Effect of Zeolite on Small Intestine Microbiota of Broiler Chickens : A Case Study

After ban of antibiotics growth promoters (AGP) in Europe in 2006, use of non-medicated gut microbiota regulators as feed additives has dramatically increased. This study aimed at describing the effects of a copper-exchanged zeolite on broiler growth performance, small intestine morphology and microbiota composition. Illumina Sequencing of the V3-V4 region of the 16S rRNA gene was employed to study the small intestine microbiota. This microbiota with copper-exchanged zeolite treated-chickens was significantly less diverse with an almost exclusive presence of Lactobacillus johnsonii and Lactobacillus reuteri. These Lactobacilli are correlated with increased diameter, length and weight of the three segments of the small intestine and decreased viscosity of the intestinal content, suggesting probiotic action. The tested copper-exchanged zeolite would act as a prebiotic, selecting a “favorable” flora for the healthy broilers development.


Introduction
Domestic chickens have a special place in society.Commercial poultry production is one of the most important sources of animal protein for human consumption, not suffering from any cultural or religious interdiction and is an important economic activity in many countries, with over 60 billion birds used in the production of meat and eggs each year.Despite their strong impact on technical and economical farm performance, little is known about microorganisms housed in the chicken gastrointestinal tract (GIT).Yet, the complex microbial community (microbiota) of the gastrointestinal tract plays an important role in the health of animal and can be considered as an important metabolic "organ".Composition of the intestinal microbiota is dynamic with spatial shifts along each GIT region in relation to environmental changes [1].The entire GIT of chicken is estimated to house 640 species of bacteria from 140 different genera [2].
Sub-therapeutic concentrations of antibiotics, known as antibiotics growth promoters (AGPs) have long been used in the food-producing industry.Their addition enhances feed efficiency, reduce mortality and improve the overall health of livestock [3].Their mode of action is thought to be due to a direct or indirect overall reduction or modification in bacterial numbers, as it is strongly suggested by their lack of effect on axenic broilers [4].The proposed mechanisms involve a reduction of microbial nutrient utilization, an enhancement of nutrient absorption due to a thinner mucus layer and healthy functional enterocytes, a decreased production of unwanted bacterial metabolites such as toxins and a reduction of intestinal infections [5].Direct action on the host's intestinal immune functions have also been suggested [6].Finally, their addition in animal's feed results in a decrease of nutrient amounts needed to produce a market-size chicken and enhances birds' growth without negative effects on meat quality if withdrawal delays are observed [7].The first evidence of AGPs performance effect dates back to 1940 [8] and has since been echoed by many studies [9] [10].The development of intensive livestock farming based on confinement of high number of birds, therefore increase risk of bacterial disease development and unbalanced gut microbiota.Subsequently, AGPs have been used in routine for decades to prevent disease and improve zootechnical performances [11].However, their overuse has contributed to the emergence of drug resistant bacteria and to the accumulation of antibiotic residues in animal products and environment [12] [13] [14] [15].Thereby, AGPs were banned from farming practices in 2006 in Europe, letting farmers to increased mortality, degraded techno-economical results and decreased animal welfare.This raised the need for safe and efficient alternatives that could increase nutrient availability for the animal, improve host immunity and intestinal microbiota [16].
As a consequence, in this study, we investigated the influence of a patented copperexchanged zeolite (B-SAFE, Pancosma) on broilers microbiota composition compared to negative control and AGP treatment.

Animals and Experimental Design
Animal experiments were performed in accordance with the Invivo NSA ethical committee in animal experimentation (CNREEA Registration Code C2EA-52, Protocol no.2012-12-20 4R).This study was designed to obtain intestinal microbiota samples from control chickens and from chickens receiving either copper-exchanged zeolite or zinc bacitracin that had significantly higher growth compared to negative control.A total of 105 unvaccinated 1-day-old male chicks (strain Ross PM3) were obtained from a local commercial hatchery and grown over a 21-day experimental period.Thirty-five chicks for each groups were randomly and equally assigned to one of the 3 dietary treatment groups: 1) "control group" exclusively fed with a corn-soybean-wheat diet "zeolite group" that received the same diet as the control group, supplemented with 6 mg of a patented copper-exchanged zeolite/kg of feed (B-SAFE, Patent no.FR 05 03671, Pancosma) and 3) a "bacitracin group" that received the same diet as the control group but was continuously supplemented over the whole trial period with sub-therapeutic doses of soluble zinc bacitracin (Bacivet S ® , Zoetis) at 30 mg/L in drinking water, equivalent to 48 ppm of zinc bacitracin in feed.Each chicken group was housed in a 1-m 2 woodchips litterpen in a controlled broiler house.Environmental conditions were adapted to animal's needs: temperature progressively dropped from 33˚C to reach 27˚C at 21 days, light progressively dropped from 24 h of light the first 4 days, to reach 6 hours of night per day in 21 days.Confinement at high stocking density (45.5 birds/m 2 ) was chosen to fit the intensive poultry farming conditions.Experimental sections were separated by two empty ones to prevent contact between animals and litter exchanges.For all groups, feed and water were provided ad libitum.On day 21, all birds were picked from each treatment, avoiding birds with extreme weight.These animals were individually weighed and slaughtered by electrical stunning and bleeding.Small batches of chickens from the same experimental groups were designed for slaughtering, in order to limit the delay between death and intestinal sampling to avoid intestinal microbiota post-mortem disturbance.Each animal was manually eviscerated and the small intestine (from the output of the gizzard to the beginning of caeca) was carefully collected, transferred into a sterile stomacher bag, weighed and rapidly frozen at −20˚C then stored at −80˚C.

Gut Morphology
Small intestine samples (N = 35 by sample) were brought back to room temperature.Gut diameters, length and weight were measured separately for each of the three small intestinal sections, distinguished using anatomical criteria; duodenum was considered from the first few centimetres to the output of the duodenal loop, the jejunum was sampled from the end of the duodenum to the Meckel's diverticulum and the ileum was sampled from the Meckel's diverticulum to the ileo-caecal junction.All measures were done, on the same day and by the same persons to ensure uniformity.Viscosity of intestinal content was observed during DNA extraction preparation and evaluated using an adaptation of a simple published procedure [33], measuring the time required for a constant volume of liquid to drain a plastic pipet, at room temperature.

DNA Extraction
Whole small intestines were suspended in 10 ml lysis buffer ASL (Qiagen, France), placed in filter stomacher bag and submitted to two one-minute consecutive cycles in an AES Smascher (BioMérieux, France).Two lysis were performed by treatment, samples were labelled A for control, B for zeolite and C for bacitracin and numbered with 1 and 2. These two extractions were pooled to give A, B and C samples numbered 3.Each group was separated and mixed in two separated batches (Figure S1).DNA was then extracted from each samples (from A1 to C3) using a bead-beating procedure.Briefly, 0.5 g of beads of diameter 425 μm and 0.5 g of beads of diameter 600 μm (Sigma Aldrich, France) were added to 1.5 ml of the lysate and homogenized on a bead-beater (Scientific industry, UK) for 10 min at full speed.The samples were then heated at 70˚C for 15 min, followed by centrifugation (9300 g, 5 min), to separate DNA from glass beads and cellular debris.Then DNA was extracted using QIAamp DNA blood mini kit (Qiagen, France) following manufacturer recommendations.

16Sr RNA Amplification and Sequencing
The V3-V4 region was amplified from each batch of each treatment group three DNA samples with the primers F343 and R784 using 30 amplification cycles with an annealing temperature of 65 degrees (an amplicon of 510 bp, although length varies depending on the organisms).Because MiSeq enables paired 250-bp reads, the ends of each read are overlapped and can be stitched together to generate extremely high-qua-lity, full-length reads of the entire V3 and V4 region in a single run.Single multiplexing was performed using homemade 6 bp index, which were added to R784 during a second PCR with 12 cycles using forward primer (5'-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGAC-3') and reverse primer (5'-CAAGCAGAAGACGGCATACGAGAT-index-GTGACTGGAGTTCAGACGTGT-3').The resulting PCR products were purified and loaded onto the Illumina MiSeq cartridge according to the manufacturer instructions.The quality of the run was checked internally using PhiX, and then each pair-end sequences were assigned to its sample with the help of the previously integrated index.Each pair-end sequences were assembled using Flash sofware using at least a 10 bp-overlap between the forward and reverse sequences, allowing 10% of mismatch.Reads that could not be assembled were discarded.Secondly, all sequences were rechecked to eliminate those corresponding to internal control PhiX, those that still contained N bases and those that correspond to the host "Gallus gallus".

Ecology Diversity and Taxonomic Identification
Chicken small intestinal microbiota were analysed using both VAMPS (Visualization and Analysis of Microbial Population Structures [34]) and MG-RAST (Metagenome Rapid Annotation using Subsystem Technology, [35]) online servers.The species observed and estimated richness were evaluated using rarefaction curves and nonparametric ACE and Chao1 tests [36] [37].Alpha-diversities were estimated using Simpson- [38] and Shannon-indices [39].Differences between microbiota were assessed using principal coordinate analysis (PCoA) on Bray-Curtis distances and unweighted Unifrac [40].Three-dimensional PCoA plots were generated using EMPeror software [41].For taxonomic assignment and differential comparison of microbial class, order, family and genus, we compare SILVA [42], RDP-II [43] and Greengenes [44] databases.Reference sequences were only Bacteria (classified and unclassified sequences).For VAMPS, we used defaults parameter whereas for MG-RAST, we set the thresholds to 95% of identity and E-value of 1 e−100 .For species determination and comparison, we used a homemade manually biocurated database, composed of V3V4 regions for all non-redundant bacterial genus identified in our samples.All these extracted sequences were aligned to determine if they can discriminate the bacterial species at a threshold value of 97%.All assignations were based on 97% or greater identity.Sequences under this threshold were not assigned and grouped under "unclassified" taxon.16S copy number correction was performed on genus representing more than 1% using rrnDB (https://rrndb.umms.med.umich.edu).

Statistical Analysis
Numeric data collected were represented as mean ± standard deviation.D'Agostino-Pearson and Shapiro tests were performed to check the normality of data distribution.Data presenting normal distribution were compared using an unpaired Student's t-Test while data with a free distribution were compared by a non-parametric Mann-Whitney test.For both statistic tests, differences were considered statistically significant at p < 0.05.Tests were carried out using GraphPad Prism v6 (GraphPad Software, San Diego, CA).

Zootechnical Conformity of the Study
The objective of the study was to compare intestinal microbiota of chickens receiving efficient growth promoting additives to microbiota of negative control birds.Birds from copper-exchanged zeolite group were significantly heavier than birds of control group (809 g vs 762 g, p < 0.05).Birds from zinc bacitracin group were also significantly heavier than control group (852 g vs 762 g, p < 0.05).As a consequence, both supplemented groups were significantly heavier than the control.Those randomly sampled animals were then qualified for microbiota investigations (Figure 1).

Comparative Gut Morphometry
Copper-exchanged zeolite and zinc bacitracin supplemented chickens had significantly increased small intestines length and weight when compared with the control group (Figure 2(a)).Interestingly, diameters of jejunum from birds that received the control diet were significantly smaller when compared with birds in those supplemented diets.We observed the same trend for ileum and duodenum segments although the differences were not statistically significant (Figure 2(b)).Noticeably, the addition of copper- exchanged zeolite to the diet resulted in a significantly higher weight and length of the jejunum and ileum when compared to both control and antibiotic treatments (Figure 3).The supernatant-fraction of stomached control small intestines were more viscous than those from supplemented groups, so much that pipetting was difficult for these samples, in accordance with previous description [33].

Microbiota Ecology
The PCoA plot showed a good clustering of triplicate samples (control triplicates: A1, A2, A3, zeolite triplicates: B1, B2, B3 bacitracin triplicates: C1, C2 C3) collected from each dietary treatment with a less defined clustering of bacitracin samples (Figure 4).This observation leads to the conclusion that both bacitracin and copper-exchanged zeolite influence the small intestine bacterial population and that the three bacterial populations are different.Interestingly and unexpectedly, the two supplemented diets (bacitracin and copper-exchanged zeolite), which lead to higher weights in the 36 individuals, have widely separated microbial populations.
To verify adequate sequencing depth of the three ureing datasets, rarefaction curves were performed, at both bacterial phylum and genus levels.These curves indicate a suitable depth of coverage.Indeed, although they do not reach the saturation phase, their slopes tend to plateau demonstrating, that a large part of the bacterial diversity was reached (Figure S2).A comparable number of phyla (12 to 13) was observed for all samples and the rarefaction curves tend to show that the bacitracin-treated community is less rich than control and copper-exchanged zeolite samples, reaching 138 versus 156 genus.This low richness is further confirmed by both observed (S) and estimated (Chao, ACE) richness curves (Figure S3) but these differences were not statistically significant.However, both Shannon and Simpson diversity indices show a small intestine microbiota of chicken treated with copper-exchanged zeolite less diverse in comparison with control and antibiotic diets.

Phylum Comparison
The predominant phylum in each group was Firmicutes, accounting for almost 90% of all sequences (Figure 5 and Figure S4).However, this number was found significantly different between animals treated with the feed additives with a high increase for copper-exchanged zeolite, reaching 99%.For control and antibiotic treatments, Proteobacteria and Actinobacteria phyla were also detected.Bacteroidetes or Bifidobacterium could not be detected in any of the treatments.Already at the higher taxonomic rank level, the three databases used (Silva, RDP and Greengenes) did not give the same results in terms of relative numbers.Since it is difficult to compare these databases and to comprehend the reasons for these differences, as described previously, a manually biocurated database dedicated to the three Phyla detected (Actinobacteria, Proteobacteria and Firmicutes) and corresponding to the V3-V4 region used in this study was created.The quantities given by our database (text in red, Figure 5) were compared to the average provided by the three public databases.Values for Actinobacteria were consistent whereas, for the Firmicutes and Proteobacteria, significant differences were observed, especially for the samples treated with bacitracin.However, no major influence in determining the phylum was observed.

Genus Comparison
As at phylum level, comparison of the different databases showed consistency for Actinobacteria whereas differences, ranging from simple to double, were observed as example for the beta-Proteobacteria and Enterococcus (text in red, Figure 6).These results do not influence the analysis because phyla have identical profiles.The comparison of genera is based on our database.For the three conditions tested, Firmicutes were largely dominated by Lactobacillus.However, two opposing tendencies were detected compared to control: 1) enrichment for the copper-exchanged zeolite diet and 2) depletion for the bacitracin group with a relative increase in favor of Enterococcus.Streptococcus and Clostridiales were found depleted in both food additives sets, virtually undetectable for the copper-exchanged zeolite treatment (Figure 6, Table 1).The Proteoacteria are mostly represented by the gamma subdivision, primarily Shigella and Escherichia and especially for chicken treated with the bacitracin (Table 1).The copper-exchanged zeolite group seems to be the only one to contain  alpha-and beta-Proteobacteria (18% and 8.8%) but in a too low proportion to be considered as significant.The Actinobacteria displayed a majority of Corynebacterium with the exception of the bacitracin set that contains a majority of Brevibacterium.A significant depletion of Brachybacterium was observed in the copper-exchanged zeolite group.However, as for the Proteobacteria, the total percentage of Actinobacteria was too low in this sample (less than 0.5%) to be interpreted (Figure 6).

Species Comparisons
For the 9 preponderant order/families/genus that summarize the three microbiota studied (Figure 7), we first checked if species could be unambiguously differentiated using the 16S rRNA V3-V4 region.Concerning the Lactobacillus, the three majority species (L.johnsonii, L. reuteri and L. gasseri) have very similar sequences but are perfectly distinguishable as shown in the tree of distance regardless the identity percentage cut-off used (Figure S5).The vast majority of assignments were completed between 100% and 98% identity.For Streptococcus, all sequences were restricted to the S. bovis group [45] but species are hardly or not even separable (Figure S6).We obtained similar results with the Enterococcus, all sequences belonging to the group of E. faecium, E.  ratti, E. villorum, E. hirae, E. durans and E. mundtii species that cannot be differentiated using this marker (100% identical, data not shown).This was the case also for, Escherichia and Shigella since the three identified species, E. coli, E. fergusonii and S. sonnei are also indistinguishable (Figure S7).Finally, for the genus Clostridium, all sequences belonged to the Peptostreptoclostridium group and separated into a major new species akin to P. bifermentans and a minoritory group (less than 1%) corresponding to P. difficile (Figure S8).Based on this information, we compared the three groups and showed that the bacitracin treatment results in depletion in L. johnsonii and disappearance of Streptococcus group bovis in favour of an enrichment in Enterobacteria (Escherichia, Shigella), Clostridium bifermentans-like and Enterococcus of faecium group (Figure 7).While there was a shift of microbiota for this treatment, there was a complete reduction of microbial complexity for the copper-exchanged zeolite treated animals with the exclusive presence of L. johnsonii and L. reuteri, the latter being strongly enriched compared to other diets (Figure 7).

Discussion
Final average body weight was 10 to 20% less than the genetic potential of the bird's strain (http://en.aviagen.com/assets/Tech_Center/Ross_Broiler/Ross-PM3-Broiler-PO-2014-EN.pdf).Commercial performances are often 10% lower than the genetic potential due to non-ideal environment of farms.In this study, stocking density was the highest that can be found in European field conditions, in order to create room of improvement for the tested additives and make sure that growth of treated groups will be significantly improved compared to the control.As a consequence, the level of performance of birds is in coherence with expectations.Animal's growth was not significantly different between copper-exchanged zeolite group and bacitracin group, and both groups were significantly heavier than the control group.Moreover, copper-exchanged zeolite seemed to enable a more homogeneous population than the 2 other groups.Anyway, these statements have to be taken with care since average weight and standard deviation were measured on birds, avoiding the extreme individuals.Therefore, the zootechnical conclusions may not be extrapolated to the whole population.Anyway, the study was designed to provide intestinal samples of animals that had significantly different weights, due to the addition of growth promoters in their diet, and this objective was achieved.
A correlation between small intestine total weight and chicken body weight was observed which is consistent with the studies of Yang et al. that showed that the weight of the intestine increased with body weight and Jamroz et al. that indicated that the increase in small intestine weight allows broiler chickens to reach a heavier body weight faster than control chickens [46] [47].In our study, both fed additives increased the diameter of the jejunum, and to a lesser extent, of duodenum and ileum.The main important morphometric changes resulted in heavier and longer jejunum and ileum segments.Thus, copper-exchanged zeolite increased the small intestine surface volume ratio and consequently may increase both the absorptive area and the mucosal hydrolysis capacity [46].It would have been interesting to measure the intestinal villi length to confirm this increase of absorption surface.In our study, the basic diet consisted mainly of wheat, corn and soybean that contain non-starch polysaccharides (NSPs).The increase in viscosity of intestine extract in the control chicken can be derived from a significant amount of soluble NSP in the bolus that would lead to poorer growth performance [48].Indeed, it has been shown that higher viscosity of the intestinal contents reduce the speed of feed bolus and digestive enzymes diffusion, delay their action and subsequently, reduce the digestibility of nutrients [49] [50]).We could therefore think that the decrease of the digesta viscosity observed in supplemented animals may contribute to better nutrient retention and could subsequently explain the enhanced growth performance [51].The reduction of intestinal contents viscosity is often associated with activity of microbial NSP-degrading enzymes such as phytases [52], xylanases [50] or beta-glucanases [51].So, we could hypothesize that quantitative and qualitative changes observed in the microbial GIT communities of both supplemented groups of animals could be responsible for a better NSP fermentation, improving nutrient digestibility and then growth performance and decrease the intestinal viscosity as previously observed in former studies [53].
The strategy choosen in this work was to study treatment-dependent pooled samples, in order to average the impact of inter-individual microbiota variability .This strategy was essential to drive effectively this pilot experience.In the light of our first results, we will soon set up an experimental protocol allowingthe study of these individual variations.In addition, the Best-Hit assignment method was preferred to OTUs approach as more accurate as the clustering approach which gave very variable results depending on the tools used (U-search, SLP or CD-hit).
As in previous studies, we showed that main phyla present in the chicken small intestine is Firmicutes and to a lesser extent Proteobacteria.It was also shown that bacitracin and copper-exchanged zeolite supplementation differentially modulate microbiota composition [54].But contrary to Wei et al, no Bacteroidetes were found in our samples [55].
Treatment with copper-exchanged zeolite significantly reduced the bacterial phyla diversity present in the small intestine of broilers and this depletion led to a predominance of Firmicutes (ca.99%) and an elimination of pathogenic genera such as Enterococcus, Shigella or Escherichia.These results suggested that copper-exchanged zeolite acts as a selector that can either inhibit the growth of pathogens or promote the growth of Lactobacillus species, supporting a healthy digestive system through "competitive exclusion" of pathogenic bacteria [56] [57], acidification of the GIT through lactic and acetic acids production [58], stimulation of the immune system [59] [60], production of antimicrobial compounds like reuterin [61], gassericin [62] or lactacin [63] and maintenance of the intestinal barrier integrity [64].The mechanism responsible for the selection of flora remains to highlight.As Lactobacillus are predominant in crop and gizzard [65], it would be interesting to investigate if copperexchanged zeolite supplementation also enriched the upper GIT of copper-exchanged zeolite treated chicken.
In our study, we have also demonstrated that, even if both treatments tend to improve, with varying intensity, the growth performance of chickens, their impact on the intestinal microbiota differed totally.Some authors showed similar results [54].Thus, unlike copper-exchanged zeolite, sub-therapeutic treatment with zinc bacitracin caused a substitution of a part of the Lactobacillus by Enteroccoccus, acting as probiotics and explaining growth improvements observed [18].
As zinc bacitracin normally targets Enterococci [66], this result suggested that the selected species of our study could be resistant to this antibiotic.However, to identify these species, we faced the fact that, if V3-V4 region of the 16s RNA is suitable to distinguish most bacterial species, it seemed insufficiently variable to identify closely related ones as previously described [67] [68].Nevertheless, for the five species of Enterococcus recruited, E. ratti and E. villorum primarily described in pigs and rats [68] [69] and E. hirae, associated with morbid events in broilers [70] can logically be excluded in favour of E. faecium and E. durans, both commensal of poultry small intestines [71]), and having a high prevalence of bacitracin-resistance [72] [73] [74].
Escherichia was previously reported to be more prevalent in small intestine of chickens that consumed zinc bacitracin [75], which supports our observation.Indeed, the antibiotic treatment increased a community of Proteobacteria indistinctly identifiedas Shigella and/or Escherichia species.
The only clear common effect of both treatments is the elimination of Streptococcus bovis Streptococcus equinus complex (SBSEC) community.Streptococcus is a complex genus with a frequently amended classification [76].Still, of the three possible species (or subspecies) present in the control samples and lacking in the two additives diets, S. infantarius remains the most probable since already described in chicken's intestine and susceptible to bacitracin [75] whereas S. macedonicus is mostly associated with food environment [77] and S. pasteurianis is mostly part of the normal flora ofhumans, also isolated from various infection sites [78].In addition, S. infantarius harbors less virulence factors than other genus of the S. gallolyticus group, which would explain their negative effect on "control" chick's growth without pathogenic symptoms [79].
It would however be interesting to duplicate the experience with a new batch of chicks from a different hatchery to see if the orientation of the flora is repeatable or if it depends on the initial flora.

Conclusion
In conclusion, our study provides new information on the compared effects of zinc bacitracin and a copper-exchanged zeolite on broiler chickens intestinal microbiota.
Tested copper-exchanged zeolite, despite differences with the antibiotic in terms of weight gain, enabled higher weights than control on birds and seems to act on chickens small intestinal morphometry and health.This additive promoted the colonization of beneficial Lactobacillus could result in a direct benefit for the chicken health and growth.Submit or recommend next manuscript to SCIRP and we will provide best service for you: Accepting pre-submission inquiries through Email, Facebook, LinkedIn, Twitter, etc.A wide selection of journals (inclusive of 9 subjects, more than 200 journals) Providing 24-hour high-quality service User-friendly online submission system Fair and swift peer-review system Efficient typesetting and proofreading procedure Display of the result of downloads and visits, as well as the number of cited articles Maximum dissemination of your research work Submit your manuscript at: http://papersubmission.scirp.org/Or contact fns@scirp.org

Figure 1 .Figure 2 .
Figure 1.Graphs showing chicken body weight (BW) distribution as well as t-test results.Control in red circles, zeolte in green squares and bacitracin in blue triangles.

Figure 3 .
Figure 3. Weight (a) and length (b) of the three segments of the small intestine presented in by graphs.The three segments (duodenum, jejunum and ileum) and three treatments (blue: Zn-bacitracin, green: B-SAFE copper-exchanged zeolite, red: control diet) were compared.*P < 0.05.

Figure 5 .
Figure 5.Comparison taxonomic assignments at phylum level using Silva, RDP and Greengenes database with VAMPS and MG-RAST online classifiers.(a), (b) and (c) show percentage of Firmicutes, Proteobacteria and Actinobacteria for each treatment (blue bacitracin, red: control and green: copper-exchanged zeolite).Numbers in black represent the mean value obtained for the three 16S reference databases and the two classifiers while values in red represent the result obtained against our dedicated database.Significant differences are shown by * (p < 0.05) and ∆ represents significant differences between the results obtained with the public and dedicated databases.

Figure 6 .
Figure 6.Comparison of order/family/genus taxonomic assignments using Silva, RDP and Greengenes database with VAMPS and MG-RAST online classifiers.(a) Firmicutes major genus: Enterococcus, Lactobacillus, Clostridiales, (b) Proteobacteria major subdivisions (c) Actinobacteria genus: Corynebacterium, Brachybacterium and Brevibacterium.(Blue bacitracin, red: control and green: copperexchanged zeolite).Numbers in black represent the mean value obtained for the three 16S reference databases and the two classifiers while values in red represent the result obtained against our dedicated database.Significant differences are shown by * (between each treatment) (p < 0.05) and ∆ (between our database and 3 others studied).

Figure 7 .
Figure 7. Contingency dots showing the distribution of the species in the small intestine microbiota of chicken for the three diets.The six most represented bacteria are colored dot-symbolized according to their proportion in each experienced-food groups.

Figure
Figure S3.Microbiota richness and comparisons between control and treated diets (blue: Zn-bacitracin, green: B-SAFE copper-exchanged zeolite, red: control diet).(a) Observed (S) and estimated (Chao/ACE) richness.(b) Shannon Weaver and Simpson index curves used to estimate alpha diversity (i.e., a combined assessment of the number of unique bacterial taxa and their abundance) of each sample.

Figure S4 .
Figure S4.Left: Rotary phylogenetic representation of the predominant microbial composition of small intestine for each feed treatment (yellow: Proteobacteria, bright green: Actinobacteria, light blue: Firmicutes and grey: minority phyla Bacteroidetes and it associated genus: Alistipes).Blue points represent the rate of presence, along a gradient from the darkest (for the high percentages) to the clearer (for the low percentages).Outside circle: copper-exchanged zeolite.Middle circle: control.Inside circle: bacitracin.Right: Phylum repartition for each diet treatment (same color code as left).

Figure S5 .
Figure S5.Distance tree showing the effectiveness of distinction of the species of Lactobacillus johnsonii, L.reuteri and L.gasseri, recruited to identity thresholds from 95 to 100%.Dot representation shows the distribution of the three species in each diets.

Figure S6 .
Figure S6.Distance tree showing that all Streptococcus species belong to group bovis and are mostly recruited at 99% identity threshold.

Figure S7 .
Figure S7.Distance tree showing that Shigella-Escherichia clustered together and are mostly recruited at 100% and 99% identity threshold.

Figure S8 .
Figure S8.Distance tree showing that Clostridia genus clustered in the Peptostreptoclostridium in two groups related to C. bifermentans and difficile.

Table 1 .
Bacterial genus reads distribution for small intestine, by treatment group using our manually biocurated database.