Diversity of Ruminococcus strains: a survey of genetic polymorphisms and plant digestibility


Dr D.O. Krause, CSIRO Tropical Agriculture, Long Pocket Laboratories, Private Bag 3, Indooroopilly, Queensland 4068, Australia (e-mail: denis.krause@tag.csiro.au).


Twenty-three strains of Ruminococcus isolated from ruminants were assessed for digestive ability on different plants and purified cellulose. Genetic diversity was assessed by ERIC, REP and 16–23S rDNA spacer polymorphisms. All ruminococci could be typed by ERIC, REP or 16–23S rDNA spacer, but all three typing methods had to be used in concert to differentiate closely related strains. Digestibility of lucerne (Medicago sativa), rhodes grass (Chloris gayana) and spear grass (Heteropogon contortus) were assessed. Dry matter (DM) digestibility was highly correlated (> 0·93) with neutral detergent fibre (NDF) digestibility, but cellulose disc digestibility was a poor indicator of DM and NDF digestibility. Studies demonstrate the wide variation in ability of ruminococci to digest forages, and some recently isolated strains (Y1, LP-9155, AR67, AR71 and AR72) were superior to reference strains (FD-1 and Ra8). Multivariate analysis showed that groupings derived from genotyping data closely resembled those determined by digestibility data. This study indicated that ruminococci are diverse in digestive ability and genotype, and this diversity suggests that there may be highly fibrolytic strains in nature that could be utilized for animal production.

Microbial diversity in the rumen is well recognized ( Krause & Russell 1996; Hespell et al. 1997 ) but there are few studies that have assessed genotypic diversity of a single genus in conjunction with a phenotype that is regarded as fundamental to the ecological fitness of the organism. Bacterial fibre digestion in the rumen is carried out primarily by Fibrobacter succinogenes, Ruminococcus (R. albus and R. flavefaciens) and Butyrivibrio fibrisolvens, but F. succinogenes and the Ruminococcus species are usually the most fibrolytic ( Hespell et al. 1997 ). Fibre fermenting ability by Ruminococcus is arguably central to its ability to compete in the rumen but it is not known if this important phenotype is related to its genotype.

Digestibility studies of different plants by a range of bacteria are few ( Dehority & Scott 1967; Morris & van Gylswyke 1980) and there are no studies which have investigated a range of ruminococci for digestive and genetic diversity. True assessments of diversity of digestive ability should be obtained from experiments using a variety of plants, and bacteria able to digest different plants efficiently have the best chance of surviving in the rumen. Cellulose digestibility has traditionally been used to evaluate activity of cellulolytic bacteria ( Halliwell & Bryant 1963; Stewart et al. 1990 ) but purified cellulose does not contain cell wall components (e.g. phenolics and lignin) that limit digestibility ( Akin et al. 1974 ; Akin 1982; McSweeney & Mackie 1997).

Several investigators ( Versalovic et al. 1991 ; van Belkum et al. 1998 ) have shown that repetitive extragenic palindromic sequences (REP) and enterobacterial repetitive intergenic consensus (ERIC) sequences can be used as bacterial typing methods. In addition, amplification of prokaryotic rDNA spacer regions with conserved 16S and 23S primers can be used to type bacteria ( Jensen & Hubner 1996). Cladistic analyses of genotype data (generated with ERIC, REP, and 16–23S rDNA spacers) and digestibility data have been used to investigate the relationship between genotype and phenotype (fibre digestibility).

Materials and methods

Strains selected

Twenty-three isolates of ruminococci were assessed for genotype and digestibility and represented isolates from North America and Australia ( Table 1). This group of ruminococci include well characterized strains (e.g. FD-1, Ra8 and SY3) that have served as standards in rumen microbiology. The AR strains, Y1, RF-1 and LP9155 have been characterized by Gram stain, substrate utilization, end-products and 16S rDNA sequence analysis ( Krause et al. 1999 ).

Table 1.  Bacterial strains
Ruminococcus albus SY3Rod Mackie (University of Illinois, USA)
R. albus Ra8Rod Mackie (University of Illinois, USA)
R. albus B199Rod Mackie (University of Illinois, USA)
R. albus AR67Keith Gregg (University of New England, Australia)
R. flavefaciens FD-1Rod Mackie (University of Illinois, USA)
R. flavefaciens R13e2Rod Mackie (University of Illinois, USA)
R. flavefaciens B146Rod Mackie (University of Illinois, USA)
R. flavefaciens Y1Geoff Gordon (CSIRO, Australia)
R. flavefaciens RF-1Colin Orpin (CSIRO, Australia)
R. flavefaciens LP-9155Chris McSweeney (CSIRO, Australia)
R. flavefaciens AR 6, 44, 45, 46, 47, 50, 69, 71, and 72Keith Gregg (University of New England, Australia)
Ruminococcus-like 48, 49, 65, 68 Keith Gregg (University of New England, Australia)

Composition of media

Bacterial cultures ( Table 1) were grown at 39 °C in basal medium containing (l−1): 10 g casitone, 2·5 g yeast extract, 150 ml clarified rumen fluid, 150 ml mineral 1 (contains 100 ml−1: 3 g K2HPO4.3H2O), 150 ml mineral 2 (contains 100 ml−1: 3 g KH2PO4, 6 g (NH4)2SO4, 6 g NaCl, 1·23 g MgSO4.7H2O, 1·58 mg CaCl2.2H2O), 2 ml trace mineral salts (contains 100 ml−1: 0·5 mg ZnSO4.7H2O, 0·15 mg MnCl2.4H2O, 1·5 mg H3BO3, 1·0 mg CoCl2.6H2O, 0·05 mg CaCl2.2H2O, 0·1 mg NiCl2.H20, 0·15 mg Na2MO4.2H2O, 7·5 mg FeCl2.4H2O), 3·1 ml volatile fatty acids (contains 100 ml−1: 17 ml acetic acid, 6 ml propionic acid, 4 ml butyric acid, 1 ml isobutyric acid, 1 ml methyl-butyric acid, 1 ml valeric acid, 1 ml isovaleric acid, 1 g phenylacetic acid), 1 g l-cysteine-HCl, 4 g Na2CO3 and 0·01% resazurin. Medium was made anaerobically according to the methods of Hungate (1950) as modified by Bryant (1972).

Digestibility of cellulose discs and plant material

Filter-paper discs (Whatman no. 1) were prepared by overnight incubation with concentrated phosphoric acid at room temperature. The discs were then washed several times with distilled water and stored at 4 °C until needed. Plant material was ground through a 1 mm screen and then washed. Treated substrate (50 mg cellulose discs or plant material) was weighed into Balch tubes, basal media (see above; 10 ml) was added under an anaerobic atmosphere (95% CO2, 5% H2), then the tubes were stoppered and autoclaved. Each bacterium was adapted to assay substrate over several transfers and a 0·1 ml inoculum (24 h culture) was used to initiate the digestibility assay. Assays were stopped at 72 h and DM and NDF disappearance was determined according to Goering & van Soest (1970). All assays were in triplicate.

Extraction of genomic DNA from Ruminococcus strains

Ruminococcus isolates were grown on cellobiose medium (basal medium plus 10 g l−1 cellobiose) and cells were centrifuged at high speed (10 000 g, 10 min). The pellet was resuspended and washed in 500 μl TE (10 mmol l−1 Tris-Cl, 1 mmol l−1 Na2-EDTA), and DNA was extracted by incubation of the pellet in lysozyme and mutanolysin. DNA was separated from cell debris by repeated phenol/chloroform and chloroform/isoamyl alchohol extractions, and precipitated in ethanol after the addition of a 1/10 volume of sodium acetate (3 mol l−1). DNA concentration was measured at 260 nm and adjusted to a final concentration of 10 ng μl−1.

PCR primers and reaction conditions

Each 50 μl PCR reaction included: 1/100 dilution of cellobiose-grown cells, 5 μl 10 × reaction buffer (Bresatec, Adelaide, Australia), 2·0 mmol l−1 (ERIC and REP) or 3 mmol l−1 (16–23S spacer), MgCl2, 0·2 mmol l−1 dNTPs, 10 pmol of each primer and 0·5 μl Taq polymerase (Promega, Sydney, Australia). For ERIC and REP-PCR, cycling conditions were denaturation for 1 min at 94 °C, annealing for 1·0 min at 47 °C (ERIC) or 40 °C (REP), and 72 °C for 2 min. This cycle was repeated 30 times. PCR cycling conditions for the 16–23S spacer were: denaturation at 94 °C for 5 min for the first cycle only and 1 min thereafter, annealing at 50 °C for 1 min, extension at 72 °C for 1·5 min for 30 cycles, followed by a final extension at 72 °C for 7 min. ERIC and REP primers were as previously described ( De Bruijn 1992). The 16–23S rDNA spacer was amplified using a conserved 16S forward primer (5′-AAG TCG TAA CAA GGT AG/AC CGT A-3′) and a conserved 23S reverse primer (5′-GGG TTT/G/C CCC CAT TCG G-3′).

Reproducibility of profiles

DNA banding patterns were based on the major as well as minor bands, but best results were obtained by running the whole of a 20 μl PCR reaction on an agarose gel. This resulted in sharp distinct patterns which were reproducible (repeated at least three times). Satisfactory banding patterns could be obtained with unextracted DNA (1/100 dilution of culture) but extracted DNA had to be used for some strains (AR47, AR65 and AR68). No differences in profiles were seen when 1/100 dilutions and extracted samples of the same bacteria were compared.

Multivariate analysis

Banding patterns from all three genotypic typing methods were scored as the presence or absence of bands and used in multivariate analysis. Euclidean distances were calculated from character data using Statistica software 5.0 (Statsoft Pty Ltd, Melbourne, Australia). Digestibility data were treated in the same way except that the data were normalized so that the mean was zero and the standard deviation was one.


ERIC, REP and 16–23S spacers

ERIC and REP primers were successfully used with all the Ruminococcus strains and gave highly polymorphic fingerprints able to differentiate strains ( Fig. 1). Strains were ordered according to their phenotype groupings ( Fig. 4b). Amplification of the 16–23S rDNA spacer resulted in PCR fragments that provided an additional means of typing ruminococci ( Fig. 1). Used in concert, these three genotyping methods could differentiate between most of the ruminococci included in this study.

Figure 1.

ERIC (a), REP (b), and 16–23S rDNA spacer (c) profiles for Ruminococcus strains

Figure 4.

Dendrograms of genotype data (a) and phenotype data (b) based on digestibility values. The linkage distance is calculated from Euclidean distances used in construction of trees

Many strains were highly similar and all three typing methods were required to separate them. AR 71 and 72 only differed in their 16–23S rDNA spacers ( Fig. 1c) and had similar digestibility values ( Table 2). B146 had a similar ERIC and 16–23S rDNA spacer to AR45, but they differed in REP patterns. AR 6, 44 and 45 had almost identical ERIC, REP and 16–23S rDNA spacers ( Fig. 1). AR 48 and AR 65 also had highly similar patterns and the ERIC, REP and 16–23S rDNA spacers are represented ( Fig. 1).

Table 2.  Digestion of plant substrates by strains of Ruminococcus*
StrainCellulose digestion (%) Lucerne DM (%) Lucerne NDF (%) Rhodes DM (%) Rhodes NDF (%) Spear DM (%) Spear NDF (%)
  • *

    Mean digestibility values are given with standard deviations. Correlations between NDF and DM for lucerne, rhodes grass and spear grass were 0·95, 0·98 and 0·94, respectively.

R. albus species AR6792·3 ± 1·653·0 ± 2·027·7 ± 1·149·0 ± 1·946·0 ± 0·140·1 ± 3·133·1 ± 1·4
SY366·7 ± 1·848·8 ± 1·222·1 ± 2·645·1 ± 2·043·1 ± 0·337·8 ± 4·129·3 ± 0·7
Ra834·8 ± 1·147·3 ± 1·921·2 ± 1·640·5 ± 1·135·7 ± 0·238·2 ± 1·124·9 ± 0·6
B199029·0 ± 2·3 1·7 ± 1·014·9 ± 2·211·3 ± 1·522·7 ± 2·4 3·2 ± 0·1
R. flavefaciens species Y186·4 ± 0·853·3 ± 0·438·2 ± 1·254·0 ± 0·953·3 ± 1·349·8 ± 0·144·8 ± 0·7
91-5-547·0 ± 1·950·0 ± 1·933·1 ± 2·361·1 ± 1·255·1 ± 1·756·3 ± 2·243·4 ± 1·4
AR7243·5 ± 1·550·5 ± 3·427·8 ± 1·454·4 ± 0·750·4 ± 1·143·8 ± 1·038·6 ± 1·1
AR7136·1 ± 1·450·6 ± 1·730·0 ± 1·749·0 ± 3·448·8 ± 0·147·9 ± 1·439·1 ± 0·3
FD-172·1 ± 1·044·3 ± 1·329·4 ± 0·548·9 ± 1·646·6 ± 1·524·3 ± 0·921·7 ± 0·3
RF-144·6 ± 1·551·8 ± 2·228·0 ± 3·749·5 ± 0·245·5 ± 0·743·4 ± 1·532·7 ± 1·0
B14671·3 ± 1·749·6 ± 2·626·0 ± 2·233·0 ± 1·028·6 ± 1·427·0 ± 1·911·2 ± 1·0
AR6948·3 ± 1·252·4 ± 2·926·0 ± 0·438·1 ± 0·629·5 ± 1·432·9 ± 1·318·5 ± 0·6
AR4560·7 ± 1·448·6 ± 2·728·3 ± 0·835·7 ± 1·528·5 ± 1·423·7 ± 1·118·7 ± 0·9
AR4448·3 ± 1·247·5 ± 0·727·6 ± 2·032·4 ± 2·429·7 ± 0·432·0 ± 1·718·3 ± 0·2
AR642·4 ± 1·651·6 ± 2·030·6 ± 2·333·9 ± 1·628·5 ± 1·433·8 ± 0·821·1 ± 1·4
R13e249·6 ± 1·542·4 ± 2·721·0 ± 1·539·2 ± 0·536·1 ± 0·428·5 ± 2·517·5 ± 1·5
AR4746·3 ± 1·545·4 ± 2·424·3 ± 2·332·7 ± 1·827·6 ± 1·224·5 ± 0·917·4 ± 0·3
AR4657·1 ± 1·342·1 ± 2·013·0 ± 2·333·0 ± 1·328·7 ± 0·311·3 ± 0·9 5·2 ± 0·8
AR5049·6 ± 1·224·4 ± 2·4037·0 ± 0·927·7 ± 2·3 9 ± 1·70
Ruminococcus-like strains AR49 1·7 ± 1·422·6 ± 2·0020·1 ± 3·213·5 ± 1·710·4 ± 1·60
AR68 1·6 ± 0·629·8 ± 2·2010·0 ± 0·5013·2 ± 0·40
AR65 2·3 ± 0·626·3 ± 1·60 5·9 ± 0·7011·4 ± 0·60
AR48026·0 ± 2·00 8·0 ± 1·80 8·4 ± 1·20

Digestibility of cellulose discs and plant substrates

Overall means (unweighted) were calculated from digestibilities of cellulose discs and plant substrates (DM and NDF, Table 2) and provided an index of digestive potency across several substrates. There was a wide range in the digestibility of cellulose discs, and varied from a high of 92·3% (AR67) to those that did not digest cellulose (B199). There was also a wide range in the digestibility values (DM and NDF) for the plant substrates, and the NDF values were always lower than the DM values.

Cellulose disc digestibility was not a good index of digestive potential and was poorly correlated with NDF digestibilities of lucerne (r= 0·65, Fig. 2a), rhodes grass (r= 0·73, Fig. 2b) and spear grass (r= 0·51, Fig. 2c). Examples are LP-9155 and RF-1, which had fairly low (47% and 44·6%) cellulose disc digestibilities but a relatively high digestibility for lucerne and grass ( Table 2). These high mean values were reflected by higher plant DM and NDF digestibility values. Plant DM and NDF digestibilities were highly correlated (> 0·93) and DM could predict NDF values (see footnote in Table 2). The rankings of bacteria were similar irrespective of plant substrate (r > 0·80) and only minor changes in rankings occurred with different plants. The overall rankings are represented by cluster analysis ( Fig. 4b).

Figure 2.

Regression analysis of cellulose digestibility against NDF digestibility of lucerne (a), rhodes grass (b), and spear grass (c). The 95% confidence interval is represented by the dotted lines

Lucerne hay had a large water soluble fraction (25·6%) and it’s dry matter digestibility was usually higher than rhodes or spear grass ( Table 2). As the water soluble components of rhodes (6·7%) and spear (8·7%) grass are lower than lucerne, the difference between DM and NDF digestibility values were less. Digestibility of lucerne NDF was fairly low and the NDF content of lucerne (57·1%) was lower than rhodes (82·8%) or spear (82·8%) Digestibilities in Table 2 are for 72 h and reflect extents of plant cell wall degradation. As rates of digestion are considered to be more relevant in the rumen, the rate and extent of rhodes grass digestibility of three ruminococci (Ra8, LP-9155 and B146) varying in digestive ability were assessed. Bacteria with the highest extents of plant cell wall degradation always had the highest rates of digestibility and this was true for both DM and NDF ( Fig. 3).

Figure 3.

Digestion of rhodes grass dry matter (DM, Fig. 2a) or neutral detergent fibre (NDF, Fig. 2b) by Ruminococcus strains (○) B146, (▵) LP-9155, and (◊) Ra8

Comparison between digestibilities and genotyping

Twenty-three isolates of Ruminococcus were examined for ERIC, REP and 16–23S rDNA spacers and fibre digestibilities. Genotyping revealed that several of these isolates (see above discussion) were highly similar and there were only minor differences in their banding patterns. These genotyping methods were able to differentiate strains below the species level and are a useful way of discriminating between closely related strains.

ERIC, REP and 16–23S rDNA spacer data were used in multivariate analysis and the 23 isolates could be separated into three groups ( Fig. 4a), but they did not always fit the 16S rDNA types ( Table 2). Digestibility data could also be used to generate a dendrogram ( Fig. 4b). Strains could be divided into three groups with group 1 having the most potent digesters and group 3, the least. In general, the strains that fell into a particular group based on genotype ( Fig. 3a) also fell into the same group when analyses were done with digestibility data.


An optimally functioning rumen is critical to animal productivity and superior performance in pasture fed animals is partly determined by efficiency of plant cell wall digestion ( Beveridge & Richards 1973; Broderick 1989; Ørskov 1991). As Ruminococcus species are one of the most fibrolytic groups of ruminal bacteria ( Weimer 1996), a study was undertaken to investigate their digestive potency and genetic diversity. These studies provide an appreciation of inherent phenotypic and genotypic variation within Ruminococcus that can be harnessed to improve ruminal function. 16S rRNA analysis differentiates strains at the species level ( Pace 1997) but to obtain an understanding of diversity within a species, subspecies genotyping methods are needed.

Rapid amplification of polymorphic DNA (RAPD) is frequently used as a typing method for bacteria and relies on the amplification of DNA with PCR primers at low stringency. RAPDs are not based on specific targets and do not require prior knowledge of sequence, but results are often variable ( Lawrence et al. 1993 ; Brikun et al. 1994 ). As ERICs, REPs and 16–23S spacers are based on conserved sites, they can be a more reliable way of typing bacteria ( Versalovic et al. 1991 ; De Bruijn 1992). REP elements may be regulatory in nature because of their ability to form stem-loop structures in transcribed RNA ( Versalovic et al. 1991 ), whereas ERIC elements contain inverted repeats located in extragenic regions ( Versalovic et al. 1991 ). Bacteria have several ribosomal operons and the spacer region between the 16S and 23S varies in sequence and length, providing an additional means of determining bacterial genotype ( Jensen & Hubner 1996).

Versalovic et al. (1991) found that the ERIC and REP elements were present in most Gram-negatives but not to the same extent in Gram-positives. They assessed more than 30 Gram-negatives and only 11 Gram-positives, but over 70% of the Gram-positives showed at least some hybridization with ERIC and REP consensus sequences. Ruminococcus are low G + C Gram-positives, related phylogenetically (by 16S rDNA analysis) to the clostridia, but no studies using ERICs, REPs and 16–23S spacers have assessed bacteria affiliated with the clostridia ( Rainey & Janssen 1995). We have surveyed (data not shown) a diverse group of Gram-positive and Gram-negative ruminal bacteria and ERIC, REP and 16–23S spacer polymorphisms can be extended to most of these bacteria.

ERICs ( Fig. 1a) and REPs ( Fig. 1b) both demonstrated the presence of highly polymorphic DNA, and primers amplified many bands. In contrast, amplification of the 16–23S rDNA spacer gave one to three bands ( Fig. 1c) but these banding patterns could be used to differentiate between bacteria. A combination of ERICs, REPs and 16–23S rDNA spacers was required to differentiate effectively between bacteria (see Results above). We found in the course of these studies that culture purity could be problematic and it was difficult to differentiate between ruminococci microscopically. These genotyping methods provide a benchmark for quality control and for differentiating among closely related strains.

Cellulose disc digestion was not the best way of evaluating organisms because of low correlations with forage NDF digestion ( Fig. 2). This observation is supported by work by Stewart et al. (1990) which showed that R. flavefaciens 007 lost its ability to degrade cotton fibres after repeated transfer on cellobiose medium. One of these colonies could be resuscitated and was slightly more active than the ‘mutated’ low cotton digesting form, but there was hardly any difference when wheat straw was the substrate. These authors concluded that the factors important in cotton digestion had little relevance to wheat straw digestion.

We considered that NDF digestion was the most important characteristic because ruminants consume a diet containing various plants and assessments of digestive ability of several plants give a more accurate understanding of variations in digestive potential. Dehority & Scott (1967) examined digestion of a wide range of forages by various species of ruminal bacteria. Digestive abilities among the fibrolytic organisms varied, but only four strains of R. flavefaciens (strains B1a, B34b, C1a and C-94) and one strain of R. albus (strain 7) were studied. Kock & Kistner (1969) and Morris & van Gylswyke (1980) studied digestibility of Eragrostis tef hay, and found similar results to Dehority & Scott (1967). However, assessments in all three studies used mostly the same strains.

As rate of passage and rate of fermentation compete in the rumen, only bacteria that ferment substrate at rates greater than the passage rate will be competitive ( Colucci et al. 1982 ). If extent of digestion does not reflect the rate of digestion, then this is an important consideration when bacteria are ranked for digestive potency. We compared the rate and extent of digestion of three ruminococci which differed in their ability to digest fibre ( Fig. 3) and found that rate and extent of digestion were highly correlated.

Lucerne has a greater soluble fraction than rhodes or spear grass and DM digestibilities were usually higher ( Table 2). These differences in the easily digestible fraction of a forage have implications for diet composition and competitive fitness of ruminococci. Fibrolytic bacteria such as Butyrivibrio fibrisolvens grow more rapidly on the soluble component of forages than ruminococci and produce bacteriocins that inhibit ruminococci growth ( Kalmokoff & Teather 1997). As B. fibrisolvens requires amino acids as a nitrogen source and ruminococci, only ammonia ( Hespell et al. 1997 ), a forage diet low in soluble nutrients (rhodes or spear) and with only a urea supplement might select for ruminococci. van Gylswyk (1970) found that a teff (Eragrostis tef) diet fed to sheep selected for ruminococci in preference to butyrivibrios, and that supplementation with urea and branched-chain volatile fatty acids increased their numbers.

Clustering based on genotype ( Fig. 4a) did not group bacteria according to their 16S rDNA determined species ( Table 2). This ‘mismatch’ is not particularly surprising because 16S rRNA phylogeny is based on highly conserved genes ( Pace 1997) while ERIC, REP and 16–23S rDNA spacers are based on chromosomal elements that are not necessarily stable and may be evolving at significantly different rates to the 16S rDNA genes ( van Belkum et al. 1998 ). There was, however, a very close relationship between phenotype ( Fig. 4b) and genotype ( Fig. 4a) based clustering, indicating that certain genetic types were associated with fibrolytic ability.

The above conclusions are supported by work done on Bradyrhizobium with ERIC, REP and 16–23S rDNA spacers ( Vinuesa et al. 1998 ). Root nodulating bacteria from a variety of geographical sources were analysed for 16S rDNA RFLP, ERIC, REP and 16–23S rDNA spacer polymorphisms and ability to nodulate legumes. There was not a good relationship between 16S rDNA grouping and grouping determined with ERIC, REP and 16–23S rDNA spacers. There was, however, a strong correlation between ERIC, REP and 16–23S rDNA spacers derived genotypes and nodulating ability, a phenotype fundamental to the competitive fitness of Bradyrhizobium.

The ruminococci are clearly a diverse group of bacteria and plant digesting ability and genotype differs widely between strains. Significant observations were that ranking’s of bacteria remained virtually the same on all forages but changed with cellulose discs ( Table 2), and that bacteria performing well on plant substrates have the best opportunity for success in the rumen. It is likely that ruminal communities are made up of a variety of ruminococci with digestive abilities not unlike those in Table 2. However, there are few studies which have indicated ruminal manipulations (dietary or otherwise) that would enable a different/introduced population to establish in the rumen. A significant research effort in this area is an essential element in furthering our understanding of the ruminal ecosystem.


The authors would like to thank Dr Keith Gregg for providing the AR strains which were originally isolated by Drs Tom Bauchop and Frank Hudman at the University of New England, Australia. Useful discussions with Drs Graeme Attwood, Brain Dalrymple and Rod Mackie are greatly appreciated. This work was partially funded by the Australian Meat Research Corporation.