The equine intestinal microbiota is complex and has an enormous impact on digestion, immune stimulation, pathogen protection and metabolism [1]. Studies investigating the equine intestinal microbiota commonly use fecal material because of the ease of sampling [2–9]. Significant differences in the microbial composition along the gastrointestinal tract have been reported in dogs and cattle [10, 11], but to date few studies have assessed whether microbial populations in feces are representative of proximal compartments of the GI tract in horses. Data from other species cannot be extrapolated to horses due to significant differences in anatomy, function, and microbiota composition. None of the studies have evaluated the entire microbial population of the equine gastrointestinal tract as only specific bacterial populations or compartments were analyzed. Furthermore, equivocal conclusions were drawn [12–15]. For proper interpretation of results it is important to understand how well the bacterial populations in feces will mimic those in proximal segments of the intestine.

As the majority of microorganisms in the intestinal tract are not cultivable with standard methods [2, 16], molecular methods have recently been adapted to study the composition of the intestinal microbiota [4, 5, 12, 16]. Terminal restriction fragment length polymorphism (TRFLP) is a 16S rRNA gene PCR based method that can be used for rapid comparison of complex microbial communities, present in the equine gastrointestinal system [17]. This method can be used to compare multiple samples, to follow spatial and temporal changes in microbial populations, and to study bacterial community dynamics in the GI tract of humans [18] and animals [11, 19], including horses [5].

The objectives of this study were to assess the similarities of the microbial populations of the duodenum, ileum, cecum, colon and rectum (feces) within and between healthy horses using TRFLP.

This study identified limited similarities in richness (Cs 0.35-0.67) of the gastrointestinal microbial populations between the small intestine, large intestine and feces within individual horses. The microbial composition in fecal samples was only partially representative of the proximal gastrointestinal compartments with the closest similarity observed between fecal and cecal samples. There were also limited similarities in richness (Cs 0.44-0.74) when comparing the same intestinal segment between horses suggesting that substantial portions of the equine intestinal microbiota are unique to each animal.

When comparing gastrointestinal compartments within a horse, the low similarity indices between compartment pairs indicated major differences in the richness of the microbiota along the gastrointestinal tract. This is consistent with results from a recent study by Dougal and co-workers (2012), who showed that that the microbial composition of the cecum is significantly different from the microbiota of the colon and feces. This is also consistent with results from other animal species; one study in cattle reported significant differences in the microbial community as ingesta travelled along the gastrointestinal tract [11]. Likewise in dogs, marked variation between compartments within dogs and within compartments between dogs was noted [10].

Rectal samples only had limited similarity to other compartments (Cs 0.39-0.67). Overall the cecum had the highest similarity index to the rectum (Cs 0.67), which could be due to separation mechanisms or actual closer similarity. These data are in agreement with findings from Milinovich et al. (2007), who found that relative abundance of selected fecal bacterial populations, were similar but not consistently representative of cecal bacterial populations. In contrast to the above study, the method used in our study included all bacterial species and further strengthens the suggestion that fecal samples are not consistently representative of proximal compartments.

Dougal et al. (2012) investigated the microbial populations of the cecum, colon and feces of three healthy horses and five healthy ponies using TRFLP. Similar to our study, the authors found that the microbial populations of the cecum were significantly different from feces; however, in contrast to our study the microbial population of the colon and the cecum were not significantly different [15]. In their study the compartment specific data was not analyzed separately for horses and ponies, and significant difference between the microbiota of horses and ponies were identified. Thus, the inclusion of ponies could have influenced the compartment specific data and could explain the difference in results of their and our study. Hastie et al. (2008) also found that the microbiota of the feces is representative of the distal colon; however, only certain bacterial species (Ruminococcus flavefaciens, Fibrobacter succinogenes and Streptococcus bovis) where tested, therefore potentially biasing results. Schoster et al. (2012) investigated the presence of C. difficile in various intestinal compartments of horses and found that rectal samples were only positive in 63% of animals from which C. difficile could be isolated from one or more proximal compartments. The data from the present study lend further credence to concerns that the fecal microbiota is not representative of the entire gastrointestinal tract. It further supports the concerns raised about the accuracy of using fecal samples for investigation and interpretation of changes in the microbiota of proximal gastrointestinal compartments, which is commonly performed [2–9].

The differences in microbiota along the gastrointestinal tract seen in our study have been observed in other studies in horses [15], as well as other animal species [10, 11], and humans [20].

Along the small intestine of the horse the secretion of sodium bicarbonate and bile neutralizes the hydrochloric acid from the stomach, thus favoring acid-tolerant bacteria by raising the pH of the ingesta to approximately 7.0 by the time it reaches the large intestine. Concurrently, an increase in the total amount of cultivable bacteria is seen [21]. The ingesta entering the cecum is rich in cellulose that is only minimally digested within the small intestine. This suggests that a largely cellulolytic and proteolytic bacterial population would be present in the cecum [22]. In the colon fiber fermentation occurs, thus requiring a shift in bacterial species [23]. The right dorsal colon has limited capacity to digest fiber [24]. Soluble carbohydrate and starch bypasses digestion in the small intestine and reaches the colon [23]. Culture independent studies have reported the total bacterial population to be higher in the colon than the cecum, which could explain the higher digestibility values in these regions [25, 26]. This suggests that the local microbial populations within different parts of the gastrointestinal tract are connected to the anatomy, function and composition as well as pH of the ingesta [15, 27].

Significant differences in the T-RF profiles of the duodenum and cecum between the Bashkir Curly horses and the remaining horses were observed (Figure 3). Thus, similarity indices between horses were only compared based on data from the six Bashkir Curly horses, which originated from the same farm and were under similar housing and dietary management. Dietary history of the remaining four horses was not available; therefore, it cannot be speculated why these differences were present. The Bashkir Curly horses were affected by equine degenerative myeloencephalopathy, which is not known to cause changes in mentation or feeding behavior potentially altering intestinal microbial populations.

Mean similarity indices for each compartment compared between animals were low suggesting substantial differences in the microbial richness between horses in all gastrointestinal compartments. This is consistent with data from other studies that found a high level of diversity between horses [4, 5, 15]. Gronvold et al. (2010) investigated the fecal microbiota of 12 healthy horses using denaturing gradient gel electrophoresis (DGGE) and also assessed the influence of penicillin and anesthesia on the fecal microbiota. The authors demonstrated a very unique composition of the fecal microbiota for individual horses, with most of the variation attributable to the statistical effect of the individual horse, not anesthesia or penicillin administration. Willing et al. (2009) reported that consistency of the microbial composition of feces within an individual fed different diets was on average 73%, with a significant effect of horse on the diversity and stability of the gastrointestinal microbiota. Taken in relation to the results of these previous studies, our results support that the equine microbiota has a substantial portion unique to each animal. Such variability needs to be taken into account when comparing treatment effects between horses.

TRFLP allows for rapid assessment of complex bacterial communities, such as those present in the equine gastrointestinal system, and permit rapid comparison of the community richness; however, there are limitations. TRFLP cannot estimate bacterial community evenness or abundance, parameters necessary to evaluate bacterial diversity [17]. Phylogeny techniques based on 16S rRNA can identify strains to a family or genus level, but are of limited utility for differentiation of species as organisms can have almost identical 16S rRNA gene sequences and still belong to separate species based on DNA hybridization [28]. A additional limitation specific to the TRFLP technique is the inability to link a peak on the electropherogram to a specific bacterial family or genus without supplementary tests which could be accomplished in tandem with 16S rDNA clone library analysis [19]. Bacterial species identification is generally limited by the lack of a generally accepted bacterial species concept. Despite the existence of a bacterial species definition, which governs when a strain can be called a new species based on DNA hybridization or 16S rRNA gene sequencing, a true concept of what constitutes a bacterial species taking into account horizontal gene transfer, phenotyping and genotypic characteristics has not been agreed upon. Recently, a genomic phylogenetic species concept for the taxonomy of prokaryotes has been proposed. This concept is based on integrating information obtained from new molecular techniques such as multilocus sequence typing and genomic approaches into the existing phylogenetic concept [28].

These recent advances in the field of metagenomics make this a more attractive option for future studies to conclusively determine the differing composition of the equine microbiota in different intestinal compartments. This technique has recently been used to characterize the fecal microbiota of two healthy horses, which showed that the equine microbiota is more diverse than the human microbiota, but less diverse than the gastrointestinal microbiota of cattle [2]. Metagenomic analysis was also recently used by Costa et al. (2012) assessing the microbiota of healthy horses and horses affected with acute colitis. Their study showed profound differences in the microbiome of diseased horses compared to healthy horses. Determining the specific composition of the gastrointestinal microbiota was not an objective of this study; therefore, TRFLP was deemed adequate.

Comparable studies in other animals and humans have reported an average of 46–50 peaks per sample [11, 18]. Similar results were expected in horses, as a prior equine study had shown 138 different TR-F peaks, although the number per sample was not reported [5]. Therefore, the low number of peaks in our study was surprising. Previous studies considered every peak as a separate one, as opposed to clustering T-RF peaks based on controls and thus obtained a larger number of T-RF peaks. This could have partially accounted for the low number of peaks seen in this study [11]. Based on results from controls we chose to group peaks within three base pairs. However, others that used similar clustering of T-RF peaks still obtained approximately 80 unique T-RF peaks from fecal samples, indicating that this alone does not account for the low number of peaks seen in our study [18]. The low number of peaks could indicate insufficient separation of very similar bacterial species, suggesting that the primer and restriction enzyme combination used lacked discriminatory power, even though two different enzymes were evaluated in a pilot trial and the enzyme creating larger numbers of peaks was chosen. Five samples did not create any T-RF peaks despite of a positive PCR result, indicating that the DNA was lost during restriction enzyme digestion or TRFLP analysis.

The primers for the amplification of the 16S rRNA gene for all bacteria were adapted from a previous study in dogs [10]. The microbiota of dogs is likely significantly different from the equine microbiota, given the differences in diet, digestion and fermentation; however, there are few studies published on the overall composition of the equine microbiota [2, 8], therefore primers could not be chosen according to this knowledge. The 16S rRNA gene is known to be extremely conserved among all bacterial species known to date, therefore it was chosen as target for this study [28].

The sample size of 10 horses in our study was comparable to similar previous studies in horses [13, 15], and despite sample size limitations, this study contributes with new findings regarding the equine gastrointestinal microbiota.

There was a large variation between the two technical replicates in some samples. This could potentially reflect the inherent bias of the methodology (biased PCR amplification) or insufficient restriction enzyme digestion; however, some variations between technical replicates can be expected.

In conclusion this study showed that the microbial populations differ between gastrointestinal compartments of healthy horses. Fecal samples showed only moderate similarity to other gastrointestinal compartments and were most similar to the cecum. Similarity between the same compartments of different horses was low, further supporting that portions of the equine intestinal microbiota are unique to each animal.