 Flavonifractor 0.193 0.125 0.007 0.011 Gallibacterium 0.321 0.276 0.001 0.002 Guggenheimella 0.654 0.704 0.012 0.028 Lachnospiracea i.s. 0.206 0.161 1.683 0.769 Mannheimia 0.018 0.016 0.000 0.000 Megamonas 0.003 0.002 0.000 0.000 Morganella 0.003 0.003 0.000 0.000 Odoribacter 0.110 0.030 0.001 0.002 Parabacteroides 1.080 0.615 0.007 0.010 Parasutterella 0.006 0.005 0.000 0.000 Pelistega 0.052 0.027 0.002 0.005 Phascolarctobacterium 0.116 0.047 0.004 0.008 Porphyromonas 5.745 3.951 0.041 0.063 Prevotella 2.124 0.483 42.945 19.788 Pseudoavonifractor 0.005 0.006 0.000 0.000 Sphingobacterium 0.005 0.002 0.000 0.000 Sphingomonas 0.014 0.016 0.001 0.002 Streptococcus 0.122 0.072 0.014 0.043 Succiniclasticum 0.170 0.150 6.052 2.195 Syntrophococcus 0.000 0.000 0.103 0.051 Turicibacter 0.003 0.003 0.000 0.000
P value 0.0028 0.0021 0.0000 0.0000 0.0000 0.0001 0.0047 0.0051 0.0072 0.0001 0.0005 0.0033 0.0066 0.0005 0.0009 0.0010 0.0000 0.0000 0.0005 0.0000 0.0000 0.0001 0.0041 0.0065 0.0000 0.0084 0.0042 0.0006 0.0047 0.0006

Table 4: Families with a signicant difference in their percentage composition in the rumen microbiota between mature dairy and beef cattle. The numbers denote mean sd (N = 4 for dairy; N = 8 for beef).
Family Acetobacteraceae Anaeroplasmataceae Bidobacteriaceae Chloroplast Clostridiales is XIII Coriobacteriaceae Desulfobulbaceae Fibrobacteraceae Lachnospiraceae Methanobacteriaceae Methylobacteriaceae Planctomycetaceae Prevotellaceae Rikenellaceae Ruminococcaceae Sphingomonadaceae Spirochaetaceae Synergistaceae Veillonellaceae Xanthomonadaceae Dairy 0.118 0.057 0.653 0.068 0.024 0.016 0.010 0.006 0.257 0.088 0.038 0.010 0.034 0.019 1.794 0.597 9.389 1.929 0.004 0.003 0.078 0.016 0.024 0.021 69.097 3.338 0.003 0.002 5.818 1.471 0.003 0.002 1.197 0.158 0.092 0.022 7.464 1.518 0.005 0.006 Beef 0.001 0.001 0.397 0.215 0.000 0.000 0.111 0.053 1.035 0.436 0.106 0.051 0.008 0.010 3.294 1.049 34.344 6.691 0.001 0.002 0.000 0.000 0.005 0.004 31.992 9.139 0.048 0.036 17.287 3.815 0.000 0.000 3.762 1.245 0.178 0.069 2.839 0.685 0.000 0.000
P value 0.0001 0.0470 0.0012 0.0040 0.0062 0.0282 0.0135 0.0259 0.0000 0.0290 0.0000 0.0318 0.0000 0.0367 0.0002 0.0016 0.0025 0.0386 0.0000 0.0401

The 42 genera with different relative abundance between dairy and beef cattle were represented by 92.2% and 87.4% of 16S sequences analyzed, respectively. Of them, 12 genera had a relative abundance greater than 1.0% in the rumen of beef steers, such as Prevotella (30.17%), Butyrivibrio (12.89%), Treponema (7.73%), Fibrobacter (7.11%), Saccharofermentans (5.43%), Succiniclasticum (5.11%), Sporobacter (4.87%), Moryella (3.61%), Pseudobutyrivibrio (2.94%), Anaerovorax (2.09%), a potentially novel genera in the family Lachnospiraceae (2.02%), and Bacteroides (1.03%). Sporobacter was a genus with a single anaerobic species isolated from the paunch of a woodfeeding termite and may contribute to the degradation of lignocellulosic matter in the digestive tract of the termite [12]. The presence of this genus in the rumen or gastrointestinal tract of ruminants has not been reported previously. Similarly, little is known about the genus Moryella, a recently described genus containing a single anaerobic species of putatively intestinal origin [5]. Our results

Figure 4: The relative composition of 2 classes in the rumen microbiome. Boxes denote the inter-quartile range between the 1st and 3rd quartiles (25 and 75%, respectively). Red: pre-ruminant dairy calves; blue: dairy cows; and green: beef steers. suggest that both genera were abundant in the rumen of beef steers. The ecological role of these bacteria in the rumen microbial ecosystem is worthy of further investigation. In the rumen of dairy cows, 3 of the 5 most abundant OTUs were related to Prevotella. Together, these 3 OTUs comprised 22.3% of the rumen microbial community in dairy cows. In the rumen of beef cattle, a single species, Butyrivibrio hungatei, accounted for 2.8% of the 16S sequences. In contrast, the abundance of this species was extremely low in the rumen of both pre-ruminant calves

Metagenomics

and dairy cows. Eight of the 10 most abundant OTUs in the rumen of dairy cows exhibited signicant differences in abundance between dairy and beef cattle. 3.4 Abomasal substrate availability had a minimal impact on the ruminal bacterial composition Two treatments were administrated per abomasum in beef steers to assess the impact of post-ruminal delivery of selected nutrients, starch and casein (N = 4, respectively). The objective of this experiment was to examine any possible systematic effect of substrate availability in the lower gastrointestinal tract on the rumen microbiome. The ruminal bacterial composition between the two groups was compared. As expected, PCA analysis failed to separate these 2 groups, suggesting that the overall bacterial community structure and composition between the 2 groups were indistinguishable (Figure 3). However, further analysis showed that the relative abundance of 7 genera displayed marginally differences between the 2 groups (P < 0.05) (Table 5). Of note, the presence of Campylobacter in the rumen was rare but reliably detectable, in agreement with the previous ndings [26]. The abundance of this genus was greater in the rumen of the beef steers with abomasal infusion of casein than with abomasal infusion of starch (Table 5). The ecological and potential pathological signicance of this observation remained unknown. On the other hand, the greater abundance of Sporobacter was seemingly associated with abomasal infusion of starch (P < 0.05). Our results suggest that down-stream infusion of substrates along the gastrointestinal tract may have a minimal effect on the bacterial composition in the rumen microbial ecosystem although nutrient recycling of nitrogen (N) could have the potential to alter the ruminal environment and cause adaptation of certain microorganisms. 4 Discussion Microorganisms in the rumen perform many important biochemical processes that are benecial to the host, including brolytic reactions that convert plant ber to short-chain fatty acids and vitamin and detoxication of secondary compounds in plants [20]. The dazzling microbial diversity and complexity of microbial interactions, both synergistic and antagonistic, in the rumen have been long appreciated [23]. As a result of co-evolution between microbial communities and their host, the rumen microbial ecosystem represents a temporal equilibrium between functional redundancy of a stable community and niche specialization. A comprehensive survey of microorganisms in the rumen is necessary to facilitate a better understanding of the biological function at the whole rumen ecosystem level. Towards this end, we investigated the bacterial composition in the rumen of both pre-ruminant dairy calves and dairy cows and beef steers using bar-coded pyrosequencing of the 16S rRNA gene.

Table 5: Substrate availability in the post-ruminal abomasum may have a minimal impact on the microbial composition in the rumen in beef steers. The numbers represent the percentage composition (mean sd; N = 4 in each group).
Genus Campylobacter Catenibacterium Dehalogenimonas Pelospora Roseburia Schwartzia Sporobacter Casein 0.046 0.03 0.006 0.00 0.006 0.00 0.000 0.00 0.305 0.14 0.020 0.01 3.603 0.76 Starch 0.007 0.01 0.000 0.00 0.000 0.00 0.010 0.01 0.092 0.05 0.000 0.00 6.144 1.92
P value

0.0392 0.0251 0.0259 0.0255 0.0303 0.0212 0.0487

It has been known that a pre-existing microbiota in pre-ruminant calves play an essential role in creating an environment favorable for the establishment of cellulolytic species [9], as well as establishing digestive functions [10]. The sequence of microbial establishment in the rumen of pre-ruminant calves has important ecological and physiological implications and recently has been investigated extensively [22]. Our previous results indicate that the rumen of pre-ruminant calves is not rudimentary because of the presence of all major functional types of microorganisms in the developing rumen. In this study, we conducted an extensive comparison in the bacterial composition between the developing and mature rumen. While the number of phyla identied in the developing rumen was signicantly smaller (12.3 1.5; mean sd) than in the mature rumen (16.21.4) (P = 0.001), the number of the genera identied was greater in the developing rumen (97.0 14.5) than in the mature rumen (75.6 11.4; P < 0.05). Interestingly, the number of OTUs at a species level in the developing rumen was much smaller (133.3 12.7) than in the mature rumen where the mean number of OTUs identied was 512.3 (39.2) (P < 0.0001). While the microbial composition in the developing rumen may not be as complex as the microbial community in the mature rumen of dairy and beef cattle, the family- and genus-level microbial diversities in the developing rumen were actually greater than in the mature rumen. These ndings were consistent with a previous report [22] that demonstrated developing rumen in pre-ruminant calves has already harbored sufcient microbial diversity to perform all major fermentation and metabolic functions. For example, sulfate is an important component in feedstuff fed to ruminants, consisting up to 1% of dry matter in grass. It has been known for many years that the rumen has considerable capacity to convert sulfate into sulde and subsequently incorporated into sulfurcontaining amino acids. The ruminal sulfate-reducing ability is strongly supported by the presence of sulfatereducing bacteria, such as species from Desulfovibrio [16].

Metagenomics

However, the relative abundance of Desulfovibrio in hayfree, sulfate-low habitats such as the developing rumen was much greater than in the mature rumen, suggesting that this group of rumen bacteria may have other biological functions in the rumen microbial ecosystem. Furthermore, plant ber was absent in the diet of pre-ruminant calves and major substrates for brolytic bacteria such as Fibrobacter and Ruminococcus are lacking in the developing rumen. These bacteria were nevertheless reliably detected in the rumen of pre-ruminant calves. This observation provided a strong piece of evidence that the developing rumen may possess sufcient microbial diversity. When substrates become available or in response to changes in the environment or diet, these bacteria can rapidly expand their populations to exploit novel niche space available. Indeed, a signicant population expansion of Fibrobacter (722 fold) and Ruminococcus (415 fold) was observed in the rumen of beef steers fed a hay-based diet, compared to in the developing rumen of pre-ruminant calves. The microbial composition in the rumen of dairy cows has been extensively surveyed using next-generation sequencing technologies [19, 24]. However, the rumen microbiota of beef cattle has not received similar attention. In this study, we conducted a detailed comparison in the rumen bacterial community composition between dairy and beef cattle. In a typical rumen microbial community of mature cattle, the mean number of taxa included approximately 16.2 phyla (1.4), 22.4 classes (2.1), 39.9 families (4.7), 75.6 genera (11.4), and 512.3 OTUs (39.2). The percentage of taxa displaying a signicant difference in the relative abundance between dairy and beef cattle ranged from 44.6% at a class level to 67.3% in an OTU level. Twenty families, 50% of all families present in a typical rumen microbial community in mature cattle, varied signicantly in abundance between dairy and beef cattle (Table 4). For example, the most abundant family in the rumen of dairy cows, Prevotellaceae, which accounted for 69% of 16S sequences, was much less in the rumen of beef cattle ( 32%), where Lachnospiraceae became the most dominant (and accounted for 34.3% of all 16S sequences). The signicant change in the abundance was also observed for several other predominant families, such as Veillonellaceae, Ruminococcaceae, Spirochaetaceae, and Fibrobacteraceae (Table 3). The 15 most abundant genera, such as Prevotella, Butyrivibrio, Treponema, Fibrobacter, and Saccharofermentans, in the descending order of the abundance, accounted for 92.9% of 16S sequences. Of them, 12 genera had signicant alterations in their relative abundance when the rumen microbial populations of dairy and beef cattle were compared. For example, like in many other gut microbial ecosystems, such as in the porcine proximal colon [26], Prevotella was the most abundant genus in the rumen of beef steers. However, its relative

abundance was much less than in the rumen of dairy cows, where it accounted for 68.5% of 16S sequences. Our ndings were quantitatively similar to a previous report in which Prevotella was reported to comprise 42% to 60% of the bacterial rRNA gene copies in lactating cows detected using quantitative PCR [36]. The signicant difference in the abundance of Prevotella between dairy and beef cattle (P = 1.60 106 ) is likely a result of their respective diets (concentrate vs hay). In agreement with several previous reports [3], the bovine rumen may harbor a much higher level of Prevotella genetic diversity because the cultured or known Prevotella species accounted for only a small portion of the observed predominance in the rumen. The genus Prevotella includes a group of bacteria with great genetic divergence and functional versatility [32]. Several species, such as P. ruminicola, are commonly thought to play a major role in initial dietary protein breakdown in the rumen [40] because they possess a rate-limiting dipeptidyl peptidase type IV. Prevotella species may contribute to ruminal brolytic capacity by acting synergistically with the cellulolytic species Fibrobacter succinogenes for efcient utilization of hemicellulose [30]. Recently, it has been found that certain Prevotella species may act as opportunistic human pathogens and are associated with dysbiosis-associated diseases [1]. Because of the fact that Prevotella may serve as a reservoir for resistance genes in the gut microbiota and its high predominance in the rumen and hindgut, this group of bacteria may deserve a special attention from the microbiological community, especially their eco-physiological roles in the rumen microbial ecosystem, relationships between the abundance and metabolic activities, and their potential pathological implications in human health. While species with a large numerical abundance obviously contribute to the rumen microbial ecosystem outputs, minor species in the rumen community may possess important and yet unrecognized ecological functions. For example, Anaerovibrio species are a group of important rumen lipolytic bacteria that are able to produce propionate via the dicarboxylic acid pathway, similarly as in propionibacteria [31]. Initial hydrolysis of ingested lipids by the lipase secreted by Anaerovibrio species must take place before biohydrogenation ensues, a process by which unsaturated fatty acids in the diet are converted to saturated fatty acids in the rumen, that eventually end up in human bodies via the consumption of meat and dairy products [23]. The relative abundance of Anaerovibrio in the rumen of steers is greater in a grass diet than in a red clover silage-based diet and is affected by sh oil supplements [18]. However, our results showed that its abundance in the rumen of beef steers fed a Timothy grass-based diet was signicantly lower than that observed in the rumen of dairy cows. The ecological implications of the observed shift remained

10

Metagenomics [6] Y. Chen, G. B. Penner, M. Li, M. Oba, and L. Guan le, Changes in bacterial diversity associated with epithelial tissue in the beef cow rumen during the transition to a high-grain diet, Appl Environ Microbiol, 77 (2011), 57705781. [7] J. E. Edwards, S. A. Huws, E. J. Kim, and A. H. Kingston-Smith, Characterization of the dynamics of initial bacterial colonization of nonconserved forage in the bovine rumen, FEMS Microbiol Ecol, 62 (2007), 323335. [8] S. C. Fernando, H. T. Purvis 2nd, F. Z. Najar, L. O. Sukharnikov, C. R. Krehbiel, T. G. Nagaraja, et al., Rumen microbial population dynamics during adaptation to a high-grain diet, Appl Environ Microbiol, 76 (2010), 74827490. [9] G. Fonty, P. Gouet, J. P. Jouany, and J. Senaud, Ecological factors determining establishment of cellulolytic bacteria and protozoa in the rumens of meroxenic lambs, J Gen Microbiol, 129 (1983), 213223. [10] G. Fonty, J. P. Jouany, M. Chavarot, F. Bonnemoy, and P. Gouet, Development of the rumen digestive functions in lambs placed in a sterile isolator a few days after birth, Reprod Nutr Dev, 31 (1991), 521528. [11] E. A. Galbraith, D. A. Antonopoulos, and B. A. White, Suppressive subtractive hybridization as a tool for identifying genetic diversity in an environmental metagenome: the rumen as a model, Environ Microbiol, 6 (2004), 928937. [12] I. Grech-Mora, M. L. Fardeau, B. K. C. Patel, B. Ollivieri, A. Rimault, G. Prensiers, et al., Isolation and characterization of sporobacter termitidis gen. nov., sp. nov., from the digestive tract of the wood-feeding termite nasutitermes lujae, Int J Syst Evol Microbiol, 46 (1996), 512518. [13] K. Gregg, Engineering gut ora of ruminant livestock to reduce forage toxicity: progress and problems, Trends Biotechnol, 13 (1995), 418421. [14] X. Hao, R. Jiang, and T. Chen, Clustering 16S rRNA for OTU prediction: a method of unsupervised Bayesian clustering, Bioinformatics, 27 (2011), 611618. [15] M. Hess, A. Sczyrba, R. Egan, T. W. Kim, H. Chokhawala, G. Schroth, et al., Metagenomic discovery of biomass-degrading genes and genomes from cow rumen, Science, 331 (2011), 463 467. [16] B. H. Howard and R. E. Hungate, Desulfovibrio of the sheep rumen, Appl Environ Microbiol, 32 (1976), 598602. [17] S. M. Huse, L. Dethlefsen, J. A. Huber, D. Mark Welch, D. A. Relman, and M. L. Sogin, Exploring microbial diversity and taxonomy using SSU rRNA hypervariable tag sequencing, PLoS Genet, 4 (2008), e1000255. [18] S. A. Huws, M. R. Lee, S. M. Muetzel, M. B. Scott, R. J. Wallace, and N. D. Scollan, Forage type and sh oil cause shifts in rumen bacterial diversity, FEMS Microbiol Ecol, 73 (2010), 396407. [19] E. Jami and I. Mizrahi, Composition and similarity of bovine rumen microbiota across individual animals, PLoS One, 7 (2012), e33306. [20] R. John Wallace, Gut microbiology broad genetic diversity, yet specic metabolic niches, Animal, 2 (2008), 661668. [21] M. Kim, M. Morrison, and Z. Yu, Status of the phylogenetic diversity census of ruminal microbiomes, FEMS Microbiol Ecol, 76 (2011), 4963. [22] R. W. Li, E. E. Connor, C. Li, R. L. Baldwin 6th, and M. E. Sparks, Characterization of the rumen microbiota of preruminant calves using metagenomic tools, Environ Microbiol, 14 (2012), 129139. [23] R. W. Li, M. Sparks, and E. E. Connor, Dynamics of the rumen microbiota, in Metagenomics and its Applications in Agriculture, Biomedicine and Environmental Studies, R. W. Li, ed., Nova Science, New York, 2011, 135164. [24] R. W. Li, S. Wu, R. L. Baldwin 6th, W. Li, and C. Li, Perturbation dynamics of the rumen microbiota in response to exogenous butyrate, PLoS One, 7 (2012), e29392.

unknown. Intriguingly, Acetobacter and Methylobacterium were moderate in their relative abundance but persistent the rumen of dairy cows. However, these genera were not detected in beef steers. Methylobacterium species are able to utilize oxalate, widespread two-carbon compounds produced in large quantity by certain plants [34]. Oxalate is toxic to mammals; and some rumen microorganisms in the rumen and colon in mammals possess the ability to degrade oxalate. Additionally, Methylobacterium has been shown to degrade nitroaromatic explosives, such as RDX [38]. Acetobacter species belong to a group of acetic acid bacteria and are aerobic. Acetobacter can be readily found in nature, especially in the alcohol-enriched environment [33]. At least one Acetobacter species is known to synthesize crystalline cellulose [42]. However, little is known about these 2 genera of rumen origin. In conclusion, we investigated the rumen microbial community composition and diversity in the bovine rumen and compared the similarities and differences of the rumen microbiota between the developing and mature rumen, as well as between dairy and beef cattle. The rumen microbiota consists of a complex community of microorganisms with a myriad of microbial interactions. Our ndings provide a glimpse of the dazzling microbial diversity in the rumen and will undoubtedly contribute to the understanding of ruminal function and ruminant nutrition.
Acknowledgments The authors thank Ashley Sperling and Alicia Beavers for their excellent technical assistance. Mention of trade names or commercial products in this publication is solely for the purpose of providing specic information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. The USDA is an equal opportunity provider and employer. Sitao Wu and Weizhong Li were partly supported by Award R01HG005978 to Weizhong Li from the National Human Genome Research Institute (NHGRI). The content is solely the responsibility of the authors and does not necessarily represent the ofcial views of the NHGRI or the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. No additional external funding was received for this study.

References
[1] C. Alauzet, H. Marchandin, and A. Lozniewski, New insights into prevotella diversity and medical microbiology, Future Microbiol, 5 (2010), 16951718. [2] A. F. Andersson, M. Lindberg, H. Jakobsson, F. B ckhed, a P. Nyr n, and L. Engstrand, Comparative analysis of human gut e microbiota by barcoded pyrosequencing, PLoS One, 3 (2008), e2836. [3] A. Z. Bekele, S. Koike, and Y. Kobayashi, Genetic diversity and diet specicity of ruminal prevotella revealed by 16s rrna genebased analysis, FEMS Microbiol Lett, 305 (2010), 4957. [4] Y. Cai and Y. Sun, Esprit-tree: hierarchical clustering analysis of millions of 16s rrna pyrosequences in quasilinear computational time, Nucleic Acids Res, 39 (2011), e95. [5] J. P. Carlier, G. Kouas, and X. Y. Han, Moryella indoligenes gen. nov., sp. nov., an anaerobic bacterium isolated from clinical specimens, Int J Syst Evol Microbiol, 57 (2007), 725729.

Metagenomics [25] R. W. Li, S. Wu, W. Li, Y. Huang, and L. C. Gasbarre, Metagenome plasticity of the bovine abomasal microbiota in immune animals in response to ostertagia ostertagi infection, PLoS One, 6 (2011), e24417. [26] R. W. Li, S. Wu, W. Li, K. Navarro, R. D. Couch, D. Hill, et al., Alterations in the porcine colon microbiota induced by the gastrointestinal nematode Trichuris suis, Infect Immun, 80 (2012). [27] N. R. McEwan, L. Abecia, M. Regensbogenova, C. L. Adam, P. A. Findlay, and C. J. Newbold, Rumen microbial population dynamics in response to photoperiod, Lett Appl Microbiol, 41 (2005), 97101. [28] R. J. Michelland, V. Monteils, S. Combes, L. Cauquil, T. Gidenne, and L. Fortun-Lamothe, Changes over time in the bacterial communities associated with uid and food particles and the ruminal parameters in the bovine rumen before and after a dietary change, Can J Microbiol, 57 (2011), 629637. [29] B. Niu, Z. Zhu, L. Fu, S. Wu, and W. Li, FR-HIT, a very fast program to recruit metagenomic reads to homologous reference genomes, Bioinformatics, 27 (2011), 17041705. [30] J. M. Osborne and B. A. Dehority, Synergism in degradation and utilization of intact forage cellulose, hemicellulose, and pectin by three pure cultures of ruminal bacteria, Appl Environ Microbiol, 55 (1989), 22472250. [31] R. A. Prins, A. Lankhorst, P. van der Meer, and C. J. Van Nevel, Some characteristics of anaerovibrio lipolytica a rumen lipolytic organism, Antonie Van Leeuwenhoek, 41 (1975), 111. [32] A. Ramsak, M. Peterka, K. Tajima, J. C. Martin, J. Wood, M. E. Johnston, et al., Unravelling the genetic diversity of ruminal bacteria belonging to the CFB phylum, FEMS Microbiol Ecol, 33 (2000), 6979. [33] P. Raspor and D. Goranovic, Biotechnological applications of acetic acid bacteria, Crit Rev Biotechnol, 28 (2008), 101124. [34] N. Sahin, Y. Kato, and F. Yilmaz, Taxonomy of oxalotrophic Methylobacterium strains, Naturwissenschaften, 95 (2008), 931 938. [35] L. Skillman, P. Evans, G. Naylor, B. Morvan, G. Jarvis, and K. Joblin, 16s ribosomal DNA-directed PCR primers for ruminal methanogens and identication of methanogens colonising young lambs, Anaerobe, 10 (2004), 277285. [36] D. M. Stevenson and P. J. Weimer, Dominance of prevotella and low abundance of classical ruminal bacterial species in the bovine rumen revealed by relative quantication real-time PCR, Appl Microbiol Biotechnol, 75 (2007), 165174. [37] Y. Sun, Y. Cai, S. M. Huse, R. Knight, W. G. Farmerie, X. Wang, et al., A large-scale benchmark study of existing algorithms for taxonomy-independent microbial community analysis, Brief Bioinform, 13 (2012), 107121. [38] B. Van Aken, J. M. Yoon, and J. L. Schnoor, Biodegradation of nitro-substituted explosives 2,4,6-trinitrotoluene, hexahydro1,3,5-trinitro-1,3,5-triazine, and octahydro-1,3,5,7-tetranitro1,3,5-tetrazocine by a phytosymbiotic Methylobacterium sp. associated with poplar tissues (Populus deltoides x nigra DN34), Appl Environ Microbiol, 70 (2004), 508517. [39] C. Vilo and Q. Dong, Evaluation of the RDP classier accuracy using 16s rRNA gene variable regions, Metagenomics, 1 (2012), Article ID 235551. [40] R. J. Wallace, Ruminal microbial metabolism of peptides and amino acids, J Nutr, 126 (1996), 1326S1334S. [41] Q. Wang, G. M. Garrity, J. M. Tiedje, and J. R. Cole, Naive Bayesian classier for rapid assignment of rRNA sequences into the new bacterial taxonomy, Appl Environ Microbiol, 73 (2007), 52615267. [42] P. J. Weimer, J. M. Hackney, H. J. Jung, and R. D. Hateld, Fermentation of a bacterial cellulose/xylan composite by mixed ruminal microora: implications for the role of polysaccharide

11 matrix interactions in plant cell wall biodegradability, J Agric Food Chem, 48 (2000), 17271733. [43] S. Wu, R. W. Li, W. Li, E. Beshah, H. D. Dawson, and J. F. Urban Jr, Worm burden-dependent disruption of the porcine colon microbiota by Trichuris suis infection, PLoS One, 7 (2012), e35470. [44] S. Wu, Z. Zhu, L. Fu, B. Niu, and W. Li, WebMGA: a customizable web server for fast metagenomic sequence analysis, BMC Genomics, 12 (2011), 444.