The animal care and use procedures followed the Guide for the Care and Use of Agricultural Animals in Research and Teaching (Federation of Animal Science Societies, 2010) under supervision of a licensed poultry veterinarian. One thousand eight hundred fifty-six (1,856) one-day-old male Ross chicks were assigned to four (4) treatment groups with eight (8) replicate pens per treatment, and 58 birds per pen raised on new litter. The treatments were: non-supplemented non-challenged group (unchallenged control); non-supplemented and challenged control; challenged and supplemented with SB (70% of sodium butyrate protected with sodium salts of palm fatty acids; ChaSB; 0.1% inclusion); and challenged and supplemented with SB plus essential oils (70% of sodium butyrate protected with sodium salts of palm fatty acids, and carvacrol and ginger oil −0.5% each in the commercial product; ChaSBEO; 0.1% inclusion). Each experimental diet was mixed separately, and the feed additives were supplemented to replace kaolin in the unsupplemented diet. The products containing SB and SBEO were provided by Norel Animal Nutrition, Madrid, Spain.

Feed consisted of a non-medicated commercial-type broiler starter, grower, and finisher diets compounded according to NRC guidelines (NRC, 1994) and contained feedstuffs commonly used in the United States (Table 1). The feeds were available ad libitum from date of chick arrival as follows: Starter, d 0 until d 13; grower, d 14 to d 34; and finisher, d 34 to d 41 (study termination). Diets were fed as crumbles (starter feed) or pellets (grower and finisher feed).

The challenge model consisted of coccidial vaccine (Coccivac®-B52) administered at d 0 by spray cabinet, and individual inoculation with ~5,000 sporulated oocysts of Eimeria maxima by oral gavage on d 13, kindly provided by Dr. Lorraine Fuller, from the Department of Poultry Science, University of Georgia. On d 18, and 19, the same birds were challenged with C. perfringens, as follow: feed was withdrawn for 4 h and water was withdrawn for 2–3 h prior to administration of C. perfringens. A measured amount of water (~200 mL with C. perfringens) that was consumed within 30 min was used for each pen. A C. perfringens culture (~1 × 10⁸ cfu/mL total) was added to this water and thoroughly mixed and given to birds in each challenge pen. Once the challenge water was consumed, treatment feed and water were returned to pen.

Two birds per pen were euthanized by cervical dislocation and the ceaca were collected into a sterile Ziploc bag, immediately put in ice, and brought to the lab. Thus, the cecal content of the two birds was pooled in a new sterile Ziploc bag, and thoroughly homogenized by hand for further microbiota analysis. The content was then diluted in a 1:10 sterile phosphate-buffered saline solution (PBS; ph of 7.4); 1 mL of the solution was transferred to a 2 mL micro-tube, centrifuged for 3 m at 3,270 g, the supernatant discarded, and 200 μg of the content used for DNA isolation.

The DNA isolation was conducted following the manufacturer recommendations (PowerViral Environmental RNA/DNA Isolation Kit—Mo Bio; QIAGEN, Carlsbad, CA, USA), with a slight modification as described by Bortoluzzi et al. (2017). The quality and presence of DNA was verified by Nanodrop (NanoDrop 1000 Spectrophotometer; Thermo Fischer Scientific, Wilminton, DE, USA) and agarose gel electrophoresis (1.5%), respectively. Only the DNA with a 260:280 ratio of 1.80 to 2.00 in the Nanodrop was used for further analysis.

The V3-V4 hypervariable region of the 16S rRNA gene was amplified using the primer FwOvAd_341f and ReOvAd_785r as previously described (Klindworth et al., 2013). All the procedures followed a standardized protocol described previously (Bortoluzzi et al., 2017). The 16S rRNA gene was then sequenced using the Illumina Miseq platform.

All sequence data processing was performed using QIIME v. 1.9.1 software package (Caporaso et al., 2010). Sequences were paired-end and quality trimmed using Geneious (Newark, NJ). High-quality sequences were aligned against the SILVA database (Ribocon GmbH, Bremen, Germany) release 119 (Pruesse et al., 2007). UCHIME software (Tiburon, CA) was used to identify and remove chimeric sequences (Edgar et al., 2011). Number of sequences per sample was normalized based on the sample with the lowest number of reads for statistical comparison (Gihring et al., 2012). Operational taxonomic units (OTUs) were assigned at a 97% identity using SILVA database. Alpha (richness: Observed Species, Chao1; diversity: Shannon, and Phylogenetic Diversity (PD) of the Whole Tree), and beta diversity indices were calculated using QIIME v1.9.1. Nonparametric statistical tests PERMANOVA and ANOSIM were used to compare categories, and the similarity between the microbiota, respectively, using the weighted (quantitative) UniFrac metric measure. Principal Coordinates Analysis (PCoA) was used to visualize the data. Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) analysis was carried out using the KEGG and Clusters of Orthologous Groups of proteins (COGs) databases. For this, a closed-reference OTU table was normalized by the 16S rDNA copy number, the metagenome was predicted, and categorized by function based on Kyoto Encyclopedia of Genes and Genomes (KEGG; Uji, Kyoto, Japan) pathway (Kanehisa and Goto, 2000). The obtained biome file was processed by STAMP (Halifax, Nova Scotia, Canada) version 2.1.3 (Parks et al., 2014).

The data was analyzed as a completely randomized design, with 8 replicates each treatment (n = 8). From each replicate, the cecal content of birds was pulled for microbiota analysis. The frequency of the main bacterial groups observed was submitted to a non-parametric one-way ANOVA (Kruskal-Wallis test) and, in case of significant difference (P ≤ 0.05), means were separated by Dunn test, using SAS 9.4 (SAS Institute, 2011). Welch's T-test was applied to compare the KEGG pathways (P ≤ 0.05).

Chao index (minimal number of OTU present in a sample), number of observed species (number of species present in a community), phylogenetic diversity (PD) of the whole tree (minimum total length of all the phylogenetic branches required to span a given set of taxa on the phylogenetic tree), and Shannon index (abundance and evenness of the species present in a sample) were evaluated as measures of alpha diversity indices (Tables 2, 3). Comparisons among timepoints were performed to evaluate the transition of the cecal microbiota whithin each experimental treatment. The different treatments differently changed the microbiota across ages, as shown by the diversity indices.

When considering the unchallenged control group (Table 2), it was observed that richness and evenness were higher, mainly on d 28 of age when compared to earlier ages. However, considering the challenged control group, only few differences were observed in terms of richness and evenness across ages (Table 2). On the other hand, when the birds were challenged and supplemented with either SB or SBEO (Table 3) many differences were observed, showing that the supplementation of these feed additives had a strong effect on the cecal microbiota across ages.

The community structure and abundance of the cecal microbiota were significantly different across the experimental groups and ages. Both community membership (unweighted) and structure (weighted) contributed to differences in cecal microbiota across ages. PERMANOVA analysis was conducted to describe the strength and significance that treatment groups and age had in determining the variation among distance matrixes (weighted UniFrac). These differences, that are shown using Principal Coordinated Analysis (PCoA) plots, were evident across ages within each experimental treatment (Figure 1). ANOSIM was used to compare the similarity among the cecal microbiota. Differences were observed across ages within different treatment groups (Table 4). The largest difference in the microbiota was observed on d 12 vs. 28 when birds were challenged and supplemented with SB (R = 0.97 and P = 0.001; “R” is the index of ANOSIM that indicates the similarity of comparison between group pairs. “R” ranges from −1 to 1: the pairs are more similar when the R index is closer to 0 and the pairs are different from each other when the R index is close to 1).

The cecal microbiota in the four ages evaluated was dominated by bacteria belonging to the Phylum Firmicutes, followed by Bacteroidetes and Proteobacteria (Table 5). The effect of age on the relative abundance of the main phylum was accessed for the four experimental groups. First, it was observed that the frequency of Firmicutes significantly decreased (87.1 vs. 61.9%; P = 0.0002), and Bacteroidetes increased (9.9 vs. 31.2%; P = 0.034) in the cecal microbiota of unchallenged control birds on d 28 compared to d 12. The same effect was observed within the challenged control group; however, a higher frequency of Proteobacteria (0.1% on d 12 vs. 1.7% on d 18; P = 0.002) was observed after the birds were challenged, which was not observed in the unchallenged control treatment. On the other hand, when the birds were challenged and supplemented with SB or SBEO, the differences in the frequency of Firmicutes and Bacteroidetes were observed at an early stage (d 21). The increase in the frequency of Proteobacteria due to the challenge when the birds were supplemented with SBEO was not as evident as in the other challenged treatments.

The most abundant genera observed in the cecal microbiota were Ruminococcus, Lactobacillus, and Bacteroides, followed by unclassified Ruminococcaceae and Clostridiales (Figure 2). Again, the effect across ages was evaluated within each experimental treatment. In the unchallenged control birds, it was observed that Ruminococcus decreased (34.1% on d 12 vs. 14.3% on d 18; P = 0.006), whereas Bacteroides (6% on d 12 vs. 23% on d 28; P = 0.02) and Clostridiales (3.4% on d 12 vs. 9.1% on d 28; P = 0.006) increased as the birds aged. In the challenged control group, however, besides similar effects on Ruminococcus (P = 0.001) and Bacteroides (P = 0.001), it was also observed that the frequency of Coprococcus (5.6% o d 12 vs. 0.9% on d 28; P = 0.01) and Blautia (4.7% on d 12 vs. 2% on 28; P = 0.02) decreased as birds aged. This was an effect caused by the challenge, as it was not observed in the unchallenged control group. Clostridiales did not increase as an effect of age in these challenged control birds (Figure 2).

In the challenged SB supplemented birds, the effect of age was also evident on the frequency of Ruminococcus (P = 0.002), Lactobacillus (P = 0.01), Bacteroides (P = 0.001), and Coprococcus (P = 0.02). The frequency of Lachnospiraceae also tended to be affected by age (P = 0.07) in this treatment. The frequency of Lactobacillus was lower at d 18 (17.1%), but not at d 21 (29.5%), compared to d 12; Lachnospiraceae tended to have higher abundance on d 18 (5.1%), compared to the other ages, and Ruminococcus and Coprococcus followed the same pattern as observed in the challenged control group. Supplementation of SB to challenged birds did not reduce the abundance of Blautia, as observed in the challenged control group (Figure 2). Lastly, looking at the effects of age in the challenged and SBEO supplemented birds, similar results were observed as the challenged SB birds. Ruminococcus decreased (39.6% on d 12 vs. 11.5% on d 28; P = 0.001), Bacteroides increased (1.2% on d 12 vs. 25% on d 28; P = 0.001), and Blautia decreased (6.5% on d 12 vs. 2.3% on d 28; P = 0.02); additionally, Lactobacillus tended to be higher at d 21 (29.2%; P = 0.06), and Ruminococcaceae had a higher abundance at d 18 compared to d 21 (11% vs. 4.2%; P = 0.02; Figure 2).

To predict microbial community functions from the microbiota data, phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt) analysis was carried out using the KEGG and Clusters of Orthologous Groups of proteins (COGs) databases. The focus of this analysis was to evaluate the changes in the predicted functions comparing d 12 (pre-challenge conditions) with d 21 (after coccidia and C. perfringens challenge) within each experimental treatment. A Venn diagram was constructed to visualize the shared and unique predicted functions among the treatments which were divided between upregulated (Figure 3A) and downregulated predictive functions (Figure 3B) on d 21 compared to d 12. Among the upregulated predicted functions, it was observed that 11, 11, and 9 were unique to the challenged, challenged supplemented with SB, and challenged supplemented with SBEO, respectively. Only few functions were downregualed on d 21 compared to d 12, wherein only 1, 4, and 2 were unique to the challenged, challenged supplemented with SB, and challenged supplemented with SBEO, respectively.

When evaluating the effect of the challenge, the predicted functions common to the unchallenged control treatment were removed. When evaluating the effect of the supplementation of SB and SBEO, the predicted functions shared with the unchallenged and challenged control were also filtered out, thereby the remaining functions were unique to SB or SBEO supplementation. In the unchallenged control treatment (Figure 4A) few changes were observed in the predictive functions of the cecal microbiota as the birds aged. A total of 14 pathways were changed (P ≤ 0.01) comparing d 21 vs. d 12. A larger number of functional categories of the cecal microbiota in the challenged control treatment were changed on d 21 vs. d 12 (Figure 4B). These changes were due mainly to the challenge since the functions shared with the unchallenged control were filtered out. A total of 30 features were unique to the treatment supplemented with SB wherein most of them were related to metabolism, and one related to penicillin and cephalosporin biosynthesis (Figure 5A). Additionally, 23 functional categories were unique to the SBEO supplemented treatment which included, besides penicillin and cephalosporin biosynthesis, biosynthesis of vancomycin group of antibiotics, and streptomycin biosynthesis that were enriched on d 21 vs. d 12 (Figure 5B).

JM, and MP are employed by Norel, Animal Nutrition who manufactures the products used in the present research. CB, MR, BV, CH, and TA have no conflict of interest to declare.