Nutraceuticals Induced Changes in the Broiler Gastrointestinal Tract Microbiota

ABSTRACT Effects of nutraceuticals on the intestinal microbiota are receiving increased attention; however, there are few studies investigating their effects on broiler meat production. The aim of this study was to implement feeding strategies and carry out a comprehensive trial examining the interplay between natural biologically active compounds such as carotenoids, anthocyanins, fermentable oligosaccharides, and synbiotics and the gastrointestinal tract microbiota. Our feeding program was applied to an intensive production system with a flock of 1,080 Ross 308 broilers. Aging induced significant changes through the feeding experiment. Nutraceuticals were shown to modulate broiler intestinal diversity and differentially enriched Lactobacillus, Enterococcus, Campylobacter, and Streptococcus in the core microbiome during the different stages of broiler rearing. Additionally, they did not remarkably affect animal growth performance; nevertheless, a positive correlation was found between body weight and Corynebacteriales and Pseudomonadales. Furthermore, a diet high in carotenoid, fermentable oligosaccharide, and anthocyanin contents affected the number of beneficial genera such as Faecalibacterium, Lactobacillus, Blautia, and Ruminococcus. With this comprehensive trial, we revealed that nutraceuticals induced modulations in broiler gastrointestinal tract microbiota. We believe that plant-derived immunostimulants, recycled from plant food waste products, can supplement antibiotic-free broiler meat production. IMPORTANCE In this trial, nutraceuticals were manufactured from waste products of food industry processing of Hungarian red sweet pepper and sour cherry and incorporated into the diet of poultry to investigate their effects on broilers’ growth and the broiler gastrointestinal tract microbiota. To avoid the generation of food waste products, we believe that this approach can be developed into a sustainable, green approach that can be implemented in commercial antibiotic-free poultry to provide safe and high-quality meat.

D uring the past 2 decades, the poultry industry has become one of the most efficient protein production systems, and it forms the basis of global protein production (1). Intensive breed selection was invented to develop chickens that convert feed into muscle mass more efficiently (2). Modern chicken breeds such as Ross 308 require less forage to achieve their desired increase (approximately 70 to 80Â) in weight (35 g to ;3 kg) throughout the production period (35 to 42 days) (3). This extreme growth rate can be associated with a range of pathological conditions (3)(4)(5), including hypertension, heart failure, insulin resistance, and increased susceptibility to infections (6)(7)(8).
The gastrointestinal tract (GIT) microbiota plays an important role in the overall health and function of the host (9)(10)(11). The GIT microbiota is the focus of major research efforts in meat production animals (12) since it has a positive impact on the immune system (12)(13)(14), GIT physiology (14,15), nutrition (11,16), and detoxification of certain compounds and productivity (16,17). It also has an important role in the poultry industry, requiring animals capable of growing rapidly (18,19).
There is growing evidence that alterations in poultry GIT microbiota composition have a pivotal role in the development of metabolic disorders (15,20,21). The diversity of the microbiota is one of the key determinants in resistance to invading pathogens (22). Higher microbial community diversity is related to a healthier host status, whereas a significant loss in complexity is associated with various diseases and susceptibility to pathogen colonization (16,(23)(24)(25). Shifts of the GIT microbiota toward beneficial bacteria could improve the health conditions of the host.
Through the past 80 years, antibiotics have been widely used to support the immunocompetence of birds against infectious diseases (26,27). For animals that grow to a great degree, application of a subtherapeutic dose of antibiotics was generally shown to improve health and productivity (28). The routine and irresponsible use of such additives is associated with undesired consequences, such as depletion of the beneficial intestinal microbiota and emergence of antibiotic-resistant microbial pathogens (29,30). The lateral exchange of genetic material across bacteria contributes to the spread of antimicrobial resistance and broadly disseminates harmful, antibiotic-resistant bacteria across the globe. This dramatic impact has been a serious threat to both human and veterinary medicine (31). Antibiotic resistance was identified by the World Health Organization (WHO) as one of the most significant global threats to public health, and their use as growth promoters was banned by the European Union (32,33).
Health-promoting probiotic bacteria can ferment prebiotics that are undigestible and nonabsorbable for the host and convert them to lactic acid and short-chain fatty acids (SCFAs) (33)(34)(35)(36)(37)(38). SCFA-producing bacteria may directly enhance the absorption of some nutrients and hence have a direct influence on metabolic functions. (39)(40)(41). It was already proven that the deterioration of community diversity and the associated alterations in SCFAs can be restored by alternative treatment strategies in both humans and animals (42), some of which may alleviate disease symptoms (36). These probiotic-based dietary supplements are increasingly considered to be effective in replacing antibiotics (43,44). Furthermore, it is also suggested that a probioticenriched diet influences the intestinal absorption of broilers, thus improving production performance (45). Additionally, numerous studies emphasize the importance of prebiotic fibers, which can enhance the effects of live beneficial microorganisms (e.g., lactic acid bacteria; Lactobacillus and Bifidobacterium) (46).
More recently, nutraceuticals have become the focus of farm animal production. These nutraceuticals are rich in plant-derived immune stimulants such as phytochemicals, vitamins, and minerals (59). Several pre-, pro-, and synbiotic-based functional medicines have already been explored thoroughly and have demonstrated the ability to rebalance dysbiotic intestinal flora and preserve animal health (60). In this trial, we focused on natural, bioactive compounds (carotenoids, anthocyanins, functional oligosaccharides, and synbiotics) obtained from reprocessed plant-based food industrial waste materials and investigated their modulatory effect on the broiler gastrointestinal tract.
By enriching the diet of a flock of 1,080 Hungarian broilers with nutraceuticals, we investigated their effect on microbiota community diversity and alterations in the baseline symbiotic microbiota. We also managed to unravel compositional shifts in the GIT microbiota and investigated how these might relate to the growth performance of Ross 308 broilers.

RESULTS
General description of sequencing results. The 16S rRNA gene-based (V3-V4 region) amplicon sequencing was carried out on the Illumina MiSeq platform, generating a total of ;11 million reads by processing 96 broiler fecal samples with a mean count of 86,470 6 24,361 reads per sample. Quality filtering with the DADA2 software resulted in an average denoised read count of 42,763 6 13,425 per sample, and after a merging process, the read count dropped to an average of 41,085 6 12,991 reads per sample. At the end, the average number of nonchimeric reads was 27,778 6 7,622 per sample.
Effects of nutraceuticals on body weight. The effects of dietary supplements on broiler growth body weight (BW) were monitored throughout the feeding trial ( Fig. 1). At the beginning, the average BW values for birds were 38.4 6 1.6 g, while by the end of this experiment broiler chicken reached 2,693 6 64.8 g on average (see Table S1 in the supplemental material ). However, by the end of the broiler productive life span, a moderate but not significant decrease in body weight was registered due to anthocyanin-based dietary supplementation in comparison to controls (ANTH BW, 2,590 6 264 g, versus BD and BGLU BW, 2,742 6 222 g).
Significant associations were found between broiler body weight and the GIT microbiota. We managed to unravel alterations induced by age (prestarter, starter, grower, finisher) and treatment (BD, BGLU, CAR, fOS, SYN, ANTH) for 11 orders in the intestinal microbiota of Ross 308 broilers, finding remarkable correlations with body weight (Fig. 2). Alterations in the strengths and directions of correlations were obtained. In this study, out of the 11 orders, there were 6 (Bacillales, Clostridiales, Corynebacteriales, Enterobacteriales, Micrococcales, Rhizobiales) where moderate positive (age-and/or diet-specific) associations were detected with BW (r value . 0.4). We estimated that during the first two phases of the feeding experiment strong and/or moderate positive correlations were found between BW and the orders Corynebacteriales, Bacillales, Clostridiales, and Micrococcales (Fig. 2a). Interestingly, in the case of the order Rhizobiales, adverse, age-dependent correlations were found between starter and grower phases. In the case of the finisher phase, only weak or very weak correlations were found.
When examining the effect of diet alone on the correlation values of these orders, we found strong positive (r value . 0.6) associations between the orders Bacillales, Corynebacteriales, Enterobacteriales, and Micrococcales and BW in ANTH-treated birds (Fig. 2b). Moderate negative (r value , 20.4) correlations were found between BW and Enterobacteriales in BGLU-treated birds. In the case of fOS-treated samples, moderate positive correlations were found with Micrococcales. Interestingly, Pseudomonadales exhibited moderate positive correlations with BW under CAR treatment and moderate negative correlations in fOS-treated samples. For the order Bacillales, moderate negative associations were shown in the CAR group.
Age and treatment induced alterations in alpha and beta diversities. Both alpha and beta diversity indices were determined to track remarkable conversions in community diversities of control (BD, BGLU) and treatment (CAR, fOS, SYN, ANTH) groups (Fig. 3). Faith's phylogenetic (Fig. 3a), Chao-1, Shannon, and Simpson (data not shown) diversity indices were applied to evaluate the species abundance, richness, and evenness of the broiler GIT microbiota. Faith's phylogenetic diversity (PD) indicated a significant increase in chicken GIT community diversity by the end of the productive life span (finisher phase), in the cases of fermentable oligosaccharide-(fOS Faith's PD: 20.3 6 4.6), synbiotic-(SYN Faith's PD: 22.5 6 0.8), and anthocyanin-treated (ANTH Faith's PD: 21.8 6 2.9) birds in comparison to those receiving basal diet (BD Faith's PD: 11.2 6 4.0) (Fig. 3a). During the grower and finisher feeding phases, fOS, SYN, and ANTH treatment caused notable increases in Faith's PD indices. Shannon and Simpson diversity indices did not show significant changes throughout the experiment due to nutraceuticals. In general, certain differences in pattern dynamics were observed in alpha diversity indices (Fig. 3b). Faith's PD, Chao-1, Shannon, and Simpson indices improved steadily with animal growth, while a deterioration was observed in these parameters during the finisher phase of the experiment. Broadly, during the grower phase, the highest community diversity was associated with CAR-treated birds, while by the end of the finisher period the community diversity proved to be the lowest in the case of animals receiving basal diet. Four beta diversity heatmaps were generated by measuring Bray-Curtis, Jaccard, and weighted and unweighted UniFrac distances (Fig. 3c) between the different experimental groups in relation to age and diet. Distance-based dissimilarity matrices showed that flock development exerted a substantial influence on overall community variations; thus, a gradual increase in community diversity was accompanied by increased heterogeneity of the GIT microbiota.
The baseline GIT microbiota reflects a dynamic equilibrium in livestock. Estimations of the healthy core microbiota were made for all experimental groups at the phylum, order, and genus taxonomic ranks, by considering taxa (order: 4; genus: 8) represented in at least 50% of the samples (Fig. 4). Characteristically, fermentable oligosaccharides, synbiotics, and anthocyanins exerted greatest community shifts in the core microbiota of starter chickens.
The most pronounced community taxonomy shifts occurred due to age. Beta diversity plots were made to investigate the age (Fig. 5a)-and diet (Fig. 5b)-induced alterations in community taxonomy. When measuring the age dependency of community taxonomy data with unweighted UniFrac metrics, principal-coordinate analysis (PCoA) resulted in two clusters (cluster 1 and cluster 2) representing different spatial ordinations between prestarter birds and older (starter, grower, and finisher) broilers FIG 4 Variations in the healthy core 50% GIT microbiota of broilers over time. Area plots visualizing the core orders and genera according to age and diet. BD negative control (basal diet with no dietary supplement) and the following dietary treatments were provided as mash feed: BGLU positive control (BD including 0.5% b-glucan), CAR (BD including 0.5% carotenoids), fOS (BD including 0.5% fermentable oligosaccharides), SYN (BD including 0.5% synbiotics), and ANTH (BD including 0.5% anthocyanins).
( Fig. 5a). Furthermore, starter, grower, and finisher birds continued to cluster less separately (Fig. 5a). When marking the samples according to diet, no distinct patterns became apparent between the treatment groups (Fig. 5b). On the basis of the PCoA plots, we concluded that age exerted more pronounced community shifts than diet. Not surprisingly, the prestarter microbiota showed less variation among samples. Additionally, the prestarter microbiota clustered distinctly in comparison to the microbiota at later time points of the experiment.
Immunostimulant-driven alterations in family taxonomy. By considering the 31 most abundant families (relative % frequencies . 0.1), we managed to explore remarkable alterations in taxonomic data during the four phases of the feeding period when comparing BGLU-and nutraceutical-treated birds (CAR, fOS, SYN, and ANTH) to nontreated controls (BD). A composite heatmap was created to show distortions in the relative abundance data normalized to that of BD animals (Fig. 6). During the prestarter phase, we observed remarkable increases in Bifidobacteriaceae due to synbiotics and anthocyanins, while fOS supplementation resulted in higher levels of Peptostreptococcaceae. Nutraceuticals increased Clostridiaceae and Lachnospiraceae. Additionally, greater abundances were observed in Erysipelotrichaceae and Ruminococcaceae in anthocyanin-challenged animals. Immunostimulants decreased the levels of Enterobacteriaceae, Leuconostocaceae, and Staphylococcaceae in comparison to their levels in the negative control (BD). In fOS-treated starter birds, remarkable increases were shown in Bacteroidaceae, Barnesiellaceae, Brevibacteriaceae, and Clostridiaceae accompanied by decreases in Bifidobacteriaceae and Burkholderiaceae. During the grower phase, carotenoids increased Barnesiellaceae and Bifidobacteriaceae and decreased Aerococcaceae, Clostridiaceae, Enterococcaceae, Moraxellaceae, and Peptostreptococcaceae. In grower animals, solid increases in Campylobacteraceae, Planococcaceae, and Pseudomonadaceae and decreases in Bacteroidaceae, Helicobacteraceae, and Marinifilaceae were registered due to anthocyanins. In the finisher phase, impressive decreases were encountered in Brevibacteriaceae in all of the treatment groups. Enrichments in Helicobacteraceae occurred through fOS, SYN, and ANTH treatments. Additionally, an increase was detected in Akkermansiaceae due to BGLU, SYN, and ANTH.
Alterations in the occurrence of SCFA-producing bacteria. Among a range of metabolites produced by the beneficious gastrointestinal tract microbiota, short-chain fatty acids (SCFAs) have received increased attention because of their important role in disease prevention and recovery (61). In this trial, appreciable alterations were found in the proportions of some genera associated with SCFA production (Fig. 7).
Diet-induced compositional differences can affect microorganisms involved in carbohydrate metabolism. Both Bacteroides and Firmicutes are associated with SCFA synthesis (1). The end products of dietary fiber fermentation have been shown to exert multiple beneficial effects on mammalian energy metabolism by enhancing the absorption of some nutrients (39)(40)(41). According to previous publications, elevated Firmicutes levels can be associated with increased nutrient absorption, whereas Bacteroidetes enrichment usually correlates with enhanced hydrolysis of glycogen, starch, and polysaccharides promoting feed utilization and digestion of the host (1,70,71). The Firmicutes-to-Bacteroidetes (F/B) ratio is important for the optimal nutritional requirements of the host (56). Under our experimental settings, a total of 7 phyla were identified. Among these, Firmicutes (R89.5% 6 7.8%), Proteobacteria (R7.3% 6 7.0%), and Bacteroidetes (R1.3% 6 2.7%) were the most predominant, followed by Actinobacteria, Proteobacteria, Tenericutes, and Verrucomicrobia. F/B ratio was biased more by age than diet (Fig. 9a). Differences in the Firmicutes-to-Bacteroides ratios may reflect alterations in (poly)saccharide utilization of flocks. Characteristically, log 2 F/B ratios represent a remarkable decrease in the course of the broiler production (prestarter phase, FIG 7 Shifts in the relative abundances of short-chain fatty acid-producing genera: Faecalibacterium, Lactobacillus, Subdoligranulum, Butyricicoccus, Streptococcus, Bacteroides, Blautia, and Ruminococcus. Age-related distributions of dedicated short-chain fatty acid-producing genera through four phases of broiler rearing: prestarter phase, starter phase, grower phase, and finisher phase. Violin plots show the influence of diet on the distribution of the short-chain fatty acid-producing genera. Asterisks indicate statistical significance: *, P # 0.05; **, P # 0.01. BD negative control (basal diet with no dietary supplement) and the following dietary treatments were provided as mash feed: BGLU positive control (BD including b-glucan), CAR (BD including carotenoids), fOS (BD including fermentable oligosaccharides), SYN (BD including synbiotics), and ANTH (BD including anthocyanins).
We also considered genera involved in carbohydrate metabolism that may include potential avian-pathogenic organisms (such as Enterococcus [72], Clostridium [24,25], and Helicobacter [73]). The probiotic genera Bacillus and Eubacterium showed the highest occurrence for the treatment with ANTHs (Fig. 9b). Regarding its age-related distribution, the genus Bacillus was least abundant during the prestarter phase and reached its highest abundances during the starter phase (prestarter, 0.008% 6 0.02%; starter, 0.05% 6 0.14%) of the experiment, while the genus Eubacterium (prestarter, 1.3% 6 3.3%, versus others, 0.5% 6 0.7%) was the most abundant genus during the prestarter phase of the experiment. The genus Corynebacterium, which can include strains causing serious outbreaks of avian infections, was not detected during the prestarter phase but peaked at the starter phase (starter, 1.8% 6 0.6%, versus grower, 0.2% 6 0.5%; finisher, 0.5% 6 0.4%; P , 0.05). Alistipes, whose members are important in the fermentation of dietary fiber, was scarce in abundance during this experiment and detected during only the grower phase (grower: 0.1% 6 0.3%) and in birds receiving basal diet (BD: 0.09% 6 0.2%) and carotenoid (CAR: 0.1% 6 0.2%) supplementation. Our data indicated that in comparison to the basal diet, nutraceuticals had decreased relative abundance of Helicobacter (nutraceuticals, 0.2% 6 0.3%, versus BD, 0.3% 6 0.8%); anthocyanins increased the abundance of Campylobacter (ANTH, 0.4% 6 1.6%, versus other, 0.05% 6 0.1%), Bacillus (ANTH, 0.1 6 0.1%, versus other, 0.01% 6 0.05%), and Eubacterium (ANTH, 1.7% 6 0.4%, versus other, 0.5% 6 0.6%); carotenoids increased Eggerthella (CAR, 0.02% 6 0.07%, versus other, 0.004% 6 0.01%); and the genus Clostridium was not detected in CARand ANTH-treated birds. We noticed a significant increase in Campylobacter and Helicobacter during the grower (P , 0.001) and finisher (P , 0.001) phases of the experiment. Clostridium was mainly detected during the prestarter phase. In the FIG 8 Differentially abundant taxonomic heat trees revealed the effects of nutraceuticals on taxa involved in lipid metabolism. The Metacoder differential heat tree illustrates the variation in microbiome phylotypes between experimental groups. The annotated tree on the left functions as a map for the unlabeled trees. Colored taxa represent the extents of log 2 differences in taxon abundances: green represents higher abundance in BD or BGLU, while brown means higher abundance in nutraceutical-treated groups. Nodes in the heat tree correspond to phylotypes, as indicated by node labels, while edges link phylotypes in accordance with the taxonomic hierarchy. Node size corresponds to the number of operational taxonomic units (OTUs) observed within a given phylotype. BD negative control (basal diet with no dietary supplement) and the following dietary treatments were provided as mash feed: BGLU positive control (BD including 0.5% b-glucan), CAR (BD including 0.5% carotenoids), fOS (BD including 0.5% fermentable oligosaccharides), SYN (BD including 0.5% synbiotics), and ANTH (BD including 0.5% anthocyanins).
Attention was also paid to the estimated relative proportions of relevant species involved in lipid and carbohydrate metabolism, such as those involved in avian infections (Fig. 9c). Noticeably, b-glucan-treated samples showed the highest species diversity for lactic acid bacteria, covering eight Lactobacillus strains. Levels of the beneficial Lactobacillus aviarius and Lactobacillus salivarius, which is one of the main suppliers of the enzyme bile salt hydrolase (BSH) (74) and is also known to provide protection against colonization by Salmonella and other pathogens, were observed in all experimental groups. L. salivarius showed enrichment in the control animals (BD-BGLU, 15.2% 6 17.8%, versus nutraceutical groups, 7.5% 6 9.2%), whereas L. aviarius showed remarkable increases due to synbiotics (SYN: 14.7% 6 15.6%) and anthocyanins (ANTH: 6.8% 6 9.2%). Lactobacillus alvi, which is frequently obtained from chicken fecal and intestine (75), was also represented uniformly and showed increases in anthocyanintreated samples (ANTH, 0.11% 6 0.2%, versus other groups, 0.01% 6 0.04%).
Microbial interconnections induced by nutraceuticals. To identify nutraceuticalinduced interconnections within the broiler intestinal microbiota, we estimated the extent to which relevant families tended to change together. Relative proportions of taxa were correlated in terms of Spearman's method (Fig. 10). We identified divergent abundance patterns by using data for the 15 most abundant families in nutraceuticalinduced treatment groups throughout the four phases of the experiment (Fig. 10a). In general, similar correlation patterns were revealed between CAR-SYN-and fOS-ANTHtreated samples. We focused on two areas. (i) First, we attempted to find correlations between families throughout the four feeding phases of the experiment. We found 13 statistically significant positive (prestarter: 5; grower: 2; finisher: 6) and 15 negative (prestarter: 4; starter: 2; grower: 1; finisher: 8) associations throughout the experiment (Fig. 10b). (ii) Second, we identified very strong correlations between families in that were exclusive to specific diets (Fig. 10c)

DISCUSSION
An extraction technology was developed (34) that is able to recycle plant-based food industrial waste to extract its bioactive compounds (anthocyanins from sour cherry and carotenoids and fermentable oligosaccharides from red sweet pepper) and conserve their beneficial, health-promoting effects. Based on this invention, our prior aim was to develop forage enriched in nutraceuticals and to investigate the effect of these natural feed additives on the broiler GIT microbiota.
The feeding program of this trial was applied according to the norms widely used in Ross 308 chicken production (34). Based on our findings, bioactive compoundenriched diets have been shown to strengthen the positive correlations between body weight and the beneficial orders Bacillales, Rhizobiales, and Corynebacteriales, which are associated with increased nutrient absorption through the improvement of the intestinal epithelium integrity (79,80). We found that, under our experimental conditions, a nutraceutical-enriched diet did not significantly improve body weight, supporting the estimations of other data (81,82). Additionally, our data did not support that probiotics enhance animal growth, which might be explained by a number of different environmental and genetic factors (36). Nutraceuticals did not significantly increase the relative proportions of Lactobacillaceae and Bifidobacteriaceae, which were previously reported to amend the utilization of prebiotic oligosaccharides in chicken (24,(83)(84)(85)(86). Furthermore, we theorize that the noticeable decrease in intestinal Clostridium and Bacteroides of anthocyanin-treated birds may be associated with alterations in bile biotransformation through which the microbiota impacts host fat digestion and utilization. Notably, we did not observe any decrease in the body gain rate of anthocyanintreated birds (ANTH finisher phase, 2,590 6 280 g, versus BD, 2,758 6 264 g).
A combined age-related view of the healthy, baseline GIT microbiota was also achieved at the phylum, order, and genus taxonomic ranks of baseline bacteria at different stages of Ross 308 broiler production. This showed that the broiler GIT microbiota was dominated by two core phyla: Firmicutes (93.0% 6 6.9%) and Proteobacteria (6.9% 6 0.9%).
We also investigated the effects of different dietary supplements on GIT community complexity through the production of Ross 308 Gallus gallus forma domestica. Based on our results, remarkable increases were detected in Faith's index due to fOS, SYN, and ANTH diet in relation to those of both controls (BD, BGLU). According to our estimations, the fOS-supplemented diet increased Faith's index, which was consistent with the results reported by Shang et al. (35). Furthermore, in accordance with a previous study (87), we found that carotenoids did not exert significant effects on community complexity. Probiotics are increasingly applied to animals in poultry industries, too (39,88). Additionally, based on our findings, b-glucan supplementation did not exert a remarkable influence on community diversity. Similar to previous reports, our data indicated that the composition of the broiler GIT microbiota diversifies remarkably as the GIT microbial population becomes more complex in aging broilers (39,89). An increase in community alpha diversity makes symbiotic communities more discordant, which was also supported by Bray-Curtis, Jaccard, and weighted and unweighted UniFrac distances. Notably, the present study revealed that appreciable beneficial effects of nutraceuticals manifested mostly by the end of the broiler productive life span, as the diversity started to decrease. This may suggest that dietary supplementation has a lesser impact on a more diverse symbiotic microbiota. Higher microbial diversity is commonly related to a healthier host status, whereas a lack of sufficient diversity in microbial community structures has been associated with intestinal diseases (10,(90)(91)(92)(93)(94). Furthermore, imbalance of the gut microbiome composition and significant losses in GIT diversity often lead to the elimination of beneficial bacteria and accompanying increases in pathogenic bacteria (95).
Additionally, we managed to investigate how nutraceuticals can shift the abundances of potential zoonotic strains. The final 2 weeks of the broiler production period is associated with elevated mortality and production losses due to localized or systemic bacterial infections. In addition to the genetic background, the performance and meat production of domestic animals (e.g., broilers) are influenced by water and feed quality, energy and nutrient content of the diet, and their relative proportions, as well as various environmental factors (ambient temperature, humidity, air speed, ventilation technique, herd density in the barn, and, moreover, environmental stress) (96,97).
Identifying symbiotic and dysbiotic taxa is not a straightforward task, and there are no obvious "good or bad guys" in complex microbial communities. However, it is essential to consider the problem of livestock contamination for both sanitation and economic reasons (85). In our experimental system with 1,080 animals, the mortality rate proved to be very low (0.56%); nonetheless, no significant differences in lethality patterns were observed between our experimental settings.
In this study, the Firmicutes-to-Bacteroides ratio was lowest in anthocyanin-fed animals, which was accompanied by a decrease in body weight in comparison to that of the controls. The potential pathogen genus Bacteroides encodes a high number of proteins involved in polysaccharide and monosaccharide metabolism, decreases colonic pH, and improves the function of epithelial cells (98). The increase in Bacteroides frequencies in the starter flock due to synbiotics supposedly modulated their polysaccharide metabolism since members of this genus are generally associated with the degradation of starch and glucan (76). However, these suggestions were not supported by our data. Acetate and propionate are mainly produced by Bacteroidetes, while Firmicutes are the main butyrate supplier (37,99,100). The highest ratios for Bacteroides gallinaceum were detected in samples receiving carotenoids and anthocyanins, while Bacteroides dorei was traceable only in CAR-, fOS-, and SYN-fed birds. Notably, in prestarter and finisher broilers, anthocyanins increased the levels of the beneficial bacteria Lachnospiraceae and Ruminococcaceae, which are usually associated with improvements in feed conversion (51). Furthermore, during the finisher phase, anthocyanins increased the levels of Akkermansiaceae, Bacteroidaceae, and Barnesiellaceae, which are in turn linked to more efficient intestinal absorption of compounds, as described previously (101). This might be suggestive of improvements in growth parameters; however, these were also not strengthened by our data.
The beneficial effects of nutraceuticals manifested in the increasing proportions of the butyrate producers Lachnospiraceae and Ruminococcaceae in finisher chickens. For colonocytes, butyrate is an important energy source that is largely metabolized in the epithelial mucosa (102). Mucin-degrading Akkermansia species are usually associated with intestinal health, due to their competitive exclusion of other, less beneficial bacteria that adhere less effectively to the mucosal surface (103,104). Additionally, Akkermansia was previously shown to decrease visceral fat deposits; thus, their abundance might be associated with decreases in body weight gain (103)(104)(105). However, in this study, no significant associations were found between Akkermansia and broiler weight. Anthocyanins enhanced the frequencies of the important butyrate producer genus Eubacterium (40,106,107), while fermentable oligosaccharides and synbiotics increased the relative abundance of the genus Clostridium during the prestarter feeding period, which might be associated with beneficial effects on animal GIT health (108).
In addition to involvement in carbohydrate metabolism, some members of the genera Helicobacter, Clostridium, and Enterococcus are important pathogens (86) that colonize the gastrointestinal tract of chickens, causing gastroenteritis (73), necrotic enteritis (24,25), and enterococcal spondylitis (72). Notorious members of the genus Clostridium also have beneficial physiological effects on various biological responses by synthesizing essential vitamins and micronutrients (thiamine, riboflavin, nicotinamide, pantothenic acid, biotin, etc.), neurotransmitters (biogenic amines), and secondary bile acids for the host (102,109,110). Furthermore, certain members are also known polyphenol producers, exhibiting antioxidant activity and decreasing inflammation (111). Lipoglycans of Clostridium and Enterococcus spp. are known to trigger inflammatory responses and insulin resistance (112). In the case of Clostridium, the highest ratios were noted in prestarter birds, treated with b-glucan, fermentable oligosaccharides, and synbiotics, whereas the highest abundances of Enterococcus were registered in prestarter and starter birds where nutraceuticals, especially fermentable oligosaccharides and anthocyanins, boosted their frequencies in comparison to controls. Previous studies reported decreased Campylobacter and Clostridium colonization measured in broilers fed fructans (113). According to our data, the proportion of the family Campylobacteraceae was significantly decreased in finisher animals receiving immunostimulants relative to those receiving the basal diet. In carotenoids-fed birds, Eggerthella increased remarkably, whereas immunostimulants (BGLU and nutraceuticals) were able to decrease the abundances of the genus Helicobacter. Interestingly, in chickens fed anthocyanins, a noticeable increase was registered for the bacterial diarrheal gastroenteritis-causing C. jejuni without affecting chicken welfare. Of note, C. jejuni can also be involved in the maintenance of intestinal epithelial integrity and the modulation of anti-inflammatory and antitumor effects (35,57,114). Although the specific mechanisms have not been fully elucidated, phytonutrients rich in antioxidants can reduce pathogenic stress (115). The Gram-negative, rod-shaped, opportunistic pathogen Alcaligenes faecalis, which can trigger infections by colonizing the respiratory tract (116), was not traceable in broilers receiving either b-glucan or nutraceuticals.
The most widely used probiotics are members of the relevant acetate-producing genus Lactobacillus (22,117), which has also been reported to positively affect the gut health of poultry by reducing inflammation and controlling enteric bacterial infections through regulating mucin composition (16,17,75,99). In this trial, carotenoids were shown to positively modulate the abundances of the genus Lactobacillus in grower and finisher animals, which might also affect certain enzymatic activities of the oligosaccharide transport system of lactobacilli (118). These data are consistent with the results of other studies reporting Lactobacillus as a major beneficial bacterium showing increases in broilers fed fructans (37,38). In control samples, elevated levels were measured for Lactobacillus salivarius in relation to that in treatment groups, which can be associated with enhanced induction of anti-inflammatory responses of chicken (99). Furthermore, the age-related oscillating patterns of the genus Lactobacillus might also be congruent with deconjugated bile acid concentrations in broiler chickens (51,119). Both human and animal studies found an association between the accumulation of lactic acids and disease states, such as colitis and gut resection (120,121). In our study, taxonomic heat trees indicated that anthocyanins remarkably decreased the relative abundance of the family Lactobacillaceae.
The most pronounced negative correlations between butyrate-producing genera such as Butyricicoccus and Ruminococcus and lactic acid-producing Lactobacillus have been revealed in anthocyanin-treated animals. According to our assumptions, this might be associated with improvements in epithelial intestinal barrier functions that are caused by decreasing lactic acid buildup and increasing osmotic load (122). Interestingly, a strong negative correlation was revealed between the lactate-and acetate-producing Bifidobacteriaceae and lactic acid-producing Staphylococcaceae (r value: 20.97) in animals fed anthocyanins. In finisher animals, very strong negative correlations were detected in birds fed nutraceuticals between Lactobacillaceae and Bacteroidaceae, whose members are known to improve metabolic efficiency and reduce colonization by undesirable microbes (36, 117, 120).
Conclusions. We report the following main results based on our data. (i) Time exerted a great influence on the chicken microbial community structure. There was a tendential increase in broiler GIT community diversity as chickens aged. Subsequent deviation from diversity can be alleviated by treating birds with fermentable oligosaccharides, synbiotics, and anthocyanins. (ii) Great emphasis was also placed on how taxonomy data correlate with enhanced bird body weight. Nutraceuticals resulted in strong positive correlations between body weight gain and the orders Bacillales, Corynebacteriales, Enterobacteriales, Micrococcales, and Pseudomonadales. (iii) The 50% core taxonomy data revealed the relations between the symbiotic broiler Ross 308 microbiota and age and diet. Fermentable oligosaccharides, synbiotics, and anthocyanins were shown to exert the greatest community shifts, especially during the prestarter and starter phases. (iv) In general, Enterobacteriaceae (prestarter, starter), Akkermansiaceae (finisher), Brevibacteriaceae (starter, finisher), Staphylococcaceae (prestarter), Bacteroidaceae (starter, grower), Bifidobacteriaceae (starter, grower), Campylobacteraceae (grower, finisher), Helicobacteraceae (finisher), Planococcaceae (grower, finisher), and Pseudomonadaceae (grower, finisher) were identified as key taxa representing significant shifts (mean log 2 fold change j$2j) in community taxon compositions due to nutraceuticals. (v) There were alterations in relative frequencies of commensal beneficial, short-chain fatty acid-producer bacteria and conditioned pathogens. The Firmicutes-to-Bacteroides ratio (F/B) proved to be the highest in b-glucantreated animals and the lowest in anthocyanin-treated animals. Coincidentally, anthocyanins were shown to increase Faecalibacterium, Blautia, and Ruminococcus in finisher birds remarkably relative to BD. Generally, fermentable oligosaccharides, synbiotics, and anthocyanins exerted a positive impact on Faecalibacterium, and the difference was more pronounced by the end of broiler rearing. Impressive alterations in Lactobacillus were mostly age related. Carotenoids were shown to increase Bifidobacteriaceae and Barnesiellaceae but reduce Enterococcaceae and Clostridiaceae in grower phase. (vi) Spearman's correlations identified mutual interconnections, i.e., very strong age-and diet-related associations of the symbiotic broiler gastrointestinal microbiota. Very strong positive correlations were revealed between body weight and the families Campylobacteraceae-Planococcaceae (CAR), Streptococcaceae-Beijerinckiaceae (CAR), Peptostreptococcaceae-Aerococcaceae (fOS), Burkholderiaceae-Rikenellaceae (fOS), Bacillaceae-Nitrosomonadaceae (SYN), Ruminococcaceae-Bifidobacteriaceae (ANTH), and Clostridiaceae-Desulfovibrionaceae (ANTH) for individual nutraceuticals. This is a unique and comprehensive trial that highlights the health benefits of bioactive compounds of recycled food waste products as potential dietary adjuncts for antibiotic-free broiler meat-production systems. Based on our observations, a nutraceutical-enriched diet did not degrade chicken development and delivered promising results in stimulating GIT health.
Additionally, this study also improves our knowledge about the effects of carotenoids, fermentable oligosaccharides, anthocyanins, and synbiotics on the composition of the broiler gastrointestinal tract microbiota.

MATERIALS AND METHODS
Birds and housing. A total of 1,080, 1-day-old Ross 308 mixed-sex broilers from a commercial hatchery in Hungary were used. The experiment was carried out on the experimental farm of the University of Debrecen. All broilers were housed in the same shed. Chickens were kept in floor pens covered with wood shavings in a thermostatically controlled house at a stocking density of 650 cm 2 /bird and reared under standard management conditions. Sampling procedures were carried out in accordance with the local (University of Debrecen) ethics committee's approved guidelines (DEMAB/12-7/2015).
Experimental design and dietary treatments. One-day-old Ross 308 hybrid chicks were randomly placed into 6 experimental groups (3 replicates/treatment, 60 birds/pen). The experiment was started at day 1 of age and lasted until 42 days. Each group was fed one of the following 6 diets: basal diet (BD), without any added supplements; basal diet including 0.5% b-glucan (BGLU); basal diet including 0.5% carotenoids (CAR); basal diet including 0.5% fermentable oligosaccharides (fOS); basal diet including 0.5% synbiotics (SYN); basal diet including 0.5% anthocyanins (ANTH). BD (negative) and BGLU (positive) were the control groups, and CAR, fOS, SYN, and ANTH were the treatment groups. Broilers were fed a commercial maize-soybean-based basal diet (BD) free of antibiotics according to four feeding periods: prestarter (1 to 9 days), starter (10 to 21 days), grower (22 to 31 days), and finisher (32 to 42 days). All diets were fed in mash form. The compounds and nutritional composition of BD are given in Table 1. The composition of nutrients in each basal diet was planned to satisfy nutritional requirements of broiler chickens according to the National Research Council (NRC) (123). Feed and water were available ad libitum during the entire experiment. Broilers were weighed at 1, 10, 21, 32, and 42 days of age. As growth performance parameters, average body weight (BW) was calculated. Mortality was monitored; it was very low (0.56%), and there was no association between mortality and feed treatments. No veterinary treatment was required for the entire duration of the experiment.
Step elution was performed with the following settings: 0 to 3 min 100% solvent A, 15 to 20 min 20% solvent A, 25 to 45 min 100% solvent B, and 48 to 50 min 100% solvent A. For detection, a diode array detector (DAD) and a 0.6-ml/min flow rate were applied. The sample was injected in a 10-ml volume, and the DAD detection was applied at 460 and 350 nm. The HPLC profile and carotenoid compounds with the greatest areas are provided in the supplemental material (Fig. S1).
Fermentable oligosaccharide (fOS) supplementation was performed as described in the work of Csernus et al. (34) (Fig. S2). Hungarian red sweet pepper was also applied to extract fermentable oligosaccharides (fOS) with high arabinogalactose content. To assess the composition of oligosaccharides, an HP 5890 gas chromatograph (GC) was applied with an SP-2380 capillary column (30 m by 0.25 mm, 0.2 mm). Samples were lyophilized and extracted with trifluoracetic acid-acetic acid-water (5:75:20) as the solvent. Oligosaccharides were turned into alditol-acetate. After the reduction step, sugars were shifted to sugar alcohols (alditols), which removed interfering isomers and anomers. Reduction was performed with NaBH 4 at alkaline pH. Acetylation was also performed with acetic anhydride in pyridine. The a CP, crude protein; AME n , apparent metabolizable energy, n = corrected for zero nitrogen balance.
Anthocyanin (ANTH) supplementation was determined as described by Nemes et al. (125) (Fig. S4). Anthocyanins were extracted from Hungarian sour cherry. Cherries were deseeded and homogenized, and then methanol-water-acetic acid solution in a 25:24:1 ratio was utilized to extract anthocyanins. The sample was mixed with a magnetic stirrer (MSH 300, BioSan, Riga, Latvia) for 1 h. Filtering and centrifugation were performed at 10,000 rpm for 5 min, and then a simple fractionation was carried out in preconditioned tubes (Superclean ENVI-18 SPE tubes). For preconditioning, 5 ml of methanol (MeOH), 5 ml of H 2 O, and 1 ml of fruit sample were used. The elution was conducted with methanol containing 20% H 2 O and vaporized at 40°C. The sample was dried in vacuum to powder. A VWR-Hitachi ChromasterUltraRs ultra-HPLC (UHPLC) instrument (Hitachi, Tokyo, Japan) was used for anthocyanin profile determination with a Phenomenex Kinetex column (2.6 mm, XB-C 18 , 100 Å, 100 Â 4.6 mm) (Phenomenex, Torrance, CA, USA). Two solvents were applied for a step elution, A (MeOH) and B (3% formic acid), with the following parameters: 0 min, 15% solvent A; 0 to 25 min, 30% solvent A; 25 to 30 min, 40% solvent A; and 30 to 40 min, 50% solvent A. UV-visible (UV-VIS) detection was applied at 534 nm, the flow rate was kept at 0.7 ml/min at 25°C, and the injection volume was 10 ml. The UHPLC profile and the main anthocyanin compounds are included in the supplemental material (Fig. S4).
Sample collection. Fecal samples were collected at 7, 19, 31, and 40 days of age (prestarter, starter, grower, and finisher sampling periods, respectively). In every experimental group (BD, BGLU, fOS, CAR, SYN, and ANTH), 4 fecal samples (1 pullet and 1 cockerel, 2 fecal pools) were collected over the whole experimental period. Fecal samples were collected freshly into specific, sterile, DNase-free stool transportation bowls and immediately placed on ice for a maximum of 3 h. Unprocessed samples were kept at 280°C until further use.
Sample preparation and mechanical cell lysis. Bacterial cell suspensions (BS) were prepared from 7 g of each broiler stool sample. Then, 7 ml of sterile PBS buffer (Thermo Fisher Scientific, MD, USA) was added to each of the samples, and they were homogenized for 4 min (by vortexing at 350 rpm) (126). The samples were centrifuged for 5 min at 500 Â g. Supernatants were collected, and the washing step was repeated 2 times. Supernatants were centrifuged for 20 min at 13,000 Â g. Finally, the supernatants were discarded, and the bacterial pellets were dissolved in 3 ml of sterile PBS buffer. One-milliliter aliquots of BS were added to PowerBead tubes (Qiagen, Hilden, Germany) for mechanical cell lysis. Bacterial cell disruption was performed with a MagNA Lyser instrument (Roche Applied Sciences, Penzberg, Germany) set to 5,000 rpm for 30 s.
DNA extraction. Total bacterial genomic DNA was extracted with the conventional isolation method. A total of 800 ml of sample lysate was mixed with 800 ml of phenol-chloroform-isoamyl alcohol (25:24:1) (Thermo Fisher Scientific, MD, USA) and vortexed thoroughly for approximately 20 s. After homogenization, the samples were incubated at room temperature for 3 min and centrifuged for 10 min at 16,000 Â g. After phase separation, the upper aqueous layer was carefully collected into a new sterile DNase-and RNase-free Eppendorf tube. For DNA precipitation, a mixture of 1 ml of glycogen (20 mg), 7.5 M NH 4 OAc (ammonium acetate in 0.5Â volume of the sample), and 100% EtOH (ethanol in 2.5Â the volume of the sample) was added to the supernatant. The samples were incubated at 220°C overnight and then centrifuged for 30 min at 16,000 Â g at 4°C to pellet the DNA. The supernatant was carefully discarded without disturbing the pellet, and 70% EtOH was added to the sample and shaken by hand for 20 s. Then, the samples were centrifuged at 4°C for 5 min at 16,000 Â g, and the supernatant was carefully removed. This washing step was repeated 2 times. The DNA pellet was dried at room temperature and then resuspended in 40 ml of nuclease-free water. DNA concentrations were determined using a Qubit fluorometric quantitation double-stranded DNA (dsDNA) assay kit (Thermo Fisher Scientific, Waltham, MA, USA) on a Clariostar microplate reader (BMG Labtech, Ortenberg, Germany). DNA quantity and quality were ascertained using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). DNA integrity (shearing/fragmentation) was measured on a 4200 TapeStation system (G2991AA; Agilent Technologies, Santa Clara, CA, USA). The eluted DNA samples were stored at 220°C.
Negative and positive DNA purification controls. To minimize laboratory contamination, sterile surgical gloves and face masks were used and all DNA extraction steps were performed with sterile or sterilized equipment under a class II laminar airflow cabinet. Negative isolation control (NIC) experiments were simultaneously conducted by substituting samples with PCR-grade water. Eluted NIC samples were used for V3-V4 PCR, and indexing was performed under DNA-free UV-sterilized AirClean PCR workstations/cabinets. At each PCR cleanup step of the library preparation, NIC amplicons were also validated on a 4200 Tape Station system (G2991AA; Agilent Technologies, Santa Clara, CA, USA) using Agilent D1000 ScreenTape (5067-5365) and Agilent genomic DNA (gDNA) reagents. Host background nucleic acid contamination was also monitored with real-time PCR using the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Sigma-Aldrich, Missouri, USA) forward primer 59-GTCTCCTCTGACTTCAACAGCG-39 and reverse primer 59-ACCACCCTGTTGCTGTAGCCAA-39 on eluted gDNAs (126). After completion of the PCR with 2Â KAPA HiFi HotStart ReadyMix, dual indexing of the samples with adaptor sequences (i7-N7xx-12 and i5-S5xx-8) was performed using the Illumina (San Diego, CA, USA) Nextera XT index kit (FC-131-1001/2). PCR cleanups and amplicon size selections were carried out with KAPA Pure Beads (KAPA Biosystems) based on the technical data sheet (KR1245-v3.16) of the manufacturer, resulting in final libraries with entries of ;580 to 630 bp. Every time, verifications were performed with PCR Agilent D1000 Screen Tape (5067-5582) and D1000 reagents (5067-5583). The 16S amplicon libraries for each sample were quantified with qPCR, normalized with respect to amplicon sizes, and pooled into a single library in equimolar quantities. Finally, 5 ml of a pooled 4 nM DNA library was used for sequencing on the Illumina MiSeq platform. The library pool was denatured with 0.2 M NaOH and diluted to 8 pM. Sequencing was carried out with a MiSeq reagent kit v3-618 cycle (MS-102-3003) following the manufacturer's protocols (Illumina, Inc., San Diego, CA, USA). Paired-end sequencing (2 Â 301 nucleotides [nt]) was performed on an Illumina MiSeq platform with a 5% PhiX spike-in quality control (PhiX control kit v3-FC-110-3001).
Sequence processing and analysis. Illumina BaseSpace software was used to demultiplex the paired-end reads and construct FASTQ files. The sequencing data were analyzed using Quantitative Insight Into Microbial Ecology (QIIME 2, v 2019.7) (127). Adaptor sequences (CTGTCTCTTATACACATCT) were found and trimmed from the 39 end of the reads with Cutadapt software integrated in the QIIME 2 pipeline. DADA2 software was used for quality trimming and filtering and for chimera removal. Sequences were clustered into amplicon sequencing variants (ASVs), with 97% similarity in sequences (128). The trimming parameters were set as follows: for the forward reads, 1 base was cropped from the start and the length was set to 300 bases; for the reverse reads, 9 bases were cropped from the start of the reads and the length was set to 223 bases.
Bioinformatic analyses. Multiple sequence alignment was performed with the MAFFT software (129), and reads were taxonomically classified using the naive Bayesian classifier trained on the SILVA (ver132) (130) reference database by selecting mapping points according to the forward-reverse primer set that was used for amplifying the 16S rRNA V3-V4 regions of the bacterial community (341F, 806R). Phylogenetic trees were constructed with the FastTree plugin (131). The QIIME2 pipeline was applied to perform alpha and beta diversity tests. For sample normalization, an 11,500 read depth was set. In the case of alpha diversity, Shannon's index (132), Faith's phylogenetic diversity index (133), Simpson evenness (134), and the Chao-1 index (135) were calculated in the QIIME2 pipeline. For beta diversity analysis, weighted/unweighted UniFrac distances (136) and Bray-Curtis dissimilarities (137) were measured. Alpha diversity differences were compared using the Kruskal-Wallis test. Beta diversity group significance was calculated with permutational multivariate analysis of variance (PERMANOVA) pseudo-F statistical test. These statistical tests were used to compare diversity between treatments; significance was P , 0.05. QIIME2 artifact files were exported from the pipeline and converted to TSV files that were used with different visualization packages. Heatmaps were generated in Python (ver3.6.5) with the Seaborn package (0.10.0); area and donut plots were constructed with pandas (0.25.3) and matplotlib (3.1.3) packages. Boxplots, violin plots, and line plots were constructed using GraphPad Prism statistical software. R (v 3.6.2) was used to visualize bubble plots and polar plots. A differential heat tree was created with the Metacoder R package (138). In the case of differential heat trees, differences were determined using a Wilcoxon rank sum test. LEfSe analysis was performed with bioBakery tools developed by the Huttenhower lab (139). Spearman correlation matrices were calculated and visualized with R statistical software using the corrplot package (https://github.com/taiyun/corrplot).
Data availability. All sequence data used in the analyses were deposited in the Sequence Read Archive (SRA) (http://www.ncbi.nlm.nih.gov/sra) under PRJNA633979. Sample IDs, metadata, and corresponding accession numbers are summarized in Fig. S1.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.

ACKNOWLEDGMENTS
We are grateful to Janos Olah and his coworkers. This study was financially supported by the Gazdaságfejlesztési és Innovációs Operatív Program (GINOP) GINOP-2.3.2-15-2016-00042 project of the Széchenyi 2020 Program given by the European Union and the Hungarian Government. M.P. designed the study plan, coordinated the study protocol, reviewed and interpreted the results, and drafted the manuscript. E.T. optimized the fecal DNA extraction protocols and performed the library preparations for next-generation sequencing (NGS). P.F. processed and analyzed the data with QIIME2 and prepared the figures. G.F. participated in the data analysis and performed the correlation and LEfSe analysis. E.T., P.F., and G.F. participated in the manuscript writing. J.R. invented the preparation of the nutraceuticals and synbiotics, and she also organized and coordinated the production of the dietary supplements. E.S. and G.P.-A. participated in the preparation of the nutraceuticals and measured the growth parameters of broilers. J.K., J.S., A.S., S.B., J.R., and L.B. reviewed the manuscript and conducted critical revision of the manuscript for important intellectual content. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.