Search
Author
Title
Vol.
Issue
Year
1st Page

Journal of Animal Science - Animal Nutrition

Investigation of bacterial diversity in the feces of cattle fed different diets1

 

This article in JAS

  1. Vol. 92 No. 2, p. 683-694
     
    Received: June 28, 2013
    Accepted: Nov 19, 2013
    Published: November 24, 2014


    2 Corresponding author(s): Jim.Wells@ars.usda.gov
 View
 Download
 Share

doi:10.2527/jas.2013-6841
  1. M. Kim*,
  2. J. Kim,
  3. L. A. Kuehn*,
  4. J. L. Bono*,
  5. E. D. Berry*,
  6. N. Kalchayanand*,
  7. H. C. Freetly*,
  8. A. K. Benson and
  9. J. E. Wells 2
  1. USDA, ARS, US Meat Animal Research Center, Clay Center, NE 689333
    Department of Food Science and Technology, University of Nebraska – Lincoln, Lincoln 68583

Abstract

The objective of this study is to investigate individual animal variation of bovine fecal microbiota including as affected by diets. Fecal samples were collected from 426 cattle fed 1 of 3 diets typically fed to feedlot cattle: 1) 143 steers fed finishing diet (83% dry-rolled corn, 13% corn silage, and 4% supplement), 2) 147 steers fed late growing diet (66% dry-rolled corn, 26% corn silage, and 8% supplement), and 3) 136 heifers fed early growing diet (70% corn silage and 30% alfalfa haylage). Bacterial 16S rRNA gene amplicons were determined from individual fecal samples using next-generation pyrosequencing technology. A total of 2,149,008 16S rRNA gene sequences from 333 cattle with at least 2,000 sequences were analyzed. Firmicutes and Bacteroidetes were dominant phyla in all fecal samples. At the genus level, Oscillibacter, Turicibacter, Roseburia, Fecalibacterium, Coprococcus, Clostridium, Prevotella, and Succinivibrio were represented by more than 1% of total sequences. However, numerous sequences could not be assigned to a known genus. Dominant unclassified groups were unclassified Ruminococcaceae and unclassified Lachnospiraceae that could be classified to a family but not to a genus. These dominant genera and unclassified groups differed (P < 0.001) with diets. A total of 176,692 operational taxonomic units (OTU) were identified in combination across all the 333 cattle. Only 2,359 OTU were shared across 3 diet groups. UniFrac analysis showed that bacterial communities in cattle feces were greatly affected by dietary differences. This study indicates that the community structure of fecal microbiota in cattle is greatly affected by diet, particularly between forage- and concentrate-based diets.



INTRODUCTION

The gastrointestinal tract serves as a habitat for a diverse and dynamic population of bacterial species that can affect growth, health, and well-being of the host. The fecal microbiota of cattle affects not only animal health but also food safety (Shanks et al., 2011). Therefore, a better understanding of the fecal microbiota structure would be important to reduce foodborne pathogens through dietary changes.

Traditional culture-based studies have allowed for the isolation of easy-to-grow bacterial strains that have contributed to underpinning metabolic functions of the fecal microbiota. However, isolates recovered from culture-based studies typically represent only a small portion of the total microbial population in an environment (Janssen, 2006).

The use of 16S rRNA gene (rrs) sequence data (Woese et al., 1983) has identified numerous unculturable microorganisms. Recently, an rrs amplification method that uses pyrosequencing was applied to compare the community structure of fecal microbiota in cattle that were fed different diets, indicating that diet greatly influences the fecal microbiota of cattle (Callaway et al., 2010; Shanks et al., 2011; Rice et al., 2012). In addition, the community structure of fecal microbiota was different among individual cattle fed the same diet (Dowd et al., 2008; Durso et al., 2010).

Although these studies provided insight into the community structure of fecal microbiota using pyrosequencing, the number of samples analyzed in these studies was limited. Lack of power in observation groups prohibits an appreciation of sources of variation, and differences in targeted regions of the rrs genes sequenced complicate comparisons between studies. In the present study, 426 fecal samples from cattle fed 3 different diets were used to analyze the community structure of the fecal microbiota using pyrosequencing. We hypothesized that the community structure of the fecal microbiota would be highly variable across animals but influenced by diet.


MATERIALS AND METHODS

Animal Diets and Fecal Sample Collection

A total of 426 cattle were used to investigate the fecal microbiota composition in cattle fed different diets (Table 1). Fecal samples were collected from 426 cattle assigned to 1 of 3 diet groups as described previously (Wells et al., 2009): 1) 143 steers were fed a late growing diet, on a DM basis, consisting of 66.0% dry-rolled corn, 26.0% corn silage, 5.85% soybean meal, and 2.15% vitamin and mineral supplement with monensin (designated as “Moderate Grain”), 2) 147 steers were fed a finishing diet, on a DM basis, consisting of 82.75% dry-rolled corn, 12.75% corn silage, and 4.5% vitamin and mineral supplement with monensin (designated as “High Grain”), and 3) 136 heifers were fed an early growing diet, on a DM basis, consisting of 70% corn silage and 30% alfalfa haylage (designated as “Silage/Forage”). All cattle were adapted to the specified diets for 1 mo before fecal sampling. Fecal samples of cattle fed Moderate Grain were collected twice (July and August 2009). The 2 samples from each animal were subjected to pyrosequencing analysis separately, and the resultant 2 sequence datasets obtained from each animal were combined into 1 sequence dataset. Fecal samples from cattle fed High Grain or Silage/Forage were collected every 2 wk for a period of 10 wk between June and September 2010, and the 6 samples from each animal were pooled into 1 composite sample for pyrosequencing analysis.


View Full Table | Close Full ViewTable 1.

Ingredient composition as percentage of diet (DM basis)1

 
Item Late growing diet (Moderate Grain) Finishing diet (High Grain) Early growing diet (Silage/Forage)
Dry-rolled corn 66 82.75
Corn silage 26 12.75 70
Soybean meal 5.85
Alfalfa haylage 30
Supplement 12 2.15
Supplement 23 4.5
1All diets were formulated to exceed National Research Council recommendations for cattle (NRC, 1996).
2Custom supplement included (% DM) 62.55% limestone, 2.38% NaCl, 32.63% urea, 0.93% trace mineral mix (13% Ca, 12% Zn, 8% Mn, 10% Zn, 1.5% Cu, 0.2% I, and 0.1% Co), 0.56% vitamin mix (A 8,818,490 IU/kg, D 881,849 IU/kg, and E 882 IU/kg), and 0.95% Rumensin-80 (Elanco Animal Health, Indianapolis, IN).
3Commercial supplement (Biegert Feeds, Bradshaw, NE) included (% DM) 32% CP (28% NPN), 7.5% Ca, 0.8% P, 4.8% NaCl, 1.8% K, 55,116 IU/kg vitamin A, 105 IU/kg vitamin E, and 0.065% sodium monensin (Elanco Animal Health, Indianapolis, IN).

Pyrosequencing

Total DNA was extracted from fecal samples using a Mini-Beadbeater-8 (BioSpec Products, Bartlesville, OK) and the QIAamp DNA Stool Mini Kit (Qiagen, Valencia, CA) as described previously (Martinez et al., 2010). Each DNA sample was amplified with universal primers 27F (5′-adaptor primer-AGAGTTTGATCMTGGCTCAG-3′) and 518R (5′-adaptor primer-barcode-ATTACCGCGGCTGCTGG-3′) targeting the V1 through V3 rrs region. Equal amounts of amplicons were pooled and gel-purified using the QIAquick Gel Extraction Kit (Qiagen) and then sequenced using the 454 GS FLX Titanium system by the Core for Applied Genomics and Ecology (University of Nebraska).

Sequence Processing and Phylogenetic Analysis

All sequence processing and phylogenetic analysis were conducted using the programs in QIIME software package 1.4.0 (Caporaso et al., 2010b). After barcode and primer sequences were trimmed off, sequences that had read lengths shorter than 200 nucleotides, mean quality score below Q25, and homopolymer stretches longer than 8 nucleotides were removed. The remaining sequences were aligned using PyNAST against the Greengenes Core reference alignment (DeSantis et al., 2006; Caporaso et al., 2010a), and sequences that could not be aligned were removed. Chimeric sequences were checked using the ChimeraSlayer program (Haas et al., 2011). Fecal samples represented by less than 2,000 cleaned sequences were excluded from analysis. All cleaned sequences across the 3 diet groups were merged and then classified into taxa using the Ribosomal Database Project (RDP) naïve Bayesian rRNA Classifier 2.2 (Wang et al., 2007). Percentage of total sequences was used to compare taxa among the 3 diet groups. Operational taxonomic units (OTU) were calculated at 0.03 dissimilarity using the uclust method (Edgar, 2010). The number of OTU was normalized by randomly subsampling 2,000 sequences from each fecal sample. A phylogenetic tree was built using FastTree 2.1.3 (Price et al., 2010) to calculate α diversity and β diversity indices.

Statistical Analysis

This study was done to better understand the variation in bacterial community structure and composition among animals. As such, individual animal was used as the experimental unit. To characterize variation across diet cohort groups, observational differences in the experimental units were analyzed. The mean abundance of each taxon was compared among the 3 diet groups using 1-way ANOVA followed by the Tukey’s test in SAS (SAS Inst. Inc., Cary, NC). Significant difference was determined at P < 0.05. Principal coordinates analysis (PCoA) was performed using both the weighted and the unweighted UniFrac methods (Lozupone and Knight, 2005).


RESULTS

Collective Data Summary

Of samples from all 426 animals, 333 samples represented by ≥2,000 cleaned rrs sequences were used for phylogenetic analysis. A total of 2,149,008 cleaned rrs sequences with 500-nucleotide average read length were obtained from the 333 samples composed of 142 Moderate Grain samples (1,035,186 sequences), 130 High Grain samples (877,232 sequences), and 61 Silage/Forage samples (236,590 sequences; Table 2). The number of cleaned rrs sequences in individual samples ranged from 2,004 to 14,770. The 2,149,008 rrs sequences were classified into 21 phyla, 40 classes, 71 orders, 152 families, and 434 genera. Firmicutes and Bacteroidetes represented 63.41 and 23.13% of all cleaned rrs sequences, respectively (Table 3). The rest of the phyla each represented less than 2% of all cleaned rrs sequences except for Proteobacteria (3.79%). Candidate division TM7 and Actinobacteria accounted for 1.56 and 1.17% of all cleaned rrs sequences, respectively. Acidobacteria, Chloroflexi, Cyanobacteria, Deferribacteres, Deinococcus-Thermus, Fibrobacteres, Fusobacteria, Gemmatimonadetes, Lentisphaerae, Nitrospira, Planctomycetes, candidate division SR1, Spirochaetes, Synergistetes, Tenericutes, and Verrucomicrobia represented less than 1.00% of all cleaned rrs sequences. However, about 6% of all cleaned rrs sequences could not be classified into a known phylum and were found across all the 333 samples. Abundant genera that represented more than 0.50% of all the 2,149,008 rrs sequences were Prevotella (7.82%), Oscillibacter (5.21%), Turicibacter (4.47%), Roseburia (3.58%), Fecalibacterium (2.65%), Coprococcus (2.37%), Succinivibrio (2.36%), Clostridium (1.90%), Lactobacillus (0.89%), Blautia (0.81%), Bacteroides (0.72%), Parabacteroides (0.64%), and Anaerovibrio (0.56%).


View Full Table | Close Full ViewTable 2.

Diversity statistics

 
Maximum no. of OTU
Diet type Sample type No. of cleaned sequences No. of observed OTU1 Chao1 ACE Shannon diversity index
Moderate Grain (n = 142) Pooled sample 1,035,186 109,984 214,992 303,111 8.64
Range of individual samples 3,132–11,702 1,432–5,048 4,048–20,171 7,165–46,185 6.33–7.67
Range of individual subsamples2 2,000 1,184 ± 11a 5,734 ± 90a 13,929 ± 254a 6.53 ± 0.03a
High Grain (n = 130) Pooled sample 877,232 67,469 167,616 278,239 7.01
Range of individual samples 2,055–14,770 584–3,643 1,101–11,977 1,330–23,750 4.34–7.00
Range of individual subsamples2 2,000 690 ± 12c 2,253 ± 94c 4,382 ± 265c 5.40 ± 0.03c
Silage/Forage (n = 61) Pooled sample 236,590 35,341 101,211 183,344 7.49
Range of individual samples 2,004–13,598 632–4,238 1,575–15,101 2,543–35,795 4.15–7.39
Range of individual subsamples2 2,000 943 ± 17b 3,570 ± 138b 7,566 ± 387b 5.88 ± 0.05b
a,b,cWithin a column, means for the range of individual subsamples having differing superscripts differ (P < 0.05).
1OTU = operational taxonomic units.
2Means among 3 diet groups were compared using ANOVA followed by the Tukey’s test.

View Full Table | Close Full ViewTable 3.

Relative abundance of dominant taxa in 3 diet groups

 
Percentage of total sequences1
Classification Collective data2 Moderate Grain High Grain Silage/Forage SEM P-value3 No. of cattle with detectable taxon4
Firmicutes 64.33 50.31c 76.91a 70.15b 0.82 0.001 333
    Oscillibacter 5.20 4.57b 8.10a 0.65c 0.21 0.001 332
    Turicibacter 4.62 1.26c 8.40a 4.37b 0.29 0.001 333
    Roseburia 3.39 3.94a 4.21a 0.38b 0.11 0.001 327
    Fecalibacterium 2.43 4.06a 1.78b 0.05c 0.14 0.001 274
    Coprococcus 2.33 2.37b 2.94a 0.95c 0.07 0.001 330
    Clostridium 1.86 1.13b 2.92a 1.29b 0.07 0.001 333
    Sporacetigenium 0.40 0.19c 0.36b 1.01a 0.02 0.001 332
    Blautia 0.78 0.22b 1.73a 0.06b 0.06 0.001 294
    Lactobacillus 0.72 0.27b 1.50a 0.13b 0.09 0.001 315
    Subdoligranulum 0.35 0.29b 0.56a 0.05c 0.02 0.001 273
    Microbacterium 0.35 0.05b 0.54a 0.63a 0.04 0.001 188
    Anaerovibrio 0.52 0.96a 0.25b 0.05c 0.03 0.001 275
    Anaerovorax 0.31 0.12c 0.31b 0.75a 0.02 0.001 318
    Ruminococcus 0.24 0.11b 0.32a 0.39a 0.02 0.001 325
    Unclassified Ruminococcaceae 15.19 8.12c 14.63b 32.86a 0.59 0.001 333
    Unclassified Lachnospiraceae 12.15 12.98a 12.71a 9.06b 0.19 0.001 333
    Unclassified Clostridiales 6.78 5.78c 6.75b 9.20a 0.13 0.001 333
    Unclassified Peptostreptococcaceae 2.57 0.68c 4.40a 3.07b 0.14 0.001 333
    Unclassified Erysipelotrichaceae 0.45 0.23c 0.41b 1.03a 0.02 0.001 333
    Unclassified Veillonellaceae 0.31 0.60a 0.12b 0.05b 0.02 0.001 258
    Unclassified Clostridia 0.39 0.20b 0.57a 0.45a 0.02 0.001 332
    Unclassified Clostridiaceae 0.40 0.48a 0.31b 0.39b 0.01 0.001 330
    Unclassified Firmicutes 1.16 0.90c 1.06b 1.96a 0.03 0.001 333
Bacteroidetes 21.28 37.39a 12.82b 1.83c 0.89 0.001 333
    Prevotella 6.99 14.39a 2.15b 0.09c 0.42 0.001 318
    Bacteroides 0.72 0.97a 0.77a 0.06b 0.05 0.001 277
    Parabacteroides 0.61 0.88a 0.59b 0.05c 0.03 0.001 274
    Unclassified Prevotellaceae 5.89 10.69a 3.37b 0.07c 0.29 0.001 301
    Unclassified Porphyromonadaceae 2.23 2.84a 2.55a 0.12b 0.11 0.001 301
    Unclassified Bacteroidales 1.95 3.44a 1.09b 0.32c 0.09 0.001 323
    Unclassified Bacteroidetes 2.67 3.81a 2.15b 1.16c 0.13 0.001 331
Proteobacteria 3.53 6.04a 1.30b 2.43b 0.21 0.001 333
    Succinivibrio 2.08 4.46a 0.43b 0.05b 0.14 0.001 267
    Pantoea 0.28 0.58a 0.05b 0.05b 0.03 0.001 111
TM7 2.32 0.24b 0.56b 10.92a 0.24 0.001 329
    TM7 genera incertae sedis 2.32 0.24b 0.56b 10.92a 0.24 0.001 329
Actinobacteria 1.28 0.14c 1.77b 2.93a 0.11 0.001 330
    Propionibacterium 0.18 0.05b 0.08b 0.68a 0.04 0.001 123
    Unclassified Coriobacteriaceae 0.20 0.06c 0.22b 0.51a 0.01 0.001 269
Cyanobacteria 0.54 1.14a 0.09b 0.09b 0.06 0.001 189
    Streptophyta 0.54 1.14a 0.09b 0.09b 0.06 0.001 189
Verrucomicrobia 0.27 0.06b 0.10b 1.15a 0.04 0.001 123
    Akkermansia 0.27 0.06b 0.10b 1.14a 0.04 0.001 111
Unclassified Bacteria 6.22 4.51c 6.28b 10.07a 0.17 0.001 333
a,b,cWithin a row, means with a different subscript were different (P < 0.05).
1Values indicate means.
2Sequences obtained from all 333 fecal samples.
3A P-value of 0.001 is less than or equal to 0.001.
4Percentage of total sequences for cattle with nondetectable taxon was treated as 0.05%.

A total of 176,692 OTU were calculated at a 0.03 dissimilarity cutoff in combination across all the 333 samples. The number of singletons was 98,526 and accounted for 56% of all the OTU. The number of OTU shared across the 3 diet groups was only 2,359 whereas the number of OTU specific to only 1 of the 3 diet groups was 145,299 that consisted of 77,444 (Moderate Grain), 36,788 (High Grain), and 28,717 (Silage/Forage), respectively (Fig. 1A). To compare OTU shared among the 3 diet groups, we normalized the 3 diet groups based on the smallest number of sequences in Silage/Forage (236,590 sequences × 3 dietary groups = 709,770 sequences; Fig. 1B). The number of shared OTU was greater between Moderate Grain and High Grain than between Silage/Forage and Moderate Grain (or High Grain; Fig. 1B). The community structure of the fecal microbiota in Silage/Forage appears to be distinct compared to Moderate Grain and High Grain.

Figure 1.
Figure 1.

Venn diagrams showing the number of observed operational taxonomic units (OTU) shared across 3 diet groups. A) Only 2,359 OTU were shared in combination across 3 diet groups, and each group was represented by numerous unique OTU. B) The number of observed OTU was normalized based on the smallest number of sequences in Silage/Forage. The number of shared OTU was lower between Silage/Forage and Moderate Grain (or High Grain) than between Moderate Grain and High Grain. See online version for figure in color.

 

Taxonomic Composition of the Fecal Microbiota

Taxonomic composition of the fecal microbiota among the 3 diet groups was compared based on the mean of the relative abundance (reads of a taxon/total reads in a sample) in the 3 diet groups as described previously (Benson et al., 2010). Firmicutes was the first dominant phylum in all the 3 diet groups as shown in collective data. The abundance of Firmicutes was different (P < 0.001) among the 3 diet groups with the greatest percentage (76.9%) in High Grain and the lowest percentage (50.3%) in Moderate Grain (Table 3). Only 13 genera within Firmicutes represented ≥0.5% of all the sequences in at least 1 diet group. The abundance of Oscillibacter, Turicibacter, Coprococcus, Clostridium, Blautia, Lactobacillus, and Subdoligranulum was different (P < 0.001) among the 3 diet groups with the High Grain diet having the greatest abundance. The abundance of Roseburia was greater (P < 0.001) in High Grain than in Silage/Forage but not different from Moderate Grain. The abundance of Microbacterium was greater (P < 0.001) in High Grain than in Moderate Grain but not different from Silage/Forage. The abundance of Fecalibacterium and Anaerovibrio was different (P < 0.001) among the 3 diet groups with the Moderate Grain diet having the greatest abundance. The abundance of Sporacetigenium and Anaerovorax was different (P < 0.001) among the 3 diet groups with the Silage/Forage diet having the greatest abundance. However, most of the Firmicutes sequences could not be classified into a known genus in all 3 groups. Three dominant unclassified groups in collective data were unclassified Ruminococcaceae (15.19%), unclassified Lachnospiraceae (12.15%), and unclassified Clostridiales (6.78%) as shown in the composition of rumen microbiota (Kim et al., 2011). These 3 unclassified groups represented approximately 50% of all sequences in Silage/Forage. The abundance of unclassified Ruminococcaceae and unclassified Clostridiales was different (P < 0.001) among the 3 diet groups with the Silage/Forage diet having the greatest abundance whereas that of unclassified Lachnospiraceae was lower (P < 0.001) in Silage/Forage than in the other 2 groups. The rest of unclassified groups did not represent more than 1.20% of all sequences in collective data.

Bacteroidetes was the second dominant phylum in both Moderate Grain and High Grain whereas it was the fourth dominant in Silage/Forage and represented only 1.83% of all the sequences (Table 3). In Silage/Forage, Firmicutes was the first dominant phylum followed by TM7 and Proteobacteria. Prevotella was the most dominant among all known genera and represented 6.99% of all sequences in collective data. The abundance of Prevotella was different (P < 0.001) among the 3 diet groups and the greatest (14.46%) in Moderate Grain. The rest of the known genera within Bacteroidetes did not represent ≥0.5% of all sequences except for Bacteroides and Parabacteroides, representing 0.7 and 0.6% of all sequences, respectively. Both Bacteroides and Parabacteroides were lower in Silage/Forage than in the other 2 groups. Many Bacteroidetes sequences also could not be classified into a known genus, and unclassified Prevotellaceae was the first dominant group among unclassified groups. Unclassified Prevotellaceae, unclassified Bacteroidales, and unclassified Bacteroidetes were different (P < 0.001) among the 3 diet groups and the greatest in Moderate Grain with the Silage/Forage diet having the lowest abundance.

Of 19 minor phyla, only Proteobacteria, TM7, Actinobacteria, Cyanobacteria, and Verrucomicrobia represented ≥0.5% in at least 1 diet group (Table 3). The abundance of Proteobacteria and Cyanobacteria was greater (P < 0.001) in Moderate Grain than in the other 2 groups whereas the abundance of TM7, Actinobacteria, and Verrucomicrobia was greater (P < 0.001) in Silage/Forage than in the other 2 groups. At the genus level, the abundance of Succinivibrio and Streptophyta were greater (P < 0.001) in Moderate Grain than in the other 2 groups whereas the abundance of Akkermansia was greater (P < 0.001) in Silage/Forage than in the other 2 groups.

Analysis of Bacterial Diversity

The 333 individual subsamples with 2,000 sequences (333 × 2,000 = 666,000) were obtained to compare diversity of the fecal microbiota among the 3 diet groups. We calculated species-level OTU at a 0.03 dissimilarity cutoff from the 333 subsamples to estimate diversity of the fecal microbiota in the 3 diet groups. We identified a total of 90,386 OTU in combination across all the 333 subsamples. Fifty-six of the 90,386 OTU represented ≥0.5% of all sequences in at least 1 diet group, and 33 of the 56 OTU could not be classified to a known genus (Table 4).


View Full Table | Close Full ViewTable 4.

Dominant operational taxonomic units (OTU) calculated from 333 subsamples in 3 diet groups

 
Percentage of total sequences1
No. OTU ID2 Classification Collective data3 Moderate Grain High Grain Silage/Forage SEM P-value4 No. of cattle with detectable taxon5
OTU-1153 Unclassified Bacteria 0.14 0.05b 0.05b 0.54a 0.02 0.001 96
OTU-1157 Unclassified Bacteria 0.36 0.08b 0.81a 0.05b 0.06 0.001 192
OTU-1158 Unclassified Bacteria 0.44 0.15b 0.93a 0.05b 0.04 0.001 263
OTU-1209 Propionibacterium 0.15 0.05b 0.07b 0.55a 0.04 0.001 98
OTU-1646 Unclassified Bacteroidetes 0.15 0.05b 0.05b 0.59a 0.06 0.001 29
OTU-1648 Unclassified Bacteroidetes 0.51 0.65a 0.59a 0.05b 0.03 0.001 268
OTU-1649 Unclassified Bacteroidetes 0.55 0.86a 0.44b 0.05c 0.05 0.001 256
OTU-1194 Microbacterium 0.34 0.05b 0.54a 0.62a 0.04 0.001 181
OTU-2601 Unclassified Porphyromonadaceae 0.29 0.20b 0.52a 0.05b 0.03 0.001 234
OTU-2602 Unclassified Porphyromonadaceae 0.39 0.54a 0.38b 0.05c 0.02 0.001 257
OTU-3869 Unclassified Prevotellaceae 0.37 0.72a 0.15b 0.05b 0.03 0.001 238
OTU-3870 Unclassified Prevotellaceae 0.39 0.05b 0.93a 0.05b 0.07 0.001 97
OTU-3872 Unclassified Prevotellaceae 0.65 1.39a 0.12b 0.05b 0.05 0.001 227
OTU-5060 Prevotella 0.35 0.70a 0.09b 0.05b 0.02 0.001 221
OTU-5061 Prevotella 0.39 0.77a 0.13b 0.05b 0.03 0.001 247
OTU-5062 Prevotella 0.68 1.33a 0.27b 0.05b 0.05 0.001 258
OTU-5135 Streptophyta 0.49 1.04a 0.08b 0.08b 0.06 0.001 171
OTU-5500 Lactobacillus 0.57 0.19b 1.22a 0.08b 0.08 0.001 296
OTU-7122 Unclassified Clostridiales 0.26 0.07c 0.16b 0.90a 0.02 0.001 320
OTU-7331 Clostridium 1.12 0.45b 2.19a 0.39b 0.06 0.001 331
OTU-9520 Unclassified Lachnospiraceae 0.15 0.05b 0.06b 0.57a 0.02 0.001 71
OTU-9522 Unclassified Lachnospiraceae 0.16 0.05b 0.05b 0.64a 0.02 0.001 117
OTU-9527 Unclassified Lachnospiraceae 0.30 0.19b 0.52a 0.11b 0.03 0.001 306
OTU-9528 Unclassified Lachnospiraceae 0.36 0.29b 0.57a 0.05c 0.03 0.001 259
OTU-9529 Unclassified Lachnospiraceae 0.40 0.05b 0.06b 1.94a 0.05 0.001 134
OTU-9530 Unclassified Lachnospiraceae 0.43 0.75a 0.26b 0.05c 0.02 0.001 263
OTU-9531 Unclassified Lachnospiraceae 1.65 0.86b 3.25a 0.05c 0.12 0.001 274
OTU-9760 Coprococcus 0.15 0.05b 0.06b 0.58a 0.02 0.001 106
OTU-9761 Coprococcus 0.36 0.58a 0.27b 0.05c 0.02 0.001 265
OTU-9762 Coprococcus 1.10 0.89b 1.82a 0.05c 0.06 0.001 272
OTU-10257 Roseburia 0.36 0.11b 0.77a 0.05b 0.04 0.001 263
OTU-10377 Unclassified Peptostreptococcaceae 0.28 0.09c 0.35b 0.55a 0.02 0.001 330
OTU-10378 Unclassified Peptostreptococcaceae 1.88 0.39c 3.79a 1.29b 0.13 0.001 333
OTU-12960 Unclassified Ruminococcaceae 0.19 0.05b 0.05b 0.82a 0.02 0.001 61
OTU-12962 Unclassified Ruminococcaceae 0.23 0.06b 0.10b 0.92a 0.03 0.001 201
OTU-12965 Unclassified Ruminococcaceae 0.25 0.17b 0.13b 0.67a 0.02 0.001 289
OTU-12966 Unclassified Ruminococcaceae 0.29 0.07b 0.63a 0.05b 0.02 0.001 248
OTU-12968 Unclassified Ruminococcaceae 0.35 0.09b 0.78a 0.05b 0.03 0.001 264
OTU-12969 Unclassified Ruminococcaceae 0.54 0.06b 0.07b 2.69a 0.07 0.001 158
OTU-12970 Unclassified Ruminococcaceae 0.53 0.09b 1.23a 0.05b 0.05 0.001 249
OTU-12971 Unclassified Ruminococcaceae 0.60 0.20b 1.25a 0.16b 0.04 0.001 327
OTU-12972 Unclassified Ruminococcaceae 0.61 0.67b 0.82a 0.05c 0.03 0.001 271
OTU-12973 Unclassified Ruminococcaceae 1.50 0.05b 0.12b 7.81a 0.21 0.001 159
OTU-13387 Fecalibacterium 1.02 1.38a 1.09b 0.05c 0.06 0.001 274
OTU-13994 Oscillibacter 0.27 0.15b 0.52a 0.05c 0.02 0.001 270
OTU-13995 Oscillibacter 0.33 0.14b 0.66a 0.05b 0.03 0.001 267
OTU-13996 Oscillibacter 0.36 0.33b 0.54a 0.05c 0.02 0.001 266
OTU-13997 Oscillibacter 0.44 0.16b 0.91a 0.05b 0.03 0.001 267
OTU-13998 Oscillibacter 1.00 0.29b 2.23a 0.05b 0.08 0.001 271
OTU-14605 Turicibacter 3.56 0.71c 6.72a 3.44b 0.24 0.001 333
OTU-14737 Helicobacter 0.26 0.05b 0.05b 1.18a 0.14 0.001 19
OTU-14977 Succinivibrio 1.26 2.60a 0.37b 0.05b 0.08 0.001 264
OTU-15034 Pantoea 0.24 0.50a 0.05b 0.05b 0.03 0.001 108
OTU-15452 TM7 genera incertae sedis 0.20 0.05b 0.08b 0.81a 0.02 0.001 171
OTU-15453 TM7 genera incertae sedis 0.22 0.05b 0.05b 0.94a 0.02 0.001 131
OTU-15486 Akkermansia 0.17 0.05b 0.10b 0.60a 0.03 0.001 84
a,b,cWithin a row, means with a different subscript were different (P < 0.05).
1Values represent means.
2A total of 176,692 OTU were numbered in serial order.
3Sequences obtained from all 333 fecal samples.
4A P-value of 0.001 is less than or equal to 0.001.
5Percentage of total sequences for cattle with nondetectable taxon was treated as 0.05%.

One OTU (OTU-14605) classified to Turicibacter was the most dominant among the 56 OTU in collective data, and its abundance was the greatest (P < 0.001) in High Grain and the lowest (P < 0.001) in Moderate Grain (Table 4). Operational taxonomic unit-13387 (Fecalibacterium) and OTU-14977 (Succinivibrio) were the most abundant (P < 0.001) in Moderate Grain while OTU-7731 (Clostridium) was the most abundant (P < 0.001) in High Grain. These 3 OTU were also dominant and accounted for >1.0% of total sequences in collective data. Five OTU (from OTU-13994 to OTU-13998) assigned to Oscillibacter were more abundant (P < 0.001) in High Grain than in the other 2 groups and rarely detected in Silage/Forage. Three Prevotella OTU (from OTU-5060 to OTU-5062) were more abundant (P < 0.001) in Moderate Grain than in the other 2 groups and rarely detected in Silage/Forage. Coprococcus was represented by 3 OTU (from OTU-9760 to OTU-9762), and their abundances were different (P < 0.001) among the 3 diet groups. Operational taxonomic unit-9760, OTU-9761, and OTU-9762 were the most abundant in Silage/Forage, Moderate Grain, and High Grain, respectively. Operational taxonomic unit-5500 (Lactobacillus) and OTU-10257 (Roseburia) were the most abundant (P < 0.001) in High Grain and did not differ (P > 0.05) between Moderate Grain and Silage/Forage. Operational taxonomic unit-1194 (Microbacterium) was rarely detected in Moderate Grain while OTU-5135 (Streptophyta) and OTU-15034 (Pantoea) was rarely detected in High Grain and Silage/Forage.

Of the 33 OTU unclassified at the genus level, 10 OTU were assigned to unclassified Ruminococcaceae, and OTU-12973 of the 10 OTU accounted for 1.5% of total sequences in collective data (Table 4). Five of the 10 OTU were the most abundant (P < 0.001) in High Grain while another 5 of the 10 OTU were the most abundant (P < 0.001) in Silage/Forage. Unclassified Lachnospiraceae was represented by 7 of the 33 OTU, and OTU-9531 of the 7 OTU accounted for 1.65% of total sequences in collective data. Three of the 7 OTU were the most abundant (P < 0.001) in High Grain while another 3 of the 7 OTU were the most abundant (P < 0.001) in Silage/Forage. The remaining 1 OTU was the most abundant (P < 0.001) in Moderate Grain. Unclassified Prevotellaceae was represented by 3 of the 33 OTU, and 2 of the 3 OTU were the most abundant (P < 0.001) in Moderate Grain and rarely detected in Silage/Forage. The remaining 1 OTU was the most abundant (P < 0.001) in High Grain and rarely detected in Moderate Grain and Silage/Forage. Unclassified Bacteroidetes was represented by 3 of the 33 OTU, and 2 of the 33 OTU were rarely detected in Silage/Forage. The remaining 1 OTU was the most abundant (P < 0.001) in Silage/Forage and rarely detected in Moderate Grain and High Grain. The TM7 genera incertae sedis was represented by 2 OTU that were the most abundant (P < 0.001) in Silage/Forage and rarely detected in Moderate Grain and High Grain. The abundances of 2 OTU classified to unclassified Porphyromonadaceae were the lowest (P < 0.001) in Silage/Forage while those of 2 OTU classified to unclassified Peptostreptococcaceae were the lowest (P < 0.001) in Moderate Grain. Unclassified Clostridiales was represented by 1 OTU that was the most abundant (P < 0.001) in Silage/Forage. Unclassified Bacteria were represented by 3 of the 33 OTU, and 2 of the 3 OTU were more abundant (P < 0.001) in High Grain than in the other 2 groups. The remaining 1 OTU was the most abundant (P < 0.001) in Silage/Forage and rarely detected in Moderate Grain and High Grain.

Observed OTU, Chao1, ACE, and Shannon’s diversity index were calculated from the 333 subsamples, and then their average values were compared among 3 diet groups. All these indices differed (P < 0.001) among the 3 diet groups and were the greatest in Moderate Grain and the lowest in High Grain (Table 2). These results indicate that bacterial diversity in the feces of cattle was the greatest in Moderate Grain but the lowest in High Grain. We also examined correlations among the 333 subsamples using UniFrac PCoA (Fig. 2). In both weighted and unweighted UniFrac, principal coordinate (PC) 1 separated bacterial communities based on diet. Unweighted UniFrac showed that spots of Silage/Forage were phylogenetically distinct from those of Moderate Grain and High Grain. Spots were more divergent in High Grain than in the other 2 groups, indicating that bacterial communities among individual animals are less similar in High Grain than in the other 2 groups.

Figure 2.
Figure 2.

UniFrac principal coordinates analysis (PCoA) showing correlations among 3 diet groups. A) Weighted PCoA analyzed from 333 subsamples with 2,000 reads. B) Unweighted PCoA analyzed from 333 subsamples with 2,000 reads. PC = principal coordinate. See online version for figure in color.

 


DISCUSSION

Great diversity of the bovine fecal microbiota may be attributed to various factors such as diet, age, and geographic region. Large variation in fecal microbiota was observed among the 3 diet groups. While this project was not specifically designed to experimentally test the role of the diet on bacterial community composition, the observed differences among the groups were striking. As an observational study, the ANOVA methods and resulting statistical comparisons are not a test of an experimental null hypothesis regarding treatment differences; rather, they are presented as a tool to interpret the data. This is the first study of this magnitude of which we are aware that clearly demonstrates bacterial similarities within and compositional differences between groups of animals. Gender and year of study are additional compounding factors across these groups that could play a minor role in affecting bacterial community composition. Bacterial diversity was greater in Moderate Grain than in High Grain but not different among pens (data not shown). The diet composed of mostly dry-rolled corn appears to reduce bacterial diversity. Nonetheless, for all diets, Firmicutes was the first dominant phylum accounting for >50% of the total sequences. The dominance of Firmicutes corroborates previous studies (Durso et al., 2010; Shanks et al., 2011; Rice et al., 2012), indicating that this phylum is a resident member of the bovine feces irrespective of the factors affecting diversity of the bovine fecal microbiota. The present study indicates that diet is an important factor affecting diversity of the bovine fecal microbiota relative to other factors.

Durso et al. (2012) indicated that Prevotella and Bacteroides were commonly found in cattle feces and associated with dietary differences. Prevotella was more abundant in cattle fed a corn-based finishing diet than in cattle fed the diet including 40% wet distillers’ grain with solubles (Durso et al., 2012). In the present study, Prevotella was the most abundant in cattle fed moderate grain with 66% dry-rolled corn. Although it seems that corn-based diets positively affect the abundance of Prevotella (Durso et al., 2012; Rice et al., 2012), Prevotella abundance was reduced in cattle fed High Grain with the greater level of dry-rolled corn (82.75%) compared to cattle fed the 66% dry-rolled corn. Both Prevotella and Bacteroides were rarely detected in cattle fed Silage/Forage (70% corn silage and 30% alfalfa haylage). Durso et al. (2012) observed an increase in abundance for Bacteroides when diets high in fat were fed, and in the present study, Bacteroides abundance seems to be negatively correlated with the diet rich in forage.

Oscillibacter was the most dominant genus in the phylum Firmicutes and the most abundant in cattle fed high grain (Table 3). Oscillibacter seems to be positively correlated with starch content. In humans, Oscillibacter increased in the diet of resistant starch (Walker et al., 2011). Cattle fed high starch diets can have high bypass starch from rumen (Wells et al., 2009) and Oscillibacter in the feces of cattle may be associated with the high levels of bypass starch. In the RDP database, rrs sequences recovered from ruminal Oscillospira were classified to Oscillibacter. Ruminal Oscillospira are positively correlated with diets rich in forage (Mackie et al., 2003), but the abundance of fecal Oscillibacter in the current study was the lowest in animals fed the diet rich in forage. Therefore, Oscillibacter present in the feces of cattle might not be represented by rrs sequences recovered from Oscillospira found in the rumen. Turicibacter also was a dominant genus as shown in mammals including humans (Cuiv et al., 2011), and its isolate strain was presumed to be a pathogen (Rettedal et al., 2009). Because Turicibacter increased in the diet rich in dry-rolled corn, potential negative affects on cattle health or performance need to be evaluated. Sporacetigenium and Anaerovorax were abundant in the feces of cattle fed 40% wet distillers’ grain with solubles compared to the corn-based finishing diet (Durso et al., 2012). Sporacetigenium ferments pentoses (Chen et al., 2006) and is likely abundant in Silage/Forage because forage diets are sources for hemicellulose that are rich in pentoses (Durso et al., 2012). The type strain of Anaerovorax ferments putrescine produced from amino acids and proteins (Matthies et al., 2000). Anaerovorax is likely abundant in Silage/Forage because the Silage/Forage diet includes the protein-rich alfalfa (Durso et al., 2012).

Roseburia, Fecalibacterium, and Coprococcus may contribute to producing butyrate that is used as the energy source for the mucosa (Pryde et al., 2002), and these 3 genera seem to be reduced by the diet rich in forage (Table 3). Anaerovibrio has been associated with lipid degradation (Henderson, 1971) and was abundant in cattle fed dry-rolled corn but rarely detected in cattle fed the diet rich in forage. Blautia can contribute to lactate and acetate production as described previously (Park et al., 2012) and was rarely found in cattle fed the diet rich in forage. Mucin-degrading Akkermansia (Derrien et al., 2004) was abundant in animals fed the diet rich in forage. The diet rich in forage may result in degradation of the protective mucus layer by Akkermansia in cattle.

Because rrs sequences that were recovered from the family Chloroplast originate from plant cells, Streptophyta placed within Chloroplast would be a taxon group present in consumed forage diets and was expected to not only be detected but also have a possible positive relationship with the amount of corn silage in the diet. Cattle fed the Moderate Grain diet, which had corn silage added to the diet, had abundant levels of this taxon group. However, the abundance of Streptophyta was lower in Silage/Forage (70% corn silage) than in Moderate Grain (26% corn silage). Factors affecting the abundance of Streptophyta in cattle feces will need to be elucidated in future studies.

Core taxa have been used to understand microbial ecology and hypothesize their function within a habitat (Shade and Handelsman, 2012). Although collective sequences were assigned to numerous taxa, only a small number of core taxa, which are 3 phyla, 3 classes, 2 orders, 6 families, and 2 genera, were detected across all the 333 cattle. Firmicutes and Bacteroidetes were major core phyla while Proteobacteria was a minor core phylum as described previously (Shanks et al., 2011; Rice et al., 2012). Only Clostridium and Turicibacter were identified as core taxa across all the 333 cattle. Prevotella, Oscillibacter, Roseburia, Coprococcus, Lactobacillus, Anaerovorax, Sporacetigenium, and Ruminococcus were detected across ≥95% of all the 333 cattle (Table 3). These 8 genera also are thought to be core taxa. A total of 10 core genera seem to be commonly found in cattle feces irrelevant to diets. These core genera might play a more central role in the fecal microbial ecosystem.

Core taxa for the 272 steers fed feedlot concentrate diets with Moderate Grain or High Grain diets were 3 phyla, 5 classes, 3 orders, 7 families, and 7 genera. The 7 genera were Trucibacter, Subdoligranulum, Oscillibacter, Fecalibacterium, Roseburia, Coprococcus, and Clostridium and were detected across all the 272 steers. In addition, Sporacetigenium, Blautia, Prevotella, Coprobacillus, Lactobacillus, Bacteroides, Parabacteroides, Anaerovibrio, Ruminococcus, and Succinivibrio were detected across ≥95% of all the 272 steers and appear to be core taxa. These 17 genera might play an important role in the fecal microbial ecosystem of cattle fed feedlot concentrate diets. This result is supported by the previous study (Durso et al., 2010) indicating that Fecalibacterium, Ruminococcus, Roseburia, Clostridium, Prevotella, and Bacteroides were core taxa for 6 cattle fed diet including 21% corn. Because Durso et al. (2010) used small numbers of sequences (11,171 sequences) for microbial analysis, the other genera seem not to be identified as core taxa.

Only 8 OTU were observed across ≥95% of all the 333 cattle and were composed of 2 Clostridium, 1 Turicibacter, 2 unclassified Clostridiales, 2 unclassified Peptostreptococcaceae, and 1 unclassified Ruminococcaceae. Species associated with these 8 core OTU might greatly contribute to the fecal microbial ecosystem irrespective of diet. Nine OTU including 6 of the 8 core OTU were observed across ≥95% of the 61 cattle fed Silage/Forage and were composed of 3 Clostridium, 1 Turicibacter, 1 Clostridiales, 3 Peptostreptococcaceae, and 1 Ruminococcaceae. Species associated with these 9 OTU might play an important role in the fecal microbial ecosystem of cattle fed Silage/Forage. Thirty-three OTU including the 8 core OTU were observed across ≥95% of the 272 cattle fed concentrate diets and were composed of 2 Clostridium, 2 Coprococcus, 5 Roseburia, 6 Oscillibacter, 1 Turicibacter, 1 Fecalibacterium, 1 Succinivibrio, 1 unclassified Bacteria, 1 unclassified Bacteroidetes, 2 unclassified Clostridiales, 5 unclassified Ruminococcaceae, 4 unclassified Lachnospiraceae, and 2 unclassified Peptostreptococcaceae. Species associated with these 33 OTU might play an important role in the fecal microbial ecosystem of cattle fed concentrated diets.

Numerous sequences could not be classified into a known genus. Dominant unclassified groups were unclassified Ruminococcaceae and unclassified Lachnospiraceae (Table 3). Most of the dominant OTU also were assigned to these 2 unclassified groups. The dominance of OTU assigned to these unclassified groups will need to be identified using quantitative real-time PCR as described previously (Stiverson et al., 2011). These 2 unclassified groups also were dominant in the rumen (Kim et al., 2011), and some strains in cattle feces may have originated from the rumen. This assumption is supported by the study by Durso et al. (2013), where some OTU assigned to these 2 unclassified groups were shared between the rumen microbiota and the fecal microbiota. The OTU shared between the rumen microbiota and the fecal microbiota within the same cattle will need to be examined in future studies. Species associated with these dominant OTU might greatly affect animal health and meat safety, and isolation and characterization of these strains will need to be conducted to elucidate their function in the cattle gastrointestinal microbial ecosystem.

Hierarchical taxa classification based on naïve Bayesian rRNA Classifier (Wang et al., 2007) seems to be unreliable for some taxa. To minimize this issue, we searched for sequences recovered from isolates in the RDP database (release 10, update 29). Roseburia included 2 sequences recovered from Butyrivibrio fibrisolvens isolated from a Greenland ice sample (Sheridan et al., 2003) and the rumen (Wallace et al., 2006). The dominance of Roseburia in both Moderate Grain and High Grain might result from the increase of B. fibrisolvens in cattle feces. Blautia included many sequences recovered from Ruminococcus spp. isolated from human feces (Hayashi et al., 2002) and the rumen (Rieu-Lesme et al., 1996). Therefore, Blautia, which was dominant in High Grain, might be associated with the increase of Ruminococcus. Anaerovorax and Coprococcus included some sequences recovered from Clostridium spp. of nonfeces origin (Collins et al., 1994; Sheridan et al., 2003; Gourgue-Jeannot et al., 2006). The abundance of Anaerovorax and Coprococcus in cattle feces might be associated with the abundance of Clostridium. Unclassified Lachnospiraceae, which was highly abundant across all samples, included some sequences recovered from Blautia spp., Clostridium spp., Eubacterium spp., and Ruminococcus spp. isolated from feces (Lan et al., 2002; Brooks et al., 2003; Ballard et al., 2005; Roger et al., 2010; Park et al., 2012). The dominance of unclassified Lachnospiraceae in cattle feces might be associated with these 4 genera. This issue will need to be considered to investigate the taxonomic composition of the fecal microbiota in future studies.

Fecal samples of cattle fed Moderate Grain were collected in July and August under the same diet, and pyrosequencing analysis for the 2 replications was conducted separately. We compared bacterial communities between the 2 replications. The abundance of the phylum Firmicutes was greater (P < 0.001) in samples collected in August than in July. Some genera were different between the 2 replications, and most of the genera were assigned to Firmicutes. Prevotella, Clostridium, Coprococcus, Roseburia, Fecalibacterium, Succinivibrio, Blautia, Subdoligranulum, Sporacetigenium, Ruminococcus, Coprobacillus, Lactobacillus, and Xylanibacter were greater (P < 0.05) in samples collected in August than in July whereas Parabacteroides, Oscillibacter, Bacteroides, Anaerovibrio, Anaerovorax, Alistipes, and Escherichia/Shigella were lower (P < 0.05) in samples collected in August than in July. The taxonomic composition at the genus level appears to be affected by environmental factors. Although some genera were different between the 2 replications, unweighted PCoA based on species-level OTU separated the 2 replications by only 2% variation, indicating slight shifts at the species level. The taxonomic composition at the species level appears to be only slightly changed by normal environmental factors within the same group of animals over time.

Although we analyzed more than 2 million sequences across all the 333 samples, the percent coverage in collective data was still incomplete (77%) based on the rarefaction estimate of maximum number OTU (Larue et al., 2005). In the pooled sample of Moderate Grain, the number of sequences per a sample on the average was 7,300 and the percent coverage was 80%. On the other hand, 3,900 sequences per sample on the average resulted in 64% coverage in the pooled sample of Silage/Forage. To investigate minor taxa and OTU in the feces of cattle, it is apparent from current research that more than 7,300 sequences per sample will need to be obtained in future studies. Nonetheless, the present study had sufficient power in observation groups to enable a better understanding of the detailed list of fecal microbiota influenced by diet, core taxa irrespective of diets, and variation of fecal microbiota among individual cattle. This is the first study of this magnitude of which we are aware that clearly demonstrates bacterial similarities within and striking compositional differences between groups of animals. Diet appeared to have a large effect on fecal microbiota, particularly when comparing forage versus grain diets. In future studies, the role of diet will need to be considered when evaluating fecal microbiota and host relationships.

 

References

Footnotes


Comments
Be the first to comment.



Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.